Professional Documents
Culture Documents
of Nanotechnology
Springer Handbooks provide
a concise compilation of approved
key information on methods of
research, general principles, and
functional relationships in physi-
cal sciences and engineering. The
world’s leading experts in the
fields of physics and engineer-
ing will be assigned by one or
several renowned editors to write
the chapters comprising each vol-
ume. The content is selected by
these experts from Springer sources
(books, journals, online content)
and other systematic and approved
recent publications of physical and
technical information.
The volumes are designed to be
useful as readable desk reference
books to give a fast and comprehen-
sive overview and easy retrieval of
essential reliable key information,
including tables, graphs, and bibli-
ographies. References to extensive
sources are provided.
Springer
Handbook
of Nanotechnology
Bharat Bhushan (Ed.)
123
Editor
Professor Bharat Bhushan
Nanoprobe Laboratory
for Bio- and Nanotechnology and Biomimetics (NLB2 )
Ohio State University
201 W. 19th Avenue
Columbus, OH 43210-1142
USA
62/3180/YL 543210
V
In a January 2000 speech at the California Institute of can sense, control, and activate on
Technology, former President W.J. Clinton talked about the micro/nanoscale and can func-
the exciting promise of nanotechnology and the impor- tion individually or in arrays to gen-
tance of expanding research in nanoscale science and erate effects on the macroscale. Due
engineering and, more broadly, in the physical sciences. to the enabling nature of these sys-
Later that month, he announced in his State of the Union tems and the significant impact they
Address an ambitious US$ 497 million federal, multia- can have on both the commercial
gency national nanotechnology initiative (NNI) in the and defense applications, industry
fiscal year 2001 budget; and he made the NNI a top as well as the federal government Prof. Neal Lane
science and technology priority within a budget that em- have taken special interest in seeing Malcolm Gillis University
phasized increased investment in US scientific research. growth nurtured in this field. Micro- Professor,
Department of Physics
With strong bipartisan support in Congress, most of this and nanosystems are the next logical and Astronomy,
request was appropriated, and the NNI was born. Often, step in the silicon revolution. Senior Fellow,
James A. Baker III Institute
federal budget initiatives only last a year or so. It is most The discovery of novel mater- for Public Policy
encouraging that the NNI has remained a high priority ials, processes, and phenomena at Rice University
Houston, Texas
of the G.W. Bush Administration and Congress, reflect- the nanoscale and the development Served in the Clinton Admin-
ing enormous progress in the field and continued strong of new experimental and theoreti- istration as Assistant to the
interest and support by industry. cal techniques for research provide President for Science and Tech-
nology and Director of the White
Nanotechnology is the ability to manipulate indi- fresh opportunities for the develop- House Office of Science and
vidual atoms and molecules to produce nanostructured ment of innovative nanosystems and Technology Policy (1998–2001)
and, prior to that, as Director of
materials and submicron objects that have applica- nanostructured materials. There is the National Science Foundation
tions in the real world. Nanotechnology involves the an increasing need for a multidis- (1993–1998). While at the White
House, he was a key figure in
production and application of physical, chemical and ciplinary, systems-oriented approach the creation of the NNI.
biological systems at scales ranging from individual to manufacturing micro/nanodevices
atoms or molecules to about 100 nm, as well as the which function reliably. This can only be achieved
integration of the resulting nanostructures into larger through the cross-fertilization of ideas from different
systems. Nanotechnology is likely to have a profound disciplines and the systematic flow of information and
impact on our economy and society in the early 21st people among research groups.
century, perhaps comparable to that of information Nanotechnology is a broad, highly interdisciplinary,
technology or cellular and molecular biology. Science and still evolving field. Covering even the most im-
and engineering research in nanotechnology promises portant aspects of nanotechnology in a single book
breakthroughs in areas such as materials and manu- that reaches readers ranging from students to active
facturing, electronics, medicine and healthcare, energy researchers in academia and industry is an enormous
and the environment, biotechnology, information tech- challenge. To prepare such a wide-ranging book on
nology and national security. Clinical trials are already nanotechnology, Prof. Bhushan has harnessed his own
underway for nanomaterials that offer the promise of knowledge and experience, gained in several indus-
cures for certain cancers. It is widely felt that nanotech- tries and universities, and has assembled internationally
nology will be the next industrial revolution. recognized authorities from four continents to write
Nanometer-scale features are built up from their el- chapters covering a wide array of nanotechnology top-
emental constituents. Micro- and nanosystems compo- ics, including the latest advances. The authors come
nents are fabricated using batch-processing techniques from both academia and industry. The topics include
that are compatible with integrated circuits and range in major advances in many fields where nanoscale science
size from micro- to nanometers. Micro- and nanosys- and engineering is being pursued and illustrate how the
tems include micro/nanoelectro-mechanical systems field of nanotechnology has continued to emerge and
(MEMS/NEMS), micromechatronics, optoelectronics, blossom. Given the accelerating pace of discovery and
microfluidics and systems integration. These systems applications in nanotechnology, it is a challenge to cap-
VI
ture it all in one volume. As in earlier editions, professor third edition. It is increasingly important that scientists
Bhushan does an admirable job. and engineers, whatever their specialty, have a solid
Professor Bharat Bhushan’s comprehensive book grounding in the fundamentals and potential applica-
is intended to serve both as a textbook for university tions of nanotechnology. This third edition addresses
courses as well as a reference for researchers. The first that need by giving particular attention to the widening
and second editions were timely additions to the litera- audience of readers. It also includes a discussion of the
ture on nanotechnology and stimulated further interest social, ethical and political issues that tend to surround
in this important new field, while serving as invaluable any emerging technology.
resources to members of the international scientific and The editor and his team are to be warmly congrat-
industrial community. The increasing demand for up- ulated for bringing together this exclusive, timely, and
to-date information on this fast moving field led to this useful nanotechnology handbook.
VII
On December 29, 1959 at the California Institute of across five federal agencies (National Science Foun-
Technology, Nobel Laureate Richard P. Feynman gave dation (NSF), Department of Energy (DOE), NASA,
at talk at the Annual meeting of the American Physical National Institute of Standards and Technology (NIST),
Society that has become one of the 20th century clas- and Environmental Protection Agency (EPA)), estab-
sic science lectures, titled There’s Plenty of Room at lish interdisciplinary research centers, and accelerate
the Bottom. He presented a technological vision of ex- technology transfer into the private sector. In addition,
treme miniaturization in 1959, several years before the Department of Defense (DOD), Homeland Security,
word chip became part of the lexicon. He talked about Agriculture and Justice as well as the National Insti-
the problem of manipulating and controlling things on tutes of Health (NIH) also fund large R&D activities.
a small scale. Extrapolating from known physical laws, They currently account for more than one-third of the
Feynman envisioned a technology using the ultimate federal budget for nanotechnology.
toolbox of nature, building nanoobjects atom by atom European Union (EU) made nanosciences and nan-
or molecule by molecule. Since the 1980s, many in- otechnologies a priority in Sixth Framework Program
ventions and discoveries in fabrication of nanoobjects (FP6) in 2002 for a period of 2003–2006. They had
have been testament to his vision. In recognition of dedicated small funds in FP4 and FP5 before. FP6 was
this reality, National Science and Technology Council tailored to help better structure European research and
(NSTC) of the White House created the Interagency to cope with the strategic objectives set out in Lis-
Working Group on Nanoscience, Engineering and Tech- bon in 2000. Japan identified nanotechnology as one of
nology (IWGN) in 1998. In a January 2000 speech at its main research priorities in 2001. The funding lev-
the same institute, former President W.J. Clinton talked els increases sharply from US$ 400 million in 2001 to
about the exciting promise of nanotechnology and the around US$ 950 million in 2004. In 2003, South Ko-
importance of expanding research in nanoscale science rea embarked upon a ten-year program with around
and technology, more broadly. Later that month, he US$ 2 billion of public funding, and Taiwan has com-
announced in his State of the Union Address an am- mitted around US$ 600 million of public funding over
bitious US$ 497 million federal, multi-agency national six years. Singapore and China are also investing on
nanotechnology initiative (NNI) in the fiscal year 2001 a large scale. Russia is well funded as well.
budget, and made the NNI a top science and technol- Nanotechnology literally means any technology
ogy priority. The objective of this initiative was to form done on a nanoscale that has applications in the
a broad-based coalition in which the academe, the pri- real world. Nanotechnology encompasses production
vate sector, and local, state, and federal governments and application of physical, chemical and biological
work together to push the envelop of nanoscience and systems at scales, ranging from individual atoms or
nanoengineering to reap nanotechnology’s potential so- molecules to submicron dimensions, as well as the
cial and economic benefits. integration of the resulting nanostructures into larger
The funding in the US has continued to increase. systems. Nanotechnology is likely to have a pro-
In January 2003, the US senate introduced a bill to found impact on our economy and society in the
establish a National Nanotechnology Program. On De- early 21st century, comparable to that of semiconduc-
cember 3, 2003, President George W. Bush signed tor technology, information technology, or cellular and
into law the 21st Century Nanotechnology Research molecular biology. Science and technology research
and Development Act. The legislation put into law in nanotechnology promises breakthroughs in areas
programs and activities supported by the National such as materials and manufacturing, nanoelectronics,
Nanotechnology Initiative. The bill gave nanotechnol- medicine and healthcare, energy, biotechnology, infor-
ogy a permanent home in the federal government mation technology and national security. It is widely
and authorized US$ 3.7 billion to be spent in the four felt that nanotechnology will be the next industrial
year period beginning in October 2005, for nanotech- revolution.
nology initiatives at five federal agencies. The funds There is an increasing need for a multidisciplinary,
would provide grants to researchers, coordinate R&D system-oriented approach to design and manufactur-
X
ing of micro/nanodevices which function reliably. This national research labs, and industry, and from all over
can only be achieved through the cross-fertilization the world.
of ideas from different disciplines and the system- This handbook is intended for three types of read-
atic flow of information and people among research ers: graduate students of nanotechnology, researchers in
groups. Reliability is a critical technology for many academia and industry who are active or intend to be-
micro- and nanosystems and nanostructured materials. come active in this field, and practicing engineers and
A broad based handbook was needed, and the first scientists who have encountered a problem and hope
edition of Springer Handbook of Nanotechnology was to solve it as expeditiously as possible. The handbook
published in April 2004. It presented an overview of should serve as an excellent text for one or two semester
nanomaterial synthesis, micro/nanofabrication, micro- graduate courses in nanotechnology in mechanical en-
and nanocomponents and systems, scanning probe mi- gineering, materials science, applied physics, or applied
croscopy, reliability issues (including nanotribology chemistry.
and nanomechanics) for nanotechnology, and indus- We embarked on the development of third edition
trial applications. When the handbook went for sale in in June 2007, and we worked very hard to get all the
Europe, it was sold out in ten days. Reviews on the chapters to the publisher in a record time of about 12
handbook were very flattering. months. I wish to sincerely thank the authors for offer-
Given the explosive growth in nanoscience and ing to write comprehensive chapters on a tight schedule.
nanotechnology, the publisher and the editor decided This is generally an added responsibility in the hec-
to develop a second edition after merely six months tic work schedules of researchers today. I depended on
of publication of the first edition. The second edition a large number of reviewers who provided critical re-
(2007) came out in December 2006. The publisher and views. I would like to thank Dr. Phillip J. Bond, Chief of
the editor again decided to develop a third edition af- Staff and Under Secretary for Technology, US Depart-
ter six month of publication of the second edition. This ment of Commerce, Washington, D.C. for suggestions
edition of the handbook integrates the knowledge from for chapters as well as authors in the handbook. Last but
nanostructures, fabrication, materials science, devices, not the least, I would like to thank my secretary Cate-
and reliability point of view. It covers various industrial rina Runyon-Spears for various administrative duties
applications. It also addresses social, ethical, and polit- and her tireless efforts are highly appreciated.
ical issues. Given the significant interest in biomedical I hope that this handbook will stimulate further in-
applications, and biomimetics a number of additional terest in this important new field, and the readers of this
chapters in this arena have been added. The third edi- handbook will find it useful.
tion consists of 53 chapters (new 10, revised 28, and as
is 15). The chapters have been written by 139 interna- February 2010 Bharat Bhushan
tionally recognized experts in the field, from academia, Editor
XI
On 29 December 1959 at the California Institute of agencies (National Science Foundation (NSF), Depart-
Technology, Nobel Laureate Richard P. Feynman gave ment of Energy (DOE), NASA, National Institute of
at talk at the Annual meeting of the American Physical Standards and Technology (NIST), and Environmental
Society that has become one of the 20th century clas- Protection Agency (EPA)), establish interdisciplinary
sic science lectures, titled “There’s Plenty of Room at research centers, and accelerate technology transfer into
the Bottom.” He presented a technological vision of ex- the private sector. In addition, Department of Defense
treme miniaturization in 1959, several years before the (DOD), Homeland Security, Agriculture and Justice as
word “chip” became part of the lexicon. He talked about well as the National Institutes of Health (NIH) would
the problem of manipulating and controlling things on also fund large R&D activities. They currently account
a small scale. Extrapolating from known physical laws, for more than one-third of the federal budget for nano-
Feynman envisioned a technology using the ultimate technology.
toolbox of nature, building nanoobjects atom by atom The European Union made nanosciences and nan-
or molecule by molecule. Since the 1980s, many inven- otechnologies a priority in the Sixth Framework Pro-
tions and discoveries in the fabrication of nanoobjects gram (FP6) in 2002 for the period of 2003-2006. They
have been a testament to his vision. In recognition of had dedicated small funds in FP4 and FP5 before. FP6
this reality, the National Science and Technology Coun- was tailored to help better structure European research
cil (NSTC) of the White House created the Interagency and to cope with the strategic objectives set out in Lis-
Working Group on Nanoscience, Engineering and Tech- bon in 2000. Japan identified nanotechnology as one of
nology (IWGN) in 1998. In a January 2000 speech at its main research priorities in 2001. The funding levels
the same institute, former President W. J. Clinton talked increased sharply from $400 million in 2001 to around
about the exciting promise of “nanotechnology” and $950 million in 2004. In 2003, South Korea embarked
the importance of expanding research in nanoscale sci- upon a ten-year program with around $2 billion of pub-
ence and, more broadly, technology. Later that month, lic funding, and Taiwan has committed around $600
he announced in his State of the Union Address an million of public funding over six years. Singapore and
ambitious $497 million federal, multiagency national China are also investing on a large scale. Russia is well
nanotechnology initiative (NNI) in the fiscal year 2001 funded as well.
budget, and made the NNI a top science and technol- Nanotechnology literally means any technology
ogy priority. The objective of this initiative was to form done on a nanoscale that has applications in the
a broad-based coalition in which the academe, the pri- real world. Nanotechnology encompasses production
vate sector, and local, state, and federal governments and application of physical, chemical and biological
work together to push the envelope of nanoscience and systems at scales, ranging from individual atoms or
nanoengineering to reap nanotechnology’s potential so- molecules to submicron dimensions, as well as the
cial and economic benefits. integration of the resulting nanostructures into larger
The funding in the U.S. has continued to increase. systems. Nanotechnology is likely to have a pro-
In January 2003, the U. S. senate introduced a bill to found impact on our economy and society in the
establish a National Nanotechnology Program. On 3 early 21st century, comparable to that of semiconduc-
December 2003, President George W. Bush signed into tor technology, information technology, or cellular and
law the 21st Century Nanotechnology Research and De- molecular biology. Science and technology research
velopment Act. The legislation put into law programs in nanotechnology promises breakthroughs in areas
and activities supported by the National Nanotechnol- such as materials and manufacturing, nanoelectronics,
ogy Initiative. The bill gave nanotechnology a perma- medicine and healthcare, energy, biotechnology, infor-
nent home in the federal government and authorized mation technology and national security. It is widely
$3.7 billion to be spent in the four year period begin- felt that nanotechnology will be the next industrial
ning in October 2005, for nanotechnology initiatives at revolution.
five federal agencies. The funds would provide grants There is an increasing need for a multidisciplinary,
to researchers, coordinate R&D across five federal system-oriented approach to design and manufactur-
XII
ing of micro/nanodevices that function reliably. This academia and industry who are active or intend to be-
can only be achieved through the cross-fertilization come active in this field, and practicing engineers and
of ideas from different disciplines and the system- scientists who have encountered a problem and hope
atic flow of information and people among research to solve it as expeditiously as possible. The handbook
groups. Reliability is a critical technology for many should serve as an excellent text for one or two semester
micro- and nanosystems and nanostructured materials. graduate courses in nanotechnology in mechanical en-
A broad-based handbook was needed, and thus the first gineering, materials science, applied physics, or applied
edition of Springer Handbook of Nanotechnology was chemistry.
published in April 2004. It presented an overview of We embarked on the development of the second edi-
nanomaterial synthesis, micro/nanofabrication, micro- tion in October 2004, and we worked very hard to get all
and nanocomponents and systems, scanning probe mi- the chapters to the publisher in a record time of about 7
croscopy, reliability issues (including nanotribology months. I wish to sincerely thank the authors for offer-
and nanomechanics) for nanotechnology, and industrial ing to write comprehensive chapters on a tight schedule.
applications. When the handbook went for sale in Eu- This is generally an added responsibility to the hec-
rope, it sold out in ten days. Reviews on the handbook tic work schedules of researchers today. I depended
were very flattering. on a large number of reviewers who provided criti-
Given the explosive growth in nanoscience and cal reviews. I would like to thank Dr. Phillip J. Bond,
nanotechnology, the publisher and the editor decided to Chief of Staff and Under Secretary for Technology, US
develop a second edition merely six months after publi- Department of Commerce, Washington, D.C. for chap-
cation of the first edition. This edition of the handbook ter suggestions as well as authors in the handbook. I
integrates the knowledge from the nanostructure, fabri- would also like to thank my colleague, Dr. Zhenhua Tao,
cation, materials science, devices, and reliability point whose efforts during the preparation of this handbook
of view. It covers various industrial applications. It also were very useful. Last but not the least, I would like
addresses social, ethical, and political issues. Given the to thank my secretary Caterina Runyon-Spears for vari-
significant interest in biomedical applications, a number ous administrative duties; her tireless efforts are highly
of chapters in this arena have been added. The sec- appreciated.
ond edition consists of 59 chapters (new: 23; revised: I hope that this handbook will stimulate further in-
27; unchanged: 9). The chapters have been written by terest in this important new field, and the readers of this
154 internationally recognized experts in the field, from handbook will find it useful.
academia, national research labs, and industry.
This book is intended for three types of readers: May 2005 Bharat Bhushan
graduate students of nanotechnology, researchers in Editor
XIII
On December 29, 1959 at the California Institute of that nanotechnology will lead to the next industrial
Technology, Nobel Laureate Richard P. Feynman gave revolution.
a talk at the Annual meeting of the American Physic- Reliability is a critical technology for many micro-
al Society that has become one classic science lecture and nanosystems and nanostructured materials. No
of the 20th century, titled “There’s Plenty of Room book exists on this emerging field. A broad based
at the Bottom.” He presented a technological vision handbook is needed. The purpose of this handbook
of extreme miniaturization in 1959, several years be- is to present an overview of nanomaterial synthe-
fore the word “chip” became part of the lexicon. He sis, micro/nanofabrication, micro- and nanocomponents
talked about the problem of manipulating and con- and systems, reliability issues (including nanotribology
trolling things on a small scale. Extrapolating from and nanomechanics) for nanotechnology, and indus-
known physical laws, Feynman envisioned a technology trial applications. The chapters have been written by
using the ultimate toolbox of nature, building nanoob- internationally recognized experts in the field, from
jects atom by atom or molecule by molecule. Since academia, national research labs and industry from all
the 1980s, many inventions and discoveries in fabri- over the world.
cation of nanoobjects have been a testament to his The handbook integrates knowledge from the fab-
vision. In recognition of this reality, in a January 2000 rication, mechanics, materials science and reliability
speech at the same institute, former President W. J. points of view. This book is intended for three types
Clinton talked about the exciting promise of “nanotech- of readers: graduate students of nanotechnology, re-
nology” and the importance of expanding research in searchers in academia and industry who are active or
nanoscale science and engineering. Later that month, intend to become active in this field, and practicing
he announced in his State of the Union Address an engineers and scientists who have encountered a prob-
ambitious $ 497 million federal, multi-agency national lem and hope to solve it as expeditiously as possible.
nanotechnology initiative (NNI) in the fiscal year 2001 The handbook should serve as an excellent text for one
budget, and made the NNI a top science and technology or two semester graduate courses in nanotechnology
priority. Nanotechnology literally means any technol- in mechanical engineering, materials science, applied
ogy done on a nanoscale that has applications in the physics, or applied chemistry.
real world. Nanotechnology encompasses production We embarked on this project in February 2002, and
and application of physical, chemical and biological we worked very hard to get all the chapters to the
systems at size scales, ranging from individual atoms publisher in a record time of about 1 year. I wish to
or molecules to submicron dimensions as well as the sincerely thank the authors for offering to write compre-
integration of the resulting nanostructures into larger hensive chapters on a tight schedule. This is generally
systems. Nanofabrication methods include the manipu- an added responsibility in the hectic work schedules of
lation or self-assembly of individual atoms, molecules, researchers today. I depended on a large number of re-
or molecular structures to produce nanostructured ma- viewers who provided critical reviews. I would like to
terials and sub-micron devices. Micro- and nanosystems thank Dr. Phillip J. Bond, Chief of Staff and Under Sec-
components are fabricated using top-down lithographic retary for Technology, US Department of Commerce,
and nonlithographic fabrication techniques. Nanotech- Washington, D.C. for suggestions for chapters as well
nology will have a profound impact on our economy as authors in the handbook. I would also like to thank
and society in the early 21st century, comparable to my colleague, Dr. Huiwen Liu, whose efforts during the
that of semiconductor technology, information technol- preparation of this handbook were very useful.
ogy, or advances in cellular and molecular biology. I hope that this handbook will stimulate further in-
The research and development in nanotechnology will terest in this important new field, and the readers of this
lead to potential breakthroughs in areas such as ma- handbook will find it useful.
terials and manufacturing, nanoelectronics, medicine
and healthcare, energy, biotechnology, information September 2003 Bharat Bhushan
technology and national security. It is widely felt Editor
XV
Editors Vita
Dr. Bharat Bhushan received an M.S. in mechanical en- chanics of Magnetic Storage Sys-
gineering from the Massachusetts Institute of Technol- tems in 1984 and the first inter-
ogy in 1971, an M.S. in mechanics and a Ph.D. in me- national symposium on Advances
chanical engineering from the University of Colorado at in Information Storage Systems in
Boulder in 1973 and 1976, respectively, an MBA from 1990, both of which are now held
Rensselaer Polytechnic Institute at Troy, NY in 1980, annually. He is the founder of an
Doctor Technicae from the University of Trondheim at ASME Information Storage and Pro-
Trondheim, Norway in 1990, a Doctor of Technical Sci- cessing Systems Division founded
ences from the Warsaw University of Technology at in 1993 and served as the found-
Warsaw, Poland in 1996, and Doctor Honouris Causa ing chair during 1993–1998. His biography has been
from the National Academy of Sciences at Gomel, listed in over two dozen Who’s Who books in-
Belarus in 2000. He is a registered professional en- cluding Who’s Who in the World and has received
gineer. He is presently an Ohio Eminent Scholar and more than two dozen awards for his contributions to
The Howard D. Winbigler Professor in the College of science and technology from professional societies,
Engineering, and the Director of the Nanoprobe Labo- industry, and US government agencies. He is also
ratory for Bio- and Nanotechnology and Biomimetics the recipient of various international fellowships in-
(NLB²) at the Ohio State University, Columbus, Ohio. cluding the Alexander von Humboldt Research Prize
His research interests include fundamental studies with for Senior Scientists, Max Planck Foundation Re-
a focus on scanning probe techniques in the interdisci- search Award for Outstanding Foreign Scientists, and
plinary areas of bio/nanotribology, bio/nanomechanics the Fulbright Senior Scholar Award. He is a foreign
and bio/nanomaterials characterization, and applica- member of the International Academy of Engineer-
tions to bio/nanotechnology and biomimetics. He is an ing (Russia), Byelorussian Academy of Engineering
internationally recognized expert of bio/nanotribology and Technology and the Academy of Triboengineer-
and bio/nanomechanics using scanning probe mi- ing of Ukraine, an honorary member of the Society
croscopy, and is one of the most prolific authors. He is of Tribologists of Belarus, a fellow of ASME, IEEE,
considered by some a pioneer of the tribology and me- STLE, and the New York Academy of Sciences,
chanics of magnetic storage devices. He has authored 6 and a member of ASEE, Sigma Xi and Tau Beta
scientific books, more than 90 handbook chapters, more Pi.
than 700 scientific papers (h factor – 45+; ISI Highly Dr. Bhushan has previously worked for the R&D
Cited in Materials Science, since 2007), and more than Division of Mechanical Technology Inc., Latham, NY;
60 technical reports, edited more than 45 books, and the Technology Services Division of SKF Industries
holds 17 US and foreign patents. He is co-editor of Inc., King of Prussia, PA; the General Products Divi-
Springer NanoScience and Technology Series and co- sion Laboratory of IBM Corporation, Tucson, AZ; and
editor of Microsystem Technologies. He has given more the Almaden Research Center of IBM Corporation, San
than 400 invited presentations on six continents and Jose, CA. He has held visiting professor appointments
more than 140 keynote/plenary addresses at major in- at University of California at Berkeley, University of
ternational conferences. Cambridge, UK, Technical University Vienna, Aus-
Dr. Bhushan is an accomplished organizer. He or- tria, University of Paris, Orsay, ETH Zurich and EPFL
ganized the first symposium on Tribology and Me- Lausanne.
XVII
List of Authors
Ruti Kapon
Kerstin Koch
Weizmann Institute of Science
Rhine-Waal University of Applied Science
Department of Biological Chemistry
Department of Life Science, Biology
Rehovot 76100, Israel
and Nanobiotechnology
e-mail: ruti.kapon@weizmann.ac.il
Landwehr 4
47533 Kleve, Germany
Josef Käs
e-mail: kerstin.koch@hochschule.rhein-waal.de
University of Leipzig
Institute of Experimental Physics I
Division of Soft Matter Physics Jing Kong
Linnéstr. 5 Massachusetts Institute of Technology
04103 Leipzig, Germany Department of Electrical Engineering
e-mail: jkaes@physik.uni-leipzig.de and Computer Science
Cambridge, MA, USA
Horacio Kido e-mail: jingkong@mit.edu
University of California at Irvine
Mechanical and Aerospace Engineering Tobias Kraus
Irvine, CA, USA Leibniz-Institut für Neue Materialien gGmbH
e-mail: hkido@uci.edu Campus D2 2
66123 Saarbrücken, Germany
Tobias Kießling e-mail: tobias.kraus@inm-gmbh.de
University of Leipzig
Institute of Experimental Physics I Anders Kristensen
Division of Soft Matter Physics Technical University of Denmark
Linnéstr. 5 DTU Nanotech
04103 Leipzig, Germany 2800 Kongens Lyngby, Denmark
e-mail: Tobias.Kiessling@uni-leipzig.de e-mail: anders.kristensen@nanotech.dtu.dk
XXII List of Authors
Christophe Laurent
Université Paul Sabatier Othmar Marti
CIRIMAT UMR 5085 CNRS Ulm University
118 Route de Narbonne Institute of Experimental Physics
31062 Toulouse, France Albert-Einstein-Allee 11
e-mail: laurent@chimie.ups-tlse.fr 89069 Ulm, Germany
e-mail: othmar.marti@uni-ulm.de
Abraham P. Lee
University of California Irvine Jack Martin
Department of Biomedical Engineering 66 Summer Street
Department of Mechanical Foxborough, MA 02035, USA
and Aerospace Engineering e-mail: jack.martin@alumni.tufts.edu
Irvine, CA 92697, USA
e-mail: aplee@uci.edu
Shinji Matsui
Stephen C. Lee University of Hyogo
Ohio State University Laboratory of Advanced Science
Biomedical Engineering Center and Technology for Industry
Columbus, OH 43210, USA Hyogo, Japan
e-mail: lee@bme.ohio-state.edu e-mail: matsui@lasti.u-hyogo.ac.jp
List of Authors XXIII
Ozgur Sahin
Oded Rabin
The Rowland Institute at Harvard
University of Maryland
100 Edwin H. Land Blvd
Department of Materials Science and Engineering
Cambridge, MA 02142, USA
College Park, MD, USA
e-mail: sahin@rowland.harvard.edu
e-mail: oded@umd.edu
Akira Sasahara
Françisco M. Raymo Japan Advanced Institute
University of Miami of Science and Technology
Department of Chemistry School of Materials Science
1301 Memorial Drive 1-1 Asahidai
Coral Gables, FL 33146-0431, USA 923-1292 Nomi, Japan
e-mail: fraymo@miami.edu e-mail: sasahara@jaist.ac.jp
Helmut Schift
Manitra Razafinimanana
Paul Scherrer Institute
University of Toulouse III (Paul Sabatier)
Laboratory for Micro- and Nanotechnology
Centre de Physique des Plasmas
5232 Villigen PSI, Switzerland
et leurs Applications (CPPAT)
e-mail: helmut.schift@psi.ch
Toulouse, France
e-mail: razafinimanana@cpat.ups-tlse.fr André Schirmeisen
University of Münster
Ziv Reich Institute of Physics
Weizmann Institute of Science Ha’Nesi Ha’Rishon Wilhelm-Klemm-Str. 10
Department of Biological Chemistry 48149 Münster, Germany
Rehovot 76100, Israel e-mail: schirmeisen@uni-muenster.de
e-mail: ziv.reich@weizmann.ac.il
Christian Schulze
Beiersdorf AG
John A. Rogers
Research & Development
University of Illinois
Unnastr. 48
Department of Materials Science and Engineering
20245 Hamburg, Germany
Urbana, IL, USA
e-mail: christian.schulze@beiersdorf.com;
e-mail: jrogers@uiuc.edu
christian.schulze@uni-leipzig.de
Olav Solgaard
Stanford University Benjamin Sullivan
E.L. Ginzton Laboratory TearLab Corp.
450 Via Palou 11025 Roselle Street
Stanford, CA 94305-4088, USA San Diego, CA 92121, USA
e-mail: solgaard@stanford.edu e-mail: bdsulliv@TearLab.com
Dan Strehle
University of Leipzig Paul Swanson
Institute of Experimental Physics I Nexogen, Inc.
Division of Soft Matter Physics Engineering
Linnéstr. 5 8360 C Camino Santa Fe
04103 Leipzig, Germany San Diego, CA 92121, USA
e-mail: dan.strehle@uni-leipzig.de e-mail: pswanson@nexogentech.com
XXVI List of Authors
Contents
1 Introduction to Nanotechnology
Bharat Bhushan ...................................................................................... 1
1.1 Nanotechnology – Definition and Examples ................................... 1
1.2 Background and Research Expenditures ......................................... 4
1.3 Lessons from Nature (Biomimetics)................................................. 6
1.4 Applications in Different Fields ...................................................... 9
1.5 Various Issues ............................................................................... 10
1.6 Research Training .......................................................................... 11
1.7 Organization of the Handbook ....................................................... 11
References .............................................................................................. 12
4 Nanowires
Mildred S. Dresselhaus, Yu-Ming Lin, Oded Rabin, Marcie R. Black,
Jing Kong, Gene Dresselhaus .................................................................... 119
4.1 Synthesis ...................................................................................... 121
4.2 Characterization and Physical Properties of Nanowires .................... 130
4.3 Applications .................................................................................. 152
4.4 Concluding Remarks ...................................................................... 159
References .............................................................................................. 159
8 Introduction to Micro-/Nanofabrication
Babak Ziaie, Antonio Baldi, Massood Z. Atashbar ...................................... 231
8.1 Basic Microfabrication Techniques.................................................. 232
8.2 MEMS Fabrication Techniques......................................................... 244
8.3 Nanofabrication Techniques .......................................................... 256
8.4 Summary and Conclusions ............................................................. 265
References .............................................................................................. 265
Contents XXIX
35 Cellular Nanomechanics
Roger Kamm, Jan Lammerding, Mohammad Mofrad ................................. 1171
35.1 Overview....................................................................................... 1171
35.2 Structural Components of a Cell...................................................... 1173
35.3 Experimental Methods................................................................... 1179
35.4 Theoretical and Computational Descriptions ................................... 1185
35.5 Mechanics of Subcellular Structures ................................................ 1188
35.6 Current Understanding and Future Needs ....................................... 1196
References .............................................................................................. 1196
Contents XXXV
Part F Biomimetics
41 Multifunctional Plant Surfaces and Smart Materials
Kerstin Koch, Bharat Bhushan, Wilhelm Barthlott ..................................... 1399
41.1 The Architecture of Plant Surfaces .................................................. 1402
41.2 Multifunctional Plant Surfaces ....................................................... 1417
41.3 Technical Uses of Superhydrophobicity ........................................... 1426
41.4 Conclusions ................................................................................... 1430
References .............................................................................................. 1431
45 The Millipede –
A Nanotechnology-Based AFM Data-Storage System
Gerd K. Binnig, Giovanni Cherubini, Michel Despont, Urs T. Dürig,
Evangelos Eleftheriou, Haralampos Pozidis, Peter Vettiger ......................... 1601
45.1 The Millipede Concept ................................................................... 1603
45.2 Thermomechanical AFM Data Storage ............................................. 1604
45.3 Array Design, Technology, and Fabrication ..................................... 1606
45.4 Array Characterization ................................................................... 1607
45.5 Three-Terminal Cantilever Design................................................... 1609
45.6 x,y,z Medium Microscanner ........................................................... 1610
45.7 First Write/Read Results with the 32×32 Array Chip........................... 1613
45.8 Polymer Medium ........................................................................... 1614
45.9 Read Channel Model...................................................................... 1621
45.10 System Aspects .............................................................................. 1624
45.11 Conclusions ................................................................................... 1629
References .............................................................................................. 1630
46 Nanorobotics
Bradley J. Nelson, Lixin Dong ................................................................... 1633
46.1 Overview of Nanorobotics .............................................................. 1634
46.2 Actuation at Nanoscales ................................................................ 1635
46.3 Nanorobotic Manipulation Systems ................................................ 1637
XXXVIII Contents
List of Abbreviations
K MC microcapillary
MCM multi-chip module
KASH Klarsicht, ANC-1, Syne Homology MD molecular dynamics
KPFM Kelvin probe force microscopy ME metal-evaporated
MEMS microelectromechanical system
MExFM magnetic exchange force microscopy
L MFM magnetic field microscopy
MFM magnetic force microscope
LA lauric acid MFM magnetic force microscopy
LAR low aspect ratio MHD magnetohydrodynamic
LB Langmuir–Blodgett MIM metal–insulator–metal
LBL layer-by-layer MIMIC micromolding in capillaries
LCC leadless chip carrier MLE maximum likelihood estimator
LCD liquid-crystal display MOCVD metalorganic chemical vapor deposition
LCoS liquid crystal on silicon MOEMS microoptoelectromechanical system
LCP liquid-crystal polymer MOS metal–oxide–semiconductor
LDL low-density lipoprotein MOSFET metal–oxide–semiconductor field-effect
LDOS local density of states transistor
LED light-emitting diode MP metal particle
LFA-1 leukocyte function-associated antigen-1 MPTMS mercaptopropyltrimethoxysilane
LFM lateral force microscope MRFM magnetic resonance force microscopy
LFM lateral force microscopy MRFM molecular recognition force microscopy
LIGA Lithographie Galvanoformung MRI magnetic resonance imaging
Abformung MRP molecular recognition phase
LJ Lennard-Jones MscL mechanosensitive channel of large
LMD laser microdissection conductance
LMPC laser microdissection and pressure MST microsystem technology
catapulting MT microtubule
LN liquid-nitrogen mTAS micro total analysis system
LoD limit-of-detection MTTF mean time to failure
LOR lift-off resist MUMP multiuser MEMS process
LPC laser pressure catapulting MVD molecular vapor deposition
LPCVD low-pressure chemical vapor deposition MWCNT multiwall carbon nanotube
LSC laser scanning cytometry MWNT multiwall nanotube
LSN low-stress silicon nitride MYD/BHW Muller–Yushchenko–Derjaguin/Burgess–
LT-SFM low-temperature scanning force Hughes–White
microscope
LT-SPM low-temperature scanning probe
microscopy
LT-STM low-temperature scanning tunneling N
microscope
LT low-temperature NA numerical aperture
LTM laser tracking microrheology NADIS nanoscale dispensing
LTO low-temperature oxide NASA National Aeronautics and Space
LTRS laser tweezers Raman spectroscopy Administration
LUMO lowest unoccupied molecular orbital NC-AFM noncontact atomic force microscopy
LVDT linear variable differential transformer NEMS nanoelectromechanical system
NGL next-generation lithography
NHS N-hydroxysuccinimidyl
M NIH National Institute of Health
NIL nanoimprint lithography
MALDI matrix assisted laser desorption ionization NIST National Institute of Standards and
MAP manifold absolute pressure Technology
MAPK mitogen-activated protein kinase NMP no-moving-part
MAPL molecular assembly patterning by lift-off NMR nuclear magnetic resonance
MBE molecular-beam epitaxy NMR nuclear mass resonance
MC microcantilever NNI National Nanotechnology Initiative
List of Abbreviations XLV
TMS trimethylsilyl V
TNT trinitrotoluene
TP track pitch VBS vinculin binding site
TPE-FCCS two-photon excitation fluorescence VCO voltage-controlled oscillator
cross-correlation spectroscopy VCSEL vertical-cavity surface-emitting laser
TPI threads per inch vdW van der Waals
TPMS tire pressure monitoring system VHH variable heavy–heavy
TR torsional resonance VLSI very large-scale integration
TREC topography and recognition VOC volatile organic compound
TRIM transport of ions in matter VPE vapor-phase epitaxy
TSDC thermally stimulated depolarization VSC vehicle stability control
current
TTF tetrathiafulvalene X
TV television
XPS x-ray photon spectroscopy
U XRD x-ray powder diffraction
UAA unnatural AA Y
UHV ultrahigh vacuum
ULSI ultralarge-scale integration YFP yellow fluorescent protein
UML unified modeling language
UNCD ultrananocrystalline diamond Z
UV ultraviolet
UVA ultraviolet A Z-DOL perfluoropolyether
1
Introduction
1. Introduction to Nanotechnology
Introduction
Bharat Bhushan
1.1 Nanotechnology –
A biological system can be exceedingly small. Definition and Examples ....................... 1
Many of the cells are very tiny, but they are
1.2 Background and Research Expenditures . 4
very active; they manufacture various sub-
stances; they walk around; they wiggle; and 1.3 Lessons from Nature (Biomimetics)......... 6
they do all kinds of marvelous things – all 1.4 Applications in Different Fields .............. 9
on a very small scale. Also, they store in-
formation. Consider the possibility that we 1.5 Various Issues ...................................... 10
too can make a thing very small which 1.6 Research Training ................................. 11
does what we want – that we can man-
ufacture an object that maneuvers at that 1.7 Organization of the Handbook............... 11
level. References .................................................. 12
(From the talk There’s Plenty of Room at the
Bottom, delivered by Richard P. Feynman at the at the California Institute of Technology; Pasadena,
annual meeting of the American Physical Society December 29, 1959).
impetus for the development of experimental and com- changes occur, until finally when the size drops be-
Introduction
putational tools. low 100 nm, dramatic changes in properties can occur.
The discovery of novel materials, processes, and If only one length of a three-dimensional nanostruc-
phenomena at the nanoscale and the development of ture is of nanodimension, the structure is referred to
new experimental and theoretical techniques for re- as a quantum well; if two sides are of nanometer
search provide fresh opportunities for the development length, the structure is referred to as a quantum wire.
of innovative nanosystems and nanostructured mater- A quantum dot has all three dimensions in the nano
ials. The properties of materials at the nanoscale can range. The term quantum is associated with these three
be very different from those at a larger scale. When types of nanostructures because the changes in prop-
the dimension of a material is reduced from a large erties arise from the quantum-mechanical nature of
size, the properties remain the same at first, then small physics in the domain of the ultrasmall. Materials can
DMD 12 µm
500 nm
Quantum-dots transistor
300 nm Red blood cell
8 µm
Molecular gear
10–100 nm
A
SWNT
1 µm
SWCNT chemical sensor
2 nm
C atom DNA
0.16 nm 2.5 nm
0.1 1 10 100 1000 10 000 100 000
Size (nm)
Fig. 1.1 Dimensions of MEMS/NEMS and BioNEMS in perspective. Examples shown are a single-walled carbon
nanotube (SWCNT) chemical sensor [1.1], molecular dynamic simulations of carbon-nanotube-based gears [1.2],
quantum-dot transistor obtained from [1.3], and digital microdevice (DMD) obtained from www.dlp.com. For com-
parison, dimensions and weights of various biological objects found in nature are also presented
Introduction to Nanotechnology 1.1 Nanotechnology – Definition and Examples 3
Table 1.1 Characteristic dimensions and weights in per- MEMS/NEMS (RF-MEMS/RF-NEMS). MEMS/
Introduction
spective NEMS for biological applications are referred to as
Characteristic dimensions in perspective
bioMEMS/bioNEMS.
To put the dimensions of MEMS/NEMS and
NEMS characteristic length < 100 nm
BioNEMS in perspective, see Fig. 1.1 and Table 1.1. In-
MEMS characteristic length < 1 mm and > 100 nm
dividual atoms are typically a fraction of a nanometer
SWCNT chemical sensor ≈ 2 nm
in diameter, DNA molecules are about 2.5 nm wide, bi-
Molecular gear ≈ 10 nm
ological cells are in the range of thousands of nm in
Quantum-dot transistor 300 nm
diameter, and human hair is about 75 μm in diameter.
Digital micromirror 12 000 nm
The smallest length of BioNEMS shown in the figure is
Individual atoms Typically a fraction of a nm about 2 nm, NEMS ranges in size from 10 to 300 nm,
in diameter
and the size of MEMS is 12 000 nm. The mass of a mi-
DNA molecules ≈ 2.5 nm wide
cromachined silicon structure can be a low as 1 nN, and
Biological cells In the range of thousands
of nm in diameter NEMS can be built with mass as low as 10−20 N with
Human hair ≈ 75 000 nm in diameter cross-sections of about 10 nm. In comparison, the mass
Weight in perspective
NEMS built with cross- As low as 10−20 N a) Global MEMS market segment
sections of about 10 nm US $ (billions)
Micromachine silicon As low as 1 nN
Microbolometers
structure 8 RF MEMS
Eyelash ≈ 100 nN Bio- and microfluidics components
Water droplet ≈ 10 μN 7 DLP (micromirrors)
MOEMS
6 Gyroscopes
Accelerometers
be nanostructured for new properties and novel perfor- 5
Pressure sensors
Inkjet head
mance. This field is opening new avenues in science and
technology. 4
Micro- and nanosystems include micro/nanoelec-
tromechanical systems (MEMS/NEMS). MEMS refers 3
to microscopic devices that have a characteristic length
2
of less than 1 mm but more than 100 nm and that
combine electrical and mechanical components. NEMS 1
refers to nanoscopic devices that have a character-
istic length of less than 100 nm and that combine 0
2003 2004 2005 2006 2007
electrical and mechanical components. In mesoscale de- Year
vices, if the functional components are on the micro- b) Global nanotechnology market segment
or nanoscale, they may be referred to as MEMS or US $ (billions)
NEMS, respectively. These are referred to as intelligent
30
miniaturized systems, comprising sensing, processing, Nanodevices
and/or actuating functions and combining electrical and Nanotools
mechanical components. The acronym MEMS origi- 20
nated in the USA. The term commonly used in Europe
is microsystem technology (MST), and in Japan the
term micromachines is used. Another term generally 10 Nanomaterials
used is micro/nanodevices. The terms MEMS/NEMS
are also now used in a broad sense and include
electrical, mechanical, fluidic, optical, and/or biolog- 0
ical function. MEMS/NEMS for optical applications 2002 2003 2004 2005 2006 2007 2008
Year
are referred to as micro/nanooptoelectromechanical
systems (MOEMS/NOEMS). MEMS/NEMS for elec- Fig. 1.2 Global MEMS and nanotechnology market seg-
tronic applications are referred to as radiofrequency ments (DLP – digital light processing)
4 Introduction
of a drop of water is about 10 μN, and the mass of an may slow down and that of nonsilicon MEMS may
Introduction
eyelash is about 100 nN. pick up during the next decade. It is expected to ex-
MEMS/NEMS and BioMEMS/BioNEMS are ex- pand in this decade, for nanomaterials and biomedical
pected to have a major impact on our lives, com- applications as well as nanoelectronics or molecu-
parable to that of semiconductor technology, infor- lar electronics. For example, miniaturized diagnostics
mation technology, or cellular and molecular biol- could be implanted for early diagnosis of illness. Tar-
ogy [1.4, 5]. MEMS/NEMS and BioMEMS/BioNEMS geted drug-delivery devices are under development.
are used in electromechanical, electronics, informa- Due to the enabling nature of these systems and be-
tion/communication, chemical, and biological applica- cause of the significant impact they can have on both
tions. The MEMS industry in 2004 was worth about commercial and defense applications, industry as well
US$ 4.5 billion, with a projected annual growth rate of as federal governments have taken special interest in
17% (Fig. 1.2) [1.6]. The NEMS industry was worth seeing growth in this field nurtured. MEMS/NEMS and
about US$ 10 billion in 2004, mostly in nanomateri- BioMEMS/BioNEMS are the next logical step in the
als (Fig. 1.2) [1.7]. Growth of Si-based MEMS/NEMS silicon revolution.
Introduction
In January 2003, the US Senate introduced a bill to a) Worldwide publications
in nanotechnology
establish a National Nanotechnology Program. On De- (1997– 1999) Europe
cember 3, 2003, President George W. Bush signed 34%
into law the 21st Century Nanotechnology Research
Others USA and
and Development Act. This legislation put into law 5% Canada
programs and activities supported by the National 28%
Nanotechnology Initiative. The bill gave nanotechnol- EU candidate
countries
ogy a permanent home in the federal government and and Russia Asia
authorized US$ 3.7 billion to be spent in the 4 year pe- 8% 25%
riod beginning in October 2005 for nanotechnology
initiatives at five federal agencies. The funds would b) Worldwide patents
provide grants to researchers, coordinate research and in nanotechnology
development (R&D) across five federal agencies [the Europe
National Science Foundation (NSF), the Department 39%
Others
of Energy (DOE), the National Aeronautics and Space 3%
Administration (NASA), the National Institute of Stan-
Asia
dards and Technology (NIST), and the Environmental 13%
Protection Agency (EPA)], establish interdisciplinary
research centers, and accelerate technology transfer USA and
Canada
into the private sector. In addition, the Departments of 45%
Defense (DOD), Homeland Security, Agriculture, and
Justice as well as the National Institutes of Health (NIH)
also fund large R&D activities. They currently account Fig. 1.5a,b Breakdown of (a) worldwide publications and
(b) worldwide patents (source: European Commission,
2003)
Public and private expenditure in nanotechnology R&D in 2004
R&D expenditure (US $ billions)
for more than one-third of the federal budget for nano-
4 Government/institutional technology.
Corporate/private The European Union (EU) made nanosciences and
2004 global total nanotechnologies a priority in the Sixth Framework
Corporate/private – US $ 6.6 billion Program (FP6) in 2002 for the period 2003–2006.
3 Government/institutional – US $ 3.7 billion There were also small dedicated funds in FP4 and
FP5 before. FP6 was tailored to help better structure
European research and to cope with the strategic ob-
jectives set out in Lisbon in 2000. Japan identified
2 nanotechnology as one of its main research priorities
in 2001. The funding levels increased sharply from
US$ 400 million in 2001 to around US$ 950 million in
2004. In 2003, South Korea embarked upon a 10 year
1 program with around US$ 2 billion of public funding,
and Taiwan has committed around US$ 600 million of
public funding over 6 years. Singapore and China are
also investing on a large scale. Russia is well funded as
0 well.
n
SA
an
ea
lia
ce
ly
in
pa
an
Ita
or
an
w
tra
Ch
U
Ja
Fr
Ta
us
er
A
G
ut
So
Start-up companies
world (1997–2002) (source: CEA, Bureau d’Etude Mar- in nanotechnology
keting) (1997 – 2002)
USA
55%
25, and Japan are about US$ 3.7 billion, US$ 2.4 billion,
and US$ 6.2 billion, respectively [1.11]. Figure 1.3b Asia
4%
shows the breakdown of expenditure in 2004 by pub-
lic and private sources, with more than US$ 10 billion Rest of
spent in nanotechnology research. Two-thirds of this world
came from corporate and private funding. Private ex- 11% Europe
penditure in the USA and Japan was slightly larger 29%
than that from public sources, whereas in Europe it was
about one-third. Figure 1.4 shows the public and private
expenditure breakdown in 2004 in various countries.
Japan and USA had the largest expenditure, followed Switzerland France
by Germany, Taiwan, South Korea, the UK, Australia, 4% 4%
UK
China, France, and Italy. Figure 1.5 shows a break- 6%
down of worldwide publications and patents. USA and
Canada led, followed by Europe and Asia. Figure 1.6 Others
5% Germany
shows the breakdown in start-up companies around the 11%
world (1997–2002). Entrepreneurship in USA is clearly
evident, followed by Europe.
Introduction
Overview of various objects from nature and their selected functions
Adhesion
Motion
Structural
coloration
Fig. 1.7 Overview of various objects from nature and their selected function (after [1.13])
and self-cleaning surfaces are of interest in various the characteristics of gecko feet would enable the de-
applications, including self-cleaning windows, wind- velopment of a superadhesive polymer tape capable
shields, exterior paints for buildings and navigation of clean, dry adhesion which is reversible [1.25–27].
ships, utensils, roof tiles, textiles, and applications (It should be noted that common manmade adhesives
requiring a reduction of drag in fluid flow, e.g., in mi- such as tape or glue involve the use of wet adhesives
cro/nanofluidics. Superhydrophobic surfaces can also that permanently attach two surfaces.) The reusable
be used for energy conversion and conservation [1.20]. gecko-inspired adhesives have the potential for use in
Nonwetting surfaces also reduce stiction at contacting everyday objects such as tapes, fasteners, and toys, and
interfaces in machinery [1.21, 22]. in high technology such as microelectronic and space
The leg attachment pads of several creatures, includ- applications. Replication of the dynamic climbing and
ing many insects (e.g., beetles and flies), spiders, and peeling ability of geckoes could find use in the treads of
lizards (e.g., geckoes), are capable of attaching to a va- wall-climbing robots. Incidentally, Velcro was invented
riety of surfaces and are used for locomotion [1.23]. based on the action of the hooked seeds of the burdock
Biological evolution over a long period of time has led plant [1.28].
to the optimization of their leg attachment systems. The Many aquatic animals can move in water at high
attachment pads have the ability to cling to different speeds with low energy input. Drag is a major hindrance
smooth and rough surfaces and detach at will [1.24,25]. to movement. Most shark species move through water
This dynamic attachment ability is referred to as re- with high efficiency and maintain buoyancy. Through
versible adhesion or smart adhesion. Replication of its ingenious design, their skin turns out to be an essen-
8 Introduction
Introduction
a) b) c)
2 µm
100 µm
e)
20 µm
d) g)
0.5 mm
BR
h) 500 µm
f)
SP
5 µm
Fig. 1.8a–h Montage of some examples from nature: (a) lotus effect [1.30], (b) glands of carnivorous plant that secrete
adhesive to trap insects [1.17], (c) water strider walking on water [1.31], (d) gecko foot exhibits reversible adhesion [1.32]
(BR – branch, SP – spatula), (e) scale structure of shark reduces drag [1.33], (f) wings of a bird in landing approach,
(g) spider web made of silk material [1.12], (h) moth’s eyes are antireflective [1.34]
tial aid to this behavior by reducing friction drag and organisms to their bodies. In addition, mucus on the
autocleaning ectoparasites from their surface [1.29]. skin of aquatic animals, including sharks, acts as an
The very small individual tooth-like scales of shark osmotic barrier against the salinity of seawater and pro-
skin, called dermal denticles, are ribbed with longi- tects the creature from parasites and infections. It also
tudinal grooves, which result in water moving very acts as a drag-reducing agent. Artificial derivatives of
efficiently over their surface. The scales also minimize fish mucus (polymer additives) are used to propel crude
the collection of barnacles and algae. Speedo created oil in the Alaska pipeline. The compliant skin of dol-
the whole-body swimsuit called Fastskin, modeled on phins allows them to swim at high speed. By interacting
shark skin, for elite swimming. Boat, ship, and air- with the water flowing over the body’s surface it stabi-
craft manufacturers are trying to mimic shark skin to lizes the flow and delays the transition to turbulence.
reduce friction drag and minimize the attachment of Dolphins possess an optimum shape for drag reduc-
Introduction to Nanotechnology 1.4 Applications in Different Fields 9
tion of submerged bodies. Submarines use the shape reduce reflection. This antireflective design led to the
Introduction
of dolphins. The streamlined form of boxfish (Ostra- discovery of antireflective surfaces [1.35].
cion meleagris) has inspired Mercedes Benz’s bionic A remarkable property of biological tissues is their
concept car with low aerodynamic drag. The beak of ability for self-healing. In biological systems, chemical
the kingfisher was used to model the nose cone of the signals released at the site of a fracture initiate a sys-
Japanese Shinkansen bullet train. Power is generated temic response that transports repair agents to the site of
by the scalloped edges of a humpback whale, and this an injury and promotes healing. Various artificial self-
design is exploited for wind turbine blades. healing materials are being developed [1.36]. Human
Bird feathers make the body water repellant, and skin is sensitive to impact, leading to purple-colored
movable flaps create wing and tail for aerodynamic marks in areas that are hit. This idea has led to the de-
lift during flying [1.29]. Birds and butterflies create velopment of coatings indicating impact damage [1.12].
brilliant hues by refracting light through millions of re- Another interesting and promising idea involves the ap-
peated structures that bend light to make certain colors. plication of an array of sensors to develop an artificial
Seashells are natural nanocomposites with a laminated nose or an artificial tongue.
structure and exhibit superior mechanical properties. Other lessons from nature include the wings of fly-
Spider web consists of silk fiber which is very strong. ing insects, abalone shell with high-impact ceramic
The materials and structures used in these objects have properties, strong spider silk, ultrasonic detection by
led to the development of various materials and fibers bats, infrared detection by beetles, and silent flying of
with high mechanical strength. Moth eyes have a mul- owls because of frayed feathers on the edges of their
tifaceted surface on the nanoscale and are structured to wings.
store bits of information, molecular switches, molecu- called DNA arrays, they can identify thousands of genes
Introduction
lar or nanotube transistors, nanotube flat-panel displays, simultaneously [1.60]. They include a microarray of sil-
nanotube integrated circuits, fast logic gates, switches, icon nanowires, roughly a few nm in size, to selectively
nanoscopic lasers, and nanotubes as electrodes in fuel bind and detect even a single biological molecule, such
cells. as DNA or protein, by using nanoelectronics to detect
BioMEMS/BioNEMS are increasingly used in com- the slight electrical charge caused by such binding, or
mercial and defense applications [1.57–63]. They are a microarray of carbon nanotubes to electrically detect
used for chemical and biochemical analyses (biosen- glucose.
sors) in medical diagnostics (e.g., DNA, RNA, proteins, After the tragedy of September 11, 2001, concern
cells, blood pressure and assays, and toxin iden- about biological and chemical warfare has led to the
tification) [1.63, 64], tissue engineering [1.65], and development of handheld units with bio- and chemical
implantable pharmaceutical drug delivery [1.66, 67]. sensors for detection of biological germs, chemical or
Biosensors, also referred to as biochips, deal with nerve agents, and mustard agents, and chemical precur-
liquids and gases. There are two types of biosen- sors to protect subways, airports, water supplies, and the
sors. A large variety of biosensors are based on population at large [1.68].
micro/nanofluidics. Micro/nanofluidic devices offer the BioMEMS/BioNEMS are also being developed for
ability to work with smaller reagent volumes and shorter minimal invasive surgery, including endoscopic surgery,
reaction times, and perform analyses multiple times laser angioplasty, and microscopic surgery. Other ap-
at once. The second type of biosensors includes mi- plications include implantable drug-delivery devices
cro/nanoarrays which perform one type of analysis (micro/nanoparticles with drug molecules encapsulated
thousands of times. Micro/nanoarrays are a tool used in functionalized shells for site-specific targeting ap-
in biotechnology research to analyze DNA or pro- plications) and a silicon capsule with a nanoporous
teins to diagnose diseases or discover new drugs. Also membrane filled with drugs for long-term delivery.
Obviously, studies of the determination and sup- is carried out to study the effects of surface rough-
Introduction
pression of active failure mechanisms affecting this new ness and scratches on stresses in nanostructures. When
and promising technology are critical to high reliabil- nanostructures are smaller than a fundamental physical
ity of MEMS/NEMS and are determining factors for length scale, conventional theory may no longer apply,
successful practical application. and new phenomena emerge. Molecular mechanics is
Adhesion between a biological molecular layer and used to simulate the behavior of a nanoobject.
the substrate, referred to as bioadhesion, and reduction The societal, ethical, political, and health/safety im-
of friction and wear of biological layers, biocompat- plications of nanotechnology are also attracting major
ibility, and biofouling for BioMEMS/BioNEMS are attention [1.11]. One of the prime reasons is to avoid
important. some of the public skepticism that surrounded the de-
Mechanical properties are known to exhibit a de- bate over biotechnology advances such as genetically
pendence on specimen size. Mechanical property eval- modified foods, while at the same time dispelling some
uation of nanoscale structures is carried out to help of the misconceptions the public may already have
design reliable systems since good mechanical prop- about nanotechnology. Health/safety issues need to be
erties are of critical importance in such applications. addressed as well. For example, one key question is
Some of the properties of interest are: Young’s mod- what happens to nanoparticles (such as buckyballs or
ulus of elasticity, hardness, bending strength, fracture nanotubes) in the environment and whether they are
toughness, and fatigue life. Finite-element modeling toxic in the human body, if digested.
References
Introduction
1.1 R.J. Chen, H.C. Choi, S. Bangsaruntip, E. Ye- biomimetics (invited), Prog. Mater. Sci. 54, 137–178
nilmex, X. Tang, Q. Wang, Y.L. Chang, H. Dai: (2009)
An investigation of the mechanisms of electrode 1.18 M. Nosonovsky, B. Bhushan: Multiscale Dissipa-
sensing of protein adsorption on carbon nano- tive Mechanisms and Hierarchical Surfaces: Friction,
tube devices, J. Am. Chem. Soc. 126, 1563–1568 Superhydrophobicity, and Biomimetics (Springer,
(2004) Berlin, Heidelberg 2008)
1.2 D. Srivastava: Computational nanotechnology of 1.19 M. Nosonovsky, B. Bhushan: Roughness-induced
carbon nanotubes. In: Carbon Nanotubes: Science superhydrophobicity: A way to design non-adhesive
and Applications, ed. by M. Meyyappan (CRC, Boca surfaces, J. Phys. D 20, 225009 (2008)
Raton 2004) pp. 25–36 1.20 M. Nosonovsky, B. Bhushan: Multiscale effects and
1.3 W.G. van der Wiel, S. De Franceschi, J.M. Elzerman, capillary interactions in functional biomimetic sur-
T. Fujisawa, S. Tarucha, L.P. Kauwenhoven: Electron faces for energy conversion and green engineering,
transport through double quantum dots, Rev. Mod. Philos. Trans. R. Soc. Lond. Ser. A 367, 1511–1539
Phys. 75, 1–22 (2003) (2009)
1.4 Anonymous: Microelectromechanical Systems: Ad- 1.21 B. Bhushan: Principles and Applications of Tribology
vanced Materials and Fabrication Methods (National (Wiley, New York 1999)
Academy Press, Washington 1997), NMAB-483 1.22 B. Bhushan (Ed.): Introduction to Tribology (Wiley,
1.5 M. Roukes: Nanoelectromechanical systems face the New York 2002)
future, Phys. World 14, 25–31 (2001) 1.23 S. Gorb (Ed.): Attachment Devices of Insect Cuticle
1.6 J.C. Eloy: Status of the MEMS Industry (Yole Devel- (Kluwer, Dordrecht 2001)
oppement, Lyon 2005), presented at SPIE Photonics 1.24 K. Autumn, Y.A. Liang, S.T. Hsieh, W. Zesch,
West, San Jose (2005) W.P. Chan, T.W. Kenny, R. Fearing, R.J. Full: Ad-
1.7 S. Lawrence: Nanotech grows up, Technol. Rev. hesive force of a single gecko foot-hair, Nature 405,
108(6), 31 (2005) 681–685 (2000)
1.8 R.P. Feynman: There’s plenty of room at the bottom, 1.25 B. Bhushan: Adhesion of multi-level hierarchical
Eng. Sci. 23, 22–36 (1960), http://www.zyvex.com/ attachment systems in gecko feet, J. Adhes. Sci.
nanotech/feynman.html Technol. 21, 1213–1258 (2007)
1.9 I. Amato: Nanotechnology (2000), http://www.ostp. 1.26 A.K. Geim, S.V. Dubonos, I.V. Grigorieva, K.S. Novo-
gov/nstc/html/iwgn/iwgn.public.brochure/welcome. selov, A.A. Zhukov, S.Y. Shapoval: Microfabricated
htm or http://www.nsf.gov/home/crssprgm/nano/ adhesive mimicking gecko foot-hair, Nat. Mater. 2,
nsfnnireports.htm 461–463 (2003)
1.10 Anonymous: National Nanotechnology Initia- 1.27 B. Bhushan, R.A. Sayer: Surface characterization and
tive (2000), http://www.ostp.gov/nstc/html/iwgn. friction of a bio-inspired reversible adhesive tape,
fy01budsuppl/nni.pdf or http://www.nsf.gov/home/ Microsyst. Technol. 13, 71–78 (2007)
crssprgm/nano/nsfnnireports.htm 1.28 S.A. Velcro: Improvements in or relating to a method
1.11 Anonymous: Towards a European Strategy for and a device for producing velvet type fabric,
Nanotechnology (European Commission Research Switzerland Patent 721338 (1995)
Directorate General, Brussels 2004) 1.29 D.W. Bechert, M. Bruse, W. Hage, R. Meyer: Fluid
1.12 Y. Bar-Cohen (Ed.): Biomimetics – Biologically In- mechanics of biological surfaces and their techno-
spired Technologies (CRC, Boca Raton 2005) logical application, Naturwissenschaften 87, 157–171
1.13 B. Bhushan: Biomimetics: Lessons from nature – An (2000)
overview, Philos. Trans. R. Soc. Lond. Ser. A 367, 1.30 B. Bhushan, Y.C. Jung, K. Koch: Micro-, nano-,
1445–1486 (2009) and hierarchical structures for superhydrophobic-
1.14 C.J. Jones, S. Aizawa: The bacterial flagellum and ity, self-cleaning, and low adhesion, Philos. Trans.
flagellar motor: Structure, assembly, and functions, R. Soc. Lond. Ser. A 367, 1631–1672 (2009)
Adv. Microb. Physiol. 32, 109–172 (1991) 1.31 X.F. Gao, L. Jiang: Biophysics: Water-repellent legs
1.15 B. Bhushan, Y.C. Jung: Wetting, adhesion and fric- of water striders, Nature 432, 36 (2004)
tion of superhydrophobic and hydrophilic leaves 1.32 H. Gao, X. Wang, H. Yao, S. Gorb, E. Arzt: Mechanics
and fabricated micro-/nanopatterned surfaces, of hierarchical adhesion structures of geckos, Mech.
J. Phys. D 20, 225010 (2008) Mater. 37, 275–285 (2005)
1.16 K. Koch, B. Bhushan, W. Barthlott: Diversity of struc- 1.33 W.E. Reif: Squamation and Ecology of Sharks,
ture, morphology, and wetting of plant surfaces, Courier Forschungsinst. Senckenberg, Vol. 78
Soft Matter 4, 1943–1963 (2008) (Schweizerbart, Stuttgart 1985)
1.17 K. Koch, B. Bhushan, W. Barthlott: Multifunctional 1.34 J. Genzer, K. Efimenko: Recent developments in
surface structures of plants: An inspiration for superhydrophobic surfaces and their relevance to
Introduction to Nanotechnology References 13
marine fouling: A review, Biofouling 22, 339–360 Engineering,and Technology (CRC, Boca Raton
Introduction
(2006) 2002)
1.35 C.G. Bernhard, W.H. Miller, A.R. Möller: The in- 1.55 H.S. Nalwa (Ed.): Nanostructured Materials and
sect corneal nipple array: A biological, broad-band Nanotechnology (Academic, San Diego 2002)
impedance transformer that acts as a antireflection 1.56 C.P. Poole, F.J. Owens: Introduction to Nanotechnol-
coating, Acta Physiol. Scand. 63, 1–79 (1965) ogy (Wiley, New York 2003)
1.36 J.P. Youngblood, N.R. Sottos: Bioinspired materials 1.57 A. Manz, H. Becker (Eds.): Microsystem Technology
for self-cleaning and self-healing, MRS Bulletin 33, in Chemistry and Life Sciences, Top. Curr. Chem.,
732–738 (2008) Vol. 194 (Springer, Berlin, Heidelberg 1998)
1.37 Anonymous: Small Tech 101 – An Introduction to 1.58 J. Cheng, L.J. Kricka (Eds.): Biochip Technology (Har-
Micro and Nanotechnology (Small Times, 2003) wood, Philadephia 2001)
1.38 M. Schulenburg: Nanotechnology – Innovation for 1.59 M.J. Heller, A. Guttman (Eds.): Integrated Microfab-
Tomorrow’s World (European Commission Research ricated Biodevices (Marcel Dekker, New York 2001)
Directorate General, Brussels 2004) 1.60 C. Lai Poh San, E.P.H. Yap (Eds.): Frontiers in Human
1.39 R.S. Muller, R.T. Howe, S.D. Senturia, R.L. Smith, Genetics (World Scientific, Singapore 2001)
R.M. White: Microsensors (IEEE, New York 1991) 1.61 C.H. Mastrangelo, H. Becker (Eds.): Microfluidics and
1.40 I. Fujimasa: Micromachines: A New Era in Mechanical BioMEMS, Proc. SPIE, Vol. 4560 (SPIE, Bellingham
Engineering (Oxford Univ. Press, Oxford 1996) 2001)
1.41 W.S. Trimmer (Ed.): Micromachines and MEMS, Clas- 1.62 H. Becker, L.E. Locascio: Polymer microfluidic de-
sical and Seminal Papers to 1990 (IEEE, New York vices, Talanta 56, 267–287 (2002)
1997) 1.63 A. van der Berg (Ed.): Lab-on-a-Chip: Chemistry in
1.42 B. Bhushan: Tribology Issues and Opportunities in Miniaturized Synthesis and Analysis Systems (Else-
MEMS (Kluwer, Dordrecht 1998) vier, Amsterdam 2003)
1.43 G.T.A. Kovacs: Micromachined Transducers Source- 1.64 P. Gravesen, J. Branebjerg, O.S. Jensen: Microflu-
book (WCB McGraw-Hill, Boston 1998) idics – A review, J. Micromech. Microeng. 3, 168–182
1.44 M. Elwenspoek, R. Wiegerink: Mechanical Microsen- (1993)
sors (Springer, Berlin Heidelberg 2001) 1.65 R.P. Lanza, R. Langer, J. Vacanti (Eds.): Principles of
1.45 S.D. Senturia: Microsystem Design (Kluwer, Boston Tissue Engineering, 2nd edn. (Academic, San Diego
2000) 2000)
1.46 T.R. Hsu: MEMS and Microsystems: Design and Man- 1.66 K. Park (Ed.): Controlled Drug Delivery: Challenges
ufacture (McGraw-Hill, Boston 2002) and Strategies (American Chemical Society, Wash-
1.47 M. Madou: Fundamentals of Microfabrication: The ington 1997)
Science of Miniaturization, 2nd edn. (CRC, Boca Ra- 1.67 P.Å. Öberg, T. Togawa, F.A. Spelman: Sensors in
ton 2002) Medicine and Health Care (Wiley, New York 2004)
1.48 A. Hierlemann: Integrated Chemical Microsensor 1.68 M. Scott: MEMS and MOEMS for national security
Systems in CMOS Technology (Springer, Berlin Hei- applications, Proc. SPIE 4980, xxxvii–xliv (2003)
delberg 2005) 1.69 B. Bhushan: Handbook of Micro/Nanotribology (CRC,
1.49 B. Bhushan: Tribology and Mechanics of Magnetic Boca Raton 1999)
Storage Devices, 2nd edn. (Springer, Berlin Heidel- 1.70 B. Bhushan (Ed.): Nanotribology and Nanomechan-
berg 1996) ics – An Introduction, 2nd edn. (Springer, Berlin,
1.50 K.E. Drexler: Nanosystems: Molecular Machinery, Heidelberg 2008)
Manufacturing and Computation (Wiley, New York 1.71 Anonymous: Current status and future needs, Proc.
1992) Workshop Res. Train. Nanosci. Nanotechnol. (Eu-
1.51 G. Timp (Ed.): Nanotechnology (Springer, New York ropean Commission Research Directorate General,
1999) Brussels 2005)
1.52 M.S. Dresselhaus, G. Dresselhaus, P. Avouris: Carbon 1.72 M. Di Ventra, S. Evoy, J.R. Heflin: Introduction to
Nanotubes – Synthesis, Structure, Properties, and Nanoscale Science and Technology (Springer, Berlin
Applications (Springer, Berlin Heidelberg 2001) Heidelberg 2004)
1.53 E.A. Rietman: Molecular Engineering of Nanosys- 1.73 A. Hett: Nanotechnology – Small Matter, Many Un-
tems (Springer, Berlin, Heidelberg 2001) knowns (Swiss Reinsurance Company, Zurich 2004)
1.54 W.A. Goddard, D.W. Brenner, S.E. Lyshevski, 1.74 M. Köhler, W. Fritzsche: Nanotechnology (Wiley, New
G.J. Iafrate (Eds.): Handbook of Nanoscience, York 2004)
15
Part A
Nanostru Part A Nanostructures, Micro-/Nanofabrication
and Materials
Nanomateria
2. Nanomaterials Synthesis and Applications:
Part A 2
Molecule-Based Devices
Françisco M. Raymo
The fabrication of conventional devices relies on the Every single nucleotide of a polynucleotide strand
assembly of macroscopic building blocks with spe- carries one of the four heterocyclic bases shown in
cific configurations. The shapes of these components Fig. 2.1b. For a strand incorporating 100 nucleotide re-
are carved out of larger materials by exploiting phys- peating units, a total of 4100 unique polynucleotide
ical methods. This top-down approach to engineered sequences are possible. It follows that nature can fab-
building blocks is extremely powerful and can deliver ricate a huge number of closely related nanostructures
effectively and reproducibly microscaled objects. This relying only on four building blocks. The heterocyclic
strategy becomes increasingly challenging, however, as bases appended to the main backbone of alternating
the dimensions of the target structures approach the phosphate and sugar units can sustain hydrogen bonding
nanoscale. Indeed, the physical fabrication of nanosized and [π · · · π] stacking interactions. Hydrogen bonds,
features with subnanometer precision is a formidable formed between [N−H] donors and either N or O ac-
technological challenge. ceptors, encourage the pairing of adenine (A) with
thymine (T) and of guanine (G) with cytosine (C).
2.1.1 From Molecular Building Blocks The stacking interactions involve attractive contacts be-
to Nanostructures tween the extended π-surfaces of heterocyclic bases.
In the B conformation of deoxyribonucleic acid
Nature efficiently builds nanostructures by relying on (DNA), the synergism of hydrogen bonds and [π · · · π]
chemical approaches. Tiny molecular building blocks stacking glues pairs of complementary polynucleotide
are assembled with a remarkable degree of structural strands in fascinating double helical supermolecules
control in a variety of nanoscaled materials with defined (Fig. 2.1c) with precise structural control at the sub-
shapes, properties, and functions. In contrast to the nanometer level. The two polynucleotide strands wrap
top-down physical methods, small components are con- around a common axis to form a right-handed double
nected to produce larger objects in these bottom-up helix with a diameter of ≈ 2 nm. The hydrogen bonded
chemical strategies. It is becoming apparent that the and [π · · · π] stacked base pairs lie at the core of the
limitations of the top-down approach to artificial nano- helix with their π-planes perpendicular to the main
structures can be overcome by mimicking nature’s axis of the helix. The alternating phosphate and sugar
bottom-up processes. Indeed, we are starting to see units define the outer surface of the double helix. In
emerge beautiful and ingenious examples of molecule- B-DNA, ≈ 10 base pairs define each helical turn cor-
based strategies to fabricate chemically nanoscaled responding to a rise per turn or helical pitch of ≈ 3 nm.
building blocks for functional materials and innovative Considering that these molecules can incorporate up to
devices. ≈ 1011 base pairs, extended end-to-end lengths span-
ning from only few nanometers to hundreds of meters
2.1.2 Nanoscaled Biomolecules: are possible.
Nucleic Acids and Proteins Nature’s operating principles to fabricate nano-
structures are not limited to nucleic acids. Proteins are
Nanoscaled macromolecules play a fundamental role also built joining simple molecular building blocks, the
in biological processes [2.1]. Nucleic acids, for exam- amino acids, by strong covalent bonds [2.1]. More pre-
ple, ensure the transmission and expression of genetic cisely, nature relies on 20 amino acids differing in their
information. These particular biomolecules are lin- side chains to assemble linear polymers, called polypep-
ear polymers incorporating nucleotide repeating units tides, incorporating an extended backbone of robust
(Fig. 2.1a). Each nucleotide has a phosphate bridge and [C−N] and [C−C] bonds (Fig. 2.2a). For a single poly-
a sugar residue. Chemical bonds between the phosphate mer strand of 100 repeating amino acid units, a total
of one nucleotide and the sugar of the next ensures of 20100 unique combinations of polypeptide sequences
the propagation of a polynucleotide strand from the 5 are possible. Considering that proteins can incorporate
to the 3 end. Along the sequence of alternating sugar more than one polypetide chain with over 4000 amino
and phosphate fragments, an extended chain of robust acid residues each, it is obvious that nature can assemble
covalent bonds involving carbon, oxygen, and phospho- an enormous number of different biomolecules relying
rous atoms forms the main backbone of the polymeric on the same fabrication strategy and a relatively small
strand. pool of building blocks.
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.1 Chemical Approaches to Nanostructured Materials 19
Part A 2.1
Sugar c)
residue (a) incorporates alternating phosphate
Heterocyclic and sugar residues joined by covalent
5' end HO base
O bonds. Each sugar carries one of four
heterocyclic bases (b). Noncovalent
O interactions between complemen-
Phosphate O P O O tary bases in two independent
bridge Nucleotide
O– polynucleotide strands encourage
repeating
O unit the formation of nanoscaled double
3 nm
O P O O helixes (c)
O–
a) O
O P O O
O– n
O
O P O O
O–
Polynucleotide O
strand O P O 3' end
O–
2 nm
B-DNA
double
b) NH2 O NH2 O helix
N N Me
N NH N NH
N N N N NH2 N O N O
A G C T
a) Amino acid b)
repeating
unit
O R O R O
H H
Ammonium H3N + N N
end N N O– Carboxylate end
R H O R H O R
n 2 nm
c) Polypeptide
strand
2 nm
3 nm 0.5 nm
Polypeptide sheet Polypeptide helix
Fig. 2.2a–c A polypeptide strand (a) incorporates amino acid residues differing in their side chains and joined by covalent
bonds. Hydrogen bonding interactions curl a single polypeptide strand into a helical arrangement (b) or lock pairs of
strands into nanoscaled sheets (c)
20 Part A Nanostructures, Micro-/Nanofabrication and Materials
The covalent backbones of the polypeptide strands ties of the assembling components in order to shape
Part A 2.1
form the main skeleton of a protein molecule. In ad- the geometry of the final product with the assistance of
dition, a myriad of secondary interactions, involving noncovalent interactions. For example, five bipyridine
noncovalent contacts between portions of the amino building blocks (Fig. 2.3) can be connected in five syn-
acid residues, control the arrangement of the individ- thetic steps to produce an oligobipyridine strand [2.4].
ual polypeptide chains. Intrastrand hydrogen bonds curl The five repeating units are bridged by C−O bonds and
single polypeptide chains around a longitudinal axis can chelate metal cations in the bay regions defined by
in a helical fashion to form tubular nanostructures their two nitrogen atoms. The spontaneous assembly of
≈ 0.5 nm wide and ≈ 2 nm long (Fig. 2.2b). Similarly, two organic strands in a double helical arrangement oc-
interstrand hydrogen bonds can align from 2 up to curs in the presence of inorganic cations. In the resulting
15 parallel or antiparallel polypeptide chains to form helicate, the two oligobipyridine strands wrap around
nanoscaled sheets with average dimensions of 2 × 3 nm2 an axis defined by five Cu(I) centers. Each inorganic
(Fig. 2.2c). Multiple nanohelices and/or nanosheets cation coordinates two bipyridine units with a tetra-
combine into a unique three-dimensional arrangement hedral geometry imposing a diameter of ≈ 0.6 nm on
dictating the overall shape and dimensions of a protein. the nanoscaled helicate [2.5]. The overall length from
one end of the helicate to the other is ≈ 3 nm [2.6].
2.1.3 Chemical Synthesis The analogy between this artificial double helix and the
of Artificial Nanostructures B-DNA double helix shown in Fig. 2.1c is obvious. In
both instances, a supramolecular glue combines two in-
Nature fabricates complex nanostructures relying on dependent molecular strands into nanostructures with
simple criteria and a relatively small pool of mo- defined shapes and dimensions.
lecular building blocks. Robust chemical bonds join The chemical synthesis of nanostructures can bor-
the basic components into covalent scaffolds. Non- row nature’s design criteria as well as its molecular
covalent interactions determine the three-dimensional building blocks. Amino acids, the basic components of
arrangement and overall shape of the resulting assem- proteins, can be assembled into artificial macrocycles.
blies. The multitude of unique combinations possible In the example of Fig. 2.4, eight amino acid residues
for long sequences of chemically connected building are joined through the formation of C−N bonds in
blocks provides access to huge libraries of nanoscaled multiple synthetic steps [2.7]. The resulting covalent
biomolecules. backbone defines a circular cavity with a diameter of
Modern chemical synthesis has evolved consider- ≈ 0.8 nm [2.8]. In analogy to the polypeptide chains
ably over the past few decades [2.2]. Experimental of proteins, the amino acid residues of this artificial
procedures to join molecular components with struc- oligopeptide can sustain hydrogen bonding interactions.
tural control at the picometer level are available. It follows that multiple macrocycles can pile on top of
A multitude of synthetic schemes to encourage the for- each other to form tubular nanostructures. The walls of
mation of chemical bonds between selected atoms in the resulting nanotubes are maintained in position by
reacting molecules have been developed. Furthermore, the cooperative action of at least eight primary hydrogen
the tremendous progress of crystallographic and spec- bonding contacts per macrocycle. These noncovalent
troscopic techniques has provided efficient and reliable interactions maintain the mean planes of independent
tools to probe directly the structural features of artifi- macrocycles in an approximately parallel arrangement
cial inorganic and organic compounds. It follows that with a plane-to-plane separation of ≈ 0.5 nm.
designed molecules with engineered shapes and dimen-
sions can be now prepared in a laboratory relying on 2.1.4 From Structural Control
the many tricks of chemical synthesis and the power of to Designed Properties and Functions
crystallographic and spectroscopic analyses.
The high degree of sophistication reached in this The examples in Figs. 2.3 and 2.4 demonstrate that
research area translates into the possibility of mimick- modular building blocks can be assembled into target
ing the strategies successfully employed by nature to compounds with precise structural control at the pico-
fabricate chemically nanostructures [2.3]. Small mo- meter level through programmed sequences of synthetic
lecular building blocks can be synthesized and joined steps. Indeed, modern chemical synthesis offers access
covalently following routine laboratory procedures. It to complex molecules with nanoscaled dimensions and,
is even possible to design the stereoelectronic proper- thus, provides cost-effective strategies for the pro-
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.1 Chemical Approaches to Nanostructured Materials 21
Part A 2.1
Me Me
nanostructures in parallel. Furthermore, the high degree
N N Bipyridine
of structural control is accompanied by the possibility ×2 ligand
of designing specific properties into the target nano- N N
structures. Electroactive and photoactive components
can be integrated chemically into functional molecular OH + Br O
machines [2.9]. Extensive electrochemical investiga-
tions have demonstrated that inorganic and organic N N
compounds can exchange electrons with macroscopic 3× Synthesis Cu(I) 3 nm
electrodes [2.10]. These studies have unraveled the pro- N N
cesses responsible for the oxidation and reduction of
numerous functional groups and indicated viable design Br O
criteria to adjust the ability of molecules to accept or do-
nate electrons [2.11]. Similarly, detailed photochemical N
and photophysical investigations have elucidated the
N
mechanisms responsible for the absorption and emis-
sion of photons at the molecular level [2.12]. The
O
vast knowledge established on the interactions between
0.6 nm
light and molecules offers the opportunity to engineer
Oligobipyridine N Synthetic
chromophoric and fluorophoric functional groups with double
strand
defined absorption and emission properties [2.11, 13]. N helix
The power of chemical synthesis to deliver func-
tional molecules is, perhaps, better illustrated by the O
molecular motor shown in Fig. 2.5. The preparation
of this [2]rotaxane requires 12 synthetic steps starting N
from known precursors [2.14]. This complex molecule
incorporates a Ru(II)-trisbipyridine stopper bridged to N
a linear tetracationic fragment by a rigid triaryl spacer. Me
The other end of the tetracationic portion is terminated
by a bulky tetraarylmethane stopper. The bipyridinium Fig. 2.3 An oligobipyridine strand can be synthesized joining five
unit of this dumbbell-shaped compound is encircled by bipyridine subunits by covalent bonds. The tetrahedral coordi-
a macrocyclic polyether. No covalent bonds join the nation of pairs of bipyridine ligands by Cu(I) ions encourages
macrocyclic and linear components. Rather, hydrogen the assembly two oligobipyridine strands into a double helical
bonding and [π · · · π] stacking interactions maintain the arrangement
0.8 nm
Me O
H2NOC O H CO2H
CO2H N NH
4× Me D O 0.5 nm
NH2 O HN
0.8 nm
NH
+ Synthesis Me Me Self-assembly
HN
CO2H NH O
4 × HO2C L N
NH2 H N
HO2C H O CONH2
O Me
Fig. 2.4 Cyclic oligopeptides can be synthesized joining eight amino acid residues by covalent bonds. The resulting
macrocycles self-assemble into nanoscaled tubelike arrays
22 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 2.2
5 nm O
Et
t-Bu
Fig. 2.5 This nanoscaled [2]rotaxane incorporates a photoactive Ru(II)-trisbipyridine stopper and two electroactive bipyridinium
units. Photoinduced electron transfer from the photoactive stopper to the encircled electroactive unit forces the macrocyclic
polyether to shuttle to the adjacent bipyridinium dication
macrocyclic polyether around the bipyridinium unit. In in the photoactive stopper is filled, and undesired
addition, mechanical constrains associated with the bulk back electron transfer is suppressed. The permanent
of the two terminal stoppers prevent the macrocycle to and light-induced reduction of the dicationic bipyri-
slip off the thread. The approximate end-to-end distance dinium unit to a radical cation depresses significantly
for this [2]rotaxane is ≈ 5 nm. the magnitude of the noncovalent interactions hold-
The bipyridinium and the 3,3 -dimethyl bipyri- ing the macrocyclic polyether in position. As a result,
dinium units within the dumbbell-shaped component the macrocycle shuttles from the reduced unit to the
undergo two consecutive and reversible monoelectronic adjacent dicationic 3,3 -dimethyl bipyridinum. After
reductions [2.14]. The two methyl substituents on the the diffusion of molecular oxygen into the acetoni-
3,3 -dimethyl bipyridinium dication make this elec- trile solution, oxidation occurs restoring the dicationic
troactive unit more difficult to reduce. In acetonitrile, form of the bipyridinium unit and its ability to sustain
its redox potential is ≈ 0.29 V more negative than strong noncovalent bonds. As a result, the macrocyclic
that of the unsubstituted bipyridinium dication. Under polyether shuttles back to its original position. This
irradiation at 436 nm in degassed acetonitrile, the ex- amazing example of a molecular shuttle reveals that
citation of the Ru(II)-trisbipyridine stopper is followed dynamic processes can be controlled reversibly at the
by electron transfer to the unsubstituted bipyridinium molecular level relying on the clever integration of elec-
unit. In the presence of a sacrificial electron donor troactive and photoactive fragments into functional and
(triethanolamine) in solution, the photogenerated hole nanoscaled molecules.
of a railroad can divert trains from one track to an- exchanged optically between remote locations [2.18].
Part A 2.2
other. Similarly, a faucet in a lavatory pipe can block Data processing and communication require the encod-
or release the flow of water. Of course, the nature of ing of information in electrical and optical signals in the
the control stimulations and the character of the final form of binary digits. Using arbitrary assumptions, logic
outcome vary significantly from case to case, but the thresholds can be established for each signal and, then,
operating principle behind each switching device is the 0 and 1 digits can be encoded following simple con-
same. In all cases, input stimulations reach the switch ventions. Sequences of electronic devices manipulate
changing its physical state and producing a specific the encoded bits executing logic functions as a result
output. of basic switching operations.
The development of nanoscaled counterparts to con- The three basic AND, NOT, and OR operators com-
ventional switches is expected to have fundamental bine binary inputs into binary outputs following precise
scientific and technological implications. For instance, logic protocols [2.17]. The NOT operator converts an
one can envisage practical applications for ultraminia- input signal into an output signal. When the input is 0,
turized switches in areas ranging from biomedical the output is 1. When the input is 1, the output is 0.
research to information technology. The major chal- Because of the inverse relationship between the input
lenge in the quest for nanoswitches, however, is the and output values, the NOT gate is often called in-
identification of reliable design criteria and operat- verter [2.19]. The OR operator combines two input
ing principles for these innovative and fascinating signals into a single output signal. When one or both
devices. Chemical approaches to implement molecule- inputs are 1, the output is 1. When both inputs are 0,
sized switches appear to be extremely promising. The the output is 0. The AND gate also combines two input
intrinsically small dimensions of organic molecules signals into one output signal. In this instance, however,
coupled with the power of chemical synthesis are the output is 1 only when both inputs are 1. When at
the main driving forces behind these exploratory least one input is 0, the output is 0.
investigations. The output of one gate can be connected to one of
Certain organic molecules adjust their structural and the inputs of another operator. A NAND gate, for ex-
electronic properties when stimulated with chemical, ample, is assembled connecting the output of an AND
electrical, or optical inputs. Generally, the change is operator to the input of a NOT gate. Now the two
accompanied by an electrochemical or spectroscopic input signals are converted into the final output after
response. Overall, these nanostructures transduce input two consecutive logic operations. In a similar fashion,
stimulations into detectable outputs and, appropriately, a NOR gate can be assembled connecting the output
are called molecular switches [2.15, 16]. The chemical of an OR operator to the input of a NOT gate. Once
transformations associated with these switching pro- again, two consecutive logic operations determine the
cesses are often reversible. The chemical system returns relation between two input signals and a single output.
to the original state when the input signal is turned off. The NAND and NOR operations are termed univer-
The interconverting states of a molecular switch can sal functions because any conceivable logic operation
be isomers, an acid and its conjugated base, the oxi- can be implemented relying only on one of these two
dized and reduced forms of a redox active molecule, or gates [2.17]. In fact, digital circuits are fabricated rou-
even the complexed and uncomplexed forms of a recep- tinely interconnecting exclusively NAND or exclusively
tor [2.9, 13, 15, 16]. The output of a molecular switch NOR operators [2.19].
can be a chemical, electrical, and/or optical signal that The logic gates of conventional microprocessors are
varies in intensity with the interconversion process. For assembled interconnecting transistors, and their input
example, changes in absorbance, fluorescence, pH, or and output signals are electrical [2.19]. But the concepts
redox potential can accompany the reversible transfor- of binary logic can be extended to chemical, mechan-
mation of a molecular switch. ical, optical, pneumatic, or any other type of signal.
First it is necessary to design devices that can respond
2.2.2 Digital Processing to these stimulations in the same way transistors re-
and Molecular Logic Gates spond to electrical signals. Molecular switches respond
to a variety of input stimulations producing specific out-
In present computer networks, data are elaborated elec- puts and can, therefore, be exploited to implement logic
tronically by microprocessor systems [2.17] and are functions [2.13, 20, 21].
24 Part A Nanostructures, Micro-/Nanofabrication and Materials
2.2.3 Molecular AND, NOT, and OR Gates protonated form. Thus, a change in H+ concentration
Part A 2.2
1 2 O 3
O
O
O O
O
N
O O
N O
N N O
–O2 C
CN
I1 I1
I O O O
I2 I2
NOT OR AND
I1 I2 O I1 I2 O
I O
0 0 0 0 0 0
0 1 0 1 1 0 1 0
1 0 1 0 1 1 0 0
1 1 1 1 1 1
Fig. 2.6 The fluorescence intensity of the pyrazoline derivative 1 is high when the concentration of H+ is low, and vice
versa. The fluorescence intensity of the anthracene derivative 2 is high when the concentration of Na+ and/or K+ is high.
The emission is low when both concentrations are low. The fluorescence intensity of the anthracene 3 is high only when
the concentrations of H+ and Na+ are high. The emission is low in the other three cases. The signal transductions of the
molecular switches 1, 2, and 3 translate into the truth tables of NOT, OR, and AND gates, respectively, if a positive logic
convention is applied to all inputs and outputs (low = 0, high = 1)
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.2 Molecular Switches and Logic Gates 25
addition of 1000 equivalents of either Na+ or K+ , the logic operations and simple combinations of them [2.13,
Part A 2.2
quantum yield raises to 0.053 and 0.14, respectively. 20,21]. Most of these molecular switches convert chem-
Similarly, the quantum yield is 0.14 when both metal ical inputs into optical outputs. But the implementation
cations are present in solution. The complexation of one of logic operations at the molecular level is not limited
of the two metal cations inside the azacrown receptor to the use of chemical inputs. For example, electrical
depresses the efficiency of the photoinduced electron signals and reversible redox processes can be exploited
transfer enhancing the fluorescence. Thus, changes in to modulate the output of a molecular switch [2.24]. The
the concentrations of Na+ (I1) and/or K+ (I2) from low supramolecular assembly 4 (Fig. 2.7) executes a XNOR
to high values switch the emission intensity (O) from function relying on these operating principles. The π-
a low to a high value. The relationship between the electron rich tetrathiafulvalene (TTF) guest threads the
chemical inputs I1 and I2 and the optical output O trans- cavity of a π-electron deficient bipyridinium (BIPY)
lates into the truth table of an OR operation if a positive host. In acetonitrile, an absorption band associated with
logic convention (low = 0, high = 1) is applied to all the charge-transfer interactions between the comple-
signals. The emission intensity is low (O = 0) only mentary π-surfaces is observed at 830 nm. Electrical
when the concentration of Na+ and K+ are low (I1 = 0, stimulations alter the redox state of either the TTF or
I2 = 0). The emission intensity is high (O = 1) for the the BIPY units encouraging the separation of the two
other three input combinations. components of the complex and the disappearance of
The anthracene derivative 3 (Fig. 2.6) is a molecular the charge-transfer band. Electrolysis at a potential of
AND gate. It transduces two chemical inputs (con- + 0.5 V oxidizes the neutral TTF unit to a monoca-
centrations of H+ and Na+ ) into an optical output tionic state. The now cationic guest is expelled from
(emission intensity). In a mixture of methanol and iso- the cavity of the tetracationic host as a result of elec-
propanol, the fluorescence quantum yield is only 0.011 trostatic repulsion. Consistently, the absorption band
in the absence of H+ or Na+ [2.23]. Photoinduced at 830 nm disappears. The charge-transfer band, how-
electron transfer from either the tertiary amino group ever, is restored after the exhaustive back reduction of
or the catechol fragment to the anthracene fluorophore the TTF unit at a potential of 0 V. Similar changes
quenches the emission. After the addition of either 100 in the absorption properties can be induced address-
equivalents of H+ or 1000 equivalents of Na+ , a modest ing the BIPY units. Electrolysis at − 0.3 V reduces the
change of the quantum yield to 0.020 and 0.011, respec- dicationic BIPY units to their monocationic forms en-
tively, is observed. Instead, the quantum yield increases couraging the separation of the two components of the
to 0.068 when both species are present in solution. The complex and the disappearance of the absorption band.
protonation of the amino group and the insertion of the The original absorption spectrum is restored after the
metal cation in the benzocrown ether receptor depress exhaustive back oxidation of the BIPY units at a poten-
the efficiency of the photoinduced electron transfer pro- tial of 0 V. Thus, this supramolecular system responds
cesses enhancing the fluorescence. Thus, changes in the to electrical stimulations producing an optical output.
concentrations of H+ (I1) and Na+ (I2) from low to high One of the electrical inputs (I1) controls the redox state
values switch the emission intensity (O) from a low to of the TTF unit switching between 0 and + 0.5 V. The
a high value. The relationship between the chemical in- other (I2) determines the redox state of the bipyridinium
puts I1 and I2 and the optical output O translates into units switching between − 0.3 and 0 V. The optical out-
the truth table of an AND operation if a positive logic put (O) is the absorbance of the charge-transfer band.
convention (low = 0, high = 1) is applied to all signals. A positive logic convention (low = 0, high = 1) can be
The emission intensity is high (O = 1) only when the applied to the input I1 and output O. A negative logic
concentration of H+ and Na+ are high (I1 = 1, I2 = 1). convention (low = 1, high = 0) can be applied to the
The emission intensity is low (O = 0) for the other three input I2. The resulting truth table corresponds to that
input combinations. of a XNOR circuit (Fig. 2.7). The charge-transfer ab-
sorbance is high (O = 1) only when one voltage input
2.2.4 Combinational Logic is low and the other is high (I1 = 0, I2 = 0) or vice
at the Molecular Level versa (I1 = 1, I2 = 1). It is important to note that the
input string with both I1 and I2 equal to 1 implies that
The fascinating molecular AND, NOT, and OR gates input potentials of + 0.5 and − 0.3 V are applied simul-
illustrated in Fig. 2.6 have stimulated the design of re- taneously to a solution containing the supramolecular
lated chemical systems able to execute the three basic assembly 4 and not to an individual complex. Of course,
26 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 2.2
4
HO I1
+ + O
N N
O O
S S
O
S O O
S O
O
N
+
N+
O XNOR
OH I2
I1 I2 O
TTF BIPY Absorbance
0 0 1
Low Low High
0 1 0
Low High Low
High Low Low 1 0 0
High High High 1 1 1
Fig. 2.7 The charge-transfer absorbance of the complex 4 is high when the voltage input addressing the tetrathiafulvalene
(TTF) unit is low and that stimulating the bipyridinium (BIPY) units is high and vice versa. If a positive logic convention
is applied to the TTF input and to the absorbance output (low = 0, high = 1) while a negative logic convention is applied
to the BIPY input (low = 0, high = 1), the signal transduction of 4 translates into the truth table of a XNOR circuit
the concomitant oxidation of the TTF guest and reduc- encoded on each signal applying positive logic conven-
tion of the BIPY units in the very same complex would tions (low = 0, high = 1). It follows that the three-state
be unrealistic. In bulk solution, instead, some com- molecular switch converts input strings of three binary
plexes are oxidized while others are reduced, leaving digits into output strings of two binary digits. The cor-
the average solution composition unaffected. Thus, the responding truth table (Fig. 2.8) reveals that the optical
XNOR operation executed by this supramolecular sys- output O1 is high (O1 = 1) when only the input I3 is
tem is a consequence of bulk properties and not a result applied (I1 = 0, I2 = 0, I3 = 1), when only the input I2
of unimolecular signal transduction. is not applied (I1 = 1, I2 = 0, I3 = 0), or when all three
Optical inputs can be employed to operate the inputs are applied (I1 = 1, I2 = 0, I3 = 0). The optical
three-state molecular switch of Fig. 2.8 in acetonitrile output O2 is high (O2 = 1) when only the input I1 is
solution [2.25]. This chemical system responds to three applied (I1 = 1, I2 = 0, I3 = 0) or when only the input
inputs producing two outputs. The three input stimu- I3 is not applied (I1 = 1, I2 = 0, I3 = 0). The combi-
lations are ultraviolet light (I1), visible light (I2), and national logic circuit (Fig. 2.8) equivalent to this truth
the concentration of H+ (I3). One of the two optical table shows that all three inputs determine the output
outputs is the absorbance at 401 nm (O1), which is O1, while only I1 and I3 control the value of O2.
high when the molecular switch is in the yellow-green
state 6 and low in the other two cases. The other opti- 2.2.5 Intermolecular Communication
cal output is the absorbance at 563 nm (O2), which is
high when the molecular switch is in the purple state 7 The combinational logic circuits in Figs. 2.7 and 2.8
and low in the other two cases. The colorless spiropy- are arrays of interconnected AND, NOT, and OR op-
ran state 5 switches to the merocyanine form 7 upon erators. The digital communication between these basic
irradiation with ultraviolet light. It switches to the pro- logic elements ensures the execution of a sequence of
tonated merocyanine from 6 when treated with H+ . simple logic operations that results in the complex logic
The colored state 7 isomerizes back to 5 in the dark function processed by the entire circuit. It follows that
or upon irradiation with visible light. Alternatively, the logic function of a given circuit can be adjusted
7 switches to 6 when treated with H+ . The colored altering the number and type of basic gates and their
state 6 switches to 5, when irradiated with visible light, interconnection protocol [2.17]. This modular approach
and to 7, after the removal of H+ . In summary, this to combinational logic circuits is extremely powerful.
three-state molecular switch responds to two optical in- Any logic function can be implemented connecting the
puts (I1 and I2) and one chemical input (I3) producing appropriate combination of simple AND, NOT, and OR
two optical outputs (O1 and O2). Binary digits can be gates.
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.2 Molecular Switches and Logic Gates 27
Part A 2.2
Me
Me 5 communication between distinct gates is needed since
they are built in the same molecular entity. Though ex-
NO
tremely elegant, this strategy does not have the same
t NO2
gh versatility of a modular approach. A different molecule
li
Vi
rd
vio
sib
gle time a different logic function has to be realized. In
to
Ultra
le lig
Aci
ligh
d
Visible
ht
with only one molecular switch is fairly limited. The
– connection of the input and output terminals of indepen-
O HO
Me Me dent molecular AND, NOT, and OR operators, instead,
Me Me would offer the possibility of assembling any combina-
+N NO2 tional logic circuit from three basic building blocks.
+N NO2
In digital electronics, the communication between
7 OH Bas
e 6 two logic gates can be realized connecting their ter-
OH
minals with a wire [2.19]. Methods to transmit binary
A ci data between distinct molecular switches are not so ob-
d
vious and must be identified. Recently we developed
I1
two strategies to communicate signals between compat-
I2 O1
ible molecular components. In one instance, a chemical
O2 signal is communicated between two distinct molecular
I3 switches [2.26]. They are the three-state switch illus-
trated in Fig. 2.8 and the two-state switch of Fig. 2.9.
I1 I2 I3 O1 O2 The merocyanine form 7 is a photogenerated base. Its
0 0 0 0 0 p-nitrophenolate fragment, produced upon irradiation
0 0 1 1 0
of the colorless state 5 with ultraviolet light, can abstract
a proton from an acid present in the same solution. The
0 1 0 0 0
resulting protonated form 6 is a photoacid. It releases
1 0 0 0 1
a proton upon irradiation with visible light and can pro-
0 1 1 0 0 tonate a base co-dissolved in the same medium. The
1 0 1 1 0 orange azopyridine 8 switches to the red-purple azopy-
1 1 0 0 1 ridinium 9 upon protonation. This process is reversible,
1 1 0 1 0 and the addition of a base restores the orange state 8.
It follows that photoinduced proton transfer can be ex-
Fig. 2.8 Ultraviolet light (I1), visible light (I2), and H+ ploited to communicate a chemical signal from 6 to 8
(I3) inputs induce the interconversion between the three and from 9 to 7. The two colored states 8 and 9 have
states 5, 6, and 7. The colorless state 5 does not absorb different absorption properties in the visible region. In
in the visible region. The yellow-green state 6 absorbs at acetonitrile, the orange state 8 absorbs at 422 nm, and
401 nm (O1). The purple state 7 absorbs at 563 nm (O2). the red-purple state 9 absorbs at 556 nm. The changes
The truth table illustrates the conversion of input strings in absorbance of these two bands can be exploited to
of three binary digits (I1, I2, and I3) into output strings of monitor the photoinduced exchange of protons between
two binary digits (O1 and O2) operated by this three-state the two communicating molecular switches.
molecular switch. A combinational logic circuit incorpo- The three-state molecular switch and the two-state
rating nine AND, NOT, and OR operators correspond to molecular switch can be operated sequentially when
this particular truth table dissolved in the same acetonitrile solution. In the pres-
ence of one equivalent of H+ , the two-state molecular
The strategies followed so far to implement com- switch is in state 9 and the absorbance at 556 nm is
plex logic functions with molecular switches are based high (O = 1). Upon irradiation with ultraviolet light
on the careful design of the chemical system and on the (I1 = 0), 5 switches to 7. The photogenerated base
judicious choice of the inputs and outputs [2.13, 20, 21]. deprotonates 9 producing 8 and 6. As a result, the ab-
A specific sequence of AND, NOT, and OR operations sorbance at 556 nm decreases (O = 0). Upon irradiation
28 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 2.2
Ultraviolet light source I1 I2 I3 O1 O2 O3
Visible light
source H+ Detector
Excitation 0 0 0 0 1 1
I3
source I1 0 0 1 0 0 1
0 1 0 0 1 1
I2
1 0 0 0 0 0
O1
O2 0 1 1 0 1 1
O3
1 0 1 0 0 1
1 1 0 0 0 0
Molecular switch
1 1 1 0 0 1
Naphthalene
+ I1
Anthracene O3
+ I2 O2
Tetracene I3
Fig. 2.10 The excitation source sends three monochromatic light beams (275, 357, and 441 nm) to a quartz cell con-
taining an equimolar acetonitrile solution of naphthalene, anthracene and tetracene. The three fluorophores absorb the
exciting beams and reemit at 305, 401, and 544 nm, respectively. The light emitted in the direction perpendicular to the
exciting beams passes through another quartz cell containing an acetonitrile solution of the three-state molecular switch
shown in Fig. 2.7. Ultraviolet (I1), visible (I2), and H+ (I3) inputs control the interconversion between the three states
of the molecular switch. They determine the intensity of the optical outputs reaching the detector and correspond to the
naphthalene (O1), anthracene (O2), and tetracene (O3) emissions. The truth table and equivalent combinational logic il-
lustrate the relation between the three inputs and the three outputs. The output O1 is always 0, and it is not influenced by
the three inputs. Only two inputs determine the value of O3, while all of them control the output O2
respectively, pass unaffected and high intensity outputs two states 5 and 7 can be controlled relying exclusively
(O2 and O3) reach the detector. When the molecular on ultraviolet inputs. Indeed, ultraviolet irradiation in-
switch is in state 6, the naphthalene and anthracene duces the isomerization of the colorless form 5 to the
emissions are absorbed and only the tetracene emission colored species 7, which reisomerizes to the original
reaches the detector (O1 = 0, O2 = 0, O3 = 1). When state in the dark. Thus, a single ultraviolet source is suf-
the molecular switch is state 7, the emission of all three ficient to control the switching from 5 to 7 and vice
fluorophores is absorbed (O1 = 0, O2 = 0, O3 = 0). The versa. On the basis of these considerations, all-optical
interconversion of the molecular switch between the NAND, NOR, and NOT gates can be implemented op-
three states is induced addressing the second cell with erating sequentially or in parallel from one to three
ultraviolet (I1), visible (I2) and H+ (I3) inputs. Thus, independent switching elements [2.28]. For example,
three independent optical outputs (O1, O2 and O3) can the all-optical network illustrated in Fig. 2.11 is a three-
be modulated stimulating the molecular switch with input NOR gate. A monochromatic optical signal travels
two optical and one chemical input. The truth table in from a visible source to a detector. Three switching el-
Fig. 2.10 illustrates the relation between the three inputs ements are aligned along the path of the traveling light.
(I1, I2 and I3) and the three outputs (O1, O2 and O3), They are quartz cells containing an acetonitrile solution
when positive logic conventions are applied to all sig- of the molecular switch shown in Fig. 2.8. The intercon-
nals. The equivalent logic circuit shows that all three version of the colorless form 5 into the purple isomer 7
inputs control the anthracene channel O2, but only I1 is induced stimulating the cell with an ultraviolet input.
and I3 influence the tetracene channel O3. Instead, the The reisomerization from 7 to 5 occurs spontaneously,
intensity of the naphthalene channel O1 is always low, as the ultraviolet sources is turned off. Using three dis-
and it is not affected by the three inputs. tinct ultraviolet sources, the three switching elements
The operating principles of the optical network in can be controlled independently.
Fig. 2.10 can be simplified to implement all-optical The colorless form 5 does not absorb in the vis-
logic gates. The chemical input inducing the formation ible region, while the purple isomer 7 has a strong
of the protonated form 6 of the molecular switch can be absorption band at 563 nm. Thus, a 563 nm optical sig-
eliminated. The interconversion between the remaining nal leaving the visible source can reach the detector
30 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 2.3
Fig. 2.11 The visible source sends a monochromatic beam (563 nm) to the detector. The traveling light is forced to pass
through three quartz cells containing the molecular switch illustrated in Fig. 2.7. The three switching elements are oper-
ated by independent ultraviolet inputs. When at least one of them is on, the associated molecular switch is in the purple
form 7, which can absorb and block the traveling light. The truth table and equivalent logic circuit illustrate the relation
between the three inputs I1, I2, and I3 and the optical output O
unaffected only if all three switching elements are in binary digits can be encoded in the three optical inputs
the nonabsorbing state 5. If one of the three ultravio- and in the optical output. The resulting truth table is il-
let inputs I1, I2, or I3 is turned on, the intensity of the lustrated in Fig. 2.11. The output O is 1 only if all three
optical output O drops to 3–4% of its original value. inputs I1, I2, or I3 are 0. The output O is 0 if at least one
If two or three ultraviolet inputs are turned on simul- of the three inputs I1, I2, or I3 is 1. This signal transduc-
taneously, the optical output drops to 0%. Indeed, the tion corresponds to that executed by a three-input NOR
photogenerated state 7 absorbs and blocks the traveling gate, which is a combination of one NOT and two OR
light. Applying positive logic conventions to all signals, operators.
electrodes. Two main approaches for the deposition adecyl tail attached to a hydrophilic bipyridinium
Part A 2.3
of organized molecular arrays on inorganic supports dication [2.36, 37]. This compound dissolves in mix-
have emerged so far. In one instance, amphiphilic tures of chloroform and methanol, but it is not
molecular building blocks are compressed into orga- soluble in moderately concentrated aqueous solutions
nized monolayers at air/water interfaces. The resulting of sodium perchlorate. Thus the spreading of an organic
films can be transferred on supporting solids employing solution of 10 on an aqueous sodium perchlorate sub-
the Langmuir–Blodgett technique [2.34]. Alternatively, phase affords a collection of disorganized amphiphiles
certain molecules can be designed to adsorb sponta- floating on the water surface (Fig. 2.12), after the
neously on the surfaces of compatible solids from liquid organic solvent has evaporated. The molecular build-
or vapor phases. The result is the self-assembly of or- ing blocks can be compressed into a monolayer with
ganic layers on inorganic supports [2.35]. the aid of a moving barrier. The hydrophobic tails
align away from the aqueous phase. The hydrophilic
2.3.2 Langmuir–Blodgett Films dicationic heads and the accompanying perchlorate
counterions pack to form an organized monolayer at
Films of amphiphilic molecules can be deposited on the air/water interface. The compression process can
a variety of solid supports employing the Langmuir– be monitored recording the surface pressure (π)-area
Blodgett technique [2.34]. This method can be ex- per molecule (A) isotherm, which indicates a limit-
tended to electroactive compounds incorporating hy- ing molecular area of ≈ 50 Å2 . This value is larger
drophilic and hydrophobic groups. For example, the than the projected area of an oligomethylene chain.
amphiphile 10 (Fig. 2.12) has a hydrophobic hex- It correlates reasonably, however, with the overall
Air
Moving
barrier
Water
Electrode Compression
Air
Moving Me
barrier
Water
Electrode
Transfer CLO–4
Air
Me
anions.
The monolayer prepared at the air/water interface Ca2+
(Fig. 2.12) can be transferred on the surface of a indium-
tin oxide electrode pre-immersed in the aqueous phase.
The slow lifting of the solid support drags the mono-
layer away from the aqueous subphase. The final result Fe
is the coating of the electrode with an organic film con-
taining electroactive bipyridinium building blocks. The
modified electrode can be integrated in a conventional
electrochemical cell to probe the redox response of the
–
electroactive layer. The resulting cyclic voltammograms
reveal the characteristic waves for the first reduction 11 +
process of the bipyridinium dications, confirming the
successful transfer of the electroactive amphiphiles N N+
+ –
from the air/water interface to the electrode surface. The O2 C
integration of the redox waves indicates a surface cov-
erage of ≈ 4 × 1010 mol cm−2 . This value corresponds
to a molecular area of ≈ 40 Å2 and is in excellent – –
agreement with the limiting molecular area of the π–A
+ +
isotherm.
These seminal experiments demonstrate that elec-
troactive amphiphiles can be organized into uniform Gold
monolayers at the air/water interface and then trans-
ferred efficiently on the surface of appropriate sub- Fig. 2.13 Mixed monolayers of the amphiphile 11 and
strates to produce electrode/monolayer junctions. The arachidic acid can be transferred from the air/water inter-
resulting electroactive materials can become the func- face to the surface of an electrode to generate a molecule-
tional components of molecule-based devices. For based photodiode
example, bipyridinium-based photodiodes can be fabri-
cated following this approach [2.38,39]. Their operating ops at a potential of 0 V relative to a saturated calomel
principles rely on photoinduced electron transfer from electrode. Indeed, the illumination of the electroactive
chromophoric units to bipyridinium acceptors. The monolayer induces the electron transfer from the pyrene
electroactive and photoactive amphiphile 11 (Fig. 2.13) appendage to the bipyridinium acceptor and then from
incorporates hydrophobic ferrocene and pyrene tails the reduced acceptor to the electrode. A second in-
and a hydrophilic bipyridinium head. Chloroform solu- tramolecular electron transfer from the ferrocene donor
tions of 11 containing ten equivalents of arachidic acid to the oxidized pyrene fills its photogenerated hole.
can be spread on an aqueous calcium chloride subphase Overall, a unidirectional flow of electrons across the
in a Langmuir trough. The amphiphiles can be com- monolayer/electrode junction is established under the
pressed into a mixed monolayer, after the evaporation influence of light.
of the organic solvent. Pronounced steps in the corre- The ability to transfer electroactive monolayers
sponding π–A isotherm suggest that the bulky ferrocene from air/water interfaces to electrode surfaces can
and pyrene groups are squeezed away from the wa- be exploited to fabricate molecule-based electronic
ter surface. In the final arrangement, both photoactive devices. In particular, arrays of interconnected elec-
groups align above the hydrophobic dication. trode/monolayer/electrode tunneling junctions can be
A mixed monolayer of 11 and arachidic acid can be assembled combining the Langmuir–Blodgett tech-
transferred from the air/water interface to the surface nique with electron beam evaporation [2.33]. Fig-
of a transparent gold electrode following the methodol- ure 2.14 illustrates a schematic representation of the
ogy illustrated for the system in Fig. 2.12. The coated resulting devices. Initially, parallel fingers are patterned
electrode can be integrated in a conventional electro- on a silicon wafer with a silicon dioxide overlayer by
chemical cell. Upon irradiation at 330 nm under an inert electron beam evaporation. The bottom electrodes de-
atmosphere, an anodic photocurrent of ≈ 2 nA devel- posited on the support can be either aluminum wires
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.3 Solid State Devices 33
covered by an aluminum oxide or n-doped silicon lines backbone [2.40, 41]. The two bipyridinium dications
Part A 2.3
with silicon dioxide overlayers. Their widths are ≈ 6 are bridged by a m-phenylene spacer and terminated by
or 7 μm, respectively. The patterned silicon chip is im- tetraarylmethane appendages. These two bulky groups
mersed in the aqueous subphase of a Langmuir trough trap mechanically the macrocycle preventing its dis-
prior to monolayer formation. After the compression of sociation from the tetracationic backbone. In addition,
electroactive amphiphiles at the air/water interface, the their hydrophobicity complements the hydrophilicity of
substrate is pulled out of the aqueous phase to encour- the two bipyridinium dications imposing amphiphilic
age the transfer of the molecular layer on the parallel character on the overall molecular assembly. This com-
bottom electrodes as well as on the gaps between them. pound does not dissolve in aqueous solutions and can
Then, a second set of electrodes orthogonal to the first be compressed into organized monolayers at air/water
is deposited through a mask by electron beam evapo- interfaces. The corresponding π–A isotherm reveals
ration. They consist of a titanium underlayer plus an a limiting molecular area of ≈ 130 Å2 . This large
aluminum overlayer. Their thicknesses are ≈ 0.05 and value is a consequence of the bulk associated with the
1 μm, respectively, and their width is ≈ 10 μm. In the hydrophobic tetraarylmethane tails and the macrocycle
final assembly, portions of the molecular layer become encircling the tetracationic backbone.
sandwiched between the bottom and top electrodes. Monolayers of the [2]rotaxane 12 can be transferred
The active areas of these electrode/monolayer/electrode from the air/water interface to the surfaces of the bottom
junctions are ≈ 60–70 μm2 and correspond to ≈ 106 aluminum/aluminum oxide electrodes of a patterned sil-
molecules. icon chip with the hydrophobic tetraarylmethane groups
The [2]rotaxane 12 (Fig. 2.14) incorporates a macro- pointing away from the supporting substrate. The sub-
cyclic polyether threaded onto a bipyridinium-based sequent assembly of a top titanium/aluminum electrode
O 12 O
O O
O O Support Molecular layer
O
O O
+
N N+ O O O O 14
O O 13
O O O O
+
N N+ O
+
N O N+
O O Oxidation
S S S•S
+N N+
O O S S Reduction S + S
O
O O
+N N+ +N N+
HO O O O O
O O O O
Fig. 2.14 The [2]rotaxane 12 and the [2]catenane 13 can be compressed into organized monolayers at air/water interfaces. The
resulting monolayers can be transferred on the bottom electrodes of a patterned silicon support. After the deposition of a top
electrode, electrode/monolayer/electrode junctions can be assembled. Note that only the portion of the monolayer sandwiched
between the top and bottom electrodes is shown in the diagram. The oxidation of the tetrathiafulvalene unit of the [2]catenane 13
is followed by the circumrotation of the macrocyclic polyether to afford the [2]catenane 14. The process is reversible, and the
reduction of the cationic tetrathiafulvalene unit restores the original state
34 Part A Nanostructures, Micro-/Nanofabrication and Materials
affords electrode/monolayer/electrode junctions. Their applied to the current output (low = 0, high = 1). The
Part A 2.3
current/voltage signature can be recorded grounding the output O is 1 when at least one of the two inputs is 1.
top electrode and scanning the potential of the bottom The [2]catenane 13 (Fig. 2.14) incorporates a macro-
electrode. A pronounced increase in current is observed cyclic polyether interlocked with a tetracationic cyclo-
when the potential is lowered below − 0.7 V. Under phane [2.42, 43]. Organic solutions of the hexafluo-
these conditions, the bipyridinium-centered LUMOs rophosphate salt of this [2]catenane and six equivalents
mediate the tunneling of electrons from the bottom to of the sodium salt of dimyristoylphosphatidic acid
the top electrode leading to a current enhancement. can be co-spread on the water surface of a Langmuir
A similar current profile is observed if the potential is trough [2.44]. The sodium hexafluorophosphate formed
returned to 0 and then back to −2 V. Instead, a modest dissolves in the supporting aqueous phase, while the
increase in current in the opposite direction is observed hydrophilic bipyridinium cations and the amphiphilic
when the potential is raised above + 0.7 V. Presum- anions remain at the interface. Upon compression, the
ably, this trend is a result of the participation of the anions align their hydrophobic tails away from the
phenoxy-centered HOMOs in the tunneling process. Af- water surface forming a compact monolayer above the
ter a single positive voltage pulse, however, no current cationic bipyridinium derivatives. The corresponding
can be detected if the potential is returned to negative π–A isotherm indicates limiting molecular areas of
values. In summary, the positive potential scan sup- ≈ 125 Å2 . This large value is a consequence of the bulk
presses irreversibly the conducting ability of the elec- associated with the two interlocking macrocycles.
trode/molecule/electrode junction. The behavior of this Monolayers of the [2]catenane 13 can be trans-
device correlates with the redox response of the [2]ro- ferred from the air/water interface to the surfaces
taxane 12 in solution. Cyclic voltammograms reveal re- of the bottom n-doped silicon/silicon dioxide elec-
versible monoelectronic reductions of the bipyridinium trodes of a patterned silicon chip with the hydrophobic
dications. But they also show two irreversible oxida- tails of the amphiphilic anions pointing away from
tions associated, presumably, with the phenoxy rings of the supporting substrate [2.45, 46]. The subsequent
the macrocycle and tetraarylmethane groups. These ob- assembly of a top titanium/aluminum electrode affords
servations suggest that a positive voltage pulse applied electrode/monolayer/electrode arrays. Their junction
to the electrode/monolayer/electrode junction oxidizes resistance can be probed grounding the top electrode
irreversibly the sandwiched molecules suppressing their and maintaining the potential of the bottom electrode
ability to mediate the transfer of electrons from the bot- at + 0.1 V. If a voltage pulse of +2 V is applied to the
tom to the top electrode under a negative bias. bottom electrode before the measurement, the junction
The device incorporating the [2]rotaxane 13 can be resistance probed is ≈ 0.7 GΩ. After a pulse of −2 V
exploited to implement simple logic operations [2.40]. applied to the bottom electrode, the junction resistance
The two bottom electrodes can be stimulated with probed at + 0.1 V drops ≈ 0.3 GΩ. Thus, alternating
voltage inputs (I1 and I2) while measuring a current positive and negative voltage pulses can switch re-
output (O) at the common top electrode. When at least versibly the junction resistance between high and low
one of the two inputs is high (0 V), the output is low values. This intriguing behavior is a result of the redox
(< 0.7 nA). When both inputs are low (−2 V), the out- and dynamic properties of the [2]catenane 13.
put is high (≈ 4 nA). If a negative logic convention is Extensive spectroscopic and crystallographic stud-
applied to the voltage inputs (low = 1, high = 0) and ies [2.42, 43] demonstrated that the tetrathiafulvalene
a positive logic convention is applied to the current unit resides preferentially inside the cavity of the tetra-
output (low = 0, high = 1), the signal transduction be- cationic cyclophane of the [2]catenane 13 (Fig. 2.14).
havior translates into the truth table of an AND gate. Attractive [π · · · π] stacking interactions between the
The output O is 1 only when both inputs are 1. In- neutral tetrathiafulvalene and the bipyridinium dicat-
stead, an OR operation can be executed if the logarithm ions are responsible for this co-conformation. Oxi-
of the current is considered as the output. The loga- dation of the tetrathiafulvalene generates a cationic
rithm of the current is −12 when both voltage inputs are form that is expelled from the cavity of the tetra-
0 V. It raises to ≈ −9 when one or both voltage inputs cationic cyclophane. After the circumrotation of the
are lowered to −2 V. This signal transduction behavior macrocyclic polyether, the oxidized tetrathiafulvalene
translates into the truth table of an OR gate if a neg- is exchanged with the neutral 1,5-dioxynaphthalene
ative logic convention is applied to the voltage inputs producing the [2]catenane 14 (Fig. 2.14). The reduc-
(low = 1, high = 0) and a positive logic convention is tion of the tetrathiafulvalene back to its neutral state
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.3 Solid State Devices 35
Part A 2.3
polyether, which restores the original state 14. The Gold
voltage pulses applied to the bottom electrode of the
electrode/monolayer/junction oxidize and reduce the
tetrathiafulvalene unit inducing the interconversion be-
Silicon Gold Monolayer
tween the forms 13 and 14. The difference in the nitride of 15
stereoelectronic properties of these two states translates
into distinct current/voltage signatures. Indeed, their SH
ability to mediate the tunneling of electrons across the
junction differs significantly. As a result, the junction
resistance probed at a low voltage after an oxidizing
pulse is significantly different from that determined un-
der the same conditions after a reducing pulse. NH2
O2N
2.3.3 Self-Assembled Monolayers
15
In the examples illustrated in Figs. 2.12–2.14, mono-
layers of amphiphilic and electroactive derivatives are
assembled at air/water interfaces and then transferred
on the surfaces of appropriate substrates. An alterna- Fig. 2.15 A monolayer of the thiol 15 is embedded between
tive strategy to coat electrodes with molecular layers two gold electrodes maintained in position by a silicon
relies on the ability of certain compounds to adsorb nitride support
spontaneously on solid supports from liquid or vapor
phases [2.35]. In particular, the affinity of certain sul- Under the influence of voltage pulses applied to one
furated functional groups for gold can be exploited to of the two gold electrodes in Fig. 2.15, the conductivity
encourage the self-assembly of organic molecules on of the sandwiched monolayer switches reversibly be-
microscaled and nanoscaled electrodes. tween low and high values [2.48]. In the initial state,
The electrode/monolayer/electrode junction in the monolayer is in a low conducting mode. A current
Fig. 2.15 incorporates a molecular layer between two output of only 30 pA is detected, when a probing volt-
gold electrodes mounted on a silicon nitride support. age of + 0.25 V is applied to the bowl-shaped electrode.
This device can be fabricated combining chemical vapor If the same electrode is stimulated with a short volt-
deposition, lithography, anisotropic etching, and self- age pulse of +5 V, the monolayer switches to a high
assembly [2.47]. Initially, a silicon wafer is coated conducting mode. Now a current output of 150 pA is
with a 50 nm thick layer of silicon nitride by low measured at the same probing voltage of + 0.25 V.
pressure chemical vapor deposition. Then, a square of Repeated probing of the current output at various in-
400 × 400 μm2 is patterned on one side of the coated tervals of time indicates that the high conducting state
wafer by optical lithography and reactive ion etch- is memorized by the molecule-based device, and it is
ing. Anisotropic etching of the exposed silicon up to retained for more than 15 min. The low conducting
the other side of the wafer leaves a suspended sil- mode is restored after either a relatively long period
icon nitride membrane of 40 × 40 μm2 . Electron beam of time or the stimulation of the bowl-shaped elec-
lithography and reactive ion etching can be used to trode with a reverse voltage pulse of −5 V. Thus the
carve a bowl-shaped hole (diameter = 30–50 nm) in the current output switches from a low to a high value,
membrane. Evaporation of gold on the membrane fills if a high voltage input is applied. It switches from
the pore producing a bowl-shaped electrode. Immer- a high to a low value, under the influence of a low
sion of the substrate in a solution of the thiol 15 voltage pulse. This behavior offers the opportunity to
results in the self-assembly of a molecular layer on store and erase binary data in analogy to a conven-
the narrow part of the bowl-shaped electrode. The tional random access memory [2.17]. Binary digits can
subsequent evaporation of a gold film on the organic be encoded on the current output of the molecule-based
monolayer produces an electrode/monolayer/electrode device applying a positive logic convention (low = 0,
junction (Fig. 2.15) with a contact area of less than high = 1). It follows that a binary 1 can be stored
2000 nm2 and ≈ 1000 molecules. in the molecule-based device applying a high voltage
36 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 2.3
a) assemble spontaneously on pre-
+ +
N N functionalized indium-tin oxide
O + electrodes. (b) Electrostatic interac-
O Si NH3 Gold 18 tions encourage the adsorption of
O the tetracationic cyclophane 17 on
+N N+ the surface-confined nanoparticles.
Indium-tin (c) An additional layer of nano-
oxide particles assembles on the cationic
Adsorption O 19 organic coating. Similar composite
O
of 18 O films can be prepared using the tetra-
Me
O cationic [2]catenane 18 instead of
b) O
+ the cyclophane 17. (d) Phosphonate
N N+ Me
N groups can be used to anchor molecu-
O + N N lar building blocks to titanium dioxide
O Si NH3 Gold Ru2+
N N nanoparticles
O N
+N N+ Me
Indium-tin O
O
oxide O Me
O
Adsorption of O
gold nanoparticles
c)
O +
O Si NH3 Gold Gold
O
Indium-tin
oxide
PO3H
d) +
O N
O P
N
N N N+ Me
Titanium O
dioxide O Ru2+
N N
O P N Me
Tin O
oxide
PO3H
gold nanoparticles having a diameter of ≈ 13 nm [2.53, and the reduction of the bipyridinium units. The peak
54]. Electrostatic interactions promote the adsorption of current for both processes increases with the number
the nanoparticles on the organic layer (Fig. 2.17a). The of alternating layers. Comparison of these values indi-
treatment of the composite film with the bipyridinium cates that the ratio between the number of tetracationic
cyclophane 17 produces an organic layer on the gold cyclophanes and that of the nanoparticles is ≈ 100 : 1.
nanoparticles (Fig. 2.18b). Following this approach, The tetracationic cyclophane 17 binds dioxyarenes
alternating layers of inorganic nanoparticles and organic in solution [2.55, 56]. Attractive supramolecular forces
building blocks can be assembled on the indium-tin between the electron deficient bipyridinium units and
oxide support. Cyclic voltammograms of the resulting the electron rich guests are responsible for complexa-
materials show the oxidation of the gold nanoparticles tion. This recognition motif can be exploited to probe
38 Part A Nanostructures, Micro-/Nanofabrication and Materials
the ability of the composite films in Fig. 2.17b,c to sense the characteristic band of the radical cation in the ab-
Part A 2.3
electron rich analytes. In particular, hydroquinone is sorption spectrum. This band persists for hours under
expected to enter the electron deficient cavities of the open circuit conditions. But it fades in ≈ 15 s under
surface-confined cyclophanes. Cyclic voltammograms a voltage bias of +1 V, as the radical cation is oxidized
consistently reveal the redox waves associated with the back to the dicationic form. In summary, an optical
reversible oxidation of hydroquinone even when very stimulation accompanied by a negative voltage bias re-
small amounts of the guest (≈ 1 × 10−5 M) are added duces the bipyridinium building block. The state of the
to the electrolyte solution [2.53, 54]. No redox response photogenerated form can be read optically, recording
can be detected with a bare indium-tin oxide electrode the absorption spectrum in the visible region, and erased
under otherwise identical conditions. The supramolecu- electrically, applying a positive voltage pulse.
lar association of the guest and the surface confined
cyclophanes increases the local concentration of hydro- 2.3.4 Nanogaps and Nanowires
quinone at the electrode/solution interface enabling its
electrochemical detection. The operating principles of the electroactive and pho-
Following a related strategy, the [2]catenane 18 toactive devices illustrated in Figs. 2.12–2.17 exploit
(Fig. 2.17) can be incorporated into similar composite the ability of small collections of molecular compo-
arrays [2.57, 58]. This interlocked molecule incorpo- nents to manipulate electrons and photons. Designed
rates a Ru(II)/trisbipyridine sensitizer and two bipyri- molecules are deposited on relatively large electrodes
dinium acceptors. Upon irradiation of the composite and can be addressed electrically and/or optically by
material at 440 nm, photoinduced electron transfer from controlling the voltage of the support and/or illumi-
the sensitizer to the appended acceptors occurs. The nating its surface. The transition from devices relying
photogenerated hole in the sensitizer is filled after the on collections of molecules to unimolecular devices re-
transfer of an electron from a sacrificial electron donor quires the identification of practical methods to contact
present in the electrolyte solution. Under a positive volt- single molecules. This fascinating objective demands
age bias applied to the supporting electrode, an electron the rather challenging miniaturization of contacting
flow from the bipyridinium acceptors to the indium- electrodes to the nanoscale.
tin oxide support is established. The resulting current A promising approach to unimolecular devices re-
switches between high and low values as the light lies on the fabrication of nanometer-sized gaps in
source is turned on and off. metallic features followed by the insertion of indi-
Another photoresponsive device, assembled com- vidual molecules between the terminals of the gap.
bining inorganic nanoparticles with molecular building This strategy permits the assembly of nanoscaled
blocks, is illustrated in Fig. 2.17d. Phosphonate groups three-terminal devices equivalent to conventional tran-
can be used to anchor a Ru(II)/trisbipyridine complex sistors [2.61–63]. A remarkable example is illustrated
with an appended bipyridinium dication to titanium in Fig. 2.18a [2.61]. It incorporates a single molecule
dioxide nanoparticles deposited on a doped tin ox- in the nanogap generated between two gold electrodes.
ide electrode [2.59, 60]. The resulting composite array Initially electron beam lithography is used to pat-
can be integrated in a conventional electrochemi- tern a gold wire on a doped silicon wafer covered
cal cell filled with an aqueous electrolyte containing by an insulating silicon dioxide layer. Then the gold
triethanolamine. Under a bias voltage of − 0.45 V feature is broken by electromigration to generate the
and irradiation at 532 nm, 95% of the excited ruthe- nanogap. The lateral size of the separated electrodes
nium centers transfer electrons to the titanium dioxide is ≈ 100 nm and their thickness is ≈ 15 nm. Scanning
nanoparticles. The other 5% donate electrons to the electron microcopy indicates that the facing surfaces
bipyridinium dications. All the electrons transferred to of the separated electrodes are not uniform and that
the bipyridinium acceptors return to the ruthenium cen- tiny gaps between their protrusions are formed. Cur-
ters, while only 80% of those accepted by the nanopar- rent/voltage measurements suggest that the size of the
ticles return to the transition metal complexes. The smallest nanogap is ≈ 1 nm. When the breakage of the
remaining 15% reach the bipyridinium acceptors, while gold feature is preceded by the deposition of a dilute
electron transfer from sacrificial triethanolamine donors toluene solution of C60 (19), junctions with enhanced
fills the photogenerated holes left in the ruthenium conduction are obtained. This particular molecule has
sensitizers. The photoinduced reduction of the bipyri- a diameter of ≈ 0.7 nm and can insert in the nanogap
dinium dication is accompanied by the appearance of facilitating the flow of electrons across the junction.
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.3 Solid State Devices 39
Part A 2.3
a) Gold Molecule Gold
a gold wire with a width of ≈ 200 nm and a thicknesses
source drain
of ≈ 10–15 nm is patterned on the silicon dioxide over-
layer by electron beam lithography. After extensive
Silicon gate
washing of the substrate with acetone and methylene
chloride and cleaning with oxygen plasma, the gold
Silicon dioxide insulator wire is exposed to a solution of the bisthiol 20. The
formation of thiolate–gold bonds promotes the self-
assembly of the molecular building block on the gold
19 surface. At this point, electromigration-induced break-
age produces a gap of 1–2 nm in the gold wire. The
surface-confined bisthiol 20 is only 0.24 nm long and,
therefore, it can insert in the nanogap producing an
N N electrode/molecule/electrode junction.
The cobalt center in 20 can be oxidized/reduced
N Co2+ N
HS SH reversibly between Co(II) and Co(III) [2.63]. When
N N this electroactive molecule is inserted in a nanogap
20
(Fig. 2.18a), its ability to accept and donate electrons
dictates the current/voltage profile of the resulting elec-
b) Platinum DNA Silicon trode/molecule/electrode junction. More precisely, no
nitride
current flows across the junction below a certain volt-
age threshold. As the source voltage is raised above
this particular value, the drain current increases in
steps. The threshold associated with the source volt-
age varies in magnitude with the gate voltage. This
Silicon
intriguing behavior is a consequence of the finite
Silicon dioxide energy necessary to oxidize/reduce the cobalt cen-
ter and of a change in the relative stabilities of the
Fig. 2.18 (a) Nanoscaled transistors can be fabricated in- oxidized and reduced forms Co(II) and Co(III) with
serting a single molecule (19 or 20) between source and the gate voltage. In summary, the conduction of the
drain electrodes mounted on a silicon/silicon dioxide sup- electrode/molecule/electrode junction can be tuned ad-
port. (b) A DNA nanowire can bridge nanoelectrodes justing the voltage of the silicon support. The behavior
suspended above a silicon dioxide support of this molecule-based nanoelectronic device is equiva-
lent to that of a conventional transistor [2.19]. In both
The unique configuration of the molecule-based instances, the gate voltage regulates the current flowing
device in Fig. 2.18a can reproduce the functions of from the source to the drain.
a conventional transistor [2.19] at the nanoscale. The The electromigration-induced breakage of pre-
two gold terminals of the junction are the drain and formed metallic features successfully produces nano-
source of this nanotransistor, and the underlying silicon gaps by moving apart two fragments of the same wire.
wafer is the gate. At a temperature of 1.5 K, the junc- Alternatively, nanogaps can be fabricated reducing the
tion conductance is very small, when the gate bias is separation of the two terminals of much larger gaps.
low, and increases in steps at higher voltages [2.61]. The For example, gold electrodes separated by a distance
conductance gap is a consequence of the finite energy of 20–80 nm can be patterned on a silicon/silicon diox-
required to oxidize/reduce the single C60 positioned in ide substrate by electron beam lithography [2.64]. The
the junction. It is interesting that the zero-conductance relatively large gap between them can be reduced sig-
window also changes with the gate voltage and can be nificantly by the electrochemical deposition of gold on
opened and closed reversibly adjusting the gate bias. the surfaces of both electrodes. The final result is the
A similar strategy can be employed to fabricate fabrication of two nanoelectrodes separated by ≈ 1 nm
a nanoscaled transistor incorporating the Co(II) com- and with a radius of curvature of 5–15 nm. The two
plex 20 shown in Fig. 2.18 [2.63]. In this instance, terminals of this nanogap can be contacted by organic
a silicon dioxide layer with a thickness of ≈ 30 nm nanowires grown between them [2.65]. In particular, the
40 Part A Nanostructures, Micro-/Nanofabrication and Materials
electropolymerization of aniline produces polyaniline below 1 pA at low voltage biases. Under these condi-
Part A 2.3
bridges between the gold nanoelectrodes. The conduc- tions, the DNA nanowire is an insulator. Above a certain
tance of the resulting junction can be probed immersing voltage threshold, however, the nanowire becomes con-
the overall assembly in an electrolyte solution. Em- ducting and currents up to 100 nA can flow across the
ploying a bipotentiostat, the bias voltage of the two junction through a single nanowire. Assuming that di-
terminals of the junction can be maintained at 20 mV, rect tunneling from electrode to electrode is extremely
while their potentials are scanned relative to that of a sil- unlikely for a relatively large gap of 8 nm, the intrigu-
ver/silver chloride reference electrode. Below ≈ 0.15 V, ing current/voltage behavior has to be a consequence
the polymer wire is in an insulating state and the current of the participation of the molecular states in the elec-
flowing across the junction is less than 0.05 nA. At this tron transport process. Two possible mechanisms can
voltage threshold, however, the current raises abruptly be envisaged. Sequential hopping of the electrons be-
to ≈ 30 nA. This value corresponds to a conductivity tween states localized in the DNA base pairs can allow
for the polymer nanojunction of 10–100 S cm−1 . When the current flow above a certain voltage threshold. But
the potential is lowered again below the threshold, the this mechanism would presumably result in a Coulomb
current returns back to very low values. The abrupt de- blockade voltage gap that is not observed experimen-
crease in current in the backward scan is observed at tally. More likely, electronic states delocalized across
a potential that is slightly more negative than that caus- the entire length of the DNA nanowire are produc-
ing the abrupt current increase in the forward scan. In ing a molecular conduction band. The off-set between
summary, the conductance of this nanoscaled junction the molecular conduction band and the Fermi levels of
switches on and off as a potential input is switched the electrodes is responsible for the insulating behav-
above and below a voltage threshold. ior at low biases. Above a certain voltage threshold,
It is interesting to note that the influence of organic the molecular band and one of the Fermi levels align
bridges on the junction conductance can be exploited facilitating the passage of electrons across the junction.
for chemical sensing. Nanogaps fabricated following Carbon nanotubes are extremely versatile build-
a similar strategy but lacking the polyaniline bridge ing blocks for the assembly of nanoscaled electronic
alter their conduction after exposure to dilute solutions devices. They can be used to bridge nanogaps [2.69–72]
of small organic molecules [2.66]. Indeed, the organic and assemble nanoscaled cross junctions [2.73–75].
analytes dock into the nanogaps producing a marked In Fig. 2.19a, a single-wall carbon nanotube crosses
decrease in the junction conductance. The magnitude of over another one in an orthogonal arrangement [2.73].
the conductance drop happens to be proportional to the Both nanotubes have electrical contacts at their ends.
analyte–nanoelectrode binding strength. Thus the pres- The fabrication of this device involves three main
ence of the analyte in solution can be detected probing steps. First, alignment marks for the electrodes are
the current/voltage characteristics of the nanogaps. patterned on a silicon/silicon dioxide support by elec-
Nanogaps between electrodes patterned on sil- tron beam lithography. Then the substrate is exposed
icon/silicon dioxide supports can be bridged also by to a dichloromethane suspension of single-wall SWNT
DNA double strands [2.67,68]. The device in Fig. 2.18b carbon nanotubes. After washing with isopropanol,
has a 10.4 nm long poly(G)–poly(C) DNA oligomer crosses of carbon nanotubes in an appropriate alignment
suspended between two nanoelectrodes. It can be fab- relative to the electrode marks are identified by tapping
ricated patterning a 30 nm wide slit in a silicon nitride mode atomic force microscopy. Finally chromium/gold
overlayer covering a silicon/silicon dioxide support by electrodes are fabricated on top of the nanotube ends,
electron beam evaporation. Underetching the silicon again, by electron beam lithography. The conductance
dioxide layer leaves a silicon nitride finger, which can of individual nanotubes can be probed by exploiting
be sputtered with a platinum layer and chopped to leave the two electric contacts at their ends. These two-
a nanogap of 8 nm. At this point, a microdroplet of terminal measurements reveal that certain nanotubes
a dilute solution of DNA is deposited on the device have metallic behavior, while others are semiconduct-
and a bias of 5 V is applied between the two elec- ing. It follows that three distinct types of cross junctions
trodes. Electrostatic forces encourage the deposition of differing in the nature of their constituent nano-
a single DNA wire on top of the nanogap. As soon tubes can be identified on the silicon/silicon dioxide
as the nanowire is in position, current starts to flow support. Four terminal current/voltage measurements
across the junction. The current/voltage signature of indicate that junctions formed by two metallic nano-
the electrode/DNA/electrode junction shows currents tubes have high conductance and ohmic behavior.
Nanomaterials Synthesis and Applications: Molecule-Based Devices 2.3 Solid State Devices 41
Part A 2.3
ior is observed when two semiconducting nanotubes a)
cross. The current/voltage signature of junctions formed Chromium/
gold
when a metallic nanotube crosses a semiconducting one electrodes
are, instead, completely different. The metallic nano-
tube depletes the semiconducting one at the junction
region producing a nanoscaled Schottky barrier with
a pronounced rectifying behavior.
Carbon
Similar fabrication strategies can be exploited to nanotubes
assemble nanoscaled counterparts of conventional tran-
sistors. The device in Fig. 2.19b is assembled patterning
an aluminum finger on a silicon/silicon dioxide sub-
strate by electron beam lithography [2.75]. After b) Source Carbon Drain
exposure to air, an insulating aluminum oxide layer gold nanotubes gold
Aluminum
forms on the aluminum finger. Then a dichloromethane oxide
suspension of single-wall carbon nanotubes is deposited Aluminum gate insulator
on the resulting substrate. Atomic force microscopy
Silicon
can be used to select carbon nanotubes with a diam- dioxide
Silicon
eter of ≈ 1 nm positioned on the aluminum finger. After
registering their coordinates relative to alignment mark-
ers, gold contacts can be evaporated on their ends
c) Nano- Output
by electron beam lithography. The final assembly is transistor voltage
a nanoscaled three-terminal device equivalent to a con-
ventional field effect transistor [2.19]. The two gold – 1.5 V
contacts are the source and drain terminals, while the
Input voltage
underlying aluminum finger reproduces the function of
the gate. At a source to drain bias of ≈ − 1.3 V, the
drain current jumps from ≈ 0 to ≈ 50 nA when the gate d) Input voltage Nanotransistor
voltage is lowered from − 1.0 to − 1.3 V. Thus mod-
Output voltage
erate changes in the gate voltage vary significantly the
current flowing through the nanotube-based device in – 1.5 V
analogy to a conventional enhancement mode p-type
field effect transistor [2.19].
The nanoscaled transistor in Fig. 2.18a has a micro- Input voltage Nanotransistor
scaled silicon gate that extends under the entire
chip [2.61, 63]. The configuration in Fig. 2.19b, in- Fig. 2.19 (a) Nanoscaled junctions can be assembled on
stead, has nanoscaled aluminum gates for every single silicon/silicon dioxide supports crossing pairs of or-
carbon nanotube transistor fabricated on the same thogonally arranged single-wall carbon nanotubes with
support [2.75]. It follows that multiple nanoscaled tran- chromium/gold electrical contacts at their ends. (b) Nano-
sistors can be fabricated on the same chip and operated transistors can be fabricated contacting the two ends
independently following this strategy. This unique fea- of a single-wall carbon nanotube deposited on an alu-
ture offers the possibility of fabricating nanoscaled minum/aluminum oxide gate with gold sources and drain.
digital circuits by interconnecting the terminals of One or two nanotube transistors can be integrated into
independent nanotube transistors. The examples in nanoscaled NOT (c) and NOR (d) logic gates
Fig. 2.19c,d illustrate the configurations of nanoscaled
NOT and NOR gates implemented using one or two nal. In particular, a voltage input of − 1.5 V lowers the
nanotube transistors. In Fig. 2.19c, an off-chip bias re- nanotube resistance (26 MΩ) below that of the bias re-
sistor is connected to the drain terminal of a single sistor (100 MΩ). As a result, the voltage output drops
transistor while the source is grounded. A voltage input to 0 V. When the voltage input is raised to 0 V, the
applied to the gate modulates the nanotube conductance nanotube resistance increases above that of the bias re-
altering the voltage output probed at the drain termi- sistor and the voltage output becomes − 1.5 V. Thus
42 Part A Nanostructures, Micro-/Nanofabrication and Materials
the output of this nanoelectronic device switches from age input and the voltage output of the device can be
Part A 2.4
a high (0 V) and to a low ( − 1.5 V) level as the in- probed at their interconnected drain terminals. When
put shifts from a low ( − 1.5 V) to a high (0 V) value. the resistance of at least one of the two nanotubes is
The inverse relation between input and output trans- below that of the resistor, the output is 0 V. When
lates into a NOT operation if a negative logic convention both nanotubes are in a nonconducting mode, the out-
(low = 1, high = 0) is applied to both signals. put voltage is − 1.5 V. Thus if a low voltage input
In Fig. 2.15d, the source terminals of two indepen- − 1.5 V is applied to one or both transistors, the out-
dent nanotube transistors fabricated on the same chip put is high (0 V). When both voltage inputs are high
are connected by a gold wire and grounded [2.75]. Simi- (0 V), the output is low ( − 1.5 V). If a negative logic
larly, the two drain terminals are connected by another convention (low = 1, high = 0) is applied to all signals,
gold wire and contacted to an off-chip bias resistors. the signal transduction behavior translates in to a NOR
The gate of each nanotube can be stimulated with a volt- operation.
integrated within functioning nanoelectronic devices rate molecules into reliable device architectures. As we
Part A 2
capable of simple logic operations. continue to gather further insights in these directions,
The remarkable examples of molecule-based mater- design criteria for a wide diversity of molecule-based
ials and devices now available demonstrate the great po- devices will emerge. It is not unrealistic to foresee the
tential and promise for this research area. At this stage, evolution of an entire generation of nanoscaled devices,
the only limit left to the design of functional molecules based on engineered molecular components, that will
is the imagination of the synthetic chemist. All sort find applications in a variety of fields ranging from
of molecular building blocks with tailored dimensions, biomedical research to information technology. Perhaps
shapes, and properties are more or less accessible with nature can once again illuminate our path, teaching us
the assistance of modern chemical synthesis. Now, the not only how to synthesize nanostructured molecules
major challenges are (1) to master the operating prin- but also how to use them. After all, nature is replete
ciples of the molecule-based devices that have been with examples of extremely sophisticated molecule-
and continue to be assembled and (2) to expand and based devices. From tiny bacteria to higher animals, we
improve the fabrication strategies available to incorpo- are all a collection of molecule-based devices.
References
2.1 D. Voet, J.G. Voet: Biochemistry (Wiley, New York 2.13 V. Balzani, M. Venturi, A. Credi: Molecular Devices
1995) and Machines (Wiley-VCH, Weinheim 2003)
2.2 K.C. Nicolau, E.C. Sorensen: Classics in Total Synthesis 2.14 P.R. Ashton, R. Ballardini, V. Balzani, A. Credi,
(VCH, Weinheim 1996) K.R. Dress, E. Ishow, C.J. Kleverlaan, O. Ko-
2.3 J.-M. Lehn: Supramolecular Chemistry: Concepts and cian, J.A. Preece, N. Spencer, J.F. Stoddart,
Perspectives (VCH, Weinheim 1995) M. Venturi, S. Wenger: A photochemically driven
2.4 M.M. Harding, U. Koert, J.-M. Lehn, A. Marquis- molecular-level abacus, Chem. Eur. J. 6, 3558–3574
Rigault, C. Piguet, J. Siegel: Synthesis of unsub- (2000)
stituted and 4,4 -substituted oligobipyridines as 2.15 M. Irié (Ed.): Photochromism: memories and
ligand strands for helicate self-assembly, Helv. switches, Chem. Rev. 100, 1683–1890 (2000)
Chim. Acta 74, 594–610 (1991) 2.16 B.L. Feringa (Ed.): Molecular Switches (Wiley-VCH,
2.5 J.-M. Lehn, A. Rigault, J. Siegel, B. Harrowfield, Weinheim 2001)
B. Chevrier, D. Moras: Spontaneous assembly of 2.17 R.J. Mitchell: Microprocessor Systems: An Introduc-
double-stranded helicates from oligobipyridine lig- tion (Macmillan, London 1995)
ands and copper(I) cations: Structure of an inorganic 2.18 D.R. Smith: Digital Transmission Systems (Van Nos-
double helix, Proc. Natl. Acad. Sci. USA 84, 2565– trand Reinhold, New York 1993)
2569 (1987) 2.19 S. Madhu: Electronics: Circuits and Systems (SAMS,
2.6 J.-M. Lehn, A. Rigault: Helicates: Tetra- and Indianapolis 1985)
pentanuclear double helix complexes of Cu(I) and 2.20 F.M. Raymo: Digital processing and communication
poly(bipyridine) strands, Angew. Chem. Int. Ed. with molecular switches, Adv. Mater. 14, 401–414
Engl. 27, 1095–1097 (1988) (2002)
2.7 J.D. Hartgerink, J.R. Granja, R.A. Milligan, M.R. Gha- 2.21 A.P. de Silva: Molecular computation – Molecular
diri: Self-assembling peptide nanotubes, J. Am. logic gets loaded, Nat. Mater. 4, 15–16 (2005)
Chem. Soc. 118, 43–50 (1996) 2.22 A. Aviram: Molecules for memory, logic and ampli-
2.8 M.R. Ghadiri, J.R. Granja, R.A. Milligan, D.E. McRee, fication, J. Am. Chem. Soc. 110, 5687–5692 (1988)
N. Khazanovich: Self-assembling organic nanotubes 2.23 A.P. de Silva, H.Q.N. Gunaratne, C.P. McCoy: A mo-
based on a cyclic peptide architecture, Nature 366, lecular photoionic AND gate based on fluorescent
324–327 (1993) signaling, Nature 364, 42–44 (1993)
2.9 V. Balzani, A. Credi, F.M. Raymo, J.F. Stoddart: Ar- 2.24 M. Asakawa, P.R. Ashton, V. Balzani, A. Credi,
tificial molecular machines, Angew. Chem. Int. Ed. G. Mattersteig, O.A. Matthews, M. Montalti,
39, 3348–3391 (2000) N. Spencer, J.F. Stoddart, M. Venturi: Electro-
2.10 A.J. Bard, L.R. Faulkner: Electrochemical Methods: chemically induced molecular motions in pseu-
Fundamentals and Applications (Wiley, New York dorotaxanes: A case of dual-mode (oxidative and
2000) reductive) dethreading, Chem. Eur. J. 3, 1992–1996
2.11 V. Balzani (Ed.): Electron Transfer in Chemistry (1997)
(Wiley-VCH, Weinheim 2001) 2.25 F.M. Raymo, S. Giordani, A.J.P. White, D.J. Williams:
2.12 J.D. Coyle: Principles and Applications of Photo- Digital processing with a three-state molecular
chemistry (Wiley, New York 1988) switch, J. Org. Chem. 68, 4158–4169 (2003)
44 Part A Nanostructures, Micro-/Nanofabrication and Materials
2.26 F.M. Raymo, S. Giordani: Signal communication be- turi, A.J.P. White, D.J. Williams: A chemically and
Part A 2
tween molecular switches, Org. Lett. 3, 3475–3478 electrochemically switchable [2]catenane incorpo-
(2001) rating a tetrathiafulvalene unit, Angew. Chem. Int.
2.27 F.M. Raymo, S. Giordani: Multichannel digital trans- Ed. 37, 333–337 (1998)
mission in an optical network of communicating 2.43 V. Balzani, A. Credi, G. Mattersteig, O.A. Matthews,
molecules, J. Am. Chem. Soc. 124, 2004–2007 (2002) F.M. Raymo, J.F. Stoddart, M. Venturi, A.J.P. White,
2.28 F.M. Raymo, S. Giordani: All-optical processing with D.J. Williams: Switching of pseudorotaxanes and
molecular switches, Proc. Natl. Acad. Sci. USA 99, catenanes incorporating a tetrathiafulvalene unit
4941–4944 (2002) by redox and chemical inputs, J. Org. Chem. 65,
2.29 A.J. Bard: Integrated Chemical Systems: A Chemical 1924–1936 (2000)
Approach to Nanotechnology (Wiley, New York 1994) 2.44 M. Asakawa, M. Higuchi, G. Mattersteig, T. Naka-
2.30 C. Joachim, J.K. Gimzewski, A. Aviram: Electronics mura, A.R. Pease, F.M. Raymo, T. Shimizu, J.F. Stod-
using hybrid-molecular and mono-molecular de- dart: Current/voltage characteristics of monolayers
vices, Nature 408, 541–548 (2000) of redox-switchable [2]catenanes on gold, Adv.
2.31 J.M. Tour: Molecular electronics. Synthesis and test- Mater. 12, 1099–1102 (2000)
ing of components, Acc. Chem. Res. 33, 791–804 2.45 C.P. Collier, G. Mattersteig, E.W. Wong, Y. Luo,
(2000) K. Beverly, J. Sampaio, F.M. Raymo, J.F. Stod-
2.32 A.R. Pease, J.O. Jeppesen, J.F. Stoddart, Y. Luo, dart, J.R. Heath: A [2]catenane based solid-state
C.P. Collier, J.R. Heath: Switching devices based on electronically reconfigurable switch, Science 289,
interlocked molecules, Acc. Chem. Res. 34, 433–444 1172–1175 (2000)
(2001) 2.46 C.P. Collier, J.O. Jeppesen, Y. Luo, J. Perkins,
2.33 R.M. Metzger: Unimolecular electrical rectifiers, E.W. Wong, J.R. Heath, J.F. Stoddart: Molecular-
Chem. Rev. 103, 3803–3834 (2003) based electronically switchable tunnel junction
2.34 M.C. Petty: Langmuir–Blodgett Films: An Introduc- devices, J. Am. Chem. Soc. 123, 12632–12641
tion (Cambridge Univ. Press, Cambridge 1996) (2001)
2.35 A. Ulman: An Introduction to Ultrathin Organic Films 2.47 J. Chen, M.A. Reed, A.M. Rawlett, J.M. Tour: Large
(Academic, Boston 1991) on-off ratios and negative differential resistance in
2.36 C. Lee, A.J. Bard: Comparative electrochemical a molecular electronic device, Science 286, 1550–
studies of N-methyl-N -hexadecyl viologen mono- 1552 (1999)
molecular films formed by irreversible adsorption 2.48 M.A. Reed, J. Chen, A.M. Rawlett, D.W. Price,
and the Langmuir–Blodgett method, J. Electroanal. J.M. Tour: Molecular random access memory cell,
Chem. 239, 441–446 (1988) Appl. Phys. Lett. 78, 3735–3737 (2001)
2.37 C. Lee, A.J. Bard: Cyclic voltammetry and Langmuir 2.49 D.I. Gittins, D. Bethell, R.J. Nichols, D.J. Schiffrin:
film isotherms of mixed monolayers of N-docosoyl- Redox-controlled multilayers of discrete gold par-
N -methyl viologen with arachidic acid, Chem. Phys. ticles: A novel electroactive nanomaterial, Adv.
Lett. 170, 57–60 (1990) Mater. 9, 737–740 (1999)
2.38 M. Fujihira, K. Nishiyama, H. Yamada: Photo- 2.50 D.I. Gittins, D. Bethell, R.J. Nichols, D.J. Schiffrin:
electrochemical responses of optically transparent Diode-like electron transfer across nanostructured
electrodes modified with Langmuir–Blodgett films films containing a redox ligand, J. Mater. Chem. 10,
consisting of surfactant derivatives of electron 79–83 (2000)
donor, acceptor and sensitizer molecules, Thin Solid 2.51 D.I. Gittins, D. Bethell, D.J. Schiffrin, R.J. Nichols:
Films 132, 77–82 (1985) A nanometer-scale electronic switch consisting of
2.39 M. Fujihira: Photoelectric conversion with Lang- a metal cluster and redox-addressable groups, Na-
muir–Blodgett films. In: Nanostructures Based on ture 408, 67–69 (2000)
Molecular Materials, ed. by W. Göpel, C. Ziegler (VCH, 2.52 A.N. Shipway, M. Lahav, I. Willner: Nanostructured
Weinheim 1992) pp. 27–46 gold colloid electrodes, Adv. Mater. 12, 993–998
2.40 C.P. Collier, E.W. Wong, M. Belohradsky, F.M. Raymo, (2000)
J.F. Stoddart, P.J. Kuekes, R.S. Williams, J.R. Heath: 2.53 A.N. Shipway, M. Lahav, R. Blonder, I. Willner: Bis-
Electronically configurable molecular-based logic bipyridinium cyclophane receptor–Au nanoparticle
gates, Science 285, 391–394 (1999) superstructure for electrochemical sensing applica-
2.41 E.W. Wong, C.P. Collier, M. Belohradsky, F.M. Raymo, tions, Chem. Mater. 11, 13–15 (1999)
J.F. Stoddart, J.R. Heath: Fabrication and transport 2.54 M. Lahav, A.N. Shipway, I. Willner, M.B. Nielsen,
properties of single-molecule-thick electrochemi- J.F. Stoddart: An enlarged bis-bipyridinum cyclo-
cal junctions, J. Am. Chem. Soc. 122, 5831–5840 phane–Au nanoparticle superstructure for selective
(2000) electrochemical sensing applications, J. Electroanal.
2.42 M. Asakawa, P.R. Ashton, V. Balzani, A. Credi, Chem. 482, 217–221 (2000)
C. Hamers, G. Mattersteig, M. Montalti, A.N. Ship- 2.55 R.E. Gillard, F.M. Raymo, J.F. Stoddart: Controlling
way, N. Spencer, J.F. Stoddart, M.S. Tolley, M. Ven- self-assembly, Chem. Eur. J. 3, 1933–1940 (1997)
Nanomaterials Synthesis and Applications: Molecule-Based Devices References 45
2.56 F.M. Raymo, J.F. Stoddart: From supramolecular 2.65 H. He, J. Zhu, N.J. Tao, L.A. Nagahara, I. Amlani,
Part A 2
complexes to interlocked molecular compounds, R. Tsui: A conducting polymer nanojunction switch,
Chemtracts Org. Chem. 11, 491–511 (1998) J. Am. Chem. Soc. 123, 7730–7731 (2001)
2.57 M. Lahav, T. Gabriel, A.N. Shipway, I. Willner: As- 2.66 A. Bogozi, O. Lam, H. He, C. Li, N.J. Tao, L.A. Na-
sembly of a Zn(II)-porphyrin-bipyridinium dyad and gahara, I. Amlani, R. Tsui: Molecular adsorption
Au-nanoparticle superstructures on conductive sur- onto metallic quantum wires, J. Am. Chem. Soc. 123,
faces, J. Am. Chem. Soc. 121, 258–259 (1999) 4585–4590 (2001)
2.58 M. Lahav, V. Heleg-Shabtai, J. Wasserman, E. Katz, 2.67 A. Bezryadin, C.N. Lau, M. Tinkham: Quan-
I. Willner, H. Durr, Y. Hu, S.H. Bossmann: Photo- tum suppression of superconductivity in ultrathin
electrochemistry with integrated photosensitizer- nanowires, Nature 404, 971–974 (2000)
electron acceptor Au-nanoparticle arrays, J. Am. 2.68 D. Porath, A. Bezryadin, S. de Vries, C. Dekker: Dir-
Chem. Soc. 122, 11480–11487 (2000) ect measurement of electrical transport through DNA
2.59 G. Will, S.N. Rao, D. Fitzmaurice: Heterosupramo- molecules, Nature 403, 635–638 (2000)
lecular optical write-read-erase device, J. Mater. 2.69 S.J. Tans, M.H. Devoret, H. Dai, A. Thess, E.E. Smal-
Chem. 9, 2297–2299 (1999) ley, L.J. Geerligs, C. Dekker: Individual single-wall
2.60 A. Merrins, C. Kleverlann, G. Will, S.N. Rao, carbon nanotubes as quantum wires, Nature 386,
F. Scandola, D. Fitzmaurice: Time-resolved optic- 474–477 (1997)
al spectroscopy of heterosupramolecular assemblies 2.70 A.F. Morpurgo, J. Kong, C.M. Marcus, H. Dai:
based on nanostructured TiO2 films modified by Gate-controlled superconducting proximity ef-
chemisorption of covalently linked ruthenium and fect in carbon nanotubes, Nature 286, 263–265
viologen complex components, J. Phys. Chem. B 105, (1999)
2998–3004 (2001) 2.71 J. Nygård, D.H. Cobden, P.E. Lindelof: Kondo physics
2.61 H. Park, J. Park, A.K.L. Lim, E.H. Anderson, in carbon nanotubes, Nature 408, 342–346 (2000)
A.P. Alivisatos, P.L. McEuen: Nanomechanical oscil- 2.72 W. Liang, M. Bockrath, D. Bozovic, J.H. Hafner,
lations in a single C60 transistor, Nature 407, 57–60 M. Tinkham, H. Park: Fabry–Perot interference in
(2000) a nanotube electron waveguide, Nature 411, 665–
2.62 W. Liang, M.P. Shores, M. Bockrath, J.R. Long, 669 (2001)
H. Park: Kondo resonance in a single-molecule tran- 2.73 M.S. Fuhrer, J. Nygård, L. Shih, M. Forero, Y.-G. Yoon,
sistor, Nature 417, 725–729 (2002) M.S.C. Mazzoni, H.J. Choi, J. Ihm, S.G. Louie, A. Zettl,
2.63 J. Park, A.N. Pasupathy, J.I. Goldsmith, C. Chang, P.L. McEuen: Crossed nanotube junctions, Science
Y. Yaish, J.R. Petta, M. Rinkoski, J.P. Sethna, 288, 494–497 (2000)
H.D. Abruna, P.L. McEuen, D.C. Ralph: Coulomb 2.74 T. Rueckes, K. Kim, E. Joselevich, G.Y. Tseng, C.-
blockade and the Kondo effect in single-atom tran- L. Cheung, C.M. Lieber: Carbon nanotube-based
sistors, Nature 417, 722–725 (2002) nonvolatile random access memory for molecular
2.64 C.Z. Li, H.X. He, N.J. Tao: Quantized tunneling current computing, Science 289, 94–97 (2000)
in the metallic nanogaps formed by electrodeposi- 2.75 A. Bachtold, P. Hadley, T. Nakanishi, C. Dekker: Logic
tion and etching, Appl. Phys. Lett. 77, 3995–3997 circuits with carbon nanotube transistors, Science
(2000) 294, 1317–1320 (2001)
47
Introduction
3. Introduction to Carbon Nanotubes
Marc Monthioux, Philippe Serp, Emmanuel Flahaut, Manitra Razafinimanana, Christophe Laurent,
Part A 3
Alain Peigney, Wolfgang Bacsa, Jean-Marc Broto
Carbon nanotubes are remarkable objects that by reactions and associations of all-carbon
look set to revolutionize the technological land- nanotubes with foreign atoms, molecules and
scape in the near future. Tomorrow’s society compounds, which may provide the path
will be shaped by nanotube applications, just to hybrid materials with even better prop-
as silicon-based technologies dominate so- erties than pristine nanotubes. Finally, we
ciety today. Space elevators tethered by the will describe the most important current and
strongest of cables; hydrogen-powered vehi- potential applications of carbon nanotubes,
cles; artificial muscles: these are just a few of which suggest that the future for the car-
the technological marvels that may be made bon nanotube industry looks very promising
possible by the emerging science of carbon indeed.
nanotubes.
Of course, this prediction is still some way
3.1 Structure of Carbon Nanotubes .............. 48
from becoming reality; we are still at the
3.1.1 Single-Wall Nanotubes ................. 48
stage of evaluating possibilities and poten-
3.1.2 Multiwall Nanotubes .................... 51
tial. Consider the recent example of fullerenes
– molecules closely related to nanotubes. The 3.2 Synthesis of Carbon Nanotubes .............. 53
anticipation surrounding these molecules, first 3.2.1 Solid Carbon Source-Based
reported in 1985, resulted in the bestow- Production Techniques
ment of a Nobel Prize for their discovery in for Carbon Nanotubes ................... 53
1996. However, a decade later, few applica- 3.2.2 Gaseous Carbon Source-Based
tions of fullerenes have reached the market, Production Techniques
suggesting that similarly enthusiastic predictions for Carbon Nanotubes ................... 62
about nanotubes should be approached with 3.2.3 Miscellaneous Techniques ............. 68
caution. 3.2.4 Synthesis of Carbon Nanotubes
There is no denying, however, that the ex- with Controlled Orientation ........... 68
pectations surrounding carbon nanotubes are
3.3 Growth Mechanisms of Carbon Nanotubes 70
very high. One of the main reasons for this
3.3.1 Catalyst-Free Growth .................... 71
is the anticipated application of nanotubes to
3.3.2 Catalytically Activated Growth........ 71
electronics. Many believe that current tech-
niques for miniaturizing microchips are about 3.4 Properties of Carbon Nanotubes............. 74
to reach their lowest limits, and that nanotube- 3.4.1 Overview ..................................... 74
based technologies are the best hope for further 3.4.2 General Properties of SWNTs .......... 75
miniaturization. Carbon nanotubes may therefore 3.4.3 Adsorption Properties of SWNTs ...... 75
provide the building blocks for further tech- 3.4.4 Electronic and Optical Properties .... 77
nological progress, enhancing our standards of 3.4.5 Mechanical Properties................... 79
living. 3.4.6 Reactivity .................................... 79
In this chapter, we first describe the structures,
syntheses, growth mechanisms and properties of 3.5 Carbon Nanotube-Based Nano-Objects ... 80
3.5.1 Heteronanotubes ......................... 80
carbon nanotubes. Then we discuss nanotube-
3.5.2 Hybrid Carbon Nanotubes.............. 80
related nano-objects, including those formed
3.5.3 Functionalized Nanotubes ............. 84
48 Part A Nanostructures, Micro-/Nanofabrication and Materials
Carbon nanotubes have long been synthesized as prod- (Sect. 3.5) led to the discovery – again unexpected – of
ucts of the action of a catalyst on the gaseous species single-wall carbon nanotubes (SWNTs) simultaneously
originating from the thermal decomposition of hydro- by Iijima and Ichihashi [3.8] and Bethune et al. [3.9].
carbons (Sect. 3.2) [3.1]. The first evidence that the Single-wall carbon nanotubes were really new nano-
nanofilaments produced in this way were actually nano- objects with properties and behaviors that are often
tubes – that they exhibited an inner cavity – can be quite specific (Sect. 3.4). They are also beautiful objects
found in the transmission electron microscope micro- for fundamental physics as well as unique molecules for
graphs published by Radushkevich and Lukyanovich in experimental chemistry, although they are still some-
1952 [3.2]. This was of course related to and made what mysterious since their formation mechanisms
possible by the progress in transmission electron mi- are the subject of controversy and are still debated
croscopy. It is then likely that the carbon filaments (Sect. 3.3). Potential applications seem countless, al-
prepared by Hughes and Chambers in 1889 [3.3], though few have reached marketable status so far
which is probably the first patent ever deposited in (Sect. 3.6). Consequently, about five papers a day are
the field, and whose preparation method was also currently published by research teams from around the
based on the catalytically enhanced thermal cracking world with carbon nanotubes as the main topic, an il-
of hydrocarbons, were already carbon nanotube-related lustration of how extraordinarily active – and highly
morphologies. The preparation of vapor-grown carbon competitive – this field of research is. It is an unusual
fibers was actually reported over a century ago [3.4, 5]. situation, similar to that for fullerenes, which, by the
Since then, the interest in carbon nanofilaments/nano- way, are again carbon nano-objects structurally closely
tubes has been recurrent, though within a scientific area related to nanotubes.
almost limited to the carbon material scientist com- This is not, however, only about scientific exalta-
munity. The reader is invited to consult the review tion. Economic aspects are leading the game to a greater
published by Baker and Harris [3.6] regarding the early and greater extent. According to experts, the world mar-
works. Worldwide enthusiasm came unexpectedly in ket was estimated to be more than 430 million dollars in
1991, after the catalyst-free formation of nearly per- 2004 and it is predicted to grow to several billion dol-
fect concentric multiwall carbon nanotubes (c-MWNTs, lars before 2009. That is serious business, and it will be
Sect. 3.1) was reported [3.7] as by-products of the for- closely related to how scientists and engineers deal with
mation of fullerenes via the electric-arc technique. But the many challenges found on the path from the beau-
the real breakthrough occurred two years later, when at- tiful, ideal molecule to the reliable – and it is hoped,
tempts to fill the nanotubes in situ with various metals cheap – manufactured product.
Part A 3.1
single molecules with huge aspect ratios.
T θ Two important consequences derive from the
SWNT structure as described above:
A
1. All carbon atoms are involved in hexagonal aro-
a1 matic rings only and are therefore in equivalent
O Ch
positions, except at each nanotube tip, where 6 ×
a2 5 = 30 atoms are involved in pentagonal rings (con-
sidering that adjacent pentagons are unlikely) –
Fig. 3.1 Sketch of the way to make a single-wall car- though not more, not less, as a consequence of Eu-
bon nanotube, starting from a graphene sheet (adapted ler’s rule that also governs the fullerene structure.
from [3.12]) For ideal SWNTs, chemical reactivity will therefore
be highly favored at the tube tips, at the locations of
the pentagonal rings.
a) 2. Although carbon atoms are involved in aromatic
rings, the C=C bond angles are not planar. This
means that the hybridization of carbon atoms is not
pure sp2 ; it has some degree of the sp3 character, in
a proportion that increases as the tube radius of cur-
b)
vature decreases. The effect is the same as for the
C60 fullerene molecules, whose radius of curvature
is 0.35 nm, and whose bonds therefore have 10% sp3
character [3.14]. On the one hand, this is believed
c) to make the SWNT surface a bit more reactive than
regular, planar graphene, even though it still consists
of aromatic ring faces. On the other hand, this some-
how induces variable overlapping of energy bands,
resulting in unique and versatile electronic behavior
(Sect. 3.4).
Fig. 3.2a–c Sketches of three different SWNT structures As illustrated by Fig. 3.2, there are many ways to
that are examples of (a) a zigzag-type nanotube, (b) an roll a graphene into a single-wall nanotube, with some
armchair-type nanotube, (c) a helical nanotube (adapted of the resulting nanotubes possessing planes of sym-
from [3.13]) metry both parallel and perpendicular to the nanotube
axis (such as the SWNTs from Fig. 3.2a,b), while oth-
ers do not (such as the SWNT from Fig. 3.2c). Similar
SWNTs with diameters as low as 0.4 nm have been to the terms used for molecules, the latter are commonly
synthesized successfully [3.11]. A suitable energetic called chiral nanotubes, since they are unable to be su-
compromise is therefore reached for ≈ 1.4 nm, the most perimposed on their own image in a mirror. Helical is
frequent diameter encountered regardless of the synthe- however sometimes preferred (see below). The various
sis technique (at least for those based on solid carbon ways to roll graphene into tubes are therefore mathe-
sources) when conditions ensuring high SWNT yields matically defined by the vector of helicity Ch , and the
are used. There is no such restriction on the nanotube angle of helicity θ, as follows (referring to Fig. 3.1)
length, which only depends on the limitations of the
preparation method and the specific conditions used OA = Ch = na1 + ma2
50 Part A Nanostructures, Micro-/Nanofabrication and Materials
with
√ √
a 3 a a 3 a
a1 = x + y and a2 = x− y ,
2 2 2 2
where a = 2.46 Å
and
2n + m
cos θ = √ ,
Part A 3.1
a) b)
lowing van der Waals forces to develop. Similar is approximately the same as the intergraphene dis-
reasons make fullerenes gather and order into ful- tance in turbostratic, polyaromatic solids, 0.34 nm (as
lerite crystals and SWNTs into SWNT ropes (Fig. 3.4a). opposed to 0.335 nm in genuine graphite), since the in-
Provided the SWNT diameter distribution is narrow, creasing radius of curvature imposed on the concentric
the SWNTs in ropes tend to spontaneously arrange graphenes prevents the carbon atoms from being ar-
into hexagonal arrays, which correspond to the high- ranged as in graphite, with each of the carbon atoms
est compactness achievable (Fig. 3.4b). This feature from a graphene facing either a ring center or a car-
Part A 3.1
brings new periodicities with respect to graphite or bon atom from the neighboring graphene. However, two
turbostratic polyaromatic carbon crystals. Turbostratic cases allow a nanotube to reach – totally or partially –
structure corresponds to graphenes that are stacked the 3-D crystal periodicity of graphite. One is to con-
with random rotations or translations instead of be- sider a high number of concentric graphenes: concentric
ing piled up following sequential ABAB positions, as graphenes with a long radius of curvature. In this case,
in graphite structure. This implies that no lattice atom the shift in the relative positions of carbon atoms from
plane exists other than the graphene planes themselves superimposed graphenes is so small with respect to
(corresponding to the (001) atom plane family). These that in graphite that some commensurability is possible.
new periodicities give specific diffraction patterns that
are quite different to those of other sp2 -carbon-based
crystals, although hk reflections, which account for the
hexagonal symmetry of the graphene plane, are still
present. On the other hand, 00l reflections, which ac-
count for the stacking sequence of graphenes in regular,
multilayered polyaromatic crystals (which do not ex-
ist in SWNT ropes) are absent. This hexagonal packing
of SWNTs within the ropes requires that SWNTs ex-
hibit similar diameters, which is the usual case for
SWNTs prepared by electric arc or laser vaporization
processes. SWNTs prepared using these methods are
actually about 1.35 nm wide (diameter of a (10, 10)
tube, among others), for reasons that are still unclear
but are related to the growth mechanisms specific to the
conditions provided by these techniques (Sect. 3.3).
Part A 3.2
obviously largely depend on the perfection and the ori-
L2 L1 entation of the graphenes in the tube (for example, the
spiral angles of the nanotubes constituting c-MWNTs
Fig. 3.8 Sketch explaining the various parameters ob- has little importance). Graphene orientation is a matter
tained from high-resolution (lattice fringe mode) transmis- of texture, as described above. Graphene perfection is
sion electron microscopy, used to quantify nanotexture: a matter of nanotexture, which is commonly used to de-
L 1 is the average length of perfect (distortion-free) scribe other polyaromatic carbon materials, and which
graphenes of coherent areas; N is the number of piled- is quantified by several parameters preferably obtained
up graphenes in coherent (distortion-free) areas; L 2 is the from high-resolution transmission electron microscopy
average length of continuous though distorted graphenes (Fig. 3.8). Both texture and nanotexture depend on the
within graphene stacks; β is the average distortion angle. processing conditions. While the texture type is a per-
L 1 and N are related to the la and lc values obtained from manent, intrinsic feature which can only be completely
x-ray diffraction altered upon a severe degradation treatment (such as
oxidation), the nanotexture can be improved by subse-
Unlike SWNTs, whose aspect ratios are so high quent thermal treatments at high temperatures (such as
that it is almost impossible to find the tube tips, the > 2000 ◦ C) and potentially degraded by chemical treat-
aspect ratios for MWNTs (and carbon nanofibers) are ments (such as slightly oxidizing conditions).
relatively large quantities [3.24–26]. Unlike fullerene under inert atmosphere, however, succeed in recovering
formation, which requires the presence of carbon atoms structural quality somewhat [3.29].
in high-temperature media and the absence of oxygen,
the utilization of these techniques for the synthesis of Laser Ablation
nanotubes (of SWNT type at least) requires an addi- After the first laser was built in 1960, physicists im-
tional condition: the presence of catalysts in either the mediately made use of it as a means of concentrating
electrode or the target. a large quantity of energy inside a very small volume
Part A 3.2
The different mechanisms (such as carbon mol- within a relatively short time. The consequence of this
ecule dissociation and atom recombination processes) energy input naturally depends upon the characteristics
involved in these high-temperature techniques take of the device employed. During the interaction between
place at different time scales, from nanoseconds to mi- the laser beam and the material, numerous phenomena
croseconds and even milliseconds. The formation of occur at the same time and/or follow each other within
nanotubes and other graphene-based products occurs the a certain time period, and each of these processes
afterward with a relatively long delay. are sensitive to different parameters such as the charac-
The methods of laser ablation, solar energy, and teristics of the laser beam, the incoming power density
electric arc are all based on one essential mechanism: (also termed the fluence), the nature of the target, and
the energy transfer resulting from the interaction be- the environment surrounding it. For instance, the solid
tween either the target material and an external radiation target can merely heat up, melt or vaporize depending
source (a laser beam or radiation emanating from so- on the power provided.
lar energy) or the electrode and the plasma (in case of While this technique was successfully used to syn-
an electric arc). This interaction causes target or anode thesize fullerene-related structures for the very first
erosion, leading to the formation of a plasma: an elec- time [3.23], the synthesis of SWNTs by laser ablation
trically neutral ionized gas, composed of neutral atoms, took another ten years of research [3.27].
charged particles (molecules and ionized species) and
electrons. The ionization degree of this plasma, defined Laser Ablation – Experimental Devices
by the ratio (n e /(n e + n o )), where n e and n o are the Two types of laser devices are currently utilized for
electron and that of neutral atom densities respectively, carbon nanotube production: lasers operating in pulsed
highlights the importance of energy transfer between mode and lasers operating in continuous mode, with the
the plasma and the material. The characteristics of this latter generally providing a smaller fluence.
plasma and notably the ranges in temperature and con- An example of the layout of a laser ablation device
centrations of the various species present in the plasma is given in Fig. 3.9. A graphite pellet containing the cat-
thereby depend not only on the nature and composi- alyst is placed in the middle of a quartz tube filled with
tion of the target or the electrode but also on the energy inert gas and placed in an oven maintained at a tempera-
transferred. ture of 1200 ◦ C [3.27, 28]. The energy of the laser beam
One of the advantages of these synthesis techniques focused on the pellet permits it to vaporize and sublime
is the ability to vary a large number of parameters the graphite by uniformly bombarding its surface. The
that modify the composition of the high-temperature carbon species, swept along by a flow of neutral gas, are
medium and consequently allow the most relevant pa- then deposited as soot in different regions: on the con-
rameters to be determined so that the optimal conditions
for the control of carbon nanotube formation can be
obtained. However, a major drawback of these tech- Furnace
niques – and of any other technique used to produce
SWNTs – is that the SWNTs formed are not pure: they
are associated with other carbon phases and remnants of Water-cooled
Laser beam Cu collector
the catalyst. Although purification processes have been
proposed in the literature and by some commercial com-
panies for removing these undesirable phases, they are Furnace Graphite
target
all based on oxidation (such as acid-based) processes
that are likely to significantly affect the SWNT struc- Fig. 3.9 Sketch of an early laser vaporization apparatus
ture [3.15]. Subsequent thermal treatments at ≈ 1200 ◦ C (adapted from [3.27, 28])
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 55
ical water-cooled copper collector, on the quartz tube from the graphite powder and the other from an al-
walls, and on the backside of the pellet. loy of transition metals (catalysts), and irradiated them
Various improvements have been made to this de- simultaneously.
vice in order to increase the production efficiency. For A sketch of a synthesis reactor based on the
example, Thess et al. [3.31] employed a second pulsed vaporization of a target at a fixed temperature by a con-
laser that follows the initial impulsion but at a differ- tinuous CO2 laser beam (λ = 10.6 μm) is shown in
ent frequency in order to ensure a more complete and Fig. 3.10 [3.30]. The power can be varied from 100 to
Part A 3.2
efficient irradiation of the pellet. This second impul- 1600 W. The temperature of the target is measured with
sion vaporizes the coarse aggregates issued from the an optical pyrometer, and these measurements are used
first ablation, causing them to participate in the active to regulate the laser power to maintain a constant vapor-
carbon feedstock involved in nanotube growth. Other ization temperature. The gas, heated by contact with the
modifications were suggested by Rinzler et al. [3.29], target, acts as a local furnace and creates an extended
who inserted a second quartz tube of a smaller diameter hot zone, making an external furnace unnecessary. The
coaxially inside the first one. This second tube re- gas is extracted through a silica pipe, and the solid prod-
duces the vaporization zone and so permits an increased ucts formed are carried away by the gas flow through the
amounts of sublimed carbon to be obtained. They also pipe and then collected on a filter. The synthesis yield
arranged the graphite pellet on a revolving system so is controlled by three parameters: the cooling rate of the
that the laser beam uniformly scans its whole surface. medium where the active, secondary catalyst particles
Other groups have realized that, where the target are formed, the residence time, and the temperature (in
contains both the catalyst and the graphite, the latter the range 1000–2100 K) at which SWNTs nucleate and
evaporates first and the pellet surface becomes more grow [3.33].
and more metal-rich, resulting in a decrease in the ef- However, devices equipped with facilities to gather
ficiency of nanotube formation during the course of the data such as the target temperature in situ are scarce
process. To solve this problem, Yudasaka et al. [3.32] and, generally speaking, this is one of the numerous
utilized two pellets facing each other, one made entirely variables of the laser ablation synthesis technique. The
parameters that have been studied the most are the na-
ture of the target, the nature and concentration of the
Pump catalyst, the nature of the neutral gas flow, and the tem-
Filter perature of the outer oven.
1. The collaret, which forms around the cathode 2. A graphite anode within which the catalysts are ho-
2. The weblike deposits found above the cathode mogeneously dispersed [3.48].
3. The soot deposited all around the reactor walls and
the bottom. The former are by far the most popular, due to their ease
of fabrication.
On the other hand, MWNTs are formed in a hard de- Optimizing the process in terms of the nanotube
posit adherent to the cathode whether catalysts are used yield and quality is achieved by studying the roles of
or not. The cathode deposits form under the cathode.
Part A 3.2
various parameters such as the type of doped anode
The formation of collaret and web is not systematic and (homogeneous or heterogeneous catalyst dispersion),
depends on the experimental conditions, as indicated in the nature as well as the concentration of the catalyst,
Table 3.1, as opposed to the cathode deposit and soot, the nature of the plasmagen gas, the buffer gas pres-
which are obtained consistently. sure, the arc current intensity, and the distance between
Two graphite rods of few millimeters in diameter electrodes. Investigating the influences of these param-
constitute the electrodes between which a potential dif- eters on the type and amount of carbon nanostructures
ference is applied. The dimensions of these electrodes formed is, of course, the preliminary work that has
vary according to the authors. In certain cases, the cath- been done. Although electric arc reactors equipped with
ode has a greater diameter than the anode in order to the facilities to perform such investigations are scarce
facilitate their alignment [3.37, 47]. Other authors uti- (Fig. 3.12), investigating the missing link (the effect of
lize electrodes of the same diameter [3.46]. The whole varying the parameters on the plasma characteristics –
device can be designed horizontally [3.38, 46] or verti- the species concentrations and temperature) is likely
cally [3.39, 41–43]. The advantage of the latter is the to provide a more comprehensive understanding of the
symmetry brought by the verticality with respect to phenomena involved during nanotube formation. This
gravity, which facilitates computer modeling (regarding has been recently performed using atomic and molecu-
convection flows, for instance). lar optical emission spectroscopy [3.39, 41–44, 46].
Two types of anode can be utilized when catalysts Finally, we should mention attempts to create an
are introduced: electric arc in liquid media, such as liquid nitro-
1. A graphite anode containing a coaxial hole several gen [3.49] or water [3.50, 51]. The goal here is to
centimeters in length into which a mixture of the make processing easier, since such systems should not
catalyst and the graphite powder is placed. require pumping devices or a closed volume and so
they are more likely to allow continuous synthesis. This
adaptation has not, however, reached the stage of mass
production.
Cathode
holder
Electric Arc Method – Results
In view of the numerous results obtained with this elec-
Window Window
tric arc technique, it is clear that both the morphology
and the production efficiency of nanotubes strongly
depends upon the experimental conditions used and,
in particular, upon the nature of the catalysts. It is
Cathode worth noting that the products obtained do not con-
Anode sist solely of carbon nanotubes. Nontubular forms of
carbon, such as nanoparticles, fullerenelike structures
including C60 , poorly organized polyaromatic carbons,
nearly amorphous nanofibers, multiwall shells, single-
wall nanocapsules, and amorphous carbon have all been
obtained, as reported in Table 3.1 [3.40, 42, 43]. In ad-
dition, remnants of the catalyst are found all over the
place – in the soot, the collaret, the web and the cathode
Gas inlet Motor Vacuum deposit – in various concentrations. Generally, at a he-
lium pressure of about 600 mbar, for an arc current of
Fig. 3.12 Sketch of an electric arc reactor 80 A and for an electrode gap of 1 mm, the synthesis of
58 Part A Nanostructures, Micro-/Nanofabrication and Materials
Table 3.1 Different carbon morphologies obtained by changing the type of anode, the type of catalyst and the pressure in a series
of arc discharge experiments (electrode gap = 1 mm)
Catalyst 0.6Ni + 0.6Co 0.6Ni + 0.6Co 0.5Ni + 0.5Co 4.2Ni + 1Y
(at. %) (homogeneous anode) (homogeneous anode)
Arc P ≈ 60 kPa P ≈ 40 kPa P ≈ 60 kPa P ≈ 60 kPa
conditions I ≈ 80 A I ≈ 80 A I ≈ 80 A I ≈ 80 A
Soot • MWNT + MWS + • POPAC and AC • AC and POPAC par- • POPAC and AC
POPAC or Cn ± cat- particles + catalysts ticles + catalysts + particles + catalysts
Part A 3.2
SWNTs is favored by the use of Ni/Y as coupled cata- the latter conditions (Table 3.1, column 4), Table 3.1 il-
lysts [3.8, 38, 52]. In these conditions, which give high lustrates the consequence of changing the parameters.
SWNT yields, SWNT concentrations are highest in the For instance (Table 3.1, column 3), using Ni/Co in-
collaret ( 50–70%), then in the web (≈ 50% or less) and stead of Ni/Y as catalysts prevents the formation of
then in the soot. On the other hand, c-MWNTs are found SWNTs. But when the Ni/Co catalysts are homoge-
in the cathode deposit. SWNT lengths are micrometric neously dispersed in the anode (Table 3.1, column 1),
and, typical outer diameters are around 1.4 nm. Using the formation of nanotubes is promoted again, but
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 59
MWNTs with two or three walls prevail over SWNTs, temperature. This makes a perfect sense, since carbon
among which DWNTs (double-wall nanotubes) domi- species are very emissive in the range 4500–6000 K, in-
nate. However, decreasing the ambient pressure from ducing that radiative losses are more significant when
60 to 40 kPa (Table 3.1, column 2) again suppresses plasmas are enriched in carbon species, leading to
nanotube formation. colder plasma temperatures, and vice versa. Such a fea-
Based on works dealing with the influence of the ture is again consistent with the steady erosion of
granulometry of the graphite powders which are mixed high thermal conductivity anodes. It is also worth
Part A 3.2
with the catalyst powder and placed in hollow-type noting that, again, area where CI/NiI ratios exhibit
graphite anodes, recent studies have demonstrated that
one of the control keys for growing SWNTs with en-
hanced purity and yield is for the anode to exhibit a) Temperature (K)
9000
a high thermal conductivity with as more limited ra- 1 mm-Ni/Y/graphite (100 µm)
dial and longitudinal variations as possible [3.53, 54]. 1 mm-Ni/Y/diamond (1 µm)
8000 1 mm-Ni/Y/graphite (1 µm)
This explains why similar results (i. e., enhanced pu-
rity and yield) were previously obtained when replacing
7000
the graphite powder by diamond powder [3.44, 45] in
spite of the low electrical conductivity of diamond,
6000
since graphite and diamond powders lead to the same
plasma composition once vaporized at high tempera-
5000
tures (> 4000 K).
A comparison of the plasma characteristics (i. e.,
4000
radial temperature profiles and CI/NiI concentration
ratio) obtained for anodes with different filler mater-
3000
ial features (i. e., 1/ 100 μm granulometry and sp2 /sp3 0 0.5 1 1.5 2 2.5
carbon) is presented in Fig. 3.13a,b respectively. The Radial coordinate (mm)
whole plasma temperature radial profiles obtained us- b) [CI]/[NiI] concentration ratio
ing either the Ni/Y/graphite (φ ≈ 1 μm) anode or the 1010
Ni/Y/diamond (φ ≈ 1 μm) anode is much smoother
109
than with the standard Ni/Y/graphite (φ ≈ 100 μm) an-
ode, meanwhile exhibiting less extreme temperatures 108
(≈ 6200 K for the highest as opposed to ≈ 8000 K re-
107
spectively for the standard anode). From 1 mm from
the arc axis, temperature is maintained at a constant 106
value at about 4000 K. The absence of large temper-
ature fluctuations is consistent with the fact that the 105
plasma is continuously fed by a rather constant ratio of 104
[carbon]/[catalysts] resulting from the steadier erosion
of the anode and a better powder mixture homoge- 103
nization. In this regard, it might be significant that the 102
smoothest temperature profile over the longest radial 0 0.5 1 1.5 2 2.5
Radial coordinate (mm)
distance is obtained for the Ni/Y/graphite (ϕ ≈ 1 μm)
anode, which has resulted in the highest yield [3.53,54]. Fig. 3.13a,b Radial temperature profiles (a) and radial
Likewise, the CI/NiI concentration ratio profiles re- [CI]/[NiI] concentration ratio (b) as obtained by emis-
lated to either the Ni/Y/graphite (φ ≈ 1 μm) anode or sion spectroscopy for hollowed-type anodes with various
the Ni/Y/diamond anode show a dramatic difference thermal behaviours. The thermal behaviour was varied by
with respect to the Ni/Y/graphite (φ ≈ 100 μm) an- varying the grain size (1 or 100 μm) and the carbon type
ode (Fig. 3.13b). They exhibit a fluctuation-free regime (sp2 – graphite, or sp3 – diamond) of the carbon powder
along the whole radial profile, with a unique maxi- which the hollow core of the anode is filled with (along
mum at ≈ 1.3–1.5 mm from the arc axis. The average with yttrium and nickel catalyst powder). Smaller grain
ratio is low (≈ 5 × 105 ) due to a relatively low distribu- size results in better compaction, hence in higher thermal
tion of carbon concentration leading to a higher plasma conductivity
60 Part A Nanostructures, Micro-/Nanofabrication and Materials
maximum values in Fig. 3.13b relate to area of min- blocks for MWNTs (formed at the cathode) but not for
imum temperature values in Fig. 3.13a. In addition, SWNTs [3.43, 45].
the CI/NiI concentration ratio is up to about 3–5 Although many aspects of it still need to be un-
orders of magnitude higher for the fine-grain graphite- derstood, the electric arc method is one of the three
containing anode and the diamond-containing anode methods currently used to produce SWNTs as com-
than for the large-grain graphite-containing anode. mercial products. Though not selling bare nanotubes
Moreover, the CI/NiI concentration ratio is even higher anymore, Nanoledge S.A. (Montpellier, France), for in-
Part A 3.2
as ≈ 1.5 orders of magnitude for the fine-grain graphite- stance, had a current production that reached several
containing anode than for the diamond-containing tens of kilograms per year (raw SWNTs, in other words
anode. unpurified), with a market price of ≈ 65 €/g in 2005,
Highly and homogeneously thermally conductive which was much cheaper than any other production
anodes lead to a steadier anode erosion, hence to stead- method. However, the drop of prices for raw SWNTs
ier plasma characteristics, hence to a more constant
variety of the carbon phase formed (SWNTs), finally
a) Temperature (K)
resulting in an enhanced purity and yield of the latter.
7000
Such experiments have revealed, as in the comparison
Anode
between the results from using homogeneous instead
Center
of heterogeneous anodes, that the physical phenomena
6500 Cathode
(charge and heat transfers) that occur in the anode dur-
ing the arc are of the utmost importance, a factor which
was neglected before this. 6000
It is clear that while the use of a rare earth element
(such as Y) as a single catalyst does not provide the right
conditions to grow SWNTs, associating it with a transi- 5500
tion metal (Ni/Y for instance) seems to lead to the best
combinations that give the highest SWNT yields [3.47].
On the other hand, using a single rare earth element 5000
may lead to unexpected results, such as the closure of 0 0.5 1 1.5 2 2.5 3 3.5 4
graphene edges from a c-MWNT wall with the neigh- Distance from center (mm)
boring graphene edges from the same wall side, leading b) Density of C2 (× 1015 cm–2)
to the preferred formation of telescopelike and open 16
c-MWNTs that are able to contain nested Gd crys- 14 Anode
tals [3.41, 43]. The effectiveness of bimetallic catalysts Center
is believed to be due to the transitory formation of nickel 12 Cathode
particles coated with yttrium carbide, which has a lattice 10
constant that is somewhat commensurable with that of
graphene [3.55]. 8
Figure 3.14 illustrates other interesting features of
6
the plasma. A common feature is that a huge vertical
gradient (≈ 500 K/mm) rapidly establishes (≈ 0.5 mm 4
from the center in the radial direction) from the bottom
2
to the top of the plasma, probably due to convection
phenomena (Fig. 3.14a). The zone of actual SWNT for- 0
0 0.5 1 1.5 2 2.5 3 3.5 4
mation is beyond the limit of the volume analyzable Distance from center (mm)
in the radial direction, corresponding to colder areas.
The C2 concentration increases dramatically from the Fig. 3.14a,b Typical temperature (a) and C2 concentration
anode to the cathode and decreases dramatically in the (b) profiles for plasma at the anode surface (squares), at
radial direction (Fig. 3.14b). This demonstrates that C2 the center of the plasma (dots), and at the cathode sur-
moieties are secondary products resulting from the re- face (triangles) at standard conditions (see text). Gradients
combination of primary species formed from the anode. are similar whichever catalyst is used, although absolute
It also suggests that C2 moieties may be the building values may vary
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 61
down to 2–5 €/g which was anticipated for 2007 has not been specially developed for the treatment of liquid,
been possible. Actually, Bucky USA (Houston, Texas, gaseous or dispersed materials. An electric arc is es-
USA) are still supplying raw SWNTs derived from elec- tablished between three graphite electrodes. The system
tric arcs at a market price of 250 $/g in 2006 (which is powered by a three-phase AC power supply op-
is, however, a 75% decrease in two years), which is erated at 600 Hz and at arc currents of 250–400 A.
barely lower than the ≈ 350 $/g proposed for 70–90%- Carbon precursors, gaseous, liquid or solid, are injected
purified SWNTs from Nanocarblab (Moscow, Russia). at the desired (variable) position into the plasma zone.
Part A 3.2
The reactive mixture can be extracted from the reac-
Three-Phase AC Arc Plasma tion chamber at different predetermined positions. After
An original semi-industrial three-phase AC plasma cooling down to room temperature, the aerosol passes
technology has been developed for the processing of through a filtering system. The main operating parame-
carbon nanomaterials [3.57, 58]. The technology has ters, which are freely adjustable, include the arc current,
a) Mirror c)
Graphite tube Gas mixture
F
Graphite
cylinder
Shields
Sun
beams Tablet
Mirror
Water
Water Water
Thermocouple
Gas
Solar
flux
Target
Filter
Heat exchanger
Water
Vacuum pump
L
20 cm
Fig. 3.15a–c Sketch of a solar energy reactor in use in the PROMES-CNRS Laboratory, Odeilho (France). (a) Gathering
of sun rays, focused at F; (b) side view of the experimental set-up at the focus of the 1 MW solar furnace; (c) top view of
the target graphite rod (adapted from [3.56])
62 Part A Nanostructures, Micro-/Nanofabrication and Materials
the flow rate and the nature of the plasma gas (N2 , Ar, associated with the other solid carbon source-based
H2 , He, and so on), the carbon precursor (gaseous, li- processes. When possible, specific in situ diagnostics
quid, solid, up to 3 kg/h), the injection and extraction (pyrometry, optical emission spectroscopy, and so on)
positions, and the quenching rate. This plasma tech- are also performed in order to investigate the roles of
nology has shown very high versatility and it has been various parameters (temperature measurements at the
demonstrated that it can be used to produce a wide range crucible surface, along the graphite tube acting as ther-
of carbon nanostructures ranging from carbon blacks to mal screen, C2 radical concentration in the immediate
Part A 3.2
carbon nanotubes over fullerenes with a high product vicinity of the crucible).
selectivity.
Solar Furnace – Results
Solar Furnace Some of the results obtained by different groups con-
Solar furnace devices were originally utilized by sev- cerning the influence of the catalyst can be summarized
eral groups to produce fullerenes [3.59–61]. Heben et al. as follows. With Ni/Co, and at low pressure, the sample
[3.62] and Laplaze et al. [3.63] later modified their collected contains mainly MWNTs with bamboo tex-
original devices to achieve carbon nanotube produc- ture, carbon shells, and some bundles of SWNTs [3.64].
tion. This modification consisted mainly of using more At higher pressures, only bundles of SWNTs are
powerful ovens [3.64, 65]. obtained, with fewer carbon shells. Relatively long bun-
dles of SWNTs are observed with Ni/Y and at a high
Solar Furnace – Experimental Devices pressure. Bundles of SWNTs are obtained in the soot
The principle of this technique is again based on the with Co; the diameters of the SWNTs range from 1 to
sublimation of a mixture of graphite powder and cata- 2 nm. Laplaze et al. [3.64] observed very few nanotubes
lysts placed in a crucible in an inert gas. An example but a large quantity of carbon shells.
of such a device is shown in Fig. 3.15. The solar In order to proceed to large-scale synthesis of
rays are collected by a plain mirror and reflected to- single-wall carbon nanotubes, which is still a challenge
ward a parabolic mirror that focuses them directly onto for chemical engineers, Flamant et al. [3.56] and Lux-
a graphite pellet in a controlled atmosphere (Fig. 3.15a). embourg et al. [3.66] recently demonstrated that solar
The high temperature of about 4000 K causes both the energy-based synthesis is a versatile method for ob-
carbon and the catalysts to vaporize. The vapors are taining SWNTs that can be scaled up from 0.1–0.2 to
then dragged by the neutral gas and condense onto the 10 g/h and then to 100 g/h productivity using existing
cold walls of the thermal screen. The reactor consists solar furnaces. Experiments performed on a medium
of a brass support cooled by water circulation, upon scale produced about 10 g/h of SWNT-rich material us-
which Pyrex chambers of various shapes can be fixed ing various mixtures of catalysts (Ni/Co, Ni/Y, Ni/Ce).
(Fig. 3.15b). This support contains a watertight pas- A numerical reactor simulation was performed in or-
sage permitting the introduction of the neutral gas and der to improve the quality of the product, which was
a copper rod onto which the target is mounted. The tar- subsequently observed to reach 40% SWNT in the
get is a graphite rod that includes pellets containing soot [3.67].
the catalysts, which is surrounded by a graphite tube
(Fig. 3.15c) that acts as both a thermal screen to reduce 3.2.2 Gaseous Carbon Source-Based
radiation losses (very important in the case of graphite) Production Techniques
and a duct to lead carbon vapors to a filter, which stops for Carbon Nanotubes
soot from being deposited on the Pyrex chamber wall.
The graphite rod target replaces the graphite crucible As mentioned in the Introduction, the catalysis-en-
filled with powdered graphite (for fullerene synthesis) hanced thermal cracking of a gaseous carbon source
or the mixture of graphite and catalysts (for nanotube (hydrocarbons, CO) – commonly referred to as catalytic
synthesis) that were used in the techniques we have chemical vapor deposition (CCVD) – has long been
discussed previously. known to produce carbon nanofilaments [3.4], so re-
These studies primarily investigated the target com- porting on all of the works published in the field since
position, the type and concentration of catalyst, the the beginning of the century is almost impossible. Until
flow-rate, the composition and pressure of the plas- the 1990s, however, carbon nanofilaments were mainly
magenic gas inside the chamber, and the oven power. produced to act as a core substrate for the subsequent
The objectives were similar to those of the works growth of larger (micrometric) carbon fibers – so-called
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 63
vapor-grown carbon fibers – via thickening in catalyst- and nanotubes. For multilayered fibrous morphologies
free CVD processes [3.68, 69]. We are therefore going (since single-layered fibrous morphologies can only be
to focus instead on more recent attempts to prepare gen- SWNT anyway), the exact name should be vapor-grown
uine carbon nanotubes. carbon nanofilaments (VGCNF). Whether or not the fil-
The synthesis of carbon nanotubes (either single- or aments are tubular is a matter of textural description,
multiwalled) by CCVD methods involves the catalytic which should go with other textural features such as
decomposition of a carbon-containing source on small bamboo, herringbone and concentric (Sect. 3.1.2). In
Part A 3.2
metallic particles or clusters. This technique involves the following, we will therefore use MWNTs for any
either an heterogeneous process if a solid substrate is in- hollowed nanofilament, whether they contain graphene
volved or an homogeneous process if everything takes walls oriented transversally or not. Any other nanofila-
place in the gas phase. The metals generally used for ment will be termed a nanofiber.
these reactions are transition metals, such as Fe, Co
and Ni. It is a rather low-temperature process com- Heterogeneous Processes
pared to arc discharge and laser ablation methods, with Heterogeneous CCVD processes simply involve pass-
the formation of carbon nanotubes typically occurring ing a gaseous flow containing a given proportion of
between 600 and 1000 ◦ C. Because of the low tempera- a hydrocarbon (mainly CH4 , C2 H2 , C2 H4 , or C6 H6 ,
ture, the selectivity of the CCVD method is generally usually as a mixture with either H2 or an inert gas
better for the production of MWNTs with respect to such as Ar) over small transition metal particles (Fe,
graphitic particles and amorphouslike carbon, which re- Co, Ni) in a furnace. The particles are deposited onto
main an important part of the raw arc discharge SWNT an inert substrate, by spraying a suspension of the
samples, for example. Both homogeneous and hetero- metal particles on it or by another method. The reac-
geneous processes appear very sensitive to the nature tion is chemically defined as catalysis-enhanced thermal
and the structure of the catalyst used, as well as to cracking
the operating conditions [3.70]. Carbon nanotubes pre- Cx H y → x C + 2y H2 .
pared by CCVD methods are generally much longer
(a few tens to hundreds of micrometers) than those ob- Catalysis-enhanced thermal cracking was used as
tained by arc discharge (a few micrometers). Depending long ago as the late nineteenth century. Extensive works
on the experimental conditions, it is possible to grow on this topic published before the 1990s include those
dense arrays of nanotubes. It is a general statement that by Baker et al. [3.6, 72], or Endo et al. [3.73, 74]. Sev-
MWNTs from CCVD contain more structural defects eral review papers have been published since then, such
(exhibit a lower nanotexture) than MWNTs from arc as [3.75], in addition to many regular papers.
discharge, due to the lower temperature of the reaction, CO can be used instead of hydrocarbons; the reac-
which does not allow any structural rearrangements. tion is then chemically defined as catalysis-enhanced
These defects can be removed by subsequently apply- disproportionation (the so-called the Boudouard equi-
ing heat treatments in vacuum or inert atmosphere to librium)
the products. Whether such a discrepancy is also true 2 CO C + CO2 .
for SWNTs remains questionable. CCVD SWNTs are
generally gathered into bundles that are generally of Heterogeneous Processes –
smaller diameter (a few tens of nm) than their arc dis- Experimental Devices
charge and laser ablation counterparts (around 100 nm The ability of catalysis-enhanced CO disproportiona-
in diameter). Specifically when performed in fluidized- tion to make carbon nanofilaments was reported by
bed reactor [3.71], CCVD provides reasonably good Davis et al. [3.76] as early as 1953, probably for the
perspectives on large-scale and low-cost processes for first time. Extensive follow-up work was performed by
the mass production of carbon nanotubes, a key point Boehm [3.77], Audier et al. [3.17, 78–80], and Gadelle
for their application at the industrial scale. et al. [3.81–84].
A final word concerns the nomenclature. Because Although formation mechanisms for SWNTs and
work in the field started more than a century ago, the MWNTs can be quite different (Sect. 3.3, or refer to
names of the carbon objects prepared by this method a review article such as [3.85]), many of the catalytic
have changed with time with the authors, research ar- process parameters play similar and important roles in
eas, and fashions. These same objects have been called the type of nanotubes formed: the temperature, the dura-
vapor-grown carbon fibers, nanofilaments, nanofibers tion of the treatment, the gas composition and flow rate,
64 Part A Nanostructures, Micro-/Nanofabrication and Materials
and of course the catalyst nature and size. At a given 2. The preparation of a solid solution of an oxide of
temperature, depending mainly on the nature of both the the chosen catalytic metal in a chemically inert and
catalyst and the carbon-containing gas, the catalytic de- thermally stable host oxide.
composition will take place at the surfaces of the metal
particles, followed by mass transport of the freshly pro- The catalyst is then reduced to form the metal particles
duced carbon by surface or volume diffusion until the on which the catalytic decomposition of the carbon
carbon concentration reaches the solubility limit, and source will lead to carbon nanotube growth. In most
Part A 3.2
the precipitation starts. cases, the nanotubes can then be separated from the
It is now agreed that CCVD carbon nanotubes form catalyst (Fig. 3.16).
on very small metal particles, typically in the nanometer
range [3.85]. These catalytic metal particles are pre- Heterogeneous Processes – Results with CCVD
pared mainly by reducing transition metal compounds Involving Impregnated Catalysts
(salts, oxides) by H2 prior to the nanotube formation A lot of work had been done in this area even before
step (where the carbon containing gas is required). It the discovery of fullerenes and carbon nanotubes, but
is possible, however, to produce these catalytic metal although the formation of tubular carbon structures by
particles in situ in the presence of the carbon source, catalytic processes involving small metal particles was
allowing for a one-step process [3.88]. Because con- clearly identified, the authors did not focus on the prepa-
trolling the metal particle size is the key issue (they ration of SWNTs orMWNTs with respect to the other
have to be nanosized), coalescence is generally avoided carbon species. Some examples will be given here to
by placing them on an inert support such as an ox- illustrate the most striking improvements obtained.
ide (Al2 O3 , SiO2 , zeolites, MgAl2 O4 , MgO) or more With the impregnation method, the process gener-
rarely on graphite. A low concentration of the catalytic ally involves four different and successive steps:
metal precursor is required to limit the coalescence of 1. Impregnation of the support by a solution of a salt
the metal particles, which can happen during the re- (nitrate, chloride) of the chosen metal catalyst
duction step. The supported catalysts can be used as 2. Drying and calcination of the supported catalyst to
a static phase placed within the gas flow, but can also get the oxide of the catalytic metal
be used as a fine powder suspended into and by the gas 3. Reduction in a H2 -containing atmosphere to make
phase, in a so-called fluidised bed process. In the lat- the catalytic metal particles
ter, the reactor has to be vertical so that to compensate 4. The decomposition of a carbon-containing gas over
the effect of gravity by the suspending effect of the gas the freshly prepared metal particles that leads to
flow. nanotube growth.
There are two main ways to prepare the catalyst:
For example, Ivanov et al. [3.89] prepared nano-
1. The impregnation of a substrate with a solution of tubes through the decomposition of C2 H2 (pure or
a salt of the desired transition metal catalyst mixed with H2 ) on well-dispersed transition metal par-
ticles (Fe, Co, Ni, Cu) supported on graphite or SiO2 . CH4 . The nanotubes obtained clearly depend upon the
Co-SiO2 was found to be the best catalyst/support com- nature of both the transition metal (or alloy) used and
bination for the preparation of MWNTs, but most of the the inert oxide (matrix); the latter because the Lewis
other combinations led to carbon filaments, sometimes acidity seems to play an important role [3.97]. For ex-
covered with amorphouslike carbon. The same authors ample, in the case of solid solutions containing around
have developed a precipitation-ion-exchange method 10 wt % of Fe, the amount of carbon nanotubes ob-
that provides a better dispersion of metals on silica com- tained decreases in the following order depending on the
Part A 3.2
pared to the classical impregnation technique. The same matrix oxide: MgO > Al2 O3 > MgAl2 O4 [3.86]. In the
group then proposed the use of a zeolite-supported Co case of MgO-based solid solutions, the nanotubes can
catalyst [3.90, 91], resulting in very finely dispersed be very easily separated from the catalyst by dissolving
metal particles (from 1 to 50 nm in diameter). They ob- it (in diluted HCl for example) [3.96]. The nanotubes
served MWNTs with a diameter around 4 nm and only obtained are typically gathered into small-diameter bun-
two or three walls only on this catalyst. Dai et al. [3.92] dles (less than 15 nm) with lengths of up to 100 μm.
have prepared SWNTs by CO disproportionation on The nanotubes are mainly SWNTs and DWNTs, with
nanosized Mo particles. The diameters of the nanotubes diameters of between 1 and 3 nm.
obtained are closely related to those of the original par- Obtaining pure nanotubes by the CCVD method re-
ticles and range from 1 to 5 nm. The nanotubes obtained quires, as for all the other techniques, the removal of
by this method are free of an amorphous carbon coat- the catalyst. When a catalyst supported (impregnated) in
ing. They also found that a synergetic effect occurs for a solid solution is used, the supporting – and catalytical-
the alloy instead of the components alone, and one of ly inactive – oxide is the main impurity, both in weight
the most striking examples is the addition of Mo to and volume. When oxides such as Al2 O3 or SiO2 (or
Fe [3.93] or Co [3.94]. even combinations) are used, aggressive treatments in-
volving hot caustic solutions (KOH, NaOH) for Al2 O3
Heterogeneous Processes – Results with CCVD or the use of HF for SiO2 are required. These treatments
Involving Solid Solution-Based Catalysts have no effect, however, on other impurities such as
A solid solution of two metal oxides is formed when other forms of carbon (amorphouslike carbon, graphi-
ions of one metalmix with ions of the other metal. tized carbon particles and shells, and so on). Oxidizing
For example, Fe2 O3 can be prepared in solid solu- treatments (air oxidation, use of strong oxidants such
tion in Al2 O3 to give a Al2−2x Fe2x O3 solid solution. as HNO3 , KMnO4 , H2 O2 ) are thus required and permit
The use of a solid solution allows a perfectly ho- the removal of most unwanted forms of carbon, but they
mogeneous dispersion of one oxide in the other to result in a low final yield of carbon nanotubes, which
be obtained. These solid solutions can be prepared are often quite damaged. Flahaut et al. [3.96] were the
in different ways, but coprecipitation of mixed ox- first to use a MgCoO solid solution to prepare SWNTs
alates and combustion synthesis are the most common and DWNTs that could be easily separated without in-
methods used to prepare nanotubes. The synthesis curring any damage via fast and safe washing with an
of nanotubes by the catalytic decomposition of CH4 aqueous HCl solution.
over an Al2−2x Fe2x O3 solid solution was originated by In most cases, only very small quantities of catalyst
Peigney et al. [3.88] and then studied extensively by the (typically less than 500 mg) are used, and most claims
same group using different oxides such as spinel-based of high-yield productions of nanotubes are based on
solid solutions (Mg1−x Mx Al2 O4 with M = Fe, Co, Ni, laboratory experimental data, without taking into ac-
or a binary alloy [3.86, 95]) or magnesia-based solid count all of the technical problems related to scaling
solutions [3.86, 96] (Mg1−x Mx O, with M = Fe, Co or up to a laboratory-scale CCVD reactor. At the present
Ni). Because of the very homogeneous dispersion of the time, although the production of MWNTs is possi-
catalytic oxide, it is possible to produce very small cat- ble on an industrial scale, the production of affordable
alytic metal particles at the high temperature required SWNTs is still a challenge, and controlling the arrange-
for the decomposition of CH4 (which was chosen for ment of and the number of walls in the nanotubes is
its greater thermal stability compared to other hydrocar- also problematic. For example, adding small amounts
bons). The method proposed by these authors involves of molybdenum to the catalyst [3.98] can lead to dras-
the heating of the solid solution from room temperature tic modifications of the nanotube type (from regular
to a temperature of between 850 and 1050 ◦ C in a mix- nanotubes to carbon nanofibers – Sect. 3.1). Flahaut
ture of H2 and CH4 , typically containing 18 mol % of et al.have recently shown that the method used to pre-
66 Part A Nanostructures, Micro-/Nanofabrication and Materials
pare a particular catalyst can play a very important Homogeneous Processes – Results
role [3.99]. Double-walled carbon nanotubes (DWNTs) The metal-organic compound decomposes in the first
represent a special case: they are at the frontier be- zone of the reactor, generating nanosized metallic par-
tween single- (SWNTs) and multiwalled nanotubes ticles that can catalyze nanotube formation. In the
(MWNTs). Because they are the MWNTs with the second part of the reactor, the carbon source is decom-
lowest possible number of walls, their structures and posed to atomic carbon, which is then responsible for
properties are very similar to those of SWNTs. Any the formation of nanotubes.
Part A 3.2
subsequent functionalization, which is often required to This technique is quite flexible and SWNTs [3.105],
improve the compatibility of nanotubes with their ex- DWNTs [3.106] and MWNTs [3.107] have been
ternal environment (composites) or to give them new obtained, in proportions depending on the carbon feed-
properties (solubility, sensors), will partially damage stock gas. The technique has also been exploited for
the external wall, resulting in drastic modifications some time in the production of vapor-grown carbon
in terms of both electrical and mechanical properties. nanofibers [3.108].
This is a serious drawback for SWNTs. In the case of The main drawback of this type of process is
DWNTs, the outer wall can be modified (functional- again that it is difficult to control the size of the
ized) while retaining the structure of the inner tube. metal nanoparticles, and thus nanotube formation is
DWNTs have been recently synthesised on a gram-scale often accompanied by the production of undesired car-
by CCVD [3.87], with a high purity and a high selectiv- bon forms (amorphous carbon or polyaromatic carbon
ity (around 80% DWNTs) (Fig. 3.16b). phases found as various phases or as coatings). In par-
ticular, encapsulated forms have been often found as the
Homogeneous Processes result of the formation of metal particles that are too
The homogenous route, also called the floating catalyst large to promote nanotube growth (and so they can end
method, differs from the other CCVD-based methods up being totally covered with graphene layers instead).
because it uses only gaseous species and does not re- The same kind of parameters have to be controlled
quire the presence of any solid phase in the reactor. as for heterogeneous processes in order to finely tune
The basic principle of this technique, similar to the this process and selectively obtain the desired morphol-
other CCVD processes, is to decompose a carbon source ogy and structure of the nanotubes formed, such as: the
(ethylene, xylene, benzene, carbon monoxide, and so choice of the carbon source; the reaction temperature;
on) on nanosized transition metal (generally Fe, Co, the residence time; the composition of the incoming
or Ni) particles in order to obtain carbon nanotubes. gaseous feedstock, with particular attention paid to the
The catalytic particles are formed directly in the reac- role played by the proportion of hydrogen, which can
tor, however, and are not introduced before the reaction, influence the orientation of the graphene with respect
as occurs in supported CCVD for instance. to the nanotube axis, thus switching from c-MWNT to
h-MWNT [3.82]; and the ratio of the metallorganic pre-
Homogeneous Processes – cursor to the carbon source [3.101]. In an independent
Experimental Devices study [3.109], it was shown that the general tendency
The typical reactor used in this technique is a quartz is:
tube placed in an oven into which the gaseous feedstock,
containing the metal precursor, the carbon source, some 1. To synthesize SWNTs when the ferrocene/benzene
hydrogen and a vector gas (N2 , Ar, or He), is passed. molar ratio is high, typically ≈ 15%
The first zone of the reactor is kept at a lower tempera- 2. To produce MWNTs when the ferrocene/benzene
ture, and the second zone, where the formation of tubes molar ratio is between ≈ 4 and ≈ 9%
occurs, is heated to 700–1200 ◦ C. The metal precursor 3. To synthesize carbon nanofibers when the fer-
is generally a metal-organic compound, such as a zero- rocene/benzene molar ratio is below ≈ 4%.
valent carbonyl compound like [Fe(CO)5 ] [3.100], or
a metallocene [3.101–103] such as ferrocene, nicke- As recently demonstrated, the overall process can be
locene or cobaltocene. The use of metal salts, such as improved by adding other compounds such as ammonia
cobalt nitrate, has also been reported [3.104]. It may be or sulfur-containing species to the reactive gas phase.
advantageous to make the reactor vertical, so that grav- The former allows aligned nanotubes and mixed C-N
ity acts symmetrically on the gaseous volume inside the nanotubes [3.110] to be obtained, while the latter results
furnace. in a significant increase in productivity [3.108, 111].
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 67
An interesting result is the increase in yield and purity a similar way by CCVD-related processes, MWNTs re-
brought about by a small input of oxygen, as achieved main far less expensive than SWNTs, reaching prices
by using alcohol vapors instead of hydrocarbons as as low as 0.055 $/g (current ASI fares for Pyrograf-III
feedstock [3.112]. It is assumed that the oxygen prefer- grade).
ably burns the poorly organized carbon out into CO2 ,
thereby enhancing the purity, and prevents the catalyst Templating
particles from being encapsulated in the carbon shells Another interesting technique, although one that is def-
Part A 3.2
too early, making them inactive, thereby enhancing the initely not suitable for mass production (and so we
nanotube yield. Moreover, it was found to promote the only touch on it briefly here), is the templating tech-
formation of SWNTs over MWNTs, since suppressing nique. It is the only other method aside from the electric
carbon shell formation suppresses MWNT formation arc technique that is able to synthesize carbon nano-
too. tubes without any catalyst. Any other work reporting the
It should be emphasized that only small amounts catalyst-free formation of nanotubes is actually likely
have been produced so far, and scale-up to industrial to have involved the presence of catalytic metallic im-
levels seems quite difficult due to the large number purities in the reactor or some other factors that caused
of parameters that must be considered. A critical one a chemical gradient in the system. Another useful aspect
is to be able to increase the quantity of metallorganic of this approach is that it allows aligned nanotubes to be
compound that is used in the reactor, in order to in- obtained naturally, without the help of any subsequent
crease production, without obtaining particles that are alignment procedure. However, the template must be re-
too big. This problem has not yet been solved. An moved (dissolved) to recover the nanotubes, in which
additional problem inherent in the process is the pos- case the alignment of the nanotubes is lost.
sibility of clogging the reactor due to the deposition of
metallic nanoparticles on the reactor walls followed by Templating – Experimental Devices
carbon deposition. An interesting alternative could be The principle of this technique is to deposit the solid
the injection, into the vertical floating reactor, of a sup- carbon coating obtained from the CVD method onto the
ported catalyst powder instead of an organometallic walls of a porous substrate whose pores are arranged in
compound. This approach has allowed the continuous parallel channels. The feedstock is again a hydrocarbon,
production of single-walled carbon nanotubes with scal- such as a common source of carbon. The substrate can
ing capability up to 220 g/h [3.114]. be alumina or zeolite for instance, which present natu-
A significant breakthrough concerning this tech- ral channel pores, while the whole system is heated to
nique could be the HiPCo process developed at Rice a temperature that cracks the hydrocarbon selected as
University, which produces SWNTs of very high pu- the carbon source (Fig. 3.17).
rity [3.115, 116]. This gas phase catalytic reaction
uses carbon monoxide to produce, from [Fe(CO)5 ],
Anodic aluminum oxide film Carbon tubes
a SWNT material that is claimed to be relatively free
of by-products. The temperature and pressure condi-
tions required are applicable to industrial plants. Upon Carbon deposition
on the pore wall HF washing
heating, the [Fe(CO)5 ] decomposes into atoms which
condense into larger clusters, and SWNT nucleate and (Propylene, 800 °C)
grow on these particles in the gas phase via CO
disproportionation (the Boudouard reaction, see Hetero-
50 –100 µm
geneous Processes in Sect. 3.1.2):
The company Carbon Nanotechnologies Inc. (Hous-
ton, USA) currently sells raw SWNT materials prepared
in this way, at a market price of 375 $/g, or 500 $/g 10 –500 µm
if purified (2005 data). Other companies that spe-
cialize in MWNTs include Applied Sciences Inc.
(Cedarville, USA), currently has a production facility of
≈ 40 tons/year of ≈ 100 nm large MWNTs (Pyrograf- Fig. 3.17 Principle of the templating technique used in the catalyst-
III), and Hyperion Catalysis (Cambridge, USA), which free formation of single-walled or concentric-type multiwalled
makes MWNT-based materials. Though prepared in carbon nanotubes (from [3.113])
68 Part A Nanostructures, Micro-/Nanofabrication and Materials
The smallest SWNTs (diameters ≈ 0.4 nm) ever ob- about the possible occurrence of by-products, but they
tained (Sect. 3.1) were actually been synthesized using are likely considering the complexity of the chemical
this technique [3.11]. The nanotube lengths are directly system (Si-C-B-N) and the high temperatures involved.
determined by the channel lengths; in other words by Terranova et al. [3.120] have investigated the cata-
the thickness of the substrate plate. One main advantage lyzed reaction between a solid carbon source and atomic
of the technique is the purity of the tubes (no catalyst hydrogen. Graphite nanoparticles (≈ 20 nm) are sent
remnants, and few other carbon phases). On the other with a stream of H2 onto a Ta filament heated at
hand, the nanotube structure is not closed at both ends, 2200 ◦ C. The species produced, whatever they are, then
which can be an advantage or a drawback depending on hit a Si polished plate warmed to 900 ◦ C that supports
the application. For instance, the porous matrix must be transition metal particles. The whole chamber is kept
dissolved using one of the chemical treatments previ- in a dynamic vacuum of 40 Torr. SWNTs are supposed
ously cited in order to recover the tubes. The fact that to form according to the authors, although their images
the tubes are open makes them even more sensitive to are not very convincing. One major drawback of the
attack from acids. method, besides its complexity compared to the others,
is that it is difficult to recover the nanotubes from the Si
3.2.3 Miscellaneous Techniques substrates to which they seem to be firmly bonded.
The final example is an attempt to prepare nano-
In addition to the major techniques described in tubes by diffusion flame synthesis [3.121]. A regular
Sects. 3.2.1 and 3.2.2, many attempts to produce nano- gaseous hydrocarbon source (ethylene, . . .) along with
tubes in various ways, often with a specific goal in mind, ferrocene vapor is passed into a laminar diffusion flame
such as looking for a low-cost or a catalyst-free produc- derived from air and CH4 of temperature 500–1200 ◦ C.
tion process, can be found in the literature. As yet, none SWNTs are formed, together with encapsulated metal
has been convincing enough to be presented as a serious particles, soot, and so on. In addition to a low yield, the
alternative to the major processes described previously. SWNT structure is quite poor.
Some examples are provided in the following.
Hsu et al. [3.117] have succeeded in prepar- 3.2.4 Synthesis of Carbon Nanotubes
ing MWNTs (including coiled MWNTs, a peculiar with Controlled Orientation
morphology resembling a spring) by a catalyst-free
(although Li was present) electrolytic method, by run- Several applications (such as field emission-based dis-
ning a 3–5 A current between two graphite electrodes plays Sect. 3.6) require that carbon nanotubes grow as
(the anode was a graphite crucible and the cathode highly aligned bunches, in highly ordered arrays, or
a graphite rod). The graphite crucible was filled with that they are located at specific positions. In this case,
lithium chloride, while the whole system was heated the purpose of the process is not mass production but
in air or argon at ≈ 600 ◦ C. As with many other controlled growth and purity, with subsequent control
techniques, by-products such as encapsulated metal par- of nanotube morphology, texture and structure. Gen-
ticles, carbon shells, amorphous carbon, and so on, are erally speaking, the more promising methods for the
formed. synthesis of aligned nanotubes are based on CCVD pro-
Cho et al. [3.118] have proposed a pure chemistry cesses, which involve the use of molecular precursors
route to nanotubes, using the polyesterification of citric as carbon sources, and the method of thermal crack-
acid onto ethylene glycol at 50 ◦ C, followed by poly- ing assisted by the catalytic activity of transition metal
merization at 135 ◦ C and then carbonization at 300 ◦ C (Co, Ni, Fe) nanoparticles deposited onto solid sup-
under argon, followed by oxidation at 400 ◦ C in air. ports. Although this approach initially produced mainly
Despite the latter oxidation step, the solid product con- MWNTs, DWNT [3.122] and SWNT [3.123] arrays
tains short MWNTs, although they obviously have poor can be selectively obtained today. Generally speaking,
Introduction to Carbon Nanotubes 3.2 Synthesis of Carbon Nanotubes 69
Part A 3.2
During the CCVD growth, nanotubes can self-
assemble into nanotube bunches aligned perpendicular
to the substrate if the catalyst film on the substrate
has a critical thickness [3.127, 128]. The driving forces
for this alignment are the van der Waals interac-
tions between the nanotubes, which allow them to
grow perpendicularly to the substrates. If the catalyst
nanoparticles are deposited onto a mesoporous sub- b)
strate, the mesoscopic pores may also have an effect
on the alignment when the growth starts, thus control-
ling the growth direction of the nanotubes. Two kinds of 2 µm
substrates have been used so far for this purpose: meso-
porous silica [3.129, 130] and anodic alumina [3.131].
Different methods of depositing metal particles onto
substrates have been reported in the literature:
1. Deposition of a thin film on alumina substrates us-
ing metallic salt precursor impregnation followed by
oxidation/reduction steps [3.132].
2. Embedding catalyst particles in mesoporous silica
by sol–gel processes [3.129].
3. Thermal evaporation of Fe, Co, Ni or Co-Ni metal
alloys on SiO2 or quartz substrates under high vac- 2 µm
uum [3.133, 134].
4. Photolithographic patterning of metal-containing Fig. 3.18 (a) Example of a controlled network of nano-
photoresist polymer using conventional black and tubes grown parallel to the substrate [3.125]; (b) example
white films as a mask [3.135] or photolithogra- of a free-standing MWNT array obtained from the pyroly-
phy and the inductive plasma deep etching tech- sis of a gaseous carbon source over catalyst nanoparticles
nique [3.136]. previously deposited onto a patterned substrate. Each
5. Electrochemical deposition into pores in anodic alu- square-base rod is a bunch of MWNTs aligned perpendic-
minum oxide templates [3.131]. ular to the surface (from [3.126])
6. Deposition of colloidal suspensions of catalyst par-
ticles with tailored diameters on a support [3.137–
Depositing the catalyst nanoparticles onto a prepat-
141], by spin-coating for instance.
terned substrate allows one to control the frequency of
7. Stamping a catalyst precursor over a patterned sil-
local occurrence and the arrangement of the nanotube
icon wafer is also possible and has been used to
bunches formed. The materials produced mainly con-
grow networks of nanotubes parallel to the sub-
sist of arrayed, densely packed, freestanding, aligned
strate (Fig. 3.18a), or more generally to localize the
MWNTs (Fig. 3.18b), which are quite suitable for
growth of individual CNTs [3.142].
field emission-based applications for instance [3.126].
A technique that combines the advantages of elec- SWNTs have also been produced, and it was reported
tron beam lithography and template methods has also that the introduction of water vapor during the CVD
been reported for the large-scale production of ordered process allows impurity-free SWNTs to be synthe-
MWNTs [3.143] or AFM tips [3.144]. sized [3.145], due to a mechanism related to that
70 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b) c)
Part A 3.3
50 µm
Low-temperature oven High-temperature oven
Fig. 3.19a–c Sketch of a double-furnace CCVD device used in the organometallic/hydrocarbon copyrolysis process.
(a) Sublimation of the precursor. (b) Decomposition of the precursor and MWNT growth onto the substrate. (c) Example
of the densely packed and aligned MWNT material obtained (from [3.146])
previously proposed for the effect of using alcohol in- the planar substrates used do not exhibit high surface
stead of hydrocarbon feedstock [3.127]. areas, the dispersion of the metal can be a key step in
When a densely packed coating of vertically the process. It has been observed that an etching pre-
aligned MWNTs is desired (Fig. 3.19c), another route treatment of the surface of the deposited catalyst thin
is the pyrolysis of hydrocarbons in the presence of film with NH3 may be critical to efficient nanotube
organometallic precursor molecules like metallocene or growth of nanotubes since it provides the appropri-
iron pentacarbonyl, operating in a dual furnace sys- ate metal particle size distribution. It may also favor
tem (Fig. 3.19a,b). The organometallic precursor (such the alignment of MWNTs and prevent the formation
as ferrocene) is first sublimed at low temperature in of amorphous carbon due to the thermal cracking of
the first furnace or injected as a solution along with acetylene [3.150]. The application of phthalocyanines
the hydrocarbon feedstock, and then the whole sys- of Co, Fe and Ni has also been reported, and in
tem is pyrolyzed at higher temperature in the second this case the pyrolysis of the organometallic precur-
furnace [3.105, 146–149]. The important parameters sors also produces the carbon for the vertically aligned
here are the heating or feeding rate of ferrocene, MWNTs [3.151].
the flow rates of the vector gas (Ar or N2 ) and the Densely packed coatings of vertically aligned
gaseous hydrocarbon, and the temperature of pyrolysis MWNTs may also be produced over metal-containing
(650–1050 ◦ C). Generally speaking, the codeposition deposits, such as iron oxides on aluminum [3.152],
process using [Fe(CO)5 ] as the catalyst source results in which case MWNT synthesis takes place on small
in thermal decomposition at elevated temperatures, pro- particles that are formed from the iron oxide deposit.
ducing atomic iron that deposits on the substrates in Interestingly, it has been recently reported that well
the hot zone of the reactor. Since nanotube growth oc- aligned MWNT arrays can be produced on a large scale
curs at the same time as the introduction of [Fe(CO)5 ], on ceramic spheres using the floating catalyst tech-
the temperatures chosen for the growth depend on the niques [3.153]. Finally, it has been recently reported
carbon feedstock utilized; for example, they can vary that the Langmuir–Blodgett method can effectively be
from 750 ◦ C for acetylene to 1100 ◦ C for methane. Mix- used to produce monolayers of aligned noncovalently
tures of [FeCp2 ] and xylene or [FeCp2 ] and acetylene functionalized SWNTs from organic solvent with dense
have also been successfully used to produce freestand- packing [3.154]. This method seems valid for bulk ma-
ing MWNTs. terials with various diameters and offers the advantage
The nanotube yield and quality are directly linked to that the SWNT monolayers are readily patterned for
the amount and size of the catalyst particles, and since device integration by microfabrication.
or a compromise between some of the proposals. An- From various results, it appears that the most im-
other reason is that the phenomena that occur during portant parameters are probably the thermodynamic
growth are pretty rapid and difficult to observe in situ. It ones (only temperature will be considered here), the
is generally agreed, however, that growth occurs such catalyst particle size, and the presence of a substrate.
that the number of dangling bonds is minimized, for Temperature is critical and basically corresponds to the
energetic reasons. discrepancy between CCVD methods and solid carbon
source-based methods.
Part A 3.3
3.3.1 Catalyst-Free Growth
Low-Temperature Conditions
As already mentioned, in addition to the templating Low-temperature conditions are typical used in CCVD,
technique, which is merely a chemical vapor infiltra- where nanotubes are frequently found to grow far below
tion mechanism for pyrolytic carbon, the growth of 1000 ◦ C. If the conditions are such that the catalyst is
c-MWNT as a deposit on the cathode in the electric a crystallized solid, the nanofilament is probably formed
arc method is a rare example of catalyst-free carbon via a mechanism similar to a VLS mechanism, in which
nanofilament growth. The driving force is obviously three steps are defined:
related to the electric field; in other words to charge
1. Adsorption then decomposition of C-containing
transfer from one electrode to the other via the par-
gaseous moieties at the catalyst surface
ticles contained in the plasma. It is not clear how the
2. Dissolution then diffusion of the C species through
MWNT nucleus is formed, but once it has, it may in-
the catalyst, thus forming a solid solution
clude the direct incorporation of C2 species into the
3. Back-precipitation of solid carbon as nanotube
primary graphene structure, as it was previously pro-
walls. The texture is then determined by the orienta-
posed for fullerenes [3.155]. This is supported by recent
tion of the crystal faces relative to the filament axis
C2 radical concentration measurements that reveal an
(Fig. 3.20), as demonstrated beyond doubt by trans-
increasing concentration of C2 from the anode being
mission electron microscopy images such as those
consumed at the growing cathode (Fig. 3.14). This indi-
in Rodriguez et al. [3.20].
cates that C2 are only secondary species and that the C2
species may actually actively participate in the growth This mechanism can therefore provide either c-MWNT,
of c-MWNTs in the arc method. h-MWNT, or platelet nanofibers. The latter, however,
are mainly formed in large particle sizes (> 100 nm
3.3.2 Catalytically Activated Growth for example). Platelet nanofibers with low diameters
(< 40 nm) have never been observed. The reasons for
Growth mechanisms involving catalysts are more diffi- this are related to graphene energetics, such as the need
cult to ascertain, since they are more diverse. Although to reach the optimal ratio between the amount of edge
it involves a more or less extensive contribution from carbon atoms (with dangling bonds) and inner carbon
a VLS (vapor–liquid–solid [3.156]) mechanism, it is atoms (where all of the σ and π orbitals are satisfied).
quite difficult to find comprehensive and plausible ex- If conditions are such that the catalyst is a liquid
planations that are able to account for both the various droplet, due to the use of high temperatures or because
conditions used and the various morphologies observed. a catalyst that melts at a low temperature is employed,
What follows is an attempt to provide overall expla- a mechanism similar to that described above can still
nations of most of the phenomena, while remaining occur, which is really VLS (vapor = gaseous C species,
consistent with the experimental data. We do not con- liquid = molten catalyst, S = graphenes), but there are
sider any hypothesis for which there is a lack of obviously no crystal faces to orient preferentially with
experimental evidence, such as the moving nanocata- the rejected graphenes. Energy minimization require-
lyst mechanism, which proposes that dangling bonds ments will therefore tend to make them concentric and
from a growing SWNT may be temporarily stabi- parallel to the filament axis.
lized by a nanosized catalyst located at the SWNT With large catalyst particles (or in the absence of
tip [3.28], or the scooter mechanism, which pro- any substrate), the mechanisms above will generally fol-
poses that dangling bonds are temporarily stabilized low a tip growth scheme: the catalyst will move forward
by a single catalyst atom which moves around the while the rejected carbon will form the nanotube be-
edge of the SWNT , allowing subsequent C atom ad- hind, whether there is a substrate or not. In this case,
dition [3.157]. there is a good chance that one end will be open. On
72 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b) c) a) b) c) d) g)
Part A 3.3
e)
SWNT
f)
CH 4 CH 4
Fe2O3
Scale bars= Substrate
10 nm
Fig. 3.21a–g High-resolution transmission electron mi-
croscopy images of several SWNTs grown from iron-based
Fig. 3.20a–c Illustration of the possible relationships be- nanoparticles using the CCVD method, showing that par-
tween the outer morphology of the catalyst crystal and ticle sizes determine SWNT diameters in this case (adapted
the inner texture of the subsequent carbon nanofila- from [3.158]). Yet the catalyst crystal imaged (the dark
ment (adapted from [3.20]). In (a), a nanotube with spot at the bottom of each tube) is different for each fig-
graphenes making the wall arranged concentrically (con- ure, considering images backward from (f) to (c) illustrates
centric MWNT). In (b), a nanofibre with graphenes what could be a sequence of growth of a SWNT from
arranged so that they make an angle with respect to the a single nanocrystal, as sketched in (g). (a) and (b) show
nanofibre axis (herringbone nanofibre). In (c), a nanofi- additional examples of fully grown SWNTs, similar to (c)
bre with the graphenes piled up perpendicularly to the
nanofibre axis (platelet nanofibre). Crystals are drawn with threshold in terms of its carbon content, it expels it quite
the projected plane perpendicular to the electron beam suddenly. Then it becomes able to incorporate a given
in a transmission electron microscope; the crystal mor- amount of carbon again without having any catalytic ac-
phologies and the subsequent graphene arrangements in the tivity for a little while. Then over-saturation is reached
out-of-plane dimension are not intended to be accurate rep- again, and so on. An exhaustive study of this phe-
resentations in these sketches (for example, the graphenes nomenon has been carried out by Jourdain et al. [3.159].
in the herringbone-type nanotubes or nanofibers in (b) can- Therefore, it is clear that 1 catalyst particle = 1
not be arranged like a pile of open books, as sketched here, nanofilament in any of the mechanisms above. This ex-
because it would leave too many dangling bonds) plains why, although it is possible to make SWNTs
by CCVD methods, controlling the catalyst particle
the other hand, when the catalyst particles deposited size is critical, since it influences the nanofilament that
onto the substrate are small enough (nanoparticles) to be grows from it. Achieving a really narrow size distribu-
held in place by interaction forces with the substrate, the tion in CCVD is quite challenging, particularly when
growth mechanism will follow a base growth scheme, nanosizes are required for the growth of SWNTs. Only
where the carbon nanofilament grows away from the particles < 2 nm are useful for this (Fig. 3.21), since
substrate, leaving the catalyst nanoparticle attached to larger SWNTs are not favored energetically [3.10]. An-
the substrate (Fig. 3.21). other distinguishing aspect of the CCVD method and
The bamboo texture that affects both the herring- its related growth mechanisms is that the process can
bone and the concentric texture may reveal a distin- occur all along the isothermal zone of the reactor fur-
guishing aspect of the dissolution-rejection mechanism: nace since it is continuously fed with a carbon-rich
the periodic, discontinuous dynamics of the phe- feedstock, which is generally in excess, with a constant
nomenon. Once the catalyst has reached the saturation composition at a given species time of flight. Roughly
Introduction to Carbon Nanotubes 3.3 Growth Mechanisms of Carbon Nanotubes 73
Table 3.2 Guidelines indicating the relationships between possible carbon nanofilament morphologies and some basic synthesis
conditions. Columns (1) and (2) mainly relate to CCVD-based methods; column (3) mainly relates to plasma-based methods
Part A 3.3
(1) (2) atoms
(3)
3 nm SWNT SWNT ? base-
Catalyst growth
particle tip- long short
3 nm MWNT
size tip- growth length length
(c,h,b) c-MWNT SWNT growth
platelet
nanofiber
catalyst)
size)
speaking, the longer the isothermal zone (in gaseous can be explained by the previous association of car-
carbon excess conditions), the longer the nanotubes. bon atoms into a pentagon, because it is the fastest way
This is why the lengths of the nanotubes can be much to limit dangling bonds at low energetic cost, thereby
longer than those obtained using solid carbon source- providing a fixation site for other carbon atoms (or
based methods. C2 ) which also will tend to close into a ring, again
Table 3.2 provides an overview of the relationship to limit dangling bonds. Since adjacent pentagons are
between general synthesis conditions and some features not energetically favored, these cycles will be hexagons.
of nanotube grown. Such a molecule is thought to be a probable precursor
for fullerenes. Fullerenes are actually always produced,
High-Temperature Conditions even in conditions that produce SWNTs. The same
High-temperature conditions are typical used in solid saturation in C described in Sect. 3.3.2 occurs for the
carbon source-based methods such as the electric carbon-metal alloy droplet as well, resulting in the pre-
arc method, laser vaporization, and the solar furnace cipitation of excess C outside the particle due to the
method (Sect. 3.2). The huge temperatures involved effect of the decreasing thermal gradient in the reac-
(several thousands of ◦ C) atomize both the carbon tor, which decreases the solubility threshold of C in
source and the catalyst. Of course, catalyst-based the metal [3.160]. Once the inner carbon atoms reach
SWNTs do not form in the areas with the highest tem- the surface of the catalyst particle, they meet the outer
peratures (contrary to c-MWNTs in the electric arc carbon species, including corannulene, that will con-
method); the medium is a mixture of atoms and radicals, tribute to capping the merging nanotubes. Once formed
some of which are likely to combine and condense into and capped, nanotubes can grow both from the inner
liquid droplets. At some distance from the atomization carbon atoms (Fig. 3.22a), according to the VLS mech-
zone, the medium is therefore made of carbon metal al- anism proposed by Saito et al. [3.160], and from the
loy droplets and of secondary carbon species that range outer carbon atoms, according the adatom mechanism
from C2 to higher order molecules such as corannulene, proposed by Bernholc et al. [3.161]. In the latter, carbon
which is made of a central pentagon surrounded by five atoms from the surrounding medium in the reactor are
hexagons. The preferred formation of such a molecule attracted then stabilized by the carbon/catalyst interface
74 Part A Nanostructures, Micro-/Nanofabrication and Materials
T° T°
Part A 3.4
Fig. 3.22 (a) Mechanism proposed for SWNT growth (see text). (b) Transmission electron microscopy image of SWNT
growing radially from the surface of a large Ni catalyst particle in an electric arc experiment. (Modified from [3.18])
at the nanotube/catalyst surface contact, promoting their those associated with CCVD methods. The reason that
subsequent incorporation at the tube base. The growth the most frequent diameter is ≈ 1.4 nm is again a mat-
mechanism therefore mainly follows the base growth ter of energy balance. Single-wall nanotubes larger than
scheme. However, once the nanotubes are capped, any ≈ 2.5 nm are not stable [3.10]. On the other hand,
C2 species that still remains in the medium that meets the strain on the C−C bond increases as the radius
the growing nanotubes far from the nanotube/catalyst of curvature decreases. The optimal diameter (1.4 nm)
interface may still incorporate the nanotubes from both should therefore correspond to the best energetic com-
the side wall or the tip, thereby giving rise to some pro- promise. Another difference from the low-temperature
portion of Stone–Wales defects [3.45]. The occurrence mechanism for CCVD is that temperature gradients in
of a nanometer-thick surface layer of yttrium carbide high temperature methods are huge, and the gas phase
(onto the main Ni-containing catalyst core), the lattice composition surrounding the catalysts droplets is also
distance of which is commensurable with that of the subjected to rapid changes (as opposed to what could
C−C distance in graphene (as recently revealed by Gav- happen in a laminar flow of a gaseous feedstock whose
illet et al. [3.55]), could possibly play a beneficial role in carbon source is in excess). This explains why nano-
stabilizing the nanotube/catalyst interface, which could tubes from arcs are generally shorter than nanotubes
explain why the SWNT yield is enhanced by bimetallic from CCVD, and why mass production by CCVD is
alloys (as opposed to single metal catalysts). favored. In the latter, the metallic particle can act as
A major difference from the low-temperature mech- a catalyst repeatedly as long as the conditions are main-
anisms described for CCVD methods is that many tained. In the former, the surrounding conditions change
nanotubes are formed from a single, relatively large continuously, and the window for efficient catalysis can
(≈ 10–50 nm) catalyst particle (Fig. 3.22b), whose size be very narrow. Decreasing the temperature gradients
distribution is therefore not as critical as it is for the low- that occur in solid carbon source-based methods of pro-
temperature mechanisms (particles that are too large, ducing SWNT, such as the electric arc reactor, should
however, induce polyaromatic shells instead of nano- therefore increase the SWNT yield and length [3.162].
tubes). This is why the diameters of SWNTs grown Amazingly, this is in opposition to what is observed
at high temperature are much more homogeneous than during arc-based fullerene production.
herringbone, bamboo) and the quality of the nanotex- a single, flat graphene sheet: ≈ 2700 m2 /g (accounting
ture (Sect. 3.1), both of which control the extent of for both sides).
anisotropy. Actually, for polyaromatic solids that con- In reality, nanotubes – specifically SWNTs – are
sist of stacked graphenes, the bond strength varies usually associated with other nanotubes in bundles,
significantly depending on whether the in-plane direc- fibers, films, papers, and so on, rather than as a single
tion is considered (characterized by very strong covalent entity. Each of these associations has a specific range of
and therefore very short – 0.142 nm – bonds) or the di- porosities that determines its adsorption properties (this
Part A 3.4
rection perpendicular to it (characterized by very weak topic is also covered in Sect. 3.6.2 on applications). It
van der Waals and therefore very loose – ≈ 0.34 nm is therefore more appropriate to discuss adsorption onto
– bonds). Such heterogeneity is not found in single the outer or the inner surface of a bundle of SWNTs.
(isolated) SWNTs. However, the heterogeneity returns, Furthermore, theoretical calculations have predicted
along with the related consequences, when SWNTs as- that the adsorption of molecules onto the surface or in-
sociate into bundles. Therefore, the properties – and side of a nanotube bundle is stronger than that onto an
applicability – of SWNTs may also change dramatically individual tube. A similar situation exists for MWNTs,
depending on whether single SWNT or SWNT ropes where adsorption can occur on or inside the tubes or
are involved. between aggregated MWNTs. It has also been shown
In the following, we will emphasize the proper- that the curvature of the graphene sheets constituting
ties of SWNTs, since their unique structures often lead the nanotube walls results in a lower heat of adsorption
to different properties to regular polyaromatic solids. compared to planar graphene (Sect. 3.1.1).
However, we will also sometimes discuss the properties
of MWNTs for comparison. Accessible Specific Surface Area of CNTs
Various studies dealing with the adsorption of nitro-
3.4.2 General Properties of SWNTs gen onto MWNTs [3.165–167] and SWNTs [3.168]
have highlighted the porous nature of these two ma-
The diameters of SWNT-type carbon nanotubes fall in terials. The pores in MWNTs can be divided mainly
the nanometer regime, but SWNTs can be hundreds into hollow inner cavities with small diameters (with
of micrometers long. SWNTs are narrower in diameter narrow size distributions, mainly 3–10 nm) and aggre-
than the thinnest line that can be obtained in electron gated pores (with wide size distributions, 20–40 nm),
beam lithography. SWNTs are stable up to 750 ◦ C in air formed by interactions between isolated MWNTs. It
(but they are usually damaged before this temperature is also worth noting that the ultrastrong nitrogen cap-
is reached due to oxidation mechanisms, as demon- illarity in the aggregated pores dominates the total
strated by the fact that they can be filled with molecules adsorption, indicating that the aggregated pores are
(Sect. 3.5). They are stable up to ≈ 1500–1800 ◦ C in much more important than the inner cavities of the
inert atmosphere, beyond which they transform into MWNTs during adsorption. The determination of the
regular, polyaromatic solids (phases built with stacked space available between a bunch of closed MWNTs
graphenes instead of single graphenes) [3.163]. They has been performed by grand canonical Monte Carlo
have half the mass density of aluminum. The properties simulation of nitrogen adsorption, resulting in a satis-
of a SWNT, like any molecule, are heavily influenced factory description of the experimental N2 adsorption
by the way that its atoms are arranged. The physical and and showing that the distance between nanotubes is
chemical behavior of a SWNT is therefore related to its in the 4–14 nm range [3.169]. Adsorption of N2 has
unique structural features [3.164]. been studied on as-prepared and acid-treated SWNTs,
and the results obtained highlight the microporous na-
3.4.3 Adsorption Properties of SWNTs ture of SWNT materials, as opposed to the mesoporous
nature of MWNT materials. Also, as opposed to iso-
An interesting feature of a SWNT is that it has the lated SWNTs (see above), surface areas that are well
highest surface area of any molecule due to the fact above 400 m2 g−1 have been measured for SWNT-
that a graphene sheet is probably the only example of bundle-containing materials, with internal surface areas
a sheetlike molecule that is energetically stable under of 300 m2 g−1 or higher.
normal conditions. If we consider an isolated SWNT The theoretical surface area of a carbon nanotube
with one open end (achieved through oxidation treat- has a broad range, from 50 to 1315 m2 g−1 depending
ment for instance), the surface area is equal to that of on the number of walls, the diameter, and the number of
76 Part A Nanostructures, Micro-/Nanofabrication and Materials
nanotubes in a bundle of SWNTs [3.170]. Experimen- erately curved and are expected to be much less reactive
tally, the surface area of a SWNT is often larger than towards dissociative chemisorption. Models have also
that of a MWNT. The total surface area of as-grown predicted enhanced reactivity at the kink sites of bent
SWNTs is typically between 400 and 900 m2 g−1 (mi- SWNTs [3.177]. Additionally, it is worth noting that
cropore volume 0.15–0.3 cm3 g−1 ), whereas values of unavoidable imperfections, such as vacancies, Stone–
200 and 400 m2 g−1 for as-produced MWNTs are often Wales defects, pentagons, heptagons and dopants, are
reported. In the case of SWNTs, the diameters of the believed to play a role in tailoring the adsorption prop-
Part A 3.4
tubes and the number of tubes in the bundle will have erties [3.178].
the most effect on the BET value. It is worth noting Considering closed-end SWNTs first, simple mol-
that opening or closing the central canal significantly in- ecules can be adsorbed onto the walls of the outer
fluences the adsorption properties of nanotubes. In the nanotubes of the bundle and preferably on the external
case of MWNTs, chemical treatments such as KOH or grooves. In the first stages of adsorption (correspond-
NaOH activation are useful for promoting micropor- ing to the most attractive sites for adsorption), it seems
osity, and surface areas as high as 1050 m2 g−1 have that adsorption or condensation in the interstitial chan-
been reported [3.171, 172]. An efficient two-step treat- nels of the SWNT bundles depends on the size of the
ment (acid +CO2 activation) has been reported to open molecule (and/or on the SWNT diameters) and on their
both ends of MWNTs [3.173]. Therefore, it appears interaction energies [3.179–181]. Opening the tubes fa-
that opening or cutting carbon nanotubes, as well as vors gas adsorption (including O2 , N2 within the inner
chemically treating them (using purification steps for walls [3.182, 183]). It was found that the adsorption
example) can considerably affect their surface area and of nitrogen on open-ended SWNT bundles is three
pore structure. times larger than that on closed-ended SWNT bun-
dles [3.184]. The significant influence that the external
Adsorption Sites and Binding Energy surface area of the nanotube bundle has on the charac-
of the Adsorbates ter of the surface adsorption isotherm of nitrogen (type
An important problem to solve when considering ad- I, II or even IV of the IUPAC classification) has been
sorption onto nanotubes is to identify the adsorption demonstrated from theoretical calculations [3.185].
sites. The adsorption of gases into a SWNT bundle Additionally, it has been shown that the analysis of the-
can occur inside the tubes (internal sites), in the in- oretical adsorption isotherms, determined from a simple
terstitial triangular channels between the tubes, on the model based on the formalism of Langmuir and Fowler,
outer surface of the bundle (external sites), or in the can help to experimentally determined the ratio of
grooves formed at the contacts between adjacent tubes open to closed SWNTs in a sample [3.186]. For
on the outside of the bundle (Fig. 3.23). Experimental hydrogen and other small molecules like CO, compu-
adsorption studies on SWNT have confirmed the ad- tational methods have shown that, for open SWNTs,
sorption on internal, external and groove sites [3.175]. the pore, interstitial and groove sites are energetically
Modeling studies have pointed out that the convex sur- more favorable than surface sites [3.187, 188]. In the
face of the SWNT is more reactive than the concave case of carbon monoxide, aside from physisorbed CO,
one and that this difference in reactivity increases as CO hydrogen bonds to hydroxyl functionalities cre-
the tube diameter decreases [3.176]. Compared to the ated on the SWNTs by acid purification have been
highly bent region in fullerenes, SWNTs are only mod- identified [3.188]. FTIR and temperature-programmed
desorption (TPD) experiments have shown that NH3 or
Pore
NO2 adsorb molecularly and that NO2 is slightly more
strongly bound than NH3 [3.189]. For NO2 , the for-
mation of nitrito (O-bonded) complexes is preferred to
Interstitial Groove nitro (N-bonded) ones. For ozone, a strong oxidizing
agent, theoretical calculations have shown that ph-
ysisorption occurs on ideal, defect-free SWNT, whereas
strong chemisorption occurs on Stone–Wales defects,
Surface highlighting the key role of defective sites in adsorption
properties [3.190]. Finally, for acetone, TPD experi-
Fig. 3.23 Sketch of a SWNT bundle, illustrating the four ments have shown that this molecule chemisorbs on
different adsorption sites (adapted from [3.174]) SWNT while physisorption occurs on graphite [3.191].
Introduction to Carbon Nanotubes 3.4 Properties of Carbon Nanotubes 77
For MWNTs, adsorption can occur in the aggre- physisorption, some experimental results, in particular
gated pores, inside the tube or on the external walls. for hydrogen, are still controversial (Sect. 3.6.2). For
In the latter case, the presence of defects, as incom- platelet nanofibers, the initial dissociation of hydro-
plete graphene layers, must be taken into consideration. gen on graphite edge sites, which constitute most of
Although adsorption between the graphenes (interca- the nanofiber surface, has been proposed [3.196]. For
lation) has been proposed in the case of hydrogen carbon nanotubes, a mechanism that involves H2 dis-
adsorption in h-MWNTs or platelet nanofibers [3.192], sociation on the residual metal catalyst followed by
Part A 3.4
it is unlikely to occur for many molecules due to steric H spillover and adsorption on the most reactive nano-
effects and should not prevail for small molecules due tube sites was envisaged [3.197]. Similarly, simply
to the long diffusion paths involved. In the case of inor- mixing carbon nanotubes with supported palladium cat-
ganic fluorides (BF3 , TiF4 , NbF5 and WF6 ), accommo- alysts increased the hydrogen uptake of the carbon by
dation of the fluorinated species into the carbon lattice a factor of three, due to hydrogen spillover from the
has been shown to result from intercalation and ad- supported catalyst [3.198]. Doping nanotubes with al-
sorption/condensation phenomena. In this case, doping- kali may enhance hydrogen adsorption, due to charge
induced charge transfer has been demonstrated [3.193]. transfer from the alkali metal to the nanotube, which
Only a few studies deal with adsorption sites in polarizes the H2 molecule and induces dipole interac-
MWNTs, but it has been shown that butane adsorbs tions [3.199].
more onto MWNTs with smaller outside diameters, Generally speaking, the adsorbates can be either
which is consistent with another statement that the charge donors or acceptors to the nanotubes. Trends
strain on curved graphene surfaces affects sorption. in the binding energies of gases with different van der
Most of the butane adsorbs to the external surface of Waals radii suggest that the groove sites of SWNTs
the MWNTs while only a small fraction of the gas con- are the preferred low coverage adsorption sites due to
denses in the pores [3.194]. Comparative adsorption of their higher binding energies. Finally, several studies
krypton or of ethylene onto MWNTs or onto graphite have shown that, at low coverage, the binding energy
has allowed scientists to determine the dependence of of the adsorbate on SWNT is between 25 and 75%
the adsorption and wetting properties of the nanotubes higher than the binding energy on a single graphene.
on their specific morphologies. Nanotubes were found This discrepancy can be attributed to an increase of ef-
to have higher condensation pressures and lower heats fective coordination at the binding sites, such as the
of adsorption than graphite [3.195]. These differences groove sites, in SWNTs bundles [3.200, 201]. Repre-
are mainly due to decreased lateral interactions between sentative results on the adsorption properties of SWNTs
the adsorbed molecules, related to the curvature of the and MWNTs are summarized in Table 3.3.
graphene sheets.
A limited number of theoretical as well as exper- 3.4.4 Electronic and Optical Properties
imental studies on the binding energies of gases onto
carbon nanotubes exist. While most of these studies re- The electronic states in SWNTs are strongly influ-
port low binding energies on SWNTs, consistent with enced by their one-dimensional cylindrical structures.
Table 3.3 Adsorption properties and sites of SWNTs and MWNTs. The letters in the Absorption sites column refer to
Fig. 3.22. The data in the last two columns are from [3.174]
Type of Porosity Surface Binding energy Adsorption Attractive Surface
` 3 −1 ´
nanotube cm g area of the adsorbate sites potential area
` 2 ´
m /g per site per site
` 2 ´
(eV) m /g
SWNT Microporous 400–900 Low, mainly Surface (A) 0.049 483
(bundle) Vmicro : physisorption Groove (B) 0.089 22
0.15– 0.3 25 –75% > graphite Pore (C) 0.062 783
Interstitial (D) 0.119 45
MWNT Mesoporous 200–400 Physisorption Surface − −
Pore
Aggregated
pores
78 Part A Nanostructures, Micro-/Nanofabrication and Materials
One-dimensional subbands are formed that have strong maps provide information about the helicity (chirality)-
singularities in the density of states (Van Hove sin- dependent transition energies and the electronic band
gularities) [3.202]. By rolling the graphene sheet to structures of CNTs [3.210]. Agglomeration of tubes
form a tube, new periodic boundary conditions are im- into ropes or bundles influences the electronic states
posed on the electronic wavefunctions, which give rise of CNTs. Photoluminescence signals are quenched for
to one-dimensional subbands: Cn K = 2q where q is agglomerated tubes.
an integer. Cn is the roll-up vector na1 + ma2 which CNTs are model systems for the study of one-
Part A 3.4
defines the helicity (chirality) and the diameter of the dimensional transport in materials. Apart from the
tube (Sect. 3.1). Much of the electronic band structure of singularities in the density of states, electron–electron
CNTs can be derived from the electronic band structure interactions are expected to show drastic changes at
of graphene by applying the periodic boundary condi- the Fermi edge; the electrons in CNTs are not de-
tions of the tube under consideration. The conduction scribed by a Fermi liquid, but instead by a Luttinger
and the valence bands of the graphene only touch at liquid model [3.211] that describes electronic transport
six corners (K points) of the Brillouin zone [3.203]. in one-dimensional systems. It is expected that the vari-
If one of these subbands passes through the K point, ation of electronic conductance vs. temperature follows
the nanotube is metallic; otherwise it is semiconduct- a power law, with zero conductance at low tempera-
ing. This is a unique property that is not found in tures. Depending on how L φ (the coherence length) on
any other one-dimensional system, which means that the one hand and L m (the electronic mean free path)
for certain orientations of the honeycomb lattice with on the other hand compare to L (the length of the
respect to the tube axis (chirality), some nanotubes nanotube), different conduction modes are observed:
are semiconducting and others are metallic. The band ballistic if L L φ , L L m , diffusive if L φ L m < L
gap for semiconducting tubes is found to be inversely and localization if L m L φ L. Fluctuations in the
proportional to the tube diameter. As pointed out in conductance can be seen when L ≈ L φ . For ballistic
Sect. 3.1, knowing (n, m) allows us, in principle, to pre- conduction (a small number of defects) [3.212–214],
dict whether the tube is metallic or not. The energy gap the predicted electronic conductance is independent of
decreases for larger tube diameters and MWNTs with the tube length. The conductance value is twice the
larger diameter are found to have properties similar to fundamental conductance unit G 0 = 4 e/h due to the
other forms of regular, polyaromatic solids. It has been existence of two propagating modes. Due to the re-
shown that electronic conduction mostly occurs through duced electron scattering observed for metallic CNTs
the external tube for MWNTs; even so, interactions with and their stability at high temperatures, CNTs can sup-
internal tubes often cannot be neglected and they de- port high current densities (max. 109 A/cm2 ): about
pend upon the helicity of the neighboring tubes [3.204]. three orders of magnitude higher than Cu. Structural
The electronic and optical properties of the tubes are defects can, however, lead to quantum interference of
considerably influenced by the environment [3.205]. the electronic wave function, which localizes the charge
Under externally applied pressure, the small interac- carriers in one-dimensional systems and increases re-
tion between the tube walls results in the internal tubes sistivity [3.211, 215, 216]. Localization and quantum
experiencing reduced pressure [3.206]. The electronic interference can be strongly influenced by applying
transition energies are in the infrared and visible spec- a magnetic field [3.217]. At low temperatures, the dis-
tral range. The one-dimensional Van Hove singularities crete energy spectrum leads to a Coulomb blockade
have a large influence on the optical properties of CNTs. resulting in oscillations in the conductance as the gate
Visible light is selectively and strongly absorbed, which voltage is increased [3.216]. In order to observe the dif-
can lead to the spontaneous burning of agglomerated ferent conductance regimes, it is important to consider
SWNTs in air at room temperature [3.207]. Strong the influence of the electrodes where Schottky barriers
Coulomb interaction in quasi-one dimension leads to are formed. Palladium electrodes have been shown to
the formation of excitons with very large binding en- form excellent junctions with nanotubes [3.218]. The
ergies in CNTs (200–400 meV), and degenerated states influence of superconducting electrodes or ferromag-
at the K , K points lead to multiple exciton states with netic electrodes on electronic transport in CNTs due to
dipole allowed (bright) and dipole forbidden transitions spin polarization has also been explored [3.219, 220].
(dark) [3.208,209]. Photoluminescence can be observed As a probable consequence of both the small
in individual SWNT aqueous suspensions stabilized number of defects (at least the kind of defects that op-
by the addition of surfactants. Detailed photoexcitation pose phonon transport) and the cylindrical topography,
Introduction to Carbon Nanotubes 3.4 Properties of Carbon Nanotubes 79
Part A 3.4
5 T 400 H
Glass M 40 J M 46 J
While tubular nanomorphology is also observed for 4 5 X 24 M5 X 50 X 70
many two-dimensional solids, carbon nanotubes are X 49 M 60 J
Kevlar T 300 P 120
P 75
unique due to the particularly strong bonding between 3 X 20 P 55 P 100
the carbons (sp2 hybridization of the atomic orbitals) Nicalon
K 1100
P 25 M 50
2 M 40
of the curved graphene sheet, which is stronger than Steel
in diamond (sp3 hybridization), as revealed by the dif- 1
Titanium
ference in C−C bond lengths (0.142 versus 0.154 nm 200 400 600 800 1,000 1,200
Tensile modulus (GPa)
for graphene and diamond respectively). This makes
carbon nanotubes – SWNTs or c-MWNTs – particu- Fig. 3.24 Plot of the tensile strength versus the tensile modulus
larly stable against deformations. The tensile strength for various fibrous materials and SWNTs. Large circles are PAN-
of SWNTs can be 20 times that of steel [3.222] and based carbon fibers, which include the fiber with the highest tensile
has actually been measured as ≈ 45 GPa [3.223]. Very strength available on the market (T1000 from Torayca); Triangles
high tensile strength values are also expected for ideal are pitch-based carbon fibers, which include the fiber with the high-
(defect-free) c-MWNTs, since combining perfect tubes est tensile modulus on the market (K1100 from Amoco)
concentrically is not supposed to be detrimental to the
overall tube strength, provided the tube ends are well that observed for regular polyaromatic carbon mater-
capped (otherwise, concentric tubes could glide relative ials due the unique shape of the nanotube, its small
to each other, inducing high strain). Tensile strength val- diameter, and its structural properties. Unlike graphite,
ues as high as ≈ 150 GPa have actually been measured perfect SWNTs have no (chemically active) dangling
for perfect MWNTs from an electric arc [3.224], al- bonds (the reactions of polyaromatic solids is known
though the reason for such a high value compared to that to occur mainly at graphene edges). Unlike fullerenes,
measured for SWNTs is not clear. It probably reveals the ratio of weak sites (C−C bonds involved in hetero-
the difficulties involved in carrying out such measure- cycles) to strong sites (C−C bonds between regular
ments in a reliable manner. The flexural modulus of hexagons) is only deviates slightly from 0 for ideal
perfect MWNTs should logically be higher than that tubes. For C60 fullerenes this ratio is 1 – C60 molecules
for SWNTs [3.222], with a flexibility that decreases as have 12 pentagons (therefore accounting for 5 × 12 = 30
the number of walls increases. On the other hand, mea- C−C bonds) and 20 hexagons, each of them with three
surements performed on defective MWNTs obtained C−C bonds not involved in an adjacent pentagon but
from CCVD exhibit a range of 3–30 GPa [3.225]. Val- shared with a neighboring hexagon (so 20 × 3 × 1/2 = 30
ues of tensile modulus are also the highest values C−C bonds are involved in hexagons only). Although
known, 1 TPa for MWNTs [3.226], and possibly even graphene faces are chemically relatively inert, the ra-
higher for SWNTs, up to 1.3 TPa [3.227, 228]. Fig- dius of curvature imposed on the graphene in nanotubes
ure 3.24 illustrates how defect-free carbon nanotubes causes the three normally planar C−C bonds caused
could spectacularly revolutionize the field of high per- by sp2 hybridization to undergo distortions, resulting in
formance fibrous materials. bond angles that are closer to three of the four C−C
bonds in diamond (characteristic of genuine sp3 hy-
3.4.6 Reactivity bridization), as the radius of curvature decreases. Even
though it is not enough to make the carbon atoms chem-
The chemical reactivities of graphite, fullerenes, and ically reactive, one consequence of this is that either
carbon nanotubes are similar in many ways. Like any nesting sites are created at the concave surface, or strong
small object, carbon nanotubes have a large surface to physisorption sites are created above each carbon atom
interact with their environment (Sect. 3.4.1). It is worth of the convex surface, both with a bonding efficiency
noting, however, that nanotube chemistry differs from that increases as the nanotube diameter decreases.
80 Part A Nanostructures, Micro-/Nanofabrication and Materials
As already pointed out in Sect. 3.1, the chemical heterocycles, two pentagons and two heptagons, ar-
reactivities of SWNTs (and c-MNWTs) are believed ranged in pairs opposite each other. Such a defect allows
to derive mainly from the caps, since they contain six localized double bonds to form between the carbon
pentagons each, as opposed to the tube body, which atoms involved in the defect (instead of these elec-
supposedly only contains hexagons. Indeed, applying trons participating in the delocalized electron cloud
oxidizing treatments to carbon nanotubes (air oxida- above the graphene as usual, enhancing the chemical
tion, wet-chemistry oxidation) selectively opens the reactivity, for example toward chlorocarbenes [3.230]).
Part A 3.5
nanotube tips [3.229]. However, that SWNTs can be This means that the overall chemical reactivity of car-
opened by oxidation methods and then filled with for- bon nanotubesshould depend strongly on how they are
eign molecules such as fullerenes (Sect. 3.5) suggests synthesized. For example, SWNTs prepared by the
the occurrence of side defects [3.15], whose iden- arc-discharge method are believed to contain fewer
tity and occurrence were discussed and then proposed structural defects than CCVD-synthesized SWNTs,
to be an average of one Stone–Wales defect every which are more chemically reactive. Of course, the
5 nm along the tube length, involving about 2% of reactivity of h-MWNT-type nanotubes is intrinsically
the carbon atoms in a regular (10,10) SWNT [3.230]. higher, due to the occurrence of accessible graphene
A Stone–Wales defect is formed from four adjacent edges at the nanotube surface.
original motivation to create such hybrids was to obtain method, and were obtained directly during processing.
metal nanowires that are likely to of interest in electron- The filling materials were easily introduced in the sys-
ics (as quantum wires). In this case, the nanotubes were tem by drilling a central hole in the anode and filling
considered to be nanomolds for the metal filler, and it it with the heteroelement. The first hybrid products ob-
was probably intended that the nanomold was to be re- tained using this approach were all reported the same
moved afterwards. However, it is likely that this removal year [3.242–246] for heteroelements such as Pb, Bi,
of the SWNT container to liberate the one-dimensional YC2 and TiC. Later on, Loiseau and Pascard [3.247]
Part A 3.5
structure inside it may destroy or at least transform this showed that MWNTs could also be filled to several μm
structure due to the stabilizing effect of interactions with in length by elements such as Se, Sb, S, and Ge, but
the nanotube wall. only with nanoparticles of elements such as Bi, B, Al
Filling nanotubes while they grow (in situ filling) and Te. Sulfur was suggested to play an important role
was one of the pioneering methods of nanotechnology. during the in situ formation of filled MWNTs using arc
In most cases, however, the filling step is separate from discharge [3.248]. This technique is no longer the pre-
nanotube synthesis. Three filling methods can then be ferred one because it is difficult to control the filling
distinguished: (a) wet chemistry procedures; capillarity- ratio and yield and to achieve mass production.
based physical procedures involving (b) a molten
material or (c) a sublimated material. Wet Chemistry Filling Method
Generally speaking, it is difficult to estimate the The wet chemistry method requires that the nanotube
filling rate, and this is usually achieved through TEM tips are opened by chemical oxidation prior to the fill-
observation, without obtaining any statistics on the ing step. This is generally achieved by refluxing the
number of tubes observed. Moreover, as far as SWNTs nanotubes in dilute nitric acid [3.249–251], although
are concerned, the fact that the nanotubes are gathered other oxidizing liquid media may work as well, such as
into bundles makes it difficult to observe the exact num- [HCl + CrO3 ] [3.252] or chlorocarbenes formed from
ber of filled tubes, as well as to estimate the filled length the photolytic dissociation of CHCl3 [3.230], a rare ex-
for each tube. It however seems that estimation of filling ample of a nonacidic liquid route to opening SWNT
rates can now be reliably obtained from x-ray studies tips. If a dissolved form (such as a salt or oxide)
and Raman spectroscopy. of the desired metal is introduced during the open-
It is also possible to fill carbon nanotubes with ma- ing step, some of it will get inside the nanotubes.
terials that could not have been introduced directly. This An annealing treatment (after washing and drying the
is done by first filling the nanotubes with an appropri- treated nanotubes) may then lead to the oxide or to the
ate precursor (one that is able to sublime, or melt or metal, depending on the annealing atmosphere [3.229].
solubilize) that will later be transformed into the re- Although the wet chemistry method initially looked
quired material by chemical reaction or by a physical promising because a wide variety of materials can be
interaction, such as electron beam irradiation for ex- introduced into nanotubes in this way and it operates
ample [3.240]. For secondary chemical transformation, at temperatures that are not much different from room
reduction by H2 is often used to obtain nanotubes filled temperature; however, close attention must be paid to
with metals [3.241]. Sulfides can also be obtained if the oxidation method that is used. The damage caused
H2 S is used as a reducing agent [3.241]. to nanotubes by severe treatments (such by using nitric
Because the inner diameters of SWNTs are gener- acid) make them unsuitable for use with SWNTs. More-
ally smaller than those of MWNTs, it is more difficult over, the filling yield is not very good, probably due to
to fill them, and the driving forces involved in this phe- the solvent molecules that also enter the tube cavity: the
nomenon are not yet totally understood (see the review filled lengths rarely exceed 100 nm. Mittal et al. [3.252]
paper by Monthioux [3.230]). This field is therefore have recently filled SWNTs with CrO3 using wet chem-
growing fairly rapidly, and so we have chosen to cite istry with an average yield of ≈ 20%.
the pioneering works and then to focus on more recent
works dealing with the more challenging topic of filling Molten State Filling Method
SWNTs. The physical filling method involving a liquid (molten)
phase is more restrictive, firstly because some ma-
In Situ Filling Method terials can decompose when they melt, and secondly
Initially, most hybrid carbon nanotubes synthesized because the melting point must be compatible with the
were based on MWNTs prepared using the electric arc nanotubes, so the thermal treatment temperature should
82 Part A Nanostructures, Micro-/Nanofabrication and Materials
remain below the temperature of transformation or the cal properties of halides crystallized within carbon
nanotubes will be damaged. Because the filling occurs nanotubes, the crystallization of molten salts within
due to capillarity, the surface tension threshold of the small-diameter SWNTs has been studied in detail, and
molten material is 100–200 N/cm2 [3.253], although the one-dimensional crystals have been shown to inter-
this threshold was proposed for MWNTs, whose inner act strongly with the surrounding graphene wall. For
diameters (5–10 nm) are generally larger than those of example, Sloan et al. [3.256] described two-layer 4 : 4
SWNTs (1–2 nm). In a typical filling experiment, the coordinated KI crystals that formed within SWNTs that
Part A 3.5
MWNTs are closely mixed with the desired amount were ≈ 1.4 nm in diameter. These two-layer crystals
of filler by gentle grinding, and the mixture is then were all surface and had no internal atoms. Signifi-
vacuum-sealed in a silica ampoule. The ampoule is then cant lattice distortions occurred compared to the bulk
slowly heated to a temperature above the melting point structure of KI, where the normal coordination is 6 : 6
of the filler and slowly cooled. This method does not (meaning that each ion is surrounded by six identical
require that the nanotubes are opened prior to the heat close neighbors). Indeed, the distance between two ions
treatment. The mechanism of nanotube opening is yet across the SWNT capillary is 1.4 times as much as the
to be clearly established, but it is certainly related to same distance along the tube axis. This suggests an ac-
the chemical reactivities of the molten materials toward commodation of the KI crystal into the confined space
carbon, and more precisely toward defects in the tube provided by the inner nanotube cavity in the constrained
structure (Sect. 3.4.4). crystal direction (across the tube axis). This implies that
Most of the works involving the application of the interactions between the ions and the surrounding
this method to SWNTs come from Oxford Univer- carbon atoms are strong. The volume available within
sity [3.254–258], although other groups have followed the nanotubes thus somehow controls the crystal struc-
the same procedure [3.249, 251, 259]. The precur- tures of inserted materials. For instance, the structures
sors used to fill the nanotubes were mainly metal and orientations of encapsulated PbI2 crystals inside
halides. Although little is known about the physi- their capillaries were found to differ for SWNTs and
DWNTs, depending on the diameter of the confining
b) 0.36 nm nanotubes [3.254]. For SWNTs, most of the encapsu-
a)
lated one-dimensional PbI2 crystals obtained exhibited
a strong preferred orientation, with their (110) planes
aligning at an angle of around 60◦ to the SWNT axes, as
- .
c) 110
- 110 shown in Fig. 3.25a,b. Due to the extremely small diam-
eters of the nanotube capillaries, individual crystallites
2 nm - . - are often only a few polyhedral layers thick, as outlined
110 110
in Fig. 3.25d–h. Due to lattice terminations enforced by
capillary confinement, the edging polyhedra must be of
d) e) h) reduced coordination, as indicated in Fig. 3.25g,h. Sim-
ilar crystal growth behavior was generally observed to
occur for PbI2 formed inside DWNTs in narrow nano-
f) tubes with diameters comparable to those of SWNTs.
As the diameter of the encapsulating capillary increases,
g) however, different preferred orientations are frequently
observed (Fig. 3.26). In this example, the PbI2 crys-
2 nm
tal is oriented with the [121] direction parallel to the
Fig. 3.25a–h HRTEM images and corresponding structural model direction of the electron beam (Fig. 3.26a–d). If the
for PbI2 filled SWNTs. (a) Image of a bundle of SWNTs, all of them PbI2 @DWNT hybrid is viewed side-on (as indicated by
being filled with PbI2 . (b) Enlargement of the portion framed in (a). the arrow in Fig. 3.26e), polyhedral slabs are seen to ar-
(c) Fourier transform obtained from (b) showing the 110 distances range along the capillary, oriented at an angle of around
at 0.36 nm of a single PbI2 crystal. (d) Image of a single PbI2 -filled 45◦ with respect to the tubule axis. High-yield filling
SWNT. (e) Enlargement of the portion framed in (b). (f) Simulated of CNTs by the capillary method is generally difficult
HRTEM image, corresponding to (e). (g) Structural model corre- but fillings of more than 60% have been reported for
sponding to (f). (h) Structural model of a SWNT filled with a PbI2 different halides, with filling lengths of up to a couple
crystal as seen in cross section (from [3.254]) of hundreds of nm [3.260]. Results from the imaging
Introduction to Carbon Nanotubes 3.5 Carbon Nanotube-Based Nano-Objects 83
Part A 3.5
of the portion framed in (a). (c) Image reconstructed by
a second Fourier transform of the inset in (a) (= filtered - - -
210 101
image). (d) Structural model corresponding to (c). (e) and
(f) Atom and structural models respectively, corresponding -
100 000 101- f)
to (d) (from [3.254]) - - 2 nm
111 210
ature, and the outer tube diameter [3.269]. The smallest the type of nanotubes used (SWNTs, c-MWNTs, h-
MWNTs have been obtained in this way. MWNTs), the oxidizing agent used (air, O2 is an inert
By synthesizing endofullerenes [3.13], it has gas, CO2 , and so on), as well as the flow rate used
been possible to use this process to synthesize make it difficult to compare published results. It is gen-
more complex nanotube-based hybrid materials such erally agreed, however, that amorphous carbon burns
as La2 @C80 @SWNTs [3.270], Gd@C82 @SWNTs first, followed by SWNTs and then multiwall mater-
[3.271], and Erx Sc3−x N@C80 @SWNT [3.272], among ials (shells, MWNTs), even if TGA is often unable to
Part A 3.5
other examples. This suggests even more potential separate the different oxidation steps clearly. Air oxida-
applications for peapods, although they are still spec- tion (static or dynamic conditions) can however be used
ulative since the related properties are still being to prepare samples of very high purity – although the
investigated [3.273–275]. yield is generally low – as monitored by in situ Raman
The last example discussed here is the successful at- spectroscopy [3.278]. Aqueous solutions of oxidizing
tempt to produce peapods by a related method, using reagents are often used for nanotube oxidation. The
accelerated fullerene ions (instead of neutral gaseous main reagent is nitric acid, either concentrated or diluted
molecules) to force the fullerenes to enter the SWNT (around 3 mol/l in most cases), but oxidants such as
structure [3.276]. potassium dichromate (K2 Cr2 O7 ), hydrogen peroxide
(H2 O2 ) or potassium permanganate (KMnO4 ) are often
3.5.3 Functionalized Nanotubes used as well. HCl, like HF, does not damage nanotubes
because it is not oxidizing.
Noting the reactivity of carbon nanotubes (Sect. 3.4.6),
nanotube functionalization reactions can be divided into Functionalization
two main groups. One is based on the chemical oxida- of Oxidized Carbon Nanotubes
tion of the nanotubes (tips, structural defects) leading to The carboxylic groups located at the nanotube tips can
carboxylic, carbonyl and/or hydroxyl functions. These be coupled to different chemical groups. Oxidized nano-
functions are then used for additional reactions, to at- tubes are usually reacted with thionyl chloride (SOCl2 )
tach oligomeric or polymeric functional entities. The to generate the acyl chloride, even if a direct reac-
second group is based on direct addition to the graphiti- tion is theoretically possible with alcohols or amines,
clike surface of the nanotubes (without any intermediate for example. The reaction of SWNTs with octadecyl-
step). Examples of the latter reactions include oxidation amine (ODA) was reported by Chen et al. [3.279]
or fluorination (an important first step for further func- after reacting oxidized SWNTs with SOCl2 . The func-
tionalization with other organic groups). The properties tionalized SWNTs are soluble in chloroform (CHCl3 ),
and applications of functionalized nanotubes have been dichloromethane (CH2 Cl2 ), aromatic solvents, and car-
reviewed in [3.277]. bon bisulfide (CS2 ). Many other reactions between
functionalized nanotubes (after reaction with SOCl2 )
Oxidation of Carbon Nanotubes and amines have been reported in the literature and
Carbon nanotubes are often oxidized and therefore will not be reviewed here. Noncovalent reactions be-
opened before chemical functionalization in order to tween the carboxylic groups of oxidized nanotubes and
increase their chemical reactivity (to create dangling octadecylammonium ions are possible [3.280], provid-
bonds). The chemical oxidation of nanotubes is mainly ing solubility in tetrahydrofuran (THF) and CH2 Cl2 .
performed using either wet chemistry or gaseous oxi- Functionalization by glucosamine using similar proce-
dants such as oxygen (typically air) or CO2 . Depending dures [3.281] produced water soluble SWNTs, which
on the synthesis used, the oxidation resistance of nano- is of special interest when considering biological appli-
tubes can vary. When oxidation is achieved using a gas cations of functionalized nanotubes. Functionalization
phase, thermogravimetric analysis (TGA) is of great with lipophilic and hydrophilic dendra (with long alkyl
use for determining at which temperature the treatment chains and oligomeric poly(ethyleneglycol) groups)
should be applied. It is important to note that TGA ac- has been achieved via amination and esterification
curacy increases as the heating rate diminishes, while reactions [3.282], leading to solubility of the func-
the literature often provides TGA analyses obtained tionalized nanotubes in hexane, chloroform, and water.
in unoptimized conditions, leading to overestimated It is interesting to note that, in the latter case, the
oxidation temperatures. Differences in the presence functional groups could be removed simply by modi-
of catalyst remnants (metals or, more rarely, oxides), fying the pH of the solution (base- and acid-catalyzed
Introduction to Carbon Nanotubes 3.6 Applications of Carbon Nanotubes 85
hydrolysis reaction conditions, [3.283]). One last ex- nanotube wall [3.292]. Silanization reactions are an-
ample is the possible interconnection of nanotubes other way to functionalize nanotubes, although only
via chemical functionalization. This has been recently tested with MWNTs. Velasco-Santos et al. [3.293] have
achieved by Chiu et al. [3.284] using the acyl chlo- reacted oxidized MWNTs with an organosilane (RSiR3 ,
ride method and a bifunctionalized amine to link where R is an organo functional group attached to sil-
the nanotubes through the formation of amide bonds. icon) and obtained nanotubes with organo functional
Finally, it has been discovered that imidazolium-ion- groups attached via silanol groups.
Part A 3.6
functionalized carbon nanotubes are highly dispersible The noncovalent sidewall functionalization of nano-
in ionic liquids of analogous chemical structure and tubes is important because the covalent bonds are
that mixtures of functionalized CNT and ionic liquids associated with changes from sp2 hybridization to
can form gels upon sonication [3.285] or waxes [3.286] sp3 carbon hybridization, which corresponds to loss
that could find applications as soft composite ma- of the graphitelike character. The physical properties
terials for electrochemistry (sensors, capacitors, or of functionalized nanotubes, specifically SWNTs, can
actuators). therefore be modified. One way to achieve the non-
covalent functionalization of nanotubes is to wrap the
Sidewall Functionalization nanotubes in a polymer [3.294], which permits sol-
of Carbon Nanotubes ubilization (enhancing processing possibilities) while
Covalent functionalization of nanotube walls is possi- preserving the physical properties of the nanotubes. One
ble through fluorination reactions. It was first reported reason to functionalize SWNTs is to make them solu-
by Mickelson et al. [3.287], based on F2 gas (the ble in regular solvents. A promising method to do this
nanotubes can then be defluorinated, if required, was found by Pénicaud et al., who made water-soluble
with anhydrous hydrazine). As recently reviewed by by adding charges to SWNTs via the transient and re-
Khabashesku et al. [3.288], it is then possible to versible formation of a nanotube salt [3.295].
use these fluorinated nanotubes to carry out subse- Finally, it is worth bearing in mind that none of
quent derivatization reactions. Thus, sidewall-alkylated these chemical reactions are specific to nanotubes and
nanotubes can be prepared by nucleophilic substitu- so they can affect most of the carbonaceous impurities
tion (Grignard synthesis or reaction with alkyllithium present in the raw materials as well, making it difficult
precursors [3.289]). These alkyl sidewall groups can to characterize the functionalized samples. The exper-
be removed by air oxidation. Electrochemical addition iments must therefore be performed with very pure
of aryl radicals (from the reduction of aryl diazo- carbon nanotube samples, which is unfortunately not al-
nium salts) to nanotubes has also been reported by ways the case for the results reported in the literature.
Bahr et al. [3.290]. Functionalizations of the external On the other hand, purifying the nanotubes to start with
wall of the nanotube by cycloaddition of nitrenes, ad- may also bias the functionalization experiments, since
dition of nuclephilic carbenes or addition of radicals purification involves chemical treatment. However a de-
have been described by Holzinger et al. [3.291]. Elec- mand for such products already exists, and purified then
trophilic addition of dichlorocarbene to SWNTs occurs fluorinated SWNTs can be bought for 900 $/g (Carbon
via a reaction with the deactivated double bonds in the Nanotechnologies Inc., 2005).
Instead, we will cover the most important applications, Field Emission-Based Devices
and divide them up according to whether they are cur- In a pioneering work by de Heer et al. [3.300], car-
rent (Sect. 3.6.1) – they are already on the market, the bon nanotubes were shown to be efficient field emitters
application is possible in the near future, or because pro- and this property is currently being used several ap-
totypes are currently being developed by profit-based plications, including flat panel displays for television
companies – or expected applications (Sect. 3.6.2). sets and computers (the first prototype of such a dis-
play was exhibited by Samsung in 1999), and devices
Part A 3.6
Part A 3.6
permit much lower electron doses than MWNTs, al- MWNTs
though they often provide a more coherent source (an
Conductive Extraction
useful feature for devices such as electron microscopes Nanotube column grid
or x-ray generators).
The market associated with this application is huge.
With major companies involved, such as Motorola, Insulator
500 nm
NEC, NKK, Samsung, Thales and Toshiba, the first flat
TV sets and computers using nanotube-based screens Glass plate
should enter the market in 2007 (Samsung data), once Fig. 3.29 (a) Principle of a field emitter-based screen. (b) Scanning
a problem with product lifetime (still only about half electron microscope image of a nanotube-based emitter system (top
that required) is fixed. On the other hand, compa- view). Round dots are MWNT tips seen through the holes corre-
nies such as Oxford Instruments and Medirad are now sponding to the extraction grid. c P. Legagneux (Thales Research
commercializing miniature x-ray generators for medic- & Technology, Orsay, France)
al applications that use nanotube-based cold cathodes
developed by Applied Nanotech Inc. Chemical Sensors
The electrical conductance of semiconductor SWNTs
was recently demonstrated to be highly sensitive to
a) G (A/V) changes in the chemical composition of the surround-
3.5 ×10–7 ing atmosphere at room temperature, due to charge
3 ×10–7 transfer between the nanotubes and the molecules from
2.5 ×10–7 the gases adsorbed onto SWNT surfaces. It has also
2 ×10–7 been shown that there is a linear dependence between
1.5 ×10–7 the concentration of the adsorbed gas and the change
1 ×10–7 in electrical properties, and that the adsorption is re-
0.5 ×10–7 versible. First tries involved NO2 or NH3 [3.304] and
0 20 ppm NO3
O2 [3.305]. SWNT-based chemical NO2 and NH3 sen-
0 120 240 360 480 600 sors are characterized by extremely short response
Time t (s)
b) G (A/V) times (Fig. 3.30), unlike conventional sensors [3.304,
2.5 ×10–6
306]. The electrical response has been measured by
exposing MWNT films to sub-ppm NO2 concentra-
2 ×10–6 tions (10–100 ppb in dry air) at different operating
1% NH3
1.5 ×10–6 temperatures ranging between 25 and 215 ◦ C [3.307].
For SWNTs, the sensor responses are linear for sim-
1 ×10–6 ilar concentrations, with detection limits of 44 ppb
0.5 ×10–6 for NO2 and 262 ppb for nitrotoluene [3.308]. High
sensitivity to water or ammonia vapor has been demon-
0 strated on a SWNT-SiO2 composite [3.309]. This
0 120 240 360 480 600 720 840 960 study indicated the presence of p-type SWNTs dis-
Time t (s)
persed among the predominantly metallic SWNTs,
Fig. 3.30a,b Demonstration of the ability of SWNTs to and that the chemisorption of gases on the surface
detect trace molecules in inert gases. (a) Increase in the of the semiconductor SWNTs is responsible for the
conductance of a single SWNT when 20 ppm of NO2 are sensing action. Determinations of CO2 and O2 concen-
added to an argon gas flow. (b) Same, but with 1% NH3 trations on a SWNT-SiO2 composite have also been
added to the argon gas flow (from [3.304]) reported [3.310]. By doping nanotubes with palladium
88 Part A Nanostructures, Micro-/Nanofabrication and Materials
nanoparticles, Kong et al. [3.311] have also shown Tropsch, polymerization and even oxidation reactions,
that the modified material can reveal the presence of hydrocarbon decomposition and use as fuel cell elec-
hydrogen at levels of up to 400 ppm, whereas the as- trocatalysts are among the most popular domains. The
grown material was totally ineffective. Miniaturized application of graphite nanofibers as direct catalysts for
gas ionization sensors, which work by fingerprinting oxidative dehydrogenation [3.320, 321] or methane de-
the ionization characteristics of distinct gases, have composition [3.322] has also been reported.
also been reported, with detection limits of 25 ppm for The morphology and size of the carbon nanotubes
Part A 3.6
Part A 3.6
Pt/MWNT Electrodeless plating with Oxygen reduction for fuel High electrocatalytic
electrodes [3.331] prefunctionalization of cell applications activity
MWNT
Rh/MWNT [3.329] Surface-mediated organo- Liquid phase hydroformyla- Higher activity of
metallic synthesis, prefunc- tion and hydrogenation Rh/MWNT compared to
tionalization of MWNT Rh/activated carbon
Ru-alkali/MWNT Liquid phase impregnation, Ammonia synthesis, gas Higher activity with MWNT
[3.332] no pretreatment of the tubes phase reaction than with graphite
Rh-phosphine/MWNT Liquid phase grafting Liquid phase Highly active and
[3.333] from [RhH(CO)(PPh3 )3 ] hydroformylation regioselective catalyst
Rh/MWNT (confined Liquid phase impregnation Conversion of CO and H2 The overall formation rate of
nanoparticles) [3.319] of oxidized MWNTs into ethanol ethanol inside the nanotubes
exceeds that on the outside
of the nanotubes by more
than one order of magnitude
nanotubes, processed by methods other than those Therefore, there is still a great need for a mater-
based on solid carbon as the source (Sect. 3.2.1). ial that can store hydrogen but is also light, compact,
Such an application also requires some surface re- relatively inexpensive, safe, easy to use, and reusable
activity, making the h-MWNT-type nanotubes, with without regeneration. Some recent articles and patents
poor nanotextures (Sect. 3.1.2), interesting candidates on the very high, reversible adsorption of hydrogen in
as starting material for preparing such catalyst supports. carbon nanotubes or platelet nanofibers have aroused
Catalysis-enhanced thermal cracking of gaseous carbon tremendous interest in the research community, stim-
Part A 3.6
precursors is therefore preferred, and pilot plants are ulating much experimental and theoretical work. Most
already being built by major chemical industrial com- of the early works done on hydrogen adsorption on car-
panies (such as Arkema in France). bon nanotubes have been reviewed in [3.334–340], from
the first report about the supposedly highly successful
3.6.2 Expected Applications storage of hydrogen in carbon layered nanostructures
Related to Adsorption at room temperature made by a group of Northeastern
University [3.192,341], to the multiple yet vain attempts
Adsorptions of various gases, liquids or metals onto to reproduce this result that followed. Actually, in spite
carbon nanotubes, and interactions between them, have of a worldwide research effort, any work published
attracted much attention recently. The applications since then claiming for a hydrogen storage in some
resulting from the adsorptive properties of carbon nanotextured carbon material with an efficiency better
nanotubes can be arbitrarily divided into two groups. than 1–2% at room temperature or close (and pressure
The first group is based on the consequences of below ≈ 300–500 bar may be regarded as suspicious.
molecular adsorption on the electronic properties of Modelling did not help, since it appeared that the
nanotubes; the main application of this is chemical calculations are closely constrained by the starting hy-
sensing (Sect. 3.6.1). The second group includes gas potheses. Actually, While considering the same (10,10)
storage, gas separation, the use of carbon nanotubes SWNT, calculations based on DFT predicted between
as adsorbants, and results from morphological inves- 14.3 and 1 wt % storage [3.342,343], calculations based
tigations of carbon nanotubes (surface areas, aspect on a geometrical model predicted 3.3% [3.334], and
ratios, and so forth). Among these latter potential ap- calculations based on a quantum mechanical molecular
plications, the possibility of storing gases – particularly dynamics model predicted 0.47% [3.344]. Therefore,
hydrogen – on carbon nanotubes has received most neither experimental results, obviously often biased by
attention. procedure problems, nor theoretical results are yet able
to demonstrate that an efficient storage of H2 is possi-
Gas Storage – Hydrogen ble for carbon nanotubes, whatever the type. However,
The development of a lightweight and safe system for a definitive statement of failure cannot yet be claimed.
hydrogen storage is necessary for the widespead use of Attempts might have failed so far because they were
highly efficient H2 -air fuel cells in transportation ve- considering by far too simplistic materials, i. e., plain
hicles. The US Department of Energy Hydrogen Plan nanotubes. Further efforts have to be made to enhance
has provided a commercially significant benchmark for the adequation of the materials to this specific purpose,
the amount of reversible hydrogen adsorption required. in particular:
This benchmark requires a system weight efficiency
(the ratio of H2 weight to system weight) of 6.5 wt % 1. By adjusting the surface properties, which can be
hydrogen, and a volumetric density of 63 kg H2 /m3 . modified by mechanical or chemical treatments, e.g.
The failure to produce a practical storage system for KOH [3.345]
hydrogen has prevented hydrogen from becoming one 2. By adjusting the texture of the material, such as the
of the most important transportation fuels. The ideal pore size [3.346] and possibly the curvature [3.347–
hydrogen storage system needs to be light, compact, 349]
relatively inexpensive, safe, easy to use, and reusable 3. By complexifying the materials, e.g., by considering
without the need for regeneration. While research and nanocomposites combining some host carbon ma-
development are continuing into such technologies as li- terials and catalyst nanoparticles so as to promote
quid hydrogen systems, compressed hydrogen systems, the dissociation of hydrogen molecules to hydro-
metal hydride systems, and superactivated carbon sys- gen atoms that can form bonds with the host [3.340,
tems, all have serious disadvantages. 350].
Introduction to Carbon Nanotubes 3.6 Applications of Carbon Nanotubes 91
In this regard, whether the best carbon material for A theoretical study has aimed to determine the ef-
H2 adsorption will still be nanotube-based is not ascer- fects of different factors such as tube diameter, density
tained. and type of the gas used on the flow of molecules in-
side nanotubes. An atomistic simulation with methane,
Gas Storage – Gases Other than Hydrogen ethane and ethylene [3.353] has shown that the molecu-
Encouraged by the potential applications related to hy- lar mobility decreases with decreasing tube for each
drogen adsorption, several research groups have tried of the three gases. Ethane and ethylene have smaller
Part A 3.6
to use carbon nanotubes as a means of stocking and mobilities due to the stronger interactions they seem
transporting other gases such as oxygen, nitrogen, noble to have with the nanotube walls. In another theoretical
gases (argon and xenon) and hydrocarbons (methane, study into the possibility of hydrocarbon mixture sep-
ethane, and ethylene). These studies have shown that aration on SWNT bundles, the authors conclude that
carbon nanotubes could become the world’s smallest carbon nanotubes can be used to separate methane/
gas cylinders, combining low weight, easy transporta- n-butane and methane/isobutene mixtures [3.354] with
bility and safe use with acceptable adsorbed quantities. an efficiency that increases as the average tube diam-
Nanotubes may also be used in medicine, where it eter decreases. Experimental work was also performed
would be extremely useful to physically confine special by the same group on the sorption of butane on
gases (133 Xe for instance) prior to injection. MWNTs [3.194]. It has been also reported that the Fick-
Kusnetzova et al. [3.351] conducted experiments ian diffusivities of CH4 /H2 mixtures in SWNT, like
with xenon and found that the storage capacities of their pure component counterparts, are extraordinar-
nanotubes can be enhanced by a tremendous amount ily large when compared with adsorbed gases in other
(a factor of 280, up to a molar ratio of NXe /NC = 0.045) nanoporous materials [3.355].
by opening the SWNT bundles via thermal activation Grand canonical Monte Carlo simulations of the
at 800 ◦ C. The gas can be adsorbed inside the nanotubes separation of hydrogen and carbon monoxide by ad-
and the rates of adsorption are also increased using this sorption on SWNTs have also been reported [3.356].
treatment. In most of the situations studied, SWNTs were found
The possibility of storing argon in carbon nano- to adsorb more CO than H2 , and excellent separation
tubes has been studied, with encouraging results, by could again probably be obtained by varying the SWNT
Gadd et al. [3.352]. Their experiments show that large average diameter.
amounts of argon can be trapped in catalytically grown
MWNTs (20–150 nm) by hot isostatic pressing (HIP- Adsorbents
ing) for 48 h at 650 ◦ C under an argon pressure of Carbon nanotubes were found to be able to adsorb some
1700 bar. Energy-dispersive x-ray spectroscopy was toxic gases such as dioxins [3.357], fluoride [3.358],
used to determine that the gas was located inside the lead [3.359] and alcohols [3.360] better than adsor-
tubes and not on the tube walls. Further studies de- bent materials in common use, such as activated carbon.
termined the argon pressure inside the tubes at room These pioneering works opened a new field of applica-
temperature. The authors estimated this to be around tions as cleaning filters for many industrial processes
600 bar, indicating that equilibrium pressure was at- with hazardous by-products. The adsorption of diox-
tained in the tubes during the HIP-ing and that MWNTs ins, which are very common and persistent carcinogenic
would be a convenient material for storing the gas. by-products of many industrial processes, is a good
example of the potential of nanotubes in this field.
Gas Separation Growing ecological awareness has resulted in the impo-
As SWNTs or MWNTs have regular geometries that sition of emission limits on dioxin-generating sources
can, to some extent, be controlled, they could be used to in many countries, but it is difficult to find mater-
develop precise separation tools. If the sorption mech- ials that can act as effective filters, even at extremely
anisms are known, it should be possible to control low concentrations. Long and Yang [3.357] found that
sorption of various gases through particular combina- nanotubes can attract and trap more dioxins than acti-
tions of temperature, pressure and nanotube morphol- vated carbons or other polyaromatic materials that are
ogy. Since the large-scale production of nanotubes is currently used as filters. This improvement is prob-
gradually progressing, and this should ultimately re- ably due to the stronger interaction forces that exist
sult in low costs, accurate separation methods based on between dioxin molecules and the curved surfaces of
carbon nanotubes are now being investigated. nanotubes compared to those for flat graphene sheets.
92 Part A Nanostructures, Micro-/Nanofabrication and Materials
MWNTs have also been used with success for the ad- of nanotubes and asbestos was investigated by Huczko
sorption of other pollutants such as volatile organic et al. [3.371] and they concluded that the tested sam-
compounds [3.361], reactive dyes [3.362], or natural ples were innocuous. However, a more recent work
organic matter in aqueous solutions [3.363]. MWNTs has shown that contact with nanotubes may lead to
show also better performances than granular activated dermal toxicity [3.372] or induce lung lesions character-
carbons for the adsorption of low molecular weight tox- ized by the presence of granulomas [3.373]. Pantarotto
ins [3.364]. et al. [3.374] reported the translocation of water-soluble
Part A 3.6
The capacity of Al2 O3 /MWNT to adsorb fluoride SWNT derivatives across cell membranes and have
from water has been reported to be 13.5 times that of shown that cell death can be induced by functionalised
activated carbon and four times that of Al2 O3 [3.358]. nanotubes (bioactive peptides), depending upon their
The same group has also reported a capacity of MWNTs concentration in the media. Recent results also indicate
to adsorb lead from water that is higher than that that nanotubes may lead to an inflammatory response
for activated carbon [3.359]. The possibility of using of the immune system by activating the complement
graphite nanofibers to purify water from alcohols has system [3.375].
also been explored [3.360]. MWNTs were found to MWNTs have been used by Mattson et al. [3.376]
be good adsorbents for the removal of dichloroben- as a substrate for neuronal growth. They have compared
zene from wastewaters over a wide range of pH. the activity of untreated MWNTs with that of MWNTs
Typically, the nanotubes adsorb 30 mg of the organic coated with a bioactive molecule (4-hydroxynonenal)
molecule per gram of MWNTs from a 20 mg/l solu- and observed that neurons elaborated multiple neurites
tion [3.365]. It has also been shown that SWNTs act on these latter functionalized nanotubes. This is an im-
as molecular sponges for molecules such as CCl4 ; the portant result that illustrates the feasibility of using
nanotubes were in contact with a support surface which nanotubes as a substrate for nerve cell growth.
also adsorbs molecules, although more weakly than Davis et al. [3.377] immobilized different proteins
the nanotubes [3.366]. Finally, oxidized carbon nano- (metallothionein, cytochrome c and c3 , β-lactamase
tubes have been successfully used for the adsorption of I) in MWNTs and checked whether these molecules
heavy metal ions such as Zn(II) [3.367], Cu(II) [3.368], were still catalytically active compared to the free
Pb(II) [3.369] or Th(IV) [3.370] from aqueous solu- ones. They have shown that confining a protein
tions. While an apolar surface might be more adapted within a nanotube provides some protection for the
for the adsorption of aromatic organic species, an oxida- external environment. Protein immobilization via non-
tion of the CNTs that provides a polar and hydrophilic covalent sidewall functionalization was proposed by
surface is highly desirable for the adsorption of heavy Chen et al. [3.378] using a bifunctional molecule
metal ions. These experimental results suggest that (1-pyrenebutanoic acid, succinimidyl ester). This mol-
carbon nanotubes may be promising adsorbents for re- ecule is tied to the nanotube wall by the pyrenyl
moving polluting agents from water. group, and amine groups or biological molecules can
react with the ester function to form amide bonds.
Biosensors This method was also used to immobilize ferritin
Attaching molecules of biological interest to carbon and streptavidin onto SWNTs. Its main advantages
nanotubes is an excellent way to produce nanometer- are that it does not modify the SWNT wall and
sized biosensors. The electrical conductivities of these that it does not perturb the sp2 structure, so the
functionalized nanotubes would depend on the inter- physical properties of the nanotubes are maintained.
action of the probe with the medium being studied, Shim et al. [3.379] have functionalized SWNTs with
which would be affected by chemical changes or inter- biotin and observed specific binding with strepta-
actions with the target species. The science of attaching vidin, suggesting biomolecular recognition possibili-
biomolecules to nanotubes is rather recent and was in- ties. Dwyer et al. [3.380] have functionalized SWNTs
spired by similar research in the fullerene area. Some by covalently coupling DNA strands to them using
results have already been patented, and so such systems EDC (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide
may become available in the near future. Using the in- hydrochloride) but did not test biomolecular recog-
ternal cavities of nanotubes to deliver drugs would be nition; other proteins such as bovine serum albumin
another amazing application, but little work has been (BSA) [3.381] have been attached to nanotubes us-
carried out so far to investigate the toxicity of nanotubes ing the same process (diimide-activated amidation with
in the human body. Comparison between the effects EDC) and most of the attached proteins remained
Introduction to Carbon Nanotubes 3.6 Applications of Carbon Nanotubes 93
bioactive. Instead of working with individual nano- material are not considered here. Filled nanotubes are
tubes (or more likely nanotube bundles in the case discussed in Sect. 3.5.2.
of SWNTs), Nguyen et al. [3.382] have function-
alized nanotubes arrayed with a nucleic acid, still Metal Matrix Composites
using EDC as the coupling agent, in order to realize Nanotube-metal matrix composites are still rarely stud-
biosensors based on protein-functionalized nanotubes. ied. Matrices include Al-, Cu-, Mg-, Ni-, Ni-P-, Ti-,
Azamian et al. [3.383] have immobilized a series of WC-Co- and Zr-based bulk metallic glasses. The ma-
Part A 3.6
biomolecules (cytochrome c, ferritin, and glucose ox- terials are generally prepared by standard powder
idase) on SWNTs, and they observed that the use of metallurgy techniques, but in this case the nanotube
EDC was not always necessary, indicating that the dispersion is not optimal. Other techniques such as
binding was predominantly noncovalent. In the case plasma spray forming [3.386], the so-called nanoscale-
of glucose oxidase, they tested the catalytic activity of dispersion method [3.387], the rapid solidification
functionalized nanotubes immobilized on a glassy car- technique [3.388] and CCVD [3.389], are being de-
bon electrode and observed a tenfold greater catalytic veloped. The spark plasma sintering (SPS) technique
response compared to that seen in the absence of modi- is sometimes used to densify the composites whilst
fied SWNTs. avoiding matrix-grain growth [3.390, 391]. The room-
Functionalization of nanotubes with biomolecules is temperature electrical resistivity of hot-pressed CCVD
still in its infancy, and their use as biosensors may lead MWNT-Al composites increases slightly upon increas-
to practical applications earlier than expected. For ex- ing the MWNT volume fraction [3.392]. The tensile
ample, functionalized nanotubes can be used as AFM strengths and elongations of unpurified arc discharge
tips (Sect. 3.6.1), allowing single-molecule measure- MWNT-Al composites are only slightly affected by an-
ments to be taken using chemical force microscopy nealing at 873 K in contrast to those of pure Al [3.393].
(CFM). Important improvements in the characteriza- The coefficient of thermal expansion (CTE) of 1 wt %
tion of biomolecules have even been achieved with MWNTs-Al composite fabricated by cold isostatic
unfunctionalized nanotube-based tips (see the review pressing and hot squeeze technique is 11% lower than
by [3.297]). Nanotube-based biosensors have now to that of pure Al or 2024Al matrix, showing some
been developed. They are based on either field ef- promises as low-CTE materials. Associated to a high
fect transistors [3.384] involving functionalized CNTs thermal conductivity, such materials would be inter-
(biomolecules) or on electrochemical detection [3.385]. esting for applications such as packaging and space
structures [3.394]. The Young’s modulus of nonpu-
3.6.3 Expected Applications rified arc discharge MWNTs-Ti composite is about
Related to Composite Systems 1.7 times that of pure Ti [3.395]. The formation
of TiC, probably resulting from a reaction between
Because of their exceptional morphological, electric- amorphous carbon and the matrix, was observed, but
al, thermal, and mechanical characteristics, carbon the MWNTs themselves were not damaged. An in-
nanotubes make particularly promising reinforcement crease in the Vickers hardness by a factor of 5.5 over
materials in composites with metals, ceramics or poly- that of pure Ti was associated with the suppression
mer matrices. Key issues to address include the good of coarsening of the Ti grains, TiC formation, and
dispersion of the nanotubes, the control of the nanotube/ the addition of MWNTs. Purified nanotube-WC-Co
matrix bonding, the densification of bulk composites nanocomposites exhibit better hardness-to-toughness
and thin films, and the possibility of aligning the relationships than pure nanocrystalline WC-Co [3.391].
nanotubes. In addition, the nanotube type (SWNT, Ni-plated MWNTs give better results than unplated
c-MWNT, h-MWNT, etc.) and origin (arc, laser, MWNTs in strength tests. Indeed, nanotube coating is
CCVD, etc.) are also important variables that control a promising way to improve the strength of bonding
the structural perfection, surface reactivity and aspect with the matrix [3.396]. Compressive testing of car-
ratio of the reinforcement. bon nanotube-reinforced Zr-based bulk metallic glass
The application of carbon nanotubes in this field is composites [3.397] shows that the composites display
expected to lead to major advances in composites. The a high fracture strength. In addition, the composites
following sections will give overviews of current work have strong ultrasonic attenuation characteristics and
on metal-, ceramic- and polymer-matrix composites excellent ability to absorb waves. This implies that such
containing nanotubes. Nanotubes coated with another composites may also be useful for shielding acous-
94 Part A Nanostructures, Micro-/Nanofabrication and Materials
tic sound or environmental noise. CCVD MWNTs-Cu of the nanotubes. The most common method is hot-
composites [3.398] also show a higher hardness and pressing (HP). Most of the works [3.406–413] report
a lower friction coefficient and wear loss. Fifty to sixty that increasing the nanotube content inhibits the densi-
percent deformation of the composites was observed. fication of the material. It has been shown for a series
Carbon nanotube-Cu composite electrodes have been of CNT-MgAl2 O4 composites [3.413] that, for a low
applied to the amperometric detection of carbohydrates, content (below 9 vol. %), CNTs favor the rearrange-
where they show an enhanced sensitivity compared ment of the grains, which is the first shrinkage step,
Part A 3.6
to detectors based on Cu or nanotubes alone [3.399]. probably owing to a lubricating role which facilitates
Hot-extruded nanotube-Mg nanocomposites showed the sliding at grain contacts or grain boundaries. By
a simultaneous increase in yield strength, ultimate ten- contrast, for higher contents, CNTs form a too rigid we-
sile strength and ductility, until a threshold of 1.3 wt % blike structure, therefore inhibiting the rearrangement
was reached [3.400]. The yield strength of SWNT-Fe process. In the second sintering step, at higher temper-
composites showed substantial enhancement relative to atures, CNTs inhibit the shrinkage, all the more when
that of similarly treated pure iron materials [3.389]. their content is increased above 5.0 vol. % only, leading
The work hardening coefficient and the Vickers hard- to decreasing densifications. Thus, composites in which
ness coefficient also significantly increased in these the nanotubes are very homogeneously dispersed may
composites. Composite films and coatings deposited be more difficult to densify. The spark-plasma sinter-
by electroless or electrodeposition techniques on var- ing (SPS) technique has been reported as an efficient
ious substrates have also been studied. The addition method to achieve the total densification of CNT-oxide
of up to 15 vol. % purified SWNTs to nanocrystalline composites without damaging the CNT [3.414–417].
Al films reduces the coefficient of thermal expansion Full densification can be reached with SPS at com-
by as much as 65% and the resulting material could paratively lower temperatures with substantial shorter
be a promising electronic packaging material [3.401]. holding time. However, the successful densification by
Ni-carbon nanotube coatings deposited on carbon steel SPS at a lower temperature than for HP supposes that
plate by electroless deposition show significantly in- matrix grains are non agglomerated and with size in the
creased resistance to corrosion [3.402] and higher range few tens of nanometers.
Vickers microhardness, higher wear resistance, and The influence of the nanotube dispersion onto me-
lower friction coefficient than SiC-reinforced compos- chanical properties, in particular on toughness, has been
ite deposits [3.403]. Ni-P-SWNT coatings prepared by controversial. Indeed, strong increases in toughness de-
electroless plating show not only higher wear resistance rived from the measure of Vickers indentation cracks
but also a lower friction coefficient and a higher corro- have been reported [3.33], but they were shown to be
sion resistance compared to Ni-P coatings [3.404]. probably widely overestimated because such materials
are very resistant to contact damage [3.418, 419]. Xia
Ceramic Matrix Composites et al. [3.420] reported microstructural investigations on
Many different ceramic matrices have been studied over MWNTs well-aligned in the pores of an alumina mem-
the years, although oxides (in particular alumina), are brane. Different possible reinforcement mechanisms
still the most studied [3.405]. There are three main induced by the MWNTs have been evidenced, such
methods for the preparation of CNT-ceramic nanocom- as crack deflection, crack bridging, MWNT pulling-
posite powders. One is mechanical milling. It usually out, and MWNT collapsing in shear bands. Indeed,
involves long times that could damage the nanotubes. although so far neither SENB nor SEVNB result have
Wet-milling is preferred but often requires the addi- evidenced that nanotubes can significantly reinforce
tion of organic additives to stabilize both the nanotubes alumina ceramics, this could be obtained with ceramic-
and the ceramic powder. This also true for a second matrix composites in which the nanotubes would have
method, i. e., the in-situ synthesis of the matrix on been properly organized. Enhanced wear resistance of
preformed nanotubes. It can lead to a good adhesion be- composites has been reported [3.421–423]. The micro-
tween the nanotubes and the ceramic, but can be rather hardness is found to either increase or decrease, and
complex to implement. A third method is the in-situ this depends greatly on the powder preparation route.
synthesis of the nanotubes within the ceramic pow- As noted in [3.419], processing-induced changes in
der using procedures closely related to those described the matrix may have greater effects on the mechani-
in Sect. 3.2.2. The densification of the nanocomposite cal properties than the actual presence of nanotubes.
powders is made difficult by the detrimental influence Regarding the thermal properties, nanotube-ceramic
Introduction to Carbon Nanotubes 3.6 Applications of Carbon Nanotubes 95
composites often show a lower thermal conductiv- walls of the MWNTs are stressed in compression,
ity than the corresponding ceramics, probably caused whereas only the outer walls are stressed in ten-
by too high thermal contact resistances at nanotube- sion because all of the inner tubes are sliding within
nanotube and nanotube-ceramic grain junctions [3.424, the outer tube. Mechanical tests performed on 5 wt %
425]. By contrast, nanotubes greatly increase the SWNT-epoxy composites [3.437] showed that SWNT
electrical conductivity of insulating ceramic nanocom- bundles were pulled out of the matrix during the
posites [3.408, 411, 426–428], with a low percolation deformation of the material. The influence of the in-
Part A 3.6
threshold (less than 1 vol. %) due to their very high as- terfacial nanotube/matrix interaction was demonstrated
pect ratio [3.427]. The electrical conductivity can be by Gong et al. [3.438]. It was also reported that coat-
tailored within several orders of magnitude directly by ing regular carbon fiber with MWNTs prior to their
the CNTs quantity and is well fitted by the scaling law dispersion into an epoxy matrix improves the interfa-
of the percolation theory with the exponent close to the cial load transfer, possibly via local stiffening of the
theoretical value characteristic of a three-dimensional matrix near the interface [3.439]. DWNTs-epoxy com-
network [3.427]. An anisotropic conductivity is ob- posites prepared by a standard calendaring technique
tained when the nanotubes are aligned within the were shown to possess higher strength, Young’s mod-
composite [3.429]. Zhan et al. [3.430] reported an ulus and strain to failure at a nanotube content of
increase of the thermoelectric power with increasing only 0.1 wt % [3.440]. A significantly improved frac-
temperature for nanotube-zirconia composites. ture toughness was also observed. The influence of
the different types of nanotubes (SWNTs, DWNTs and
Polymer Matrix Composites MWNTs) on the mechanical properties of epoxy-matrix
Nanotube-polymer composites, first reported by Ajayan composites is discussed in [3.441]. The stiffness and
et al. [3.431], are now being intensively studied; es- damping properties of SWNT- and MWNT-epoxy com-
pecially epoxy- and polymethylmethacrylate (PMMA)- posites were investigated for use in structural vibration
matrix composites. A review of the mechanical prop- applications [3.442]. It was shown that enhancement in
erties can be found in [3.432]. In terms of mechanical damping ratio is more dominant than enhancement in
characteristics, the three key issues that affect the per- stiffness, MWNTs making a better reinforcement than
formance of a fiber-polymer composite are the strength SWNTs. Indeed, up to 700% increase in damping ra-
and toughness of the fibrous reinforcement, its orien- tio is observed for MWNT-epoxy beam as compared
tation, and good interfacial bonding, which is crucial to the plain epoxy beam. Industrial epoxy loaded with
to load transfer [3.433]. The ability of the polymer 1 wt % unpurified CCVD-prepared SWNTs showed an
to form large-diameter helices around individual nano- increase in thermal conductivity of 70 and 125% at 40 K
tubes favors the formation of a strong bond with the and at room temperature, respectively [3.443]. Also,
matrix [3.433]. Isolated SWNTs may be more desirable the Vickers hardness rose by a factor of 3.5 with the
than MWNTs or bundles for dispersion in a matrix be- SWNT loading up at 2 wt %. An increase in the amount
cause of the weak frictional interactions between layers of MWNTs led to an increase of the glass transition
of MWNTs and between SWNTs in bundles [3.433]. temperature of MWNT-epoxy-composites. The effect
The main mechanisms of load transfer are microme- is stronger when using samples containing functional-
chanical interlocking, chemical bonding and van der ized MWNTs [3.444]. Pecastaings et al. [3.445] have
Waals bonding between the nanotubes and the ma- investigated the role of interfacial effects in carbon
trix. A high interfacial shear stress between the fiber nanotube-epoxy nanocomposite behavior.
and the matrix will transfer the applied load to the As for ceramic matrix composites, the electrical
fiber over a short distance [3.434]. SWNTs longer characteristics of SWNT- and MWNT-epoxy compos-
than 10–100 μm would be needed for significant load- ites are described by the percolation theory. Very low
bearing ability in the case of nonbonded SWNT-matrix percolation thresholds (much below 1 wt %) are of-
interactions, whereas the critical length for SWNTs ten reported [3.446–448]. Thermogravimetric analysis
cross-linked to the matrix is only 1 μm [3.435]. De- shows that, compared to pure PMMA, the thermal
fects are likely to limit the working length of SWNTs, degradation of PMMA films occurs at a slightly
however [3.436]. higher temperature when 26 wt % of MWNTs are
The load transfer to MWNTs dispersed in an added [3.449]. Improving the wetting between the
epoxy resin was much higher in compression than MWNTs and the PMMA by coating the MWNTs
in tension [3.434]. It was proposed that all of the with poly(vinylidene fluoride) prior to melt-blending
96 Part A Nanostructures, Micro-/Nanofabrication and Materials
with PMMA resulted in an increased storage modu- A great deal of work has also been devoted
lus [3.450]. The impact strength in aligned SWNT- to the applications of nanotube-polymer compos-
PMMA composites increased significantly with only ites as materials for molecular optoelectronics, using
0.1 wt % of SWNTs, possibly because of weak inter- primarily poly(m-phenylenevinylene-co-2,5-dioctoxy-
facial adhesion and/or of the high flexibility of the p-phenylenevinylene) (PmPV) as the matrix. This
SWNTs and/or the pullout and sliding effects of indi- conjugated polymer tends to coil, forming a helical
vidual SWNTs within bundles [3.451]. The transport structure. The electrical conductivity of the composite
Part A 3.6
properties of arc discharge SWNT-PMMA composite films (4–36 wt % MWNTs) is increased by eight or-
films (10 μm thick) were studied in great detail [3.452, ders of magnitude compared to that of PmPV [3.483].
453]. The electrical conductivity increases by nine or- Using the MWNT-PmPV composites as the electron
ders of magnitude from 0.1 to 8 wt % SWNTs. The transport layer in light-emitting diodes results in a sig-
room-temperature conductivity is again well described nificant increase in brightness [3.484]. The SWNTs act
by the standard percolation theory, confirming the good as a hole-trapping material that blocks the holes in
dispersion of the SWNTs in the matrix. The rheolog- the composites; this is probably induced through long-
ical threshold of SWNT-PMMA composites is about range interactions within the matrix [3.485]. Similar
0.12 wt %, smaller than the percolation threshold of investigations were carried out on arc discharge SWNT-
electrical conductivity, about 0.39 wt % [3.454]. This is polyethylene dioxythiophene (PEDOT) composite lay-
understood in terms of the smaller nanotube–nanotube ers [3.486] and MWNT-polyphenylenevinylene com-
distance required for electrical conductivity compared posites [3.487].
to that required to impede polymer mobility. Further- To conclude, two critical issues must be consid-
more, decreased SWNT alignment, improved SWNT ered when using nanotubes as components for ad-
dispersion and/or longer polymer chains increase the vanced composites. One is to choose between SWNTs,
elastic response of the nanocomposite. The effects of DWNTs, and MWNTs. The former seem more bene-
small quantities of SWNTs (up to 1 wt %) in PMMA ficial to mechanical strengthening, provided that they
on its flammability properties were studied [3.455]. The are isolated or arranged into cohesive yarns so that the
formation of a continuous SWNTs network layer cov- load can be conveniently transferred from one SWNT to
ering the entire surface without any cracks is critical another. Unfortunately, despite many advances [3.456–
for obtaining the lowest mass-loss rate of the nanocom- 461], this is still a technical challenge. The other issue
posites. One of the most interesting development of is to tailor the nanotube/matrix interface with respect to
nanotube-polymer composites is their use for the pro- the matrix. In this case, DWNTs and MWNTs may be
duction of spun fibers, films and textiles with extraordi- more useful than SWNTs.
nary mechanical and electrical properties [3.456–462].
Polymer composites with other matrices include Multifunctional Materials
CCVD-prepared MWNT-polyvinyl alcohol [3.463], One of the major benefits expected from incorporating
arc-prepared MWNT-polyhydroxyaminoether [3.464], carbon nanotubes into other solid or liquid materials is
arc-prepared MWNT-polyurethane acrylate [3.465, that they endow the material with some electrical con-
466], SWNT-polyurethane acrylate [3.467], SWNT- ductivity while leaving other properties or behaviors
polycarbonate [3.468], MWNT-polyaniline [3.469], unaffected. As already mentioned in the previous sec-
MWNT-polystyrene [3.470], CCVD double-walled tion, the percolation threshold is reached at very low
nanotubes-polystyrene-polymethylacrylate [3.471], nanotube loadings. Tailoring the electrical conductiv-
MWNT-polypropylene [3.472, 473], SWNT-polyethy- ity of a bulk material is then achieved by adjusting
lene [3.474–476], SWNT-poly(vinyl acetate) the nanotube volume fraction in the formerly insulat-
[3.475,476], CCVD-prepared MWNT-polyacrylonitrile. ing material while making sure that this fraction is
[3.477], SWNT-polyacrylonitrile [3.478], MWNT-oxo- not too large.As demonstrated by Maruyama [3.488],
titanium phthalocyanine [3.479], arc-prepared MWNT- there are three areas of interest regarding the electrical
poly(3-octylthiophene) [3.480], SWNT-poly(3-octyl- conductivity:
thiophene) [3.481] and CCVD MWNT-poly(3-hexyl-
thiophene) [3.482]. These works deal mainly with films 1. Electrostatic discharge (for example, preventing fire
100–200 μm thick, and aim to study the glass transition or explosion hazards in combustible environments
of the polymer, its mechanical and electrical character- or perturbations in electronics, which requires an
istics, as well as the photoconductivity. electrical resistivity of less than 1012 Ω cm)
Introduction to Carbon Nanotubes 3.6 Applications of Carbon Nanotubes 97
Dimensional
temperature
Conduction/
conductiona
dissipationc
dissipation
reductione
Part A 3.6
Through-
Strength/
thickness
stabilityd
shielding
strength
strength
Serviceb
stiffness
Specific
Surface
Static
CTE
EMI
Fiber fraction Applications
2. Electrostatic painting (which requires the material of minimal weight and volume. All of these properties
to be painted to have enough electrical conductivity should be possible with a single nanotube-containing
– an electrical resistivity below 106 Ω cm – to pre- composite material instead of complex multimateri-
vent the charged paint droplets from being repelled) als combining layers of polymers, aluminum, copper,
3. Electromagnetic interference shielding (which is and so on. Table 3.5 provides an overview of various
achieved for an electrical resistivity of less than fields in which nanotube-based multifunctional mater-
10 Ω cm. ials should find application.
a semiconductor SWNT across two electrodes (source nected gold coatings onto a pulled glass micropipette
and drain) deposited on an insulating substrate that (Fig. 3.32), and then attaching two MWNTs (or two
serves as a gate electrode [3.492, 493]. The associa- SWNT-bundles) ≈ 20–50 nm in diameter to each of the
tion of two such SWNT-based FETs makes a voltage gold electrodes. Applying a voltage (0–8.5 V) between
inverter [3.494]. the two electrodes then makes the tube tips open and
All of the latter developments are fascinating and close reversibly in a controlled manner.
provide promising outlets for nanotube-based elec- A similar experiment, again rather simple, was pro-
Part A 3.6
tronics. However, progress is obviously needed before posed by Baughman et al. the same year (1999) [3.498].
SWNT-based integrated circuits can be constructed on This consisted of mounting two SWNT-based pa-
a routine basis. A key issue is the need to be able per strips (bucky-paper) on both sides of insulating
to selectively prepare either metallic or semiconductor double-sided tape. The two bucky-paper strips had been
nanotubes. Although a method of selectively destroy- previously loaded with Na+ and Cl− , respectively.
ing metallic SWNTs in bundles of undifferentiated When 1 V was applied between the two paper strips,
SWNTs [3.496] has been proposed, the method is not both of them expanded, but the strip loaded with Na+
scalable and selective synthesis would be preferable. expanded a bit more, forcing the whole system to bend.
Also, defect-free nanotubes are required. Generally Though performed in a liquid environment, this behav-
speaking, this relates to another major challenge, which ior has inspired the authors to predict a future use for
is to be able to fabricate integrated circuits includ- their system in artificial muscles.
ing nanometer-size components (that only sophisticated Another example of amazing nanotools is the nan-
imaging methods such as AFM are able to visualize) on othermometer proposed by Gao and Bando [3.499].
an industrial scale. An overview of the issues related to A single MWNT was used, which was partially filled
the integration of carbon nanotubes into microelectron- with liquid gallium. Temperature variations in the range
ics systems has been written by Graham et al. [3.497]. 50–500 ◦ C cause the gallium to reversibly move up and
down within the nanotube cavity at reproducible levels
Nanotools, Nanodevices and Nanosystems with respect to the temperature values applied.
Due to the ability of graphene to expand slightly when Of course, nanotools such as nanotweezers or
electrically charged, nanotubes have been found to act nanothermometers are hardly commercial enough to
as actuators. Kim and Lieber [3.495] demonstrated this justify industrial investment. But such experiments are
by designing nanotweezers, which are able to grab, more than just amazing laboratory curiosities. They
manipulate and release nano-objects (the nanobead demonstrate the ability of carbon nanotubes to pro-
that was handled for the demonstration was actually vide building blocks for future nanodevices, including
closer to a micrometer in size than a nanometer), as nanomechanical systems.
well as to measure their electrical properties. This was
made possible by simply depositing two nonintercon- Supercapacitors
Supercapacitors consist of two electrodes immersed
in an electrolyte (such as 6 M KOH), separated by
an insulating ion-permeable membrane. Charging the
capacitors is achieved by applying a potential be-
Deposit independent tween the two electrodes, which makes the cations
metal coatings and the anions move toward the oppositely charged
electrode. Suitable electrodes should exhibit high elec-
trical conductivities and high surface areas, since the
Attach
carbonnanotubes capacitance is proportional to these parameters. Actu-
+ ally, the surface area should consist of an appropriate
– combination of mesopores (to allow the electrolyte
components to circulate well, which is related to the
Fig. 3.32 Sketch explaining how the first nanotweezers charging speed) and micropores (whose walls pro-
were designed. The process involves modifying a glass vide the attractive surfaces and fixation sites for the
micropipette (dark cone, top). Two Au coatings (in gray, ions). Based on early work by Niu et al. [3.500],
middle) are deposited so that they are not in contact. Then such a combination was found to be provided by
a voltage is applied to the electrodes (from [3.495]) the specific architecture offered by packed and entan-
Introduction to Carbon Nanotubes 3.7 Toxicity and Environmental Impact of Carbon Nanotubes 99
gled h-MWNTs with poor nanotextures (Sect. 3.1.2). Capacitors including nanotubes have already shown
However, activation pretreatments were necessary. For capacitances as high as 180–200 F/g, equivalent to
instance, a capacitor made from nanotubes with a sur- those obtained with electrodes built from regular car-
face area of 220 m2 /g exhibited a capacitance of 20 F/g, bon materials, but they have the advantage of faster
which increased to 100 F/g after an activation treat- charging [3.171]. Current work in this area will cer-
ment was applied to the nanotubes so that their surface tainly lead to further optimization of both the nanotube
area increased to 880 m2 /g [3.171]. Alternatively, again material architecture and the nanotube-supported con-
Part A 3.7
due to their remarkable architectures derived from ductive polymers, meaning that the outlook for the
their huge aspect ratios, nanotubes can also be used commercial use of nanotubes as components for su-
as supports for conductive polymer coatings, such percapacitors is positive, and this is ignoring the
as polypyrrole or polyaniline [3.501], or additives to potential application of second-generation nanotubes
regular carbon electrodes [3.502], which make the (such as nanotube-based nano-objects) in this field.
material more open, allowing easier circulation and A first attempt to use hybrid nanotubes (Sect. 3.5.2)
penetration of ions. Supercapacitors built from such has already resulted in improved properties with
composites can survive more than 2000 charging cycles, respect to genuine (undoped) nanotube-based sys-
with current densities as high as 350 mA/g [3.503]. tems [3.504].
then digest) cellular debris and pathogens as well as to very quickly by adsorption of proteins (complement
stimulate lymphocytes and other immune cells to re- system [3.506], surfactants [3.507], etc.); this adsorp-
spond to the pathogens). Taking into account the small tion can be very specific [3.506, 507], and is likely to be
size of macrophages as compared to that of agglom- dynamic and controlled by the affinity of the molecules
erates, bundles or even individual CNTs, macrophages for the surface of the CNTs (pristine or functionalised).
usually do not manage to get rid of the CNTs by phago- It is thus obvious that the surface chemistry of the CNTs
cytose. However, they try to do so and thus release will play a very important role.
Part A 3.8
reactive oxygen species (ROS), enzymes, cytokines (in- The potential use of CNTs in commercial products
terferons (IFN)), etc. and agglomerate around them to (Sect. 3.5) begs the question of their fate at the end of
isolate them from the body. Proteins present in the blood their lifecycle. If the impact of CNTs on human health
and most biological fluids (complement system – in- is under investigation for already a few years now, it
nate immunity) will play a similar role by labelling the is noteworthy that the environmental impact has almost
CNTs (opsonisation) and possibly generating some in- not been taken into account. Only a few publications
flammatory reactions. The complement system strongly (less than 15) are available to date and the concentra-
interacts with the lymphocytes. These natural phenom- tion at which ecotoxic effects are evidenced is usually
ena have deleterious consequences on the surrounding much higher than what could be reasonably found in the
tissues: inflammation in a first instance, formation of environment (unless very local and specific conditions).
granuloma (commonly observed in the lungs after expo- Due to the potentially very high specific surface area of
sure to CNTs). Each target organ has its own phagocyte CNTs, they could act as vectors for pollutants adsorbed
cells (Kupffer cells in the liver, Langerhans cells in the on their surface (PAH, polycyclic aromatic hydrocar-
skin, etc.). bons for example), even if themselves do not show any
Toxicity can be assessed both by in vitro and in vivo sign of toxicity.
experiments. In the case of in vitro assays, cell cultures There is currently no consensus about the toxicity
(usually immortalised cancer cells, but also primary cul- of CNTs [3.505], although more than 500 papers have
tures or even stem cells) are exposed to suspensions of now been published already on this topic within the
CNTs. In the case of in vivo assays, the animals (mice, last 5 years. Despite the worldwide effort devoted to
rats, worms, amphibians, fishes, etc.) are exposed either this field of research, the huge variety of CNT types,
to aerosols (inhalation) or mainly again to suspensions shapes, composition, etc. will make very difficult to an-
of CNTs which will be administrated according to dif- swer this simple question: are CNT toxic? The principle
ferent protocols depending on the study (intra-tracheal of precaution should not stop all research in this area but
instillation, injection, contact with the skin, etc.). Ex- only draw the attention to a more responsible attitude
trapolating the toxicity results from animals (or even for people working on their synthesis or manipulating
worse, from cells) to humans is very delicate but the them, and industrials willing to include them in con-
data are however very useful for the sake of comparison sumer products. Gloves should be worn at any time as
in a given system and with given experimental condi- well as an adapted (FFP3 type) disposable dust mask.
tions. As soon as CNTs are in contact with a biological Wearing a lab coat is recommended to limit contamina-
fluid, their surface chemistry is likely to be modified tion of clothes. CNT wastes should be burnt.
References
3.1 M. Monthioux, V.L. Kuznetsov: Who should be given 3.19 P.J.F. Harris: Carbon Nanotubes and Related Struc-
the credit for the discovery of carbon nanotubes?, tures (Cambridge Univ. Press, Cambridge 1999)
Carbon 44, 1621–1623 (2006) 3.20 N.M. Rodriguez, A. Chambers, R.T. Baker: Catalytic
3.2 L.V. Radushkevich, V.M. Lukyanovich: O strukture engineering of carbon nanostructures, Langmuir 11,
ugleroda, obrazujucegosja pri termiceskom ra- 3862–3866 (1995)
Part A 3
zlozenii okisi ugleroda na zeleznom kontakte, Zurn. 3.21 M. Monthioux, L. Noé, L. Dussault, J.-C. Dupin, N. La-
Fis. Chim. 26, 88–95 (1952), in Russian torre, T. Ubieto, E. Romeo, C. Royo, A. Monzón,
3.3 T.V. Hughes, C.R. Chambers: Manufacture of Carbon C. Guimon: Texturising and structurising mech-
Filaments, US Patent 405480 (1889) anisms of carbon nanofilament during growth,
3.4 P. Schützenberger, L. Schützenberger: Sur quelques J. Mater. Chem. 17, 4611–4618 (2007)
faits relatifs à l’histoire du carbone, C. R. Acad. Sci. 3.22 J. Vera-Agullo, H. Varela-Rizo, J.A. Conesa, C. Al-
Paris 111, 774–778 (1890), in French mansa, C. Merino, I. Martin-Gullon: Evidence for
3.5 C. Pélabon, H. Pélabon: Sur une variété de car- growth mechanism and helix-spiral cone structure
bone filamenteux, C. R. Acad. Sci. Paris 137, 706–708 of stacked-cup carbon nanofibers, Carbon 45, 2751–
(1903), in French 2758 (2007)
3.6 R.T.K. Baker, P.S. Harris: The formation of filamen- 3.23 H.W. Kroto, J.R. Heath, S.C. O’Brien, R.F. Curl,
tous carbon. In: Chemistry and Physics of Carbon, R.E. Smalley: C60 Buckminsterfullerene, Nature 318,
Vol. 14, ed. by P.L. Walker Jr., P.A. Thrower (Dekker, 162–163 (1985)
New York 1978) pp. 83–165 3.24 W. Krätschmer, L.D. Lamb, K. Fostiropoulos,
3.7 S. Iijima: Helical microtubules of graphite carbon, D.R. Huffman: Solid C60 : A new form of carbon,
Nature 354, 56–58 (1991) Nature 347, 354–358 (1990)
3.8 S. Iijima, T. Ichihashi: Single-shell carbon nano- 3.25 L. Fulchieri, Y. Schwob, F. Fabry, G. Flamant,
tubes of 1 nm diameter, Nature 363, 603–605 (1993) L.F.P. Chibante, D. Laplaze: Fullerene production in
3.9 D.S. Bethune, C.H. Kiang, M.S. de Vries, G. Gorman, a 3-phase AC plasma process, Carbon 38, 797–803
R. Savoy, J. Vazquez, R. Bayers: Cobalt-catalysed (2000)
growth of carbon nanotubes with single-atomic- 3.26 K. Saidane, M. Razafinimanana, H. Lange, A. Huczko,
layer walls, Nature 363, 605–607 (1993) M. Baltas, A. Gleizes, J.L. Meunier: Fullerene syn-
3.10 J. Tersoff, R.S. Ruoff: Structural properties of thesis in the graphite electrode arc process: local
a carbon-nanotube crystal, Phys. Rev. Lett. 73, 676– plasma characteristics and correlation with yield,
679 (1994) J. Phys. D Appl. Phys. 37, 232–239 (2004)
3.11 N. Wang, Z.K. Tang, G.D. Li, J.S. Chen: Single-walled 3.27 T. Guo, P. Nikolaev, A.G. Rinzler, D. Tomanek,
4 Å carbon nanotube arrays, Nature 408, 50–51 D.T. Colbert, R.E. Smalley: Self-assembly of tubular
(2000) fullerenes, J. Phys. Chem. 99, 10694–10697 (1995)
3.12 N. Hamada, S.I. Sawada, A. Oshiyama: New one- 3.28 T. Guo, P. Nikolaev, A. Thess, D.T. Colbert, R.E. Smal-
dimensional conductors, graphite microtubules, ley: Catalytic growth of single-walled nanotubes by
Phys. Rev. Lett. 68, 1579–1581 (1992) laser vaporisation, Chem. Phys. Lett. 243, 49–54
3.13 M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund: Sci- (1995)
ence of Fullerenes and Carbon Nanotubes (Academic, 3.29 A.G. Rinzler, J. Liu, H. Dai, P. Nikolaev, C.B. Huffman,
San Diego 1995) F.J. Rodriguez-Macias, P.J. Boul, A.H. Lu, D. Hey-
3.14 R.C. Haddon: Chemistry of the fullerenes: The man- mann, D.T. Colbert, R.S. Lee, J.E. Fischer, A.M. Rao,
ifestation of strain in a class of continuous aromatic P.C. Eklund, R.E. Smalley: Large scale purification of
molecules, Science 261, 1545–1550 (1993) single wall carbon nanotubes: Process, product and
3.15 M. Monthioux, B.W. Smith, B. Burteaux, A. Claye, characterization, Appl. Phys. A 67, 29–37 (1998)
J. Fisher, D.E. Luzzi: Sensitivity of single-wall 3.30 L.M. Chapelle, J. Gavillet, J.L. Cochon, M. Ory,
nanotubes to chemical processing: An electron mi- S. Lefrant, A. Loiseau, D. Pigache: A continuous wave
croscopy investigation, Carbon 39, 1261–1272 (2001) CO2 laser reactor for nanotube synthesis, Proc. Elec-
3.16 H. Allouche, M. Monthioux: Chemical vapor depo- tron. Prop. Nov. Mater. – XVI Int. Wintersch. – AIP
sition of pyrolytic carbon onto carbon nanotubes. Conf. Proc., ed. by H. Kuzmany, J. Fink, M. Mehring,
Part II – Structure and texture, Carbon 43, 1265–1278 S. Roth (Springer, Berlin Heidelberg 1999) pp. 237–
(2005) 240
3.17 M. Audier, A. Oberlin, M. Oberlin, M. Coulon, L. Bon- 3.31 A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit,
netain: Morphology and crystalline order in catalytic J. Robert, C. Xu, Y.H. Lee, S.G. Kim, D.T. Colbert,
carbons, Carbon 19, 217–224 (1981) G. Scuseria, D. Tomanek, J.E. Fischer, R.E. Smal-
3.18 Y. Saito: Nanoparticles and filled nanocapsules, Car- ley: Crystalline ropes of metallic carbon nanotubes,
bon 33, 979–988 (1995) Science 273, 487–493 (1996)
102 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.32 M. Yudasaka, T. Komatsu, T. Ichihashi, S. Iijima: characteristics and the morphology and structure of
Single wall carbon nanotube formation by laser ab- the carbon phases synthesised by electric arc dis-
lation using double targets of carbon and metal, charge, Proc. 25th Bienn. Conf. Carbon, Lexington
Chem. Phys. Lett. 278, 102–106 (1997) 2001, ed. by F. Derbyshire (American Carbon Society
3.33 M. Castignolles, A. Foutel-Richard, A. Mavel, J.L. Co- 2001), Extend. Abstr. (CD-ROM), Novel/14.1
chon, D. Pigache, A. Loiseau, P. Bernier: Combined 3.44 M. Pacheco, M. Monthioux, M. Razafinimanana,
experimental and numerical study of the param- L. Donadieu, H. Allouche, N. Caprais, A. Gleizes:
eters controlling the C-SWNT synthesis via laser New factors controlling the formation of single-
Part A 3
vaporization, Proc. Electron. Prop. Nov. Mater. – XVI wall carbon nanotubes by arc plasma, Proc. Carbon
Int. Wintersch. – AIP Conf. Proc., ed. by H. Kuz- 2002 Int. Conf., Beijing 2002, ed. by H.-M. Cheng
many, J. Fink, M. Mehring, S. Roth (Springer, Berlin (Shanxi Chunqiu Audio-Visual Press, Beijing 2002),
Heidelberg 2002) pp. 385–389 CD-ROM/Oral/I014
3.34 T.W. Ebbesen, P.M. Ajayan: Large-scale synthesis of 3.45 M. Monthioux, M. Pacheco, H. Allouche, M. Razafin-
carbon nanotubes, Nature 358, 220–221 (1992) imanana, N. Caprais, L. Donnadieu, A. Gleizes: New
3.35 D. Ugarte: Morphology and structure of graphitic data about the formation of SWNTs by the electric
soot particles generated in arc-discharge C60 pro- arc method. In: Electronic Properties of Molecular
duction, Chem. Phys. Lett. 198, 596–602 (1992) Nanostructures, AIP Conf. Proc., ed. by H. Kuzmany,
3.36 T.W. Ebbesen: Carbon nanotubes, Ann. Rev. Mater. J. Fink, M. Mehring, S. Roth (Springer, Berlin Heidel-
Sci. 24, 235–264 (1994) berg 2002) pp. 182–185
3.37 T. Beltz, J. Find, D. Herein, N. Pfänder, T. Rühle, 3.46 H. Lange, A. Huczko, M. Sioda, M. Pacheco,
H. Werner, M. Wohlers, R. Schlögl: On the production M. Razafinimanana, A. Gleizes: Influence of
of different carbon forms by electric arc graphite gadolinium on carbon arc plasma and formation
evaporation, Ber. Bunsen. Phys. Chem. 101, 712–725 of fullerenes and nanotubes, Plasma Chem. Plasma
(1997) Process 22, 523–536 (2002)
3.38 C. Journet, W.K. Maser, P. Bernier, A. Loiseau, 3.47 C. Journet: La production de nanotubes de carbone.
L.M. de la Chapelle, S. Lefrant, P. Deniard, R. Lee, Ph.D. Thesis (University of Montpellier II, Montpellier
J.E. Fischer: Large-scale production of single-walled 1998)
carbon nanotubes by the electric-arc technique, 3.48 T. Sogabe, T. Masuda, K. Kuroda, Y. Hirohaya,
Nature 388, 756–758 (1997) T. Hino, T. Ymashina: Preparation of B4 C-mixed
3.39 K. Saïdane, M. Razafinimanana, H. Lange, M. Baltas, graphite by pressureless sintering and its air oxi-
A. Gleizes, J.J. Gonzalez: Influence of the carbon arc dation behavior, Carbon 33, 1783–1788 (1995)
current intensity on fullerene synthesis, Proc. 24th 3.49 M. Ishigami, J. Cumings, A. Zettl, S. Chen: A sim-
Int. Conf. Phenom. Ioniz. Gases, ed. by P. Pisarczyk, ple method for the continuous production of carbon
T. Pisarczyk, J. Wotowski (Institute of Plasma Physics nanotubes, Chem. Phys. Lett. 319, 457–459 (2000)
and Laser Microfusion, Warsaw 1999) pp. 203–204 3.50 Y.L. Hsin, K.C. Hwang, F.R. Chen, J.J. Kai: Produc-
3.40 H. Allouche, M. Monthioux, M. Pacheco, M. Razafin- tion and in-situ metal filling of carbon nanotube in
imanana, H. Lange, A. Huczko, T.P. Teulet, water, Adv. Mater. 13, 830–833 (2001)
A. Gleizes, T. Sogabe: Physical characteristics of 3.51 H.W. Zhu, X.S. Li, B. Jiang, C.L. Xu, C.L. Zhu, Y.F. Zhu,
the graphite-electrode electric-arc as parameters D.H. Wu, X.H. Chen: Formation of carbon nanotubes
for the formation of single-wall carbon nano- in water by the electric arc technique, Chem. Phys.
tubes, Proc. Eurocarbon, Vol. 2 (Deutsche Keramische Lett. 366, 664–669 (2002)
Gesellschaft, 2000) pp. 1053–1054 3.52 W.K. Maser, P. Bernier, J.M. Lambert, O. Stephan,
3.41 M. Razafinimanana, M. Pacheco, M. Monthioux, P.M. Ajayan, C. Colliex, V. Brotons, J.M. Planeix,
H. Allouche, H. Lange, A. Huczko, A. Gleizes: Spec- B. Coq, P. Molinie, S. Lefrant: Elaboration and
troscopic study of an electric arc with Gd and Fe characterization of various carbon nanostructures,
doped anodes for the carbon nanotube formation, Synth. Met. 81, 243–250 (1996)
Proc. 25th Int. Conf. Phenom. Ioniz. Gases, ed. by 3.53 A. Mansour, M. Razafinimanana, M. Monthioux,
E. Goto (Nagoya Univ., Nagoya 2001) pp. 297–298 M. Pacheco, A. Gleizes: A significant improvement
3.42 M. Razafinimanana, M. Pacheco, M. Monthioux, of both yield and purity during SWCNT synthesis via
H. Allouche, H. Lange, A. Huczko, P. Teulet, the electric arc process, Carbon 45, 1651–1661 (2007)
A. Gleizes, C. Goze, P. Bernier, T. Sogabe: Influence of 3.54 A. Mansour: Caractérisation expérimentale d’un
doped graphite electrode in electric arc for the for- plasma d’arc électrique en vue du contrôle de la
mation of single wall carbon nanotubes, Proc. 6th synthèse des nanotubes de carbone monoparois.
Eur. Conf. Therm. Plasma Process. – Prog. Plasma Ph.D. Thesis (University Paul Sabatier, Toulouse 2007)
Process. Mater., New York 2000, ed. by P. Fauchais 3.55 J. Gavillet, A. Loiseau, J. Thibault, A. Maigné,
(Begell House, New York 2001) pp. 649–654 O. Stéphan, P. Bernier: TEM study of the influence
3.43 M. Pacheco, H. Allouche, M. Monthioux, A. Razafini- of the catalyst composition on the formation and
manana, A. Gleizes: Correlation between the plasma growth of SWNT, Proc. Electron. Prop. Nov. Mater. –
Introduction to Carbon Nanotubes References 103
XVI Int. Wintersch. – AIP Conf. Proc., ed. by H. Kuz- 3.68 G.G. Tibbetts, M. Endo, C.P. Beetz: Carbon fibers
many, J. Fink, M. Mehring, S. Roth (Springer, Berlin grown from the vapor phase: A novel material,
Heidelberg 2002) pp. 202–206 SAMPE Journal 22, 30 (1989)
3.56 G. Flamant, J.F. Robert, S. Marty, J.M. Gineste, J. Gi- 3.69 R.T.K. Baker: Catalytic growth of carbon filaments,
ral, B. Rivoire, D. Laplaze: Solar reactor scaling up. Carbon 27, 315–323 (1989)
The fullerene synthesis case study, Energy 29, 801– 3.70 E. Lamouroux, P. Serp, P. Kalck: Catalytic chemical
809 (2004) vapor deposition routes towards single-walled and
3.57 T.M. Gruenberger, J. Gonzalez-Aguilar, F. Fabry, double-walled carbon nanotubes, Catal. Rev. Sci.
Part A 3
L. Fulchieri, E. Grivei, N. Probst, G. Flamant, Eng. 49, 341–405 (2007)
H. Okuno, J.C. Charlier: Production of carbon nano- 3.71 R. Philippe, A. Morançais, M. Corrias, B. Caussat,
tubes and other nanostructures via continuous Y. Kihn, P. Kalck, D. Plee, P. Gaillard, D. Bernard,
3-phase AC plasma processing, Fuller. Nanotub. Car- P. Serp: Catalytic production of carbon nanotubes
bon Nanostruct. 12, 571–581 (2004) by fluidized-bed CVD, Chem. Vap. Depos. 13, 447–457
3.58 H. Okuno, E. Grivel, F. Fabry, T.M. Gruenberger, (2007)
J.J. Gonzalez-Aguilar, A. Palnichenko, L. Fulchieri, 3.72 R.T.K. Baker, P.S. Harris, R.B. Thomas, R.J. Waite:
N. Probst, J.C. Chalier: Synthesis of carbon nanotubes Formation of filamentous carbon from iron, cobalt,
and nano-necklaces by thermal plasma process, and chromium catalyzed decomposition of acety-
Carbon 42, 2543–2549 (2004) lene, J. Catal. 30, 86–95 (1973)
3.59 L.P.F. Chibante, A. Thess, J.M. Alford, M.D. Diener, 3.73 T. Koyama, M. Endo, Y. Oyuma: Carbon fibers
R.E. Smalley: Solar generation of the fullerenes, obtained by thermal decomposition of vaporized
J. Phys. Chem. 97, 8696–8700 (1993) hydrocarbon, Jpn. J. Appl. Phys. 11, 445–449 (1972)
3.60 C.L. Fields, J.R. Pitts, M.J. Hale, C. Bingham, 3.74 M. Endo, A. Oberlin, T. Koyama: High resolution
A. Lewandowski, D.E. King: Formation of fullerenes electron microscopy of graphitizable carbon fiber
in highly concentrated solar flux, J. Phys. Chem. 97, prepared by benzene decomposition, Jpn. J. Appl.
8701–8702 (1993) Phys. 16, 1519–1523 (1977)
3.61 P. Bernier, D. Laplaze, J. Auriol, L. Barbedette, 3.75 N.M. Rodriguez: A review of catalytically grown car-
G. Flamant, M. Lebrun, A. Brunelle, S. Della-Negra: bon nanofibers, J. Mater. Res. 8, 3233–3250 (1993)
Production of fullerenes from solar energy, Synth. 3.76 W.R. Davis, R.J. Slawson, G.R. Rigby: An unusual
Met. 70, 1455–1456 (1995) form of carbon, Nature 171, 756 (1953)
3.62 M.J. Heben, T.A. Bekkedhal, D.L. Schultz, K.M. Jones, 3.77 H.P. Boehm: Carbon from carbon monoxide dis-
A.C. Dillon, C.J. Curtis, C. Bingham, J.R. Pitts, proportionation on nickel and iron catalysts;
A. Lewandowski, C.L. Fields: Production of single morphological studies and possible growth mech-
wall carbon nanotubes using concentrated sun- anisms, Carbon 11, 583–590 (1973)
light, Proc. Symp. Recent Adv. Chem. Phys. Fuller. 3.78 M. Audier, A. Oberlin, M. Coulon: Crystallographic
Rel. Mater., Pennington 1996, ed. by K.M. Kadish, orientations of catalytic particles in filamentous car-
R.S. Ruoff (Electrochemical Society, Pennington bon; case of simple conical particles, J. Cryst. Growth
1996) pp. 803–811 55, 546–549 (1981)
3.63 D. Laplaze, P. Bernier, C. Journet, G. Vié, G. Flamant, 3.79 M. Audier, M. Coulon: Kinetic and microscopic as-
E. Philippot, M. Lebrun: Evaporation of graphite us- pects of catalytic carbon growth, Carbon 23, 317–323
ing a solar furnace, Proc. 8th Int. Symp. Solar Conc. (1985)
Technol., Köln 1996, ed. by M. Becker, M. Balmer 3.80 M. Audier, A. Oberlin, M. Coulon: Study of biconic
(Müller, Heidelberg 1997) pp. 1653–1656 microcrystals in the middle of carbon tubes obtained
3.64 D. Laplaze, P. Bernier, W.K. Maser, G. Flamant, by catalytic disproportionation of CO, J. Cryst. Growth
T. Guillard, A. Loiseau: Carbon nanotubes: The solar 57, 524–534 (1981)
approach, Carbon 36, 685–688 (1998) 3.81 A. Thaib, G.A. Martin, P. Pinheiro, M.C. Schouler,
3.65 T. Guillard, S. Cetout, L. Alvarez, J.L. Sauvajol, P. Gadelle: Formation of carbon nanotubes from the
E. Anglaret, P. Bernier, G. Flamant, D. Laplaze: Pro- carbon monoxide disproportionation reaction over
duction of carbon nanotubes by the solar route, Eur. Co/Al2 O3 and Co/SiO2 catalysts, Catal. Lett. 63, 135–141
Phys. J. 5, 251–256 (1999) (1999)
3.66 D. Luxembourg, G. Flamant, A. Guillot, D. Laplaze: 3.82 P. Pinheiro, M.C. Schouler, P. Gadelle, M. Mermoux,
Hydrogen storage in solar produced single-walled E. Dooryhée: Effect of hydrogen on the orientation of
carbon nanotubes, Mater. Sci. Eng. B 108, 114–119 carbon layers in deposits from the carbon monoxide
(2004) disproportionation reaction over Co/Al2 O3 catalysts,
3.67 G. Flamant, M. Bijeire, D. Luxembourg: Modelling of Carbon 38, 1469–1479 (2000)
a solar reactor for single wall nanotubes synthesis, 3.83 P. Pinheiro, P. Gadelle: Chemical state of a sup-
ASME J. Solar Energy Eng. 128, 1–124 (2006) ported iron-cobalt catalyst during CO disproportion-
104 Part A Nanostructures, Micro-/Nanofabrication and Materials
ation. I. Thermodynamic study, J. Phys. Chem. Solids (Mg,Co, Mo)O catalysts: Influence of the proportions
62, 1015–1021 (2001) of cobalt and molybdenum, J. Mater. Chem. 14,
3.84 P. Pinheiro, P. Gadelle, C. Jeandey, J.L. Oddou: 646–653 (2004)
Chemical state of a supported iron-cobalt cata- 3.99 E. Flahaut, C. Laurent, A. Peigney: Catalytic CVD
lyst during CO disproportionation. II. Experimental synthesis of double and triple-walled carbon nano-
study, J. Phys. Chem. Solids 62, 1023–1037 (2001) tubes by the control of the catalyst preparation,
3.85 C. Laurent, E. Flahaut, A. Peigney, A. Rousset: Metal Carbon 43, 375–383 (2005)
nanoparticles for the catalytic synthesis of carbon 3.100 R. Marangoni, P. Serp, R. Feurrer, Y. Kihn, P. Kalck,
Part A 3
nanotubes, New J. Chem. 22, 1229–1237 (1998) C. Vahlas: Carbon nanotubes produced by substrate
3.86 E. Flahaut: Synthèse par voir catalytique et carac- free metalorganic chemical vapor deposition of iron
térisation de composites nanotubes de carbone- catalyst and ethylene, Carbon 39, 443–449 (2001)
metal-oxyde Poudres et matériaux denses. Ph.D. 3.101 R. Sen, A. Govindaraj, C.N.R. Rao: Carbon nanotubes
Thesis (Univers. Paul Sabatier, Toulouse 1999) by the metallocene route, Chem. Phys. Lett. 267,
3.87 E. Flahaut, R. Bacsa, A. Peigney, C. Laurent: 276–280 (1997)
Gram-scale CCVD synthesis of double-walled carbon 3.102 Y.Y. Fan, H.M. Cheng, Y.L. Wei, G. Su, S.H. Shen: The
nanotubes, Chem. Commun., 1442–1443 (2003) influence of preparation parameters on the mass
3.88 A. Peigney, C. Laurent, F. Dobigeon, A. Rousset: Car- production of vapor grown carbon nanofibers, Car-
bon nanotubes grown in situ by a novel catalytic bon 38, 789–795 (2000)
method, J. Mater. Res. 12, 613–615 (1997) 3.103 L. Ci, J. Wei, B. Wei, J. Liang, C. Xu, D. Wu: Carbon
3.89 V. Ivanov, J.B. Nagy, P. Lambin, A. Lucas, nanofibers and single-walled carbon nanotubes
X.B. Zhang, X.F. Zhang, D. Bernaerts, G. Van Ten- prepared by the floating catalyst method, Carbon
deloo, S. Amelinckx, J. Van Landuyt: The study of 39, 329–335 (2001)
nanotubules produced by catalytic method, Chem. 3.104 M. Glerup, H. Kanzow, R. Almairac, M. Castignolles,
Phys. Lett. 223, 329–335 (1994) P. Bernier: Synthesis of multi-walled carbon nano-
3.90 V. Ivanov, A. Fonseca, J.B. Nagy, A. Lucas, P. Lambin, tubes and nano-fibres using aerosol method with
D. Bernaerts, X.B. Zhang: Catalytic production and metal-ions as the catalyst precursors, Chem. Phys.
purification of nanotubules having fullerene-scale Lett. 377, 293–298 (2003)
diameters, Carbon 33, 1727–1738 (1995) 3.105 O.A. Nerushev, M. Sveningsson, L.K.L. Falk,
3.91 K. Hernadi, A. Fonseca, J.B. Nagy, D. Bernaerts, F. Rohmund: Carbon nanotube films obtained by
A. Fudala, A. Lucas: Catalytic synthesis of carbon thermal vapour deposition, J. Mater. Chem. 11, 1122–
nanotubes using zeolite support, Zeolites 17, 416–423 1132 (2001)
(1996) 3.106 Z. Zhou, L. Ci, L. Song, X. Yan, D. Liu, H. Yuan,
3.92 H. Dai, A.G. Rinzler, P. Nikolaev, A. Thess, D.T. Col- Y. Gao, J. Wang, L. Liu, W. Zhou, G. Wang, S. Xie:
bert, R.E. Smalley: Single-wall nanotubes produced Producing cleaner double-walled carbon nanotubes
by metal-catalysed disproportionation of carbon in a floating catalyst system, Carbon 41, 2607–2611
monoxide, Chem. Phys. Lett. 260, 471–475 (1996) (2003)
3.93 A.M. Cassel, J.A. Raymakers, J. Kong, H. Dai: Large 3.107 F. Rohmund, L.K.L. Falk, F.E.B. Campbell: A simple
scale CVD synthesis of single-walled carbon nano- method for the production of large arrays of aligned
tubes, J. Phys. Chem. B 109, 6484–6492 (1999) carbon nanotubes, Chem. Phys. Lett. 328, 369–373
3.94 B. Kitiyanan, W.E. Alvarez, J.H. Harwell, D.E. Re- (2000)
sasco: Controlled production of single-wall carbon 3.108 G.G. Tibbetts, C.A. Bernardo, D.W. Gorkiewicz,
nanotubes by catalytic decomposition of CO on R.L. Alig: Role of sulfur in the production of car-
bimetallic Co-Mo catalysts, Chem. Phys. Lett. 317, bon fibers in the vapor phase, Carbon 32, 569–576
497–503 (2000) (1994)
3.95 A. Govindaraj, E. Flahaut, C. Laurent, A. Peigney, 3.109 S. Bai, F. Li, Q.H. Yang, H.-M. Cheng, J.B. Bai:
A. Rousset, C.N.R. Rao: An investigation of car- Influence of ferrocene/benzene mole ratio in the
bon nanotubes obtained from the decomposition of synthesis of carbon nanostructures, Chem. Phys.
methane over reduced Mg1−x Mx Al2 O4 spinel cata- Lett. 376, 83–89 (2003)
lysts, J. Mater. Res. 14, 2567–2576 (1999) 3.110 W.Q. Han, P. Kholer-Riedlich, T. Seeger, F. Ernst,
3.96 E. Flahaut, A. Peigney, C. Laurent, A. Rousset: Syn- M. Ruhle, N. Grobert, W.K. Hsu, B.H. Chang, Y.Q. Zhu,
thesis of single-walled carbon nanotube-Co-MgO H.W. Kroto, M. Terrones, H. Terrones: Aligned CNx
composite powders and extraction of the nano- nanotubes by pyrolysis of ferrocene under NH3 at-
tubes, J. Mater. Chem. 10, 249–252 (2000) mosphere, Appl. Phys. Lett. 77, 1807–1809 (2000)
3.97 J. Kong, A.M. Cassel, H. Dai: Chemical vapor de- 3.111 L. Ci, Z. Rao, Z. Zhou, D. Tang, X. Yan, Y. Liang,
position of methane for single-walled carbon D. Liu, H. Yuan, W. Zhou, G. Wang, W. Liu, S. Xie:
nanotubes, Chem. Phys. Lett. 292, 567–574 (1998) Double wall carbon nanotubes promoted by sulfur
3.98 E. Flahaut, A. Peigney, W.S. Bacsa, R.R. Bacsa, in a floating iron catalyst CVD system, Chem. Phys.
C. Laurent: CCVD synthesis of carbon nanotubes from Lett. 359, 63–67 (2002)
Introduction to Carbon Nanotubes References 105
3.112 S. Maruyama, R. Kojima, Y. Miyauchi, S. Chi- 3.127 Y.Y. Wei, G. Eres, V.I. Merkulov, D.H. Lowdens: Effect
ashi, M. Kohno: Low-temperature synthesis of of film thickness on carbon nanotube growth by se-
high-purity single-walled carbon nanotubes from lective area chemical vapor deposition, Appl. Phys.
alcohol, Chem. Phys. Lett. 360, 229–234 (2002) Lett. 78, 1394–1396 (2001)
3.113 T. Kyotani, L.F. Tsai, A. Tomita: Preparation of ul- 3.128 I.T. Han, B.K. Kim, H.J. Kim, M. Yang, Y.W. Jin,
trafine carbon tubes in nanochannels of an anodic S. Jung, N. Lee, S.K. Kim, J.M. Kim: Effect of Al and
aluminum oxide film, Chem. Mater. 8, 2109–2113 catalyst thickness on the growth of carbon nano-
(1996) tubes and application to gated field emitter arrays,
Part A 3
3.114 E. Mora, T. Tokune, A.R. Harutyunyan: Continuous Chem. Phys. Lett. 400, 139–144 (2004)
production of single-walled carbon nanotubes us- 3.129 W.Z. Li, S.S. Xie, L.X. Qian, B.H. Chang, B.S. Zou,
ing a supported floating catalyst, Carbon 45, 971–977 W.Y. Zhou, R.A. Zha, G. Wang: Large scale synthesis
(2007) of aligned carbon nanotubes, Science 274, 1701–1703
3.115 R.E. Smalley, J.H. Hafner, D.T. Colbert, K. Smith: Cat- (1996)
alytic growth of single-wall carbon nanotubes from 3.130 F. Zheng, L. Liang, Y. Gao, J.H. Sukamto, L. Aardahl:
metal particles, US Patent 19980601010903 (1998) Carbon nanotubes synthesis using mesoporous silica
3.116 P. Nikolaev: Gas-phase production of single-walled templates, Nano Lett. 2, 729–732 (2002)
carbon nanotubes from carbon monoxide: A review 3.131 S.H. Jeong, O.-K. Lee, K.H. Lee, S.H. Oh, C.G. Park:
of the HiPco process, J. Nanosci. Nanotechnol. 4, Preparation of aligned carbon nanotubes with
307–316 (2004) prescribed dimension: Template synthesis and soni-
3.117 W.K. Hsu, J.P. Hare, M. Terrones, H.W. Kroto, cation cutting approach, Chem. Mater. 14, 1859–1862
D.R.M. Walton, P.J.F. Harris: Condensed-phase (2002)
nanotubes, Nature 377, 687 (1995) 3.132 N.S. Kim, Y.T. Lee, J. Park, H. Ryu, H.J. Lee, S.Y. Choi,
3.118 W.S. Cho, E. Hamada, Y. Kondo, K. Takayanagi: Syn- J. Choo: Dependence of vertically aligned growth of
thesis of carbon nanotubes from bulk polymer, Appl. carbon nanotubes on catalyst, J. Phys. Chem. B 106,
Phys. Lett. 69, 278–279 (1996) 9286–9290 (2002)
3.119 Y.L. Li, Y.D. Yu, Y. Liang: A novel method for 3.133 C.J. Lee, D.W. Kim, T.J. Lee, Y.C. Choi, Y.S. Park,
synthesis of carbon nanotubes: Low tempera- Y.H. Lee, W.B. Choi, N.S. Lee, G.-S. Park, J.M. Kim:
ture solid pyrolysis, J. Mater. Res. 12, 1678–1680 Synthesis of aligned carbon nanotubes using ther-
(1997) mal chemical vapor deposition, Chem. Phys. Lett.
3.120 M.L. Terranova, S. Piccirillo, V. Sessa, P. Sbornicchia, 312, 461–468 (1999)
M. Rossi, S. Botti, D. Manno: Growth of single-walled 3.134 W.D. Zhang, Y. Wen, S.M. Liu, W.C. Tjiu, G.Q. Xu,
carbon nanotubes by a novel technique using nano- L.M. Gan: Synthesis of vertically aligned carbon
sized graphite as carbon source, Chem. Phys. Lett. nanotubes on metal deposited quartz plates, Carbon
327, 284–290 (2000) 40, 1981–1989 (2002)
3.121 R.L. Vander Wal, T. Ticich, V.E. Curtis: Diffusion flame 3.135 S. Huang, L. Dai, A.W.H. Mau: Controlled fabrication
synthesis of single-walled carbon nanotubes, Chem. of large scale aligned carbon nanofiber/nanotube
Phys. Lett. 323, 217–223 (2000) patterns by photolithography, Adv. Mater. 14, 1140–
3.122 I. Gunjishima, T. Inoue, S. Yamamuro, K. Sumiyama, 1143 (2002)
A. Okamoto: Synthesis of vertically aligned, double- 3.136 T. Sun, G. Wang, H. Liu, L. Feng, D. Zhu: Control over
walled carbon nanotubes from highly active Fe-V-O the wettability of an aligned carbon nanotube film,
nanoparticles, Carbon 45, 1193–1199 (2007) J. Am. Chem. Soc. 125, 14996–14997 (2003)
3.123 G. Zhong, T. Iwasaki, J. Robertson, H. Kawarada: 3.137 Y. Huh, J.Y. Lee, J. Cheon, Y.K. Hong, J.Y. Koo,
Growth kinetics of 0.5 cm vertically aligned single- T.J. Lee, C.J. Lee: Controlled growth of carbon nano-
walled carbon nanotubes, J. Phys. Chem. B 111, 1907– tubes over cobalt nanoparticles by thermal chemical
1910 (2007) vapor deposition, J. Mater. Chem. 13, 2297–2300
3.124 H. Cui, G. Eres, J.Y. Howe, A. Puretzki, M. Varela, (2003)
D.B. Geohegan, D.H. Lowndes: Growth behavior of 3.138 Y. Kobayashi, H. Nakashima, D. Takagi, Y. Homma:
carbon nanotubes on multilayered metal catalyst CVD growth of single-walled carbon nanotubes us-
film in chemical vapor deposition, Chem. Phys. Lett. ing size-controlled nanoparticle catalyst, Thin Solid
374, 222–228 (2003) Films 464/465, 286–289 (2004)
3.125 A.M. Cassel, N.R. Franklin, T.W. Tombler, E.M. Chan, 3.139 C.L. Cheung, A. Kurtz, H. Park, C.M. Lieber: Diameter-
J. Han, H. Dai: Directed growth of free-standing controlled synthesis of carbon nanotubes, J. Phys.
single-walled carbon nanotubes, J. Am. Chem. Soc. Chem. B 106, 2429–2433 (2002)
121, 7975–7976 (1999) 3.140 Y. Huh, J.Y. Lee, J. Cheon, Y.K. Hong, J.Y. Koo,
3.126 S. Fan, M. Chapline, N. Franklin, T. Tombler, T.J. Lee, C.J. Lee: Controlled growth of carbon nano-
A.M. Cassel, H. Dai: Self-oriented regular arrays of tubes over cobalt nanoparticles by thermal chemical
carbon nanotubes and their field emission proper- vapor deposition, J. Mater. Chem. 13, 2297–2300
ties, Science 283, 512–514 (1999) (2003)
106 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.141 M. Paillet, V. Jourdain, P. Poncharal, J.-L. Sauvajol, 3.154 X. Li, L. Zhang, X. Wang, I. Shimoyama, X. Sun,
A. Zahab, J.C. Meyer, S. Roth, N. Cordente, C. Amiens, W.-S. Seo, H. Dai: Assembly of densely aligned
B. Chaudret: Versatile synthesis of individual single- single-walled carbon nanotubes from bulk ma-
walled carbon nanotubes from nickel nanoparticles terials Langmuir–Blodgett, J. Am. Chem. Soc. 129,
for the study of their physical properties, J. Phys. 4890–4891 (2007)
Chem. B 108, 17112–17118 (2004) 3.155 M. Endo, H.W. Kroto: Formation of carbon
3.142 S. Casimirius, E. Flahaut, C. Laurent, C. Vieu, nanofibers, J. Phys. Chem. 96, 6941–6944 (1992)
F. Carcenac, C. Laberty-Robert: Optimized microcon- 3.156 R.S. Wagner: VLS mechanisms of crystal growth. In:
Part A 3
tact printing process for the patterned growth of Whisker Technology, ed. by P.A. Levit (Wiley, New
individual SWNTs, Microelectron. Eng. 73/74, 564– York 1970) pp. 47–72
569 (2004) 3.157 Y.H. Lee, S.G. Kim, D. Tomanek: Catalytic growth of
3.143 Y. Lei, K.S. Yeong, J.T.L. Thong, W.K. Chim: Large- single-wall carbon nanotubes: An ab initio study,
scale ordered carbon nanotubes arrays initiated Phys. Rev. Lett. 78, 2393–2396 (1997)
from highly ordered catalyst arrays on silicon sub- 3.158 H. Dai: Carbon Nanotubes: Synthesis, integration,
strates, Chem. Mater. 16, 2757–2761 (2004) and properties, Acc. Chem. Res. 35, 1035–1044 (2002)
3.144 Q. Ye, A.M. Cassel, H. Liu, K.J. Chao, J. Han, 3.159 V. Jourdain, H. Kanzow, M. Castignolles, A. Loiseau,
M. Meyyappan: Large-scale fabrication of carbon P. Bernier: Sequential catalytic growth of carbon
nanotube probe tips for atomic force microscopy nanotubes, Chem. Phys. Lett. 364, 27–33 (2002)
critical dimension imaging applications, Nano Lett. 3.160 Y. Saito, M. Okuda, N. Fujimoto, T. Yoshikawa,
4, 1301–1308 (2004) M. Tomita, T. Hayashi: Single-wall carbon nanotubes
3.145 K. Hata, D.N. Futaba, K. Mizuno, T. Namai, M. Yu- growing radially from Ni fine particles formed by arc
mara, S. Iijima: Ware-assisted highly efficient evaporation, Jpn. J. Appl. Phys. 33, L526–L529 (1994)
synthesis of impurity-free single-walled carbon 3.161 J. Bernholc, C. Brabec, M. Buongiorno Nardelli,
nanotubes, Science 306, 1362–1364 (2004) A. Malti, C. Roland, B.J. Yakobson: Theory of growth
3.146 R. Andrews, D. Jacques, A.M. Rao, F. Derbyshire, and mechanical properties of nanotubes, Appl.
D. Qian, X. Fan, E.C. Dickey, J. Chen: Continous pro- Phys. A 67, 39–46 (1998)
duction of aligned carbon nanotubes: A step closer 3.162 M. Pacheco: Synthèse des nanotubes de carbone par
to commercial realization, Chem. Phys. Lett. 303, arc electrique. Ph.D. Thesis (Université Toulouse III,
467–474 (1999) Toulouse 2003)
3.147 C.N.R. Rao, R. Sen, B.C. Satishkumar, A. Govindaraj: 3.163 K. Méténier, S. Bonnamy, F. Béguin, C. Journet,
Large aligned carbon nanotubes bundles from fer- P. Bernier, L.M. de la Chapelle, O. Chauvet, S. Lefrant:
rocene pyrolysis, Chem. Commun., 1525–1526 (1998) Coalescence of single walled nanotubes and forma-
3.148 X. Zhang, A. Cao, B. Wei, Y. Li, J. Wei, C. Xu, D. Wu: tion of multi-walled carbon nanotubes under high
Rapid growth of well-aligned carbon nanotube ar- temperature treatments, Carbon 40, 1765–1773 (2002)
rays, Chem. Phys. Lett. 362, 285–290 (2002) 3.164 P.G. Collins, P. Avouris: Nanotubes for electronics,
3.149 X. Zhang, A. Cao, Y. Li, C. Xu, J. Liang, D. Wu, B. Wei: Sci. Am. 283, 38–45 (2000)
Self-organized arrays of carbon nanotube ropes, 3.165 Q.-H. Yang, P.X. Hou, S. Bai, M.Z. Wang, H.M. Cheng:
Chem. Phys. Lett. 351, 183–188 (2002) Adsorption and capillarity of nitrogen in aggregated
3.150 K.S. Choi, Y.S. Cho, S.Y. Hong, J.B. Park, D.J. Kim: multi-walled carbon nanotubes, Chem. Phys. Lett.
Effects of ammonia on the alignment of carbon 345, 18–24 (2001)
nanotubes in metal-assisted chemical vapor depo- 3.166 S. Inoue, N. Ichikuni, T. Suzuki, T. Uematsu,
sition, J. Eur. Ceram. Soc. 21, 2095–2098 (2001) K. Kaneko: Capillary condensation of N2 on multiwall
3.151 N.S. Kim, Y.T. Lee, J. Park, J.B. Han, Y.S. Choi, carbon nanotubes, J. Phys. Chem. 102, 4689–4692
S.Y. Choi, J. Choo, G.H. Lee: Vertically aligned car- (1998)
bon nanotubes grown by pyrolysis of iron, cobalt, 3.167 S. Agnihotri, J.P. Mota, M. Rostam-Abadi, M.J. Rood:
and nickel phthalocyanines, J. Phys. Chem. B 107, Structural characterization of single-walled carbon
9249–9255 (2003) nanotube bundles by experiment and molecular
3.152 C. Emmeger, J.M. Bonard, P. Mauron, P. Sudan, simulation, Langmuir 21, 896–904 (2005)
A. Lepora, B. Grobety, A. Züttel, L. Schlapbach: 3.168 M. Eswaramoorthy, R. Sen, C.N.R. Rao: A study of
Synthesis of carbon nanotubes over Fe catalyst on micropores in single-walled carbon nanotubes by
aluminum and suggested growth mechanism, Car- the adsorption of gases and vapors, Chem. Phys.
bon 41, 539–547 (2003) Lett. 304, 207–210 (1999)
3.153 Q. Zhang, J. Huang, F. Wei, G. Xu, Y. Wang, W. Qian, 3.169 S. Furmaniak, A.P. Terzyk, P.A. Gauden, K. Lota,
D. Wang: Large scale production of carbon nano- E. Frackowiak, F. Beguin, P. Kowalczyk: Determina-
tubes arrays on the sphere surface from liquefied tion of the space between closed multiwalled carbon
petroleum gas at low cost, Chin. Sci. Bull. 52, 2896– nanotubes by GCMC simulation of nitrogen adsorp-
2902 (2007) tion, J. Colloid Interface Sci. 317, 442–448 (2008)
Introduction to Carbon Nanotubes References 107
3.170 A. Peigney, C. Laurent, E. Flahaut, R.R. Bacsa, ended nanotube bundles, J. Phys. Chem. B 107,
A. Rousset: Specific surface area of carbon nano- 1540–1542 (2003)
tubes and bundles of carbon nanotubes, Carbon 39, 3.185 J. Jiang, S.I. Sandler: Nitrogen adsorption on carbon
507–514 (2001) nanotubes bundles: Role of the external surface,
3.171 E. Frackowiak, S. Delpeux, K. Jurewicz, K. Szostak, Phys. Rev. B 68, 245412–1–245412–9 (2003)
D. Cazorla-Amoros, F. Béguin: Enhanced capacitance 3.186 M. Arab, F. Picaud, C. Ramseyer, M.R. Babaa, F. Val-
of carbon nanotubes through chemical activation, saque, E. McRae: Characterization of single wall
Chem. Phys. Lett. 336, 35–41 (2002) carbon nanotubes by means of rare gas adsorption,
Part A 3
3.172 E. Raymundo-Piñero, P. Azaïs, T. Cacciaguerra, J. Chem. Phys. 126, 054709 (2007)
D. Cazorla-Amorós, A. Linares-Solano, F. Béguin: 3.187 J. Zhao, A. Buldum, J. Han, J.P. Lu: Gas molecule
KOH and NaOH activation mechanisms of multi- adsorption in carbon nanotubes and nanotube bun-
walled carbon nanotubes with different structural dles, Nanotechnology 13, 195–200 (2002)
organisation, Carbon 43, 786–795 (2005) 3.188 C. Matranga, B. Bockrath: Hydrogen-bonded and
3.173 S. Delpeux, K. Szostak, E. Frackowiak, F. Béguin: physisorbed CO in single-walled carbon nanotubes
An efficient two-step process for producing opened bundles, J. Phys. Chem. B 109, 4853–4864 (2005)
multi-walled carbon nanotubes of high purity, 3.189 M.D. Ellison, M.J. Crotty, D. Koh, R.L. Spray, K.E. Tate:
Chem. Phys. Lett. 404, 374–378 (2005) Adsorption of NH3 and NO2 on single-walled car-
3.174 K.A. Williams, P.C. Eklund: Monte Carlo simulation of bon nanotubes, J. Phys. Chem. B 108, 7938–7943
H2 physisorption in finite diameter carbon nanotube (2004)
ropes, Chem. Phys. Lett. 320, 352–358 (2000) 3.190 S. Picozzi, S. Santucci, L. Lozzi, L. Valentin, B. Delley:
3.175 U. Burghaus, D. Bye, K. Cosert, J. Goering, A. Guer- Ozone adsorption on carbon nanotubes: The role of
ard, E. Kadossov, E. Lee, Y. Nadoyama, N. Richter, Stone–Wales defects, J. Chem. Phys. 120, 7147–7152
E. Schaefer, J. Smith, D. Ulness, B. Wymore: (2004)
Methanol adsorption in carbon nanotubes, Chem. 3.191 N. Chakrapani, Y.M. Zhang, S.K. Nayak, J.A. Moore,
Phys. Lett. 442, 344–347 (2007) D.L. Carrol, Y.Y. Choi, P.M. Ajayan: Chemisorption of
3.176 Z. Chen, W. Thiel, A. Hirsch: Reactivity of the con- acetone on carbon nanotubes, J. Phys. Chem. B 107,
vex and concave surfaces of single-walled carbon 9308–9311 (2003)
nanotubes (SWCNTs) towards addition reactions: de- 3.192 A. Chambers, C. Park, R.T.K. Baker, N. Rodriguez:
pendence on the carbon-atom pyramidalization, Hydrogen storage in graphite nanofibers, J. Phys.
ChemPhysChem 1, 93–97 (2003) Chem. B 102, 4253–4256 (1998)
3.177 S. Park, D. Srivastava, K. Cho: Generalized reactivity 3.193 J. Giraudet, M. Dubois, D. Claves, J.P. Pinheiro,
of curved surfaces: carbon nanotubes, Nano Lett. 3, M.C. Schouler, P. Gadelle, A. Hamwi: Modifying the
1273–1277 (2003) electronic properties of multi-wall carbon nano-
3.178 X. Lu, Z. Chen, P. Schleyer: Are Stone–Wales defect tubes via charge transfer, by chemical doping with
sites always more reactive than perfect sites in the some inorganic fluorides, Chem. Phys. Lett. 381,
sidewalls of single-wall carbon nanotubes?, J. Am. 306–314 (2003)
Chem. Soc. 127, 20–21 (2005) 3.194 J. Hilding, E.A. Grulke, S.B. Sinnott, D. Qian, R. An-
3.179 M. Muris, N. Dupont-Pavlosky, M. Bienfait, P. Zep- drews, M. Jagtoyen: Sorption of butane on carbon
penfeld: Where are the molecules adsorbed on multiwall nanotubes at room temperature, Lang-
single-walled nanotubes?, Surf. Sci. 492, 67–74 muir 17, 7540–7544 (2001)
(2001) 3.195 K. Masenelli-Varlot, E. McRae, N. Dupont-Pavlosky:
3.180 R.B. Hallock, Y.H. Yang: Adsorption of helium and Comparative adsorption of simple molecules on
other gases to carbon nanotubes and nanotubes carbon nanotubes. Dependence of the adsorption
bundles, J. Low Temp. Phys. 134, 21–30 (2004) properties on the nanotube morphology, Appl. Surf.
3.181 J. Zhu, Y. Wang, W. Li, F. Wei, Y. Yu: Density Sci. 196, 209–215 (2002)
functional study of nitrogen adsorption in single- 3.196 D.J. Browning, M.L. Gerrard, J.B. Lakeman, I.M. Mel-
wall carbon nanotubes, Nanotechnology 18, 095707 lor, R.J. Mortimer, M.C. Turpin: Studies into the
(2007) storage of hydrogen in carbon nanofibers: Proposal
3.182 A. Fujiwara, K. Ishii, H. Suematsu, H. Kataura, of a possible mechanism, Nano Lett. 2, 201–205
Y. Maniwa, S. Suzuki, Y. Achiba: Gas adsorption (2002)
in the inside and outside of single-walled carbon 3.197 F.H. Yang, R.T. Yang: Ab initio molecular orbital
nanotubes, Chem. Phys. Lett. 336, 205–211 (2001) study of adsorption of atomic hydrogen on graphite:
3.183 C.M. Yang, H. Kanoh, K. Kaneko, M. Yudasaka, insight into hydrogen storage in carbon nanotubes,
S. Iijima: Adsorption behaviors of HiPco single- Carbon 40, 437–444 (2002)
walled carbon nanotubes aggregates for alcohol 3.198 A.D. Lueking, R.T. Yang: Hydrogen spillover to en-
vapors, J. Phys. Chem. 106, 8994–8999 (2002) hance hydrogen storage – Study of the effect of
3.184 D.H. Yoo, G.H. Rue, M.H.W. Chan, Y.W. Hwang, carbon physicochemical properties, Appl. Catal. A
H.K. Kim: Study of nitrogen adsorbed on open- 265, 259–268 (2004)
108 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.199 G.E. Froudakis: Why alkali-metal-doped carbon 3.216 K. Liu, S. Roth, G.S. Duesberg, G.T. Kim, D. Popa,
nanotubes possess high hydrogen uptake, Nano K. Mukhopadhyay, R. Doome, J. B’Nagy: Antilocal-
Lett. 1, 531–533 (2001) ization in multiwalled carbon nanotubes, Phys. Rev.
3.200 H. Ulbricht, G. Moos, T. Hertel: Physisorption of B 61, 2375–2379 (2000)
molecular oxygen on single-wall carbon nanotube 3.217 G. Fedorov, B. Lassagne, M. Sagnes, B. Raquet,
bundles and graphite, Phys. Rev. B 66, 075404–1– J.M. Broto, F. Triozon, S. Roche, E. Flahaut:
075404–7 (2002) Gate-dependent magnetoresistance phenomena in
3.201 H. Ulbricht, J. Kriebel, G. Moos, T. Hertel: Desorp- carbon nanotubes, Phys. Rev. Lett. 94, 66801–
Part A 3
M.E. McHenry, H.W. Kroto, D.R.M. Walton: Metallic 3.247 A. Loiseau, H. Pascard: Synthesis of long carbon
behaviour of boron-containing carbon nanotubes, nanotubes filled with Se, S, Sb, and Ge by the arc
Chem. Phys. Lett. 323, 572–579 (2000) method, Chem. Phys. Lett. 256, 246–252 (1996)
3.232 R. Czerw, M. Terrones, J.C. Charlier, X. Blasé, B. Foley, 3.248 N. Demoncy, O. Stephan, N. Brun, C. Colliex,
R. Kamalakaran, N. Grobert, H. Terrones, D. Tekleab, A. Loiseau, H. Pascard: Filling carbon nanotubes
P.M. Ajayan, W. Blau, M. Rühle, D.L. Caroll: Identifi- with metals by the arc discharge method: The key
cation of electron donor states, in N-doped carbon role of sulfur, Eur. Phys. J. B 4, 147–157 (1998)
nanotubes, Nano Lett. 1, 457–460 (2001) 3.249 C.H. Kiang, J.S. Choi, T.T. Tran, A.D. Bacher: Molecular
Part A 3
3.233 O. Stephan, P.M. Ajayan, C. Colliex, P. Redlich, nanowires of 1 nm diameter from capillary filling of
J.M. Lambert, P. Bernier, P. Lefin: Doping graphitic single-walled carbon nanotubes, J. Phys. Chem. B
and carbon nanotube structures with boron and 103, 7449–7551 (1999)
nitrogen, Science 266, 1683–1685 (1994) 3.250 Z.L. Zhang, B. Li, Z.J. Shi, Z.N. Gu, Z.Q. Xue, L.M. Peng:
3.234 A. Loiseau, F. Willaime, N. Demoncy, N. Schram- Filling of single-walled carbon nanotubes with sil-
chenko, G. Hug, C. Colliex, H. Pascard: Boron nitride ver, J. Mater. Res. 15, 2658–2661 (2000)
nanotubes, Carbon 36, 743–752 (1998) 3.251 A. Govindaraj, B.C. Satishkumar, M. Nath, C.N.R. Tao:
3.235 C.C. Tang, L.M. de la Chapelle, P. Li, Y.M. Liu, Metal nanowires and intercalated metal layers in
H.Y. Dang, S.S. Fan: Catalytic growth of nanotube single-walled carbon nanotubes bundles, Chem.
and nanobamboo structures of boron nitride, Chem. Mater. 12, 202–205 (2000)
Phys. Lett. 342, 492–496 (2001) 3.252 J. Mittal, M. Monthioux, H. Allouche: Room temper-
3.236 K. Suenaga, C. Colliex, N. Demoncy, A. Loiseau, ature filling of single-wall carbon nanotubes with
H. Pascard, F. Willaime: Synthesis of nanoparticles chromium oxide in open air, Chem. Phys. Lett. 339,
and nanotubes with well separated layers of boron- 311–318 (2001)
nitride and carbon, Science 278, 653–655 (1997) 3.253 E. Dujardin, T.W. Ebbesen, H. Hiura, K. Tanigaki:
3.237 D. Golberg, Y. Bando, L. Bourgeois, K. Kurashima, Capillarity and wetting of carbon nanotubes, Science
T. Sato: Large-scale synthesis and HRTEM analysis 265, 1850–1852 (1994)
of single-walled B- and N-doped carbon nanotube 3.254 E. Flahaut, J. Sloan, K.S. Coleman, V.C. Williams,
bundles, Carbon 38, 2017–2027 (2000) S. Friedrichs, N. Hanson, M.L.H. Green: 1D p-block
3.238 R.S. Lee, J. Gavillet, M. Lamy de la Chapelle, halide crystals confined into single walled carbon
A. Loiseau, J.-L. Cochon, D. Pigache, J. Thibault, nanotubes, Proc. Mater. Res. Soc. Symp., Vol. 633
F. Willaime: Catalyst-free synthesis of boron nitride (2001) pp. A13.15.1–A13.15.6
single-wall nanotubes with a preferred zig-zag 3.255 J. Sloan, A.I. Kirkland, J.L. Hutchison, M.L.H. Green:
configuration, Phys. Rev. B 64, 121405.1–121405.4 Integral atomic layer architectures of 1D crystals in-
(2001) serted into single walled carbon nanotubes, Chem.
3.239 B. Burteaux, A. Claye, B.W. Smith, M. Monthioux, Commun., 1319–1332 (2002)
D.E. Luzzi, J.E. Fischer: Abundance of encapsulated 3.256 J. Sloan, M.C. Novotny, S.R. Bailey, G. Brown, C. Xu,
C60 in single-wall carbon nanotubes, Chem. Phys. V.C. Williams, S. Friedrichs, E. Flahaut, R.L. Cal-
Lett. 310, 21–24 (1999) lender, A.P.E. York, K.S. Coleman, M.L.H. Green,
3.240 D. Ugarte, A. Châtelain, W.A. de Heer: Nanocapillar- R.E. Dunin-Borkowski, J.L. Hutchison: Two layer 4 : 4
ity and chemistry in carbon nanotubes, Science 274, co-ordinated KI crystals grown within single walled
1897–1899 (1996) carbon nanotubes, Chem. Phys. Lett. 329, 61–65
3.241 J. Cook, J. Sloan, M.L.H. Green: Opening and filling (2000)
carbon nanotubes, Fuller. Sci. Technol. 5, 695–704 3.257 G. Brown, S.R. Bailey, J. Sloan, C. Xu, S. Friedrichs,
(1997) E. Flahaut, K.S. Coleman, J.L. Hutchinson, R.E. Dunin-
3.242 P.M. Ajayan, S. Iijima: Capillarity-induced filling of Borkowski, M.L.H. Green: Electron beam induced
carbon nanotubes, Nature 361, 333–334 (1993) in situ clusterisation of 1D ZrCl4 chains within
3.243 P.M. Ajayan, T.W. Ebbesen, T. Ichihashi, S. Iijima, single-walled carbon nanotubes, Chem. Commun.,
K. Tanigaki, H. Hiura: Opening carbon nanotubes 845–846 (2001)
with oxygen and implications for filling, Nature 362, 3.258 J. Sloan, D.M. Wright, H.G. Woo, S. Bai-
522–525 (1993) ley, G. Brown, A.P.E. York, K.S. Coleman,
3.244 S. Seraphin, D. Zhou, J. Jiao, J.C. Withers, R. Loufty: J.L. Hutchison, M.L.H. Green: Capillarity and sil-
Yttrium carbide in nanotubes, Nature 362, 503 (1993) ver nanowire formation observed in single walled
3.245 S. Seraphin, D. Zhou, J. Jiao, J.C. Withers, R. Loufty: carbon nanotubes, Chem. Commun., 699–700
Selective encapsulation of the carbides of yttrium (1999)
and titanium into carbon nanoclusters, Appl. Phys. 3.259 X. Fan, E.C. Dickey, P.C. Eklund, K.A. Williams,
Lett. 63, 2073–2075 (1993) L. Grigorian, R. Buczko, S.T. Pantelides, S.J. Penny-
3.246 R.S. Ruoff, D.C. Lorents, B. Chan, R. Malhotra, cook: Atomic arrangement of iodine atoms inside
S. Subramoney: Single-crystal metals encapsulated single-walled carbon nanotubes, Phys. Rev. Lett.
in carbon nanoparticles, Science 259, 346–348 (1993) 84, 4621–4624 (2000)
110 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.260 G. Brown, S.R. Bailey, M. Novotny, R. Carter, E. Fla- nanotube peapod structures, Science 295, 828–831
haut, K.S. Coleman, J.L. Hutchison, M.L.H. Green, (2002)
J. Sloan: High yield incorporation and washing 3.274 H. Kondo, H. Kino, T. Ohno: Transport properties
properties of halides incorporated into single walled of carbon nanotubes encapsulating C60 and related
carbon nanotubes, Appl. Phys. A 76, 457–462 (2003) materials, Phys. Rev. B 71, 115413 (2005)
3.261 J. Sloan, D.E. Luzzi, A.I. Kirkland, J.L. Hutchi- 3.275 S.H. Jhang, S.W. Lee, D.S. Lee, Y.W. Park, G.H. Jeong,
son, M.L.H. Green: Imaging and characterization T. Hirata, R. Hatakeyama, U. Dettlaff, S. Roth,
of molecules and one-dimensional crystals formed M.S. Kabir, E.E.B. Campbell: Random telegraph noise
Part A 3
within carbon nanotubes, Mater. Res. Soc. Bull. 29, in carbon nanotube peapod transistors, Fuller. Nan-
265–271 (2004) otub. Carbon Nanostruct. 13, 195–198 (2005)
3.262 J. Chancolon, F. Archaimbault, A. Pineau, S. Bon- 3.276 G.H. Jeong, R. Hatakeyama, T. Hirata, K. Tohji,
namy: Confinement of selenium into carbon nano- K. Motomiya, N. Sato, Y. Kawazoe: Structural de-
tubes, Fuller. Nanotub. Carbon Nanostruct. 13, formation of single-walled carbon nanotubes and
189–194 (2005) fullerene encapsulation due to magnetized plasma
3.263 B.W. Smith, M. Monthioux, D.E. Luzzi: Encapsulated ion irradiation, Appl. Phys. Lett. 79, 4213–4215 (2001)
C60 in carbon nanotubes, Nature 396, 323–324 (1998) 3.277 Y.P. Sun, K. Fu, Y. Lin, W. Huang: Functionalized
3.264 B.W. Smith, D.E. Luzzi: Formation mechanism of carbon nanotubes: Properties and applications, Acc.
fullerene peapods and coaxial tubes: A path to large Chem. Res. 35, 1095–1104 (2002)
scale synthesis, Chem. Phys. Lett. 321, 169–174 (2000) 3.278 S. Osswald, E. Flahaut, H. Ye, Y. Gogotsi: Elimination
3.265 K. Hirahara, K. Suenaga, S. Bandow, H. Kato, of D-band in Raman spectra of double-wall carbon
T. Okazaki, H. Shinohara, S. Iijima: One-dimensional nanotubes by oxidation, Chem. Phys. Lett. 402, 422–
metallo-fullerene crystal generated inside single- 427 (2005)
walled carbon nanotubes, Phys. Rev. Lett. 85, 3.279 J. Chen, M.A. Hamon, M. Hui, C. Yongsheng,
5384–5387 (2000) A.M. Rao, P.C. Eklund, R.C. Haddon: Solution prop-
3.266 B.W. Smith, M. Monthioux, D.E. Luzzi: Carbon nano- erties of single-walled carbon nanotubes, Science
tube encapsulated fullerenes: A unique class of 282, 95–98 (1998)
hybrid material, Chem. Phys. Lett. 315, 31–36 (1999) 3.280 J. Chen, A.M. Rao, S. Lyuksyutov, M.E. Itkis, M.A. Ha-
3.267 D.E. Luzzi, B.W. Smith: Carbon cage structures in sin- mon, H. Hu, R.W. Cohn, P.C. Eklund, D.T. Colbert,
gle wall carbon nanotubes: A new class of materials, R.E. Smalley, R.C. Haddon: Dissolution of full-length
Carbon 38, 1751–1756 (2000) single-walled carbon nanotubes, J. Phys. Chem. B
3.268 S. Bandow, M. Takisawa, K. Hirahara, M. Yudasoka, 105, 2525–2528 (2001)
S. Iijima: Raman scattering study of double-wall 3.281 F. Pompeo, D.E. Resasco: Water solubilization of
carbon nanotubes derived from the chains of single-walled carbon nanotubes by functionaliza-
fullerenes in single-wall carbon nanotubes, Chem. tion with glucosamine, Nano Lett. 2, 369–373 (2002)
Phys. Lett. 337, 48–54 (2001) 3.282 Y.P. Sun, W. Huang, Y. Lin, K. Fu, A. Kitay-
3.269 Y. Sakurabayashi, M. Monthioux, K. Kishita, gorodskiy, L.A. Riddle, Y.J. Yu, D.L. Carroll: Soluble
Y. Suzuki, T. Kondo, M. Le Lay: Tayloring double wall dendron-functionalized carbon nanotubes: Prepa-
carbon nanotubes?. In: Molecular Nanostructures, ration, characterization, and properties, Chem.
Am. Inst. Phys. Conf. Proc., Vol. 685, ed. by H. Kuz- Mater. 13, 2864–2869 (2001)
many, J. Fink, M. Mehring, S. Roth (Springer, Berlin 3.283 K. Fu, W. Huang, Y. Lin, L.A. Riddle, D.L. Carroll,
Heidelberg 2003) pp. 302–305 Y.P. Sun: Defunctionalization of functionalized car-
3.270 B.W. Smith, D.E. Luzzi, Y. Achiba: Tumbling atoms bon nanotubes, Nano Lett. 1, 439–441 (2001)
and evidence for charge transfer in La2 @C80 @SWNT, 3.284 P.W. Chiu, G.S. Duesberg, U. Dettlaff-Weglikowska,
Chem. Phys. Lett. 331, 137–142 (2000) S. Roth: Interconnection of carbon nanotubes by
3.271 K. Suenaga, M. Tence, C. Mory, C. Colliex, H. Kato, chemical functionalization, Appl. Phys. Lett. 80,
T. Okazaki, H. Shinohara, K. Hirahara, S. Bandow, 3811–3813 (2002)
S. Iijima: Element-selective single atom imaging, 3.285 T. Fukushima, T. Aida: Ionic liquids for soft func-
Science 290, 2280–2282 (2000) tional materials with carbon nanotubes, Chem. Eur.
3.272 D.E. Luzzi, B.W. Smith, R. Russo, B.C. Satishkumar, J. 13, 5048–5058 (2007)
F. Stercel, N.R.C. Nemes: Encapsulation of metallo- 3.286 Y. Lei, C. Xiong, L. Dong, H. Guo, X. Su, J. Yao, Y. You,
fullerenes and metallocenes in carbon nanotubes, D. Tian, X. Shang: Ionic liquid of ultralong carbon
Proc. Electron. Prop. Nov. Mater. – XVI Int. Winter- nanotubes, Small 3, 1889–1893 (2007)
sch. – AIP Conf. Proc., ed. by H. Kuzmany, J. Fink, 3.287 E.T. Mickelson, C.B. Huffman, A.G. Rinzler,
M. Mehring, S. Roth (Springer, Berlin Heidelberg R.E. Smalley, R.H. Hauge, J.L. Margrave: Fluorina-
2001) pp. 622–626 tion of single-wall carbon nanotubes, Chem. Phys.
3.273 D.J. Hornbaker, S.-J. Kahng, S. Misra, B.W. Smith, Lett. 296, 188–194 (1998)
A.T. Johnson, E.J. Mele, D.E. Luzzi, A. Yazdani: 3.288 V.N. Khabashesku, W.E. Billups, J.L. Margrave: Flu-
Mapping the one-dimensional electronic states of orination of single-wall carbon nanotubes and
Introduction to Carbon Nanotubes References 111
subsequent derivatization reactions, Acc. Chem. Res. wall carbon nanotube films, Appl. Phys. Lett. 73,
35, 1087–1095 (2002) 918–920 (1998)
3.289 P.J. Boul, J. Liu, E.T. Michelson, C.B. Huffman, 3.302 W. Zhu, C. Bower, O. Zhou, G. Kochanski, S. Jin: Large
L.M. Ericson, I.W. Chiang, K.A. Smith, D.T. Colbert, curent density from carbon nanotube field emitters,
R.H. Hauge, J.L. Margrave, R.E. Smalley: Reversible Appl. Phys. Lett. 75, 873–875 (1999)
side-wall functionalization of buckytubes, Chem. 3.303 Y. Saito, R. Mizushima, T. Tanaka, K. Tohji, K. Uchida,
Phys. Lett. 310, 367–372 (1999) M. Yumura, S. Uemura: Synthesis, structure, and
3.290 J.L. Bahr, J. Yang, D.V. Kosynkin, M.J. Bronikowski, field emission of carbon nanotubes, Fuller. Sci.
Part A 3
R.E. Smalley, J.M. Tour: Functionalization of car- Technol. 7, 653–664 (1999)
bon nanotubes by electrochemical reduction of aryl 3.304 J. Kong, N.R. Franklin, C. Zhou, M.G. Chapline,
diazonium salts: A bucky paper electrode, J. Am. S. Peng, K. Cho, H. Dai: Nanotube molecular wire
Chem. Soc. 123, 6536–6542 (2001) as chemical sensors, Science 287, 622–625 (2000)
3.291 M. Holzinger, O. Vostrowsky, A. Hirsch, F. Hennrich, 3.305 P.G. Collins, K. Bradley, M. Ishigami, A. Zettl:
M. Kappes, R. Weiss, F. Jellen: Sidewall functional- Extreme oxygen sensitivity of electronic proper-
ization of carbon nanotubes, Angew. Chem. Int. Ed. ties of carbon nanotubes, Science 287, 1801–1804
40, 4002–4005 (2001) (2000)
3.292 Y. Chen, R.C. Haddon, S. Fang, A.M. Rao, P.C. Eklund, 3.306 H. Chang, J.D. Lee, S.M. Lee, Y.H. Lee: Adsorption of
W.H. Lee, E.C. Dickey, E.A. Grulke, J.C. Pender- NH3 and NO2 molecules on carbon nanotubes, Appl.
grass, A. Chavan, B.E. Haley, R.E. Smalley: Chemical Phys. Lett. 79, 3863–3865 (2001)
attachment of organic functional groups to single- 3.307 C. Cantalini, L. Valentini, L. Lozzi, I. Armentano,
walled carbon nanotube material, J. Mater. Res. 13, J.M. Kenny, S. Santucci: NO2 gas sensitivity of carbon
2423–2431 (1998) nanotubes obtained by plasma enhanced chem-
3.293 C. Velasco-Santos, A.L. Martinez-Hernandez, M. Lo- ical vapor deposition, Sens. Actuators B 93, 333–337
zada-Cassou, A. Alvarez-Castillo, V.M. Castano: (2003)
Chemical functionalization of carbon nanotubes 3.308 J. Li, Y. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan:
through an organosilane, Nanotechnology 13, 495– Carbon nanotubes sensors for gas and organic vapor
498 (2002) detection, Nano Lett. 3, 929–933 (2003)
3.294 A. Star, J.F. Stoddart, D. Steuerman, M. Diehl, 3.309 O.K. Varghese, P.D. Kichambre, D. Gong, K.G. Ong,
A. Boukai, E.W. Wong, X. Yang, S.W. Chung, H. Choi, E.C. Dickey, C.A. Grimes: Gas sensing characteristics
J.R. Heath: Preparation and properties of polymer- of multi-wall carbon nanotubes, Sens. Actuators B
wrapped single-walled carbon nanotubes, Angew. 81, 32–41 (2001)
Chem. Int. Ed. 41, 1721–1725 (2002) 3.310 K.G. Ong, K. Zeng, C.A. Grimes: A wireless, passive
3.295 A. Pénicaud, P. Poulin, A. Derré, E. Anglaret, P. Pe- carbon nanotube-based gas sensor, IEEE Sens. J.
tit: Spontaneous dissolution of a single-wall carbon 2(2), 82–88 (2002)
nanotube salt, J. Am. Chem. Soc. 127, 8–9 (2005) 3.311 J. Kong, M.G. Chapline, H. Dai: Functionalized car-
3.296 R. Stevens, C. Nguyen, A. Cassel, L. Delzeit, M. Meyya- bon nanotubes for molecular hydrogen sensors, Adv.
pan, J. Han: Improved fabrication approach for Mater. 13, 1384–1386 (2001)
carbon nanotube probe devices, Appl. Phys. Lett. 3.312 A. Modi, N. Koratkar, E. Lass, B. Wei, P.M. Ajayan:
77, 3453–3455 (2000) Miniaturized gas ionisation sensors using carbon
3.297 J.H. Hafner, C.L. Cheung, A.T. Wooley, C.M. Lieber: nanotubes, Nature 424, 171–174 (2003)
Structural and functional imaging with carbon 3.313 F. Rodriguez-Reinoso: The role of carbon mater-
nanotube AFM probes, Progr. Biophys. Mol. Biol. 77, ials in heterogeneous catalysis, Carbon 36, 159–175
73–110 (2001) (1998)
3.298 S.S. Wong, E. Joselevich, A.T. Woodley, C.L. Cheung, 3.314 E. Auer, A. Freund, J. Pietsch, T. Tacke: Carbon as
C.M. Lieber: Covalently functionalized nanotubes as support for industrial precious metal catalysts, Appl.
nanometre-size probes in chemistry and biology, Catal. A 173, 259–271 (1998)
Nature 394, 52–55 (1998) 3.315 J.M. Planeix, N. Coustel, B. Coq, B. Botrons,
3.299 C.L. Cheung, J.H. Hafner, C.M. Lieber: Carbon nano- P.S. Kumbhar, R. Dutartre, P. Geneste, P. Bernier,
tube atomic force microscopy tips: Direct growth by P.M. Ajayan: Application of carbon nanotubes as
chemical vapor deposition and application to high- supports in heterogeneous catalysis, J. Am. Chem.
resolution imaging, Proc. Natl. Acad. Sci. USA 97, Soc. 116, 7935–7936 (1994)
3809–3813 (2000) 3.316 P. Serp, M. Corrias, P. Kalck: Carbon nanotubes and
3.300 W.A. de Heer, A. Châtelain, D. Ugarte: A carbon nanofibers in catalysis, Appl. Catal. A 253, 337–358
nanotube field-emission electron source, Science (2003)
270, 1179–1180 (1995) 3.317 K.P. De Jong, J.W. Geus: Carbon nanofibers: catalytic
3.301 J.M. Bonard, J.P. Salvetat, T. Stockli, W.A. de Heer, synthesis and applications, Catal. Rev. 42, 481–510
L. Forro, A. Chatelâin: Field emission from single- (2000)
112 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.318 N.F. Goldshleger: Fullerene and fullerene-based 3.332 H.-B. Chen, J.D. Lin, Y. Cai, X.Y. Wang, J. Yi, J. Wang,
materials in catalysis, Fuller. Sci. Technol. 9, 255– G. Wei, Y.Z. Lin, D.W. Liao: Novel multi-walled
280 (2001) nanotube-supported and alkali-promoted Ru cat-
3.319 X. Pan, Z. Fan, W. Chen, Y. Ding, H. Luo, X. Bao: En- alysts for ammonia synthesis under atmospheric
hanced ethanol production inside carbon-nanotube pressure, Appl. Surf. Sci. 180, 328–335 (2001)
reactors containing catalytic particles, Nat. Mater. 6, 3.333 Y. Zhang, H.B. Zhang, G.D. Lin, P. Chen, Y.Z. Yuan,
507–511 (2007) K.R. Tsai: Preparation, characterization and catalytic
3.320 M.F.R. Pereira, J.L. Figueiredo, J.J.M. Órfão, P. Serp, hydroformylation properties of carbon nanotubes-
Part A 3
P. Kalck, Y. Kihn: Catalytic activity of carbon supported Rh-phosphine catalyst, Appl. Catal. A 187,
nanotubes in the oxidative dehydrogenation of 213–224 (1999)
ethylbenzene, Carbon 42, 2807–2813 (2004) 3.334 M.S. Dresselhaus, K.A. Williams, P.C. Eklund: Hydro-
3.321 G. Mestl, N.I. Maksimova, N. Keller, V.V. Roddatis, gen adsorption in carbon materials, Mater. Res. Soc.
R. Schlögl: Carbon nanofilaments in heterogeneous Bull. 24, 45–50 (1999)
catalysis: An industrial application for new carbon 3.335 H.-M. Cheng, Q.-H. Yang, C. Liu: Hydrogen storage
materials?, Angew. Chem. Int. Ed. Engl. 40, 2066– in carbon nanotubes, Carbon 39, 1447–1454 (2001)
2068 (2001) 3.336 G.G. Tibbetts, G.P. Meisner, C.H. Olk: Hydrogen stor-
3.322 N. Muradov: Catalysis of methane decomposition age capacity of carbon nanotubes, filaments, and
over elemental carbon, Catal. Commun. 2, 89–94 vapor-grown fibers, Carbon 39, 2291–2301 (2001)
(2001) 3.337 F.L. Darkrim, P. Malbrunot, G.P. Tartaglia: Review of
3.323 J.E. Fischer, A.T. Johnson: Electronic properties of hydrogen storage adsorption in carbon nanotubes,
carbon nanotubes, Curr. Opin. Solid State Mater. Sci. Int. J. Hydrogen Energy 27, 193–202 (2002)
4, 28–33 (1999) 3.338 G.E. Froudakis: Hydrogen interaction with carbon
3.324 M. Menon, A.N. Andriotis, G.E. Froudakis: Curvature nanotubes: a review of ab initio studies, J. Phys.
dependence of the metal catalyst atom interaction Condens. Matter 14, R453–R465 (2002)
with carbon nanotubes walls, Chem. Phys. Lett. 320, 3.339 M. Hirscher, M. Becher: Hydrogen storage in car-
425–434 (2000) bon nanotubes, J. Nanosci. Nanotechnol. 3(1/2), 3–17
3.325 N. Ishigami, H. Ago, Y. Motoyama, M. Takasaki, (2003)
M. Shinagawa, K. Takahashi, T. Ikuta, M. Tsuji: 3.340 P. Kowalczyk, R. Holyst, M. Terrones, H. Terrones:
Microreactor utilizing a vertically-aligned carbon Hydrogen storage in nanoporous carbon materials:
nanotube array grown inside the channels, Chem. myth and facts, PhysChemChemPhys 9(15), 1786–1792
Commun., 1626 (2007) (2007)
3.326 G.G. Wildgoose, C.E. Banks, R.G. Compton: Metal 3.341 C. Park, P.E. Anderson, C.D. Tan, R. Hidalgo,
nanoparticles and related materials supported on N. Rodriguez: Further studies of the interaction of
carbon nanotubes: methods and applications, Small hydrogen with graphite nanofibers, J. Phys. Chem.
2, 182–193 (2006) B 103, 10572–10581 (1999)
3.327 T. Kyotani, S. Nakazaki, W.-H. Xu, A. Tomita: Chem- 3.342 S.M. Lee, H.Y. Lee: Hydrogen storage in single-
ical modification of the inner walls of carbon walled carbon nanotubes, Appl. Phys. Lett. 76,
nanotubes by HNO3 oxidation, Carbon 39, 782–785 2877–2879 (2000)
(2001) 3.343 X. Zhang, D. Cao, J. Chen: Hydrogen adsorption stor-
3.328 Z.J. Liu, Z.Y. Yuan, W. Zhou, L.M. Peng, Z. Xu: age on single-walled carbon nanotube arrays by
Co/carbon nanotubes monometallic system: The ef- a combination of classical potential and density
fects of oxidation by nitric acid, PhysChemChemPhys functional theory, J. Phys. Chem. B 107, 4942–4950
3, 2518–2521 (2001) (2003)
3.329 R. Giordano, P. Serp, P. Kalck, Y. Kihn, J. Schreiber, 3.344 H.M. Cheng, G.P. Pez, A.C. Cooper: Mechanism of hy-
C. Marhic, J.-L. Duvail: Preparation of rhodium sup- drogen sorption in single-walled carbon nanotubes,
ported on carbon canotubes catalysts via surface J. Am. Chem. Soc. 123, 5845–5846 (2001)
mediated organometallic reaction, Eur. J. Inorg. 3.345 C.-H. Chen, C.-C. Huang: Hydrogen storage by
Chem. 2003, 610–617 (2003) KOH-modified multi-walled carbon nanotubes, Int.
3.330 A. Carrillo, J.A. Swartz, J.M. Gamba, R.S. Kane, J. Hydrogen Energy 32, 237–246 (2007)
N. Chakrapani, B. Wei, P.M. Ajayan: Noncovalent 3.346 M.A. de la Casa-Lillo, F. Lamari-Darkrim, D. Cazorla-
functionalization of graphite and carbon nanotubes Amoros, A. Linares-Solano: Hydrogen storage in
with polymer multilayers and gold nanoparticles, activated carbons and activated carbon fibers,
Nano Lett. 3, 1437–1440 (2003) J. Phys. Chem. B 106, 10930–10934 (2002)
3.331 Z. Liu, X. Lin, J.Y. Lee, W. Zhang, M. Han, 3.347 P. Marinelli, R. Pellenq, J. Conard: H stocké dans les
L.M. Gan: Preparation and characterization of carbones un site légèrement métastable, Natl. Conf.
platinum-based electrocatalysts on multiwalled Mater., Tours (2002), AF-14-020
carbon nanotubes for proton exchange membrane 3.348 G. Mpourmpakis, G.E. Froudakis, G.P. Lithoxoos,
fuel cells, Langmuir 18, 4054–4060 (2002) J. Samios: Effect of curvature and chirality for hy-
Introduction to Carbon Nanotubes References 113
drogen storage in single-walled carbon nanotubes: 3.363 C. Lu, F. Su: Adsorption of natural organic matter by
a combined ab initio and Monte Carlo investigation, carbon nanotubes, Sep. Purif. Technol. 58, 113–121
J. Chem. Phys. 126, 144704 (2007) (2007)
3.349 C.I. Weng, S.P. Ju, K.C. Fang, F.P. Chang: Atom- 3.364 C. Ye, Q.-M. Gong, F.-P. Lu: Adsorption of uraemic
istic study of the influences of size, VDW distance toxins on carbon nanotubes, Sep. Purif. Technol. 58,
and arrangement of carbon nanotubes on hy- 2–6 (2007)
drogen storage, Comput. Mater. Sci. 40, 300–308 3.365 X. Peng, Y. Li, Z. Luan, Z. Di, H. Wang, B. Tian, Z. Jia:
(2007) Adsorption of 1,2-dichlorobenzene from water to
Part A 3
3.350 A.L.M. Reddy, S. Ramaprabhu: Hydrogen storage carbon nanotubes, Chem. Phys. Lett. 376, 154–158
properties of nanocrystalline Pt dispersed multi- (2003)
walled carbon nanotubes, Int. J. Hydrogen Energy 3.366 P. Kondratyuk, J.T. Yates: Nanotubes as molecular
32, 3998–4004 (2007) sponges: the adsorption of CCl4 , Chem. Phys. Lett.
3.351 A. Kusnetzova, D.B. Mawhinney, V. Naumenko, 383, 314–316 (2004)
J.T. Yates, J. Liu, R.E. Smalley: Enhancement of 3.367 C. Lu, H. Chiu: Adsorption of zinc(II) from water with
adsorption inside of single-walled nanotubes: purified carbon nanotubes, Chem. Eng. Sci. 61, 1138
Opening the entry ports, Chem. Phys. Lett. 321, 292– (2006)
296 (2000) 3.368 A. Stafiej, K. Pyrzynska: Adsorption of heavy metal
3.352 G.E. Gadd, M. Blackford, S. Moricca, N. Webb, ions with carbon nanotubes, Sep. Purif. Technol. 58,
P.J. Evans, A.M. Smith, G. Jacobsen, S. Leung, A. Day, 49–52 (2007)
Q. Hua: The world’s smallest gas cylinders?, Science 3.369 H. Wang, A. Zhou, F. Peng, H. Yu, J. Yang: Mechanism
277, 933–936 (1997) study on adsorption of acidified multiwalled carbon
3.353 Z. Mao, S.B. Sinnott: A computational study of nanotubes to Pb(II), J. Colloid Interf. Sci. 316, 277–283
molecular diffusion and dynamic flow through car- (2007)
bon nanotubes, J. Phys. Chem. B 104, 4618–4624 3.370 C. Chen, X. Li, D. Zhao, X. Tan, X. Wang: Adsorp-
(2000) tion kinetic, thermodynamic and desorption studies
3.354 Z. Mao, S.B. Sinnott: Separation of organic molecular of Th(IV) on oxidized multi-wall carbon nanotubes,
mixtures in carbon nanotubes and bundles: Mo- Colloids Surf. A Physicochem. Eng. Asp. 302, 449–454
lecular dynamics simulations, J. Phys. Chem. B 105, (2007)
6916–6924 (2001) 3.371 A. Huczko, H. Lange, E. Calko, H. Grubek-Jaworska,
3.355 H. Chen, D.S. Sholl: Rapid diffusion of CH4 /H2 mix- P. Droszcz: Physiological testing of carbon nano-
tures in single-walled carbon nanotubes, J. Am. tubes: Are they asbestos-like?, Fuller. Sci. Technol.
Chem. Soc. 126, 7778–7779 (2004) 9, 251–254 (2001)
3.356 C. Gu, G.-H. Gao, Y.X. Yu, T. Nitta: Simulation 3.372 A.A. Shvedova, V. Castranova, E.R. Kisin, D. Schweg-
for separation of hydrogen and carbon monoxide ler-Berry, A.R. Murray, V.Z. Gandelsman, A.M. May-
by adsorption on single-walled carbon nanotubes, nard, P. Baron: Exposure to carbon nanotube
Fluid Phase Equilib. 194/197, 297–307 (2002) material: assessment of nanotube cytotoxicity using
3.357 R.Q. Long, R.T. Yang: Carbon nanotubes as superior human keratinocyte cells, Toxical Environ. Health A
sorbent for dioxine removal, J. Am. Chem. Soc. 123, 66, 1909–1926 (2003)
2058–2059 (2001) 3.373 C.W. Lam, J.T. James, R. McCluskey, R.L. Hunter: Pul-
3.358 Y.H. Li, S. Wang, A. Cao, D. Zhao, X. Zhang, C. Xu, monary toxicity of single-wall carbon nanotubes in
Z. Luan, D. Ruan, J. Liang, D. Wu, B. Wei: Adsorption mice 7 and 90 days after intratracheal instillation,
of fluoride from water by amorphous alumina sup- Toxicol. Sci. 77, 126–134 (2004)
ported on carbon nanotubes, Chem. Phys. Lett. 350, 3.374 D. Pantarotto, J.P. Briand, M. Prato, A. Bianco:
412–416 (2001) Translocation of bioactive peptides across cell mem-
3.359 Y.H. Li, S. Wang, J. Wei, X. Zhang, C. Xu, Z. Luan, branes by carbon nanotubes, Chem. Commun., 16–17
D. Wu, B. Wei: Lead adsorption on carbon nano- (2004)
tubes, Chem. Phys. Lett. 357, 263–266 (2002) 3.375 C. Salvador-Morales, E. Flahaut, E. Sim, J. Sloan,
3.360 C. Park, E.S. Engel, A. Crowe, T.R. Gilbert, N.M. Ro- M.L.H. Green, R.B. Sim: Complement activation and
driguez: Use of carbon nanofibers in the removal protein adsorption by carbon nanotubes, Mol. Im-
of organic solvents from water, Langmuir 16, 8050– mun. 43, 193–201 (2006)
8056 (2000) 3.376 M.P. Mattson, R.C. Haddon, A.M. Rao: Molecular
3.361 E. Diaz, S. Ordonez, A. Vega: Adsorption of volatile functionalization of carbon nanotubes and use as
organic compounds onto carbon nanotubes, car- substrates for neuronal growth, J. Mol. Neurosci. 14,
bon nanofibers, and high-surface-area graphites, 175–182 (2000)
J. Colloid Interf. Sci. 305, 7–16 (2007) 3.377 J.J. Davis, M.L.H. Green, H.A.O. Hill, Y.C. Leung,
3.362 C.-H. Wu: Adsorption of reactive dye onto carbon P.J. Sadler, J. Sloan, A.V. Xavier, S.C. Tsang: The im-
nanotubes: equilibrium, kinetics and thermody- mobilization of proteins in carbon nanotubes, Inorg.
namics, J. Hazard. Mater. 144, 93–100 (2007) Chim. Acta 272, 261–266 (1998)
114 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.378 R.J. Chen, Y. Zhang, D. Wang, H. Dai: Noncovalent 3.394 C.F. Deng, Y.X. Ma, P. Zhang, X.X. Zhang, D.Z. Wang:
sidewall functionalization of single-walled carbon Thermal expansion behaviors of aluminum compos-
nanotubes for protein immobilization, J. Am. Chem. ite reinforced with carbon nanotubes, Mater. Lett.
Soc. 123, 3838–3839 (2001) 62, 2301–2303 (2008)
3.379 M. Shim, N.W.S. Kam, R.J. Chen, Y. Li, H. Dai: 3.395 T. Kuzumaki, O. Ujiie, H. Ichinose, K. Ito: Mechanical
Functionalization of carbon nanotubes for biocom- characteristics and preparation of carbon nanotube
patibility and biomolecular recognition, Nano Lett. fiber-reinforced Ti composite, Adv. Eng. Mater. 2,
2, 285–288 (2002) 416–418 (2000)
Part A 3
3.380 C. Dwyer, M. Guthold, M. Falvo, S. Washburn, R. Su- 3.396 E. Carreno-Morelli, J. Yang, E. Couteau, K. Hernadi,
perfine, D. Erie: DNA-functionalized single-walled J.W. Seo, C. Bonjour, L. Forro, R. Schaller: Carbon
carbon nanotubes, Nanotechnology 13, 601–604 nanotube/magnesium composites, Phys. Status So-
(2002) lidi (a) 201, R53–R55 (2004)
3.381 H. Huang, S. Taylor, K. Fu, Y. Lin, D. Zhang, 3.397 Z. Bian, R.J. Wang, W.H. Wang, T. Zhang, A. Inoue:
T.W. Hanks, A.M. Rao, Y. Sun: Attaching proteins to Carbon-nanotube-reinforced Zr-based bulk metal-
carbon nanotubes via diimide-activated amidation, lic glass composites and their properties, Adv. Funct.
Nano Lett. 2, 311–314 (2002) Mater. 14, 55–63 (2004)
3.382 C.V. Nguyen, L. Delzeit, A.M. Cassell, J. Li, J. Han, 3.398 S.R. Dong, J.P. Tu, X.B. Zhang: An investigation of
M. Meyyappan: Preparation of nucleic acid func- the sliding wear behavior of Cu-matrix composite
tionalized carbon nanotube arrays, Nano Lett. 2, reinforced by carbon nanotubes, Mater. Sci. Eng. A
1079–1081 (2002) 313, 83–87 (2001)
3.383 B.R. Azamian, J.J. Davis, K.S. Coleman, C.B. Bag- 3.399 J. Wang, G. Chen, M. Wang, M.P. Chatrathi: Carbon-
shaw, M.L.H. Green: Bioelectrochemical single- nanotube/copper composite electrodes for capillary
walled carbon nanotubes, J. Am. Chem. Soc. 124, electrophoresis microchip detection of carbohy-
12664–12665 (2002) drates, Analyst (Cambridge) 129, 512–515 (2004)
3.384 E. Katz, I. Willner: Biomolecule-functionalized car- 3.400 C.S. Goh, J. Wei, L.C. Lee, M. Gupta: Simultaneous
bon nanotubes: Applications in nanobioelectronics, enhancement in strength and ductility reinforcing
ChemPhysChem 5, 1084–1104 (2004) magnesium with carbon nanotubes, Mater. Sci. Eng.
3.385 J. Wang: Carbon-nanotube based electrochemical A 423, 153–156 (2006)
biosensors: a review, Electroanalysis 17, 7–14 (2005) 3.401 Q. Ngo, B.A. Cruden, A.M. Cassell, M.D. Walker, Q. Ye,
3.386 T. Laha, A. Agarwal, T. McKechnie, S. Seal: Synthesis J.E. Koehne, M. Meyyappan, J. Li, C.Y. Yang: Thermal
and characterization of plasma spray formed carbon conductivity of carbon nanotube composite films,
nanotube reinforced aluminum composite, Mater. Mater. Res. Soc. Symp. Proc., Vol. 812 (2004) pp. 179–
Sci. Eng. A 381, 249–258 (2004) 184
3.387 T. Noguchi, A. Magario, S. Fukazawa, S. Shimizu, 3.402 X.H. Chen, C.S. Chen, H.N. Xiao, F.Q. Cheng, G. Zhang,
J. Beppu, M. Seki: Carbon nanotube/aluminium G.J. Yi: Corrosion behavior of carbon nanotubes-Ni
composites with uniform dispersion, Mater. Trans. composite coating, Surf. Coat. Technol. 191, 351–356
45, 602–604 (2004) (2005)
3.388 Y.B. Li, Q. Ya, B.Q. Wei, J. Liang, D.H. Wu: Process- 3.403 X.H. Chen, C.S. Chen, H.N. Xiao, H.B. Liu, L.P. Zhou,
ing of a carbon nanotubes-Fe82 P18 metallic glass S.L. Li, G. Zhang: Dry friction and wear characteris-
composite, J. Mater. Sci. Lett. 17, 607–609 (1998) tics of nickel/carbon nanotube electroless composite
3.389 A. Goyal, D.A. Wiegand, F.J. Owens, Z. Iqbal: En- deposits, Tribol. Int. 39, 22–28 (2006)
hanced yield strength in iron nanocomposite with in 3.404 Z. Yang, H. Xu, M.-K. Li, Y.-L. Shi, Y. Huang, H.-L. Li:
situ grown single-wall carbon nanotubes, J. Mater. Preparation and properties of Ni-P/single-walled
Res. 21, 522–528 (2006) carbon nanotubes composite coatings by means
3.390 K.T. Kim, K.H. Lee, S.I. Cha, C.-B. Mo, S.H. Hong: of electroless plating, Thin Solid Films 466, 86–91
Characterization of carbon nanotubes/Cu nanocom- (2004)
posites processed by using nano-sized Cu powders, 3.405 A. Peigney, C. Laurent: Carbon nanotubes ceramic
Mater. Res. Soc. Symp. Proc. 821, 111–116 (2004) composites. In: Ceramic Matrix Composites: Mi-
3.391 F. Zhang, J. Shen, J. Sun: Processing and proper- crostructure, Properties and Applications, ed. by
ties of carbon nanotubes-nano-WC-Co composites, I.M. Low (Woodhead, Cambridge 2006) pp. 309–333
Mater. Sci. Eng. A 381, 86–91 (2004) 3.406 C. Laurent, A. Peigney, O. Dumortier, A. Rousset: Car-
3.392 C.L. Xu, B.Q. Wei, R.Z. Ma, J. Liang, X.K. Ma, D.H. Wu: bon nanotubes-Fe-alumina nanocomposites. Part
Fabrication of aluminum-carbon nanotube com- II: Microstructure and mechanical properties of the
posites and their electrical properties, Carbon 37, hot-pressed composites, J. Eur. Ceram. Soc. 18,
855–858 (1999) 2005–2013 (1998)
3.393 T. Kuzumaki, K. Miyazawa, H. Ichinose, K. Ito: Pro- 3.407 A. Peigney, C. Laurent, E. Flahaut, A. Rousset: Carbon
cessing of carbon nanotube reinforced aluminum nanotubes in novel ceramic matrix nanocomposites,
composite, J. Mater. Res. 13, 2445–2449 (1998) Ceram. Int. 26, 677–683 (2000)
Introduction to Carbon Nanotubes References 115
3.408 E. Flahaut, A. Peigney, C. Laurent, C. Marlière, nanotube/ceramic composites, Scr. Mater. 58, 223–
F. Chastel, A. Rousset: Carbon nanotube-metal- 226 (2008)
oxide nanocomposites: Microstructure, electrical 3.424 G.-D. Zhan, J.D. Kuntz, H. Wang, C.-M. Wang,
conductivity and mechanical properties, Acta Mater. A.K. Mukherjee: Anisotropic thermal properties of
48, 3803–3812 (2000) single-wall-carbon-nanotube-reinforced nanoce-
3.409 J.W. An, D.H. You, D.S. Lim: Tribological properties ramics, Philos. Mag. Lett. 84, 419–423 (2004)
of hot-pressed alumina-CNT composites, Wear 255, 3.425 Q. Huang, L. Gao, Y. Liu, J. Sun: Sintering
677–681 (2003) and thermal properties of multiwalled carbon
Part A 3
3.410 J. Ning, J. Zhang, Y. Pan, J. Guo: Surfactants assisted nanotube-BaTiO3 composites, J. Mater. Chem. 15,
processing of carbon nanotube-reinforced SiO2 ma- 1995–2001 (2005)
trix composites, Ceram. Int. 30, 63–67 (2004) 3.426 G.-D. Zhan, J.D. Kuntz, J.E. Garay, A.K. Mukherjee:
3.411 Q. Huang, L. Gao: Manufacture and electrical Electrical properties of nanoceramics reinforced with
properties of multiwalled carbon nanotube/BaTiO3 ropes of single-walled carbon nanotubes, Appl.
nanocomposite ceramics, J. Mater. Chem. 14, 2536– Phys. Lett. 83, 1228–1230 (2003)
2541 (2004) 3.427 S. Rul, F. Lefevre-Schlick, E. Capria, C. Laurent,
3.412 J. Fan, D. Zhao, M. Wu, Z. Xu, J. Song: Prepa- A. Peigney: Percolation of single-walled carbon
ration and microstructure of multi-wall carbon nanotubes in ceramic matrix nanocomposites, Acta
nanotubes-toughened Al2 O3 composite, J. Am. Ce- Mater. 52, 1061–1067 (2004)
ram. Soc. 89, 750–753 (2006) 3.428 S.-L. Shi, J. Liang: Electronic transport proper-
3.413 A. Peigney, S. Rul, F. Lefevre-Schlick, C. Laurent: ties of multiwall carbon nanotubes/yttria-stabilized
Densification during hot-pressing of carbon nano- zirconia composites, J. Appl. Phys. 101, 023708–5
tube metal-ceramic composites, J. Eur. Ceram. Soc. (2007)
27, 2183–2193 (2007) 3.429 A. Peigney, E. Flahaut, C. Laurent, F. Chas-
3.414 J. Sun, L. Gao, W. Li: Colloidal processing of car- tel, A. Rousset: Aligned carbon nanotubes in
bon nanotube/alumina composites, Chem. Mater. ceramic-matrix nanocomposites prepared by high-
14, 5169–5172 (2002) temperature extrusion, Chem. Phys. Lett. 352, 20–25
3.415 G.D. Zhan, J.D. Kuntz, J. Wan, A.K. Mukherjee: (2002)
Single-wall carbon nanotubes as attractive tough- 3.430 G.-D. Zhan, J.D. Kuntz, A.K. Mukherjee, P. Zhu,
ening agents in alumina-based composites, Nat. K. Koumoto: Thermoelectric properties of carbon
Mater. 2, 38–42 (2003) nanotube/ceramic nanocomposites, Scr. Mater. 54,
3.416 S.I. Cha, K.T. Kim, K.H. Lee, C.B. Mo, S.H. Hong: 77–82 (2006)
Strengthening and toughening of carbon nano- 3.431 P.M. Ajayan, O. Stephan, C. Colliex, D. Trauth:
tube reinforced alumina nanocomposite fabricated Aligned carbon nanotube arrays formed by cutting
by molecular level mixing process, Scr. Mater. 53, a polymer resin-nanotube composite, Science 265,
793–797 (2005) 1212–1214 (1994)
3.417 C.B. Mo, S.I. Cha, K.T. Kim, K.H. Lee, S.H. Hong: 3.432 J.N. Coleman, U. Khan, Y.K. Gun’ko: Mechanical re-
Fabrication of carbon nanotube reinforced alumina inforcement of polymers using carbon nanotubes,
matrix nanocomposite by sol-gel process, Mater. Sci. Adv. Mater. 18, 689–706 (2006)
Eng. A 395, 124–128 (2005) 3.433 R. Haggenmueller, H.H. Gommans, A.G. Rinzler,
3.418 X. Wang, N.P. Padture, H. Tanaka: Contact-damage- J.E. Fischer, K.I. Winey: Aligned single-wall car-
resistant ceramic/single-wall carbon nanotubes and bon nanotubes in composites by melt processing
ceramic/graphite composites, Nat. Mater. 3, 539– methods, Chem. Phys. Lett. 330, 219–225 (2000)
544 (2004) 3.434 L.S. Schadler, S.C. Giannaris, P.M. Ajayan: Load
3.419 W.A. Curtin, B.W. Sheldon: CNT-reinforced ceramics transfer in carbon nanotube epoxy composites,
and metals, Mater. Today 7, 44–49 (2004) Appl. Phys. Lett. 73, 3842–3844 (1998)
3.420 Z. Xia, L. Riester, W.A. Curtin, H. Li, B.W. Sheldon, 3.435 S.J.V. Frankland, A. Caglar, D.W. Brenner, M. Griebel:
J. Liang, B. Chang, J.M. Xu: Direct observation of Molecular simulation of the influence of chem-
toughening mechanisms in carbon nanotube ce- ical cross-links on the shear strength of carbon
ramic matrix composites, Acta Mater. 52, 931–944 nanotube-polymer interfaces, J. Phys. Chem. B 106,
(2004) 3046–3048 (2002)
3.421 D.S. Lim, J.W. An, H.J. Lee: Effect of carbon 3.436 H.D. Wagner: Nanotube-polymer adhesion: A mech-
nanotube addition on the tribological behavior of anics approach, Chem. Phys. Lett. 361, 57–61 (2002)
carbon/carbon composites, Wear 252, 512–517 (2002) 3.437 P.M. Ajayan, L.S. Schadler, C. Giannaris, A. Rubio:
3.422 D.-S. Lim, D.-H. You, H.-J. Choi, S.-H. Lim, H. Jang: Single-walled carbon nanotube-polymer compos-
Effect of CNT distribution on tribological behavior of ites: Strength and weakness, Adv. Mater. 12, 750–753
alumina-CNT composites, Wear 259, 539–544 (2005) (2000)
3.423 Z.H. Xia, J. Lou, W.A. Curtin: A multiscale ex- 3.438 X. Gong, J. Liu, S. Baskaran, R.D. Voise, J.S. Young:
periment on the tribological of aligned carbon Surfactant-assisted processing of carbon nano-
116 Part A Nanostructures, Micro-/Nanofabrication and Materials
tube/polymer composites, Chem. Mater. 12, 1049– 3.452 J.M. Benoit, B. Corraze, S. Lefrant, W.J. Blau,
1052 (2000) P. Bernier, O. Chauvet: Transport properties of
3.439 E.T. Thostenson, W.Z. Li, D.Z. Wang, Z.F. Ren, PMMA-carbon nanotubes composites, Synth. Met.
T.W. Chou: Carbon nanotube/carbon fiber hybrid 121, 1215–1216 (2001)
multiscale composites, J. Appl. Phys. 91, 6034–6037 3.453 J.M. Benoit, B. Corraze, O. Chauvet: Localization,
(2002) Coulomb interactions, and electrical heating in
3.440 F.H. Gojny, M.H.G. Wichmann, U. Kopke, B. Fiedler, single-wall carbon nanotubes/polymer composites,
K. Schulte: Carbon nanotube-reinforced epoxy- Phys. Rev. B 65, 241405/1–241405/4 (2002)
Part A 3
composites: enhanced stiffness and fracture tough- 3.454 F. Du, R.C. Scogna, W. Zhou, S. Brand, J.E. Fis-
ness at low nanotube content, Compos. Sci. Technol. cher, K.I. Winey: Nanotube networks in polymer
64, 2363–2371 (2004) nanocomposites: Rheology and electrical conduc-
3.441 F.H. Gojny, M.H.G. Wichmann, B. Fiedler, K. Schulte: tivity, Macromolecules 37, 9048–9055 (2004)
Influence of different carbon nanotubes on the me- 3.455 T. Kashiwagi, F. Du, K.I. Winey, K.M. Groth,
chanical properties of epoxy matrix composites – J.R. Shields, R.H. Harris Jr., J.F. Douglas: Flamma-
A comparative study, Compos. Sci. Tech. 65, 2300– bility properties of PMMA-single walled carbon
2313 (2005) nanotube nanocomposites, Polym. Mater. Sci. Eng.
3.442 H. Rajoria, N. Jalili: Passive vibration damping 91, 90–91 (2004)
enhancement using carbon nanotube-epoxy rein- 3.456 B. Vigolo, A. Pénicaud, C. Coulon, C. Sauder,
forced composites, Compos. Sci. Tech. 65, 2079–2093 R. Pailler, C. Journet, P. Bernier, P. Poulin: Macro-
(2005) scopic fibers and ribbons of oriented carbon
3.443 M.J. Biercuk, M.C. Llaguno, M. Radosavljevic, nanotubes, Science 290, 1331–1334 (2000)
J.K. Hyun, A.T. Johnson, J.E. Fischer: Carbon nano- 3.457 B. Vigolo, P. Poulin, M. Lucas, P. Launois, P. Bernier:
tube composites for thermal management, Appl. Improved structure and properties of single-wall
Phys. Lett. 80, 2767–2769 (2002) carbon nanotube spun fibers, Appl. Phys. Lett. 11,
3.444 F.H. Gojny, K. Schulte: Functionalisation effect on 1210–1212 (2002)
the thermo-mechanical behaviour of multi-wall 3.458 P. Poulin, B. Vigolo, P. Launois: Films and fibers of
carbon nanotube/epoxy-composites, Compos. Sci. oriented single wall nanotubes, Carbon 40, 1741–
Technol. 64, 2303–2308 (2004) 1749 (2002)
3.445 G. Pecastaings, P. Delhaes, A. Derre, H. Saadaoui, 3.459 K. Jiang, Q. Li, S. Fan: Spinning continuous carbon
F. Carmona, S. Cui: Role of interfacial effects in nanotube yarn, Nature 419, 801 (2002)
carbon nanotube/epoxy nanocomposite behavior, 3.460 M. Zhang, K.R. Atkinson, R.H. Baughman: Multi-
J. Nanosci. Nanotechnol. 4, 838–843 (2004) functional carbon nanotube yarns by downsizing an
3.446 S. Barrau, P. Demont, A. Peigney, C. Laurent, C. La- ancient technology, Science 306, 1356–1361 (2004)
cabanne: Effect of palmitic acid on the electrical 3.461 J. Steinmetz, M. Glerup, M. Paillet, P. Bernier,
conductivity of carbon nanotubes-polyepoxy com- M. Holzinger: Production of pure nanotube fibers
posite, Macromolecules 36, 9678–9680 (2003) using a modified wet-spinning method, Carbon 43,
3.447 S. Barrau, P. Demont, A. Peigney, C. Laurent, 2397–2400 (2005)
C. Lacabanne: DC and AC conductivity of carbon 3.462 A.B. Dalton, S. Collins, E. Munoz, J.M. Razal,
nanotubes-polyepoxy composites, Macromolecules V.H. Ebron, J.P. Ferraris, J.N. Coleman, B.G. Kim,
36, 5187–5194 (2003) R.H. Baughman: Super-tough carbon-nanotube
3.448 J. Sandler, M.S.P. Shaffer, T. Prasse, W. Bauhofer, fibers, Nature 423, 703 (2003)
K. Schulte, A.H. Windle: Development of a dispersion 3.463 M.S.P. Shaffer, A.H. Windle: Fabrication and char-
process for carbon nanotubes in an epoxy matrix acterization of carbon nanotube/poly(vinyl alcohol)
and the resulting electrical properties, Polymer 40, composites, Adv. Mater. 11, 937–941 (1999)
5967–5971 (1999) 3.464 L. Jin, C. Bower, O. Zhou: Alignment of carbon nano-
3.449 Z. Jin, K.P. Pramoda, G. Xu, S.H. Goh: Dynamic tubes in a polymer matrix by mechanical stretching,
mechanical behavior of melt-processed multi- Appl. Phys. Lett. 73, 1197–1199 (1998)
walled carbon nanotube/poly(methyl methacrylate) 3.465 H.D. Wagner, O. Lourie, Y. Feldman, R. Tenne:
composites, Chem. Phys. Lett. 337, 43–47 (2001) Stress-induced fragmentation of multiwall carbon
3.450 Z. Jin, K.P. Pramoda, S.H. Goh, G. Xu: Poly(vinylidene nanotubes in a polymer matrix, Appl. Phys. Lett. 72,
fluoride)-assisted melt-blending of multi-walled 188–190 (1998)
carbon nanotube/poly(methyl methacrylate) com- 3.466 H.D. Wagner, O. Lourie, X.F. Zhou: Macrofrag-
posites, Mater. Res. Bull. 37, 271–278 (2002) mentation and microfragmentation phenomena in
3.451 C.A. Cooper, D. Ravich, D. Lips, J. Mayer, H.D. Wag- composite materials, Compos. Part A 30, 59–66
ner: Distribution and alignment of carbon nano- (1998)
tubes and nanofibrils in a polymer matrix, Compos. 3.467 J.R. Wood, Q. Zhao, H.D. Wagner: Orientation of car-
Sci. Technol. 62, 1105–1112 (2002) bon nanotubes in polymers and its detection by
Introduction to Carbon Nanotubes References 117
Raman spectroscopy, Compos. Part A 32, 391–399 3.480 I. Musa, M. Baxendale, G.A.J. Amaratunga,
(2001) W. Eccleston: Properties of regular poly(3-octyl-
3.468 Q. Zhao, J.R. Wood, H.D. Wagner: Using carbon thiophene)/multi-wall carbon nanotube compos-
nanotubes to detect polymer transitions, J. Polym. ites, Synth. Met. 102, 1250 (1999)
Sci. B 39, 1492–1495 (2001) 3.481 E. Kymakis, I. Alexandou, G.A.J. Amaratunga:
3.469 M. Cochet, W.K. Maser, A.M. Benito, M.A. Callejas, Single-walled carbon nanotube-polymer compos-
M.T. Martinesz, J.M. Benoit, J. Schreiber, O. Chauvet: ites: Electrical, optical and structural investigation,
Synthesis of a new polyaniline/nanotube composite: Synth. Met. 127, 59–62 (2002)
Part A 3
In-situ polymerisation and charge transfer through 3.482 K. Yoshino, H. Kajii, H. Araki, T. Sonoda, H. Take,
site-selective interaction, Chem. Commun., 1450– S. Lee: Electrical and optical properties of conducting
1451 (2001) polymer-fullerene and conducting polymer-carbon
3.470 D. Qian, E.C. Dickey, R. Andrews, T. Rantell: Load nanotube composites, Fuller. Sci. Technol. 7, 695–711
transfer and deformation mechanisms in carbon (1999)
nanotube-polystyrene composites, Appl. Phys. Lett. 3.483 S.A. Curran, P.M. Ajayan, W.J. Blau, D.L. Carroll,
76, 2868–2870 (2000) J.N. Coleman, A.B. Dalton, A.P. Davey, A. Drury,
3.471 V. Datsyuk, C. Guerret-Piecourt, S. Dagreou, L. Bil- B. McCarthy, S. Maier, A. Strevens: A composite
lon, J.-C. Dupin, E. Flahaut, A. Peigney, C. Laurent: from poly(m-phenylenevinylene-co-2,5-dioctoxy-
Double walled carbon nanotube/polymer compos- p-phenylenevinylene) and carbon nanotubes.
ites via in-situ nitroxide mediated polymerisation of A novel material for molecular optoelectronics, Adv.
amphiphilic block copolymers, Carbon 43, 873–876 Mater. 10, 1091–1093 (1998)
(2005) 3.484 P. Fournet, D.F. O’Brien, J.N. Coleman, H.H. Horhold,
3.472 R. Blake, Y.K. Gun’ko, J. Coleman, M. Cadek, A. Fon- W.J. Blau: A carbon nanotube composite as an
seca, J.B. Nagy, W.J. Blau: A generic organometallic electron transport layer for M3EH-PPV based light-
approach toward ultra-strong carbon nanotube emitting diodes, Synth. Met. 121, 1683–1684 (2001)
polymer composites, J. Am. Chem. Soc. 126, 10226– 3.485 H.S. Woo, R. Czerw, S. Webster, D.L. Carroll, J. Bal-
10227 (2004) lato, A.E. Strevens, D. O’Brien, W.J. Blau: Hole
3.473 T. Kashiwagi, E. Grulke, J. Hilding, K. Groth, R. Har- blocking in carbon nanotube-polymer composite
ris, K. Butler, J. Shields, S. Kharchenko, J. Douglas: organic light-emitting diodes based on poly(m-
Thermal and flammability properties of polypropy- phenylene vinylene-co-2,5-dioctoxy-p-phenylene
lene/carbon nanotube nanocomposites, Polymer 45, vinylene), Appl. Phys. Lett. 77, 1393–1395 (2000)
4227–4239 (2004) 3.486 H.S. Woo, R. Czerw, S. Webster, D.L. Carroll, J.W. Park,
3.474 C. Wei, D. Srivastava, K. Cho: Thermal expansion and J.H. Lee: Organic light emitting diodes fabricated
diffusion coefficients of carbon nanotube-polymer with single wall carbon nanotubes dispersed in
composites, Los Alamos Nat. Lab., Preprint Archive, a hole conducting buffer: The role of carbon nano-
Condensed Matter (archiv:cond-mat/0203349), 1–11 tubes in a hole conducting polymer, Synth. Met. 116,
(2002) 369–372 (2001)
3.475 J.C. Grunlan, M.V. Bannon, A.R. Mehrabi: Latex- 3.487 H. Ago, K. Petritsch, M.S.P. Shaffer, A.H. Windle,
based, single-walled nanotube composites: pro- R.H. Friend: Composites of carbon nanotubes and
cessing and electrical conductivity, Polym. Prepr. 45, conjugated polymers for photovoltaic devices, Adv.
154–155 (2004) Mater. 11, 1281–1285 (1999)
3.476 J.C. Grunlan, A.R. Mehrabi, M.V. Bannon, J.L. Bahr: 3.488 B. Maruyama, K. Alam: Carbon nanotubes and
Water-based single-walled-nanotube-filled poly- nanofibers in composite materials, SAMPE Journal
mer composite with an exceptionally low percola- 38, 59–70 (2002)
tion threshold, Adv. Mater. (Weinheim) 16, 150–153 3.489 P. Lambin, A. Fonseca, J.P. Vigneron, J. B’Nagy,
(2004) A.A. Lucas: Structural and electronic properties of
3.477 C. Pirlot, I. Willems, A. Fonseca, J.B. Nagy, J. Delhalle: bent carbon nanotubes, Chem. Phys. Lett. 245, 85–
Preparation and characterization of carbon nano- 89 (1995)
tube/polyacrylonitrile composites, Adv. Eng. Mater. 3.490 L. Chico, V.H. Crespi, L.X. Benedict, S.G. Louie,
4, 109–114 (2002) M.L. Cohen: Pure carbon nanoscale devices: Nano-
3.478 H. Lam, H. Ye, Y. Gogotsi, F. Ko: Structure and tube heterojunctions, Phys. Rev. Lett. 76, 971–974
properties of electrospun single-walled carbon (1996)
nanotubes reinforced nanocomposite fibrils by co- 3.491 Z. Yao, H.W.C. Postma, L. Balents, C. Dekker: Car-
electrospinning, Polym. Prepr. 45, 124–125 (2004) bon nanotube intramolecular junctions, Nature 402,
3.479 L. Cao, H. Chen, M. Wang, J. Sun, X. Zhang, 273–276 (1999)
F. Kong: Photoconductivity study of modified carbon 3.492 S.J. Tans, A.R.M. Verschueren, C. Dekker: Room
nanotube/oxotitanium phthalocyanine composites, temperature transistor based on single carbon
J. Phys. Chem. B 106, 8971–8975 (2002) nanotube, Nature 393, 49–52 (1998)
118 Part A Nanostructures, Micro-/Nanofabrication and Materials
3.493 R. Martel, T. Schmidt, H.R. Shea, T. Hertel, P. Avouris: 3.501 E. Frackowiak, F. Béguin: Electrochemical storage
Single and multi-wall carbon nanotube field effect of energy in carbon nanotubes and nanostructured
transistors, Appl. Phys. Lett. 73, 2447–2449 (1998) carbons, Carbon 40, 1775–1787 (2002)
3.494 V. Derycke, R. Martel, J. Appenzeller, P. Avouris: Car- 3.502 C. Portet, P.L. Taberna, P. Simon, E. Flahaut:
bon nanotube inter- and intramolecular logic gates, Influence of carbon nanotubes addition on carbon–
Nano Lett. 1, 453–456 (2001) carbon supercapacitor performances in organic
3.495 P. Kim, C.M. Lieber: Nanotube nanotweezers, Sci- electrolyte, J. Power. Sources 139, 371–378 (2005)
ence 286, 2148–2150 (1999) 3.503 E. Frackowiak, K. Jurewicz, K. Szostak, S. Delpeux,
Part A 3
3.496 P.G. Collins, M.S. Arnold, P. Avouris: Engineering F. Béguin: Nanotubular materials as electrodes for
carbon nanotubes using electrical breakdown, Sci- supercapacitors, Fuel Process. Technol. 77, 213–219
ence 292, 706–709 (2001) (2002)
3.497 A.P. Graham, G.S. Duesberg, W. Hoenlein, F. Kreupl, 3.504 G. Lota, E. Frackowiak, J. Mittal, M. Monthioux: High
M. Liebau, R. Martin, B. Rajasekharan, W. Pamler, performance supercapacitor from chromium oxide-
R. Seidel, W. Steinhoegl, E. Unger: How do carbon nanotubes based electrodes, Chem. Phys. Lett. 434,
nanotubes fit into the semiconductor roadmap?, 73–77 (2007)
Appl. Phys. A 80, 1141–1151 (2005) 3.505 R. Hurt, M. Monthioux, A. Kane (Eds.): Toxicology
3.498 R.H. Baughman, C. Changxing, A.A. Zakhidov, of carbon nanomaterials, Carbon 44(6), 1028–1033
Z. Iqbal, J.N. Barisci, G.M. Spinks, G.G. Wallace, (2006), Special issue
A. Mazzoldi, D. de Rossi, A.G. Rinzler, O. Jaschinki, 3.506 C. Salvador-Morales, E. Flahaut, E. Sim, J. Sloan,
S. Roth, M. Kertesz: Carbon nanotubes actuators, M.L.H. Green, R.B. Sim: Complement activation and
Science 284, 1340–1344 (1999) protein adsorption by carbon nanotubes, Mol. Im-
3.499 Y. Gao, Y. Bando: Carbon nanothermometer con- mun. 43, 193–201 (2006)
taining gallium, Nature 415, 599 (2002) 3.507 C. Salvador-Morales, P. Townsend, E. Flahaut,
3.500 C. Niu, E.K. Sichel, R. Hoch, D. Moy, H. Tennent: High C. Vénien-Bryan, A. Vlandas, M.L.H. Green, R.B. Sim:
power electro-chemical capacitors based on carbon Binding of pulmonary surfactant proteins to carbon
nanotube electrodes, Appl. Phys. Lett. 70, 1480–1482 nanotubes; potential for damage to lung immune
(1997) defence mechanisms, Carbon 45, 607–617 (2007)
119
Nanowires
4. Nanowires
Mildred S. Dresselhaus, Yu-Ming Lin, Oded Rabin, Marcie R. Black, Jing Kong, Gene Dresselhaus
Part A 4
structures, small in diameter, and large in surface-
to-volume ratio. Thus, their physical properties are 4.2 Characterization and Physical Properties
different than those of structures of different scale of Nanowires ....................................... 130
and dimensionality. While the study of nanowires 4.2.1 Structural Characterization ............ 130
is particularly challenging, scientists have made 4.2.2 Mechanical Properties................... 135
immense progress in both developing synthetic 4.2.3 Transport Properties ..................... 136
methodologies for the fabrication of nanowires, 4.2.4 Optical Properties ......................... 147
and developing instrumentation for their charac-
4.3 Applications ......................................... 152
terization. The chapter is divided into three main
4.3.1 Electrical Applications ................... 152
sections: Sect. 4.1 the synthesis, Sect. 4.2 the char- 4.3.2 Thermoelectric Applications ........... 154
acterization and physical properties, and Sect. 4.3 4.3.3 Optical Applications ...................... 154
the applications of nanowires. Yet, the reader will 4.3.4 Chemical and Biochemical
discover many links that make these aspects of Sensing Devices ........................... 157
nanoscience intimately interdepent. 4.3.5 Magnetic Applications................... 158
Nanowires are attracting much interest from those seek- the ways in which nanowires differ from their corre-
ing to apply nanotechnology and (especially) those sponding bulk materials. Yet the sizes of nanowires are
investigating nanoscience. Nanowires, unlike other typically large enough (> 1 nm in the quantum-confined
low-dimensional systems, have two quantum-confined direction) to result in local crystal structures that are
directions but one unconfined direction available for closely related to their parent materials, allowing the-
electrical conduction. This allows nanowires to be used oretical predictions about their properties to be made
in applications where electrical conduction, rather than based on knowledge of their bulk properties.
tunneling transport, is required. Because of their unique Not only do nanowires exhibit many properties that
density of electronic states, in the limit of small diam- are similar to, and others that are distinctly different
eters nanowires are expected to exhibit significantly from, those of their bulk counterparts, nanowires also
different optical, electrical and magnetic properties to have the advantage from an applications standpoint in
their bulk 3-D crystalline counterparts. Increased sur- that some of the materials parameters critical for certain
face area, very high density of electronic states and properties can be independently controlled in nanowires
joint density of states near the energies of their van but not in their bulk counterparts. Certain properties
Hove singularities, enhanced exciton binding energy, can also be enhanced nonlinearly in small-diameter
diameter-dependent bandgap, and increased surface nanowires, by exploiting the singular aspects of the 1-D
scattering for electrons and phonons are just some of electronic density of states.
120 Part A Nanostructures, Micro-/Nanofabrication and Materials
Furthermore, nanowires have been shown to pro- search is of significant broad interest at the present time.
vide a promising framework for applying the bottom-up It is the aim of this review to focus on nanowire proper-
approach [4.1] to the design of nanostructures for ties that differ from those of their parent crystalline bulk
nanoscience investigations and for potential nanotech- materials, with an eye toward possible applications that
nology applications. might emerge from the unique properties of nanowires
Driven by (1) these new research and development and from future discoveries in this field.
opportunities, (2) the smaller and smaller length scales For quick reference, examples of typical nanowires
now being used in the semiconductor, optoelectronics that have been synthesized and studied are listed in Ta-
and magnetics industries, and (3) the dramatic develop- ble 4.1. Also of use to the reader are review articles that
ment of the biotechnology industry where the action is focus on a comparison between nanowire and nanotube
also at the nanoscale, the nanowire research field has properties [4.56] and the many reviews that have been
developed with exceptional speed in the last few years. written about carbon nanotubes [4.57–59], which can be
Therefore, a review of the current status of nanowire re- considered as a model one-dimensional system.
Nanowires 4.1 Synthesis 121
4.1 Synthesis
In this section we survey the most common synthetic of parallel and nearly cylindrical channels, as shown
approaches that have successfully afforded high-quality in Fig. 4.1a. The self-organization of the pore structure
nanowires of a large variety of materials (Table 4.1). in an anodic alumina template involves two coupled
In Sect. 4.1.1, we discuss methods which make use processes: pore formation with uniform diameters and
of various templates with nanochannels to confine the pore ordering. The pores form with uniform diam-
nanowire growth in two dimensions. In Sect. 4.1.2, we eters because of a delicate balance between electric
present the synthesis of nanowires by the vapor–liquid– field-enhanced diffusion which determines the growth
solid mechanism and its many variations. In Sect. 4.1.3, rate of the alumina, and dissolution of the alumina
examples of other synthetic methods of general ap- into the acidic electrolyte [4.68]. The pores are be-
plicability are presented. The last part of this section lieved to self-order because of mechanical stress at
(Sect. 4.1.4) features several approaches that have been the aluminum–alumina interface due to expansion dur-
Part A 4.1
developed to organize nanowires into simple architec- ing the anodization. This stress produces a repulsive
tures. force between the pores, causing them to arrange in
a hexagonal lattice [4.69]. Depending on the anodiza-
4.1.1 Template-Assisted Synthesis tion conditions, the pore diameter can be systematically
varied from ≤ 10 up to 200 nm with a pore density in
The template-assisted synthesis of nanowires is a con- the range of 109 –1011 pores/cm2 [4.13, 25, 65, 66]. It
ceptually simple and intuitive way to fabricate nano- has been shown by many groups that the pore size dis-
structures [4.62–64]. These templates contain very tribution and the pore ordering of the anodic alumina
small cylindrical pores or voids within the host ma- templates can be significantly improved by a two-
terial, and the empty spaces are filled with the chosen step anodization technique [4.60, 70, 71], where the
material, which adopts the pore morphology, to form aluminum oxide layer is dissolved after the first an-
nanowires. In this section, we describe the templates odization in an acidic solution followed by a second
first, and then describe strategies for filling the tem- anodization under the same conditions.
plates to make nanowires. Another type of porous template commonly used
for nanowire synthesis is the template type fabricated
Template Synthesis by chemically etching particle tracks originating from
In template-assisted synthesis of nanostructures, the ion bombardment [4.72], such as track-etched polycar-
chemical stability and mechanical properties of the tem- bonate membranes (Fig. 4.1b) [4.73, 74], and also mica
plate, as well as the diameter, uniformity and density films [4.39].
of the pores are important characteristics to consider. Other porous materials can be used as host tem-
Templates frequently used for nanowire synthesis in- plates for nanowire growth, as discussed by Ozin [4.62].
clude anodic alumina (Al2 O3 ), nanochannel glass, ion Nanochannel glass (NCG), for example, contains a reg-
track-etched polymers and mica films. ular hexagonal array of capillaries similar to the pore
Porous anodic alumina templates are produced by structure in anodic alumina with a packing density
anodizing pure Al films in selected acids [4.65–67]. as high as 3 × 1010 pores/cm2 [4.63]. Porous Vycor
Under carefully chosen anodization conditions, the re- glass that contains an interconnected network of pores
sulting oxide film possesses a regular hexagonal array less than 10 nm was also employed for the early
a) b)
study of nanostructures [4.75]. Mesoporous molecular aligned with their main axis perpendicular to the film.
sieves [4.76], termed MCM-41, possess hexagonally- Selective removal of the PMMA component afforded
packed pores with very small channel diameters which the preparation of 14 nm diameter ordered pore arrays
can be varied between 2 and 10 nm. Conducting organic with a packing density of 1.9 × 1011 cm−3 .
filaments have been fabricated in the nanochannels of
MCM-41 [4.77]. Recently, the DNA molecule has also Nanowire Template-Assisted Growth
been used as a template for growing nanometer-sized by Pressure Injection
wires [4.3]. The pressure injection technique is often employed for
Diblock copolymers, polymers that consist of two fabricating highly crystalline nanowires from a low-
chain segments different properties, have also been uti- melting point material and when using porous templates
lized as templates for nanowire growth. When two with robust mechanical strength. In the high-pressure
components are immiscible in each other, phase seg- injection method, the nanowires are formed by pressure-
regation occurs, and depending on their volume ratio, injecting the desired material in liquid form into the
evacuated pores of the template. Due to the heating
Part A 4.1
which are denoted by [101̄1] and [011̄2] when using marily occur in the more accessible cracks, leaving
a hexagonal unit cell, suggesting a wire diameter- most of the nanopores unfilled. Particle track-etched
dependent crystal growth direction. On the other hand, mica films or polymer membranes are typical tem-
30 nm Bi nanowires produced using a much higher plates used in simple DC electrolysis. To use anodic
pressure of > 20 000 psi show a different crystal ori- aluminum oxide films in the DC electrochemical de-
entation of (001) along the wire axis [4.14], indicating position, the insulating barrier layer which separates
that the preferred crystal orientation may also depend the pores from the bottom aluminum substrate has
on the applied pressure, with the most dense packing to be removed, and a metal film is then evaporated
direction along the wire axis for the highest applied onto the back of the template membrane [4.87]. Com-
pressure. pound nanowire arrays, such as Bi2 Te3 , have been
fabricated in alumina templates with a high filling fac-
Electrochemical Deposition tor using the DC electrochemical deposition [4.16].
The electrochemical deposition technique has attracted Figure 4.3a,b, respectively, shows the top view and
Part A 4.1
increasing attention as a versatile method for fabricating the axial cross-sectional SEM images of a Bi2 Te3
nanowires in templates. Traditionally, electrochemistry nanowire array [4.16]. The light areas are associ-
has been used to grow thin films on conducting surfaces. ated with Bi2 Te3 nanowires, the dark regions denote
Since electrochemical growth is usually controllable empty pores, and the surrounding gray matrix is
in the direction normal to the substrate surface, this alumina.
method can be readily extended to fabricate 1-D or Surfactants are also used with electrochemical de-
0-D nanostructures, if the deposition is confined within position when necessary. For example, when using
the pores of an appropriate template. In the electro- templates derived from PMMA/PS diblock copolymers,
chemical methods, a thin conducting metal film is a methanol surfactant is used to facilitate pore fill-
first coated on one side of the porous membrane to ing [4.79], thereby achieving a ≈ 100% filling factor.
serve as the cathode for electroplating. The length of It is also possible to employ an ac electrodeposi-
the deposited nanowires can be controlled by vary- tion method in anodic alumina templates without the
ing the duration of the electroplating process. This removal of the barrier layer, by utilizing the rectifying
method has been used to synthesize a wide variety of properties of the oxide barrier. In ac electrochemical
nanowires, such as metals (Bi [4.9, 74]; Co [4.81, 82]; deposition, although the applied voltage is sinusoidal
Fe [4.25, 83]; Cu [4.73, 84]; Ni [4.39, 81]; Ag [4.85]; and symmetric, the current is greater during the ca-
Au [4.5]); conducting polymers [4.9, 61]; superconduc- thodic half-cycles, making deposition dominant over
tors (Pb [4.86]); semiconductors (CdS [4.19]); and even the stripping, which occurs in the subsequent anodic
superlattice nanowires with A/B constituents (such as half-cycles. Since no rectification occurs at defect sites,
Cu/Co [4.73, 84]) have been synthesized electrochemi- the deposition and stripping rates are equal, and no
cally (Table 4.1). material is deposited. Hence, the difficulties associated
In the electrochemical deposition process, the cho- with cracks are avoided. In this fashion, metals, such as
sen template has to be chemically stable in the Co [4.82] and Fe [4.25, 83], and semiconductors, such
electrolyte during the electrolysis process. Cracks as CdS [4.19], have been deposited into the pores of an-
and defects in the templates are detrimental to the odic aluminum oxide templates without removing the
nanowire growth, since the deposition processes pri- barrier layer.
a) b)
Vapor Deposition
Vapor deposition of nanowires includes physical va-
por deposition (PVD) [4.8], chemical vapor deposition
(CVD) [4.29], and metallo-organic chemical vapor de-
position (MOCVD) [4.32]. Like electrochemical depo-
sition, vapor deposition is usually capable of preparing
smaller-diameter (≤ 20 nm) nanowires than pressure in-
jection methods, since it does not rely on the high
0.1µm 15 nm pressure and the surface tension involved to insert the
material into the pores.
Part A 4.1
Fig. 4.4 (a) TEM image of a single Co(10 nm)/Cu(10 nm) In the physical vapor deposition technique, the ma-
multilayered nanowire. (b) A selected region of the sample terial to be filled is first heated to produce a vapor, which
at high magnification (after [4.84]) is then introduced through the pores of the template and
cooled to solidify. Using a specially designed experi-
In contrast to nanowires synthesized by the pressure mental setup [4.8], nearly single-crystal Bi nanowires
injection method, nanowires fabricated by the elec- in anodic aluminum templates with pore diameters as
trochemical process are usually polycrystalline, with small as 7 nm have been synthesized, and these Bi
no preferred crystal orientations, as observed by XRD nanowires were found to possess a preferred crystal
studies. However, some exceptions exist. For exam- growth orientation along the wire axis, similar to the
ple, polycrystalline CdS nanowires, fabricated by an Bi nanowires prepared by pressure injection [4.8, 13].
ac electrodeposition method in anodic alumina tem- Compound materials that result from two reacting
plates [4.19], possibly have a preferred wire growth gases have also be prepared by the chemical vapor de-
orientation along the c-axis. In addition, Xu et al. have position (CVD) technique. For example, single-crystal
prepared a number of single-crystal II–VI semiconduc- GaN nanowires have been synthesized in anodic alu-
tor nanowires, including CdS, CdSe and CdTe, by DC mina templates through a gas reaction of Ga2 O vapor
electrochemical deposition in anodic alumina templates with a flowing ammonia atmosphere [4.28, 29]. A dif-
with a nonaqueous electrolyte [4.18, 22]. Furthermore, ferent liquid/gas phase approach has been used to
single-crystal Pb nanowires were formed by pulse elec- prepare polycrystalline GaAs and InAs nanowires in
trodeposition under overpotential conditions, but no a nanochannel glass array [4.32]. In this method, the
specific crystal orientation along the wire axis was ob- nanochannels are filled with one liquid precursor (such
served [4.86]. The use of pulse currents is believed to as Me3 Ga or Et3 In) via a capillary effect and the
be advantageous for the growth of crystalline wires be- nanowires are formed within the template by reactions
cause the metal ions in the solution can be regenerated between the liquid precursor and the other gas reactant
between the electrical pulses and therefore uniform de- (such as AsH3 ).
position conditions can be produced for each deposition
pulse. Similarly, single-crystal Ag nanowires were fab- 4.1.2 VLS Method for Nanowire Synthesis
ricated by pulsed electrodeposition [4.4].
One advantage of the electrochemical deposition Some of the recent successful syntheses of semiconduc-
technique is the possibility of fabricating multilayered tor nanowires are based on the so-called vapor–liquid–
structures within nanowires. By varying the cathodic solid (VLS) mechanism of anisotropic crystal growth.
potentials in the electrolyte, which contains two dif- This mechanism was first proposed for the growth of
ferent kinds of ions, different metal layers can be single crystal silicon whiskers 100 nm to hundreds of
controllably deposited. Co/Cu multilayered nanowires micrometer in diameter [4.88]. The proposed growth
have been synthesized in this way [4.73, 84]. Figure 4.4 mechanism (Fig. 4.5) involves the absorption of source
shows TEM images of a single Co/Cu nanowire which material from the gas phase into a liquid droplet of cat-
is about 40 nm in diameter [4.84]. The light bands rep- alyst (a molten particle of gold on a silicon substrate
resent Co-rich regions and the dark bands represent in the original work [4.88]). Upon supersaturation of
Cu-rich layers. This electrodeposition method provides the liquid alloy, a nucleation event generates a solid
Nanowires 4.1 Synthesis 125
Part A 4.1
in the growth of germanium nanowires from gold nan- 3. Laser-assisted catalytic VLS growth is a method
oclusters by the VLS method were recorded by in situ used to generate nanowires under nonequilibrium
TEM [4.89]. conditions. Using laser ablation of a target con-
Reduction of the average wire diameter to the taining both the catalyst and the source materials,
nanometer scale requires the generation of nanosized a plasma is generated from which catalyst nan-
catalyst droplets. However, due to the balance between oclusters nucleate as the plasma cools down. Single
the liquid-vapor surface free energy and the free energy crystal nanowires grow as long as the particle re-
of condensation, the size of a liquid droplet, in equilib- mains liquid [4.48].
rium with its vapor, is usually limited to the micrometer 4. Interestingly, by optimizing the material properties
range. This obstacle has been overcome in recent years of the catalyst-nanowire system, conditions can be
by several new methodologies: achieved for which nanocrystals nucleate in a li-
quid catalyst pool supersaturated with the nanowire
1. Advances in the synthesis of metal nanoclusters material, migrate to the surface due to a large sur-
have made monodispersed nanoparticles commer- face tension, and continue growing as nanowires
cially available. These can be dispersed on a solid perpendicular to the liquid surface [4.50]. In this
substrate in high dilution so that when the temper- case, supersaturated nanodroplets are sustained on
ature is raised above the melting point, the liquid the outer end of the nanowire due to the low solubil-
clusters do not aggregate [4.47]. ity of the nanowire material in the liquid [4.91].
2. Alternatively, metal islands of nanoscale sizes can
self-form when a strained thin layer is grown or A wide variety of elemental, binary and com-
heat-treated on a nonepitaxial substrate [4.34]. pound semiconductor nanowires has been synthesized
a) b) c)
Si
[111]
SiOx
100 nm 10 nm
Fig. 4.6 (a) TEM images of Si nanowires produced after laser-ablating a Si0.9 Fe0.1 target. The dark spheres with
a slightly larger diameter than the wires are solidified catalyst clusters (after [4.48]). (b) Diffraction contrast TEM im-
age of a Si nanowire. The crystalline Si core appears darker than the amorphous oxide surface layer. The inset shows
the convergent beam electron diffraction pattern recorded perpendicular to the wire axis, confirming the nanowire crys-
tallinity (after [4.48]). (c) STEM image of Si/Si1−x Gex superlattice nanowires in the bright field mode. The scale bar is
500 nm (after [4.90])
126 Part A Nanostructures, Micro-/Nanofabrication and Materials
do not require highly sophisticated equipment (such growth along the [001] direction, inducing the growth
as scanning microscopy or lithography-based methods), of hexagonal-plate particles [4.105]. A coordinating
and exclude cases for which the nanowires are not self- alkyl-diamine solvent was used to grow polycrystalline
sustained (such as in the case of atomic rows on the PbSe nanowires at low temperatures [4.42]. Here,
surface of crystals). the surfactant-induced directional growth is believed
A solution-phase synthesis of nanowires with con- to occur through to the formation of organometallic
trollable diameters has been demonstrated [4.45, 101], complexes in which the bidentate ligand assumes the
without the use of templates, catalysts, or surfac- equatorial positions, thus hindering the ions from ap-
tants. Instead, Gates et al. make use of the anisotropy proaching each other in this plane. Additionally, the
of the crystal structure of trigonal selenium and tel- alkyl-diamine molecules coat the external surface of
lurium, which can be viewed as rows of 1-D helical the wire, preventing lateral growth. The aspect ratio
atomic chains. Their approach is based on the mass of the wires increased as the temperature was low-
transfer of atoms during an aging step from a high ered in the range 10 ◦ C < T < 117 ◦ C. Ethylenediamine
Part A 4.1
free-energy solid phase (e.g., amorphous selenium) was used to grow CdS nanowires and tetrapods by
to a seed (e.g., trigonal selenium nanocrystal) which a solvothermal recrystallization process starting with
grows preferentially along one crystallographic axis. CdS nanocrystals or amorphous particles [4.17]. While
The lateral dimension of the seed, which dictates the the coordinating solvent was crucial for the nanowire
diameter of the nanowire, can be controlled by the tem- growth, its role in the shape and phase control was not
perature of the nucleation step. Furthermore, Se/Te clarified.
alloy nanowires were synthesized by this method, and
Ag2 Se compound nanowires were obtained by treating
selenium nanowires with AgNO3 [4.102–104]. In a sep-
arate work, tellurium nanowires were transformed into
Bi2 Te3 nanowires by their reaction with BiPh3 [4.105]. Graphite
More often, however, the use of surfactants is
Elektrodeposition
necessary to promote the anisotropic 1-D growth of
of MoO2 nanowires
nanocrystals. Solution phase synthetic routes have been
optimized to produce monodispersed quantum dots,
(zero-dimensional isotropic nanocrystals) [4.106]. Sur-
factants are necessary in this case to stabilize the
interfaces of the nanoparticles and to retard oxida- Reduction to Mo0 in H2
tion and aggregation processes. Detailed studies on at 500 °C for ≈ 1h
the effect of growth conditions revealed that they can
be manipulated to induce a directional growth of the
nanocrystals, usually generating nanorods (aspect ratio
of ≈ 10), and in favorable cases, nanowires with high
aspect ratios. Heath and LeGoues synthesized germa- Cast poly(styrene) film
nium nanowires by reducing a mixture of GeCl4 and
phenyl-GeCl3 at high temperature and high pressure.
The phenyl ligand was essential for the formation of
high aspect ratio nanowires [4.33]. In growing CdSe
nanorods [4.20], Alivisatos et al. used a mixture of two
surfactants, whose concentration ratio influenced the Lift-off of embedded
Mo0 nanowires
structure of the nanocrystal. It is believed that different
surfactants have different affinities, and different ab-
sorption rates, for the different crystal faces of CdSe,
thereby regulating the growth rates of these faces. In Poly(styrene)
the liquid phase synthesis of Bi nanowires, the ad-
ditive NaN(SiMe3 )2 induces the growth of nanowires Fig. 4.8 Schematic of the electrodeposition step edge dec-
oriented along the [110] crystal direction from small oration of HOPG (highly oriented pyrolytic graphite) for
bismuth seed clusters, while water solely retarded the the synthesis of molybdenum nanowires (after [4.38, 100])
128 Part A Nanostructures, Micro-/Nanofabrication and Materials
Stress-induced crystalline bismuth nanowires have ness the full potential of nanowires for applications. We
been grown from sputtered films of layers of Bi and will first review examples of nanowires with nontrivial
CrN. The nanowires presumably grow from defects and structures, and then proceed to describe methods used
cleavage fractures in the film, and are up to several mil- to create assemblies of nanowires of a predetermined
limeters in length with diameters ranging from 30 to structure.
200 nm [4.7]. While the exploration of this technique We mentioned in Sect. 4.1.2 that the preparation of
has only begun, stress-induced unidirectional growth nanowires with a graded composition or with a super-
should be applicable to a variety of composite films. lattice structure along their main axis was demonstrated
Selective electrodeposition along the step edges in by controlling the gas phase chemistry as a function of
highly oriented pyrolytic graphite (HOPG) was used time during the growth of the nanowires by the VLS
to obtain MoO2 nanowires as shown in Fig. 4.8. The method. Control of the composition along the axial di-
site-selectivity was achieved by applying a low over- mension was also demonstrated by a template-assisted
potential to the electrochemical cell in which the HOPG method, for example by the consecutive electrochem-
Part A 4.1
served as cathode, thus minimizing the nucleation ical deposition of different metals in the pores of an
events on less favorable sites (plateaux). While these alumina template [4.110]. Alternatively, the composi-
nanowires cannot be removed from the substrate, they tion can be varied along the radial dimension of the
can be reduced to metallic molybdenum nanowires, nanowire, for example by first growing a nanowire by
which can then be released as free-standing nanowires. the VLS method and then switching the synthesis con-
Other metallic nanowires were also obtained by this ditions to grow a different material on the surface of
method [4.38, 100]. In contrast to the template syn- the nanowire by CVD. This technique was demon-
thesis approaches described above, in this method the strated for the synthesis of Si/Ge and Ge/Si coaxial
substrate only defines the position and orientation of (or core–shell) nanowires [4.111], and it was shown
the nanowire, not its diameter. In this context, other that the outer shell can be formed epitaxially on the
surface morphologies, such as self-assembled grooves inner core by a thermal annealing process. Han et al.
in etched crystal planes, have been used to generate demonstrated the versatility of MgO nanowire arrays
nanowire arrays via gas-phase shadow deposition (for grown by the VLS method as templates for the PLD
example: Fe nanowires on (110)NaCl [4.27]). The cross deposition of oxide coatings to yield MgO/YBCO,
section of artificially prepared superlattice structures MgO/LCMO, MgO/PZT and MgO/Fe3 O4 core/shell
has also been used for site-selective deposition of par- nanowires, all exhibiting epitaxial growth of the shell
allel and closely spaced nanowires [4.109]. Nanowires on the MgO core [4.37]. A different approach was
prepared on the above-mentioned substrates would adopted by Wang et al. who generated a mixture of
have semicircular, rectangular, or other unconventional coaxial and biaxial SiC-SiOx nanowires by the catalyst-
cross-sectional shapes. free high-temperature reaction of amorphous silica and
a carbon/graphite mixture [4.112].
4.1.4 Hierarchical Arrangement A different category of nontrivial nanowires is that
and Superstructures of Nanowires of nanowires with a nonlinear structure, resulting from
multiple one-dimensional growth steps. Members of
Ordering nanowires into useful structures is another this category are tetrapods, which were mentioned in
challenge that needs to be addressed in order to har- the context of the liquid phase synthesis (Sect. 4.1.3).
a) b) c) d)
Fig. 4.9a–d SEM images of (a) sixfold- (b) fourfold- and (c) twofold-symmetry nanobrushes made of an In2 O3 core and
ZnO nanowire brushes (after [4.107]), and of (d) ZnO nanonails (after [4.108]). The scale bars are (a) 1 μm, (b) 500 nm,
(c) 500 nm, and (d) 200 nm
Nanowires 4.1 Synthesis 129
Part A 4.1
field strength. The nanowires align between two iso-
lated electrodes which are capacitatively coupled to
a pair of buried electrodes biased with an AC volt-
Subphase age. Once a nanowire shorts the electrodes, the electric
field is eliminated, preventing more nanowires from
Fig. 4.10 A TEM image of a smectic phase of a BaCrO4 depositing. The above techniques have been success-
nanorod film (left inset) achieved by the Langmuir– fully used to prepare electronic circuitry and optical
Blodgett technique, as depicted by the illustration (af- devices out of nanowires (Sects. 4.3.1 and 4.3.3). Al-
ter [4.113]) ternatively, alignment and positioning of the nanowires
can be specified and controlled during their growth by
In this process, a tetrahedral quantum dot core is first the proper design of the synthesis method. For exam-
grown, and then the conditions are modified to in- ple, ZnO nanowires prepared by the VLS method were
duce one-dimensional growth of a nanowire from each grown into an array in which both their position on
one of the facets of the tetrahedron. A similar pro- the substrate and their growth direction and orienta-
cess produced high-symmetry In2 O3 /ZnO hierarchical tion were controlled [4.54]. The nanowire growth region
nanostructures. From a mixture of heat-treated In2 O3 , was defined by patterning the gold film, which serves
ZnO, and graphite powders, faceted In2 O3 nanowires as a catalyst for the ZnO nanowire growth, employing
were first obtained, on which oriented shorter ZnO soft-lithography, e-beam lithography, or photolithogra-
nanowires were crystallized [4.107]. Brushlike struc- phy. The orientation of the nanowires was achieved by
tures were obtained as a mixture of 11 structures of selecting a substrate with a lattice structure matching
different symmetries. For example, two, four, or six that of the nanowire material to facilitate the epitaxial
rows of ZnO nanorods could be found on different core growth. These conditions result in an array of nanowire
nanowires, depending on the crystallographic orienta- posts at predetermined positions, all vertically aligned
tion of the main axis of the core nanowire, as shown with the same crystal growth orientation (Fig. 4.11).
in Fig. 4.9. Comblike structures made entirely of ZnO Similar rational GaN nanowire arrays have been syn-
were also reported [4.54]. thesized epitaxially on (100)LiAlO2 and (111)MgO
Controlling the position of a nanowire in the growth single-crystal substrates. In addition, control over the
process is important for preparing devices or test struc- crystallographic growth directions of nanowires was
tures containing nanowires, especially when it involves achieved by lattice-matching to different substrates.
a large array of nanowires. Post-synthesis methods to For example, GaN nanowires on (100)LiAlO2 sub-
align and position nanowires include microfluidic chan- strates grow oriented along the [110] direction, whereas
nels [4.114], Langmuir–Blodgett assemblies [4.113], (111)MgO substrates result in the growth of GaN
and electric field-assisted assembly [4.115]. The first nanowires with an [001] orientation, due to the dif-
method involves the orientation of the nanowires by ferent lattice-matching constraints [4.117]. A similar
the liquid flow direction when a nanowire solution is structure could be obtained by the template-mediated
injected into a microfluidic channel assembly and by electrochemical synthesis of nanowires (Sect. 4.1.1),
the interaction of the nanowires with the side walls particularly if anodic alumina with its parallel and or-
of the channel. The second method involves the align- dered channels is used. The control over the location
130 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b) c)
1 µm 1 µm 100 nm
Fig. 4.11a–c SEM images of ZnO nanowire arrays grown on a sapphire substrate, where (a) shows patterned growth,
(b) shows a higher resolution image of the parallel alignment of the nanowires, and (c) shows the faceted side-walls and
the hexagonal cross section of the nanowires. For nanowire growth, the sapphire substrates were coated with a 1.0–3.5 nm
Part A 4.2
thick patterned layer of Au as the catalyst, using a TEM grid as the shadow mask. These nanowires have been used for
nanowire laser applications (after [4.122])
of the nucleation of nanowires in the electrochemi- By growing the template on a patterned conductive
cal deposition is determined by the pore positions and substrate that serves as a back-electrode [4.119–121]
the back-electrode geometry. The pore positions can different materials can be deposited in the pores at dif-
be precisely controlled by imprint lithography [4.118]. ferent regions of the template.
10 μm length scales can be probed, providing infor- gies for Si nanowires prepared by the laser ablation
mation on the size, size distribution, shapes, spatial of a Si target [4.123]: (a) spring-shaped; (b) fishbone-
distributions, density, nanowire alignment, filling fac- shaped (indicated by solid arrow) and frogs egg-shaped
tors, granularity, etc.. As another example, Fig. 4.11a (indicated by the hollow arrow), (c) pearl-shaped, while
shows an SEM image of ZnO nanowire arrays grown (d) shows the poly-sites of nanowire nucleation. The
on a sapphire substrate [4.122], which provides evi- crystal quality of nanowires is revealed from high-
dence for the nonuniform spatial distribution of the resolution TEM images with atomic resolution, along
nanowires on the substrate, which was attained by pat- with selected area electron diffraction (SAED) pat-
terning the catalyst film to define high-density growth terns. For example, Fig. 4.14 shows a TEM image of
regions and nanowire-free regions. Figure 4.11b, show- one of the GaN nanowires from Fig. 4.12, indicating
ing a higher magnification of the same system, indicates single crystallinity and showing (100) lattice planes,
that these ZnO nanowires grow perpendicular to the thus indicating the growth direction of the nanowire.
substrate, are well-aligned with approximately equal This information is supplemented by the corresponding
Part A 4.2
wire lengths, and have wire diameters in the range electron diffraction pattern in the upper right. A more
20 ≤ dW ≤ 150 nm. The SEM micrograph in Fig. 4.11c comprehensive review of the application of TEM for
provides further information about the surface of growth orientation indexing and crystal defect charac-
the nanowires, showing it to be well-faceted, form- terization in nanowires is available elsewhere [4.124].
ing a hexagonal cross section, indicative of nanowire The high resolution of the TEM also permits the
growth along the 0001 direction. Both the uniformity surface structures of the nanowires to be studied. In
of the nanowire size, their alignment perpendicular to many cases, the nanowires are sheathed with a native
the substrate, and their uniform growth direction, as oxide layer, or an amorphous oxide layer that forms dur-
suggested by the SEM data, are linked to the good epi- ing the growth process. This can be seen in Fig. 4.6b
taxial interface between the (0001) plane of the ZnO for silicon nanowires and in Fig. 4.15 for germanium
nanowire and the (110) plane of the sapphire substrate. nanowires [4.35], showing a mass–thickness contrast
(The crystal structures of ZnO and sapphire are es- TEM image and a selected-area electron diffraction pat-
sentially incommensurate, with the exception that the tern of a Ge nanowire. The main TEM image shows that
a-axis of ZnO and the c-axis of sapphire are related al- these Ge nanowires possess an amorphous GeO2 sheath
most exactly by a factor of 4, with a mismatch of less with a crystalline Ge core that is oriented in the [211]
than 0.08% at room temperature [4.122].) The well- direction.
faceted nature of these nanowires has important impli-
cations for their lasing action (Sect. 4.3.2). Figure 4.12
shows an SEM image of GaN nanowires synthesized by
a laser-assisted catalytic growth method [4.30], indicat-
ing a random spatial orientation of the nanowire axes
and a wide diameter distribution for these nanowires, in
contrast to the ZnO wires in Fig. 4.11 and to arrays of
well-aligned nanowires prepared by template-assisted
growth (Fig. 4.3).
c) 50 nm d) 300 nm
– 022
[0 1]
[01
111 –
111
[211]
)
(111)
00
(1
(01
–)
1
Oxide
(01
layer
–)
1
Part A 4.2
GeO2
Before After H2 annealing
Ge at 130 °C for 6 h
100 nm
10
Fig. 4.16 High-resolution transmission electron microscope
Fig. 4.15 A mass–thickness contrast TEM image of a Ge (HRTEM) image of a Bi nanowire (left) before and (right) after
nanowire taken along the [01̄1] zone axis and a selected- annealing in hydrogen gas at 130 ◦ C for 6 h within the environmen-
area electron diffraction pattern (upper left inset) (af- tal chamber of the HRTEM instrument to remove the oxide surface
ter [4.35]). The Ge nanowires were synthesized by laser layer (after [4.125])
ablation of a mixture of Ge and GeO2 powder. The core of
the Ge nanowire is crystalline, while the surface GeO2 is nanowires [4.90] and core–shell nanowires [4.111]
amorphous (Sect. 4.1.2).
field gradient microscopy (EFM) [4.13], magnetic field of nanowires. A scanning tunneling microscope can be
microscopy (MFM) [4.40], and scanning thermal mi- employed to reveal both topographical structural infor-
croscopy (SThM) [4.128], combined with atomic force mation, such as that illustrated in Fig. 4.18, as well as
microscopy (AFM), have been employed to study the information on the local electronic density of states of
structural, electronic, magnetic, and thermal properties a nanowire, when used in the STS (scanning tunneling
spectroscopy) mode. Figure 4.18 shows STM height im-
ages (taken in the constant current STM mode) of MoSe
a) molecular wires deposited from a methanol or acetoni-
trile solution of Li2 Mo6 Se6 onto Au substrates. The
STM image of a single MoSe wire (Fig. 4.18a) exhibits
a 0.45 nm lattice repeat distance in a MoSe molecu-
lar wire. When both STM and STS measurements are
made on the same sample, the electronic and structural
Part A 4.2
0 μm
a) b)
Part A 4.2
observed for these aligned ZnO nanowires, indicat-
ing that their preferred growth direction is (001) along 30 40 50 60 70 80 90
2θ (deg)
the wire axis. Similarly, XRD was used to confirm
the different growth directions of GaN nanowire ar- Fig. 4.20 X-ray diffraction pattern of aligned ZnO nano-
ray grown epitaxially on (100)LiAlO2 and (111)MgO wires (Fig. 4.11) grown on a sapphire substrate. Only [00]
substartes [4.117]. diffraction peaks are observed for the nanowires, owing to
EDAX has been used to determine the chem- their well-oriented growth orientation. Strong diffraction
ical compositions and stoichiometries of compound peaks for the sapphire substrate are found (after [4.122])
nanowires or impurity contents in nanowires. However,
the results from EDAX analysis should be interpreted of the pore-confined indium in these samples is broader
carefully to avoid systematic errors. in temperature than for bulk indium, as expected for the
heterogeneity in the pore diameter and in the indium
4.2.2 Mechanical Properties crystal size aspect ratio within the samples.
Sheathed nanowires provide an opportunity to
Thermal Stability study the melting and recrystallization of individual
Due to the large surface area-to-volume ratio in nanowires. The shell layer surrounding the nanowire
nanowires and other nanoparticles, the thermal stabil- provides confinement to keep the liquid phase within
ity of nanowires is anticipated to differ significantly the inner cylindrical volume. However, the shell–
from that of the bulk material. Theoretical studies of nanowire surface interaction should be taken into
materials in confined geometries show that the melt- account when analyzing the phase transition thermody-
ing point of the material is reduced in nanostructures, namics and kinetics. Yang et al. produced germanium
as is the latent heat of fusion, and that large hysteresis nanowires coated with a thin (1–5 nm) graphite sheath,
can be observed in melting–freezing cycles. These phe- by pyrolysis of organic molecules over VLS-grown
nomena have been studied experimentally in three types nanowires, and followed the melting and recrystalliza-
of nanowire systems: porous matrices impregnated with tion of the germanium by variable temperature TEM
a plurality of nanowires, individual nanowires sheathed imaging [4.130]. The melting of the nanowires was fol-
by a thin coating, and individual nanowires. lowed by the disappearance of the electronic diffraction
The melting freezing of matrix-supported nanowires pattern. It was found that the nanowires began melt-
can be studied by differential scanning calorimetry ing from their ends, with the melting front advancing
(DSC), since large volumes of samples can thus be pro- towards the center of the nanowire as the tempera-
duced. Huber et al. investigated the melting of indium ture was increased. During the cool-down part of the
in porous silica glasses with mean pore diameters rang- cycle, the recrystallization of the nanowire occurred in-
ing from 6 to 141 nm [4.129]. The melting point of stantaneously following significant supercooling. The
the pore-confined indium shows a linear dependence on authors report both the largest melting point suppres-
inverse pore diameter, with a maximum melting point sion recorded thus far for germanium (≈ 300 ◦ C), and
depression of 50 K. They also recorded a 6 K difference a large melting–recrystallization hysteresis of up to
in the melting temperature and the freezing tempera- ≈ 300 ◦ C. Similarly, carbon nanotubes have been filled
ture of 12.8 nm diameter indium. The melting profile with various low-temperature metals [4.131]. A nanoth-
136 Part A Nanostructures, Micro-/Nanofabrication and Materials
ermometer has been demonstrated using a 10 nm liquid eter, (important for both classical and quantum size
gallium filled-carbon nanotube, showing an expansion effects), material composition, surface conditions, crys-
coefficient that is linear in temperature and identical to tal quality, and the crystallographic orientation along
the bulk value [4.132]. the wire axis for materials with anisotropic material pa-
A different behavior was observed in free-standing rameters, such as the effective mass tensor, the Fermi
copper nanowires [4.134]. In this system, there is lit- surface, or the carrier mobility.
tle interaction between the nanowire surface and the Electronic transport phenomena in low-dimensional
surroundings, and the nanowire is not confined in its systems can be roughly divided into two categories:
diameter, as in the case of the sheathed nanowires. Ther- ballistic transport and diffusive transport. Ballistic
mal treatment of the free-standing nanowires leads to transport phenomena occur when the electrons can
their fragmentation into a linear array of metal spheres. travel across the nanowire without any scattering. In
Thinner nanowires were more vulnerable than thicker this case, the conduction is mainly determined by the
nanowires to the thermal treatment, showing constric- contacts between the nanowire and the external circuit,
Part A 4.2
tions and segmentation at lower temperatures. Analysis and the conductance is quantized into an integral num-
of the temperature response of the nanowires indi- ber of universal conductance units G 0 = 2e2 /h [4.135,
cates that the nanowire segmentation is a result of 136]. Ballistic transport phenomena are usually ob-
the Rayleigh instability, starting with oscillatory per- served in very short quantum wires, such as those
turbations of the nanowire diameter, leading to long produced using mechanically controlled break junctions
cylindrical segments, that become more separated and (MCBJ) [4.137, 138] where the electron mean free path
more spherical at higher temperatures. These observa- is much longer than the wire length and the conduction
tions indicate that annealing and melting are dominated is a pure quantum phenomenon. To observe ballistic
by the surface diffusion of atoms on the entire surface transport, the thermal energy must also obey the relation
of the nanowire (versus tip-initiated melting). kB T ε j − ε j−1 , where ε j − ε j−1 is the energy separa-
tion between subband levels j and j − 1. On the other
4.2.3 Transport Properties hand, for nanowires with lengths much larger than the
carrier mean free path, the electrons (or holes) undergo
The study of electrical transport properties of nanowires numerous scattering events when they travel along the
is important for nanowire characterization, electronic wire. In this case, the transport is in the diffusive regime,
device applications, and the investigation of unusual and the conduction is dominated by carrier scattering
transport phenomena arising from one-dimensional within the wires, due to phonons (lattice vibrations),
quantum effects. Important factors that determine the boundary scattering, lattice and other structural defects,
transport properties of nanowires include the wire diam- and impurity atoms.
Conductance Quantization
a) b) Counts
in Metallic Nanowires
Gold Nano- 8 ×105 The ballistic transport of 1-D systems has been ex-
wires contact Au – Au
tensively studied since the discovery of quantized
6 ×105 conductance in 1-D systems in 1988 [4.135, 136]. The
phenomena of conductance quantization occur when the
4 ×105 diameter of the nanowire is comparable to the electron
Fermi wavelength, which is on the order of 0.5 nm for
2 ×105
most metals [4.139]. Most conductance quantization ex-
0 periments up to the present were performed by bringing
1 2 3 4 together and separating two metal electrodes. As the
Conductance (2e 2/h)
two metal electrodes are slowly separated, a nanocon-
Fig. 4.21 (a) Schematic representation of the last stages of the con- tact is formed before it breaks completely (Fig. 4.21a),
tact breakage process (after [4.133]). (b) Histogram of conductance and conductance in integral multiple values of G 0
values built with 18 000 gold contact breakage experiments in air at is observed through these nanocontacts. Figure 4.21b
room temperature, showing conductance peaks at integral values of shows the conductance histogram built with 18 000
G0 . In this experiment the gold electrodes approach and separate at contact breakage curves between two gold electrodes
89 000 Å/s (after [4.133]) at room temperature [4.133], with the electrode sep-
Nanowires 4.2 Characterization and Physical Properties of Nanowires 137
aration up to ≈ 1.8 nm. The conductance quantization context have usually been made from conventional
behavior is found to be independent of the contact ma- semiconducting materials, such as group IV and III–V
terial, and has been observed in various metals, such compound semiconductors, via the VLS growth method
as Au [4.133], Ag, Na, Cu [4.140], and Hg [4.141]. (Sect. 4.1.2), and their nanowire properties have been
For semimetals such as Bi, conductance quantization compared to their well-established bulk properties. In-
has also been observed for electrode separations as terestingly, the physical principles for describing bulk
long as 100 nm at 4 K because of the long Fermi semiconductor devices also hold for devices based on
wavelength (≈ 26 nm) [4.139], indicating that the con- these semiconducting nanowires with wire diameters of
ductance quantization may be due to the existence of tens of nanometers. For example, Fig. 4.22 shows the
well-defined quantum states localized at a constriction current–voltage (I –V ) behavior of a 4-by-1 crossed p-
instead of resulting from the atom rearrangement as the Si/n-GaN junction array at room temperature [4.142].
electrodes separate. Since conductance quantization is The long horizontal wire in the figure is a p-Si nanowire
only observed in breaking contacts, or for very narrow (10–25 nm in diameter) and the four short vertical wires
Part A 4.2
and very short nanowires, most nanowires of practi- are n-GaN nanowires (10–30 nm in diameter). Each of
cal interest (possessing lengths of several micrometer) the four nanoscale cross points independently forms
lie in the diffusive transport regime, where the carrier a p-n junction with current rectification behavior, as
scattering is significant and should be considered. shown by the I –V curves in Fig. 4.22, and the junc-
tion behavior (for example the turn-on voltage) can
I–V Characterization be controlled by varying the oxide coating on these
of Semiconducting Nanowires nanowires [4.142].
The electronic transport behavior of nanowires may be Huang et al. have demonstrated nanowire junc-
categorized based on the relative magnitudes of three tion diodes with a high turn-on voltage (≈ 5 V) by
length scales: carrier mean free path W , the de Broglie increasing the oxide thickness at the junctions. The
wavelength of electrons λe , and the wire diameter dW . high turn-on voltage enables the use of the junction in
For wire diameters much larger than the carrier mean
free path (dW W ), the nanowires exhibit transport
Current (nA)
properties similar to bulk materials, which are indepen-
2000
dent of the wire diameter, since the scattering due to
the wire boundary is negligible compared to other scat-
tering mechanisms. For wire diameters comparable to
or smaller than the carrier mean free path (dW ≈ W 1500
or dW < W ), but still much larger than the de Broglie
wavelength of the electrons (dW λe ), the transport in
nanowires is in the classical finite size regime, where 1000
the band structure of the nanowire is still similar to that
of bulk, while the scattering events at the wire bound-
ary alter their transport behavior. For wire diameters
comparable to the electronic wavelength dW ≈ λe , the 500
electronic density of states is altered dramatically and
quantum subbands are formed due to the quantum con-
finement effect at the wire boundary. In this regime, the 0
transport properties are further influenced by the change –4 –2 0 2 4
in the band structure. Therefore, transport properties for Bias (V)
nanowires in the classical finite size and quantum size Fig. 4.22 I –V behavior for a 4(p) by 1(n) crossed p-Si/n-
regimes are highly diameter-dependent. GaN junction array shown in the inset. The four curves
Researchers have investigated the transport prop- represent the I –V response for each of the four junc-
erties of various semiconducting nanowires and have tions, showing similar current rectifying characteristics in
demonstrated their potential for diverse electronic de- each case. The length scale bar between the two mid-
vices, such as for p-n diodes [4.142, 143], field effect dle junctions is 2 μm (after [4.142]). The p-Si and n-GaN
transistors [4.142], memory cells, and switches [4.144] nanowires are 10–25 and 10–30 nm in diameter, respec-
(Sect. 4.3.1). So far, the nanowires studied in this tively
138 Part A Nanostructures, Micro-/Nanofabrication and Materials
The very small electron effective mass components curves in Fig. 4.24a show a nonmonotonic trend for
and the long carrier mean free paths in Bi facilitate the large-diameter (70 and 200 nm) nanowires, although
study of quantum size effects in the transport properties R(T ) becomes monotonic with T for small-diameter
of nanowires. Quantum size effects are expected to be- (≤ 48 nm) nanowires. This dramatic change in the be-
come significant in bismuth nanowires with diameters havior of R(T ) as a function of dW is attributed to
smaller than 50 nm [4.147], and the fabrication of crys- a unique semimetal–semiconductor transition phenom-
talline nanowires with this diameter range is relatively ena in Bi [4.78], induced by quantum size effects. Bi
easy. is a semimetal in bulk form, in which the T -point
Figure 4.24a shows the T dependence of the re- valence band overlaps with the L-point conduction
sistance R(T ) for Bi nanowires (7 ≤ dW < 200 nm) band by 38 meV at 77 K. As the wire diameter de-
synthesized by vapor deposition and pressure injec- creases, the lowest conduction subband increases in
tion [4.8], illustrating the quantum effects in their energy and the highest valence subband decreases in
temperature-dependent resistance. In Fig. 4.24a, the energy. Model calculations predict that the band over-
Part A 4.2
R(T ) behavior of Bi nanowires is dramatically dif- lap should vanish in Bi nanowires (with their wire
ferent from that of bulk Bi, and is highly sensi- axes along the trigonal direction) at a wire diameter
tive to the wire diameter. Interestingly, the R(T ) ≈ 50 nm [4.147].
28 nm 2 μm
1.5 1.5
7 nm
1
0 50 100 150 200 250 300
200 nm
c) R (T )/R (300 K) T (K)
1 4
70 nm 36 nm
Bulk Bi 3
70 nm (polycrystalline)
0.5 2
T (K) 70 nm
0 0
1 10 100 10 100
T (K) T (K)
Fig. 4.24 (a) Measured temperature dependence of the resistance R(T ) normalized to the room temperature (300 K)
resistance for bismuth nanowire arrays of various wire diameters dW (after [4.8]). (b) R(T )/R(290 K) for bismuth wires
of larger dW and lower mobility (after [4.10]). (c) Calculated R(T )/R(300 K) of 36 and 70 nm bismuth nanowires. The
dashed curve refers to a 70 nm polycrystalline wire with increased boundary scattering (after [4.78])
140 Part A Nanostructures, Micro-/Nanofabrication and Materials
effects, various scattering mechanisms related to the a temperature dependence of T −1/2 at low tempera-
transport properties of nanowires can be accounted tures, consistent with 1-D localization theory. Thus, due
for by Matthiessen’s rule. Furthermore, the transport to this localization effect, the use of nanowires with
model has also been generalized to predict the trans- very small diameters for transport applications may be
port properties of Te-doped Bi nanowires [4.78], Sb limited.
nanowires [4.155], and BiSb alloy nanowires [4.156],
and good agreement between experiment and theory has Magnetoresistance
also been obtained for these cases. Magnetoresistance (MR) measurements provide an
For nanowires with diameters comparable to the informative technique for characterizing nanowires,
phase-breaking length, their transport properties may because these measurements yield a great deal of
be further influenced by localization effects. It has information about the electron scattering with wire
been predicted that in disordered systems, the extended boundaries, the effects of doping and annealing on scat-
electronic wavefunctions become localized near defect tering, and localization effects in the nanowires [4.150].
Part A 4.2
sites, resulting in the trapping of carriers and giving rise For example, at low fields the MR data show a quadratic
to different transport behavior. Localization effects are dependence on the B field from which carrier mobility
also expected to be more pronounced as the dimension- estimates can be made (Fig. 4.27 at low B field).
ality and sample size are reduced. Localization effects Figure 4.27 shows the longitudinal magnetoresis-
on the transport properties of nanowire systems have tance (B parallel to the wire axis) for 65 and 109 nm
been studied on Bi nanowires [4.158] and, more re- Bi nanowire samples (before thermal annealing) at 2 K.
cently, on Zn nanowires [4.52]. Figure 4.26 shows the The MR maxima in Fig. 4.27a are due to the classical
measured R(T )/R(300 K) of Zn nanowires fabricated size effect, where the wire boundary scattering is re-
by vapor deposition in porous silica or alumina [4.52]. duced as the cyclotron radius becomes smaller than the
While 15 nm Zn nanowires exhibit an R(T ) behav- wire radius in the high field limit, resulting in a decrease
ior with a T 1 dependence as expected for a metallic in the resistivity. This behavior is typical for the longi-
wire, the R(T ) of 9 and 4 nm Zn nanowires exhibits tudinal MR of Bi nanowires in the diameter range of 45
a) ΔR(B)/R(0) b) Bm (T)
0.16 5
0.14 4
65 nm
0.12
109 nm 3
0.1
2
0.08
1
0 20 40 60 80 100
0.06 T (K)
c) Bm (T)
3
0.04
0.02 2
0
1
– 0.02
– 0.04 0
0 1 2 3 4 5 0 0.01 0.02
B (T) 1/d w (nm–1)
Fig. 4.27 (a) Longitudinal magnetoresistance, ΔR(B)/R(0), at 2 K as a function of B for Bi nanowire arrays with diam-
eters of 65 and 109 nm before thermal annealing. (b) The peak position Bm as a function of temperature for the 109 nm
diameter Bi nanowire array after thermal annealing. (c) The peak position Bm of the longitudinal MR (after thermal
annealing) at 2 K as a function of 1/dW , the reciprocal of the nanowire diameter (after [4.157])
142 Part A Nanostructures, Micro-/Nanofabrication and Materials
to 200 nm [4.8, 149, 150, 157], and the peak position Bm (28–70 nm) for T < 5 K [4.150]. In these curves, a sub-
moves to lower B field values as the wire diameter in- tle steplike feature is seen at low magnetic fields, which
creases, as shown in Fig. 4.27c [4.157], where Bm varies is found to depend only on the wire diameter, and is
linearly with 1/dW . The condition for the occurrence of independent of temperature, the orientation of the mag-
Bm is approximately given by Bm ≈ 2ckF /edW where netic field, and even on the nanowire material (see for
kF is the wave vector at the Fermi energy. The peak example Sb nanowires [4.155]). The lack of a depen-
position Bm is found to increase linearly with increas- dence of the magnetic field at which the step appears
ing temperature in the range of 2–100 K, as shown on temperature, field orientation, and material type in-
in Fig. 4.27b [4.157]. As T is increased, phonon scat- dicates that the phenomenon is related to the magnetic
tering becomes increasingly important, and therefore field length L H = (/eB)1/2 . The characteristic length
a higher magnetic field is required to reduce the resis- L H is the spatial extent of the wave function of electrons
tivity associated with boundary scattering sufficiently to in the lowest Landau level, and L H is independent of
change the sign of the MR. Likewise, increasing the the carrier effective masses. Setting L H (Bc ) equal to the
Part A 4.2
grain boundary scattering is also expected to increase diameter dW of the nanowire defines a critical magnetic
the value of Bm at a given T and wire diameter. field strength Bc below which the wavefunction is con-
The presence of the peak in the longitudinal MR of
nanowires requires a high crystal quality with long car-
R (B)/(B = 0 T)
rier mean free paths along the nanowire axis, so that 1.012
most scattering events occur at the wire boundary in- 1.85 K Bi J1A
0.13 T
stead of at a grain boundary, at impurity sites, or at 70 nm
1.006 3K
defect sites within the nanowire. Liu et al. have inves-
tigated the MR of 400 nm Bi nanowires synthesized by 4K
electrochemical deposition [4.74], and no peak in the 1.004
longitudinal MR is observed. The absence of a magne-
toresistance peak may be attributed to a higher defect 1
level in the nanowires produced electrochemically and 1.008 1.35 K Bi J4A
0.3 T 2K 48 nm
to a large wire diameter, much longer than the carrier 4K
mean free path. The negative MR observed for the Bi
nanowire arrays above Bm (Fig. 4.27) shows that wire 1.004
boundary scattering is a dominant scattering process
for the longitudinal magnetoresistance, thereby estab- 1
lishing that the mean free path is larger than the wire
1.02
diameter and that a ballistic transport behavior is indeed 0.54 T
1.39 K
observed in the high field regime.
In addition to the longitudinal magnetoresistance 1.01
4.32 K
measurements, transverse magnetoresistance measure- 1.97 K Bi J2B
ments (B perpendicular to the wire axis) have also been 36 nm
performed on Bi nanowire array samples [4.8,150,157], 1
where a monotonically increasing B 2 dependence over 1.1
0.87 T
1.39 K
the entire range 0 ≤ B ≤ 5.5 T is found for all Bi
nanowires studied thus far. This is as expected, since the 1.97 K
wire boundary scattering cannot be reduced by a mag- 1.05
4.33 K
netic field perpendicular to the wire axis. The transverse Bi J5A
28 nm
magnetoresistance is also found to be always larger than
1
the longitudinal magnetoresistance in nanowire arrays. 0 1 2 3 4 5
By applying a magnetic field to nanowires at very B (T)
low temperatures (≤ 5 K), one can induce a transition Fig. 4.28 Longitudinal magnetoresistance as a function of
from a 1-D confined system at low magnetic fields magnetic field for Bi nanowires of the diameters indicated.
to a 3-D confined system as the field strength in- The vertical bars indicate the critical magnetic field Bc at
creases, as shown in Fig. 4.28 for the longitudinal MR which the magnetic length equals the nanowire diameter
of Bi nanowire arrays of various nanowire diameters (after [4.150])
Nanowires 4.2 Characterization and Physical Properties of Nanowires 143
fined by the nanowire boundary (the 1-D regime), and diameter nanowires, nor for nanowires that are heavily
above which the wavefunction is confined by the mag- doped.
netic field (the 3-D regime). The physical basis for this
phenomenon is associated with confinement of a single Thermoelectric Properties
magnetic flux quantum within the nanowire cross sec- Nanowires are predicted to hold great promise for ther-
tion [4.150]. This phenomenon, though independent of moelectric applications [4.147, 160], due to their novel
temperature, is observed for T ≤ 5 K, since the phase band structure compared to their bulk counterparts and
breaking length has to be larger than the wire diameter. the expected reduction in thermal conductivity associ-
This calculated field strength Bc indicated in Fig. 4.28 ated with enhanced boundary scattering (see below).
by vertical lines for the appropriate nanowire diameters, Due to the sharp density of states at the 1-D sub-
provides a good fit to the steplike features in these MR band edges (where the van Hove singularities occur),
curves. nanowires are expected to exhibit enhanced Seebeck
The Shubnikov–de Haas (SdH) quantum oscillatory coefficients compared to their bulk counterparts. Since
Part A 4.2
effect, which results from the passage of the quantized the Seebeck coefficient measurement is intrinsically in-
Landau levels through the Fermi energy as the field dependent of the number of nanowires contributing to
strength varies, should, in principle, provide the most the signal, the measurements on nanowire arrays of uni-
direct measurement of the Fermi energy and carrier den- form wire diameter are, in principle, as informative as
sity. For example, Heremans et al. have demonstrated single-wire measurements. The major challenge with
that SdH oscillations can be observed in Bi nanowire measuring the Seebeck coefficients of nanowires lies in
samples with diameters down to 200 nm [4.159], and the design of tiny temperature probes to accurately de-
they have demonstrated that Te doping can be used to termine the temperature difference across the nanowire.
raise the Fermi energy in Bi nanowires. Such infor- Figure 4.29a shows the schematic experimental setup
mation on the Fermi energy is important because, for for the Seebeck coefficient measurement of nanowire
certain applications based on nanowires, it is neces- arrays [4.161], where two thermocouples are placed on
sary to place the Fermi energy near a subband edge both faces of a nanowire array and a heater is attached
where the density of states has a sharp feature. How- to one face of the array to generate a temperature gra-
ever, due to the unusual 1-D geometry of nanowires, dient along the nanowire axis. Ideally, the size of the
other characterization techniques that are commonly thermocouples should be much smaller than the thick-
used in bulk materials to determine the Fermi energy ness of the nanowire array template (i. e. the nanowire
and the carrier concentration (such as Hall measure- length) to minimize error. However, due to the thin-
ment) cannot be applied to nanowire systems. The ness of most templates (≤ 50 μm) and the large size of
observation of the SdH oscillatory effect requires crys- commercially-available thermocouples (≈ 12 μm), the
tal samples of very high quality which allow carriers to measured Seebeck coefficient values are usually under-
execute a complete cyclotron orbit in the nanowire be- estimated.
fore they are scattered. For small nanowire diameters, The thermoelectric properties of Bi nanowire sys-
large magnetic fields are required to produce cyclotron tems have been investigated extensively because of
radii smaller than the wire radius. For some nanowire their potential as good thermoelectric materials. Fig-
systems, all Landau levels may have passed through ure 4.29b shows the measured Seebeck coefficients
the Fermi level at such a high field strength, and in S(T ) as a function of temperature for nanowire ar-
such a case, no oscillations can be observed. The lo- rays with diameters of 40 and 65 nm and different
calization effect may also prevent the observation of isoelectronic Sb alloy concentrations [4.154], and
SdH oscillations for very small diameter (≤ 10 nm) S(T ) results for bulk Bi are shown (solid curve) for
nanowires. Observing SdH oscillations in highly doped comparison. Thermopower enhancement is observed
samples (as may be required for certain applications) in Fig. 4.29b as the wire diameter decreases and as
may be difficult because impurity scattering reduces the Sb content increases, which is attributed to the
the mean free path, requiring high B fields to satisfy semimetal–semiconductor transition induced by quan-
the requirement that carriers complete a cyclotron orbit tum confinement and to Sb alloying effects in Bi1−x Sbx
prior to scattering. Therefore, although SdH oscilla- nanowires. Heremans et al. have observed a substan-
tions provide the most direct method of measuring the tial increase in the thermopower of Bi nanowires as the
Fermi energy and carrier density of nanowire samples, wire diameter decreases further, as shown in Fig. 4.30a
this technique may, however, not work for small- for Bi(15 nm)/silica and Bi(9 nm)/alumina nanocom-
144 Part A Nanostructures, Micro-/Nanofabrication and Materials
Bulk Bi Zn 4 nm/Vycor
1×102
– 100 T1
1
1×10
Fig. 4.30 (a) Absolute value of the Seebeck coefficient of two Bi(15 nm)/silica and two Bi(9 nm)/alumina nanocomposite
samples, in comparison to bulk Bi and 200 nm Bi nanowires in the pores of alumina templates (after [4.52]). The full line
on top part of the figure is a fit to a T −1 law. The Seebeck coefficient of the Bi(9 nm)/alumina composite is positive; the
rest are negative. (b) The Seebeck coefficient of Zn(9 nm)/Al2 O3 and Zn(4 nm)/Vycor glass nanocomposite samples in
comparison to bulk Zn (after [4.52])
Nanowires 4.2 Characterization and Physical Properties of Nanowires 145
Part A 4.2
(Sect. 4.3.3) [4.92], one-dimensional waveguides, and
resonant tunneling diodes [4.94, 163]. Figure 4.31a length for 10 nm diameter PbSe/PbS nanowires at 77 K,
shows a schematic structure of a superlattice nanowire where optimal refers to the placement of the Fermi level
consisting of interlaced quantum dots of two different to optimize ZT . The optimal ZT for 10 nm diameter PbSe,
materials, as denoted by A and B. Various tech- PbS, and PbSe0.5 S0.5 nanowires are 0.33, 0.22, and 0.48,
niques have been developed to synthesize superlattice respectively (after [4.153])
nanowire structures with different interface conditions,
as mentioned in Sects. 4.1.1 and 4.1.2. nant tunneling process in one-dimensional structures,
In this superlattice (SL) nanowire structure, the demonstrating that transport phenomena occur in su-
electronic transport along the wire axis is made possi- perlattice nanowires via tunneling and the possibility
ble by the tunneling between adjacent quantum dots, of controlling the electronic band structure of the SL
while the uniqueness of each quantum dot and its 0- nanowires by carefully selecting the constituent mater-
D characteristic behavior is maintained by the energy ials. This new kind of structure is especially attractive
difference of the conduction or valence bands be- for thermoelectric applications, because the interfaces
tween quantum dots of different materials (Fig. 4.31b), between the nanodots can reduce the lattice ther-
which provides some amount of quantum confine- mal conductivity by blocking the phonon conduction
ment. Recently, Björk et al. have observed interesting along the wire axis, while electrical conduction may
nonlinear I –V characteristics with a negative dif- be sustained and even benefit from the unusual elec-
ferential resistance in one-dimensional heterogeneous tronic band structures due to the periodic potential
structures made of InAs and InP, where InP serves as perturbation. For example, Fig. 4.32 shows the cal-
the potential barrier [4.94, 163]. The nonlinear I –V culated dimensionless thermoelectric figure of merit
behavior is associated with the double barrier reso- ZT = S2 σ T/κ (Sect. 4.3.2) where κ is the total thermal
conductivity (including both the lattice and electronic
contributions) of 10 nm diameter PbS/PbSe superlattice
a) LA LB nanowires as a function of the segment length. A higher
A B
thermoelectric performance than for PbSe0.5 S0.5 al-
D
loy nanowires can be achieved for a 10 nm diameter
superlattice nanowire with segment lengths ≤ 7 nm.
However, the localization effect, which may become
b) important for very short segment lengths, may jeop-
ECB +εnm
B
ardize this enhancement in the ZT of superlattice
ECA +εnm
A nanowires [4.153].
mA mB
Fig. 4.31 (a) Schematic diagram of superlattice (seg- Thermal Conductivity of Nanowires
mented) nanowires consisting of interlaced nanodots A and Experimental measurements of the temperature depen-
B of the indicated length and wire diameter. (b) Schematic dence of the thermal conductivity κ(T ) of individual
potential profile of the subbands in the superlattice suspended nanowires have been carried out on study the
nanowire (after [4.162]) dependence of κ(T ) on wire diameter. In this context,
146 Part A Nanostructures, Micro-/Nanofabrication and Materials
0
0 100 200 300 decreases, indicating a growing importance of boundary
T (K)
scattering and a corresponding decreasing importance
Fig. 4.33 Predicted thermal conductivities of Si nanowires of phonon–phonon scattering. At the smallest wire
of various diameters (after [4.168]) diameter of 22 nm, a linear κ(T ) dependence is found
experimentally, consistent with a linear T dependence
measurements have been made on nanowires down to of the specific heat for a 1-D system, and a temperature-
only 22 nm in diameter [4.164]. Such measurements are independent mean free path and velocity of sound. Fur-
very challenging and are now possible due to techno- ther insights are obtained through studies of the thermal
logical development in the micro- and nanofabrication conductivity of Si/SiGe superlattice nanowires [4.170].
of miniature thermal sensors, and the use of nanometer- Model calculations for κ(T ) based on a radia-
size thermal scanning probes [4.128, 165, 166]. The tive heat transfer model have been carried out for
experiments show that the thermal conductivity of small Si nanowires [4.168]. These results show that the
homogeneous nanowires may be more than one order predicted κ(T ) behavior for Si nanowires is simi-
of magnitude smaller than in the bulk, due mainly to lar to that observed experimentally in the range of
strong boundary scattering effects [4.167]. Phonon con- 37 ≤ dW ≤ 115 nm regarding both the functional form
finement effects may eventually become important in of κ(T ) and the magnitude of the relative decrease
nanowires with even smaller diameters. Measurements in the maximum thermal conductivity κmax as a func-
on mats of nanowires (Fig. 4.12) do not generally give tion of dW . However, the model calculations predict
a) b) Gth (T )/16g0
100
10
0.1
60 100 600 1000 6000
Temperature T (mK)
Fig. 4.34 (a) Suspended mesoscopic phonon device used to measure ballistic phonon transport. The device consists of
an 4 × 4 μm2 phonon cavity (center) connected to four Si3 N4 membranes, 60 nm thick and less than 200 nm wide. The
two bright C-shaped objects on the phonon cavity are thin film heating and sensing Cr/Au resistors, whereas the dark
regions are empty space. (b) Log–log plot of the temperature dependence of the thermal conductance G 0 of the structure
in (a) normalized to 16g0 (see text) (after [4.169])
Nanowires 4.2 Characterization and Physical Properties of Nanowires 147
a substantially larger magnitude for κ(T ) (by 50% or is not always straightforward. The wavelength of light
more) than is observed experimentally. Furthermore, used to probe the sample is usually smaller than the wire
the model calculations (Fig. 4.34) do not reproduce the length, but larger than the wire diameter. Hence, the
experimentally observed linear T dependence for the probe light used in an optical measurement cannot be
22 nm nanowires, but rather predict a 3-D behavior for focused solely onto the wire, and the wire and the sub-
both the density of states and the specific heat in 22 nm strate on which the wire rests (or host material, if the
nanowires [4.168, 171, 172]. wires are embedded in a template) are probed simulta-
Thermal conductance measurements on GaAs neously. For measurements, such as photoluminescence
nanowires below 6 K show a power law dependence, (PL), if the substrate does not luminescence or absorb
but the T dependence becomes somewhat less pro- in the frequency range of the measurements, PL mea-
nounced below ≈ 2.5 K [4.165]. This deviation from sures the luminescence of the nanowires directly and
the power law temperature dependence led to a more the substrate can be ignored. However, in reflection and
detailed study of the quantum limit for the thermal transmission measurements, even a nonabsorbing sub-
Part A 4.2
conductance. To carry out these more detailed experi- strate can modify the measured spectra of nanowires.
ments, a mesoscopic phonon resonator and waveguide In this section we discuss the determination of the
device were constructed that included four ≈ 200 nm dielectric function for nanowires in the context of effec-
wide and 85 nm thick silicon nitride nanowirelike tive medium theories. We then discuss various optical
nanoconstrictions (Fig. 4.33a), and this was used to techniques with appropriate examples that sensitively
establish the quantized thermal conductance limit differentiate nanowire properties from those also found
of g0 = π 2 kB2 T/(3h) (Fig. 4.33b) for ballistic phonon in the parent bulk material, placing particular empha-
transport [4.169, 173]. For temperatures above 0.8 K, sis on electronic quantum confinement effects. Finally,
the thermal conductance in Fig. 4.33b follows a T 3 law, phonon confinement effects are reviewed.
but as T is further reduced, a transition to a linear
T dependence is observed, consistent with a phonon The Dielectric Function
mean free path of ≈ 1 μm, and a thermal conductance In this subsection, we review the use of effective
value approaching 16g0 , corresponding to four mass- medium theory as a method to handle the optical
less phonon modes per channel and four channels in properties of nanowires whose diameters are typically
their phonon waveguide structure (Fig. 4.33a). Ballis- smaller than the wavelength of light, noting that ob-
tic phonon transport occurs when the thermal phonon servable optical properties of materials can be related to
wavelength (380 nm for the experimental structure) is the complex dielectric function [4.174, 175]. Effective
somewhat greater than the width of the phonon waveg- medium theories [4.176, 177] can be applied to model
uide at the waveguide constriction. the nanowire and substrate as one continuous compos-
ite with a single complex dielectric function (
1 + i
2 ),
4.2.4 Optical Properties where the real and imaginary parts of the dielectric
function
1 and
2 are related to the index of refraction
Optical methods provide an easy and sensitive tool for (n) and the absorption coefficient (K ) by the relation
measuring the electronic structures of nanowires, since
1 + i
2 = (n + iK )2 . Since photons at visible or infrared
optical measurements require minimal sample prepara- wavelengths see a dielectric function for the compos-
tion (for example, contacts are not required) and the ite nanowire array/substrate system that is different
measurements are sensitive to quantum effects. Opti- from that of the nanowire itself, the optical transmis-
cal spectra of 1-D systems, such as carbon nanotubes, sion and reflection are different from what they would
often show intense features at specific energies near sin- be if the light were focused only on the nanowire. One
gularities in the joint density of states that are formed commonly observed consequence of effective medium
under strong quantum confinement conditions. A vari- theory is the shift in the plasma frequency in accordance
ety of optical techniques have shown that the properties with the percentage of nanowire material that is con-
of nanowires are different to those of their bulk coun- tained in the composite [4.178]. The plasma resonance
terparts, and this section of the review focuses on these occurs when
1 (ω) becomes zero, and the plasma fre-
differences in the optical properties of nanowires. quency of the nanowire composite will shift to lower
Although optical properties have been shown to (higher) energies when the magnitude of the dielectric
provide an extremely important tool for characteriz- function of the host materials is larger (smaller) than
ing nanowires, the interpretation of these measurements that of the nanowire.
148 Part A Nanostructures, Micro-/Nanofabrication and Materials
Although reflection and transmission measurements providing direct information on the effective bandgap.
probe both the nanowire and the substrate, the optical As the wire diameter of an InP nanowire is decreased
properties of the nanowires can be determined indepen- so that it becomes smaller than the bulk exciton diam-
dently. One technique for separating out the dielectric eter of 19 nm, quantum confinement effects set in,
function of the nanowires from the host is to use an and the band gap is increased. This results in an
effective medium theory in reverse. Since the dielec- increase in the PL peak energy. The smaller the effec-
tric function of the host material is often known, and tive mass, the larger the quantum confinement effects.
the dielectric function of the composite material can be When the shift in the peak energy as a function of
measured by the standard method of using reflection nanowire diameter Fig. 4.35 is analyzed using an ef-
and transmission measurements in combination with fective mass model, the reduced effective mass of the
either the Kramer–Kronig relations or Maxwell’s equa- exciton is deduced to be 0.052 m 0 , which agrees quite
tions, the complex dielectric function of the nanowires well with the literature value of 0.065 m 0 for bulk
can be deduced. An example where this approach has InP. Although the linewidths of the PL peak for the
Part A 4.2
been used successfully is for the determination of the small-diameter nanowires (10 nm) are smaller at low
frequency dependence of the real and imaginary parts temperature (7 K), the observation of strong quantum
of the dielectric function
1 (ω) and
2 (ω) for a paral- confinement and bandgap tunability effects at room
lel array of bismuth nanowires filling the pores of an temperature are significant for photonics applications of
alumina template [4.179]. nanowires (Sect. 4.3.3).
The resolution of photoluminescence (PL) optical
Characteristic Optical Properties of Nanowires imaging of a nanowire is, in general, limited by the
A wide range of optical techniques are available for the wavelength of light. However, when a sample is placed
characterization of nanowires, to distinguish their prop- very close to the detector, the light is not given a chance
erties from those of their parent bulk materials. Some to diffract, and so samples much smaller than the wave-
differences in properties relate to geometric differences, length of light can be resolved. This technique is known
such as the small diameter size and the large length-to- as near-field scanning optical microscopy (NSOM) and
diameter ratio (also called the aspect ratio), while others has been used to successfully image nanowires [4.183].
focus on quantum confinement issues. For example, Fig. 4.36 shows the topographical (a) and
Probably the most basic optical technique is (b) NSOM PL images of a single ZnO nanowire.
to measure the reflection and/or transmission of Magnetooptics can be used to measure the elec-
a nanowire to determine the frequency- dependent real tronic band structure of nanowires. For example,
and imaginary parts of the dielectric function. This tech- magnetooptics in conjunction with photoconductance
nique has been used, for example, to study the band has been proposed as a tool to determine band parame-
gap and its temperature dependence in gallium nitride ters for nanowires, such as the Fermi energy, electron
nanowires in the 10–50 nm range in comparison to bulk effective masses, and the number of subbands to be
values [4.180]. The plasma frequency, free carrier den- considered [4.184]. Since different nanowire subbands
sity, and donor impurity concentration as a function have different electrical transmission properties, the
of temperature were also determined from the infrared electrical conductivity changes when light is used to
spectra, which is especially useful for nanowire re- excite electrons to higher subbands, thereby provid-
search, since Hall effect measurements cannot be made ing a method for studying the electronic structure of
on nanowires. nanowires optically. Magnetooptics can also be used to
Another common method used to study nanowires is study the magnetic properties of nanowires in relation
photoluminescence (PL) or fluorescence spectroscopy. to bulk properties [4.27, 185]. For example, the surface
Emission techniques probe the nanowires directly and magnetooptical Kerr effect has been used to measure the
the effect of the host material does not have to be dependence of the magnetic ordering temperature of Fe-
considered. This characterization method has been Co alloy nanowires on the relative concentration of Fe
used to study many properties of nanowires, such and Co [4.185], and it was used to find that, unlike in
as the optical gap behavior, oxygen vacancies in the case of bulk Fe-Co alloys, cobalt in nanowires in-
ZnO nanowires [4.55], strain in Si nanowires [4.181], hibits magnetic ordering. Nickel nanowires were found
and quantum confinement effects in InP nanowires to have a strong increase in their magnetooptical activity
[4.182]. Figure 4.35 shows the photoluminescence of with respect to bulk nickel. This increase is attributed to
InP nanowires as a function of wire diameter, thereby the plasmon resonance in the wires [4.186].
Nanowires 4.2 Characterization and Physical Properties of Nanowires 149
15 nm 15 nm
20 nm
20 nm
Part A 4.2
50 nm 50 nm
1.5 1.5
1.4 1.45
10 30 50 10 30 50
Diameter (nm) Diameter (nm)
Nonlinear optical properties of nanowires have (THG) in a single nanowire using near-field optical
received particular attention since the nonlinear be- microscopy [4.187]. ZnO nanowires were shown to
havior is often enhanced compared to bulk materials have strong SHG and THG effects that are highly
and the nonlinear effects can be utilized for many polarization-sensitive, and this polarization sensitivity
applications. One such study measured the second har- can be explained on the basis of optical and geomet-
monic generation (SHG) and third harmonic generation rical considerations. Some components of the second
150 Part A Nanostructures, Micro-/Nanofabrication and Materials
harmonic polarization tensor are found to be enhanced point valence band transition (Fig. 4.37). Since both the
in nanowires while others are suppressed as the wire initial and final states for the indirect L–T point valence
diameter is decreased, and such effects could be of inter- band transition downshift in energy as the wire diam-
est for device applications. The authors also showed that eter dW is decreased, the shift in the absorption peak
the second-order nonlinearities are mostly wavelength- results from a difference between the effective masses
independent for λ < 400 nm, which is in the transparent and not from the actual value of either of the masses.
regime for ZnO, below the onset of band gap absorp- Hence the diameter dependence of the absorption peak
tion, and this observation is also of interest for device energy is an order of magnitude less for a valence to va-
applications. lence band indirect transition than for a direct interband
Reflectivity and transmission measurements have L-point transition. Furthermore, the band-tracking ef-
also been used to study the effects of quantum con- fect for the indirect transition gives rise to a large value
finement and surface effects on the low-energy indirect for the joint density of states, thus accounting for the
transition in bismuth nanowires [4.189]. Black et al. high intensity of this feature. The enhancement in the
Part A 4.2
investigated an intense and sharp absorption peak in absorption resulting from this indirect transition may
bismuth nanowires, which is not observed in bulk bis- arise from a gradient in the dielectric function, which
muth. The energy position E p of this strong absorption is large at the bismuth–air or bismuth–alumina inter-
peak increases with decreasing diameter. However, the faces, or from the relaxation of momentum conservation
rate of increase in energy with decreasing diameter rules in nanosystems. It should be noted that, in con-
|∂E p /∂dW | is an order of magnitude less than that pre- trast to the surface effect for bulk samples, the whole
dicted for either a direct interband transition or for nanowire contributes to the optical absorption due to the
intersubband transitions in bismuth nanowires. On the spatial variation in the dielectric function, since the pen-
other hand, the magnitude of |∂E p /∂dW | agrees well etration depth is larger than or comparable to the wire
with that predicted for an indirect L-point valence to T - diameter. In addition, the intensity can be quite signif-
[012]
Γ 0.6
0.05
Bisectrix (y)
Binary (x)
0.4
L-point T-point
B
0.2
0 0
1000 2000 3000 4000 1000 2000 3000 4000
1/λ (cm–1) 1/λ (cm–1)
Fig. 4.37 (a) The measured optical transmission spectra as a function of wavenumber (1/λ) of a ≈ 45 nm diameter
bismuth nanowire array. (b) The simulated optical transmission spectrum resulting from an indirect transition of an L-
point electron to a T -point valence subband state. The insert in (a) shows the bismuth Brillouin zone, and the locations
of the T -point hole and the three L-point electron pockets, including the nondegenerate A, and the doubly-degenerate B
pockets. The insert in (b) shows the indirect L–T point electronic transition induced by a photon with an energy equal
to the energy difference between the initial and final states minus the phonon energy (about 100 cm−1 ) needed to satisfy
conservation of energy in a Stokes process (after [4.188])
Nanowires 4.2 Characterization and Physical Properties of Nanowires 151
icant because there are abundant initial state electrons, Intensity (arb. units)
final state holes, and appropriate phonons for making 2000
an indirect L–T point valence band transition at room T = 300 K E2
temperature. Interestingly, the polarization dependence
λi = 514.5 nm
of this absorption peak is such that the strong absorp- A1 (LO)
tion is present when the electric field is perpendicular
1500
to the wire axis, but is absent when the electric field is
parallel to the wire axis, contrary to a traditional polar-
izer, such as a carbon nanotube where the optical E field
is polarized by the nanotube itself and is aligned along
the carbon nanotube axis. The observed polarization 1000
dependence for bismuth nanowires is consistent with
E1(TO)
a surface-induced effect that increases the coupling be- GaN
Part A 4.2
tween the L-point and T -point bands throughout the full nanowires
volume of the nanowire. Figure 4.37 shows the exper- E2
500
imentally observed transmission spectrum in bismuth
A1 (TO) E1 (LO)
nanowires of ≈ 45 nm diameter (a), and the simulated Sapphire
optical transmission from an indirect transition in bis- GaN
film
muth nanowires of ≈ 45 nm diameter is also shown for
comparison in (b). The indirect L–T point valence band 0
0 200 400 600 800 1000
transition mechanism [4.188] is also consistent with Raman shift (cm –1)
observations of the effect on the optical spectra of a de-
crease in the nanowire diameter and of n-type doping of Fig. 4.38 Room-temperature Raman scattering spectra of
bismuth nanowires with Te. GaN nanowires and of a 5 μm thick GaN epilayer film with
green (514.5 nm) laser excitation. The Raman scattering re-
Phonon Confinement Effects sponse was obtained by dividing the measured spectra by
Phonons in nanowires are spatially confined by the the Bose–Einstein thermal factor [4.190]
nanowire cross-sectional area, crystalline boundaries
and surface disorder. These finite size effects give rise observable in the Raman spectra for the GaN nanowires
to phonon confinement, causing an uncertainty in the but are not found in the corresponding spectra for
phonon wavevector which typically gives rise to a fre- bulk GaN.
quency shift and lineshape broadening. Since zone The experimental spectra in Fig. 4.38 show the
center phonons tend to correspond to maxima in the four A1 + E 1 + 2E 2 modes expected from symme-
phonon dispersion curves, the inclusion of contribu- try considerations for bulk GaN crystals. Two types
tions from a broader range of phonon wave vectors of quantum confinement effects are observed. The
results in both a downshift in frequency and an asym- first type is the observation of the downshift and the
metric broadening of the Raman line, which develops asymmetric broadening effects discussed above. Obser-
a low frequency tail. These phonon confinement ef- vations of such downshifts and asymmetric broadening
fects have been theoretically predicted [4.191, 192] and have also been recently reported in 7 nm diameter Si
experimentally observed in GaN [4.190], as shown nanowires [4.193]. A second type of confinement effect
in Fig. 4.38 for GaN nanowires with diameters in the found in Fig. 4.38 for GaN nanowires is the appearance
range 10–50 nm. The application of these theoreti- of additional Raman features not found in the corre-
cal models indicates that broadening effects should be sponding bulk spectra and associated with combination
noticeable as the wire diameter in GaN nanowires de- modes, and a zone boundary mode. Resonant enhance-
creases to ≈ 20 nm. When the wire diameter decreases ment effects were also observed for the A1 (LO) phonon
further to ≈ 10 nm, the frequency downshift and asym- at 728 cm−1 (Fig. 4.38) at higher laser excitation ener-
metric Raman line broadening effects should become gies [4.190].
152 Part A Nanostructures, Micro-/Nanofabrication and Materials
4.3 Applications
In the preceding sections we have reviewed many of of typical nanowires that are discussed in this review.
the central characteristics that make nanowires in some In addition, unlike traditional silicon processing, dif-
cases similar to and in some cases very different from ferent semiconductors can be used simultaneously in
their parent materials. We have also shown that some nanowire devices to produce diverse functionalities. Not
properties are diameter-dependent, and these proper- only can wires of different materials be combined, but
ties are therefore tunable during synthesis. Thus, it is a single wire can be made of different materials. For
of great interest to find applications that could bene- example, junctions of GaAs and GaP show rectifying
fit in unprecedented ways from both the unique and behavior [4.92], thus demonstrating that good electronic
tunable properties of nanowires and the small sizes of interfaces between two different semiconductors can be
these nanostructures, especially in the miniaturization achieved in the synthesis of multicomponent nanowires.
of conventional devices. As the synthetic methods for Transistors made from nanowires could also hold ad-
Part A 4.3
the production of nanowires are maturing (Sect. 4.1) vantages due to their unique morphology. For example,
and nanowires can be made in reproducible and cost- in bulk field effect transistors (FETs), the depletion
effective ways, it is only a matter of time before layer formed below the source and drain region results
applications will be seriously explored. This is a timely in a source–drain capacitance which limits the oper-
development, as the semiconductor industry will soon ation speed. However, in nanowires, the conductor is
be reaching what seems to be its limit in feature size surrounded by an oxide and thus the depletion layer
reduction, and approaching a classical-to-quantum size cannot be formed. Thus, depending on the device de-
transition. At the same time, the field of biotechnol- sign, the source–drain capacitance in nanowires could
ogy is expanding through the availability of tremendous be greatly minimized and possibly eliminated.
genome information and innovative screening assays. Device functionalities common in conventional
Since nanowires are similar in size to the shrinking semiconductor technologies, such as p-n junction
electronic components and to cellular biomolecules, it diodes [4.142], field-effect transistors [4.144], logic
is only natural for nanowires to be good candidates gates [4.142], and light-emitting diodes [4.92, 194],
for applications in these fields. Commercialization of have been recently demonstrated in nanowires, show-
nanowire devices, however, will require reliable mass ing their promise as building blocks that could be used
production, effective assembly techniques and quality to construct complex integrated circuits by employ-
control methods. ing the bottom-up paradigm. Several approaches have
In this section, applications of nanowires to elec- been investigated to form nanowire diodes (Sect. 4.2.2).
tronics (Sect. 4.3.1), thermoelectrics (Sect. 4.3.2), op- For example, Schottky diodes can be formed by con-
tics (Sect. 4.3.3), chemical and biochemical sensing tacting a GaN nanowire with Al electrodes [4.143].
(Sect. 4.3.4), and magnetic media (Sect. 4.3.5) are dis- Furthermore, p-n junction diodes can be formed at
cussed. the crossing of two nanowires, such as the crossing
of n- and p-type InP nanowires doped by Te and Zn,
4.3.1 Electrical Applications respectively [4.194], or Si nanowires doped by phos-
phorus (n-type) and boron (p-type) [4.195]. In addition
The microelectronics industry continues to face tech- to the crossing of two distinctive nanowires, hetero-
nological (in lithography for example) and economic geneous junctions have also been constructed inside
challenges as the device feature size is decreased, es- a single wire, either along the wire axis in the form
pecially below 100 nm. The self-assembly of nanowires of a nanowire superlattice [4.92], or perpendicular to
might present a way to construct unconventional de- the wire axis by forming a core–shell structure of sil-
vices that do not rely on improvements in photolithogra- icon and germanium [4.111]. These various nanowire
phy and, therefore, do not necessarily imply increasing junctions not only possess the current rectifying proper-
fabrication costs. Devices made from nanowires have ties (Fig. 4.22) expected of bulk semiconductor devices,
several advantages over those made by photolithog- but they also exhibit electroluminescence (EL) that may
raphy. A variety of approaches have been devised to be interesting for optoelectronic applications, as shown
organize nanowires via self-assembly (Sect. 4.1.4), thus in Fig. 4.39 for the electroluminescence of a crossed
eliminating the need for the expensive lithographic tech- junction of n- and p-type InP nanowires [4.194]
niques normally required to produce devices the size (Sect. 4.3.3).
Nanowires 4.3 Applications 153
Part A 4.3
nanowire junctions [4.142], as shown in Fig. 4.40 for the
OR and AND logic gates constructed from 2-by-1 and Fig. 4.39a,b Optoelectrical characterization of a crossed nanowire
1-by-3 nanowire p-n junctions, respectively. By func- junction formed between 65 nm n-type and 68 nm p-type InP
tionalizing nanowires with redox-active molecules to nanowires. (a) Electroluminescence (EL) image of the light emit-
store charge, nanowire FETs were demonstrated with ted from a forward-biased nanowire p-n junction at 2.5 V. Inset,
two-level [4.144] and with eight-level [4.201] mem- photoluminescence (PL) image of the junction. (b) EL intensity as
ory effects, which may be used for nonvolatile memory a function of operation voltage. Inset, the SEM image and the I –V
or as switches. In another advance, In2 O3 nanowire characteristics of the junction (after [4.194]). The scale bar in the
FETs with high-k dielectric material were demon- inset is 5 μm
strated, and substantially enhanced performance was
obtained due to the highly efficient coupling of the Nanowires have also been proposed for applications
gate [4.202]. A vertical FET with a surrounding gate associated with electron field emission [4.204], such as
geometry has also been demonstrated, which has the flat panel displays, because of their small diameter and
potential for high-density nanoscale memory and logic large curvature at the nanowire tip, which may reduce
devices [4.203]. the threshold voltage for electron emission [4.205]. In
a) b) V0(V)
Fig. 4.40a–d Nanowire logic gates:
Vi1 Vi2 11
5 01 10 (a) Schematic of logic OR gate con-
V0 V0(V) structed from a 2 (p-Si) by 1 (n-GaN)
4 5
4
crossed nanowire junction. The in-
Vi1 Vi2
3 3 set shows the SEM image (scale
2
1 bar: 1 μm) of an assembled OR gate
p V0 2
n V0 0 and the symbolic electronic circuit.
1 0 1 2 3 4 5
Vi1 Vi2 00 Vi(V) (b) The output voltage of the cir-
OR
Silicon oxide 0 cuit in (a) versus the four possible
OR address level logic address level inputs: (0,0); (0,1);
c) d) V0(V) (1,0); (1,1), where logic 0 input is 0 V
Vc1
V0 (V) and logic 1 is 5 V (same for below).
R 5 (c) Schematic of logic AND gate con-
5 11
Vc1 V0 4 structed from a 1 (p-Si) by 3 (n-GaN)
4
3
2 crossed nanowire junction. The in-
Vi1 V Vi2 Vc2 3 1 set shows the SEM image (scale bar:
p
n
Vc1 Vi1 Vi2 2 0 1 2 3 4 5 1 μm) of an assembled AND gate
Vi (V)
Vi1 V0 Vi2 Vc2 and the symbolic electronic circuit.
1
AND 00 01 10 (d) The output voltage of the circuit
Silicon oxide
AND address level in (c) versus the four possible logic
address level inputs (after [4.142])
154 Part A Nanostructures, Micro-/Nanofabrication and Materials
this regard, the demonstration of very high field emis- For example, Figs. 4.29 and 4.30a in Sect. 4.2.3 show
sion currents from the sharp tip (≈ 10 nm radius) of a Si an enhanced in S for bismuth and bismuth-antimony
cone [4.204], from carbon nanotubes [4.206], from Si nanowires as the wire diameter decreases. In addition to
nanowires inside a carbon nanotube [4.207], and from alleviating the undesired connections between σ , S and
Co nanowires [4.208], has stimulated interest in this the electronic contribution to the thermal conductivity,
potential area of application for nanowires. nanowires also have the advantage that the phonon con-
The concept of constructing electronic devices tribution to the thermal conductivity is greatly reduced
based on nanowires has already been demonstrated, because of boundary scattering (Sect. 4.2), thereby
and the next step for electronic applications would be achieving a high ZT . Figure 4.41a shows the theoret-
to devise a feasible method for integration and mass ical values for ZT versus sample size for both bismuth
production. We expect that, in order to maintain the thin films (2-D) and nanowires (1-D) in the quantum-
growing rate of device density and functionality in the confined regime, exhibiting a rapidly increasing ZT as
existing electronic industry, new kinds of complemen- the quantum size effect becomes more and more im-
Part A 4.3
tary electronic devices will emerge from this bottom-up portant [4.210]. In addition, the quantum size effect in
scheme for nanowire electronics, different from what nanowires can be combined with other parameters to
has been produced by the traditional top-down approach tailor the band structure and electronic transport behav-
pursued by conventional electronics. ior (for instance, Sb alloying in Bi) to further optimize
ZT . For example, Fig. 4.41b shows the predicted ZT
4.3.2 Thermoelectric Applications for p-type Bi1−x Sbx alloy nanowires as a function of
wire diameter and Sb content x [4.211]. The occurrence
One proposed application for nanowires is for thermo- of a local ZT maxima in the vicinity of x ≈ 0.13 and
electric cooling and for the conversion between thermal dW ≈ 45 nm is due to the coalescence of ten valence
and electrical energy [4.171, 209]. The efficiency of bands in the nanowire and the resulting unusual high
a thermoelectric device is measured in terms of a di- density of states for holes, which is a phenomenon ab-
mensionless figure of merit ZT , where Z is defined as sent in bulk Bi1−x Sbx alloys. For nanowires with very
small diameters, it is speculated that localization effects
σ S2
Z= , (4.2) will eventually limit the enhancement of ZT . However,
κ in bismuth nanowires, localization effects are not signif-
where σ is the electrical conductivity, S is the See- icant for wires with diameters larger than 9 nm [4.52].
beck coefficient, κ is the thermal conductivity, and T In addition to 1-D nanowires, ZT values as high as
is the temperature. In order to achieve a high ZT and ≈ 2 have also been experimentally demonstrated in
therefore efficient thermoelectric performance, a high macroscopic samples containing PbSe quantum dots (0-
electrical conductivity, a hugh Seebeck coefficient and D) [4.212] and stacked 2-D films [4.167].
a low thermal conductivity are required. In 3-D systems, Although the application of nanowires to thermo-
the electronic contribution to κ is proportional to σ in electrics appears very promising, these materials are
accordance with the Wiedemann–Franz law, and nor- still in the research phase of the development cycle and
mally materials with high S have a low σ. Hence an are far from being commercialized. One challenge for
increase in the electrical conductivity (for example by thermoelectric devices based on nanowires lies in find-
electron donor doping) results in an adverse variation in ing a suitable host material that will not reduce ZT
both the Seebeck coefficient (decreasing) and the ther- too much due to the unwanted heat conduction through
mal conductivity (increasing). These two trade-offs set the host material. Therefore, the host material should
the upper limit for increasing ZT in bulk materials, with have a low thermal conductivity and occupy a volume
the maximum ZT remaining ≈ 1 at room temperature percentage in the composite material that is as low as
for the 1960–1995 time frame. possible, while still providing the quantum confinement
The high electronic density of states in quantum- and the support for the nanowires.
confined structures is proposed as a promising possibil-
ity to bypass the Seebeck/electrical conductivity trade- 4.3.3 Optical Applications
off and to control each thermoelectric-related variable
independently, thereby allowing for increased electri- Nanowires also hold promise for optical applications.
cal conductivity, relatively low thermal conductivity, One-dimensional systems exhibit a singularity in their
and a large Seebeck coefficient simultaneously [4.210]. joint density of states, allowing quantum effects in
Nanowires 4.3 Applications 155
2 70
1-D 1
60
Bi at 77 K 50
Trigonal direction 1.2
1 40
2-D 30
20
Part A 4.3
1
0 10
0 10 20 30 40 50 60 0 5 10 15 20 25 30
d W (nm) Antimony content (at %)
Fig. 4.41 (a) Calculated ZT of 1-D (nanowire) and 2-D (quantum well) bismuth systems at 77 K as a function of dW ,
denoting the wire diameter or film thickness. The thermoelectric performance (ZT ) is expected to improve greatly when
the wire diameter is small enough for the nanowire to become a one-dimensional system. (b) Contour plot of optimal ZT
values for p-type Bi1−x Sbx nanowires versus wire diameter and antimony concentration calculated at 77 K (after [4.211])
nanowires to be optically observable, sometimes even that in forward bias, light characteristic of lasing was
at room temperature. Since the density of states of observed at the end of the wire [4.213]. LEDs have
a nanowire in the quantum limit (small wire diameter) also been achieved with core–shell structured nanowires
is highly localized in energy, the available states quickly made of n-GaN/InGaN/p-GaN [4.214].
fill up with electrons as the intensity of the incident light Light emission from quantum wire p-n junctions
is increased. This filling up of the subbands, as well is especially interesting for laser applications, because
as other effects that are unique to low-dimensional ma- quantum wires can form lasers with lower excitation
terials, lead to strong optical nonlinearities in quantum thresholds than their bulk counterparts and they also
wires. Quantum wires may thus yield optical switches exhibit decreased sensitivity of performance to tempera-
with a lower switching energy and increased switching ture [4.215]. Furthermore, the emission wavelength can
speed compared to currently available optical switches. be tuned for a given material composition by simply
Light emission from nanowires can be achieved by altering the geometry of the wire.
photoluminescence (PL) or electroluminescence (EL), Lasing action has been reported in ZnO nanowires
distinguished by whether the electronic excitation is with wire diameters that are much smaller than the
achieved by optical illumination or by electrical stim- wavelength of the light emitted (λ = 385 nm) [4.122]
ulation across a p-n junction, respectively. PL is often (Fig. 4.42). Since the edges and lateral surfaces of ZnO
used for optical property characterization, as described nanowires are faceted (Sect. 4.2.1), they form optical
in Sect. 4.2.4, but from an applications point of view, EL cavities that sustain desired cavity modes. Compared
is a more convenient excitation method. Light-emitting to conventional semiconductor lasers, the exciton laser
diodes (LEDs) have been achieved in junctions between action employed in zinc oxide nanowire lasers ex-
a p-type and an n-type nanowire (Fig. 4.39) [4.194] hibits a lower lasing threshold (≈ 40 kW/cm2 ) than
and in superlattice nanowires with p-type and n-type their 3-D counterparts (≈ 300 kW/cm2 ). In order to uti-
segments [4.92]. The light emission was localized lize exciton confinement effects in the lasing action,
to the junction area, and was polarized in the su- the exciton binding energy (≈ 60 meV in ZnO) must
perlattice nanowire. An electrically driven laser was be greater than the thermal energy (≈ 26 meV at 300 K).
fabricated from CdS nanowires. The wires were as- Decreasing the wire diameter increases the excitation
sembled by evaporating a metal contact onto an n-type binding energy and lowers the threshold for lasing. PL
CdS nanowire which resided on a p+ silicon wafer. The NSOM imaging confirmed the waveguiding properties
cleaved ends of the wire formed the laser cavity, so of the anisotropic and the well-faceted structure of ZnO
156 Part A Nanostructures, Micro-/Nanofabrication and Materials
6 1 µm
0
Part A 4.3
limit, and not by synthesis limitations, and thus can be cient and their tunable bandgap are also characteristics
as small as 145 nm. Figure 4.43a shows an optical im- that can be used to enhance the energy conversion effi-
age of many Au-Ag-Au-Ag barcoded wires, where the ciency of solar cells.
silver segments show higher reflectivity. Figure 4.43b
is a backscattering mode FE-SEM image of a single 4.3.4 Chemical and Biochemical
nanowire, highlighting the composition and segment Sensing Devices
length variations along the nanowire.
Both the large surface area and the high conductiv- Sensors for chemical and biochemical substances with
ity along the length of a nanowire are favorable for its nanowires as the sensing probe are a very attrac-
use in inorganic–organic solar cells [4.225], which offer tive application area. Nanowire sensors will potentially
promise from a manufacturing and cost-effectiveness be smaller, more sensitive, demand less power, and
standpoint. In a hybrid nanocrystal–organic solar cell, react faster than their macroscopic counterparts. Ar-
the incident light forms bound electron–hole pairs (ex- rays of nanowire sensors could, in principle, achieve
citons) in both the inorganic nanocrystal and in the nanometer-scale spatial resolution and therefore pro-
surrounding organic medium. These excitons diffuse to vide accurate real-time information regarding not only
the inorganic–organic interface and disassociate to form the concentration of a specific analyte but also its spa-
an electron and a hole. Since conjugated polymers usu- tial distribution. Such arrays could be very useful, for
ally have poor electron mobilities, the inorganic phase example, for dynamic studies on the effects of chemical
is chosen to have a higher electron affinity than the or- gradients on biological cells. The operation of sen-
ganic phase so that the organic phase carries the holes sors made with nanowires, nanotubes, or nanocontacts
and the semiconductor carries the electrons. The sepa- is based mostly on the reversible change in the con-
rated electrons and holes drift to the external electrodes ductance of the nanostructure upon absorption of the
through the inorganic and organic materials, respec- agent to be detected, but other detection methods, such
tively. However, only those excitons formed within an as mechanical and optical detection, are conceptually
exciton diffusion length from an interface can disas- plausible. The increased sensitivity and faster response
sociate before recombining, and therefore the distance time of nanowires are a result of the large surface-to-
between the dissociation sites limits the efficiency of volume ratio and the small cross section available for
a solar cell. A solar cell prepared from a composite conduction channels. In the bulk, on the other hand,
of CdSe nanorods inside poly(3-ethylthiophene) [4.225] the abundance of charges can effectively shield external
yielded monochromatic power efficiencies of 6.9% and fields, and the abundance of material can afford many
power conversion efficiencies of 1.7% under A.M. 1.5 alternative conduction channels. Therefore, a stronger
illumination (equal to solar irradiance through 1.5 times chemical stimulus and longer response time are neces-
the air mass of the Earth at direct normal incidence). sary to observe changes in the physical properties of
The nanorods provide a large surface area with good a 3-D sensor in comparison to a nanowire.
chemical bonding to the polymer for efficient charge It is often necessary to modify the surface of the
transfer and exciton dissociation. Furthermore, they nanowires to achieve a strong interaction with the an-
provide a good conduction path for the electrons to alytes that need to be detected. Surface modifications
reach the electrode. Their enhanced absorption coeffi- utilize the self-assembly, chemisorption or chemical re-
158 Part A Nanostructures, Micro-/Nanofabrication and Materials
activity of selected organic molecules and polymers ical detection devices were made in a field effect tran-
towards metal and oxide surfaces. Examples include: sistor geometry, so that the back-gate potential could be
thiols on gold, isocyanides on platinum, and siloxanes used to regulate the conductance in conjugation with
on silica. These surface coatings regulate the binding the chemical detection and to provide a real-time di-
and chemical reactivity of other molecules towards the rect read-out [4.227]. The extension of this device to
nanowire in a predictable manner [4.226]. detect multiple analytes using multiple nanowires, each
Cui et al. placed silicon nanowires made by the VLS sensitized to a different analyte, could provide for fast,
method (Sect. 4.1.2) between two metal electrodes and sensitive, and in situ screening procedures.
modified the silicon oxide coating of the wire through A similar approach was used by Favier et al., who
the addition of molecules that are sensitive to the ana- made a nanosensor for the detection of hydrogen from
lyte to be detected [4.227]. For example, a pH sensor of an array of palladium nanowires between two metal
was made by covalently linking an amine-containing contacts [4.44]. They demonstrated that nanogaps were
silane to the surface of the nanowire. Variations in the present in their nanowire structure, and upon absorp-
Part A 4.3
pH of the solution into which the nanowire was im- tion of H2 and formation of Pd hydride, the nanogap
mersed caused protonation and deprotonation of the structure would close and improve the electrical con-
−NH2 and the −SiOH groups on the surface of the tact, thereby increasing the conductance of the nanowire
nanowire. The variation in surface charge density regu- array. The response time of these sensors was 75 ms,
lates the conductance of the nanowire; due to the p-type and they could operate in the range 0.5–5% H2 before
characteristics of a silicon wire, the conductance in- saturation occurred.
creases with the addition of negative surface charge.
The combined acid and base behavior of the surface 4.3.5 Magnetic Applications
groups results in an approximately linear dependence
of the conductance on pH in the pH range 2 to 9, thus It has been demonstrated that arrays of single-domain
leading to a direct readout pH meter. This same type magnetic nanowires can be prepared with controlled
of approach was used for the detection of the binding nanowire diameter and length, aligned along a common
of biomolecules, such as streptavidin using biotin- direction and arranged in a close-packed ordered array
modified nanowires (Fig. 4.44). This nanowire-based (Sect. 4.1), and that the magnetic properties (coercivity,
device has high sensitivity and could detect streptavidin remanence and dipolar magnetic interwire interaction)
binding down to a concentration of 10 pM (10−12 mol). can be controlled to achieve a variety of magnetic ap-
Subsequent results demonstrated the capabilities of plications [4.40, 79].
these functionalized Si nanowire sensors as DNA sen- The most interesting of these applications is for
sors down to the femtomolar range [4.228]. The chem- magnetic storage, where the large nanowire aspect ra-
tio (length/diameter) is advantageous for preventing
the onset of the superparamagnetic limit at which the
a) b) Conductance (nS) magnetization direction in the magnetic grains can be
1600
reversed by the thermal energy kB T , thereby resulting in
SiNW loss of recorded data in the magnetic recording medium.
1500
1 3 The magnetic energy in a grain can be increased by
1400 increasing either the volume or the anisotropy of the
2
grain. If the volume is increased, the particle size in-
1300 creases, so the resolution is decreased. For spherical
magnetized grains, the superparamagnetic limit at room
SiNW 1200 temperature is reached at 70 Gbit/in2 . In nanowires,
0 200 400 the anisotropy is very large and yet the wire diam-
Time (s)
eters are small, so that the magnetostatic switching
Fig. 4.44 (a) Streptavidin molecules bind to a silicon nanowire energy can easily be above the thermal energy while
functionalized with biotin. The binding of streptavidin to biotin the spatial resolution is large. For magnetic data stor-
causes the nanowire to change its resistance. (b) The conductance age applications, a large aspect ratio is needed for the
of a biotin-modified silicon nanowire exposed to streptavidin in nanowires in order to maintain a high coercivity, and
a buffer solution (regions 1 and 3) and with the introduction of a so- a sufficient separation between nanowires is needed
lution of antibiotin monoclonal antibody (region 2) (after [4.227]) to suppress interwire magnetic dipolar coupling. Thus
Nanowires References 159
nanowires can form stable and highly dense magnetic 35 nm diameters [4.40] or diblock copolymer tem-
memory arrays with packing densities in excess of plates [4.79] to make Co nanowires, with mean
1011 wires/cm2 . diameters of 14 nm and 100% filling of the template
The onset of superparamagnetism can be pre- pores (Sect. 4.1.1). The ordered magnetic nanowire ar-
vented in the single-domain magnetic nanowire ar- rays that have already been demonstrated offer the
rays that have already been fabricated using either exciting promise of systems permitting 1012 bits/in2
porous alumina templates to make Ni nanowires with data storage.
Part A 4
ization and physical properties of nanowires, placing counterparts, we can expect future research empha-
particular emphasis on nanowire properties that differ sis to be increasingly focused on smaller diameter
from those of the bulk counterparts and potential appli- nanowires, where new unexplored physical phenom-
cations that might result from the special structures and ena related to quantum confinement effects are more
properties of nanowires. likely to be found. We can also expect the develop-
We have shown that the newly emerging field of ment of applications to follow, some coming sooner and
nanowire research has developed very rapidly over the others later. Many promising applications are now at
past few years, driven by the development of a vari- the early demonstration stage (Sect. 4.3), but are mov-
ety of complementary nanowire synthesis methods and ing ahead rapidly because of their promise of new
effective tools for measuring nanowire structure and functionality, not previously available, in the fields of
properties (Sects. 4.1 and 4.2). At present, much of the electronics, optoelectronics, biotechnology, magnetics,
progress is at the demonstration-of-concept level, with and energy conversion and generation, among others.
many gaps in knowledge remaining to be elucidated, Many exciting challenges remain in advancing both the
theoretical models to be developed, and new nanowire nanoscience and the nanotechnological promise already
systems to be explored. Having demonstrated that many demonstrated by the nanowire research described in this
of the most interesting discoveries to date relate to review.
References
4.1 R.P. Feynman: There’s plenty of room at the bottom, Au nanowire arrays, J. Mater. Chem. 11, 1732–1734
Caltech Eng. Sci. 23, 22 (1960) (2001)
4.2 Y. Mao, S.S. Wong: General, room-temperature 4.7 Y.-T. Cheng, A.M. Weiner, C.A. Wong, M.P. Balogh,
method for the synthesis of isolated as well as arrays M.J. Lukitsch: Stress-induced growth of bis-
of single-crystalline ABO4 -type nanorods, J. Am. muth nanowires, Appl. Phys. Lett. 81, 3248–3250
Chem. Soc. 126, 15245–15252 (2004) (2002)
4.3 E. Braun, Y. Eichen, U. Sivan, G. Ben-Yoseph: DNA- 4.8 J. Heremans, C.M. Thrush, Y.-M. Lin, S. Cronin,
templated assembly and electrode attachment of Z. Zhang, M.S. Dresselhaus, J.F. Mansfield: Bismuth
a conducting silver wire, Nature 391, 775–778 (1998) nanowire arrays: Synthesis, galvanomagnetic prop-
4.4 G. Sauer, G. Brehm, S. Schneider, K. Nielsch, erties, Phys. Rev. B 61, 2921–2930 (2000)
R.B. Wehrspohn, J. Choi, H. Hofmeister, U. Gösele: 4.9 L. Piraux, S. Dubois, J.L. Duvail, A. Radulescu,
Highly ordered monocrystalline silver nanowire ar- S. Demoustier-Champagne, E. Ferain, R. Legras:
rays, J. Appl. Phys. 91, 3243–3247 (2002) Fabrication and properties of organic, metal
4.5 G.L. Hornyak, C.J. Patrissi, C.M. Martin: Fabrica- nanocylinders in nanoporous membranes, J. Mater.
tion, characterization and optical properties of gold Res. 14, 3042–3050 (1999)
nanoparticle/porous alumina composites: The non- 4.10 K. Hong, F.Y. Yang, K. Liu, D.H. Reich, P.C. Sear-
scattering Maxwell–Garnett limit, J. Phys. Chem. B son, C.L. Chien, F.F. Balakirev, G.S. Boebinger: Giant
101, 1548–1555 (1997) positive magnetoresistance of Bi nanowire arrays in
4.6 X.Y. Zhang, L.D. Zhang, Y. Lei, L.X. Zhao, Y.Q. Mao: high magnetic fields, J. Appl. Phys. 85, 6184–6186
Fabrication and characterization of highly ordered (1999)
160 Part A Nanostructures, Micro-/Nanofabrication and Materials
4.11 A.J. Yin, J. Li, W. Jian, A.J. Bennett, J.M. Xu: Fab- 4.24 T. Gao, G.W. Meng, J. Zhang, Y.W. Wang, C.H. Liang,
rication of highly ordered metallic nanowire arrays J.C. Fan, L.D. Zhang: Template synthesis of single-
by electrodeposition, Appl. Phys. Lett. 79, 1039–1041 crystal Cu nanowire arrays by electrodeposition,
(2001) Appl. Phys. A 73, 251–254 (2001)
4.12 Z. Zhang, J.Y. Ying, M.S. Dresselhaus: Bismuth 4.25 D. Al-Mawlawi, N. Coombs, M. Moskovits: Magnetic-
quantum-wire arrays fabricated by a vacuum melt- properties of Fe deposited into anodic aluminum-
ing and pressure injection process, J. Mater. Res. 13, oxide pores as a function of particle-size, J. Appl.
1745–1748 (1998) Phys. 70, 4421–4425 (1991)
4.13 Z. Zhang, D. Gekhtman, M.S. Dresselhaus, J.Y. Ying: 4.26 F. Li, R.M. Metzger: Activation volume of α-Fe par-
Processing and characterization of single-crystalline ticles in alumite films, J. Appl. Phys. 81, 3806–3808
ultrafine bismuth nanowires, Chem. Mater. 11, 1659– (1997)
1665 (1999) 4.27 A. Sugawara, T. Coyle, G.G. Hembree, M.R. Schein-
4.14 T.E. Huber, M.J. Graf, P. Constant: Processing and fein: Self-organized Fe nanowire arrays prepared
characterization of high-conductance bismuth wire by shadow deposition on NaCl(110) templates, Appl.
array composites, J. Mater. Res. 15, 1816–1821 (2000) Phys. Lett. 70, 1043–1045 (1997)
Part A 4
4.15 L. Li, G. Li, Y. Zhang, Y. Yang, L. Zhang: Pulsed 4.28 G.S. Cheng, L.D. Zhang, Y. Zhu, G.T. Fei, L. Li, C.M. Mo,
electrodeposition of large-area, ordered Bi1−x Sbx Y.Q. Mao: Large-scale synthesis of single crystalline
nanowire arrays from aqueous solutions, J. Phys. gallium nitride nanowires, Appl. Phys. Lett. 75,
Chem. B 108, 19380–19383 (2004) 2455–2457 (1999)
4.16 M.S. Sander, A.L. Prieto, R. Gronsky, T. Sands, 4.29 G.S. Cheng, L.D. Zhang, S.H. Chen, Y. Li, L. Li,
A.M. Stacy: Fabrication of high-density, high aspect X.G. Zhu, Y. Zhu, G.T. Fei, Y.Q. Mao: Ordered nano-
ratio, large-area bismuth telluride nanowire arrays structure of single-crystalline GaN nanowires in
by electrodeposition into porous anodic alumina a honeycomb structure of anodic alumina, J. Mater.
templates, Adv. Mater. 14, 665–667 (2002) Res. 15, 347–350 (2000)
4.17 M. Chen, Y. Xie, J. Lu, Y.J. Xiong, S.Y. Zhang, Y.T. Qian, 4.30 Y. Huang, X. Duan, Y. Cui, C.M. Lieber: Gallium nitride
X.M. Liu: Synthesis of rod-, twinrod-, and tetrapod- nanowire nanodevices, Nano Lett. 2, 101–104 (2002)
shaped CdS nanocrystals using a highly oriented 4.31 X. Duan, C.M. Lieber: Laser-assisted catalytic growth
solvothermal recrystallization technique, J. Mater. of single crystal GaN nanowires, J. Am. Chem. Soc.
Chem. 12, 748–753 (2002) 122, 188–189 (2000)
4.18 D. Xu, Y. Xu, D. Chen, G. Guo, L. Gui, Y. Tang: 4.32 A.D. Berry, R.J. Tonucci, M. Fatemi: Fabrication of
Preparation of CdS single-crystal nanowires by elec- GaAs, InAs wires in nanochannel glass, Appl. Phys.
trochemically induced deposition, Adv. Mater. 12, Lett. 69, 2846–2848 (1996)
520–522 (2000) 4.33 J.R. Heath, F.K. LeGoues: A liquid solution synthesis
4.19 D. Routkevitch, T. Bigioni, M. Moskovits, J.M. Xu: of single-crystal germanium quantum wires, Chem.
Electrochemical fabrication of CdS nanowire ar- Phys. Lett. 208, 263–268 (1993)
rays in porous anodic aluminum oxide templates, 4.34 Y. Wu, P. Yang: Germanium nanowire growth via
J. Phys. Chem. 100, 14037–14047 (1996) simple vapor transport, Chem. Mater. 12, 605–607
4.20 L. Manna, E.C. Scher, A.P. Alivisatos: Synthesis of sol- (2000)
uble and processable rod-, arrow-, teardrop-, and 4.35 Y.F. Zhang, Y.H. Tang, N. Wang, C.S. Lee, I. Bello,
tetrapod-shaped CdSe nanocrystals, J. Am. Chem. S.T. Lee: Germanium nanowires sheathed with an
Soc. 122, 12700–12706 (2000) oxide layer, Phys. Rev. B 61, 4518–4521 (2000)
4.21 D. Routkevitch, A.A. Tager, J. Haruyama, D. Al- 4.36 S.J. May, J.-G. Zheng, B.W. Wessels, L.J. Lauhon:
Mawlawi, M. Moskovits, J.M. Xu: Nonlithographic Dendritic nanowire growth mediated by a self-
nano-wire arrays: fabrication, physics, and device assembled catalyst, Adv. Mater. 17, 598–602
applications, IEEE Trans. Electron. Dev. 43, 1646– (2005)
1658 (1996) 4.37 S. Han, C. Li, Z. Liu, B. Lei, D. Zhang, W. Jin, X. Liu,
4.22 D.S. Xu, D.P. Chen, Y.J. Xu, X.S. Shi, G.L. Guo, L.L. Gui, T. Tang, C. Zhou: Transition metal oxide core-shell
Y.Q. Tang: Preparation of II–VI group semiconductor nanowires: Generic synthesis and transport studies,
nanowire arrays by dc electrochemical deposition Nano Lett. 4, 1241–1246 (2004)
in porous aluminum oxide templates, Pure Appl. 4.38 M.P. Zach, K.H. Ng, R.M. Penner: Molybdenum
Chem. 72, 127–135 (2000) nanowires by electrodeposition, Science 290, 2120–
4.23 R. Adelung, F. Ernst, A. Scott, M. Tabib-Azar, 2123 (2000)
L. Kipp, M. Skibowski, S. Hollensteiner, E. Spiecker, 4.39 L. Sun, P.C. Searson, L. Chien: Electrochemical de-
W. Jäger, S. Gunst, A. Klein, W. Jägermann, position of nickel nanowire arrays in single-crystal
V. Zaporojtchenko, F. Faupel: Self-assembled mica films, Appl. Phys. Lett. 74, 2803–2805 (1999)
nanowire networks by deposition of copper onto 4.40 K. Nielsch, R. Wehrspohn, S. Fischer, H. Kronmüller,
layered-crystal surfaces, Adv. Mater. 14, 1056–1061 J. Barthel, J. Kirschner, U. Gösele: Magnetic proper-
(2002) ties of 100 nm nickel nanowire arrays obtained from
Nanowires References 161
ordered porous alumina templates, MRS Symp. Proc. sonidze, G. Dresselhaus: Nanowires and nanotubes,
636, D1.9–1–D1.9–6 (2001) Mater. Sci. Eng. C 23, 129–140 (2003), also in: Current
4.41 Y. Wang, X. Jiang, T. Herricks, Y. Xia: Single crystalline trends in nanotechnologies: From materials to sys-
nanowires of lead: Large-scale synthesis, mecha- tems, Eur. Mater. Res. Soc. Symp. Proc., Vol. 140, ed.
nistic studies, and transport measurements, J. Phys. by W. Jantsch, H. Grimmeiss, G. Marietta (Elsevier,
Chem. B 108, 8631–8640 (2004) Amsterdam 2002)
4.42 E. Lifshitz, M. Bashouti, V. Kloper, A. Kigel, 4.57 R. Saito, G. Dresselhaus, M.S. Dresselhaus: Physi-
M.S. Eisen, S. Berger: Synthesis and characteriza- cal Properties of Carbon Nanotubes (Imperial College
tion of PbSe quantum wires, multipods, quantum Press, London 1998)
rods, cubes, Nano Lett. 3, 857–862 (2003) 4.58 M.S. Dresselhaus, G. Dresselhaus, P. Avouris: Carbon
4.43 W. Lu, P. Gao, W.B. Jian, Z.L. Wang, J. Fang: Perfect nanotubes: Synthesis, structure, properties and ap-
orientation ordered in-situ one-dimensional self- plications, Springer Ser. Top. Appl. Phys. 80, 1–447
assembly of Mn-doped PbSe nanocrystals, J. Am. (2001)
Chem. Soc. 126, 14816–14821 (2004) 4.59 R.C. Haddon: Special issue on carbon nanotubes,
4.44 F. Favier, E.C. Walter, M.P. Zach, T. Benter, Acc. Chem. Res. 35, 997–1113 (2002)
Part A 4
R.M. Penner: Hydrogen sensors and switches from 4.60 Y.-M. Lin, X. Sun, S. Cronin, Z. Zhang, J.Y. Ying,
electrodeposited palladium mesowire arrays, Sci- M.S. Dresselhaus: Fabrication, transport properties
ence 293, 2227–2231 (2001) of Te-doped bismuth nanowire arrays. In: Molecular
4.45 B. Gates, B. Mayers, B. Cattle, Y. Xia: Synthesis, Electronics: MRS Symposium Proceedings, Vol. 582,
characterization of uniform nanowires of trigonal ed. by S.T. Pantelides, M.A. Reed, J. Murday, A. Avi-
selenium, Adv. Funct. Mater. 12, 219–227 (2002) ran (Materials Research Society Press, Pittsburgh
4.46 C.A. Huber, T.E. Huber, M. Sadoqi, J.A. Lubin, S. Man- 2000) pp. 1–6, Chap. H10.3
alis, C.B. Prater: Nanowire array composites, Science 4.61 C.R. Martin: Nanomaterials: A membrane-based
263, 800–802 (1994) synthetic approach, Science 266, 1961–1966 (1994)
4.47 Y. Cui, L.J. Lauhon, M.S. Gudiksen, J. Wang, 4.62 G.A. Ozin: Nanochemistry: synthesis in diminishing
C.M. Lieber: Diameter-controlled synthesis of sin- dimensions, Adv. Mater. 4, 612–649 (1992)
gle crystal silicon nanowires, Appl. Phys. Lett. 78, 4.63 R.J. Tonucci, B.L. Justus, A.J. Campillo, C.E. Ford:
2214–2216 (2001) Nanochannel array glass, Science 258, 783–785
4.48 A.M. Morales, C.M. Lieber: A laser ablation method (1992)
for the synthesis of crystalline semiconductor 4.64 J.Y. Ying: Nanoporous systems and templates, Sci.
nanowires, Science 279, 208–211 (1998) Spec. 18, 56–63 (1999)
4.49 N. Wang, Y.F. Zhang, Y.H. Tang, C.S. Lee, S.T. Lee: 4.65 J.W. Diggle, T.C. Downie, C.W. Goulding: Anodic ox-
SiO2 -enhanced synthesis of Si nanowires by laser ide films on aluminum, Chem. Rev. 69, 365–405
ablation, Appl. Phys. Lett. 73, 3902–3904 (1998) (1969)
4.50 M.K. Sunkara, S. Sharma, R. Miranda, G. Lian, 4.66 J.P. O’Sullivan, G.C. Wood: The morphology and
E.C. Dickey: Bulk synthesis of silicon nanowires us- mechanism of formation of porous anodic films
ing a low-temperature vapor-liquid-solid method, on aluminum, Proc. R. Soc. Lond. A 317, 511–543
Appl. Phys. Lett. 79, 1546–1548 (2001) (1970)
4.51 S. Vaddiraju, H. Chandrasekaran, M.K. Sunkara: Va- 4.67 A.P. Li, F. Müller, A. Birner, K. Neilsch, U. Gösele:
por phase synthesis of tungsten nanowires, J. Am. Hexagonal pore arrays with a 50-420nm interpore
Chem. Soc. 125, 10792–10793 (2003) distance formed by self-organization in anodic alu-
4.52 J.P. Heremans, C.M. Thrush, D.T. Morelli, M.-C. Wu: mina, J. Appl. Phys. 84, 6023–6026 (1998)
Thermoelectric power of bismuth nanocomposites, 4.68 J.P. Sullivan, G.C. Wood: The morphology, mech-
Phys. Rev. Lett. 88, 216801–1–216801–4 (2002) anism of formation of porous anodic films on
4.53 Y. Li, G.S. Cheng, L.D. Zhang: Fabrication of highly aluminum, Proc. R. Soc. Lond. A 317, 511–543 (1970)
ordered ZnO nanowire arrays in anodic alumina 4.69 O. Jessensky, F. Müller, U. Gösele: Self-organized
membranes, J. Mater. Res. 15, 2305–2308 (2000) formation of hexagonal pore arrays in anodic alu-
4.54 P. Yang, H. Yan, S. Mao, R. Russo, J. Johnson, mina, Appl. Phys. Lett. 72, 1173–1175 (1998)
R. Saykally, N. Morris, J. Pham, R. He, H.-J. Choi: 4.70 F. Li, L. Zhang, R.M. Metzger: On the growth of highly
Controlled growth of ZnO nanowires and their optical ordered pores in anodized aluminum oxide, Chem.
properties, Adv. Funct. Mater. 12, 323–331 (2002) Mater. 10, 2470–2480 (1998)
4.55 M.J. Zheng, L.D. Zhang, G.H. Li, W.Z. Shen: Fabrica- 4.71 H. Masuda, M. Satoh: Fabrication of gold nanodot
tion and optical properties of large-scale uniform array using anodic porous alumina as an evap-
zinc oxide nanowire arrays by one-step electro- oration mask, Jpn. J. Appl. Phys. 35, L126–L129
chemical deposition technique, Chem. Phys. Lett. (1996)
363, 123–128 (2002) 4.72 E. Ferain, R. Legras: Track-etched membrane – dy-
4.56 M.S. Dresselhaus, Y.-M. Lin, O. Rabin, A. Jorio, namics of pore formation, Nucl. Instrum. Methods B
A.G. Souza Filho, M.A. Pimenta, R. Saito, G.G. Sam- 84, 331–336 (1993)
162 Part A Nanostructures, Micro-/Nanofabrication and Materials
4.73 A. Blondel, J.P. Meier, B. Doudin, J.-P. Ansermet: Gi- 4.89 Y. Wu, P. Yang: Direct observation of vapor-liquid-
ant magnetoresistance of nanowires of multilayers, solid nanowire growth, J. Am. Chem. Soc. 123, 3165–
Appl. Phys. Lett. 65, 3019–3021 (1994) 3166 (2001)
4.74 K. Liu, C.L. Chien, P.C. Searson, Y.Z. Kui: Structural 4.90 Y. Wu, R. Fan, P. Yang: Block-by-block growth
and magneto-transport properties of electrode- of single-crystalline Si/SiGe superlattice nanowires,
posited bismuth nanowires, Appl. Phys. Lett. 73, Nano Lett. 2, 83–86 (2002)
1436–1438 (1998) 4.91 S. Sharma, M.K. Sunkara, R. Miranda, G. Lian,
4.75 C.A. Huber, T.E. Huber: A novel microstructure: E.C. Dickey: A novel low temperature synthesis
semiconductor-impregnated porous Vycor glass, method for semiconductor nanowires. In: Syn-
J. Appl. Phys. 64, 6588–6590 (1988) thesis, Functional Properties and Applications of
4.76 J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, Nanostructures: Mat. Res. Soc. Symp. Proc., San
C.T. Kresge, K.D. Schmitt, C.T.-W. Chu, D.H. Ol- Francisco, Vol. 676, ed. by H.W. Hahn, D.L. Feldheim,
son, E.W. Sheppard, S.B. McCullen, J.B. Higgins, C.P. Kubiak, R. Tannenbaum, R.W. Siegel (Materials
J.L. Schlenker: A new family of mesoporous mo- Research Society Press, Pittsburgh 2001) p. Y1.6
lecular sieves prepared with liquid crystal templates, 4.92 M.S. Gudiksen, L.J. Lauhon, J. Wang, D.C. Smith,
Part A 4
J. Am. Chem. Soc. 114, 10834–10843 (1992) C.M. Lieber: Growth of nanowire superlattice struc-
4.77 C.-G. Wu, T. Bein: Conducting polyaniline filaments tures for nanoscale photonics and electronics,
in a mesoporous channel host, Science 264, 1757– Nature 415, 617–620 (2002)
1759 (1994) 4.93 Y. Wu, J. Xiang, C. Yang, W. Lu, C.M. Lieber: Single-
4.78 Y.-M. Lin, S.B. Cronin, J.Y. Ying, M.S. Dresselhaus, crystal metallic nanowires and metal/semiconductor
J.P. Heremans: Transport properties of Bi nanowire nanowire heterostructures, Nature 430, 61–65
arrays, Appl. Phys. Lett. 76, 3944–3946 (2000) (2004)
4.79 T. Thurn-Albrecht, J. Schotter, G.A. Kästle, N. Em- 4.94 M.T. Björk, B.J. Ohlsson, T. Sass, A.I. Persson, C. The-
ley, T. Shibauchi, L. Krusin-Elbaum, K. Guarini, lander, M.H. Magnusson, K. Deppert, L.R. Wallen-
C.T. Black, M.T. Tuominen, T.P. Russell: Ultrahigh- berg, L. Samuelson: One-dimensional steeplechase
density nanowire arrays grown in self-assembled for electrons realized, Nano Lett. 2, 87–89 (2002)
diblock copolymer templates, Science 290, 2126–2129 4.95 N. Wang, Y.H. Tang, Y.F. Zhang, C.S. Lee, S.T. Lee:
(2000) Nucleation and growth of Si nanowires from silicon
4.80 A.W. Adamson: Physical Chemistry of Surfaces (Wiley, oxide, Phys. Rev. B 58, R16024–R16026 (1998)
New York 1982) p. 338 4.96 Y.F. Zhang, Y.H. Tang, N. Wang, C.S. Lee, I. Bello,
4.81 R. Ferré, K. Ounadjela, J.M. George, L. Piraux, S.T. Lee: One-dimensional growth mechanism of
S. Dubois: Magnetization processes in nickel and crystalline silicon nanowires, J. Cryst. Growth 197,
cobalt electrodeposited nanowires, Phys. Rev. B 56, 136–140 (1999)
14066–14075 (1997) 4.97 S.T. Lee, Y.F. Zhang, N. Wang, Y.H. Tang, I. Bello,
4.82 H. Zeng, M. Zheng, R. Skomski, D.J. Sellmyer, Y. Liu, C.S. Lee, Y.W. Chung: Semiconductor nanowires from
L. Menon, S. Bandyopadhyay: Magnetic properties oxides, J. Mater. Res. 14, 4503–4507 (1999)
of self-assembled Co nanowires of varying length 4.98 D.D.D. Ma, C.S. Lee, Y. Lifshitz, S.T. Lee: Periodic ar-
and diameter, J. Appl. Phys. 87, 4718–4720 (2000) ray of intramolecular junctions of silicon nanowires,
4.83 Y. Peng, H.L. Zhang, S.-L. Pan, H.-L. Li: Mag- Appl. Phys. Lett. 81, 3233–3235 (2002)
netic properties and magnetization reversal of α-Fe 4.99 D. Whang, S. Jin, C.M. Lieber: Large-scale hier-
nanowires deposited in alumina film, J. Appl. Phys. archical organization of nanowires for functional
87, 7405–7408 (2000) nanosystems, Jpn. J. Appl. Phys. 43, 4465–4470
4.84 L. Piraux, J.M. George, J.F. Despres, C. Leroy, (2004)
E. Ferain, R. Legras, K. Ounadjela, A. Fert: Gi- 4.100 M.P. Zach, K. Inazu, K.H. Ng, J.C. Hemminger,
ant magnetoresistance in magnetic multilayered R.M. Penner: Synthesis of molybdenum nanowires
nanowires, Appl. Phys. Lett. 65, 2484–2486 (1994) with millimeter-scale lengths using electrochemical
4.85 S. Bhattacharrya, S.K. Saha, D. Chakravorty: step edge decoration, Chem. Mater. 14, 3206–3216
Nanowire formation in a polymeric film, Appl. Phys. (2002)
Lett. 76, 3896–3898 (2000) 4.101 B. Gates, Y. Yin, Y. Xia: A solution-phase approach
4.86 G. Yi, W. Schwarzacher: Single crystal superconduc- to the synthesis of uniform nanowires of crystalline
tor nanowires by electrodeposition, Appl. Phys. Lett. selenium with lateral dimensions in the range of
74, 1746–1748 (1999) 10-30 nm, J. Am. Chem. Soc. 122, 12582–12583 (2000)
4.87 D. Al-Mawlawi, C.Z. Liu, M. Moskovits: Nanowires 4.102 B. Mayers, B. Gates, Y. Yin, Y. Xia: Large-scale syn-
formed in anodic oxide nanotemplates, J. Mater. thesis of monodisperse nanorods of Se/Te alloys
Res. 9, 1014–1018 (1994) through a homogeneous nucleation and solution
4.88 R.S. Wagner, W.C. Ellis: Vapor-liquid-solid mecha- growth process, Adv. Mater. 13, 1380–1384 (2001)
nism of single crystal growth, Appl. Phys. Lett. 4, 4.103 B. Gates, Y. Wu, Y. Yin, P. Yang, Y. Xia: Single-
89–90 (1964) crystalline nanowires of Ag2 Se can be synthesized by
Nanowires References 163
templating against nanowires of trigonal Se, J. Am. self-assembled alumina templates. In: Nonlitho-
Chem. Soc. 123, 11500–11501 (2001) graphic and Lithographic Methods for Nanofabrica-
4.104 B. Gates, B. Mayers, Y. Wu, Y. Sun, B. Cattle, P. Yang, tion: MRS Symposium Proceedings, Boston, Vol. 636,
Y. Xia: Synthesis and characterization of crystalline ed. by J.A. Rogers, A. Karim, L. Merhari, D. Norris,
Ag2 Se nanowires through a template-engaged re- Y. Xia (Materials Research Society Press, Pittsburgh
action at room temperature, Adv. Funct. Mater. 12, 2001) pp. D4.7–1–D4.7–6
679–686 (2002) 4.120 O. Rabin, P.R. Herz, Y.-M. Lin, A.I. Akinwande,
4.105 H. Yu, P.C. Gibbons, W.E. Buhro: Bismuth, tellurium S.B. Cronin, M.S. Dresselhaus: Formation of thick
and bismuth telluride nanowires, J. Mater. Chem. porous anodic alumina films and nanowire arrays
14, 595–602 (2004) on silicon wafers and glass, Adv. Funct. Mater. 13,
4.106 X. Peng, J. Wickham, A.P. Alivisatos: Kinetics of II–VI, 631–638 (2003)
III–V colloidal semiconductor nanocrystal growth: 4.121 O. Rabin, P.R. Herz, Y.-M. Lin, S.B. Cronin, A.I. Akin-
’Focusing’ of size distributions, J. Am. Chem. Soc. wande, M.S. Dresselhaus: Arrays of nanowires on
120, 5343–5344 (1998) silicon wafers, 21st Int. Conf. Thermoelectr. Proc.
4.107 J.Y. Lao, J.G. Wen, Z.F. Ren: Hierarchical ZnO nano- ICT ’02 Long Beach (IEEE, Piscataway 2002) pp. 276–
Part A 4
structures, Nano Lett. 2, 1287–1291 (2002) 279
4.108 J.Y. Lao, J.Y. Huang, D.Z. Wang, Z.F. Ren: ZnO 4.122 M.H. Huang, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind,
nanobridges and nanonails, Nano Lett. 3, 235–238 E. Weber, R. Russo, P. Yang: Room-temperature
(2003) ultraviolet nanowire nanolasers, Science 292, 1897–
4.109 N.A. Melosh, A. Boukai, F. Diana, B. Gerardot, 1899 (2001)
A. Badolto, P.M. Petroff, J.R. Heath: Ultrahigh- 4.123 Y.H. Tang, Y.F. Zhang, N. Wang, C.S. Lee, X.D. Han,
density nanowire lattices and circuits, Science 300, I. Bello, S.T. Lee: Morphology of Si nanowires
112–115 (2003) synthesized by high-temperature laser ablation,
4.110 S.R. Nicewarner-Peña, R.G. Freeman, B.D. Reiss, J. Appl. Phys. 85, 7981–7983 (1999)
L. He, D.J. Peña, I.D. Walton, R. Cromer, C.D. Keat- 4.124 Y. Ding, Z.L. Wang: Structure analysis of nanowires
ing, M.J. Natan: Submicrometer metallic barcodes, and nanobelts by transmission electron microscopy,
Science 294, 137–141 (2001) J. Phys. Chem. B 108, 12280–12291 (2004)
4.111 L.J. Lauhon, M.S. Gudiksen, D. Wang, C.M. Lieber: 4.125 S.B. Cronin, Y.-M. Lin, O. Rabin, M.R. Black,
Epitaxial core-shell and core-multishell nanowire G. Dresselhaus, M.S. Dresselhaus, P.L. Gai: Bismuth
heterostructures, Nature 420, 57–61 (2002) nanowires for potential applications in nanoscale
4.112 Z.L. Wang, Z.R. Dai, R.P. Gao, Z.G. Bai, J.L. Gole: Side- electronics technology, Microsc. Microanal. 8, 58–63
by-side silicon carbide-silica biaxial nanowires: (2002)
Synthesis, structure and mechanical properties, 4.126 M.S. Sander, R. Gronsky, Y.-M. Lin, M.S. Dressel-
Appl. Phys. Lett. 77, 3349–3351 (2000) haus: Plasmon excitation modes in nanowire arrays,
4.113 P. Yang, F. Kim: Langmuir–Blodgett assembly of J. Appl. Phys. 89, 2733–2736 (2001)
one-dimensional nanostructures, ChemPhysChem 3, 4.127 L. Venkataraman, C.M. Lieber: Molybdenum selenide
503–506 (2002) molecular wires as one-dimensional conductors,
4.114 B. Messer, J.H. Song, P. Yang: Microchannel net- Phys. Rev. Lett. 83, 5334–5337 (1999)
works for nanowire patterning, J. Am. Chem. Soc. 4.128 A. Majumdar: Scanning thermal microscopy, Annu.
122, 10232–10233 (2000) Rev. Mater. Sci. 29, 505–585 (1999)
4.115 P.A. Smith, C.D. Nordquist, T.N. Jackson, T.S. Mayer, 4.129 K.M. Unruh, T.E. Huber, C.A. Huber: Melting and
B.R. Martin, J. Mbindyo, T.E. Mallouk: Electric- freezing behavior of indium metal in porous glasses,
field assisted assembly and alignment of metallic Phys. Rev. B 48, 9021–9027 (1993)
nanowires, Appl. Phys. Lett. 77, 1399–1401 (2000) 4.130 Y.Y. Wu, P.D. Yang: Melting and welding semi-
4.116 S. Jin, D.M. Whang, M.C. McAlpine, R.S. Friedman, conductor nanowires in nanotubes, Adv. Mater. 13,
Y. Wu, C.M. Lieber: Scalable interconnection and in- 520–523 (2001)
tegration of nanowire devices without registration, 4.131 P.M. Ajayan, S. Iijima: Capillarity-induced fill-
Nano Lett. 4, 915–919 (2004) ing of carbon nanotubes, Nature 361, 333–334
4.117 T. Kuykendall, P.J. Pauzauskie, Y.F. Zhang, J. Gold- (1993)
berger, D. Sirbuly, J. Denlinger, P.D. Yang: Crystal- 4.132 Y. Gao, Y. Bando: Carbon nanothermometer con-
lographic alignment of high-density gallium nitride taining gallium, Nature 415, 599 (2002)
nanowire arrays, Nat. Mater. 3, 524–528 (2004) 4.133 J.L. Costa-Krämer, N. Garcia, H. Olin: Conductance
4.118 H. Masuda, H. Yamada, M. Satoh, H. Asoh, M. Nakao, quantization histograms of gold nanowires at 4 K,
T. Tamamura: Highly ordered nanochannel-array Phys. Rev. B 55, 12910–12913 (1997)
architecture in anodic alumina, Appl. Phys. Lett. 71, 4.134 M.E.T. Morales, A.G. Balogh, T.W. Cornelius, R. Neu-
2770–2772 (1997) mann, C. Trautmann: Fragmentation of nanowires
4.119 O. Rabin, P.R. Herz, S.B. Cronin, Y.-M. Lin, A.I. Ak- driven by Rayleigh instability, Appl. Phys. Lett. 84,
inwande, M.S. Dresselhaus: Nanofabrication using 5337–5339 (2004)
164 Part A Nanostructures, Micro-/Nanofabrication and Materials
4.135 D.A. Wharam, T.J. Thornton, R. Newbury, M. Pep- single-crystalline bismuth nanowire arrays, Appl.
per, H. Ahmed, J.E.F. Frost, D.G. Hasko, D.C. Peacock, Phys. Lett. 73, 1589–1591 (1998)
D.A. Ritchie, G.A.C. Jones: One-dimensional trans- 4.150 J. Heremans, C.M. Thrush, Z. Zhang, X. Sun,
port and the quantization of the ballistic resistance, M.S. Dresselhaus, J.Y. Ying, D.T. Morelli: Magne-
J. Phys. C 21, L209–L214 (1988) toresistance of bismuth nanowire arrays: A pos-
4.136 B.J. van Wees, H. van Houten, C.W.J. Beenakker, sible transition from one-dimensional to three-
J.G. Williamson, L.P. Kouvenhoven, D. van der Marel, dimensional localization, Phys. Rev. B 58, R10091–
C.T. Foxon: Quantized conductance of point contacts R10095 (1998)
in a two-dimensional electron gas, Phys. Rev. Lett. 4.151 L. Sun, P.C. Searson, C.L. Chien: Finite-size effects
60, 848–850 (1988) in nickel nanowire arrays, Phys. Rev. B 61, R6463–
4.137 C.J. Muller, J.M. van Ruitenbeek, L.J. de- R6466 (2000)
Jongh: Conductance and supercurrent disconti- 4.152 Y.-M. Lin, S.B. Cronin, O. Rabin, J.Y. Ying,
nuities in atomic-scale metallic constrictions of M.S. Dresselhaus: Transport properties and ob-
variable width, Phys. Rev. Lett. 69, 140–143 servation of semimetal-semiconductor transition
(1992) in Bi-based nanowires. In: Quantum Confined
Part A 4
4.138 C.J. Muller, J.M. Krans, T.N. Todorov, M.A. Reed: Semiconductor Nanostructures: MRS Symposium
Quantization effects in the conductance of metal- Proceedings, Boston, Vol. 737-C, ed. by J.M. Buriak,
lic contacts at room temperature, Phys. Rev. B 53, D.D.M. Wayner, F. Priolo, B. White, V. Klimov,
1022–1025 (1996) L. Tsybeskov (Materials Research Society Press, Pitts-
4.139 J.L. Costa-Krämer, N. Garcia, H. Olin: Conductance burgh 2003) p. F3.14
quantization in bismuth nanowires at 4 K, Phys. Rev. 4.153 Y.-M. Lin, M.S. Dresselhaus: Transport properties
Lett. 78, 4990–4993 (1997) of superlattice nanowires and their potential for
4.140 C.Z. Li, H.X. He, A. Bogozi, J.S. Bunch, N.J. Tao: thermoelectric applications. In: Quantum Confined
Molecular detection based on conductance quanti- Semiconductor Nanostructures: MRS Symposium
zation of nanowires, Appl. Phys. Lett. 76, 1333–1335 Proceedings, Boston, Vol. 737-C, ed. by J.M. Buriak,
(2000) D.D.M. Wayner, F. Priolo, B. White, V. Klimov,
4.141 J.L. Costa-Krämer, N. Garcia, P. Garcia-Mochales, L. Tsybeskov (Materials Research Society Press, Pitts-
P.A. Serena, M.I. Marques, A. Correia: Conductance burgh 2003) p. F8.18
quantization in nanowires formed between micro 4.154 Y.-M. Lin, O. Rabin, S.B. Cronin, J.Y. Ying, M.S. Dres-
and macroscopic metallic electrodes, Phys. Rev. B selhaus: Semimetal-semiconductor transition in
55, 5416–5424 (1997) Bi1−x Sbx alloy nanowires and their thermoelectric
4.142 Y. Huang, X. Duan, Y. Cui, L.J. Lauhon, K.-H. Kim, properties, Appl. Phys. Lett. 81, 2403–2405 (2002)
C.M. Lieber: Logic gates and computation from as- 4.155 J. Heremans, C.M. Thrush, Y.-M. Lin, S.B. Cronin,
sembled nanowire building blocks, Science 294, M.S. Dresselhaus: Transport properties of antimony
1313–1317 (2001) nanowires, Phys. Rev. B 63, 085406–1–085406–8
4.143 J.-R. Kim, H. Oh, H.M. So, J.-J. Kim, J. Kim, C.J. Lee, (2001)
S.C. Lyu: Schottky diodes based on a single GaN 4.156 Y.-M. Lin, S.B. Cronin, O. Rabin, J.Y. Ying, M.S. Dres-
nanowire, Nanotechnology 13, 701–704 (2002) selhaus: Transport properties of Bi1−x Sbx alloy
4.144 X. Duan, Y. Huang, C.M. Lieber: Nonvolatile mem- nanowires synthesized by pressure injection, Appl.
ory and programmable logic from molecule-gated Phys. Lett. 79, 677–679 (2001)
nanowires, Nano Lett. 2, 487–490 (2002) 4.157 Z. Zhang, X. Sun, M.S. Dresselhaus, J.Y. Ying,
4.145 E.C. Walter, R.M. Penner, H. Liu, K.H. Ng, M.P. Zach, J. Heremans: Electronic transport properties of sin-
F. Favier: Sensors from electrodeposited metal gle crystal bismuth nanowire arrays, Phys. Rev. B 61,
nanowires, Surf. Interface Anal. 34, 409–412 4850–4861 (2000)
(2002) 4.158 D.E. Beutler, N. Giordano: Localization and electron-
4.146 E.C. Walter, K.H. Ng, M.P. Zach, R.M. Penner, electron interaction effects in thin Bi wires and
F. Favier: Electronic devices from electrodeposited films, Phys. Rev. B 38, 8–19 (1988)
metal nanowires, Microelectron. Eng. 61/62, 555– 4.159 J. Heremans, C.M. Thrush: Thermoelectric power of
561 (2002) bismuth nanowires, Phys. Rev. B 59, 12579–12583
4.147 Y.-M. Lin, X. Sun, M.S. Dresselhaus: Theoretical in- (1999)
vestigation of thermoelectric transport properties of 4.160 L.D. Hicks, M.S. Dresselhaus: Thermoelectric figure of
cylindrical Bi nanowires, Phys. Rev. B 62, 4610–4623 merit of a one-dimensional conductor, Phys. Rev. B
(2000) 47, 16631–16634 (1993)
4.148 K. Liu, C.L. Chien, P.C. Searson: Finite-size effects in 4.161 Y.-M. Lin, S.B. Cronin, O. Rabin, J. Heremans,
bismuth nanowires, Phys. Rev. B 58, R14681–R14684 M.S. Dresselhaus, J.Y. Ying: Transport properties
(1998) of Bi-related nanowire systems. In: Anisotropic
4.149 Z. Zhang, X. Sun, M.S. Dresselhaus, J.Y. Ying, J. Here- Nanoparticles: Synthesis, Characterization and Ap-
mans: Magnetotransport investigations of ultrafine plications: MRS Symposium Proceedings, Boston,
Nanowires References 165
Vol. 635, ed. by S. Stranick, P.C. Searson, L.A. Lyon, 4.179 M.R. Black, Y.-M. Lin, S.B. Cronin, O. Rabin,
C. Keating (Materials Research Society Press, Pitts- M.S. Dresselhaus: Infrared absorption in bismuth
burgh 2001) pp. C4301–C4306 nanowires resulting from quantum confinement,
4.162 Y.-M. Lin, M.S. Dresselhaus: Thermoelectric prop- Phys. Rev. B 65, 195417–1–195417–9 (2002)
erties of superlattice nanowires, Phys. Rev. B 68, 4.180 M.W. Lee, H.Z. Twu, C.-C. Chen, C.H. Chen: Op-
075304 (2003) tical characterization of wurtzite gallium nitride
4.163 M.T. Björk, B.J. Ohlsson, C. Thelander, A.I. Persson, nanowires, Appl. Phys. Lett. 79, 3693–3695 (2001)
K. Deppert, L.R. Wallenberg, L. Samuelson: Nanowire 4.181 D.M. Lyons, K.M. Ryan, M.A. Morris, J.D. Holmes:
resonant tunneling diodes, Appl. Phys. Lett. 81, Tailoring the optical properties of silicon nanowire
4458–4460 (2002) arrays through strain, Nano Lett. 2, 811–816 (2002)
4.164 D. Li, Y. Wu, P. Kim, L. Shi, P. Yang, A. Ma- 4.182 M.S. Gudiksen, J. Wang, C.M. Lieber: Size-depent
jumdar: Thermal conductivity of individual silicon photoluminescence from single indium phosphide
nanowires, Appl. Phys. Lett. 83, 2934–2936 (2003) nanowires, J. Phys. Chem. B 106, 4036–4039 (2002)
4.165 T.S. Tighe, J.M. Worlock, M.L. Roukes: Direct ther- 4.183 J.C. Johnson, H. Yan, R.D. Schaller, L.H. Haber,
mal conductance measurements on suspended R.J. Saykally, P. Yang: Single nanowire lasers,
Part A 4
monocrystalline nanostructures, Appl. Phys. Lett. 70, J. Phys. Chem. B 105, 11387–11390 (2001)
2687–2689 (1997) 4.184 S. Blom, L.Y. Gorelik, M. Jonson, R.I. Shekhter,
4.166 S.T. Huxtable, A.R. Abramson, C.-L. Tien, A. Ma- A.G. Scherbakov, E.N. Bogachek, U. Landman:
jumdar, C. LaBounty, X. Fan, G. Zeng, J.E. Bowers, Magneto-optics of electronic transport in nanowires,
A. Shakouri, E.T. Croke: Thermal conductivity of Phys. Rev. B 58, 16305–16314 (1998)
Si/SiGe and SiGe/SiGe superlattices, Appl. Phys. Lett. 4.185 J.P. Pierce, E.W. Plummer, J. Shen: Ferromagnetism
80, 1737–1739 (2002) in cobalt-iron alloy nanowire arrays on w(110), Appl.
4.167 R. Venkatasubramanian, E. Siivola, T. Colpitts, Phys. Lett. 81, 1890–1892 (2002)
B. O’Quinn: Thin-film thermoelectric devices with 4.186 S. Melle, J.L. Menendez, G. Armelles, D. Navas,
high room-temperature figures of merit, Nature 413, M. Vazquez, K. Nielsch, R.B. Wehrsphon, U. Gösele:
597–602 (2001) Magneto-optical properties of nickel nanowire ar-
4.168 C. Dames, G. Chen: Modeling the thermal conduc- rays, Appl. Phys. Lett. 83, 4547–4549 (2003)
tivity of a SiGe segmented nanowire, 21st Int. Conf. 4.187 J.C. Johnson, H. Yan, R.D. Schaller, P.B. Petersen,
Thermoelectr. Proc. ICT ’02, Long Beach (IEEE, Piscat- P. Yang, R.J. Saykally: Near-field imaging of non-
away 2002) pp. 317–320 linear optical mixing in single zinc oxide nanowires,
4.169 K. Schwab, J.L. Arlett, J.M. Worlock, M.L. Roukes: Nano Lett. 2, 279–283 (2002)
Thermal conductance through discrete quantum 4.188 M.R. Black, Y.-M. Lin, S.B. Cronin, M.S. Dresselhaus:
channels, Physica E 9, 60–68 (2001) Using optical measurements to improve electronic
4.170 D. Li, Y. Wu, R. Fan, P. Yang, A. Majumdar: Thermal models of bismuth nanowires, 21st Int. Conf. Ther-
conductivity of Si/SiGe superlattice nanowires, Appl. moelectr. Proc. ICT ’02, Long Beach, ed. by T. Caillat,
Phys. Lett. 83, 3186–3188 (2003) J. Snyder (IEEE, Piscataway 2002) pp. 253–256
4.171 G. Chen, M.S. Dresselhaus, G. Dresselhaus, J.- 4.189 M.R. Black, P.L. Hagelstein, S.B. Cronin, Y.-M. Lin,
P. Fleurial, T. Caillat: Recent developments in M.S. Dresselhaus: Optical absorption from an indi-
thermoelectric materials, Int. Mater. Rev. 48, 45–66 rect transition in bismuth nanowires, Phys. Rev. B
(2003) 68, 235417 (2003)
4.172 C. Dames, G. Chen: Theoretical phonon thermal con- 4.190 H.-L. Liu, C.-C. Chen, C.-T. Chia, C.-C. Yeh, C.-
ductivity of Si-Ge superlattice nanowires, J. Appl. H. Chen, M.-Y. Yu, S. Keller, S.P. DenBaars: Infrared
Phys. 95, 682–693 (2004) and Raman-scattering studies in single-crystalline
4.173 K. Schwab, E.A. Henriksen, J.M. Worlock, M.L. Roukes: GaN nanowires, Chem. Phys. Lett. 345, 245–251
Measurement of the quantum of thermal conduc- (2001)
tance, Nature 404, 974–977 (2000) 4.191 H. Richter, Z.P. Wang, L. Ley: The one phonon
4.174 M. Cardona: Light Scattering in Solids (Springer, Raman-spectrum in microcrystalline silicon, Solid
Berlin Heidelberg 1982) State Commun. 39, 625–629 (1981)
4.175 P.Y. Yu, M. Cardona: Fundamentals of Semiconduc- 4.192 I.H. Campbell, P.M. Fauchet: The effects of mi-
tors (Springer, Berlin Heidelberg 1995), Chap. 7 crocrystal size and shape on the one phonon
4.176 J.C.M. Garnett: Colours in metal glasses, in metallic Raman-spectra of crystalline semiconductors, Solid
films, and in metallic solutions, Philos. Trans. R. Soc. State Commun. 58, 739–741 (1986)
Lond. A 205, 237–288 (1906) 4.193 R. Gupta, Q. Xiong, C.K. Adu, U.J. Kim, P.C. Eklund:
4.177 D.E. Aspnes: Optical properties of thin films, Thin Laser-induced Fano resonance scattering in silicon
Solid Films 89, 249–262 (1982) nanowires, Nano Lett. 3, 627–631 (2003)
4.178 U. Kreibig, L. Genzel: Optical absorption of small 4.194 X. Duan, Y. Huang, Y. Cui, J. Wang, C.M. Lieber:
metallic particles, Surf. Sci. 156, 678–700 (1985) Indium phosphide nanowires as building blocks
166 Part A Nanostructures, Micro-/Nanofabrication and Materials
for nanoscale electronic and optoelectronic devices, 4.209 G. Dresselhaus, M.S. Dresselhaus, Z. Zhang, X. Sun,
Nature 409, 66–69 (2001) J. Ying, G. Chen: Modeling thermoelectric behavior
4.195 Y. Cui, C.M. Lieber: Functional nanoscale electronic in Bi nano-wires, 17th Int. Conf. Thermoelectr. Proc.
devices assembled using silicon nanowire building ICT’98, Nagoya, ed. by K. Koumoto (IEEE, Piscataway
blocks, Science 291, 851–853 (2001) 1998) pp. 43–46
4.196 Y. Cui, X. Duan, J. Hu, C.M. Lieber: Doping and elec- 4.210 L.D. Hicks, M.S. Dresselhaus: The effect of quantum
trical transport in silicon nanowires, J. Phys. Chem. well structures on the thermoelectric figure of merit,
B 104, 101–104 (2000) Phys. Rev. B 47, 12727–12731 (1993)
4.197 G.F. Zheng, W. Lu, S. Jin, C.M. Lieber: Synthesis 4.211 O. Rabin, Y.-M. Lin, M.S. Dresselhaus: Anoma-
and fabrication of high-performance n-type sili- lously high thermoelectric figure of merit in Bi1−x Sbx
con nanowire transistors, Adv. Mater. 16, 1890–1891 nanowires by carrier pocket alignment, Appl. Phys.
(2004) Lett. 79, 81–83 (2001)
4.198 J. Goldberger, D.J. Sirbuly, M. Law, P. Yang: ZnO 4.212 T.C. Harman, P.J. Taylor, M.P. Walsh, B.E. LaForge:
nanowire transistors, J. Phys. Chem. B 109, 9–14 Quantum dot superlattice thermoelectric materials
(2005) and devices, Science 297, 2229–2232 (2002)
Part A 4
4.199 D.H. Kang, J.H. Ko, E. Bae, J. Hyun, W.J. Park, 4.213 X. Duan, Y. Huang, R. Agarwal, C.M. Lieber: Single-
B.K. Kim, J.J. Kim, C.J. Lee: Ambient air effects on nanowire electrically driven lasers, Nature 421, 241
electrical characteristics of gap nanowire transistors, (2003)
J. Appl. Phys. 96, 7574–7577 (2004) 4.214 F. Qian, Y. Li, S. Gradecak, D.L. Wang, C.J. Barrelet,
4.200 S.-W. Chung, J.-Y. Yu, J.R. Heath: Silicon nanowire C.M. Lieber: Gallium nitride-based nanowire radial
devices, Appl. Phys. Lett. 76, 2068–2070 (2000) heterostructures for nanophotonics, Nano Lett. 4,
4.201 C. Li, W. Fan, B. Lei, D. Zhang, S. Han, T. Tang, X. Liu, 1975–1979 (2004)
Z. Liu, S. Asano, M. Meyyappan, J. Han, C. Zhou: Mul- 4.215 V. Dneprovskii, E. Zhukov, V. Karavanskii, V. Poborchii,
tilevel memory based on molecular devices, Appl. I. Salamatini: Nonlinear optical properties of semi-
Phys. Lett. 84, 1949–1951 (2004) conductor quantum wires, Superlattice. Microst.
4.202 B. Lei, C. Li, D.Q. Zhang, Q.F. Zhou, K. Shung, 23(6), 1217–1221 (1998)
C.W. Zhou: Nanowire transistors with ferroelec- 4.216 J.C. Johnson, K.P. Knutsen, H. Yan, M. Law, Y. Zhang,
tric gate dielectrics: Enhanced performance and P. Yang, R.J. Saykally: Ultrafast carrier dynamics in
memory effects, Appl. Phys. Lett. 84, 4553–4555 single ZnO nanowire and nanoribbon lasers, Nano
(2004) Lett. 4, 197–204 (2004)
4.203 H.T. Ng, J. Han, T. Yamada, P. Nguyen, Y.P. Chen, 4.217 J.X. Ding, J.A. Zapien, W.W. Chen, Y. Lifshitz, S.T. Lee,
M. Meyyappan: Single crystal nanowire vertical X.M. Meng: Lasing in ZnS nanowires grown on anodic
surround-gate field-effect transistor, Nano Lett. 4, aluminum oxide templates, Appl. Phys. Lett. 85, 2361
1247–1252 (2004) (2004)
4.204 M. Ding, H. Kim, A.I. Akinwande: Observation of va- 4.218 J.C. Johnson, H.-J. Choi, K.P. Knutsen, R.D. Schaller,
lence band electron emission from n-type silicon P. Yang, R.J. Saykally: Single gallium ni-
field emitter arrays, Appl. Phys. Lett. 75, 823–825 tride nanowire lasers, Nat. Mater. 1, 106–110
(1999) (2002)
4.205 F.C.K. Au, K.W. Wong, Y.H. Tang, Y.F. Zhang, 4.219 H.J. Choi, J.C. Johnson, R. He, S.K. Lee, F. Kim,
I. Bello, S.T. Lee: Electron field emission from sil- P. Pauzauskie, J. Goldberger, R.J. Saykally, P. Yang:
icon nanowires, Appl. Phys. Lett. 75, 1700–1702 Self-organized GaN quantum wire UV lasers, J. Phys.
(1999) Chem. B 107, 8721–8725 (2003)
4.206 P.M. Ajayan, O.Z. Zhou: Applications of carbon 4.220 C.J. Barrelet, A.B. Greytak, C.M. Lieber: Nanowire
nanotubes. In: Carbon Nanotubes: Synthesis, Struc- photonic circuit elements, Nano Lett. 4, 1981–1985
ture, Properties and Applications, Springer Ser. (2004)
Top. Appl. Phys., Vol. 80, ed. by M.S. Dresselhaus, 4.221 M. Law, D.J. Sirbuly, J.C. Johnson, J. Goldberger,
G. Dresselhaus, P. Avouris (Springer, Berlin Heidel- R.J. Saykally, P. Yang: Ultralong nanoribbon wave-
berg 2001) pp. 391–425 guides for sub-wavelength photonics integration,
4.207 M. Lu, M.K. Li, Z.J. Zhang, H.L. Li: Synthesis of car- Science 305, 1269–1273 (2004)
bon nanotubes/si nanowires core-sheath structure 4.222 H. Kind, H. Yan, B. Messer, M. Law, P. Yang:
arrays and their field emission properties, Appl. Surf. Nanowire ultraviolet photodetectors and optical
Sci. 218, 196–202 (2003) switches, Adv. Mater. 14, 158–160 (2002)
4.208 L. Vila, P. Vincent, L. Dauginet-DePra, G. Pirio, 4.223 B.M.I. van der Zande, M.R. Böhmer, L.G.J. Fokkink,
E. Minoux, L. Gangloff, S. Demoustier-Champagne, C. Schöneberger: Colloidal dispersions of gold rods:
N. Sarazin, E. Ferain, R. Legras, L. Piraux, P. Legag- synthesis and optical properties, Langmuir 16, 451–
neux: Growth and field-emission properties of 458 (2000)
vertically aligned cobalt nanowire arrays, Nano Lett. 4.224 B.M.I. van der Zande, G.J.M. Koper, H.N.W. Lekker-
4, 521–524 (2004) kerker: Alignment of rod-shaped gold particles by
Nanowires References 167
electric fields, J. Phys. Chem. B 103, 5754–5760 4.227 Y. Cui, Q. Wei, H. Park, C. Lieber: Nanowire nanosen-
(1999) sors for highly sensitive and selective detection of
4.225 W.U. Huynh, J.J. Dittmer, A.P. Alivisatos: Hybrid biological and chemical species, Science 293, 1289–
nanorod-polymer solar cells, Science 295, 2425–2427 1292 (2001)
(2002) 4.228 J. Hahm, C. Lieber: Direct ultra-sensitive electri-
4.226 L.A. Bauer, N.S. Birenbaum, G.J. Meyer: Biologi- cal detection of DNA and DNA sequence variations
cal applications of high aspect ratio nanoparticles, using nanowire nanosensors, Nano Lett. 4, 51–54
J. Mater. Chem. 14, 517–526 (2004) (2004)
Part A 4
169
Template-Ba
5. Template-Based Synthesis of Nanorod
Part A 5
capillary force and centrifugation, and chemical of Fullerenes and Metallofullerenes 180
conversion. In each section, the relevant funda-
mentals are first introduced, and then examples 5.4 Template Filling ................................... 180
are given to illustrate the specific details of each 5.4.1 Colloidal Dispersion (Sol) Filling ..... 180
technique. 5.4.2 Melt and Solution Filling ............... 181
5.4.3 Centrifugation ............................. 181
5.1 Template-Based Approach .................... 170 5.5 Converting from Reactive Templates....... 182
5.2 Electrochemical Deposition.................... 171 5.6 Summary and Concluding Remarks......... 182
5.2.1 Metals ........................................ 172 References .................................................. 183
Syntheses, characterizations and applications of nano- even more versatile method for various materials. Sub-
wires, nanorods, nanotubes and nanobelts (also often strate ledge or step-induced growth of nanowires or
referred to as one-dimensional nanostructures) are sig- nanorods has also been investigated intensively [5.7].
nificant areas of current endeavor in nanotechnology. Except for VLS and template-based growth, most of
Many techniques have been developed in these areas, the above-mentioned methods result in randomly ori-
and our understanding of the field has been signif- ented nanowires or nanorods (commonly in the form
icantly enhanced [5.1–5]. The field is still evolving of powder). The VLS method provides the ability to
rapidly with new synthesis methods and new nanowires grow well oriented nanorods or nanowires directly at-
or nanorods reported in the literature. Evaporation– tached to substrates, and is therefore often advantageous
condensation growth has been successfully applied to for characterization and applications; however, catalysts
the synthesis of various oxide nanowires and nanorods. are required to form a liquid capsule at the advanc-
Similarly, the dissolution–condensation method has ing surface during growth at elevated temperatures.
been widely used for the synthesis of various metal- In addition, the possible incorporation of catalyst into
lic nanowires from solutions. The vapor–liquid–solid nanowires and the difficulty removing such capsules
(VLS) growth method is a highly versatile approach; from the tips of nanowires or nanorods are two dis-
various elementary and compound semiconductor nano- advantages of this technique. Template-based growth
wires have been synthesized using this method [5.6]. often suffers from the polycrystalline nature of the
Template-based growth of nanowires or nanorods is an resultant nanowires and nanorods, in addition to the dif-
170 Part A Nanostructures, Micro-/Nanofabrication and Materials
ficulties involved in finding appropriate templates with ties of individual nanostructures. However, such studies
pore channels of a desired diameter, length and sur- generally require a lot of experimental preparation. For
face chemistry and in removing the template completely example, for electrical conductivity measurements, pat-
without compromising the integrity of grown nanowires terned electrodes are first created on a substrate, and
or nanorods. The discussion in this chapter will focus on then nanowires or nanorods are dispersed in an ap-
nanorod and nanowire arrays, although nanotube arrays propriate solvent or solution. This nanowire colloidal
are mentioned briefly in conjunction with nanorod and dispersion is then cast on the substrate containing pat-
nanowire fabrication. In addition, the terms of nanorod tern electrodes. Measurements are carried out after
and nanowire are used interchangeably without special identifying individual nanowires or nanorods bridging
distinction in this chapter; this is commonplace in the two electrodes. The options for manipulating nanowires
literature. or nanorods are limited, and it is difficult to improve the
In comparison with nanostructured materials in contact between the sample and the electrodes to ensure
other forms, nanorod arrays offer several advantages the desired ohmic contact. For practical applications,
for studying properties and for practical applications. the output or signal generated by single nanowire- or
Significant progress has been made in studies of nanorod-based devices is small, and the signal-to-noise
the physical properties of individual nanowires and ratio is small, which means that highly sensitive instru-
nanorods performed by directly measuring the proper- mentation is required to accommodate such devices.
Part A 5.1
of using PC as the template are its easy handling and ity and resistance to high temperatures, which allows
easy removal by means of pyrolysis at elevated tempera- the nanorods to densify completely before removal.
tures, but the flexibility of PC is more prone to distortion This results in fairly free-standing and unidirectionally-
during the heating process, and removal of the template aligned nanorod arrays with a larger surface area than
occurs before complete densification of the nanorods. for PC. The problem with AAM is the complete removal
These factors result in broken and deformed nanorods. of the template after nanorod growth, which is yet to be
The advantage of using AAM as the template is its rigid- achieved when using wet chemical etching.
Part A 5.2
RT
this method is only applicable to electrically conductive E = E0 + ln (ai ) , (5.1)
materials such as metals, alloys, semiconductors, and ni F
electrically conductive polymers. After the initial de- where E 0 is the standard electrode potential (or the
position, the electrode is separated from the depositing potential difference between the electrode and the so-
solution by the deposit and so the deposit must conduct lution) when the activity ai of the ions is unity, F is
in order to allow the deposition process to continue. Faraday’s constant, R is the gas constant, and T is the
When the deposition is confined to the pores of tem- temperature. When the electrode potential is higher than
a)
e
Vacant
MO
Potential
Energy level
of electrons
Occupied
MO
A+e A–
the energy level of a vacant molecular orbital in the The system used to perform electrolysis is called an
electrolyte, electrons will transfer from the electrode electrolytic cell. In this cell, the electrode connected to
to the solution and the electrolyte will be reduced, as the positive side of the power supply, termed the an-
shown in Fig. 5.1a [5.24]. On the other hand, if the ode, is where an oxidation reaction takes place, whereas
electrode potential is lower than the energy level of the electrode connected to the negative side of the
an occupied molecular orbital in the electrolyte, the power supply, the cathode, is where a reduction reac-
electrons will transfer from the electrolyte to the elec- tion proceeds, accompanied by deposition. Therefore,
trode, resulting in electrolyte oxidation, as illustrated in electrolytic deposition is also called cathode deposition,
Fig. 5.1b [5.24]. These reactions stop when equilibrium but it is most commonly referred to as electrochemical
is achieved. deposition or electrodeposition.
When an external electric field is applied between
two dissimilar electrodes, charged species flow from 5.2.1 Metals
one electrode to the other, and electrochemical reactions
occur at both electrodes. This process, called electrol- The growth of nanowires of conductive materials in
ysis, converts electrical energy to chemical potential. an electric field is a self-propagating process [5.27].
Once the small rods form, the electric field and the den-
a) sity of current lines between the tips of nanowires and
V the opposing electrode are greater than that between
two electrodes, due to the shorter distances between
the nanowires and the electrodes. This ensures that the
Part A 5.2
I
species being deposited is constantly attracted prefer-
entially to the nanowire tips, resulting in continued
growth. To better control the morphology and size, tem-
plates containing channels in the desired shape are used
to guide the growth of nanowires. Figure 5.2 illustrates
a common setup used for the template-based growth
Porous membrane of nanowires [5.26]. The template is attached to the
cathode, which is brought into contact with the depo-
Copper film
sition solution. The anode is placed in the deposition
b) Current (mA) solution, parallel to the cathode. When an electric field
is applied, cations diffuse through the channels and
2 deposit on the cathode, resulting in the growth of nano-
wires inside the template. This figure also shows the
III current density at different stages of deposition when
a constant electric field is applied. The current does
not change significantly until the pores are completely
II filled, at which point the current increases rapidly due
1
to improved contact with the electrolyte solution. The
current saturates once the template surface is com-
pletely covered. This approach has yielded nanowires
I made from different metals, including Ni, Co, Cu and
0 Au, with nominal pore diameters of between 10 and
0 1000 200 nm. The nanowires were found to be true repli-
Time (s)
cas of the pores [5.28]. Possin [5.11] prepared various
Fig. 5.2a,b Common experimental setup for the template- metallic nanowires using radiation track-etched mica.
based growth of nanowires using electrochemical deposi- Likewise, Williams and Giordano [5.29] produced silver
tion. (a) Schematic illustration of the arrangement of the nanowires with diameters of less than 10 nm. Whitney
electrodes for nanowire deposition. (b) Current–time curve et al. [5.26] fabricated arrays of nickel and cobalt nano-
for electrodeposition of Ni into a polycarbonate membrane wires, also using PC templates. Single crystal bismuth
with 60 nm diameter pores at − 1.0 V. Insets depict the nanowires have been grown in AAM using pulsed elec-
different stages of the electrodeposition (after [5.26]) trodeposition and Fig. 5.3 shows SEM and TEM images
Template-Based Synthesis of Nanorod or Nanowire Arrays 5.2 Electrochemical Deposition 173
a) b) c) d)
[110]
0.32nm
0.32 nm
60 nm
–
102
10
–
102
211
113 3 nm
500 μm 200 μm 015
Fig. 5.3a–d SEM images of Bi nanowire arrays: (a) top view, (b) tilt view. (c) TEM image of a typical Bi single nanowire.
(d) HRTEM image of a typical Bi single nanowire. The inset is the corresponding ED pattern (after [5.30])
of the bismuth nanowires [5.30]. Single crystal copper thickness of nanotubules. An increase in deposition
and lead nanowires were prepared by DC electrodeposi- time leads to a thick wall, but sometimes the hol-
tion and pulse electrodeposition, respectively [5.31,32]. low tubule morphology persists even after prolonged
The growth of single crystal lead nanowires required deposition.
a greater departure from equilibrium conditions (greater Although many research groups have reported on
overpotential) compared to the conditions required for the growth of uniformly sized nanorods and nanowires
Part A 5.2
polycrystalline ones. on PC template membranes, Schönenberger et al. [5.38]
Hollow metal tubules can also be prepared [5.33, reported that the channels of carbonate membranes were
34]. In this case the pore walls of the template are chem- not always uniform in diameter. They grew Ni, Co, Cu,
ically modified by anchoring organic silane molecules and Au nanowires using polycarbonate membranes with
so that the metal will preferentially deposit onto the nominal pore diameters of between 10 and 200 nm by
pore walls instead of the bottom electrode. For exam- an electrolysis method. From both a potentiostatic study
ple, the porous surface of an anodic alumina template of the growth process and a SEM analysis of nanowire
was first covered with cyanosilanes; subsequent elec- morphology, they concluded that the pores were gen-
trochemical deposition resulted in the growth of gold erally not cylindrical with a constant cross section, but
tubules [5.35]. An electroless electrolysis process has instead were rather cigarlike. For pores with a nominal
also been investigated for the growth of nanowires and diameter of 80 nm, the middle section of the pores was
nanorods [5.16, 33, 36]. Electroless deposition is actu- wider by up to a factor of 3.
ally a chemical deposition process and it involves the
use of a chemical agent to coat a material onto the 5.2.2 Semiconductors
template surface [5.37]. The main differences between
electrochemical deposition and electroless deposition Semiconductor nanowire and nanorod arrays have been
are that the deposition begins at the bottom electrode synthesized using AAM templates, such as CdSe and
and the deposited materials must be electrically con- CdTe [5.39]. The synthesis of nanowire arrays of bis-
ductive in the former. The electroless method does not muth telluride (Bi2 Te3 ) provide a good example of
require the deposited materials to be electrically con- the synthesis of compound nanowire arrays by elec-
ductive, and the deposition starts from the pore wall trochemical deposition. Bi2 Te3 is of special interest
and proceeds inwardly. Therefore, in general, electro- as a thermoelectric material and Bi2 Te3 nanowire ar-
chemical deposition results in the formation of solid rays are believed to offer high figures of merit for
nanorods or nanowires of conductive materials, whereas thermal-electrical energy conversion [5.40, 41]. Both
electroless deposition often results in hollow fibrils or polycrystalline and single crystal Bi2 Te3 nanowire ar-
nanotubules. For electrochemical deposition, the length rays have been grown by electrochemical deposition
of nanowires or nanorods can be controlled by the inside anodic alumina membranes [5.42, 43]. Sander
deposition time, whereas in electroless deposition the and coworkers [5.42] fabricated Bi2 Te3 nanowire ar-
length of the nanotubules is solely dependent on the rays with diameters as small as ≈ 25 nm from a solution
length of the deposition channels or pores. Variation of 0.075 M Bi and 0.1 M Te in 1 M HNO3 by electro-
of deposition time would result in a different wall chemical deposition at − 0.46 V versus a Hg/Hg2 SO4
174 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b) larly, large area Sb2 Te3 nanowire arrays have also been
successfully grown by template-based electrochemical
deposition, but the nanowires grown are polycrystalline
and show no clear preferred growth direction [5.45].
0.219 nm
(220)
[110]
20 30 40 50 60 70 80 90 100
200nm 2θ (deg)
Fig. 5.5a–c TEM images and XRD pattern of a single Bi2 Te3 nanowire. (a) TEM image and (b) HRTEM image of the
same nanowire. The inset is the corresponding ED pattern. (c) XRD pattern of Bi2 Te3 nanowire array (electrodeposition
time: 5 min) (after [5.43])
Part A 5.3
500 nm
50
electrode (and therefore the subsequent growth surface) It is obvious that the pH and the concentration of VO2+
and the electrolyte solution, the ionic cluster (VO2+ ) clusters in the vicinity of the growth surface shift away
is oxidized and solid V2 O5 is deposited through the from that in the bulk solution; both the pH and the VO2+
following reaction concentration decrease.
ZnO nanowire arrays were fabricated by a one-
2VO2+ + 3H2 O → V2 O5 + 6H+ + 2e− . (5.2) step electrochemical deposition technique based on an
A reduction reaction takes place at the counter elec- ordered nanoporous alumina membrane [5.55]. The
trode ZnO nanowire array is uniformly assembled into the
nanochannels of an anodic alumina membrane and con-
2H+ + 2e− → H2 (g) . (5.3) sists of single crystal particles.
Counter -ion
Co-ion
Flow-lines
Adsorbed Diffuse layer
layer with excess
negative charges
Electric field E
Particle Particle
Particle velocity υ
Shear surface
Electric Ions moving
potential ζ ζ0 with the particle
Immobile ions
1/κ
Distance from the surface x
Fig. 5.7 Schematic illustrating electrical double layer structure and the electric potential near the solid surface with both
the Stern and Gouy layers indicated. Surface charge is assumed to be positive (after [5.59])
Part A 5.3
species such as polymers. A combination of elec- the diffusion layer. Taken together, the Stern layer and
trostatic forces, Brownian motion and osmotic forces diffusion layer is known as the double layer structure in
results in the formation of a double layer structure, the classical theory of electrostatic stabilization.
schematically illustrated in Fig. 5.7. The figure depicts Upon the application of an external electric field,
a positively charged particle surface, the concentra- charged particles are set in motion, as schematically
tion profiles of negative ions (counterions) and positive illustrated in Fig. 5.8 [5.59]. This type of motion is re-
ions (surface charge-determining ions), and the electric ferred to as electrophoresis. When a charged particle
potential profile. The concentration of counterions grad- moves, some of the solvent or solution surrounding the
ually decreases with distance from the particle surface, particle will also move with it, since part of the sol-
whereas that of charge-determining ions increases. As vent or solution is tightly bound to the particle. The
a result, the electric potential decreases with distance. plane that separates the tightly bound liquid layer from
Near the particle surface, the electric potential decreases the rest of the liquid is called the slip plane (Fig. 5.7).
linearly, in the region known as the Stern layer. Outside The electric potential at the slip plane is known as the
of the Stern layer, the decrease follows an exponential zeta potential, which is an important parameter when
relationship. The region between the Stern layer and the determining the stability and transport of a colloidal dis-
point where the electric potential equals zero is called persion or a sol. A zeta potential of more than about
25 mV is typically required to stabilize a system [5.60].
The zeta potential ζ around a spherical particle can be
described using the relation [5.61]
Q
ζ=
4πεr a (1 + κa)
with
1/2
e2 n i z i2
κ= , (5.4)
εr ε0 k B T
Fig. 5.8 Schematic showing electrophoresis. Upon application of where Q is the charge on the particle, a is the radius of
an external electric field to a colloidal system or a sol, the charged the particle out to the shear plane, εr is the relative di-
nanoparticles or nanoclusters are set in motion (after [5.1]) electric constant of the medium, and n i and z i are the
Template-Based Synthesis of Nanorod or Nanowire Arrays 5.3 Electrophoretic Deposition 177
bulk concentration and valence of the i-th ion in the of an electrode. If the particles are positively charged
system, respectively. (or more precisely, they have a positive zeta potential),
The mobility of a nanoparticle in a colloidal disper- deposition of solid particles will occur at the cathode.
sion or a sol μ, is dependent on the dielectric constant of Otherwise, deposition will be at the anode. The elec-
the liquid medium εr , the zeta potential of the nanopar- trostatic double layers collapse at the electrodes and the
ticle ζ , and the viscosity η of the fluid. Several forms for particles coagulate, producing porous materials made of
this relationship have been proposed, such as the Hückel compacted nanoparticles. Typical packing densities are
equation [5.59, 61, 63–65] far less than the theoretical density of 74 vol.% [5.66].
2εr ε0 ζ Many theories have been proposed to explain the pro-
μ= . (5.5) cesses at the deposition surface during electrophoretic
3πη deposition. However, the evolution of structure on the
Electrophoretic deposition simply uses the oriented mo- deposition surface is not well understood. The elec-
tion of charged particles in an electrical field to grow trochemical processes that take place at the deposition
films or monoliths by transferring the solid particles surface and at the electrodes are complex and vary from
from a colloidal dispersion or a sol onto the surface system to system. The final density is dependent upon
a) b)
Part A 5.3
1 µm 1 µm
d) Relative intensity
c)
(101) TiO2 nanorods
TiO2 powder
(200)
(004) (105)
(211)
(204)
(220) (215)
(103) (112) (213)
(116) (107) (301)
1 µm
20 30 40 50 60 70 80
2θ (deg)
Fig. 5.9a–d SEM micrograph of TiO2 nanorods grown by template-based electrochemically induced sol–gel deposition.
The diameters of the nanorods are approximately: (a) 180 nm (for the 200 nm polycarbonate membrane); (b) 90 nm (for
the 100 nm membrane); (c) 45 nm (for the 50 nm membrane). (d) XRD patterns of both the grown nanorods and a powder
derived from the same sol. Both samples consist of the anatase phase only and no peak position shift was observed
(after [5.62])
178 Part A Nanostructures, Micro-/Nanofabrication and Materials
the concentration of particles in sols or colloidal dis- vanadium pentoxide nanorod arrays from TiO2+ and
persions, the zeta potential, the external electric field, VO+ 2 solutions respectively. Miao et al. [5.71] prepared
and the reaction kinetics between the surfaces of the single crystalline TiO2 nanowires by electrochemically
particles. A slow reaction and a slow arrival of nanopar- induced sol–gel deposition. Titania electrolyte solution
ticles onto the surface would allow sufficient particle was prepared by dissolving Ti powder into a H2 O2 and
relaxation on the deposition surface, so a high packing NH4 OH aqueous solution to form TiO2+ ionic clus-
density would be expected. ters [5.72]. When an electric field was applied, the
TiO2+ ionic clusters diffused to the cathode and under-
5.3.1 Polycrystalline Oxides went hydrolysis and condensation reactions, resulting
in the deposition of nanorods of amorphous TiO2 gel.
Limmer et al. [5.62, 67–69] combined sol–gel prepara- After heating at 240 ◦ C for 24 h in air, single crystal
tion with electrophoretic deposition to prepare nanorods anatase nanorods with diameters of 10, 20, and 40 nm
of various complex oxides. One of the advantages and lengths ranging from 2 to 10 μm were synthesized.
of this technique is the ability to synthesize complex The formation of single crystal TiO2 nanorods here is
oxides and organic–inorganic hybrids with desired sto- different to that reported by Martin’s group [5.73]. It is
ichiometric compositions. Another advantage is their suggested that the nanoscale crystallites generated dur-
applicability to a variety of materials. In their ap- ing heating assembled epitaxially to form single crystal
proach, conventional sol–gel processing was applied nanorods.
to the synthesis of various sols. By controlling the During typical sol–gel processing, nanoclusters are
sol preparation appropriately, nanometer particles of formed through homogeneous nucleation and subse-
Part A 5.3
a desired stoichiometric composition were formed, quent growth through sequential yet parallel hydrolysis
and electrostatically stabilized by pH adjustment. Us- and condensation reactions. Sol electrophoretic deposi-
ing radiation-tracked etched polycarbonate membranes tion enriches and deposits these formed nanoclusters at
with an electric field of ≈ 1.5 V/cm, they have grown an appropriate electrode surface under an external elec-
nanowires with diameters ranging from 40 to 175 nm tric field. The modified process is to limit and induce the
and lengths of 10 μm, corresponding to the thickness condensation reaction at the growth surface by chang-
of the membrane. The materials include anatase TiO2 , ing local pH value, which is a result of partial water
amorphous SiO2 , perovskite BaTiO3 and Pb(Ti, Zr)O3 , hydrolysis at the electrode or growth surface
and layered perovskite Sr2 Nb2 O7 . Figure 5.9 shows 2H2 O + 2e− → H2 + 2OH− , (5.6)
SEM micrographs and XRD patterns of TiO2 nanorod
arrays [5.62]. 2VO+ −
2 + 2OH → V2 O5 + H2 O . (5.7)
Wang et al. [5.70] used electrophoretic deposition Reaction (5.6), or the electrolysis of water, plays a very
to make nanorods of ZnO from colloidal sols. ZnO important role here. As the reaction proceeds, hy-
colloidal sol was prepared by hydrolyzing an alco- droxyl groups are produced, resulting in increased pH
holic solution of zinc acetate with NaOH, with a small near to the deposition surface. This increase in pH
amount of zinc nitrate added as a binder. This so- value near to the growth surface initiated and pro-
lution was then introduced into the pores of anodic motes the precipitation of V2 O5 , or reaction (5.7).
alumina membranes at voltages of 10–400 V. It was The initial pH of the VO+ 2 solution is ≈ 1.0, meaning
found that lower voltages led to dense, solid nanorods, that VO+ 2 is stable. However, when the pH increases
while higher voltages caused the formation of hollow to ≈ 1.8, VO+ 2 is no longer stable and solid V2 O5
tubules. They suggested that the higher voltages cause forms. Since the change in pH occurs near to the
dielectric breakdown of the anodic alumina, causing it growth surface, reaction (5.7) or deposition is likely
to become as positively charged as the cathode. Electro- to occur on the surface of the electrode through
static attraction between the ZnO nanoparticles and the heterogeneous nucleation and subsequent growth. It
pore walls then leads to tubule formation. should be noted that the hydrolysis of water has
another effect on the deposition of solid V2 O5 . Reac-
5.3.2 Single Crystal Oxide Nanorod Arrays tion (5.6) produces hydrogen on the growth surface.
Obtained by Changing the Local pH These molecules may poison the growth surface be-
fore dissolving into the electrolyte or by forming a gas
A modified version of sol electrophoretic deposition has bubble, which may cause the formation of porous
been used to grow single crystalline titanium oxide and nanorods.
Template-Based Synthesis of Nanorod or Nanowire Arrays 5.3 Electrophoretic Deposition 179
The formation of single crystal nanorods from 5.3.3 Single Crystal Oxide Nanorod Arrays
solutions by pH change-induced surface condensa- Grown by Homoepitaxial Aggregation
tion has been proven by TEM analyses, including
high-resolution imaging showing the lattice fringes Single crystal nanorods can also be grown directly
and electron diffraction. The growth of single crystal by conventional electrophoretic deposition. However,
nanorods by pH change-induced surface condensation several requirements must be met for such growth.
is attributed to evolution selection growth, which is First, the nanoclusters or particles in the sol must have
briefly summarized below. The initial heterogeneous a crystalline structure extended to the surface. Second,
nucleation or deposition onto the substrate surface the deposition of nanoclusters on the growth surface
results in the formation of nuclei with random orien- must have a certain degree of reversibility so that the
tations. The subsequent growth of various facets of nanoclusters can rotate or reposition prior to their irre-
a nucleus is dependent on the surface energy, and varies versible incorporation into the growth surface. Thirdly,
significantly from one facet to another [5.74]. For one- the deposition rate must be slow enough to permit
dimensional growth, such as film growth, only the sufficient time for the nanoclusters to rotate or repo-
highest growth rate with a direction perpendicular to sition. Lastly, the surfaces of the nanoclusters must
the growth surface will be able to continue to grow. be free of strongly attached alien chemical species.
The nuclei with the fastest growth direction perpen- Although precise control of all these parameters re-
dicular to the growth surface will grow larger, while mains a challenge, the growth of single crystal nanorods
nuclei with slower growth rates will eventually cease through homoepitaxial aggregation of nanocrystals has
to grow. Such a growth mechanism results in the for- been demonstrated [5.79, 80]. The formation of single
Part A 5.3
mation of columnar structured films where all of the crystalline vanadium pentoxide nanorods by template-
grains have the same crystal orientation (known as tex- based sol electrophoretic deposition can be attributed to
tured films) [5.75, 76]. In the case of nanorod growth homoepitaxial aggregation of crystalline nanoparticles.
inside a pore channel, such evolution selection growth Thermodynamically it is favorable for the crystalline
is likely to lead to the formation of a single crystal nanoparticles to aggregate epitaxially; this growth be-
nanorod or a bundle of single crystal nanorods per pore havior and mechanism is well documented in the
channel. Figure 5.10 shows typical TEM micrographs literature [5.81, 82]. In this growth mechanism, an ini-
and selected-area electron diffraction patterns of V2 O5 tial weak interaction between two nanoparticles allows
nanorods. It is well known that [010] (the b-axis) is rotation and migration relative to each other. Obviously,
the fastest growth direction for a V2 O5 crystal [5.77, homoepitaxial aggregation is a competitive process and
78], which would explain why single crystal vanadium porous structure is expected to form through this ho-
nanorods or a bundle of single crystal nanorods grow moepitaxial aggregation (as schematically illustrated in
along the b-axis. Fig. 5.11). Vanadium oxide particles present in a typ-
a) b) c)
[010]
5 µm 5 nm 2 nm
Fig. 5.10 (a) SEM image of V2 O5 nanorod arrays on an ITO substrate grown in a 200 nm carbonate membrane by sol
electrophoretic deposition; (b) TEM image of a V2 O5 nanorod with its electron diffraction pattern; (c) high-resolution
TEM image of the V2 O5 nanorod showing the lattice fringes (after [5.54])
180 Part A Nanostructures, Micro-/Nanofabrication and Materials
ical sol are known to easily form ordered crystalline with an externally applied electric field of 100–150 V
structure [5.83], so it is reasonable to expect that ho- with a distance of 5 mm between the two electrodes.
moepitaxial aggregation of vanadium nanocrystals from Both nanorods and nanotubes of fullerenes or met-
sol results in the formation of single crystal nanorods. allofullerenes can form and it is believed that initial
Single crystal nanorods formed in this way are likely to deposition occurs along the pore surface. A short de-
undergo significant shrinkage when fired at high tem- position time results in the formation of nanotubes,
peratures due to its original porous nature; 50% lateral whereas extended deposition leads to the formation of
shrinkage has been observed in vanadium pentoxide solid nanorods. These nanorods possess either crys-
nanorods formed by this method. In addition, it might talline or amorphous structure.
ing processes, significant shrinkage would be expected. prepared by injecting molten liquid into anodic alumina
However, the actual shrinkage observed is small when templates [5.93].
compared with the pore size. These results indicate Polymeric fibrils have been made by filling the tem-
that an (unknown) mechanism is acting to enrich the plate pores with a monomer solution containing the
concentration of solid inside the pores. One possible desired monomer and a polymerization reagent, fol-
mechanism could be the diffusion of solvent through lowed by in situ polymerization [5.14, 94–97]. The
the membrane, leading to the enrichment of solid on the polymer preferentially nucleates and grows on the pore
internal surfaces of the template pores, similar to what walls, resulting in tubules at short deposition times.
happens during ceramic slip casting [5.90]. Figure 5.12a Metal, oxide and semiconductor nanowires have
is a top view of ZnO nanotubules array, Fig. 5.12b recently been synthesized using self-assembled meso-
is TiO2 nanotubules array, Fig. 5.12c is hollow nan- porous silica as the template. For example, Han
tube array. The observed formation of nanotubules (in et al. [5.98] have synthesized Au, Ag and Pt nano-
Part A 5.4
Fig. 5.12 [5.73]) may imply that this process is indeed wires in mesoporous silica templates. The mesoporous
present. However, considering the fact that the tem- templates were first filled with aqueous solutions of
plates were typically emerged into sol for just a few the corresponding metal salts (such as HAuCl4 ). Af-
minutes, diffusion through the membrane and enrich- ter drying and treatment with CH2 Cl2 , the samples
ment of the solid inside the pores must be rather were reduced under H2 flow to form metallic nano-
rapid processes. It was also noticed that the nanorods wires. Liu et al. [5.99] carefully studied the interface
made by template filling are commonly polycrystalline between these nanowires and the matrix using high-
or amorphous, although single crystal TiO2 nanorods resolution electron microscopy and electron energy loss
were sometimes observed for nanorods smaller than spectroscopy techniques. A sharp interface only exists
20 nm [5.73]. between noble metal nanowires and the matrix. For
magnetic nickel oxide, a core–shell nanorod structure
5.4.2 Melt and Solution Filling containing a nickel oxide core and a thin nickel sil-
icate shell was observed. The magnetic properties of
Metallic nanowires can also be synthesized by filling the templated nickel oxide were found to be signifi-
a template with molten metals [5.91]. One example is cantly different from nickel oxide nanopowders due to
the preparation of bismuth nanowires using pressure in- the alignment of the nanorods. In another study, Chen
jection of molten bismuth into the nanochannels of an et al. filled the pores of a mesoporous silica template
anodic alumina template [5.92]. The anodic alumina with an aqueous solution of Cd and Mn salts, dried the
template was degassed and immersed in the liquid bis- sample, and reacted it with H2 S gas to convert it to
muth at 325 ◦ C (Tm = 271.5 ◦ C for Bi), and then high (Cd,Mn)S [5.100].
pressure Ar gas of ≈ 300 bar was applied in order to
inject liquid Bi into the nanochannels of the template 5.4.3 Centrifugation
for 5 h. Bi nanowires with diameters of 13–110 nm
and large aspect ratios (of up to several hundred) have Filling the template with nanoclusters via centrifugation
been obtained. Individual nanowires are believed to be forces is another inexpensive method for mass produc-
single-crystal. When exposed to air, bismuth nanowires ing nanorod arrays. Figure 5.13 shows SEM images of
are readily oxidized. An amorphous oxide layer ≈ 4 nm lead zirconate titanate (PZT) nanorod arrays with uni-
in thickness was observed after 48 h. After 4 weeks, the form sizes and unidirectional alignment [5.101]. These
bismuth nanowires were completely oxidized. Nano- nanorod arrays were grown in polycarbonate mem-
wires of other metals, such as In, Sn and Al, and the brane from PZT sol by centrifugation at 1500 rpm
semiconductors Se, Te, GaSb, and Bi2 Te3 , were also for 60 min. The samples were attached to silica glass
182 Part A Nanostructures, Micro-/Nanofabrication and Materials
and fired at 650 ◦ C in air for 60 min. Nanorod ar- bility to any colloidal dispersion system, including
rays of other oxides (silica and titania) were prepared. those consisting of electrolyte-sensitive nanoclusters or
The advantages of centrifugation include its applica- molecules.
consumable templates, although the resultant nanowires sors, (NH4 )2 MoS4 and (NH4 )2 Mo3 S13 , into the pores
and nanorods are generally not ordered to form aligned of alumina membrane templates. Then the template
arrays. Nanowires of compounds can be prepared us- filled with the molecular precursors was heated to
ing a template-directed reaction. First nanowires or an elevated temperature and the molecular precursors
nanorods of one constituent element are prepared, were thermally decomposed into MoS2 [5.105]. Cer-
and then these are reacted with chemicals contain- tain polymers and proteins were also used to direct
ing the other element desired in order to form the the growth of nanowires of metals or semiconductors.
final product. Gates et al. [5.102] converted single For example, Braun et al. [5.106] reported a two-step
crystalline trigonal selenium nanowires into single crys- procedure using DNA as a template for the vecto-
talline nanowires of Ag2 Se by reacting Se nanowires rial growth of a silver nanorods 12 μm in length and
with aqueous AgNO3 solutions at room temperature. 100 nm in diameter. CdS nanowires were prepared by
Nanorods can also be synthesized by reacting volatile polymer-controlled growth [5.107]. For the synthesis of
metal halides or oxide species with carbon nanotubes CdS nanowires, cadmium ions were well distributed in
to form solid carbide nanorods with diameters of be- a polyacrylamide matrix. The Cd2+ -containing polymer
tween 2 and 30 nm and lengths of up to 20 μm [5.103]. was treated with thiourea (NH2 CSNH2 ) solvothermally
ZnO nanowires were prepared by oxidizing metal- in ethylenediamine at 170 ◦ C, resulting in degradation
lic zinc nanowires [5.104]. Hollow nanotubules of of polyacrylamide. Single crystal CdS nanowires 40 nm
MoS2 ≈ 30 μm long and 50 nm in external diam- in diameter and up to 100 μm in length were obtained
eter with wall thicknesses of 10 nm were prepared with preferential [001] orientations.
colloidal dispersion is the method used to synthesize the lengths and the diameters of the resultant nano-
dielectric nanorods and nanowires. Template filling is tubes are dependent on the templates used for the
conceptually straightforward, although complete filling synthesis, the wall thicknesses of the nanotubes can be
is often very difficult. Converting reactive templates is readily controlled through the duration of growth. An-
a method used to achieve both nanorod arrays and ran- other great advantage of the template-based synthesis
domly oriented nanowires or nanorods, and it is often of nanotubes is the possibility of multilayered hollow
combined with other synthetic methods. nanotube or solid nanocable structures. For example,
This chapter has focused on the growth of solid Ni@V2 O5 ·nH2 O nanocable arrays have been synthe-
nanorod and nanowire arrays by template-based syn- sized by a two-step approach [5.109]. First, Ni nanorod
thesis; however, the use of template-based synthesis to arrays were grown in a PC template by electrochemical
synthesize nanotubes, and in particular nanotube ar- deposition, and then the PC template was removed by
rays, has received increasing attention [5.108]. One pyrolysis, followed by sol electrophoretic deposition of
of the greatest advantages using template-based syn- V2 O5 ·nH2 O on the surfaces of the Ni nanorod arrays. It
thesis to grow of nanotubes and nanotube arrays is is obvious that there is a lot of scope for more research
the independent control of the lengths, diameters, and into template-based syntheses of nanorod, nanotube and
the wall thicknesses of the nanotubes available. While nanocable arrays, and their applications.
References
5.1 G.Z. Cao: Nanostructures and Nanomaterials: Syn- 5.13 S. Fan, M.G. Chapline, N.R. Franklin, T.W. Tombler,
Part A 5
thesis, Properties and Applications (Imperial College, A.M. Cassell, H. Dai: Self-oriented regular arrays of
London 2004) carbon nanotubes and their field emission proper-
5.2 Z.L. Wang: Nanowires and Nanobelts: Materials, ties, Science 283, 512–514 (1999)
Properties and Devices, Nanowires and Nonobelts 5.14 P. Enzel, J.J. Zoller, T. Bein: Intrazeolite assem-
of Functional Materials, Vol. 2 (Kluwer, Boston bly and pyrolysis of polyacrylonitrile, J. Chem. Soc.
2003) Chem. Commun. 8, 633–635 (1992)
5.3 Y. Xia, P. Yang, Y. Sun, Y. Wu, Y. Yin, F. Kim, 5.15 C. Guerret-Piecourt, Y. Le Bouar, A. Loiseau, H. Pas-
H. Yan: One-dimensional nanostructures: Synthe- card: Relation between metal electronic structure
sis, characterization and applications, Adv. Mater. and morphology of metal compounds inside carbon
15, 353–389 (2003) nanotubes, Nature 372, 761–765 (1994)
5.4 A. Huczko: Template-based synthesis of nanomate- 5.16 P.M. Ajayan, O. Stephan, P. Redlich, C. Colliex: Car-
rials, Appl. Phys. A 70, 365–376 (2000) bon nanotubes as removable templates for metal
5.5 C. Burda, X. Chen, R. Narayanan, M.A. El-Sayed: oxide nanocomposites, nanostructures, Nature 375,
Chemistry and properties of nanocrystals of different 564–567 (1995)
shapes, Chem. Rev. 105, 1025–1102 (2005) 5.17 M. Knez, A.M. Bittner, F. Boes, C. Wege, H. Jeske,
5.6 X. Duan, C.M. Lieber: General synthesis of compound E. Maiâ, K. Kern: Biotemplate synthesis of 3-nm
semiconductor nanowires, Adv. Mater. 12, 298–302 nickel and cobalt nanowires, Nano Lett. 3, 1079–1082
(2000) (2003)
5.7 M.P. Zach, K.H. Ng, R.M. Penner: Molybdenum 5.18 R. Gasparac, P. Kohli, M.O.M.L. Trofin, C.R. Martin:
nanowires by electrodeposition, Science 290, 2120– Template synthesis of nano test tubes, Nano Lett. 4,
2123 (2000) 513–516 (2004)
5.8 R.C. Furneaux, W.R. Rigby, A.P. Davidson: The 5.19 C.F. Monson, A.T. Woolley: DNA-templated construc-
formation of controlled-porosity membranes from tion of copper nanowires, Nano Lett. 3, 359–363
anodically oxidized aluminium, Nature 337, 147–149 (2003)
(1989) 5.20 Y. Weizmann, F. Patolsky, I. Popov, I. Willner:
5.9 R.L. Fleisher, P.B. Price, R.M. Walker: Nuclear Tracks Telomerase-generated templates for the growing of
in Solids (Univ. of California Press, Berkeley 1975) metal nanowires, Nano Lett. 4, 787–792 (2004)
5.10 R.J. Tonucci, B.L. Justus, A.J. Campillo, C.E. Ford: 5.21 A. Despic, V.P. Parkhuitik: Modern Aspects of Elec-
Nanochannel array glass, Science 258, 783–787 trochemistry, Vol. 20 (Plenum, New York 1989)
(1992) 5.22 D. Al Mawiawi, N. Coombs, M. Moskovits: Magnetic
5.11 G.E. Possin: A method for forming very small diam- properties of Fe deposited into anodic aluminum
eter wires, Rev. Sci. Instrum. 41, 772–774 (1970) oxide pores as a function of particle size, J. Appl.
5.12 C. Wu, T. Bein: Conducting polyaniline filaments in Phys. 70, 4421–4425 (1991)
a mesoporous channel host, Science 264, 1757–1759 5.23 C.A. Foss, M.J. Tierney, C.R. Martin: Template-
(1994) synthesis of infrared-transparent metal microcylin-
184 Part A Nanostructures, Micro-/Nanofabrication and Materials
ders: Comparison of optical properties with the tion of cadmium chalcogenide microdiode arrays,
predictions of effective medium theory, J. Phys. Chem. Mater. 5, 902–904 (1993)
Chem. 96, 9001–9007 (1992) 5.40 L.D. Hicks, M.S. Dresselhaus: Thermoelectric figure of
5.24 A.J. Bard, L.R. Faulkner: Electrochemical Methods merit of a one-dimensional conductor, Phys. Rev. B
(Wiley, New York 1980) 47, 679–682 (1993)
5.25 J.B. Mohler, H.J. Sedusky: Electroplating for the 5.41 M.S. Dresselhaus, G. Dresselhaus, X. Sun, Z. Zhang,
Metallurgist, Engineer and Chemist (Chemical Pub- S.B. Cronin, T. Koga: Low-dimensional thermo-
lishing, New York 1951) electric materials, Phys. Solid State 41, 679–682
5.26 T.M. Whitney, J.S. Jiang, P.C. Searson, C.L. Chien: (1999)
Fabrication and magnetic properties of arrays of 5.42 M.S. Sander, R. Gronsky, T. Sands, A.M. Stacy: Struc-
metallic nanowires, Science 261, 1316–1319 (1993) ture of bismuth telluride nanowire arrays fabricated
5.27 F.R.N. Nabarro, P.J. Jackson: Growth of crystal by electrodeposition into porous anodic alumina
whiskers – A review. In: Growth and Perfection of templates, Chem. Mater. 15, 335–339 (2003)
Crystals, ed. by R.H. Doremus, B.W. Roberts, D. Turn- 5.43 C. Lin, X. Xiang, C. Jia, W. Liu, W. Cai, L. Yao, X. Li:
bull (Wiley, New York 1958) pp. 11–102 Electrochemical fabrication of large-area, ordered
5.28 B.Z. Tang, H. Xu: Preparation, alignment and Bi2 Te3 nanowire arrays, J. Phys. Chem. B 108, 1844–
optical properties of soluble poly(phenylacetylene)- 1847 (2004)
wrapped carbon nanotubes, Macromolecules 32, 5.44 D.S. Xu, Y.J. Xu, D.P. Chen, G.L. Guo, L.L. Gui,
2567–2569 (1999) Y.Q. Tang: Preparation of CdS single-crystal nano-
5.29 W.D. Williams, N. Giordano: Fabrication of 80 Å wires by electrochemically induced deposition, Adv.
metal wires, Rev. Sci. Instrum. 55, 410–412 (1984) Mater. 12, 520–522 (2000)
5.30 C.G. Jin, G.W. Jiang, W.F. Liu, W.L. Cai, L.Z. Yao, 5.45 C. Lin, G. Zhang, T. Qian, X. Li, Z. Yao: Large-area
Part A 5
Z. Yao, X.G. Li: Fabrication of large-area single crys- Sb2 Te3 nanowire arrays, J. Phys. Chem. B 109, 1430–
tal bismuth nanowire arrays, J. Mater. Chem. 13, 1432 (2005)
1743–1746 (2003) 5.46 C. Jérôme, R. Jérôme: Electrochemical synthesis of
5.31 M.E.T. Molares, V. Buschmann, D. Dobrev, R. Neu- polypyrrole nanowires, Angew. Chem. Int. Ed. 37,
mann, R. Scholz, I.U. Schuchert, J. Vetter: Single- 2488–2490 (1998)
crystalline copper nanowires produced by elec- 5.47 A.G. MacDiarmid: Nobel lecture: “Synthetic metals”:
trochemical deposition in polymeric ion track A novel role for organic polymers, Rev. Mod. Phys.
membranes, Adv. Mater. 13, 62–65 (2001) 73, 701–712 (2001)
5.32 G. Yi, W. Schwarzacher: Single crystal superconduc- 5.48 K. Doblhofer, K. Rajeshwar: Handbook of Conducting
tor nanowires by electrodeposition, Appl. Phys. Lett. Polymers (Marcel Dekker, New York 1998), Chap. 20
74, 1746–1748 (1999) 5.49 L. Dauginet, A.-S. Duwez, R. Legras, S. Demoustier-
5.33 C.J. Brumlik, V.P. Menon, C.R. Martin: Synthesis Champagne: Surface modification of polycarbonate
of metal microtubule ensembles utilizing chemical, and poly(ethylene terephthalate) films and mem-
electrochemical and vacuum deposition techniques, branes by polyelectrolyte, Langmuir 17, 3952–3957
J. Mater. Res. 268, 1174–1183 (1994) (2001)
5.34 C.J. Brumlik, C.R. Martin: Template synthesis of 5.50 C.R. Martin: Membrane-based synthesis of nano-
metal microtubules, J. Am. Chem. Soc. 113, 3174–3175 materials, Chem. Mater. 8, 1739–1746 (1996)
(1991) 5.51 C.R. Martin: Template synthesis of polymeric and
5.35 C.J. Miller, C.A. Widrig, D.H. Charych, M. Majda: Mi- metal microtubules, Adv. Mater. 3, 457–459 (1991)
croporous aluminum oxide films at electrodes. 4. 5.52 J.C. Hulteen, C.R. Martin: A general template-
Lateral charge transport in self-organized bilayer based method for the preparation of nanomaterials,
assemblies, J. Phys. Chem. 92, 1928–1936 (1988) J. Mater. Chem. 7, 1075–1087 (1997)
5.36 W. Han, S. Fan, Q. Li, Y. Hu: Synthesis of gallium 5.53 L. Liang, J. Liu, C.F. Windisch Jr., G.J. Exarhos,
nitride nanorods through a carbon nanotube- Y. Lin: Assembly of large arrays of oriented conduct-
confined reaction, Science 277, 1287–1289 (1997) ing polymer nanowires, Angew. Chem. Int. Ed. 41,
5.37 G.O. Mallory, J.B. Hajdu (Eds.): Electroless Plating: 3665–3668 (2002)
Fundamentals and Applications (AESF, Orlando 1990) 5.54 K. Takahashi, S.J. Limmer, Y. Wang, G.Z. Cao: Growth
5.38 C. Schönenberger, B.M.I. van der Zande, L.G.J. Fok- and electrochemical properties of single-crystalline
kink, M. Henny, C. Schmid, M. Krüger, A. Bachtold, V2 O5 nanorod arrays, Jpn. J. Appl. Phys. B 44, 662–
R. Huber, H. Birk, U. Staufer: Template synthesis 668 (2005)
of nanowires in porous polycarbonate membranes: 5.55 M.J. Zheng, L.D. Zhang, G.H. Li, W.Z. Shen: Fabrica-
Electrochemistry and morphology, J. Phys. Chem. B tion and optical properties of large-scale uniform
101, 5497–5505 (1997) zinc oxide nanowire arrays by one-step electro-
5.39 J.D. Klein, R.D. Herrick II, D. Palmer, M.J. Sailor, chemical deposition technique, Chem. Phys. Lett.
C.J. Brumlik, C.R. Martin: Electrochemical fabrica- 363, 123–128 (2002)
Template-Based Synthesis of Nanorod or Nanowire Arrays References 185
5.56 I. Zhitomirsky: Cathodic electrodeposition of ceramic diamond films by the addition of nitrogen, J. Appl.
and organoceramic materials. Fundamental aspects, Phys. 79, 1357–1364 (1996)
Adv. Colloid Interf. Sci. 97, 279–317 (2002) 5.76 M. Ohring: Materials Science of Thin Films (Academic,
5.57 O.O. Van der Biest, L.J. Vandeperre: Electrophoretic San Diego 2001)
deposition of materials, Annu. Rev. Mater. Sci. 29, 5.77 D. Pan, Z. Shuyuan, Y. Chen, J.G. Hou: Hydrothermal
327–352 (1999) preparation of long nanowires of vanadium oxide,
5.58 P. Sarkar, P.S. Nicholson: Electrophoretic deposi- J. Mater. Res. 17, 1981–1984 (2002)
tion (EPD): Mechanism, kinetics, and application to 5.78 V. Petkov, P.N. Trikalitis, E.S. Bozin, S.J.L. Billinge,
ceramics, J. Am. Ceram. Soc. 79, 1987–2002 (1996) T. Vogt, M.G. Kanatzidis: Structure of V2 O5 ·nH2 O xe-
5.59 A.C. Pierre: Introduction to Sol-Gel Processing rogel solved by the atomic pair distribution function
(Kluwer, Norwell 1998) technique, J. Am. Chem. Soc. 124, 10157–10162 (2002)
5.60 J.S. Reed: Introduction to the Principles of Ceramic 5.79 K. Takahashi, S.J. Limmer, Y. Wang, G.Z. Cao: Syn-
Processing (Wiley, New York 1988) thesis, electrochemical properties of single crystal
5.61 R.J. Hunter: Zeta Potential in Colloid Science: Prin- V2 O5 nanorod arrays by template-based electrode-
ciples and Applications (Academic, London 1981) position, J. Phys. Chem. B 108, 9795–9800 (2004)
5.62 S.J. Limmer, T.P. Chou, G.Z. Cao: A study on the 5.80 G.Z. Cao: Growth of oxide nanorod arrays through
growth of TiO2 using sol electrophoresis, J. Mater. sol electrophoretic deposition, J. Phys. Chem. B 108,
Sci. 39, 895–901 (2004) 19921–19931 (2004)
5.63 C.J. Brinker, G.W. Scherer: Sol-Gel Science: the 5.81 R.L. Penn, J.F. Banfield: Morphology development
Physics and Chemistry of Sol-Gel Processing (Aca- and crystal growth in nanocrystalline aggregates
demic, San Diego 1990) under hydrothermal conditions: Insights from tita-
5.64 J.D. Wright, N.A.J.M. Sommerdijk: Sol-Gel Materials: nia, Geochim. Cosmochim. Acta 63, 1549–1557 (1999)
Part A 5
Chemistry and Applications (Gordon and Breach, 5.82 C.M. Chun, A. Navrotsky, I.A. Aksay: Aggregation
Amsterdam 2001) growth of nanometer-sized BaTiO3 particles, Proc.
5.65 D.H. Everett: Basic Principles of Colloid Science (The Microsc. Microanal. (1995) pp. 188–189
Royal Society of Chemistry, London 1988) 5.83 J. Livage: Synthesis of polyoxovanadates via chimie
5.66 W.D. Callister: Materials Science and Engineering: An douce, Coord. Chem. Rev. 178–180, 999–1018 (1998)
Introduction (Wiley, New York 1997) 5.84 K.V. Saban, J. Thomas, P.A. Varughese, G. Varghese:
5.67 S.J. Limmer, S. Seraji, M.J. Forbess, Y. Wu, T.P. Chou, Thermodynamics of crystal nucleation in an exter-
C. Nguyen, G.Z. Cao: Electrophoretic growth of lead nal electric field, Cryst. Res. Technol. 37, 1188–1199
zirconate titanate nanorods, Adv. Mater. 13, 1269– (2002)
1272 (2001) 5.85 D. Grier, E. Ben-Jacob, R. Clarke, L.M. Sander:
5.68 S.J. Limmer, S. Seraji, M.J. Forbess, Y. Wu, T.P. Chou, Morphology and microstructure in electrochemical
C. Nguyen, G.Z. Cao: Template-based growth of var- deposition of zinc, Phys. Rev. Lett. 56, 1264–1267
ious oxide nanorods by sol-gel electrophoresis, Adv. (1986)
Funct. Mater. 12, 59–64 (2002) 5.86 C.J. Li, Y.G. Guo, B.S. Li, C.R. Wang, L.J. Wan, C.L. Bai:
5.69 S.J. Limmer, G.Z. Cao: Sol-gel electrophoretic depo- Template synthesis of Sc@C82 (I) nanowires and
sition for the growth of oxide nanorods, Adv. Mater. nanotubes at room temperature, Adv. Mater. 17,
15, 427–431 (2003) 71–73 (2005)
5.70 Y.C. Wang, I.C. Leu, M.N. Hon: Effect of colloid char- 5.87 Y.G. Guo, C.J. Li, L.J. Wan, D.M. Chen, C.R. Wang,
acteristics on the fabrication of ZnO nanowire arrays C.L. Bai, Y.G. Wang: Well-defined fullerene
by electrophoretic deposition, J. Mater. Chem. 12, nanowire arrays, Adv. Funct. Mater. 13, 626–630
2439–2444 (2002) (2003)
5.71 Z. Miao, D. Xu, J. Ouyang, G. Guo, Z. Zhao, Y. Tang: 5.88 B.B. Lakshmi, C.J. Patrissi, C.R. Martin: Sol-gel tem-
Electrochemically induced sol-gel preparation of plate synthesis of semiconductor oxide micro- and
single-crystalline TiO2 nanowires, Nano Lett. 2, 717– nanostructures, Chem. Mater. 9, 2544–2550 (1997)
720 (2002) 5.89 Q. Lu, F. Gao, S. Komarneni, T.E. Mallouk: Ordered
5.72 C. Natarajan, G. Nogami: Cathodic electrodeposi- SBA-15 nanorod arrays inside a porous alumina
tion of nanocrystalline titanium dioxide thin films, membrane, J. Am. Chem. Soc. 126, 8650–8651 (2004)
J. Electrochem. Soc. 143, 1547–1550 (1996) 5.90 J.S. Reed: Introduction to Principles of Ceramic Pro-
5.73 B.B. Lakshmi, P.K. Dorhout, C.R. Martin: Sol-gel cessing (Wiley, New York 1988)
template synthesis of semiconductor nanostruc- 5.91 C.A. Huber, T.E. Huber, M. Sadoqi, J.A. Lubin, S. Man-
tures, Chem. Mater. 9, 857–863 (1997) alis, C.B. Prater: Nanowire array composite, Science
5.74 A. van der Drift: Evolutionary selection, a principle 263, 800–802 (1994)
governing growth orientation in vapor-deposited 5.92 Z. Zhang, D. Gekhtman, M.S. Dresselhaus, J.Y. Ying:
layers, Philips Res. Rep. 22, 267–288 (1968) Processing and characterization of single-crystalline
5.75 G.Z. Cao, J.J. Schermer, W.J.P. van Enckevort, ultrafine bismuth nanowires, Chem. Mater. 11, 1659–
W.A.L.M. Elst, L.J. Giling: Growth of {100} textured 1665 (1999)
186 Part A Nanostructures, Micro-/Nanofabrication and Materials
5.93 E.G. Wolff, T.D. Coskren: Growth, morphology of 5.101 T. Wen, J. Zhang, T.P. Chou, S.J. Limmer, G.Z. Cao:
magnesium oxide whiskers, J. Am. Ceram. Soc. 48, Template-based growth of oxide nanorod arrays
279–285 (1965) by centrifugation, J. Sol-Gel Sci. Tech. 33, 193–200
5.94 W. Liang, C.R. Martin: Template-synthesized poly- (2005)
acetylene fibrils show enhanced supermolecular 5.102 B. Gates, Y. Wu, Y. Yin, P. Yang, Y. Xia: Single-
order, J. Am. Chem. Soc. 112, 9666–9668 (1990) crystalline nanowires of Ag2 Se can be synthesized by
5.95 S.M. Marinakos, L.C. Brousseau III, A. Jones, templating against nanowires of trigonal Se, J. Am.
D.L. Feldheim: Template synthesis of one- Chem. Soc. 123, 11500–11501 (2001)
dimensional Au, Au-poly(pyrrole) and poly(pyrrole) 5.103 E.W. Wong, B.W. Maynor, L.D. Burns, C.M. Lieber:
nanoparticle arrays, Chem. Mater. 10, 1214–1219 Growth of metal carbide nanotubes, nanorods,
(1998) Chem. Mater. 8, 2041–2046 (1996)
5.96 H.D. Sun, Z.K. Tang, J. Chen, G. Li: Polarized Raman 5.104 Y. Li, G.S. Cheng, L.D. Zhang: Fabrication of highly
spectra of single-wall carbon nanotubes mono- ordered ZnO nanowire arrays in anodic alumina
dispersed in channels of AlPO4 -5 single crystals, membranes, J. Mater. Res. 15, 2305–2308 (2000)
Solid State Commun. 109, 365–369 (1999) 5.105 C.M. Zelenski, P.K. Dorhout: The template syn-
5.97 Z. Cai, J. Lei, W. Liang, V. Menon, C.R. Martin: thesis of monodisperse microscale nanofibers,
Molecular and supermolecular origins of enhanced nanotubules of MoS2 , J. Am. Chem. Soc. 120, 734–742
electronic conductivity in template-synthesized (1998)
polyheterocyclic fibrils. 1. Supermolecular effects, 5.106 E. Braun, Y. Eichen, U. Sivan, G. Ben-Yoseph: DNA-
Chem. Mater. 3, 960–967 (1991) templated assembly and electrode attachment of
5.98 Y.J. Han, J.M. Kim, G.D. Stucky: Preparation of noble a conducting silver wire, Nature 391, 775–778 (1998)
metal nanowires using hexagonal mesoporous silica 5.107 J. Zhan, X. Yang, D. Wang, S. Li, Y. Xie, Y. Xia, Y. Qian:
Part A 5
SBA-15, Chem. Mater. 12, 2068–2069 (2000) Polymer-controlled growth of CdS nanowires, Adv.
5.99 J. Liu, G.E. Fryxell, M. Qian, L.-Q. Wang, Y. Wang: Mater. 12, 1348–1351 (2000)
Interfacial chemistry in self-assembled nanoscale 5.108 Y. Wang, K. Takahashi, H.M. Shang, G.Z. Cao:
materials with structural ordering, Pure Appl. Chem. Synthesis, electrochemical properties of vanadium
72, 269–279 (2000) pentoxide nanotube arrays, J. Phys. Chem. B109,
5.100 L. Chen, P.J. Klar, W. Heimbrodt, F. Brieler, M. Fröba: 3085–3088 (2005)
Towards ordered arrays of magnetic semiconduc- 5.109 K. Takahashi, Y. Wang, G.Z. Cao: Ni-V2 O5 ·n H2 O core-
tor quantum wires, Appl. Phys. Lett. 76, 3531–3533 shell nanocable arrays for enhanced electrochemical
(2000) intercalation, J. Phys. Chem. B 109, 48–51 (2005)
187
Part A 6
arrangement. In reality, it is often a challenge to 6.4 Processes and Setups ............................ 205
reach good yields, high precision, and good speci- 6.4.1 Setups for Particle Assembly .......... 205
ficity, in particular for very small particles. Since the 6.4.2 Particle Printing and Processing ..... 206
method is very general, particles of various mater-
ials such as oxides, metals, semiconductors, and 6.5 Conclusions .......................................... 206
polymers can be arranged for applications ranging
from microelectronics to optics and biochemistry.
References .................................................. 207
Solid particles with sub-μm diameters are intriguing There are two prerequisites: first, particles with narrow
objects. They have a well-defined surface which is size distribution and well-defined structures and sur-
large compared with their volume, so that they interact faces have to be available from different materials in
strongly with their environment. At the same time, par- sufficient quantities. Second, these particles have to be
ticles are clearly defined entities which can be mixed, arranged such that they provide the desired functional-
purified, modified, and arranged into larger structures. ity. Templated particle assembly is one way to do so.
This combination has made them popular in fields rang- A template defines the particle arrangement in advance
ing from biology (where they carry analyte-binding according to the designer’s wishes.
molecules) to semiconductor fabrication (where they Producing particles of sufficient quality to be used
confine electrons) [6.1]. as building blocks is not necessarily simple, but it can be
It is tempting to try and use such particles as done efficiently. Chemical methods are known to pro-
nanoscale building blocks to create functional devices, duce particles from very small clusters (with diameters
be their function electronic, mechanical or chemical. in the low nanometer regime), various shapes of sin-
188 Part A Nanostructures, Micro-/Nanofabrication and Materials
gle crystals with diameters from 10 to about 100 nm, ticles are used as memory elements [6.6], they need to
and larger particles with diameters up to microme- be electrically addressed – a task that is greatly simpli-
ters. Some syntheses produce particles that are rather fied if their positions are well known in advance.
monodisperse, the best methods reaching coefficients of Interacting particles can exhibit collective proper-
variation below 3%. This is still worse than, say, the ties that depend on their relative arrangement. In the
relative size distribution of bricks in most buildings, field of metamaterials, for example, the activity of many
but good enough for the particle to arrange sponta- particles with sizes well below the wavelength of an in-
neously into ordered supercrystals [6.2]. The particles cident electromagnetic wave leads to unusual far-field
can be simple crystals or complex structures with a shell behavior [6.7]. From afar, the bulk metamaterial appears
that differs from the core, for example, to protect the to have, for example, a negative refractive index. Opti-
surface of the core [6.3]. Chemical methods readily cal metamaterials also include photonic crystals, which
produce such core–shell structures which would be exhibit a photonic bandgap much like the electronic
exceedingly complicated to make using conventional bandgap of semiconductors due to a periodic poten-
methods. tial caused by regular crystals of spherical, diffracting
As for their arrangement, particles down to about particles. Templated assembly can create such dense
100 μm in diameter are routinely handled using con- structures with well-defined boundaries, and it can
ventional pick-and-place techniques, a method widely influence the packing itself by imposing a desired ge-
used in industrial processes. Such serial methods be- ometry on the first layer.
come very time consuming at smaller scales, and they More complex structures, possibly including more
fail in the sub-μm regime, where adhesion forces ren- than one particle type, offer even more complex func-
der the simple maneuver of putting down a particle tionalities. One popular target is smart materials, which
very challenging [6.4]. In this size range, particles are react to a stimulus in a coherent and useful way.
dominated by Brownian motion. They move randomly Much like the electronic properties of a semiconductor
in their suspensions, and alternative assembly meth- microchip lead to extremely complex electronic behav-
ods become necessary for their placement. Templated ior, patterned materials formed from arranged particles
particle assembly is such an alternative strategy, based might exhibit useful mechanical, thermal or other prop-
Part A 6
on a predefined surface that carries the information on erties. Another application of such complex structures
the final particle placement. It can produce a variety (which are hard to produce) is anticounterfeiting, where
of particle arrangements in parallel and over large ar- an object is protected by a small particle structure with
eas (with typical lateral dimensions up to 106 particle a unique property that can be detected.
diameters). Templated assembly is, of course, competing with
Templated assembly utilizes the strong interactions more traditional means of micro- and nanofabrica-
of particles with interfaces and their tendency to pro- tion, as covered in other chapters of this Handbook.
duce dense packings to create predictable arrangements Templated assembly is advantageous in that it takes
on a patterned surface. Since the desirable arrangement advantage of the chemically produced small dimen-
depends on the desired material properties, it is an ad- sions of nanoparticles, and it is more general than
vantage of templated assembly to give the user great traditional methods in that it can process a wide vari-
flexibility in attainable particle arrangements. ety of available colloids. The actual assembly process
There are rather different motivations for the use can be rather simple and compatible with continuous
of well-defined particle arrangements. If single-particle processing, even under ambient conditions. The most
properties are to be exploited (for example, their small challenging prerequisite is usually the template, which
size, large surface-to-volume ratio or optical proper- has to be fabricated to provide sufficient definition of
ties), it is often critical to know in advance the exact the assembled structure.
particle positions. Particles are then commonly arranged A process that arranges particles into a regular struc-
into spaced arrays, possibly with alignment marks. In ture without any template is often called self-assembly.
a biological assay, for example, a fluorescence reader Here, the information on the arrangement is not con-
can find the individual particles in a regular array tained in a template but in the properties of the particles
according to their position and record their optical prop- themselves. The problem of programming the assem-
erties to gain information on an analyte that had come bly process is thus shifted to the particles, which have
into contact with the particles [6.5]. Similarly, if par- to be chosen (or modified) such that they assemble into
Templated Self-Assembly of Particles 6.1 The Assembly Process 189
a certain structure. This is not an easy task, and there eral patents covering the integration of semiconductor
are few examples so far of rational materials design us- pieces into polymers and other carriers, which today
ing engineered particles. A template, on the other hand, it mainly uses for the production of radiofrequency
can be defined using classical top-down methods, which identification (RFID) chips, in which small electronic
provide great flexibility. radiofrequency components are mounted on a paper or
Still, templates become hard to fabricate if the polymer label which is attached to an item for wireless
particles are small and high patterning resolution is re- identification. Similar methods for much smaller par-
quired. A combination of self-assembly and templated ticles are currently being developed, but have not yet
assembly is then useful: boundaries are defined by been applied industrially.
the template, but additional effects such as particle– The challenges that occur when going down in par-
particle, particle–surface or particle–solvent interac- ticle size are mostly due to the greater influence of
tions lead to a predictable particle arrangement inside Brownian motion, which disturbs any order formed;
the boundaries. strong adhesion to surfaces, which increases unspe-
We will limit ourselves here to processes with cific adsorption and makes pick-and-place difficult; and
surface-bound templates and disregard supramolecu- the problem of process control as the particles become
lar assembly, although molecular cages might also be harder to resolve with conventional optical methods.
regarded as templates. Likewise, biomineralization pro- In addition, the dimensions of the targeted nanos-
cesses which can be templated using certain surfaces tructured materials are often comparable to those
will not be covered here. The main focus is on sub-μm produced with larger particles, but the number of
particles that are hard to place using any other method particles involved is now very much higher (scaling
but can be assembled with high quality by means of inversely with the particle volume). Even assembly
templated assembly processes. methods with very high yields are therefore bound to
Even today, larger particles (between ≈ 1 and produce defects, which might hinder the function of
100 μm) are assembled using templated assembly meth- the material. In some interesting applications (such as
ods, mostly from slurries in an approach called fluidic optical metamaterials), the absolute placement accura-
assembly [6.8]. Illumina, Inc. arranges 3 μm-diameter cies required to create a discernible optical effect are
Part A 6.1
glass beads functionalized with short DNA strands strict. Templated assembly is in principle able to pro-
into a regular grid, which can then be used for DNA vide such accuracies – even for many particles – and we
sequencing. Alien Technology Corporation holds sev- will discuss its prerequisites in the next section.
Table 6.1 Interactions that can drive particle assembly pro- free energy of the particle will change as the interac-
cesses tion with the binding sites changes. If there is a gradient
Interaction Typical range present, a directing force will act on the particle and bias
(order of magnitude) its random motion towards an energy minimum. This
Covalent 0.1 nm
energy landscape, formed by the superposition of the
van der Waals 1 nm
interaction, governs the particle’s motion.
Some interactions are strong but short-ranged, for
Coulomb (electrostatic) 1 nm (polar)–100 nm (apolar)
example, covalent bonds. In the energy landscape pic-
Hydrophobic 1 nm
ture shown in Fig. 6.1, they will resemble a steep well
Capillary 1 mm
into which the particle falls and from which it can
hardly escape. On the other hand, the particle can be
The goal of an assembly process is to overcome this in close proximity to such steep wells and still not feel
random, diffusional motion (with an energy scale of their presence. More precisely, the probability distribu-
kB T and characterized by the diffusion coefficient D) tion of its presence will only be affected locally. When
by a bias that induces drift so that the probability of the particle is trapped inside the well, and if the en-
finding a particle at the desired position is markedly in- trapment can be reasonably modeled using a harmonic
creased. Particles are then held in place until the system oscillator, its deviations from the minimum at x = 0
is quenched in some way, for example, by exchanging equals [6.10]
its environment. k T
B
In order to arrange the particles, templated assembly x2 = (6.2)
processes use potentials with minima at the particles’ mω20
target positions. Such potential wells can be defined us- for a particle with mass m that is bound
√ as in a harmonic
ing various particle–surface interactions, some of which oscillator with a frequency ω = k/m, the square root
are listed in Table 6.1. These interactions act over differ- of the spring constant over the particle mass. Thus,
ent lengths, have different strengths, and form minima a steep potential minimum can trap a particle with high
with different geometries, all of which can influence the accuracy: if the oscillator has a frequency of 1 GHz,
Part A 6.1
subject to shielding by ions from the solvent; their Various forces can occur in combination or subse-
strength can also depend on the hydrodynamic situa- quently during an assembly process. For example, in
tion. van der Waals interactions depend on the dielectric the classical example of the convective assembly of par-
properties of the solvent: their interaction length is gen- ticles in a thin wetting film, hydrodynamic drag and
erally so short that they do not funnel particles from capillary forces act in different stages of the assembly,
the bulk but trap particles that randomly hit the sur- yielding two-dimensional crystals of particles [6.12].
face or were attracted by other forces. Supramolecular In templated assembly, one can use such combined ef-
interactions are not included here because they are too fects to cause additional confinement. Aizenberg and
diverse; in general, such interactions tend to be similarly her group have shown that the combination of capil-
short-ranged as van der Waals interactions. In three- lary and electrostatic forces produces a focusing effect
phase systems, capillary forces can occur and exert very when particles are assembled on larger patches [6.11].
long-ranged forces even on small particles. The energy landscape changes with time: its minimum
An important practical limit of the assembly accu- becomes narrower as the liquid evaporates and centers
racy is the template. The template has to be fabricated, the assembled particle on the binding site.
often using top-down methods, to define the final par- The formation of a potential funnel that guides the
ticle positions. It may have binding sites that are large particle to its desired position is desirable for suc-
enough for many particles to be trapped inside, either cessful templated particle assembly. A properly chosen
in ordered arrangements or in disordered layers. On the energy landscape ensures high placement accuracy, as
other hand, it may have binding sites that are small discussed above. It also increases the yield of assem-
enough to accommodate only a single particle. If so, bly by attracting particles from a larger volume towards
the area of a binding site usually has to be on the order the binding site. Assembly is more rapid if particles
of the particle’s projected area. A particle that comes are guided from a larger volume instead of randomly
into contact with the binding site might be irreversibly diffusing until they accidentally arrive at the binding
adsorbed immediately. In the energy landscape picture, site. On the other hand, unspecific deposition is avoided
this would correspond to a well with steep walls and if secondary minima on the energy landscape are kept
a flat base. On the other hand, if the well has slanted shallow and are connected to the global minima (the
Part A 6.1
walls and a small base, the particle can align with the binding sites) via low-energy pathways.
binding site with better placement accuracy (Fig. 6.2).
If it is not possible to pattern the template with very 6.1.2 Mobility, Stability, and Yield
small binding sites, one either has to accept limited
placement accuracy or employ an additional focusing If the energy landscape is appropriate, a particle with
mechanism so that the particle will be deposited at sufficient mobility can explore it and assemble. With in-
a well-defined position inside the binding site. One ex- creasing mobility, it will (on average) find the binding
ample is the combination of electrostatic and capillary site more rapidly and escape from secondary minima
interactions [6.11]. more easily, but it will also have a larger probability
of escaping from the desired minimum. The probability
for a particle with mass m to escape from the bind-
Free energy ing site that produces a potential well with local shape
ω2 x 2 /2 surrounded by valleys of height Q equals, in
unit time [6.13],
ω
P= e−m Q/(kB T ) , (6.3)
2π
a result widely used in transition-state theory. This rate
can be limiting for the assembly process, but it is more
frequently the initial adsorption that requires most time.
Colloidal particles and solvated molecules gain the mo-
bility required to find binding sites through Brownian
motion by collisions with the solvent molecules. Equa-
tion (6.1) describes the ideal situation of an infinitely
Fig. 6.2 Placement accuracy and yield depend on the ge- dilute particle suspension, where no interaction between
ometry and potential shape of the binding sites the particles exists. In practice, interactions are very
192 Part A Nanostructures, Micro-/Nanofabrication and Materials
common at higher particle concentrations, and they in- is the most widespread regime in templated assembly
fluence the particle mobility according to [6.14] that uses chemical patterns on a submerged template
c (∂μ/∂c) p,T with sparse binding sites.
D(c) = , (6.4) If multiple particles are to be assembled in a single
1−c f (c) step in close proximity, the same repulsive interactions
which gives the diffusion constant D as a function that prevent agglomeration in the colloid have to be
of the number concentration c, the chemical poten- overcome to pack the particles densely. These forces
tial μ, and the friction coefficient f . If the particles are considerable. In a stable, aqueous colloid, the elec-
interact strongly, the chemical potential will increase trostatic repulsion, characterized by eψs , will generally
with increasing concentration, and so will the diffusion be greater than 10 kB T and often around 100 kB T . Over-
constant. At increased concentrations, the assembly be- coming this barrier to reach the energy minimum caused
havior will then change. For gold colloids, the apparent by van der Waals interactions therefore requires a large
diffusivity can be increased by two orders of magnitude driving force. Alternatively, the ionic strength can be in-
at increased concentrations, but drop radically when the creased locally to lower the electrostatic interaction and
range of stable concentrations is exceeded. create a funnel through which the particles can reach the
Equation (6.4) is a thermodynamic expression, and densely packed stage. If none of the above is present,
the link to the microscopic events at a binding site in sparse packing will result, described by modified ran-
templated assembly is not trivial. A reduced diffusion dom sequential adsorption models, as discussed later.
constant can indicate a reduced escape rate from a bind- Many technological applications of self-assembled
ing site, but the thermodynamic value obviously does nanostructures, in particular those in electronics, require
not hold when regarding a single particle, particularly high yields of assembly and well-defined arrangements.
if it is in the proximity of a binding site and encounters This is in contrast to biological systems, where defects
additional interactions. Statistical effects (as addressed can be repaired through error-correction mechanisms.
in Sect. 6.1.4) are also not covered. Two consequences In the absence of such mechanisms, however, the yield
of particle interaction are particularly important: the po- of assembly has to be very high. This yield depends on
tential required to increase the concentration of a colloid the nature of the binding sites, the concentration of par-
Part A 6.1
locally and the limited stability of colloidal suspensions. ticles, and the characteristics of the assembly process. In
Colloid scientists have long studied the case of in- particular, we can differentiate between abrupt assem-
teracting particles to derive expressions for colloidal bly processes, where the actual particle deposition and
stability. Smoluchowski and others derived expressions its final immobilization (or quenching) occur almost si-
for the rate of agglomeration as a function of particle multaneously, and gradual assembly processes, where
mobility and interaction, arriving at a characteristic time the two steps are not coupled. If the assembly takes
for doublet formation of place in the front of a receding meniscus that moves
πμa3 W over a solid template, it will leave the particle dry and
tp = , (6.5) immobile, and if a binding site stays empty, there is no
φkB T second chance for it to be filled. If the template is en-
which depends on temperature T and viscosity μ [6.14]. tirely submerged in the liquid, on the other hand, we
The value is inversely proportional to the volume frac- can at least theoretically wait until every binding site is
tion φ of the particles and depends strongly on their filled. For a rough estimate of the assembly rate, we can
diameter a and the interaction potentials (expressed via use Schurr’s expression [6.15] for the particle flux Js to
the stability ratio W). Doublets can thus form at time a surface
scales ranging from milliseconds to many hours, a very
wide range that is reflected in the qualitative statement kB T
that a colloid is stable or unstable towards flocculation. Js = cs , (6.6)
2πm
A similar time scale will govern the templated assem-
bly of particles. Different regimes occur, also depending which assumes a Maxwell–Boltzmann-distribution to
on the hydrodynamic situation. The assembly can be derive the flux from the particle number concentration at
purely Brownian (if there is no flow present), diffusion the surface cs and their mass m. If we know the sticking
limited (for a rapid and efficient adsorption process at probability S of the binding site, i. e., the probability for
low concentrations and high surface densities) or reac- a particle to be adsorbed upon contact, we can directly
tion limited, if sufficient particles are present. The latter calculate the half-life of an empty binding site.
Templated Self-Assembly of Particles 6.1 The Assembly Process 193
If particles readily desorb from binding sites, an by tuning the strength of the interaction, particle ar-
equilibrium situation will finally develop. The yield will rangements between well-ordered layers and randomly
then never reach unity, and its value will fluctuate over distributed submonolayers can be obtained.
time.
6.1.4 Thermodynamics, Kinetics,
6.1.3 Large Binding Sites and Statistics
Consider a particle that hits a binding site with area A. Diffusion constants scale inversely with the par-
If the particle gets sufficiently close to the site and if its ticle radius. The diffusion constants of nanoparticles
interaction with the site is sufficiently large to overcome are therefore much smaller than those of molecules.
Brownian motion, the particle will be adsorbed. When A 100 nm-diameter sphere moving in water will exhibit
we have a large number of such binding sites, particles a diffusion constant D of approximately 10−12 m2 /s.
will be randomly arranged inside the various A, so that Diffusion-limited processes with particles are thus slow,
the precision of arrangement is limited by the minimum equilibrium situations can often not be reached in ob-
size that (a) the template patterning can produce and (b) servable times, and the kinetics of the assembly process
allows for sufficiently rapid particle assembly. influences the assembly results. From an energy land-
If, on the other hand, a funneling effect of the kind scape point of view, it is not sufficient to provide
discussed above is present, the distribution of the par- a well-defined minimum in an appropriate position; the
ticles might be biased towards a certain part of A. Then, pathway to this minimum also has to be taken into ac-
the width of the position distribution is the result of count.
the competition between a stochastic force (Brownian Most real template–particle systems will have com-
in general) and the directing force. plex energy landscapes with a variety of secondary
If the area A is large enough to accommodate mul- minima and kinetic traps. A well-known example is
tiple particles, particles can either arrange into random a chemically functionalized surface onto parts of which
submonolayers or into ordered dense layers. The first particles should bind specifically. In practice, one finds
case, particle adsorption on strongly adsorbing surfaces, unspecific deposition and a certain degree of particle
Part A 6.1
is described reasonably well by the random sequen- accumulation, both caused typically by unspecific van
tial adsorption (RSA) model, which predicts a random der Waals-type attractions. Countermeasures include
particle distribution. Adsorption ceases when there is stirring, which increases particle mobility and keeps
no space left in the binding area that could accommo- them from settling in secondary minima; rapid process-
date an additional particle. The final packing density ing, which decreases the number of undesired particle
is called the jamming limit, which can be numerically collisions and thus the probability of reaching such
found to be θ∞ ≈ 0.547 for two-dimensional, circular a minimum; and washing, which removes weakly bound
particles [6.16]. Random sequential adsorption is the particles.
subject of numerous reviews, which also discuss its ap- There is one limitation, however, that cannot be
plication to anisotropic particle such as rods [6.17, 18]. overcome by such mobility-increasing measures. When
The RSA model accurately describes many molecu- the number of particles in the volume affected by a bind-
lar adsorption problems, in particular the adsorption of ing site is small, the probability of finding at least one
proteins on surfaces. It does not cover processes that particle inside this volume will be small too. In the sim-
result in dense ordered arrangements, for example, self- ple Poisson model of the situation, a volume V would
assembled monolayers (SAM). In contrast to the RSA contain a certain number n of particles with probability
model, the molecules that constitute a SAM retain some νn
mobility even after they are adsorbed on the surface. W(n) = e−ν , (6.7)
They interact with other molecules even before they ad- n!
sorb, and they interact with the underlying metal film. where υ is the average number of particles in the vol-
Larger particles sometimes behave similarly. The rear- ume, ν = Vc in the homogenous case. The probability
rangement of particles in an evaporating liquid film due of finding at least a single particle in this volume is
to capillary forces is a well-known example. therefore smaller than
When dense ordered packings are desired, the ∞
particle–surface interaction has to be appropriate to W(n ≥ 1) ≤ W(n) = 1 − e−ν , (6.8)
avoid RSA-like adsorption. It turns out [6.19] that, n=1
194 Part A Nanostructures, Micro-/Nanofabrication and Materials
and the particle concentration has to be above mum yield, say, 90%, the colloid concentration at the
binding sites therefore has to be at least c = 2.3/V ,
ln(1 − γ ) independent of any further process details. This con-
cc ≥ − (6.9)
V centration can be provided either by an overall larger
colloid concentration or (often more practical) by an
to guarantee a certain probability γ for a particle to additional, long-range force that acts on many particles,
be present. This limits the yield in assembly methods much like a funnel again. Electrostatic or hydrodynamic
which only capture particles during a short period of forces can increase the particle concentration locally,
time from the volume V : when there is no particle for example, at a three-phase boundary line, and enable
present, none can be assembled. When we regard a large sufficient assembly yields. We will see how this is done
number of binding sites and require a certain mini- experimentally in the next section.
ticles tend to agglomerate into larger clusters due to film electret by contacting a nanopatterned electrode
strong van der Waals interactions, thus making it almost with the target substrate [6.20, 21]. The electret mater-
impossible to arrange patterns of individual particles. ial can be a polymer (poly(methylmethacrylate) PMMA
Therefore, submicron-sized particles are often delivered or a fluorocarbon layer) or SiO2 . The flexible patterned
as suspensions in a liquid medium, especially when they electrode is made from a patterned silicone elastomer
were synthesized in liquid phase. Usually, nanoparticles (polydimethylsiloxane PDMS) with a thin conductive
are easier to stabilize in liquid, and particle agglomera- gold layer evaporated on top [6.20] or from thin pat-
tion is prevented by surface chemicals creating a surface terned silicon on top of a flat PDMS sheet [6.22]. The
charge or by the addition of surfactants. flexible electrode is brought into direct contact with
For assembly from the liquid phase we differentiate the electret and the charge image is produced by an
two cases: assembly from the bulk liquid onto the solid electrical pulse. Charge patterns can also be produced
template (Fig. 6.3b) or assembly at the solid–liquid–gas by sequentially writing with a conducting atomic force
boundary, i. e., at the meniscus of a liquid front moving microscopy (AFM) tip [6.23], although in these exam-
Part A 6.2
on the substrate. The electric field of the charged resist 230 nm
pattern guides the nanoparticles into the areas of free sil-
icon substrate. The additional ions improve the contrast Fig. 6.5 Ag particles (10 nm) assembled in 230 nm-wide
between deposition in desired and undesired areas of the lines. The inset shows the funneling effect which reduces
template (Fig. 6.5). By controlling the amount of ions it the actual width of the assembled particle lines to only
is possible to create and control a funneling effect which 50 nm. The scale bar corresponds to 1 μm (after [6.25],
c
focuses the nanoparticles into structures much smaller Macmillan 2006)
than the actual template pattern. Among the smallest
structures that have been realized by this method are
35 nm features assembled from 10 nm Ag nanoparti-
cles in 200 nm holes (Fig. 6.6) [6.25]. In the majority of
35 nm
these assemblies, multiple nanoparticles are deposited
into one assembly site and it is difficult if not impossible
to assemble single nanoparticles with high yield.
100 nm
6.2.2 Assembly in the Liquid Phase
Electroplated connection
Fig. 6.7a,b Schematic description of the multibatch assembly process (1–4) with SAM-covered binding sites that can
be deactivated selectively (a). A two-batch assembly result fabricated according to the scheme (b) (after [6.32],
c IOP
2003)
tions [6.30], and form factor [6.8, 31] have been used. switched off and reactivated later for a second assembly
Also, electric fields can be applied to direct the particles step (Fig. 6.7) [6.32,36]. In this way, different objects or
or nanoobjects towards the targeted adsorption sites. particles can be assembled onto the same template se-
quentially. For this purpose, the assembly sites consist
Wetting Contrast of gold electrodes which are covered by a hydrophobic
For larger particles and objects, ranging from mil- alkanethiol SAM. The alkanethiol SAM can be elec-
limeters down to several tens of micrometers, wetting trochemically removed from individual electrodes in
contrast in combination with capillary forces is applied a selective manner. When dipped into an adhesive, only
Part A 6.2
for the assembly [6.27, 32, 33]. Topographic three- the hydrophobic SAM-covered sites of the template are
dimensional (3-D) features on the template may support wetted and covered with an adhesive layer. In the subse-
the assembly in addition and introduce selectivity in quent assembly step, only the adhesive-covered sites are
a multicomponent assembly [6.34]. The template has active and can grab an object from solution. After the
hydrophobic assembly sites which can be selectively first particle assembly, all vacant electrodes can be mod-
covered by a layer of adhesive or solder. The objects ified with a SAM, simply by dipping into an alkanethiol
to be assembled are agitated in a fluid. In the simplest solution, and the process can begin again.
case – when a low-melting solder or a liquid organic ad-
hesive is used – the fluid is water [6.35]. When higher Electrostatic Nanoparticle Adsorption
temperatures are necessary to melt the solder, ethylene In liquid suspensions, particles are usually stabilized by
glycol [6.33, 34] can be used as a fluid. The suspended surface charges. These surface charges prevent the par-
objects selectively adhere to the solder or adhesive when ticles from agglomerating and can be exploited to guide
they come into contact. Objects to be assembled may the particles by electrostatic interaction to adsorption
also have a combination of hydrophilic and hydropho- sites of opposite charge. The template needs to dis-
bic faces, which makes them adsorb with a preferred play a contrast in surface charge. This can be achieved
side or orientation. The strong capillary forces of the by microcontact printing of SAMs with charged end-
solder or adhesive guide the assembled objects into groups [6.11, 37]. The pattern contrast can be further
the desired orientation. The geometry of the adsorp- enhanced through layer-by-layer (LBL) adsorption of
tion sites and of the attached surfaces play a crucial polyelectrolyte multilayers onto the printed monolay-
role in this last step because local energy minima might ers [6.26,38]. Microcontact printing of a polyelectrolyte
freeze the assembled objects into undesired orientations pattern onto LBL multilayers also results in a pattern of
on the template if the binding sites are not designed different surface charge on the template [6.39]. Other
carefully. methods based on nanoimprint lithography (NIL) and
Böhringer and coworkers devised a method in subsequent monolayer formation have been described
which hydrophobic assembly sites can be selectively as well (Sect. 6.3) [6.40].
Templated Self-Assembly of Particles 6.2 Classes of Directed Particle Assembly 197
Part A 6.2
Sagiv and coworkers fabricated charged adsorption ate assembly sites for selective adsorption of nanoparticles
sites by means of writing with a conductive AFM into (after [6.41],
c American Chemical Society)
a self-assembled silane monolayer [6.41]. The other-
wise inert monolayer is activated by the charged AFM can either carry a positive charge to attract negatively
tip, and functionalized molecules can be coupled onto charged nanoparticles (Fig. 6.9) or carry a thiol group
the patterned areas (Fig. 6.8). The added molecules which binds to gold nanoparticles [6.28]. The latter
800 20
20
600 15
400 10
10
200 5
0 0 0
0 200 400 600 800 (nm) 0 200 400 600 800 1000 (nm)
Fig. 6.9 Au particles (17 nm) assembled on amino-terminated bilayer templates (after [6.41],
c American Chemical
Society 2004)
198 Part A Nanostructures, Micro-/Nanofabrication and Materials
such a template with 450 nm-wide holes (b) (after [6.42], appropriately coated particles. Of course, such specific
c Wiley-VCH 2007) interactions are ultimately based on the standard set
of interactions such as Coulombic and van der Waals
case constitutes an example of an adsorption mecha- interactions or hydrogen bonds. However, by steric con-
nism where the particles are fixed to their adsorption site formation of the interacting entities, selectivity towards
by a chemical bond and not by Coulomb interaction. In a small set of binding partners or even towards a single
this case the directing force of the surface charge is not species arises. As examples, we discuss supramolec-
present. Due to the high lateral resolution of the AFM it ular interactions of small hydrophobic groups with
is possible to create very small patterns of just a few tens cyclodextrins [6.30, 43] and DNA hybridization [6.29].
of nanometers. Still, the adsorption of several nanopar- Both types of interactions are short-range compared
ticles per site is routinely observed on these templates. with electrostatic forces. Thus, the particles have to
Aizenberg and coworkers [6.11] published an as- come into close vicinity to the binding sites either
sembly method that makes use of a combination of by diffusion or by other transport mechanisms, e.g.,
electrostatic adsorption and capillary forces. First, the through agitation. While the interaction of molecular
particles are attracted towards charged monolayer sites. subunits with the cyclodextrins is selective and depends
Then they are focused onto the sites by capillary forces on the size and polarity of the guest unit, DNA hy-
when the solvent dries. In this work, the charged mono- bridization is selective towards the exact composition,
layer sites were also prepared by microcontact printing and only the exact counterpart to the offered sequence
of charged thiol molecules. is recognized and adsorbed onto the binding site on the
Charges deposited into a surface of an electret can template.
also act as a template for the assembly of nanoparti- In the case of cyclodextrin as recognition species,
cles. As in nanoxerography (see the previous section), a monolayer of the cyclodextrin units is patterned onto
the electret can be a polymer or a surface oxide layer a surface. This can be done by nanoimprint lithogra-
and the charge pattern can be formed by a confor- phy, which can expose just a fraction of the substrate
mal electrode [6.20]. To create very small features in surface for the formation of the cyclodextrin mono-
Templated Self-Assembly of Particles 6.2 Classes of Directed Particle Assembly 199
Part A 6.2
for the adsorption of nanoparticles functionalized with The electrodes were covered with photoresist, which
the complementary strand [6.29,49]. DNA tiles can also had openings at the desired binding sites (Fig. 6.11).
carry a pattern of other specific binding sites such as bi- Nanowires were directed to and adsorbed onto those
otin functionalities. In this case templated assembly of sites. The topographic structure of the assembly sites
streptavidin-functionalized nanoparticles can be carried helped to maintain the wires in the correct positions
out [6.50]. The adsorption of nanoparticles onto spe- upon drying. The assembled wires were fixed in a plat-
cific sites of DNA tiles allows for very high resolution ing process and lift-off of the resist layer removed those
in the assembly. However, the problem of assembling nanowires that adsorbed onto undesired positions. This
the DNA tiles and DNA origami structures themselves combination of methods can significantly reduce the er-
onto solid supports at specific locations with designated ror count and increase the yield of the assembly process
orientations is not yet fully solved. An interesting ap- (Fig. 6.12).
proach to this problem using dielectrophoretic assembly
is described by Kuzyk et al. [6.51].
Dielectrophoretic Assembly
In dielectrophoresis, particles in a solvent are attracted
to or repelled from a nonuniform alternating-current
(AC) electrical field. The strength and direction of the
dielectrophoretic force depends on the dielectric prop-
erties of the particles, solvent, electrode configuration,
voltage, and frequency. By appropriate design of the
10 µm
electrodes, particles can be forced to desired areas on
the template. However, if there is no additional persist- Fig. 6.12 SEM image of rhodium nanowires assembled us-
ing force, the particles leave their positions as soon as ing the process depicted in Fig. 6.11. The scale bar is 10 μm
the AC field is turned off. Suzuki et al. applied a combi- (after [6.44],
c Macmillan 2008)
200 Part A Nanostructures, Micro-/Nanofabrication and Materials
A very versatile variant of dielectrophoretic assem- onto the substrate. Convective flow of water trans-
bly was demonstrated by Chiou and coworkers [6.53]. ports even more particles towards the edge of the
They fabricated an assembly setup with rewritable elec- drop, thus forming the well-known coffee-stain-like
trode patterns on a photoconductive surface. Simply by patterns [6.57]. When the convective flow of water
projecting an image through a microscope lens, they towards the meniscus can be controlled, it is pos-
could define their electrode pattern for dielectrophoretic sible to assemble particle monolayers or multilayers
assembly of 4.5 μm latex beads. in a reproducible manner [6.56, 58]. Particles can
When NPs are assembled from aqueous suspen- even be assembled in spaced arrays when the menis-
sions, drying is always a critical step where strong cus only gets pinned at some specific locations on
capillary forces of the drying droplet may act on the an otherwise smooth and nonwetting substrate. Such
assembled particles and destroy or alter the assembled pinning locations can be formed by geometric fea-
pattern. On the other hand, the strength and directing tures on the substrate, by a pattern of wetting spots,
capacity of capillary forces may be exploited to control or by spots of increased particle–substrate interac-
the assembly of nanoparticles very accurately, as shown tion. Many researchers have exploited this mechanism
in the next section. for templated particle assembly with different setups
(Fig. 6.13).
6.2.3 Assembly at Gas–Liquid Interfaces In most examples of this kind of assembly, the par-
ticles are dispersed in an aqueous colloidal suspension.
At the phase boundary between a colloidal suspen- Often, these suspensions contain surfactants to further
sion, the template, and the surrounding air, very strong stabilize the colloids and prevent them from agglom-
capillary forces may act, depending on the solvent com- eration and precipitation. When the meniscus of such
position used. In many microelectromechanical systems a particle suspension sweeps over a flat nonwetting sur-
(MEMS) those capillary forces are detrimental to the face, no particles are left on the substrate. The meniscus
fabricated microstructures and drying is a very criti- acts like a doctor blade, moving the particles over the
cal step in MEMS fabrication. However, those strong surface. At geometrical features on the substrate such as
directing forces can be exploited very well for the as- a hole or the step of a raised structure, the water menis-
Part A 6.2
sembly of particles onto a template. cus gets pinned and capillary forces can drive particles
When the meniscus of an aqueous particle suspen- into holes or corners.
sion gets pinned on a surface it deposits the particles In the simplest experimental setup, a drop of col-
at the phase boundary in monolayers and multilayers loidal suspension is left drying on a topographically
a) b) c) Optical
Patterned microscope
photoresist
x y
Peltier element
Flow
Fe Heat exchanger
Fc
F0
Motorized
Stepper motor translation stage
Fig. 6.13a–c Schematic depictions of capillary assembly setups: (a) dipping the template into the particle suspension and slowly
pulling it out (after [6.54],
c Wiley-VCH 2005); (b) assembly in a fluidic cell with a constant flow of particle suspension
(after [6.55],
c American Chemical Society 2001); (c) assembly on a motorized stage with controllable assembly speed and
temperature (after [6.56],
c American Chemical Society 2007)
Templated Self-Assembly of Particles 6.2 Classes of Directed Particle Assembly 201
a)
2 µm 3 µm
Part A 6.2
ticles into geometric features on the template. Here,
there is only minimal control of the yield and evolution
of the deposition process. At the start of the process,
a low concentration of particles will be present at the 10 µm
meniscus. Then, with increasing evaporation, more par-
ticles are driven to the edge of the drop with the flux
b)
of water, and the assembly yield will increase. Finally,
as the particle concentration in the drying drop reaches
higher values, particles start to agglomerate and deposit
in large aggregates. Thus, this simple method only sup-
plies a relative small fraction of the template area with
the desired assembly result.
Better assembly yield is achieved by placing the
template (almost) vertically into a container of the col-
loidal suspension (Fig. 6.13a) [6.60]. As the solvent
slowly evaporates, the meniscus moves over the tem-
plate and deposits the particles in a controlled manner.
Particles as small as 2 nm in diameter have been suc- 1 µm
cessfully assembled into template features of several
10 nm by this method [6.60]. Still, there is no direct Fig. 6.15a,b Optical micrograph of the assembly of 60 nm
control of particle concentration during the assembly Au particles into 3 μm-spaced holes. The bright accumula-
and little possibility to react to changing parameters. tion zone is clearly visible. (a) Optical micrograph (inset)
Better control can be gained by pulling the template in and SEM image of 60 nm Au particles assembled in a setup
a controlled manner out of the flask of colloidal suspen- as illustrated in Fig. 6.13c and transferred to a silicon wafer
sion [6.54, 61]. (after [6.59], c Macmillan 2007)
202 Part A Nanostructures, Micro-/Nanofabrication and Materials
and particles towards the meniscus. Lowering the tem- This can be regarded as a kind of hierarchical assembly,
perature reduces the water flux and allows the particles where the assembly mechanism helps to form a sub-
to diffuse away from this so-called accumulation zone structure with features smaller than those of the actual
into the bulk solution. Consequently, assembly yield template.
6.3 Templates
The template carries the positional information on par- Some assembly processes are more complex. A par-
ticle arrangement. In most templates used today, there is ticle might only be deposited on one side of a binding
a simple relation between the position of a binding site site, only in its center (even if there is enough space
and the final particle position. The binding site might for multiple particles around it) or multiple particles
be larger than the particle’s footprint or differ from its might fill a larger binding site with a regular structure.
shape; it might accommodate just a single particle or This can be desirable: the fabrication of the template
a large number of particles. When working with very is usually easier if the critical dimensions do not have
small particles, its shape might be irregular due to the to be identical to the particle’s diameter. In the ideal
limited resolution of the patterning process. The as- case, the template pattern would be easy to fabricate but
sembly process translates this geometry into a particle define the desired assembly of very small particles un-
arrangement, as discussed in Sect. 6.1. Still, the rela- ambiguously, whereupon the assembly process would
tion between template and particle position is simple: translate it into a very high-resolution energy landscape
there has to be a feature on the template exactly where for the particles to occupy. If the desired arrangement
a particle is intended to be placed. Thus, the pattern- is very complex, the template will generally have to be
ing technique used to make the template has to define rather complex too, but most practical arrangements are
the particle locations with an accuracy that is within the highly repetitive and modular, and could be encoded
range of the particle’s dimensions. efficiently.
Templated Self-Assembly of Particles 6.3 Templates 203
Many assembly templates are fabricated by means terference patterns, but they can do so over large areas in
of top-down micro- and nanofabrication. The patterns a single exposure step. If small structures are to be cre-
are usually designed in a computer, transferred to ated, the radiation wavelength has to be low, its intensity
a mask (often via electron-beam or laser direct writing), high, and its coherence sufficient. An excellent source
and converted to chemical or topographical patterns on is synchrotron radiation. Larger patterns are straight-
the template surface. The typical resolution limit for forward to create using simple laser interference. Other
templates formed using typical ultraviolet (UV) lithog- efficient routes to certain template geometries include
raphy is around 1 μm, although smaller structures are wrinkling patterns and step edges that form when crys-
achievable if artifacts are acceptable. Smaller structures tals are cleaved along high-index planes. All these
can be written using electron-beam (e-beam) lithog- templates are limited to specific geometries and, thus,
raphy. While photomasks for UV photolithography create specific particle arrangements.
(produced in large e-beam writers) are readily avail-
able commercially, sub-μm e-beam patterns in other 6.3.1 Chemical Templates
materials are less common, and the sequential nature
of e-beam writing makes the patterning of large areas Chemical templates display a pattern of selective
time consuming and costly. Electron-beam patterning is surface chemistries with areas that prevent particle
very flexible, however, and is widely used for research adsorption and others that support it. Additional geo-
purposes. metrical features are not necessary per se but can be
If the primary patterning method is costly, repli- helpful to increase selectivity. A simple wetting contrast
cation techniques are useful both to produce multiple (e.g., hydrophilic patches on a hydrophobic substrates)
templates from one primary pattern and to cover larger can be sufficient to assemble colloidal particles at
areas with a repeated pattern. For patterns below the the three-phase boundary [6.61]. In this case, the hy-
optical diffraction limit, molding and printing are pop- drophilic spots have to be large enough (> 25 μm) to
ular methods. The primary structure is used either to cause a significant lowering of the receding contact an-
imprint a polymer layer at increased temperature and gle and deposition of particles. Chemical patterns with
under pressure (nanoimprint lithography) or to shape features of several micrometers can be fabricated by op-
Part A 6.3
a liquid prepolymer while it is curing (molding, UV tical lithography. After exposure and development of
imprint lithography, and others). Although some top- the photoresist on a Si/SiO2 surface, the exposed sub-
down methods directly produce chemical patterns, for strate areas can be treated with a silane molecule. Upon
example, by oxidizing UV-sensitive monolayers in pho- removal of the remaining photoresist, a chemical pattern
tolithography or with an e-beam, the most common (bare Si/SiO2 surface versus silane surface) is achieved.
product is a topographical pattern. Soft lithography is For higher-resolution features, nanoimprint lithography
a route both to replicate such a pattern and to convert it or e-beam lithography can be utilized to pattern a poly-
to chemical contrast: a silicone prepolymer mixture is mer resist layer. Again, the accessible substrate areas
cured on the topographical pattern, cured to a solid rub- can then be patterned by a specific surface chemistry,
ber, and used as a stamp with the inverse pattern. This and subsequent removal of the polymer resist provides
stamp can then print molecules on various surfaces, the template. The areas of bare substrate may also be
a process named microcontact printing. covered by a surface chemistry orthogonal to the first
Even e-beam writing is limited to minimum fea- one (hydrophobic–hydrophilic, anionic–cationic) in or-
ture sizes in the range of tens of nanometers. Smaller der to increase assembly contrast [6.40]. Alternatively,
structures can be formed using probe methods. For the polymer resist might not be removed at all, thus pro-
example, the tip of an atomic force microscope can me- viding an additional 3-D feature to support assembly
chanically remove (scratch) a monolayer or oxidize its onto the template [6.40].
functional groups locally. The resulting template cannot Microcontact printing of organic monolayers is also
be replicated easily, which makes the process rather un- a viable method for the fabrication of surface-chemical
economical, but it provides extremely high resolution, patterns on oxide or noble-metal surfaces. Depending
for example, for the arrangement of metal clusters in on the quality of the stamp material and architecture,
lines to investigate their conductivity. even sub-μm patterns are attainable [6.64].
An efficient alternative to e-beam writing for the Many examples of chemical templates do not only
patterning of larger areas is the use of interference rely on a wetting contrast, but provide real adsorption
lithography techniques. They only produce regular in- sites for the particles. This can be achieved by pattern-
204 Part A Nanostructures, Micro-/Nanofabrication and Materials
ing a charged molecule or a polyelectrolyte layer, which terns, topographical templates are simple in structure
then attracts the particles by electrostatic forces or with and fabrication and can be replicated via molding and
specific supramolecular interactions [6.43, 54]. imprinting techniques.
Self-assembled structures of block copolymers on Topographical templates are used in convec-
surfaces can also act as a chemical template when tive [6.56] and capillary particle assembly [6.62],
nanoparticles selectively adsorb onto one block of the as an additional guide to the crystal structure in
polymer [6.65]. electrophoretic particle assembly [6.68], and as an ad-
ditional guide in dielectrophoretic assembly [6.44]. In
6.3.2 Charges and Electrodes all these cases, the geometries are very simple: holes
of uniform depth in an otherwise continuous layer. In
Electric charge can be brought onto a surface by means most cases, the structures are formed in photoresist by
of an electrode. This is the principle of xerography, standard UV lithography and used without further pro-
which has been scaled down using electrically con- cessing. If the resist sits on top of a wafer, it is simple to
ductive AFM tips to write very small charged areas create an additional wetting contrast between the (poly-
onto the surface of a dielectric. The charged regions mer) top surface and the (silicon oxide) bottom surface
attract particles, which then assemble on the written pat- of the template holes.
terns [6.23]. Small tips (readily available in an AFM) More complex geometries are required to precisely
enable high resolution to be attained, although the actual tune the forces in capillary assembly. Step edges,
charge pattern can deviate from the intended design. crosses, corner shapes, and other well-defined obsta-
Actively driven electrodes are the most versatile cles trap particles in reproducible arrangements. Such
option for electric-field templates. They require consid- templates are harder to fabricate than holes. They can,
erable effort in terms of interconnections and electrode however, be replicated in polymers, such as poly-
design and one has to avoid particle attractions to the dimethylsiloxane (PDMS). A single silicon master can
wiring, but they can be actively switched. What is more, then produce many (up to several hundred) single-use
AC potentials can be applied, so that dielectrophore- assembly templates.
sis takes place. Nanoparticles [6.66] and nanowires can Polymer molding is also the basis of the microflu-
Part A 6.3
thus be aligned with two electrodes and can then be idic ducts used as templates in the micromolding in
connected [6.67]. The electric leads for both assem- capillaries (MIMIC) process. These channels, in which
bly electrodes and device interconnects (which can be particles are arranged from a microfluidic flow, are first
identical) are fabricated using standard microfabrica- formed as lines in ultrathick resist and then replicated in
tion techniques. If necessary, they can be combined PDMS. The soft silicone replica adheres to flat surfaces,
with additional topographical features, for example, to forming channels into which the particle suspension is
improve alignment in the assembly of anisotropic par- sucked by capillary forces.
ticles [6.44].
Dielectrophoresis can also be driven by an ex- 6.3.4 Advanced Templates
ternal electromagnetic field that is projected onto an
appropriate substrate [6.53]. The projected image, mi- More than one force can be involved in particle assem-
croscopically demagnified, causes the assembly forces. bly, thus assembly templates can guide the assembly in
Such an image can be modulated and is far more flexible more than one way. Advanced templates combine, for
than patterned electrodes; it can even be time dependent example, a long-range force caused by electrostatic or
to further optimize the assembly process. Its resolution dielectrophoretic interactions with short-range interac-
is, however, diffraction limited. tions due to topography that provide high accuracy in
the last moments of assembly [6.44]. An electrode array
6.3.3 Topographical Templates can be created on a flat substrate and a polymer resist
patterned on top of the array, so that a hole in the shape
Purely topographical templates guide particle assem- of the particle remains. In a similar vein, a hydrophilic
bly by geometrical exclusion and by modulating other substrate can be coated with a hydrophobic resist, which
forces, for example, capillary interactions. Geometrical creates a wetting difference that helps to capture a liquid
confinement can be very precise – the particle cannot volume in the binding site [6.60].
enter a template wall – but it is limited by template If one of the particle–template interactions is con-
precision. Compared with electrodes or chemical pat- trollable (as is dielectrophoresis), such templates could
Templated Self-Assembly of Particles 6.4 Processes and Setups 205
be addressable and certain sites turned off during as- is correct. Together with a feedback loop that con-
sembly [6.32]. In the style of a raster, a general-purpose trols process parameters, the yield of assembly would
master with a relatively dense, regular array of bind- then be automatically optimized. This might be eas-
ing sites could then be modulated to produce arbitrary ier to realize than it seems: if electrodes are present
particle arrangements. in a template, it seems feasible to measure the dielec-
The ultimate template would not only be control- tric properties of the binding site, which will likely
lable, but could also sense whether a given binding site depend on the presence of a particle (and possibly its
is occupied, and if so, whether the particle alignment alignment).
Part A 6.4
assembly, however. The purpose of these setups is either be done in simple immersion setups, where the
to bring surface and particles into contact in a con- liquid slowly evaporates and the boundary line moves
trollable way, where convection and other disturbing over the vertically immersed patterned surface, or in
influences are minimized, and to monitor and control more involved setups, for example, the Capillary As-
conditions relevant for the assembly process, such as sisted Particle Assembly tool of Malaquin et al. [6.56].
contact angle, ionic strength, temperature, and field The speed of the moving meniscus, the contact angle,
strengths. Excess particles are removed without de- and the hydrodynamic situation inside the liquid (all rel-
stroying the particles’ order and might be reintroduced evant for the assembly process) are more or less well
to increase the assembly yield. Some setups allow controlled in the different setups.
inducing a bias, for example, to align anisotropic Hydrodynamic forces are also used for assembly
particles. in the absence of capillary bridges but in the pres-
If the assembly takes place at the liquid–solid in- ence of gravity, for larger particles well beyond the
terface, the final removal of the solvent (if so desired) limits of Brownian behavior, at about one micrometer.
is a critical step. Capillary forces have frequently been In so-called fluidic assembly, appropriately shaped ob-
found to destroy or change the particle arrangement. In jects fall into complementary shaped binding sites from
general, a final quenching of the assembled particles is a moving liquid [6.8]. This process is not very efficient,
required to permanently retain their order and enable and the objects have to be brought to the surface re-
further processing or integration into a device. peatedly. Setups have been devised to move the object
The integration of the assembled particles into slurry, agitate it, and recycle it.
a functioning device can involve electrical connections, Liquid flows also drive the assembly in micro-
optical coupling, thermal joining, and many other pro- molding in capillaries (MIMIC), a term coined by
cessing steps. The interfaces that are created often Whiteside’s group [6.70]. In this process, microfluidic
govern the device performance. Surface analysis and channels are filled with particles (usually in densely
modification are therefore common subsequent steps in packed structures). Xia introduced a similar method to
the integration process. assemble polystyrene spheres first on flat, but later also
206 Part A Nanostructures, Micro-/Nanofabrication and Materials
on patterned surfaces: water is drained slowly from his Blodgett films) or by bringing it into contact with the
cell through scratches or small ducts at the side, while surface horizontally (the Schaefer approach) [6.72]. Al-
ultrasonic agitation provides the mobility needed by the ternatively, a stamp is coated with the layer and then
particles to arrange themselves [6.71]. prints it onto a target.
Less widespread are vacuum setups for particle de- Printing has also been demonstrated for single-
position. Here, particle production can be coupled to particle arrays in a process called self-assembly, trans-
particle assembly and size selection. Particles are first fer, and integration (SATI), which uses a multistep
created (usually directly from a metal with thermal adhesion cascade [6.73]. In this approach, particles
methods), separated by size in an electric field, and as- leave one surface in favor of the other due to differences
sembled in a large vacuum chamber [6.25]. While such in adhesion, which has to be tuned. The strategy works
setups are considerably more complex than most liquid- with particles covering a wide size range; printing of
phase techniques, they allow particles to be deposited both 100 μm and 60 nm particles has been demon-
on very clean surfaces. strated [6.59].
Postprocessing of the (printed or directly assem-
6.4.2 Particle Printing and Processing bled) particles is generally required if they are to be
electrically connected or need to be embedded, pro-
It is often desirable to carry out the particle assem- tected, or to act either as a template or building block of
bly process away from the target substrate. Assembly a further structure. Parallel electric contacting of many
usually requires binding sites, which might be unde- assembled nanowires (e.g., for sensing purposes) has
sirable to have on the target; it often involves solution been demonstrated using conventional technology on
chemistry, which might contaminate the target; or it unconventionally assembled particles [6.44]. Other par-
requires specific surface properties, which the target ticles have been used as nucleation sites for nanowire
simply might not have. If the particles are assembled on growth [6.59], templates for etches and deposition pro-
a secondary surface, these requirements are lifted, but cesses, and transistor and memory elements. In all these
a transfer step becomes necessary to bring the particle cases, the particle surfaces were modified or covered
arrangement to its final destination. by layers of material. A frequent task is the removal of
Part A 6.5
In the classical case of a Langmuir–Blodgett trough, organic adlayers that remain following liquid-phase as-
the formed monolayer is transferred by drawing the tar- sembly, which can be effected using plasma ashing or
get surface through the interface vertically (Langmuir– thermal annealing.
6.5 Conclusions
Building devices and materials from nanoparticles has cesses. The main challenges here include the typically
been proven to be a feasible idea. It resembles in- very large number of particles that require extremely
dustrial production from standardized components. The high yields of assembly, the limited quality of even
advantages are similar: building blocks, here nanopar- the best chemically synthesized nanoparticles, and the
ticles, can be produced efficiently in large quantities preparation of suitable templates. Once such obstacles
if they are not too complex, and they can be mod- are overcome, templated assembly enables hierarchical,
ified and inspected and then used to build different complex structures to be made in large quantities using
products. As long as it is impossible to build complex relatively simple equipment.
structures directly from atoms and molecules, particle All these challenges are currently being addressed.
assembly will be one of the most interesting routes to Mechanistic understanding of particle synthesis is in-
creating nanostructures. Templated particle assembly creasing as synthesis protocols are being analyzed in
reduced the process complexity even further and ex- detail, scale-up is investigated, and alternative routes
isting methods already provide exquisite control over become available for many popular nanoparticles. An
particle positioning. increasing range of particles is now commercially avail-
There are some niches where templated assembly able in consistent quality.
is used industrially today, be it in the fabrication of Templates can already be fabricated with high qual-
RFID tags or for bead-based assays, but sub-μm particle ity on small areas, and various researchers are working
assembly is yet to be introduced into production pro- on large-scale nanopatterning and replication methods
Templated Self-Assembly of Particles References 207
to produce large areas of identical patterns efficiently. (CMOS)-type components, the assembly and transfer
Nanoimprint lithography, for example, can replicate precision has to be adequate to match the underly-
a master many times and is even compatible with roll- ing structures. In most cases, short-range accuracy
to-roll-type fabrication, where long plastic sheets are is governed by the assembly process and the pre-
continuously patterned by a rotating drum. As with cision of the template, while long-range order is
molding, nanoimprint lithography is not an alternative influenced mainly by the template and the trans-
to templated nanoparticle assembly, since it can only fer process. All three may have to be optimized
handle a very limited set of materials, but it is ideal for to meet the stringent requirements of semiconductor
the production of templates. fabrication.
Finally, the assembly processes themselves are In addition to such improvements, the develop-
constantly improving. Improved understanding of the ment of templated assembly processes for increasingly
interactions during assembly allows researchers to tune smaller particles with very high accuracy will continue.
the interaction strengths and thereby engineer the en- An important goal here is the assembly of particles
ergy landscape of the assembly process. Thus, both the well below 10 nm in diameter, which exhibit electronic
stability of the original particles (for example, the col- quantum effects, with a precision that is sufficient to
loidal suspension) and their behavior during assembly connect them electronically. Ideally, this would be pos-
are optimized towards high yield. In addition, better sible on areas far above the square centimeters that
control of the process parameters during assembly is have so far been demonstrated, if possible on standard
now possible in modified versions of classical dip- 300 mm wafers. Finally, the assembly (and, if necessary,
coating setups. When combined with in situ analysis the transfer) should be compatible with different par-
methods, yields and assembly qualities can be opti- ticle materials and substrates. A truly versatile process
mized by adjusting parameters such as temperature and would accept any colloidal particle and thus be able to
template velocity. handle a very wide range of materials including oxides,
Today, coatings containing nanoparticles are com- semiconductors, metals, and polymers, amongst many
monly applied using dip-coating, spin-coating or spray- others.
coating techniques. Such methods are comparatively The ideal process would also handle very small par-
Part A 6
simple and compatible with a variety of relevant ge- ticles. How small? We do not know at present. Gold-55
ometries. If templated assembly could be performed clusters that resemble molecules rather than particles
using the same deposition techniques, this would render have already been arranged using templated assembly
it compatible with established technology and simplify processes, albeit with a precision far worse than the par-
its introduction into other processes. Alternatively, if ticle diameter [6.28]. Will it be possible at some point to
specialized deposition techniques are required, or if arrange single atoms and molecules on a surface using
the template cannot be applied to the substrate, as- a reasonably simple template? That such arrangements
sembly can be performed on a specialized template are stable and lead to interesting effects has already
and the particles subsequently transferred onto the tar- been demonstrated using high-vacuum scanning tun-
get surface. Together, these processes bridge the gap neling microscopy [6.74]. Whether templated assembly
between particle assembly and current standard meth- can provide a realistic route to such patterning with ul-
ods of fabrication. If particles are to be combined timate precision will remain an active topic of research
with, say, complementary metal–oxide–semiconductor for years to come.
References
6.1 A.N. Shipway, E. Katz, I. Willner: Nanoparticle ar- 6.3 F. Caruso: Nanoengineering of particle surfaces, Adv.
rays on surfaces for electronic, optical, and sensor Mater. 13(1), 11 (2001)
applications, Chem. Phys. Chem. 1(1), 18–52 (2000) 6.4 K. Molhave, T.M. Hansen, D.N. Madsen, P. Boggild:
6.2 C.B. Murray, C.R. Kagan, M.G. Bawendi: Synthesis Towards pick-and-place assembly of nanostruc-
and characterization of monodisperse nanocrystals tures, J. Nanosci. Nanotechnol. 4(3), 279–282 (2004)
and close-packed nanocrystal assemblies, Annu. 6.5 P. Alivisatos: The use of nanocrystals in biological
Rev. Mater. Sci. 30, 545–610 (2000) detection, Nat. Biotechnol. 22(1), 47–52 (2004)
208 Part A Nanostructures, Micro-/Nanofabrication and Materials
6.6 S. Paul, C. Pearson, A. Molloy, M.A. Cousins, generated by a conductive atomic force microscope
M. Green, S. Kolliopoulou, P. Dimitrakis, P. Nor- tip, Adv. Mater. 13(18), 1395–1398 (2001)
mand, D. Tsoukalas, M.C. Petty: Langmuir–Blodgett 6.24 C.R. Barry, H.O. Jacobs: Fringing field directed
film deposition of metallic nanoparticles and their assembly of nanomaterials, Nano Lett. 6(12), 2790–
application to electronic memory structures, Nano 2796 (2006)
Lett. 3(4), 533–536 (2003) 6.25 H. Kim, J. Kim, H.J. Yang, J. Suh, T. Kim, B.W. Han,
6.7 J.D. Joannopoulos, S.G. Johnson, J.N. Winn, S. Kim, D.S. Kim, P.V. Pikhitsa, M. Choi: Parallel pat-
R.D. Meade: Photonic Crystals: Molding the Flow of terning of nanoparticles via electrodynamic focusing
Light (Princeton Univ. Press, Princeton 2008) of charged aerosols, Nat. Nanotechnol. 1(2), 117–121
6.8 H.J.J. Yeh, J.S. Smith: Fluidic self-assembly for the (2006)
integration of GaAs light-emitting-diodes on Si sub- 6.26 I. Lee, H.P. Zheng, M.F. Rubner, P.T. Hammond:
strates, IEEE Photon. Technol. Lett. 6(6), 706–708 Controlled cluster size in patterned particle arrays
(1994) via directed adsorption on confined surfaces, Adv.
6.9 A. Einstein: The theory of the Brownian motion, Ann. Mater. 14(8), 572–577 (2002)
Phys. 19(2), 371–381 (1906) 6.27 U. Srinivasan, D. Liepmann, R.T. Howe: Microstruc-
6.10 R.M. Mazo: Brownian Motion. Fluctuations, Dynam- ture to substrate self-assembly using capillary
ics, and Applications (Clarendon, Oxford 2002) forces, J. Microelectromech. Syst. 10(1), 17–24 (2001)
6.11 J. Aizenberg, P.V. Braun, P. Wiltzius: Patterned col- 6.28 S.T. Liu, R. Maoz, G. Schmid, J. Sagiv: Template
loidal deposition controlled by electrostatic and guided self-assembly of [Au(55)] clusters on nano-
capillary forces, Phys. Rev. Lett. 84(13), 2997–3000 lithographically defined monolayer patterns, Nano
(2000) Lett. 2(10), 1055–1060 (2002)
6.12 N.D. Denkov, O.D. Velev, P.A. Kralchevsky, I.B. Ivanov, 6.29 J.D. Le, Y. Pinto, N.C. Seeman, K. Musier-Forsyth,
H. Yoshimura, K. Nagayama: Mechanism of forma- T.A. Taton, R.A. Kiehl: DNA-templated self-assembly
tion of 2-dimensional crystals from latex-particles of metallic nanocomponent arrays on a surface,
on substrates, Langmuir 8(12), 3183–3190 (1992) Nano Lett. 4(12), 2343–2347 (2004)
6.13 S. Chandrasekhar: Stochastic problems in physics 6.30 P. Maury, M. Peter, O. Crespo-Biel, X.Y. Ling,
and astronomy, Rev. Mod. Phys. 15(1), 1–89 (1943) D.N. Reinhoudt, J. Huskens: Patterning the molecu-
6.14 W.R. Russel, D.A. Saville, W.R. Schowalter: Colloidal lar printboard: patterning cyclodextrin monolayers
Dispersions (Cambridge Univ. Press, Cambridge 1989) on silicon oxide using nanoimprint lithography and
its application in 3-D multilayer nanostructuring,
Part A 6
6.38 K.M. Chen, X.P. Jiang, L.C. Kimerling, P.T. Hammond: ing with microparticle monolayers covalently linked
Selective self-organization of colloids on patterned to glass surfaces, Langmuir 20, 11005–11011 (2004)
polyelectrolyte templates, Langmuir 16(20), 7825– 6.53 P.Y. Chiou, A.T. Ohta, M.C. Wu: Massively parallel
7834 (2000) manipulation of single cells and microparticles using
6.39 H.P. Zheng, M.F. Rubner, P.T. Hammond: Particle optical images, Nature 436(7049), 370–372 (2005)
assembly on patterned "plus/minus" polyelectrolyte 6.54 P. Maury, M. Escalante, D.N. Reinhoudt, J. Huskens:
surfaces via polymer-on-polymer stamping, Lang- Directed assembly of nanoparticles onto polymer-
muir 18(11), 4505–4510 (2002) imprinted or chemically patterned templates fab-
6.40 P. Maury, M. Peter, V. Mahalingam, D.N. Reinhoudt, ricated by nanoimprint lithography, Adv. Mater.
J. Huskens: Patterned self-assembled monolayers 17(22), 2718–2723 (2005)
on silicon oxide prepared by nanoimprint lithogra- 6.55 Y.D. Yin, Y. Lu, B. Gates, Y.N. Xia: Template-assisted
phy and their applications in nanofabrication, Adv. self-assembly: A practical route to complex aggre-
Funct. Mater. 15(3), 451–457 (2005) gates of monodispersed colloids with well-defined
6.41 S.T. Liu, R. Maoz, J. Sagiv: Planned nanostructures sizes, shapes, and structures, J. Am. Chem. Soc.
of colloidal gold via self-assembly on hierarchi- 123(36), 8718–8729 (2001)
cally assembled organic bilayer template patterns 6.56 L. Malaquin, T. Kraus, H. Schmid, E. Delamarche,
with in-situ generated terminal amino functional- H. Wolf: Controlled particle placement through
ity, Nano Lett. 4(5), 845–851 (2004) convective and capillary assembly, Langmuir 23,
6.42 Y.H. Kim, J. Park, P.J. Yoo, P.T. Hammond: Selective 11513–11521 (2007)
assembly of colloidal particles on a nanostructured 6.57 R.D. Deegan, O. Bakajin, T.F. Dupont, G. Huber,
template coated with polyclectrolyte multilayers, S.R. Nagel, T.A. Witten: Capillary flow as the cause
Adv. Mater. 19(24), 4426 (2007) of ring stains from dried liquid drops, Nature
6.43 X.Y. Ling, I.Y. Phang, D.N. Reinhoudt, G.J. Vancso, 389(6653), 827–829 (1997)
J. Huskens: Supramolecular layer-by-layer assembly 6.58 B.G. Prevo, O.D. Velev: Controlled, rapid deposition
of 3-D multicomponent nanostructures via multi- of structured coatings from micro- and nanoparticle
valent molecular recognition, Int. J. Mol. Sci. 9, suspensions, Langmuir 20(6), 2099–2107 (2004)
486–497 (2008) 6.59 T. Kraus, L. Malaquin, H. Schmid, W. Riess,
6.44 M.W. Li, R.B. Bhiladvala, T.J. Morrow, J.A. Sioss, N.D. Spencer, H. Wolf: Nanoparticle printing with
K.K. Lew, J.M. Redwing, C.D. Keating, T.S. Mayer: single-particle resolution, Nat. Nanotechnol. 2,
570–576 (2007)
Part A 6
Bottom-up assembly of large-area nanowire res-
onator arrays, Nat. Nanotechnol. 3(2), 88–92 (2008) 6.60 Y. Cui, M.T. Bjork, J.A. Liddle, C. Sonnichsen,
6.45 B. Kannan, R.P. Kulkarni, A. Majumdar: DNA-based B. Boussert, A.P. Alivisatos: Integration of colloidal
programmed assembly of gold nanoparticles on nanocrystals into lithographically patterned de-
lithographic patterns with extraordinary specificity, vices, Nano Lett. 4(6), 1093–1098 (2004)
Nano Lett. 4, 1521–1524 (2004) 6.61 C.A. Fustin, G. Glasser, H.W. Spiess, U. Jonas: Param-
6.46 H. Zhang, Z. Li, C.A. Mirkin: Dip–pen nanoli- eters influencing the templated growth of colloidal
thography-based methodology for preparing arrays crystals on chemically patterned surfaces, Langmuir
of nanostructures functionalized with oligonu- 20, 9114–9123 (2004)
cleotides, Adv. Mater. 14, 1472–1474 (2002) 6.62 Y.N. Xia, Y.D. Yin, Y. Lu, J. McLellan: Template-
6.47 E. Winfree, F.R. Liu, L.A. Wenzler, N.C. Seeman: De- assisted self-assembly of spherical colloids into
sign and self-assembly of two-dimensional DNA complex and controllable structures, Adv. Funct.
crystals, Nature 394(6693), 539–544 (1998) Mater. 13(12), 907–918 (2003)
6.48 P.W.K. Rothemund: Folding DNA to create nanoscale 6.63 M.J. Gordon, D. Peyrade: Separation of colloidal
shapes and patterns, Nature 440(7082), 297–302 nanoparticles using capillary immersion forces,
(2006) Appl. Phys. Lett. 89(5), 053112 (2006)
6.49 J. Sharma, R. Chhabra, Y. Liu, Y. Ke, H. Yan: DNA- 6.64 H. Schmid, B. Michel: Siloxane polymers for
templated self-assembly of two-dimensional and high-resolution, high-accuracy soft lithography,
periodical gold nanoparticle arrays, Angew. Chem. Macromolecules 33(8), 3042–3049 (2000)
Int. Ed. 45, 730–735 (2006) 6.65 C. Minelli, C. Hinderling, H. Heinzelmann, R. Pugin,
6.50 I. Cheng, B. Wei, X. Zhang, Y. Wang, Y. Mi: Patterning M. Liley: Micrometer-long gold nanowires fabricated
of gold nanoparticles on DNA self-assembled scaf- using block copolymer templates, Langmuir 21(16),
folds, Res. Lett. Nanotechnol. 2008, 827174 (2008) 7080–7082 (2005)
6.51 A. Kuzyk, B. Yurke, J.J. Toppari, V. Linko, P. Törmä: 6.66 O.D. Velev, E.W. Kaler: In situ assembly of colloidal
Dielectrophoretic trapping of DNA origami, Small 4, particles into miniaturized biosensors, Langmuir
447–450 (2008) 15(11), 3693–3698 (1999)
6.52 M. Suzuki, T. Yasukawa, Y. Mase, D. Oyamatsu, 6.67 P.A. Smith, C.D. Nordquist, T.N. Jackson, T.S. Mayer,
H. Shiku, T. Matsue: Dielectrophoretic micropattern- B.R. Martin, J. Mbindyo, T.E. Mallouk: Electric-
210 Part A Nanostructures, Micro-/Nanofabrication and Materials
field assisted assembly and alignment of metallic idic cells fabricated using non-photolithographic
nanowires, Appl. Phys. Lett. 77(9), 1399–1401 (2000) methods, Langmuir 17(20), 6344–6350 (2001)
6.68 N.V. Dziomkina, G.J. Vancso: Colloidal crystal assem- 6.72 I. Langmuir, V.J. Schaefer: Activities of urease and
bly on topologically patterned templates, Soft Matter pepsin monolayers, J. Am. Chem. Soc. 60, 1351–1360
1(4), 265–279 (2005) (1938)
6.69 G. Markovich, C.P. Collier, J.R. Heath: Re- 6.73 T. Kraus, L. Malaquin, E. Delamarche, H. Schmid,
versible metal-insulator transition in ordered metal N.D. Spencer, H. Wolf: Closing the gap between self-
nanocrystal monolayers observed by impedance assembly and microsystems using self-assembly,
spectroscopy, Phys. Rev. Lett. 80(17), 3807–3810 transfer, and integration of particles, Adv. Mater.
(1998) 17(20), 2438–2442 (2005)
6.70 E. Kim, Y.N. Xia, G.M. Whitesides: Micromolding in 6.74 F. Rosei, M. Schunack, Y. Naitoh, P. Jiang, A. Gour-
capillaries: Applications in materials science, J. Am. don, E. Laegsgaard, I. Stensgaard, C. Joachim, F. Be-
Chem. Soc. 118(24), 5722–5731 (1996) senbacher: Properties of large organic molecules
6.71 Y. Lu, Y.D. Yin, B. Gates, Y.N. Xia: Growth of large on metal surfaces, Prog. Surf. Sci. 71(5–8), 95–146
crystals of monodispersed spherical colloids in flu- (2003)
Part A 6
211
Three-Dimen
7. Three-Dimensional Nanostructure Fabrication
by Focused Ion Beam Chemical Vapor Deposition
Shinji Matsui
7.1 Three-Dimensional
In this chapter, we describe three-dimensional
Nanostructure Fabrication ..................... 212
nanostructure fabrication using 30 keV Ga+ focused
7.1.1 Fabrication Process....................... 212
ion beam chemical vapor deposition (FIB-CVD) and
7.1.2 Three-Dimensional
a phenanthrene (C14 H10 ) source as a precursor. We Pattern-Generating System ........... 214
also consider microstructure plastic art, which is
a new field that has been made possible by mi- 7.2 Nanoelectromechanics .......................... 215
crobeam technology, and we present examples of 7.2.1 Measuring Young’s Modulus .......... 215
such art, including a micro wine glass with an ex- 7.2.2 Free-Space Nanowiring................. 217
ternal diameter of 2.75 µm and height of 12 µm. The 7.2.3 Nanomechanical Switch ................ 220
film deposited during such a process is diamond- 7.2.4 Nanoelectrostatic Actuator ............ 221
like amorphous carbon, which has a Young’s
modulus exceeding 600 GPa, appearing to make
7.3 Nanooptics:
Brilliant Blue Observation
it highly desirable for various applications. The
from a Morpho Butterfly Scale
production of three-dimensional nanostructure is
Quasistructure ...................................... 223
discussed. The fabrication of microcoils, nanoelec-
trostatic actuators, and 0.1 µm nanowiring – all 7.4 Nanobiology ........................................ 224
potential components of nanomechanical systems 7.4.1 Nanoinjector ............................... 224
– is explained. The chapter ends by describing 7.4.2 Nanomanipulator......................... 225
the realization of nanoinjectors and nanomanip-
ulators, novel nanotools for manipulation and 7.5 Summary ............................................. 228
analyzing subcellular organelles.
References .................................................. 228
Part A 7
Electron beams (EBs) and focused ion beam (FIBs) ing EB-CVD [7.3]. Blauner et al. demonstrated pillars
have been used to fabricate various two-dimensional and walls with high aspect ratios achieved using FIB-
nanostructure devices such as single-electron transis- CVD [7.4].
tors and metal–oxide–semiconductor (MOS) transistors The deposition rate of FIB-CVD is much higher
with nanometer gate lengths. Ten-nanometer structures than that of EB-CVD due to factors such as the differ-
can be formed by using a commercially available EB ence in mass between an electron and an ion. Further-
or FIB system with 5–10 nm-diameter beams and high- more, the smaller penetration depth of ions compared
resolution resist [7.1]. Two-dimensional nanostructure with electrons makes it easier to create complicated
fabrication is therefore already an established process. three-dimensional nanostructures. For example, when
There are various approaches to three-dimensional fab- we attempt to make a coil nanostructure with line width
rication using a laser, an EB, or a FIB to perform of 100 nm, 10–50 keV electrons pass through the ring
chemical vapor deposition (CVD). FIB- and EB-CVD of the coil and reach the substrate because of the large
are superior to laser-CVD [7.2] in terms of spatial reso- range of electrons (at least a few microns), which makes
lution and beam-scan control. Koops et al. demonstrated it difficult to create a coil nanostructure using EB-CVD.
some applications such as an atomic force microscopy On the other hand, since the range of ions is a few
(AFM) tip and a field emitter that were realized us- tens of nanometers or less, the ions are deposited in-
212 Part A Nanostructures, Micro-/Nanofabrication and Materials
side the ring. Up to now, the realization of complicated Therefore, this chapter reports on complicated three-
nanostructures using FIB-CVD has not been reported. dimensional nanostructure fabrication using FIB-CVD.
(200 nm thick) was deposited on a silicon substrate by Beam-induced chemical vapor deposition (CVD) is
30 keV Ga+ FIB using phenanthrene precursor gas. The widely used in the electrical device industry for repair of
chips and masks. This type of deposition is mainly done
on two-dimensional (2-D) pattern features, but it can
Primary ion also be used to fabricate a three-dimensional (3-D) ob-
Ga+ (30 keV)
ject. Koops et al. demonstrated nanoscale 3-D structure
Phenanthrene (C14H10) construction [7.3] by applying electron-beam-induced
Dissociation amorphous carbon deposition onto a micro vacuum
of adsorbed molecules
3 tube. However, focused ion beam (FIB)-induced CVD
2
1 seems to have many advantages for the fabrication of
Secondary
3-D nanostructures [7.4–6]. The key issue to realizing
electron such 3-D nanostructures is the short penetration depth
of the ions (a few tens of nm) into the target mater-
ial, being much shorter than that of electrons (several
DLC hundreds of μm). This short penetration depth reduces
Si the dispersion area of the secondary electrons, and so
the deposition area is restricted to roughly several tens
Fig. 7.1 Fabrication process for three-dimensional nanostructure by of nanometers. A 3-D structure usually contains over-
FIB-CVD hang structures and hollows. Gradual position scanning
3-D Nanostructure Fabrication by FIB-CVD 7.1 Three-Dimensional Nanostructure Fabrication 213
a) Wine glass b) Coil c) Micro Colosseum Fig. 7.2 (a) Micro wine glass with
an external diameter of 2.75 μm and
a height of 12 μm. (b) Microcoil with
coil diameter of 0.6 μm, coil pitch of
0.7 μm, and line width of 0.08 μm.
(c) Micro Colosseum
1 μm 1 μm 1 μm
of the ion beam during the CVD process causes the The process of fabricating three-dimensional struc-
position of the growth region around the beam point tures by FIB-CVD is illustrated in Fig. 7.1 [7.7]. In
to shift. When the beam point reaches the edge of the FIB-CVD processes, the beam is scanned in digi-
wall, secondary electrons appear at the side of the wall tal mode. First, a pillar is formed on the substrate
and just below the top surface. The DLC then starts by fixing the beam position (position 1). After that,
to grow laterally; the width of the lateral growth is the beam position is moved to within a diameter of
also about 80 nm. Therefore, by combining the lateral the pillar (position 2) and then fixed until the de-
growth mode with rotating beam scanning, it is possible posited terrace thickness exceeds the range of the ions
to obtain 3-D structures with rotational symmetry, such (a few tens of nm). This process is repeated to make
as a wine glass. three-dimensional structures. The key point to mak-
ing three-dimensional structures is to adjust the beam
scan speed so that the ion beam remains within the de-
posited terrace, which means that the terrace thickness
always exceeds the range of the ions. Growth in the
x- and y-directions is controlled by both beam deflec-
tors. The growth in the z-direction is determined by the
deposition rate; that is, the height of the structure is pro-
portional to the irradiation time when the deposition rate
Part A 7.1
is constant.
We intend to open up a new field of microstructure
plastic art using FIB-CVD. To demonstrate the possi-
bilities of this field, a micro wine glass created on a Si
2.75 μm
5 μm
Fig. 7.3 Micro wine glass with an external diameter of
2.75 μm and a height of 12 μm on a human hair Fig. 7.4 Micro Leaning Tower of Pisa
214 Part A Nanostructures, Micro-/Nanofabrication and Materials
Establish
priority
3-D CAD model Slice data Voxel data
Side view
7 6 5 4 3 2 1
Blanking
data
Blanking
Scan data
Scan
3-D pattern
generator
of 0.4 pA. The results show that FIB-CVD is a highly beam must irradiate the correct positions in the cor-
promising technique for realizing parts of a microsys- rect order. If the ion beam irradiates a voxel located in
tem, although their mechanical performance must be mid-air without a support layer, the ions intended for
measured. the voxel will be deposited on the substrate. Therefore,
1 μm 1 μm
Fig. 7.6a,b Micro Starship Enterprise NCC-1701D, 8.8 μm long Fig. 7.7 T-4 bacteriophage
3-D Nanostructure Fabrication by FIB-CVD 7.2 Nanoelectromechanics 215
the sequence of irradiation is determined, as shown in ship Enterprise NCC-1701D (from the television series
Fig. 7.5. Star Trek), which was fabricated by FIB-CVD at
The scan data and blanking signal therefore include 10 ∼ 20 pA [7.8]. The nanospaceship is 8.8 μm long
the scan sequence, the dwell time, the interval time, and and was realized at about 1 : 100 000 000 scale on
the irradiation pitch. These parameters are calculated silicon substrate. The dwell time (t d ), interval time
from the beam diameter, xy-resolution, and z-resolution (ti ), irradiation pitch ( p), and total process time (tp ),
of fabrication. The z-resolution is proportional to the were 80 μs, 150 μs, 2.4 nm, and 2.5 h, respectively.
dwell time and inversely proportional to the square of The horizontal overhang structure was successfully
the irradiation pitch. The scan data are passed to the fabricated.
beam deflector of the FIB-CVD, as are the blanking Figure 7.7 shows a nano T4 bacteriophage, which
data. The blanking signal controls the dwell time and is an artificial version of the virus fabricated by FIB-
interval time of the ion beam. CVD on silicon surface. The size of the artificial nano
Figure 7.6 shows a 3-D CAD model and an T4 bacteriophage is about ten times that of the real
scanning ion microscope (SIM) image of the star- virus.
7.2 Nanoelectromechanics
7.2.1 Measuring Young’s Modulus modulus, determined using resonant vibration and the
SEM monitoring technique [7.10, 11].
An evaluation of the mechanical characteristics of such The system setup for monitoring mechanical vi-
nanostructures is needed for material physics. Buks and bration is shown in Fig. 7.8b. There were two ways
Roukes reported a simple but useful technique [7.9] for of measuring the pillar vibrations. One is active mea-
measuring the resonant frequencies of nanoscale ob- surement, where the mechanical vibration is induced
jects using a scanning electron microscope (SEM). The by a thin piezoelectric device, 300 μm thick and 3 mm
secondary electron detector in the SEM can detect fre- square. The piezo device was bonded to the sidewall of
quencies up to around 4 MHz, so the sample vibration the SEM’s sample holder with silver paste. The sam-
is measured as the oscillatory output signal of the de- ple holder was designed to observe cross sections in
tector. Buks and Roukes used this technique to evaluate the SEM (S5000, Hitachi) system. Therefore, the pil-
the Casimir attractive force between two parallel beams lar’s vibration was observed as a side-view image, as
fabricated on a nanoscale. We evaluated the mechanical shown in Fig. 7.8a. The range of vibration frequencies
characteristics of DLC pillars in terms of the Young’s involved was 10 kHz up to 2 MHz, which is much faster
Part A 7.2
a) b)
Electron
beam
Secondary
electron
detector
Oscilloscope
Sample
holder
than the SEM raster scanning speed. Thus the resonant irradiation. The relationship of the resonant frequency
vibrations of the pillars can be taken as the trace of the to the Young’s modulus, which depends on the ratio of
pillar’s vibration in the SEM image. The resonant fre- the pillar diameter to the squared length, is summarized
quency and amplitude were controlled by adjusting the in Fig. 7.9. All of the pillars evaluated in this figure were
power of the driving oscillator. fabricated using the SMI9200 FIB system under rapid
The other way to measure pillar vibrations is pas- growth conditions. Typical growth rates were about
sive measurement using a spectrum analyzer (Agilent 3–5 μm/min for the 100 nm-diameter and 240 μm-wide
4395A), where most of the vibration seemed to de- pillars, and 0.9 μm/min for the 480 nm-wide pillars.
rive from environmental noise from rotary pumps and When calculating the data shown in Fig. 7.9, we as-
air conditioners. Some parts of the vibration result sumed that the density of the DLC pillars was about
from spontaneous vibration associated with thermal ex- 2.3 g/cm3 , which is almost identical to that of graphite
citations [7.9]. Because of the excitation and residual and quartz. The slope of the line in Fig. 7.9 indicates
noise, the pillars on the SEM sample holder always the Young’s modulus for each pillar. The Young’s mod-
vibrated at a fundamental frequency, even if noise iso- uli of the pillars were distributed over a range from 65
lation is enforced on the SEM system. The amplitude of to 140 GPa, which is almost identical to that of nor-
these spontaneous vibrations was on the order of a few mal metals. Wider pillars tended to have larger Young’s
nanometers at the top of the pillar, and high-resolution moduli.
SEM can easily detect it at a magnification of 300 000. We found that the stiffness increases significantly as
We arranged several pillars with varying diameters the local gas pressure decreases, as shown in Fig. 7.10.
and lengths. The DLC pillars with the smallest diameter While the absolute value of the local gas pressure at the
of 80 nm were grown using point irradiation. While we beam point is very difficult to determine, we found that
used two FIB systems for pillar fabrication, slight dif- the growth rate can be a useful parameter for describing
ferences in the beam diameters of the two systems did the dependence of Young’s modulus on pressure. All
not affect the diameters of the pillars. Larger-diameter data points indicated in Fig. 7.10 were obtained from
pillars were fabricated using an area-limited raster scan pillars grown using point irradiation. Therefore, the pil-
mode. Raster scanning of a 160 nm2 region produced lar diameters did vary slightly from 100 nm but not by
a pillar with a cross section of about 240 nm2 , and more than 5%. A relatively low gas pressure, with good
a 400 nm2 scan resulted in a pillar with a cross sec- uniformity, was obtained by using a single gas nozzle
tion of 480 nm2 . A typical SEM image taken during and gas reflector. We used a cleaved side-wall of an Si
resonance is shown in Fig. 7.8a. The FIB-CVD pillars tip as the gas reflector, which was placed 10–50 μm
seemed very durable against mechanical vibration. This away from the beam point so as to face the gas noz-
kind of measurement usually requires at least 30 min, zle. The growth rate was controlled by changing the
Part A 7.2
including spectrum analysis and photo recording, but distance to the wall. While there is a large distribu-
the pillars still survived without any change in reso-
nance characteristics. This durability of the DLC pillars
Resonant frequency (kHz)
should be useful in nanomechanical applications.
The resonant frequency f of the pillar is defined 4000
∅ = 0.1 μm
by (7.1) for a pillar with a square cross section, and (7.2) ∅ = 0.24 μm
for a circular cross section ∅ = 0.48 μm
3000
aβ 2 E E = 100 GPa
f square = , (7.1) E = 140 GPa
2π L 2 12ρ 2000
aβ 2 E
f circular = , (7.2)
1000
2π L 2 16ρ
where a is the width of the square pillar or the diameter E = 650 GPa
of the circular-shaped pillar, L is the length of the pil- 0
0 1 2 3 4
lar, ρ is the density, and E is the Young’s modulus. The
a/L2 (× 10–3)
coefficient β defines the resonant mode; β = 1.875 for
the fundamental mode. We used (7.1) for pillars 240 and Fig. 7.9 Dependence of resonant frequency on pillar
480 nm wide, and (7.2) for pillars grown by point-beam length
3-D Nanostructure Fabrication by FIB-CVD 7.2 Nanoelectromechanics 217
Part A 7.2
reaches that of carbon nanotubes and natural diamond
crystals. We think that this high Young’s modulus is due upon introducing C14 H10 and W(CO)6 as a source gas
to surface modification caused by the direct ion impact. was 1 × 10−4 and 1.5 × 10−3 Pa, respectively. If a Ga+
Ga+ FIB
Beam scanning direction
Faster
Source gas 3-D nanowiring
growth direction
ion beam is irradiated onto the substrate, C14 H10 or the beam at a certain fixed speed in a plane. How-
W(CO)6 molecules adsorbed on the substrate surface ever, if the beam scanning speed is faster than the
are decomposed, and carbon (C) is mainly deposited nanowiring growth speed, it grows downward or drops;
onto the surface of the substrate. The direction of depo- conversely, if the scanning speed is too slow, the de-
sition growth can be controlling through the scanning position grows slanting upward. Therefore, it is very
direction of the beam. The material deposited using important to control the beam scanning speed care-
C14 H10 gas was diamond-like carbon, as confirmed by fully when growing a nanowire horizontally. It turns
Raman spectra, and it had a very large Young’s modulus out that the optimal beam scanning speed to realize
of 600 GPa [7.7, 10]. a nanowire growing horizontally, using two C14 H10 gas
After the two walls shown in Fig. 7.11 were formed, guns, was about 190 nm/s. The expected pattern resolu-
free-space nanowiring was performed by adjusting the tion archived using FIB-CVD is around 80 nm, because
beam scanning speed. The ion beam used was a 30 kV both the primary Ga+ ion and secondary-electron scat-
Ga+ FIB, and the irradiation current was 0.8–2.3 pA. tering occur over distances of around 20 nm [7.10, 13].
The x- and y-scanning directions and the beam scan- Figures 7.12 and 7.13 show examples of free-space
ning speed were controlled by the CPG. The height in nanowiring fabricated by FIB-CVD and CPG. All of
the z-direction was proportional to the irradiation time. the structures shown were fabricated using C14 H10 as
Deposition is made to occur horizontally by scanning a precursor gas.
Figure 7.12a shows nanobridge free-space wiring.
a) The growth time was 1.8 min and the wiring width was
80 nm. Figure 7.12b shows free-space nanowires with
parallel resistances. The growth time was 2.8 min, and
the wiring width was also 80 nm.
1 μm Figure 7.13a shows free-space nanowiring grown
80 nm
in 16 directions from the center. Figure 7.13b shows
a scanning ion microscope (SIM) image of an induc-
b) tor (L), a resistor (R), and a capacitor (C) in a parallel
circuit structure with free-space nanowiring. A coiled
structure was fabricated by circle-scanning of the Ga+
FIB. The growth times of the L, R, and C structures
were about 6, 2, and 12 min, and all the nanowiring is
about 110 nm wide. From these structures, one can see
80 nm 1 μm that it is possible to fabricate nanowiring at an arbitrary
Part A 7.2
Fig. 7.12 (a) DLC free-space wiring with a bridge shape. position using FIB-CVD and CPG. These results also
(b) DLC free-space wiring with parallel resistances indicate that various circuit structures can be formed by
combining L, C, and R.
The free-space wiring structures were observed us-
a) b)
ing 200 keV TEM. The analyzed area was 20 nm in
diameter. Figure 7.14a,b shows TEM images of DLC
free-space wiring and a pillar. It became clear from
these energy-dispersive x-ray (EDX) measurements that
the dark part (A) of Fig. 7.14a corresponds to the Ga
core, while the outside part (B) of Fig. 7.14a corre-
sponds to amorphous carbon. This free-space wiring
therefore consists of amorphous carbon with a Ga core.
The center position of the Ga core is actually located
below the center of the wiring. However, in the case of
the DLC pillar, the Ga core is located at the center of the
1 μm 1 μm
pillar. To investigate the difference between these core
Fig. 7.13 (a) Radial DLC free-space wiring grown in 16 directions positions, the Ga core distribution in free-space wiring
from the center. (b) Scanning ion microscope (SIM) micrograph of was observed in detail by TEM. The center position of
inductance (L), resistance (R), and capacitor (C) structure the Ga core was about 70 nm from the top, which was
3-D Nanostructure Fabrication by FIB-CVD 7.2 Nanoelectromechanics 219
Part A 7.2
gas to create a gas mixture containing a metal in or- (e)
ρ = 4 × 10–4 Ω cm
der to obtain lower electrical resistivity. Figure 7.15b–d C = 22%
corresponds to increasing W(CO)6 contents in the gas Ga = 25% (b)
W = 53% ρ = 16 Ω cm
mixture. The W(CO)6 content rate was controlled by C = 85%
100 (d) Ga = 13%
the sublimation temperature of the C14 H10 gas. As the ρ = 2 × 10–2 Ω cm
W = 2%
W(CO)6 content was increased, the nanowiring growth C = 74 %
Ga = 19 %
time and width become longer, being 195 s and 120 nm W = 7%
for Fig. 7.15b, 237 s and 130 nm for Fig. 7.15c, and
10
296 s and 140 nm for Fig. 7.15d. Finally, we tried to fab-
(c)
ricate free-space nanowiring using only W(CO)6 , but ρ = 4 × 10–2 Ω cm
(a)
did not obtain continuous wiring, because the deposi- C = 78 %
ρ = 100 Ω cm
Ga = 18 %
tion rate for a source gas of just W(CO)6 was very slow. C = 90 %
1 W = 4%
Ga = 10 %
The electrical resistivity (Fig. 7.15a) for nano- W = 0%
wiring fabricated using only C14 H10 source gas was
1 × 102 Ω cm. The elemental contents were 90% C and 10–4 10–3 10–2 10–1 100 101
Voltage (V)
10% Ga, as measured using a SEM-EDX spot beam.
The I –V curves in Fig. 7.15b–d correspond to increas- Fig. 7.15 Electrical resistivity measurement for nanowiring. The
ing W(CO)6 content in the gas mixture. As the W(CO)6 electrical resistivity ρ was calculated from the I –V curve. Elemen-
content increases, the electrical resistivity decreases, tal contents of C, Ga, and W were measured by SEM-EDX
220 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) Intensity (au)
3
2.5
1.5
1
Part A 7.2
0.5
0 50 100 150 200 250 300 350
2.5
1.5
0.5
0 50 100 150 200 250 300 350
Part A 7.2
ten nanomechanical switches. The threshold voltage pulled using a micropipette puller (PC-10: Narishige
was around 20 V in each case. The switching function Co.). The dimensions of the glass capillary were 90 mm
was confirmed by performing on/off operations at an ap- in length and 1 mm in diameter. Using this process,
40
50 Fig. 7.19 (a) I –V curve for the
20
nanomechanical switch. (b) Pulsed
0 0 current to on/off operation for the
0 5 10 15 20 25 30 0 2 4 6 8
nanomechanical switch at an applied
Voltage (V) Time (s)
voltage of 30 V
222 Part A Nanostructures, Micro-/Nanofabrication and Materials
Glass capillary III. 3-D structure fabrication by FIB-CVD a) SIM image b) Moving principle
Ga+ FIB
I. Pulling
Gas nozzle
Glass capillary
1 μm
Au
(electrode)
Fig. 7.21a,b Laminated pleats-type electrostatic actuator.
Fig. 7.20 Fabrication process of 3-D nanoelectrostatic actuators (a) SIM image of a laminated pleats-type electrostatic ac-
tuator fabricated on the tip of a Au-coated glass capillary.
we obtained a glass capillary tip with a 1 μm diam- (b) Illustration of moving principle of the actuator
eter. Next, we coated the glass capillary surface with
Au by direct-current (DC) sputtering. The Au thickness A laminated pleats-type electrostatic actuator was
was ≈ 30 nm. This Au coating serves as the electrode fabricated by FIB-CVD. Figure 7.21a shows an SIM
that controls the actuator and manipulator. Then, the image of a laminated pleats-type electrostatic actuator
3-D nanoelectrostatic actuators and manipulators were fabricated at 7 pA and 60 min exposure time. Fig-
fabricated by FIB-CVD. This process was carried out ure 7.21b shows the principle behind the movement of
in a commercially available FIB system (SIM9200: SII this actuator. The driving force is the repulsive force
NanoTechnology Inc.) with a Ga+ ion beam operating due to the accumulation of electric charge. This elec-
at 30 keV. FIB-CVD was carried out using a phenan- tric charge can be stored in the pleats structures of the
threne (C14 H10 ) precursor as the source material. The actuator by applying a voltage across the glass capil-
beam diameter was about 7 nm. The inner diameter of lary. The pillar structure of this actuator bends due to
each nozzle was 0.3 mm. The phenanthrene gas pres- charge repulsion, as shown in Fig. 7.21b. Figure 7.22
sure during growth was typically 5 × 10−5 Pa in the shows the dependence of the bending distance on the
specimen chamber. The Ga+ ion beam was controlled
Part A 7.2
0.6
0.4 a
1 μm
0.2
0 Fig. 7.23a,b Coil-type electrostatic actuator. (a) SIM im-
0 250 500 750 1000 1250 1500 age of a coil-type electrostatic actuator fabricated on the
Voltage (V)
tip of a Au-coated glass capillary. (b) Illustration of moving
Fig. 7.22 Dependence of bending distance on applied voltage principle for the actuator
3-D Nanostructure Fabrication by FIB-CVD 7.3 Nanooptics: Brilliant Blue Observation from a Morpho Butterfly Scale Quasistructure 223
Part A 7.3
(SEM) (Fig. 7.25a), we found three-dimensional (3-D)
nanostructures 2 μm in height, 0.7 μm in width, and
with a 0.22 μm grating pitch on the scales. These nano-
structures cause a similar optical phenomenon to the
iridescence produced by a jewel beetle.
We duplicated the Morpho butterfly scale qua-
sistructure with a commercially available FIB system
(SMI9200: SII Nanotechnology Inc.) using a Ga+ ion
beam operating at 30 kV [7.18]. The beam diameter was
about 7 nm at 0.4 pA. The FIB-CVD was performed
using phenanthrene (C14 H10 ) as a precursor.
In this experiment, we used a computer-controlled 1 μm
pattern generator, which converted 3-D computer-aided
design (CAD) data into a scanning signal, which was Fig. 7.25a,b Morpho butterfly scales. (a) Optical microscope image
passed to an FIB scanning apparatus in order to fabri- showing top view of Morpho butterfly. SEM image showing a cross-
cate a 3-D mold [7.8]. The scattering range of the Ga sectional view of Morpho butterfly scales. (b) SIM image showing
primary ions is about 20 nm, and the range of the sec- inclined view of Morpho butterfly scale quasistructure fabricated by
ondary electrons induced by the Ga ion beam is about FIB-CVD
224 Part A Nanostructures, Micro-/Nanofabrication and Materials
pho butterfly scale quasistructure fabricated using istics as real Morpho butterfly scales.
7.4 Nanobiology
7.4.1 Nanoinjector ported that diamond-like carbon (DLC) pillars grown
by FIB-CVD with a precursor of phenanthrene vapor
Three-dimensional nanostructures on a glass capillary have very large Young’s moduli, exceeding 600 GPa,
have a number of useful applications, such as manipu- which potentially makes them useful for various appli-
lators and sensors in various microstructures. We have cations [7.10]. These characteristics are applicable to
demonstrated the fabrication of a nozzle nanostructure the fabrication of various biological devices.
on a glass capillary for a bioinjector using 30 keV Ga+ In one experiment, nozzle nanostructure fabrication
focused ion beam assisted deposition with a precur- for biological nanoinjector research was studied. The tip
sor of phenanthrene vapor and etching [7.19]. It has diameters of conventional bioinjectors are greater than
been demonstrated that nozzle nanostructures of vari- 100 nm and the tip shapes cannot be controlled. A bio-
ous shapes and sizes can be successfully fabricated. An nanoinjector with various nanostructures on the top of
inner tip diameter of 30 nm on a glass capillary and a tip a glass capillary has the following potential applications
shape with an inclined angle have been realized. We re- (shown in Fig. 7.27):
3-D Nanostructure Fabrication by FIB-CVD 7.4 Nanobiology 225
Manipulator
Conventionally, the tip shape of a microinjector
made by pulling a glass capillary, and which is used
as an injector into a cell, is controlled by applying me-
chanical grinding (or not). However, the reliability of
this technique for controlling tip shape is very poor and
Cell requires experienced workers.
A bionanoinjector tip was fabricated on a glass cap-
illary by FIB-CVD, as shown in Fig. 7.28a–c. First,
FIB etching made the tip surface of the glass capil-
lary smooth. Then, a nozzle structure was fabricated at
Sensor
the tip by FIB-CVD. Figure 7.28a shows the surface of
a chip smoothed at 120 pA and after 30 s exposure time
Injector by FIB etching, with inner hole diameter of 870 nm.
A nozzle structure fabricated by FIB-CVD with inner
Fig. 7.27 Potential uses for a bionanoinjector hole diameter of 220 nm is shown in Fig. 7.28b. Fig-
ure 7.28c shows a cross section of Fig. 7.28b. These
1. Injection of various reagents into a specific or- results demonstrate that a bionanoinjector could be suc-
ganelle in a cell cessfully fabricated by 3-D nanostructure fabrication
2. Selective manipulation of a specific organelle out- using FIB-CVD. The bionanoinjector was used to in-
side of a cell by using the nanoinjector as an ject dye into a egg cell (Ciona intestinalis) as shown in
aspirator Fig. 7.29.
3. Reducing the mechanical stress produced when op-
erating in the cell by controlling the shape and size 7.4.2 Nanomanipulator
of the bionanoinjector
4. Measurement of the electric potential of a cell, an An electrostatic 3-D nanomanipulator that can manipu-
organelle, and an ion channel exiting on a mem- late nanoparts and operate on cells has been developed
brane, by fabricating an electrode by FIB-CVD. This 3-D nanomanipulator has four fin-
gers so that it can manipulate a variety of shapes.
Thus far, 3-D nanostructure fabrications on a glass To move the nanomanipulator, electric charge is ac-
capillary have not been reported. We present nozzle cumulated in the structure by applying voltage to the
nanostructure fabrication on a glass capillary by FIB- four-fingered structure, and electric charge repulsion
CVD and etching in order to confirm the possibility of causes them to move. Furthermore, we succeeded in
Part A 7.4
bionanoinjector fabrication. catching a microsphere (made from polystyrene latex)
The nozzle structures of the nanoinjector were fab- with a diameter of 1 μm using this 3-D nanomanipulator
ricated using a function generator (Wave Factory: NF with four fingers [7.20].
Electronic Instruments, Yokohama, Japan). Conven- The glass capillary (GD-1; Narishige Co.) was
tional microinjectors are fabricated by pulling a glass pulled using a micropipette puller (PC-10; Narishige
capillary (GD-1: Narishige Co.) using a micropipette Co.). A tip diameter of about 1.0 μm could be obtained
puller (PC-10: Narishige Co.). The glass capillary was using this process. Then, the glass capillary surface was
90 mm in length and 1 mm in diameter. coated with Au in order to fabricate an electrode for
Tip
ip of
bio-nano
injector Fig. 7.28a–c SIM images of a bio-
nanoinjector fabricated on a glass
Glass capillary by FIB-CVD. (a) Before
capillary
1 μm 1 μm 1 μm
FIB-CVD, (b) after FIB-CVD, and
(c) cross section of (b)
226 Part A Nanostructures, Micro-/Nanofabrication and Materials
Glass
capillary
Optical
microscope
3-D nanomanipulator
Nano-
injector Glass capillary
Cell; 100 μm
Fig. 7.29 Injection into an egg cell (Ciona intestinalis) us-
ing a bionanoinjector
x
nanomanipulator control. The thickness of the Au coat- y
ing was about 30 nm. Finally, a 3-D nanomanipulator
structure with four fingers (Fig. 7.30) was fabricated Microsphere z
(polystyrene latex with a diameter of 1 μm)
by FIB-CVD on the tip of the glass capillary with an
electrode. Fig. 7.31 Illustration of 1 μm polystyrene microsphere ma-
Microsphere (a polystyrene latex ball with a diam- nipulation by using a 3-D electrostatic nanomanipulator
eter of 1 μm) manipulation was carried out using the with four fingers
3-D nanomanipulator with four fingers. An illustration
of this manipulation experiment is shown in Fig. 7.31.
By connecting the manipulator fabricated by FIB-CVD
to a commercial manipulator (MHW-3; Narishige Co.),
the direction of movement along the x-, y-, and z-axis
could be controlled. The microsphere target was fixed to
the side of a glass capillary, and the manipulation was
observed from the top with an optical microscope.
The optical microscope image of Fig. 7.32 shows 10 μm
the situation during manipulation. First, the 3-D
Fig. 7.32 In situ observation of 1 μm polystyrene micro-
nanomanipulator was made to approach the micro-
sphere manipulation using a 3-D electrostatic nanomanip-
sphere; no voltage was applied. Next, the four fingers
ulator with four fingers
Part A 7.4
a) b)
1 μm
Fig. 7.30a,b SIM image of the 3-D electrostatic nanomanipulator that the action of catching the microsphere occurs due
with four fingers before manipulation. (a) Side view, (b) top view to the elastic force of the manipulator’s structure. We
3-D Nanostructure Fabrication by FIB-CVD 7.4 Nanobiology 227
Part A 7.4
beam, and therefore to fabricate the nanonet structure spheres. This proves that we successfully captured the
on the glass capillary. microspheres.
The flexibility and practicality of the nanonet is en-
hanced by fabricating it on a glass capillary, since this
is used in many fields, including biology and medicine.
The FIB-CVD deposition time was about 40 min at
a beam current of 7 pA. The diameter of the ring used
to hang the net was about 7 μm, and the width of the net
was about 300 nm.
We performed an experiment under an optical mi-
croscope in which polystyrene microspheres were cap-
tured with the nanonet. Figure 7.34 shows a schematic
of the experimental apparatus. Polystyrene micro- 1 μm
spheres with a diameter of 2 μm were dispersed in
distilled water to simulate subcellular organelles. The Fig. 7.35 SIM image of nanonet holding three micro-
x-, y-, and z-axis movements of the FIB-CVD nanonet spheres after capture
228 Part A Nanostructures, Micro-/Nanofabrication and Materials
7.5 Summary
Three-dimensional nanostructure fabrication using sirable for various applications. A nanoelectrostatic
30 keV Ga+ FIB-CVD and a phenanthrene (C14 H10 ) actuator and 0.1 μm nanowiring were fabricated and
source as a precursor has been demonstrated. The evaluated as parts of nanomechanical system. Fur-
film deposited on a silicon substrate was characterized thermore, a nanoinjector and nanomanipulator were
using a transmission microscope and Raman spec- fabricated as novel nanotools for manipulation and
tra. This characterization indicated that the deposited analysis of subcellular organelles. These results demon-
film is diamond-like amorphous carbon, which has strate that FIB-CVD is one of the key technologies
attracted attention due to its hardness, chemical in- needed to make 3-D nanodevices that can be used
ertness, and optical transparency. Its large Young’s in the field of electronics, mechanics, optics, and
modulus, which exceeds 600 GPa, makes it highly de- biology.
References
7.1 S. Matsui: Nanostructure fabrication using electron 7.11 M. Ishida, J. Fujita, Y. Ochiai: Density estimation for
beam and its application to nanometer devices, amorphous carbon nanopillars grown by focused ion
Proc. IEEE 85, 629 (1997) beam assisted chemical vapor deposition, J. Vac. Sci.
7.2 O. Lehmann, F. Foulon, M. Stuke: Surface and three- Technol. B 20, 2784 (2002)
dimensional processing by laser chemical vapor 7.12 T. Morita, R. Kometani, K. Watanabe, K. Kanda,
deposition, NATO ASI Ser. Appl. Sci. 265, 91 (1994) Y. Haruyama, T. Hoshino, K. Kondo, T. Kaito,
7.3 H.W. Koops, J. Kretz, M. Rudolph, M. Weber, T. Ichihashi, J. Fujita, M. Ishida, Y. Ochiai,
G. Dahm, K.L. Lee: Characterization and appli- T. Tajima, S. Matsui: Free-space-wiring fabrica-
cation of materials grown by electron-beam- tion in nano-space by focused-ion-beam chemical
induced deposition, Jpn. J. Appl. Phys. 33, 7099 vapor deposition, J. Vac. Sci. Technol. B 21, 2737
(1994) (2003)
7.4 A. Wagner, J.P. Levin, J.L. Mauer, P.G. Blauner, 7.13 J. Fujita, M. Ishida, Y. Ochiai, T. Ichihashi, T. Kaito,
S.J. Kirch, P. Long: X-ray mask repair with focused S. Matsui: Graphitization of Fe-doped amorphous
ion beams, J. Vac. Sci. Technol. B 8, 1557 (1990) carbon pillars grown by focused ion-beam-induced
7.5 I. Utke, P. Hoffmann, B. Dwir, K. Leifer, E. Kapon, chemical-vapor deposition, J. Vac. Sci. Technol. B
P. Doppelt: Focused electron beam induced de- 20, 2686 (2002)
position of gold, J. Vac. Sci. Technol. B 18, 3168 7.14 K. Nakamatsu, K. Yamamoto, T. Hirayama, S. Matsui:
(2000) Fabrication of fine electron biprism filament by free-
7.6 A.J. DeMarco, J. Melngailis: Lateral growth of fo- space-nanowiring technique of focused-ion-beam
cused ion beam deposited platinum for stencil mask + chemical vapor deposition for accurate off-axis
Part A 7
repair, J. Vac. Sci. Technol. B 17, 3154 (1999) electron holography, Appl. Phys. Express 1, 117004
7.7 S. Matsui, T. Kaito, J. Fujita, M. Komuro, K. Kanda, (2008)
Y. Haruyama: Three-dimensional nanostructure 7.15 T. Morita, K. Nakamatsu, K. Kanda, Y. Haruyama,
fabrication by focused-ion-beam chemical vapor K. Kondo, T. Hoshino, T. Kaito, J. Fujita, T. Ichihashi,
deposition, J. Vac. Sci. Technol. B 18, 3181 (2000) M. Ishida, Y. Ochiai, T. Tajima, S. Matsui: Nanome-
7.8 T. Hoshino, K. Watanabe, R. Kometani, T. Morita, chanical switch fabrication by focused-ion-beam
K. Kanda, Y. Haruyama, T. Kaito, J. Fujita, chemical vapor deposition, J. Vac. Sci. Technol. B
M. Ishida, Y. Ochiai, S. Matsui: Development 22, 3137 (2004)
of three-dimensional pattern-generating system 7.16 R. Kometani, T. Hoshino, K. Kondo, K. Kanda,
for focused-ion-beam chemical-vapor deposition, Y. Haruyama, T. Kaito, J. Fujita, M. Ishida, Y. Ochiai,
J. Vac. Sci. Technol. B 21, 2732 (2003) S. Matsui: Characteristics of nano-electrostatic ac-
7.9 E. Buks, M.L. Roukes: Stiction, adhesion energy, and tuator fabricated by focused ion beam chemical
the Casimir effect in micromechanical systems, Phys. vapor deposition, Jpn. J. Appl. Phys. 43, 7187
Rev. B 63, 033402 (2001) (2004)
7.10 J. Fujita, M. Ishida, T. Sakamoto, Y. Ochiai, 7.17 P. Vukusic, J.R. Sambles: Photonic structures in bi-
T. Kaito, S. Matsui: Observation and characteristics of ology, Nature 424, 852 (2003)
mechanical vibration in three-dimensional nano- 7.18 K. Watanabe, T. Hoshino, K. Kanda, Y. Haruyama,
structures and pillars grown by focused ion beam S. Matsui: Brilliant blue observation from a Morpho-
chemical vapor deposition, J. Vac. Sci. Technol. B 19, butterfly-scale quasi-structure, Jpn. J. Appl. Phys.
2834 (2001) 44, L48 (2005)
3-D Nanostructure Fabrication by FIB-CVD References 229
7.19 R. Kometani, T. Morita, K. Watanabe, K. Kanda, 7.21 S. Akita, Y. Nakayama, S. Mizooka, Y. Takano,
Y. Haruyama, T. Kaito, J. Fujita, M. Ishida, Y. Ochiai, T. Okawa, K.Y. Miyatake, S. Yamanaka, M. Tsuji,
S. Matsui: Nozzle-nanostructure fabrication on glass T. Nosaka: Nanotweezers consisting of carbon nano-
capillary by focused-ion-beam chemical vapor de- tubes operating in an atomic force microscope, Appl.
position and etching, Jpn. J. Appl. Phys. 42, 4107 Phys. Lett. 79, 1691 (2001)
(2003) 7.22 R. Kometani, T. Hoshino, K. Kanda, Y. Haruyama,
7.20 R. Kometani, T. Hoshino, K. Kondo, K. Kanda, T. Kaito, J. Fujita, M. Ishida, Y. Ochiai, S. Mat-
Y. Haruyama, T. Kaito, J. Fujita, M. Ishida, Y. Ochiai, sui: Three-dimensional high-performance nano-
S. Matsui: Performance of nanomanipulator fab- tools fabricated using focused-ion-beam chemical-
ricated on glass capillary by focused-ion-beam vapor-deposition, Nucl. Instrum. Methods Phys.
chemical vapor deposition, J. Vac. Sci. Technol. B Res. B 232, 362 (2005)
23, 298 (2005)
Part A 7
231
Introduction
8. Introduction to Micro-/Nanofabrication
Part A 8
Recent innovations in the area of micro/nanofabrication the nm to several hundred μm range. We will mainly fo-
have created a unique opportunity for manufacturing cus on the most important and widely used techniques
structures in the nm–mm range. The available six or- and will not discuss specialized methods. After a brief
ders of magnitude dimensional span can be used to introduction to basic microfabrication, we will discuss
fabricate novel electronic, optical, magnetic, mechan- MEMS fabrication techniques used to build microstruc-
ical, and chemical/biological devices with applications tures down to about 1 μm in dimensions. Following
ranging from sensors to computation and control. In this this, we will discuss several top-down and bottom-up
chapter, we will introduce major micro/nanofabrication nanofabrication methods not discussed in other chapters
techniques currently used to fabricate structures from of this Handbook.
232 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 8.1
e) Etch the oxide
• Mechanical structure
• Electrical isolation
Substrate
• Electrical connection
• Sensing or actuating
• Mask for etching and doping
f) Strip the photoresist
• Support or mold during deposition of other mater-
ials (sacrificial materials)
Substrate • Passivation
Most of the deposited thin films have properties dif-
Fig. 8.2a–f Schematic drawing of the photolithographic ferent from those of their corresponding bulk forms (for
steps with a positive PR example, metals shows higher resistivities in thin-film
234 Part A Nanostructures, Micro-/Nanofabrication and Materials
Quartz
Wafers
cassette
c) d)
Quartz tube
Part A 8.1
The penetration of impurities into the material follows characterize the internal stress of deposited thin films
a Gaussian distribution. After implantation, an anneal- for any specific equipment and deposition conditions.
ing process is needed to activate the impurities and The PECVD process is performed in plasma sys-
repair the damage in the crystal structure produced by tems such as that represented in Fig. 8.6. The use of
ion collisions. A drive-in process to redistribute the im- radiofrequency (RF) energy to create highly reactive
purities, done in a standard furnace such as those used species in the plasma allows for the use of lower tem-
for oxidation or diffusion, may be required as well. perature at the substrate (150–350 ◦ C). Parallel-plate
plasma reactors normally used in microfabrication can
Chemical Vapor Deposition and Epitaxy only process a limited number of wafers per batch. The
As its name suggests, chemical vapor deposition (CVD) wafers are positioned horizontally on top of the lower
includes all deposition techniques using the reaction of electrode so only one side gets deposited. Typical ma-
chemicals in gas phase to form the deposited thin film. terials deposited with PECVD include silicon oxide,
236 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 8.1
able to deposit more than one material simultane- frequent case) a thin film of the same metal, called
ously or sequentially. This capability is very useful the seed layer, is previously deposited on the surface.
to obtain alloys and multilayer films (e.g., multilayer Masking of the seed layer with a resist permits selective
magnetic recording heads are sputtered). For certain electroplating of the patterned areas. Figure 8.10 illus-
low-reactivity metals such as Au and Pt the previous trates a typical sequence of the steps required to obtain
deposition of a thin layer of another metal is needed isolated metal structures.
to improve adhesion. Ti and Cr are two frequently
used adhesion promoters. Stress in evaporated or sput- Pulsed Laser and Atomic Layer Deposition
tered layers is typically tensile. The deposition rates Pulsed laser and atomic layer deposition techniques
are much higher than for most CVD techniques. How- have attracted a considerable amount of attention re-
ever, due to stress accumulation and cracking, thickness cently. These two techniques offer several unique
beyond 2 μm is rarely deposited with these processes. advantages compared with other thin-film deposition
238 Part A Nanostructures, Micro-/Nanofabrication and Materials
plume
acted vapor is pumped out and a vapor dose of a second
Laser pulse
reactant is introduced. This reacts with the precursor on
the surface in a self-limiting fashion. Finally, the extra
unreacted vapor is pumped out and the cycle is repeated.
Target
Part A 8.1
roughness. This has been extensively used in fabricating
Wet Etching microfluidic components (mainly channels). Although
Historically, wet etching techniques preceded the dry Cr/Au is usually used as the masking layer, long etch
ones. These still constitute an important group of times require a more robust mask (bonded silicon has
etchants for micro/nanofabrication in spite of their been used for this purpose).
less frequent application in the VLSI processes. Wet Silicon anisotropic wet etch constitutes an impor-
etchants are by and large isotropic and show supe- tant technique in bulk micromachining. The three most
rior selectivity to the masking layer as compared with important silicon etchants in this category are potas-
various dry techniques. In addition, due to the lateral sium hydroxide (KOH), ethylenediamine pyrochate-
undercut, the minimum feature size achievable with chol (EDP), and tetramethyl ammonium hydroxide
wet etchants is limited to > 3 μm. Silicon dioxide is (TMAH). These are all anisotropic etchants which
commonly etched in dilute (6 : 1, 10 : 1, or 20 : 1 by attack silicon along preferred crystallographic direc-
240 Part A Nanostructures, Micro-/Nanofabrication and Materials
Dry Etching
Most of the dry etching techniques are plasma based.
They have several advantages when compared with
wet etching. These include smaller undercut (allowing
smaller lines to be patterned) and higher anisotropy (al-
lowing high-aspect-ratio vertical structures). However,
the selectivity of dry etching techniques is lower than
KOH that of wet etchants, and one must take into account the
finite etch rate of the masking materials. The three basic
dry etching techniques, namely high-pressure plasma
etching, reactive-ion etching (RIE), and ion milling, uti-
lize different mechanisms to obtain directionality.
n-epi p-substrate Counterelectrode
Ion milling is a purely physical process which
utilizes accelerated inert ions (e.g., Ar+ ) striking per-
Fig. 8.15 Electrochemical etch setup pendicular to the surface to remove the material
( p ≈ 10−4 –10−3 Torr) (Fig. 8.16a). The main charac-
regions are exposed to the solution an oxide passivation teristics of this technique are very low etch rates (in the
layer is formed and etching is stopped. This tech- order of a few nm/min) and poor selectivity (close to
nique can be used to fabricate single-crystalline silicon 1 : 1 for most materials); hence it is generally used to
membranes for pressure sensors and other mechanical etch very thin layers. In high-pressure (10−1 –5 Torr)
transducers. plasma etchers highly reactive species are created that
a)
Ion
Mask
Material atom
b)
Ion
Mask Material atom
Nonvolatile species
Volatile product
Part A 8.1
c)
Ion
Mask
Material atom
Activated material atom
Volatile product
Fig. 8.16a–c Simplified representation of etching mechanisms for: (a) ion milling, (b) high-pressure plasma etching, and
(c) RIE
242 Part A Nanostructures, Micro-/Nanofabrication and Materials
recipes [8.14]. For each chemistry the etch rate, di- ature; however, in order to increase the bond strength,
rectionality, and selectivity with respect to the mask a high-temperature (800–1200 ◦ C) anneal is usually re-
materials depend on parameters such as the flow rates quired. A major advantage of silicon fusion bonding is
of the gases entering the chamber, the working pressure, the thermal matching of the substrates.
and the RF power applied to the plasma.
Anodic Bonding
8.1.4 Substrate Bonding Silicon–glass anodic (electrostatic) bonding is another
major substrate joining technique which has been ex-
Substrate (wafer) bonding (silicon–silicon, silicon– tensively used for microsensor packaging and device
glass, and glass–glass) is among the most important fab- fabrication. The main advantage of this technique
rication techniques in microsystem technology [8.15, is its lower bonding temperature, which is around
16]. It is frequently used to fabricate complex 3-D struc- 300–400 ◦ C. Figure 8.18 shows the bonding setup.
tures both as a functional unit and as a part of the A glass wafer (usually Pyrex 7740 because of thermal
final microsystem package and encapsulation. The two expansion match to silicon) is placed on top of a sili-
most important bonding techniques are silicon–silicon con wafer and the sandwich is heated to 300–400 ◦ C.
fusion (or silicon direct bonding) and silicon–glass Subsequently, a voltage of ≈ 1000 V is applied to the
electrostatic (or anodic) bonding. In addition to these glass–silicon sandwich with the glass connected to the
techniques, several other alternative methods which uti- cathode. The bond starts immediately after the appli-
lize an intermediate layer (eutectic, adhesive, and glass cation of the voltage and spreads outward from the
frit) have also been investigated. All these techniques cathode contact point. The bond can be observed visu-
can be used to bond the substrates at the wafer level. In ally as a dark-grayish front which expands across the
this section we will only discuss wafer-level techniques wafer.
and will not treat device-level bonding methods (e.g., The bonding mechanism is as follows. During the
e-beam and laser welding). heating period, glass sodium ions move toward the cath-
ode and create a depletion layer at the silicon–glass
Silicon Direct Bonding interface. A strong electrostatic force is therefore cre-
Direct silicon or fusion bonding is used in the fab- ated at the interface, which pulls the substrates into
rication of micromechanical devices and silicon-on- intimate contact. The exact chemical reaction respon-
insulator (SOI) substrates. Although it is mostly used to sible for anodic bonding is not yet clear, but covalent
bond two silicon wafers with or without an oxide layer, silicon–oxygen bonds at the interface seem to be re-
it has also been used to bond different semiconductors sponsible for the bond. Silicon–silicon anodic bonding
such as GaAs and InP [8.16]. One main requirement using sputtered or evaporated glass interlayer is also
for a successful bond is sufficient planarity (< 10 Å possible.
surface roughness and < 5 μm bow across a 4 inch
wafer) and cleanliness of the surfaces. In addition, ther- Bonding with Intermediate Layers
mal expansion mismatch also needs to be considered Various other wafer bonding techniques utilizing an
if bonding of two dissimilar materials is contemplated. intermediate layer have also been investigated [8.16].
The bonding procedure is as follows: the silicon or Among the most important ones are adhesive, eutectic,
oxide-coated silicon wafers are first thoroughly cleaned. and glass frit bonds. Adhesive bonding using a poly-
Part A 8.1
Subsequently the surfaces are hydrated (activated) in
HF or boiling nitric acid (Radio Corporation of Amer-
Pyrex
ica (RCA) clean also works). This renders the surfaces Cathode
hydrophilic by creating an abundance of hydroxyl ions.
+
Then the substrates are brought together in close prox- Depletion Na 1000 V
imity (starting from the center to avoid void formation). layer
The close approximation of the bonding surfaces al- Silicon
Anode
lows the short-range attractive van der Waals forces to
bring the surfaces into intimate contact on the atomic
scale. Following this step, hydrogen bonds between the
Hot plate
two hydroxyl-coated silicon wafers bond the substrates
together. These steps can be performed at room temper- Fig. 8.18 Glass–silicon anodic bonding setup
244 Part A Nanostructures, Micro-/Nanofabrication and Materials
mer (e.g., polyimides, epoxies, thermoplastic adhesives, must be removed from the silicon surface prior to the
and photoresists) in between the wafers has been used to deposition of the gold. In addition, all organic contam-
join different wafer substrates [8.17]. Complete curing inants on the gold surface must be removed (using UV
(in an oven or using dielectric heating) of the polymer light) prior to the bond. Pressure must also be applied
before or during the bonding process prevents subse- in order to achieve a better contact. Although eutec-
quent solvent outgassing and void formation. Although tic bonding can be accomplished at low temperatures,
reasonably high bonding strengths can be obtained, achieving uniformity over large areas has proven to be
these bonds are nonhermetic and unstable over a period challenging.
of time. Glass frit can also be used as an interlayer in sub-
In eutectic bonding process, gold-coated silicon strate bonding. In this technique, first a thin layer of
wafers are bonded together at temperatures greater than glass is deposited and preglazed. The glass-coated sub-
the silicon–gold eutectic point (363 ◦ C, 2.85% silicon strates are then brought into contact and the sandwich is
and 97.1% Au) [8.18]. This process can achieve high heated to above the glass melting temperature (typically
bonding strength and good stability at relatively low < 600 ◦ C). As for the eutectic process, pressure must be
temperatures. For good bond uniformity silicon dioxide applied for adequate contact [8.19].
Part A 8.2
compatibility and yield have to be carefully consid-
c ) p ++ etch stop without n-epi ered. Among the most popular techniques in this
category is postprocessing of CMOS integrated cir-
cuits by front-side etching in TMAH solutions. As was
mentioned previously, silicon-rich TMAH does not at-
Si p ++ tack aluminum and therefore can be used to undercut
microstructures in an already processed CMOS chip.
Figure 8.24 shows a schematic of such a process in
which a front-side wet etch and electrochemical etch
Fig. 8.20a–c Wet micromachining etch-stop techniques: stop are used to produce suspended beams. This tech-
(a) electrochemical with n-epi on p-substrate, (b) p++ etch nique has been extensively used to fabricate a variety
stop with n-epi, and (c) p++ etch stop without n-epi of microsensors (e.g., humidity, gas, chemical, and
246 Part A Nanostructures, Micro-/Nanofabrication and Materials
a)
Silicon
b)
p ++ p ++
Silicon
c)
Silicon
Fig. 8.23 SEM image of a microaccelerometer fabricated
using the dissolved wafer process (after [8.22])
d)
Silicon Suspension Electronics
beam
n-well
Glass
P-substrate
e)
Fig. 8.24 Suspended island created on a prefabricated
CMOS chip using front-side wet etch and electrochemical
etch stop
Fig. 8.22a–e Dissolved wafer process sequence: (a) KOH
etch, (b) deep B diffusion, (c) shallow B diffusion,
(d) silicon–glass anodic bond, and (e) release in EDP
Part A 8.2
fabricated CMOS chip is used as a mask to etch the
underlying oxide layers anisotropically all the way to
f) the silicon (CHF3 /O2 ) (Fig. 8.28b). This is followed by
an anisotropic silicon etch to create a recess in the sili-
con, which will be used in the final step to facilitate the
undercut and release (Fig. 8.28c). Finally, an isotropic
silicon etch is used to undercut and release the structures
Fig. 8.26a–f Cross section of the SCREAM process (Fig. 8.28d). Figure 8.29 shows an SEM photograph of
(a) silicon wafer, (b) anisotropic silicon etch, (c) confor- a comb-drive actuator fabricated using this technology.
mal passivation, (d) anisotropic etching of the passivation In addition to the methods described above, recent
(hence protecting the sidewall), (e) isotropic silicon etch, advancements in the development of deep reactive-ion
and (f) metal deposition etching of silicon (DRIE) have created new op-
248 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) Top a)
metal
layer SOI wafer
DRIE
b)
b)
c)
c)
d)
Part A 8.2
ials are phosphosilicate glass (PSG) and polysilicon,
respectively (low-temperature oxide, LTO, is also fre-
c) quently used as the sacrificial layer). However, there
are several other sacrificial/structural combinations
that have been used to create a variety of surface-
Silicon micromachined structures. Important design issues
related to the choice of the sacrificial layer are:
Two examples of commercially available surface- (poly-Si1−x Gex ) can be deposited at temperatures as
micromachined devices illustrate various successful low as 350 ◦ C (poly-Ge deposition temperature is
System Sacrificial layer Structural layer Structural layer etchant Sacrificial layer etchant
1 PSG or LTO Poly-Si RIE Wet or vapor HF
2 Photoresist, polyimide Metals (Al, Ni, Co, Ni-Fe) Various metal etchants Organic solvents, plasma O2
3 Poly-Si Nitride RIE KOH
4 PSG or LTO Poly-Ge H2 O2 or RCA1 Wet or vapor HF
5 PSG or LTO Poly-Si-Ge H2 O2 or RCA1 Wet or vapor HF
Introduction to Micro-/Nanofabrication 8.2 MEMS Fabrication Techniques 251
usually lower than that for poly-SiGe). Table 8.2 (TI aluminum micromirrors), silicone polymeric layers
summarizes important surface-micromachined sacrifi- (Analog Devices accelerometers), and siloxane self-
cial/structural combinations. assembled monolayers.
An important consideration in the design and pro-
cessing of surface-micromachined structures is the issue Surface-Micromachining Integration
of stiction [8.11, 37, 38]. This can happen during the with Active Electronics
release step if a wet etchant is used to remove the sac- Integration of surface-micromachined structures with
rificial layer or during the device lifetime. The reason on-chip circuitry can increase performance and simplify
for stiction during release is the surface tension of the packaging. However, issues related to process com-
liquid etchant, which can hold the microstructure down patibility and yield have to be carefully considered.
and cause stiction. This usually happens when the struc- The two most common techniques are MEMS-first and
ture is compliant and does not possess enough spring MEMS-last techniques. In the MEMS-last technique,
constant to overcome the surface tension force of the the integrated circuit is first fabricated and surface-
rinsing liquid (i. e., water). There are several ways one micromachined structures are subsequently built on top
can alleviate the release-related stiction problem. These of the silicon wafer. An aluminum structural layer with
include: a sacrificial photoresist layer is an attractive combina-
tion due to the low thermal budget of the process (TI
1. The use of dry or vapor phase etchant
micromirror array). However, in applications where me-
2. The use of solvents with lower surface tension
chanical properties of Al are not adequate, polysilicon
3. Geometrical modifications
structural material with an LTO or PSG sacrificial layer
4. CO2 critical drying
must be used. Due to the rather high deposition temper-
5. Freeze-drying
ature of polysilicon, this combination requires special
6. Self-assembled monolayer (SAM) or organic thin-
attention with regard to the thermal budget. For ex-
film surface modification
ample, aluminum metallization must be avoided and
The first technique prevents stiction by not using substituted with refractory metals such as tungsten. This
a wet etchant, although in the case of vapor-phase re- can only be achieved at the cost of greater process com-
lease, condensation is still possible and can cause some plexity and lower transistor performance.
stiction. The second method uses rinsing solvents such The MEMS-first technique alleviates these difficul-
as methanol with a lower surface tension than water. ties by fabricating the microstructures at the very be-
This is usually followed by rapid evaporation of the ginning of the process. However, if the microstructures
solvent on a hot-plate. However, this technique is not are processed first, they have to be buried in a sealed
optimum and many structures still stick. The third tech- trench to eliminate the interference of microstructures
nique is geometrical, providing dimples in the structural with subsequent CMOS processes. Figure 8.35 shows
layer in order to reduce the contact surface area and a cross section of a MEMS-first fabrication process
hence reduce the attractive force. The fourth and fifth developed at the Sandia National Laboratory [8.39].
techniques rely on phase change (in one case CO2 and The process starts with shallow anisotropic etching
the other butyl alcohol) which avoids the liquid phase
altogether by directly going to the gas phase. The last
CMOS device area Micromechanical device area
technique uses self-assembled monolayers or organic
Part A 8.2
thin films to coat the surfaces with a hydrophobic layer. Passivation
The stiction that occurs during the operating lifetime of Metal Polysilicon
the device (in-use stiction) is due to condensation of Structural
moisture on the surfaces, electrostatic charge accumu- polysilicon
lation, or direct chemical bonding. Surface passivation
using self-assembled monolayers or organic thin films
Epitaxial layer
can be used to reduce the surface energy and reduce
or eliminate capillary forces and direct chemical bond- Silicon substrate
ing. These organic coatings also reduce electrostatic
Sacrificial oxide Silicon nitride
forces if a thin layer is applied directly to the semicon-
ductor (without an intervening oxide layer). Commonly Fig. 8.35 Cross section of the Sandia MEMS-first inte-
used organic coatings include fluorinated fatty acids grated fabrication process
252 Part A Nanostructures, Micro-/Nanofabrication and Materials
of trenches in a silicon substrate to accommodate the to move these structures out of the silicon plane and
height of the polysilicon structures fabricated later on. into position. Figure 8.36 shows an SEM photograph
A silicon nitride layer is then deposited to provide iso- of a bar-code microscanner using a silicon optical mi-
lation at the bottom of the trenches. Next, several layers crobench with 3-D surface-micromachined structures.
of polysilicon and sacrificial oxide are deposited and
patterned in a standard surface-micromachining pro- 8.2.3 High-Aspect-Ratio Micromachining
cess. Subsequently, the trenches are completely filled
with sacrificial oxide and the wafers are planarized The bulk and surface micromachining technologies pre-
with chemical–mechanical polishing (this avoids com- sented in the previous sections fulfill the requirements
plication in the following lithographic steps). After an of a large group of applications. Certain applications,
annealing step, the trenches are sealed with a nitride however, require the fabrication of high-aspect-ratio
cap. At this point, a standard CMOS fabrication pro- structures that is not possible with the aforementioned
cess is performed. At the end of the CMOS process the technologies. In this section, we describe three tech-
nitride cap is etched and the buried structures released nologies, LIGA, HEXSIL, and HARPSS, capable of
by etching the sacrificial oxide. producing structures with vertical dimensions much
larger than their lateral dimensions by means of x-ray
Three-Dimensional Microstructures lithography (LIGA) and DRIE etching (HEXSIL and
Using Surface Micromachining HARPSS).
Three-dimensional surface microstructures can be fab-
ricated using surface micromachining. The fabrica- LIGA
tion of hinges for the vertical assembly of MEMS LIGA is a high-aspect-ratio micromachining process
was a major advance towards achieving 3-D mi- which relies on x-ray lithography and electroplating
crostructures [8.41]. Optical microsystems have greatly (in German: Lithographie, Galvanoformung, Abfor-
benefited from surface-micromachined 3-D structures. mung) [8.43, 44]. We already introduced the con-
These microstructures are used as passive or active com- cept of the plating-through-mask technique in Surface
ponents (micromirror, Fresnel lens, optical cavity, etc.) Micromachining (Fig. 8.10). With standard UV pho-
on a silicon optical bench (silicon microphotonics). An tolithography and photoresists, the maximum thickness
example is a Fresnel lens that has been surface micro- achievable is on the order of a few tens of microns and
machined in polysilicon and then erected using hinge the resulting metal structures show tapered walls. LIGA
structures and locked in place using micromachined is a technology based on the same plating-through-mask
tabs, thus liberating the structure from the horizontal idea but can be used to fabricate metal structures of
plane of the wafer [8.40, 42]. Various microactuators thickness ranging from a few microns to a few millime-
(e.g., comb drive, and vibromotors) have been used ters with almost vertical side-walls. This is achieved
using x-ray lithography and special photoresists. Due
to their short wavelength, x-rays are able to penetrate
Back support through a thick photoresist layer with no scattering and
define features with lateral dimensions down to 0.2 μm
Microhinge (aspect ratio > 100 : 1).
The photoresists used in LIGA should comply with
Part A 8.2
a) c) Electroplated metal
UV light
UV mask
Photoresist
Sacrificial
Seed Substrate layer
layer (with isolated layer) d)
b)
Part A 8.2
X-ray
X-ray mask
e) Fixed part Movable part
X-ray resist
Fig. 8.38a–e Sacrificial LIGA process: (a) UV lithography for sacrificial layer patterning, (b) x-ray lithography, (c) elec-
troplating, (d) structure releasing, and (e) top view of the movable structure
254 Part A Nanostructures, Micro-/Nanofabrication and Materials
HEXSIL
The second method for fabricating high-aspect-ratio
d) Etch structural layer from the surface structures, which is based on a template replica-
tion technology, is hexagonal honeycomb polysilicon
(HEXSIL) [8.49]. Figure 8.39 shows a simplified pro-
cess flow. A high-aspect-ratio template is first formed
in a silicon substrate using DRIE. Next, a sacrificial
multilayer is deposited to allow the final release of the
structures. The multilayer is composed of one or more
e) Etch sacrificial layer and pull the structure out
PSG nonconformal layers to provide fast etch release
(≈ 20 μm/min in 49% HF) alternated with conformal
layers of either oxide or nitride to provide enough
thickness for proper release of the structures. The to-
tal thickness of the sacrificial layer has to be larger than
the shrinkage or elongation of the structures caused by
Part A 8.2
f) Example of a
HEXSIL-fabricated structure
tios larger than 20 : 1 have been demonstrated. A major Fig. 8.40 SEM micrograph of an angular microactuator
problem of this alternative is the removal of the SU-8 fabricated using the HEXSIL process (after [8.47])
Introduction to Micro-/Nanofabrication 8.2 MEMS Fabrication Techniques 255
Nitride
Fig. 8.42 SEM photograph of a microgyroscope fabri-
cated using the HARPSS process (after [8.51])
Silicon
ductive surface is needed for the deposition to occur.
After deposition of structural materials a blanket etch
b) Poly 1 deposition and etch back, oxide patterning and
poly 2 deposition and patterning (poly-Si or nitride) or a mechanical lapping (nickel)
Poly 2 is performed to remove the excess materials from the
Poly 1
surface. Finally, a 49% HF with surfactant is used to dis-
solve the sacrificial layers. The process can be repeated
many times using the same template, thus consider-
ably lowering fabrication costs. Figure 8.40 shows an
SEM photograph of a microactuator fabricated using the
HEXSIL process.
Part A 8.2
done at the end by means of a directional silicon etch
followed by an isotropic etch. The small vertical gaps
and thick structures possible with this technology find
application in the fabrication of a variety of MEMS de-
vices, particularly inertial sensors [8.53] and RF beam
resonators [8.54]. Figure 8.41 shows the process flow
in a cross section of a single-crystal silicon beam res-
Fig. 8.41a–d HARPSS process flow: (a) nitride deposi- onator. The HARPSS process starts with deposition and
tion and patterning, DRIE etching and oxide deposition, patterning of a silicon nitride layer that will be used
(b) poly 1 deposition and etch back, oxide patterning, and to isolate the poly structure’s connection pads from the
poly 2 deposition and patterning, (c) DRIE etching, and substrate. High-aspect-ratio trenches (≈ 5 μm wide) are
(d) silicon isotropic etching then etched into the substrate using a DRIE etch. Next,
256 Part A Nanostructures, Micro-/Nanofabrication and Materials
a conformal oxide layer (LPCVD) is deposited. This Finally, the structures are released using a DRIE step
layer has two functions, to: followed by an isotropic silicon etch through a pho-
toresist mask that exposes only the areas of silicon
1. Protect the structures during the dry etch release
substrate surrounding the structures. It should be noted
2. Define the submicrometer gap between silicon and
that single-crystal silicon structures are not protected at
polysilicon structures
the bottom during the isotropic etch. This causes the
Following the oxide deposition, the trenches are com- single-crystal silicon structures to be etched vertically
pletely filled with LPCVD polysilicon. The polysilicon from the bottom, and thus be shorter than the polysili-
is etched back and the underlying oxide is patterned con structures. Figure 8.42 shows an SEM photograph
to provide anchor points for the structures. A second of a microgyroscope fabricated using the HARPSS
layer of polysilicon is then deposited and patterned. process.
techniques that are not covered in other chapters of this well and superlattice structures using epitaxial growth
Handbook. These include: is a mature and well-developed field and therefore will
not be discussed in this chapter. Instead, we will concen-
1. E-beam nanofabrication trate on quantum wire and dot nanostructure fabrication
2. Epitaxy and strain engineering using basic epitaxial techniques [8.61, 62].
3. Scanning-probe techniques
4. Self-assembly and template manufacturing Quantum Structure Nanofabrication
Using Epitaxy on Patterned Substrates
8.3.1 E-Beam Nanofabrication There have been several different approaches to the
fabrication of quantum wires and dots using epitaxial
In previous sections, we discussed several important layers. The most straightforward technique involves e-
lithography techniques used commonly in MEMS and beam lithography and etching of an epitaxial grown
microfabrication. These included various forms of UV layer (e.g., InGaAs on GaAs substrate) [8.63]. However,
(regular, deep, and extreme) and x-ray lithography. due to damage and/or contamination during lithography,
However, due to the lack of resolution (in case of this method is not very suitable for active device fabri-
the UV) or difficultly in manufacturing mask and cation (e.g., quantum dot lasers). Several other methods
radiation sources (x-ray), these techniques are not suit- involving regrowth of epitaxial layers over nonplanar
able for nm-scale fabrication. E-beam lithography is surfaces such as step-edge, cleaved-edge, and patterned
an alternative and attractive technique for fabricating substrate have been used to fabricate quantum wires
nanostructures [8.56]. It uses an electron beam to ex- and dots without the need for lithography and etch-
pose an electron-sensitive resist such as poly(methyl ing of the quantum confinement structure [8.62, 64].
methacrylate) (PMMA) dissolved in trichlorobenzene These nonplanar surface templates can be fabricated
(positive) or polychloromethylstyrene (negative) [8.57]. in a variety of ways such as etching through a mask
The e-beam gun is usually part of a scanning electron or cleavage along crystallographic planes. Subsequent
microscope (SEM), although transmission electron mi- epitaxial growth on top of these structures results in
croscopes (TEM) can also be used. Although electron a set of planes with different growth rates depending
wavelengths of the order of 1 Å can be easily achieved, on the geometry or surface diffusion and adsorption ef-
electron scattering in the resist limits the attainable fects. These effects can significantly enhance or limit
resolutions to > 10 nm. Beam control and pattern gener- the growth rate on certain planes, resulting in lateral
ation is achieved through a computer interface. E-beam patterning and confinement of deposited epitaxial layers
lithography is serial and hence has low throughput. Al- and formation of quantum wires (in V-grooves) and dots
though this is not a major concern in fabricating devices
used in studying fundamental microphysics, it severely InP cap layer
a)
limits large-scale nanofabrication. E-beam lithography
InGaAs
in conjunction with processes such as lift-off, etching, Q-wells
and electrodeposition can be used to fabricate various
nanostructures. InGaAs
Q-wire
8.3.2 Epitaxy and Strain Engineering InP
Part A 8.3
Atomic-precision deposition techniques such as molec- b) AlGaAs GaAs
ular-beam epitaxy (MBE) and metalorganic chemical
vapor deposition (MOCVD) have proven to be effective Mask
tools in fabricating a variety of quantum confine-
ment structures and devices (quantum well lasers,
photodetector, resonant tunneling diodes, etc.) [8.58– GaAs
60]. Although quantum wells and superlattices are the
structures that lend themselves most easily to these
techniques (Fig. 8.43a), quantum wires and dots have Fig. 8.44 (a) InGaAs quantum wire fabricated in V-groove
also been fabricated by adding subsequent steps such InP, and (b) AlGaAs quantum wire fabricated by epitaxial
as etching and selective growth. Fabrication of quantum growth on a masked GaAs substrate
258 Part A Nanostructures, Micro-/Nanofabrication and Materials
(in inverted pyramids). Figure 8.44a shows a schematic quantum dots or wires. Figure 8.45 shows a schematic
cross section of an InGaAs quantum wire fabricated in of the strain-induced self-assembly process. When the
a V-groove in InP. As can be seen the growth rate on the lattice constant of the substrate and the epitaxial layer
side-walls is lower than that of the top and bottom sur- differ considerably, only the first few deposited mono-
faces. Therefore the thicker InGaAs layer at the bottom layers crystallize, in the form of an epitaxial strained
of the V-groove forms a quantum wire confined from the layer in which the lattice constants are equal. When
sides by a thinner layer having a wider bandgap. Fig- a critical thickness is exceeded, a significant strain that
ure 8.44b shows a quantum wire formed using epitaxial occurs in the layer leads to the breakdown of this or-
growth over a dielectric patterned planar substrate. In dered structure and to the spontaneous formation of
both of these techniques it is relatively easy to create randomly distributed islets of regular shape and simi-
quantum wells; however, in order to create quantum lar size (usually < 30 nm in diameter). This mode of
wires and dots one still needs e-beam lithography to growth is usually referred to as the Stranski–Krastanow
pattern the grooves and window templates. mode. The quantum dot size, separation, and height de-
pend on the deposition parameters (i. e., total deposited
Quantum Structure Nanofabrication material, growth rate, and temperature) and material
Using Strain-Induced Self-Assembly combinations. As can be seen, this is a very convenient
A more recent technique for fabricating quantum wires method to grow perfect crystalline nanostructures over
and dots involves strain-induced self-assembly [8.62, a large area without any lithography and etching. One
65]. The term self-assembly represents a process where major drawback of this technique is the randomness of
a strained two-dimensional (2-D) system reduces its the quantum dot distribution. It should be mentioned
energy by a transition into a 3-D morphology. The that this technique can also be used to fabricate quantum
most commonly used material combination for this wires by strain relaxation bunching at step edges.
technique is the Inx Ga1−x As/GaAs system, which of-
fers a large lattice mismatch (7.2% between InAs and 8.3.3 Scanning Probe Techniques
GaAs) [8.66, 67], although recently Ge dots on Si sub-
strate have also attracted considerable attention [8.68]. The invention of scanning probe microscopy in the
This method relies on lattice mismatch between an epi- 1980s revolutionized atomic-scale imaging and spec-
taxially grown layer and its substrate to form an array of troscopy. In particular scanning tunneling and atomic
force microscopes (STM and AFM) have found
widespread applications in physics, chemistry, mater-
a) ial science, and biology. The possibility of atomic-scale
manipulation, lithography, and nanomachining using
such probes was considered from the beginning and
has matured considerably over the past decade. In this
section after a brief introduction to scanning probe
microscopes, we will discuss several important nano-
b)
lithography and machining techniques which have been
used to create nm-sized structures.
Scanning probe microscopy (SPM) systems are ca-
Part A 8.3
Part A 8.3
cantilever and is brought into contact with the surface
(Fig. 8.46b). The force between the tip and the sur-
face is detected by sensing the cantilever deflection.
A topographic image of the surface is obtained by plot-
ting the deflection as a function of the x–y-position.
In a more common mode of operation a feedback loop
is used to maintain a constant deflection while the to-
pographic information is obtained from the cantilever 1 µm
vertical displacement. Some scanning probe systems
use a combination of the AFM and the STM modes, Fig. 8.47 SEM image of an inverted truncated pyramid ar-
i. e., the tip is mounted on a cantilever with electrical ray fabricated on a silicon SOI wafer by SPM oxidation and
connection so that both the surface forces and tunneling subsequent etch in TMAH (pitch is 500 nm) (after [8.69])
260 Part A Nanostructures, Micro-/Nanofabrication and Materials
temperature on top of many materials. The deposited the case of silicon or poly) is removed with a HF dip.
silicon layer can be patterned and used as, for example, Subsequently 35–100 nm-thick resist is spin-coated on
the gate of a 0.1 μm CMOS transistor [8.74], or it can be top of the surface. The exposure is done by moving the
used as a mask to pattern an underlying film. The ma- SPM tip over the surface while applying a bias voltage
jor drawback of this technique is poor reproducibility sufficiently high to produce emission of electrons from
due to tip wear during the anodization. However, using the tip (a few tens of volts). Development of the resist is
AFM in noncontact mode has proved to overcome this performed in standard solutions following the exposure.
problem [8.70]. Features below 50 nm in width have been achieved with
this procedure.
Probe Resist Exposure and Lithography
Electrons emitted from a biased SPM tip can be used to Dip-Pen Nanolithography
expose a resist in the same way e-beam lithography that In dip-pen nanolithography (DPN) the tip of an AFM
does (Fig. 8.48) [8.74]. Various systems have been used operated in air is inked with a chemical of interest and
for this lithographic technique; these include constant- brought into contact with a surface. The ink molecules
current STM, noncontact AFM, and AFM with constant flow from the tip onto the surface as with a fountain pen.
tip–resist force and constant current. The systems using The water meniscus that naturally forms between the tip
AFM cantilevers have the advantage that they can per- and the surface enables the diffusion and transport of the
form imaging and alignment tasks without exposing the molecules, as shown in Fig. 8.49. Inking can be done by
resist. Resists well characterized for e-beam lithography dipping the tip into a solution containing a low concen-
(e.g., PMMA or SAL601) have been used with scan- tration of the molecules followed by a drying step (e.g.,
ning probe lithography to achieve reliable sub-100 nm blow-drying with compressed difluoroethane). Line
lithography. The procedure for this process is as fol- widths down to 12 nm with spatial resolution of 5 nm
lows. The wafers are cleaned and the native oxide (in have been demonstrated with this technique [8.75].
Species patterned with DPN include conducting poly-
mers, gold, dendrimers, DNA, organic dyes, antibodies,
I and alkanethiols. Alkanethiols have been also used as an
organic monolayer mask to etch a gold layer and sub-
sequently etch the exposed silicon substrate. One can
Organic resist also use a heated AFM cantilever to control the deposi-
V tion of a solid organic ink. This technique was recently
reported by Sheehan et al. in which 100 nm lines of
Exposure Conducting substrate octadecylphosphonic acid (melting point 100 ◦ C) were
written using a heated AFM probe [8.76].
Electric fields strong enough to induce the emission assembly) [8.88]. Molecular self-assembly is one of the
of atoms from the tip can be easily generated by ap- most important techniques used in biology for the de-
plying voltage pulses above 3 V. This phenomenon has velopment of complex functional structures. Since these
been used to transfer material from the tip to the surface techniques require that the target structures be ther-
and vice versa. Mounds (10–20 nm) of metals such as modynamically stable, it tends to produce structures
Au, Ag, and Pt have been deposited or removed from that are relatively defect-free and self-healing. Self-
a surface in this fashion [8.79]. The same approach has assembly is by no means limited to molecules or the
been used to extract single atoms from a semiconductor nanodomain and can be carried out on just about any
surface and redeposit them elsewhere [8.80]. Manip- scale, making it a powerful bottom-up assembly and
ulation of nanoparticles, molecules, and single atoms manufacturing method (multiscale ordering). Another
on top of a surface has also been achieved by sim- attractive feature of this technique relates to the possi-
ply pushing or sliding them with the SPM tip [8.81]. bility of combining self-assembly properties of organic
Metals can also be locally deposited by the STM chem- molecules with the electronic, magnetic, and photonic
ical vapor deposition technique [8.82]. In this technique properties of inorganic components. Template manu-
a precursor organometallic gas is introduced into the facturing is another bottom-up technique which utilizes
STM chamber. A voltage pulse applied between the tip material deposition (electroplating, CVD, etc.) into nan-
and the surface dissociates the precursor gas into a thin otemplates in order to fabricate nanostructures. Due
layer of metal. Local electrochemical etching [8.83] and to the simplicity and flexibility of electrochemistry for
electrodeposition [8.84] is also possible using SPM sys- plating and surface finishing of a broad range of ma-
tems. A droplet of suitable solution is first placed on the terials, its principle has recently been widely used for
substrate. Then the STM tip is immersed into the droplet electrochemical fabrication of various metallic nano-
and a voltage is applied. In order to reduce Faradaic cur- structures based on various templates. For example,
rents the tip is coated with wax such that only the very electrochemical deposition has been used to deposit
end is exposed to the solution. Sub-100 nm feature size large arrays of nanostructures in nanoporous templates,
has been achieved using this technique. such as porous alumina. This template-based deposition
Using a single tip to serially produce the desired typically provides metal nanowires as small as 25 nm in
modification in a surface leads to very slow fabrica- diameter and a few micrometers in length [8.89].
tion processes that are impractical for mass production. The nanotemplates used to fabricate nanostructures
Many of the scanning probe techniques developed so are usually prepared using self-assembly techniques.
far, however, could also be performed by an array In the following sections, we will discuss various
of tips, which would increase throughput and make important self-assembly and template manufacturing
them more competitive compared with other paral- techniques currently under heavy investigation.
lel nanofabrication processes. This approach has been
demonstrated for imaging, lithography [8.85], and data Physical and Chemical Self-Assembly
storage [8.86] using both one- and two-dimensional The central theme behind the self-assembly process is
arrays of scanning probes. With the development of spontaneous (physical) or chemical aggregation of col-
larger arrays with advances in individual control of loidal nanoparticles [8.90]. Spontaneous self-assembly
force, vertical position, and current, we might see these exploits the tendency of monodispersed nano or submi-
techniques being incorporated as standard fabrication cro colloidal spheres to organize into a face-centered
Part A 8.3
processes in the industry. cubic (fcc) lattice. The force driving this process is
the desire of the system to achieve a thermodynami-
8.3.4 Self-Assembly cally stable state (minimum free energy). In addition to
and Template Manufacturing spontaneous thermal self-assembly, gravitational, con-
vective, and electrohydrodynamic forces can also be
Self-assembly is a nanofabrication technique that in- used to induce aggregation into complex 3-D struc-
volves aggregation of colloidal nanoparticles into the tures. Chemical self-assembly requires the attachment
final desired structure [8.87]. This aggregation can be of a single-molecular organic layer (self-assembled
either spontaneous (entropic) and due to the thermo- monolayer or SAM) to the colloidal particles (organic
dynamic minima (energy minimization) constraints or or inorganic) and subsequent self-assembly of these
chemical and due to the complementary binding of components into a complex structures using molecular
organic molecules and supramolecules (molecular self- recognition and binding.
262 Part A Nanostructures, Micro-/Nanofabrication and Materials
the maximum number of colloidal particles determined dry from the solution, whereas monolayers made of
by the geometrical confinement. This method can be ester-terminated alkylsilanes emerge wet from the so-
used to fabricate a variety of polygonal and polyhedral lution used in their formation. The disadvantage of this
aggregates which are difficult to generate [8.97]. method is that, if the alkyltrichlorosilane in the solvent
adhering to the substrate is exposed to water, a cloudy
Chemical Self-Assembly film is deposited on the surface due to the formation
Organic and supramolecular SAMs play a critical role of a gel of polymeric siloxane. One solution to this
in colloidal particle self-assembly. SAMs are robust or- problem is the use of alkyldimethylchlorosilanes, which
ganic molecules which are chemically adsorbed onto have a single anchoring point, and thus cannot form
solid substrates [8.98]. Most often they have a hy- polymers. Chlorosilanes are sometimes preferred over
drophilic (polar) head which can be bonded to various alkoxysilanes because of their higher reactivity. How-
solid surfaces and a long hydrophobic (nonpolar) tail ever, the reactivity of chlorosilanes severely limits the
which extends outward. SAMs are formed by the range of functional groups that can be introduced at the
immersion of a substrate in a dilute solution of the mol- end of the hydrocarbon tail. On the contrary, methoxysi-
ecule in an organic solvent or water (liquid phase) or by lanes and ethoxysilanes are commonly available with
exposure to an atmosphere containing such a molecule many functional groups including amino, mercapto,
(gas phase). The resulting film is a dense organiza- epoxy, and thiocyanate groups, which are often neces-
tion of molecules arranged to expose the end group. sary for subsequent binding of colloidal particles and
The durability of the SAM is highly dependent on biomolecules. Gas-phase deposition of these molecules
the effectiveness of the anchoring to the surface of yields more uniform layers compared with liquid-phase
the substrate. SAMs have been widely studied because procedures [8.101]. Soft-lithography-like molds have
the end group can be functionalized to form precisely been used to obtain reactive silane patterns from the gas
arranged molecular arrays for various applications rang- phase by taking advantage of the characteristic perme-
ing from simple, ultrathin insulators and lubricants to ability of PDMS to volatile molecules [8.102].
complex biological sensors. Chemical self-assembly Another important organic SAM system is the alka-
uses organic or supramolecular SAMs as the binding nethiols (X(CH2 )n SH, where X is the end group) on
and recognition sites for fabricating complex 3-D struc- gold [8.98, 103–105]. A major advantage of using gold
tures from colloidal nanoparticles. The most commonly as the substrate material is that it does not have a stable
used organic monolayers include: oxide and thus can be handled in ambient conditions.
When a fresh, clean, hydrophilic gold substrate is im-
1. Organosilicon compounds on glass and oxidized sil-
icon
2. Alkanethiols, dialkyl disulfides, and dialkyl sulfides a) R
on gold
3. Fatty acids on alumina and other metal oxides
4. DNA (CH2)n
Octadecyltrichlorosilane (OTS) is the most common
organosilane used in the formation of SAMs, mainly be-
SiO3
cause of the fact that it is simple, readily available, and
Part A 8.3
forms good, dense layers [8.99, 100]. Alkyltrichlorosi-
lane monolayers can be prepared on clean silicon wafers b)
whose surface is SiO2 (with almost 5 × 1014 SiOH
groups/cm2 ). Figure 8.53 shows a schematic represen-
SiO2 SiO2
tation of the formation of alkylsiloxane monolayers by
adsorption of alkyltrichlorosilane from solution onto
Si/SiO2 substrates. Since the silicon–chloride bond is Si Si
susceptible to hydrolysis, a limited amount of water
has to be present in the system in order to obtain
good-quality monolayers. Monolayers made of methyl- Fig. 8.53 (a) Alkylsiloxane formed from the adsorption of
and vinyl-terminated alkylsilanes are autophobic to alkyltrichlorosilane on Si/SiO2 substrates. (b) Schematic
the hydrocarbon solution and hence emerge uniformly representation of the process
264 Part A Nanostructures, Micro-/Nanofabrication and Materials
mersed (several min to several h) into a dilute solution bind specifically to another nucleotide when arranged
(10−3 M) of the organic sulfur compound (alkanethi- in the two-strand double helix (A to T, and C to G).
ols) in an inorganic solvent a close-packed, oriented This specific bonding capability can be used to assem-
monolayers can be obtained. Sulfur is used as the ble nanophase material and nanostructures [8.109]. For
head group because of its strong interaction with the example, nucleotide-functionalized nanogold particles
gold substrate (44 kcal/mol), resulting in the forma- have been assembled into complex 3-D structures by
tion of a close-packed, ordered monolayer. The end attaching DNA strands to the gold via an enabler or
group of alkanethiol can be modified to render the linker [8.110]. In a separate work DNA was used to as-
adsorbed layer hydrophobic or hydrophilic. Another semble nanoparticles into macroscopic materials. This
method for depositing alkanethiol SAM is soft lithog- method uses alkane dithiol as the linker molecule to
raphy. This technique is based on inking a PDMS stamp connect the DNA template to the nanoparticle. The thiol
with alkanethiol and its subsequent transfer to pla- groups at each end of the linker molecule covalently
nar or nonplanar substrates. Alkanethiol-functionalized attach themselves to the colloidal particles to form ag-
surfaces (planar, nonplanar, spherical) can also be gregate structures [8.111].
used to self-assemble a variety of intricate 3-D struc-
tures [8.106]. Template Manufacturing
Carboxylic acid derivatives self-assemble on sur- Template manufacturing refers to a set of techniques
faces (e.g., glass, Al2 O3 , and Ag2 O) through an that can be used to fabricate 3-D organic or inorganic
acid–base reaction, giving rise to monolayers of fatty structures from a nanotemplate. These templates dif-
acids [8.107]. The time required for the formation of fer in material, pattern, feature size, overall template
a complete monolayer increases with decreasing con- size, and periodicity. Although nanotemplates can be
centration. Higher concentration of the carboxylic acid fabricated using e-beam lithography, the serial nature
is required to form a monolayer on gold as compared of this technique prohibits its widespread application.
with on Al2 O3 . This is due to differences in the affin- Self-assembly is the preferred technique to produce
ity of the COOH groups (more affinity to Al2 O3 and large-area nanotemplates in a massively parallel fash-
glass than gold) and also the surface concentration of ion. Several nanotemplates have been investigated for
the salt-forming oxides of the two substrates. In the case use in template manufacturing. These include polymer
of amorphous metal oxide surfaces, the chemisorption colloidal spheres, alumina membranes, and nuclear-
of alkanoic acids is not unique. For example, on Ag2 O, track etched membranes. Colloidal spheres can be
the two oxygen atoms of the carboxylate bind to the deposited in a regular 3-D array using the techniques
substrate in a nearly symmetrical manner, thus resulting described in the previous section (Figs. 8.50–8.52).
in ordered monolayers with a chain tilt angle of 15−25◦ Porous aluminum oxide membranes can be fabricated
from the surface normal. However, on CuO and Al2 O3 , by the anodic oxidation of aluminum [8.112]. The ox-
the oxygen atoms bind themselves symmetrically and idized film consists of columnar arrays of hexagonal
the chain tilt angle is close to 0◦ . The structure of the close-packed pores with separation comparable to the
monolayers is thus a balance of the various interactions pore size. By controlling the electrolyte species, tem-
taking place in the polymer chains. perature, anodizing voltage, and time, different pore
Deoxyribonucleic acid (DNA), the framework on sizes, densities, and heights can be obtained. The pore
which all life is built, can be used to self-assemble nano- size and depth can further be adjusted by etching the
Part A 8.3
materials into useful macroscopic aggregates that dis- oxide in an appropriate acid. Templates of porous poly-
play a number of desirable physical properties [8.108]. carbonate or mica membranes can be fabricated by
DNA consists of two strands, which are coiled around nuclear-track etched membranes [8.113]. This tech-
each other to form a double helix. When the two strands nique is based on the passage of high-energy decay
are uncoiled singular strands of nucleotides are left. fragments from a radioactive source through a dielectric
These nucleotides consist of a sugar (pentose ring) material. The particles leave behind chemically ac-
a phosphate (PO4 ), and a nitrogenous base. The order tive damaged tracks which can subsequently be etched
and architecture of these components is essential for to create pores through the thickness of the mem-
the proper structure of a nucleotide. There are typically brane [8.114, 115]. Unlike other methods, the pore
four nucleotides found in DNA: adenine (A), guanine separation and hence the pore density is independent
(G), cytosine (C), and thymine (T). A key property of the pore size. The pore density is only determined
of the DNA structure is that the nucleotides described by the irradiation process. More recently, electrochem-
Introduction to Micro-/Nanofabrication References 265
ical deposition of metallic nanowires on the step edges can be selectively etched, resulting in an air-filled
of highly oriented pyrolytic graphite (HOPG) templates 3-D complex structure. For example, nickel [8.119],
have also been demonstrated, which produces metal- iron [8.120], and cobalt [8.121] nanowires have been
lic nanowires with diameters as small as 15 nm. This electrochemically grown into porous template matri-
method has been successfully used for metals such as ces. Three-dimensional photonic crystals have been
Cu, Ni, Au, and Pd [8.116, 117]. fabricated by electrochemical deposition of CdSe and
Subsequent to template fabrication, the intersti- silicon into polystyrene and silica colloidal assem-
tial spaces (in the case of colloidal spheres) or pores bly templates [8.91, 122]. An interesting example of
(in the case of alumina and polycarbonate mem- template-assisted manufacturing is the synthesis of
branes) in the template are filled with the desired nanometer metallic barcodes [8.123]. These nanobar-
material [8.93, 118]. This can be done by using a va- codes are prepared by electrochemical reduction of
riety of deposition techniques such as electroplating metallic ions into the pores of an aluminum oxide mem-
and CVD. The final structure can be a composite of brane, followed by their release through etching of the
nanotemplate and deposited material or the template template [8.93].
References
Part A 8
8.1 S.A. Campbell: The Science and Engineering of Mi-
croelectronic Fabrication (Oxford Univ. Press, New (PLD) – A versatile thin film technique, Adv. Solid
York 2001) State Phys. 43, 505–517 (2003)
8.2 C.J. Jaeger: Introduction to Microelectronic Fabrica- 8.7 M. Leskela, M. Ritala: Atomic layer deposition chem-
tion (Prentice Hall, New Jersey 2002) istry: recent developments and future challenges,
8.3 J.D. Plummer, M.D. Deal, P.B. Griffin: Silicon VLSI Angew. Chem. Int. Ed. 42, 5548–5554 (2003)
Technology (Prentice-Hall, New Jersey 2000) 8.8 J.L. Vossen: Thin Film Processes (Academic, New York
8.4 J.E. Bjorkholm: EUV lithography: the successor to 1976)
optical lithography, Intel Technol. J. 2, 1–8 (1998) 8.9 M. Gad-el-Hak (Ed.): The MEMS Handbook (CRC, Boca
8.5 S. Owa, H. Nagasaka: Advantage and feasibility of Raton 2002)
immersion lithography, J. Microlithogr. Microfabr. 8.10 T.-R. Hsu: MEMS and Microsystems Design and Man-
Microsyst. 3, 97–103 (2004) ufacture (McGraw Hill, New York 2002)
8.6 H.U. Krebs, M. Störmer, J. Faupel, E. Süske, 8.11 G.T.A. Kovacs: Micromachined Transducers Source-
T. Scharf, C. Fuhse, N. Seibt, H. Kijewski, D. Nelke, book (McGraw Hill, New York 1998)
266 Part A Nanostructures, Micro-/Nanofabrication and Materials
8.12 P. Rai-Choudhury (Ed.): Handbook of Microlithog- 8.29 X. Huikai, L. Erdmann, Z. Xu, K.J. Gabriel, G.K. Fed-
raphy, Micromachining and Microfabrication. Vol. der: Post-CMOS processing for high-aspect-ratio
2: Micromachining and Microfabrication (SPIE/IEE, integrated silicon microstructures, J. Microelec-
Bellingham, Washington/London 1997) tromech. Syst. 11, 93–101 (2002)
8.13 M.F. Aimi, M.P. Rao, N.C. MacDonald, A.S. Zu- 8.30 J.M. Bustillo, R.S. Muller: Surface micromachin-
ruzi, D.P. Bothman: High-aspect-ratio bulk mi- ing for microelectromechanical systems, Proc. IEEE
cromachining of titanium, Nat. Mater. 3, 103–105 86(8), 1552–1574 (1998)
(2004) 8.31 H.C. Nathanson, W.E. Newell, R.A. Wickstrom,
8.14 T.J. Cotler, M.E. Elta: Plasma-etch technology, IEEE J.R. Davis: The resonant gate transistor, IEEE Trans.
Circuits Devices Mag. 6, 38–43 (1990) Electron Devices 14, 117–133 (1967)
8.15 U. Gosele, Q.Y. Tong: Semiconductor wafer bonding, 8.32 R.T. Howe, R.S. Muller: Polycrystalline silicon mi-
Annu. Rev. Mater. Sci. 28, 215–241 (1998) cromechanical beams, Proc. Electrochem. Soc.
8.16 Q.Y. Tong, U. Gosele: Semiconductor Wafer Bonding: Spring Meet., Montreal (1982) pp. 184–185
Science and Technology (Wiley, New York 1999) 8.33 J.A. Geen, S.J. Sherman, J.F. Chang, S.R. Lewis:
8.17 F. Niklaus, P. Enoksson, E. Kalveston, G. Stemme: Single-chip surface-micromachined integrated gy-
Void-free full-wafer adhesive bonding, J. Mi- roscope with 50 /h root Allan deviation, IEEE
cromech. Microeng. 11, 100–107 (2000) J. Solid-State Circuits 37, 1860–1866 (2002)
8.18 C.A. Harper: Electronic Packaging and Interconnec- 8.34 P.F. Van Kessel, L.J. Hornbeck, R.E. Meier, M.R. Dou-
tion Handbook (McGraw-Hill, New York 2000) glass: A MEMS-based projection display, Proc. IEEE
8.19 W.H. Ko, J.T. Suminto, G.J. Yeh: Bonding tech- 86(8), 1687–1704 (1998)
niques for microsensors. In: Micromachining and 8.35 A.E. Franke, D. Bilic, D.T. Chang, P.T. Jones,
Micropackaging for Transducers, ed. by W.H. Ko R.T. Howe, G.C. Johnson: Post-CMOS integration
(Elsevier, Amsterdam 1985) of germanium microstructures, Proc. Micro Electro
8.20 G.T.A. Kovacs, N.I. Maluf, K.A. Petersen: Bulk mi- Mech. Syst., Orlando (1999) pp. 630–637
cromachining of silicon, Proc. IEEE 86(8), 1536–1551 8.36 S. Sedky, P. Fiorini, M. Caymax, S. Loreti, K. Baert,
(1998) L. Hermans, R. Mertens: Structural and mechanical
8.21 K. Najafi, K.D. Wise, T. Mochizuki: A high-yield properties of polycrystalline silicon germanium for
IC-compatible multichannel recording array, IEEE micromachining applications, J. Microelectromech.
Trans. Electron Devices 32, 1206–1211 (1985) Syst. 7, 365–372 (1998)
8.22 A. Selvakumar, K. Najafi: A high-sensitivity z-axis 8.37 N. Tas, T. Sonnenberg, H. Jansen, R. Legtenberg,
capacitive silicon microaccelerometer with a tor- M. Elwenspoek: Stiction in surface micromachining,
tional suspension, J. Microelectromech. Syst. 7, J. Micromech. Microeng. 6, 385–397 (1996)
192–200 (1998) 8.38 R. Maboudian, R.T. Howe: Critical review: Adhesion
8.23 H. Baltes, O. Paul, O. Brand: Micromachined ther- in surface micromechanical structures, J. Vac. Sci.
mally based CMOS microsensors, Proc. IEEE 86(8), Technol. B 15(1), 1–20 (1997)
1660–1678 (1998) 8.39 J.H. Smith, S. Montague, J.J. Sniegowski, J.R. Murray,
8.24 N.C. MacDonald: SCREAM MicroElectroMechanical P.J. McWhorter: Embedded micromechanical devices
Systems, Microelectron. Eng. 32, 51–55 (1996) for the monolithic integration of MEMS with CMOS,
8.25 B. Eyre, K.S.J. Pister, W. Gekelman: Multi-axis mi- Proc. Int. Electron Devices Meet., Washington (1995)
crocoil sensors in standard CMOS, Proc. SPIE Conf. pp. 609–612
Micromach. Devices Compon., Austin (1995) pp. 183– 8.40 R.S. Muller, K.Y. Lau: Surface-micromachined mi-
191 crooptical elements and systems, Proc. IEEE 86(8),
8.26 K.A. Shaw, Z.L. Zhang, N.C. MacDonnald: SCREAM: 1705–1720 (1998)
A single mask, single-crystal silicon process for 8.41 K.S.J. Pister, M.W. Judy, S.R. Burgett, R.S. Fearing:
Part A 8
microelectromechanical structures, Proc. IEEE Work- Microfabricated hinges: 1 mm vertical features with
shop Micro Electro Mech. Syst., Fort Lauderdale surface micromachining, Proc. 6th Int. Conf. Solid-
(1993) pp. 155–160 State Sens. Actuators, San Francisco (1991) pp. 647–
8.27 G.K. Fedder, S. Santhanam, M.L. Reed, S.C. Eagle, 650
D.F. Guillo, M.S.C. Lu, L.R. Carley: Laminated high- 8.42 L.Y. Lin, S.S. Lee, M.C. Wu, K.S.J. Pister: Microma-
aspect-ratio microstructures in a conventional CMOS chined Integrated optics for free space interconnec-
process, Proc. IEEE Workshop Micro Electro Mech. tion, Proc. IEEE Micro Electro Mech. Syst. Workshop,
Syst., San Diego (1996) pp. 13–18 Amsterdam (1995) pp. 77–82
8.28 B.P. Van Drieenhuizen, N.I. Maluf, I.E. Opris, 8.43 E.W. Becker, W. Ehrfeld, P. Hagmann, A. Maner,
G.T.A. Kovacs: Force-balanced accelerometer with D. Munchmeyer: Fabrication of microstructures with
mG resolution fabricated using silicon fusion bond- high aspect ratios and great structural heights by
ing and deep reactive ion etching, Proc. Int. Conf. synchrotron radiation lithography, galvanoforming,
Solid-State Sens. Actuators, Chicago (1997) pp. 1229– and plastic molding (LIGA process), Microelectron.
1230 Eng. 4, 35–56 (1986)
Introduction to Micro-/Nanofabrication References 267
8.44 H. Guckel: High-aspect-ratio micromachining via 8.61 S. Kim, M. Razegi: Advances in quantum dot struc-
deep x-ray lithography, Proc. IEEE 86(8), 1586–1593 tures. In: Processing and Properties of Compound
(1998) Semiconductors, ed. by K. Willardson, E.R. Weber
8.45 K.Y. Lee, N. LaBianca, S.A. Rishton, S. Zolgharnain, (Academic, New York 2001)
J.D. Gelorme, J. Shaw, T.H.P. Chang: Micromachining 8.62 D. Bimberg, M. Grundmann, N.N. Ledentsov: Quan-
applications of a high resolution ultra-thick pho- tum Dot Heterostructures (Wiley, New York 1999)
toresist, J. Vac. Sci. Technol. B 13, 3012–3016 (1995) 8.63 G. Seebohm, H.G. Craighead: Lithography and pat-
8.46 K. Roberts, F. Williamson, G. Cibuzar, L. Thomas: terning for nanostructure fabrication. In: Quantum
The fabrication of an array of microcavities utilizing Semiconductor Devices and Technologies, ed. by
SU-8 photoresist as an alternative ‘LIGA’ technol- T.P. Pearsall (Kluwer, Boston 2000)
ogy, Proc. 13th Bienn. Univ./Gov./Ind. Microelectron. 8.64 E. Kapon: Lateral patterning of quantum well het-
Symp. (IEEE), Minneapolis (1999) pp. 139–141 erostructures by growth on nonplanar substrates.
8.47 D.A. Horsley, M.B. Cohn, A. Singh, R. Horowitz, In: Epitaxial Microstructures, ed. by A.C. Gossard
A.P. Pisano: Design and fabrication of an angular (Academic, New York 1994)
microactuator for magnetic disk drives, J. Microelec- 8.65 F. Guffarth, R. Heitz, A. Schliwa, O. Stier,
tromech. Syst. 7, 141–148 (1998) N.N. Ledentsov, A.R. Kovsh, V.M. Ustinov, D. Bim-
8.48 C. Burbaum, J. Mohr, P. Bley, W. Ehrfeld: Fabrica- berg: Strain engineering of self-organized InAs
tion of capacitive acceleration sensors by the LIGA quantum dots, Phys. Rev. B 64, 085305–1–085305–7
technique, Sens. Actuators A Phys. A27, 559–563 (2001)
(1991) 8.66 M. Sugawara: Self-Assembled InGaAs/GaAs Quantum
8.49 C.G. Keller, R.T. Howe: Hexsil bimorphs for vertical Dots (Academic, New York 1999)
actuation, Dig. Tech. Pap. 8th Int. Conf. Solid- 8.67 B.C. Lee, S.D. Lin, C.P. Lee, H.M. Lee, J.C. Wu,
State Sens. Actuators Eurosens. IX, Stockholm (1995) K.W. Sun: Selective growth of single InAs quantum
pp. 99–102 dots using strain engineering, Appl. Phys. Lett. 80,
8.50 C.G. Keller, R.T. Howe: Nickel-filled hexsil thermally 326–328 (2002)
actuated tweezers, Dig. Tech. Pap. 8th Int. Conf. 8.68 K. Brunner: Si/Ge nanostructures, Rep. Prog. Phys.
Solid-State Sens. Actuators Eurosens. IX, Stockholm 65, 27–72 (2002)
(1995) pp. 376–379 8.69 F.S.S. Chien, W.F. Hsieh, S. Gwo, A.E. Vladar,
8.51 N. Yazdi, F. Ayazi, K. Najafi: Micromachined inertial J.A. Dagata: Silicon nanostructures fabricated by
sensors, Proc. IEEE 86, 1640–1659 (1998) scanning-probe oxidation and tetra-methyl am-
8.52 F. Ayazi, K. Najafi: High aspect-ratio combined poly monium hydroxide etching, J. Appl. Phys. 91,
and single-crystal silicon (HARPSS) MEMS technology, 10044–10050 (2002)
J. Microelectromech. Syst. 9, 288–294 (1998) 8.70 M. Calleja, J. Anguita, R. Garcia, K. Birkelund,
8.53 F. Ayazi, K. Najafi: A HARPSS polysilicon vibrating F. Perez-Murano, J.A. Dagata: Nanometer-scale
ring gyroscope, J. Microelectromech. Syst. 10, 169– oxidation of silicon surfaces by dynamic force
179 (2001) microscopy: reproducibility, kinetics and nanofab-
8.54 Y.S. No, F. Ayazi: The HARPSS process for fabri- rication, Nanotechnology 10, 34–38 (1999)
cation of nano-precision silicon electromechanical 8.71 E.S. Snow, P.M. Campbell, F.K. Perkins: Nanofabri-
resonators, Proc. 2001 1st IEEE Conf. Nanotechnol., cation with proximal probes, Proc. IEEE 85, 601–611
Maui (2001) pp. 489–494 (1997)
8.55 G. Timp: Nanotechnology (Springer, New York 1998) 8.72 H. Sugimura, T. Uchida, N. Kitamura, H. Masuhara:
8.56 P. Rai-Choudhury (Ed.): Handbook of Microlithog- Tip-induced anodization of titanium surfaces by
raphy, Micromachining and Microfabrication. Vol. scanning tunneling microscopy: A humidity effect
1: Microlithography (SPIE/IEE, Bellingham, Washing- on nanolithography, Appl. Phys. Lett. 63, 1288–1290
Part A 8
ton/London 1997) (1993)
8.57 L. Ming, C. Bao-qin, Y. Tian-Chun, Q. He, X. Qiuxia: 8.73 N. Kramer, J. Jorritsma, H. Birk, C. Schonenberger:
The sub-micron fabrication technology, Proc. 6th Nanometer lithography on silicon and hydrogenated
Int. Conf. Solid-State Integr.-Circuit Technol., San amorphous silicon with low energy electrons, J. Vac.
Francisco, Vol. 1 (2001) pp. 452–455 Sci. Technol. B 13, 805–811 (1995)
8.58 M.A. Herman: Molecular Beam Epitaxy: Fundamen- 8.74 H.T. Soh, K.W. Guarini, C.F. Quate: Scanning Probe
tals and Current Status (Springer, New York 1996) Lithography (Kluwer Academic, Boston 2001)
8.59 J.S. Frood, G.J. Davis, W.T. Tsang: Chemical Beam 8.75 C.A. Mirkin: Dip-pen nanolithography: automated
Epitaxy and Related Techniques (Wiley, New York fabrication of custom multicomponent, sub-100-
1997) nanometer surface architectures, MRS Bulletin 26,
8.60 S. Mahajan, K.S. Sree Harsha: Principles of Growth 535–538 (2001)
and Processing of Semiconductors (McGraw Hill, New 8.76 P.E. Sheehan, L.J. Whitman, W.P. King, B.A. Nelson:
York 1999) Nanoscale deposition of solid inks via thermal dip
268 Part A Nanostructures, Micro-/Nanofabrication and Materials
pen nanolithography, Appl. Phys. Lett. 85, 1589– 8.93 J.C. Hulteen, C.R. Martin: A general template-
1591 (2004) based method for the preparation of nanomaterials,
8.77 L.L. Sohn, R.L. Willett: Fabrication of nanostructures J. Mater. Chem. 7, 1075–1087 (1997)
using atomic-force-microscope-based lithography, 8.94 J.D. Joannopoulos, P.R. Villeneuve, S. Fan: Photonic
Appl. Phys. Lett. 67, 1552–1554 (1995) crystals: putting a new twist on light, Nature 386,
8.78 H.J. Mamin, B.D. Terris, L.S. Fan, S. Hoen, R.C. Bar- 143–149 (1997)
rett, D. Rugar: High-density data storage using 8.95 T.D. Clark, R. Ferrigno, J. Tien, K.E. Paul, G.M. White-
proximal probe techniques, IBM J. Res. Dev. 39, sides: Template-directed self-assembly of 10-µm-
681–699 (1995) sized hexagonal plates, J. Am. Chem. Soc. 124,
8.79 K. Bessho, S. Hashimoto: Fabricating nanoscale 5419–5426 (2002)
structures on Au surface with scanning tunneling 8.96 S.A. Sapp, D.T. Mitchell, C.R. Martin: Using template-
microscope, Appl. Phys. Lett. 65, 2142–2144 (1994) synthesized micro- and nanowires as building
8.80 I.W. Lyo, P. Avouris: Field-induced nanometer- to blocks for self-assembly of supramolecular archi-
atomic-scale manipulation of silicon surfaces with tectures, Chem. Mater. 11, 1183–1185 (1999)
the STM, Science 253, 173–176 (1991) 8.97 Y. Yin, Y. Lu, B. Gates, Y. Xia: Template assisted self-
8.81 M.F. Crommie, C.P. Lutz, D.M. Eigler: Confinement assembly: A practical route to complex aggregates
of electrons to quantum corrals on a metal surface, of monodispersed colloids with well-defined sizes,
Science 262, 218–220 (1993) shapes and structures, J. Am. Chem. Soc. 123, 8718–
8.82 A. de Lozanne: Pattern generation below 0.1 mi- 8729 (2001)
cron by localized chemical vapor deposition with the 8.98 J.L. Wilbur, G.M. Whitesides: Self-assembly and
scanning tunneling microscope, Jpn. J. Appl. Phys. self-assembles monolayers in micro- and nanofab-
33, 7090–7093 (1994) rication. In: Nanotechnology, ed. by G. Timp
8.83 L.A. Nagahara, T. Thundat, S.M. Lindsay: Nano- (Springer, New York 1999)
lithography on semiconductor surfaces under an 8.99 S.R. Wasserman, Y.T. Tao, G.M. Whitesides: Structure
etching solution, Appl. Phys. Lett. 57, 270–272 (1990) and reactivity of alkylsiloxane monolayers formed
8.84 T. Thundat, L.A. Nagahara, S.M. Lindsay: Scan- by reaction of alkyltrichlorosilanes on silicon sub-
ning tunneling microscopy studies of semiconductor strates, Langmuir 5, 1074–1087 (1989)
electrochemistry, J. Vac. Sci. Technol. A 8, 539–543 8.100 C.P. Tripp, M.L. Hair: An infrared study of the reac-
(1990) tion of octadecyltrichlorosilane with silica, Langmuir
8.85 S.C. Minne, S.R. Manalis, A. Atalar, C.F. Quate: Inde- 8, 1120–1126 (1992)
pendent parallel lithography using the atomic force 8.101 S. Fiorilli, P. Rivolo, E. Descrovi, C. Ricciardi,
microscope, J. Vac. Sci. Technol. B 14, 2456–2461 L. Pasquardini, L. Lunelli, L. Vanzetti, C. Pederzolli,
(1996) B. Onida, E. Garrone: Vapor-phase self-assembled
8.86 M. Lutwyche, C. Andreoli, G. Binnig, J. Brugger, monolayers of aminosilane on plasma-activated sil-
U. Drechsler, W. Häberle, H. Rohrer, H. Rothuizen, icon substrates, J. Colloid Interface Sci. 321, 235–241
P. Vettiger: Microfabrication and parallel operation (2008)
of 5 × 5 2-D AFM cantilever arrays for data storage 8.102 R. de la Rica, A. Baldi, E. Mendoza, A. San Paulo,
and imaging, Proc. MEMS 98, 8–11 (1998) A. Llobera, C. Fernandez-Sanchez: Silane nanopat-
8.87 G.M. Whitesides, B. Grzybowski: Self-assembly at all terns via gas-phase soft lithography, Small 4(8),
scales, Science 295, 2418–2421 (2002) 1076–1079 (2008)
8.88 P. Kazmaier, N. Chopra: Bridging size scales with 8.103 D.R. Walt: Nanomaterials: Top-to-bottom functional
self-assembling supramolecular materials, MRS Bul- design, Nature 1, 17–18 (2002)
letin 25, 30–35 (2000) 8.104 J. Noh, T. Murase, K. Nakajima, H. Lee, M. Hara:
8.89 G.A. Gelves, Z.T.M. Murakami, M.J. Krantz, Nanoscopic investigation of the self-assembly pro-
Part A 8
J.A. Haber: Multigram synthesis of copper nanowires cesses of dialkyl disulfides and dialkyl sulfides on
using ac electrodeposition into porous aluminium Au(111), J. Phys. Chem. B 104, 7411–7416 (2000)
oxide templates, J. Mater. Chem. 16, 3075–3083 8.105 M. Himmelhaus, F. Eisert, M. Buck, M. Grunze: Self-
(2006) assembly of n-alkanethiol monolayers: A study by
8.90 R. Plass, J.A. Last, N.C. Bartelt, G.L. Kellogg: Self- IR-visible sum frequency spectroscopy (SFG), J. Phys.
assembled domain patterns, Nature 412, 875 (2001) Chem. 104, 576–584 (1999)
8.91 Y.A. Vlasov, X.-Z. Bo, J.G. Sturm, D.J. Norris: On-chip 8.106 A.K. Boal, F. Ilhan, J.E. DeRouchey, T. Thurn-
natural self-assembly of silicon photonic bandgap Albrecht, T.P. Russell, V.M. Rotello: Self-assembly of
crystals, Nature 414, 289–293 (2001) nanoparticles into structures spherical and network
8.92 C. Gigault, K. Dalnoki-Veress, J.R. Dutcher: Changes aggregates, Nature 404, 746–748 (2000)
in the morphology of self-assembled polystyrene 8.107 A. Ulman: An Introduction to Ultrathin Organic
microsphere monolayers produced by annealing, Films: From Langmuir–Blodgett to Self-Assembly
J. Colloid Interface Sci. 243, 143–155 (2001) (Academic, New York 1991)
Introduction to Micro-/Nanofabrication References 269
8.108 E. Winfree, F. Liu, L.A. Wenzler, N.C. Seeman: Design arrays by electrodeposition, ChemPhysChem 4(2),
and self-assembly of two-dimensional DNA crystals, 131–138 (2003)
Nature 394, 539–544 (1998) 8.117 M.Z. Atashbar, D. Banerji, S. Singamaneni,
8.109 J.H. Reif, T.H. LaBean, N.C. Seeman: Programmable V. Bliznyuk: Deposition of parallel arrays of palla-
assembly at the molecular scale: self-assembly of dium nanowires and electrical characterization us-
DNA lattices, Proc. 2001 IEEE Int. Conf. Robot. Autom., ing microelectrode contacts, Nanotechnology 15(3),
Seoul (2001) pp. 966–971 374–378 (2004)
8.110 A.P. Alivisatos, K.P. Johnsson, X. Peng, T.E. Wilson, 8.118 Y. Du, W.L. Cai, C.M. Mo, J. Chen, L.D. Zhang,
C.J. Loweth, M.P. Bruchez Jr, P.G. Schultz: Organi- X.G. Zhu: Preparation and photoluminescence of
zation of ‘nanocrystal molecules’ using DNA, Nature alumina membranes with ordered pore arrays, Appl.
382, 609–611 (1996) Phys. Lett. 74, 2951–2953 (1999)
8.111 H. Cao, Z. Yu, J. Wang, J.O. Tegenfeldt, R.H. Austin, 8.119 M. Guowen, C. Anyuan, C. Ju-Yin, A. Vijayaragha-
E. Chen, W. Wu, S.Y. Chou: Fabrication of 10 nm en- van, J.J. Yung, M. Shima, P.M. Ajayan: Ordered
closed nanofluidic channels, Appl. Phys. Lett. 81, Ni nanowire tip arrays sticking out of the anodic
174–176 (2002) aluminum oxide template, J. Appl. Phys. 97, 64303–
8.112 H. Masuda, H. Yamada, M. Satoh, H. Asoh: Highly 1–64303–5 (2005)
ordered nanochannel-array architecture in anodic 8.120 S. Yang, H. Zhu, D. Yu, Z. Jin, S. Tang, Y. Du: Prepa-
alumina, Appl. Phys. Lett. 71, 2770–2772 (1997) ration and magnetic property of Fe nanowire array,
8.113 R.L. Fleischer: Nuclear Tracks in Solids: Principles and J. Magn. Magn. Mater. 222, 97–100 (2000)
Applications (Univ. of California Press, Berkeley 1976) 8.121 M. Sun, G. Zangari, R.M. Metzger: Cobalt island ar-
8.114 R.E. Packard, J.P. Pekola, P.B. Price, R.N.R. Spohr, rays with in-plane anisotropy electrodeposited in
K.H. Westmacott, Y.Q. Zhu: Manufacture observation highly ordered alumina, IEEE Trans. Magn. 36, 3005–
and test of membranes with locatable single pores, 3008 (2000)
Rev. Sci. Instrum. 57, 1654–1660 (1986) 8.122 P.V. Braun, P. Wiltzius: Electrochemically grown
8.115 L. Sun, P.C. Searson, C.L. Chien: Electrochemical de- photonic crystals, Nature 402, 603–604 (1999)
position of nickel nanowire arrays in single-crystal 8.123 S.R. Nicewarner-Pena, R.G. Freeman, B.D. Reiss,
mica films, Appl. Phys. Lett. 74, 2803–2805 (1999) L. He, D.J. Pena, I.D. Walton, R. Cromer, C.D. Keat-
8.116 E.C. Walter, M.P. Zach, F. Favier, B.J. Murray, K. In- ing, M.J. Natan: Submicrometer metallic barcodes,
azu, J.C. Hemminger, R.M. Penner: Metal nanowire Science 294, 137–141 (2001)
Part A 8
271
Nanoimprint
9. Nanoimprint Lithography –
Patterning of Resists Using Molding
Helmut Schift, Anders Kristensen
cepts of molding processes can be observed in daily but at the expense of a possible reduction of feature res-
life, such as imprinting a footprint into snow or clay, olution due to deformation of the stamp. These issues
making waffles in a pressure process with subsequent are treated in more detail in [9.8, 9].
thermocuring, or replicating a seal into wax (Fig. 9.1). In this chapter we provide an overview of the differ-
Even these examples show the variety of molding pro- ent processes currently called nanoimprinting, from hot
cesses. One common important prerequisite of these embossing of thermoplastic materials to imprinting and
Nanoimprint Lithography – Patterning of Resists Using Molding 9.1 Emerging Nanopatterning Methods 273
curing of liquid resins. After this Introduction into the processes. We conclude with an outlook in Sect. 9.5, in
basics of molding, Sect. 9.1 places the two main NIL which we discuss the prospects of NIL and aspects of
techniques into the context of the emerging nanopat- its commercialization. Further information can be found
terning methods for lithography. Section 9.2 is the main in the references, in publications dealing with so-called
section, where the NIL process is described in detail, lithography, electroforming, and molding (LIGA, from
beginning with a discussion about polymer properties, its German abbreviation) technology [9.10] and optical
giving an insight into squeeze flow of thin films. As storage fabrication, but not least within this Handbook
a first step towards applications major pattern trans- in Chaps. 8 and 10 about silicon micromachining and
fer techniques used in NIL are presented. Section 9.3 soft lithography. In this chapter we restrict ourselves to
presents materials and tools for NIL, ranging from lithographic patterning of thin films on hard substrates.
materials for stamps and resists, to imprint machines. We present basic concepts rather than state-of-the-art
Section 9.4 presents typical applications which are cur- tools and hot scientific issues. As a complement to this
rently envisaged both at an industrial and at laboratory chapter, the reader is advised to refer to two publica-
scale. Although for many people the main driving force tions: A recent review on NIL [9.11] deals with a range
behind NIL is its use as next-generation lithography of process issues relevant for research and industry, and
(NGL) for complementary metal–oxide–semiconductor a deeper insight into advanced concepts of printing.
(CMOS) chip fabrication, the reader will be introduced Specific NIL processes and process flows for a vari-
to different other applications which do not have the ety of applications are presented in the NaPa Library
demanding overlay requirements imposed by multilevel of processes (NaPa LoP) [9.12].
far by molding is much higher than that used in in- other lithographic techniques, the concept of resist pat-
dustrial fabrication of processors and memory chips terning (which can also be found in different chapters
with high-end photolithography (PL). This makes NIL in this Handbook). The resist patterns are generated by
a promising technology for NGL [9.14]. Apart from molding of a viscous material and fixed by cooling and
these advantages molding offers more: By creating curing, while in PL the resist is patterned by selective lo-
a three-dimensional (3-D) resist pattern by mechani- cal chemical modification of a positive or negative resist
274 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b)
Hard stamp
Transparent
Liquid resin Tg
Hard stamp Tg
Solid polymer
Expose through
Substrate stamp
Tg
Tg
Viscous polymer Gap 0.5 bar
80 bar
Tg
Crosslinked
Solid polymer
polymer
Fig. 9.3a,b Schematic of NIL process: (a) thermal NIL (hot embossing) and (b) UV-NIL. In both cases a thickness profile
is generated in the thin polymer layer. After removing the residual layer, the remaining polymer can serve as a masking
layer which can be used as a resist for pattern transfer
by exposure and wet development. The two main NIL gies, and to establish standard processes which can be
methods are outlined in Fig. 9.3. For lithographic appli- scaled up to common wafer sizes.
cations, as needed in microelectronics and hard disks,
NIL is in competition with other emerging patterning 9.1.1 Next-Generation Lithography
techniques. Its success will mainly depend on the abil-
ity to solve processing issues such as resolution and With its integration into the International Technology
throughput. It is also important to develop reliable tools Roadmap for Semiconductors (ITRS) on NGL in 2003
with a long lifetime, which are available and can be used for the 32 nm node and beyond, NIL has become more
in combination with other cleanroom process technolo- than a simple high-resolution method [9.14] (Table 9.1).
Table 9.1 ITRS roadmap showing the resolution of different lithographic patterning techniques, with focus on large area,
parallel techniques, and practical and actual resolution limits for different lithography methods (after [9.15], revised and
updated (state of the art 2009))
Lithography type Practical resolution limit (nm) Ultimate resolution limit (nm)
UV-proximity photolithography (365 nm) 2500 125 (hard contact)
Deep-UV projection (DUV, 193 nm) 45 20–30 (immersion)
EUV projection (soft X-rays, 13.6 nm, with reflective mask) 45 20–30
EUV interference lithography (with diffraction grating) 20 10
X-ray proximity (0.8 nm, 1 : 1 mask) 70 10
Electron beam (low-energy beam arrays) 40 –50 Resist: 7 –20
Ion beam projection (mask-less patterning) 25 20
Part A 9.1
It is now considered a candidate for replacing or com- must satisfy the requirements for full compatibility,
plementing advanced optical lithographic methods for similar specifications, yield, and throughput. The in-
the fabrication of processors and solid-state memory vestments are expected to be lower than for the current
chips, which over the years have been developed and frontrunners: extreme-ultraviolet (EUV) lithography or
pushed to higher resolution with a vast investment of parallel electron-beam exposure.
resources. Over more than 40 years, Moore’s law has
described with amazing accuracy the reduction of fea- 9.1.2 Variants of Nanoimprint Lithography
ture size (and cost per transistor), and therefore serves
as a roadmap for the developments needed for future Molding of Thermoplastic Resists
microchips [9.15–17]. This development is driven by by Thermal NIL
economic considerations, and leads to competition be- NIL was first reported as thermoplastic molding [9.20–
tween different candidate fabrication methods. These do 22], and is therefore often referred to as thermal NIL
not only have to provide the resolution of the smallest (here also named NIL or T-NIL) or hot embossing
feature size (node), but also satisfy issues such as align- lithography (HEL) [9.7, 22, 23]. The unique advantage
ment (overlay of several masking levels), critical dimen- of a thermoplastic material is that the viscosity can
sions (CD), simple mask fabrication, high throughput be changed to a large extent by simply varying the
(mass fabrication), and low cost of ownership (CoO, temperature. Figure 9.5 shows viscosity plotted against
e.g. no dependence of expensive machines such as syn- temperature for various thermoplastic polymers, i. e.,
chrotrons, back-ups and tool and mask redundancy), poly(methyl methacrylate) (PMMA) and polystyrene
which become increasingly difficult to meet if smaller (PS) with different molecular weights, and some com-
exposure wavelengths have to be used (Fig. 9.4).
The financial and physical barriers to these tech-
Zero-shear viscosity η0 (Pa s)
niques are now so great that alternatives such as NIL
1010
are considered as a way out of this spiral of rising in-
mr-I 8000E
vestments for next-generation chips with even smaller 109
feature sizes. This means that all technical issues con- mr-I 8000
nected with NIL for integration into chip manufacturing 108 mr-I
7000E
107
2
Exposure rate (cm /s)
102 106
g-line (436 nm)
KrF PL 105
101 EUV (248 nm) i-line (365 nm)
PS 353 k
(13.6 nm)
0 ArF 104 mr-NIL
10 F2 (193 nm) 6000
X-ray PMMA
(157 nm) DUV 103
–1
UV-NIL Cell 75 k
10 PMMA
e-beam PS 58 k 25 k
Large Ion
area 102
beam 60 80 100 120 140 160 180 200 220 240 260
10–2
T-NIL NIL Temperature T (°C)
Single Fig. 9.5 Graph showing zero-shear viscosities for some standard
10–3 point
EBL resists for thermal NIL for different polymers, taken from differ-
10–4 ent sources: PMMA with Mw of 25 and 75 k [9.18], PS of 58
SPM
and 353 k [9.19], and the commercial resists mr-I 7000E, 8000,
10–5 8000E, and mr-NIL6000 [9.11], showing the potential of rheology
10 100 1000 and of the large variation of viscosity of thermoplastic polymers
Resolution (nm)
Part A 9.1
Table 9.2 Comparison of hot embossing (NIL) and UV imprinting (UV-NIL), with typical parameters of current pro-
cesses
Type of NIL (properties) Thermal NIL (hot embossing) UV-NIL (hard stamp)
Basic process sequence (Fig. 9.6) 1) Spin-coat thermoplastic film 1) Dispense liquid resin
2) Place stamp on film 2) Parallel alignment of stamp with defined gap
3) Heat until viscous 3) Imprint at low pressure
4) Emboss at high pressure 4) Expose with UV light through stamp
5) Cool until solid and crosslink
6) Demold stamp 5) Demold stamp
Pressure p 20–100 bar 0 – 5 bar
Temperature Tmold 100–200 ◦ C 20 ◦ C (ambient)
Temperature Tdemold 20–80 ◦ C 20 ◦ C (ambient)
Resist Solid, thermoplastic Liquid, UV-curable
Tg ≈ 60–100 ◦ C
Viscosity η 103 –107 Pa s 10−2 –10−3 Pa s
Stamp material Si, SiO2 Glass, SiO2
opaque transparent
Stamp area Full wafer, > 200 mm diameter 25 × 25 mm2 , limited by control of gap
Stamp contact Facilitated by bending Planarization layer
Embossing time From s to min < 1 min (per exposure)
Advantage Low-cost, large-area equipment and stamps Low viscosity, low pressure,
alignment through stamp
Challenge Process time, Step and repeat needed for large areas
thermal expansion due to thermal cycle
Part A 9.1
sists [9.24–27]. The basic difference between UV-NIL thickness used. In a second step, the thickness profile
and NIL is that a resin that is liquid at room tem- of the polymer film can now be used as a resist for pat-
perature is shaped by a moderate pressure, and by tern transfer. For this, the residual layer remaining in
exposing light through the transparent stamp the resin the thin areas of the resist has to be removed, which is
is cross-linked and hardened. The stamp either sinks done by homogeneously thinning down the entire re-
down to the substrate or must be kept at a constant sist using an (ideally) anisotropic etching process. In
distance from the substrate during both filling and ex- this way, process windows are opened to the substrate
posure, due to the low resist viscosities. The mechanical and the polymer can be used as a masking layer for fur-
setup has to be able to compensate for wedge errors ther processing steps. There are an increasing number
in a low-imprint-pressure process. Patterning on nonflat of process variations, which are mostly variants of the
substrates or over topography therefore requires a pla- established thermal NIL and UV-NIL processes, partic-
narization strategy and often small stamps. Because of ularly those using special methods of pattern transfer
the small pressures used, both hard stamps or stamps (e.g., reversal imprint) and hybrid processes (combi-
with protrusions made by soft material on a rigid back- nations of different processes). All the processes have
bone can be used. their specific advantages, e.g., while UV-NIL can be
performed at room temperature, hot embossing is low
Resist Window Opening for Pattern Transfer cost since nontransparent stamps can be used. The ma-
A basic characteristic of NIL is the patterning of a thin jor characteristics of typical processes, along with those
layer of material, in which the dimensions (lateral of hybrid approaches, are summarized in Table 9.2 and
structure size and height) become similar to the film presented in more detail in [9.11].
properties can be changed without affecting the chem- by their chemical composition and physical proper-
ical ones. While squeeze flow governs high-viscosity ties. In the case of molding these are often long-chain
molding (where pressure is the driving force to displace molecules with molecular weight Mw . The polymer
the viscous material), in UV-NIL low pressure or even Mw is important because it determines many physical
mold filling by simple capillary action (where surface properties. Some examples include the temperatures for
energy controls the wetting and spreading of the vis- transitions from liquid to viscoelastic rubber to solid,
278 Part A Nanostructures, Micro-/Nanofabrication and Materials
and mechanical properties such as stiffness, strength, 500 kg/mol (also denoted 500 k), is normally used for
viscoelasticity, toughness, and viscosity. However, if the EBL, since the development contrast between exposed
Mw is too low, the transition temperatures will be too and unexposed areas increases with Mw [9.33, 34].
low and the mechanical properties of the polymer ma- A lower Mw , of some tens of kg/mol, is patterned
terial will be insufficient to be useful as a hard resist for in NIL, due to the strong increase in temperature-
pattern transfer. The examples given in this section are dependent viscosity with Mw [9.35]. Apart from their
simple and meant to illustrate the specific terms needed mobility it is expected that shorter chains, which in the
to understand polymer behavior in molding. case of amorphous polymers are normally present as en-
It has been known for a long time that polymers tangled coils, can move more easily into small mold
can replicate topographies with high fidelity. Up to now cavities. A convenient way of expressing the size of
5 nm resolution of polymer ridges with a pitch of 14 nm a macromolecule present as a statistical coil aggregate
has been demonstrated [9.32]. In contrast with meth- is by its radius of gyration Rg , which is calculated from
ods such as electron-beam lithography (EBL), where the statistical mean path of the chain in a random-walk
nanoscale chemical contrast can be produced by lo- model using a self-avoiding walk. Rg can be measured
cal irradiation-induced chain scission, polymer chains directly in experiments by small-angle neutron scatter-
are only moved and deformed during molding, thus re- ing [9.36]. It can also be defined not only for a linear
taining their chemical properties such as Mw . Molding chain but also for polymers with branched structure, etc.
topographic details down to a few nanometers means It also equals the square of the average distance between
that single polymer chains have to deform or flow. This the segments and the center of mass of the poly-
deformation can be illustrated by comparing the poly- mer [9.28], which means it can be used to give a rough
mer with a pot full of cooked spaghetti, and instead estimate of the mean distance between different coils.
of the viscosity change with temperature we simply Since entire coils are both moved and deformed, Rg will
take the different mobility of the filaments when wet only give a rough estimate of the achievable minimum
or dry. When a water glass, representing the 10 nm pil- resolution of a pattern in an amorphous polymer film.
lar stamp shown in Fig. 9.2a, is pressed into this pot, As an example we take a PMMA macromolecule with
single spaghetti filaments have to be moved before the Mw of 25 kg/mol. Here the chain contains N = 250
glass can sink into the entangled network. If the poly- MMA monomer elements [C5 H8 O2 ] with a weight
mers can slide along each other, the deformation can of 100 g/mol each and has a total length of about
be permanent after drying and demolding. If stress is L = 80 nm. Both with simple considerations based on
frozen, the matrix around the cylindrical hole will re- the volume of a single molecule in the bulk PMMA and
lax after demolding. Note that this simple example can formulas for the random walk [Rg = (N/6)1/2 · (L/N)],
also be used to illustrate the difference between totally a Rg value of 2 nm can be calculated.
amorphous and semicrystalline polymers. A polymeric liquid, whilst retaining the properties
A polymer is a large molecule made up of of a liquid, follows a rubber-like elasticity. An exam-
many small, simple chemical units, joined together ple is the melted cheese on a pizza. If melted cheese
by chemical reaction. For example, polyethylene is dripped vertically, it flows slowly, just like a li-
[CH3 −(CH2 )n −CH3 ] is a long chain-like molecule quid. However, if it is pulled and then the tension
composed of ethylene molecules [CH2 =CH2 ]. Most removed, melted cheese will contract just like rubber.
artificially produced polymers are a repetitive se- In other words, although melted cheese is a liquid, it
quence of particular atomic groups, and take the form also has elasticity. Substances like this, which have both
[−A−A−A−]. The basic unit A of this sequence is viscous and elastic properties, are called viscoelastic
called the monomer unit, and the number of units n substances. In order to calculate the flow of a fluid when
in the sequence is called the degree of polymerization. an external force is applied, we need an equation relat-
The molecular weight Mw of a polymer is defined by ing the stress in the fluid to its deformation. This type
the weight of a molecule expressed in atomic mass of equation is called a constitutive equation. For ex-
Part A 9.2
units (amu). It may be calculated from the molecular ample, if a polymeric liquid undergoing a steady flow
formula of the substance; it is the sum of the atomic is stopped, the stress does not immediately become 0,
weights of the atoms making up the molecule. For ex- but decays with a relaxation time τ. Here τ depends
ample, poly(methyl methacrylate) (PMMA), a classic strongly on the Mw of the polymer and the tempera-
resist material, exhibits very good resolution for both ture, and can be on the order of several minutes to hours
EBL and NIL. A high-Mw PMMA, typically above in some cases.
Nanoimprint Lithography – Patterning of Resists Using Molding 9.2 Nanoimprint Process 279
Demolding
Stamp molding, and (2) after demolding. In the case
wi wi +1 wi wi +1
hr s0 si sN = s0 si of viscous molding, where volume conser-
vation can be anticipated, the residual layer
h0 Thermoplastic polymer hr thickness can be calculated from geometrical
hf
Hard substrate parameters such as the initial film thickness
and the size and density of cavities
280 Part A Nanostructures, Micro-/Nanofabrication and Materials
stamp geometry can therefore be optimized by reducing total volume of the cavity array. This simplification can
the dimensions of the protrusions. While stamps with be used for the calculation of embossing times. The fill
nanopillar arrays, as shown in Fig. 9.2, would allow fast factor should be kept constant, both locally (at length
embossing of some microseconds, using standard NIL scales corresponding to the cavity dimensions) and also
process parameters, already protrusions of some hun- across the wafer, i. e., for large stamp protrusions, to en-
dreds of microns would increase embossing times to sure better flow of the polymer and shorter embossing
Nanoimprint Lithography – Patterning of Resists Using Molding 9.2 Nanoimprint Process 281
ing. As seen before, this is a result of the molding Figure 9.10a shows such a case, in which a grating area,
process slowing down due to the squeeze flow. For typically of some square millimeters, is surrounded by
many applications, when pattern transfer has to be a large nonstructured area. In the ideal case of a totally
achieved after the embossing, it is important to deter- rigid stamp, the final thickness would be determined by
mine the final residual thickness h f of this polymer the fill factor of the grating averaged over the whole
layer (Fig. 9.8) before the next process step. Further- stamp area, which could be calculated by the simple rule
282 Part A Nanostructures, Micro-/Nanofabrication and Materials
of conservation of polymer volume. This can only be a thin PDMS relief coated on a harder substrate – is par-
achieved if the polymer can flow easily over large dis- ticularly useful in full wafer concepts. It combines the
tances; otherwise, parts of the grating will not be filled. complementary mechanical properties of a soft surface
In the other extreme case of a totally flexible stamp relief for the achievement of local conformal contact
and low lateral transport of polymer, both stamp areas and a rigid but bendable backbone, which can be used
could be calculated independently. While the center of for mounting and alignment. A process working with
the grating would sink to half of the depth of the cavi- moderate resist viscosities (with η0 = 50 mPa s and be-
ties (assuming a fill factor of 50%), in the nonstructured low) for providing liquid films by spin-coating has been
area almost no sinking would occur. In between, at the developed and can be applied at reduced environmental
border of the grating, the stamp tries to accommodate pressure [9.69, 83].
this mismatch by bending. Depending on the thickness
and elastic behavior of the stamp, as well as the design 9.2.4 Demolding
of the stamp, characteristic distances can be calculated.
In many cases rules of thumb for design and process During demolding the rigid stamp is detached from
optimizations are sufficient for achieving homogenous the molded structure, which can be done in a parallel
molding. However, for more complex cases, simulations way when using small rigid stamps, or by delamination
are needed to predict the filling of both small and large if thin wafer-like substrates are used. If fully molded,
structures in the vicinity of one another. Furthermore the the thickness profile in the resist exhibits the inverse
dynamic behavior of filling has to be taken into account. polarity of the relief of the stamp surface. The de-
The task becomes even more challenging if emboss- molding process, also called de-embossing, is normally
ing over topography has to be considered. In this case, performed in the frozen state, i. e., when both the mold
a planarization layer can be used below the NIL resist. and molded material are considered solid. For thermo-
plastic materials this happens at temperature well below
Resist Density Adaptation Tg , but high enough that frozen stress due to thermal
in Low-Pressure Imprinting contraction does not lead to damage during demolding.
UV-NIL processes are performed at room temperature, In cases in which the resist is cured before demold-
at which resist precursors are present as liquid films or ing, i. e., cross-linked by exposure or heat, demolding
droplets. When using hard stamps as in step and flash can take place at temperatures similar to the molding
imprint lithography (SFIL, a step and repeat UV-NIL temperature. A successful demolding process relies on
process [9.26, 27]), or jet-and-flash imprint lithography a controlled balance of forces at the interfaces between
(JFIL, for single step wafer scale imprint), a homo- the stamp, substrate, and molded polymer film. There-
geneous residual layer thickness can be achieved by fore mechanical, physical, and chemical mechanisms
locally varying the amount of liquid resin necessary responsible for adhesion have to be overcome. The fol-
to fill the cavities of the stamp. Particularly suitable lowing effects have to be avoided or reduced [9.11, 18,
for this is an array of droplets formed by dispensing 53, 80, 81] (Fig. 9.11):
low-viscosity UV curable monomer (with η0 ≤ 5 mPa s)
onto the substrate surface prior to imprinting. By • Undercuts or negative slopes in the stamp may lead
contact of the stamp with the dense droplet array, a con- to mechanical interlocking of structures, which in
tinuous film is formed by capillary action. To handle the frozen state are elastically elongated and de-
pattern density variations, the drop-on-demand UV-NIL formed before ripping. Sidewalls with positive or
process at atmospheric environmental pressure has been at best vertical inclination are a prerequisite for de-
developed [9.82]. molding without distortion.
• Friction due to surface roughness may occur during
Soft Lithography with Conformable Stamps the sliding of molded structures along vertical cavity
in Low-Pressure Imprinting walls. The effect of this can only be overcome if the
Part A 9.2
The forces on a stamp protrusion with liquid resists are surface of the molded material is elastic and enables
induced by capillary action rather than by squeeze flow gliding of the wall without sticking.
and are therefore low. Therefore in UV-NIL, compliant • The enlarged surface area of the patterned stamp
stamps made from elastomeric materials, e.g., poly- leads to an increase of hydrogen bridges and van der
dimethylsiloxane (PDMS), a UV-transparent rubber, Waals forces, or other chemical bonding effects due
can also be applied. The concept of layered stamps – to ionic, atomic, and metallic binding. This effect
Nanoimprint Lithography – Patterning of Resists Using Molding 9.2 Nanoimprint Process 283
More information about curing and multilayer resists underlying substrate are generated. This is also
can be found in Sect. 9.3. called a window-opening or breakthrough etch. In
the second case the thickness contrast of the remain-
9.2.6 Pattern Transfer ing polymer is used to mask the substrate against the
etching medium. Both processes have to be highly
In many cases the lithographic process is only complete anisotropic, i. e., during the transfer step the lateral
when the resist pattern is transferred to another ma- size of the structure has to be preserved, including
terial. This process, in which the resist is transformed the slope of the original pattern. Apart from open-
into a patterned masking layer, allows the substrate to ing windows using reactive-ion etching (RIE), other
be attacked by plasma, etching solutions, electroplat- pattern transfer strategies have been found which
ing, deposition of materials, and other substrate-altering circumvent the residual layer problem.
processes. A unique advantage of molding instead of • Lift-off is a patterning technique adding thin lay-
exposure is that complex stamp profiles, such as stair ers of a solid material (e.g., metal) locally to the
cases, V-grooves, and pyramids, both convex and con- window openings in the resist [9.107–114]. Un-
cave, can be replicated. They can be used for the dercuts, as can be generated in PL and EBL, are
generation of 3-D structures such as for T-gate transis- a prerequisite for good lift-off. However, in NIL,
tors or contact holes, or serve for stepwise etching of where sidewalls are at best vertical, a high thickness
underlying layers with variation of the opening width. contrast (aspect ratio) of the structures is needed.
As long as undercuts and 3-D patterning are not nec- Lift-off resists are a means to generate defined un-
essary, in most cases this pattern transfer is therefore dercuts using a bilayer resist system, by selectively
similar to in EBL. However, in this section we em- dissolving a sacrificial bottom layer through the
phasize methods where NIL has some specific process structured openings of a top layer.
advantages over conventional lithographic methods, or • Electroforming and electroplating, like lift-off, are
where the use of NIL implies some major changes in the processes that add material to the areas not cov-
fabrication process or properties of the devices: ered by the resist [9.97, 115, 116]. Electroforming
provides a good alternative to the lift-off process
• In NIL, etching is used for both the removal of the because metal structures can be generated with
residual layer and the pattern transfer of the resist considerable height and good surface quality. If
pattern to the underlying substrate [9.7, 97–106]. a conductive seed layer is deposited below the resist,
In the first case the polymer layer has to be ho- during electroplating the metal layer starts to grow
mogeneously thinned down until openings to the from within the window regions and conforms to the
Stamp Electroforming
with nanorelief
PMMA SiO2 ≈ 20 nm
Imprinting Si or SiO2
Fig. 9.13a,b NIL and lift-off for
Demolding Fluorinated silane
≈ 50 nm 0 nm 0V the generation of nanopatterns
100 nm
HP = 35.5 nm with chemical contrast. (a) Process
RIE etching scheme for local silane deposition
3 nm 0.15V
Deposit from gas phase and (b) AFM/LFM
Part A 9.2
a) b) c) d) e)
Transfer layer Etch barrier
Release layer solution Curred etch barrier
Template Residual layer
UV
Part A 9.2
Fig. 9.15a–e Process scheme of UV imprinting and pattern transfer, using a double layer (also called direct SFIL). The
molded top layer, also called the etch barrier, is coated on a transfer layer, which serves as a planarization layer. It has
also antireflective properties for the UV exposure through the stamp. (a) dispensing of viscous resist droplet, (b) imprint,
(c) UV-exposure and curing, (d) demolding of hardened resist, (e) residual layer etch and transfer into bottom layer
(breakthrough etch/window opening)
Nanoimprint Lithography – Patterning of Resists Using Molding 9.2 Nanoimprint Process 287
b) 5)
Bottom layer Seed layer
0) (with wires) deposition
1) Stamp
(transparent)
Resist 6) Metallization
bottom layer (copper
(with seal) electroplating)
2) Molding and 7)
exposure of Metal
resist through thinning
stamp (CMP
process)
is used for pattern transfer. The tone can be preserved total, the reduction from 128 process steps down to 56
if another tone reversal process is used. This can be results in a cost reduction that justifies the introduction
achieved by imprinting a pattern into the thick trans- of a new technology and serves as an example that the
fer layer, and by spin-coating a silicon-containing resin 3-D pattern capability can be a decisive argument over
on top of it. If the top residual layer of the pla- resolution for the introduction of NIL into chip manu-
narized film is etched away, the high etch resistance facturing. Figure 9.16 shows the pattering scheme for
of the silicon remaining in the trenches of the bottom one level of the contact layer of an IBM power PC mi-
layer will enable the patterning of the transfer layer croprocessor. Obducat has used a similar process for the
with reversed tone. This strategy has the advantage generation of micrometer-sized contact holes in printed
that stamp contamination containing silicon residues is circuit boards (PCB).
avoided [9.136].
The 3-D patterning capability of NIL makes it pos- 9.2.9 Reversal NIL
sible to reduce the number of process steps in contact
layer fabrication of microchips by using innovative pat- In contrast to NIL, in reversal NIL the resist is pat-
tern transfer. The connection of the transistors is done terned either directly onto the stamp or onto an auxiliary
using several levels of lateral wires, each contacted ver- substrate, e.g., by spin-coating, casting or imprint, and
tically by through-holes. This contact layer of a chip is then transferred from the mold to a different substrate
fabricated using lithography and copper electroplating. by bonding. Thus patterned resist structures are ob-
For the wiring scheme of a chip, as shown in Fig. 9.16, tained as in direct NIL, and even embedded channels
eight levels of wiring layers are needed, each of which can be created. The concept is well presented in [9.139–
is done in a so-called dual hard damascene process. 143]. In reversal NIL it is possible to transfer patterns
A process has been proposed which reduces the num- onto substrates that are not suitable for spin-coating or
Part A 9.2
ber of process steps necessary for one level from 16 to have surface topographies. However, complete trans-
7 [9.137]. A two-tiered stamp with three height levels fer does not only depend on a good balance of the
makes it possible to pattern the through-holes as well as surface energies, but also on the pattern density and
the wires in one step [9.138]. In this way, several ex- roughness of the structures. As an example, embed-
posure steps can be replaced by a single imprint with ded channels generated by reversal NIL are shown in
patterns of different residual polymer layer thickness. In Fig. 9.17 [9.140].
288 Part A Nanostructures, Micro-/Nanofabrication and Materials
mr-I 8000 [9.144] and E Safe solvent 115 (170 –190) 7 × 104 (E grade n = 1.415, NIL resist with thermal properties
at 180 ◦ C) similar to PMMA, but higher etch resistance
mr-I 9000 [9.144] and E Safe solvent 65 (140 –160) n = 1.417, thermocurable NIL resist [9.90]
NEB22 [9.158] PGMEA 80 (95 –130) Negative EBL resist based on poly(hydroxystyren),
[9.159] [9.160] high etch resistance in fluoro- and chloro-based
plasmas [9.150], low Mw (3k)
Nanoimprint Lithography – Patterning of Resists Using Molding 9.3 Tools and Materials for Nanoimprinting 289
for a specific application. Both the processing proper- mized for greater etching resistance or better flow at
ties as well as those for the final application purpose lower temperatures. In Table 9.3 we give an overview
have to be considered. Many of the resists, as used of NIL resists with references to further information
for PL and EBL [9.161, 162], exhibit thermoplastic be- on these materials. Further information can be found
havior. A typical example is PMMA, a regular linear in [9.90, 91, 163–170].
homopolymer, with a short side-chain. It is used as UV-curable NIL materials are composed of a mix-
a high-resolution standard material for EBL and also as ture of monomers (or prepolymers) and a suitable
a bulk material for hot embossing and injection mold- photoinitiator, and often chemicals are added which
ing. For a long time it has been known that sub-10 nm decrease the effect of radical scavengers on photopoly-
resolution can be achieved [9.35]. PMMA is a low- merization [9.11, 48, 171–178]. Immediately during
cost material, and available with different Mw values. contact of the stamp with the liquid mixture, filling
It is compatible with other cleanroom processes, ex- of the mold starts by capillary forces, which pulls the
hibits good coating properties using safer solvents, and stamp towards the substrate. Therefore, the general
can be coated from solution to a thickness ranging from strategy is that low viscosities are needed for both rapid
20 nm to several μm. It has well-characterized optical, dispensing and filling of mold cavities. Thin resin lay-
mechanical, and chemical properties, and proved relia- ers on top of a thicker transfer layer are used to achieve
bility in many different applications. When used as an
etching mask, e.g., for Si, it exhibits a sufficient, but
not high etching resistance. The glass-transition tem- a)
perature Tg of PMMA (105 ◦ C) is low enough to enable
molding at temperatures below 200 ◦ C, but high enough
to ensure sufficient thermal stability in etching pro-
cesses. Acrylate-based polymers can also be used with
cross-linking agents. A further advantage of PMMA b) Grating
is that the process window, defined as the temperature
range between the lower temperature for viscoelastic
molding where relaxation due to frozen-in strain has
to be expected and the higher temperature where the Cavities Si
viscosity is so low that the onset of capillary bridges 1 µm
(viscous fingering) will affect the residual layer homo-
Part A 9.3
geneity [9.19], is quite large. This enables imprinting Fig. 9.17a,b Reverse microfluidic channels fabricated by
to be optimized by using tradeoffs between differ- double-sided imprinting: (a) 3-D schematic of a re-
ent parameters according to Stefan’s equation. Apart sist with a top grating and embedded channels. SEM
from PMMA, during the first 10 years of NIL, a num- micrographs of cross-sections of imprinted nanofluidic
ber of resists have been developed and characterized; channels: (b) 3000 nm (width) × 200 nm (height) channels,
they exhibit different Tg values, and have been opti- with a 700 nm-pitch grating on top (after [9.140])
290 Part A Nanostructures, Micro-/Nanofabrication and Materials
a homogeneous film thickness. Cross-linking and pho- stability (lifetime and wear), thermal expansion coeffi-
topolymer conversion is adapted to achieve high curing cient and Poisson’s ratio (dimension mismatch leading
speed and high etch resistance in the following break- to distortions during demolding), roughness (higher de-
through plasma etching process. molding force and greater damage), Young’s modulus
In UV-NIL a chemical reaction between the stamp (bending), and notch resistance (lifetime and handling).
and resist cannot be excluded. Small feature sizes along Issues related to fabrication are processibility (etching
with high silicon content and a large degree of cross- processes, selectivity, cleanroom environment) and sur-
linking make any residual imprint polymer left on the face quality (resolution). Use of a stamp material in
mold almost impossible to remove from the template a NIL process is determined by additional properties
without damaging the expensive quartz template. It such as transparency, conductivity, antisticking prop-
has been shown that a fluorosilane release layer ap- erties (with or without an antiadhesive coating, e.g.,
plied to a UV-NIL stamp undergoes attack by acrylate, a covalent coating), availability and cost (standard ma-
methacrylate, and vinyl ether UV-curable resist sys- terials and sizes, tolerances, processing equipment and
tems, indicating that its degradation is intrinsic to the time), and how easy it is to employ in NIL (e.g., fix-
chemistries involved. Future resist chemistries have to ing by clamping, thermobonding, gluing). In Table 9.4
satisfy the criterion of low reactivity toward antiadhe- we give a brief overview of the mechanical and ther-
sive coatings and stamp materials [9.179, 180]. mal properties of materials used for stamps. Further
information can be found in [9.27, 100, 156, 181–194].
9.3.2 Stamp Materials
9.3.3 Stamp Fabrication
Not only the mechanical but also the optical and chem-
ical properties are important when choosing a stamp Any kind of process generating a surface profile in
material for NIL. Critical mechanical parameters and a hard material can be used to fabricate stamps for NIL.
their implications for NIL are hardness and thermal The most common lithographic processes are based
on resist patterning with subsequent pattern transfer.
Therefore the requirements for these processes such as
resolution, aspect ratio, depth homogeneity, sidewall
roughness, and sidewall inclination are similar to the
processes presented before in this chapter. For highest
resolution, both serial and parallel fabrication methods
are available, however, with different area, through-
NIL NIL put, and freedom of design. The processes are standard
0th generation 1st generation 2nd generation
stamp(positive) stamp(negative)
processes for nanolithography, which also can be used
stamp(positive)
directly for patterning. When using them for the fabri-
cation of stamps, apart from higher throughput, greater
flexibility and reproducibility can be achieved. Using
stamp copies instead of the original is a way to enhance
Low aspect ratio Low aspect ratio High aspect ratio the lifetime of a stamp, simply because the original
is reserved for the copying process. There are differ-
Sieve device ent methods to generate copies from hard masters with
3rd generation
(negative) proved resolutions below 100 nm:
Fig. 9.18 Process chain from stamp origination to application: the thousands of molding cycles without significant
example of a porous membrane chip as shown in Fig. 9.14. The low- wear. The original, a patterned resist or etched relief
aspect-ratio original stamp fabricated by EBL and RIE is transferred on a glass master, is often lost during the transfer to
into a high-aspect-ratio stamp by two consecutive NIL copying nickel, therefore only after a first-generation nickel
steps, providing increased lifetime of the original and greater flexi- copy is drawn can further generations be repeatedly
bility copied from it.
Nanoimprint Lithography – Patterning of Resists Using Molding 9.3 Tools and Materials for Nanoimprinting 291
• Using the hard master with an etched surface relief sidewall roughness should be elastically absorbed by
directly as a mold is a straightforward approach if the molded material, while the surface maintains its an-
the mechanical setup allows or favors the use of sil- tisticking properties. Because the molded polymer film
icon (Si) or fused silica (SiO2 ). Stamp copies can be is squeezed between the two surfaces of stamp and sub-
fabricated using NIL and subsequent pattern trans- strate, they need to exhibit opposite surface properties.
fer (Fig. 9.18). Molds made from silicon wafers are The adhesion at both interfaces must be different to an
well suited to use as stamps in NIL, and have even extent that, while the polymer film adheres perfectly to
shown their mass-fabrication capability in CD in- the substrate surface, the stamp can be separated from
jection molding. For UV-NIL such as SFIL, molds the structures without any damage at any location of the
were successfully made using standard mask blanks stamp. If the stamp material does not exhibit good an-
(fused silica). tisticking properties to the molded material, the stamp
• As a third solution a polymer or sol–gel layer with has to be coated with a thin antiadhesive layer. A low-
an imprinted surface relief can be directly used as surface-energy release layer on stamp surfaces not only
a replication tool. This is possible if the thermo- helps to improve imprinting quality, but it also signif-
mechanical replication process does not exert high icantly increases stamp lifetime by preventing surface
forces on the relief structure. Resist hardened by contamination. An antiadhesive coating has to be chem-
light, heat or by chemical initiation may support ically inert and hydrophobic but at the same time allow
high temperatures and can be used repeatedly in filling of the mold cavities when the polymer is in its
NIL. However, the lifetime of polymeric molds is viscous state.
still low, and good solutions for antiadhesive coat- One of the major advantages of using Si or SiO2
ings have to be found. stamps for NIL is that they can be coated with anti-
• Hybrid molds use different materials for the surface sticking films using silane chemistry. Damage to the
relief and the support. They consist of a sub- molded structure during demolding is highly depen-
strate plate as a mechanical support covered with dent on the quality of the antiadhesive layer. Fluorinated
a thin polymer layer with nanostructured relief. trichlorosilanes with different carbon chain lengths are
In the case of NIL they have the advantage that commonly used due to their low surface energy, high
a substrate material can be chosen with thermome- surface reactivity, and high resistance to temperature
chanical properties adapted to the substrate to be and pressure. They support multiple long embossing se-
patterned. Furthermore this approach is useful if quences with repeated temperature cycles higher than
thin flexible substrates are needed. 200 ◦ C. Currently it seems that, as long as mechanical
abrasion can be avoided, the silanes match the normal
The methods differ mostly in the properties of the use lifetime of a Si stamp, which is some tens of cy-
materials used for the stamps (mechanical robustness, cles for NIL in a laboratory environment or thousands
thermal expansion coefficient, transparency, fabrication if automated step-and-repeat imprinting processes or
tolerance) and the surface properties of the patterned injection molding processes are used. However, the low-
relief (antiadhesive coating possibility). Although for energy surface that a fluorosilane layer presents is not
many applications electroplating of metal molds is fa- unreactive, and it is rapidly and easily degraded during
vored because of their great flexibility and robustness use, particularly at high temperatures (above 200 ◦ C)
compared with silicon, the effort to fabricate high- and by chemical attack by the abundant free radicals
quality mold inserts with defined outlines is often only present in curable resists. Therefore not only the chem-
justified for production tools. istry of resists has to be taken into account to improve
the lifetime of stamps, but strategies such as recoating
9.3.4 Antiadhesive Coatings have to be considered. Apart from silicon wafers, which
have the advantage that they are suitable for standard
One of the most important tasks for NIL is to cleanroom processing, other materials to be used as NIL
Part A 9.3
provide stamps with good antisticking surface prop- stamps, e.g., nickel (Ni) shim or duroplastic polymers,
erties [9.195–202]. The stamp surface should allow can also be coated with silanes if an intermediate SiO2
the molded surfaces to detach easily from the mold, layer is deposited onto the materials. The silane coating
and once released, provide a low friction resulting in can be performed by immersion in a solution of iso-
a continuous vertical sliding movement without stick- octane, or by chemical vapor deposition (CVD), either
ing. Nanoscopic interlocking of structures caused by at ambient pressure by heating the silane on a hot plate
292 Part A Nanostructures, Micro-/Nanofabrication and Materials
a) b) c)
Part A 9.3
Fig. 9.20a–c Three examples of NIL presses. (a) Simple hydraulic press, with temperature-controlled pressing plates.
(b) Semiautomated, hydraulic full-wafer NIL press, based on an anodic bonder. (c) Automated step-and-flash UV-NIL
production tool
Nanoimprint Lithography – Patterning of Resists Using Molding 9.3 Tools and Materials for Nanoimprinting 293
a) b) c1)
1. Press Stamp
Polymer
Roller
Polymer
2. Lift
Substrate
Substrate
4. Press
3. Step c2)
Imprint +
RIE
Roller
5. Lift Polymer
Substrate
Fig. 9.21a–c Outline of the three most common types of NIL machines: (a) full-wafer parallel press, (b) step-and-repeat
press, and (c) two roll-embossing setups
previously structured areas should not be affected by cushion balances thickness variations due to both tol-
imprints in the close vicinity (e.g., reheating of already erances of the setup and the nature of the molding
molded resist over Tg in thermal NIL and cross-linking process. The latter is caused by the fact that the size
of resist outside the stamp area in UV-NIL). In PL, step- and shape of the stamp surface relief leads to local
ping was needed because of the limitation of the maxi- pressure variations during the squeeze flow and, if the
mum field size to be exposed, and because the continu- stamp can bend, to local differences in the sinking
ous reduction of structure sizes and diffraction effects velocity. When using thick polymer plates, for which
was only possible by optical reduction of the mask- molding leads to surface modulation of a bulk mater-
ing structures into the resist by high-resolution optics. ial, the cushion is formed by the viscous material itself.
Furthermore this enabled a noncontact process to be es- However, in NIL, a thickness profile has to be gener-
tablished, while 1 : 1 imaging of a mask structure would ated in a resist whose thickness is often lower than the
have lead to unwanted reduction of the proximity gap. thickness tolerances of the substrates and mechanical
setup used. Furthermore height defects in the range up
Single-Step Wafer-Scale NIL to some μm, such as dust particles, have to be equi-
Single-step NIL machines pattern the surface on an en- librated. Therefore the cushioning has to be achieved
tire wafer in one step. Thus the stamp must have the by the pressing mechanism, and its ability to compen-
same size as the wafer to be patterned. The simplest sate has to be larger than the defects and tolerances of
mechanism for full wafer imprinting is a parallel-plate the stamps and substrates. Lateral spreading and disper-
embossing system. A linear movement of the piston be- sion of the applied pressure can be achieved by using
hind the stamp leads to local thinning of the polymer a spring mechanism, which can consist of an additional
under the stamp protrusions, which is possible because plastic or elastic layer; e.g., a mattress made of rubber
the polymer is moved from squeezed areas into voids (silicone, polydimethylsiloxane (PDMS), Viton), poly-
in the stamp. This movement can be generated us- tetrafluoroethylene (PTFE, Teflon) or elastic graphite
ing pneumatic, hydraulic, or motor-driven pistons. The can be used. The thickness has to be chosen in order to
Part A 9.3
pressure must be maintained during the whole mold- achieve equilibration of a few micrometers, for which
ing process, until the voids are filled, and the molded some 100 μm are sufficient. Due to the high pressure
structures are fixed during the cooling or curing step, used in NIL, compensation of small wedges, i. e., non-
depending on the method used. However, under nor- parallel alignment, is not needed. The applied pressure
mal process conditions, embossing with a hard master of the large backing plate is then spread into infinites-
does not work without a cushioning mechanism. This imal small area elements behind the stamp, and is able
294 Part A Nanostructures, Micro-/Nanofabrication and Materials
resist film is patterned by means of electromagnetic ra- ing thin films of organic light-emitting materials and
diation, electrons or by mechanical deformation. Only polymers doped with laser dyes to create organic light-
a few steps in the process flow are different, for exam- emitting devices (OLED) [9.220,221] and lasers [9.155,
ple, the dry etch step to remove the 10–100 nm-thick 222, 223]. NIL is also suitable for nanoscale patterning
residual polymer layer after the imprint. Both additive of conducting organic films for cost-effective organic
and subtractive processes have been demonstrated, as electronics [9.224].
discussed in Sect. 9.2. Sometimes even the resist is the Within the rapidly growing field of lab-on-a-chip
same, for example, PMMA, which is a widely used re- applications [9.225], NIL offers an attractive, cost-
sist for both EBL and NIL. The advantages of NIL come effective method for molding of complex structures,
into play if high resolution is needed over a large area. integrating micro- and nanofluidics, optics, mechan-
For such applications, NIL is a cost-effective alternative ics, and electronics on a single chip [9.226]; for
to current cutting-edge lithography techniques such as example, the micro- to nanoscale fabrication capabil-
deep-ultraviolet (DUV) lithography [9.212], dedicated ities are used to create single-use polymer devices
to CMOS chip manufacturing. The cost of owner- containing nanopillar arrays [9.227] and nanofluidic
ship for NGL technologies, such as extreme-ultraviolet channels [9.228] for DNA separation and sequencing.
(EUV) lithography [9.213], is reaching a level that In this section we will give an overview of differ-
requires extremely high production volumes to be eco- ent fields of applications. We start with two examples
nomically viable. This development has already forced of pattern-transfer applications that are close to pro-
several branches of the electronics industry to explore duction: patterned media for HDD, and subwavelength
NIL as an alternative fabrication method. Examples of metal wire gratings for HDTV projectors. We then
such products are patterned media for hard-disk drives discuss a few examples of laboratory-scale potential
(HDD) [9.214, 215], surface acoustic wave (SAW) fil- high-impact applications of NIL. These examples were
ters for cell phones [9.27,216], and subwavelength wire selected from a large number of NIL applications. The
grid polarizers for high-definition TV (HDTV) pro- number of laboratory-scale NIL applications is rapidly
jectors [9.217]. Even the semiconductor industry was growing, reflecting a wealth of new possible device
considering NIL as possible NGL to deliver the 32 nm architectures becoming feasible by NIL. Some of the
node and beyond [9.14]. For chip manufacture the abil- applications are directly relevant for industrial produc-
ity to print smaller features sizes is the most important tion, and others are directed towards research. Even
issue, because NIL simply does not have the restrictions in research the nanostructuring capability of replica-
encountered by optical methods and already now of- tion processes are needed. Further insight into this field
fers a resolution higher than the next technical nodes. is given in Sect. 9.2.6 about pattern transfer and in
Among other the major technological challenges to be Sect. 9.5 about commercialization aspects of NIL.
solved are: overlay accuracy, low defect density, er-
ror detection in high resolution stamps and imprints, 9.4.2 Patterned Magnetic Media
fast imprint cycles, and critical dimension (CD) control. for Hard-Disk Drives
In addition to the high resolution, the NIL technique
also offers capability for 3-D or multilevel imprinting, Since the first demonstration of NIL, patterned mag-
when the stamp is patterned with structures of different netic media for HDD has been a key application for NIL
heights (Sect. 9.2.8). technology [9.229]. After the invention of the HDD in
The NIL process offers new possibilities to form 1957, the storage capacity, quantified in areal density
polymer devices with microscale to nanoscale fea- of bits, has been increased to the current (2008) level
tures. Nanoscale-patterned polymer films find a wide of 178 Gb/inch2 in data storage applications. The size
range of applications within optics, electronics, and of the individual bits, defined by local magnetization of
nanobiotechnology. The capability to form 3-D polymer a homogeneous (unpatterned) thin magnetic film, was
structures, with curved surfaces and high aspect ratio, reduced, and the bit density increased, by the applica-
Part A 9.4
paves the way for new classes of polymer-based passive tion of multilayer magnetic films as recording media;
optical devices, such as lenses and zone plates [9.126], the sensitivity of the read head was increased by ex-
photonic crystals (PhC) [9.100,218,219], and integrated ploiting the giant-magnetoresistance effect in multilayer
polymer optics [9.147]. The NIL technique allows for thin-film conductors [9.230]; and the magnetization was
choosing a wide range of polymers with optimized applied perpendicular to the surface of the recording
optical properties [9.153, 219], and allows for pattern- media, while microelectromechanical systems (MEMS)
296 Part A Nanostructures, Micro-/Nanofabrication and Materials
technology for the mechanical parts has been developed mer was removed by oxygen plasma, and the disk was
to a rather extreme level: In current HDDs the read– then sputter-coated with a CrX/Co-alloy double-layer
write head flies at a height of 2–3 nm above the surface magnetic thin film. These devices were designed for
of the disk plate. An overview of HDD technology is in-plane, i. e., longitudinal magnetic polarization, but
given in [9.231]. This current level of storage density is DTR media for perpendicular polarization have also
projected to increase by three orders of magnitude over been realized by EBL and RIE etching of the magnetic
the next 10 years, in order to meet market requirements. film [9.234].
The possibilities to increase the bit density with The DTR media technology offers the possibility to
current technology, where bits are written by local mag- regain the loss in electrical signal-to-noise ratio, as the
netization of an unpatterned thin magnetic film, are magnetic bit size is reduced. However, with decreas-
mainly limited by the read–write width, the positioning ing bit size that is necessary to follow the roadmap,
of the magnetic head, and by thermal instability induced the technology will be limited by thermal instabilities,
by superparamagnetism in the grains of the magnetic or superparamagnetism. The magnetic film consists of
film. These challenges are addressed by patterning the small, weakly coupled magnetic grains, which behave
magnetic film. as single-domain magnetic particles. Each bit con-
Discrete track recording (DTR) media [9.232], sists of the order of 100 grains (domains with single
where the magnetic film is patterned with a spiral land crystalline orientation) to obtain a reasonable signal-
and groove track, have been developed to overcome the to-noise ratio. In order to keep this ratio of grains per
problems associated with the read–write width and po- bit, the grain size must be reduced with the bit size.
sitioning of the magnetic head (Fig. 9.23a). The idea The magnetic energy of a single grain scales with the
of DTR media is more than 40 years old [9.232], but volume of the grain. This implies that the bit can be
has not been implemented in production due to the lack erased thermally, when the grain size becomes suffi-
of a nanolithography process that meets the demanding ciently small and weakly coupled to neighboring grains.
requirements for the surface smoothness of the disk sur- This is referred to as the superparamagnetic limit.
face [9.233] and that is suitable for large-scale low-cost The superparamagnetic limit can be overcome
fabrication. Researchers at WD Media (formerly Ko- by lithographically defining each bit, as a magnetic
mag Inc.) have demonstrated a cost-effective process for nanoparticle, or nanomagnet [9.230, 235, 236]. In such
volume manufacturing of DTR media, based on double- a quantized magnetic disk [9.235] each magnetic
sided thermal NIL with a commercially available resist nanoparticle is a single magnetic domain with a well-
and wet etching on a 95 nm-diameter nickel phospho- defined shape and uniaxial magnetic anisotropy, so the
rous (NiP)-plated Al:Mg disk [9.214]. The process steps magnetization only has two possible stable states, equal
are outlined in Fig. 9.23b. The nickel stamps with track in magnitude but opposite in direction, as illustrated in
pitches down to 127 nm, corresponding to an areal den- Fig. 9.24. Such defined bits can be thermally stable for
sity of 200 Gb/inch2 , were electroformed from a silicon sizes down below 10 nm [9.215].
master, which was patterned either by laser-beam or The feasibility of NIL for fabrication of patterns of
electron-beam writing, equipped with a rotating stage magnetic nanostructures for quantized magnetic disks
with radial beam positioning. After etching, the poly- has been investigated by several research groups, as re-
a) b) 1 Ni stamper
perpendicular to the grating) light GaN-based light-emitting diodes (LEDs) have large po-
tential as energy-efficient, long-lifetime, environmental-
d d
n 2p = n 21 + 1 − n 22 , friendly, and stable light sources, and are currently
a a entering a range of applications, such as full-color dis-
n2n2 plays and projectors, traffic lights, and automotive and
n 2s = d 2 1 2 d 2 ,
Part A 9.4
Part A 9.4
Polymer
b)
Thermal SiO2
Si substrate
Buffered HF etch
Si substrate
large protrusion areas, implying that a large polymer to nanostructured surface topography, which require
flow is needed to fill the stamp cavities. A thin residual nanometer-scale patterned surface areas in the mm2 to
polymer layer is obtained by combining a high imprint cm2 range.
pressure, a high process temperature, and a long imprint Nanofluidic channels can be used to stretch
time. The mode confinement in the PS waveguides is DNA [9.255, 256] for high-throughput linear analysis,
enhanced by etching the substrate oxide layer isotrop- measuring the length L of individual DNA molecules,
ically in hydrofluoric acid (HF), to create a pedestal or possibly sequencing by detection of fluorescent la-
structure. The Q-factor of the resonator device de- bels attached to specific DNA sequences [9.228]. The
pends critically on the surface scattering losses in the linear analysis relies on uniform stretching of DNA
waveguides. The surface roughness of the polymer molecules without coiling as they are driven through
waveguides can be reduced by a controlled thermal a narrow channel. This implies that the nanofluidic
reflow. The device is heated to 10–20 ◦ C below the channel should have cross-sectional dimensions D
glass transition, and the surface reflows under the ac- close to or smaller than the persistence length of
tion of surface tension. A loss reduction of more than DNA, L p ≈ 50 nm [9.257]. The assumption of uniform
70 dB/cm was achieved by this approach [9.250]. stretching of the molecule also puts strong requirements
on channel sidewall smoothness.
9.4.6 Bio Applications Tegenfeldt et al. [9.255] investigated the dynam-
ics of genomic-length DNA molecules in 100 nm-wide
Micro- and nanofabrication technology has enabled nanochannels, defined by NIL. The device layout is
methods to manipulate and probe individual molecules shown in Fig. 9.32. Two microfluidic channels, A–B and
and cells on a chip [9.251–255]. This type of application D–E, are connected by a 5 × 1 mm2 array of 100 nm-
often requires a large area covered with nanostruc- wide nanofluidic channels. The nanofluidic channel
tures. Sometimes a large number of identical devices array is defined by thermal NIL, and the pattern is
Part A 9.4
are needed for statistical evaluation, or to give redun- transferred into the silica substrate by metallization,
dancy, e.g., against clogging of nanofluidic channels. lift-off, and CF4 : H2 RIE. The microfluidic channels
With these requirements, NIL is advantageous, or some- are defined on a second silica substrate by UV pho-
times the only viable lithography method, even for tolithography (PL) and RIE, and fluidic access ports
laboratory-scale experiments and prototyping. Another are sandblasted. The two silica substrates are bonded by
example is devices for investigation of cell response cleaning the surfaces, using the so-called RCA protocol
Nanoimprint Lithography – Patterning of Resists Using Molding 9.4 Nanoimprinting Applications 301
(standard wet chemical process for removal of par- by Thamdrup et al. [9.259] (Fig. 9.33). The devices
ticles and organic surface contamination [9.258]) before were fabricated by thermal NIL in low-Mw (50 k)
bonding at room temperature, and annealing at 100 ◦ C. PMMA using a 100 mm-diameter two-level hybrid
The microfluidic channels allow for fast transport of the stamp. The fluidic structures were sealed using ther-
DNA from the input port to the nanofluidic channels. mal fusion bonding. The line array of stamp protrusions
External electrodes are fitted in the access ports A–E, to imprint the (250 × 250 nm2 ) nanochannels was de-
in order to apply a driving electric field, pulling the fined by EBL in SU-8 [9.135] and RIE etching in
DNA through the nanochannels. The DNA is marked a thermally grown oxide layer on a silicon wafer. The
with fluorescent dye molecules, which makes it possi- 1 μm-high, 50 μm-wide stamp protrusions for the mi-
ble to detect individual DNA molecules optically in the crofluidic load channels were subsequently formed by
nanochannels, by means of an optical microscope. UV-PL in a sol–gel process, using an organic-inorganic
Similar nanofluidic devices for DNA stretching hybrid polymer commercialized under the name Ormo-
can be fabricated in polymer at low cost and high comp [9.144]. The stamp is compatible with molecular
throughput in a single NIL process, as demonstrated vapor deposition (MVD), used for applying a durable
chlorosilane-based antistiction coating, and allows for
imprinting up to a temperature of 270 ◦ C. To benchmark
a) c) the device performance to conventional fused-silica
devices the extension of YOYO-1-stained T4 GT7
bacteriophage DNA inside the PMMA nanochannels
was experimentally investigated using epifluorescence
10 µm microscopy. The measured average extension length
amounts to 20% of the full contour length, with
d) a standard deviation of 4%. These results are in good
agreement with results obtained by stretching DNA in
b) conventional fused-silica nanochannels.
Cell growth and adhesion can be strongly influenced
10 µm
by surface topography on the micrometer to nanometer
length scale [9.260]. This has been exploited by Gade-
Fig. 9.33a–d Nanofluidic channels fabricated in PMMA gaard et al. [9.261] to create a three-dimensional tubular
by a single thermal NIL step using a two-level stamp (af- scaffold for tissue engineering of blood vessels that re-
ter [9.259]). (a) V-shaped, microfluidic channels (50 μm produce the basic structure of natural blood vessels:
wide and 1 μm deep) are connected by an array of nanoflu- a layer of smooth muscle cells (fibroblasts) coaxially
idic channels, 250 nm wide and 250 nm deep. (b) Schemat- embedded between an outer collagen mesh and an inner
Part A 9.4
ics showing the conformation of linear DNA when linen of endothelial cells (Fig. 9.34). Such artificially
confined inside the poly(methyl methacrylate) (PMMA) grown blood vessels with tight control of cellular at-
nanochannels (de Gennes regime). (c) SEM picture of two tachment, migration, and growth are expected to reduce
nanoimprint stamp; (d) SEM picture of the imprinted de- problems with cellular debris and inflammation. This
vice before the channels are sealed by thermal polymer would be a major improvement compared with cur-
bonding of lid (after [9.259]) rent medical procedures, where polymer tubes are used
302 Part A Nanostructures, Micro-/Nanofabrication and Materials
100 nm
Fig. 9.34 Swiss roll tubular construct
for vascular tissue engineering
E
A, B
D
C
100 µm
for vascular grafting. To facilitate cell growth which thick film polycaprolactone (PCL), which is a US Food
mimics the structure of blood vessels, the 3-D tubu- and Drug Administration (FDA)-approved biodegrad-
lar scaffold consists of a coaxial polymer layers with able polymer, and a thermoplastic with glass-transition
different surface topography, which selectively stim- temperature Tg ≈ 60 ◦ C, which was subsequently rolled
ulates growth of a particular cell type: muscle and up to form the required 3-D tubular structure. The
endothelial cells. A surface topography with nanometer- desired surface structure was realized on flat silicon
scale features on the inner surface favors adhesion and stamps by UV-PL and EBL, and negative stamp repli-
growth of endothelial cells. The bulk of the tube wall cas were formed in PDMS by casting and peel-off.
consists of microchannels with embedded micrometer- The PDMS stamp replicas were used for double-side
sized grooves which stimulate growth and adhesion of embossing on the PCL sheet. After embossing, the
muscle cells (fibroblasts). The scaffold is fabricated PCL sheet was rolled up to form the tubular scaffold
by multilevel thermal NIL in an approximately 30 μm- structure.
cess chain already established in a silicon cleanroom • The 3-D patterning capability (Sect. 9.2.8) makes it
environment. Furthermore substrate sizes, through- possible to develop innovative pattern-transfer pro-
put, and yield have to correspond to the production cesses, thus leading to significant cost reduction.
needs. As in research, many of the machines already Similar advances could be achieved if materials with
available can be used for moderate-scale produc- new properties are patterned. This is mainly due to
tion. They can be scaled up to substrate sizes of the fact that NIL has the concept of displacing ma-
200 mm and higher in combination with batch-mode terial at the nanoscale rather than removing material
operation. More sophisticated are machines based selectively and locally. Often this goes along with
on step-and-repeat NIL, which can help to solve some tradeoffs on resolution and alignment, which
equilibration and overlay issues. Further improve- is justified depending on the application.
ments can be expected if new resists and process
schemes are developed. In order to achieve a criti- NIL has now passed a barrier from the laboratory
cal mass of technological expertise, the integration scale to industrial preproduction. Although it seems
of NIL into a consortium of technology providers is that room-temperature processes based on UV exposure
of advantage, making it possible for manufacturers have an advantage over processes based on thermo-
to buy standard equipment and materials, along with cycles, to date it is difficult to say which process
process knowhow. will become a standard process and make it to the
production line. For example, isothermal processes at
Until now NIL was considered as a very promis- elevated temperatures using hybrid processes that use
ing patterning method, because it combines resolution both thermal NIL and hardening by UV curing have
with large area and throughput. As long as it is seen been established [9.273]. With state-of-the-art UV-NIL
as an alternative to establish high-end photolithogra- equipment [9.268], more than six wafers per hour with
phies, the strategy will most likely be to replace a diameter of 200 mm can now be achieved in a step-
single lithographic steps by imprinting. The only con- and-repeat modus (using a stamp area of 45 × 60 mm2 ).
sequence in the multimasking process sequence needed Single-step wafer-scale hot embossing has similar capa-
in microchip fabrication would then be modifying the bilities, and can even push throughput further if heatable
pattern transfer process, e.g., by adding the resid- stamps with low thermal mass are used [9.210, 274].
ual layer etch (Sect. 9.2.6). The requirements of the However, NIL is currently such a fast-moving field that
ITRS roadmap are so high that other more estab- prejudgment about the final success of one technique is
lished NGLs might make faster advances towards the not possible or advisable. Innovative solutions are still
next node, and the introduction of NIL into large- needed to solve process and stamp lifetime issues for
scale fabrication would be further postponed (NIL many different applications. Probably, not only a sin-
was first added to the 2003 ITRS roadmap for the gle NIL process will be successfully implemented, but
32 nm node [9.14, 272]). However, NIL has other ca- many variants of NIL. This includes hybrid approaches,
pabilities, as demonstrated in Sect. 9.4, even if not e.g., NIL in combination with other lithographic pro-
all requirements of the ITRS roadmap are met at cesses, or the fabrication and copying of stamps using
once: NIL.
The aim of this chapter was to give an insight into
• Enterprises with applications ranging from tem- the concepts used in NIL, along with presenting the ad-
plates for hard-disk production to SAW filters for vantages and limitations of processes ranging from tool
mobile phones, polarizers for flat-panel screens, and fabrication to pattern transfer. Although more referring
templates for biodevices are now heading into repli- to the older thermoplastic molding process, which is
cation techniques based on NIL processes. Most of the authors’ original field of expertise, it was intended
these processes are based on single layers covered to be general enough that future developments can be
with nanostructures, mostly regular high-resolution judged. The interested reader, however, will find more
Part A 9.5
gratings and dot arrays, and need single-step wafer- detailed information at technological conferences and in
scale replication tools for large areas. scientific publications, and also in the patent literature.
304 Part A Nanostructures, Micro-/Nanofabrication and Materials
References
9.1 E. Berliner: Gramophone, US Patent 372786 (1887), used for hot embossing lithography, Microelectron.
http://www.audioannals.com/berlinere.htm Eng. 54, 229–245 (2000)
9.2 E. Berliner: Process for producing records of sound, 9.19 H. Schulz, M. Wissen, N. Bogdanski, H.-C. Scheer,
US Patent 382790 (1888), http://www.audioannals. K. Mattes, C. Friedrich: Impact of molecular weight
com/berlinere.htm of polymers and shear rate effects for nanoimprint
9.3 J.C. Ruda: Record manufacturing: making sound for lithography, Microelectron. Eng. 83, 259–280 (2006)
everyone, J. Audio Eng. Soc. 25(10/11), 702–711 (1977) 9.20 S.Y. Chou, P.R. Krauss, P.J. Renstrom: Imprint of sub-
9.4 K.C. Pohlmann: The Compact Disc Handbook, Com- 25 nm vias and trenches in polymers, Appl. Phys.
put. Music Dig. Audio Ser., Vol. 5, 2nd edn. (A-R Lett. 67(21), 3114–3116 (1995)
Editions, Madison 1992) 9.21 S.Y. Chou, P.R. Krauss, P.J. Renstrom: Nanoimprint
9.5 H. Schift, C. David, M. Gabriel, J. Gobrecht, lithography, J. Vac. Sci. Technol. B 14(6), 4129–4133
L.J. Heyderman, W. Kaiser, S. Köppel, L. Scandella: (1996)
Nanoreplication in polymers using hot embossing 9.22 S.Y. Chou: Nanoimprint lithography, US Patent
and injection molding, Microelectron. Eng. 53, 171– 5772905 (1995)
174 (2000) 9.23 L. Baraldi, R. Kunz, J. Meissner: High-precision
9.6 S.Y. Chou, P.R. Krauss: Imprint lithography with molding of integrated optical structures, Proc. SPIE
sub-10 nm feature size and high throughput, Mi- 1992, 21–29 (1993)
croelectron. Eng. 35, 237–240 (1997) 9.24 J. Haisma, M. Verheijen, K. van den Heuvel, J. van
9.7 R.W. Jaszewski, H. Schift, J. Gobrecht, P. Smith: Hot den Berg: Mold-assisted lithography: A process for
embossing in polymers as a direct way to pattern reliable pattern replication, J. Vac. Sci. Technol. B
resist, Microelectron. Eng. 41/42, 575–578 (1998) 14, 4124–4128 (1996)
9.8 Y. Xia, G.M. Whitesides: Soft lithography, Angew. 9.25 M. Colburn, S. Johnson, M. Stewart, S. Damle,
Chem. Int. Ed. 37, 550–575 (1998) T. Bailey, B. Choi, M. Wedlake, T. Michealson,
9.9 B. Michel, A. Bernard, A. Bietsch, E. Delamarche, S.V. Sreenivasan, J. Ekerdt, C.G. Willson: Step and
M. Geissler, D. Juncker, H. Kind, J.-P. Renault, flash imprint lithography: A new approach to high-
H. Rothuizen, H. Schmid, P. Schmidt-Winkel, resolution patterning, Proc. SPIE 3676, 379–389
R. Stutz, H. Wolf: Printing meets lithography: Soft (1999)
approaches to high-resolution, IBM J. Res. Dev. 9.26 D.J. Resnick, W.J. Dauksher, D. Mancini, K.J. Nord-
45(5), 697–719 (2001) quist, T.C. Bailey, S. Johnson, N. Stacey, J.G. Ekerdt,
9.10 W. Menz, J. Mohr, O. Paul: Microsystem Technology C.G. Willson, S.V. Sreenivasan, N. Schumaker: Im-
(Wiley-VCH, Weinheim 2001) print lithography: Lab curiosity or the real NGL?,
9.11 H. Schift: Nanoimprint lithography: An old story in Proc. SPIE 5037, 12–23 (2003)
modern times? A review, J. Vac. Sci. Technol. B 26(2), 9.27 D.J. Resnick, S.V. Sreenivasan, C.G. Willson: Step and
458–480 (2008) flash imprint lithography, Mater. Today 8, 34–42
9.12 H. Schift (Ed.): NaPa Library of Processes (NaPa- (2005)
consortium, 2008), http://www.napanil.org (last 9.28 M. Doi: Introduction to Polymer Physics (Clarendon,
access December 2009) Oxford 1996)
9.13 C. Sotomayor Torres: Alternative lithography – 9.29 D.W. van Krevelen: Properties of Polymers (Elsevier,
Unleashing the potential of nanotechnology. In: Amsterdam 1990)
Nanostructure Science and Technology, ed. by 9.30 H. Schift, L.J. Heyderman: Nanorheology – squeezed
D.J. Lockwood (Kluwer, New York 2003) flow in hot embossing of thin films. In: Alternative
9.14 International Technology Roadmap for Semiconduc- Lithography, Nanostruct. Sci. Technol., ed. by C. So-
tors, http://public.itrs.net/ (last accessed May 8, tomayor Torres (Kluwer, New York 2003) pp. 46–76
2008) 9.31 H.-C. Scheer, H. Schulz, T. Hoffmann, C.M. Sotomayor
9.15 R. Compaño (Ed.): Technology Roadmap for Nano- Torres: Nanoimprint techniques. In: Handbook of
electronics, European Commission IST Programme, Thin Film Materials, Vol. 5, ed. by H.S. Nalva (Aca-
Future and Emerging Technologies, 2nd edn. (Eu- demic, New York 2002) pp. 1–60, Chap. 1
ropean Commission, Brussels 2000) 9.32 M.D. Austin, H. Ge, W. Wu, M. Li, Z. Yu, D. Wasser-
9.16 H. Moore: Cramming more components onto inte- man, S.A. Lyon, S.Y. Chou: Fabrication of 5 nm
Part A 9
grated circuits, Electronics 38(8), 114–117 (1965) linewidth and 14 nm pitch features by nanoimprint
9.17 S. Okazaki: Resolution limits of optical lithography, lithography, Appl. Phys. Lett. 84(26), 5299–5301
J. Vac. Sci. Technol. B 9(6), 2829–2833 (1991) (2004)
9.18 L.J. Heyderman, H. Schift, C. David, J. Gobrecht, 9.33 E.A. Dobisz, S.L. Brandow, R. Bass, J. Mitterender:
T. Schweizer: Flow behaviour of thin polymer films Effects of molecular properties on nanolithography
Nanoimprint Lithography – Patterning of Resists Using Molding References 305
in polymethyl methacrylate, J. Vac. Sci. Technol. B 9.50 A. Lebib, Y. Chen, J. Bourneix, F. Carcenac,
18, 107–111 (2000) E. Cambril, L. Couraud, H. Launois: Nanoimprint
9.34 A. Olzierski, I. Raptis: Development and molecular- lithography for a large area pattern replication, Mi-
weight issues on the lithographic performance croelectron. Eng. 46, 319–322 (1999)
of poly (methyl methacrylate), Microelectron. Eng. 9.51 C. Gourgon, J.H. Tortai, F. Lazzarino, C. Perret,
73/74, 244–251 (2004) G. Micouin, O. Joubert, S. Landis: Influence of resid-
9.35 M. Khoury, D.K. Ferry: Effect of molecular weight ual solvent in polymers patterned by nanoimprint
on poly(methyl methacrylate) resolution, J. Vac. Sci. lithography, J. Vac. Sci. Technol. B 22(6), 602–606
Technol. B 14, 75–79 (1996) (2004)
9.36 L.J. Fetters, D.J. Lohse, D. Richter, T.A. Witten, 9.52 Y. Hirai, M. Fujiwara, T. Okuno, Y. Tanaka, M. Endo,
A. Zirkel: Connection between polymer molecular S. Irie, K. Nakagawa, M. Sasago: Study of the resist
weight, density, chain dimensions, and melt vis- deformation in nanoimprint lithography, J. Vac. Sci.
coelastic properties, Macromolecules 27, 4639–4647 Technol. B 19(6), 2811–2815 (2001)
(1994) 9.53 Y. Hirai, T. Konishi, T. Yoshikawa, S. Yoshida:
9.37 A. Franck: Kunststoff-Kompendium, 4th edn. (Vogel, Simulation and experimental study of polymer de-
Würzburg 1996) p. 255, in German formation in nanoimprint lithography, J. Vac. Sci.
9.38 C.B. Roth, J.R. Dutcher: Mobility on different length Technol. B 22(6), 3288–3293 (2002)
scales in thin polymer films. In: Soft Mater- 9.54 H.D. Rowland, W.P. King: Polymer deformation and
ials: Structure and Dynamics, ed. by J.R. Dutcher, filling modes during microembossing, J. Micromech.
A.G. Marangoni (Dekker, New York 2004) Microeng. 14, 1625–1632 (2004)
9.39 J.N. D’Amour, U. Okoroanyanwu, C.W. Frank: Influ- 9.55 S. Zankovych, T. Hoffmann, J. Seekamp, J.-U. Bruch,
ence of substrate chemistry on the properties of C.M. Sotomayor Torres: Nanoimprint lithography:
ultrathin polymer films, Microelectron. Eng. 73/74, challenges and prospects, Nanotechnology 12(2), 91–
209–217 (2004) 95 (2001)
9.40 R.B. Bird, C.F. Curtis, R.C. Armstrong, O. Hassager: 9.56 M. Beck, M. Graczyk, I. Maximov, E.-L. Sarwe,
Dynamics of Polymeric Liquids (Wiley, New York 1987) T.G.I. Ling, M. Keil, L. Montelius: Improving stamps
9.41 L.G. Baraldi: Heißprägen in Polymeren für die Her- for 10 nm level wafer scale nanoimprint, lithogra-
stellung integriert-optischer Systemkomponenten. phy, Microelectron. Eng. 61/62, 441–448 (2002)
Ph.D. Thesis (ETH Zurich, Zurich 1994), Vol. 10762, in 9.57 D.-Y. Khang, H.H. Lee: Room-temperature imprint
German lithography by solvent vapor treatment, Appl. Phys.
9.42 M.J. Stefan: Parallel Platten Rheometer, Akad. Wiss. Lett. 76(7), 870–872 (2000)
Math.-Naturwiss. Vienna 2(69), 713–735 (1874), in 9.58 D.-Y. Khang, H. Yoon, H.H. Lee: Room-temperature
German imprint lithography, Adv. Mater. 13(10), 749–751
9.43 J.-H. Jeong, Y.-S. Choi, Y.-J. Shin, J.-J. Lee, K.- (2001)
T. Park, E.-S. Lee, S.-R. Lee: Flow behavior at the 9.59 D.-Y. Khang, H. Kang, T.-I. Kim, H.H. Lee: Low-
embossing stage of nanoimprint lithography, Fibers pressure nanoimprint lithography, Nano Lett. 4(4),
Polym. 3(3), 113–119 (2002) 633–637 (2004)
9.44 H. Schift, S. Park, J. Gobrecht: Nano-imprint – 9.60 H. Lee, G.Y. Jung: Full wafer scale near zero resid-
Molding resists for lithography, J. Photopolym. Sci. ual nano-imprinting lithography using UV curable
Technol. 16(3), 435–438 (2003) monomer solution, Microelectron. Eng. 77(1), 42–47
9.45 H.-C. Scheer, H. Schulz, T. Hoffmann, C.M. Sotomayor (2005)
Torres: Problems of the nanoimprinting technique 9.61 L. Tan, Y.P. Kong, S.W. Pang, A.F. Yee: Imprinting
for nanometer scale pattern definition, J. Vac. Sci. of polymer at low temperature and pressure, J. Vac.
Technol. B 16, 3917–3921 (1998) Sci. Technol. B 22(5), 2486–2492 (2004)
9.46 H.-C. Scheer, H. Schulz: A contribution to the flow 9.62 C. Finder, C. Mayer, H. Schulz, H.-C. Scheer, M. Fink,
behaviour of thin polymer films during hot em- K. Pfeiffer: Non-contact fluorescence measurements
bossing lithography, Microelectron. Eng. 56, 311–332 for inspection and imprint depth control in nanoim-
(2001) print lithography, Proc. SPIE 4764, 218–223 (2002)
9.47 L.J. Guo: Recent progress in nanoimprint technology 9.63 D. Jucius, V. Grigaliunas, A. Guobiene: Rapid evalu-
and its applications, J. Phys. D 37, R123–R141 (2004) ation of imprint quality using optical scatterometry,
9.48 L.J. Guo: Nanoimprint lithography: Methods and Microelectron. Eng. 71, 190–196 (2004)
Part A 9
material requirements, Adv. Mater. 19, 495–513 9.64 A. Fuchs, B. Vratzov, T. Wahlbrink, Y. Georgiev,
(2007) H. Kurz: Interferometric in situ alignment for UV-
9.49 C. Gourgon, C. Perret, G. Micouin, F. Lazzarino, based nanoimprint, J. Vac. Sci. Technol. B 22(6),
J.H. Tortai, O. Joubert, J.-P.E. Grolier: Influence of 3242–3245 (2002)
pattern density in nanoimprint lithography, J. Vac. 9.65 Z. Yu, H. Gao, S.Y. Chou: In situ real time process
Sci. Technol. B 21(1), 98–105 (2003) characterisation in nanoimprint lithography using
306 Part A Nanostructures, Micro-/Nanofabrication and Materials
time-resolved diffractive scatterometry, Appl. Phys. 9.82 M. Colburn, B.J. Choi, S.V. Sreenivasan, R.T. Bon-
Lett. 85(18), 4166–4168 (2004) necaze, C.G. Willson: Ramifications of lubrication
9.66 F. Lazzarino, C. Gourgon, P. Schiavone, C. Perret: theory on imprint lithography, Microelectron. Eng.
Mold deformation in nanoimprint lithography, J. 75, 321–329 (2004)
Vac. Sci. Technol. B 22(6), 3318–3322 (2002) 9.83 A. Fuchs, M. Bender, U. Plachetka, U. Hermanns,
9.67 C. Perret, C. Gourgon, F. Lazzarino, J. Tallal, S. Landis, H. Kurz: Ultraviolet-based nanoimprint at reduced
R. Pelzer: Characterization of 8 in wafers printed by environmental pressure, J. Vac. Sci. Technol. B 23(6),
nanoimprint lithography, Microelectron. Eng. 73/74, 2925–2928 (2005)
172–177 (2004) 9.84 M. Colburn, I. Suez, B.J. Choi, M. Meissl, T. Bai-
9.68 C. Gourgon, C. Perret, J. Tallal, F. Lazzarino, S. Lan- ley, S.V. Sreenivasan, J.G. Ekerdt, C.G. Willson:
dis, O. Joubert, R. Pelzer: Uniformity across 200 mm Characterization and modelling of volumetric and
silicon wafers printed by nanoimprint lithography, mechanical properties for step and flash imprint
J. Phys. D 38, 70–73 (2005) lithography photopolymers, J. Vac. Sci. Technol. B
9.69 U. Plachetka, M. Bender, A. Fuchs, B. Vratzov, 19(6), 2685–2689 (2001)
T. Glinsner, F. Lindner, H. Kurz: Wafer scale 9.85 D.J. Resnick, W.J. Dauksher, D. Mancini, K.J. Nord-
patterning by soft UV-nanoimprint lithography, Mi- quist, T.C. Bailey, S. Johnson, N. Stacey, J.G. Ekerdt,
croelectron. Eng. 73/74, 167–171 (2004) C.G. Willson, S.V. Sreenivasan, N. Schumaker: Im-
9.70 N. Roos, M. Wissen, T. Glinsner, H.-C. Scheer: Im- print lithography for integrated circuit fabrication,
pact of vacuum environment on the hot embossing J. Vac. Sci. Technol. B 21(6), 2624–2631 (2003)
process, Proc. SPIE 5037, 211–218 (2003) 9.86 M. Otto, M. Bender, B. Hadam, B. Spangenberg,
9.71 D. Pisignano, A. Melcarne, D. Mangiullo, R. Cin- H. Kurz: Characterization and application of a UV-
golani, G. Gigli: Nanoimprint lithography of chro- based imprint technique, Microelectron. Eng. 57/58,
mophore molecules under high-vacuum conditions, 361–366 (2001)
J. Vac. Sci. Technol. B 22(1), 185–188 (2004) 9.87 B. Vratzov, A. Fuchs, M. Lemme, W. Henschel,
9.72 H. Schift, L.J. Heyderman, M. Auf der Maur, J. Go- H. Kurz: Large scale ultraviolet-based nanoimprint
brecht: Pattern formation in hot embossing of thin lithography, J. Vac. Sci. Technol. B 21(6), 2760–2764
polymer films, Nanotechnology 12, 173–177 (2001) (2003)
9.73 S.Y. Chou, L. Zhuang: Lithographically induced self- 9.88 M. Komuro, J. Taniguchi, S. Inoue, N. Kimura,
assembly of periodic polymer micropillar arrays, J. Y. Tokano, H. Hiroshima, S. Matsui: Imprint char-
Vac. Sci. Technol. B 17, 3197–3202 (1999) acteristics by photo-induced solidification of liquid
9.74 S.Y. Chou, L. Zhuang, L.J. Guo: Lithographically in- polymer, Jpn. J. Appl. Phys. 39, 7075–7079 (2000)
duced self-construction of polymer microstructures 9.89 H. Schulz, H.-C. Scheer, T. Hoffmann, C.M. Sotomayor
for resistless patterning, Appl. Phys. Lett. 75, 1004– Torres, K. Pfeiffer, G. Bleidießel, G. Grützner, C. Car-
1006 (1999) dinaud, F. Gaboriau, M.-C. Peignon, J. Ahopelto,
9.75 L. Wu, S.Y. Chou: Electrohydrodynamic instability B. Heidari: New polymer materials for nanoimprint-
of a thin film of viscoelastic polymer underneath ing, J. Vac. Sci. Technol. B 18(4), 1861–1865 (2000)
a lithographically manufactured mask, J. Non- 9.90 H. Schulz, D. Lyebyedyev, H.-C. Scheer, K. Pfeiffer,
Newton. Fluid Mech. 125, 91–99 (2005) G. Bleidießel, G. Grützner, J. Ahopelto: Mas-
9.76 E. Schäffer, T. Thurn-Albrecht, T.P. Russell, ter replication into thermosetting polymers for
U. Steiner: Electrically induced structure formation nanoimprinting, J. Vac. Sci. Technol. B 18(6), 3582–
and pattern transfer, Nature 403, 874–877 (2000) 3585 (2000)
9.77 E. Schäffer, T. Thurn-Albrecht, T.P. Russell, 9.91 K. Pfeiffer, M. Fink, G. Bleidießel, G. Grützner,
U. Steiner: Method and apparatus for forming sub- H. Schulz, H.-C. Scheer, T. Hoffmann, C.M. So-
micron patterns on films, US Patent 07880075001 tomayor Torres, F. Gaboriau, C. Cardinaud: Novel
(1999) linear and crosslinking polymers for nanoimprint-
9.78 E. Schäffer, U. Steiner: Methods and apparatus for ing with high etch resistance, Microelectron. Eng.
the formation of patterns in films using temperature 53, 411–414 (2000)
gradients, European Patent PCT 124205.6 (2000) 9.92 S. Rudschuck, D. Hirsch, K. Zimmer, K. Otte, A. Braun,
9.79 K.Y. Suh, H.H. Lee: Capillary force lithography: large- R. Mehnert, F. Bigl: Replication of 3-D-micro- and
area patterning, self-organization, and anisotropic nanostrucutures using different UV-curable poly-
dewetting, Adv. Funct. Mater. 6/7, 405–413 (2002) mers, Microelectron. Eng. 53, 557–560 (2000)
Part A 9
9.80 Y. Hirai, S. Yoshida, N. Takagi: Defect analysis in 9.93 M. Sagnes, L. Malaquin, F. Carcenac, C. Vieu,
thermal nanoimprint lithography, J. Vac. Sci. Tech- C. Fournier: Imprint lithography using thermo-
nol. B 21(6), 2765–2770 (2003) polymerisation of MMA, Microelectron. Eng. 61/62,
9.81 Y. Hirai, T. Yoshikawa, N. Takagi, S. Yoshida: Me- 429–433 (2002)
chanical properties of poly-methyl methacrylate 9.94 A. Abdo, S. Schuetter, G. Nellis, A. Wei, R. Engelstad,
(PMMA) for nanoimprint lithography, J. Photopolym. V. Truskett: Predicting the fluid behavior during
Sci. Technol. 16(4), 615–620 (2003) the dispensing process for step-and-flash imprint
Nanoimprint Lithography – Patterning of Resists Using Molding References 307
lithography, J. Vac. Sci. Technol. B 22(6), 3279–3282 mix&match process for fabrication of interdigitated
(2002) nanobiosensors, Microelectron. Eng. 53, 521–524
9.95 Y. Hirai, H. Kikuta, T. Sanou: Study on optical (2000)
intensity distribution in photocuring nanoimprint 9.109 M. Beck, F. Persson, P. Carlberg, M. Graczyk, I. Max-
lithography, J. Vac. Sci. Technol. B 21(6), 2777–2782 imov, T.G.I. Ling, L. Montelius: Nanoelectrochemical
(2003) transducers for (bio-) chemical sensor applications
9.96 C.-H. Chang, R.K. Heilmann, R.C. Fleming, J. Carter, fabricated by nanoimprint lithography, Microelec-
E. Murphy, M.L. Schattenburg, T.C. Bailey, J.G. Ek- tron. Eng. 73/74, 837–842 (2004)
erdt, R.D. Frankel, R. Voisin: Fabrication of sawtooth 9.110 H. Schift, L.J. Heyderman, C. Padeste, J. Gobrecht:
diffraction gratings using nanoimprint lithography, Chemical nano-patterning using hot embossing
J. Vac. Sci. Technol. B 21(6), 2755–2759 (2003) lithography, Microelectron. Eng. 61/62, 423–428
9.97 L.J. Heyderman, H. Schift, C. David, B. Ketterer, (2002)
M. Auf der Maur, J. Gobrecht: Nanofabrication us- 9.111 S. Park, H. Schift, C. Padeste, J. Gobrecht: Nanos-
ing hot embossing lithography and electroforming, tructuring of anti-adhesive layer by hot embossing
Microelectron. Eng. 57/58, 375–380 (2001) lithography, Microelectron. Eng. 67/68, 252–258
9.98 P.R. Krauss, S.Y. Chou: Nano-compact disks with 400 (2003)
Gbit/in2 storage density fabricated using nanoim- 9.112 S. Park, S. Saxer, C. Padeste, H.H. Solak, J. Gobrecht,
print lithography and read with proximal probe, H. Schift: Chemical patterning of sub 50 nm half
Appl. Phys. Lett. 71(21), 3174–3176 (1997) pitches via nanoimprint lithography, Microelectron.
9.99 W. Wu, B. Cui, X. Sun, W. Zhang, L. Zhuang, Eng. 78/79, 682–688 (2005)
L. Kong, S.Y. Chou: Large area high density quan- 9.113 D. Falconnet, D. Pasqui, S. Park, R. Eckert, H. Schift,
tized magnetic disks fabricated using nanoimprint J. Gobrecht, R. Barbucci, M. Textor: A novel
lithography, J. Vac. Sci. Technol. B 16(6), 3825–3829 approach to produce protein nanopatterns by com-
(1998) bining nanoimprint, lithography and molecular
9.100 H. Schift, S. Park, C.-G. Choi, C.-S. Kee, S.-P. Han, K.- self-assembly, Nano Lett. 4(10), 1909–1914 (2004)
B. Yoon, J. Gobrecht: Fabrication process for polymer 9.114 J.D. Hoff, L.-J. Cheng, E. Meyhofer, L.J. Guo,
photonic crystals using nanoimprint lithography, A.J. Hunt: Nanoscale protein patterning by imprint
Nanotechnology 16, S261–S265 (2005) lithography, Nano Lett. 4(5), 853–857 (2004)
9.101 M. Hartney, D. Hess, D. Soane: Oxygen plasma etch- 9.115 T. Schliebe, G. Schneider, H. Aschoff: Nanostructur-
ing for resist stripping and multilayer lithography, ing high resolution phase zone plates in nickel and
J. Vac. Sci. Technol. B 7, 1–13 (1989) germanium using cross-linked polymers, Microelec-
9.102 W. Pilz, J. Janes, K.P.M. Müller, J. Pelka: Oxygen re- tron. Eng. 30, 513–516 (1996)
active ion etching of polymers – Profile evolution 9.116 G. Simon, A.M. Haghiri-Gosnet, F. Carcenac,
and process mechanisms, Proc. SPIE 1392, 84–94 H. Launois: Electroplating: an alternative transfer
(1990) technology in the 20 nm range, Microelectron. Eng.
9.103 B. Heidari, I. Maximov, E.-L. Sarwe, L. Montelius: 35, 51–54 (1997)
Large scale nanolithography using imprint lithogra- 9.117 K. Pfeiffer, M. Fink, G. Grützner, G. Bleidießel,
phy, J. Vac. Sci. Technol. B 17, 2961–2964 (1999) H. Schulz, H.-C. Scheer: Multistep profiles by mix
9.104 D. Lyebyedyev, H.-C. Scheer: Mask definition by and match of nanoimprint and UV-lithography, Mi-
nanoimprint lithography, Proc. SPIE 4349, 82–85 croelectron. Eng. 57/58, 381–387 (2001)
(2001) 9.118 X. Cheng, L.J. Guo: A combined-nanoimprint-and-
9.105 X.-M. Yan, S. Kwon, A.M. Contreras, J. Bokor, photolithography patterning technique, Microelec-
G.A. Somorjai: Fabrication of large number density tron. Eng. 3/4, 277–282 (2004)
platinum nanowire arrays by size reduction lithog- 9.119 X. Cheng, L.J. Guo: One-step lithography for various
raphy and nanoimprint lithography, Nano Lett. 5(4), size patterns with a hybrid mask-mold, Microelec-
745–748 (2005) tron. Eng. 3/4, 288–293 (2004)
9.106 L.J. Heyderman, B. Ketterer, D. Bächle, F. Glaus, 9.120 N. Kehagias, S. Zankovych, A. Goldschmidt, R. Kian,
B. Haas, H. Schift, K. Vogelsang, J. Gobrecht, L. Tie- M. Zelsmann, C.M. Sotomayor Torres, K. Pfeif-
fenauer, O. Dubochet, P. Surbled, T. Hessler: High fer, G. Ahrens, G. Grützner: Embedded polymer
volume fabrication of customised nanopore mem- waveguides: design and fabrication approaches,
brane chips, Microelectron. Eng. 67/68, 208–213 Superlattices Microstruct. 36(1-3), 201–210 (2004)
Part A 9
9.123 I. Martini, J. Dechow, M. Kamp, A. Forchel, J. Koeth: 9.138 S. Johnson, D.J. Resnick, D. Mancini, K.J. Nordquist,
GaAs field effect transistors fabricated by imprint W.J. Dauksher, K. Gehoski, J.H. Baker, L. Dues,
lithography, Microelectron. Eng. 60(3-4), 451–455 A. Hooper, T.C. Bailey, S.V. Sreenivasan, J.G. Ekerdt,
(2002) C.G. Willson: Fabrication of multi-tiered structures
9.124 A.P. Kam, J. Seekamp, V. Solovyev, C. Clavijo Cedeño, on step and flash imprint lithography templates,
A. Goldschmidt, C.M. Sotomayor Torres: Nanoim- Microelectron. Eng. 67/68, 221–228 (2003)
printed organic field-effect transistors: fabrication, 9.139 D. Suh, J. Rhee, H.H. Lee: Bilayer reversal imprint
transfer mechanism and solvent effects on device lithography: direct metal–polymer transfer, Nano-
characteristics, Microelectron. Eng. 73/74, 809–813 technology 15, 1103–1107 (2004)
(2004) 9.140 Y.P. Kong, H.Y. Lowa, S.W. Pang, A.F. Yee: Duo-mold
9.125 H. Schulz, A.S. Körbes, H.-C. Scheer, L.J. Balk: imprinting of three-dimensional polymeric struc-
Combination of nanoimprint and scanning force tures, J. Vac. Sci. Technol. B 22(6), 3251–3265 (2004)
lithography for local tailoring of sidewalls of 9.141 T. Borzenko, M. Tormen, G. Schmidt, L.W. Molen-
nanometer devices, Microelectron. Eng. 53, 221–224 kamp: Polymer bonding process for nano-
(2000) lithography, Appl. Phys. Lett. 79(14), 2246–2248
9.126 M. Tormen, L. Businaro, M. Altissimo, F. Romanato, (2001)
S. Cabrini, F. Perennes, R. Proietti, H.-B. Sun, 9.142 X.D. Huang, L.-R. Bao, X. Cheng, L.J. Guo, S.W. Pang,
S. Kawata, E. Di Fabrizio: 3-D patterning by means A.F. Yee: Reversal imprinting by transferring polymer
of nanoimprinting, x-ray and two-photon lithog- from mold to substrate, J. Vac. Sci. Technol. B 20(6),
raphy, Microelectron. Eng. 73/74, 535–541 (2004) 2872–2876 (2002)
9.127 X. Sun, L. Zhuang, W. Zhang, S.Y. Chou: Multilayer 9.143 N. Kehagias, V. Reboud, G. Chansin, M. Zelsmann,
resist methods for nanoimprint lithography on non- C. Jeppesen, C. Schuster, M. Kubenz, F. Reuther,
flat surfaces, J. Vac. Sci. Technol. B 16(6), 3922–3925 G. Grützner, C.M. Sotomayor Torres: Reverse-contact
(1998) UV nanoimprint lithography for multilayered struc-
9.128 F. van Delft: Bilayer resist used in e-beam lithogra- ture fabrication, Nanotechnology 18, 175303 (2007)
phy for deep narrow structures, Microelectron. Eng. 9.144 micro resist technology GmbH:
46, 369–373 (1999) http://www.microresist.de/ (micro resist technology
9.129 L. Tan, Y.P. Kong, L.-L. Bao, X.D. Huang, L.J. Guo, GmbH, Berlin 2009)
S.W. Pang, A.F. Yee: Imprinting polymer film on 9.145 Polysciences Inc.: http://www.polysciences.com
patterned substrates, J. Vac. Sci. Technol. B 21(6), (Polysciences Inc., Warrington 2009)
2742–2748 (2003) 9.146 Allresist GmbH: http://www.allresist.de (Allresist
9.130 B. Faircloth, H. Rohrs, R. Tiberio, R. Ruoff, R.R. Krch- GmbH, Strausberg 2009)
navek: Bilayer nanoimprint lithography, J. Vac. Sci. 9.147 C.-Y. Chao, L.J. Guo: Polymer microring resonators
Technol. B 18(4), 1866–1873 (2000) fabricated by nanoimprint technique, J. Vac. Sci.
9.131 A. Lebib, M. Natali, S.P. Li, E. Cambril, L. Manin, Technol. B 20, 2862–2866 (2002)
Y. Chen, H.M. Janssen, R.P. Sijbesma: Control of 9.148 Bayer AG: http://plastics.bayer.com (Bayer Material
the critical dimension with a trilayer nanoimprint Science, Leverkusen 2009)
lithography procedure, Microelectron. Eng. 57/58, 9.149 LG Dow Polycarbonate Ltd.: http://www.lg-dow.com
411–416 (2001) (LG Dow Polycarbonate Ltd., Yeosu Chunnam 2009)
9.132 Y. Chen, K. Peng, Z. Cui: A lift-off process for high 9.150 J. Tallal, D. Peyrade, F. Lazzarino, K. Berton, C. Perret,
resolution patterns using PMMA/LOR resist stack, Mi- M. Gordon, C. Gourgon, P. Schiavone: Replication of
croelectron. Eng. 73/74, 278–281 (2004) sub-40 nm gap nanoelectrodes over an 8 in. sub-
9.133 P. Carlberg, M. Graczyk, E.-L. Sawe, I. Maxi- strate by nanoimprint lithography, Microelectron.
mov, M. Beck, L. Montelius: Lift-off process for Eng. 78/79, 676–681 (2005)
nanoimprint lithography, Microelectron. Eng. 67/68, 9.151 Zeon Chemicals L. P.:
203–207 (2003) http://www.zeonchemicals.com (Zeon Chemicals L.
9.134 W. Li, J.O. Tegenfeldt, L. Chen, R.H. Austin, S.Y. Chou, P., Louisville 2009)
P.A. Kohl, J. Krotine, J.C. Sturm: Sacrificial polymers 9.152 Topas Advanced Polymers: http://www.topas.com/
for nanofluidic channels in biological applications, (Topas Advanced Polymers, Florence 2009)
Nanotechnology 14, 578–583 (2003) 9.153 T. Nielsen, D. Nilsson, F. Bundgaard, P. Shi, P. Szabo,
9.135 MicroChem Corp.: http://www.microchem.com/ (Mi- O. Geschke, A. Kristensen: Nanoimprint lithography
Part A 9
croChem Corp., Newton 2009) in the cyclic olefin copolymer, Topas, a highly UV-
9.136 M.W. Lin, H.-L. Chao, J. Hao, E.K. Kim, F. Palmieri, transparent and chemically resistant thermoplast, J.
W.C. Kim, M. Dickey, P.S. Ho, C.G. Willson: Pla- Vac. Sci. Technol. B 22, 1770–1775 (2004)
narization for reverse-tone step and flash imprint 9.154 B. Simmons, B. Lapizco-Encinas, R. Shediac,
lithography, Proc. SPIE 6151, 688–699 (2006) J. Hachman, J. Chames, J. Brazzle, J. Cere-
9.137 W. Trybula: Sematech, AMRC, and nano, Nanoprint muga, G. Fiechtner, E. Cummings, Y. Fintschenko:
Nanoimpr. Technol. (NNT) Conf., Vienna (2004) Polymeric insulator-based (electrodeless) dielec-
Nanoimprint Lithography – Patterning of Resists Using Molding References 309
trophoresis (iDEP) for the monitoring of water-borne nanoimprint lithography, Mater. Sci. Eng. C 15(1/2),
pathogens, Proc. MicroTAS 2, 171–173 (2004) 241–243 (2001)
9.155 D. Nilsson, S. Balslev, A. Kristensen: A microfluidic 9.168 K. Pfeiffer, F. Reuther, M. Fink, G. Grützner,
dye laser fabricated by nanoimprint lithography in P. Carlberg, I. Maximov, L. Montelius, J. Seekamp,
a highly transparent and chemically resistant cyclo- S. Zankovych, C.M. Sotomayor Torres, H. Schulz,
olefin copolymer (COC), J. Micromech. Microeng. 15, H.-C. Scheer: A comparison of thermally and
296–300 (2005) photochemically cross-linked polymers for nanoim-
9.156 K. Pfeiffer, M. Fink, G. Ahrens, G. Grützner, printing, Microelectron. Eng. 67/68, 266–273 (2003)
F. Reuther, J. Seekamp, S. Zankovych, C.M. So- 9.169 C.D. Schaper, A. Miahnahri: Polyvinyl alcohol tem-
tomayor Torres, I. Maximov, M. Beck, M. Graczyk, plates for low cost, high resolution, complex
L. Montelius, H. Schulz, H.-C. Scheer, F. Steingrüber: printing, J. Vac. Sci. Technol. B 22(6), 3323–3326
Polymer stamps for nanoimprinting, Microelectron. (2002)
Eng. 61/62, 393–398 (2002) 9.170 R.M. Reano, Y.P. Kong, H.Y. Low, L. Tan, F. Wang,
9.157 M. Wissen, H. Schulz, N. Bogdanski, H.-C. Scheer, S.W. Pang, A.F. Yee: Stability of functional polymers
Y. Hirai, H. Kikuta, G. Ahrens, F. Reuther, K. Pfeiffer: after plasticizer-assisted imprint lithography, J. Vac.
UV curing of resists for warm embossing, Microelec- Sci. Technol. B 22(6), 3294–3299 (2002)
tron. Eng. 73/74, 184–189 (2004) 9.171 B.K. Long, B.K. Keitz, C.G. Willson: Materials for
9.158 Sumitomo Chemical Corp.: http://www.sumitomo- step and flash imprint lithography (S-FIL), J. Mater.
chem.co.jp/ (Sumitomo Chemical Corp., Sendai 2009) Chem. 17, 3575–3580 (2007)
9.159 S. Landis, N. Chaix, C. Gourgon, C. Perret, T. Leveder: 9.172 J. Hao, M. Lin, F. Palmieri, Y. Nishimura, H.-
Stamp design effect on 100 nm feature size for 8 L. Chao, M.D. Stewart, A. Collins, K. Jen, C.G. Willson:
inch nanoimprint lithography, Nanotechnology 17, Photocurable silicon-base material for imprinting
2701–2709 (2006) lithography, Proc. SPIE 6517, 6517–6580 (2007)
9.160 N. Chaix, C. Gourgon, S. Landis, C. Perret, M. Fink, 9.173 S. Johnson, R. Burns, E.K. Kim, M. Dickey, G. Schmid,
F. Reuther, D. Mecerreyes: Influence of the molecular J. Meiring, S. Burns, C.G. Willson, D. Convey, Y. Wei,
weight and imprint conditions on the formation of P. Fejes, K. Gehoski, D. Mancini, K. Nordquist,
capillary bridges in nanoimprint lithography, Nano- W.J. Dauksher, D.J. Resnick: Effects of etch barrier
technology 17, 4082–4087 (2006) densification on step and flash imprint lithography,
9.161 C.G. Willson, R.A. Dammel, A. Reiser: Photoresist J. Vac. Sci. Technol. B 23(6), 2553–2556 (2005)
materials: A historical perspective, Proc. SPIE 3049, 9.174 F. Xu, N. Stacey, M. Watts, V. Truskett, I. McMackin,
28–41 (1997) J. Choi, P. Schumaker, E. Thompson, D. Babbs,
9.162 M.D. Stewart, C.G. Willson: Photoresists. In: En- S.V. Sreenivasan, G. Willson, N. Schumaker: Devel-
cyclopedia of Materials: Science and Technology, opment of imprint materials for the step and flash
ed. by K.H.J. Buschow, R.W. Cahn, M.C. Flemings, imprint lithography process, Proc. SPIE 5374, 232–241
B. Ilschner, E.J. Kramer, H.E.H. Meijer, S. Mahajan (2004)
(Routledge, New York 2001) pp. 6973–6978 9.175 M. Vogler, S. Wiedenberg, M. Mühlberger, I. Berg-
9.163 K. Pfeiffer, G. Bleidießel, G. Grützner, H. Schulz, mair, T. Glinsner, H. Schmidt, E.-B. Kley,
T. Hoffmann, H.-C. Scheer, C.M. Sotomayor Torres, G. Grützner: Development of a novel, low-viscosity
J. Ahopelto: Suitability of new polymer materials UV-curable polymer system for UV-nanoimprint
with adjustable glass temperature for nano- lithography, Microelectron. Eng. 84, 984–988 (2007)
imprinting, Microelectron. Eng. 46, 431–434 (1999) 9.176 P. Voisin, M. Zelsmann, R. Cluzel, E. Pargon, C. Gour-
9.164 F. Gaboriau, M.-C. Peignon, A. Barreau, G. Turban, gon, J. Boussey: Characterisation of ultraviolet
C. Cardinaud, K. Pfeiffer, G. Bleidießel, G. Grutzner: nanoimprint dedicated resists, Microelectron. Eng.
High density fluorocarbon plasma etching of new 84, 967–972 (2007)
resists suitable for nanoimprint lithography, Micro- 9.177 H. Schmitt, L. Frey, H. Ryssel, M. Rommel, C. Lehrer:
electron. Eng. 53, 501–505 (2000) UV nanoimprint materials: surface energies, residual
9.165 F. Gottschalch, T. Hoffmann, C.M. Sotomayor Tor- layers, and imprint quality, J. Vac. Sci. Technol. B
res, H. Schulz, H.-C. Scheer: Polymer issues in 25(3), 785–790 (2007)
nanoimprinting technique, Solid-State Electron. 43, 9.178 W.-C. Liao, S.L.-C. Hsu: A novel liquid thermal poly-
1079–1083 (1999) merization resist for nanoimprint lithography with
9.166 H. Schulz, H.-C. Scheer, T. Hoffmann, C.M. Sotomayor low shrinkage and high flowability, Nanotechnology
Part A 9
nanoimprint resists, J. Vac. Sci. Technol. B 25(4), 9.195 R.W. Jaszewski, H. Schift, B. Schnyder, A. Schneuwly,
1179–1185 (2007) P. Gröning: The deposition on anti-adhesive ultra-
9.181 M. Köhler: Etching in Microsystem Technology thin teflon-like films and their interaction with
(Wiley-VCH, Weinheim 1999) polymers during hot embossing, Appl. Surf. Sci. 143,
9.182 H. Schift, J. Gobrecht, B. Satilmis, J. Söchtig, 301–308 (1999)
F. Meier, W. Raupach: Nanoreplikation im Ver- 9.196 R.W. Jaszewski, H. Schift, P. Gröning, G. Margari-
bund: Ein Schweizer Netzwerk, Kunststoffe 94, 22–26 tondo: Properties of thin anti-adhesive films used
(2004), in German (English vers.: Nanoreplication in for the replication of microstructures in polymers,
a Network, Kunstst. Plast Eur. 94, 1-4 (2004)) Microelectron. Eng. 35, 381–384 (1997)
9.183 S. Park, H. Schift, H.H. Solak, J. Gobrecht: Stamps 9.197 U. Srinivasan, M.R. Houston, R.T. Howe, R. Mabou-
for nanoimprint lithography by extreme ultraviolet dian: Alkyltrichlorosilane-based self-assembled
interference lithography, J. Vac. Sci. Technol. B 22(6), monolayer films for stiction reduction in silicon mi-
3246–3250 (2004) cromachines, J. Microelectromech. Syst. 7, 252–260
9.184 K.A. Lister, B.G. Casey, P.S. Dobson, S. Thoms, (1998)
D.S. Macintyre, C.D.W. Wilkinson, J.M.R. Weaver: 9.198 H. Schulz, F. Osenberg, J. Engemann, H.-C. Scheer:
Pattern transfer of a 23 nm-period grating and sub- Mask fabrication by nanoimprint lithography using
15 nm dots into CVD diamond, Microelectron. Eng. antisticking layers, Proc. SPIE 3996, 244–249 (2000)
73/74, 319–322 (2004) 9.199 M. Beck, M. Graczyk, I. Maximov, E.-L. Sarwe,
9.185 J. Taniguchi, Y. Tokano, I. Miyamoto, M. Komuro, T.G.I. Ling, M. Keil, L. Montelius: Improving stamps
H. Hiroshima: Diamond nanoimprint lithography, for 10 nm level wafer scale nanoimprint lithography,
Nanotechnology 13, 592–596 (2002) Microelectron. Eng. 61/62, 441–448 (2002)
9.186 Y. Hirai, S. Yoshida, N. Takagi, Y. Tanaka, H. Yabe, 9.200 H. Schift, S. Saxer, S. Park, C. Padeste, U. Pieles,
K. Sasaki, H. Sumitani, K. Yamamoto: High aspect J. Gobrecht: Controlled co-evaporation of silanes for
pattern fabrication by nano imprint lithography us- nanoimprint stamps, Nanotechnology 16, S171–S175
ing fine diamond mold, Jpn. J. Appl. Phys. 42(6B), (2005)
3863–3866 (2003) 9.201 M. Keil, M. Beck, G. Frennesson, E. Theander,
9.187 S.W. Pang, T. Tamamura, M. Nakao, A. Ozawa, H. Ma- E. Bolmsjö, L. Montelius, B. Heidari: Process devel-
suda: Direct nano-printing on Al substrate using SiC opment and characterization of antisticking layers
mold, J. Vac. Sci. Technol. B 16, 1145–1149 (1998) on nickel-based stamps designed for nanoimprint
9.188 J. Gao, M.B. Chan-Park, D. Xie, Y. Yan, W. Zhou, lithography, J. Vac. Sci. Technol. B 22(6), 3283–3287
B.K.A. Ngoi, C.Y. Yue: UV embossing of submicron (2002)
patterns on biocompatible polymeric films using 9.202 S. Park, H. Schift, C. Padeste, B. Schnyder, R. Kötz,
a focused ion beam fabricated mold, Chem. Mater. J. Gobrecht: Anti-adhesive layers on nickel stamps
16(6), 956–958 (2004) for nanoimprint lithography, Microelectron. Eng.
9.189 M.M. Alkaisi, R.J. Blaikie, S.J. McNab: Low temper- 73/74, 196–201 (2004)
ature nanoimprint lithography using silicon nitride 9.203 ABCR GmbH: http://www.abcr.de/ (ABCR GmbH,
molds, Microelectron. Eng. 57/58, 367–373 (2001) Karlsruhe 2009)
9.190 Y. Hirai, S. Harada, S. Isaka, M. Kobayashi, Y. Tanaka: 9.204 B. Heidari, I. Maximov, E.-L. Sarwe, L. Montelius:
Nano-imprint lithography using replicated mold by Large scale nanolithography using imprint lithogra-
Ni electroforming, Jpn. J. Appl. Phys. 41(6B), 4186– phy, J. Vac. Sci. Technol. B 17, 2961–2964 (1999)
4189 (2002) 9.205 B. Heidari, I. Maximov, L. Montelius: Nanoimprint
9.191 Z. Yu, L. Chen, W. Wu, H. Ge, S.Y. Chou: Fabrica- lithography at the 6 in. wafer scale, J. Vac. Sci.
tion of nanoscale gratings with reduced line edge Technol. B 18(6), 3557–3560 (2000)
roughness using nanoimprint lithography, J. Vac. 9.206 N. Roos, T. Luxbacher, T. Glinsner, K. Pfeiffer,
Sci. Technol. B 21(5), 2089–2092 (2003) H. Schulz, H.-C. Scheer: Nanoimprint lithography
9.192 N. Roos, H. Schulz, L. Bendfeldt, M. Fink, K. Pfeif- with a commercial 4 inch bond system for hot em-
fer, H.-C. Scheer: First and second generation purely bossing, Proc. SPIE 4343, 427–436 (2001)
thermoset stamps for hot embossing, Microelectron. 9.207 L. Bendfeldt, H. Schulz, N. Roos, H.-C. Scheer: Groove
Eng. 61/62, 399–405 (2002) design of vacuum chucks for hot embossing lithog-
9.193 N. Roos, H. Schulz, M. Fink, K. Pfeiffer, F. Osen- raphy, Microelectron. Eng. 61/62, 455–459 (2002)
berg, H.-C. Scheer: Performance of 4” wafer-scale 9.208 T. Haatainen, J. Ahopelto, G. Grützner, M. Fink,
Part A 9
thermoset working stamps in hot embossing lithog- K. Pfeiffer: Step and stamp imprint lithography us-
raphy, Proc. SPIE 4688, 232–239 (2002) ing a commercial flip chip bonder, Proc. SPIE 3997,
9.194 H. Schift, S. Park, J. Gobrecht, S. Saxer, F. Meier, 874–879 (2000)
W. Raupach, K. Vogelsang: Hybrid bendable stamp 9.209 H. Tana, A. Gilbertson, S.Y. Chou: Roller nanoimprint
copies for molding fabricated by nanoimprint, Mi- lithography, J. Vac. Sci. Technol. B 16(6), 3926–3928
croelectron. Eng. 78/79, 605–611 (2005) (1998)
Nanoimprint Lithography – Patterning of Resists Using Molding References 311
9.210 M. Tormen: A nano impression lithographic process print lithography, Rev. Sci. Instrum. 75, 4481–4486
which involves the use of a die having a re- (2004)
gion able to generate heat, European Patent PCT/IB 9.224 C. Clavijo Cedeño, J. Seekamp, A.P. Kam, T. Hoff-
2004/002120 (2004) mann, S. Zankovych, C.M. Sotomayor Torres,
9.211 S.Y. Chou, C. Keimel, J. Gu: Ultrafast and direct C. Menozzi, M. Cavallini, M. Murgia, G. Ruani, F. Bis-
imprint of nanostructures in silicon, Nature 417, carini, M. Behl, R. Zentel, J. Ahopelto: Nanoimprint
835–837 (2002) lithography for organic electronics, Microelectron.
9.212 J.J. Shamaly, V.F. Bunze: I-line to DUV transition for Eng. 61/62, 25–31 (2002)
critical levels, Microelectron. Eng. 30, 87–93 (1996) 9.225 A. Manz, N. Graber, H.M. Widmer: Miniaturized
9.213 J.E. Bjorkholm: EUV lithography – The successor to total chemical analysis systems: A novel concept
optical lithography?, Intel Technol. J. Q3 (1998), for chemical sensing, Sens. Actuators B 1, 244–248
http://www.intel.com/technology/itj/q31998/articles (1990)
/art_4.htm 9.226 E. Verpoorte, N.F. De Rooij: Microfluidics meets
9.214 D. Wachenschwanz, W. Jiang, E. Roddick, A. Homola, MEMS, Proc. IEEE 91, 930–953 (2003)
P. Dorsey, B. Harper, D. Treves, C. Bajorek: Design of 9.227 A. Pepin, P. Youinou, V. Studer, A. Lebib, Y. Chen:
a manufacturable discrete track recording medium, Nanoimprint lithography for the fabrication of DNA
IEEE Trans. Mag. 41, 670–675 (2005) electrophoresis chips, Microelectron. Eng. 61/62,
9.215 G.M. McClelland, M.W. Hart, C.T. Rettner, M.E. Best, 927–932 (2002)
K.R. Carter, B.D. Terris: Nanoscale patterning of 9.228 J.O. Tegenfeldt, C. Prinz, H. Cao, R.L. Huang,
magnetic islands by imprint lithography using R.H. Austin, S.Y. Chou, E.C. Cox, J.C. Sturm: Micro-
a flexible mold, Appl. Phys. Lett. 81, 1483–1485 and nanofluidics for DNA analysis, Anal. Bioanal.
(2002) Chem. 378, 1678–1692 (2004)
9.216 G.F. Cardinale, J.L. Skinner, A.A. Talin, R.W. Bro- 9.229 S.Y. Chou: Patterned magnetic nanostructures and
cato, D.W. Palmer, D.P. Mancini, W.J. Dauksher, quantized magnetic disks, Proc. IEEE 85, 652–671
K. Gehoski, N. Le, K.J. Nordquist, D.J. Resnick: Fab- (1997)
rication of a surface acoustic wave-based correlator 9.230 M.N. Baibich, J.M. Broto, A. Fert, F. Nguyen
using step-and-flash imprint lithography, J. Vac. Van Dau, F. Petroff, P. Eitenne, G. Creuzet,
Sci. Technol. B 22, 3265–3270 (2004) A. Friederich, J. Chazelas: Giant magnetoresistance
9.217 S.-W. Ahn, K.-D. Lee, J.-S. Kim, S.H. Kim, S.H. Lee, of (001)Fe/(001)Cr magnetic superlattices, Phys. Rev.
J.-D. Park, P.-W. Yoon: Fabrication of subwave- Lett. 61, 2472–2475 (1988)
length aluminum wire grating using nanoimprint 9.231 Y. Li, A.K. Menon: Magnetic recording technologies:
lithography and reactive ion etching, Microelectron. Overview. In: Encyclopedia of Materials: Science and
Eng. 78/79, 314–318 (2005) Technology, ed. by K.H.J. Buschow (Elsevier, Amster-
9.218 J. Seekamp, S. Zankovych, A.H. Helfer, P. Maury, dam 2001) pp. 4948–4957
C.M. Sotomayor Torres, G. Böttger, C. Liguda, M. Eich, 9.232 L.F. Shew: Discrete tracks for saturation magnetic
B. Heidari, L. Montelius, J. Ahopelto: Nanoimprinted recording, IEEE Trans. Broadcast Telev. Receiv. 9, 56–
passive optical devices, Nanotechnology 13, 581–586 62 (1963)
(2002) 9.233 A.K. Menon: Interface tribology for 100 Gb/in2 , Tribol.
9.219 C.M. Sotomayor Torres, S. Zankovych, J. Seekamp, Int. 33, 299–308 (2000)
A.P. Kam, C. Clavijo Cedeño, T. Hoffmann, 9.234 Y. Soeno, M. Moriya, K. Ito, K. Hattori, A. Kaizu,
J. Ahopelto, F. Reuther, K. Pfeiffer, G. Blei- T. Aoyama, M. Matsuzaki, H. Sakai: Feasibility of
dießel, G. Grützner, M.V. Maximov, B. Heidari: discrete track perpendicular media for high track
Nanoimprint lithography: An alternative nanofab- density recording, IEEE Trans. Magn. 39, 1967–1971
rication approach, Mater. Sci. Eng. C 23, 23–31 (2003)
(2003) 9.235 S.Y. Chou, M. Wei, P.R. Krauss, P.B. Fisher: Study
9.220 J. Wang, X. Sun, L. Chen, S.Y. Chou: Direct nanoim- of nanoscale magnetic structures fabricated using
print of submicron organic light-emitting structures, electron beam lithography and quantum magnetic
Appl. Phys. Lett. 75, 2767–2769 (1999) disk, J. Vac. Sci. Technol. B 12, 3695–3698 (1994)
9.221 X. Cheng, Y. Hong, J. Kanicki, L.J. Guo: High- 9.236 R.L. White, R.M.H. Newt, R.F.W. Pease: Patterned
resolution organic polymer light-emitting pixels media: A viable route to 50 Gbit/in2 and up for
fabricated by imprinting technique, J. Vac. Sci. Tech- magnetic recording?, IEEE Trans. Magn. 33, 990–995
Part A 9
9.239 A. Kikitsu, Y. Kamata, M. Sakurai, K. Naito: Recent 9.255 J.O. Tegenfeldt, C. Prinz, H. Cao, S. Chou, W.W. Reis-
progress of patterned media, IEEE Trans. Magn. 43, ner, R. Riehn, Y.M. Wang, E.C. Cox, J.C. Sturm,
3685–3688 (2007) P. Silberzan, R.H. Austin: The dynamics of genomic-
9.240 M. Natali, A. Lebib, E. Cambril, Y. Chen, I.L. Prej- length DNA molecules in 100-nm channels, Proc.
beanu, K. Ounadjela: Nanoimprint lithography of Natl. Acad. Sci. USA 101, 10979–10983 (2004)
high-density cobalt dot patterns for fine tuning 9.256 L.J. Guo, X. Cheng, C.-F. Chou: Fabrication of
of dipole interactions, J. Vac. Sci. Technol. B 19, size-controllable nanofluidic channels by nanoim-
2779–2783 (2001) printing and its application for DNA stretching, Nano
9.241 J. Moritz, B. Dieny, J.P. Nozieres, S. Landis, A. Lebib, Lett. 4, 69–73 (2004)
Y. Chen: Domain structure in magnetic dots prepared 9.257 C. Bustamante, J.F. Marko, E.D. Siggia, S. Smith:
by nanoimprint and e-beam lithography, J. Appl. Entropic elasticity of λ-phage DNA, Science 265,
Phys. 91, 7314–7316 (2002) 1599–1600 (1994)
9.242 P. Lalanne, M. Hutley: Artificial media optical 9.258 W. Kern, D.A. Puotinen: RCA Rev. 31, 187–206 (1970)
properties-subwavelength scale. In: Enclopedia of 9.259 L.H. Thamdrup, A. Klukowska, A. Kristensen:
Optical Engineering, ed. by R.G. Driggers (Dekker, Stretching DNA in polymer nanochannels fabricated
New York 2003) pp. 62–71 by thermal imprint in PMMA, Nanotechnology 19,
9.243 http://www.pcmag.com/ (last accessed December 9, 125301 (2008)
2009) 9.260 M.J. Dalby, N. Gadegaard, R. Tare, A. Andar,
9.244 Z. Yu, W. Wu, L. Chen, S. Chou: Fabrication of large M.O. Riehle, P. Herzyk, C.D.W. Wilkinson, R.O.C. Or-
area 100 nm pitch grating by spatial frequency effo: The control of human mesenchymal cell
doubling an nanoimprint lithography for subwave- differentiation using nanoscale symmetry and dis-
length optical applications, J. Vac. Sci. Technol. B 19, order, Nat. Mater. 6, 997–1003 (2007)
2816–2819 (2001) 9.261 K. Seunarine, D.O. Meredith, M.O. Riehle, C.D.W. Wil-
9.245 MOXTEK Inc.: http://www.moxtek.com/ (last ac- kinson, N. Gadegaard: Biodegradable polymer tubes
cessed December 9, 2009) with lithographically controlled 3-D micro- and
9.246 NanoOpto, API Nanotronics Corp.: nanotopography, Microelectron. Eng. 85(5/6), 1350–
http://www.nanoopto.com/ (last accessed Decem- 1354 (2008)
ber 9, 2009) 9.262 A. Kapr: Johann Gutenberg: The Man and His Inven-
9.247 A.A. Erchak, D.J. Ripin, S. Fan, P. Rakich, tion (Scolar, London 1996), http://www.gutenberg.
J.D. Joannopoulos, E.P. Ippen, G.S. Petrich, de/publ.htm
L.A. Kolodziejski: Enhanced coupling to vertical ra- 9.263 EVGroup: http://www.evgroup.com/ (EVGroup, St.
diation using a two-dimensional photonic crystal in Florian 2009)
a semiconductor light-emitting diode, Appl. Phys. 9.264 SÜSS Microtec: http://www.suss.com/ (SÜSS Microtec,
Lett. 78, 563–565 (2001) Garching 2009)
9.248 S.H. Kim, K.-D. Lee, J.-Y. Kim, M.-K. Kwon, S.- 9.265 Obducat: http://www.obducat.com/ (Obducat, Mal-
J. Park: Fabrication of photonic crystal structures on mö 2009)
light emitting diodes by nanoimprint lithography, 9.266 Smart Equipment Technology S.A.S.:
Nanotechnology 18, 055306 (2007) http://www.set-sas.fr/ (Smart Equipment Technol-
9.249 L.J. Guo, X. Cheng, C.Y. Chao: Fabrication of pho- ogy S.A.S., Saint Jeoire 2009)
tonic nanostructures in nonlinear optical polymers, 9.267 Jenoptik: http://www.jenoptik.com (Jenoptik, Jena
J. Mod. Opt. 49, 663–673 (2002) 2009)
9.250 C.-Y. Chao, L.J. Guo: reduction of surface scatter- 9.268 Molecular Imprints:
ing loss in polymer microrings using thermal-reflow http://www.molecularimprints.com/ (Molecular Im-
technique, IEEE Photon. Technol. Lett. 16, 1498–1500 prints, Austin 2009)
(2004) 9.269 Nanonex: http://www.nanonex.com/ (Nanonex,
9.251 H.C. Hoch, L.W. Jelinski, H.G. Craighead (Eds.): Monmouth Junction 2009)
Nanofabrication and Biosystems: Integrating Ma- 9.270 Eulitha: http://www.eulitha.com/ (Eulitha, Villigen
terials Science, Engineering, and Biology (Cam- 2009)
bridge Univ. Press, Cambridge 1996) 9.271 NIL Technology: http://www.nilt.com/ (NIL Technol-
9.252 H.G. Craighead: Nanoelectromechanical systems, ogy, Kongens Lyngby 2009)
Science 290, 1532–1535 (2000) 9.272 Sematech: http://www.sematech.org/ (Sematech,
Part A 9
9.253 L.R. Huang, J.O. Tegenfeldt, J.J. Kraeft, J.C. Sturm, Austin 2009)
R.H. Austin, E.C. Cox: A DNA prism for high-speed 9.273 M. Beck, B. Heidari: Nanoimprint lithography for
continous frationation of large DNA molecules, Nat. high volume HDI manufacturing, OnBoard Technol.
Biotechnol. 20, 1048–1051 (2002) Sept., 52–55 (2006),
9.254 H.G. Craighead: Nanostructure science and technol- http://www.onboard-technology.com/
ogy: Impact and prospects for biology, J. Vac. Sci. 9.274 L. Olsson: Method and device for transferring a pat-
Technol. A 21, S216–S221 (2003) tern, European Patent PCT/SE 2003/001003 (2002)
313
Part A 10
and Nanofabrication
Etienne Menard, John A. Rogers
There is considerable interest in methods that can be techniques also have the disadvantage of high capi-
used to build structures that have micron or nanometer tal and operational costs. As a result, some of the
dimensions. Historically, research and development in oldest and conceptually simplest forms of lithogra-
this area has been driven mainly by the needs of the phy – embossing, molding, stamping, writing, and so
microelectronics industry. The spectacularly success- on – are now being reexamined for their potential
ful techniques that have emerged from those efforts – to serve as the basis for nanofabrication techniques
such as photolithography and electron beam lithogra- that can avoid these limitations [10.1]. Considerable
phy – are extremely well suited to the tasks for which progress has been made in the last few years, mainly
they were principally designed: forming structures of by combining these approaches or variants of them with
radiation-sensitive materials (including photoresists or new materials, chemistries, and processing techniques.
electron beam resists) on ultraflat glass or semicon- This chapter highlights some recent advances in high-
ductor surfaces. Significant challenges exist in adapting resolution printing methods, in which a stamp forms
these methods for new emerging applications and areas a pattern of ink on a surface that it contacts. It focuses
of research that require patterning of unusual sys- on approaches whose capabilities, level of development,
tems and materials, (including those in biotechnology and demonstrated applications indicate a strong po-
and plastic electronics), structures with nanometer di- tential for widespread use, especially in areas where
mensions (below 50–100 nm), large areas in a single conventional methods are unsuitable.
step (larger than a few square centimeters), or non- Contact printing involves the use of an element
planar (rough or curved) surfaces. These established with surface relief (the stamp) to transfer material ap-
314 Part A Nanostructures, Micro-/Nanofabrication and Materials
plied to its surface (the ink) to locations on a substrate a pattern of functional material. This chapter focuses
Part A 10.1
that it contacts. The printing press, one of the earliest on (1) printing techniques that are capable of micron
manufacturable implementations of this approach, was and nanometer resolution, and (2) their use for fabricat-
introduced by Gutenberg in the fifteenth century. Since ing key elements of active electronic or optical devices
then, this general approach has been used almost exclu- and subsystems. It begins with an overview of some
sively for producing printed text or images with features methods for fabricating high-resolution stamps and then
that are 100 μm or larger in their smallest dimension. illustrates two different ways that these stamps can be
The resolution is determined by the nature of the ink and used to print patterns of functional materials. Applica-
its interaction with the stamp and/or substrate, the res- tions that highlight the capabilities of these techniques
olution of the stamp, and the processing conditions that and the performances of systems that are constructed
are used for printing or to convert the pattern of ink into with them are also presented.
Part A 10.1
Pour over and cure PDMS stamp Master
Peel back
PDMS stamp ≈ 1 nm
3.6 nm
Imprint and UV ≈ 2 µm
cure photopolymer
Replica
Peel back
PDMS stamp
or proximity-mode photolithography with a mask pro- Fig. 10.3 Atomic force micrographs (top picture) of a
duced by direct write photolithography represents master that consists of a submonolayer of single-walled
a convenient method of fabricating the master. For carbon nanotubes (SWNTs; diameter between 0.5 and
features smaller than ≈ 2 μm, several different tech- 5 nm) grown on a SiO2 /Si wafer. The bottom atomic force
niques can be used [10.2], including: (1) projection micrograph shows a replica of the relief structures in
mode photolithography [10.3], (2) direct write elec- poly(urethane). These results indicate effective operation
tron beam (or focused ion beam) lithography [10.4, 5], of a PDMS stamp for soft imprint lithography at the single
(3) scanning probe lithography [10.6–9] or (4) laser nanometer scale
interference lithography [10.10]. The first approach re-
quires a photomask generated by some other method, the projection optics relaxes the resolution require-
such as direct write photolithography or electron beam ments on the mask and enables features as small as
lithography. The reduction (typically 4×) provided by ≈ 90 nm when deep ultraviolet radiation and phase
316 Part A Nanostructures, Micro-/Nanofabrication and Materials
shifting masks are used. The costs for these systems In order to evaluate the ultimate resolution limit
Part A 10.2
are, however, very high and their availability for gen- of the soft lithography methods, masters with relief
eral research purposes is limited. The second method structures in the single nanometer range must be fab-
is flexible in the geometry of patterns that can be pro- ricated. A simple method, presented in Fig. 10.2, uses
duced, and the writing systems are highly developed: submonolayer coverage of single-walled carbon nano-
30–50 nm features can be achieved with commercial tubes (SWNT grown, by established chemical vapor
systems [10.11], and < 10 nm features are possible deposition techniques, on an ultraflat silicon wafer. The
with research tools, as first demonstrated more than SWNT, which have diameters (heights and widths) in
25 years ago by Broers [10.12]. The main drawbacks the 0.5–5 nm range, are molded on the bottom sur-
of this method are that it is relatively slow and it face of a PDMS stamp generated by casting and curing
is difficult to pattern large areas. Like projection- against this master. Such a mold can be used to replicate
mode photolithography, it can be expensive. The third the relief structure into a variety of photocurable poly-
method, scanning probe lithography, is quite powerful mers in a kind of soft nanoimprinting technique [10.13–
in principle, but the tools are not as well established as 15]. A single mold can give exceedingly high reso-
those for other approaches. This technique has atomic lution, approaching the single nanometer scale range
resolution, but its writing speed can be lower and (comparable to a few bond lengths in the polymer back-
the areas that can be patterned are smaller than elec- bone), as can be seen in Fig. 10.3 [10.16]. These results
tron beam systems. Interference lithography provides demonstrate the extreme efficiency of the basic soft
a powerful, low-cost tool for generating periodic ar- lithographic procedure for generating and using elas-
rays of features with dimensions down to 100–200 nm; tomeric elements. The ultimate limits are difficult to
smaller sizes demand ultraviolet lasers, and patterns predict, due to substantial uncertainties surrounding the
with aperiodic or nonregular features are difficult to polymer physics and chemistry that dominates in the
produce. nanometer regime.
Part A 10.2
Stamp with HDT
Gold-coated
substrate 2 µm 10 µm
Part A 10.3
Form Si stamp, fluorinate surface
relief features in the geometry of the desired pattern, CF3 CF3 CF3 CF3
(2) a method for depositing a thin layer of solid mater- (CF2)5 (CF2)5 (CF2)5 (CF2)5
ial onto the raised features of this stamp, (3) a means (CF2)2 (CF2)2 (CF2)2 (CF2)2
OH OH OH OH OH OH
Oxidized
Substrate
substrate
Vapor deposit SAM
HS
100 nm HS HS HS HS HS HS
Stamp Si
H3CO OCH3 O Si O Si O Si O Si O Si O Si O
OCH3
O O O O O O
3-mercaptopropyl- Si Si Si Si Si Si
trimethoxysilane
Bring Au-coated
stamp into contact
100 nm Stamp
1 µm Au
Printed pattern
Au Au
Fig. 10.9 Scanning electron micrograph (SEM) of a pat-
HS HS HS HS HS HS
tern produced by nanotransfer printing. The structure
consists of a bilayer of Au (20 nm)/Ti (1 nm) (white) in
O Si O Si O Si O Si O Si O Si O
the geometry of a photonic bandgap waveguide printed
O O O O O O
onto a thin layer of polydimethylsiloxane on a sheet of Si Si Si Si Si Si
plastic (black). Electron beam lithography and etching of
a GaAs wafer produced the stamp that was used in this Remove stamp;
transfer complete
case. The transfer chemistry relied on condensation reac-
Au Au Au Au
tions between titanol groups on the surface of the Ti and S S HS HS S S
silanol groups on the surface of the PDMS. The frames on
the right show SEMs of the Au/Ti-coated stamp (top) be-
O Si O Si O Si O Si O Si O Si O
fore printing and on the substrate (bottom) after printing. O O O O O O
The electron beam lithography and etching used to fabri-
cate the stamp limit the minimum feature size (≈ 70 nm)
and the edge resolution (≈ 5–10 nm) of this pattern Fig. 10.10 Schematic illustration of steps involved in nan-
otransfer printing a pattern of a thin layer of Au onto
regions of the stamp (to which the metal has extremely a silicon wafer using a self-assembled monolayer (SAM)
poor adhesion) to the substrate. Complete pattern trans- surface chemistry. Plasma oxidizing the surface of the
fer from an elastomeric stamp to a thin elastomeric wafer generates OH groups. Solution or vapor phase ex-
substrate occurs readily at room temperature in open air posure of the wafer to 3-mercaptopropyltrimethoxysilane
with contact times of less than 15 s. When a rigid stamp yields a SAM with exposed thiol groups. Contacting an
is employed, slight heating is needed to induce transfer. Au-coated stamp to this surface produces thiol linkages
While the origin of this difference is unclear, it may re- that bond the gold to the substrate. Removing the stamp
flect the comparatively poor contact when rigid stamps completes the transfer printing process
are used; similar differences are also observed in cold
welding of gold films [10.40]. A wide range of surface chemistries can be used for
Figure 10.9 shows scanning electron micrographs of the transfer. SAMs are particularly attractive due to their
a pattern produced using a GaAs stamp generated by chemical flexibility. Figure 10.10 illustrates the use of
electron beam lithography and etching. The frames on a thiol-terminated SAM and nTP for forming patterns
the right show images of the metal-coated stamp before of Au on a silicon wafer [10.38]. Here, the vapor phase
printing (top) and the transferred pattern (bottom). The cocondensation of the methoxy groups of molecules of
resolution appears to be limited only by the resolution 3-mercaptopropyltrimethoxysilane (MPTMS) with the
of the stamp itself, and perhaps by the grain size of the −OH-terminated surface of the wafer produces a SAM
metal films. Although the accuracy in multilevel regis- of MPTMS with exposed thiol (−SH) groups. PDMS
tration that is possible with nTP has not yet been quanti- stamps can be prepared for printing on this surface by
fied, its performance is likely similar to that of emboss- coating them with a thin film (≈ 15 nm) of Au using
ing techniques when rigid stamps are used [10.41]. conditions (thermal evaporation 1.0 nm/s; ≈ 10−7 Torr
Stamping Techniques for Micro- and Nanofabrication 10.3 Nanotransfer Printing 321
Part A 10.3
a)
5 µm 500 nm
50 µm 100 µm b)
1 µm 500 nm
Figure 10.11 shows some optical micrographs of typical in the regions of contact. This reaction produces per-
Part A 10.4
printed patterns in this case [10.38]. manent Au−S bonds at the stamp/substrate interface
Similar surface chemistries can guide transfer to (see insets in Fig. 10.4 for idealized chemical reac-
other substrates. Alkanedithiols, for example, are useful tion schemes). Figure 10.12 schematically illustrates
for printing Au onto GaAs wafers [10.39]. Immers- how this procedure can generate continuous 3-D metal
ing these substrates (freshly etched with 37% HCl for patterns using stamps with nonvertical side walls. Sev-
≈ 2 min to remove the surface oxide) in a 0.05 M so- eral layers can be transferred on top of each other by
lution of 1,8-octanedithiol in ethanol for 3 h produces successively cold-welding the different gold metal lay-
a monolayer of dithiol on the surface. Although the ers. Figure 10.13 shows high- and low-magnification
chemistry of this system is not completely clear, it is scanning electron microscope images of nanotransfer
generally believed that the thiol end groups bond chem- printed single- and multilayer 3-D metal structures
ically to the surface. Surface spectroscopy suggests [10.43, 44]. The integrity of these free-standing 3-D
the formation of Ga−S and As−S bonds. Contacting metal structures is remarkable but depends critically on
an Au-coated PDMS stamp with the treated substrate the careful optimization of the metal evaporation condi-
causes the exposed thiol endgroups to react with Au tions and stamp & substrate surface chemistries.
10.4 Applications
Although conventional patterning techniques, such as 10.4.1 Unconventional Electronic Systems
photolithography or electron beam lithography, have the
required resolution, they are not appropriate because A relatively new direction in electronics research seeks
they are expensive and generally require multiple pro- to establish low-cost plastic materials, substrates and
cessing steps with resists, solvents and developers that printing techniques for large-area flexible electronic
can be difficult to use with organic active materials devices, such as paperlike displays. These types of
and plastic substrates. Microcontact and nanotransfer novel devices can complement those (including high-
printing are both particularly well suited for this appli- density memories and high-speed microprocessors)
cation. They can be combined and matched with other that are well suited to existing inorganic (such as
techniques, such as ink-jet or screen printing, to form
a complete system for patterning all layers in practical
plastic electronic devices [10.45]. We have focused our Semiconductor
efforts partly on unusual electronic systems such as flex-
ible plastic circuits and devices that rely on electrodes L
Source Drain
patterned on curved objects such as microcapillaries
and optical fibers. We have also explored photonic sys- Dielectric
tems such as distributed feedback structures for lasers
and other integrated optical elements that demand sub- Gate
micron features. The sections below highlight several
examples in each of these areas. Substrate
Isd (µA)
Fig. 10.14 Schematic cross-sectional view (left) and elec-
trical performance (right) of an organic thin film transistor 6 Vg from 0 to –50 V
with microcontact printed source and drain electrodes. The in steps of –10 V
structure consists of a substrate (PET), a gate electrode 4
(indium tin oxide), a gate dielectric (spin-cast layer of
organosilsesquioxane), source and drain electrodes (20 nm 2
Au and 1.5 nm Ti), and a layer of the organic semiconduc-
tor pentacene. The electrical properties of this device are
0
comparable to or better than those that use pentacene with
photolithographically defined source/drain electrodes and 0 –10 –20 –30 –40 –50
Vsd (V)
inorganic dielectrics, gates and substrates
Stamping Techniques for Micro- and Nanofabrication 10.4 Applications 323
silicon) electronics technologies. High-resolution pat- eas, (3) strategies for achieving densities of defects
Part A 10.4
terning methods for defining the separation between the that are as good as those observed with photolithogra-
source and drain electrodes (the channel length) of tran- phy when the patterning is performed outside of clean
sistors in these plastic circuits are particularly important room facilities, (4) methods for removing the printed
because this dimension determines current output and SAMs to allow good electrical contact of the electrodes
other important characteristics [10.46].
Figure 10.14 illustrates schematically a cross-
sectional view of a typical organic transistor. The frame a) Clean stam p Stamp
on the right shows the electrical switching charac-
teristics of a device that uses source/drain electrodes
of Au patterned by μCP, a dielectric layer of an
organosilsesquioxane, a gate of indium tin oxide (ITO),
and a PET substrate. The effective semiconductor mo- b) Ink, align Plastic sheet
bility extracted from these data is comparable to those
measured in devices that use the same semiconductor
(pentacene in this case) with inorganic substrates and Au/Ti
dielectrics, and gold source/drain electrodes defined by
photolithography. Our recent work [10.1, 31, 47] with
μCP in the area of plastic electronics demonstrates:
(1) methods for using cylindrical roller stamps mounted
c) Initiate contact
on fixed axles for printing in a continuous reel-to-reel
fashion, high-resolution source/drain electrodes in ul-
trathin gold and silver deposited from solution at room
temperature using electroless deposition, (2) techniques
for performing registration and alignment of the printed
features with other elements of a circuit over large ar- d) Peel sheet away
25 µm
Fig. 10.16a–d Schematic illustration of fabrication steps
for microcontact printing over large areas onto plastic
sheets. The process begins with cleaning the stamp using
a conventional adhesive roller lint remover. This procedure
effectively removes dust particles. To minimize distortions,
the stamp rests face-up on a flat surface and it is not
manipulated directly during the printing. Alignment and
registration are achieved with alignment marks on one side
of the substrate and the stamp. By bending the plastic sheet,
contact is initiated on one side of the stamp; the contact line
is then allowed to progress gradually across the stamp. This
Fig. 10.15 Image of a flexible plastic active matrix back- approach avoids formation of air bubbles that can frus-
plane circuit whose finest features (transistor source/drain trate good contact. After the substrate is in contact with the
electrodes and related interconnects) are patterned by stamp for a few seconds, the plastic substrate is separated
microcontact printing. The circuit rests partly on the elas- from the stamp by peeling it away beginning in one corner.
tomeric stamp that was used for printing. The circuit Good registration (maximum cumulative distortions of less
consists of a square array of interconnected organic tran- than 50 μm over an area of 130 cm2 ) and low defect den-
sistors, each of which acts locally as a voltage-controlled sity can be achieved with this simple approach. It is also
switch to control the color of an element in the display. The well suited for use with rigid composite stamps designed
inset shows an optical micrograph of one of the transistors to reduce the level of distortions even further
324 Part A Nanostructures, Micro-/Nanofabrication and Materials
with organic semiconductors deposited on top of them, ment that controls the color of a display pixel [10.25,
Part A 10.4
and (5) materials and fabrication sequences that can ef- 48]. The transistors themselves have the layout illus-
ficiently exploit these printed electrodes for working trated in Fig. 10.11, and they use similar materials. The
organic TFTs in large-scale circuits. semiconductor in this image is blue (pentacene), the
Figure 10.15 provides an image of a large-area plas- source/drain level is Au, the ITO appears green in the
tic circuit with critical features defined by μCP. This optical micrograph in the inset. Part of the circuit rests
circuit is a flexible active matrix backplane for a dis- on the stamp that was used for μCP. The smallest
play. It consists of a square array of interconnected features are the source and drain electrodes (≈ 15 μm
transistors, each of which serves as a switching ele- lines), the interconnecting lines (≈ 15 μm lines), and
the channel length of the transistor (≈ 15 μm). This
circuit incorporates five layers of material patterned
+ with good registration of the source/drain, gate, and
E semiconductor levels. The simple printing approach is
illustrated in Fig. 10.16 [10.25]. Just before use, the
–
surface of the stamp is cleaned using a conventional ad-
hesive roller lint remover; this procedure removes dust
from the stamp in such a way that does not contaminate
or damage its surface. Inking the stamp and placing it
face-up on a flat surface prepares it for printing. Match-
ing the cross-hair alignment marks on the corners of
one edge of the stamp with those patterned in the ITO
brings the substrate into registration with the stamp.
During this alignment, features on the stamp are viewed
directly through the semitransparent substrate. By bend-
ing the PET sheet, contact with the stamp is initiated
on the edge of the substrate that contains the cross-hair
marks. Gradually unbending the sheet allows contact
to progress across the rest of the surface. This printing
procedure is attractive because it avoids distortions that
can arise when directly manipulating the flexible rubber
stamp. It also minimizes the number and size of trapped
air pockets that can form between the stamp and sub-
strate. Careful measurements performed after etching
the unprinted areas of the gold show that over the entire
6 × 6 area of the circuit, (1) the overall alignment ac-
curacy for positioning the stamp relative to the substrate
Fig. 10.17 Schematic exploded view of the components of a pixel in (the offset of the center of the distribution of registration
an electronic paperlike display (bottom frame) that uses a microcon- errors) is ≈ 50–100 μm, even with the simple approach
tact printed flexible active matrix backplane circuit (illustration near used here, and (2) the distortion in the positions of fea-
the bottom frame). The circuit is laminated against an unpatterned tures in the source/drain level, when referenced to the
thin sheet of electronic ink (top frame) that consists of a monolayer gate level, can be as small as ≈ 50 μm (the full width at
of transparent polymer microcapsules (diameter ≈ 100 μm). These half maximum of the distribution of registration errors).
capsules contain a heavily dyed black fluid and a suspension of These distortions represent the cumulative effects of de-
charged white pigment particles (see right inset). When one of the formations in the stamp and distortions in the gate and
transistors turns on, electric fields develop between an unpatterned column electrodes that may arise during the patterning
transparent frontplane electrode (indium tin oxide) and a backplane and processing of the flexible PET sheet. The density
electrode that connects to the transistor. Electrophoretic flow drives of defects in the printed patterns is comparable to (or
the pigment particles to the front or the back of the display, depend- smaller than) that in resist patterned by contact-mode
ing on the polarity of the field. This flow changes the color of the photolithography when both procedures are performed
pixel, as viewed from the front of the display, from black to white outside of a clean-room facility (when dust is the domi-
or vica versa nant source of defects).
Stamping Techniques for Micro- and Nanofabrication 10.4 Applications 325
Part A 10.4
Isd (µA)
40
100 µm
30
Vg = 100 V
20
10
0
0 20 40 60 80 100
Vsd (V)
Vout (V)
40
30
20
10
0
0 10 20 30 40
Vin (V)
Au/SiNX /Ti-Au
Ethylbenzene
800 Trilayer
Photolith
Part A 10.4
Coupling coefficient
ing this structure into the core of a larger microcapillary
1 that also supports a printed microcoil completes the
0.8 transformer [10.55]. This type of device shows good
0.6 coupling coefficients up to relatively high frequencies.
Examples of other optoelectronic components appear
0.4 in fiber optics where microfabricated on-fiber structures
0.3
0.2 Photolithography,
reactive ion etching
100 µm ≈ 1 mm
300 nm
0.1 Oxide
1 10 100 1000 ≈ 1 mm Silicon
Frequency (kHz)
Fig. 10.22 The inset shows a concentric microtransformer
formed using microcoils printed onto two different micro- 50–150 nm
capillary tubes. The smaller of the tubes (outer diameter
135 μm) has a ferromagnetic wire threaded through its Cast elastomer against
core. The larger one (outer diameter 350 μm) has the etched oxide, remove from oxide
smaller tube threaded through its core. The resulting
structure is a microtransformer that shows good coupling Elastomer
coefficients at frequencies up to ≈ 1 MHz. The graph
shows its performance
Fig. 10.24 The top frames gives a schematic illustration Intensity (arb. units)
of steps for microcontact printing high-resolution gratings
directly onto the top surfaces of ridge waveguides. The
printing defines a sacrificial etch mask of gold which is
subsequently removed. Producing this type of structure
with photolithography is difficult because of severe thick-
ness nonuniformities that appear in photoresist spin-cast on
this type of nonplanar substrate. The upper bottom frame
shows a top view optical micrograph of printed gold lines
on the ridge waveguides. The lower bottom frame shows
the emission output of a plastic photopumped laser that
uses the printed structure and a thin evaporated layer of 620 630 640 650 660
Wavelength (nm)
gain media
Stamping Techniques for Micro- and Nanofabrication 10.5 Conclusions 329
Part A 10.5
Intensity (arb. units)
plastic lasers that use microcontact printed resonators
based on surface relief distributed Bragg reflectors (DBRs)
and distributed feedback gratings (DFBs) on glass sub-
≈ 3 mm ≈ 3 mm
≈ 1 mm strates. The grating periods are ≈ 600 nm in both cases.
The lasers use thin film plastic gain media deposited onto
the printed gratings. This layer forms a planar waveguide
that confines the light to the surface of the substrate. The
laser shows emission over a narrow wavelength range, with
a width that is limited by the resolution of the spectrom-
eter used to characterize the output. In both cases, the
emission profiles, the lasing thresholds and other charac-
640 645 650 655 660 665 teristics of the devices are comparable to similar lasers that
Wavelength (nm) use resonators formed by high-resolution projection-mode
Intensity (arb. units)
photolithography
≈ 1 mm
with photolithography. For example, μCP can be used
≈ 1 mm to form DFB resonators directly on the top surfaces
of ridge waveguides [10.58]. Figure 10.25 illustrates
the procedures. The bottom left frame shows an opti-
cal micrograph of the printed gold lines. Sublimation
of a ≈ 200 nm film of tris(8-hydroxyquinoline) alu-
minum (Al) doped with 0.5–5.0 wt. % of the laser dye
DCMII onto the resonators produces waveguide DFB
lasers. The layer of gain material itself provides a planar
615 620 625 630 635 640 645 650 waveguide with air and polymer as the cladding lay-
Wavelength (nm)
ers. The relief waveguide provides lateral confinement
of the light. Photopumping these devices with the output
observed at wavelengths corresponding to the third har- of a pulsed nitrogen laser (≈ 2 ns, 337 nm) causes lasing
monic of the gratings. These characteristics are similar due to Bragg reflections induced by the DFB structures
to those observed in lasers that use resonators generated on the top surfaces of the ridge waveguides. Some of the
with photolithography and are better than those that use laser emission scatters out of the plane of the waveguide
imprinted polymers [10.57]. at an angle allowed by phase matching conditions. In
Contact printing not only provides a route to this way, the grating also functions as an output coupler
low-cost equivalents of gratings fabricated with other and offers a convenient way to characterize the laser
approaches, but also allows the fabrication of struc- emission. The bottom right frame of Fig. 10.22 shows
tures that would be difficult or impossible to generate the emission profile.
10.5 Conclusions
This chapter provides an overview of two contact print- makes it crucial to develop new methods for fabricating
ing techniques that are capable of micrometer and the relevant test structures and devices. The simplic-
submicrometer resolution. It also illustrates some ap- ity of these techniques together with the interesting and
plications of these methods that may provide attractive subtle materials science, chemistry, and physics associ-
alternatives to more established lithographic methods. ated with them make this a promising area for basic and
The growing interest in nanoscience and technology applied study.
330 Part A Nanostructures, Micro-/Nanofabrication and Materials
References
Part A 10
10.1 C.A. Mirkin, J.A. Rogers: Emerging methods for lowed by chemical etching, Appl. Phys. Lett. 63,
micro- and nanofabrication, MRS Bulletin 26, 506– 2002–2004 (1993)
507 (2001) 10.18 Y. Xia, G.M. Whitesides: Soft lithography, Angew.
10.2 H.I. Smith, H.G. Craighead: Nanofabrication, Phys. Chem. Int. Ed. 37, 550–575 (1998)
Today 43, 24–43 (1990) 10.19 Y. Xia, J.A. Rogers, K.E. Paul, G.M. Whitesides:
10.3 W.M. Moreau (Ed.): Semiconductor Lithography: Unconventional methods for fabricating and pat-
Principles and Materials (Plenum, New York 1988) terning nanostructures, Chem. Rev. 99, 1823–1848
10.4 S. Matsui, Y. Ochiai: Focused ion beam applications (1999)
to solid state devices, Nanotechnology 7, 247–258 10.20 J.A. Rogers, R.J. Jackman, J.L. Wagener, A.M. Veng-
(1996) sarkar, G.M. Whitesides: Using microcontact print-
10.5 J.M. Gibson: Reading and writing with electron ing to generate photomasks on the surface of
beams, Phys. Today 50, 56–61 (1997) optical fibers: A new method for producing in-fiber
10.6 L.L. Sohn, R.L. Willett: Fabrication of nano- gratings, Appl. Phys. Lett. 70, 7–9 (1997)
structures using atomic-force microscope-based 10.21 B. Michel, A. Bernard, A. Bietsch, E. Delamarche,
lithography, Appl. Phys. Lett. 67, 1552–1554 (1995) M. Geissler, D. Juncker, H. Kind, J.P. Renault,
10.7 E. Betzig, K. Trautman: Near-field optics – Mi- H. Rothuizen, H. Schmid, P. Schmidt-Winkel,
croscopy, spectroscopy, and surface modification R. Stutz, H. Wolf: Printing meets lithography: soft
beyond the diffraction limit, Science 257, 189–195 approaches to high-resolution printing, IBM J. Res.
(1992) Dev. 45, 697–719 (2001)
10.8 A.J. Bard, G. Denault, C. Lee, D. Mandler, D.O. Wipf: 10.22 J.A. Rogers: Rubber stamping for plastic electronics
Scanning electrochemical microscopy: A new tech- and fiber optics, MRS Bulletin 26, 530–534 (2001)
nique for the characterization and modification of 10.23 N.B. Larsen, H. Biebuyck, E. Delamarche, B. Michel:
surfaces, Acc. Chem. Res. 23, 357 (1990) Order in microcontact printed self-assembled
10.9 J.A. Stroscio, D.M. Eigler: Atomic and molecular monolayers, J. Am. Chem. Soc. 119, 3017–3026 (1997)
manipulation with the scanning tunneling micro- 10.24 H.A. Biebuyck, G.M. Whitesides: Self-organization
scope, Science 254, 1319–1326 (1991) of organic liquids on patterned self-assembled
10.10 J. Nole: Holographic lithography needs no mask, monolayers of alkanethiolates on gold, Langmuir
Laser Focus World 33, 209–212 (1997) 10, 2790–2793 (1994)
10.11 A.N. Broers, A.C.F. Hoole, J.M. Ryan: Electron beam 10.25 J.A. Rogers, Z. Bao, K. Baldwin, A. Dodabalapur,
lithography – Resolution limits, Microelectron. B. Crone, V.R. Raju, V. Kuck, H. Katz, K. Amundson,
Eng. 32, 131–142 (1996) J. Ewing, P. Drzaic: Paper-like electronic displays:
10.12 A.N. Broers, W. Molzen, J. Cuomo, N. Wittels: Large area, rubber stamped plastic sheets of elec-
Electron-beam fabrication of 80 metal structures, tronics and electrophoretic inks, Proc. Natl. Acad.
Appl. Phys. Lett. 29, 596 (1976) Sci. USA 98, 4835–4840 (2001)
10.13 G.D. Aumiller, E.A. Chandross, W.J. Tomlinson, 10.26 J.L. Wilbur, H.A. Biebuyck, J.C. MacDonald,
H.P. Weber: Submicrometer resolution replication G.M. Whitesides: Scanning force microscopies
of relief patterns for integrated optics, J. Appl. can image patterned self-assembled monolayers,
Phys. 45, 4557–4562 (1974) Langmuir 11, 825–831 (1995)
10.14 Y. Xia, J.J. McClelland, R. Gupta, D. Qin, X.-M. Zhao, 10.27 J.C. Love, D.B. Wolfe, M.L. Chabinyc, K.E. Paul,
L.L. Sohn, R.J. Celotta, G.M. Whiteside: Replica G.M. Whitesides: Self-assembled monolayers of
molding using polymeric materials: A practical step alkanethiolates on palladium are good etch resists,
toward nanomanufacturing, Adv. Mater. 9, 147–149 J. Am. Chem. Soc. 124, 1576–1577 (2002)
(1997) 10.28 H. Schmid, B. Michel: Siloxane polymers for
10.15 T. Borzenko, M. Tormen, G. Schmidt, L.W. Molen- high-resolution, high-accuracy soft lithography,
kamp, H. Janssen: Polymer bonding process for Macromolecules 33, 3042–3049 (2000)
nanolithography, Appl. Phys. Lett. 79, 2246–2248 10.29 K. Choi, J.A. Rogers: A photocurable poly(dimethyl-
(2001) siloxane) chemistry for soft lithography in the
10.16 H. Hua, Y. Sun, A. Gaur, M.A. Meitl, L. Bilhaut, nanometer regime, J. Am. Chem. Soc. 125, 4060–
L. Rotinka, J. Wang, P. Geil, M. Shim, J.A. Rogers: 4061 (2003)
Polymer imprint lithography with molecular-scale 10.30 J.A. Rogers, K.E. Paul, G.M. Whitesides: Quantifying
resolution, Nano Lett. 4(12), 2467–2471 (2004) distortions in soft lithography, J. Vac. Sci. Technol.
10.17 A. Kumar, G.M. Whitesides: Features of gold hav- B 16, 88–97 (1998)
ing micrometer to centimeter dimensions can be 10.31 J. Tate, J.A. Rogers, C.D.W. Jones, W. Li, Z. Bao,
formed through a combination of stamping with D.W. Murphy, R.E. Slusher, A. Dodabalapur,
an elastomeric stamp and an alkanethiol ink fol- H.E. Katz, A.J. Lovinger: Anodization and micro-
Stamping Techniques for Micro- and Nanofabrication References 331
contact printing on electroless silver: solution- dimensional and multilayer nanostructures formed
Part A 10
based fabrication procedures for low voltage by nanotransfer printing, Nano Lett. 3, 1223–1227
organic electronic systems, Langmuir 16, 6054– (2003)
6060 (2000) 10.45 Z. Bao, J.A. Rogers, H.E. Katz: Printable organic and
10.32 Y. Xia, E. Kim, G.M. Whitesides: Microcontact print- polymeric semiconducting materials and devices,
ing of alkanethiols on silver and its application to J. Mater. Chem. 9, 1895–1904 (1999)
microfabrication, J. Electrochem. Soc. 143, 1070– 10.46 J.A. Rogers, Z. Bao, A. Dodabalapur, A. Makhija:
1079 (1996) Organic smart pixels and complementary inverter
10.33 Y.N. Xia, X.M. Zhao, E. Kim, G.M. Whitesides: A se- circuits formed on plastic substrates by casting,
lective etching solution for use with patterned printing and molding, IEEE Electron Dev. Lett. 21,
self-assembled monolayers of alkanethiolates on 100–103 (2000)
gold, Chem. Mater. 7, 2332–2337 (1995) 10.47 J.A. Rogers, Z. Bao, A. Makhija: Non-photo-
10.34 R.J. Jackman, J. Wilbur, G.M. Whitesides: Fab- lithographic fabrication sequence suitable for
rication of submicrometer features on curved reel-to-reel production of high performance or-
substrates by microcontact printing, Science 269, ganic transistors and circuits that incorporate
664–666 (1995) them, Adv. Mater. 11, 741–745 (1999)
10.35 R.J. Jackman, S.T. Brittain, A. Adams, M.G. Pren- 10.48 P. Mach, S. Rodriguez, R. Nortrup, P. Wiltzius,
tiss, G.M. Whitesides: Design and fabrication J.A. Rogers: Active matrix displays that use printed
of topologically complex, three-dimensional mi- organic transistors and polymer dispersed liquid
crostructures, Science 280, 2089–2091 (1998) crystals on flexible substrates, Appl. Phys. Lett. 78,
10.36 J.A. Rogers, R.J. Jackman, G.M. Whitesides: Mi- 3592–3594 (2001)
crocontact printing and electroplating on curved 10.49 J.A. Rogers: Toward paperlike displays, Science 291,
substrates: A new means for producing free- 1502–1503 (2001)
standing three-dimensional microstructures with 10.50 Y.-L. Loo, T. Someya, K.W. Baldwin, P. Ho, Z. Bao,
possible applications ranging from micro-coil A. Dodabalapur, H.E. Katz, J.A. Rogers: Soft, con-
springs to coronary stents, Adv. Mater. 9, 475–477 formable electrical contacts for organic transistors:
(1997) High resolution circuits by lamination, Proc. Natl.
10.37 Y.-L. Loo, R.W. Willett, K. Baldwin, J.A. Rogers: Ad- Acad. Sci. USA 99, 10252–10256 (2002)
ditive, nanoscale patterning of metal films with 10.51 C. Kim, P.E. Burrows, S.R. Forrest: Micropattern-
a stamp and a surface chemistry mediated transfer ing of organic electronic devices by cold-welding,
process: applications in plastic electronics, Appl. Science 288, 831–833 (2000)
Phys. Lett. 81, 562–564 (2002) 10.52 C. Kim, M. Shtein, S.R. Forrest: Nanolithography
10.38 Y.-L. Loo, R.W. Willett, K. Baldwin, J.A. Rogers: based on patterned metal transfer and its applica-
Interfacial chemistries for nanoscale transfer print- tion to organic electronic devices, Appl. Phys. Lett.
ing, J. Am. Chem. Soc. 124, 7654–7655 (2002) 80, 4051–4053 (2002)
10.39 Y.-L. Loo, J.W.P. Hsu, R.L. Willett, K.W. Baldwin, 10.53 J.A. Rogers, R.J. Jackman, G.M. Whitesides, D.L. Ol-
K.W. West, J.A. Rogers: High-resolution transfer son, J.V. Sweedler: Using microcontact printing to
printing on GaAs surfaces using alkane dithiol self- fabricate microcoils on capillaries for high reso-
assembled monolayers, J. Vac. Sci. Technol. B 20, lution 1 H-NMR on nanoliter volumes, Appl. Phys.
2853–2856 (2002) Lett. 70, 2464–2466 (1997)
10.40 G.S. Ferguson, M.K. Chaudhury, G.B. Sigal, 10.54 J.A. Rogers, R.J. Jackman, G.M. Whitesides: Con-
G.M. Whitesides: Contact adhesion of thin gold- structing single and multiple helical microcoils and
films on elastomeric supports – cold welding under characterizing their performance as components of
ambient conditions, Science 253, 776–778 (1991) microinductors and microelectromagnets, J. Micro-
10.41 W. Zhang, S.Y. Chou: Multilevel nanoimprint electromech. Syst. 6, 184–192 (1997)
lithography with submicron alignment over 4 in Si 10.55 R.J. Jackman, J.A. Rogers, G.M. Whitesides: Fab-
wafers, Appl. Phys. Lett. 79, 845–847 (2001) rication and characterization of a concentric,
10.42 N. Bowden, S. Brittain, A.G. Evans, J.W. Hutchin- cylindrical microtransformer, IEEE Trans. Magn. 33,
son, G.M. Whitesides: Spontaneous formation of 2501–2503 (1997)
ordered structures in thin films of metals supported 10.56 J.A. Rogers, M. Meier, A. Dodabalapur: Using
on an elastomeric polymer, Nature 393, 146–149 stamping and molding techniques to produce dis-
(1998) tributed feedback and Bragg reflector resonators
10.43 E. Menard, L. Bilhaut, J. Zaumseil, J.A. Rogers: Im- for plastic lasers, Appl. Phys. Lett. 73, 1766–1768
proved surface chemistries, thin film deposition (1998)
techniques, and stamp designs for nanotransfer 10.57 M. Berggren, A. Dodabalapur, R.E. Slusher,
printing, Langmuir 20, 6871–6878 (2004) A. Timko, O. Nalamasu: Organic solid-state lasers
10.44 J. Zaumseil, M.A. Meitl, J.W.P. Hsu, B. Acharya, with imprinted gratings on plastic substrates, Appl.
K.W. Baldwin, Y.-L. Loo, J.A. Rogers: Three- Phys. Lett. 72, 410–411 (1998)
332 Part A Nanostructures, Micro-/Nanofabrication and Materials
10.58 J.A. Rogers, M. Meier, A. Dodabalapur: Distributed nanoscale printing and molding on non-planar
Part A 10
feedback ridge waveguide lasers fabricated by substrates, Appl. Phys. Lett. 74, 3257–3259 (1999)
333
Material Aspe
Christian A. Zorman, Mehran Mehregany
11. Material Aspects of Micro-
and Nanoelectromechanical Systems
Part A 11
11.1 Silicon ................................................. 333
One of the more significant technological
11.1.1 Single-Crystal Silicon .................... 333
achievements during the last 20 years has been
11.1.2 Polycrystalline
the development of MEMS and its new offshoot,
and Amorphous Silicon ................. 336
NEMS. These developments were made possible 11.1.3 Porous Silicon .............................. 338
by significant advancements in the mater- 11.1.4 Silicon Dioxide ............................. 339
ials and processing technologies used in the 11.1.5 Silicon Nitride .............................. 340
fabrication of MEMS and NEMS devices. While
initial developments capitalized on a mature 11.2 Germanium-Based Materials ................. 340
Si infrastructure built for the integrated cir- 11.2.1 Polycrystalline Ge ......................... 340
11.2.2 Polycrystalline SiGe ...................... 341
cuit (IC) industry, recent advances have come
about using materials and processes not as- 11.3 Metals ................................................. 341
sociated with IC fabrication, a trend that is
11.4 Harsh-Environment Semiconductors ...... 343
likely to continue as new application areas
11.4.1 Silicon Carbide ............................. 343
emerge. 11.4.2 Diamond ..................................... 346
A well-rounded understanding of MEMS
and NEMS technology requires a basic knowl- 11.5 GaAs, InP, and Related III–V Materials .... 349
edge of the materials used to construct the 11.6 Ferroelectric Materials........................... 350
devices, since material properties often gov-
ern device performance and dictate fabrication 11.7 Polymer Materials ................................. 351
approaches. An understanding of the materials 11.7.1 Polyimide.................................... 351
11.7.2 SU-8........................................... 351
used in MEMS and NEMS involves an under-
11.7.3 Parylene ..................................... 352
standing of material systems, since such devices
11.7.4 Liquid Crystal Polymer................... 352
are rarely constructed of a single material but
rather a collection of materials working in con- 11.8 Future Trends....................................... 352
junction with each other to provide critical References .................................................. 353
functions. It is from this perspective that the
following chapter is constructed. A preview of
the materials selected for inclusion in this chap- significant size. It does, however, present a se-
ter is presented in Table 11.1. It should be clear lection of some of the more important material
from this table that this chapter is not a sum- systems, and especially those that illustrate the
mary of all materials used in MEMS and NEMS, importance of viewing MEMS and NEMS in terms of
as such a work would itself constitute a text of material systems.
11.1 Silicon
11.1.1 Single-Crystal Silicon the piezoresistive effect in germanium (Ge) and Si was
first identified [11.1]. It was discovered that the piezore-
Use of silicon (Si) as a material for microfabricated sen- sistive coefficients of Si were significantly higher than
sors dates back to the middle of the 20th century when those associated with metals used in conventional strain
334 Part A Nanostructures, Micro-/Nanofabrication and Materials
Table 11.1 Distinguishing characteristics and application examples of selected materials for MEMS and NEMS
electrostatic actuation
Silicon dioxide (SiO2 ) Insulating, etched by HF, Sacrificial layer in polysilicon
compatible with polysilicon surface micromachining,
passivation layer for devices
Silicon nitride (Si3 N4 , Six Ny ) Insulating, chemically resistant, Isolation layer for electrostatic
mechanically durable devices, membrane and bridge
material
Polycrystalline germanium Deposited at low temperatures Integrated surface micromachined
(polyGe), Polycrystalline MEMS
silicon-germanium (poly SiGe)
Gold (Au), aluminum (Al) Conductive thin films, flexible Innerconnect layers, masking
deposition techniques layers, electromechanical switches
Bulk Ti High strength, corrosion resistant Optical MEMS
Nickel-iron (NiFe) Magnetic alloy Magnetic actuation
Titanium-nickel (TiNi) Shape-memory alloy Thermal actuation
Silicon carbide (SiC) diamond Electrically and mechanically Harsh-environment MEMS,
stable at high temperatures, high-frequency MEMS/NEMS
chemically inert, high Young’s
modulus to density ratio
Gallium arsenide (GaAs), indium Wide bandgap, epitaxial growth on RF MEMS, optoelectronic devices,
phosphide (InP), indium arsenide related ternary compounds single-crystal bulk and surface
(InAs) and related materials micromachining
Lead zirconate titanate (PZT) Piezoelectric material Mechanical sensors and actuators
Polyimide Chemically resistant, high- Mechanically flexible MEMS,
temperature polymer bioMEMS
SU-8 Thick, photodefinable resist Micromolding, High-aspect-ratio
structures
Parylene Biocompatible polymer, deposited Protective coatings, molded
at room temperature by CVD polymer structures
Liquid crystal polymer Chemically resistant, low moisture bioMEMS, RF MEMS
permeability, insulating
gauges; and this finding initiated the development of tures could be integrated with Si IC devices marked
Si-based strain gauge devices, and along with Si bulk the advent of MEMS with Si firmly positioned as the
micromachining techniques, piezoresistive Si pressure primary MEMS material.
sensors during the 1960s and 1970s. The subsequent For MEMS applications, single-crystal Si serves
development of Si surface micromachining techniques several key functions. Single-crystal Si is one of the
along with the recognition that micromachined Si struc- most versatile materials for bulk micromachining due
Material Aspects of Micro- and Nanoelectromechanical Systems 11.1 Silicon 335
to the availability of anisotropic etching processes in a highly effective anisotropic Si etching technique that
conjunction with good mechanical properties. Single- can be used to generate patterns that are independent of
crystal Si has favorable mechanical properties (i. e., crystalline orientation. Fluorinated compounds such as
a Young’s modulus of about 190 GPa), enabling its use CF4 , SF6 , and NF3 , or chlorinated compounds such as
as a material for membranes, resonant beams, and other CCl4 or Cl2 , sometimes mixed with He, O2 , or H2 , are
such structures. For surface micromachining applica- commonly used in Si RIE. The RIE process is highly
tions, single-crystal Si substrates are used primarily as directional, which enables direct lateral pattern trans-
Part A 11.1
mechanical platforms on which device structures are fer from an overlying masking material to the etched
fabricated, although the advent of silicon-on-insulator Si surface. SiO2 thin films are often used as masking
(SOI) substrates enables the fabrication of single-crystal and sacrificial layers owing to its chemical durability
Si surface micromachined structures by using the buried under these plasma conditions. Process limitations (i. e.,
oxide as a sacrificial layer. Use of high-quality single- etch rates) restrict the etch depths of conventional Si
crystal wafers enables the fabrication of integrated RIE to less than 10 μm; however, a process called deep
MEMS devices, at least for materials and processes that reactive ion etching (DRIE) has extended the use of
are compatible with Si ICs. anisotropic dry etching to depths well beyond several
From the materials perspective, single-crystal Si hundred micrometer.
is a relatively easy material to bulk micromachine Using the aforementioned processes and tech-
due to the availability of anisotropic etchants such as niques, a wide variety of microfabricated devices have
potassium hydroxide (KOH) and tetramethyl-aluminum been made from single-crystal Si, such as piezoresis-
hydroxide (TMAH) that attack the (100) and (110) Si tive pressure sensors, accelerometers, and mechanical
crystal planes significantly faster than the (111) crys- resonators, to name a few. Using nearly the same ap-
tal planes. For example, the etching rate ratio of (100) proaches but on a smaller scale, top-down nanomachin-
to (111) planes in Si is about 400:1 for a typical ing techniques have been used to fabricate nanoelec-
KOH/water etching solution. Silicon dioxide (SiO2 ), tromechanical devices from single-crystal Si. Single-
silicon nitride (Si3 N4 ), and some metallic thin films crystal Si is particularly well suited for nanofabrication
(e.g., Cr, Au, etc.) provide good etch masks for most Si because high crystal quality substrates with very smooth
anisotropic etchants. Heavily boron-doped Si is an ef- surfaces are readily available. By coupling electron-
fective etch stop for some liquid reagents. Boron-doped beam (e-beam) lithographic techniques with conven-
etch stops are often less than 10 μm thick, since the tional Si etching, device structures with submicrome-
boron concentration in Si must exceed 7 × 1019 cm3 for ter dimensions have been fabricated. Submicrometer,
the etch stop to be effective and the doping is done by single-crystal Si nanomechanical structures have been
thermal diffusion. Ion implantation can be used to cre- successfully micromachined from bulk Si wafers [11.2]
ate a subsurface etch stop layer; however, the practical
limit is a few micrometer.
In contrast to anisotropic etching, isotropic etch-
ing exhibits no selectivity to the various crystal planes.
Commonly used isotropic Si etchants consist of hy-
drofluoric (HF) and nitric (HNO3 ) acid mixtures in
water or acetic acid (CH3 COOH), with the etch rate
dependent on the ratio of HF to HNO3 . From a pro-
cessing perspective, isotropic etching of Si is commonly
used for removal of work-damaged surfaces, creation
of structures in single-crystal slices, and patterning of
single-crystal or polycrystalline films.
Well-established dry etching processes are routinely
used to pattern single-crystal Si. The process spectrum
ranges from physical techniques such as sputtering and
ion milling to chemical techniques such as plasma etch-
ing. Reactive ion etching (RIE) is the most commonly Fig. 11.1 A collection of Si nanoelectromechanical beam
used dry etching technique for Si patterning. By com- resonators fabricated from a single-crystal Si substrate
bining both physical and chemical processes, RIE is (courtesy M. Roukes, Caltech)
336 Part A Nanostructures, Micro-/Nanofabrication and Materials
and silicon-on-insulator (SOI) wafers [11.3]. In the For MEMS and IC applications, polysilicon films
former case, an isotropic Si etch was performed to re- are commonly deposited using a process known as low-
lease the device structures, whereas in the latter case, pressure chemical vapor deposition (LPCVD). The typ-
the 50–200 nm structures were released by dissolving ical polysilicon LPCVD reactor is based on a hot-wall,
the underlying oxide layer in HF. An example of na- resistance-heated furnace. Typical processes are per-
noelectromechanical beam structures fabricated from formed at temperatures ranging from 580 to 650 ◦ C and
a single-crystal Si substrate is shown in Fig. 11.1. pressures from 100 to 400 mtorr. The most commonly
Part A 11.1
polysilicon films in piezoresistive strain gauges, achiev- Many device designs require polysilicon thick-
ing gauge factors as high as 15 for a single strip nesses that are not readily achievable using conven-
sensor. tional LPCVD polysilicon due to the low deposition
The thermal conductivity of polysilicon is a strong rates associated with such systems. For these appli-
function of its microstructure, and therefore the con- cations, epitaxial Si reactors can be used to grow
ditions used during deposition [11.4]. For fine-grained polysilicon films. Unlike conventional LPCVD pro-
films, the thermal conductivity is about 25% of the value cesses with deposition rates of less than 100 Å/min,
Part A 11.1
of single-crystal Si. For thick films with large grains, the epitaxial processes have deposition rates on the order of
thermal conductivity ranges between 50% and 85% of 1 μm/min [11.11]. The high deposition rates result from
the single-crystal value. the much higher substrate temperatures (> 1000 ◦ C)
Like the electrical and thermal properties of poly- and deposition pressures (> 50 torr) used in these pro-
silicon, the as-deposited residual stress in polysilicon cesses. The polysilicon films are usually deposited on
films depends on microstructure. For films deposited SiO2 sacrificial layers to enable surface micromachin-
under typical conditions (200 mtorr, 625 ◦ C), the as- ing. An LPCVD polysilicon seed layer is sometimes
deposited polysilicon films have compressive residual used in order to control nucleation, grain size, and
stresses. The highest compressive stresses are found in surface roughness. As with conventional polysilicon,
amorphous Si films and polysilicon films with a strong, the microstructure and residual stress of the epi-poly
columnar (110) texture. For films with fine-grained films, as they are known, are related to deposition con-
microstructures, the stress tends to be tensile. Annealing ditions. Compressive films generally have a mixture
can be used to reduce the compressive stress in as- of [110] and [311] grains [11.12, 13], while tensile
deposited polysilicon films. For instance, compressive films have a random mix of [110], [100], [111], and
residual stresses on the order of 500 MPa can be re- [311] grains [11.12]. The Young’s modulus of epi-
duced to less than 10 MPa by annealing the as-deposited poly measured from micromachined test structures is
films at 1000 ◦ C in a N2 ambient [11.7, 8]. Rapid ther- comparable with LPCVD polysilicon [11.13]. Mech-
mal annealing (RTA) provides an effective method of anical properties test structures [11.11–13], thermal
stress reduction in polysilicon films on temperature- actuators [11.11], electrostatically actuated accelerom-
sensitive substrates. Zhang et al. [11.9] reported that eters [11.11], and gryoscopes [11.14] have been fabri-
a 10 s anneal at 1100 ◦ C was sufficient to completely re- cated from these films.
lieve the stress in films that originally had a compressive As a low-temperature alternative to LPCVD polysil-
stress of about 340 MPa. RTA is particularly attractive icon, physical vapor deposition (PVD) techniques
in situations where the process parameters require a low have been developed to produce Si thin films on
thermal budget. temperature-sensitive substrates. Abe et al. [11.15] and
As an alternative to high-temperature annealing, Honer et al. [11.16] have developed sputtering pro-
Yang et al. [11.10] have developed an approach that
actually utilizes the residual stress characteristics of
Dep. rate (Å/min)
polysilicon deposited under various conditions to con- 800
struct polysilicon multilayers that have the desired
700
thickness and stress values. The multilayers are com-
prised of alternating tensile and compressive polysilicon 600
layers that are deposited in a sequential manner. The 500
tensile layers consist of fine-grained polysilicon grown 400
at a temperature of 570 ◦ C, while the compressive lay-
300
ers are made up of columnar polysilicon deposited at
615 ◦ C. The overall stress in the composite film depends 200
on the number of alternating layers and the thickness of 100
each layer. With the proper set of parameters, a com- 0
posite polysilicon multilayer can be deposited with near 500 550 600 650 700 750 800 850 900
Temperature (°C)
zero residual stress and no stress gradient. The process
achieves stress reduction without high-temperature an- Fig. 11.3 Deposition rate versus substrate temperature for in situ
nealing, a considerable advantage for integrated MEMS boron-doped ( ) and undoped ( ) polysilicon films grown by at-
processes. mospheric pressure chemical vapor deposition (after [11.5])
338 Part A Nanostructures, Micro-/Nanofabrication and Materials
cesses for polysilicon. Early work [11.15] emphasized gion, leading to the eventual removal of the fluorinated
the ability to deposit very smooth (2.5 nm) polysili- Si. The quality of the etched surface is related to the
con films on thermally oxidized wafers at reasonable density of holes at the surface, which is controlled by
deposition rates (19.1 nm/min) and with low resid- the applied current density. For high current densities,
ual compressive stresses. The process involved DC the density of holes is high and the etched surface is
magnitron sputtering from a Si target using an Ar smooth. For low current densities, the hole density is
sputtering gas, a chamber pressure of 5 mtorr, and low and clustered in highly localized regions associated
Part A 11.1
a power of 100 W. The authors reported that a post- with surface defects. Surface defects become enlarged
deposition anneal at 700 ◦ C in N2 for 2 h was needed by etching, which leads to the formation of pores. Pore
to crystallize the deposited film and perhaps lower size and density are related to the type of Si used and
the stress. Honer et al. [11.16] sought to develop the conditions of the electrochemical cell. Both single-
a polymer-friendly, Si-based surface micromachining crystal and polycrystalline Si can be converted to porous
process based on polysilicon sputtered onto polyimide Si.
and PSG sacrificial layers. To improve the conductiv- The large surface-to-volume ratios make porous Si
ity of the micromachined Si structures, the sputtered Si attractive for gaseous and liquid applications, including
films were sandwiched between two TiW cladding lay- filter membranes and absorbing layers for chemical and
ers. The device structures on polyimide were released mass sensing [11.18]. When single-crystal substrates
using oxygen plasma etching. The processing step with are used, the unetched porous layer remains single crys-
the highest temperature was, in fact, the polyimide talline and is suitable for epitaxial Si growth. It has
cure at 350 ◦ C. To test the robustness of the process, been shown that CVD coatings do not generally pen-
sputter-deposited Si microstructures were fabricated on etrate the porous regions, but rather overcoat the pores
substrates containing CMOS devices. As expected from at the surface of the substrate [11.19]. The formation
thermal budget considerations, the authors reported no of localized Si-on-insulator structures is therefore pos-
measurable degradation of device performance. sible by simply combining pore formation with epitaxial
PECVD has emerged as an alternative to LPCVD growth, followed by dry etching to create access holes
for the production of Si-based surface microma- to the porous region and thermal oxidation of the un-
chined structures on temperature-sensitive substrates. derlying porous region. A third application uses porous
Gaspar et al. [11.17] recently reported on the de- Si as a sacrificial layer for polysilicon and single-
velopment of surface micromachined microresonators crystalline Si surface micromachining. As shown by
fabricated from hydrogenated amorphous Si (a-Si:H) Lang et al. [11.19], the process involves the electri-
thin films deposited by PECVD. The vertically actuated cal isolation of the solid structural Si layer by either
resonators consisted of doubly-clamped microbridges pn-junction formation through selective doping or use
suspended over fixed Al electrodes. The a-Si:H films of electrically insulating thin films since the formation
were deposited using SiH4 and H2 precursors and PH3 of pores only occurs on electrically charged surfaces.
as a doping gas. The substrate temperature was held A weak Si etchant will aggressively attack the porous
to around 100 ◦ C, which enabled the use of photore- regions with little damage to the structural Si layers and
sist as a sacrificial layer. The microbridges consisted of can be used to release the devices.
a large paddle suspended by two thin paddle supports, Porous polysilicon is currently being developed as
with the paddle providing a large reflective surface for a structural material for chip-level vacuum packaging
optical detection of resonant frequency. The megahertz- [11.20]. In this example, a 1.5 μm thick polysilicon is
frequency resonators exhibited quality factors in the deposited onto a supporting PSG sacrificial layer, elec-
1 × 105 range when tested in vacuum. trochemically etched in an HF solution to render it
porous, and then annealed by RTA to reduce stress in
11.1.3 Porous Silicon the porous layer. When fabricated locally over a prefab-
ricated device structure (prior to release), the porous Si
Porous Si is produced by room temperature electro- forms a localized shell that will serve as a mechanical
chemical etching of Si in HF. If configured as an support for the main packaging structure. The porous
electrode in an HF-based electrochemical circuit, pos- structure enables an HF etch to remove the supporting
itive charge carriers (holes) at the Si surface facilitate PSG layer as well as any sacrificial oxide layers asso-
the exchange of F atoms with H atoms terminating the ciated with the prefabricated MEMS device. After the
Si surface. The exchange continues in the subsurface re- sacrificial etch, the packaging sequence is completed by
Material Aspects of Micro- and Nanoelectromechanical Systems 11.1 Silicon 339
depositing a polysilicon film by LPCVD at 179 mtorr result in LTO and PSG films that are slightly less dense
on the porous shell, thus fully encapsulating the device than thermal oxides due to the incorporation of hydro-
under vacuum conditions. This technique was used to gen in the films. LTO films can, however, be densified
package a microfabricated Pirani vacuum gauge, which by an annealing step at high temperature (1000 ◦ C). The
enabled an in situ measurement of pressure versus time. low density of LTO and PSG films is partially responsi-
The authors found no detectable change in pressure over ble for the increased etch rate in HF.
a 3-month period. Thermal SiO2 and LTO are electrical insulators
Part A 11.1
used in numerous MEMS applications. The dielectric
11.1.4 Silicon Dioxide constants of thermal oxide and LTO are 3.9 and 4.3,
respectively. The dielectric strength of thermal SiO2 is
Silicon dioxide (SiO2 ) is one of the most widely used 1.1 × 106 V/cm, and for LTO it is about 80% of that
materials in the fabrication of MEMS. In polysilicon value [11.21]. The stress in thermal SiO2 is compressive
surface micromachining, SiO2 is used as a sacrificial with a magnitude of about 300 MPa [11.21]. For LTO,
material since it can be easily dissolved using etchants however, the typical as-deposited residual stress is ten-
that do not attack polysilicon. SiO2 is widely used as sile, with a magnitude of about 100–400 MPa [11.21].
an etch mask for dry etching of thick polysilicon films The addition of phosphorus to LTO decreases the ten-
since it is chemically resistant to dry etching processes sile residual stress to about 10 MPa for phosphorus
for polysilicon. SiO2 films are also used as passiva- concentrations of 8% [11.22]. As with polysilicon, the
tion layers on the surfaces of environmentally sensitive properties of LTO and PSG are dependent on processing
devices. conditions.
The most common processes used to produce SiO2 Plasma enhanced chemical vapor deposition
films for polysilicon surface micromachining are ther- (PECVD) is another common method to produce oxides
mal oxidation and LPCVD. Thermal oxidation of Si of silicon. Using a plasma to dissociate the gaseous pre-
is performed at temperatures of 900–1200 ◦ C in the cursors, the deposition temperatures needed to deposit
presence of oxygen or steam. Since thermal oxidation PECVD oxide films is lower than for LPCVD films. For
is a self-limiting process, the maximum practical film this reason, PECVD oxides are quite commonly used as
thickness that can be obtained is about 2 μm, which is masking, passivation, and protective layers, especially
sufficient for many sacrificial applications. As noted by on devices that have been coated with metals.
its name, thermal oxidation of Si can only be performed Quartz is the crystalline form of SiO2 and has
on Si surfaces. interesting properties for MEMS. Quartz is optically
SiO2 films can be deposited on a wide variety of transparent, piezoelectric, and electrically insulating.
substrate materials by LPCVD. In general, LPCVD Like single-crystal Si, quartz substrates are available
provides a means for depositing thick (> 2 μm) SiO2 as high-quality, large-area wafers that can be bulk
films at temperatures much lower than thermal oxida- micromachined using anisotropic etchants. A short re-
tion. Known as low-temperature oxides, or LTO for view of the basics of quartz etching was written by
short, these films have a higher etch rate in HF than Danel et al. [11.23] and is recommended for those
thermal oxides, which translates to significantly faster interested in the subject. Quartz has recently become
release times when LTO films are used as sacrificial lay- a popular substrate material for microfluidic devices due
ers. Phosphosilicate glass (PSG) can be formed using to its optical, electronic, and chemical properties.
nearly the same deposition process as LTO by adding Another SiO2 -related material that has recently
a phosphorus-containing gas to the precursor flows. found uses in MEMS is spin-on-glass (SOG). SOG
PSG films are useful as sacrificial layers since they gen- is a polymeric material with a viscosity suitable for
erally have higher etching rates in HF than LTO films. spin coating. Two recent publications illustrate the po-
PSG and LTO films are deposited in hot-wall, low- tential for SOG in MEMS fabrication. In the first
pressure, fused-silica furnaces in systems similar to example, Yasseen et al. [11.24] detailed the develop-
those described previously for polysilicon. Precursor ment of SOG as a thick-film sacrificial molding material
gases include SiH4 as a Si source, O2 as an oxygen for thick polysilicon films. The authors reported a pro-
source, and, in the case of PSG, PH3 as a source of cess to deposit, polish, and etch SOG films that were
phosphorus. LTO and PSG films are typically deposited 20 μm thick. The thick SOG films were patterned into
at temperatures of 425–450 ◦ C and pressures ranging molds and filled with 10 μm thick LPCVD polysilicon
from 200 to 400 mtorr. The low deposition temperatures films, planarized by selective CMP, and subsequently
340 Part A Nanostructures, Micro-/Nanofabrication and Materials
dissolved in a wet etchant containing HCl, HF, and of 107 V/cm. LPCVD Si3 N4 films are deposited in
H2 O to reveal the patterned polysilicon structures. The horizontal furnaces similar to those used for polysil-
cured SOG films were completely compatible with the icon deposition. Typical deposition temperatures and
polysilicon deposition process. In the second example, pressures range between 700 and 900 ◦ C and 200 and
Liu et al. [11.25] fabricated high aspect ratio channel 500 mtorr, respectively. The standard source gases are
plate microstructures from SOG. Electroplated nickel dichlorosilane (SiH2 Cl2 ) and ammonia (NH3 ). To pro-
(Ni) was used as a molding material, with Ni channel duce stoichiometric Si3 N4 a NH3 to SiH2 Cl2 ratio 10:1
Part A 11.2
plate molds fabricated using a conventional LIGA pro- is commonly used. The microstructure of films de-
cess. The Ni molds were then filled with SOG, and the posited under these conditions is amorphous.
sacrificial Ni molds were removed in a reverse elec- The residual stress in stoichiometric Si3 N4 is
troplating process. In this case, the fabricated SOG large and tensile, with a magnitude of about 1 GPa.
structures (over 100 μm tall) were micromachined glass Such a large residual stress causes films thicker than
structures fabricated using a molding material more a few thousand angstroms to crack. Nonetheless thin
commonly used for structural components. stoichiometric Si3 N4 films have been used as mechan-
ical support structures and electrical insulating layers
11.1.5 Silicon Nitride in piezoresistive pressure sensors [11.26]. To enable
the use of Si3 N4 films for applications that require
Silicon nitride (Si3 N4 ) is widely used in MEMS for micrometer-thick, durable, and chemically resistant
electrical isolation, surface passivation, etch masking, membranes, Six Ny films can be deposited by LPCVD.
and as a mechanical material typically for membranes These films, often referred to as Si-rich or low-stress ni-
and other suspended structures. Two deposition meth- tride, are intentionally deposited with an excess of Si
ods are commonly used to deposit Si3 N4 thin films, by simply decreasing the ratio of NH3 to SiH2 Cl2 dur-
LPCVD, and PECVD. PECVD silicon nitride is gener- ing deposition. Nearly stress-free films can be deposited
ally nonstoichiometric (sometimes denoted as Six Ny :H) using a NH3 -to-SiH2 Cl2 ratio of 1/6, a deposition tem-
and may contain significant concentrations of hydro- perature of 850 ◦ C, and a pressure of 500 mtorr [11.27].
gen. Use of PECVD silicon nitride in micromachining The increase in Si content not only leads to a reduc-
applications is somewhat limited because it has a high tion in tensile stress, but also a decrease in the etch rate
etch rate in HF (e.g., often higher than that of thermally in HF. Such properties have enabled the development
grown SiO2 ). However, PECVD offers the ability to de- of fabrication techniques that would otherwise not be
posit nearly stress-free silicon nitride films, an attractive feasible with stoichiometric Si3 N4 . For example, low-
property for encapsulation and packaging. stress silicon nitride has been surface micromachined
Unlike its PECVD counterpart, LPCVD Si3 N4 is using polysilicon as the sacrificial material [11.28]. In
extremely resistant to chemical attack, thereby mak- this case, Si anisotropic etchants such as KOH and
ing it the material of choice for many Si bulk and EDP were used for dissolving the sacrificial polysili-
surface micromachining applications. LPCVD Si3 N4 is con. French et al. [11.29] used PSG as a sacrificial layer
commonly used as an insulating layer because it has to surface micromachine low-stress nitride, capitalizing
a resistivity of 1016 Ω cm and field breakdown limit on the HF resistance of the nitride films.
be reduced to nearly zero after short anneals at mod- Ge-rich films, a thin polysilicon seed layer is sometimes
est temperatures (30 s at 600 ◦ C). Poly-Ge is essentially used on SiO2 surfaces since Ge does not readily nucle-
impervious to KOH, TMAH, and BOE, enabling the ate on oxide surfaces. Like many compound materials,
fabrication of Ge membranes on Si substrates [11.30]. variations in film composition can change the physi-
The mechanical properties of poly-Ge are compara- cal properties of the material. For instance, etching of
ble to those of polysilicon, having a Young’s modulus poly-SiGe by H2 O2 becomes significant for Ge concen-
of 132 GPa and a fracture stress ranging between 1.5 trations over 70%. Sedky et al. [11.33] has shown that
Part A 11.3
and 3.0 GPa [11.31]. Mixtures of HNO3 , H2 O, and the microstructure, film conductivity, residual stress,
HCl and H2 O, H2 O2 , and HCl, as well as the RCA and residual stress gradient are related to the concen-
SC-1 cleaning solution, isotropically etch Ge. Since tration of Ge in the material. With respect to residual
these mixtures do not etch Si, SiO2 , Si3 N4 , and SiN, stress, Franke et al. [11.32] produced in situ boron-
poly-Ge can be used as a sacrificial substrate layer in doped films with residual compressive stresses as low
polysilicon surface micromachining. Using these tech- as 10 MPa.
niques, devices such as poly-Ge-based thermistors and The poly-SiGe, poly-Ge material system is par-
Si3 N4 membrane-based pressure sensors made using ticularly attractive for surface micromachining since
poly-Ge sacrificial layers have been fabricated [11.30]. H2 O2 can be used as a release agent. It has been re-
Franke et al. [11.31] found no performance degradation ported that poly-Ge etches at a rate of 0.4 μm/min in
in Si CMOS devices following the fabrication of surface H2 O2 , while poly-SiGe with Ge concentrations below
micromachined poly-Ge structures, thus demonstrating 80% have no observable etch rate after 40 h [11.34].
the potential for on-chip integration of Ge electrome- The ability to use H2 O2 as a sacrificial etchant makes
chanical devices with Si circuitry. the combination of poly-SiGe and poly-Ge extremely
attractive for surface micromachining from process-
11.2.2 Polycrystalline SiGe ing, safety, and materials compatibility points of view.
Due to the conformal nature of LPCVD processing,
Like poly-Ge, polycrystalline SiGe (poly-SiGe) is a ma- poly-SiGe structural elements, such as gimbal-based
terial that can be deposited at temperatures lower microactuator structures have been made by high-
than polysilicon. Deposition processes include LPCVD, aspect-ratio micromolding [11.34]. Capitalizing on the
APCVD, and RTCVD (rapid thermal CVD) using SiH4 low deposition temperatures, poly-SiGe MEMS inte-
and GeH4 as precursor gases. Deposition temperatures grated with Si ICs has been demonstrated [11.32]. In
range between 450 ◦ C for LPCVD [11.32] and 625 ◦ C this process, CMOS structures are first fabricated on Si
by rapid thermal CVD (RTCVD) [11.33]. In general, the wafers. Poly-SiGe mechanical structures are then sur-
deposition temperature is related to the concentration of face micromachined using a poly-Ge sacrificial layer.
Ge in the films, with higher Ge concentrations result- A significant advantage of this design lies in the fact
ing in lower deposition temperatures. Like polysilicon, that the MEMS structure is positioned directly above the
poly-SiGe can be doped with boron and phosphorus to CMOS structure, thus reducing the parasitic capacitance
modify its conductivity. In situ boron doping can be and contact resistance characteristic of interconnects
performed at temperatures as low as 450 ◦ C [11.32]. associated with side-by-side integration schemes. Use
Sedky et al. [11.33] showed that the deposition tem- of H2 O2 as the sacrificial etchant eliminates the need
perature of conductive films doped with boron could be for layers to protect the underlying CMOS structure
further reduced to 400 ◦ C if the Ge content was kept at during release. In addition to its utility as a mater-
or above 70%. ial for integrated MEMS devices, poly-SiGe has been
Unlike poly-Ge, poly-SiGe can be deposited identified as a material well suited for micromachined
on a number of sacrificial substrates, including thermopiles [11.35] to its lower thermal conductivity
SiO2 [11.33], PSG [11.31], and poly-Ge [11.31]. For relative to Si.
11.3 Metals
It can be argued that of all the material categories as- enabling, since metallic thin films are used in many
sociated with MEMS, metals may be among the most different capacities, from etch masks used in device
342 Part A Nanostructures, Micro-/Nanofabrication and Materials
fabrication to interconnects and structural elements in cantly out of the plane of the wafer due to the stress
microsensors and microactuators. Metallic thin films differences in the bimorph. The switch is drawn closed
can be deposited using a wide range of techniques, by electrostatic attraction. The bimorph consists of met-
including evaporation, sputtering, CVD, and electro- als that can easily be processed with GaAs wafers, thus
plating. Since a complete review of the metals used making integration with GaAs devices possible. The re-
in MEMS is far beyond the scope of this chapter, the leased switches were relatively slow, at 10 ms, but an
examples presented in this section were selected to rep- actuation voltage of only 26 V was needed to close the
Part A 11.3
in a recent report by Shih et al. [11.39]. In this study, galvanoforming, and abformung), this process has been
TiNi films were deposited by cosputtering elemental used to produce a wide variety of high-aspect-ratio
Ti and Ni targets and cosputtering TiNi alloy and ele- structures from plateable materials, such as nickel-iron
mental Ti targets. It was reported that cosputtering from (NiFe) magnetic alloys [11.41] and Ni [11.42].
TiNi and Ti targets produced better films due to pro- In addition to elemental metals and simple com-
cess variations related to roughening of the Ni target in pound alloys, more complex metallic alloys commonly
the case of Ti and Ni cosputtering. The TiNi/Ti cosput- used in commerical macroscopic applications are find-
Part A 11.4
tering process has been used to produce shape-memory ing their way into MEMS applications. One such
material for a silicon spring-based microvalve [11.40]. example is an alloy of titanium known as Ti-6Al-4V.
Use of thin-film metal alloys in magnetic actua- Composed of 88% titanium, 6% aluminum, and 4%
tor systems is another example of the versatility of vanadium, this alloy is widely used in commercial
metallic materials in MEMS. Magnetic actuation in mi- avation due to its weight, strength, and temperature tol-
crodevices generally requires the magnetic layers to be erance. Pornsin-Sirirak et al. [11.43] have explored the
relatively thick (tens to hundreds of micrometer) to gen- use of this alloy in the manufacture of MEMS-based
erate magnetic fields of sufficient strength to generate winged structures for micro aerial vehicles. The authors
the desired actuation. To this end, magnetic materials considered this alloy not only because of its weight
are often deposited by thick-film methods such as elec- and strength, but also because of its ductility and its
troplating. The thicknesses of these layers exceeds what etching rate at room temperature. The designs for the
can feasibly be patterned by etching, so plating is often wing prototype were modeled after the wings of bats
performed in microfabricated molds made from ma- and various flying insects. For this application, Ti-alloy
terials such as polymethylmethacrylate (PMMA). The structures patterned from bulk (250 μm thick) material
PMMA mold thickness can exceed several hundred mi- by an HF/HO3 /H2 O etching solution were used rather
crometer, so x-rays are used as the exposure source than thin films. Parylene-C (detailed in a later section)
during the patterning steps. When necessary a metallic was deposited on the patterned alloy to serve as the
thin-film seed layer is deposited prior to plating. After wing membrane. The miniature micromachined wings
plating, the mold is dissolved, which frees the metal- were integrated into a test setup, and several prototypes
lic component. Known as LIGA (short for lithography, actually demonstrated short duration flight.
at modifying the conductivity of SiC that growth of un- the substrate itself [11.50]. By comparison, poly-
doped SiC films is extremely challenging because the SiC films deposited by LPCVD from SiH2 Cl2 and
concentrations of residual nitrogen in typical deposition C2 H2 are highly textured (111) films with a columnar
systems are sufficient for n-type doping. microstructure [11.48], while films deposited from disi-
APCVD and LPCVD can also be used to deposit labutane have a distribution of orientations [11.44]. This
3C-SiC on Si substrates. Heteroepitaxy is possible de- variation suggests that device performance can be tai-
spite a 20% lattice mismatch because 3C-SiC and lored by selecting the proper substrate and deposition
Part A 11.4
of the Si mold; then the substrate is immersed in these plasmas generally prohibits the use of photoresist
a Si etchant to completely dissolve the mold. This as a masking material; therefore, hard masks made of
process has been used to fabricate solid SiC fuel atom- Al, Ni, and ITO are often used. RIE-based SiC surface
izers [11.42], and a variant has been used to fabricate micromachining processes with polysilicon and SiO2
SiC structures for micropower systems [11.57]. Re- sacrificial layers have been developed for single-layer
cently, Min et al. [11.58] reported a process to fabricate devices [11.61, 62]. ICP RIE of SiC using SF6 plasmas
reusable glass press molds made from SiC structures and Ni or ITO etch masks has been developed for bulk
Part A 11.4
that were patterned using Si molding masters. SiC was micromachining SiC substrates, with structural depths
selected as the material for the glass press mold because in excess of 100 μm reported [11.63].
the application requires a hard, mechanically strong, Until recently, multilayer thin-film structures were
and chemically stable material that can withstand and very difficult to fabricate by direct RIE because the etch
maintain its properties at temperatures between 600 rates of the sacrificial layers were much higher than the
and 1400 ◦ C. SiC structural layers, making dimensional control very
In addition to CVD processes, bulk micromachined difficult. To address this issue, a micromolding process
SiC structures can be fabricated using sintered SiC for patterning SiC films on sacrificial-layer substrates
powders. Tanaka et al. [11.59] describe a process was developed [11.64]. In essence, the micromolding
where SiC components, such as micro gas turbine en- technique is the thin-film analog to the molding-based,
gine rotors, can be fabricated from SiC powders using bulk micromachining technique presented earlier. The
a microreaction-sintering process. The molds are mi- micromolding process utilizes polysilicon and SiO2
crofabricated from Si using DRIE and filled with SiC films as both molds and sacrificial substrate layers, with
and graphite powders mixed with a phenol resin. The SiO2 molds used with polysilicon sacrificial layers and
molds are then reaction-sintered using a hot isostatic vice versa. These films are deposited and patterned
pressing technique. The SiC components are then re- using conventional methods, thus leveraging the well-
leased from the Si mold by wet chemical etching. The characterized and highly selective processes developed
authors reported that the component shrinkage was less for polysilicon MEMS. Poly-SiC films are simply de-
than 3%. The bending strength and Vickers hardness of posited into the micromolds and mechanical polishing
the microreaction-sintered material was roughly 70 to is used to remove poly-SiC from atop the molds. Appro-
80% of commercially available reaction-sintered SiC, priate etchants are then used to dissolve the molds and
the difference being attributed to the presence of unre- sacrificial layers. The micromolding method utilizes the
acted Si in the microscale components. differences in chemical properties of the three materials
In a related process, Liew et al. [11.60] detail a tech- in this system in a way that bypasses the difficulties as-
nique to create silicon carbon nitride (SiCN) MEMS sociated with chemical etching of SiC. This technique
structures by molding injectable polymer precursors. has been developed specifically for multilayer process-
Unlike the aforementioned processes, this technique ing and has been used successfully to fabricate SiC
uses SU-8 photoresists for the molds. To be detailed micromotors [11.64] and the lateral resonant structure
later in this chapter, SU-8 is a versatile photodefinable shown in Fig. 11.4 [11.65].
polymer in which thick films (hundreds of micrometer) Recent advancements in the area of SiC RIE show
can be patterned using conventional UV photolitho- that significant progress has been made in developing
graphic techniques. After patterning, the molds are etch recipes with selectivities to nonmetal mask and sac-
filled with the SiCN-containing polymer precursor, rificial layers that are suitable for multilayer SiC surface
lightly polished, and then subjected to a multistep heat- micromachining. For instance, Gao et al. [11.66] have
treating process. During the thermal processing steps, developed a transformer-coupled RIE process using
the SU-8 mold decomposes and the SiCN structure is a HBr-based chemistry for thin-film poly-SiC etching.
released. The resulting SiCN structures retain many of The recipe exhibits a SiC-to-SiO2 selectivity of 20:1
the same properties of stoichiometric SiC. and a SiC-to-Si3 N4 selectivity of 22:1, which are the
Although SiC cannot be etched using conventional highest reported thus far. In addition, the anisotropy
wet etch techniques, SiC can be patterned using con- of the etch was quite high, and micromasking, a com-
ventional dry etching techniques. RIE processes using mon problem when metal masks are used, was not an
fluorinated compounds such as CHF3 and SF6 com- issue. This process has since been used to fabricate
bined with O2 and sometimes with an inert gas or H2 are multilayered lateral resonant structures that utilize poly-
used to pattern thin films. The high oxygen content in SiC as the main structural material and polysilicon as
346 Part A Nanostructures, Micro-/Nanofabrication and Materials
11.4.2 Diamond
Part A 11.4
tics, laser-to-fiber alignment, and power device cooling and patterned. The diamond films are then etched in
structures [11.70]. the O2 ion-beam plasma, and the structures are released
Due to the nucleation processes associated with by etching the polysilicon with KOH. This process has
diamond film growth, surface micromachining of poly- been used to create lateral resonant structures, but a sig-
crystalline diamond thin films requires modifications to nificant stress gradient in the films rendered the devices
conventional micromachining to facilitate film growth inoperable.
on sacrificial substrates. Initially, conventional RIE In general, conventional HFCVD requires that the
Part A 11.4
methods were generally ineffective, so work was fo- substrate be pretreated with a seeding layer prior to
cused on developing selective deposition techniques. diamond film growth. However, a method called biased
One early method used selective seeding to form pat- enhanced nucleation (BEN) has been developed that en-
terned templates for diamond nucleation. The selective ables the growth of diamond on unseeded Si surfaces.
seeding process employed the lithographic patterning Wang et al. [11.74] have shown that if Si substrates are
of photoresist that contained diamond powders [11.71]. masked with patterned SiO2 films, selective diamond
The diamond-loaded photoresist was deposited and pat- growth will occur primarily on the exposed Si surfaces,
terned onto a Cr-coated Si wafer. During the onset of and a slight HF etch is sufficient to remove the adventi-
diamond growth, the patterned photoresist rapidly evap- tious diamond from the SiO2 mask. This group was able
orates, leaving behind the diamond seed particles in to use this method to fabricate diamond micromotor
the desired locations. A patterned diamond film is then rotors and stators on Si surfaces.
selectively grown on these locations. Diamond is a difficult, but not impossible, mater-
A second process utilized selective deposition di- ial to etch using conventional RIE techniques. It is well
rectly on sacrificial substrate layers. This process known that diamond can be etched in oxygen plasmas,
combined conventional diamond seeding with pho- but these plasmas can be problematic for device fabrica-
tolithographic patterning and etching to fabricate mi- tion because the etching tends to be isotropic. A recent
cromachined diamond structures on SiO2 sacrificial development, however, suggests that RIE processes for
layers [11.72]. The process was performed in one of diamond are close at hand. Wang et al. [11.74] describe
two ways. The first approach begins with the seeding a process to fabricate a vertically actuated, doubly
of an oxidized Si wafer. The wafer is coated with a pho- clamped micromechanical diamond beam resonator us-
toresist and photolithographically patterned. Unmasked ing RIE. The process outlined in this paper addresses
regions of the seeded SiO2 film are then partially etched, two key issues related to diamond surface micromachin-
forming a surface unfavorable for diamond growth. The ing, namely, residual stress gradients in the diamond
photoresist is then removed and a diamond film is de- films and diamond patterning techniques. A microwave
posited on the seeded regions. The second approach also plasma CVD (MPCVD) reactor was used to grow the
begins with an oxidized Si wafer. The wafer is coated diamond films on sacrificial SiO2 layers pretreated
with a photoresist, photolithographically patterned, and with a nanocrystalline diamond powder, resulting in
then seeded with diamond particles. The photoresist is a uniform nucleation density at the diamond/SiO2 inter-
removed, leaving behind a patterned seed layer suitable face. The diamond films were etched in a CF4 /O2
for selective growth. These techniques have been suc- plasma using Al as a hard mask. Reasonably straight
cessfully used to fabricate cantilever beams and bridge sidewalls were created, with roughness attributable to
structures. the surface roughness of the faceted diamond film.
A third method to surface micromachine polycrys- An Au/Cr drive electrode beneath the sacrificial oxide
talline diamond films follows the conventional approach remained covered throughout the diamond-patterning
of film deposition, dry etching, and release. The chem- steps and thus was undamaged during the diamond-
ical inertness of diamond renders most conventional etching process. This work has since been extended
plasma chemistries useless; however, oxygen-based to develop a 1.51 GHz diamond micromechanical disk
ion-beam plasmas can be used to etch diamond thin resonator [11.74]. In this instance, the nanocrystalline
films [11.73]. A simple surface micromachining process diamond film was deposited my MPCVD, coated with
begins with the deposition of a polysilicon sacrificial an oxide film that had been patterned into an etch
layer on a Si3 N4 -coated Si wafer. The polysilicon layer mask, and then etched in a O2 /CF4 RIE plasma under
is seeded using diamond slurry, and a diamond film is conditions that yielded a fairly anisotropic etch with
deposited by HFCVD. Since photoresists are not resis- a diamond-to-oxide selectivity of 15:1. The disk was
tant to O2 plasmas, an Al masking film is deposited suspended over the substrate on a polysilicon stem using
348 Part A Nanostructures, Micro-/Nanofabrication and Materials
Part A 11.5
electronic applications. Like Si, significant research in on the creation of a highly resistive near-surface GaAs
bulk crystal growth has led to the development of GaAs layer on an n-type GaAs substrate by low-dose nitro-
and InP substrates that are commercially available as gen implantation in the MeV energy range. A pulsed
high-quality, single-crystal wafers. Unlike compound electrochemical etch method using an H2 PtCl6 , H3 PO4 ,
semiconductors such as SiC, III–V materials can be H2 SO4 platinum electrolytic solution at 40 ◦ C with
deposited as ternary and quaternary alloys with lattice 17 V, 100 ms pulses is sufficient to selectively remove
constants that closely match the binary compounds from n-type GaAs at about 3 μm/min. Using this method,
which they are derived (i. e., Alx Ga1−x As and GaAs), stress-free, tethered membranes could readily be fab-
thus permitting the fabrication of a wide variety of het- ricated from the highly resistive GaAs layer. The high
erostructures that facilitate device performance. implant energies enable the fabrication of membranes
Crystalline GaAs has a zinc blend crystal structure several micrometer thick. Moreover, the authors demon-
with an electronic bandgap of 1.4 eV, enabling GaAs strated that if the GaAs wafer were etched in such a way
electronic devices to function at temperatures as high as to create an undulating surface prior to ion implanta-
as 350 ◦ C [11.80]. High-quality, single-crystal wafers tion, corrugated membranes could be fabricated. These
are commercially available, as are well-developed met- structures can sustain much higher deflection ampli-
alorganic chemical vapor deposition (MOCVD) and tudes than flat structures.
molecular beam epitaxy (MBE) growth processes for Micromachining of InP closely resembles the tech-
epitaxial layers of GaAs and its alloys. GaAs does not niques used for GaAs. Many of the properties of InP are
outperform Si in terms of mechanical properties; how- similar to GaAs in terms of crystal structure, mechanical
ever, its stiffness and fracture toughness are still suitable stiffness, and hardness; however, the optical properties
for micromechanical devices. of InP make it particularly attractive for microop-
Micromachining of GaAs is relatively straightfor- tomechanical devices to be used in the 1.3–1.55 μm
ward, since many of its lattice-matched ternary and wavelength range [11.90]. Like GaAs, single-crystal
quaternary alloys have sufficiently different chem-
ical properties to allow their use as sacrificial lay-
ers [11.81]. For example, the most common ternary
alloy for GaAs is Alx Ga1−x As. For values of x
less than or equal to 0.5, etchants containing mix-
tures of HF and H2 O will etch Alx Ga1−x As without
attacking GaAs, while etchants containing NH4 OH
and H2 O2 attack GaAs isotropically but do not etch
Alx Ga1−x As. Such selectivity enables the microma-
chining of GaAs wafers using lattice-matched etch
stops and sacrificial layers. Devices fabricated using
these methods include comb drive lateral resonant
structures [11.81], pressure sensors [11.82, 83], ther-
mopile sensors [11.83], Fabry–Perot detectors [11.84], 1 µm
and cantilever-based sensors and actuators [11.85, 86].
In addition, nanoelectromechanical devices, such as
suspended micromechanical resonators [11.87] and
tethered membranes [11.88], have been fabricated using Fig. 11.7 SEM micrograph of a GaAs nanomechanical
these techniques. An example of a nanoelectromechan- beam resonator fabricated by epitaxial growth, electron-
ical beam structure fabricated from GaAs is shown in beam lithography, and selective etching (courtesy of
Fig. 11.7. M. Roukes, Caltech)
350 Part A Nanostructures, Micro-/Nanofabrication and Materials
wafers of InP are readily available, and ternary and qua- into device structures. Despite a 7% lattice mis-
ternary lattice-matched alloys, such as InGaAs, InAlAs, match between InAs and (111) GaAs, high-quality
InGaAsP, and InGaAlAs, can be used as either etch stop epitaxial layers can be grown on GaAs substrates.
and/or sacrificial layers depending on the etch chemistry As described by Yamaguchi et al. [11.92], the sur-
[11.81]. For instance, InP structural layers deposited face Fermi level of InAs/GaAs structures is pinned
on In0,53 Al0,47 As sacrificial layers can be released us- in the conduction band, enabling the fabrication of
ing etchants containing C6 H8 O7 , H2 O2 , and H2 O. In very thin conductive membranes. In fact, the authors
Part A 11.6
addition, InP films and substrates can be etched in so- have successfully fabricated free-standing InAs struc-
lutions containing HCl and H2 O using In0,53 Ga0,47 As tures that range in thickness from 30 to 300 nm.
films as etch stops. Using InP-based micromachining The thin InAs films were grown directly on GaAs
techniques, multiair gap filters [11.91] bridge structures substrates by MBE and etched using a solution con-
[11.90], and torsional membranes [11.84] have been taining H2 O, H2 O2 , and H2 SO4 . The structures, mainly
fabricated from InP and its related alloys. doubly clamped cantilevers, were released by etch-
In addition to GaAs and InP, materials such ing the GaAs substrate using an H2 O/H2 O2 /NH4 OH
as indium arsenide (InAs) can be micromachined solution.
Part A 11.7
polyimides can be acquired in bulk or deposited as thin cated on these substrates using conventional techniques.
films by spin coating, and they can be patterned us- Thin polyimide layers were deposited between various
ing conventional dry etching techniques and processed metal layers to serve as insulating layers. A capping
at relatively high temperatures. These attributes make polyimide layer was then deposited on the top of the
polyimides an attractive group of polymers for MEMS substrates, and then the entire polyimide/metal struc-
that require polymer structural and/or substrate layers, ture was peeled off the Si carrier wafers. Backside
such as microfabricated biomedical devices where in- processing was then performed on the free-standing
ertness and flexibility are important parameters. polyimide structures to create devices that have ex-
Shearwood et al. [11.96] explored the use of poly- posed electrodes on both surfaces. In a later paper,
imides as a robust mechanical material for microfabri- Stieglitz et al. [11.101] describe a variation of this pro-
cated audio membranes. The authors fabricated 7 μm cess for neural prostheses.
thick, 8 mm diameter membranes on GaAs substrates
by bulk micromachining the GaAs substrate using 11.7.2 SU-8
a NH3 /H2 O2 solution. They realized 100% yield and,
despite a low Young’s modulus (≈ 3 GPa), observed flat SU-8 is a negative-tone epoxylike photoresist that is
membranes to within 1 nm after fabrication. receiving much attention for its versatility in MEMS
Jiang et al.[11.97] capitalized on the strength and processing. It is a high-aspect-ratio, UV-sensitive resist
flexibility of polyimides to fabricate a flexible sheer- designed for applications requiring single-coat resists
stress sensor array based on Si sensors. The sensor with thicknesses on the order of 500 μm [11.102]. SU-8
array consisted of a collection of Si islands linked by has favorable chemical properties that enable it to be
two polyimide layers. Each Si sensor island was 250 × used as a molding material for high-aspect-ratio elec-
250 μm2 in area and 80 μm in thickness. Al was used troplated structures (as an alternative to LIGA) and
as an electrical innerconnect layer. The two polyimide as a structural material for microfluidics [11.102]. In
layers served as highly flexible hinges, making it pos- terms of mechanical properties, Lorenz et al. [11.103]
sible to mount the sensor array on curved surfaces. The reported that SU-8 has a modulus of elasticity of
sensor array was successful in profiling the shear-stress 4.02 GPa, which compares favorably with a commonly-
distribution along the leading edge of a rounded delta used polyamid (3.4 GPa).
wing. In addition to the above-mentioned conven-
The chemical and temperature durability of poly- tional uses for SU-8, several interesting alterna-
imides enables their use as a sacrificial layer for tive uses are beginning to appear in the literature.
a number of commonly used materials, such as evapo- Conradie et al. [11.104] have used SU-8 to trim the
rated or sputter-deposited metals. Memmi et al. [11.98] mass of silicon paddle oscillators as a means to adjust
developed a fabrication process for capacitive mi- the resonant frequency of the beams. The trimming pro-
cromechanical ultrasonic transducers using a polyimide cess involves the patterning of SU-8 posts on Si paddles.
as a sacrificial layer. The authors showed that the The process capitalizes on the relative chemical stabil-
polyimide could withstand the conditions used to ity of the SU-8 resin in conjunction with the relatively
deposit silicon monoxide by evaporation and sili- large masses that can be patterned using standard UV
con nitride by PECVD at 400 ◦ C. Recent work by exposure processes.
Bagolini et al. [11.99] has shown that polyimides can SU-8 is also of interest as a bonding layer mater-
even be used as sacrificial layers for PECVD SiC. ial for wafer bonding processes using patterned bonding
In the area of microfabricated biomedical de- layers. Pan et al. [11.105] compared several UV pho-
vices, polyimides are receiving attention as a sub- todefinable polymeric materials and found that SU-8
strate material for implantable devices, owing to their exhibited the highest bonding strength (20.6 MPa) for
potential biocompatiblity and mechanical flexibility. layer thicknesses up to 100 μm.
352 Part A Nanostructures, Micro-/Nanofabrication and Materials
sition process is conformal, which enables parylene board and electronics packaging applications and has
coatings to be applied to prefabricated structures, such recently been investigated for use in MEMS applica-
as Si microneedles [11.106], low-stress silicon nitride tions requiring a material that is mechanically flexible,
membrane particle filters [11.73], and micromachined electrically insulating, chemically durable, and imper-
polyimide/Au optical scanners [11.107]. In the former meable to moisture. LCP can be bonded to itself and
cases, the parylene coating served to strengthen the other substrate materials such as glass and Si by thermal
microfabricated structures, while in the latter case it lamination. It can be micromachined using an oxygen
served to protect the structure from condensing water plasma and yet is highly resistant to HF and many
vapor. metal etchants [11.110]. The moisture absorption is
In addition to its function as a protective coat- less than 0.02% as compared with about 1% for poly-
ing, parylene can actually be micromachined into imide [11.111], making it well suited as a packaging
free-standing components. Noh et al. [11.108] demon- material.
strated a method to create bulk micromachined parylene Applications where LCPs are used as a key com-
microcolumns for miniature gas chromatographs. The ponent in a MEMS device are beginning to emerge.
structure is fabricated using a micromolding technique Faheem et al. [11.112] reported on the use of LCP
where Si molds are fabricated by DRIE and coated for encapsulation of variable RF MEMS capacitors.
with parylene to form three sides of the microcol- In this example, LCP, dispensed in liquid form, was
umn. A second wafer is coated with parylene, and the used to join and seal a glass microcap to a prefab-
two are bonded together via a fusion bonding pro- ricated, microbridge capacitor. LCP was chosen in
cess. After bonding, the structure is released from part because in addition to the aforementioned prop-
the Si mold by KOH etching. In a second exam- erties, it has very low RF loss characteristics, making
ple, Yao et al. [11.109] describe a dry release process it very well suited as an RF MEMS packaging mater-
for parylene surface micromachining. In this process, ial. Wang et al. [11.110] showed that LCP is a very
sputtered Si is used as a sacrificial layer onto which versatile material that is highly compatible with many
a thick sacrificial photoresist is deposited. Parylene is standard Si-based processing techniques. They also
then deposited on the photoresist and patterned into showed that micromachining techniques can be used
the desired structural shape. The release procedure is to make LCP cantilever flow sensors that incorporate
a two-step process. First the photoresist is dissolved in metal strain gauges and LCP membrane tactile sensors
acetone. This results in the parylene structure sticking using NiCr strain gauges. Lee et al. [11.111] has devel-
to the sputtered Si. Next, a dry BrF3 etch is per- oped a LCP-based, mechanically flexible, multichannel
formed that dissolves the Si and releases the parylene microelectrode array structure for neural stimulation
structures. Parylene beams that were 1 mm long and and recording.
lenge facing device design and fabrication engineers proaches to create devices and systems that cannot be
alike will be how to marry top-down and bottom-up ap- made using either process alone.
References
11.1 C.S. Smith: Piezoresistive effect in germanium and strength evaluated with microstructures, J. Mi-
Part A 11
silicon, Phys. Rev. 94, 1–10 (1954) cromech. Microeng. 9, 245–251 (1999)
11.2 A.N. Cleland, M.L. Roukes: Fabrication of high fre- 11.14 K. Funk, H. Emmerich, A. Schilp, M. Offenberg,
quency nanometer scale mechanical resonators R. Neul, F. Larmer: A surface micromachined silicon
from bulk Si crystals, Appl. Phys. Lett. 69, 2653– gyroscope using a thick polysilicon layer, Proc. 12th
2655 (1996) Int. Conf. Microelectromech. Syst. (IEEE, Piscataway
11.3 D.W. Carr, H.G. Craighead: Fabrication of nano- 1999) pp. 57–60
electromechanical systems in single crystal silicon 11.15 T. Abe, M.L. Reed: Low strain sputtered polysili-
using silicon on insulator substrates and electron con for micromechanical structures, Proc. 9th Int.
beam lithography, J. Vac. Sci. Technol. B 15, 2760– Workshop Microelectromech. Syst. (IEEE, Piscat-
2763 (1997) away 1996) pp. 258–262
11.4 T. Kamins: Polycrystalline Silicon for Integrated Cir- 11.16 K. Honer, G.T.A. Kovacs: Integration of sputtered
cuits and Displays, 2nd edn. (Kluwer, Boston 1988) silicon microstructures with pre-fabricated CMOS
11.5 J.J. McMahon, J.M. Melzak, C.A. Zorman, J. Chung, circuitry, Sens. Actuators A 91, 392–403 (2001)
M. Mehregany: Deposition and characterization of 11.17 J. Gaspar, T. Adrega, V. Chu, J.P. Conde: Thin-film
in-situ boron doped polycrystalline silicon films paddle microresonators with high quality factors
for microelectromechanical systems applications, fabricated at temperatures below 110 ◦ C, Proc. 18th
Mater. Res. Symp. Proc. 605, 31–36 (2000) Int. Conf. Microelectromech. Syst. (IEEE, Piscataway
11.6 L. Cao, T.S. Kin, S.C. Mantell, D. Polla: Simulation 2005) pp. 125–128
and fabrication of piezoresistive membrane type 11.18 R. Anderson, R.S. Muller, C.W. Tobias: Porous
MEMS strain sensors, Sens. Actuators 80, 273–279 polycrystalline silicon: A new material for MEMS,
(2000) J. Microelectromech. Syst. 3, 10–18 (1994)
11.7 H. Guckel, T. Randazzo, D.W. Burns: A simple tech- 11.19 W. Lang, P. Steiner, H. Sandmaier: Porous silicon:
nique for the determination of mechanical strain A novel material for microsystems, Sens. Actuators
in thin films with application to polysilicon, J. Appl. A 51, 31–36 (1995)
Phys. 57, 1671–1675 (1983) 11.20 R. He, C.J. Kim: On-chip hermetic packaging en-
11.8 R.T. Howe, R.S. Muller: Stress in polysilicon and abled by post-deposition electrochemical etching
amorphous silicon thin films, J. Appl. Phys. 54, of polysilicon, Proc. 18th Int. Conf. Microelec-
4674–4675 (1983) tromech. Syst. (IEEE, Piscataway 2005) pp. 544–547
11.9 X. Zhang, T.Y. Zhang, M. Wong, Y. Zohar: Rapid 11.21 S.K. Ghandhi: VLSI Fabrication Principles – Silicon
thermal annealing of polysilicon thin films, J. Mi- and Gallium Arsenide (Wiley, New York 1983)
croelectromech. Syst. 7, 356–364 (1998) 11.22 W.A. Pilskin: Comparison of properties of dielectric
11.10 J. Yang, H. Kahn, A.-Q. He, S.M. Phillips, films deposited by various methods, J. Vac. Sci.
A.H. Heuer: A new technique for producing large- Technol. 21, 1064–1081 (1977)
area as-deposited zero-stress LPCVD polysilicon 11.23 J.S. Danel, F. Michel, G. Delapierre: Micromachin-
films: The multipoly process, J. Microelectromech. ing of quartz and its application to an acceleration
Syst. 9, 485–494 (2000) sensor, Sens. Actuators A 21–23, 971–977 (1990)
11.11 P. Gennissen, M. Bartek, P.J. French, P.M. Sarro: 11.24 A. Yasseen, J.D. Cawley, M. Mehregany: Thick
Bipolar-compatible epitaxial poly for smart sen- glass film technology for polysilicon surface micro-
sors: Stress minimization and applications, Sens. machining, J. Microelectromech. Syst. 8, 172–179
Actuators A 62, 636–645 (1997) (1999)
11.12 P. Lange, M. Kirsten, W. Riethmuller, B. Wenk, 11.25 R. Liu, M.J. Vasile, D.J. Beebe: The fabrication of
G. Zwicker, J.R. Morante, F. Ericson, J.A. Schweitz: nonplanar spin-on glass microstructures, J. Micro-
Thick polycrystalline silicon for surface-micro- electromech. Syst. 8, 146–151 (1999)
mechanical applications: deposition, structuring, 11.26 B. Folkmer, P. Steiner, W. Lang: Silicon nitride
and mechanical characterization, Sens. Actuators membrane sensors with monocrystalline transduc-
A 54, 674–678 (1996) ers, Sens. Actuators A 51, 71–75 (1995)
11.13 S. Greek, F. Ericson, S. Johansson, M. Furtsch, 11.27 M. Sekimoto, H. Yoshihara, T. Ohkubo: Silicon ni-
A. Rump: Mechanical characterization of thick tride single-layer x-ray mask, J. Vac. Sci. Technol.
polysilicon films: Young’s modulus and fracture 21, 1017–1021 (1982)
354 Part A Nanostructures, Micro-/Nanofabrication and Materials
11.28 D.J. Monk, D.S. Soane, R.T. Howe: Enhanced 11.41 S.D. Leith, D.T. Schwartz: High-rate through-mold
removal of sacrificial layers for silicon surface electrodeposition of thick (> 200 micron) NiFe MEMS
micromachining, 7th Int. Conf. Solid State Sens. components with uniform composition, J. Micro-
Actuators, Technical Digest (Institute of Electrical electromech. Syst. 8, 384–392 (1999)
Engineers of Japan, Tokyo 1993) pp. 280–283 11.42 N. Rajan, M. Mehregany, C.A. Zorman, S. Ste-
11.29 P.J. French, P.M. Sarro, R. Mallee, E.J.M. Fakkeldij, fanescu, T. Kicher: Fabrication and testing of
R.F. Wolffenbuttel: Optimization of a low-stress micromachined silicon carbide and nickel fuel
silicon nitride process for surface micromachining atomizers for gas turbine engines, J. Microelec-
Part A 11
applications, Sens. Actuators A 58, 149–157 (1997) tromech. Syst. 8, 251–257 (1999)
11.30 B. Li, B. Xiong, L. Jiang, Y. Zohar, M. Wong: Germa- 11.43 T. Pornsin-Sirirak, Y.C. Tai, H. Nassef, C.M. Ho:
nium as a versatile material for low-temperature Titanium-alloy MEMS wing technology for a mi-
micromachining, J. Microelectromech. Syst. 8, croaerial vehicle application, Sens. Actuators A 89,
366–372 (1999) 95–103 (2001)
11.31 A. Franke, D. Bilic, D.T. Chang, P.T. Jones, T.J. King, 11.44 C.R. Stoldt, C. Carraro, W.R. Ashurst, D. Gao,
R.T. Howe, C.G. Johnson: Post-CMOS integration of R.T. Howe, R. Maboudian: A low temperature CVD
germanium microstructures, Proc. 12th Int. Conf. process for silicon carbide MEMS, Sens. Actuators A
Microelectromech. Syst. (IEEE, Piscataway 1999) 97/98, 410–415 (2002)
pp. 630–637 11.45 M. Eickhoff, H. Moller, G. Kroetz, J. von Berg,
11.32 A.E. Franke, Y. Jiao, M.T. Wu, T.J. King, R.T. Howe: R. Ziermann: A high temperature pressure sensor
Post-CMOS modular integration of poly-SiGe mi- prepared by selective deposition of cubic silicon
crostructures using poly-Ge sacraficial layers, Solid carbide on SOI substrates, Sens. Actuators 74, 56–
State Sens. Actuator Workshop, Technical Digest 59 (1999)
(Transducers Research Foundation, Hilton Head 11.46 Y.T. Yang, K.L. Ekinci, X.M.H. Huang, L.M. Schi-
2000) pp. 18–21 avone, M.L. Roukes, C.A. Zorman, M. Mehregany:
11.33 S. Sedky, P. Fiorini, M. Caymax, S. Loreti, K. Baert, Monocrystalline silicon carbide nanoelectrome-
L. Hermans, R. Mertens: Structural and mechanical chanical systems, Appl. Phys. Lett. 78, 162–164
properties of polycrystalline silicon germanium for (2001)
micromachining applications, J. Microelectromech. 11.47 D. Young, J. Du, C.A. Zorman, W.H. Ko: High-
Syst. 7, 365–372 (1998) temperature single crystal 3C-SiC capacitive pres-
11.34 J.M. Heck, C.G. Keller, A.E. Franke, L. Muller, sure sensor, IEEE Sens. J. 4, 464–470 (2004)
T.-J. King, R.T. Howe: High aspect ratio polysilicon- 11.48 C.A. Zorman, S. Rajgolpal, X.A. Fu, R. Jezeski,
germanium microstructures, Proc. 10th Int. Conf. J. Melzak, M. Mehregany: Deposition of polycrys-
Solid State Sens. Actuators (Institute of Electrical talline 3C-SiC films on 100 mm-diameter (100) Si
Engineers of Japan, Tokyo 1999) pp. 328–334 wafers in a large-volume LPCVD furnace, Elec-
11.35 P. Van Gerwen, T. Slater, J.B. Chevrier, K. Baert, trochem. Solid State Lett. 5, G99–G101 (2002)
R. Mertens: Thin-film boron-doped polycrystalline 11.49 I. Behrens, E. Peiner, A.S. Bakin, A. Schlachetzski:
silicon70% -germanium30% for thermopiles, Sens. Micromachining of silicon carbide on silicon fabri-
Actuators A 53, 325–329 (1996) cated by low-pressure chemical vapor deposition,
11.36 D. Hyman, J. Lam, B. Warneke, A. Schmitz, T.Y. Hsu, J. Micromech. Microeng. 12, 380–384 (2002)
J. Brown, J. Schaffner, A. Walson, R.Y. Loo, 11.50 C.A. Zorman, S. Roy, C.H. Wu, A.J. Fleischman,
M. Mehregany, J. Lee: Surface micromachined RF M. Mehregany: Characterization of polycrystalline
MEMS switches on GaAs substrates, Int. J. Radio silicon carbide films grown by atmospheric pres-
Freq. Microw. Commun. Eng. 9, 348–361 (1999) sure chemical vapor deposition on polycrystalline
11.37 C. Chang, P. Chang: Innovative micromachined mi- silicon, J. Mater. Res. 13, 406–412 (1996)
crowave switch with very low insertion loss, Sens. 11.51 C.H. Wu, C.H. Zorman, M. Mehregany: Growth of
Actuators 79, 71–75 (2000) polycrystalline SiC films on SiO2 and Si3 N4 by APCVD,
11.38 M.F. Aimi, M.P. Rao, N.C. MacDonald, A.S. Zuruzi, Thin Solid Films 355/356, 179–183 (1999)
D.P. Bothman: High-aspect-ratio bulk microma- 11.52 P. Sarro: Silicon carbide as a new MEMS technol-
chining of Ti, Nat. Mater. 3, 103–105 (2004) ogyctuators, Sens. Actuators A 82, 210–218 (2000)
11.39 C.L. Shih, B.K. Lai, H. Kahn, S.M. Phillips, 11.53 N. Ledermann, J. Baborowski, P. Muralt, N. Xan-
A.H. Heuer: A robust co-sputtering fabrication pro- topoulos, J.M. Tellenbach: Sputtered silicon car-
cedure for TiNi shape memory alloys for MEMS, bide thin films as protective coatings for MEMS
J. Microelectromech. Syst. 10, 69–79 (2001) applications, Surf. Coat. Technol. 125, 246–250
11.40 G. Hahm, H. Kahn, S.M. Phillips, A.H. Heuer: Fully (2000)
microfabricated silicon spring biased shape mem- 11.54 X.A. Fu, R. Jezeski, C.A. Zorman, M. Mehregany:
ory actuated microvalve, Solid State Sens. Actuator Use of deposition pressure to control the residual
Workshop, Technical Digest (Transducers Research stress in polycrystalline SiC films, Appl. Phys. Lett.
Foundation, Hilton Head Island 2000) pp. 230–233 84, 341–343 (2004)
Material Aspects of Micro- and Nanoelectromechanical Systems References 355
11.55 J. Trevino, X.A. Fu, M. Mehregany, C. Zorman: 11.69 P. Rangsten, H. Bjorkman, K. Hjort: Microfluidic
Low-stress, heavily-doped polycrystalline silicon components in diamond, Proc. 10th Int. Conf. Solid
carbide for MEMS applications, Proc. 18th Int. Conf. State Sens. Actuators (IEEE, Sendai 1999) pp. 190–
Microelectromech. Syst. (IEEE, Piscataway 2005) 193
pp. 451–454 11.70 H. Bjorkman, P. Rangsten, K. Hjort: Diamond mi-
11.56 R.S. Okojie, A.A. Ned, A.D. Kurtz: Operation of a 6H- crostructures for optical microelectromechanical
SiC pressure sensor at 500 ◦ C, Sens. Actuators A 66, systems, Sens. Actuators 78, 41–47 (1999)
200–204 (1998) 11.71 M. Aslam, D. Schulz: Technology of diamond micro-
Part A 11
11.57 K. Lohner, K.S. Chen, A.A. Ayon, M.S. Spearing: electromechanical systems, Proc. 8th Int. Conf.
Microfabricated silicon carbide microengine struc- Solid State Sens. Actuators (IEEE, Stockholm 1995)
tures, Mater. Res. Soc. Symp. Proc. 546, 85–90 pp. 222–224
(1999) 11.72 R. Ramesham: Fabrication of diamond microstruc-
11.58 K.O. Min, S. Tanaka, M. Esashi: Micro/nano glass tures for microelectromechanical systems (MEMS)
press molding using silicon carbide molds fabri- by a surface micromachining process, Thin Solid
cated by silicon lost molding, Proc. 18th Int. Conf. Films 340, 1–6 (1999)
Microelectromech. Syst. (IEEE, Miami 2005) pp. 475– 11.73 X. Yang, J.M. Yang, Y.C. Tai, C.M. Ho: Microma-
478 chined membrane particle filters, Sens. Actuators
11.59 S. Tanaka, S. Sugimoto, J.-F. Li, R. Watanabe, 73, 184–191 (1999)
M. Esashi: Silicon carbide micro-reaction-sintering 11.74 X.D. Wang, G.D. Hong, J. Zhang, B.L. Lin, H.Q. Gong,
using micromachined silicon molds, J. Microelec- W.Y. Wang: Precise patterning of diamond films for
tromech. Syst. 10, 55–61 (2001) MEMS application, J. Mater. Process. Technol. 127,
11.60 L.A. Liew, W. Zhang, V.M. Bright, A. Linan, 230–233 (2002)
M.L. Dunn, R. Raj: Fabrication of SiCN ceramic 11.75 A.R. Krauss, O. Auciello, D.M. Gruen, A. Jayatissa,
MEMS using injectable polymer-precursor tech- A. Sumant, J. Tucek, D.C. Mancini, N. Moldovan,
nique, Sens. Actuators A 89, 64–70 (2001) A. Erdemire, D. Ersoy, M.N. Gardos, H.G. Busmann,
11.61 A.J. Fleischman, S. Roy, C.A. Zorman, M. Mehre- E.M. Meyer, M.Q. Ding: Ultrananocrystalline dia-
gany: Polycrystalline silicon carbide for surface mond thin films for MEMS and moving mechanical
micromachining, Proc. 9th Int. Workshop Mi- assembly devices, Diam. Relat. Mater. 10, 1952–1961
croelectromech. Syst. (IEEE, San Diego 1996) (2001)
pp. 234–238 11.76 X. Xiao, J. Birrell, J.E. Gerbi, O. Auciello, J.A. Carlisle:
11.62 A.J. Fleischman, X. Wei, C.A. Zorman, M. Mehre- Low temperature growth of ultrananocrystalline
gany: Surface micromachining of polycrystalline diamond, J. Appl. Phys. 96, 2232–2239 (2004)
SiC deposited on SiO2 by APCVD, Mater. Sci. Forum 11.77 T.A. Friedmann, J.P. Sullivan, J.A. Knapp, D.R. Tal-
264–268, 885–888 (1998) lant, D.M. Follstaedt, D.L. Medlin, P.B. Mirkarimi:
11.63 G. Beheim, C.S. Salupo: Deep RIE process for silicon Thick stress-free amorphous-tetrahedral carbon
carbide power electronics and MEMS, Mater. Res. films with hardness near that of diamond, Appl.
Soc. Symp. Proc. 622, T8.8.1–T8.8.6 (2000) Phys. Lett. 71, 3820–3822 (1997)
11.64 A. Yasseen, C.H. Wu, C.A. Zorman, M. Mehregany: 11.78 J.P. Sullivan, T.A. Friedmann, K. Hjort: Diamond
Fabrication and testing of surface micromachined and amorphous carbon MEMS, MRS Bulletin 26,
polycrystalline SiC micromotors, Electron. Device 309–311 (2001)
Lett. 21, 164–166 (2000) 11.79 J.R. Webster, C.W. Dyck, J.P. Sullivan, T.A. Fried-
11.65 X. Song, S. Rajgolpal, J.M. Melzak, C.A. Zorman, mann, A.J. Carton: Performance of amorphous
M. Mehregany: Development of a multilayer SiC diamond RF MEMS capacitive switch, Electron. Lett.
surface micromachining process with capabilities 40, 43–44 (2004)
and design rules comparable with conventional 11.80 K. Hjort, J. Soderkvist, J.-A. Schweitz: Galium ar-
polysilicon surface micromachining, Mater. Sci. Fo- senide as a mechanical material, J. Micromech.
rum 389–393, 755–758 (2001) Microeng. 4, 1–13 (1994)
11.66 D. Gao, M.B. Wijesundara, C. Carraro, R.T. Howe, 11.81 K. Hjort: Sacrificial etching of III–V compounds for
R. Maboudian: Recent progress toward and micromechanical devices, J. Micromech. Microeng.
manufacturable polycrystalline SiC surface micro- 6, 365–370 (1996)
machining technology, IEEE Sens. J. 4, 441–448 11.82 K. Fobelets, R. Vounckx, G. Borghs: A GaAs pres-
(2004) sure sensor based on resonant tunnelling diodes,
11.67 T. Shibata, Y. Kitamoto, K. Unno, E. Makino: Micro- J. Micromech. Microeng. 4, 123–128 (1994)
machining of diamond film for MEMS applications, 11.83 A. Dehe, K. Fricke, H.L. Hartnagel: Infrared
J. Microelectromech. Syst. 9, 47–51 (2000) thermopile sensor based on AlGaAs-GaAs micro-
11.68 H. Bjorkman, P. Rangsten, P. Hollman, K. Hjort: machining, Sens. Actuators A 46/47, 432–436 (1995)
Diamond replicas from microstructured silicon 11.84 A. Dehe, J. Peerlings, J. Pfeiffer, R. Riemenschnei-
masters, Sens. Actuators 73, 24–29 (1999) der, A. Vogt, K. Streubel, H. Kunzel, P. Meissner,
356 Part A Nanostructures, Micro-/Nanofabrication and Materials
H.L. Hartnagel: III–V compound semiconductor mi- 11.98 D. Memmi, V. Foglietti, E. Cianci, G. Caliano,
cromachined actuators for long resonator tunable M. Pappalardo: Fabrication of capacitive mi-
Fabry-Perot detectors, Sens. Actuators A 68, 365– cromechanical ultrasonic transducers by low-
371 (1998) temperature process, Sens. Actuators A 99, 85–91
11.85 T. Lalinsky, S. Hascik, Z. Mozolova, E. Burian, (2002)
M. Drzik: The improved performance of GaAs mi- 11.99 A. Bagolini, L. Pakula, T.L.M. Scholtes, H.T.M. Pham,
cromachined power sensor microsystem, Sens. P.J. French, P.M. Sarro: Polyimide sacrificial layer
Actuators 76, 241–246 (1999) and novel materials for post-processing surface
Part A 11
11.86 T. Lalinsky, E. Burian, M. Drzik, S. Hascik, Z. Mo- micromachining, J. Micromech. Microeng. 12, 385–
zolova, J. Kuzmik, Z. Hatzopoulos: Performance of 389 (2002)
GaAs micromachined microactuator, Sens. Actua- 11.100 T. Stieglitz: Flexible biomedical microdevices with
tors 85, 365–370 (2000) double-sided electrode arrangements for neural
11.87 H.X. Tang, X.M.H. Huang, M.L. Roukes, M. Bichler, applications, Sens. Actuators A 90, 203–211 (2001)
W. Wegscheider: Two-dimensional electron-gas 11.101 T. Stieglitz, G. Matthias: Flexible BioMEMS with
actuation and transduction for GaAs nanoelec- electrode arrangements on front and back side as
tromechanical systems, Appl. Phys. Lett. 81, key component in neural prostheses and biohybrid
3879–3881 (2002) systems, Sens. Actuators B 83, 8–14 (2002)
11.88 T.S. Tighe, J.M. Worlock, M.L. Roukes: Direct ther- 11.102 H. Lorenz, M. Despont, N. Fahrni, J. Brugger, P. Vet-
mal conductance measurements on suspended tiger, P. Renaud: High-aspect-ratio, ultrathick,
monocrystalline nanostructure, Appl. Phys. Lett. negative-tone-near-UV photoresist and its appli-
70, 2687–2689 (1997) cations in MEMS, Sens. Actuators A 64, 33–39 (1998)
11.89 J. Miao, B.L. Weiss, H.L. Hartnagel: Micro- 11.103 H. Lorenz, M. Despont, N. Fahrni, N. LaBianca,
machining of three-dimensional GaAs mem- P. Renaud, P. Vettiger: SU-8: A low-cost negative
brane structures using high-energy nitrogen im- resist for MEMS, J. Micromech. Microeng. 7, 121–124
plantation, J. Micromech. Microeng. 13, 35–39 (1997)
(2003) 11.104 E.H. Conradie, D.F. Moore: SU-8 thick photore-
11.90 C. Seassal, J.L. Leclercq, P. Viktorovitch: Fabrica- sist processing as a functional material for MEMS
tion of inp-based freestanding microstructures by applications, J. Micromech. Microeng. 12, 368–374
selective surface micromachining, J. Micromech. (2002)
Microeng. 6, 261–265 (1996) 11.105 C.T. Pan, H. Yang, S.C. Shen, M.C. Chou, H.P. Chou:
11.91 J. Leclerq, R.P. Ribas, J.M. Karam, P. Viktorovitch: A low-temperature wafer bonding technique using
III–V micromachined devices for microsystems, Mi- patternable materials, J. Micromech. Microeng. 12,
croelectron. J. 29, 613–619 (1998) 611–615 (2002)
11.92 H. Yamaguchi, R. Dreyfus, S. Miyashita, Y. Hi- 11.106 P.A. Stupar, A.P. Pisano: Silicon, parylene, and
rayama: Fabrication and elastic properties of InAs silicon/parylene micro-needles for strength and
freestanding structures based on InAs/GaAs(111) toughness, 11th Int. Conf. Solid State Sens. Ac-
a heteroepitaxial systems, Physica E 13, 1163–1167 tuators, Technical Digest (Springer, Berlin 2001)
(2002) pp. 1368–1389
11.93 C. Lee, T. Itoh, T. Suga: Micromachined piezoelectric 11.107 J.M. Zara, S.W. Smith: Optical scanner using a MEMS
force sensors based on PZT thin films, IEEE Trans. actuator, Sens. Actuators A 102, 176–184 (2002)
Ultrason. Ferroelectr. Freq. Control 43, 553–559 11.108 H.S. Noh, P.J. Hesketh, G.C. Frye-Mason: Parylene
(1996) gas chromatographic column for rapid thermal cy-
11.94 B. Xu, L.E. Cross, J.J. Bernstein: Ferroelectric and cling, J. Microelectromech. Syst. 11, 718–725 (2002)
antiferroelectric films for microelectromechanical 11.109 T.J. Yao, X. Yang, Y.C. Tai: BrF3 dry release technol-
systems applications, Thin Solid Films 377/378, 712– ogy for large freestanding parylene microstructures
718 (2000) and electrostatic actuators, Sens. Actuators A
11.95 S.P. Beeby, A. Blackburn, N.M. White: Process- 97/98, 771–775 (2002)
ing of PZT piezoelectric thick films on silicon 11.110 X. Wang, J. Engel, C. Liu: Liquid crystal polymer
for microelectromechanical systems, J. Micromech. (LCP) for MEMS: Processing and applications, J. Mi-
Microeng. 9, 218–229 (1999) cromech. Microeng. 13, 628–633 (2003)
11.96 C. Shearwood, M.A. Harradine, T.S. Birch, 11.111 C.J. Lee, S.J. Oh, J.K. Song, S.J. Kim: Neural signal
J.C. Stevens: Applications of polyimide membranes recording using microelectrode arrays fabricated
to MEMS technology, Microelectron. Eng. 30, 547– on liquid crystal polymer material, Mater. Sci. Eng.
550 (1996) C 4, 265–268 (2004)
11.97 F. Jiang, G.B. Lee, Y.C. Tai, C.M. Ho: A flexible 11.112 F.F. Faheem, K.C. Gupta, Y.C. Lee: Flip-chip as-
micromachine-based shear-stress sensor array sembly and liquid crystal polymer encapsulation
and its application to separation-point detection, for variable MEMS capacitors, IEEE Trans. Microw.
Sens. Actuators 79, 194–203 (2000) Theory Tech. 51, 2562–2567 (2003)
357
Part B
MEMS/NE Part B MEMS/NEMS and BioMEMS/NEMS
MEMS/NEMS D
12. MEMS/NEMS Devices and Applications
Part B 12
Darrin J. Young, Christian A. Zorman, Mehran Mehregany
Microelectromechanical Systems, generally referred to microassembled with electronics on the same substrate
as MEMS, has had a history of research and devel- or package, achieving high-performance functional sys-
opment over a few decades. Besides the traditional tems. These devices and systems have played key roles
microfabricated sensors and actuators, the field covers in many important areas such as transportation, com-
micromechanical components and systems integrated or munication, automated manufacturing, environmental
360 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12
Travel direction
Fig. 12.2 SEM micrograph of polysilicon microgears (af-
ter [12.3])
ficult to provide an overview covering every aspect of amples in this chapter. For a wide-ranging discussion
Part B 12.1
the topic. It is the authors’ intent to illustrate key as- of nearly all types of micromachined sensors and actu-
pects of MEMS technology and its impact to specific ators, books by Kovacs [12.4] and Senturia [12.5] are
applications by selecting a few demonstrative device ex- recommended.
Capacitive Sensor
Fig. 12.3 Cross-sectional schematic of a piezoresistive Capacitive pressure sensors are attractive because they
pressure sensor are virtually temperature independent and consume zero
362 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12.1
Vacuum cavity
Silicon wafer
Silicon wafer form the sensor diaphragm, which serves as the device
top electrode. Contact pads are formed by metallization
Fig. 12.5 Cross-sectional schematic of a touch-mode ca-
and patterning. This type of pressure sensor exhibits
pacitive pressure sensor
a nonlinear characteristic and a limited dynamic range.
These phenomena, however, can be alleviated through
applying an electrostatic force-balanced feedback ar-
DC power. The devices do not exhibit initial turn-on chitecture. A common practice is to introduce another
drift and are stable over time. Furthermore, CMOS mi- electrode above the sensing diaphragm through wafer
croelectronic circuits can be readily interfaced with the bonding [12.7], thus forming two capacitors in se-
sensors to provide advanced signal conditioning and ries with the diaphragm being the middle electrode.
processing, thus improving overall system performance. The capacitors are interfaced with electronic circuits,
An example of a capacitive pressure sensor is shown in which convert the sensor capacitance value to an output
Fig. 12.4. The device consists of an edge clamped sil- voltage corresponding to the diaphragm position. This
icon diaphragm suspended over a vacuum cavity. The voltage is further processed to generate a feedback sig-
diaphragm can be square or circular with a typical nal to the top electrode, thus introducing an electrostatic
thickness of a few micrometers and a length or radius pull up force to maintain the deflectable diaphragm at
of a few hundreds micrometers, respectively. The vac- its nominal position. This negative feedback loop would
uum cavity typically has a depth of a few micrometers. substantially minimize the device nonlinearity and also
The diaphragm and substrate form a pressure depen- extend the sensor dynamic range.
dent air-gap variable capacitor. An increased external A capacitive pressure sensor achieving an inherent
pressure causes the diaphragm to deflect towards the linear characteristic response and a wide dynamic range
substrate, thus resulting in an increase in the capac- can be implemented by employing a touch-mode ar-
itance value. A simplified fabrication process can be chitecture [12.8]. Figure 12.5 shows the cross-sectional
outlined as follows. A silicon wafer is first patterned view of a touch-mode pressure sensor. The device con-
and etched to form the cavity. The wafer is then ox- sists of an edge-clamped silicon diaphragm suspended
idized followed by bonding to a second silicon wafer over a vacuum cavity. The diaphragm deflects under an
with a heavily-doped boron layer, which defines the di- increasing external pressure and touches the substrate,
aphragm thickness, at the surface. The bonding process causing a linear increase in the sensor capacitance value
can be performed in vacuum to realize the vacuum cav- beyond the touch point pressure. Figure 12.6 shows
ity. If the vacuum bonding is not performed at this stage, a typical device characteristic curve. The touch point
a low pressure sealing process can be used to form the pressure can be designed through engineering the sen-
vacuum cavity after patterning the sensor diaphragm, sor geometric parameters such as the diaphragm size,
provided that sealing channels are available. The sil- thickness, cavity depth, etc., for various application
icon substrate above the boron layer is then removed requirements. The device can be fabricated using a pro-
through a wet etching process, followed by patterning to cess flow similar to the flow outlined for the basic
MEMS/NEMS Devices and Applications 12.1 MEMS Devices and Applications 363
Part B 12.1
Suspended diaphragm (0.8 mm diameter) Anchor Sacrificial layer
Substrate
Structural layer
Substrate
Air gap
Substrate
Accelerometer
Part B 12.1
Top and Silicon
An accelerometer generally consists of a proof mass bottom proof
Silicon
suspended by compliant mechanical suspensions an- frame stiffener mass
chored to a fixed frame. An external acceleration supports
displaces the support frame relative to the proof mass.
The displacement can result in an internal stress change
in the suspension, which can be detected by piezoresis-
tive sensors as a measure of the external acceleration.
The displacement can also be detected as a capacitance
change in capacitive accelerometers. Capacitive sensors
are attractive for various applications because they ex-
hibit high sensitivity and low temperature dependence,
turn-on drift, power dissipation, and noise. The sensors
can also be readily integrated with CMOS electronics Top and bottom
to perform advanced signal processing for high system support beams
performance. Capacitive accelerometers may be divided Metal pads
into two categories as vertical and lateral type sensors.
Figure 12.11 shows sensor structures for the two ver- 1 mm
sions. In a vertical device, the proof mass is suspended
above the substrate electrode by a small gap typically on Fig. 12.13 SEM micrograph of a MEMS z-axis accelerom-
the order of a micrometer, forming a parallel-plate sense eter fabricated by using a combined surface and bulk
capacitance. The proof mass moves in the direction micromachining technology (after [12.15])
perpendicular to the substrate (z-axis) upon a vertical
input acceleration, thus changing the gap and hence in a plane parallel to the substrate when subjected to
the capacitance value. The lateral accelerometer con- a lateral input acceleration, thus changing the over-
sists of a number of movable fingers attached to the lap area of these fingers; hence the capacitance value.
proof mass, forming a sense capacitance with an array Figure 12.12 shows an SEM top view of a surface-
of fixed parallel fingers. The sensor proof mass moves micromachined polysilion z-axis accelerometer [12.14].
The device consists of a 400 × 400 μm2 proof mass with
a thickness of 2 μm suspended above the substrate elec-
trode by four folded beam suspensions with an air gap
around 2 μm, thus achieving a sense capacitance of
≈ 500 fF. The visible holes are used to ensure com-
plete removal of the sacrificial oxide underneath the
proof mass at the end of the fabrication process. The
sensor can be interfaced with a microelectronic charge
amplifier converting the capacitance value to an out-
put voltage for further signal processing and analysis.
Force feedback architecture can be applied to stabilize
the proof mass position. The combs around the periph-
ery of the proof mass can exert an electrostatic levitation
force on the proof mass to achieve the position con-
trol, thus improving the system frequency response and
linearity performance [12.14].
Surface micromachined accelerometers typically
suffer from severe mechanical thermal vibration, com-
monly referred to as Brownian motion [12.16], due
to the small proof mass, thus resulting in a high me-
chanical noise floor which ultimately limits the sensor
Fig. 12.12 SEM micrograph of a polysilicon surface- resolution. Vacuum packaging can be employed to min-
micromachined z-axis accelerometer (after [12.14]) imize this adverse effect but with a penalty of increasing
366 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12.1
100 µm
Fig. 12.14 SEM micrograph of a polysilicon surface-micromachined
lateral accelerometer (
c Analog Devices Inc.)
Part B 12.1
sensing capacitance configuration is attractive for in- Structural anchor to substrate
terfacing with differential electronic detection circuits
to suppress common-mode noise and other undesir-
able signal coupling. Monolithic accelerometers with z-Axis
a three-axis sensing capability integrated with on-chip
Input rotation
electronic detection circuits have been realized using Sense mode
surfacing micromachining and CMOS microelectron- Comb drives to
ics fabrication technologies [12.18]. Figure 12.16 shows sustain oscillation
a photo of one of these microsystem chips, which has an
area of 4 × 4 mm2 . One vertical accelerometer and two
lateral accelerometers are placed at the chip center with
corresponding detection electronics along the periphery. Driven mode
A z-axis reference device, which is not movable, is used Interdigitated comb finger
deflection sense capacitors
with the vertical sensor for electronic interfacing. The
prototype system achieves a sensing resolution on the Fig. 12.17 Schematic of a vibratory rate gyroscope
order of 1 mG with a 100 Hz bandwidth along each axis.
The level of performance is adequate for automobile
safety activation systems, vehicle stability and active
suspension control, and various consumer products.
Recently, monolithic MEMS accelerometers fabricated
by using post-CMOS surface micromachining fabri-
cation technology have been developed√ to achieve an
acceleration noise floor of 50 μg/ Hz [12.19]. This
technology can enable MEMS capacitive inertial sen-
sors to be integrated with interface electronics in
a commercial CMOS process, thus minimizing proto-
typing cost.
Gyroscope
Most of micromachined gyroscopes employ vibrating
mechanical elements to sense rotations. The sensors
rely on energy transfer between two vibration modes 4 mm
of a structure caused by Coriolis acceleration. Fig-
ure 12.17 presents a schematic of a z-axis vibratory Fig. 12.18 Photo of a monolithic polysilicon surface-micromachined
rate gyroscope. The device consists of an oscillat- z-axis vibratory gyroscope with integrated interface and control
ing mass electrostatically driven into resonance along electronics (after [12.20])
the drive-mode axis using comb fingers. An angu- √
lar rotation along the vertical axis (z-axis) introduces of ≈ 1◦ /(s Hz) under a vacuum pressure around
a Coriolis acceleration, which results in a structure 50 mTorr. Other MEMS single-axis gyroscopes inte-
deflection along the sense-mode axis, shown in the grated in commercial IC processes were demonstrated
figure. The deflection changes the differential sense recently achieving an enhance performance [12.21, 22].
capacitance value, which can be detected as a mea- A dual-axis gyroscope based on a rotational disk
sure of input angular rotation. A z-axis vibratory rate at its resonance can be used to sense angular rotation
gyroscope operating upon this principle is fabricated along two lateral axes (x- and y-axis). Figure 12.19
using surface micromachining technology and inte- shows a device schematic demonstrating the operating
grated together with electronic detection circuits, as principle. A rotor disk supported by four mechani-
illustrated in Fig. 12.18 [12.20]. The micromachined cal suspensions can be driven into angular resonance
sensor is fabricated using polysilicon structural ma- along the z-axis. An input angular rotation along the
terial with a thickness around 2 μm and occupies an x-axis will generate a Coriolis acceleration causing the
area of 1 × 1 mm2 . The sensor achieves a resolution disk to rotate along the y-axis, and vice versa. This
368 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12.1
z-Axis drive
x-Axis
coriolis y-Axis coriolis
output output
oscillations oscillations
x,y-Axis
gyro
Fig. 12.21 Photo of a surface-micromachined inertial measurement system with a six-degree sensing capability
MEMS/NEMS Devices and Applications 12.1 MEMS Devices and Applications 369
Part B 12.1
Surface micromachining has served as a key enabling
technology to realize microeletromechancal optical
devices for various applications ranging from so-
phisticated visual information displays and fiber-optic
telecommunication to bar-code reading. Most of the
existing optical systems are implemented using conven-
tional optical components, which suffer from bulky size,
high cost, large power consumption, poor efficiency
and reliability issues. MEMS technology is promising
for producing miniaturized, reliable, inexpensive opti-
cal components to revolutionize conventional optical
systems [12.25]. In this section of the chapter, a few
selected MEMS optical devices will be presented to Fig. 12.23 SEM micrograph of a close-up view of a DMD
illustrate their impact in the fields of visual display, pre- pixel array (after [12.24])
cision optical platform, and data switching for optical
communication. ure 12.23 shows an SEM micrograph of a close-up view
of a DMD pixel array [12.24]. Each mirror is capable of
Visual Display rotating by ±10◦ corresponding to either the on or off
An early MEMS device successfully used for various position due to an electrostatic actuation force. Light
display applications is the Texas Instruments Digital reflected from any on-mirrors passes through a projec-
Micromirror Device (DMD). The DMD technology tion lens and creates images on a large screen. Light
can achieve higher performance in terms of resolution, from the remaining off-mirrors is reflected away from
brightness, contrast ratio, and convergence than the con- the projection lens to an absorber. The proportion of
ventional cathode ray tube and is critical for digital time during each video frame that a mirror remains in
high-definition television applications. A DMD con- the on-state determines shades of gray, from black for
sists of a large array of small mirrors with a typical zero on-time to white for a hundred percent on-time.
area of 16 × 16 μm2 as illustrated in Fig. 12.22. A probe Color can be added by a color wheel or a three-DMD
tip is shown in the figure for a size comparison. Fig- chip setup. The three DMD chips are used for pro-
jecting red, green and blue colors. Each DMD pixel
consists of a mirror connected by a mirror support post
to an underlying yoke. The yoke in turn is connected
by torsion hinges to hinge support posts, as shown in
Fig. 12.24 [12.26]. The support post and hinges are
hidden under the mirror to avoid light diffraction and
thus improve contrast ratio and optical efficiency. There
are two gaps on the order of a micrometer, one be-
tween the mirror and the underlying hinges and address
electrodes, and a second between the coplanar address
electrodes and hinges and an underlying metal layer
from the CMOS static random access memory (SRAM)
structure. The yoke is tilted over the second gap by
an electrostatic actuation force, thus rotating the mir-
ror plate. The SRAM determines which angle the mirror
needs to be tilted by applying proper actuation volt-
ages to the mirror and address electrodes. The DMD is
fabricated using an aluminum-based surface microma-
chining technology. Three layers of aluminum thin film
Fig. 12.22 Photo of a digital micromirror device (DMD) are deposited and patterned to form the mirror and its
array (c Texas Instruments) suspension system. Polymer material is used as sacrifi-
370 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12.1
Layers
Mirror
Mirror
Yoke Mirror
and address
hinge electrode
Via 2
Yoke address contact
electrode to CMOS
2.5 nm
Part B 12.1
Comb resonator Slider Support beam Mirror
microreflector to redirect an optical beam in a desir-
able location. Each of the two slides is actuated with
an integrated microvibromotor shown in Fig. 12.27b.
The vibromotor consists of four electrostatic comb res-
onators with attached impact arms driving a slider
through oblique impact. The two opposing impacters
are used for each travel direction to balance the forces.
The resonator is a capacitively driven mass anchored to
the substrate through a folded beam flexure. The flex-
ure compliance determines the resonant frequency and
travel range of the resonator. When the comb struc-
tures are driven at their resonant frequency (around
8 kHz), the slider exhibits a maximum velocity of over
1 mm/s. Characterization of the vibromotor also shows
that a slider step resolution of less than 0.3 μm can
be achieved [12.31], making it attractive for precision
alignment of various optical components. The prototype
Impact arms Slider
microreflector can obtain an angular travel range over
90◦ and a translational travel range of 60 μm. By using
this device, beam steering, fiber coupling, and optical
scanning have been demonstrated.
Unactivated
mirror
Focusing
optics Outputs
Inputs Optical
fiber arrays
Fig. 12.28 Schematic of a two-dimensional micromirror-based fiber
optic switching matrix
100 µm
to corresponding output fibers. As described in the Fig. 12.30 SEM micrograph of a polysilicon surface-
previous sections, these micromirrors can be batch fab- micromachined vertical torsion mirror (after [12.32])
ricated using silicon micromachining technologies, thus
achieving an integrated solution with the potential for extended to a 64 × 64 matrix. The micromirrors are
low cost. A significant reduction in power dissipation is moved between two fixed stops by digital control, thus
also expected. eliminating the need for precision motion control. Fig-
Figure 12.28 shows an architecture of a two- ure 12.29 presents an SEM micrograph of a simple
dimensional micromirror array forming a switching 2 × 2 MEMS fiber optic switching network prototype for
matrix with rows of input fibers and columns of output an illustration purpose [12.32]. The network includes
fibers (or vice versa). An optical beam from an input a mirror chip passively integrated with a silicon sub-
fiber can be directed to an output fiber through activat- mount, which contains optical fibers and ball lenses.
ing the corresponding reflecting micromirror. Switches The mirror chip consists of four surface-micromachined
with eight inputs and eight outputs can be readily im- vertical torsion mirrors. The four mirrors are arranged
plemented using this technique, which can be further such that in the reflection mode, the input beams are re-
flected by two 45◦ vertical torsion mirrors and coupled
into the output fibers located on the same side of the
Vertical torsion mirror devices Ball lenses (D = 300 µm)
chip. In the transmission mode, the vertical torsion mir-
rors are rotated out of the optical paths, thus allowing
the input beams to be coupled into the opposing out-
put fibers. Figure 12.30 shows an SEM micrograph of
a polysilicon vertical torsion mirror. The device con-
sists of a mirror plate attached to a vertical supporting
frame by torsion beams and a vertical back electrode
Mirror chip
plate. The mirror plate is ≈ 200 μm wide, 160 μm long,
and 1.5 μm thick. The mirror surface is coated with
a thin layer of gold to improve the optical reflectiv-
ity. The back plate is used to electrostatically actuate
the mirror plate so that the mirror can be rotated out
of the optical path in the transmission mode. Surface
micromachining with microhinge technology is used to
1 mm realize the overall structure. The back electrode plate
is integrated with a scratch drive actuator array [12.33]
Silicon submount Optical fibers for self-assembly. The self-assembly approach is criti-
Fig. 12.29 SEM micrograph of a 2 × 2 MEMS fiber optic cal when multiple vertical torsion mirrors are used to
switching network (after [12.32]) implement more advanced functions.
MEMS/NEMS Devices and Applications 12.1 MEMS Devices and Applications 373
Part B 12.1
Input/output fibers Focusing optics Fixed mirror
with a large scaling potential can be implemented by us-
ing a three-dimensional (3-D) switching architecture as
shown in Fig. 12.31. The network consists of arrays of
two-axis mirrors to steer optical beams from input fibers
to output fibers. A precision analog closed-loop mirror
position control is required to accurately direct a beam
along two angles so that one input fiber can be opti-
cally connected to any output fiber. The optical length
depends little on which set of fibers are connected,
thus achieving a more uniform switching characteristic,
which is critical for implementing large scale network.
Two-axis mirrors are the crucial components for imple-
menting the 3-D architecture. Figure 12.32 shows an
SEM micrograph of a surface-micromachined two-axis
beam-steering mirror positioned by using self-assembly
technique [12.34]. The self-assembly is accomplished
during the final release step of the mirror processing
sequence. Mechanical energy is stored in a special high-
stress layer during the deposition, which is put on top
of the four assembly arms. Immediately after the as-
sembly arms are released, the tensile stress in this layer
causes the arms to bend up, pushing the mirror frame
and lifting it above the silicon substrate. All mirrors
Two-axis gimbaled mirror array
used in the switching network can be fabricated simul-
taneously without any human intervention or external Fig. 12.31 Schematic of a three-dimensional micromirror-based
power supply. fiber optic switching matrix
12.1.4 RF MEMS
RF PLL PLL IF
VCO VCO
RF
switch
Crystal
RF PLL
RF filter VCO
(ceramic) Power amplifier
a superheterodyne radio architecture, in which discrete able matching networks, etc. Capacitors with high
components are shaded in dark color. Theses off-chip quality factors (Q), large tuning range and linear
devices occupy the majority of the system area, thus characteristics are crucial for achieving system perfor-
severely hindering transceiver miniaturization. MEMS mance requirements. On-chip silicon PN junction and
technology, however, offers a potential solution to inte- MOS based variable capacitors suffer from low qual-
grate these discrete components onto silicon substrates ity factors (below 10 at 1 GHz), limited tuning range
with microelectronics, achieving a size reduction of and poor linearity, thus are inadequate for building
a few orders of magnitude. It is therefore expected to be- high-performance transceivers. MEMS technology has
come an enabling technology to ultimately miniaturize demonstrated monolithic variable capacitors achieving
radio transceivers for future wireless communications. stringent performance requirements. These devices typ-
ically reply on an electrostatic actuation method to vary
MEMS Variable Capacitors the air gap between a set of parallel plates [12.35–38]
Integrated high-performance variable capacitors are or vary the capacitance area between a set of conduc-
critical for low noise VCOs, antenna tuning, tun- tors [12.39] or mechanically displace a dielectric layer
a) b)
Fig. 12.34 (a) SEM micrograph of a top view of an aluminum surface-micromachined variable capacitor; (b) SEM
micrograph of a cross-sectional view of the variable capacitor (after [12.35])
MEMS/NEMS Devices and Applications 12.1 MEMS Devices and Applications 375
Part B 12.1
have been achieved with various device configurations.
Figure 12.34 shows SEM micrographs of an aluminum
micromachined variable capacitor fabricated on a sil-
icon substrate [12.35]. The device consists of a 200 ×
200 μm2 aluminum plate with a thickness of 1 μm sus-
pended above the bottom electrode by an air gap of
1.5 μm. Aluminum is selected as the structural mater-
ial due to its low resistivity, critical for achieving a high
quality factor at high frequencies. A DC voltage ap-
plied across the top and bottom electrodes introduces an
electrostatic pull-down force, which pulls the top plate
towards the bottom electrode, thus changing the device Fig. 12.35 SEM micrograph of four MEMS aluminum variable ca-
capacitance value. The capacitors are fabricated using pacitors connected in parallel (after [12.35])
aluminum-based surface micromachining technology.
Sputtered aluminum is used for building the capaci- of 62 has been demonstrated at 1 GHz, which matches
tor top and bottom electrodes. Photoresist is served as or exceeds that of discrete varactor diodes and is at
the sacrificial layer, which is then removed through least an order of magnitude larger than that of a typ-
an oxygen-based plasma dry etch to release the mi- ical junction capacitor implemented in a standard IC
crostructure. The processing technology requires a low process.
thermal budget, thus allowing the variable capacitors to MEMS tunable capacitors based upon varying ca-
be fabricated on top of wafers with completed electronic pacitance area between a set of conductors have been
circuits without degrading the performance of active de- demonstrated. Figure 12.36 shows an SEM micrograph
vices. Figure 12.35 presents an SEM micrograph of four of a such device [12.39]. The capacitor comprises
MEMS tunable capacitors connected in parallel. This arrays of interdigitated electrodes, which can be elec-
device achieves a nominal capacitance value of 2 pF trostatically actuated to vary the electrode overlap
and a tuning range of 15% with 3 V. A quality factor area. A close-up view of the electrodes is shown in
Fig. 12.37. The capacitor is fabricated using a silicon-
500 µm
5 µm
Fig. 12.36 SEM micrograph of a silicon tunable capacitor Fig. 12.37 SEM micrograph of a close view of a tunable
using a comb drive actuator (after [12.39]) capacitor comb fingers (after [12.39])
376 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 12.1
0V 4V
2V 5V
200 µm
16 µm
RF Tuning
Fig. 12.39 SEM micrograph of a copper surface-microma-
Fig. 12.38 Photos of comb fingers at different actuation voltages chined tunable capacitor with a movable dielectric layer
(after [12.39]) (after [12.40])
Part B 12.1
is mainly caused by substrate loss and metal resistive
loss at high frequencies. Micromachining technology
provides an attractive solution to minimize these loss
contributions; hence enhancing the device quality fac-
tors. Figure 12.40 shows an SEM micrograph of a 3-D
coil inductor fabricated on a silicon substrate [12.41].
The device consists of 4-turn 5 μm thick copper traces
electroplated around an insulting core with a 650 μm
by 500 μm cross section. Compared to spiral induc-
tors, this geometry minimizes the coil area which is in
close proximity to the substrate and hence the eddy-
current loss, resulting in a maximized Q-factor and
device self-resonant frequency. Copper is selected as 1 mm
the interconnect metal because of its low sheet re-
sistance, critical for achieving a high Q-factor. The
inductor achieves a 14 nH inductance value with a qual- Fig. 12.42 SEM micrograph of a self-assembled out-of-
ity factor of 16 at 1 GHz. A single-turn 3-D device plane coil inductor (after [12.43])
exhibits a Q-factor of 30 at 1 GHz, which matches
the performance of discrete counterparts. The high-Q
3-D inductor and MEMS tunable capacitors, shown in
Fig. 12.35, have been employed to implement a RF
CMOS VCO achieving a low phase noise performance
suitable for typical wireless communication applica-
tions such as GMS cellular telephony [12.42].
Other 3-D inductor structures such as the levi-
tated spiral inductors have been demonstrated using
micromachining fabrication technology. Figure 12.41
shows an SEM micrograph of a levitated copper in-
ductor, which is suspended above the substrate through
supporting posts [12.44]. The levitated geometry can
minimize the substrate loss, thus achieving an im-
proved quality factor. The inductor shown in the figure
100 µm
MEMS Switches
The microelectromechanical switch is another poten-
tially attractive miniaturized component enabled by
micromachining technologies. These switches offer su-
perior electrical performance in terms of insertion loss,
isolation, linearity, etc., and are intended to replace
off-chip solid-state counterparts, which provide switch-
ing between the receiver and transmitter signal paths.
They are also critical for building phase shifters, tun-
able antennas, and filters. The MEMS switches can
be characterized into two categories: capacitive and
Ground Dielectric Lower electrode
metal-to-metal contact types. Figure 12.44 presents
a cross-sectional schematic of an RF MEMS capacitive Fig. 12.45 Top view photo of a fabricated RF MEMS ca-
switch. The device consists of a conductive membrane, pacitive switch (after [12.46])
typically made of aluminum or gold alloy suspended
above a coplanar electrode by an air gap of a few mi- a strong signal coupling (low insertion loss). The ca-
crometers. For RF or microwave applications, actual pacitive switch consumes near-zero power, which is
metal-to-metal contact is not necessary; rather, a step attractive for low power portable applications. Switch-
change in the plate-to-plate capacitance realizes the ing cycles over millions for this type of device have
switching function. A thin silicon nitride layer with been demonstrated. Figure 12.45 shows a top view
a thickness on the order of 1000 Å is typically de- photo of a fabricated MEMS capacitive switch [12.46].
posited above the bottom electrode. When the switch Surface micromachining technology, using metal for the
is in the on-state, the membrane is high resulting in electrodes and polymer as the sacrificial layer, is used
a small plate-to-plate capacitance; hence, a minimum to fabricate the device. The switch can be actuated with
high-frequency signal coupling (high isolation) between a DC voltage on the order of 50 V and exhibits a low
the two electrodes. In the off-state (with a large enough insertion loss of ≈ − 0.28 dB at 35 GHz and a high
applied DC voltage), the switch provides a large ca- isolation of −35 dB at the same frequency.
pacitance due to the thin dielectric layer, thus causing Metal-to-metal contact switches are important for
interfacing large bandwidth signals including DC. This
type of device typically consists of a cantilever beam
a) Suspended Silicon nitride or clamped-clamped bridge with a metallic contact pad
membrane positioned at the beam tip or underneath bridge cen-
Post Switch up Post
ter. Through an electrostatic actuation, a contact can
be formed between the suspended contact pad and an
Input Output underlying electrode on the substrate [12.47–49]. Fig-
ure 12.46 shows a cross-sectional schematic of a metal-
b)
to-metal contact switch [12.49]. The top view of the
fabricated device is presented in Fig. 12.47. The switch
exhibits an actuation voltage of 30 V, a response time
Input
Switch down
Output
of 20 μs, and mechanical strength to withstand 109 actu-
ations. An isolation greater than 50 dB below 2 GHz and
insertion loss less than 0.2 dB from DC through 40 GHz
Fig. 12.44 Cross-sectional schematics of an RF MEMS ca- has been demonstrated. Metal-to-metal contact switches
pacitive switch relying on electrothermal actuation method have also
MEMS/NEMS Devices and Applications 12.1 MEMS Devices and Applications 379
Part B 12.1
Up state Sustaining CMOS circuitry Micromechanical resonator
Down state
Spring coupler Electrode Anchor µResonators Output electrode Anchor Transducer Plated input
gap electrodes
R = 17 µm
20 μm
Electrode Electrode
Fig. 12.49 SEM micrograph of a polysilicon surface-
micromachined two-resonator spring-coupled bandpass
micromechanical filter (after [12.54])
by sequential assembly using atomic and/or molecu- shape of nanomechanical beams (Fig. 12.1b). Ni was
Part B 12.2
lar building blocks. While advancements in bottom-up then deposited and patterned by lifting off the PMMA
approaches are developing at a very rapid pace, most (Fig. 12.1c). Next, the underlying oxide film was pat-
advanced NEMS devices are currently created utiliz- terned by RIE using the Ni film as an etch mask. After
ing top-down techniques that combine existing process oxide etching, nanomechanical beams were patterned
technologies, such as electron-beam lithography, con- by etching the Si substrate using RIE, as shown in
ventional film growth and chemical etching. Top-down Fig. 12.1d. Following Si RIE, the Ni etch mask was
approaches make integration with microscale packag- removed and the sidewalls of the Si nanomechanical
ing relatively straightforward since the only significant beams were lightly oxidized in order to protect them
difference between the nanoscale and microscale pro- during the release step (Fig. 12.1e). After performing an
cessing steps is the method used to pattern the various anisotropic SiO2 etch to clear any oxide from the field
features. areas, the Si beams were released using an isotropic Si
In large measure, NEMS has followed a de- RIE step, as shown in Fig. 12.1f. After release, the pro-
velopmental path similar to the route taken in the tective SiO2 film was removed by wet etching in HF
development of MEMS in that both have leveraged (Fig. 12.1g). Using this process, the authors reported the
existing processing techniques from the IC industry. successful fabrication of nanomechanical Si beams with
For instance, the electron-beam lithographic techniques micrometer-scale lengths (≈ 8 μm) and submicrometer
used in top-down NEMS fabrication are the same widths (330 nm) and heights (800 nm).
techniques that have become standard in the fabri- The advent of silicon-on-insulator (SOI) substrates
cation of submicrometer transistors. Furthermore, the with high quality, submicrometer-thick silicon top lay-
materials used in many of the first generation, top- ers enables the fabrication of nanomechanical Si beams
down NEMS devices, (Si, GaAs, Si3 N4 , SiC) were with fewer processing steps than the aforementioned
first used in ICs and then in MEMS. Like the first technique, since the buried oxide layer makes these
MEMS devices, the first generation NEMS structures device structures relatively easy to pattern and re-
consisted of free-standing nanomechanical beams, pad- lease. Additionally, the buried SiO2 layer electrically
dle oscillators, and tethered plates made using simple isolates the beams from the substrate. Carr and Craig-
bulk and single layer surface nanomachining processes. head [12.60] detail a process that uses SOI substrates
Recent advancements have focused on incorporating to fabricate submicrometer clamped-clamped mechan-
nanomaterials such as nanotubes and nanowires synthe- ical beams and suspended plates with submicrometer
sized using bottom-up approaches into NEMS devices tethers. The process, presented in Fig. 12.2, begins
by integrating these materials into top-down nano- and with the deposition of PMMA on an SOI substrate.
micromachining processes. The following text serves The SOI substrate has a top Si layer that was either
only at a brief introduction to the technology, high- 50 or 200 nm in thickness. The PMMA is patterned
lighting the key materials, fabrication approaches, and into a metal lift-off mask by electron beam lithog-
emerging application areas. For additional details and raphy (Fig. 12.2b). An Al film is then deposited and
perspectives, readers are encouraged to consult an ex- patterned by lift-off into a Si etch mask, as shown
cellent review on the subject [12.58]. in Fig. 12.2c. The nanomechanical beams are then
patterned by Si RIE and released by etching the under-
12.2.1 Materials and Fabrication Techniques lying SiO2 in a buffered hydrofluoric acid solution as
shown in Fig. 12.2d,e, respectively. Using this process,
Like Si MEMS, Si NEMS capitalizes on well-developed nanomechanical beams that were 7–16 μm in length,
processing techniques for Si and the availability of high- 120–200 nm in width and 50 or 200 nm in thickness
quality substrates. Cleland and Roukes [12.59] reported were successfully fabricated.
a relatively simple process to fabricate nanomechanical Fabrication of NEMS structures is not limited to
clamped-clamped beams directly from single-crystal Si. In fact, III–V compounds, such as gallium ar-
(100) Si substrates. As illustrated in Fig. 12.1, the pro- senide (GaAs), make particularly good NEMS materials
cess begins with the thermal oxidation of a Si substrate from a fabrication perspective because thin epitaxial
(Fig. 12.1a). Large Ni contact pads were then fabricated GaAs films can be grown on lattice-matched materials
using optical lithography and lift-off. A polymethyl that can be used as sacrificial release layers. A col-
methacrylate (PMMA) lift-off mold was then deposited lection of clamped-clamped nanomechanical GaAs
and patterned using electron-beam lithography into the beams fabricated on lattice-matched sacrificial layers
382 Part B MEMS/NEMS and BioMEMS/NEMS
having micrometer-scale lengths and submicrometer Nanomechanical resonators have also been fash-
Part B 12.2
widths and thicknesses is shown in Fig. 12.3. Tighe ioned out of thin nanocrystalline diamond thin
et al. [12.61] reported on the fabrication of GaAs plates films [12.65]. In this case, the diamond films were
suspended with nanomechanical tethers. The structures deposited on SiO2 -coated Si substrates by microwave
were made from single-crystal GaAs films that were plasma chemical vapor deposition using CH4 and H2 as
epitaxially grown on aluminum arsenide (AlAs) sac- feedstock. The diamond films were patterned by RIE us-
rificial layers. Ni etch masks were fabricated using ing metal masks patterned by e-beam lithography. The
electron beam lithography and lift-off as described pre- plasma chemistry in this case was based on CF4 and O2 .
viously. The GaAs films were patterned into beams The devices were then released in a buffered HF solu-
using a chemically assisted ion beam etching process tion. It is noteworthy that the structures did not require
and released using a highly selective AlAs etchant. In a critical-point drying step after the wet chemical re-
a second example, Tang et al. [12.62] has capitalized lease, owing to the chemical inertness of the diamond
on the ability to grow high-quality GaAs layers on surface.
ternary compounds such as Alx Ga1−x As to fabricate NEMS structures are not restricted to those that
complex GaAs-based structures, such as submicrometer can be made from patterned thin films using top-
clamped-clamped beams from GaAs/AlGaAs quantum down techniques. In fact, carbon nanotubes (CNT) have
well heterostructures. As with the process described been incorporated into NEMS devices using an ap-
by Tighe et al. [12.61], this process exploits a lat- proach that combines both bottom-up and top-down
tice matched sacrificial layer, in this case Al0.8 Ga0.2 As, processing techniques. An example illustrating the
which can be selectively etched to release the het- promise and challenges of merging bottom-up with
erostructure layers. top-down techniques is the CNT-based electrostatic ro-
Silicon carbide and diamond NEMS structures have tational actuator developed by Fennimore et al. [12.66].
been developed for applications requiring a material In this example, multiwalled CNTs (MWCNT) are
with a higher acoustic velocity and/or a higher degree grown using a conventional arc discharge process,
of chemical inertness than Si. Silicon carbide nanome- which typically produces an assortment of CNTs.
chanical resonators have been successfully fabricated The CNTs are then transferred to a suitable SiO2
from both epitaxial 3C-SiC films grown on Si sub- coated Si in a 1,2-dichlorobenzene suspension. An
strates [12.63] and bulk 6H-SiC substrates [12.64]. In AFM or SEM is then used to select a properly posi-
the case of the 3C-SiC devices, the ultrathin epitaxial tioned CNT as determined by prefabricated alignment
films were grown by atmospheric pressure chem- marks on the substrate. Conventional electron-beam
ical vapor deposition (APCVD) on (100)Si substrates. lithography and lift-off techniques are then used to
Nanomechanical beams were patterned using a metal pattern an Au film into contact/anchor pads on the
RIE mask that was itself patterned by e-beam lithog- two ends of the CNT, a rotor pad at its center and
raphy. Reactive ion etching was performed using two two counter electrodes at 90◦ to the anchor pads.
NH3 -based plasma chemistries, with the first recipe per- Anchoring is accomplished by sandwiching the CNT
forming an anisotropic SiC etch down to the Si substrate between the Au contact and the underlying SiO2 film.
and the second performing an isotropic Si etch used The rotor is released by simply etching the sacrifi-
to release the SiC beams. The two etches were per- cial SiO2 layer, taking care not to completely undercut
formed sequentially, thereby eliminating a separate wet the anchors yet allowing for adequate clearance for
or dry release step. For the 6H-SiC structures, a suitable the rotor. Under proper conditions, the outer wall
sacrificial layer was not available since the structures of the MWCNTs could be detached from the inner
were fabricated directly on commercially available bulk walls in order to allow for free rotation of the rotor
wafers. To fabricate the structures, a metal etch mask plate.
was lithographically patterned by e-beam techniques on
the 6H-SiC surface. The anisotropic SiC etch mentioned 12.2.2 Transduction Techniques
above was then performed, but with the substrate tilted
roughly 45◦ with respect to the direction of the plasma Several unique approaches have been developed to
using a special fixture to hold the wafer. A second such actuate and sense the motion of NEMS devices. Elec-
etch was performed on the substrate tilted back 90◦ with trostatic actuation can be used to actuate beams [12.67],
respect to the first etch, resulting in released beams with tethered meshes [12.68], and paddle oscillators [12.69].
triangular cross sections. Sekaric et al. [12.70] has shown that low power
MEMS/NEMS Devices and Applications 12.3 Current Challenges and Future Trends 383
lasers can be used to drive paddle oscillators into 12.2.3 Application Areas
Part B 12.3
self-oscillation by induced thermal effects on the struc-
tures. In these examples, an optical detection scheme For the most part, NEMS technology is still in the
based on the modulation of incident laser light by initial stage of development. Technological challenges
a vibrating beam is used to detect the motion of the related to fabrication and packaging will require inno-
beams. Cleland and Roukes [12.59] describe a mag- vative solutions before such devices make a significant
netomotive transduction technique that capitalizes on commercial impact. Nevertheless, NEMS devices have
a time-varying Lorentz force created by an alternat- already been used for precision measurements [12.71]
ing current in the presence of a strong magnetic field. enabling researchers to probe the properties of matter
In this case, the nanomechanical beam is positioned on a nanoscopic level [12.72, 73]. Sensor technologies
in the magnetic field so that an AC current pass- based on NEMS structures, most notably for attogram
ing through the beam is transverse to the field lines. scale mass detection [12.74, 75], attonewton force de-
The resulting Lorentz force causes the beam to oscil- tection [12.76], virus detection [12.77], and gaseous
late, which creates an electromotive force along the chemical detection [12.78] have emerged and will con-
beam that can be detected as a voltage. Thus, in this tinue to mature. Without question, NEMS structures
method, the excitation and detection are performed will prove to be useful platforms for a host of ex-
electrically. In all of the above-mentioned cases, the periments and scientific discoveries in fields ranging
measurements were performed in vacuum, presum- from physics to biology, and with advancements in pro-
ably to minimize the effects of squeeze-film damping cess integration and packaging, there is little doubt that
as well as mass loading due to adsorbates from the NEMS technology will find its way into commercial
environment. micro/nanosystems as well.
crofabrication techniques into a unified process, since tion of MEMS devices, especially in situations where
Part B 12
NEMS devices are likely to consist of both nanoscale the environment exerts mechanical stresses on the
and microscale structures. Integration will be particu- package, which in turn results in a long-term device per-
larly challenging for nanoscale devices fabricated using formance drift. To address many of these issues, wafer
a bottom-up approach, since no analog is found in level packaging schemes that are customized to the de-
microfabrication. Nevertheless, hybrid systems con- vice of interest will likely become more common. In
sisting of nanoscale and microscale components will essence, packaging of MEMS will move away from
become increasingly common as the field continues to the conventional IC methods that utilize independently
expand. manufactured packages toward custom packages, which
Fabrication issues notwithstanding, packaging is are created specifically for the device as a part of the
and will continue to be a significant challenge to the im- batch fabrication process.
plementation of MEMS. MEMS is unlike IC packaging Without question, the increasing advancement of
which benefits from a high degree of standardiza- MEMS will open many new potential application areas
tion. MEMS devices inherently require interaction with to the technology. In most cases, MEMS will be one
the environment, and since each application has in of several alternatives available for implementation. For
some way a unique environment, standardization of cost sensitive applications, the trade off between techni-
packaging becomes extremely difficult. This lack of cal capabilities and cost will challenge those who desire
standardization tends to drive up the costs associated to commercialize the technology. The biggest challenge
with packaging, making MEMS less competitive with to the field will be to identify application areas that are
alternative approaches. In addition, packaging tends well suited for MEMS/NEMS technology and have no
to negate the effects of miniaturization based upon serious challengers. As MEMS technology moves away
microfabrication, especially for MEMS devices requir- from component level and more towards microsystems
ing protection from certain environmental conditions. solutions, it is likely that such application areas will
Moreover, packaging can cause performance degrada- come to the fore.
References
12.1 M. Mehregany, S.F. Bart, L.S. Tavrow, J.H. Lang, 12.9 D.J. Young, J. Du, C.A. Zorman, W.H. Ko: High-
S.D. Senturia: Principles in design and microfabri- temperature single-crystal 3C-SiC capacitive pres-
cation of variable-capacitance side-drive motors, sure sensor, IEEE Sens. J. 4, 464–470 (2004)
J. Vac. Sci. Technol. A 8, 3614–3624 (1990) 12.10 H. Kapels, R. Aigner, C. Kolle: Monolithic surface-
12.2 Y.-C. Tai, R.S. Muller: IC-processed electrostatic micromachined sensor system for high pressure
synchronous micromotors, Sens. Actuators 20, 49– applications, Int. Conf. Solid-State Sens. Actuators
55 (1989) (2001) pp. 56–59
12.3 J.J. Sniegowski, S.L. Miller, G.F. LaVigne, M.S. Roders, 12.11 J.M. Bustillo, R.T. Howe, R.S. Muller: Surface mi-
P.J. McWhorter: Monolithic geared-mechanisms cromachining for microelectromechanical systems,
driven by a polysilicon surface-micromachined Proc. IEEE 86(8), 1552–1574 (1998)
on-chip electrostatic microengine, IEEE Solid-State 12.12 J.H. Smith, S. Montague, J.J. Sniegowski, J.R. Mur-
Sens. Actuators Workshop (1996) pp. 178–182 ray, P.J. McWhorter: Embedded micromechanical
12.4 G.T.A. Kovacs: Micromachined Transducer Source- devices for the monolithic integration of MEMS with
book (McGraw-Hill, Boston 1998) CMOS, IEEE Int. Electron Dev. Meet. (1993) pp. 609–
12.5 S.D. Senturia: Microsystem Design (Kluwer, Dord- 612
recht 1998) 12.13 T.A. Core, W.K. Tsang, S.J. Sherman: Fabri-
12.6 J.E. Gragg, W.E. McCulley, W.B. Newton, C.E. Der- cation technology for an integrated surface-
rington: Compensation and calibration of a mono- micromachined sensor, Solid State Technol. 36(10),
lithic four terminal silicon pressure transducer, IEEE 39–40, 42, 46–47 (1993)
Solid-State Sens. Actuators Workshop (1984) pp. 21– 12.14 C. Lu, M. Lemkin, B.E. Boser: A monolithic surface
27 micromachined accelerometer with digital output,
12.7 Y. Wang, M. Esashi: A novel electrostatic servo ca- IEEE Int. Solid-State Circuits Conf. (1995) pp. 160–161
pacitive vacuum sensor, IEEE Int. Conf. Solid-State 12.15 N. Yazdi, K. Najafi: An all-silicon single-wafer
Sens. Actuators (1997) pp. 1457–1460 fabrication technology for precision microac-
12.8 W.H. Ko, Q. Wang: Touch mode capacitive pressure celerometers, IEEE Int. Conf. Solid-State Sens.
densors, Sens. Actuators 75, 242–251 (1999) Actuators (1997) pp. 1181–1184
MEMS/NEMS Devices and Applications References 385
12.16 T.B. Gabrielson: Mechanical-thermal noise in mi- 12.31 O. Solgaard, M. Daneman, N.C. Tien, A. Friedberger,
Part B 12
cromachined acoustic and vibration sensors, IEEE R.S. Muller, K.Y. Lau: Optoelectronic packaging
Trans. Electron Dev. 40(5), 903–909 (1993) using silicon surface-micromachined alignment
12.17 J. Chae, H. Kulah, K. Najafi: A monolithic three- mirrors, IEEE Photon. Technol. Lett. 7(1), 41–43
axis micro-g micromachined silicon capacitive (1995)
accelerometer, IEEE J. Solid-State Circuits 14, 235– 12.32 S.S. Lee, L.S. Huang, C.J. Kim, M.C. Wu: 2 × 2 MEMS
242 (2005) fiber optic switches with silicon sub-mount for
12.18 M. Lemkin, M.A. Ortiz, N. Wongkomet, B.E. Boser, P low-cost packaging, IEEE Solid-State Sens. Actu-
J.H. Smith: A 3-axis surface micromachined Δ ators Workshop (1998) pp. 281–284
accelerometer, IEEE Int. Solid-State Circuits Conf. 12.33 T. Akiyama, H. Fujita: A quantitative analysis of
(1997) pp. 202–203 scratch drive actuator using Buckling motion, Tech.
12.19 J. Wu, G.K. Fedder, L.R. Carley: A low-noise low-√ Dig., 8th IEEE Int. MEMS Workshop (1995) pp. 310–
offset capacitive sensing amplifier for a 50 µg/ Hz 315
monolithic CMOS MEMS accelerometer, IEEE J. Solid- 12.34 V.A. Aksyuk, F. Pardo, D.J. Bishop: Stress-induced
State Circuits 39, 722–730 (2004) curvature engineering in surface-micromachined
12.20 W.A. Clark, R.T. Howe: Surface micromachined Z- devices, Proc. SPIE 3680, 984 (1999)
axis vibratory rate gyroscope, IEEE Solid-State Sens. 12.35 D.J. Young, B.E. Boser: A micromachined vari-
Actuators Workshop (1996) pp. 283–287 able capacitor for monolithic low-noise VCOs,
12.21 J.A. Geen, S.J. Sherman, J.F. Chang, S.R. Lewis: IEEE Solid-State Sens. Actuators Workshop (1996)
Single-chip surface micromachined integrated pp. 86–89
gyroscope with 50◦ /h Allan deviation, IEEE J. Solid- 12.36 A. Dec, K. Suyama: Micromachined electro-
State Circuits 37, 1860–1866 (2002) mechanically tunable capacitors and their appli-
12.22 H. Xie, G.K. Fedder: Fabrication, characterization, cations to RF IC’s, IEEE Trans. Microw. Theory Tech.
and analysis of a DRIE CMOS-MEMS gyroscope, IEEE 46, 2587–2596 (1998)
Sens. J. 3, 622–631 (2003) 12.37 Z. Li, N.C. Tien: A high tuning-ratio silicon-
12.23 T. Juneau, A.P. Pisano: Micromachined dual in- micromachined variable capacitor with low driving
put axis angular rate sensor, IEEE Solid-State Sens. voltage, IEEE Solid-State Sens. Actuators Workshop
Actuators Workshop (1996) pp. 299–302 (2002) pp. 239–242
12.24 L.J. Hornbeck: Current status of the digital mi- 12.38 Z. Xiao, W. Peng, R.F. Wolffenbuttel, K.R. Farmer:
cromirror device (DMD) for projection television Micromachined variable capacitor with wide tun-
applications, IEEE Int. Electron Dev. Meet. (1993) ing range, IEEE Solid-State Sens. Actuators Work-
pp. 381–384 shop (2002) pp. 346–349
12.25 R.S. Muller, K.Y. Lau: Surface-micromachined mi- 12.39 J.J. Yao, S.T. Park, J. DeNatale: High tuning-ratio
crooptical elements and systems, Proc. IEEE 86(8), MEMS-based tunable capacitors for RF com-
1705–1720 (1998) munications applications, IEEE Solid-State Sens.
12.26 P.F. Van Kessel, L.J. Hornbeck, R.E. Meier, Actuators Workshop (1998) pp. 124–127
M.R. Douglass: A MEMS-based projection display, 12.40 J.B. Yoon, C.T.-C. Nguyen: A high-Q tunable mi-
Proc. IEEE 86(8), 1687–1704 (1998) cromechanical capacitor with movable dielectric
12.27 M.S. Cohen, M.F. Cina, E. Bassous, M.M. Opy- for RF applications, IEEE Int. Electron Dev. Meet.
rsko, J.L. Speidell, F.J. Canora, M.J. DeFranza: (2000) pp. 489–492
Packaging of high density fiber/laser modules 12.41 D.J. Young, V. Malba, J.J. Ou, A.F. Bernhardt,
using passive alignment techniques, IEEE Trans. B.E. Boser: Monolithic high-performance three-
Compon. Hybrids Manuf. Technol. 15, 944–954 dimensional coil inductors for wireless communi-
(1992) cation applications, IEEE Int. Electron Dev. Meet.
12.28 M.J. Wale, C. Edge: Self-aligned flip-chip assem- (1997) pp. 67–70
bly of photonic devices with electrical and optical 12.42 D.J. Young, B.E. Boser, V. Malba, A.F. Bernhardt:
connections, IEEE Trans. Compon. Hybrids Manuf. A micromachined RF low phase noise voltage-
Technol. 13, 780–786 (1990) controlled oscillator for wireless communication,
12.29 M.J. Daneman, N.C. Tien, O. Solgaard, K.Y. Lau, Int. J. RF Microw. Comput.-Aided Eng. 11(5), 285–
R.S. Muller: Linear vibromotor-actuated micro- 300 (2001)
machined microreflector for integrated optical 12.43 C.L. Chua, D.K. Fork, K.V. Schuylenbergh, J.P. Lu:
systems, IEEE Solid-State Sens. Actuators Workshop Self-assembled out-of-plane high Q inductors,
(1996) pp. 109–112 IEEE Solid-State Sens. Actuators Workshop (2002)
12.30 K.S.J. Pister, M.W. Judy, S.R. Burgett, R.S. Fearing: pp. 372–373
Microfabricated hinges, Sens. Actuators 33(3), 249– 12.44 J.B. Yoon, C.H. Han, E. Yoon, K. Lee, C.K. Kim:
256 (1992) Monolithic high-Q overhang inductors fabricated
386 Part B MEMS/NEMS and BioMEMS/NEMS
on silicon and glass substrates, IEEE Int. Electron from bulk Si crystals, Appl. Phys. Lett. 69, 2653–
Part B 12
12.72 K. Schwab, E.A. Henriksen, J.M. Worlock, M.L. Rou- noelectromechanical oscillators, J. Appl. Phys. 95,
Part B 12
kes: Measurement of the quantum of thermal 3694–3703 (2004)
conductance, Nature 404, 974–977 (2000) 12.76 T.D. Stowe, K. Yasumura, T.W. Kenny, D. Botkin,
12.73 S. Evoy, A. Olkhovets, L. Sekaric, J.M. Parpia, K. Wago, D. Rugar: Attonewton force detection us-
H.G. Craighead, D.W. Carr: Temperature-dependent ing ultrathin silicon cantilevers, Appl. Phys. Lett.
internal friction in silicon nanoelectromechanical 71, 288–290 (1997)
systems, Appl. Phys. Lett. 77, 2397–2399 (2000) 12.77 B. Illic, Y. Yang, H.G. Craighead: Virus detection
12.74 K.L. Ekinci, X.M.H. Huang, M.L. Roukes: Ultrasensi- using nanoelectromechanical devices, Appl. Phys.
tive nanoelectromechanical mass detection, Appl. Lett. 85, 2604–2606 (2004)
Phys. Lett. 84, 4469–4471 (2004) 12.78 H. Liu, J. Kameoka, D.A. Czaplewski, H.G. Craig-
12.75 B. Illic, H.G. Craighead, S. Krylov, W. Senaratne, head: Polymeric nanowire chemical sensor, Nano
C. Ober, P. Neuzil: Attogram detection using na- Lett. 4, 617–675 (2004)
389
Next-Genera
13. Next-Generation DNA Hybridization
and Self-Assembly Nanofabrication Devices
Michael J. Heller, Benjamin Sullivan, Dietrich Dehlinger, Paul Swanson, Dalibor Hodko
Part B 13
13.1 Electronic Microarray Technology ........... 391
The new era of nanotechnology presents many
13.1.1 400 Test Site CMOS Microarray ........ 392
challenges and opportunities. One area of consid-
13.1.2 Electric Field Technology
erable challenge is nanofabrication, in particular
Description .................................. 394
the development of fabrication technologies that 13.1.3 Electronic DNA Hybridization
can evolve into viable manufacturing processes. and Assay Design ......................... 395
Considerable efforts are being expended to re- 13.1.4 DNA Genotyping Applications......... 396
fine classical top-down approaches, such as 13.1.5 On-Chip Strand Displacement
photolithography, to produce silicon-based elec- Amplification............................... 396
tronics with nanometer-scale features. So-called 13.1.6 Cell Separation
bottom-up or self-assembly processes are also on Microelectronic Arrays .............. 397
being researched and developed as new ways of 13.2 Electric Field-Assisted Nanofabrication
producing heterogeneous nanostructures, nano- Processes ............................................. 397
materials and nanodevices. It is also hoped that 13.2.1 Electric Field-Assisted
there are novel ways to combine the best as- Self-Assembly Nanofabrication ...... 397
pects of both top-down and bottom-up processes
13.3 Conclusions .......................................... 399
to create a totally unique paradigm change for
the integration of heterogeneous molecules and References .................................................. 400
nanocomponents into higher order structures.
Over the past decade, sophisticated microelec- have been used primarily for DNA diagnostic ap-
trode array devices produced by the top-down plications, they do have the intrinsic ability to
process (photolithography) have been developed transport almost any type of charged molecule
and commercialized for DNA diagnostic genotyp- or other entity to or from any site on the surface
ing applications. These devices have the ability to of the array. These include other molecules with
produce electric field geometries on their surfaces self-assembling properties such as peptides and
that allow DNA molecules to be transported to or proteins, as well as nanoparticles, cells and even
from any site on the surface of the array. Such de- micron-scale semiconductor components. Micro-
vices are also able to assist in the self-assembly electronic arrays thus have the potential to be
(via hybridization) of DNA molecules at specific used in a highly parallel electric field pick and
locations on the array surface. Now a new genera- place fabrication process allowing a variety of
tion of these microarray devices are available that molecules and nanostructures to be organized into
contain integrated CMOS components within their higher order two- and three-dimensional struc-
underlying silicon structure. The integrated CMOS tures. This truly represents a synergy of combining
allows more precise control over the voltages and the best aspects of top-down and bottom-up
currents sourced to the individual microelectrode technologies into a novel nanomanufacturing
sites. While such microelectronic array devices process.
Nanotechnology and nanoscience are producing a wide terials, energy conversion processes and a new gener-
range of new ideas and concepts, and are likely to ation of biomaterials, biosensors and other biomedical
enable novel nanoelectronics, nanophotonics, nanoma- devices. Many of the challenges and opportunities in
390 Part B MEMS/NEMS and BioMEMS/NEMS
Table 13.1 Some challenges for nanotechnology and nano- simple structures with higher order electronic prop-
fabrication (National Nanotechnology Initiative & NSF) erties [13.2–4]. Nevertheless, the issue of developing
a viable cost-effective self-assembly nanofabrication
(1) Better understanding of scaling problems process that allows billions of nanocomponents to be
and phenomena assembled into useful logic and memory devices still
(2) Better synthetic methods for nano build- remains a considerable challenge. In addition to the na-
ing blocks noelectronic applications, other new nanomaterials and
Part B 13
(3) Control of nanoscale building blocks nanodevices with higher order photonic, mechanical,
(such as size and shape) mechanistic, sensory, chemical, catalytic and therapeu-
(4) Enormous complexity, heterogeneous ma- tic properties are also envisioned as an outcome of
terials and sizes nanotechnology efforts [13.1–4]. Again, a key problem
in enabling such new materials and devices will most
(5) Surfaces for nanostructure assembly
likely be in developing effective nanomanufacturing
(6) Directed hierarchical self-assembly technologies for organizing and integrating heteroge-
(mimic biological) neous components of different sizes and compositions
(7) Need for highly parallel processes into these higher level structures and devices.
(8) Equipment for parallel directed self-as- Living systems provide some of the best exam-
sembly ples of self-assembly or self-organization processes
(9) Integrate bottom-up and top-down that should be considered very closely when devel-
approaches oping strategies for bottom-up nanofabrication. The
molecular biology of living systems includes many
(10) Need better analytical capabilities
molecules which have high fidelity recognition prop-
(11) Tools for modeling and simulations erties such as DNA, RNA, and many types of protein
(12) Scale-up issues for manufacturing macromolecules. Proteins can serve as structural ele-
ments, as binding recognition moieties (antibodies), and
as highly efficient chemomechanical catalytic macro-
nanotechnology have been identified through the efforts molecules (enzymes). Such biomolecules are able
of the National Nanotechnology Initiative [13.1]. While to interact and organize into second-order macro-
many opportunities exist, there are also considerable molecules and nanostructures which store and trans-
challenges that must be met and overcome in order to late genetic information (involving DNA and RNA
obtain the benefits (Table 13.1). Most challenging will structural proteins as well as enzymes), and per-
be those areas that relate to nanofabrication, in particu- form biomolecular syntheses and energy conversion
lar the development of viable fabrication technologies metabolic processes (involving enzymes and structural
which will lead to cost-effective nanomanufacturing proteins). All of these biomolecules, macromolecules,
processes. Enormous efforts are now being carried nanostructures and nanoscale processes are integrated
out to refine classical top-down or photolithography and contained within higher order membrane-encased
processes to produce silicon (CMOS) integrated elec- structures called cells. Cells in turn can then replicate
tronic devices with nanometer-scale features. While this and differentiate (via these nanoscale processes) to form
goal is being achieved, this type of process requires and maintain living organisms. Thus, biology has devel-
billion-dollar fabrication facilities and it appears to be oped the ultimate bottom-up nanofabrication processes
reaching some fundamental limits. So-called bottom- that allow component biomolecules and nanostructures
up self-assembly processes are also being studied and with intrinsic self-assembly and catalytic properties to
developed as possible new ways of producing nanoelec- be organized into highly intricate living systems.
tronics as well as new nanomaterials and nanodevices. Of all the different biomolecules that could be use-
Generally, self-assembly-based nanoelectronics are en- ful for nanofabrication, nucleic acids, with their high
visioned as one of the more revolutionary outcomes fidelity recognition and intrinsic self-assembly prop-
of nanotechnology. There are now numerous exam- erties, represent a most promising material that can
ples of promising nanocomponents such as organic be used to create nanoelectronic, nanophotonic and
electron transfer molecules, quantum dots, carbon nano- many other types of organized nanostructures [13.5–8].
tubes and nanowires, and also some limited success The nucleic acids, which include deoxyribonucleic acid
in first-level assembly of such nanocomponents into (DNA), ribonucleic acid (RNA) and other synthetic
Next-Generation DNA Hybridization and Self-Assembly Nanofabrication Devices 13.1 Electronic Microarray Technology 391
DNA analogs (peptide nucleic acids and so on) are biomolecules), the high-fidelity recognition properties
programmable molecules, which have intrinsic molecu- of DNA molecules are overcome by nonspecific bind-
lar recognition and self-assembly properties via their ing and other entropy-related factors, and the specificity
nucleotide base (A, T, G, C) sequence. Short DNA and efficiency of DNA hybridization is considerably re-
sequences called oligonucleotides are readily synthe- duced. Under in vivo conditions (inside living cells),
sized by automated techniques. They can additionally the binding interactions of high-fidelity recognition
be modified with a variety of functional groups such molecules like DNA are much more controlled and
Part B 13.1
as amines, biotin moieties, fluorescent or chromophore compartmentalized, and the DNA hybridization process
groups, and charge transfer molecules. Additionally, is assisted by structural protein elements and active dy-
synthetic DNA molecules can be attached to quantum namic enzyme molecules. Thus, new nanofabrication
dots, metallic nanoparticles and carbon nanotubes, as processes based on self-assembly or self-organization
well as surfaces like glass, silicon, gold and semicon- using high-fidelity recognition molecules like DNA
ductor materials. Synthetic DNA molecules (oligonu- should also incorporate strategies for assisting and con-
cleotides) represent an ideal type of molecular Lego trolling the overall process.
for the self-assembly of nanocomponents into more Active microelectronic arrays have been developed
complex two- and three-dimensional higher order struc- for a number of applications in bioresearch and DNA
tures. Initially, DNA sequences can be used as a kind clinical diagnostics [13.8–17]. These active microarrays
of template for assembly on solid surfaces. The tech- are able to produce electric fields on the array surface
nique involves taking complementary DNA sequences that allow charged reagent molecules (DNA, RNA, pro-
and using them as a kind of selective glue to bind other teins, enzymes), nanostructures, cells and microscale
DNA-modified macromolecules or nanostructures to- structures to be transported to any of the microscopic
gether. The base pairing property of DNA allows one sites on the device surface. When DNA hybridization
single strand of DNA with a unique base sequence is carried out on the microarray, the device allows
to recognize and bind together with its complemen- electric fields to direct the self-assembly of the DNA
tary DNA strand to form a stable double-stranded DNA hybrid at the test site. In principle, these active microar-
structures. While high-fidelity recognition molecules ray devices can serve as motherboards or hostboards
like DNA allow one to self-assemble higher order to assist in the self-assembly of DNA molecules, as
structures, the process has some significant limitations. well as other moieties such as nanostructures or even
First, for in vitro applications (in a test tube), DNA microscale components [13.18–25]. Active microarray
and other high-fidelity recognition molecules like an- electric field assembly is thus a type of pick and place
tibodies, streptavidins, and lectins work most efficiency process that has the potential to be used for hetero-
when the complexity of the system is relatively low. In geneous integration and nanofabrication of molecular
other words, as the complexity of the system increases and nanoscale components into higher order materials,
(more unrelated DNA sequences, proteins, and other structures and devices [13.24].
a)
Part B 13.1
b) c)
Part B 13.1
Fig. 13.3 Photograph of the 400-site CMOS ACV400-chip
array. Four counter-electrodes, two positioned longitudi-
20 kV X300 50 µm 26/Dec/03
nally and two horizontally, surround the active working
electrode array Fig. 13.4 Close-up of the current 400-site chip array. The
CMOS chip has an array of 16 × 25 (400) sites; each elec-
communication, memory, temperature sensing, volt- trode is 50 μm in diameter with a 150 μm center-to-center
age/current sourcing and measuring circuits on-chip. distance
After standard CMOS fabrication within the underlying
silicon, thin film deposition and patterning techniques by 7 mm in size. Figure 13.4 shows a close up of
were used to fabricate the platinum microelectrodes the 400 site microarray with several of the 50 μm
on the surface of the standard CMOS chip [13.30]. diameter platinum microelectrodes. Underneath each of
The CMOS microelectrode array chip developed for the 400 microelectrodes is an analog sample and hold
the Nanochip 400 system consists of a 16 by 25 ar- circuit which maintains a predefined voltage on the
ray of 50 μm diameter microelectrodes spaced 150 μm electrode. A digital to analog converter sequentially in-
center to center. Figure 13.3 shows the 400 test site terrogates the digitally stored bias value for each of the
CMOS microelectronic array device which is only 5 microelectrodes and refreshes each sample and hold cir-
Analog to Tempe-
Array digital rature
Test converter sensor
port
V to Control
array to array
Digital to Control
Trans-
analog
ceiver
converter 8 8 Serial
Dual i/o
port Logic block
ram
Modulo 9 9
402 EEPROM
counter
Control
Fig. 13.5 Block diagram of the 400 CMOS chip control circuitry
394 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 13.1
Fluidic
chamber
cuit accordingly. A separate loop sequentially measures +5 V and ground for the analog circuits, digital sig-
the voltage and current at each of the microelectrodes. nal in, digital signal out, clock signal, reset, and two
Microelectrodes can be operated at a fixed voltage, or terminals for an external current sampling resistor. For
by means of a feed back loop, at constant current or structural support, 76 flip-chip solder bonds are used be-
at a fixed voltage offset from a reference electrode. tween the chip and a ceramic substrate which doubles
The microchip also contains a p-n junction temperature as both a fluidic chamber and as an electrical contact
sensor and EEPROM memory to store thermal calibra- interface (Fig. 13.6). In order prevent any compromise
tion coefficients, serial numbers, and assay-related data. in the quality of either the circuitry or the microelec-
Figure 13.5 shows the Nanochip 400 CMOS circuitry trode array, the CMOS is fabricated at one foundry
block diagram. The chip has only 12 external electrical while the platinum microelectrode array is fabricated
connections, +5 V and ground for the digital circuits, at a second foundry. The CMOS process requires 16
masking step followed by an additional three masking
steps for the platinum electrodes. Figure 13.7 shows
Dielectric via profile with Overlay oxide over metals
overlay oxide on top (PECVD oxide) a high-magnification cross section of an individual mi-
croelectrode and the underlying CMOS circuitry. After
the finished wafers have been inspected for defects, the
wafers are diced into individual chips and then flip-chip
Ti/Pt metal
bonded onto the ceramic substrates. The flip-chip on
substrates (FCOS) are thermally calibrated and then fi-
nally assembled into the NanoChip 400 plastic cartridge
housing which provides the fluidic delivery system and
outside electrical connections (Fig. 13.2, inset).
Oxidation H2 O → 2H+ + 12 O2 + 2 e− ,
(13.1)
Electronic Electrode Reduction 2 e− + 2H2 O → 2OH− + H2 . (13.2)
array site
Part B 13.1
energized Important to the operation of the device is a thin 1–2 μm
hydrogel permeation layer which covers the platinum
microelectrode surface. This hydrogel permeation layer
Charged is designed to protect the more sensitive DNA and other
biomolecules biomolecules from the electrolysis reactions and prod-
Negatively ucts that occur on the platinum microarray surface. The
charged permeation layer is usually made of either agarose or
DNA
polyacrylamide. The layer also contains streptavidin
Analyte which facilitates the capture and binding of biotinylated
molecules DNA probes or DNA target molecules onto the hydrogel
concentrate surface.
The electric field microarray technology also takes
Fig. 13.8 Diagram of negatively charged DNA molecules advantage of the H+ ions (low pH) generated at the
being transported to a positively charged microelectrode positive electrode to carry out a unique electronic hy-
bridization process. Electronic hybridization is carried
on the microarray surface. Depending on the charge out under low salt and low conductance conditions
of the molecule, they will be rapidly transported and it allows denatured DNA molecules to be hy-
to the oppositely biased microelectrode. Figure 13.8 bridized only in the microscopic area around the
shows a diagram of negatively charged DNA molecules activated microelectrode test site. A second electronic
being transported to a positively charged microelec- process called electronic stringency is achieved by re-
trode. The electronic addressing of DNA and other versing the polarity at the microelectrode (negative
biomolecules onto the microarray test sites can ac- activation) causing nonspecifically bound (negatively
celerate hybridization and other molecular binding charged) DNA molecules to be driven away from the
processes by up to 1000 times compared to tradi- test site. Electronic stringency can also aid in the dif-
tional passive methods (such as those of Affymetrix ferentiation of DNA binding strengths, allowing better
and GeneChip) (Table 13.2). By way of example, hy- single base discrimination for the determination of sin-
bridization on a passive microarray may take up to gle nucleotide polymorphisms (SNPs).
several hours for the low concentrations of target DNA
frequently found in many clinical samples. During the 13.1.3 Electronic DNA Hybridization
operation of the electronic microarray, the bias potential and Assay Design
is sufficiently high (> 1.2 V) to cause the electroly-
sis of water. Oxidation occurs on the positively biased The electronic hybridization technology and the
microelectrode and reduction occurs on the negatively Nanochip 400 System provide the researcher with
Table 13.2 Comparison of electronic hybridization speed with that of conventional passive hybridization on microarrays
Hybridization Concentration Concentration factor Stringency control
time of targets at a site
NanoChip 10–100 s Directed > 1000 times Electronic thermal
and localized chemical
Active hybridization At the array sites;
ability to control
individual sites
Passive hybridization 1–2 h Undirected; Low, diffusion- Thermal chemical
technologies sites cannot be con- dependent
trolled independently
396 Part B MEMS/NEMS and BioMEMS/NEMS
a completely flexible and open platform for perform- 1. Determination of multiple genes in one sample ad-
ing DNA hybridizations. It offers users the option to dressed on the chip
either generate or create user-defined microarrays spe- 2. Determination of multiple samples with one gene of
cific for their own targets using electronic addressing interest addressed on the chip
of the biotinylated probes on the array, or to perform 3. Determination of multiple samples with multiple
target- or single-base-specific analysis of DNA or RNA genes of interest addressed on the chip
molecules. Both processes are performed very rapidly 4. Single site multiplexing where several targets are
Part B 13.1
because the electronic addressing enables hybridization discriminated on the same site using different fluor-
of the target DNA to occur within 30 to 60 s, as com- escent probes.
pared to competitive microarray technologies where
hybridization is performed within one to several hours. 13.1.4 DNA Genotyping Applications
Electronic microarrays provide an open platform and
offer flexibility in the assay design [13.31–36]. This in- Electronic hybridization technology has been developed
cludes either the PCR amplicon down format to screen and commercialized and is now used for a number
one or more patients for one or more SNPs, or the cap- of practical applications including clinical diagnos-
ture probe down format to screen many patients for tic DNA genotyping [13.31–40]. In all applications,
one or more SNPs. In addition, sandwich-type assays the DNA probe detection step is accomplished with
are easy to perform where oligonucleotide discrimina- fluorophore reporters and laser-based fluorescence de-
tors and/or fluorescent probe labeled oligonucleotides tection. In addition, due to its open platform character,
are hybridized, electronically or passively, to captured Nanogen’s users have developed about 200 additional
PCR amplicons or probes. These assay design tools en- assays using the electronic microarray system and tech-
able the users to increase the discrimination at single- nology. The DNA assay areas developed by the users
base resolution as well as to minimize the nonspe- include diagnostics related to the following diseases and
cific binding. Multiple probes or discriminators can applications: coronary artery disease, cardiovascular
be attached at a single site which facilitates the de- disease, hypertension, cardiac function, venous throm-
tection of multiple targets on a single electrode site. botic disease, metabolism, drug metabolism/cancer,
This enables the worker to use over 1000 characteris- cancer, cytokine, transcription factor, bacterial ID, Rett
tic genes or single nucleotide polymorphisms (SNPs) syndrome, thrombophilia, thalassemia and deafness.
on a single array. In multiple DNA target detection
the user can use blocker nucleotides that block SNPs 13.1.5 On-Chip Strand Displacement
which are not reported at a particular location. This Amplification
allows the same universal reporter and specific dis-
criminators to be used to report one or more SNPs on It has also been demonstrated that a complex DNA
another site. Two lasers are used to recognize fluor- amplification technique can be performed on separate
escent signal from two reporter dyes (green and red). test sites of the microarray chip. This approach sig-
All of the steps are extremely fast, so the users can nificantly reduces the time for DNA analysis because
design and generate their own assays and incorporate it incorporates DNA amplification and detection into
the preparation of the array with oligonucleotides spe- a single platform. The assays were demonstrated us-
cific to multiple targets and samples as a part of the ing an isothermal strand displacement amplification
assay. (SDA is licensed technology from Becton, Dickinson
Other flexible electronic hybridization format de- and Co., Franklin Lakes, USA). In the SDA amplifi-
signs allow several types of multiplexed DNA analyses cation, DNA polymerase recognizes the nicked strand
to be carried out, such as the determination of multiple of DNA and initiates resynthesis of that strand, dis-
genes in one sample, multiple samples with one gene, placing the original strand. The released amplicons
or multiple samples with multiple genes. The ability then travel in solution to primers of the complemen-
to control individual test sites permits genetically un- tary strand which are either in solution or anchored on
related DNA molecules to be used simultaneously on the test site. Oligonucleotide primers without nicking
the same microchip. In contrast, sites on a conventional sites, called bumper primers, are synthesized in the re-
DNA array cannot be controlled separately, and all pro- gions flanking the amplicons that were just produced,
cess steps must be performed on an entire array. Types and assist in strand displacement and initial template
of multiplexed analysis include: replication [13.41].
Next-Generation DNA Hybridization and Self-Assembly Nanofabrication Devices 13.2 Electric Field-Assisted Nanofabrication 397
13.1.6 Cell Separation ria bacterial cells (≈ 1 μm) from whole blood cells
on Microelectronic Arrays (≈ 10 μm) in a highly parallel manner. At an ac fre-
quency of about 10 kHz the Listeria bacterial cells can
Microelectronic arrays have also used been for cell be positioned on specific microlocations at high-field
separation applications. Disease diagnostics frequently regions and the blood cells can be positioned in the
involve identifying a small number of specific bacte- low-field regions between the microelectrodes. The rel-
ria or viruses in a blood sample (infectious disease), ative positioning of the cells between the high- and
Part B 13.2
fetal cells in maternal blood (genetic diseases) or tu- low-field regions is based on dielectric differences be-
mor cells among a background of normal cells (early tween the cell types. While maintaining the ac field,
cancer detection). One powerful electric field tech- the microarray can be washed with a buffer solution
nique used for cell separation is called dielectrophoresis that removes the blood cells (low-field regions) from
(DEP). The DEP process involves the application of the more firmly bound bacteria (high-field regions)
an asymmetric alternating current (AC) electric field near the microelectrodes. The bacteria can then be re-
to the cell population. Active microelectronic arrays leased and collected or electronically lysed to release
have been used to achieve the separation of bacte- the genomic DNA or RNA for further manipulation and
ria from whole blood [13.42], for the separation of analysis [13.42]. DEP represents a particularly useful
cervical carcinoma cells from blood [13.43], and for process that allows difficult cell separation applications
gene expression analysis [13.44]. Microelectronic ar- to be carried out rapidly and with high selectivity. The
ray devices utilizing high frequency ac fields have DEP process may also be useful for nanofabrication
been used to carry out the DEP separation of Liste- purposes [13.23, 24].
parallel electric field pick and place process for the 500 nm
heterogeneous integration of molecular, nanoscale and particles
microscale components into complex three-dimensional
structures. If desired, this process can be used to as-
semble molecules and/or nanocomponents within the
defined perimeters of larger silicon or other semicon-
5 µm
ductor structures. Electric field-assisted self-assembly particles
technology is based on three key physical principles:
1. The use of functionalized DNA or other high-
fidelity recognition components as molecular Lego
blocks for nanofabrication
2. The use of DNA or other high-fidelity recogni-
tion components as a selective glue that provides Fig. 13.10 Parallel electronic addressing of two different
intrinsic self-assembly properties to other molecu- types of microspheres to the microelectronic array test sites
lar, nanoscale or microscale components (metallic
nanoparticles, quantum dots, carbon nanotubes, or-
ganic molecular electronic switches, micrometer
and submicrometer silicon lift-off devices and com-
ponents)
3. The use of active microelectronic array devices 500 µm
to provide electric field assistance or control of particles
the intrinsic self-assembly of any modified elec-
tronic/photonic components and structures [13.18–
1 µm
24]. particles
Microelectronic arrays have been used to direct the
binding of derivatized nanospheres and microspheres
onto selected locations on the microarray surface. In
this case, fluorescent and nonfluorescent polystyrene 80 µm
nanospheres and microspheres derivatized with spe-
cific DNA oligonucleotides are transported and bound Fig. 13.11 Electronic addressing of two different layers of
to selected test sites or microlocations derivatized microspheres to the microelectronic array test sites
Next-Generation DNA Hybridization and Self-Assembly Nanofabrication Devices 13.3 Conclusions 399
Part B 13.3
B B
and addressing to form multiple layers of particles
and other materials, allowing fabrication in the third c) W
dimension.
Present nanofabrication methods do not allow most
nanostructures to be modified in a controlled or precise
manner. For example, it would be extremely difficult X
to attach different DNA sequences or different kinds of
protein molecules in precise locations around quantum Z
dots or other nanoparticles (Fig. 13.12a,b). Unfortu-
nately, without this first-order property it becomes
even more difficult to then assemble these nanostruc-
tures into higher order heterogeneous 3-D structures, Y
even though the core structure is derivatized with Fig. 13.12 (a,b) Precision nanosphere functionalization
high-fidelity recognition components (Fig. 13.12c). Mi- scheme. (c) Type of heterogeneous 3-D higher order
croelectronic array devices may offer the opportunity structure that can only be obtained using precision nano-
to develop processes that will allow core nanostructures structures
to be selectively modified in a precise fashion [13.24].
The proposed electric field microarray techniques may the nanostructures, it will be possible to functional-
provide the ability to carry out the precision function- ize the core structure selectively with most biological
alization of nanostructures by processes which involve and/or chemical groups. Such devices and processes
transporting and orienting the nanostructures onto sur- allow one to design and create functionalized nano-
faces containing the selected ligand molecules which structures with binding groups arranged in tetrahedral,
are then reacted only with a selected portion of the hexagonal or other coordinate positions around the core
nanostructure. By repeating the process and reorienting nanostructure.
13.3 Conclusions
Active microelectronic array technology provides a probe loading station and fluorescent detection sys-
a number of advantages for carrying out DNA hy- tem. Active microelectronic arrays are fundamentally
bridization diagnostics and other affinity-based assays different from other DNA chip or microarray devices,
for molecular biology research and clinical diagnostic which are essentially passive. Active microelectronic
applications. The technology has also demonstrated the arrays allow DNA molecules, RNA, oligonucleotide
potential for assisted self-assembly and other nanofab- probes, PCR amplicons, proteins, nanostructures, cells
rication applications. Microelectronic arrays have been and even microscale devices to be rapidly transported
designed and fabricated with 25 to 10 000 microscopic and selectively addressed to any of the test sites on
test sites, and devices with 100 and 400 test sites have the microelectronic array surface. Active microarray de-
been commercialized. The newer 400-test site devices vices have considerable potential for nanofabrication
have CMOS elements incorporated into the underlying by directed self-assembly of molecular, nanoscale and
silicon structure that provide on-board control of cur- microscale components into higher order mechanisms,
rent and voltage to each of the test sites on the device. structures, and devices. This electric field technol-
Microelectronic chips are incorporated into a cartridge- ogy makes possible a type of pick and place process
type device so that they can be conveniently used with for the heterogeneous integration of diverse molecular
400 Part B MEMS/NEMS and BioMEMS/NEMS
and nanoscale components into higher order structures the inherent hierarchical logic of allowing one to con-
within defined perimeters of larger silicon or semi- trol the organization and assembly of components from
conductor structures. The technology provides the best the molecular level to the nanoscale level to microscale
aspects of a top-down and bottom-up process, and has three-dimensional integrated structures and devices.
References
Part B 13
13.1 National Research Council: Small Wonders, End- electrophoretic fields for multiplex DNA hybridiza-
less Frontiers: Review of the Nanotional Nano- tion and genomic applications, Electrophoresis 21,
technology Initiative (National Research Council, 157–164 (2000)
Washington 2002) 13.14 C. Gurtner, E. Tu, N. Jamshidi, R. Haigis, T. Onofrey,
13.2 M.P. Hughes (Ed.): Nanoelectromechanics in En- C.F. Edman, R. Sosnowski, B. Wallace, M.J. Heller:
gineering and Biology (CRC, Boca Raton 2003) Microelectronic array devices and techniques for
13.3 W.A. Goddard, D. Brenner, S. Lyshevski, G. Lafrate electric field enhanced DNA hybridization in low-
(Eds.): Handbook of Nanoscience, Engineering and conductance buffers, Electrophoresis 23, 1543–1550
Technology (CRC, Boca Raton 2003) (2002)
13.4 V. Balzani, M. Venturi, A. Credi (Eds.): Molecular De- 13.15 M.J. Heller: DNA microarray technology: devices,
vices and Mechanics – Journey into the Nanoworld systems and applications, Ann. Rev. Biomed. Eng.
(Wiley-VCH, Weinheim 2003) 4, 129–153 (2002)
13.5 R. Bashir: Biologically mediated assembly of artifi- 13.16 M.J. Heller, E. Tu, R. Martinsons, R.R. Anderson,
cial nanostructures and microstructures. In: Hand- C. Gurtner, A. Forster, R. Sosnowski: Active micro-
book of Nanoscience, Engineering and Technology, electronic array systems for DNA hybridization,
ed. by W.A. Goddard, D. Brenner, S. Lyshevski, genotyping, pharmacogenomics and nanofabrica-
G. Lafrate (CRC, Boca Raton 2003) pp. 15–1–15–31, tion applications. In: Integrated Microfabricated
Chap. 15 Devices, ed. by M.J. Heller, A. Guttman (Marcel
13.6 M.J. Heller, R.H. Tullis: Self-organizing molecular Dekker, New York 2002) pp. 223–270, Chap. 10
photonic structures based on functionalized syn- 13.17 S.K. Kassengne, H. Reese, D. Hodko, J.M. Yang,
thetic DNA polymers, Nanotechnology 2, 165–171 K. Sarkar, P. Swanson, D.E. Raymond, M.J. Heller,
(1991) M.J. Madou: Numerical modeling of transport
13.7 D.M. Hartmann, D. Schwartz, G. Tu, M. Hellerand, and accumulation of DNA on electronically active
S.C. Esener: Selective DNA attachment of particles biochips, Sens. Actuators B 94, 81–98 (2003)
to substrates, J. Mater. Res. 17, 473–478 (2002) 13.18 S.C. Esener, D. Hartmann, M.J. Heller, J.M. Cable:
13.8 M.J. Heller: An active microelectronics device for DNA assisted micro-assembly: A heterogeneous in-
multiplex DNA analysis, IEEE Eng. Med. Biol. 15, tegration technology for optoelectronics, Proc. SPIE
100–103 (1996) 70, 113–140 (1998)
13.9 R.G. Sosnowski, E. Tu, W.F. Butler, J.P. O’Connell, 13.19 C. Gurtner, C.F. Edman, R.E. Formosa, M.J. Heller:
M.J. Heller: Rapid determination of single base Photoelectrophoretic transport and hybridization
mismatch in DNA hybrids by direct electric field of DNA on unpatterned silicon substrates, J. Am.
control, Proc. Natl. Acad. Sci. USA 94, 1119–1123 Chem. Soc. 122(36), 8589–8594 (2000)
(1997) 13.20 Y. Huang, K.L. Ewalt, M. Tirado, R. Haigis, A. Forster,
13.10 C.F. Edman, D.E. Raymond, D.J. Wu, E. Tu, R.G. Sos- D. Ackley, M.J. Heller, J.P. O’Connell, M. Krihak:
nowski, W.F. Butler, M. Nerenberg, M.J. Heller: Electric manipulation of bioparticles and macro-
Electric field directed nucleic acid hybridization on molecules on microfabricated electrodes, Anal.
microchips, Nucl. Acids Res. 25, 4907–4914 (1997) Chem. 73, 1549–1559 (2001)
13.11 M.J. Heller: An integrated microelectronic hy- 13.21 C.F. Edman, C. Gurtner, R.E. Formosa, J.J. Coleman,
bridization system for genomic research and M.J. Heller: Electric-field-directed pick-and-place
diagnostic applications. In: Micro Total Analysis assembly, HDI 3(10), 30–35 (2000)
Systems, ed. by D.J. Harrison, A. van den Berg 13.22 C.F. Edman, R.B. Swint, C. Gurthner, R.E. For-
(Kluwer Academic, Dordrecht 1998) pp. 221–224 mosa, S.D. Roh, K.E. Lee, P.D. Swanson, D.E. Ackley,
13.12 M.J. Heller, E. Tu, A. Holmsen, R.G. Sosnowski, J.J. Colman, M.J. Heller: Electric field directed as-
J.P. O’Connell: Active microelectronic arrays for DNA sembly of an InGaAs LED onto silicon circuitry, IEEE
hybridization analysis. In: DNA Microarrays: A Prac- Photon. Tech. Lett. 12(9), 1198–1200 (2000)
tical Approach, ed. by M. Schena (Univ. Press, 13.23 C.F. Edman, M.J. Heller, R. Formosa, C. Gurtner:
Oxford 1999) pp. 167–185 Methods and apparatus for the electronic homo-
13.13 M.J. Heller, A.H. Forster, E. Tu: Active micro- geneous assembly and fabrication of devices, US
electronic chip devices which utilize controlled Patent 6569382 (2003)
Next-Generation DNA Hybridization and Self-Assembly Nanofabrication Devices References 401
13.24 M.J. Heller, J.M. Cable, S.C. Esener: Methods for the 13.36 W.A. Thistlethwaite, L.M. Moses, K.C. Hoffbuhr,
electronic assembly and fabrication of devices, US J.M. Devaney, E.P. Hoffman: Rapid genotyping of
Patent 6652808 (2003) common MeCP2 mutations with an electronic DNA
13.25 C.F. Edman, M.J. Heller, C. Gurtner, R. Formosa: microchip using serial differential hybridization,
Systems and devices for the photoelectrophoretic J. Mol. Diagn. 5(2), 121–126 (2003)
transport and hybridization of oligonucleotides, US 13.37 V.R. Mas, R.A. Fisher, D.G. Maluf, D.S. Wilkinson,
Patent 6706473 (2004) T.G. Carleton, A. Ferreira-Gonzalez: Hepatic artery
13.26 A. Taton, C. Mirkin, R. Letsinger: Scanometric DNA thrombosis after liver transplantation and ge-
Part B 13
array detection with nanoparticle probes, Science netic factors: Prothrombin G20210A polymorphism,
289, 1757–1760 (2000) Transplantation 76(1), 247–249 (2003)
13.27 M. Chee, R. Yang, E. Hubbell, A. Berno, X. Huang, 13.38 R. Santacroce, A. Ratti, F. Caroli, B. Foglieni,
D. Stern, J. Winkler, D. Lockhart, M. Morris, S. Fodor: A. Ferraris, L. Cremonesi, M. Margaglione, M. Seri,
Accessing genetic information with high-density R. Ravazzolo, G. Restagno, B. Dallapiccola,
DNA arrays, Science 274, 610–614 (1996) E. Rappaport, E.S. Pollak, S. Surrey, M. Ferrari,
13.28 A. Pease, D. Solas, E. Sullivan, M. Cronin, C. Holmes, P. Fortina: Analysis of clinically relevant single-
S. Fodor: Light-generated oligonucleotide arrays nucleotide polymorphisms by use of microelectric
for rapid DNA sequence analysis, Proc. Natl. Acad. array technology, Clin. Chem. 48(12), 2124–2130
Sci. USA 99, 5022–5026 (1994) (2002)
13.29 R.J. Lipshutz, D. Morris, M. Chee, E. Hubbell, 13.39 A. Åsberg, K. Thorstensen, K. Hveem, K. Bjerve:
M.J. Kozal, N. Shah, N. Shen, R. Yang, S.P. Fodor: Hereditary hemochromatosis: The clinical signif-
Using oligonucleotide probe arrays to access ge- icance of the S64C mutation, Genet. Test. 6(1),
netic diversity, Biotechniques 19(3), 442–447 (1995) 59–62 (2002)
13.30 P. Swanson, R. Gelbart, E. Atlas, L. Yang, T. Gro- 13.40 J.G. Evans, C. Lee-Tataseo: Determination of the
gan, W.F. Butler, D.E. Ackley, E. Sheldon: A fully factor V Leiden single-nucleotide polymorphism in
multiplexed CMOS biochip for DNA analysis, Sens. a commercial clinical laboratory by use of NanoChip
Actuators B 64, 22–30 (2000) microelectric array technology, Clin. Chem. 48(9),
13.31 P.N. Gilles, D.J. Wu, C.B. Foster, P.J. Dillion, 1406–1411 (2002)
S.J. Channock: Single nucleotide polymorphic dis- 13.41 T. Walker, J. Nadeau, P. Spears, J. Schram, C. Nycz,
crimination by an electronic dot blot assay on D. Shank: Multiplex strand displacement amplifi-
semiconductor microchips, Nat. Biotechnol. 17(4), cation (SDA) and detection of DNA sequences from
365–370 (1999) Mycobacterium tuberculosis and other mycobacte-
13.32 N. Narasimhan, D. O’Kane: Validation of SNP geno- ria, Nucl. Acids Res. 22(13), 2670–2677 (1994)
typing for human serum paraoxonase gene, Clin. 13.42 J. Cheng, E.L. Sheldon, L. Wu, A. Uribe, L.O. Gerrue,
Chem. 34(7), 589–592 (2001) J. Carrino, M.J. Heller, J.P. O’Connell: Electric field
13.33 R. Sosnowski, M.J. Heller, E. Tu, A. Forster, controlled preparation and hybridization analy-
R. Radtkey: Active microelectronic array system for sis of DNA/RNA from E. coli on microfabricated
DNA hybridization, genotyping and pharmacoge- bioelectronic chips, Nat. Biotechnol. 16, 541–546
nomic applications, Psychiatr. Genet. 12, 181–192 (1998)
(2002) 13.43 J. Cheng, E.L. Sheldon, L. Wu, M.J. Heller,
13.34 Y.R. Sohni, J.R. Cerhan, D.J. O’Kane: Microarray and J. O’Connell: Isolation of cultured cervical carci-
microfluidic methodology for genotyping cytokine noma cells mixed with peripheral blood cells on
gene polymorphisms, Hum. Immunol. 64, 990–997 a bioelectronic chip, Anal. Chem. 70, 2321–2326
(2003) (1998)
13.35 E.S. Pollak, L. Feng, H. Ahadian, P. Fortina: 13.44 Y. Huang, J. Sunghae, M. Duhon, M.J. Heller,
Microarray-based genetic analysis for studying B. Wallace, X. Xu: Dielectrophoretic separation and
susceptibility to arterial and venous thrombotic gene expression profiling on microelectronic chip
disorders, Ital. Heart J. 2, 569–572 (2001) arrays, Anal. Chem. 74, 3362–3371 (2002)
403
Single-Walle
14. Single-Walled Carbon Nanotube Sensor Concepts
Part B 14
single-walled carbon nanotubes (SWNTs); virtually 14.2 Fabrication of SWNT Sensors .................. 412
any property of SWNTs (e.g., electronic, electri- 14.2.1 Methods for SWNT Production ........ 412
cal, mechanical, and optical) can result or has 14.2.2 Strategies for SWNT Assembly
already resulted in sensor concept demonstra- into Devices................................. 413
tors. The basic questions that this chapter will 14.3 Example
attempt to address are: why use SWNTs, and how State-of-the-Art Applications................ 416
can SWNTs be used in sensing applications? A tour 14.3.1 Chemical and Biochemical Sensors . 416
through the gallery of basic nanotube properties is 14.3.2 Piezoresistive Sensors ................... 418
used to reveal the richness and uniqueness of this 14.3.3 Resonant Sensors ......................... 420
material’s intrinsic properties. Together with ex- 14.4 Concluding Remarks ............................. 421
amples from the literature showing performance
of SWNT-based sensors at least comparable to (and References .................................................. 421
sometimes surpassing) that of state-of-the-art
micro- or macrodevices, these nanotube proper-
ties should explain why so much effort is currently a brief survey of SWNT properties useful for sens-
being invested in this field. Because nanotubes, ing. The CNFET is introduced in Sect. 14.1.2 as
like any other nanoobject, are not easy to probe, a platform enabling access to individual SWNT
a versatile strategy for accessing their properties, properties during the sensing process. The cur-
via the carbon nanotube field-effect transistor rent status of CNFET-based sensor characterization
(CNFET) concept, will be described in this chapter. is captured in Sect. 14.1.3. Methods for fabricat-
Fabricating CNFET devices, together with exam- ing, or supporting the fabrication of, SWNT FETs
ples of SWNT sensor demonstrators utilizing the are reviewed in Sect. 14.2. Finally, Sect. 14.3 will
CNFET principle, will outline a proposal for how be devoted to examples of CNT-based sensors,
nanotubes can be utilized in sensors. encompassing three main case studies, namely
In Sect. 14.1 design considerations for SWNT (bio)chemical, piezoresistive, and resonator
sensors are brought into attention, starting with sensors.
Sensors are only one possible application of SWNTs. tor manufacturers and academic institutions, to have
Other notable applications include field-emission de- greater potential for post-complementary metal–oxide–
vices, energy storage, composites, and nanoelectron- semiconductor (CMOS) device concepts than any other
ics [14.1]. For example, in nanoelectronics, CNTs on the horizon (e.g., molecular electronic devices, fer-
have been assessed by the International Technology romagnetic logic devices, and spin transistors). For
Roadmap for Semiconductors 2007 (ITRS), edited sensing devices, carbon nanotubes present several key
by a group of scientists from all major semiconduc- advantages, including:
404 Part B MEMS/NEMS and BioMEMS/NEMS
1. Nanometer feature size: useful for building highly mism worldwide. Many of these investigations have
localized sensing units (active spots) and for large- been published in prestigious research journals. There
scale integration (sensor arrays) are, however, still some challenges to overcome before
2. High sensitivity to stimuli: from their unique struc- broader acceptance in industrial product development
tural and electronic properties (Sect. 14.1.1) and activities will be observed. In fact, carbon nanotubes
high surface-to-volume ratio are today at the crossroads between basic science and
3. Low power consumption: whether operating as tran- engineering. CNT device demonstrators and theoretical
sistors or mechanical resonators, the power used to extrapolations surpass in performance state-of-the-art
excite or probe a SWNT is on the order of 10 nW. devices, more than motivating any future attempts to
solve the remaining issues. Based on the steep evolu-
Because of these and other advantages, at rela- tion slope experienced so far, it may not take long until
tively short time after their discovery, SWNTs have carbon-nanotube-based sensors will appear on device
resulted in sensor device demonstrators fueling opti- and product roadmaps.
Part B 14.1
block available for engineering sensing devices. Fur- this book. In the following, we condense the most rele-
thermore, most of the knowledge gained in studying the vant SWNT properties, supported by examples of how
rich variety of SWNT FET sensors is transferable to these have been employed in sensor devices.
more complex devices involving MWNTs, bundles or
networks. SWNTs as Nanometer-Thin Semiconducting
In the next subsection, some of the most important or Metallic Wires
SWNT properties and property modulation mechanisms Structurally, SWNTs are molecular cylinders with
useful for sensing are listed and exemplified. The op- monoatomic-thick walls, resembling a honeycomb lat-
eration and particularities of CNFETs are reviewed in tice of carbon atoms (graphene) rolled into a tube
Sect. 14.1.2, whereas in Sect. 14.1.3 the current status (Fig. 14.1). The structure of a SWNT is uniquely iden-
of sensor characterization is briefly discussed. tified by its chiral indices (n, m) that give the so-called
chiral vector C h = na1 + ma2 determining the circum-
14.1.1 CNT Properties for Sensing ference of the nanotube upon rolling [a1,2 are the
Part B 14.1
two-dimensional (2-D) graphene lattice vectors as in
SWNTs have many interesting properties for nanode- Fig. 14.1a]. The (n, m) indices can be interchanged
vices in general, and nanosensors in particular. The with (dt , θ), where
√ dt is the nanotube diameter given
properties of SWNTs are discussed in many text- by dt ≈ (a/π) n 2 + m 2 + nm [nm] √ and θ is the chi-
books [14.22–24] and are also surveyed in Chap. 3 of ral angle defined by tan θ = 3m/(2n + m) (where
a) b) E (eV)
8
4
a1
a2 0
–4
θ
Ch = na1 + ma2
–8
–π/T 0 +π/T DOS (arb. units)
(8,4)
k
c) E (eV) d) E (eV)
8 8
4 4
0 0
–4 –4
–8 –8
–π/T 0 +π/T DOS (arb. units) –π/T 0 +π/T DOS (arb. units)
(11,0) (6,6)
k k
Fig. 14.1a–d Structural and electronic properties of SWNTs. (a) The unfolded unit cell of a chiral (8,4) CNT show-
ing chiral vector C h , chiral angle θ, and 2-D graphene lattice vectors a1,2 (in inset). (b) The folded (8,4) structure,
electronic band structure, and density of states (DOS). Bands are obtained via the zone-folding procedure applied to
tight-binding dispersion relations [14.21]. (c,d) Band structure and DOS for zigzag (11,0) and armchair (6,6) nanotubes,
respectively
406 Part B MEMS/NEMS and BioMEMS/NEMS
SWNTs), the rest ( p = ±1) being semiconducting (two- beam and a resonance in the frequency response spec-
thirds of all SWNTs). Structure, bands, and densities trum corresponding to the fundamental bending mode ω0 .
of states for three selected SWNTs (a chiral, a zigzag, (b) With applied tension the resonance frequency ω shifts
and an armchair tube) are displayed in Fig. 14.1b–d. As upwards, whereas (c) with attached particles (mass load-
for any one-dimensional (1-D) structure, the density of ing) ω shifts downwards
states of SWNTs is singular at energies corresponding
to subband extrema (van Hove singularities). A rough be treated as a continuum elastic beam, the resonance
estimation for√the electronic band gap of s-SWNTs is frequency of the fundamental flexural mode is given
E g ≈ (2at0 )/( 3dt ) [eV] [14.21] (where t0 = 2.6 eV is by [14.33]
the so-called hopping tight-binding parameter for π or-
bitals). More accurate electronic structure calculations, 4π 2 E I L 2σ
ω= 2 1+ 2 , (14.1)
taking into account the surface curvature of nanotubes, L 3ρ A 4π E I
revealed that in fact only armchair (n = m) SWNTs are where L is the tube length, I is the moment of inertia, A
truly metallic, whereas other tubes with p = 0 actually is the cross-sectional area, ρ is the mass density (ρL =
have a small bandgap E g ≈ 40/dt2 [meV] [14.26]; these ρ A), and σ is the initial tension in the beam.
tubes are labeled small-gap semiconducting (SGS)- The dependence of ω on tension σ has been uti-
SWNTs. lized by Sazonova et al. [14.16] and others [14.17, 18]
The mentioned structural and electronic properties to demonstrate tunable CNT resonators. These devices
have resulted in a few SWNT sensor concepts. For ex- can thus be employed as sensitive strain/stress sensors.
ample, FETs based on s-SWNTs have been utilized as On the other hand, the attachment of a small mass to
charge detectors in flow meters [14.27] or (bio)chemical the tube increases the effective mass of the beam and
sensors [14.2, 7] (see Sect. 14.3.1 for more details). The as a result downshifts ω (via the ρ term in (14.1)). This
high curvature (dt ≈ 1 nm) of carbon nanotubes (both principle, known as resonant inertial balance, has been
tips and bodies) has been exploited, for example, in utilized [14.14,15,17] to detect minute mass loading on
gas ionization sensors [14.28] and capacitance gas sen- a CNT. Sensing mechanisms for strain sensors and iner-
sors [14.29]. tial balances are sketched in Fig. 14.2, and are discussed
in more detail in Sect. 14.3.3.
SWNTs as Diamond-Stiff, Ultralight Strings
The basic inertial and mechanical properties of Mechanically Tunable Electronic
SWNTs are: linear mass density ρL = 2.33dt [zg/nm] Properties of SWNTs
(1 zg = 10−21 g, and dt is the tube diameter in nm, In analogy to crystalline semiconductors such as silicon,
as given above), Young’s modulus E in the range of straining carbon nanotubes modifies their electronic
≈ 1.25 TPa [14.30], maximum tensile strain of 6%, and bandgap. With an extended version of the model
strength of ≈ 45 GPa [14.31, 32]. These properties pro- that was used to obtain the bands in Fig. 14.1, Yang
mote carbon nanotubes as ideal nanosized beams for et al. [14.34] calculated the modulation of bandgaps E g
mechanical sensors. For example, consider a straight, of SWNTs of different chiralities under tensile (axial)
doubly clamped SWNT. Assuming that the SWNT can and shear (torsional) strain (Fig. 14.3c,d). A linear ap-
Single-Walled Carbon Nanotube Sensor Concepts 14.1 Design Considerations for SWNT Sensors 407
proximation of the bandgap change ΔE g with axial ε chiral angle; as such, armchair tubes (θ = π6 ) are in-
and torsional γ strain, was derived by the same group sensitive to axial strain but maximally sensitive to
as [14.35] torsional strain. Zigzag tubes (θ = 0) are just the oppo-
ΔE g (ε, γ ) ≈ 3t0 sgn (2 p + 1) site, whereas chiral tubes are sensitive to both types of
strain.
× [(1 + ν)ε cos 3θ + γ sin 3θ] [eV] , For ordinary semiconductors such as silicon the
with hopping parameter t0 , family p, and chirality θ conductance varies exponentially with bandgap, i. e.,
introduced earlier, and ν = 0.2 being the Poisson’s ra- G ∝ exp(−E g /kB T ). This insight has led to many
tio. The sign of ΔE g depends on p, being positive for sensor concepts (strain, torsion, force, pressure, etc.)
p ∈ {0, 1} (bandgap increases with strain), and nega- that involve SWNTs as strain gauges [14.8–12]. More
tive for p = −1 (bandgap decreases with strain). Also, details about piezoresistive properties of SWNT in CN-
the sensitivity with respect to strain depends on the FET configurations will follow in Sect. 14.3.2.
Part B 14.1
a) c) Bandgap Eg /t0
0.6 (8,0)
EF
(7,2)
DOS
0.1
Eg E (8,2)
(5,5)
0
–3 –2 –1 0 1 2 3
Axial strain (%)
b) d) Bandgap Eg /t0
0.6
EF
DOS
0.5
(8,0)
Eg E
0.4 (8,1)
(7,2)
Torsional strain with ΔEg < 0 (9,1)
(10,0)
EF (8,3)
(6,5)
0.2
DOS
(5,5)
Eg E 0.1
(8,2)
0
–3 –2 –1 0 1 2 3
Torsion (deg)
Fig. 14.3a–d Influence of strain on electronic properties of SWNTs. (a,b) Density of states modulation by strain. The
example in (a) corresponds to tensile strain applied to a metallic zigzag CNT with p = 0 (bandgap opening), and the one
in (b) to torsional strain applied to a chiral CNT with p = 1 (bandgap closing). (c,d) Calculated bandgaps as a function
of tensile (c) and torsional (d) strain, for several chiralities belonging to each of the three families ( p = 0 dotted, p = 1
dashed, and p = −1 solid lines). ((c,d) after [14.34])
408 Part B MEMS/NEMS and BioMEMS/NEMS
SWNTs as Optically Active, Direct-Gap Materials in devices such as polarized photodetectors [14.19,
Inspecting Fig. 14.1 reveals that subband extrema come 20].
in pairs at the same lattice wavenumber k (electron–
hole symmetry). SWNTs are therefore optically ac- SWNT Sensitivity to Molecule Adsorption
tive materials. The absorption of photons generates SWNTs are hollow structures, made of surface only.
electron–hole ( e− –h+ ) pairs, and conversely, e− –h+ Molecular adsorption is therefore expected to have
recombination occurs over the nanotube bandgap via a huge impact on CNT electronic properties. The-
photon emission. Resonant absorption of photons hap- oretically, the interaction of molecules with carbon
pens at energies corresponding to transitions between nanotubes is a complex topic that can only be treated
symmetric (with respect to E F ) van Hove singu- ab initio within quantum mechanics. Generally speak-
larities. These transition energies are labeled E nn s/m , ing, molecules either physisorb or chemisorb at the
where “s/m” denotes semiconducting or metallic and CNT surface. Bond breaking is rare, as the sp2 hy-
n = 1, 2, . . . is the index of the van Hove singular- bridized carbon network is very stable. However, local
s/m > E nn
s/m , with sp3 hybridization is possible, and this adsorption mech-
Part B 14.1
s m
E 22 E 11
s
E 11 EF EF EF E EF E
Low coverage density High coverage density
New states
DOS
as in the valence or conduction band of s-SWNTs) ad- crystalline, low-defect-density conductors for which
sorbing molecules may create a disorder potential which quantum-mechanical effects are expected to play a role.
results in bandgap opening at the Fermi level. In a general conductor with length below the elas-
Chemical and biochemical sensors based on the tic mean free path (
e ) and phase relaxation length
modulation of SWNT electronic properties upon expo- (L ϕ ), at low temperature and low bias (quasiequilib-
sure to target molecules have been demonstrated [14.2, rium), each mode (Bloch wave) carries a current of
7]. The most widely used method for reading out these 2e/h. Therefore, the total source and drain current Isd ,
modulations is via transport measurements (CNFETs). can be obtained by summing the average number of
However, capacitive [14.29] and optical [14.39] detec- modes M, going from source to drain, with energies in
tion have also been employed. Chemical sensors will be the range (eVs , eV d ) centered at the Fermi level E F . This
treated in Sect. 14.3.1. yields Isd = (2e2 /h)MVsd [14.40], where e and h are
the usual electron charge and Planck’s constant, respec-
14.1.2 Carbon Nanotube FET Structures tively, and Vsd = Vs − V d is the applied source–drain
Part B 14.1
(low) bias. In other words, the sample conductance
The preceding section focused on the properties of G = Isd /Vsd is simply MG 0 , where the quantum con-
SWNTs and property modulation subject to mechani- ductance G 0 = (2e2 /h) = (1/12.9) kΩ−1 is introduced.
cal, chemical or optical stimuli. However, these prop- For a m-SWNT, at any energy close to E F (to within
erties are intrinsic to isolated SWNTs, and in order to ±1 eV), there are precisely two subbands (Fig. 14.1d),
become useful they need to be probed from the macro- meaning that M = 2 and the predicted conductance
scopic world. A simple way to probe CNT properties of a metallic tube is G = 2G 0 (= 1/6.5 kΩ−1 ). Con-
is to contact the nanotube electrically with metallic ductance values approaching 2G 0 have indeed been
leads. In combination with a nearby gate electrode this measured experimentally [14.41, 42] for m-SWNTs.
configuration is known as the carbon nanotube field- However, in most situations, G is well below 2G 0 . One
effect transistor (CNFET). A basic CNFET is sketched reason for this is that the metallic leads used to contact
in Fig. 14.6, consisting of an individual SWNT (the the tube differ in geometry, lattice, and electronic prop-
channel) placed on an insulator (typically SiO2 ), con- erties (most notably work function and effective mass)
tacted by source and drain electrodes, and gated by from SWNTs. The heterojunction between a metal and
the conducting substrate (typically highly doped Si). the SWNT incorporates interface barriers that intro-
The fabrication of such a structure (and of more ad- duce scattering and reduce the electron transfer rates.
vanced CNFETs) is covered in Sect. 14.2.2. Before Another reason for reduced conductance G < 2G 0 is
giving examples of sensor demonstrators in Sect. 14.3 disorder in the nanotube, caused by structural defects,
we review the basic operation and particularities of CN- adsorbed molecules or simply the random potential pro-
FETs. Understanding CNFET operation is important in file of the underlying oxide (trapped charges, dangling
identifying to what extent contacting SWNTs by metal bonds, etc.). Accordingly, in order to include interface
leads affects their properties. and disorder, the conductance is reduced by a multi-
The right frameset to study transport in carbon plicative factor T (G → T 2G 0 ), describing the average
nanotubes is mesoscopic physics, since nanotubes are transmission probability for an electron to propagate
from source to drain. At room temperature, for long
enough nanotubes, phonon scattering leads to incoher-
Metal electrodes
ent, classically diffusive transport, respecting Ohm’s
Source Drain law (T ∝ L1 ).
Vs Vd
SWNT (channel) As mentioned previously, for m-SWNTs the trans-
mission T can approach 1, which corresponds to thin
SiO2 (oxide) contact barriers through which electrons can easily
tunnel, and clean SWNT surfaces. For s-SWNTs, the
Doped Si (back gate) situation is different. The physics of heterostructures
predicts that, at the interface between a metal and
Vg
a semiconductor, Schottky barriers decaying slowly
within the semiconductor may arise. Figure 14.7a de-
Fig. 14.6 Schematic representation of a simple SWNT picts schematically the band diagram at the interface
field-effect transistor in cross-sectional view between a metal of work function φm , and an intrinsic
410 Part B MEMS/NEMS and BioMEMS/NEMS
a) b) c) d)
Isd
Vacuum Vacuum Vacuum Ec
χs Ec Ei
φm φnSB
EF Ei EF Ev
Ec
Ev φpSB
Ei
Ev 0 Vgs
Fig. 14.7a–d Band-structure diagram at the interface between a metal and an s-SWNT. (a) No gate bias applied Vgs = 0,
(b) Vgs > 0 just above the threshold for electron conduction, and (c) Vgs < 0 just below the threshold for hole conduction.
Part B 14.1
(d) The Isd (Vgs ) characteristic of the transistor, emphasizing the asymmetric electron versus hole conductivity, due to
p
φSB > φSB
n
s-SWNT of electron affinity χs (χs < φm < χs + E g /2). thermionic emission (see later). Above this threshold,
There are two such barriers, one at the source and the increasing Vgs further thins down the barrier, increasing
other at the drain. For simplicity, assume low Vsd bias Isd . The barrier width depends, besides on the position
(quasithermal equilibrium), which allows a flat Fermi of E F , on the geometry of the leads and oxide thick-
level to be defined throughout the structure. Assume ness [14.44]. Similarly, for negative, large enough gate
as well that the conductance of the device can be writ- voltage, the bulk valence band E V becomes aligned to
p p
ten as 1/G = 1/G s + 1/G t + 1/G d , where G s,t,d is the E F (Fig. 14.7c). In this case φB ≡ φSB = E g + χs − φm ,
conductance of the source barrier, nanotube channel, φB = E g , and hole tunneling dominates. The inequality
n
and drain barrier, respectively. The latter assumption ig- between the electron and hole barriers (in this example
p
nores interference effects and holds only if the length L φSB > φSB n ) translates into asymmetric electron versus
of the nanotube segment exceeds L ϕ [14.40]. Although hole conductivity [the two branches of the Isd (Vgs ) char-
still debated, this seems to be the case for SWNTs with acteristic], sketched qualitatively in Fig. 14.7d.
L ≥ 1 μm at room temperature [14.43]. Within this pic- As opposed to normal, planar junctions (e.g., metal–
ture, G d = G s and we can focus on just one interface. silicon), Fermi-level pinning is proposed not to exist in
At finite temperature (T = 0), the charge flow across metal–CNT junctions, because of the particular tubu-
any metal–semiconductor junction, characterized by lar geometry of the CNTs [14.45]. As a result, different
some interface barrier φB , has two components: tunnel- metals lead to barriers of different heights, which should
ing through φB and thermionic emission over φB . At also depend on the CNT diameter dt (dt determines the
T = 0 only direct tunneling at E F is possible. There are bandgap E g as discussed in Sect. 14.1.1). This fact, evi-
two such barriers, one for electrons φBn and one for holes denced experimentally [14.46], is a strong confirmation
p
φB . In Fig. 14.7a, corresponding to Vgs = 0 and an in- of the Schottky barrier (SB) model of CNFETs. Con-
p
trinsic s-SWNT, φBn = E g /2, φB = E g + χs − φm , and sequently, it seems possible to eliminate the Schottky
only thermionic emission is possible. In this situation barrier completely by either choosing metals with large
Isd is practically 0. A positive voltage Vgs > 0 applied work function such as Au, Pt or Pd (φm ≈ 5.1 eV) to
p
to the gate (Fig. 14.7b) downshifts the bands in the bulk cancel φSB , or metals with low work function such as
of the nanotube, but not the position of the band edges Al (φm ≈ 4.2 eV) to cancel φSB n , knowing that the mid-
at the interface with respect to E F . Thermionic emission gap work function of SWNTs is around 4.5 eV. Based
increases for electrons and decreases for holes. At some on this principle Javey et al. [14.47] have demonstrated
point, when the nanotube conduction band E C becomes G within 50% of the ideal 2G 0 . More experimen-
n = φ − χ and φp = E . The
aligned to E F , φBn ≡ φSB tal evidence in support of the SB-CNFET model has
m s B g
barrier φB is also a lot thinner than at Vgs = 0 because
n been supplied by Appenzeller et al. [14.48, 49]. By
the electron density inside the nanotube is higher, more measuring the temperature dependence of the Isd (Vgs )
effectively screening any interface potential. Tunnel- characteristic [14.49], they concluded that tunneling is
ing of electrons increases and will eventually exceed the main injection mechanism in SB-CNFETs and not
Single-Walled Carbon Nanotube Sensor Concepts 14.1 Design Considerations for SWNT Sensors 411
thermionic emission. However, the work function alone first reported detection of NO2 and NH3 with individual
is not enough, as for example Pt (φm ≈ 5.6 eV) yields SWNT FETs. In 2008 Cho et al. (in collaboration with
poor contacts, likely due to poor wetting or native-oxide Kong) published the first results for a hybrid CNT chem-
tunnel barriers [14.50]. ical sensor with a CMOS interface chip [14.54]. The
For certain CNFET sensor devices, the Schottky fabrication of CNT sensor arrays is still far from scal-
barrier FET model is essential in understanding the de- able, yet by this approach the interface captures signals
vice operation and sensing mechanism. This is the case from up to 24 CNFETs in parallel, further multiplexed
of (bio)chemical sensors to be discussed in Sect. 14.3.1. and sent to analysis. The authors manage to measure
In other devices, Schottky barriers, if present at all, con- 414 devices, which results in a histogram of the distribu-
tribute only to the contact resistance, which to a first tion of resistance, spread over six orders of magnitude.
approximation is constant during the sensing process, This large dispersion is the result of varying number
and can be thus factored out from the sensing mech- and type of CNTs in the devices, which is not con-
anism. This is the case with piezoresistive gauges and trolled during fabrication. Dispersion is also visible in
Part B 14.1
resonators as presented in Sects. 14.3.2 and 14.3.3, the recorded sensor response on the measurand. This
respectively. For these devices, the transport picture in- kind of platform is a first step towards capturing device
troduced in this section, in terms of thermionic emission performance statistics, a prerequisite for process control
and tunneling, is still useful for explaining their opera- and optimization.
tion. The only difference is that potential barriers for Acquiring sensor signal-to-noise ratio is currently in
electrons and holes will be found on the nanotube and progress. Electronic noise in CNFETs has received a lot
not only at the interface (Fig. 14.14c,d). Further insight of attention lately. Collins et al. [14.55] were the first
into the physics of CNFETs in different regimes (large- to measure 1/ f noise in SWNTs, both individual tubes
bias, low-temperature phenomena such as Luttinger and mats. The noise spectrum was found to agree with
liquid, Coulomb blockade or orbital Kondo effects, etc.) the classical noise power formula SV ( f ) = (A/ f β )V 2 ,
have also been described [14.51–53]. where A is the noise amplitude, f is the frequency,
β ≈ 1, and V is the bias voltage. Furthermore, the
14.1.3 Sensor Characterization noise amplitude seems to obey the empirical Hooge
law A = αH /N, where αH is the Hooge constant and
Sensor characterization refers to assessing the perfor- N is the number of carriers in the channel, because
mance metrics for a certain sensing technology. Some on average A decreases with the number of tubes in
of the most important performance metrics are sen- the sample [14.55]. Interestingly, the SV ( f ) formula
sitivity, signal-to-noise ratio (SNR), limit-of-detection is expected to hold for classically diffusive transport,
(LoD), cross-sensitivity/selectivity, signal rise/fall time which is questionable for CNFETs. However, Appen-
(speed), repeatability, offset/sensitivity, drift, hysteresis, zeller et al. [14.56] have calculated the average number
and lifetime/robustness. So far, mainly sensitivity and of carriers inside SWNT ballistic FETs, for different
dynamic properties, e.g., signal rise/fall time, have been contact metals, and found the Hooge law to be valid
targeted by the CNT sensor community (Sect. 14.3). even in this regime, with a fitted αH value of ≈ 2 × 10−3
This is a natural situation, considering that the tech- (the same as for bulk silicon). Upon closer examina-
nology for fabricating individual SWNT FETs is not tion, it has been observed by Tersoff [14.57] that the
yet scalable (Sect. 14.2). Efforts have then been focused Hooge law is not accurate in the subthreshold region
on demonstrating that CNT nanosensors are superior to of a CNFET. He proposed adding a phenomenological
existing devices, to motivate and justify future invest- term to the noise power, proportional to ( dIsd / dVgs )2 ,
ment. On the other hand, the lack of batch fabrication to account for gating of the CNFET by fluctuating
processes has hindered gathering statistical information charges in the vicinity of the tube. This model has re-
about the characteristics of CNT sensors in operation; cently been confirmed experimentally [14.58]. Overall,
most publications on individual SWNT sensors (in par- regardless of the model utilized, the noise in CNFETs
ticular FET devices) refer to data obtained from just is found to be significant, which limits the SNR of
a few fabricated samples. Therefore the reproducibility CNT sensors. The fact that noise is proposed to be
of sensor characteristics has been ignored, or marginally mainly extrinsic [14.56], i. e., caused by external fluc-
addressed. tuations, is however encouraging since it sets clear
To better illustrate matters, consider the story of technical objectives in achieving better control of the
chemical CNFET sensors. In 2000 Kong et al. [14.2] CNT environment.
412 Part B MEMS/NEMS and BioMEMS/NEMS
of nanotubes, orientation, and slack (straightness). Post ter dispersing nanotubes in liquid, different techniques
assembly, other processing steps such as nanotube elec- can be employed to separate (semiconducting from
trical contacting or device encapsulation may follow. metallic) and sort (by diameter) CNTs, as explained
The development of complete processes for the fabrica- below.
tion of micro- and nanosystems that integrate nanotube In catalytic chemical vapor deposition, SWNTs
devices/sensors with acceptable yield is currently one of are synthesized from metallic catalyst particles from
the key topics in CNT research. This section attempts a carbon-containing gas feedstock. Since CCVD is
to survey some of the available methods and processes a catalytic process, patterning of a catalyst-containing
aimed at controlled diameter, chirality, location, and layer can be utilized to grow nanotubes at selected
orientation. locations directly on preprocessed silicon chips (see
As discussed in Sect. 14.1, most of the intrinsic Sect. 14.2.2 for more details). The key observation for
properties and property modulations of SWNTs depend CCVD is that the size of the catalyst particle correlates
on their diameter dt . For example, the bandgap E g of with the SWNT diameter [14.59], and the catalyst par-
an s-SWNT is inversely proportional to dt , as are the ticle density determines the final CNT density [14.60].
n/p
Schottky barriers φSB in CNFETs (Sect. 14.1.2). Other In [14.59] Cheung et al. prepared Fe nanoparticles,
properties, most notably piezoresistance, depend as well with narrow diameter distributions centered around 3,
on chirality θ. Since dt and θ of a SWNT are defined 9, and 13 nm, on SiO2 . The grown CNT diameter
during synthesis (production), methods for nanotube distribution mirrors the initial particle size distribu-
production are briefly surveyed in Sect. 14.2.1. Location tion, with a standard deviation of roughly 30% of
and orientation control during assembly are important the average. However, 3 nm particles produce SWNTs
for building complex structures, for developing batch with double-walled CNTs (DWNT), 9 nm particles pro-
fabrication processes, and for large-scale device inte- duce D/MWNT with just a few SWNTs, and 13 nm
gration (sensor arrays). Section 14.2.2 will thus review particles produce only MWNTs, showing that a par-
processes for the assembly of SWNTs into devices, ticle size < 3 nm is required for CCVD synthesis of
together with some general strategies for achieving lo- SWNTs. For SWNT-based FETs, a good compromise
n/p
cation and orientation control. between low contact resistance (low φSB ) and large
gap E g (large on/off ratio) would correspond to diam-
14.2.1 Methods for SWNT Production eters in the range 1.5–2 nm [14.46]. Obtaining such
small catalyst nanoparticles, while keeping a narrow
Carbon nanotube production methods can be classi- size distribution, is challenging. Li et al. [14.61] have
fied into two categories, namely high temperature (arc demonstrated SWNT diameter distributions of either
discharge and laser ablation; T = 1200–3000 ◦ C) and (1.5 ± 0.4) nm or (3.0 ± 0.9) nm (again standard devi-
medium temperature (catalytic chemical vapor depo- ation/average ≈ 30%). Their method involved loading
sition, CCVD; T = 400–1100 ◦ C). Details about each the hollow cavity of the iron-storage protein ferritin
method can be obtained from textbooks [14.22, 24] and (internal cavity diameter 8 nm) with Fe, followed by de-
also Chap. 3 of this book. Here, only those aspects position on a substrate and calcination of the protein
which are relevant for CNT devices and integration are shell. Particle size was controlled by the iron loading
discussed. time.
Single-Walled Carbon Nanotube Sensor Concepts 14.2 Fabrication of SWNT Sensors 413
Currently, much activity is going into trying to Liquid Suspension Assembly Methods
achieve SWNTs with acceptable diameter distributions. Historically liquid suspension assembly methods were
For CCVD-grown tubes, the most promising approach the first to appear. The very first CNFETs, demonstrated
seems to remain improving particle size distribution. in 1998 [14.68, 69], were built through a similar, simple
Post synthesis, for SWNT dispersed in solution, re- process. In [14.69] laser-ablation SWNTs were dis-
gardless of their origin (arc discharge, laser ablation persed by sonication in dichloroethane and then spread
or CCVD), several methods have been proposed to over Au electrodes predefined using electron-beam
narrow down the diameter distribution. For exam- lithography (EBL) on a doped Si wafer substrate cov-
ple, Tromp et al. [14.62] have recently demonstrated ered by a thick gate oxide film (SiO2 ). The Si substrate
diameter separation of SWNTs by noncovalent func- served as a back gate to all devices. A schematic cross-
tionalization with anchor molecules optimally attaching
to specific tube diameters (1.2 nm). This functionaliza- Table 14.2 A comparison between liquid suspension and
tion improves solubility of the selected tubes, which in situ growth assembly methods
Part B 14.2
are subsequently separated from the insoluble bulk via
sonication and centrifugation. By using DNA wrapping SWNT source Liquid In situ
and density gradient ultracentrifugation (DGU), Arnold suspensiona growthb
et al. [14.63] have shown good separation by diameter, CNT qualityc + +/− d
for SWNTs with diameters below 1 nm. Soon after- Average diameter controle −f +g
wards, the same group [14.64] extended the technique Diameter standard deviationh + −
to surfactants, and managed to isolate narrow distribu- Chirality controli − −
tions of SWNTs with > 97% within a 0.02 nm-diameter FET performancej −k +
range, with, e.g., 84% of the tubes being of (6,5) chiral Alignment precision −/+ l +
indices. Furthermore, by using mixtures of surfactants, Ease of integration +/− m +n
they were able to produce predominantly semiconduct- a CNTs as produced by arc discharge or laser ablation
ing or metallic SWNTs. Combined with the recently b In situ growth mainly by CCVD (Sect. 14.2.1)
proposed continued growth [14.65] the DGU method c SWNT quality is attributed to crystallinity and defect density:
might one day achieve macroquantities of essentially Arc discharge or laser ablation SWNTs are still considered to
single-chirality SWNTs. However, to date, achieving produce better CNTs (Sect. 14.2.1)
d Progress in CCVD leads to constant increase in CNT quality.
SWNTs with a single chirality remains the key chal- Measures include post annealing, H2 treatment
lenge in CNT production. e Average SWNT diameters determine Schottky barrier heights
in CNFETs (Sect. 14.1.2)
f The average diameter of arc discharge and laser ablation
14.2.2 Strategies for SWNT Assembly
SWNTs is between 1.2 and 1.5 nm and is difficult to control
into Devices g The average diameter of SWNTs correlates with catalyst par-
ticle size and growth conditions and can be tuned (Sect. 14.2.1)
Perhaps the simplest way to survey SWNT assembly h A small diameter standard deviation should reduce perfor-
strategies is to split the discussion into two parts by con- mance variations in CNFET sensor devices
i Functionalization of SWNTs, sonication and centrifugation
sidering the nanotube source: liquid suspension (laser,
post synthesis narrows diameter distribution. Individual chiral-
arc or CCVD SWNTs dispersed in liquid) or in situ ities may be selected (Sect. 14.2.1). Only applicable to SWNTs
growth (CCVD SWNTs grown directly on the target assembled from liquid suspension
j FET performance for sensing applications refers to low con-
substrate). In-depth reviews of the various assembly
strategies can be found elsewhere [14.66, 67]. Here, tact resistance, low noise, and high on/off ratio
k Liquid suspension involves CNT surface treatment by
we focus only on those techniques resulting directly in (strongly binding) surfactants, sonication, and centrifugation.
CNFET device structures or that have some relevance These steps may degrade the CNFET performance
l Localized surface functionalization and dielectrophoresis
to this topic. Regarding liquid suspension or in situ
growth, there is an ongoing debate as to which one increases the alignment precision. However, bundling of indi-
vidual CNTs is still an issue in these processes
produces best results. Table 14.2 compares the two in m Wet processing is more compatible with CMOS integrated
several aspects, showing a good balance of advantages circuit (IC) substrates, but wet processing may influence the
and disadvantages. In general, fluid-dispersed tubes of- SWNTs’ electronic properties
n In situ growth is compatible with MEMS substrates in-
fer better SWNTs (because of laser or arc production
cluding suspended MEMS and nanoelectromechanical systems
and separation techniques), but assembly is more com-
(NEMS) structures
plex than for in situ growth.
414 Part B MEMS/NEMS and BioMEMS/NEMS
a) b)
SiO2 substrate
SiO2 SiO2
AFM/EBL SWNT
deposition
SWNT
SiO2
-NH2 SAM
deposition SiO2
0 0.25 0.5 0.75 1
(µm)
Fig. 14.9 (a) Schematic diagram of process for depositing SWNTs on functionalized surfaces. (b) AFM image of a device
made on Au electrodes (after [14.71])
Single-Walled Carbon Nanotube Sensor Concepts 14.2 Fabrication of SWNT Sensors 415
providing a basic method to separate metallic from alyst layer on top, transferred by contact printing.
semiconducting tubes. Electric fields were applied during growth via outer
poly-Si pads contacted by metal leads (electrodes). Fig-
In Situ Growth Assembly Methods ure 14.10 shows both the process flow and growth
The second major class of assembly methods is based results, with and without applied fields. Field-directed
on in situ growth (via CCVD) of SWNTs directly growth was subsequently demonstrated on flat sur-
onto silicon chips/wafers. CCVD has one major ad- faces as well [14.80], and catalyst patterning and
vantage over liquid suspension deposition regarding field-directed growth have been combined to build CN-
location control, namely that the catalyst can be pat- FETs by Dittmer et al. [14.81]. Another approach
terned [14.77]. A first process for fabricating CNFETs for orienting CNT growth was proposed in [14.82],
using CCVD has been proposed by Soh et al. [14.78]. where a strong correlation between nanotube orienta-
EBL was first used to pattern wells, into which cata- tion and feedstock gas flow direction was observed.
lyst was deposited. A resist lift-off left isolated catalyst Finally, numerous recent studies show that surface-
Part B 14.2
islands, from which SWNTs were grown by CCVD. directed growth is also possible on substrates such as
Metal electrodes are defined on top of the catalyst is- A- and R-planes of sapphire [14.83] or miscut C-plane
lands by EBL, contacting some of the grown tubes. sapphire [14.84]. The oriented SWNTs on sapphire
The location of tubes is thus approximately controlled, have been contacted into FET configurations by Liu
but not the orientation nor the number of nanotubes et al. [14.85].
between two electrodes, although by changing the cat- As discussed in Sect. 14.1.1, SWNTs have interest-
alyst particle density and tuning the growth parameters ing electromechanical properties, which often require
the yield can be improved. Nevertheless, considerable free-standing (suspended) nanotube segments in order
progress has been made lately in catalyst deposition to manifest. Assembly of SWNTs into suspended mi-
and patterning. Javey and Dai [14.79] have developed crostructures tends to be easier with in situ growth
a method that allows positioning of individual catalyst than liquid suspension, because of complications aris-
particles with EBL resolution. Furthermore tube-to- ing from capillary forces. Directed growth of SWNTs
particle number ratio approaching 1/1, i. e., close to from predefined silicon towers was first shown by Cas-
100% SWNT growth yield, was achieved. sell et al. [14.86, 87], via a process that preceded the
Since nanotube location control can be achieved field-directed growth shown in Fig. 14.10. Basic elec-
by catalyst patterning, a lot of work has focused sub- tromechanical structures were presented later by the
sequently on orienting CCVD growth. Field-directed same group, with a process involving patterned CCVD
growth has been pioneered by Zhang et al. [14.76], ex- growth directly from the surface of Mo, a refrac-
ploiting the large anisotropic polarizability of SWNTs. tory metal capable of withstanding CCVD temperatures
Elevated polysilicon structures were first defined onto (≈ 900 ◦ C) [14.88] and inhibiting catalyst particle dif-
a quartz substrate by optical lithography, with a cat- fusion. Integration of CCVD nanotube growth into
a) Poly-Si b) c)
Quartz
Poly-Si Catalyst
Quartz
Electrode
Carbon nanotube
Quartz
5 µm 10 µm
Fig. 14.10 (a) Schematic diagram of the process flow for electric-field-directed growth of SWNTs. (b,c) SEM images
of SWNTs, grown in various fields. At zero field, tubes grow randomly (b), whereas alignment is seen in 0.5 V/μm
(c) (after [14.76])
416 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 14.3
alyzed the adsorption of molecules either on the
b) Conductance (A/V)
2.5×10–6
nanotube body or onto the metal contacts. Transport
NH3 in the two situations is expected to have differ-
2×10 –6 ent signatures. In the former case, molecules dope
1% in the nanotube, shifting its charge-neutrality level E F .
1.5×10–6 If this process does not change the electron affin-
ity of the SWNT χs (valid at least at low doping
1×10–6 concentrations), then the electron and hole Schottky
n/p
barriers φSB remain unchanged; just the bulk bands
5×10–7 of the CNT shift up or down depending on the dop-
ing polarity. The result is simply a rigid shift of the
0 Isd (Vgs ) characteristic, as sketched in Fig. 14.13b. On
0 120 240 360 480 600 720 840 960 the other hand, molecules attaching mainly to the
Time (s) leads modify the metal work function φm . This mod-
c) Isd (µA)
ifies the relative height of the two Schottky barriers
n/p
1.5 φSB , which translates into changed polarity of the de-
vice, as illustrated by the Isd (Vgs ) characteristic in
Fig. 14.13a. These signatures have been observed ex-
1 perimentally [14.44] and assigned to one of the two
mechanisms with the help of device simulations. Ex-
posing a CNFET to potassium showed the tube doping
0.5 signature, whereas oxygen showed the contact adsorp-
After After tion signature (Fig. 14.13c,d).
NH3 Before NO2 In attempting to identify the sensing mechanisms,
0 other groups have devised means to expose just parts of
–6 –4 –2 0 2 4 6 8
the CNFET via selective passivation [14.93–95], with
Vgs (V)
so far inconclusive results. The sensing mechanism gets
Fig. 14.12a–c SWNT FET chemical sensor measurements. further convoluted by the presence of the gate oxide.
(a,b) Time response of the same device upon exposure Helbling et al. [14.96] showed that etching away the
to 200 ppm NO2 (a) and after recovery to 1% NH3 (b). underlying SiO2 from a CNFET results in chemically
(c) Isd (Vgs ) characteristics, before and after exposure to sensitive devices. On the other hand, Auvray et al. [14.3]
gas, showing a gate threshold shift (after [14.2]) observed that SWNTs on aminosilane are 4–5 times
more sensitive to triethylamine than are SWNTs on
described in [14.78]. Such CNFET devices were ex- SiO2 .
posed to NO2 and NH3 diluted in Ar or air, resulting in For CNFETs to work as biochemical sensors, some
a significant change in their conductance. Figure 14.12a modifications are required. First, because of operation
shows the response of a sensor to 200 ppm of NO2 . in aqueous solutions that contain ions, the back gate is
The conductance increased over three orders of mag- changed to a reference electrode known as the liquid
418 Part B MEMS/NEMS and BioMEMS/NEMS
a) E'C Isd
c) Isd (nA) d) Isd (nA)
3 3
E'F EC
2.5 2.5
EF E'V
2 2
EV
0 Vgs 1.5 1.5
b) Isd 1 1
EC
0.5 0.5
E'C
EF
EV 0 0
E'V –10 0 10 –10 0 10
0 Vgs Vgs (V) Vgs (V)
Part B 14.3
Fig. 14.13a–d Mechanisms encountered in CNT chemical sensors. (a,b) Band diagrams and corresponding Isd (Vgs ) char-
acteristics, before (solid line) and after (dashed line) adsorption of molecules, for the cases of adsorption at the contacts
(a) and nanotube doping (b). In (a) the metal work function φm is modified, whereas in (b) the charge-neutrality level
E F changes in the CNT. (c,d) Measured Isd (Vgs ) characteristics subject to oxygen exposure (c), assigned to mecha-
nism (a), and potassium doping (d), assigned to mechanism (b). Arrows in (a–d) indicate the change trend in Isd (Vgs )
characteristics. ((c,d) after [14.44])
gate [14.97]. With this modification, CNFETs exhibit 14.3.2 Piezoresistive Sensors
similar Isd (Vlg ) characteristics to devices in normal,
gaseous environment (Vlg ⇔ Vgs ). Second, to facilitate The electronic structure and thus the electrical con-
adsorption of biomolecules onto SWNTs, functionaliza- ductance of all three classes of SWNTs (metallic,
tion is often required [14.98]. The first biochemical CN- semiconducting, and small-bandgap semiconducting)
FET sensor was demonstrated by Besteman et al. [14.7]. depend strongly on the mechanical deformation of the
Glucose oxidase was attached to a SWNT, which sub- SWNT structure, as theoretically predicted in [14.26,
sequently showed response to both pH and glucose 35] and briefly outlined in Sect. 14.1.1 for axial and
concentration (0.1 M). In this study functionalization torsional strain.
is an enabler; unfunctionalized devices simply do not This theoretical expectation was first corrobo-
respond. More generally, functionalization is essential rated experimentally in pioneering work by Tombler
both in liquid- and gas-phase sensors for increasing et al. [14.8], for a suspended m-SWNT clamped to and
sensitivity and selectivity [14.99]. For biosensors, the contacted by two metal electrodes. Upon deforming and
debate concerning the sensing mechanism mirrors the straining the nanotube by using an AFM tip, the overall
one for chemical sensors. Recently, combining simu- conductance of the device decreased reversibly by two
lation and measurements, Heller et al. [14.100] found orders of magnitude (strain ε ≈ 3%). The authors have
that both electrostatic gating and Schottky barriers are attributed this large conductance change to a mechani-
contributing, and that the substrate also influences the cally induced modification of the atomic structure of the
response. SWNT region immediately below the AFM tip, which
Concerning the sensing performance of (bio)chem- could raise a tunnel barrier in the path of free carriers.
ical CNFETs, Kong et al. [14.2] report concentration A suspended doubly clamped CNFET configura-
limits of detection of ≈ 2 ppm for NO2 and ≈ 0.1% tion to measure the piezoresistance (i. e., the change of
for NH3 , Auvray et al. [14.3] report ≈ 10 ppb for tri- the bandgap as a function of axial strain dE g / dε) was
ethylamine, whereas Besteman et al. [14.7] speculate first proposed by Minot et al. [14.9]. The experimental
that CNFET sensors have the potential to measure setup was similar to that of Tombler. The CNFETs were
the enzymatic activity of even a single redox enzyme. strained by an AFM tip while the tip was simultaneously
These sensitivities, combined with ≈ 10 nW power con- used as a local gate in the center region of the suspended
sumption and fast response times, give SWNT FET SWNT (Fig. 14.14a). The overall CNFET conductance
(bio)chemical sensors key advantages over competing was tuned by the gate voltage at the tip (Vtip ). Fig-
technologies. ure 14.14b shows a series of sweeps of Vtip for an
Single-Walled Carbon Nanotube Sensor Concepts 14.3 Example State-of-the-Art Applications 419
intrinsic m-SWNT when no strain is applied. By grad- forms in the gate characteristic at Vtip ≈ 1 V. In this con-
ually straining the suspended doubly clamped CNFET figuration it is assumed that the tip gate is acting locally
a small bandgap opens ( dE g / dε > 0), according to the in the middle part of the SWNT, while the sections close
theory of SGS-SWNTs ( p = 0), and a conductance dip to the contacts are p-type (this is because, for Vtip = 0 V,
the CNFET is in the on-state for hole conductance). The
strained CNFET characteristic can be explained start-
a)
ing from this initial situation. For a small positive gate
voltage (0 V ≤ Vtip < 1 V) a barrier for holes is created,
Gold Gold
leading to a reduction of the total current (Fig. 14.14c).
θ
z<0
The hole current reaches its minimum (Imin ) when the
barrier formed by Vtip reaches E g . Up to this point only
L0 thermionic emission is accounted for, since no tunnel-
ing is expected due to the width of the barrier formed
by the tip gate (> 20 nm). For Vtip > 1 V the CNFET
Part B 14.3
b) G (e2/h)
section of the SWNT reaches inversion and transport
0 increases due to tunneling through the thinned barri-
0.09
ers ζ (Fig. 14.14d). Since the barrier height for holes
at Imin is approximately equal to the bandgap E g of
0.08 Rmax (h /e2) the CNFET, and since thermionic emission dominates,
the overall change in the bandgap of the CNFET due
0.07 16 to strain can be extracted by employing the thermally
12 ε activated conductance formula G ∝ exp(−E g (ε)/kB T ).
0.06 2% As a result an overall change of the bandgap of
0 0.01 0.02
ε dE g / dε ≈ 35 meV/% was extracted. This value is be-
–1 0 1 2
low the theoretical prediction of a maximum bandgap
Vtip (V) change of dE g max / dε ≈ 94 meV/% [14.35].
c) In contrast to doubly clamped suspended SWNTs
Vtip ≈ 1V EC the electromechanical properties of CNFETs adhering
Eg to thin-film membranes were investigated in two stud-
EV
ies [14.10, 11], confirming piezoresistivity also in this
situation. Grow et al. [14.10] apply the gate voltage at
the metalized backside of a SiNx membrane. For SGS-
ζ
SWNTs FETs the Isd (Vgs ) characteristic shows that the
x tube axis unstrained device is p-type, with no or only small Schot-
tky barriers for the holes (small on-state resistance). The
d)
Vtip > 1V EC applied strain increases the bandgap ( dE g / dε > 0). By
increasing the back-gate voltage Vgs , a small but wide
EV hole barrier (zero tunneling) is formed along the length
of the CNFET channel, reducing the hole current and
reaching Imin when the barrier is close to E g . The largest
ζ change in channel conductance at various strain levels
therefore happens at Imin .
Fig. 14.14a–d Strain-dependent conductance measurement The sensitivity of the devices can be expressed
on a doubly clamped fully suspended CNFET. (a) Schematic as | dE g / dε| in meV/% or alternatively by the gauge
of the experimental setup; an AFM tip strains the nano- factor (GF) which is usually used for classical piezore-
tube. (b) G(Vtip ) measurements for an intrinsic metallic sistors. The GF is defined by the relative change of
SWNT for 0–2% strain (arrow indicates increasing strain). resistance divided by the strain of the SWNT and is
Inset shows Rmax (ε) for Vtip ≈ 1 V. (c,d) Band diagrams at only constant for small strains ( dε) around a work-
the center of the SWNT at Vtip ≈ 1 V (c) and Vtip > 1 V ing point ε0 . Minot et al. [14.9] reported a maximum
(d) displaying a barrier of width ζ below the AFM tip sensitivity of 54 meV/% for the suspended CNFET,
(after [14.9]) while Grow et al. [14.10] extracted 180 meV/% for
420 Part B MEMS/NEMS and BioMEMS/NEMS
a SGS-SWNT adhering to a substrate. For a metallic For measuring mass via ω(ρ) either a cantilever
SWNT a sensitivity of 340–430 meV/% was re- or doubly clamped beam configuration can be em-
ported [14.11]. Such high sensitivities of CNFETs in ployed. Indeed, the cantilever configuration has been
contact with the substrate may be explained by ad- utilized from the beginning to acquire the Young’s
ditional mechanical distortions of the SWNTs due to moduli of CNT [14.101] or to demonstrate mass sens-
interaction with the dielectric substrate. The maximum ing. Poncharal et al. [14.14] were the first to show
reported GF for a membrane-based unstrained CN- that, with a MWNT cantilever, a mass as small as
FET is 856 [14.10], and in a prestrained suspended 22 ± 6 fg (1 fg = 10−15 g) can be measured. Recently,
metallic SWNT with ε0 ≈ 0.4% a GF of 2900 was Jensen et al. [14.15] claim to have detected only
shown [14.12]. 51 Au atoms, corresponding to a mass of roughly 17 zg
In general the piezoresistive effect of SWNTs has (1 zg = 10−21 g), with a DWNT cantilever inside a field-
been experimentally manifested in CNFETs by loading emitting device.
the tubes with axial strain. Piezoresistive CNFETs ex- On the other hand, the doubly clamped beam con-
Part B 14.3
cel in terms of their very high sensitivity even at small figuration is useful for measuring stress and strain via
strain (< 0.2%), nanoscale size, and low power con- ω(σ ), hardly possible with cantilevers which have a free
sumption (nW), in stark contrast to the mW of classical end. Moreover, for doubly clamped beams, biasing and
piezoresistors. measuring currents through the nanotube is straight-
forward, which is a prerequisite for electronic readout.
14.3.3 Resonant Sensors The first electromechanical resonator based on a doubly
clamped carbon nanotube was proposed by Sazonova
Along with piezoresistive sensors, resonant sensors are et al. [14.16]. Structurally, the device is a standard back-
useful for measuring mechanical stimuli such as strain gate CNFET with a SWNT channel doubly clamped
and stress. In addition, resonators can be utilized to by top electrodes and partly or totally suspended over
measure inertial properties (mainly mass), in which a trench (Figs. 14.11d and 14.15a [14.18]). The sus-
case they are known as inertial balances. The core pended region is the vibrating part of the nanotube.
principle, briefly explained in Sect. 14.1.1, exploits the The mechanical actuation of the nanotube is electro-
dependence of the resonant frequency (typically the static. An applied gate voltage Vgs induces a capacitive
fundamental flexural mode) of a SWNT beam on quan- Coulomb force F, deflecting the suspended nanotube
tities such as tension σ and mass density ρ, ω = ω(σ , ρ). segment. Vgs contains a direct-current (DC) compo-
a) b) Frequency (MHz)
90
70
W
Source Drain 50
δz
L
Gate 30
–4 –2 0 2 4
Vgs (V)
Fig. 14.15a,b Suspended SWNT resonator. (a) SEM image and cross-section schematic (scale bar 300 nm). (b) Detected
current as a function of gate voltage and frequency. Inset shows the extracted positions of the resonance frequency ω0
(after [14.16])
Single-Walled Carbon Nanotube Sensor Concepts References 421
nent to prestrain the nanotube (σ = 0), and a harmonic 10 kHz. In Fig. 14.15b, an intensity map of the current
at frequency ω to excite the nanotube beam vibra- amplitude δIsd (Vgs , ω) is shown [14.16]. As expected,
tion. At frequencies close to the fundamental flexural ω0 increases with Vgs , since the static gate voltage
mode ω0 , the nanotube behaves like a forced clas- changes the tension σ in the nanotube. Branches visi-
sical beam [14.102], oscillating with an amplitude ble below the fundamental ω0 in this map are due to
δz that follows a classical Lorentzian centered at ω0 initial slack. Indeed, Witkamp et al. [14.18] have shown
(Fig. 14.2). that, for straight tubes (no slack), the modes below the
Readout of the nanotube displacement is done via fundamental are suppressed.
the channel conductance. For a CNFET the conductance As for sensing capabilities of CNT resonators,
G depends on both the gate voltage Vgs and the dis- Sazonova et al. [14.16] have
√ estimated the force sensi-
tance of the nanotube from the gate z through the gate tivity to be around 1 fN/ Hz (1 fN = 10−15 N), within
capacity Cgs . This is so because the conductance is pro- a factor of ten of the best measured sensitivities at room
portional to the free-carrier concentration in the channel temperature. In [14.17], Peng et al. have loaded a CN-
(G ∝ Q = Cgs Vgs ). At frequencies close to ω0 , the os-
Part B 14
FET resonator with thermally evaporated Fe. Assuming
cillation amplitude δz becomes important, resulting in a uniform coating of 2 nm of Fe (≈ 3.5 × 10−17 g), they
a detectable conductance oscillation δG(t). Theoreti- extrapolate a minimum detectable mass on the order of
cally, δG(t) is as easy to read as applying a source–drain 1 ag (10−18 g). One of the limiting factors so far for
voltage Vsd and measuring the current Isd . However, be- both inertial balances and force sensors has been the
cause f 0 = ω0 /(2π) > 10 MHz, Sazonova et al. [14.16] quality factor Q, which in all reports has not exceeded
have proposed using the CNFET as a mixer to downcon- 300 [14.16–18]. However Q optimization has not been
vert δG(t) from ω to Δω, conveniently chosen at around addressed so far.
References
14.1 R.H. Baughman, A.A. Zakhidov, W.A. de Heer: Car- 14.3 S. Auvray, V. Derycke, M. Goffman, A. Filoramo,
bon nanotubes – The route toward applications, O. Jost, J.-P. Bourgoin: Chemical optimization of
Science 297, 787–792 (2002) self-assembled carbon nanotube transistors, Nano
14.2 J. Kong, N.R. Franklin, C. Zhou, M.G. Chapline, Lett. 5, 451–455 (2005)
S. Peng, K. Cho, H. Dai: Nanotube molecular 14.4 P.G. Collins, K. Bradley, M. Ishigami, A. Zettl: Ex-
wires as chemical sensors, Science 287, 622–625 treme oxygen sensitivity of electronic properties of
(2000) carbon nanotubes, Science 287, 1801–1804 (2000)
422 Part B MEMS/NEMS and BioMEMS/NEMS
14.5 J. Li, Y. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan: 14.19 M. Freitag, Y. Martin, J.A. Misewich, R. Mar-
Carbon nanotube sensors for gas and organic vapor tel, P. Avouris: Photoconductivity of single carbon
detection, Nano Lett. 3, 929–933 (2003) nanotubes, Nano Lett. 3, 1067–1071 (2003)
14.6 Q.F. Pengfei, O. Vermesh, M. Grecu, A. Javey, 14.20 X. Qiu, M. Freitag, V. Perebeinos, P. Avouris: Photo-
Q. Wang, H. Dai, S. Peng, K.J. Cho: Toward large ar- conductivity spectra of single-carbon nanotubes:
rays of multiplex functionalized carbon nanotube Implications on the nature of their excited states,
sensors for highly sensitive and selective molecular Nano Lett. 5, 749–752 (2005)
detection, Nano Lett. 3, 347–351 (2003) 14.21 C.T. White, J.W. Mintmire: Density of states reflects
14.7 K. Besteman, J.O. Lee, F.G.M. Wiertz, H.A. Heering, diameter in nanotubes, Nature 394, 29–30 (1998)
C. Dekker: Enzyme-coated carbon nanotubes as 14.22 M.S. Dresselhaus, G. Dresselhaus, P. Avouris: Car-
single-molecule biosensors, Nano Lett. 3, 727–730 bon Nanotubes: Synthesis, Structure, Properties
(2003) and Applications (Springer, Berlin, Heidelberg
14.8 T.W. Tombler, C. Zhou, L. Alexeyev, J. Kong, H. Dai, 2001)
W. Liu, C.S. Jayanthi, M. Tang, S.Y. Wu: Reversible 14.23 R. Saito, G. Dresselhaus, M.S. Dresselhaus: Physical
electromechanical characteristics of carbon nano- Properties of Carbon Nanotubes (Imperial College
Part B 14
14.36 P. Ruffieux, O. Gröning, M. Bielmann, P. Mauron, 14.51 M.P. Anantram, F. Léonard: Physics of carbon
L. Schlapbach, P. Gröning: Hydrogen adsorption nanotube electronic devices, Rep. Prog. Phys. 69,
on sp2 -bonded carbon: Influence of the local 507–561 (2006)
curvature, Phys. Rev. B 66, 245416–1–245416–8 14.52 M.J. Biercuk, S. Ilani, C.M. Marcus, P.L. McEuen:
(2002) Electrical transport in single-wall carbon nano-
14.37 S. Picozzi, S. Santucci, L. Lozzi, L. Valentini, B. Del- tubes. In: Carbon Nanotubes, ed. by A. Jorio,
ley: Ozone adsorption on carbon nanotubes: The G. Dresselhaus, M.S. Dresselhaus (Springer, Berlin,
role of Stone–Wales defects, J. Chem. Phys. 120, Heidelberg 2008) pp. 455–493
7147–7152 (2004) 14.53 S. Roche, E. Akkermans, O. Chauvet, F. Hekking, J.-
14.38 D.L. Carroll, P. Redlich, X. Blase, J.-C. Charlier, P. Issi, R. Martel, G. Montambaux, P. Poncharal:
S. Curran, P.M. Ajayan, S. Roth, M. Rühle: Effects Transport properties. In: Understanding Carbon
of nanodomain formation on the electronic struc- Nanotubes, ed. by A. Loiseau, P. Launois, P. Petit,
ture of doped carbon nanotubes, Phys. Rev. Lett. S. Roche, J.-P. Salvetat (Springer, Berlin, Heidel-
81, 2332–2335 (1998) berg 2006) pp. 335–437
14.39 R.J. Chen, N.R. Franklin, J. Kong, J. Cao, 14.54 T.S. Cho, K.-J. Lee, J. Kong, A.P. Chandrakasan: The
Part B 14
T.W. Tombler, Y. Zhang, H. Dai: Molecular pho- design of a low power carbon nanotube chemical
todesorption from single-walled carbon nano- sensor system, Des. Autom. Conf. 2008 (DAC 2008),
tubes, Appl. Phys. Lett. 79, 2258–2260 (2001) Anaheim (2008) pp. 84–89
14.40 S. Datta: Electronic Transport in Mesoscopic Systems 14.55 P.G. Collins, M.S. Fuhrer, A. Zettl: 1/f noise in
(Cambridge Univ. Press, Cambridge 1997) carbon nanotubes, Appl. Phys. Lett. 76, 894–896
14.41 J. Kong, E. Yenilmez, T.W. Tombler, W. Kim, H. Dai, (2000)
R.B. Laughlin, L. Liu, C.S. Jayanthi, S.Y. Wu: Quan- 14.56 J. Appenzeller, Y.M. Lin, J. Knoch, Z.H. Chen,
tum interference and ballistic transmission in P. Avouris: 1/f noise in carbon nanotube devices
nanotube electron waveguides, Phys. Rev. Lett. 87, – On the impact of contacts and device geometry,
106801–1–106801–4 (2001) IEEE Trans. Nanotechnol. 6, 368–373 (2007)
14.42 W. Liang, M. Bockrath, D. Bozovic, J.H. Hafner, 14.57 J. Tersoff: Low-frequency noise in nanoscale bal-
M. Tinkham, H. Park: Fabry–Perot interference in listic transistors, Nano Lett. 7, 194–198 (2007)
a nanotube electron waveguide, Nature 411, 665– 14.58 J. Männik, I. Heller, A.M. Janssens, S.G. Lemay,
669 (2001) C. Dekker: Charge noise in liquid-gated single-
14.43 J.-C. Charlier, X. Blase, S. Roche: Electronic and wall carbon nanotube transistors, Nano Lett. 8,
transport properties of nanotubes, Rev. Mod. Phys. 685–688 (2007)
79, 677–732 (2007) 14.59 C.L. Cheung, A. Kurtz, H. Park, C.M. Lieber:
14.44 S. Heinze, J. Tersoff, R. Martel, V. Derycke, J. Appen- Diameter-controlled synthesis of carbon nano-
zeller, P. Avouris: Carbon nanotubes as Schottky tubes, J. Phys. Chem. B 106, 2429–2433 (2002)
barrier transistors, Phys. Rev. Lett. 89, 106801–1– 14.60 Y. Tu, Z.P. Huang, D.Z. Wang, J.G. Wen, Z.F. Ren:
106801–4 (2002) Growth of aligned carbon nanotubes with con-
14.45 F. Leonard, J. Tersoff: Role of Fermi-level pinning trolled site density, Appl. Phys. Lett. 80, 4018–4020
in nanotube Schottky diodes, Phys. Rev. Lett. 84, (2002)
4693–4696 (2000) 14.61 Y. Li, W. Kim, Y. Zhang, M. Rolandi, D. Wang, H. Dai:
14.46 Z.H. Chen, J. Appenzeller, J. Knoch, Y.M. Lin, Growth of single-walled carbon nanotubes from
P. Avouris: The role of metal-nanotube contact in discrete catalytic nanoparticles of various sizes,
the performance of carbon nanotube field-effect J. Phys. Chem. B 105, 11424–11431 (2001)
transistors, Nano Lett. 5, 1497–1502 (2005) 14.62 R.M. Tromp, A. Afzali, M. Freitag, D.B. Mitzi,
14.47 A. Javey, J. Guo, Q. Wang, M. Lundstrom, H. Dai: Z. Chen: Novel strategy for diameter-selective
Ballistic carbon nanotube field-effect transistors, separation and functionalization of single-walled
Nature 424, 654–657 (2003) carbon nanotubes, Nano Lett. 8, 469–472 (2008)
14.48 J. Appenzeller, J. Knoch, V. Derycke, R. Martel, 14.63 M.S. Arnold, S.I. Stupp, M.C. Hersam: Enrichment
S. Wind, P. Avouris: Field-modulated carrier trans- of single-walled carbon nanotubes by diameter in
port in carbon nanotube transistors, Phys. Rev. density gradients, Nano Lett. 5, 713–718 (2005)
Lett. 89, 126801–1–126801–4 (2002) 14.64 M.S. Arnold, A.A. Green, J.F. Hulvat, S.I. Stupp,
14.49 J. Appenzeller, M. Radosavljević, J. Knoch, M.C. Hersam: Sorting carbon nanotubes by elec-
P. Avouris: Tunneling versus thermionic emission tronic structure using density differentiation, Nat.
in one-dimensional semiconductors, Phys. Rev. Nanotechnol. 1, 60–65 (2006)
Lett. 92, 048301–1–048301–4 (2004) 14.65 R.E. Smalley, Y. Li, V.C. Moore, B.K. Price, R. Col-
14.50 D. Mann, A. Javey, J. Kong, Q. Wang, H. Dai: Ballistic orado, H.K. Schmidt, R.H. Hauge, A.R. Barron,
transport in metallic nanotubes with reliable Pd J.M. Tour: Single wall carbon nanotube am-
ohmic contacts, Nano Lett. 3, 1541–1544 (2003) plification: en route to a type-specific growth
424 Part B MEMS/NEMS and BioMEMS/NEMS
mechanism, J. Am. Chem. Soc. 128, 15824–15829 14.80 E. Joselevich, C.M. Lieber: Vectorial growth of
(2006) metallic and semiconducting single-wall carbon
14.66 E. Joselevich, H. Dai, J. Liu, K. Hata, A.H. Windle: nanotubes, Nano Lett. 2, 1137–1141 (2002)
Carbon nanotube synthesis and organization. In: 14.81 S. Dittmer, J. Svensson, E.E.B. Campbell: Elec-
Carbon Nanotubes, ed. by A. Jorio, G. Dresselhaus, tric field aligned growth of single-walled carbon
M.S. Dresselhaus (Springer, Berlin, Heidelberg nanotubes, Curr. Appl. Phys. 4, 595–598 (2004)
2008) pp. 101–164 14.82 S. Huang, X. Cai, J. Liu: Growth of millimeter-long
14.67 Y. Yan, M.B. Chan-Park, Q. Zhang: Advances in and horizontally aligned single-walled carbon
carbon-nanotube assembly, Small 3, 24–42 (2007) nanotubes on flat substrates, J. Am. Chem. Soc.
14.68 S.J. Tans, A.R.M. Verschueren, C. Dekker: Room- 125, 5636–5637 (2003)
temperature transistor based on a single carbon 14.83 S. Han, X. Liu, C. Zhou: Template-free directional
nanotube, Nature 393, 49–52 (1998) growth of single-walled carbon nanotubes on a-
14.69 R. Martel, T. Schmidt, H.R. Shea, T. Hertel, and r-plane sapphire, J. Am. Chem. Soc. 127, 5294–
P. Avouris: Single- and multi-wall carbon nano- 5295 (2005)
tube field-effect transistors, Appl. Phys. Lett. 73, 14.84 A. Ismach, L. Segev, E. Wachtel, E. Joselevich:
Part B 14
14.95 X.L. Liu, Z.C. Luo, S. Han, T. Tang, D.H. Zhang, bilization, J. Am. Chem. Soc. 123, 3838–3839
C.W. Zhou: Band engineering of carbon nanotube (2001)
field-effect transistors via selected area chemical 14.99 K. Balasubramanian, M. Burghard: Chemically
gating, Appl. Phys. Lett. 86, 243501–1–243501–3 functionalized carbon nanotubes, Small 1, 180–192
(2005) (2005)
14.100 I. Heller, A.M. Janssens, J. Männik, E.D. Minot,
14.96 T. Helbling, R. Pohle, L. Durrer, C. Stampfer, C. Ro-
S.G. Lemay, C. Dekker: Identifying the mechanism
man, A. Jungen, M. Fleischer, C. Hierold: Sensing
of biosensing with carbon nanotube transistors,
NO2 with individual suspended single-walled car-
Nano Lett. 8, 591–595 (2008)
bon nanotubes, Sens. Actuators B 132, 491–497
14.101 M.M.J. Treacy, T.W. Ebbesen, J.M. Gibson: Ex-
(2008)
ceptionally high Young’s modulus observed for
14.97 M. Krüger, M.R. Buitelaar, T. Nussbaumer, C. Schö- individual carbon nanotubes, Nature 381, 678–680
nenberger, L. Forró: Electrochemical carbon nano- (1996)
tube field-effect transistor, Appl. Phys. Lett. 78, 14.102 D. Garcia-Sanchez, A. San Paulo, M.J. Esplandiu,
1291–1293 (2001) F. Perez-Murano, L. Forró, A. Aguasca, A. Bach-
14.98 R.J. Chen, Y. Zhang, D. Wang, H. Dai: Non-
Part B 14
told: Mechanical detection of carbon nanotube
covalent sidewall functionalization of single- resonator vibrations, Phys. Rev. Lett. 99, 085501–
walled carbon nanotubes for protein immo- 1–085501–4 (2007)
427
Nanomechan
15. Nanomechanical Cantilever Array Sensors
Part B 15
try, weighing, photothermal spectroscopy, as
well as for monitoring chemical reactions such 15.3.1 Static Mode ................................. 430
15.3.2 Dynamic Mode ............................. 432
as catalysis on small surfaces. In the past few
15.3.3 Heat Mode .................................. 433
years, the cantilever-sensor concept has been ex-
15.3.4 Further Operation Modes............... 434
tended to biochemical applications and as an
analytical device for measurements of biomateri- 15.4 Microfabrication ................................... 434
als. Because of the label-free detection principle 15.5 Measurement Setup .............................. 434
of cantilever sensors, their small size and scal- 15.5.1 Measurements in Gaseous
ability, this kind of device is advantageous for or Liquid Environments................. 434
diagnostic applications and disease monitoring, 15.5.2 Readout Principles ....................... 436
as well as for genomics or proteomics purposes. 15.6 Functionalization Techniques ................ 438
The use of microcantilever arrays enables de- 15.6.1 General Strategy .......................... 438
tection of several analytes simultaneously and 15.6.2 Functionalization Methods ............ 438
solves the inherent problem of thermal drift often 15.7 Applications ......................................... 439
present when using single microcantilever sen- 15.7.1 Chemical Detection....................... 439
sors, as some of the cantilevers can be used as 15.7.2 Biochemical Environment ............. 442
sensor cantilevers for detection, and other can- 15.7.3 Microcantilever Sensors
tilevers serve as passivated reference cantilevers to Measure Physical Properties....... 444
that do not exhibit affinity to the molecules to be 15.8 Conclusions and Outlook ....................... 445
detected.
References .................................................. 446
15.1 Technique
Sensors are devices that detect, or sense, a sig- that produce a change in response when an external
nal. Moreover, a sensor is also a transducer, i. e. it physical parameter is changed. However, many more
transforms one form of energy into another or re- types of transducers exist, such as electrochemical
sponds to a physical parameter. Most people will (pH probe), electromechanical (piezoelectric actua-
associate sensors with electrical or electronic devices tor, quartz, strain gauge), electroacoustic (gramophone
428 Part B MEMS/NEMS and BioMEMS/NEMS
pick-up, microphone), photoelectric (photodiode, solar By scanning the tip across a conductive or non-
cell), electromagnetic (antenna), magnetic (Hall-effect conductive surface using an x-y-z actuator system (e.g.
sensor, tape or hard-disk head for storage applications), a piezoelectric scanner), an image of the topography is
electrostatic (electrometer), thermoelectric (thermocou- obtained by recording the correction signal that has to
ple, thermoresistors), and electrical (capacitor, resistor). be applied to the z-actuation drive to keep the inter-
Here we want to concentrate on a further type of sensor action between tip and sample surface constant. SFM
not yet mentioned: the mechanical sensor. It responds methods are nowadays well established in scientific
to changes of an external parameter, such as tempera- research, education and, to a certain extent, also in in-
ture changes or molecule adsorption, by a mechanical dustry. Beyond imaging of surfaces, cantilevers have
response, e.g. by bending or deflection. been used for many other purposes. However, here we
focus on their application as sensor devices.
15.1.1 Cantilevers
15.1.2 History of Cantilever Sensors
Mechanical sensors consist of a fixed and a movable
part. The movable part can be a thin membrane, a plate The idea of using beams of silicon as sensors to mea-
or a beam, fixed at one or both ends. The structures sure deflections or changes in resonance frequency is
described here are called cantilevers. A cantilever is re- actually quite old. First reports go back to 1968, when
garded here as a microfabricated rectangular bar-shaped Wilfinger et al. [15.2] investigated silicon cantilever
structure that is longer than it is wide and has a thick- structures of 50 × 30 × 8 mm3 , i. e. quite large struc-
ness that is much smaller than its length or width. It tures, for detecting resonances. On the one hand, they
Part B 15.1
is a horizontal structural element supported only at one used localized thermal expansion in diffused resistors
end on a chip body; the other end is free (Fig. 15.1). (piezoresistors) located near the cantilever support to
Most often it is used as a mechanical probe to image create a temperature gradient for actuating the can-
the topography of a sample using a technique called tilever at its resonance frequency. On the other hand,
atomic force microscopy (AFM) or scanning force mi- the piezoresistors could also be used to sense me-
croscopy (SFM) [15.1], invented by Binnig et al. in the chanical deflection of the cantilever. This early report
mid 1980s [15.1]. For AFM a microfabricated sharp already contains concepts for sensing and actuation
tip is attached to the apex of the cantilever and serves of cantilevers. In the following years only a few
as a local probe to scan the sample surface. The dis- reports are available on the use of cantilevers as sen-
tance between tip and surface is controlled via sensitive sors, e.g. Heng [15.3], who fabricated gold cantilevers
measurement of interatomic forces in the piconewton capacitively coupled to microstrip lines in 1971 to
range. mechanically trim high-frequency oscillator circuits.
In 1979, Petersen [15.4] constructed cantilever-type
micromechanical membrane switches in silicon that
1 should have filled the gap between silicon transistors
3 and mechanical electromagnetic relays. Kolesar [15.5]
suggested the use of cantilever structures as electronic
2 nerve-agent detectors in 1985.
4 l Only with the availability of microfabricated can-
5
tilevers for AFM [15.1] did reports on the use of
t
cantilevers as sensors become more frequent. In 1994,
w Itoh and Suga [15.6] presented a cantilever coated with
a thin film of zinc oxide and proposed piezoresistive
Fig. 15.1 Schematic of a cantilever: (1) rigid chip body, (2) deflection readout as an alternative to optical beam-
solid cantilever-support structure, (3) hinge of cantilever, deflection readout. Cleveland et al. [15.7] reported the
(4) upper surface of the cantilever, which is usually func- tracking of cantilever resonance frequency to detect
tionalized with a sensor layer for detection of molecules, nanogram changes in mass loading when small par-
(5) lower surface of the cantilever, usually passivated in or- ticles are deposited onto AFM probe tips. Thundat
der not to show affinity to the molecules to be detected. The et al. [15.8] showed that the resonance frequency as well
geometrical dimensions, length l, width w and thickness t, as static bending of microcantilevers are influenced by
are indicated ambient conditions, such as moisture adsorption, and
Nanomechanical Cantilever Array Sensors 15.2 Cantilever Array Sensors 429
that deflection of metal-coated cantilevers can be fur- the cantilever can sense water vapor with picogram
ther influenced by thermal effects (bimetallic effect). mass resolution.
The first chemical sensing applications were presented The deflection of individual cantilevers can easily
by Gimzewski et al. [15.9], who used static cantilever be determined using AFM-like optical beam-deflection
bending to detect chemical reactions with very high sen- electronics. However, single cantilever responses can
sitivity. Later Thundat et al. [15.10] observed changes be prone to artifacts such as thermal drift or unspe-
in the resonance frequency of microcantilevers due to cific adsorption. For this reason the use of passivated
adsorption of analyte vapor on exposed surfaces. Fre- reference cantilevers is desirable. The first use of can-
quency changes have been found to be caused by mass tilever arrays with sensor and reference cantilevers was
loading or adsorption-induced changes in the cantilever reported in 1998 [15.11], and represented significant
spring constant. By coating cantilever surfaces with hy- progress for the understanding of true (difference) can-
groscopic materials, such as phosphoric acid or gelatin, tilever responses.
For the use of a cantilever as a sensor, neither a sharp Given a cantilever coated with gold on its upper sur-
Part B 15.2
tip at the cantilever apex nor a sample surface is re- face for adsorption of alkanethiol molecules and left
quired. The cantilever surfaces serve as sensor surfaces uncoated on its lower surface (consisting of silicon and
and allow the processes taking place on the surface of silicon oxide), the adsorption of thiol molecules will
the beam to be monitored with unprecedented accuracy, take place on the upper surface of the cantilever, re-
in particular the adsorption of molecules. The forma- sulting in a downward bending of the cantilever due
tion of molecule layers on the cantilever surface will to the formation of surface stress. We will call this
generate surface stress, eventually resulting in a bending process development of compressive surface stress, be-
of the cantilever, provided the adsorption preferentially cause the forming self-assembled monolayer produces
occurs on one surface of the cantilever. Adsorption is a downward bending of the cantilever (away from the
controlled by coating one surface (typically the upper gold coating). In the opposite situation, i. e. when the
surface) of a cantilever with a thin layer of a material cantilever bends upwards, we would speak of tensile
that exhibits affinity to molecules in the environment stress. If both the upper and lower surfaces of the can-
(sensor surface). This surface of the cantilever is re- tilevers are involved in the reaction, then the situation
ferred to as the functionalized surface. The other surface will be much more complex, as a predominant com-
of the cantilever (typically the lower surface) may be pressive stress formation on the lower cantilever surface
left uncoated or be coated with a passivation layer, i. e. might appear like tensile stress on the upper surface.
a chemical surface that does not exhibit significant affin- For this reason, it is of utmost importance that the
ity to the molecules in the environment to be detected. lower cantilever surface is passivated in order that ide-
To enable functionalized surfaces to be established, of- ally no processes take place on the lower surface of the
ten a metal layer is evaporated onto the surface designed cantilever.
as sensor surface. Metal surfaces, e.g. gold, may be
used to covalently bind a monolayer that represents the 15.2.3 Disadvantages
chemical surface sensitive to the molecules to be de- of Single Microcantilevers
tected from environment. Frequently, a monolayer of
thiol molecules covalently bound to a gold surface is Single microcantilevers are susceptible to parasitic
used. The gold layer is also favorable for use as a re- deflections that may be caused by thermal drift or chem-
flection layer if the bending of the cantilever is read out ical interaction of a cantilever with its environment,
via an optical beam-deflection method. in particular if the cantilever is operated in a liquid.
430 Part B MEMS/NEMS and BioMEMS/NEMS
the sensing cantilevers is crucial [15.11]. As the differ- tilever array (Fig. 15.2c), in which several cantilevers
ence in signals from the reference and sensor cantilevers are used either as sensor or as reference cantilevers
shows the net cantilever response, even small sen- so that multiple difference signals can be evaluated
sor responses can be extracted from large cantilever simultaneously. Thermal drift is canceled out as one
deflections without being dominated by undesired ef- surface of all cantilevers, typically the lower one,
fects. When only single microcantilevers are used, is left uncoated or coated with the same passivation
no thermal-drift compensation is possible. To obtain layer.
a) d) g)
b) e) h)
c) f) i) Heat source
Heat sink
Fig. 15.3a–i Basic cantilever operation modes: (a) static bending of a cantilever on adsorption of a molecular layer.
(b) Diffusion of molecules into a polymer layer leads to swelling of the polymer and eventually to a bending of the
Part B 15.3
cantilever. (c) Highly specific molecular recognition of biomolecules by receptors changes the surface stress on the upper
surface of the cantilever and results in bending. (d) Oscillation of a cantilever at its resonance frequency (dynamic mode)
allows information on mass changes taking place on the cantilever surface to be obtained (application as a microbalance).
(e) Changing the temperature while a sample is attached to the apex of the cantilever allows information to be gathered
on decomposition or oxidation process. (f) Dynamic-mode measurements in liquids yield details on mass changes during
biochemical processes. (g) In the heat mode, a bimetallic cantilever is employed. Here bending is due to the difference
in the thermal expansion coefficients of the two materials. (h) A bimetallic cantilever with a catalytically active surface
bends due to heat production during a catalytic reaction. (i) A tiny sample attached to the apex of the cantilever is
investigated, taking advantage of the bimetallic effect. Tracking the deflection as a function of temperature allows the
observation of phase transitions in the sample in a calorimeter mode
lar microcantilever of length l, thickness t and width w where E is Young’s modulus, t the thickness of the
is calculated as cantilever, ν the Poisson’s ratio (νSi = 0.24), and R the
bending radius of the cantilever.
Ewt 3 Static-mode operation has been reported in various
k= , (15.1)
4l 3 environments. In its simplest configuration, molecules
from the gaseous environment adsorb on the func-
where E is the Young’s modulus [E Si = 1.3 × 1011 N/m2 tionalized sensing surface and form a molecular layer
for Si(100)]. (Fig. 15.3a), provided the molecules exhibit some affin-
As a response to surface stress, e.g. owing to adsorp- ity to the surface. In the case of alkanethiol covalently
tion of a molecular layer, the microcantilever bends, and binding to gold, the affinity is very high, resulting
its shape can be approximated as part of a circle with in a fast bending response within minutes [15.13].
radius R. This radius of curvature is given by [15.15,16] Polymer sensing layers only exhibit a partial sensitiv-
1 6(1 − ν) ity, i. e. polymer-coated cantilevers always respond to
= . (15.2) the presence of volatile molecules, but the magnitude
R Et 2 and temporal behavior are specific to the chemistry of
The resulting surface stress change is described using the polymer. Molecules from the environment diffuse
Stoney’s formula [15.15] into the polymer layer at different rates, mainly de-
pending on the size and solubility of the molecules
Et 2 in the polymer layer (Fig. 15.3b). A wide range of
Δσ = , (15.3)
hydrophilic/hydrophobic polymers can be selected, dif-
6R(1 − ν)
432 Part B MEMS/NEMS and BioMEMS/NEMS
fering in their affinity to polar/unpolar molecules. Thus, Mass-change determination can be combined
the polymers can be chosen according to what an appli- with varying environment temperature conditions
cation requires. (Fig. 15.3e) to obtain a method introduced in the lit-
Static-mode operation in liquids, however, usually erature as micromechanical thermogravimetry [15.20].
requires rather specific sensing layers, based on molecu- A tiny piece of sample to be investigated has to be
lar recognition, such as DNA hybridization [15.17] or mounted at the apex of the cantilever. Its mass should
antigen–antibody recognition (Fig. 15.3c). Cantilevers not exceed several hundred nanograms. Adsorption,
functionalized by coating with biochemical sensing desorption and decomposition processes, occurring
layers respond very specifically using biomolecular while changing the temperature, produce mass changes
key–lock principles of molecular recognition. How- in the picogram range that can be observed in real time
ever, whether molecular recognition will actually lead by tracking the resonance-frequency shift.
to a bending of the cantilever depends on the efficiency Dynamic-mode operation in a liquid environment is
of transduction, because the surface stress has to be more difficult than in air, because of the large damp-
generated very close to the cantilever surface to pro- ing of the cantilever oscillation due to the high viscosity
duce bending. By just scaling down standard gene-chip of the surrounding media (Fig. 15.3f). This results in
strategies to cantilever geometry utilizing long spacer a low quality factor Q of the oscillation, and thus the
molecules so that DNA molecules become more acces- resonance frequency shift is difficult to track with high
sible for hybridization, the hybridization takes place at resolution. The quality factor is defined as
a distance of several nanometers from the cantilever sur-
face. In such experiments, no cantilever bending was Q = 2Δ f/ f 0 . (15.6)
Part B 15.3
Part B 15.3
φres = (γ2)
quency f 2 . The corresponding phase would be ϕres (γ2 ),
which is not equal to π/2, as would be expected in φ0 = π/2
the undamped case. If direct resonance-frequency track-
ing or a phase-locked loop is used to determine the
3
frequency of the oscillating cantilever, then only its 1 2
resonance frequency is detected, but not its eigenfre- 0
quency. Remember that the eigenfrequency, and not 0
the resonance frequency, is required to determine mass f res (γ2) f 0 Frequency
changes.
Fig. 15.4 (a) Resonance curve with no damping (0), and
increasing damping (1)–(3). The undamped curve with
15.3.3 Heat Mode resonance frequency f 0 exhibits a very high amplitude,
whereas the resonance peak amplitude decreases with
If a cantilever is coated with metal layers, thermal ex-
damping. This also involves a shift in resonance frequen-
pansion differences in the cantilever and the coating
cies from f 1 to f 3 to lower values. (b) Corresponding phase
layer will further influence cantilever bending as a func-
curves showing no damping (0), and increasing damp-
tion of temperature. This mode of operation is referred
ing (1)–(3). The steplike phase jump at resonance of the
to as the heat mode and causes cantilever bending be-
undamped resonance gradually broadens with increasing
cause of differing thermal expansion coefficients in the
damping
sensor layer and cantilever materials [15.9] (Fig. 15.3g)
5 t1 + t2 l3 reactions (Fig. 15.3h), or are due to material proper-
Δz = (α1 − α2 ) 2 P . (15.10) ties of a sample attached to the apex of the cantilever
4 t2 κ (λ1 t1 + λ2 t2 )w
(micromechanical calorimetry, Fig. 15.3i). The sensitiv-
Here α1 , α2 are the thermal expansion coefficients of ity of the cantilever heat mode is orders of magnitude
the cantilever and coating materials, respectively, λ1 , λ2 higher than that of traditional calorimetric methods
their thermal conductivities, t1 , t2 the material thick- performed on milligram samples, as it only requires
nesses, P is the total power generated on the cantilever, nanogram amounts of sample and achieves nano-
and κ is a geometry parameter of the cantilever device. joule [15.20], picojoule [15.22] and femtojoule [15.23]
Heat changes are either caused by external in- sensitivity.
fluences (change in temperature, Fig. 15.3g), occur These three measurement modes have established
directly on the surface by exothermal, e.g. catalytic, cantilevers as versatile tools to perform experiments
434 Part B MEMS/NEMS and BioMEMS/NEMS
in nanoscale science with very small amounts of reference electrode, usually made of a noble metal. If
material. the voltage between the measurement and the reference
electrode is changed, electrochemical processes on the
15.3.4 Further Operation Modes measurement electrode (cantilever) are induced, such as
adsorption or desorption of ions from the electrolyte so-
Photothermal Spectroscopy lution onto the measurement electrode. These processes
When a material adsorbs photons, a fraction of the en- lead to a bending of the cantilever due to changes in
ergy is converted into heat. This photothermal heating surface stress and in the electrostatic forces [15.24].
can be measured as a function of the light wave-
length to provide optical absorption data of the material. Detection of Electrostatic and Magnetic Forces
The interaction of light with a bimetallic microcan- The detection of electrostatic and magnetic forces is
tilever creates heat on the cantilever surface, resulting possible if charged or magnetic particles are deposited
in a bending of the cantilever [15.14]. Such bimetallic- on the cantilever [15.25, 26]. If the cantilever is placed
cantilever devices are capable of detecting heat flows in the vicinity of electrostatic charges or magnetic par-
due to an optical heating power of 100 pW, which is ticles, attractive or repulsion forces occur according to
two orders of magnitude better than in conventional the polarity of the charges or magnetic particles present
photothermal spectroscopy. on the cantilever. These forces will result in an upward
or a downward bending of the cantilever. The magnitude
Electrochemistry of the bending depends on the distribution of charged
A cantilever coated with a metallic layer (measure- or magnetic particles on both the cantilever and in the
Part B 15.5
ment electrode) on one side is placed in an electrolytic surrounding environment according to the laws of elec-
medium, e.g. a salt solution, together with a metallic trostatics and magnetism.
15.4 Microfabrication
Silicon cantilever sensor arrays have been microfab-
ricated using a dry-etching silicon-on-insulator (SOI)
fabrication technique developed in the micro-/nano-
mechanics department at the IBM Zurich Research
Laboratory. One chip comprises eight cantilevers, hav-
ing a length of 500 μm, a width of 100 μm, and
a thickness of 0.5 μm, and arranged on a pitch of
250 μm. For dynamic-mode operation, the cantilever 1 mm
thickness may be up to 7 μm. The resonance frequen-
cies of the cantilevers vary by 0.5% only, demonstrating Fig. 15.5 Scanning electron micrograph of a cantilever-
the high reproducibility and precision of cantilever sensor array. c Viola Barwich, University of Basel,
fabrication. A scanning electron microscopy image of Switzerland
a cantilever sensor-array chip is shown in Fig. 15.5.
Bypass
Cantilever array
b) G ADC PC
I to V Amplifier
conv.
VCSEL Syringe
y1 y2 pump
Part B 15.5
Liquid samples PSD
Multiplexing: 0.1–1 s
λ = 760 nm, 0.1 mW
reproducibly into the measurement chamber and purge face of a position-sensing detector (PSD). PSDs are
the chamber. light-sensitive photopotentiometer-like devices that pro-
Figure 15.6 shows the schematic set-up for experi- duce photocurrents at two opposing electrodes. The
ments performed in a gaseous (Fig. 15.6a) and a liquid, magnitude of the photocurrents depends linearly on
biochemical (Fig. 15.6b) environment for the optical the distance of the impinging light spot from the elec-
beam-deflection embodiment of the measurement set- trodes. Thus the position of an incident light beam can
up. The cantilever sensor array is located in an analysis easily be determined with micrometer precision. The
chamber with a volume of 3–90 μl, which has inlet photocurrents are transformed into voltages and am-
and outlet ports for gases or liquids. The cantilever plified in a preamplifier. As only one PSD is used,
deflection is determined by means of an array of the eight lasers cannot be switched on simultane-
eight vertical-cavity surface-emitting lasers (VCSELs) ously. Therefore, a time-multiplexing procedure is used
arranged at a linear pitch of 250 μm that emit at a wave- to switch the lasers on and off sequentially at typ-
length of 760 nm into a narrow cone of 5 to 10◦ . ical intervals of 10–100 ms. The resulting deflection
The light of each VCSEL is collimated and fo- signal is digitized and stored together with time in-
cused onto the apex of the corresponding cantilever by formation on a personal computer (PC), which also
a pair of achromatic doublet lenses, 12.5 mm in diam- controls the multiplexing of the VCSELs as well as
eter. This size has to be selected in such a way that the switching of the valves and mass flow controllers
all eight laser beams pass through the lens close to its used for setting the composition ratio of the analyte
center to minimize scattering, chromatic and spherical mixture.
aberration artifacts. The light is then reflected off the The measurement setup for liquids (Fig. 15.6b) con-
gold-coated surface of the cantilever and hits the sur- sists of a polyetheretherketone (PEEK) liquid cell,
436 Part B MEMS/NEMS and BioMEMS/NEMS
which contains the cantilever array and is sealed by 15.5.2 Readout Principles
a viton O-ring and a glass plate. The VCSELs and
the PSD are mounted on a metal frame around the This section describes various ways to determine the de-
liquid cell. After preprocessing the position of the de- flection of cantilever sensors. They differ in sensitivity,
flected light beam in a current-to-voltage converter and effort for alignment and setup, robustness and ease of
amplifier stage, the signal is digitized in an analog-to- readout as well as their potential for miniaturization.
digital converter and stored on a PC. The liquid cell is
equipped with inlet and outlet ports for liquids. They Piezoresistive readout
are connected via 0.18 mm inner-diameter Teflon tubing Piezoresistive cantilevers [15.6, 20] are usually U-
to individual thermally equilibrated glass containers, shaped, having diffused piezoresistors in both of the
in which the biochemical liquids are stored. A six- legs close to the hinge (Fig. 15.7a). The resistance in
position valve allows the inlet to the liquid chamber to the piezoresistors is measured by a Wheatstone-bridge
be connected to each of the liquid-sample containers technique employing three reference resistors, one of
separately. The liquids are pulled (or pushed) through which is adjustable. The current flowing between the
the liquid chamber by means of a syringe pump con- two branches of the Wheatstone bridge is initially
nected to the outlet of the chamber. A Peltier element nulled by changing the resistance of the adjustable resis-
is situated very close to the lumen of the chamber to tor. If the cantilever bends, the piezoresistor changes its
allow temperature regulation within the chamber. The value and a current will flow between the two branches
entire experimental setup is housed in a temperature- of the Wheatstone bridge. This current is converted
controlled box regulated with an accuracy of 0.01 K to via a differential amplifier into a voltage for static-
Part B 15.5
a) Osc. amplifier
b) Passivation layer SiO2 0.2 μm
1 Upper electrode Au/Cr 0.2 μm
Piezoelectric PZT layer 1.25 μm
2 2 Frequency Vac
Thermal SiO2 layer 1.8 μm
3 4 Lower electrode Pt/Ti 0.35 μm
Freq. Rigid
R + ΔR 8 gen. chip
+
body
Vout
Rref Rref G
– Lock-in e)
6
7 L
5 VWB Laser beam
c) d) 1
l Δd
1 10 2
θ
2 +
3
V
–
θ/2 Δx
13 12 11 9 Cantilever
3 4 4
R
5 s
7 8
6
Fig. 15.7 (a) Piezoresistive readout: (1) cantilever, (2) piezoresistors, (3) Au contact pads, (4) external piezocrystal for
actuation, (5) Wheatstone-bridge circuit, (6) differential amplifier, (7) lock-in amplifier, (8) function generator. (b) Piezo-
electric readout. (c) Capacitive readout: (1) solid support, (2) rigid beam with counter-electrode, (3) insulation layer
(SiO2 ), (4) flexible cantilever with electrode. (d) Interferometric readout: (1) laser diode, (2) polarizer, (3) nonpolarizing
beam splitter, (4) Wollaston prism, (5) focusing lens, (6) cantilever, (7) reference beam (near cantilever hinge), (8) ob-
ject beam (near cantilever apex), (9) diaphragm and λ/4 plate, (10) focusing lens, (11) Wollaston prism, (12) quadrant
photodiode, (13) differential amplifier. (e) Beam-deflection readout
Nanomechanical Cantilever Array Sensors 15.5 Measurement Setup 437
the piezoresistive cantilever is externally actuated via object and reference beam is established. Another Wol-
a frequency generator connected to a piezocrystal. The laston prism separates the reference and object beams
alternating current (AC) actuation voltage is fed as ref- again for analysis with a four-quadrant photodiode.
erence voltage into a lock-in amplifier and compared A differential amplifier is used to obtain the cantilever
with the response of the Wheatstone-bridge circuit. This deflection with high accuracy. However, the interfero-
technique allows one to sweep resonance curves and to metric setup is quite bulky and difficult to handle.
determine shifts in resonance frequency.
Optical (Beam-Deflection) Readout
Piezoelectric Readout The most frequently used approach to read out can-
Piezoelectric cantilevers [15.27] are actuated by apply- tilever deflections is optical beam deflection [15.32],
ing an electric AC voltage via the inverse piezoelectric because it is a comparatively simple method with an ex-
effect (self-excitation) to the piezoelectric material cellent lateral resolution. A schematic of this method is
(PZT or ZnO). Sensing of bending is performed by shown in Fig. 15.7e.
recording the piezoelectric current change due to the The actual cantilever deflection Δx scales with the
fact that the PZT layer may produce a sensitive field cantilever dimensions; therefore the surface stress Δσ
response to weak stress through the direct piezoelectric in N/m is a convenient quantity to measure and com-
effect. Such cantilevers are multilayer structures con- pare cantilever responses. It takes into account the
sisting of an SiO2 cantilever and the PZT piezoelectric cantilever material properties, such as Poisson’s ratio ν,
layer. Two electrode layers, insulated from each other, Young’s modulus E and the cantilever thickness t. The
provide electrical contact. The entire structure is pro- radius of curvature R of the cantilever is a measure of
Part B 15.5
tected using passivation layers (Fig. 15.7b). An identical bending, (15.2). As shown in the drawing in Fig. 15.7e,
structure is usually integrated into the rigid chip body the actual cantilever displacement Δx is transformed
to provide a reference for the piezoelectric signals from into a displacement Δd on the PSD. The position of
the cantilever. a light spot on a PSD is determined by measuring
the photocurrents from the two facing electrodes. The
Capacitive Readout movement of the light spot on the linear PSD is calcu-
For capacitive readout (Fig. 15.7c), a rigid beam lated from the two currents I1 and I2 and the size L of
with an electrode mounted on the solid support and the PSD by
a flexible cantilever with another electrode layer are
I1 − I2 L
used [15.28, 29]. Both electrodes are insulated from Δd = · . (15.11)
each other. Upon bending of the flexible cantilever the I1 + I2 2
capacitance between the two electrodes changes and As all angles are very small, it can be assumed that the
allows the deflection of the flexible cantilever to be bending angle of the cantilever is equal to half of the
determined. Both static- and dynamic-mode measure- angle θ of the deflected laser beam, i. e. θ/2. Therefore,
ments are possible. the bending angle of the cantilever can be calculated to
be
Optical (Interferometric) Readout
Interferometric methods [15.30, 31] are most accurate θ Δd
= , (15.12)
for the determination of small movements. A laser beam 2 2s
passes through a polarizer plate (polarization 45◦ ) and where s is the distance between the PSD and the can-
is partially transmitted by a nonpolarized beam splitter tilever. The actual cantilever deflection Δx is calculated
(Fig. 15.7d). The transmitted beam is divided in a Wol- from the cantilever length l and the bending angle θ/2
laston prism into a reference and an object beam. These by
mutually orthogonally polarized beams are then focused
onto the cantilever. Both beams (the reference beam θ/2
Δx = ·l . (15.13)
from the hinge region and the object beam from the 2
apex region of the cantilever) are reflected back to the
Combination of (15.12) and (15.13) relates the actual
objective lens, pass the Wollaston prism, where they
cantilever deflection Δx to the PSD signal
are recombined into one beam, which is then reflected
into the other arm of the interferometer, where after the lΔd
λ/4 plate a phase shift of a quarter wavelength between Δx = . (15.14)
4s
438 Part B MEMS/NEMS and BioMEMS/NEMS
The relation between the radius of curvature and the and after substitution becomes
deflection angle is 2ls
R= , (15.16)
θ l Δd
= , (15.15)
2 R or R = 2Δx
l2
.
with 2 nm of titanium and 20 nm of gold, which yields filled easily using a laboratory pipette, so that the fluid
a reflective surface and an interface for attaching func- with the probe molecules for coating the cantilever is
tional groups of probe molecules, e.g. for anchoring guided through the channels towards openings at a pitch
molecules with a thiol group to the gold surface of the matched to the distance between individual cantilevers
cantilever. Such thin metal layers are believed not to in the array (Fig. 15.8a).
contribute significantly to bimetallic bending, because The cantilever array is then introduced into the open
the temperature is kept constant. channels of the μFN that are filled with a solution of
the probe molecules. The incubation of the cantilever
15.6.2 Functionalization Methods array in the channels of the μFN takes from a few
seconds (self-assembly of alkanethiol monolayers) to
There are numerous ways to coat a cantilever with ma- several tens of minutes (coating with protein solutions).
terial, both simple and more advanced ones. The method To prevent evaporation of the solutions, the channels are
of choice should be fast, reproducible, reliable and al- covered by a slice of poly(dimethylsiloxane) (PDMS).
low one or both of the surfaces of a cantilever to be In addition, the microfluidic network may be placed in
coated separately. an environment filled with saturated vapor of the solvent
used for the probe molecules.
Simple Methods
Obvious methods to coat a cantilever are thermal or Array of Dimension-Matched Capillaries
electron-beam-assisted evaporation of material, elec- A similar approach is insertion of the cantilever ar-
trospray or other standard deposition methods. The ray into an array of dimension-matched disposable
disadvantage of these methods is that they only are glass capillaries. The outer diameter of the glass cap-
suitable for coating large areas, but not individual illaries is 240 μm so that they can be placed neatly
cantilevers in an array, unless shadow masks are next to each other to accommodate the pitch of the
used. Such masks need to be accurately aligned to cantilevers in the array (250 μm). Their inner diam-
the cantilever structures, which is a time-consuming eter is 150 μm, providing sufficient room to insert the
process. cantilevers (width: 100 μm) safely (Fig. 15.8b). This
Other methods to coat cantilevers use manual place- method has been successfully applied for the deposi-
ment of particles onto the cantilever [15.9, 20, 33–35], tion of a variety of materials onto cantilevers, such
which requires skillful handling of tiny samples. Can- as polymer solutions [15.37], self-assembled monolay-
tilevers can also be coated by directly pipetting solu- ers [15.39], thiol-functionalized single-stranded DNA
tions of the probe molecules onto the cantilevers [15.36] oligonucleotides [15.40], and protein solutions [15.41].
Nanomechanical Cantilever Array Sensors 15.7 Applications 439
Inkjet Spotting
a) 2 b) c)
All of the above techniques require manual align- 7
ment of the cantilever array and functionalization 5
4 6
tool, and are therefore not ideal for coating a large
number of cantilever arrays. The inkjet-spotting tech- 3
nique, however, allows rapid and reliable coating of 1 1 1
cantilever arrays [15.42, 43]. An x-y-z positioning sys-
tem allows a fine nozzle (capillary diameter: 70 μm) Fig. 15.8 (a) Cantilever functionalization in microfluidic networks.
to be positioned with an accuracy of approximately (b) Incubation in dimension-matched microcapillaries. (c) Coat-
10 μm over a cantilever. Individual droplets (diameter: ing with an inkjet spotter: (1) cantilever array, (2) reservoir wells,
60–80 μm, volume 0.1–0.3 nl) can be dispensed indi- (3) microfluidic network with channels, (4) PDMS cover to avoid
vidually by means of a piezo-driven ejection system evaporation, (5) microcapillaries, (6) inkjet nozzle, (7) inkjet x-y-z
in the inkjet nozzle. When the droplets are spot- positioning unit
ted with a pitch smaller than 0.1 mm, they merge
and form continuous films. By adjusting the number monolayers, polymer solutions, self-assembled DNA
of droplets deposited on the cantilevers, the result- single-stranded oligonucleotides [15.43], and protein
ing film thickness can be controlled precisely. The layers has been demonstrated. In conclusion, inkjet
inkjet-spotting technique allows a cantilever to be spotting has turned out to be a very efficient and ver-
coated within seconds and yields very homogeneous, satile method for functionalization, which can even be
reproducibly deposited layers of well-controlled thick- used to coat arbitrarily shaped sensors reproducibly and
Part B 15.7
ness. Successful coating of self-assembled alkanethiol reliably [15.44, 45].
15.7 Applications
15.7.1 Chemical Detection relative humidity of ZFM5 zeolites attached to resonat-
ing microcantilevers was observed in [15.50]. Relative
Hydrogen humidity was measured with an accuracy of 1% using
Early reports on detection of gases such as hydrogen piezoresistive sensors embedded in polymer [15.51].
involve nanomechancal detection of catalytic reactions A detection limit of 10 ppm is achieved using Al2 O3
of bimetallic microcantilevers coated with aluminum coated microcantilevers [15.52].
and a top layer of platinum in thermal mode [15.9].
The catalytic reaction of oxygen present in a reaction Other Vapors
chamber with hydrogen being introduced into the cham- ZFM5 zeolites were used to detect vapor of p-nitro-
ber produces oscillatory chemical reactions resulting in aniline dye in dynamic mode with picogram sensitiv-
mechanical oscillations of the cantilever due to heat for- ity [15.50]. A freon gas sensor using a piezoelectric
mation related to catalytic conversion of H2 and O2 to microcantilever coated with MFI zeolite is described
form H2 O. By use of an array of four platinum coated in [15.53]. Ethanol vapor detection in dynamic mode
and four uncoated microcantilevers, a change of the de- is described in [15.54].
flection signal due to bending of the platinum coated
cantilever relative to the uncoated cantilevers is ob- Alkane Thiol Vapors
served upon hydrogen adsorption in the presence of Surface stress changes and kinetics were measured
oxygen [15.11]. Similar responses are obtained with in situ during the self-assembly of alkanethiols on
Pd coated glass cantilevers [15.46] and with Pd coated gold by means of a micromechanical sensor, ob-
silicon microcantilevers using dynamic mode [15.47], serving scaling of compressive surface stress with
capacitive readout [15.48] or beam-deflection readout the length of the alkane chain [15.13, 55]. 65 ppb
in static mode [15.49]. of 2-mercaptoethanol have been measured evaluating
the response of gold-coated silicon nitride micro-
Water Vapor cantilevers [15.56]. The mechanism of stress for-
First observation of microcantilever resonance fre- mation upon adsorption of thiol layers has been
quency detuning is reported in [15.8]. A dependence on studied by exposing monolayers of alkanethiols on
440 Part B MEMS/NEMS and BioMEMS/NEMS
gold to low energy Ar ions, resulting in formation at concentration of 150 ppm within seconds are re-
of a large tensile stress [15.57]. The influence of ported [15.73].
surface morphology and thickness of the gold coat-
ing of the cantilever is discussed in [15.58, 59]. Ion Sensing
A multiple-point deflection technique has been used Using microcantilevers coated with a self-assembled
to investigate stress evolution during the adsorption monolayer of triethyl-12-mercaptododecylammonium
of dodecanethiol on microcantilever sensors, allow- bromide on gold CrO2− 4 ions are detected at a con-
ing to assess the cantilever bending profile [15.60]. centration of 10−9 M. Other anions, such as Cl− , Br− ,
−
Using gold-coated, piezoelectric-excited, millimeter- CO2− 2−
3 , HCO3 and SO4 do not deflect such modi-
sized cantilevers exposed to 1-hexadecanethiol (HDT) fied cantilevers significantly [15.74]. Hg2+ has been
in ethanol, a detection range between 1 fM to 1 mM measured at a concentration of 10−11 M using a mi-
is claimed [15.61]. The formation of alkanedithiol crocantilever coated with gold. Almost no affinity to
(HS–(CH)SH) monolayers on gold in solution is other cations exists, such as K+ , Na+ Pb2+ , Zn2+ ,
monitored using microcantilever sensors [15.62]. The Ni2+ , Cd2+ , Cu2+ , and Ca2+ [15.75]. Adsorption char-
nanomechanical bending of microfabricated cantilevers acteristics of Ca2+ ions as a function of concentration
during the immobilization of alkanethiols of differ- in aequous CaCl2 solution was investigated in static
ent chain lengths has been investigated in the liquid and dynamic mode [15.76]. Microcantilevers function-
phase [15.63]. alized with metal-binding protein AgNt84-6 are able
to detect heavy metal ions like Hg2+ and Zn2+ , but
Metal Vapors are insensitive to Mn2+ [15.77]. Hydrogels containing
Part B 15.7
Detection of mercury vapor was one of the first benzo-18-crown-6 have been used to modify microcan-
applications of microcantilever sensors in dynamic tilevers for measurements of the concentration of Pb2+
mode [15.10]. 20 ppb of Hg vapor was detected in aqueous solutions [15.78]. Using different thiolated
using a microcantilever with an integrated piezoelec- ligands as self-assembled monolayers (SAMs) func-
tric film [15.64]. A monolayer of 1,6-hexanedithiol tionalized on silicon microcantilevers (MCs) coated
has been identified as a unusually specific recog- with gold allows to detect Cs+ , Co2+ and Fe3+ [15.79].
nition agent for CH3 Hg+ [15.65]. 15 ppb detection A gold coated microcantilever is utilized as the
limit for mercury is reported using microcantilevers working electrode to detect Cr(VI) [15.80]. Others
that are thermally excited at the fundamental and use 11-undecenyltriethylammonium bromide [15.81]
first three higher order modes [15.66, 67]. Cs ion or sol–gel layers [15.82] for detection of Cr(VI).
concentrations in the range of 10−11 –10−7 M were Based on the EDTA-Cd(II) complex and its bind-
detected using a 1,3-alternate 25,27-bis(11-mercapto- ing capability to bovine serum albumine (BSA) and
1-undecanoxy)-26,28-calix[4]benzo-crown-6 caesium antibody-based Cd(II) sensor using microcantilevers is
recognition agent bound to a gold coated microcan- presented [15.83].
tilever [15.68]. The crown cavity is highly selective
to Cs, as compared to K or Na. An atomic force mi- Volatile Organic Compounds
croscope cantilever has been used as a bending-beam A microcantilever-based alcohol vapor sensor is de-
sensor to measure surface stress changes which occur scribed in [15.84] using the piezoresistive technique and
during electrochemical processes, such as the formation polymer coating. They also present a simple evapora-
of a Pb layer on Au [15.69]. tion model that allows determining the concentration.
The detection limit found is 10 ppm for methanol,
HF and HCN ethanol and 2-propanol. In [15.85] an integrated com-
Microcantilevers have been used as a gas sensor plementary metal oxide semiconductor (CMOS) chem-
to detect hydrofluoric acid (HF) at a threshold of ical microsensor with piezoresistive detection (Wheat-
0.2 ppm [15.70]. Femtomolar HF concentrations, which stone bridge configuration) using poly(etherurethane)
is also a decomposition component of nerve agents, (PEUT) as the sensor layer is presented. They are able
were detected using a SiO2 microcantilever. The high to reversibly detect volatile organic compounds (VOCs)
sensitivity is considered to be due to the reaction such as toluene, n-octane, ethyl acetate and ethanol
of HF with SiO2 [15.71]. The etching rate is de- with a sensitivity level down to 200 ppm. An improved
termined to 0.05 nm/min for SiO2 and 0.7 nm/min version of that device is described in [15.86]. The
for Si3 N4 [15.72]. Sensor responses towards HCN at sensitivity could be increased to 5 ppm for n-octane.
Nanomechanical Cantilever Array Sensors 15.7 Applications 441
Later the technique has been refined by using elec- phtalocyanine and methyl-β-cyclodextrin. Analytes de-
tromagnetic rather than electrothermal actuation and tected include pentane, methanol, acetonitrile, acetone,
transistor-based readout reducing power dissipation on ethanol and trichloroethylene. In [15.95] results are pre-
the cantilever [15.87]. Piezoelectric readout in dynamic sented on independent component analysis (ICA) of
mode and electromagnetic actuation of cantilevers ethanol, propanol and DIMP using cantilever coated
spray-coated with PEUT is reported in [15.88], achiev- with molecular recognition phases (MRP), whereby
ing a sensitivity of 14 ppm for ethanol. In [15.89] ICA has proven its feature extraction ablility for com-
a study is published how to prepare polyethylene glycol ponents in mixtures.
(PEG) coated microcantilever sensors using a microcap-
illary pipette assisted method. PEG coating is suitable Toxins
for ethanol sensing as ethanol quickly forms hydrogen Detection of the organochlorine insecticide compound
bonds with the OH groups of the PEG. Sensor op- dichlorodiphenyltrichloroethane (DDT) is reported us-
eration is reported to be reversible and reproducible. ing a synthetic hapten of the pesticide as recognition
In [15.90] artificial neural networks are used for analyte site conjugated with bovine serum albumin (BSA) co-
species and concentration identification with polymer valently immobilised on the gold-coated side of the
coated optically read-out microcantilevers. The analytes cantilever by using thiol self assembled monolay-
detected are carbon dioxide, dichloromethane, diiso- ers [15.96].
propylmethylphosphonate (DIMP), dioxane, ethanol,
water, 2-propanol, methanol, trichloroethylene and Explosives, Chemical Warfare and Biohazards
trichloromethylene. In [15.91] the chemical sensing Security measures require inexpensive, highly selective
Part B 15.7
performance of a silicon reconant microcantilever sen- and very sensitive small sensors that can be mass-
sor is investigated in dependence on the thickness produced and microfabricated. Such low cost sensors
of the sensitive coating. For a coating thickness of could be arranged as a sensor grid for large area cov-
1, 4 and 21 μm of PEUT a limit of detection of erage of sensitive infrastructure, like airports, public
30 ppm was found for ethanol. A new concept of buildings, or traffic infrastructure. Threats can be of
parylene micromembrane array for chemical sensing chemical, biological, radioactive or explosive nature.
is presented [15.92] using the capacitive method. The Microcantilever sensors are reported to offer very high
parylene membrane is suspended over a metal pad sensitivities of explosives detection. Photomechanical
patterned on the substrate. The pad and part of the chemical microsensors based on adsorption-induced
membrane that is metal-coated serve as electrodes and photoinduced stress changes due to the presence
for capacitive measurement. The top electrode lo- of diisopropyl methyl phosphonate (DIMP), which is
cated on the membrane is chemically modified by a model compound for phosphorous-containing chem-
applying a gold layer and self-assembled thiol mono- ical warfare agents, and trinitrotoluene (TNT), an
layers (–COOH, –CH3 and –OH) for detection of explosive are reported [15.97]. Further explosives fre-
analyte molecules. Successful detection of 2-propanol quently used include pentaerythritol tetranitrate (PETN)
and toluene is reported. In [15.93] a sensitive self- and hexahydro-1,3,5-triazine (RDX), often also with
oscillating cantilever array is described for quantitative plastic fillers [15.98]. These compounds are very sta-
and qualitative analysis of organic vapor mixtures. The ble, if no detonator is present. Their explosive power,
cantilevers are electromagnetically actuated and the however, is very large, and moreover, the vapor pres-
resonance frequency is measured using a frequency sures of PETN and RDX are very low, in the range of
counter. Sensor response is reproducible and reversible. ppb and ppt. By functionalizing microcantilevers with
Using PEUT coating the smallest measured concentra- self-assembled monolayers of 4-mercaptobeonzoic acid
tion is 400 ppm, but the limit of detection is well below (4-MBA) PETN was detected at a level of 1400 ppt and
1 ppm. In [15.94] a combination of gas chromatogra- RDX at a level of 290 ppt [15.99]. TNT was found to
phy with a microcantilever sensor array for enhanced readily stick to Si surfaces, suggesting the use of micro-
selectivity is reported. Test VOC mixtures composed cantilevers for TNT detection, taking advantage of the
of acetone, ethanol and trichloroethylene in pentane, as respective adsorption/desorption kinetics [15.100, 101].
well as methanol with acetonitrile in pentane were first Detection of TNT via deflagration on a microcantilever
separated in a gas chromatography column and then is described by Pinnaduwage et al. [15.101]. They used
detected using micocantilevers coated with responsive piezoresistive microcantilevers where the cantilever de-
phases such as 3-aminopropyltriethoxy silane, copper flection was measured optically via beam deflection.
442 Part B MEMS/NEMS and BioMEMS/NEMS
TNT vapor from a generator placed 5 mm away from pH when coated with 4-aminobutyltriethoxysilane,
the microcantilever was observed to adsorb on its sur- 11-mercaptoundecanoic acid and Au/Al-coated over
face resulting in a decrease of resonance frequency. a pH range 2–12. Aminosilane-modified SiO2 /Au
Application of an electrical pulse (10 V, 10 ms) to cantilevers performed robustly over pH range 2–8
the piezoresistive cantilever resulted in deflagration of (49 nm deflection/pH unit), while Si3 N4 /Au can-
the TNT vapor and a bump in the cantilever bend- tilevers performed well at pH 2–6 and 8–12 (30 nm
ing signal. This bump was found to be related to the deflection/pH unit) [15.108]. Microcantilevers with
heat produced during deflagration. The amount of heat poly(methacrylic acid) (PMAA) and poly(ethylene gly-
released is proportional to the area of the bump in col) dimethacrylate coating showed to be sensitive to
the time versus bending signal diagram of the pro- pH changes [15.109]. Also hydrogel catings were found
cess. The deflagration was found to be complete, as to be sensitive to pH [15.110]. The dependence of the
the same resonance frequency as before the experi- micromechanical responses to different ionic strength
ment was observed. The amount of TNT mass involved and ion species present in the aqueous environment is
was determined as 50 pg. The technique was later ex- discussed in [15.111], highlighting the critical role of
tended to the detection of PETN and RDX, where counter- and co-ions on surface stress.
much slower reaction kinetics was observed [15.99,
102]. Traces of 2,4-dinitrotoluene (DNT) in TNT can Glucose
also be used for detection of TNT, because it is Glucose sensing via microcantilevers is achieved by
the major impurity in production grade TNT. Fur- coating the cantilevers with the enzyme glucose oxidase
thermore DNT is a decomposition product of TNT. on gold [15.112] or via polyethyleneimine (PEI) con-
Part B 15.7
The saturation concentration of DNT in air at 20 ◦ C jugation [15.113]. Glucose concentrations between 0.2
is 25 times higher than that of TNT. DNT was re- and 20 mM could be detected [15.114]. In another study
ported detectable at the 300 ppt level using polysiloxane a detection range between 2 and 50 mM is reported for
polymer layers [15.103]. Microfabrication of electro- glucose. No signal is observed for fructose, mannose
statically actuated resonant microcantilever beams in and galactose [15.115].
CMOS technology for detection of the nerve agent
stimulant dimethylmethylphosphonate (DMMP) using Hydrogen Peroxide (H2 O2 )
polycarbosilane-coated beams [15.104] is an important Hydrogen peroxide is detected at the nM level us-
step towards an integrated platform based on silicon mi- ing multilayer modified microcantilevers functionalized
crocantilevers, which besides compactness might also through a layer-by-layer nanoassembly technique via
include telemetry [15.105]. Cu2+ /L-cysteine bilayer- intercalation of the enzyme horseradish peroxidase. The
coated microcantilever demonstrated high sensitivity magnitudes of bending were found to be proportional to
and selectivity toward organo-phosphorus compounds the concentrations of hydrogen peroxide [15.116].
in aqueous solution. The microcantilever was found to
undergo bending upon exposure to nerve agent simu- DNA, RNA
lant DMMP at concentrations as low as 10−15 M due Specific DNA hybridization detection was observed
to the complexation of the phosphonyl group and the via surface stress changes related to transduction of
Cu2+ /L-cysteine bilayer on the microcantilever sur- receptor-ligand binding into a direct nanomechanical
face [15.106, 107]. response of microfabricated cantilevers without the
need for external labeling or amplification. The dif-
15.7.2 Biochemical Environment ferential deflection of the cantilevers was found to
provide a true molecular recognition signal despite large
pH responses of individual cantilevers. Hybridization of
Control of pH is often important in biochemi- complementary oligonucleotides shows that a single
cal reactions. Therefore this section concerns mea- base mismatch between two 12-mer oligonucleotides is
surement of pH using microcantilevers. The in- clearly detectable [15.17]. The findings were confirmed
terfacial stress of self-assembled monolayers of or modeled by several groups [15.117, 118]. Hy-
mercaptohexadecanoic acid and hexadecanethiol de- bridization in a complex nonspecific background was
pends on pH values and ionic strength [15.39]. observed in a complement concentration range between
SiO2 and silicon nitride microcantilevers were also 75 nM and 2 μM [15.40], following Langmuir model
found to exhibit a deflection dependence with kinetics [15.119]. Enzymatic processes were directly
Nanomechanical Cantilever Array Sensors 15.7 Applications 443
Part B 15.7
of a laser beam illuminating the cantilevers of an ar- fragment antibodies instead of complete antibodies
ray sequentially. DNA hybridization is also reported allowed to lower the limit of detection to concentra-
using polymer SU-8 cantilevers [15.125]. Mukhopad- tions of about 1 nM [15.139]. In [15.140] detection
hyay et al. report 20 nM hybridization sensitivity using of prostate specific antigen (PSA) and C-reactive pro-
piezoresistive cantilevers and DNA sequences with tein is reported. Detection of the human oestrogen
an overhang extension distal to the surface [15.126]. receptor in free and oestradiol-bound conformation can
A larger array comprising 20 microcantilevers is de- be distinguished [15.141]. The Ca2+ binding protein
scribed in [15.127]. Moreover, the authors present calmodulin changes its conformation in presence of ab-
integration of the array with microfluidics. Surface sence of Ca2+ resulting in a microcantilver deflection
stress changes in response to thermal dehybridization, change [15.142]. No effect is observed upon expo-
or melting, is reported [15.128]. The dependence of salt sure to K+ and Mg2+ . Detection of activated cyclic
concentration and hybridization efficiency is discussed adenosine monophosphate (cyclic AMP)-dependent
in [15.129]. Two different DNA-binding proteins, the protein kinase (PKA) is performed in dynamic mode
transcription factors SP1 and NF-kappa B are inves- employing a peptide derived from the heat-stable pro-
tigated [15.130]. Phase transition and stability issues tein kinase inhibitor (PKI) [15.143]. Detection of
of DNA are discussed in [15.131]. Differential gene streptavidin at 1–10 nM concentration is reported us-
expression of the gene 1-8U, a potential marker for ing biotin-coated cantilevers [15.144]. Using GST
cancer progression or viral infections, has been ob- (glutathione-S-transferase) for detection of GST anti-
served in a complex background. The measurements bodies, a sensitivity of 40 nM was obtained [15.145].
provide results within minutes at the picomolar level A two-dimensional multiplexed real-time, label-free
without target amplification, and are sensitive to base antibody–antigen binding assay by optically detecting
mismatches [15.132]. nanoscale motions of two-dimensional arrays of mi-
crocantilever beams is presented in [15.146]. Prostate
Proteins and Peptides specific antigen (PSA) was detected at 1 ng/ml using
Microfabricated cantilevers were utilized to detect ad- antibodies covalently bound to one surface of the can-
sorption of low-density lipoproteins (LDL) and their tilevers. Conformational changes in membrane protein
oxidized form (oxLDL) on heparin, an to detect patches of bacteriorhodopsin proteoliposomes were ob-
adsorption of bovine serum albumine and Immunoglob- served with microcantilevers through prosthetic retinal
uline G (IgG) [15.133]. In [15.134] the activity, removal (bleaching) [15.147]. Using an analog of the
stability, lifetime and re-usability of monoclonal an- myc-tag decapeptide, binding of anti-myc-tag antibod-
tibodies to myoglobin covalently immobilised onto ies is reported [15.148].
444 Part B MEMS/NEMS and BioMEMS/NEMS
Lipid Bilayers, Liposomes, Cells showing that the feline coronavirus (FIP) type I virus
Cantilever array sensors can sense the formation by can be detected by a microcantilever modified by FIP
vesicle fusion of supported phospholipid bilayers of type I anti-viral antiserum. A method for quantification
1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) on of a prostate cancer biomarker in urine without sample
their surface and can monitor changes in mechani- preparation using monoclonal antibodies in described
cal properties of lipid bilayers [15.149]. Liposomes in [15.160].
are detected based on their interaction with protein
C2A which recognizes the phosphatidylserine (PS) ex- 15.7.3 Microcantilever Sensors
posed on the surface of liposome [15.150]. Individual to Measure Physical Properties
Escherichia coli (E. coli) O157:H7 cell–antibody bind-
ing events using microcantilevers operated in dynamic Besides chemical and biochemical sensing, microcan-
mode are reported [15.151]. The contractile force of tilevers can also detect changes in physical properties
self-organized cardiomyocytes was measured on bio- of surrounding media, such as gas or liquid, or of layers
compatible poly(dimethylsiloxane) cantilevers, repre- deposited on the cantilever itself.
senting a microscale cell-driven motor system [15.152].
Resonanting cantilevers were used to detect individual Density and Viscosity
phospholipid vesicle adsorption in liquid. A resonance A piezoelectric unimorph cantilever as a liquid
frequency shift corresponding to an added mass of viscosity-and-density sensor was tested using water-
450 pg has been measured [15.153]. glycerol solutions of different compositions, whereby
the resonance frequency decreased while the width
Part B 15.7
Spores, Bacteria and Viruses of the resonance peak increased with increasing glyc-
Micromechanical cantilever arrays have been used for erol content [15.161]. The viscosity of complex or-
quantitative detection of vital fungal spores of As- ganic liquids with non-Newtonian behavior is studied
pergillus niger and Saccharomyces cerevisiae. The in [15.162] in a wide range from 10 to 500 mm2 /s.
specific adsorption and growth on concanavalin A, fi- Simultaneous determination of density and viscosity
bronectin or immunoglobulin G cantilever surfaces was of water/ethanol mixtures based on resonance curves
investigated. Maximum spore immobilization, germi- of microcantilevers is reported in [15.163]. A detailed
nation and mycelium growth was observed on the theoretical study of viscoelastic effects on the fre-
immunoglobulin G functionalized cantilever surfaces, quency shift of microcantilever chemical sensors is
as measured from shifts in resonance frequency within given in [15.164]. Microcantilever deflection as a func-
a few hours, being much faster than with standard tion of flow speed of viscous fluids is investigated
Petri dish cultivation [15.154]. Short peptide ligands in [15.165]. Viscosity of sugar solutions is tested using
can be used to efficiently capture Bacillus subtilis microcantilevers [15.166].
(a simulant of Bacillus anthracis) spores in liquids.
Fifth-mode resonant frequency measurements were per- Gas and Flow Sensing
formed before and after dipping microcantilever arrays Gas sensing does not only involve chemical detection,
into a static B. subtilis solution showing a substantial but also pressure and flow sensing. Brown et al. [15.167]
decrease in frequency for binding-peptide-coated mi- have studied the behavior of magnetically actuated os-
crocantilevers as compared to that for control peptide cillating microcantilevers at large deflections and have
cantilevers [15.155]. found hysteresis behavior at resonance. The ampli-
tude at the actuation frequency changes depending on
Medical pressure due to damping. The authors have used can-
A bioassay of prostate-specific antigen (PSA) using tilever in cantilever (CIC) structures, and have observed
microcantilevers has been presented in [15.156], cov- changes in deflection as gas pressure is varied. At atmo-
ering a wide range of concentrations from 0.2 ng/ml spheric pressure, damping is large and the oscillation
to 60 μg/ml in a background of human serum albumin amplitude is relative small and hysteresis effects are
(HSA). Detection has been confirmed by another group absent. At lower pressure, abrupt changes in the os-
using microcantilevers in resonant mode [15.157, 158]. cillation amplitude occur with changes in the driving
The feasibility of detecting severe acute respiratory frequency. Since the change of amplitude and driving
syndrome associated coronavirus (SARS-CoV) using frequency, at which they occur are pressure dependent,
microcantilever technology is studied in [15.159] by these quantities can be used for accurate determina-
Nanomechanical Cantilever Array Sensors 15.8 Conclusions and Outlook 445
tion of gas pressure, demonstrated in the range between cantilever using a He-Ne laser in a Michelson inter-
10−3 and 102 Torr. Brown et al. [15.168] emphasize ferometer configuration, whereby the cantilever acts as
that microelectromechanical system pressure sensors moving mirror element in one path of the interferome-
will have a wide range of applications, especially in ter. A fixed mirror serves a reference in the other arm of
the automotive industry. Piezoresistive cantilever based the interferometer.
deflection measurement has major advantages over di-
aphragms. The pressure range has been extended to Thermal Expansion
15–1450 Torr by means of design geometry adapta- The thermal expansion of TaOx N y thin films deposited
tion. Su et al. [15.169] present highly sensitive ultrathin on a microcantilever was measured to examine the
piezoresistive silicon microcantilevers for gas velocity residual stress and the thermal expansion coefficient by
sensing, whereby the deflection increases with airflow observsing the changes in radius of curvature [15.173].
distribution in a steel pipe. The detection principle is Thermal drift issues of resonaniting microcantilevers
based on normal pressure drag producing bending of are discussed in detail in [15.174].
the cantilever. The minimum flow speed measured was
7.0 cm/s, which is comparable to classical hot-wire Infrared Imaging
anemometers. Mertens et al. [15.170] have investigated Microcantilevers can also be used as uncooled,
the effects of temperature and pressure on microcan- microcantilever-based infrared (IR) imaging devices
tilever resonance response in helium and nitrogen. by monitoring the bending of the microcantilever as
Resonance response as a function of pressure showed a function of the IR radiation intensity incident on the
three different regimes, which correspond to molecu- cantilever surface. The infrared (thermal) image of the
Part B 15.8
lar flow, transition regimes and viscous flow, whereby source is obtained by rastering a single microfabricated
the frequency variation of the cantilever is mainly due cantilever across the image formed at the focal plane of
to changes in the mean free path of gas molecules. Ef- a concave mirror [15.175–177]. The method has later
fects observed allow measurement of pressures between been refined such that photons are detected using the
7.5 × 10−5 and 7500 Torr. Mortet et al. [15.171] present stress caused by photoelectrons emitted from a Pt film
a pressure sensor based on a piezoelectric bimorph mi- surface in contact with a semiconductor microstruc-
crocantilever with a measurement range between 75 and ture, which forms a Schottky barrier. The photoinduced
6400 Torr. The resonance frequency shift is constant for bending of the Schottky barrier microstructure is due
pressures below 375 Torr. For higher pressures the sen- to electronic stress produced by photoelectrons diffus-
sitivity is typically a few ppm/mbar, but depends on ing into the microstructure [15.178]. The performance
the mode number. Sievilä et al. [15.172] present a can- of IR imaging via microcantilevers has been enhanced
tilever paddle within a frame operating like a moving by one-fold leg and two-fold legs beam structures with
mirror to detect the displacements in the oscillating absorber plates [15.179–181].
drag [15.182]. Biochemical selectivity can be en- tion [15.184]. Miniaturization of microcantilevers into
hanced by using enantioselective receptors [15.183]. true nanometric dimensions, like by using single wall
Other shapes for micromechanical sensors like mi- carbon nanotubes [15.185] or graphene sheets [15.186]
crospirals could be advantageous for biochemical detec- will further increase sensitivity.
References
15.1 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi- 15.16 F.J. von Preissig: Applicability of the classical
croscope, Phys. Rev. Lett. 56, 930–933 (1986) curvature-stress relation for thin films on plate
15.2 R.J. Wilfinger, P.H. Bardell, D.S. Chhabra: The substrates, J. Appl. Phys. 66, 4262–4268 (1989)
resonistor, a frequency sensitive device utilizing 15.17 J. Fritz, M.K. Baller, H.P. Lang, H. Rothuizen,
the mechanical resonance of a silicon substrate, P. Vettiger, E. Meyer, H.-J. Güntherodt, C. Gerber,
IBM J. 12, 113–118 (1968) J.K. Gimzewski: Translating biomolecular recog-
15.3 T.M.S. Heng: Trimming of microstrip circuits utiliz- nition into nanomechanics, Science 288, 316–318
ing microcantilever air gaps, IEEE Trans. Microw. (2000)
Theory Technol. 19, 652–654 (1971) 15.18 M. Alvarez, L.G. Carrascosa, M. Moreno, A. Calle,
15.4 K.E. Petersen: Micromechanical membrane swit- A. Zaballos, L.M. Lechuga, C.-A. Martinez,
ches on silicon, IBM J. Res. Dev. 23, 376–385 (1979) J. Tamayo: Nanomechanics of the formation of
15.5 E.S. Kolesar: Electronic nerve agent detector, US DNA self-assembled monolayers, hybridization on
Patent 4549427 (1983) microcantilevers, Langmuir 20, 9663–9668 (2004)
15.6 T. Itoh, T. Suga: Force sensing microcantilever using 15.19 D. Sarid: Scanning Force Microscopy with Appli-
Part B 15
sputtered zinc-oxide thin-film, Appl. Phys. Lett. cations to Electric, Magnetic, and Atomic Forces
64, 37–39 (1994) (Oxford Univ. Press, New York 1991)
15.7 J.P. Cleveland, S. Manne, D. Bocek, P.K. Hansma: 15.20 R. Berger, H.P. Lang, C. Gerber, J.K. Gimzewski,
A nondestructive method for determining the J.H. Fabian, L. Scandella, E. Meyer, H.-
spring constant of cantilevers for scanning force J. Güntherodt: Micromechanical thermogravime-
microscopy, Rev. Sci. Instrum. 64, 403–405 try, Chem. Phys. Lett. 294, 363–369 (1998)
(1993) 15.21 J.E. Sader: Frequency response of cantilever beams
15.8 T. Thundat, R.J. Warmack, G.Y. Chen, D.P. Allison: immersed in viscous fluids with applications to the
Thermal, ambient-induced deflections of scanning atomic force microscope, J. Appl. Phys. 84, 64–76
force microscope cantilevers, Appl. Phys. Lett. 64, (1998)
2894–2896 (1994) 15.22 T. Bachels, F. Tiefenbacher, R. Schafer: Conden-
15.9 J.K. Gimzewski, C. Gerber, E. Meyer, R.R. Schlit- sation of isolated metal clusters studied with
tler: Observation of a chemical reaction using a calorimeter, J. Chem. Phys. 110, 10008–10015
a micromechanical sensor, Chem. Phys. Lett. 217, (1999)
589–594 (1994) 15.23 J.R. Barnes, R.J. Stephenson, C.N. Wood-
15.10 T. Thundat, G.Y. Chen, R.J. Warmack, D.P. Allison, burn, S.J. O’Shea, M.E. Welland, T. Rayment,
E.A. Wachter: Vapor detection using resonating mi- J.K. Gimzewski, C. Gerber: A femtojoule calorimeter
crocantilevers, Anal. Chem. 67, 519–521 (1995) using micromechanical sensors, Rev. Sci. Instrum.
15.11 H.P. Lang, R. Berger, C. Andreoli, J. Brugger, 65, 3793–3798 (1994)
M. Despont, P. Vettiger, C. Gerber, J.K. Gimzewski, 15.24 T.A. Brunt, T. Rayment, S.J. O’Shea, M.E. Welland:
J.-P. Ramseyer, E. Meyer, H.-J. Güntherodt: Measuring the surface stresses in an electrochemi-
Sequential position readout from arrays of mi- cally deposited monolayer: Pb on Au(111), Langmuir
cromechanical cantilever sensors, Appl. Phys. Lett. 12, 5942–5946 (1996)
72, 383–385 (1998) 15.25 R. Puers, D. Lapadatu: Electrostatic forces and their
15.12 H. Ibach: Adsorbate-induced surface stress, J. Vac. effects on capacitive mechanical sensors, Sens. Ac-
Sci. Technol. A 12, 2240–2245 (1994) tuators A 56, 203–210 (1996)
15.13 R. Berger, E. Delamarche, H.P. Lang, C. Gerber, 15.26 J. Fricke, C. Obermaier: Cantilever beam accelerom-
J.K. Gimzewski, E. Meyer, H.-J. Güntherodt: Sur- eter based on surface micromachining technology,
face stress in the self-assembly of alkanethiols on J. Micromech. Microeng. 3, 190–192 (1993)
gold, Science 276, 2021–2024 (1997) 15.27 C. Lee, T. Itoh, T. Ohashi, R. Maeda, T. Suga:
15.14 J.R. Barnes, R.J. Stephenson, M.E. Welland, C. Ger- Development of a piezoelectric self-excitation,
ber, J.K. Gimzewski: Photothermal spectroscopy self-detection mechanism in PZT microcantilevers
with femtojoule sensitivity based on microme- for dynamic scanning force microscopy in liquid,
chanics, Nature 372, 79–81 (1994) J. Vac. Sci. Technol. B 15, 1559–1563 (1997)
15.15 G.G. Stoney: The tension of metallic films deposited 15.28 T. Göddenhenrich, H. Lemke, U. Hartmann, C. Hei-
by electrolysis, Proc. R. Soc. 82, 172–177 (1909) den: Force microscope with capacitive displace-
Nanomechanical Cantilever Array Sensors References 447
ment detection, J. Vac. Sci. Technol. A 8, 383–387 on a nanomechanical cantilever array, Proc. Natl.
(1990) Acad. Sci. USA 99, 9783–9787 (2002)
15.29 J. Brugger, R.A. Buser, N.F. de Rooij: Microma- 15.41 Y. Arntz, J.D. Seelig, H.P. Lang, J. Zhang, P. Hun-
chined atomic force microprobe with integrated ziker, J.-P. Ramseyer, E. Meyer, M. Hegner,
capacitive read-out, J. Micromech. Microeng. 2, C. Gerber: Label-free protein assay based on
218–220 (1992) a nanomechanical cantilever array, Nanotechnol-
15.30 C. Schönenberger, S.F. Alvarado: A differential ogy 14, 86–90 (2003)
interferometer for force microscopy, Rev. Sci. In- 15.42 A. Bietsch, M. Hegner, H.P. Lang, C. Gerber: Inkjet
strum. 60, 3131–3134 (1989) deposition of alkanethiolate monolayers and DNA
15.31 M.J. Cunningham, S.T. Cheng, W.W. Clegg: A dif- oligonucleotides on gold: Evaluation of spot uni-
ferential interferometer for scanning force mi- formity by wet etching, Langmuir 20, 5119–5122
croscopy, Meas. Sci. Technol. 5, 1350–1354 (1994) (2004)
15.32 G. Meyer, N.M. Amer: Novel optical approach to 15.43 A. Bietsch, J. Zhang, M. Hegner, H.P. Lang, C. Ger-
atomic force microscopy, Appl. Phys. Lett. 53, ber: Rapid functionalization of cantilever array
2400–2402 (1988) sensors by inkjet printing, Nanotechnology 15,
15.33 R. Berger, C. Gerber, J.K. Gimzewski, E. Meyer, 873–880 (2004)
H.-J. Güntherodt: Thermal analysis using a mi- 15.44 D. Lange, C. Hagleitner, A. Hierlemann, O. Brand,
cromechanical calorimeter, Appl. Phys. Lett. 69, H. Baltes: Complementary metal oxide semi-
40–42 (1996) conductor cantilever arrays on a single chip:
15.34 L. Scandella, G. Binder, T. Mezzacasa, J. Gobrecht, mass-sensitive detection of volatile organic com-
R. Berger, H.P. Lang, C. Gerber, J.K. Gimzewski, pounds, Anal. Chem. 74, 3084–3085 (2002)
J.H. Koegler, J.C. Jansen: Combination of single 15.45 C.A. Savran, T.P. Burg, J. Fritz, S.R. Manalis: Micro-
Part B 15
crystal zeolites and microfabrication: Two appli- fabricated mechanical biosensor with inherently
cations towards zeolite nanodevices, Microporous differential readout, Appl. Phys. Lett. 83, 1659–
Mesoporous Mater. 21, 403–409 (1998) 1661 (2003)
15.35 R. Berger, C. Gerber, H.P. Lang, J.K. Gimzewski: 15.46 S. Okuyama, Y. Mitobe, K. Okuyama, K. Matsushita:
Micromechanics: a toolbox for femtoscale science: Hydrogen gas sensing using a Pd-coated can-
towards a laboratory on a tip, Microelectron. Eng. tilever, Jpn. J. Appl. Phys. 39, 3584–3590 (2000)
35, 373–379 (1997) 15.47 A. Fabre, E. Finot, J. Demoment, S. Contreras: Mon-
15.36 H.P. Lang, R. Berger, F.M. Battiston, J.-P. Ram- itoring the chemical changes in Pd induced by
seyer, E. Meyer, C. Andreoli, J. Brugger, P. Vettiger, hydrogen absorption using microcantilevers, Ul-
M. Despont, T. Mezzacasa, L. Scandella, H.-J. Gün- tramicroscopy 97, 425–432 (2003)
therodt, C. Gerber, J.K. Gimzewski: A chemical 15.48 D.R. Baselt, B. Fruhberger, E. Klaassen, S. Ce-
sensor based on a micromechanical cantilever ar- malovic, C.L. Britton, S.V. Patel, T.E. Mlsna,
ray for the identification of gases and vapors, Appl. D. McCorkle, B. Warmack: Design and performance
Phys. A 66, 61–64 (1998) of a microcantilever-based hydrogen sensor, Sens.
15.37 M.K. Baller, H.P. Lang, J. Fritz, C. Gerber, Actuator. B Chem. 88, 120–131 (2003)
J.K. Gimzewski, U. Drechsler, H. Rothuizen, M. De- 15.49 Y.-I. Chou, H.-C. Chiang, C.-C. Wang: Study on Pd
spont, P. Vettiger, F.M. Battiston, J.-P. Ramseyer, functionalization of microcantilever for hydrogen
P. Fornaro, E. Meyer, H.-J. Güntherodt: A cantilever detection promotion, Sens. Actuator. B Chem. 129,
array based artificial nose, Ultramicroscopy 82, 1–9 72–78 (2008)
(2000) 15.50 L. Scandella, G. Binder, T. Mezzacasa, J. Gobrecht,
15.38 S. Cesaro-Tadic, G. Dernick, D. Juncker, G. Bu- R. Berger, H.P. Lang, C. Gerber, J.K. Gimzewski,
urman, H. Kropshofer, B. Michel, C. Fattinger, J.H. Koegler, J.C. Jansen: Combination of single
E. Delamarche: High-sensitivity miniaturized im- crystal zeolites and microfabrication: Two appli-
munoassays for tumor necrosis factor α using cations towards zeolite nanodevices, Microporous
microfluidic systems, Lab Chip 4, 563–569 (2004) Mesoporous Mater. 21, 403–409 (1998)
15.39 J. Fritz, M.K. Baller, H.P. Lang, T. Strunz, E. Meyer, 15.51 R.L. Gunter, W.D. Delinger, T.L. Porter, R. Stewart,
H.-J. Güntherodt, E. Delamarche, C. Gerber, J. Reed: Hydration level monitoring using embed-
J.K. Gimzewski: Stress at the solid-liquid interface ded piezoresistive microcantilever sensors, Med.
of self-assembled monolayers on gold investi- Eng. Phys. 27, 215–220 (2005)
gated with a nanomechanical sensor, Langmuir 16, 15.52 X. Shi, Q. Chen, J. Fang, K. Varahramyan, H.F. Ji:
9694–9696 (2000) Al2 O3 -coated microcantilevers for detection of
15.40 R. McKendry, J. Zhang, Y. Arntz, T. Strunz, M. Heg- moisture at ppm level, Sens. Actuator. B Chem. 129,
ner, H.P. Lang, M.K. Baller, U. Certa, E. Meyer, 241–245 (2008)
H.-J. Güntherodt, C. Gerber: Multiple label-free 15.53 J. Zhou, P. Li, S. Zhang, Y.C. Long, F. Zhou,
biodetection and quantitative DNA-binding assays Y.P. Huang, P.Y. Yang, M.H. Bao: Zeolite-modified
448 Part B MEMS/NEMS and BioMEMS/NEMS
microcantilever gas sensor for indoor air quality 15.67 A.R. Kadam, G.P. Nordin, M.A. George: Comparison
control, Sens. Actuator. B Chem. 94, 337–342 (2003) of microcantilever Hg sensing behavior with ther-
15.54 A. Vidic, D. Then, C. Ziegler: A new cantilever sys- mal higher order modes for as-deposited sputtered
tem for gas and liquid sensing, Ultramicroscopy 97, and thermally evaporated Au films, J. Vac. Sci.
407–416 (2003) Technol. B 24, 2271–2275 (2006)
15.55 R. Berger, E. Delamarche, H.P. Lang, C. Gerber, 15.68 H.F. Ji, E. Finot, R. Dabestani, T. Thundat,
J.K. Gimzewski, E. Meyer, H.J. Güntherodt: Sur- G.M. Brown, P.F. Britt: A novel self-assembled
face stress in the self-assembly of alkanethiols on monolayer (SAM) coated microcantilever for low
gold probed by a force microscopy technique, Appl. level caesium detection, Chem. Commun. 36(6),
Phys. A 66, S55–S59 (1998) 457–458 (2000)
15.56 P.G. Datskos, I. Sauers: Detection of 2-mer- 15.69 T.A. Brunt, T. Rayment, S.J. O’Shea, M.E. Welland:
captoethanol using gold-coated micromachined Measuring the surface stresses in an electrochem-
cantilevers, Sens. Actuator. B Chem. 61, 75–82 ically deposited metal monolayer: Pb on Au(111),
(1999) Langmuir 12, 5942–5946 (1996)
15.57 A.N. Itakura, R. Berger, T. Narushima, M. Kitajima: 15.70 J. Mertens, E. Finot, M.H. Nadal, V. Eyraud,
Low-energy ion-induced tensile stress of self- O. Heintz, E. Bourillot: Detection of gas trace of
assembled alkanethiol monolayers, Appl. Phys. hydrofluoric acid using microcantilever, Sens. Ac-
Lett. 80, 3712–3714 (2002) tuator. B Chem. 99, 58–65 (2004)
15.58 M. Godin, P.J. Williams, V. Tabard-Cossa, 15.71 Y.J. Tang, J. Fang, X.H. Xu, H.F. Ji, G.M. Brown,
O. Laroche, L.Y. Beaulieu, R.B. Lennox, P. Grutter: T. Thundat: Detection of femtomolar concentra-
Surface stress, kinetics, and structure of alka- tions of HF using an SiO2 microcantilever, Anal.
nethiol self-assembled monolayers, Langmuir 20, Chem. 76, 2478–2481 (2004)
Part B 15
Cr(VI) using a microcantilever electrode coated with 15.93 D. Then, A. Vidic, C. Ziegler: A highly sensitive
a self-assembled monolayer, J. Vac. Sci. Technol. A self-oscillating cantilever array for the quantitative
23, 1022–1028 (2005) and qualitative analysis of organic vapor mixtures,
15.81 V.I. Boiadjiev, G.M. Brown, L.A. Pinnaduwage, Sens. Actuator. B Chem. 117, 1–9 (2006)
G. Goretzki, P.V. Bonnesen, T. Thundat: Photo- 15.94 P.J. Chapman, F. Vogt, P. Dutta, P.G. Datskos,
chemical hydrosilylation of 11-undecenyltriethylammonium G.L. Devault, M.J. Sepianiak: Facile hyphenation
bromide with hydrogen-terminated Si surfaces for of gas chromatography and a microcantilever ar-
the development of robust microcantilever sensors ray sensor for enhanced selectivity, Anal. Chem.
for Cr(VI), Langmuir 21, 1139–1142 (2005) 79, 364–370 (2007)
15.82 N.A. Carrington, L. Yong, Z.-L. Xue: Electrochem- 15.95 R. Archibald, P. Datskos, G. Devault, V. Lamberti,
ical deposition of sol-gel films for enhanced N. Lavrik, D. Noid, M. Sepianiak, P. Dutta: Inde-
chromium(VI) determination in aqueous solutions, pendent component analysis of nanomechanical
Anal. Chim. Acta 572, 17–24 (2006) responses of cantilever arrays, Anal. Chim. Acta
15.83 S. Velanki, S. Kelly, T. Thundat, D.A. Blake, H.F. Ji: 584, 101–105 (2007)
Detection of Cd(II) using antibody-modified micro- 15.96 M. Alvarez, A. Calle, J. Tamayo, L.M. Lechuga,
cantilever sensors, Ultramicroscopy 107, 1123–1128 A. Abad, A. Montoya: Development of nanome-
(2007) chanical biosensors for detection of the pesticide
15.84 H. Jensenius, J. Thaysen, A.A. Rasmussen, L.H. Veje, DDT, Biosensor. Bioelectron. 18, 649–653 (2003)
O. Hansen, A. Boisen: A microcantilever-based 15.97 P.G. Datskos, M.J. Sepaniak, C.A. Tipple, N. Lavrik:
alcohol vapor sensor-application and response Photomechanical chemical microsensors, Sens. Ac-
model, Appl. Phys. Lett. 76, 2815–2817 (2000) tuator. B Chem. 76, 393–402 (2001)
15.85 A. Hierlemann, D. Lange, C. Hagleitner, N. Kerness, 15.98 L.A. Pinnaduwage, V. Boiadjiev, J.E. Hawk,
Part B 15
A. Koll, O. Brand, H. Baltes: Application-specific T. Thundat: Sensitive detection of plastic explosives
sensor systems based on CMOS chemical microsen- with self-assembled monolayer-coated microcan-
sors, Sens. Actuator. B Chem. 70, 2–11 (2000) tilevers, Appl. Phys. Lett. 83, 1471–1473 (2003)
15.86 D. Lange, C. Hagleitner, A. Hierlemann, O. Brand, 15.99 L.A. Pinnaduwage, A. Gehl, D.L. Hedden, G. Mu-
H. Baltes: Complementary metal oxide semi- ralidharan, T. Thundat, R.T. Lareau, T. Sulchek,
conductor cantilever arrays on a single chip: L. Manning, B. Rogers, M. Jones, J.D. Adams: A mi-
Mass-sensitive detection of volatile organic com- crosensor for trinitrotoluene vapour, Nature 425,
pounds, Anal. Chem. 74, 3084–3095 (2002) 474–474 (2003)
15.87 C. Vancura, M. Rüegg, Y. Li, C. Hagleitner, A. Hi- 15.100 L.A. Pinnaduwage, D. Yi, F. Tian, T. Thundat,
erlemann: Magnetically actuated complementary L.T. Lareau: Adsorption of trinitrotoluene on un-
metal oxide semiconductor resonant cantilever gas coated silicon microcantilever surfaces, Langmuir
sensor systems, Anal. Chem. 77, 2690–2699 (2005) 20, 2690–2694 (2004)
15.88 L. Fadel, F. Lochon, I. Dufour, O. Français: Chemical 15.101 L.A. Pinnaduwage, A. Wig, D.L. Hedden, A. Gehl,
sensing: Millimeter size resonant microcantilever D. Yi, T. Thundat, R.T. Lareau: Detection of trini-
performance, J. Micromech. Microeng. 14, S23–S30 trotoluene via deflagration on a microcantilever,
(2004) J. Appl. Phys. 95, 5871–5875 (2004)
15.89 Y.I. Wright, A.K. Kar, Y.W. Kim, C. Scholz, 15.102 L.A. Pinnaduwage, T. Thundat, A. Gehl, S.D. Wilson,
M.A. George: Study of microcapillary pipette- D.L. Hedden, R.T. Lareau: Desorption characteris-
assisted method to prepare polyethylene glycol- tics, of un-coated silicon microcantilever surfaces
coated microcantilever sensors, Sens. Actuator. B for explosive and common nonexplosive vapors,
107, 242–251 (2005) Ultramicroscopy 100, 211–216 (2004)
15.90 L.R. Senesac, P. Dutta, P.G. Datskos, M.J. Sepianiak: 15.103 L.A. Pinnaduwage, T. Thundat, J.E. Hawk,
Analyte species and concentration identification D.L. Hedden, R. Britt, E.J. Houser, S. Step-
using differentially functionalized microcantilever nowski, R.A. McGill, D. Bubb: Detection of 2,4-
arrays and artificial neural networks, Anal. Chim. dinitrotoluene using microcantilever sensors, Sens.
Acta 558, 94–101 (2006) Actuator. B 99, 223–229 (2004)
15.91 F. Lochon, L. Fadel, I. Dufour, D. Rebière, J. Pistré: 15.104 I. Voiculescu, M.E. Zaghloul, R.A. McGill, E.J. Houser,
Silicon made resonant microcantilever: Depen- G.K. Fedder: Electrostatically actuated resonant
dence of the chemical sensing performances on microcantilever beam in CMOS technology for the
the sensitive coating thickness, Mater. Sci. Eng. C detection of chemical weapons, IEEE Sensor. J. 5,
26, 348–353 (2006) 641–647 (2005)
15.92 S. Satyanarayana, D.T. McCormick, A. Majumdar: 15.105 L.A. Pinnaduwage, H.F. Ji, T. Thundat: Moore’s
Parylene micro membrane capacitive sensor array law in homeland defense: An integrated sensor
for chemical and biological sensing, Sens. Actuator. platform based on silicon microcantilevers, IEEE
B Chem. 115, 494–502 (2006) Sensor. J. 5, 774–785 (2005)
450 Part B MEMS/NEMS and BioMEMS/NEMS
15.106 Y.M. Yang, H.F. Ji, T. Thundat: Nerve agents de- manipulation of DNA on microcantilever surfaces,
tection using a Cu2+ /L-cysteine bilayer-coated Langmuir 18, 8732–8736 (2002)
microcantilever, J. Am. Chem. Soc. 125, 1124–1125 15.121 M. Su, S.U. Li, V.P. Dravid: Microcantilever
(2003) resonance-based DNA detection with nanopar-
15.107 H.F. Ji, X.D. Yan, J. Zhang, T. Thundat: Mo- ticle probes, Appl. Phys. Lett. 82, 3562–3564
lecular recognition of biowarfare agents using (2003)
micromechanical sensors, Expert Rev. Mol. Diagn. 15.122 M. Alvarez, L.G. Carrascosa, M. Moreno, A. Calle,
4, 859–866 (2004) A. Zaballos, L.M. Lechuga, C. Martinez, J. Tamayo:
15.108 H.F. Ji, K.M. Hansen, Z. Hu, T. Thundat: Detec- Nanomechanics of the formation of DNA self-
tion of pH variation using modified microcantilever assembled monolayers and hybridization on mi-
sensors, Sens. Actuator. B Chem. 72, 233–238 (2001) crocantilevers, Langmuir 20, 9663–9668 (2004)
15.109 R. Bashir, J.Z. Hilt, O. Elibol, A. Gupta, N.A. Peppas: 15.123 R.L. Gunter, R. Zhine, W.G. Delinger, K. Manygoats,
Micromechanical cantilever as an ultrasensitive pH A. Kooser, T.L. Porter: Investigation of DNA sensing
microsensor, Appl. Phys. Lett. 81, 3091–3093 (2002) using piezoresistive microcantilever probes, IEEE
15.110 Y.F. Zhang, H.F. Ji, D. Snow, R. Sterling, Sensor. J. 4, 430–433 (2004)
G.M. Brown: A pH sensor based on a microcan- 15.124 M. Alvarez, J. Tamayo: Optical sequential readout
tilever coated with intelligent hydrogel, Instrum. of microcantilever arrays for biological detection,
Sci. Technol. 32, 361–369 (2004) Sens. Actuator. B Chem. 106, 687–690 (2005)
15.111 M. Watari, J. Galbraith, H.P. Lang, M. Sousa, 15.125 M. Calleja, M. Nordstrom, M. Alvarez, J. Tamayo,
M. Hegner, C. Gerber, M.A. Horton, R.A. McK- L.M. Lechuga, A. Boisen: Highly sensitive polymer-
endry: Investigating the molecular mechanisms of based cantilever-sensors for DNA detection, Ultra-
in-plane mechanochemistry on cantilever arrays, microscopy 105, 215–222 (2005)
Part B 15
J. Am. Chem. Soc. 129, 601–609 (2007) 15.126 R. Mukhopadhyay, M. Lorentzen, J. Kjems, F. Be-
15.112 A. Subramanian, P.I. Oden, S.J. Kennel, K.B. Ja- senbacher: Nanomechanical sensing of DNA se-
cobson, R.J. Warmack, T. Thundat, M.J. Doktycz: quences using piezoresistive cantilevers, Langmuir
Glucose biosensing using an enzyme-coated mi- 21, 8400–8408 (2005)
crocantilever, Appl. Phys. Lett. 81, 385–387 (2002) 15.127 L.M. Lechuga, J. Tamayo, M. Alvarez, L.G. Carras-
15.113 X.D. Yan, H.F. Ji, Y. Lvov: Modification of microcan- cosa, A. Yufera, R. Doldan, E. Peralias, A. Rueda,
tilevers using layer-by-layer nanoassembly film J.A. Plaza, K. Zinoviev, C. Dominguez, A. Zabal-
for glucose measurement, Chem. Phys. Lett. 396, los, M. Moreno, C. Martinez, D. Wenn, N. Harris,
34–37 (2004) C. Bringer, V. Bardinal, T. Camps, C. Vergnenegre,
15.114 J.H. Pei, F. Tian, T. Thundat: Glucose biosensor C. Fontaine, V. Diaz, A. Bernad: A highly sensitive
based on the microcantilever, Anal. Chem. 76, 292– microsystem based on nanomechanical biosensors
297 (2004) for genomics applications, Sens. Actuator. B Chem.
15.115 X.D. Yan, X.H.K. Xu, H.F. Ji: Glucose oxidase 118, 2–10 (2006)
multilayer modified microcantilevers for glucose 15.128 S.L. Biswal, D. Raorane, A. Chaiken, H. Birecki,
measurement, Anal. Chem. 77, 6197–6204 (2005) A. Majumdar: Nanomechanical detection of DNA
15.116 X.D. Yan, X.L. Shi, K. Hill, H.F. Ji: Microcantilevers melting on microcantilever surfaces, Anal. Chem.
modified by horseradish peroxidase intercalated 78, 7104–7109 (2006)
nano-assembly for hydrogen peroxide detection, 15.129 J.C. Stachowiak, M. Yue, K. Castelino, A. Chakraborty,
Anal. Sci. 22, 205–208 (2006) A. Majumdar: Chemomechanics of surface stresses
15.117 K.M. Hansen, H.F. Ji, G.H. Wu, R. Datar, R. Cote, induced by DNA hybridization, Langmuir 22, 263–
A. Majumdar, T. Thundat: Cantilever-based optical 268 (2006)
deflection assay for discrimination of DNA single- 15.130 F. Huber, M. Hegner, C. Gerber, H.J. Guntherodt,
nucleotide mismatches, Anal. Chem. 73, 1567–1571 H.P. Lang: Label free analysis of transcription
(2001) factors using microcantilever arrays, Biosens. Bio-
15.118 M.F. Hagan, A. Majumdar, A.K. Chakraborty: electron. 21, 1599–1605 (2006)
Nanomechanical forces generated by surface 15.131 S.L. Biswal, D. Raorane, A. Chaiken, A. Majumdar:
grafted DNA, J. Phys. Chem. B 106, 10163–10173 Using a microcantilever array for detecting phase
(2002) transitions and stability of DNA, Clin. Lab. Med. 27,
15.119 R. Marie, H. Jensenius, J. Thaysen, C.B. Christensen, 163–163 (2007)
A. Boisen: Adsorption kinetics and mechanical 15.132 J. Zhang, H.P. Lang, F. Huber, A. Bietsch,
properties of thiol-modified DNA-oligos on gold W. Grange, U. Certa, R. McKendry, H.-J. Gün-
investigated by microcantilever sensors, Ultrami- therodt, M. Hegner, C. Gerber: Rapid and label-free
croscopy 91, 29–36 (2002) nanomechanical detection of biomarker tran-
15.120 K.A. Stevenson, A. Mehta, P. Sachenko, K.M. Hansen, scripts in human RNA, Nat. Nanotechnol. 1, 214–220
T. Thundat: Nanomechanical effect of enzymatic (2006)
Nanomechanical Cantilever Array Sensors References 451
15.133 A.M. Moulin, S.J. O’Shea, M.E. Welland: Microcanti- piezoresistive cantilever arrays, Nanotechnology
lever-based biosensors, Ultramicroscopy 82, 23–31 18, 125503 (2007)
(2000) 15.146 M. Yue, J.C. Stachowiak, H. Lin, R. Datar, R. Cote,
15.134 C. Grogan, R. Raiteri, G.M. O’Connor, T.J. Glynn, A. Majumdar: Label-free protein recognition two-
V. Cunningham, M. Kane, M. Charlton, D. Leech: dimensional array using nanomechanical sensors,
Characterisation of an antibody coated microcan- Nano Lett. 8, 520–524 (2008)
tilever as a potential immuno-based biosensor, 15.147 T. Braun, N. Backmann, M. Vogtli, A. Bietsch, A. En-
Biosens. Bioelectron. 17, 201–207 (2002) gel, H.P. Lang, C. Gerber, M. Hegner: Conforma-
15.135 A. Kooser, K. Manygoats, M.P. Eastman, T.L. Porter: tional change of bacteriorhodopsin quantitatively
Investigation of the antigen antibody reaction be- monitored by microcantilever sensors, Biophys. J.
tween anti-bovine serum albumin (a-BSA) and 90, 2970–2977 (2006)
bovine serum albumin (BSA) using piezoresistive 15.148 B.H. Kim, O. Mader, U. Weimar, R. Brock, D.P. Kern:
microcantilever based sensors, Biosens. Bioelec- Detection of antibody peptide interaction using
tron. 19, 503–508 (2003) microcantilevers as surface stress sensors, J. Vac.
15.136 C.A. Savran, S.M. Knudsen, A.D. Ellington, Sci. Technol. B 21, 1472–1475 (2003)
S.R. Manalis: Micromechanical detection of pro- 15.149 I. Pera, J. Fritz: Sensing lipid bilayer formation and
teins using aptamer-based receptor molecules, expansion with a microfabricated cantilever array,
Anal. Chem. 76, 3194–3198 (2004) Langmuir 23, 1543–1547 (2007)
15.137 J.H. Lee, K.H. Yoon, K.S. Hwang, J. Park, S. Ahn, 15.150 S.J. Hyun, H.S. Kim, Y.J. Kim, H.I. Jung: Mechanical
T.S. Kim: Label free novel electrical detection using detection of liposomes using piezoresistive can-
micromachined PZT monolithic thin film cantilever tilever, Sens. Actuator. B Chem. 117, 415–419 (2006)
for the detection of C-reactive protein, Biosens. 15.151 B. Ilic, D. Czaplewski, M. Zalalutdinov, H.G. Craig-
Part B 15
Bioelectron. 20, 269–275 (2004) head, P. Neuzil, C. Campagnolo, C. Batt: Single cell
15.138 Y.F. Zhang, S.P. Venkatachalan, H. Xu, X.H. Xu, detection with micromechanical oscillators, J. Vac.
P. Joshi, H.F. Ji, M. Schulte: Micromechanical Sci. Technol. B 19, 2825–2828 (2001)
measurement of membrane receptor binding for 15.152 J. Park, J. Ryu, S.K. Choi, E. Seo, J.M. Cha, S. Ryu,
label-free drug discovery, Biosens. Bioelectron. 19, J. Kim, B. Kim, S.H. Lee: Real-time measure-
1473–1478 (2004) ment of the contractile forces of self-organized
15.139 N. Backmann, C. Zahnd, F. Huber, A. Bietsch, cardiomyocytes on hybrid biopolymer microcan-
A. Pluckthun, H.P. Lang, H.J. Güntherodt, M. Heg- tilevers, Anal. Chem. 77, 6571–6580 (2005)
ner, C. Gerber: A label-free immunosensor array 15.153 S. Ghatnekar-Nilsson, J. Lindahl, A. Dahlin,
using single-chain antibody fragments, Proc. Natl. T. Stjernholm, S. Jeppesen, F. Hook, L. Montelius:
Acad. Sci. USA 102, 14587–14592 (2005) Phospholipid vesicle adsorption measured in situ
15.140 K.W. Wee, G.Y. Kang, J. Park, J.Y. Kang, D.S. Yoon, with resonating cantilevers in a liquid cell, Nano-
J.H. Park, T.S. Kim: Novel electrical detection of technology 16, 1512–1516 (2005)
label-free disease marker proteins using piezore- 15.154 N. Nugaeva, K.Y. Gfeller, N. Backmann, H.P. Lang,
sistive self-sensing micro-cantilevers, Biosens. M. Duggelin, M. Hegner: Micromechanical can-
Bioelectron. 20, 1932–1938 (2005) tilever array sensors for selective fungal immo-
15.141 R. Mukhopadhyay, V.V. Sumbayev, M. Lorentzen, bilization and fast growth detection, Biosens.
J. Kjems, P.A. Andreasen, F. Besenbacher: Can- Bioelectron. 21, 849–856 (2005)
tilever sensor for nanomechanical detection of 15.155 B. Dhayal, W.A. Henne, D.D. Doorneweerd,
specific protein conformations, Nano Lett. 5, 2385– R.G. Reifenberger, P.S. Low: Detection of Bacillus
2388 (2005) subtilis spores using peptide-functionalized can-
15.142 X. Yan, K. Hill, H. Gao, H.F. Ji: Surface stress changes tilever arrays, J. Am. Chem. Soc. 128, 3716–3721
induced by the conformational change of proteins, (2006)
Langmuir 22, 11241–11244 (2006) 15.156 G. Wu, R.H. Datar, K.M. Hansen, T. Thundat,
15.143 H.S. Kwon, K.C. Han, K.S. Hwang, J.H. Lee, T.S. Kim, R.J. Cote, A. Majumdar: Bioassay of prostate-
D.S. Yoon, E.G. Yang: Development of a pep- specific antigen (PSA) using microcantilevers, Nat.
tide inhibitor-based cantilever sensor assay for Biotechnol. 19, 856–860 (2001)
cyclic adenosine monophosphate-dependent pro- 15.157 K.S. Hwang, J.H. Lee, J. Park, D.S. Yoon,
tein kinase, Anal. Chim. Acta, 585, 344–349 (2007) J.H. Park, T.S. Kim: In-situ quantitative analysis of
15.144 W. Shu, E.D. Laue, A.A. Seshia: Investigation of a prostate-specific antigen (PSA) using a nanome-
biotin-streptavidin binding interactions using mi- chanical PZT cantilever, Lab. Chip. 4, 547–552
crocantilever sensors, Biosens. Bioelectron. 22, (2004)
2003–2009 (2007) 15.158 J.H. Lee, K.S. Hwang, J. Park, K.H. Yoon, D.S. Yoon,
15.145 V. Dauksaite, M. Lorentzen, F. Besenbacher, T.S. Kim: Immunoassay of prostate-specific anti-
J. Kjems: Antibody-based protein detection using gen (PSA) using resonant frequency shift of piezo-
452 Part B MEMS/NEMS and BioMEMS/NEMS
I. Dufour, D. Rebière: Study of viscoelastic effect 15.178 P.G. Datskos, S. Rajic, I. Datskou: Detection of
on the frequency shift of microcantilever chem- infrared photons using the electronic stress in
ical sensors, IEEE Trans. Ultrason. Ferroelectr. 53, metal-semiconductor cantilever interfaces, Ultra-
2166–2173 (2006) microscopy 82, 49–56 (2000)
15.165 A. Jana, A. Raman, B. Dhayal, S.L. Tripp, 15.179 C.B. Li, B.B. Jiao, S.L. Shi, D.P. Chen, T.C. Ye,
R.G. Reifenberger: Microcantilever mechanics in Q.C. Zhang, Z.Y. Guo, F.L. Dong, Z.Y. Miao: A novel
flowing viscous fluids, Appl. Phys. Lett. 90, 114110 uncooled substrate-free optical-readable infrared
(2007) detector: design, fabrication and performance,
15.166 M. Hennemeyer, S. Burghardt, R. Stark: Cantilever Meas. Sci. Technol. 17, 1981–1986 (2006)
micro-rheometer for the characterization of sugar 15.180 X. Yu, Y. Yi, S. Ma, M. Liu, X. Liu, L. Dong, Y. Zhao:
solutions, Sensors 8, 10–22 (2008) Design and fabrication of a high sensitivity focal
15.167 K.B. Brown, Y. Ma, W. Allegretto, R.P.W. Lawson, plane array for uncooled IR imaging, J. Micromech.
F.E. Vermeulen, A.M. Robinson: Microstructural Microeng. 18, 057001 (2008)
pressure sensor based on an enhanced resonant 15.181 B. Jiao, C. Li, D. Chen, T. Ye, Y. Ou, L. Dong,
mode hysteresis effect, J. Vac. Sci. Technol. B 19, Q. Zhang: An optical readout method based un-
1628–1632 (2001) cooled infrared imaging system, Int. J. Infrared
15.168 K.B. Brown, W. Allegretto, F.E. Vermeulen, Millim. Waves 29, 261–271 (2008)
A.M. Robinson: Simple resonating microstructures 15.182 T.P. Burg, S.R. Manalis: Suspended microchannel
for gas pressure measurement, J. Micromech. Mi- resonators for biomolecular detection, Appl. Phys.
croeng. 12, 204–210 (2002) Lett. 83, 2698–2700 (2003)
15.169 Y. Su, A.G.R. Evans, A. Brunnschweiler, G. Ensell: 15.183 P. Dutta, C. Tipple, N. Lavrik, P. Datskos: Enan-
Characterization of a highly sensitive ultra-thin tioselective sensors based on antibody-mediated
piezoresistive silicon cantilever probe and its ap- nanomechanics, Anal. Chem. 75, 2342–2348 (2003)
plication in gas flow velocity sensing, J. Micromech. 15.184 H.F. Ji, Y.Q. Lu, H.W. Du, X.H. Xu, T. Thundat: Spiral
Microeng. 12, 780–785 (2002) springs and microspiral springs for chemical and
15.170 J. Mertens, E. Finot, T. Thundat, A. Fabre, biological sensing, Appl. Phys. Lett. 88, 063504
M.H. Nadal, V. Eyraud, E. Bourillot: Effects of (2006)
temperature and pressure on microcantilever reso- 15.185 G. Singh, P. Rice, R.L. Mahajan: Fabrication and
nance response, Ultramicroscopy 97, 119–126 (2003) mechanical characterization of a force sensor
15.171 V. Mortet, R. Petersen, K. Haenen, M. D’Olieslaeger: based on an individual carbon nanotube, Nano-
Wide range pressure sensor based on a piezoelec- technology 18, 475501 (2007)
tric bimorph microcantilever, Appl. Phys. Lett. 88, 15.186 A. Sakhaee-Pour, M.T. Ahmadian, A. Vafai: Appli-
133511 (2006) cations of single-layered graphene sheets as mass
15.172 P. Sievilä, V.P. Rytkönen, O. Hahtela, N. Chekurov, sensors and atomistic dust detectors, Solid State
J. Kauppinen, I. Tittonen: Fabrication and char- Commun. 145, 168–172R (2008)
453
Biological Mo
16. Biological Molecules in Therapeutic Nanodevices
In this chapter, we discuss the incorporation of other surface-science investigations of the in-
molecules into nanodevices as functional device terfaces revealed phenomena not previously
components. Our primary focus is on biologi- documented for nanoscale protein interfaces (lu-
cal molecules, although we also discuss the use brication by directly adsorbed proteins, increases
of organic molecules as functional components in friction force associated with polymer-mediated
of supramolecular nanodevices. Our primary de- increases in sample compliance). Furthermore,
vice interest is in devices used in human therapy the studies revealed wear of polymer and re-
and diagnosis, though when it is informative, ceptor proteins from semiconductor surfaces
we discuss other nontherapeutic nanodevices by an atomic force microscopy (AFM) tip which
containing biomolecular components. We dis- was not a concerted process, but rather depth
cuss design challenges associated with devices of wear increased with increasing load on the
built from prefabricated components (biological cantilever. These studies also revealed that the
macromolecules) but that are not as frequently as- polymer–protein interfaces were disturbed by
sociated with fully synthetic nanodevices. Some nanonewton forces, suggesting that interfaces
design challenges (abstraction of device object of immunoFET protein sensors translated to
properties, inputs, and outputs) can be addressed in vivo use must likely be protected from, or
using existing systems engineering approaches hardened to endure, abrasion from tissue. The re-
Part B 16
and tools (including unified modeling language), sults demonstrate that nanoscience (in this case,
whereas others (selection of optimal biologi- nanotribology) is needed to design and charac-
cal macromolecules from the billions available) terize functional planar immunoFET sensors, even
have not been fully addressed. We discuss various though the sensors themselves are mesoscale
assembly strategies applicable to biological macro- devices. The results further suggest that modifi-
molecules and organic molecules (self-assembly, cations made to the sensor interfaces to address
chemoselective conjugation) and their advan- these nanoscale challenges may be best accom-
tages and disadvantages. We provide an example plished by protein and interfacial engineering
of a functional mesoscale device, a planar field- approaches.
effect transistor (FET) protein sensor, that depends
on nanoscale components for its function. We also
16.1 Definitions and Scope ........................... 454
use the sensor platform to illustrate how pro-
16.1.1 Design Issues ............................. 455
tein and other molecular engineering approaches
16.1.2 Identification
can address nanoscale technological problems, of Biomolecular Components ....... 456
and argue that protein engineering is a legitimate 16.1.3 Design Paradigms....................... 457
nanotechnology in this application. In developing 16.1.4 Utility and Scope
the functional FET sensor, both direct adsorption of Therapeutic Nanodevices ......... 461
of protein analyte receptors as well as linkage of
receptors to the sensing surface through a poly- 16.2 Assembly Approaches ............................ 461
mer layer were tested. However, in the realized FET 16.2.1 Low-Throughput
Construction Methods ................. 461
sensor, interfaces consist of a polymer layer linked
16.2.2 Supramolecular Chemistry
to the semiconductor surface and to an analyte
and Self-Assembly ..................... 463
receptor (a protein). Nanotribology and
16.2.3 Chemoselective Conjugation ........ 469
454 Part B MEMS/NEMS and BioMEMS/NEMS
diffuse and amorphous. If nanotechnology embraces ev- Third, consistent with our device-centric bias, the term
erything, what then is it? Consideration of the scope of specific (as in, specific nanometer-scale elements) is in-
the field may be useful. tended to exclude materials whose utility derives solely
To frame the discussion, we will define nanotech- from properties inherent to being finely divided (high
nology as the discipline that aims to satisfy desired surface-to-volume ratios, for instance), or other ma-
objectives using materials and devices whose valuable terial, chemical, and physical properties unless those
properties are based on a specific nanometer-scale el- properties contribute to specific device function. We
ement of their structures. As opposed to nanoscience, made this exclusion based on our assessment that
nanotechnology is application oriented, so nanoscience therapeutic nanodevices are more intriguing than nano-
is important to nanotechnology primarily to the extent materials per se (see below), although we will engage
that it is relevant to device design, function or applica- these attributes where they are germane to specific de-
tion. vice or therapeutic applications. Fourth, our definition
The meaning of therapeutic is self-explanatory and implies that limited nanotechnology has been avail-
refers here to intervention in human disease processes able since the 1970s in the form of biotechnology.
(although many of the approaches discussed are equally Based on their nanoscale structures, individual biologi-
applicable to veterinary medicine). We confine our cal macromolecules (such as proteins) often exhibit the
discussion mostly to therapeutics used in vivo, be- coordinated, modular multifunctionality that is charac-
cause such applications clearly benefit from the low teristic of purpose-built devices (Fig. 16.1).
invasiveness that ultrasmall, but multipotent, nanoth- Biological macromolecules rely on the deployment
erapeutics potentially offer. It is debatable whether of specific chemical functionalities with specific rela-
imaging, diagnostic or sensing devices can be consid- tive distributions in space with nanometer (and greater)
ered therapeutic in this context, though as we will see, resolution for their function, so the inclusion of mo-
sensing/diagnostic functionalities are often inextricable lecular engineering aspects of biotechnology under the
elements of therapeutic nanodevices, and it is difficult nanotechnology rubric is legitimate, despite the discom-
to consider smart nanotherapeutics without discussion fort this may cause to traditionally trained engineers.
Biological Molecules in Therapeutic Nanodevices 16.1 Definitions and Scope 455
Part B 16.1
or substances. Devices exhibit desirable emergent prop- tibody fragment (single-chain Fv or SCFv) that is monospecific and
erties inherent to their design, properties that emerge monovalent can recognize only one antigenic determinant and is
as the result of the spatial and/or temporal organiza- engineered to lack common effector functions. This construct is
tion, coordination, and regulation of action of individual translated as a single, continuous polypeptide chain (hence the name
components. The organization of components in de- SCFv) because a peptide linker (indicated by the connecting line in
vices allows them to perform multistep, cogent work the figure) is incorporated to connect the carboxy-end of the Hc
processes that cannot be mimicked by simple ad- fragment and the amino-end of the Lc fragment
mixtures of individual components. In fact, if device
functions can be mimicked well by simple mixtures structures [16.2–8]. This bias toward free molecules
of components, the labor involved in configuring and flies in the face of the obvious importance of integrated
constructing a nanoscale device is not warranted. Our supramolecular structures in biology and, to the casual
device definition thus excludes nanomaterials used as observer, may seem an odd gap in attention and em-
drug formulation excipients (pharmacologically inert phasis on the part of biotechnologists. The bias toward
materials included in formulations that improve phar- single-molecule, protein therapeutics follows from the
macophore uptake, biodistribution, pharmacokinetic, fact that biotechnology developed as an industrial activ-
handling, storage or other properties), but embraces ity, governed by market considerations. Of the potential
those same materials as integral components of drug- therapeutics that might be realized from biotechnology,
delivery or other clinical devices. single-protein therapeutics are among the easiest to re-
alize from both technical and regulatory perspectives,
16.1.1 Design Issues and so warrant extensive industrial attention. This is
changing, however, and more complex entities (actual
The biotechnology industry historically has focused supramolecular therapeutic devices) have and will ap-
on production of individual soluble protein and nu- pear with increasing frequency in the 21st century.
cleic acid molecules for pharmaceutical use, with only New materials derived from micro-/nanotechnology
limited attention paid to functional supramolecular provide the opportunity to complement the tradi-
456 Part B MEMS/NEMS and BioMEMS/NEMS
tional limits of biotechnology by providing scaffolds the full range of functionality of individual biological
that can support higher-level organization of multiple molecules in interactions with other biological systems
biomolecules to perform work activities that could not (as in nanotherapeutics) is often not known. This makes
be performed by mixtures of free, soluble molecules. the design and prototyping of biological nanodevices an
Such supramolecular structures have been called empirically intensive, iterative process [16.2–16].
nanobiotechnological devices [16.9], nanobiological Paralleling the paucity of information typically
devices [16.2–8], and semisynthetic nanodevices, and available about individual proteins, the number of
they figure prominently in therapeutic nanotechnology. distinct natural proteins in the biologic world is un-
Some such devices have sensing and diagnostic capabil- known but exponentially high (certainly in excess of
ities, and therefore belong to the already discussed class 1013 distinct molecules [16.17]). When one consid-
of theranostic agents (see [16.9] for a recent example of ers engineered proteins, particularly those made by
a multifunctional theranostic nanoparticle, and [16.10] high-throughput mutagenesis methods, the number of
for a recent review of such multifunctional nanopar- existing protein sequences rises additional orders of
ticulate magnetic resonance imaging contrast agents in magnitude. Most of these molecules remain to be dis-
cardiovascular disease). covered, so their individual properties (that might be
critical to device designers, such as functional pH
16.1.2 Identification and temperature ranges, ionic requirements, cofactor
of Biomolecular Components requirements, radiation tolerance, resistance to degra-
dation, etc.) are mostly unknown. Among the relative
Design of nanodevices is similar to design of other handful of proteins that are known, properties are typi-
engineered structures, providing that the special prop- cally incompletely known, and those properties that are
erties of the materials (relating to their nanoscale known often are not those of greatest interest in se-
aspects such as quantum, electrical, mechanical, biolog- lecting a protein as a nanodevice component. In fact,
ical properties, etc.), as well as their pharmacological existing protein databases focus primarily on phar-
properties, are considered. Therapeutics interact with macological properties or evolutionary relationships
patients on multiple levels, ranging from organismal between proteins. We have proposed the building of
Part B 16.1
to molecular, but it is reasonable to expect that most databases useful specifically for nanobiotechnology,
nanotherapeutics will interface with patients at the though this has yet to occur [16.5, 17, 18]. Existing
nanoscale [16.2–16]. Typically, this means interaction protein databases, based as they are on phylogenetic,
between therapeutics and patient biological macro- protein sequence, protein structure or protein primary
molecules, supramolecular structures, organelles, cells function information, are not satisfactory for nano-
or tissues, which in turn often dictates the incorporation device design, if such design is to be realized as
of biological macromolecules (and other biostructures) a discipline in and of itself, and if the immense
into nanodevices [16.2–10]. Incorporating biological power of biologic nanotechnology is to be real-
structures into (nanobiological) devices presents spe- ized to any significant extent. As it is, inventors of
cial challenges that do not occur in other aspects of nanobiotechnological devices often rely on their per-
engineering. sonal knowledge of biology and biochemistry to select
Unlike fully synthetic devices, semibiological nano- biological components for devices. It is obvious then,
devices must incorporate prefabricated biological com- that the devices designed by even the most sophis-
ponents (proteins, nucleic acids or derivatives thereof), ticated biologists are very likely to be suboptimal.
and therefore intact nanodevices are seldom made en- As a corollary, this means that nanobiological device
tirely de novo. As a corollary, knowledge of properties designers must have significant biological expertise.
of nanobiological device components is usually incom- This constitutes a huge barrier to entry for nonbiolo-
plete, as the molecules were not made by human design, gists who could otherwise contribute importantly to the
so their properties are not known a priori and must be field. Thus, the lack of an appropriate protein database
discovered. Therefore the range of activities inherent and accompanying search tools implies an immense
to any nanobiological device design may be much less opportunity cost for the field of biologic nanotechnol-
obvious and less well defined than for fully synthetic ogy.
devices. Further complicating the issue, the activities Biological macromolecules have properties, partic-
of biological molecules are often multifaceted (many ularly those relating to their stability, that can limit
genes and proteins exhibit pleiotropic activities), and their use in device contexts. In general, proteins, nu-
Biological Molecules in Therapeutic Nanodevices 16.1 Definitions and Scope 457
cleic acids, lipids, and other biomolecules are more way to delimit the activity of nanotherapeutics contain-
labile to physical/biochemical insult than are many syn- ing biomolecules.
thetic materials. With the possible exceptions of topical
agents or oral delivery and endosomal uptake of nan- 16.1.3 Design Paradigms
otherapeutics (both involving exposure to low pH),
device lability in the face of physical insult is gen- Several early attempts to codify the canonical properties
erally a major consideration only in ex vivo settings of ideal nanobiological devices, and therapeutic nano-
(relating to device storage, sterilization, etc.), because devices in particular, have been made [16.1,5,11–15,19,
physical conditions that would destroy the device would 20]. These attempts to codify design constitute a lim-
be bad for the patient as well. However, living or- ited set of design guidelines that are summarized in
ganisms remodel themselves constantly in response to Table 16.1. Naturally occurring, functional biological
stress, development, pathology, and external stimuli. components generally exist in the context of higher-
For instance, epithelial tissues and blood components order systems that support the organisms of which they
are constantly eliminated and regenerated, and bone and are a part. In general, nanobiological devices contain bi-
vasculature are continuously remodeled. The metabolic ological components that retain their function in new
facilities responsible (circulating and tissue-bound pro- (device) contexts. In other words, one must incorpo-
teases and other enzymes, various clearance organs, the rate into the device enough of a functional biological
immune system, etc.) can potentially process biologi- unit (nucleic acid, protein, oligomeric protein complex,
cal components of nanobiological therapeutic devices organelle, cell, etc.) to allow that unit to perform the
as well as endogenous materials, leading to partial function for which it was selected. If one wishes, for
or complete degradation of nanotherapeutic structure, example, to appropriate the specific antigen-recognition
function, or both. Furthermore, immune and wound re- property of an antibody for a device function (say, in
sponses protect the host against pathogenic organism targeting, as discussed later), it is not necessary to in-
incursions, by mechanisms that involve sequestering corporate the entire 150 000 atomic mass unit (AMU)
and degrading pathogens. Nanobiological therapeutics antibody, the bulk of which is devoted to functions other
are subject to the actions of these host defense sys- than antigen recognition (Fig. 16.1) [16.21], but it is
Part B 16.1
tems as well as normal remodeling processes. Various critical to incorporate the approximately 20 000 AMU
strategies to stabilize biomolecules and structures in of the antibody essential for specific antibody–antigen
heterologous in vivo environments are applicable to binding (the variable domain, Fig. 16.1).
nanobiological therapeutics. Conversely, instability of Device function is the result of the summed and
active biocomponents can offer a valuable and simple various activities of biological and synthetic device
Table 16.1 Some ideal characteristics of nanodevices [16.1, 5, 11–15, 19, 20]
contribute to overall device functions that are entirely expression cassette is applied at this point, the only E. coli
different from those in which they participated in their that survive are those containing recombinant bacmid (i. e.,
original, organismal contexts. those that have received the gene for insect cell expression
All of these features are illustrated in the bacmid, or by transposition from the donor). Recombinant bacmid are
Bac-to-Bac, system, a commercially available molecu- conveniently isolated from E. coli and introduced into in-
lar cloning device ([16.22–24] and Fig. 16.2). Bacmid sect cell culture, where expression of the gene of interest
configures prokaryotic and eukaryotic genetic elements occurs (after [16.25])
from multiple sources into a device for producing
recombinant eukaryotic viral genomes in bacteria, sign approaches for building functional devices with
a function that is unprecedented in nature. The system biocomponents. Bacmid complies fully with those de-
is feasible because of the modularity of the genetic el- sign recommendations of Table 16.1 that are not
ements involved and because of the strict control of therapeutic device specific. However, Table 16.1 pro-
the relative arrangement of genetic elements allowed vides guidelines only. Systems engineering approaches
by recombinant deoxyribonucleic acid (DNA) technol- commonly used in software and computer engineer-
ogy. Analogous cloning devices based on comparable ing provide a more rigorous framework to consider
arrangements of bacterial and eukaryotic regulatory and nanodevice design [16.17, 18]. Systems tools such as
structural genes are reviewed in [16.18]. Other nu- unified modeling language (UML) allow more precise
cleic acid devices using genetic control elements from depiction of nanobiological devices than do most text
phylogenetically various sources are being developed descriptions. A simplified UML use-case of bacmid is
to preprogram the micro- and nanoscale architec- presented in Fig. 16.3.
tural properties and physiological behavior of living Most importantly, UML forces designers to abstract
things [16.25, 26]. knowledge of component (object) functions and proper-
Bacmid provides an example of a nontherapeutic ties and to express them in terms of object inputs and
nanobiological device and illustrates some specific de- outputs. This makes device designs modular by explic-
Biological Molecules in Therapeutic Nanodevices 16.1 Definitions and Scope 459
Input Output
Bacmid cloning system
Recombinant
Gene of • Replication in bacteria Baculovirus
interest (goi) • Accepts insertion of goi (containing goi)
• Replicates in insect cells
Note: Decision:
E. coli as
bacterial host «calls» Donor object
• Ts replication E. coli
«calls» • Contains goi, antibiotic
Replication object resistance gene between
• Replicates large DNAs attL and attR
in E. coli and
• Contains insertion site
Recombination enzyme object
for goi
• Mediates transposition of DNA
«uses» between attL and attR
Note: Decisions:
• F-ori as replication «calls»
object
• attTn7 as insertion site
E. coli object
«calls»
• Provides Tn7 recombination
enzymes
• Support F-ori, ts on replication
Part B 16.1
Transposition object
«uses» • Contains goi, antibiotic resistance E. coli with
gene between attL and attR recombinant bacmid
Sequence thread
«Destroys»
Unreacted donor objects
Elevated temp.
Fig. 16.3 The bacmid system represented as a use-case with unified modeling language (UML). Objects are represented
by boxes, split horizontally into compartments. The topmost compartment contains the object name, the middle com-
partment contains the object attributes, and the bottom compartment contains the object operations (if different from
the object attributes). Notes contain expository information such as design decisions. Sequence threads indicate serial
events in process flow. Process flow is indicated by solid arrows. Dashed arrows indicate communication between ob-
jects and, for the purposes of illustrating bacmid, are labeled as uses (the object utilizes a second object), calls (the
object elicits an action from a second object), and destroys (a step that causes the loss or destruction of one or more
objects). End of process is indicated. The abstraction of the function of individual objects driven by UML may make
other analogous devices (similar to bacmid, but substituting one or more isofunctional objects) obvious [16.17, 18]. In
protein-containing devices, systematic identification of isofunctional objects could be facilitated using protein databases
modified as suggested in [16.17] (after [16.18])
460 Part B MEMS/NEMS and BioMEMS/NEMS
a) S
S C S
1
2
3
4
I E E
E
Ta
E
E
EE
E E
I C C EE C Ta
1
EE 1
2 E E 2
3
4 E 1 3
4
2
E 3 E
E 4
E
I Ta
Tr C Tr
Part B 16.1
1
2
3
4
Tr
Fig. 16.4a,b A hypothetical, modular nanotherapeutic agent patterned after the dendrimer-based cluster agent for
oncology of Baker [16.27–29]. Self-assembling, multidendrimer structures as shown are sometimes referred to as
tecto(dendrimers). Each polyamidoamine (PAMAM) dendrimer subunit is grown from an initiator core (C), and the
tunable surface groups of the dendrimers are represented by Z. In this device, each dendrimer subunit has a specific,
dedicated function in the device: the central dendrimer encapsulates small-molecule therapeutics (E), whereas other
functional components are segregated to other dendrimer components. These include biochemical targeting/tethering
functions (Ta), therapeutic triggering functions to allow activation of prodrug portions of the device by an external opera-
tor (Tr), metal or other constituents for imaging (I), and sensing functions (S) to mediate intrinsically controlled activation
or release of therapeutic. This design constitutes a therapeutic platform [16.14, 16] because of its modular design. The
depicted device is only one possible configuration of an almost infinite number of analogous therapeutics that can be
tuned to fit particular therapeutic needs by interchanging functional modules (after [16.16]). (b) A more sophisticated as-
sembly strategy for dendrimer therapeutics that utilizes self-assembly of biomolecules [16.30–32]. Individual dendritic
polymers (spheres representing generation 5 and generation 7 PAMAM dendrimers) are conjugated with single-stranded
(ss) oligonucleotides (light and dark gray lines). When oligonucleotides of two dendritic polymers are complementary,
individual oligos hybridize, forming double-stranded (ds)DNA complexes, linking the two dendritic polymers. An inter-
esting feature of this assembly system is the rigidity of short dsDNA (< 50 base pair) segments. This allows assembly of
objects at precise, tunable nanoscale distances (after [16.30])
Biological Molecules in Therapeutic Nanodevices 16.2 Assembly Approaches 461
itly identifying object properties and limitations: any 16.1.4 Utility and Scope
object producing the same outputs and accepting the of Therapeutic Nanodevices
same inputs can be substituted for the original compo-
nent in the functional nanodevice. By this means, UML Therapeutic nanotechnology will be useful when the
notation transforms nanodevice designs from one-off underlying biology of the disease states involved is
individual designs into general designs for a class of amenable to intervention at the nanoscale. While several
functionally similar devices [16.18]. In general, biolo- disease states and physiological conditions (cancer, vac-
Part B 16.2
gist inventors have intuitively used these systems design cination, cardiovascular disease, etc.) are particularly
approaches to design devices, as we discuss [16.18]. accessible to nanoscale interventions, some nanotech-
However, UML-enforced explicit callout of compo- nological approaches may be applicable more broadly,
nent properties that are necessary and sufficient for in indications we currently cannot predict. Much as was
device function makes design of isofunctional devices the case with the introduction of recombinant protein
easier. An argument might be made that, in some therapeutics, nanotherapeutics may present regulatory
cases, UML expression of a device design, an ex- and pharmacoeconomic challenges related to their nov-
ample of which is shown in Fig. 16.3, may render elty and their cost of goods (COGs). However, there
related devices (that are implied in the UML depiction) is little doubt that nanobiological devices providing
obvious. If so, broad use of UML in nanobiotech- clear patient benefit, and whose production, regulatory
nology might have significant intellectual property approval, and distribution are amenable to feasible busi-
implications. ness models, will enter clinical practice.
purposes than to production of commercial therapeu- that this might make relatively low-throughput syn-
tics. For example, direct-write technologies can obtain thesis/assembly methods practical, though this remains
high (nanometer-scale) resolution; electron-beam (e- to be seen. If tailored therapeutics become the stan-
beam) lithography is a technique requiring no mask and dard, the pharmaceutical industry will be irrevocably
can yield resolutions on the order of tens of nanome- changed, with the concept of the blockbuster that can
ters, depending on the resist materials used [16.33]. produce multiple billions of dollars in revenue per year
Resolution in e-beam lithography is ultimately limited through sales of a single agent to a broad popula-
by electron scattering in the resist and electron optics, tion made no longer relevant. Business models in the
and like most direct-write approaches, e-beam lithog- pharmaceutical industry may be changed beyond recog-
raphy is limited in its throughput. Parallel approaches nition, and in the fullness of time, pharmacies could
involving simultaneous writing with up to 1000 shaped begin to resemble the formularies of old, with the ca-
e-beams are under development [16.33] and may miti- pacity to make a specific preparation for an individual
gate limitations in manufacturing rate. patient on site, but now using sophisticated bio- and
Atomic force microscopy (AFM) approaches utilize nanotechnologies.
an ultrafine cantilever tip (typically with tip diameters Individualized therapies for one patient are likely to
of 50 nm or less) in contact with, or tapping, a sur- differ from those for other patients by their incorpora-
face or stage. The technique can be used to image tion of engineered molecules that are immunologically,
molecules, analyze molecular biochemical properties toxicologically or physiologically tolerable by each
(such as ligand–receptor affinity [16.34]), or manipu- individual host. Device functional properties will be
late materials at the nanoscale. In the latter mode, force tuned to integrate with individual host physiologies.
microscopy has been used to manipulate atoms to build One facile way to achieve patient-specific theranostics
individual nanostructures since the mid-1980s, though might be to design a modular device whose properties
the manufacturing throughput of manual placement of are tuned by substitution of biochemically isofunctional
atoms and nanoscale components by force microscopy components, but whose other properties (immunologi-
is limited, even with highly multiplexed arrays of can- cal, toxicological, etc.) are suited to individual patients.
tilevers. This sounds like the modular nanodevice construction
Part B 16.2
Dip-pen nanolithography (DPN) is a force mi- strategy we advocate [16.17, 18]. Should this come to
croscopy methodology that can achieve high-resolution pass, pharmaceutical companies will have to find a way
features (features of 100 nm or less) in a single step. to claim as intellectual property immense collections
In DPN, the AFM tip is coated with molecules to be of similar function, but individually tailored, nanother-
deployed on a surface, and the molecules are trans- apeutic devices. Patenting each device individually is
ferred from the AFM tip to the surface as the coated tip economically and logistically impossible, and a patent
contacts it. DPN also can be used to functionalize sur- on a single specific nanotherapeutic might be easily in-
faces with two or more constituents and is well suited vented around by competitors. One possible solution
for deployment of functional biomolecules on synthetic may be to structure patent claims broadly, around UML
surfaces with nanoscale precision [16.35, 36]. DPN suf- representations of nanotherapeutics that explicitly call
fers the limitations of synthetic throughput typical of out the critical characteristics of objects to identify alter-
AFM construction strategies. nate, isofunctional objects that might be substituted for
Much as multibeam strategies might improve the objects of the device in its original design [16.18].
throughput in e-beam lithography [16.34], multiple This may capture a group of analogous, related thera-
tandem probes may significantly increase assembly nostics, rather than an individual nanotherapeutic, or it
throughput for construction methods that depend on may make some or all of the analogous devices obvious
force microscopy, but probably not sufficiently to allow in the eyes of the patent examiner. This would likely
manufacture of bulk quantities of nanostructures, as will preclude the proliferation of devices, each captured by
likely be needed for consumer nanotherapeutic devices. its own patent, but template off a single central inven-
As standard of care evolves increasingly toward tai- tion, as occurred with analogues of bacmid [16.18].
lored courses of therapy [16.37], individual therapeutics For the moment, though, ideal manufacturing ap-
will become increasingly multicapable and powerful. proaches for nanotherapeutic devices resemble either
Potentially, fewer copies of a nanotherapeutic may be industrial polymer chemistry, occurring in bulk, in
required per patient, and each patient’s nanotherapeu- convenient buffer systems, or in massively parallel
tic may be tailored to him/her. It is not inconceivable industrial microfabrication approaches. In any case,
Biological Molecules in Therapeutic Nanodevices 16.2 Assembly Approaches 463
therapy for a single patient may involve multiple their subunits exceeds the critical micelle concentra-
billions of individual nanotherapeutic units, so each tion (CMC) in a solvent in which one of the polymeric
individual nanotherapeutic structure must require only domains is insoluble (Fig. 16.6). The CMC is deter-
minimal input from a human technician. Self-assembly, mined by the insoluble polymeric domain, and can be
when feasible, allows device construction without on- adjusted by control of the chemistry and length of the
going human intervention. immiscible domain, as well as by control of solvent
conditions. Micelles formed at low concentrations from
16.2.2 Supramolecular Chemistry low-CMC polymers are stable at high dilution. Micelles
and Self-Assembly formed from polymer monomers with high CMCs can
dissociate upon dilution, a phenomenon that might be
Exploitation of the self-assembly properties of den- exploited to control release of therapeutic cargos. If
drimers and its exploitation to build a modular ther- desired, micelles can be stabilized by covalent cross-
apeutic has already been alluded to in Fig. 16.4a. linking to generate shell-stabilized structures [16.42–
Self-assembly has long been recognized as a poten- 44]. Size dispersity and other properties of micelles
tially critical labor-saving approach to the construction can be manipulated by control of solvent conditions,
of nanostructures [16.38], and many organic and inor- incorporation of excipients (to modulate polymer pack-
ganic materials have self-assembly properties that can ing properties), temperature, and agitation. From the
be exploited to build structures with controlled config- standpoint of size, reasonably monodisperse prepara-
urations. Self-assembly processes are usually driven by tions (polydispersities of 1–5%) of nanoscale micellar
thermodynamic forces and generally result in structures structures can be prepared.
that are not covalently linked. Intra-/intermolecular The immense versatility of industrial polymer
forces driving assembly can be electrostatic or hy- chemistry allows micellar structures to be tuned chem-
drophobic interactions, hydrogen bonds, and van der ically to suit the task at hand. They can be modified
Waals interactions between and within subunits of the for targeting (by appending ligands that recognize par-
self-assembling structures or the assembly environ- ticular targeting sites to their surfaces) or to support
ment. Thus, final configurations are limited by the higher-order assembly of micelles. They can be made
Part B 16.2
ability to tune the properties of the subunits and con- to imbibe therapeutic or other molecules for deliv-
trol the assembly environment to generate particular ery and caused to dissociate or disgorge themselves of
structures. payloads at desired times or bodily sites under the in-
Highly hydrophobic carbon nanotubes sponta- fluence of local physical or chemical conditions. The
neously assemble into higher-order [16.39] structures tunability of these and other properties at the level of
(nanoropes and multiwall carbon nanotubes) as the monomeric polymer subunits (as well as the level of
result of hydrophobic interactions between individ- assembled higher-order structures) makes micelles po-
ual tubes [16.39]. C60 fullerenes and single-wall tentially powerful nanoscale vehicles for the delivery of
carbon nanotubes (SWCNT) also spontaneously assem- drugs or imaging contrast agents.
ble (Fig. 16.5) into higher-order nanostructures called In its most sophisticated manifestation, synthetic
peapods [16.40], in which fullerene molecules are en- polymer self-assembly strategies can allow the gener-
capsulated in nanotubes. The fullerenes of peapods ation of three-dimensional structures of highly defined
modulate the local electronic properties of the SWCNT nanoscale morphology. The production of controlled
in which they are encapsulated and may allow tun- self-assembled structures can be affected by synthe-
ing of carbon nanotube electrical properties, perhaps as sis and assemble specific organic chemical compounds
in one-dimensional (1-D) carbon nanotube field-effect of controlled structure and chirality. Nanoscale mo-
transistors (FETs) [16.41]. Several self-assembled car- lecular assemblies of this nature are said to be the
bon structures and carbon-structure-containing devices result of supramolecular chemistries. Supramolecular
are depicted in Fig. 16.5. chemistries exploit designer knowledge of molecular
In drug delivery, the most familiar self-assembled geometries, intramolecular interactions (hydrophobic,
nanostructures are micelles [16.42–44]. These struc- metal chelation, hydrogen bonding, and dipole in-
tures are formed from the association of block teractions), and intramolecular packing properties to
copolymer subunits (Fig. 16.6), each individual subunit drive formation of nanoscale structures of controlled
containing hydrophobic and hydrophilic domains. Mi- shape [16.45, 46]. As the referenced reviews [16.45, 46]
celles spontaneously form when the concentration of show, supramolecular chemistry can produce space-
464 Part B MEMS/NEMS and BioMEMS/NEMS
a) b)
c)
a) b)
H2O
H2O
Dialysis
against
water
H2O H2O
c)
organic Evaporation
solvent
Part B 16.2
Polymer in water
Fig. 16.6a–d Micellar drug-delivery vehicles and their d) Incorporation in micelles (arb. units)
self-assembly from block copolymers [16.42–44]. (a) Mor-
phology of a micelle in aqueous buffer. Hydrophobic and 25
hydrophilic polymer blocks, copolymers containing the
hydrophilic polymer blocks, copolymers containing the
blocks, micelles generated from the block copolymers, 20
and (hydrophobic) drugs for encapsulation in the micelles
are indicated. (b) Micelle self-assembly and charging with
drug occurring simultaneously when the drug–polymer for- 15
mulation is transitioned from organic to aqueous solvent
by dialysis. (c) Preformed micelles can be passively im-
bibed with drugs in organic solvent. Organic solvent is 10
then removed by evaporation, resulting in compression
of the (now) drug-bearing hydrophobic core of the mi-
celle. (d) An illustration of concentration-driven micelle 5
formation. At and above the critical micelle concentration
(CMC), block copolymer monomers assemble into mi-
celles rather than exist as free block copolymer molecules. 0
The arrow indicates the CMC for this system (after [16.42– –10 –9 –8 –7 –6 –5 –4 –3
Log concentration
44])
466 Part B MEMS/NEMS and BioMEMS/NEMS
ning electron micrograph image showing the network of nanofibers Design of self-assembling polymeric units that form
in vitro; scale bar in (b) indicates 200 nm. (c,d) Representative trac- specific nanoscale structures requires organic chem-
ings of descending motor axon fibers within a distance of 500 μm ical synthetic capabilities and modeling skills that
rostral of the lesion in vehicle-injected (c) and IKVAV PA-injected are not widely distributed in the physical and bi-
(d) mice. The dotted lines demarcate the borders of the area of ological nanotechnology communities, but they do
spinal cord injury. Colored lines indicate descending motor neu- constitute genuine nanotechnologies, in as much as
rons impinging on the lesion. Scale bars in (c,d) indicate 100 μm. supramolecular chemistry facilitates multicomponent
R – rostral; C – caudal; D – dorsal; V – ventral. IKVAV PA also functional structures whose functions depend on struc-
promotes lesser, but still significant, regeneration of sensory neu- turally defined nanoscale components. The potential of
rons (not shown [16.48]). IKVAV PA-induced improvements in supramolecular structures in therapy is immense.
regeneration of axons are associated with significant behavioral im- Supramolecular chemistry of synthetic peptides
provements in mice. The self-assembly properties of IKVAV PA is has provided pharmacologically active antimicrobial
essential to its function: IKVAV peptide alone does not promote structures (Fig. 16.8) [16.49, 50]. In the anti-infective
regeneration [16.47, 48] (after [16.48]) architecture, individual peptide components are flat, cir-
cular molecules. The planar character of the toroidal
The designers incorporated the peptide isoleucine– subunits is a consequence of the alternating chirality
lysine–valine–alanine–valine (IKVAV) into the polar of alternating dl amino acids (AAs) in the primary se-
head group of each amphiphilic polymer monomer. IK- quence of the peptide rings, currently only possible for
VAV is a neuroactive peptide derived from the protein synthetic peptides. Ribosomes recognize and incorpo-
laminin, and is presented at immensely high valency on rate into nascent polypeptides only l amino acids, and
the surface of assembled IKVAV PA nanofilaments. In so, as a result of AA chirality and bond strain, peptides
the central nervous system, IKVAV peptide is known made by ribosomes cannot be flat, closed toroids like
to inhibit differentiation to glial cells and promote neu- those of the peptidyl anti-infective agents. Alternation
ronal outgrowth. Glial cells have a primary role in of d and l AAs is not possible in proteins made by
laying down scar tissue after spinal cord injury, and scar ribosomal synthesis, though as we discuss below, this
Biological Molecules in Therapeutic Nanodevices 16.2 Assembly Approaches 467
H
H R O O H
O N H N O
R D N R • •••
•••••••••••••••• •
L L O
HN • •••
O O N •••••••••••••••• •
NH
H O • •••
R D D R •••••••••••••••• •
H
• •••
HN O H •••••••••••••••• •
O H
NH • •••
O L L N O •••••••••••••••• •
D N
R R
HN N O
O R H
O N
H O
H
Fig. 16.8 A self-assembling peptide antibiotic nanostructure [16.49, 50]. Peptide linkages and the α-carbons and their
pendant R-groups are indicated. The synthetic peptide rings are planar as a result of the alternating chirality (D or L) of
their amino acid (AA) constituents. R-groups of AAs radiate out from the center of the toroid structure. Individual toroids
self-assemble (stack) as a result of hydrogen-bonding interaction between amine and carboxy groups of the peptide
backbones of adjacent toroids. The surface chemistry of multitoroid stacks is tuned at the level of the AA sequence
and, therefore, R-group content of the synthetic peptide rings. The chemical properties of the stacked toroid surfaces
allow them to intercalate into the membranes of pathogenic organisms, with lethal consequences. The specific membrane
preferences for intercalation of the compound are tuned by control of the R-group content of the toroids (after [16.49,50])
may change. Much as in α-helical domains of riboso- self-assembling d, l peptide toroids [16.49, 50], rep-
mally synthesized proteins, however, the AA R-groups resent critically needed, novel antibacterial agents.
Part B 16.2
(which are of varying hydrophobic or hydrophilic Resistance to traditional, microbially derived antibi-
chemical specificities) are arranged in the plane of the otics often is tied to detoxifying functions associated
closed d, l rings, extending out from the center of the with secondary metabolite synthesis; these detoxify-
rings. Hydrogen bonds between individual rings gov- ing functions are essential for the viability of many
ern self-assembly of the toroids into rod-like stacks, antibiotic-producing organisms [16.52]. The genes
while the R-groups dominate interactions between mul- encoding such detoxifying functions are rapidly dis-
tiple stacks of toroids and other macromolecules and seminated to other microorganisms, accounting for
structures. The planar toroidal subunits can be admin- the rapid evolution of drug-resistant organisms that
istered as monomers and self-assemble into multitoroid has bedeviled antimicrobial chemotherapy for the last
rods (Fig. 16.8) at the desired site of action (in bio- half-century. Synthetic nanoscale antibiotics, such as
logical membranes). The peptide toroid R-groups are peptide toroids [16.49, 50] and N8N antimicrobial na-
chemically tuned so that the rod structures into which noemulsion [16.51], act by mechanisms entirely distinct
they spontaneously assemble intercalate preferentially from those of traditional secondary metabolite antibi-
in specific lipid bilayers (i. e., in pathogen versus host otics, and presumably no native detoxifying genes exist
membranes). Moreover, the assembled rods may un- for these supramolecular antibiotics because they are
dergo an additional level of self-assembly into multirod structures not produced by evolution. Therefore, novel
structures, spanning pathogen membranes [16.49, 50]. nanoscale antimicrobials may not be subject to the
Intercalation of the self-assembled rods into pathogen unfortunately rapid rise in resistant organisms associ-
membranes reduces the integrity of pathogen mem- ated with most secondary metabolite antibiotics, though
branes selectively, potentially rendering agents toxic to this remains to be seen. As bacterial infection con-
pathogens but not to hosts. tinues to reemerge as a major cause of morbidity
Other self-assembling, nanoscale antibiotics with and mortality in the developed world, as a con-
morphologies more like that of micellar struc- sequence of increasing antibiotic-resistant pathogens,
tures [16.42–44] have been described (N8N antimi- novel, nanoscale antibiotics will become increasingly
crobial nanoemulsion [16.51]). They, as well as the important.
468 Part B MEMS/NEMS and BioMEMS/NEMS
Biological macromolecules undergo self-assembly structures can exhibit tightly controlled topographies
at multiple levels, and like all instances of such assem- but have limited integrity in terms of geometry [16.56],
bly, biological self-assembly processes are driven by due to the flexibility of double-stranded (ds)DNA of 200
thermodynamic forces. Some biomolecules undergo in- or more base pairs in length.
tramolecular self-assembly (as in protein folding from Often the domains of biomolecules responsible for
linear peptide sequences, Fig. 16.9). Higher-order struc- assembly and recognition are small, continuous, and
tures are, in turn, built by self-assembly of smaller discrete enough that they can be abstracted from their
self-assembled subunits (for instance, structures as- native context as assembly modules and appended to
sembled by hybridization of multiple oligonucleotides,
enzyme complexes, fluid mosaic membranes, ribo-
a) NH2
somes, organelles, cells, tissues, etc.). Assemblies of SH
higher-order structures, such as oligomeric proteins, are
driven by the same forces that drive intramolecular pro- HS
tein folding. SH
SH
Naturally occurring proteins are nonrandom copoly-
mers of 20 chemically distinct amino acid (AA) SH NH2
subunits. The precise order of AAs (i. e., via interactions
between AA side-chains) drives the linear polypeptide SH
chains to form specific secondary structures (the α-
helices and β-sheet structures seen in Fig. 16.9). The
secondary structures have their own preferences for as- COOH
SH
sociation, which in turn leads to the formation of the
tertiary and quaternary structures that constitute the Partially folded
intermediate
folded protein structures. In its entirety, this process
produces consistent structures that derive their biolog-
ical functions from strict control of the deployment of HS S S bond
Part B 16.2
other nanomaterials for direct formation of controlled carbodiimide (EDC) conjugation [16.61]] used to con-
nanoscale architectures. Antibodies and other specific jugate cytokines to nanoparticles tend to inactivate
biological affinity reagents can be used to assemble human interleukin (hIL)-3 and other cytokines. The
hybrid nanostructures (Fig. 16.9). Oligonucleotides are same protein–particle bioconjugates retain bioactivity if
particularly intriguing in this application. judiciously chosen chemoselective conjugation strate-
Unlike longer DNA segments, shorter dsDNA seg- gies are used [16.1, 59, 60].
ments (of 50 base pairs or less) are rigid. The rigidity The potential utility of chemoselective conjugation
of short ds oligonucleotides has been cleverly exploited for incorporation of active biological structures into
to build multidendrimer complexes (Fig. 16.4b) [16.30– semisynthetic nanodevices cannot be overestimated. In
32]. Dendrimers in these complexes are held at very fact, some protein engineering methods that can change
precise distances from one another, defined by the protein topography, and therefore the spatial relation-
length of the dsDNA segments. Nanometer-scale in- ships between proteins and the device nanocomponents
terdendrimer distances measured by AFM and other to which they are conjugated, can be used to control
methods are within 10% of distances predicted from protein-device spatial arrangements only in circum-
the dsDNA length [16.30, 31]. The length of the ds stances where the association between the protein and
oligonucleotide is defined by the DNA sequence, and device is nonrandom and oriented, as can occur via
is easily tuned. This assembly strategy is currently be- chemoselective conjugation.
ing used to build multidendrimer anticancer devices, Protein sequences are said to be circularly permuted
analogous to those envisioned in Fig. 16.4a, albeit with (CP) when their parental amino- and carboxy-ends are
greater synthetic control and stability. The implications ligated together, and new amino- and carboxy-ends are
of the strategy goes well beyond dendrimer complexes, introduced elsewhere in the protein sequence. There are
and might be used to impose precise distance relation- as many potential CP variants of a given protein as there
ships on nanoscale components or between nanoscale are AAs in the protein sequence. In each distinct variant,
components and surfaces in devices. the CP sequence initiates (and ends) at a different AA
of the parental sequence.
16.2.3 Chemoselective Conjugation The fact that many proteins tolerate circular per-
Part B 16.2
mutation of their primary AA sequences came as
Several chemoselective bioconjugate approaches have a surprise to molecular biologists. Nonetheless, multiple
arisen from the field of protein semisynthesis [16.20, examples of biofunctional CP proteins have been doc-
57, 58]. These protein synthetic chemistries allow site- umented [16.62–71]. We recently developed a method
specific conjugation of polypeptides to heterologous that displays all possible CP variants of a protein on the
materials in bulk, as the result of conjugation between surfaces of phage particles. Phage-presented CP pro-
exclusive, mutually reactive electrophile–nucleophile teins can be conveniently screened for function by their
pairs, one on the polypeptide, the other on its conju- affinity for a target (for a known ligand of the parent
gation partner. Chemoselective conjugation strategies protein, for instance). We call this process scanning cir-
have been applied to the synthesis of multiple nanobio- cular permutation [16.72, 73], and it allows functional
logical devices [16.1, 1–8, 15, 57–60]. screening of every possible CP variant of a protein of
Proteins are profoundly dependent on their three- interest.
dimensional shapes for their activities: chemical deriva- One of the remarkable findings of the scanning CP
tization at critical AA sites can profoundly and nega- work was the finding that a large fraction (as many
tively impact protein bioactivity. Because conjugation as 13 , and possibly more) of the distinct CP variants
can be directed to preselected sites via chemoselec- of a test protein were bioactive [16.72]. Many of the
tive approaches, and since the sites of conjugation in new amino- and carboxy-ends of functional CP variants
the protein can be chosen because the proteins tol- fell within known secondary structural domains of the
erate adducts at those positions, proteins coupled to parent protein, and yet CP variants interrupting those
nanomaterials by such chemoselective methodologies secondary structural domains exhibited full bioactivity.
often retain their biological activity. In contrast, pro- This all suggests that the linear order of protein AA
tein bioactivity in conjugates often is lost or profoundly and secondary structural domains as they occur in nat-
impaired when proteins are coupled to nanomaterials ural proteins may not be requisite for protein structure
using promiscuous chemistries. For instance, promis- and function, and further that multiple biofunctional CP
cuous chemistries [1-ethyl-3-(3-dimethylaminopropyl) variants of many proteins may be possible [16.62–72].
470 Part B MEMS/NEMS and BioMEMS/NEMS
a) b) c) N
C
N N
C C
Chemoselective
C conjugation
N
N CN
Part B 16.2
Fig. 16.10a–c Examples of protein engineering manipulations that could potentially alter the orientations of proteins
(SCFvs) deployed on micromachined surfaces. Ribbon diagrams of SCFvs (VH–VL configuration) to be deposited on
a SiO2 surface. VH is shown in dark brown, VL in light brown. Complementarity-determining regions (CDRs) of VH
and VL are shown in black. CDRs direct specific antigen recognition. Micromachined metal oxide surface is represented
by a brown bar. A polymeric layer on the surface is represented by wavy lines. N- and C-ends of SCFvs are indicated.
Chemoselective ligation between N-ends of SCFvs and polymer is indicated. (a) Affinity peptide–SCFv: surface-specific
affinity peptide selected from a display library (for instance, SiO2 - [16.74, 75] or Al2 O3 -binding [16.76] peptide) is
inserted into the SCFv antibody fragment (gray line). The affinity peptide binds to the surface, effectively orienting the
SCFv and determining the proximity of the CDRs to the SiO2 surface. (b) Parent SCFv: chemoselective conjugation of
a modified SCFv (with an N-terminal aldehyde; [16.1, 59, 60, 72, 73] to the polymer layer. Note the position of the VH
CDRs. (c) A circularly permuted variant of a SCFv: chemoselective conjugation of a circularly permuted [16.72, 73], but
otherwise comparable, SCFv. In (b) and (c), note that chemoselective conjugation produces a consistent orientation of
SCFvs and that, relative to the parent SCFv, CP alters the proximity of the CDRs to the surface (after [16.73])
Scanning CP [16.72] provides a high-throughput way relative to the surface. Circular permutation alters pro-
to identify biofunctional CP variants and to build col- tein topology to modulate distances between protein
lections of CP versions of a single parent sequence. functional domains (such as antigen-combining sites
When proteins are conjugated to a polymer interface of SCFvs or single chain fragment variables) and N-
by a single, specific amino acid (as by chemoselective termini (where conjugation uniquely occurs, and so
conjugation [16.20, 57, 58]), their orientations are more proximal to the surface). Together, chemoselective con-
or less regular (limited by the conformational freedoms jugation and circular permutation modulate protein
of the protein and interfacial polymers themselves). orientation on surfaces to allow tuning of distance to
Typically, chemoselective conjugation is limited solely surfaces and protein functional domains (like antigen
to protein N-termini, producing consistent orientation combining sites) [16.72, 73]. This effect is illustrated
Biological Molecules in Therapeutic Nanodevices 16.3 Sensing Devices 471
for a hypothetical antibody fragment (so-called single- tems [16.79], potentially allowing proteins with UAAs
chain Fv or SCFv) in Fig. 16.10. The potential utility to be synthesized in host systems of varying expres-
of changing these distances is discussed in the context sion properties or with host-specific posttranslational
of a nanobiotechnological sensor system below. In this modifications.
sensor, certain key nanoscale distances are the primary These systems for ribosomal introduction of UAAs
determinants of overall sensor function, and we will at specific sites in biologically produced proteins have
show how circular permutagenesis and other molecu- several general features in common [16.77–79]. They
lar biology approaches can be used to address problems exploit suppression of nonsense codons (of which there
occurring at the nanoscale. are three in E. coli, and which, in wild-type strains,
stop protein translation) by a mutant transfer ribonu-
16.2.4 Unnatural Amino Acids to Support cleic acid (RNA) (a so-called suppressing tRNA) whose
Chemoselective Conjugation anticodon loop is homologous to the nonsense codon
of Biologically Produced Proteins that will drive UAA insertion into nascent protein. Not
only must this mutant tRNA recognize the nonsense
Early protein chemoselective conjugation methods de- codon, but it must also have the capacity to be charged
pended on either protein or peptide chemical synthesis with the desired UAA by a corresponding aminoacyl-
to introduce a nucleophile or electrophile into the tRNA synthase. Generating aminoacyl-tRNA synthases
protein sequence for chemoselective conjugation, or that can recognize both the suppressing tRNA and the
on some postsynthesis processing of a native AA to desired UAA requires extensive genetic engineering ef-
a chemically reactive group [16.20, 57, 58]. Typically, fort. To ensure specificity of UAA incorporation, the
the reactive group has been at an end of a polypep- aminoacyl-tRNA synthase cannot recognize any native
tide (most typically, at the N-end). Recently, more AA, and cannot direct charging of any of the host’s
elegant and versatile methods to introduce reactive other tRNAs.
groups to polypeptides via ribosomal synthesis have de- As current (2008) metabolic engineering capabili-
veloped. These methods utilize biologically produced ties go, UAA-incorporating systems must rank as the
proteins made in expression host systems engineered most sophisticated described to date. Nonetheless, their
Part B 16.3
to introduce unnatural AAs (UAAs) at specific sites output is engineered protein with reactive groups at
in proteins. UAAs can have a variety of functional predetermined AA positions within them. These reac-
side-chains, including chromophores, fluors, photore- tive groups can be conjugated chemoselectively to other
active groups or chemically reactive groups. To the nanoscale components, to effect changes in the spa-
extent that reactive functionalities of UAAs can par- tial relationship of particular protein functional domains
ticipate in chemoselective conjugations with other and the components to which proteins are conjugated.
chemical groups, this approach can be used to pre- A collection of proteins differing from one another
pare proteins for chemoselective conjugation to other only by the position of the UAA that participates in
surfaces or nanoscale moieties. Importantly, since the chemoselective conjugation could have a utility to nan-
UAAs are introduced by ribosomal protein synthe- otechnologists that closely parallels that of a collection
sis, the position of the UAA, and therefore the site of circularly permuted proteins. As Fig. 16.10 shows,
of chemoselective conjugation, is not limited to the circular permuted proteins chemoselectively conjugated
end of the protein sequence. Chemoselective conju- to surfaces can be used to tune distance between
gation can potentially be made to occur at any AA underlying surfaces and protein functional domains. In-
of the polypeptide sequence. Such UAA-incorporating corporation of UAAs and chemoselective conjugation
systems have been developed in prokaryotic, lower therefore might also be used for analogous purposes in
eukaryotic [16.77, 78], and mammalian expression sys- nanobiological devices.
secondary effects of therapy, some of which can be (proteins, nucleic acids) often constitute the active parts
debilitating or life threatening. One possible approach of biosensor interfaces. Whatever the chemical nature
to this issue is the incorporation of sensing capabil- of the interface, it determines the selectivity, sensi-
ity (specifically, the capacity to recognize appropriate tivity, and stability of the sensing system and also
contexts for therapeutic activity) into nanotherapeutic is a dominant determinant of sensor operational lim-
devices. Sensing capability may allow self-regulation its. Assay methodology determines the need (or lack
of a therapeutic device, reporting to an external clin- thereof) for analyte tracers, the number of analytical
ician/device operator (as in imaging applications, see reagents, and the complexity and rapidity of the sensing
below), or both. process.
Biosensors are typically considered to be multi-
functional, multicomponent devices [16.80]. Usually 16.3.1 Planar FET Protein Sensors
a biosensor system is composed of a signal transducer,
a sensor interface, a biological detection (bioaffinity) To illustrate the potential of biotechnology to solve
agent, and an associated assay methodology, with each nanoscale problems, we will focus on protein-sensing
system component governed by its inherent operational immunoFETs (a field-effect transistor sensor which
considerations. Signal transducers are moieties that are binds analyte via an antibody molecule or fragment
sensitive to a physicochemical change in their envi- thereof on its sensing surface). Field-effect transistors
ronment and that undergo some detectable change in (FETs) consist of a current source and current drain
chemistry, structure or state as the result of analyte (the separated from each other by a semiconductor channel
thing to be sensed) recognition. Analytes for nanoth- through which current flows. A gate to which elec-
erapeutic application could be biomolecules, such as trical bias is applied is positioned above the channel
proteins, small molecules (organic or inorganic), and to modulate channel conductance. Depending on gate
ions (salts or hydrogen ions), or physical conditions bias, current flow through the FET channel is increased
(such as redox state or temperature). Interfaces are or decreased (Fig. 16.11) [16.81–85]. Electrical prop-
the sensor components that interact directly with the erties of FET sensing channels can be modulated by
analyte. For sensor use in nanotherapeutic devices, im- any proximal electric field/charge acting as a gate, al-
Part B 16.3
mobilized or otherwise captured biological molecules lowing configuration of sensors [16.81–91], including,
in principal, protein sensors. Receptor molecules (anti-
body fragments, peptides, aptamers, etc.) that recognize
Analyte
• size protein analyte molecules are deployed on the sensing
Receptor • conformation channel, and analytes bind to the interface via the re-
• charge distribution
• size
A • epitope distribution
ceptors. Charges of protein analytes induce a dipole
• topology A A
• epitope recognition between the surface and the underlying depletion region
A A A of the semiconductor, eliciting a gating effect which
FET device
Polymer affects current through the FET. Differently charged
• thickness
• combosition Sensing channel oxide • composition analytes interact differentially with charge carriers in
• capacitance layer S D
• surface chemistry Semiconductor • uniformity the FET, producing analyte-charge-specific responses.
FET sensors can be constructed from nanowires (so-
Fig. 16.11 A protein-sensing FET and some of the parameters influ- called 1-D FETs). One-dimensional structures may have
encing its function. Current source and drain (S, D) are shown, as much enhanced sensitivity due to their high surface-
is the semiconductor (gray box). Capacitance layer/sensing chan- area-to-volume ratios [16.82–85]. An example of a 1-D
nel oxide is shown in brown. Current flowing from the source to FET was shown in Fig. 16.5, and 1-D FETs much
the drain in the semiconductor (ISD ) is indicated by the black ar- like those pictured are widely accepted as nanotech-
row, and increases or decreases in response to charges brought to nological devices, though meso- to macroscale planar
the sensing channel surface as a result of analyte binding. Recep- FET sensors often are not. The reason for consid-
tors (SCFvs, streptavidin, etc.) are shown as gray C-shaped lines ering 1-D FETs as real nanotechnology is that 1-D
and bind analyte (gray balls, labeled A) specifically but reversibly devices contain a nanoscale component (the semi-
and are conjugated to the sensing channel by an interfacial polymer. conducting nanowire). They also additionally contain
Parameters that might be chosen judiciously to optimize the sensor nanoscale receptor molecules. Thus, although physi-
are shown. Note that interfacial optimization occurs primarily on the cal nanotechnologists often chafe (very unreasonably)
nanoscale, with specific changes to molecular sizes and orientations at this argument, planar FETs are nanodevices for the
Biological Molecules in Therapeutic Nanodevices 16.3 Sensing Devices 473
same reason as 1-D nanowire FETs are. Planar FETs, where [16.90, 91] and will repeat the discussion only
though they themselves may be meso- to macroscale, briefly:
depend on obligately nanoscale components (biomolec-
1. Use of intact antibody as receptors as in Fig. 16.12 is
ular receptors for analytes) for their function. Here we
unnecessary: much smaller (10 ×), antigen-binding
focus on planar FETs as a platform to demonstrate
antibody fragments are available.
how biotechnology can solve a problem occurring at the
2. Antibodies do not adsorb exclusively at C3 domain
nanoscale.
(as shown in Fig. 16.12): antibodies bind surfaces in
The potential utility of FET protein sensing in phys-
a distribution of orientations.
iological environments is obvious, but has only recently
3. Antibodies contain a flexible region at the crotch of
been realized [16.90, 91]. FET sensor sensitivity to an-
the Y called the hinge that allows antigen-combining
alyte charge is limited by multiple factors [16.81–91],
sites vast freedom in their relative positions. The
though the key problem cited is typically regarded to be
implicit assumption illustrated in Fig. 16.12 is that
shielding by buffer ions in the high-salt (150 mM Na+ )
antibodies are rigid bodies, which is wrong.
physiological environment (Fig. 16.12) [16.81–84, 86].
4. The effect of antigen (analyte) three-dimensional
This led to an assessment that planar immunoFETs,
(3-D) shape and charge distribution on charge prox-
owing to the size of antibodies (10–12 nm) and the
imity to the sensing channel is not considered in
maximal distance over which analyte charges were
classical analysis.
thought to be detectable by the FET (sometimes called
5. The diversity available in native and phage-
the Debye length, < 2 nm), were fundamentally infeasi-
presented antibody repertoires was not exploited.
ble. The classical analysis of immunoFETs reasonably
The impact on sensor properties of analyte recep-
considers ion shielding, but is so fatally flawed in its
tors (antibodies or fragments thereof) of differing
consideration of receptor protein structures on FET sur-
valency, conformations, antigen (analyte) affinity or
faces that it is irrelevant. To biochemists, the fallacies
analyte epitope recognition properties were not con-
of the classical model are obvious, and biochemists en-
sidered.
countering the classical model immediately dismiss it
6. Improvement of antibody receptor properties by
as meaningless, though much of the sensing community
protein engineering is not considered in the classical
Part B 16.3
has clung tenaciously to the moribund model. We have
analysis.
discussed the fatal flaws of the classical analysis else-
7. Data directly shows that planar FET protein sensors
can detect a biologically important analyte at con-
Antigen centrations which occur in real biological systems
Antibody
under physiological salt conditions [16.90, 91].
Suffice it to say that, if planar immunoFETs are not
V Predicted feasible, the reasoning of the classical assessment has
10– C1 Debye length little to do with why they might be infeasible. However,
12 nm C2 we know from our work that planar FET protein sen-
C3 sors operating at physiological salt concentrations are
1–2 nm
S D feasible [16.90, 91], though we show that the distance
between the analyte charge and the sensing channels is
Fig. 16.12 The biochemical basis of the classical anal- a key parameter for sensor sensitivity. We will focus
ysis interpreted as indicating the infeasibility of planar on biotechnology approaches to engineer the nanome-
immunoFETs operating in physiological buffer. Artist’s ter distance between bound analyte charges and FET
conception of antibodies on FET sensing channel. Anti- sensing surfaces as an existence proof of the utility of
body domains (constant 1, 2, 3 as C1, C2, C3 and variable biotechnology to nanotechnology.
as V) are indicated, as is the predicted Debye length in
physiological buffer (150 mM salt, about 1–2 nm). FET 16.3.2 Biotechnology Approaches
source and drain (S and D) are shown. Shown is the classi- to the “Fundamental Limitations”
cal (but wrong, see text) depiction of antibodies adsorbed of Planar ImmunoFETs
onto a FET sensing channel solely by their C3 domains.
This flawed representation is central to the classical objec- First, in the classical model [16.81–84, 86], chemose-
tion to the feasibility of immunoFETs (after [16.81]) lective conjugation of receptors (antibodies) to sensing
474 Part B MEMS/NEMS and BioMEMS/NEMS
interfaces is not considered. Chemoselective conjuga- munoFET receptors. This was a major oversight, even
tion would have the desirable effect of orienting the at the time when the classical model was promulgated
antibody layer consistently, from one antibody molecule in the early 1990s. The tools available for antibody en-
to the next. Even though the classical representation gineering are vast and sophisticated. We have discussed
(as mimicked in Fig. 16.12) shows antibodies adsorbed a few of them above (receptor circular permutagene-
onto sensing surfaces in an oriented fashion, it is sis [16.72, 73] and introduction of UAAs [16.77–79],
well known to immunologists that antibodies adsorb both useful in the context of chemoselective conjuga-
to surfaces in a stochastic distribution of orientations, tion of the antibody or fragments thereof to the sensing
binding to the surfaces via a random distribution of interface). As Fig. 16.10 illustrates, the effect of circular
antibody surface regions. That is to say that the anti- permutagenesis of a SCFv is to change the orientation
body layer as contemplated in the classical assessment of its antigen-combining sites to the site of chemoselec-
is misrepresented, and the lack of orientation of the an- tive conjugation. Effectively, the orientation (relative to
tibodies at real interfaces would interfere with many the immunoFET sensing channel) and therefore the dis-
of the available biotechnology strategies to control the tance charges of analytes bound to antibody fragments
distance between the analyte charge and the sensing is different when analytes are bound to different CP an-
surface. However, the entire classical assessment of im- tibody variants. Since charges of protein analytes are
munoFETs [16.81–84, 86] is based on the incorrect not very mobile, the critical distance between analyte
representation of the extent of order in the protein ori- charges and sensing surfaces is different with different
entation shown in Fig. 16.12. This critical flaw alone circularly permuted variants of a single antibody frag-
invalidates the classical assessment, but there is much ment. In the context of the immunoFET, it should be
more wrong with the assessment than just this point. possible to identify those circularly permuted antibody
Secondly, use of intact antibodies (IgGs) as im- fragments that produce the greatest sensor sensitivity,
munoFET receptors is unnecessary: specific epitope presumably because they minimize the distance be-
recognition function can be isolated on fragments tween bound analyte charges and FET sensing surfaces.
several-fold smaller than intact antibodies [16.21, 92– Similarly, the optimal AA position for a UAA in an an-
99]. Single-chain Fv antibody fragments (single-chain tibody fragment could be determined for an antibody
Part B 16.3
fragment variables, SCFvs) are less than half the size fragment used as a FET receptor from the standpoint
of intact antibodies. Single-domain (camelid) antibody of immunoFET sensitivity. The approach could tune
fragments, called variable heavy–heavy (VHH), are the distance between underlying surfaces and analyte
smaller still: about 10% of the mass of intact antibod- charges, much as circular permutagenesis can.
ies [16.97–99]. There are multiple methods to convert There are still many other biotechnology ap-
existing antibody genes (isolated from mammalian B- proaches that might be used to optimize receptors for
cells) to genes that will direct biologic production of immunoFETs that have been ignored to date. We and
either SCFvs or VHHs recognizing the same antigen multiple others have identified peptides that can specif-
(analyte), as well as methods to isolate de novo SCFvs ically bind microfabricated surfaces [16.74–76]. Such
or VHHs that are directed to antigens of interest that are peptides might be inserted into SCFv or VHH se-
well known to those skilled in the art [16.97–99]. The quences and used to adhere the SCFv or VHH to the
simple expedient of using an antibody fragment as an micromachined surface, as we suggest [16.73, 74]. Pre-
immunoFET receptor reduces the distance between the sumably, variable positions of peptide insertion could
sensing surface and the bound analyte charge to nearly lead to differential orientations of antibody fragments
the calculated distance over which buffer counterion on the immunoFET surface, with the potential benefits
shielding is expected to occur (Fig. 16.12 [16.81–84, in terms of FET sensitivity already described for circu-
86]). Again, failure to consider this fact is in itself suf- lar permutagenesis or UAA insertion described above.
ficient to invalidate the classical assessment. Antibody This has the potentially great advantage of adhering
fragments are convenient substrates for chemoselec- immunoFET receptors directly to the sensing surface,
tive conjugation, and we exploit the protein orientation but our experience with direct adsorption of receptor
chemoselective approaches in our SCFv layers to engi- proteins on micromachined surfaces have proven direct
neer analyte–semiconductor distances. adsorption to be unsatisfactory in our hands (see below).
Thirdly, no consideration is given in the classi- The foregoing is a very incomplete list of poten-
cal model to engineering the antibody molecules or tial biotechnology approaches to addressing a limitation
fragments thereof to optimize their function as im- correctly identified (buffer counterion shielding of an-
Biological Molecules in Therapeutic Nanodevices 16.3 Sensing Devices 475
alyte charges) but incorrectly analyzed [16.81–84, 86]. surface via a biotinylated silane polymer (3-amino-
Many more approaches are possible, although we will propyltriethoxysilane, APTES Fig. 16.13) significantly
not attempt to enumerate them here. The point is that enhanced resistance to wear [16.100, 105, 106].
planar immunoFETs have a nanoscale limitation that
can be addressed by biotechnology. Hence, in this ap- a) b)
plication, biotechnology is a bona fide nanotechnology.
NH2 NH2
16.3.3 Nanotribology O O
of Protein-Sensing Interfaces O O
on Micromachined Surfaces 1.233 nm N N
Part B 16.3
surfaces [16.100, 104–106]) and on a sputter-coated SiO2/Al2O3
aluminum layer with a native oxide (aluminum sur-
faces [16.100]). Silicon surfaces simulate the sensing
surfaces of a metal–oxide–semiconductor field-effect d) R1 R1 R1
transistor (MOSFET), whereas aluminum simulates the
sensing surface of an AlGaN (aluminum–gallium ni- H3C Si CH3 H3C Si CH3 H3C Si CH3
tride) heterojunction FET. Native Al2 O3 oxide which
O O O
develops on sputter-coated aluminum surfaces is also
the most prevalent oxide on the surfaces of AlGaN/GaN
heterojunction FETs. All studies were performed with SiO2/Al2O3
the model receptor protein streptavidin, and not with an
antibody fragment. Fig. 16.13a–d Tri- and monoalkoxy silane constituents
Lee et al. and Bhushan et al. studied the step- of the sensor interface on micromachined substrates. In
by-step morphological changes resulting on silicon (a) and (b), oxyamine-terminated polymers for chemos-
surfaces during deposition of a model protein (strep- elective conjugation are shown. (a) shows the trialkoxy
tavidin) [16.104, 105]. These studies revealed, among interfacial polymer and (b) its monoalkoxy analogue.
other things, that streptavidin adsorbed to silicon in- As determined by summing bond lengths, the length of
teracted with the surface very weakly. Most of the polymer in nm is shown. Both ellipsometry (Fig. 16.14)
adsorbed streptavidin could be removed from the sur- and AFM analysis show that interfaces made with the
face by rinsing with an aqueous buffer [16.105]. monoalkoxy version of polymers are thinner and less
Streptavidin adherence by adsorption to micropatterned rough, presumably due to reduced intramolecular silane
asperities on the silicon surface and to the edges polymerization [16.100–103]. (c) shows a hypothetical film
of silicon fragments was modestly stronger than to made from trialkoxy polymer of (a), and (d) shows a hypo-
the flat silicon surface, although adhesion was still thetical film film made with the monoalkoxy polymer of
weak [16.104,105]. Adhering streptavidin to the silicon (b). R1 is (CH2 CH2 NHCOCH2 ONH2 ) (after [16.100])
476 Part B MEMS/NEMS and BioMEMS/NEMS
Ellipsometry of APTES-coated silicon surfaces APTES films applied by our original procedure were
showed the thickness of the polymer layer to be several- roughest, followed by APTES films deposited by the
fold the expected monolayer thickness [16.100]. The more optimal procedure, and APDMES films were
Part B 16.3
unexpected thickness of the film likely represents mul- the smoothest and most regular. This roughness rank-
tilayering by APTES, a phenomenon that had been ing was conserved after subsequent steps in interface
previously reported [16.101–103]. Despite the fact that construction (biotinylation of and streptavidin binding
the sensor with a multilayered APTES layer was suf- to the biotinylated silane polymer, Fig. 16.14 [16.100,
ficiently sensitive to detect analyte [16.90, 91], thinner 106]). The results observed on silicon surfaces were
interfaces would theoretically increase sensor sensitiv- replicated on aluminum surfaces, with APDMES pro-
ity to analyte. Multilayering is related to the trivalency viding a smoother, thinner interfacial polymer layer
of APTES siloxane residues and is influenced by than did APTES [16.100]. Furthermore, the rank order
the deposition method [16.101–103]. Consequently, of roughness following biotinylation was also con-
APTES polymer films were also constructed using served for APTES films deposited by our two protocols
a more optimized deposition protocol [16.101]. The and APDMES films on aluminum [16.100].
more optimal method produced thinner APTES films, Bhushan et al. [16.100] studied friction and wear
as expected (Fig. 16.14). However, peak-to-valley val- of streptavidin deposited by physical adsorption and
ues for APTES films deposited by either protocol were via a biotinylated APTES polymer layer on the silicon
substantially greater than the summed bond lengths surface. The coefficient of friction for streptavidin-
of APTES [16.100, 104–106]. The monosiloxane adsorbed silicon surfaces is less than for uncoated sili-
reagent, aminopropyldimethylethoxysilane (APDMES, con [16.100], because the streptavidin acts as a lubricant
Fig. 16.13), which has a much diminished capacity film. Consistent with this interpretation, coefficients of
to produce multilayers, was also used to build poly- friction were dependent on the concentration of strep-
mer films on silicon. Strikingly, APDMES produced tavidin (that is, on the surface density of adsorbed
the thinnest films, with thicknesses comparable to streptavidin molecules, which increases when higher
the summed bond lengths of the APDMES polymer concentrations of streptavidin were used in coating the
(Fig. 16.14). surfaces), and decreased at higher streptavidin concen-
The thicker silane films (based on APTES) were trations. At higher densities of adsorbed streptavidin,
also substantially rougher than the APDMES film. the surface is more uniform and smoother, and the sili-
Biological Molecules in Therapeutic Nanodevices 16.3 Sensing Devices 477
con substrate is more fully covered with protein than at centration. When streptavidin is linked to the surface
lower concentration. This implies that the streptavidin through a biotinylated silane polymer, the coefficient of
forms a nearly continuous lubricant film at higher con- friction increases. This follows from the increased com-
a) After wear test in air in contact mode and PBS in tapping mode
APDMES on SiO2 in air in contact mode Biotin over APDMES on SiO2 STA-biotin over APDMES on SiO2
in PBS in tapping mode in PBS tapping mode
Surface height Friction force Surface height Phase Surface height Phase
100 nN 50% of free amplitude(≈2 nN)
(nm) (nN) (nm) (nN) (nm) (deg)
3 1 3 2 3 2
–3 –1 –3 –2 –3 –2
6 2 6 4 6 4
(nm) (nN) (nm) (deg) (nm) (deg)
0 0 0 0 0 0
0 3 0 3 0 3 0 3 0 3 0 3
(µm) (µm) (µm) (µm) (µm) (µm)
200 nN 30% of free amplitude(≈3 nN)
(nm) (nN) (nm) (deg) (nm) (deg)
3 1 3 2 3 2
–3 –1 –3 –2 –3 –2
6 2 6 4 6 4
(nm) (nN) (nm) (deg) (nm) (deg)
0 0 0 0 0 0
Part B 16.3
0 3 0 3 0 3 0 3 0 3 0 3
(µm) (µm) (µm) (µm) (µm) (µm)
300 nN 10% of free amplitude(≈2 nN)
(nm) (nN) (nm) (deg) (nm) (deg)
3 1 3 2 3 2
–3 –1 –3 –2 –3 –2
6 2 6 4 6 4
(nm) (nN) (nm) (deg) (nm) (deg)
0 0 0 0 0 0
0 3 0 3 0 3 0 3 0 3 0 3
(µm) (µm) (µm) (µm) (µm) (µm)
b)
Average phase change (deg)
5 1 5 2 5 2
Average wear depth (nm)
4 0.8 4 4
3 0.6 3 3
1 1
2 0.4 2 2
1 0.2 1 1
0 0 0 0 0 0
0 100 200 300 400 0 1 2 3 4 0 1 2 3 4
Average normal load (nN) Average normal load (nN) Average normal load (nN)
Fig. 16.15a,b Differential wear of interfacial polymer–protein films by AFM operated in tapping mode. (a) AFM surface height
and friction force/phase angle images and cross-sectional profiles obtained after wear testing in air in contact mode/phosphate
buffered saline (PBS) in tapping mode at a range of normal loads and free amplitudes, and (b) plot of average wear depth and
average friction force/phase angle as a function of average normal load, for APDMES on silicon/SiO2 , biotin over APDMES on
silicon/SiO2 , and on STA–biotin on APDMES and silicon/SiO2 (after [16.100])
478 Part B MEMS/NEMS and BioMEMS/NEMS
pliance of the underlying silane polymer. When normal if the polymer layer were a true, un-cross-linked, self-
load is applied to the surface, the surface is compressed, assembled monolayer, individual polymer monomers
owing to the increased mechanical compliance afforded would have no way to transfer force applied on them
by interfacial biotinylated polymer length, flexibility to their neighboring monomers, and so would be indi-
or layering, all resulting in a larger contact area be- vidually plucked off the interface by the tip.
tween the AFM tip and interfacial molecules. Since These studies also [16.100, 106] occasioned further
the size of streptavidin is much larger than that of the insights into the mechanism of wear of polymer–protein
biotinylated silane polymer, the tightly packed strep- interfaces on micromachined surfaces. The polymer–
tavidin molecules cause very little lateral deflection of protein interfaces (biotinylated silane polymer–strept-
the tip. In streptavidin surfaces deposited by linkage to avidin) on silicon were sufficiently delicate that their
a polymer bound to the silicon surface, high contact wear could not be investigated by contact-mode AFM
area and low lateral deflection cause the friction force at any load tested. Contact mode always immediately
to increase relative to the same applied normal load on stripped the surface, even at the lowest loads achiev-
streptavidin surfaces deposited by adsorption [16.100, able. Consequently, wear experiments were performed
106]. Together, these effects produce a cushioning at varying loads in tapping mode, wherein the free
effect that runs counter to the lubricating effect ob- amplitude (and hence the load) of the cantilever was
served when protein (streptavidin) is directly adsorbed systematically varied (Fig. 16.15). As one might expect,
to the surfaces, increasing the coefficients of friction as load increased, wear increased. The finding implies
for streptavidin deposited by a polymer interface. The that wear of APDMES–streptavidin interfaces on sili-
observations of streptavidin’s lubricating properties and con is not a concerted process, but proceeds sequentially
polymer–protein interfacial cushioning properties have more deeply as the applied force increases. That said, all
been replicated on aluminum surfaces [16.100]. We tested interfaces were very delicate, and required only
think these are likely general properties of many protein nanonewton forces to remove them from the surfaces.
interfaces on micromachined surfaces. We expect that It can reasonably be expected that unprotected im-
many proteins (such as SCFvs adsorbed onto various munoFET sensors in vivo will experience at least
metal oxide surfaces) will produce analogous lubricat- nanonewton abrasive forces. Such forces are likely to be
Part B 16.4
ing properties, and that cushioning-mediated friction sufficient to strip receptor layers from the immunoFET,
increases will occur for many polymer–protein inter- and thereby cause sensor failure. This implies that
faces, regardless of the composition of the underlying the sensor polymer–receptor interfaces must either be
smooth, micromachined surfaces. hardened somehow to resist encountered abrasive in-
Interestingly, streptavidin bound to silicon sur- sult, or that the sensor interfaces must be sheltered
faces via biotinylated silane polymers had differential from direct, abrasive insult by tissue. This problem is
resistance to wear, with APTES interfaces (which ex- not satisfactorily solved, but is currently under inves-
hibit the most multilayering) being less robust than tigation. However, the point remains that the primacy
interfaces based on APDMES (with limited/no mul- of the importance of the immunoFET interface drove
tilayering). The most highly multilayered polymer surface-science and nanotribological study of the inter-
interfaces had, by definition, larger extents of inter- face. As a consequence, we documented properties new
monomer cross-linking, and were less resistant to wear to nanoscale tribology of protein interfaces (lubricating
by cantilevers [16.100]. This was interpreted by analogy and cushioning effects). Nanotribological characteriza-
to graphene layers in graphite, in that bonds between tion of models of immunoFET surfaces also revealed
monomers of the polymer layer were more numerous another nanoscale parameter that must be addressed
than bonds between the polymer layer and the silicon before the immunoFETs can be deployed in in vivo
substrate. Hence, when the polymer layer tore under clinical or research applications: the robustness to
stress from the cantilever, a larger section of the cross- mechanical insult (or lack thereof) of FET receptor
linked polymer sheet tore free. It was expected that, layers.
Previous versions of this chapter ended with a discus- reflected the proposer’s or author’s naivete. The direct
sion of the complexity of biology (being much greater response was often that the issue of planar FET sen-
than encountered in traditional engineering disciplines), sors had been dealt with since 1991, and did not warrant
the fashion in which biologists prioritize information, further consideration. This response might have made
and cultural differences between engineers and biolog- sense had the classical assessment been theoretically
ical scientists [16.107]. The chapter discussed some of sound, or if it were unsound in some way that required
the ways in which biologists and engineers hold and dis- research to discover. However, the assessment was un-
seminate information (approaches so vastly different as sound from its first enunciation, largely because it did
to interfere with design of nanobiological devices). It not take into account knowledge about protein and an-
also discussed the (then largely unmet) need for signif- tibody structure that was available long before 1991. In
icant biological education for biomedical engineers, at the original proposal of the assessment, this may have
least those who will be involved in producing devices been a result of the (then) poorly appreciated inter-
that interact with living systems in vivo, or contain com- disciplinary nature of biosensor technology, or naivete
ponents derived from living systems. The consensus regarding antibody structures and properties.
regarding the need for biologic training for biomedi- The classical assessment was promulgated by elec-
cal engineers appears to be shifting in favor of more trical engineers who had little knowledge of antibody
biology, in no small part as a result of the increas- structure, so they made mistakes in aspects of the as-
ing prevalence of biologic components in biomedical sessment dependent on understanding antibodies. In its
nano- and microdevices and attempts to transfer those general form, this problem is a pitfall to which all in-
technologies to the clinic. However, there are still terdisciplinary scientists are vulnerable. By definition,
conundrums related to how biomedical and other engi- one cannot know what one does not know, and to
neers see themselves and see biology that result in lost some extent ignorance must be forgiven, as it is the
opportunities. default human condition. However, it should not have
taken a working planar bioFET to overturn the classi-
16.4.1 You Do Not Know cal assessment. The classical assessment was dead on
What You Do Not Know: arrival. Immunologists knew on cursory inspection that
Part B 16.4
the Consequences of Certainty the classical assessment was nearly without relevance to
real protein sensing. Unfortunately, engineers did not.
As discussed above, planar immunoFETs and similar In many cases, they failed to reexamine the assump-
bioFETs operating at physiological salt concentrations tions of the classical assessment, in favor of reiterating
were regarded as infeasible from 1991 up until re- its conclusion. If anything was done improperly, it was
cently [16.81–84,86]. However, there are now examples in failing to consider reasoned discussion of the flaws in
of such planar FET protein sensors that can detect the assessment (as above). Doing so may have cost the
protein analytes in physiological buffers at analyte con- sensor field 20 years.
centrations known to occur in vivo [16.90, 91]. While This is mentioned without the intent to embarrass
this might seem to be a breakthrough in sensing tech- persons who proposed or adhered to the classical as-
nology, this development was delayed for years longer sessment, but rather to point out that dogmatic certainty
than it needed to be by the intransigent adherence of has consequences. Error is an ongoing hazard in any in-
biosensor specialists to a flawed assessment of feasibil- terdisciplinary field. As biomedical nanotechnologists,
ity [16.81–84, 86]. we mention this story with trepidation for our own
The irony was that, with the right FET platform, pla- pronouncements.
nar FET sensing of proteins in physiological buffers has
always been possible. The sensing community could 16.4.2 Are Proteins and Molecules
have discovered this fact 20 years ago, and potentially Legitimately Part
could have moved to making functional immunoFET of Nanotechnology?
sensors for clinical use at that time, but for stubborn
adherence to a flawed model of planar immunoFET There is a school of thought that holds that nano-
interface structure. Until as little as a year ago, fund- technology deals solely with quantum and other phe-
ing agencies and journals responded to applications or nomena that manifest only when matter is subdivided
papers discussing planar FETs for protein sensing at into nanoscale particles or structures. This reasoning
physiological salt concentrations as if the documents would exclude supramolecular chemistry depending on
480 Part B MEMS/NEMS and BioMEMS/NEMS
organic molecules and devices built from protein or nu- The molecular nanocomponents of the immunoFET
cleic acid functional components from nanotechnology. and IKVAV PA are tuned to suit (at least partially)
If this reasoning is accepted as valid, nanotechnology so using systematic knowledge of nanocomponent proper-
defined may have little intersection with human health. ties. It is our opinion that engineering is any activity
As the reader might assume, such reasoning does that uses knowledge of component properties to con-
not appeal to us. We hope we have presented a com- trol the behavior of materials and devices containing
pelling case that molecules are essential components of those components to achieve a desired end. In the
nanotherapeutics. Organic or biologic macromolecules case of the biological molecules with which we work,
(proteins, nucleic acids, etc.) are necessary for many knowledge of properties is frequently fragmentary, lim-
interactions between living things and therapeutic de- iting our ability to design individual protein molecules
vices. In part, this is because these molecules are de novo, and driving us to use molecular genetic screen-
dimensionally and chemically appropriate to interact ing techniques to identify variant proteins with desired
with the biocomponent of the patient (i. e., the argu- properties. Nonetheless, we see ourselves as engineers
ment of speaking to the patient’s body in a language because we use knowledge of our materials (proteins) to
it understands). Also, synthetic nanostructures with the its limits to optimize functional devices [immunoFETs,
precise chemical and topological complexity of biolog- magnetic resonance imaging (MRI) contrast agents]. In
ical macromolecules cannot yet be produced or tuned the future, protein structure is likely to become more
with the same efficacy as biotechnology produces and amenable to quantitative modeling and design, and the
tunes the properties of proteins (i. e., the argument that inclusion of protein engineering as a genuine engineer-
biomolecules are the only show in town). Moreover, ing discipline will become less controversial.
with the examples of the FET protein sensor and the IK- At any rate, fortunately for us, and for biomedical
VAV PA amphiphile, we have shown that biotechnology nanotechnologists, the restrictive view of the definition
and organic chemistry (respectively) can be used to ad- of nanotechnology and engineering is less widely held
dress problems occurring at the nanoscale. In this sense, than in the past. These fields will likely become increas-
they fall reasonably under the rubric of nanotechnology. ingly welcoming to biologists as time passes.
Part B 16
References
16.1 S.C. Lee, R. Parthasarathy, T. Duffin, K. Botwin, Nanotechnology: The Convergence of Biotechnol-
T. Beck, G. Lange, J. Zobel, D. Jansson, D. Kun- ogy, Polymer Chemistry and Materials Science, ed.
neman, E. Rowold, C.F. Voliva: Antibodies to by S.C. Lee, L. Savage (IBC, Southborough 1998)
PAMAM dendrimers: Reagents for immune detec- pp. 3–14
tion assembly and patterning of dendrimers. In: 16.6 S.C. Lee: A biological nanodevice for drug deliv-
Dendrimers and Other Dendritic Polymers, ed. by ery, National Science and Technology Council. IWGN
D. Tomalia, J. Frechet (Wiley, London 2001) pp. 559– Workshop Report: Nanotechnology Research Direc-
566 tions. International Technology Research Institute,
16.2 S.C. Lee: Biotechnology for nanotechnology, Trends World Technology Division (Kluwer, Baltimore 1999)
Biotechnol. 16, 239–240 (1998) pp. 91–92
16.3 S.C. Lee: Engineering the protein components of 16.7 S.C. Lee, R. Parthasarathy, K. Botwin: Proteinpoly-
nanobiological devices. In: Biological Molecules in mer conjugates: Synthesis of simple nanobiotech-
Nanotechnology: The Convergence of Biotechnol- nological devices, Polym. Prepr. 40, 449–450
ogy, Polymer Chemistry and Materials Science, ed. (1999)
by S.C. Lee, L. Savage (IBC, Southborough 1998) 16.8 L. Jelinski: Biologically related aspects of nanopar-
pp. 67–74 ticles, nanostructured materials and nanodevices.
16.4 S.C. Lee: How a molecular biologist can wind up In: Nanostructure Science and Technology, ed. by
organizing nanotechnology meetings. In: Biologi- R.W. Siegel, E. Hu, M.C. Roco (Kluwer, Dordrecht
cal Molecules in Nanotechnology: The Convergence 1999) pp. 113–130
of Biotechnology, Polymer Chemistry and Materials 16.9 B.R. Smith, J. Heverhagen, M. Knopp, P. Schmal-
Science, ed. by S.C. Lee, L. Savage (IBC, Southbor- brock, J. Shapiro, M. Shiomi, N. Moldovan,
ough 1998) M. Ferrari, S.C. Lee: Magnetic resonance imag-
16.5 S.C. Lee: The nanobiological strategy for construc- ing of atherosclerosis in vivo using biochemically
tion of nanodevices. In: Biological Molecules in targeted ultrasmall superparamagnetic iron oxide
Biological Molecules in Therapeutic Nanodevices References 481
particles (SPIONs), Biomed. Microdevices 9, 719–728 uloviruses (bacmids) in E. coli, Gene 160, 191–194
(2007) (1995)
16.10 A.J. Nijdam, T.R. Nicholson III, J. Shapiro, 16.24 V.A. Luckow, S.C. Lee, G.F. Barry, P.O. Olins: Ef-
B.R. Smith, J.T. Heverhagen, P. Schmalbrock, ficient generation of infectious recombinant bac-
M.V. Knopp, A. Kebbel, D. Wang, S.C. Lee: Biochem- uloviruses by site-specific, transposon-mediated
ically targeted nanoparticulate contrast agents for insertion of foreign DNA into a baculovirus genome
magnetic resonance imaging diagnosis of cardio- propagated in E. coli, J. Virol. 67, 4566–4579
vascular disease, Curr. Nanosci. 5, 88–102 (2009) (1993)
16.11 J.R. Baker Jr.: Therapeutic nanodevices. In: Biolog- 16.25 T. Gardner, C.R. Cantor, J.J. Collins: Construction
ical Molecules in Nanotechnology: The Convergence of a genetic toggle switch in E. coli, Nature 403,
of Biotechnology, Polymer Chemistry and Materials 339–342 (2000)
Science, ed. by S.C. Lee, L. Savage (IBC, Southbor- 16.26 J. Hasty, F. Isaacs, M. Dolnik, D. McMillen,
ough 1998) pp. 173–183 J.J. Collins: Designer gene networks: Towards fun-
16.12 R. Duncan: Drug targeting: Where are we now and damental cellular control, Chaos 11, 107–220 (2001)
where are we heading?, J. Drug Target. 5, 1–4 (1997) 16.27 S. Uppuluri, D.R. Swanson, L.T. Piehler, J. Li,
16.13 R. Duncan, S. Gac-Breton, R. Keane, Y.N. Sat, G. Hagnauer, D.A. Tomalia: Core shell tecto(den-
R. Satchi, F. Searle: Polymer-drug conjugates, drimers). I. Synthesis and characterization of
PDEPT and PELT: Basic principles for design and saturated shell models, Adv. Mater. 12, 796–800
transfer from the laboratory to clinic, J. Cont. Re- (2000)
lease 74, 135–146 (2001) 16.28 A.K. Patri, I.J. Majoros, J.R. Baker Jr.: Dendritic
16.14 D.S. Goldin, C.A. Dahl, K.L. Olsen, L.H. Ostrach, polymer macromolecular carriers for drug delivery,
R.D. Klausner: Biomedicine. The NASA-NCI col- Curr. Opin. Chem. Biol. 6, 466–471 (2002)
laboration on biomolecular sensors, Science 292, 16.29 A. Quintana, E. Raczka, L. Piehler, I. Lee, A. Myc,
443–444 (2001) I. Majoros, A.K. Patri, T. Thomas, J. Mule, J.R. Baker
16.15 S.C. Lee: Dendrimers in nanobiological devices. In: Jr.: Design and function of a dendrimer-based
Dendrimers and Other Dendritic Polymers, ed. by therapeutic nanodevice targeted to tumor cells
D. Tomalia, J. Frechet (Wiley, London 2001) pp. 548– through the folate receptor, Pharma. Res. 19, 1310–
557 1316 (2002)
16.16 J.R. Baker Jr., A. Quintana, L. Piehler, M. Banazak- 16.30 Y. Choi, T. Thomas, A. Kotlyar, M. Islam, J. Baker
Part B 16
Holl, D. Tomalia, E. Racka: The synthesis and testing Jr.: Synthesis and functional evaluation of DNA-
of anti-cancer therapeutic nanodevices, Biomed. assembled polyamidoamine dendrimer clusters for
Microdevices 3, 61–69 (2001) cancer cell-specific targeting, Chem. Biol. 12, 35–43
16.17 K.D. Bhalerao, E. Eteshola, M. Keener, S.C. Lee: (2005)
Nanodevice design through the functional abstrac- 16.31 Y. Choi, A. Mecke, B.G. Orr, M.M. Banaszak Holl,
tion of biological macromolecules, Appl. Phys. Lett. J.R. Baker Jr.: DNA-directed synthesis of generation
87, 143902–143904 (2005) 7 and 5 PAMAM dendrimer nanoclusters, Nano Lett.
16.18 S.C. Lee, K. Bhalerao, M. Ferrari: Object ori- 4, 391–397 (2004)
ented design tools for supramolecular devices and 16.32 D.G. Mullen, A.M. Desai, J.N. Waddell, X.-M. Cheng,
biomedical nanotechnology, Ann. New York Acad. C.V. Kelly, D.Q. McNerny, I.J. Majoros, J.R. Baker
Sci. 1013, 110–123 (2004) Jr., L.M. Sander, B.G. Orr, M.M. Banaszak Holl:
16.19 J.M. Harris, N.E. Martin, M. Modi: Pegylation: The implications of stochastic synthesis for the
A novel process for modifying pharmacokinetics, conjugation of functional groups to nanoparticles,
Clin. Pharmacokin. 40, 539–551 (2001) Bioconjug. Chem. 19, 1748–1752 (2008)
16.20 S.B.H. Kent: Building proteins through chemistry: 16.33 T.R. Groves, D. Pickard, B. Rafferty, N. Crosland,
Total chemical synthesis of protein molecules by D. Adam, G. Schubert: Maskless electron beam
chemical ligation of unprotected protein segments. lithography: Propects, progress and challenges,
In: Biological Molecules in Nanotechnology: The Microelectron. Eng. 61, 285–293 (2002)
Convergence of Biotechnology, Polymer Chemistry 16.34 M. Guthold, R. Superfine, R. Taylor: The rules
and Materials Science, ed. by S.C. Lee, L. Savage are changing: Force measurements on single
(IBC, Southborough 1998) pp. 75–92 molecules and how they relate to bulk reaction ki-
16.21 C.A. Janeway, P. Travers, M. Walport, J.D. Capra: netics and energies, Biomed. Microdevices 3, 9–18
Immunobiology (Elsevier, London 1999) (2001)
16.22 S.C. Lee, M.S. Leusch, V.A. Luckow, P. Olins: Method 16.35 L.M. Demers, D.S. Ginger, S.-J. Park, Z. Li,
of producing recombinant viruses in bacteria, US S.-W. Chung, C.A. Mirkin: Direct patterning of mod-
Patent 5348886 (1993) ified oligonucleotides on metals and insulatos by
16.23 M.S. Leusch, S.C. Lee, P.O. Olins: A novel hostvector dip-pen nanolithography, Science 296, 1836–1838
system for direct selection of recombinant bac- (2002)
482 Part B MEMS/NEMS and BioMEMS/NEMS
16.36 K.-B. Lee, S.-J. Park, C.A. Mirkin, J.C. Smith, in Laboratory Medicine, ed. by V. Lorian (Williams
M. Mrksich: Protein nanoarrays generated by dip- and Wilkins, Baltimore 1984) pp. 474–489
pen nanolithography, Science 295, 1702–1705 (2002) 16.53 M.J. Heller: Utilization of synthetic DNA for mo-
16.37 M. Ferrari, J. Liu: The engineered course of treat- lecular electronic and photonic-based device
ment, Mech. Eng. 123, 44–47 (2001) applications. In: Biological Molecules in Nano-
16.38 K.E. Drexler: Engines of Creation: The Coming Era technology: The Convergence of Biotechnology,
of Nanotechnology (Anchor Books, New York 1986) Polymer Chemistry and Materials Science, ed.
16.39 J. Cumings, A. Zetti: Low-friction nanoscale linear by S.C. Lee, L. Savage (IBC, Southborough 1998)
bearing realized frommultiwall carbon nanotubes, pp. 59–66
Science 289, 602–604 (2000) 16.54 Z. Ma, S. Taylor: Nucleic acid triggered catalytic drug
16.40 D.J. Hornbaker, S.-J. Kahng, S. Mirsa, B.W. Smith, release, Proc. Natl. Acad. Sci. USA 97, 11159–11163
A.T. Johnson, E.J. Mele, D.E. Luzzi, A. Yazdoni: (2000)
Mapping the one-dimensional electronic states of 16.55 R.C. Merkle: Biotechnology as a route to nanotech-
nanotube peapod structures, Science 295, 828–831 nology, Trends Biotechnol. 17, 271–274 (1999)
(2002) 16.56 N.C. Seeman, J. Chen, Z. Zhang, B. Lu, H. Qiu,
16.41 C. Dekker: Carbon nanotubes as molecular quan- T.-J. Fu, Y. Wang, X. Li, J. Qi, F. Liu, L.A. Wen-
tum wires, Phys. Today 28, 22–28 (1999) zler, S. Du, J.E. Mueller, H. Wang, C. Mao, W. Sun,
16.42 M.-C. Jones, J.-C. Leroux: Polymeric micellesa Z. Shen, M.H. Wong, R. Sha: A bottom-up ap-
new generation of colloidal drug carriers, Eur. proach to nanotechnology using DNA. In: Biological
J. Pharma. Biopharma. 48, 101–111 (1999) Molecules in Nanotechnology: The Convergence of
16.43 I. Uchegbu: Parenteral drug delivery: 1, Pharma. J. Biotechnology, Polymer Chemistry and Materials
263, 309–318 (1999) Science, ed. by S.C. Lee, L. Savage (IBC, Southbor-
16.44 I. Uchegbu: Parenteral drug delivery: 2, Pharma. J. ough 1998) pp. 45–58
263, 355–359 (1999) 16.57 G. Lemieux, C. Bertozzi: Chemoselective ligation re-
16.45 J.D. Hartgerink, E.R. Zubarev, S.I. Stupp: Supramolec- actions with proteins, oligosaccharides and cells,
ular one-dimensional objects, Curr. Opin. Solid Trends Biotechnol. 16, 506–512 (1998)
State Mater. Sci. 5, 355–361 (2001) 16.58 R. Offord, K. Rose: Multicomponent synthetic
16.46 L.C. Palmer, Y.S. Velichko, M. Olvera De La Cruz, constructs. In: Biological Molecules in Nano-
S.I. Stupp: Supramolecular self-assembly codes for technology: The Convergence of Biotechnology,
Part B 16
functional structures, Philos. Trans. R. Soc. A 365, Polymer Chemistry and Materials Science, ed.
1417–1433 (2007) by S.C. Lee, L. Savage (IBC, Southborough 1998)
16.47 G.A. Silva, C. Catherine, K.L. Niece, E. Beniash, pp. 93–105
D.A. Harrington, J.A. Kessler, S.I. Stupp: Selective 16.59 S.C. Lee, R. Parthasarathy, K. Botwin, D. Kun-
differentiation of neural progenitor cells by high- neman, E. Rowold, G. Lange, J. Zobel, T. Beck,
epitope density nanofibers, Science 303, 1352–1355 T. Miller, W. Hood, J. Monahan, R. Jansson, J.P. Mc-
(2004) Kearn, C.F. Voliva: Biochemical and immunological
16.48 V.M. Tysseling-Mattiace, V. Sahni, K.L. Niece, properties of cytokines conjugated to dendritic
D. Birch, C. Czeisler, M. Fehlings, S.I. Stupp, polymers, Biomed. Microdevices Biomems Biomed.
J.A. Kessler: Self-assembling nanofibers inhibit Nanotechnol. 6, 191–201 (2004)
glial scar gormation and promote axon elongation 16.60 S.C. Lee, R. Parthasarathy, T. Duffin, K. Botwin,
after spinal cord injury, J. Neurosci. 28, 3814–3823 T. Beck, G. Lange, J. Zobel, D. Kunneman,
(2008) E. Rowold, C.F. Voliva: Recognition properties of
16.49 S. Fernandez-Lopez, H.-S. Kim, E.C. Choi, M. Del- antibodies to PAMAM dendrimers and their use
gado, J.R. Granja, A. Khasanov, K. Kraehen- in immune detection of dendrimers, Biomed. Mi-
buehl, G. Long, D.A. Weinberger, K.M. Wilcoxen, crodevices 3, 51–57 (2001)
M. Ghardiri: Antibacterial agents based on the 16.61 G.T. Hermanson: Bioconjugate Chemistry (Aca-
cyclic D,L-alpha-peptide architecture, Nature 412, demic, San Diego 1996)
452–455 (2001) 16.62 S. Topell, R. Glockshuber: Circular permutation of
16.50 A. Saghatelian, Y. Yokobayashi, K. Soltani, the green fluorescent protein, Meth. Mol. Biol. 183,
M.R. Ghadiri: A chiroselective peptide replicator, 31–48 (2002)
Nature 409, 777–778 (2001) 16.63 A. Rojas, S. Garcia-Vallve, J. Palau, A. Romeu:
16.51 T. Hamouda, A. Myc, B. Donovan, A.Y. Shih, Circular permutations in proteins, Biologia 54,
J.D. Reuter, J.R. Baker Jr.: A novel surfactant na- 255–277 (1999)
noemulsion with a unique non-irritant topical 16.64 T.U. Schwartz, R. Walczak, G. Blobel: Circular per-
antimicrobial activity against bacteria, enveloped mutation as a tool to reduce surface entropy
viruses and fungi, Microbiol. Res. 156, 1–7 (2001) triggers crystallization of the signal recognition
16.52 J. Davies: Aminoglycoside-aminocyclitol antibi- particle receptor beta subunit, Protein Sci. 13,
otics and their modifying enzymes. In: Antibiotics 2814–2818 (2004)
Biological Molecules in Therapeutic Nanodevices References 483
16.65 U. Heinemann, M. Hahn: Circular permutation of 16.78 J. Xie, P.G. Schultz: A chemical toolkit for proteins
polypeptide chains: Implications for protein fold- – an expanded genetic code, Nat. Rev. Mol. Cell
ing and stability, Prog. Biophys. Mol. Biol. 64, Biol. 7, 775–782 (2006)
121–143 (1996) 16.79 N. Hino, A. Hayashi, K. Sakamoto, S. Yokoyama:
16.66 A. Buchwalder, H. Szadkoski, K. Kirschner: A fully Site-specific incorporation of non-natural amino
active variant of dihydrofolate reductase with acids into proteins in mammalian cells with an
a circularly permuted sequence, Biochem. 31, 1621– expanded genetic code, Nat. Protoc. 1, 2957–2962
1630 (1992) (2006)
16.67 L.S. Mullins, K. Wesseling, J.M. Kuo, J.B. Garrett, 16.80 K. Rogers: Principles of affinity-based biosensors,
F.M. Raushel: Transposition of protein sequences: Mol. Biotechnol. 14, 109–129 (2000)
Circular permutation of ribonuclease T1, J. Am. 16.81 M.J. Schoning, A. Poghossian: Recent advances
Chem. Soc. 116, 5529–5533 (1994) in biologically sensitive field-effect transistors
16.68 M. Hahn, K. Piotukh, R. Borriss, U. Heinemann: (BioFETS), Analyst 127, 1137–1151 (2002)
Native-like in vivo folding of a circularly per- 16.82 P. Bergveld, J. Hendrikes, W. Olthuis: Theory and
muted jellyroll protein shown by crystal structure application of the material work function for
analysis, Proc. Natl. Acad. Sci. USA 91, 10417–10421 chemical sensors based on the field effect prin-
(1994) ciple, Meas. Sci. Technol. 9, 1801–1808 (1998)
16.69 Y.R. Yang, H.K. Schachznan: Aspartate transcar- 16.83 W. Olthius, P. Bergveld, J. Kruise: The exploita-
bamoylase containing circularly permuted catalytic tion of ISFETs to determine acid-base behavior of
polypeptide chains, Proc. Natl. Acad. Sci. USA 90, proteins, Electrochim. Acta 43, 3483–3488 (1997)
11980–11984 (1993) 16.84 R.B. Schasfoort, R.P. Kooyman, P. Bergveld,
16.70 X. Lin, G. Koelsch, J.A. Loy, J. Tang: Rearranging J. Greve: A new approach to ImmunoFET operation,
the domains of pepsinogen, Protein Sci. 4, 159–166 Biosens. Bioelectron. 5, 103–124 (1990)
(1995) 16.85 Y. Cui, Q. Wei, H. Park, C.M. Lieber: Nanowire
16.71 M.L. Vignais, C. Corbier, G. Mulliert, C. Bran- nanosensors for highly sensitive and selective de-
lant, G. Branlant: Circular permutation within tection of biological and chemical species, Science
the coenzyme binding domain of the tetrameric 293, 1289 (2001)
glyceraldehyde-3-phosphate dehydrogenase from 16.86 P. Bergveld: A critical evaluation of direct electrical
Bacillus stearothermophilus, Protein Sci. 4, 994– protein detection methods, Biosens. Bioelectron.
Part B 16
1000 (1995) 6, 55–72 (1991)
16.72 E. Eteshola, C.D. Van Valkenburgh, S. Merlin, 16.87 J.I. Hahm, C.M. Lieber: Direct ultrasensitive electri-
E. Rowold, J. Adams, R. Ibdah, L.E. Pegg, A. Donelly, cal detection of DNA and DNA sequence variations
E. Rowold, J. Klover, S.C. Lee: Screening of a li- using nanowire nanosensors, Nano Lett. 4, 51–54
brary of circularly permuted proteins on phage (2004)
to manipulate protein topography, J. Nanoeng. 16.88 A. Star, J.C.P. Gabriel, K. Bradley, G. Gruner: Elec-
Nanosyst. 219, 45–55 (2006) tronic detection of specific protein binding using
16.73 E. Eteshola, M.T. Keener, M.A. Elias, J. Shapiro, nanotube FET devices, Nano Lett. 3, 459–463 (2003)
L.J. Brillson, B. Bhushan, S.C. Lee: Engineering 16.89 G. Zheng, F. Patolsky, Y. Cui, W.U. Wang,
functional protein interfaces for immunologically C.M. Lieber: Multiplexed electrical detection of
modified field effect transistors (ImmunoFETs) by cancer markers with nanowire sensor arrays, Nat.
molecular genetic means, J. R. Soc. Interface 5, Biotechnol. 23, 1294–1301 (2005)
123–127 (2008) 16.90 J. Shapiro, S. Gupta, E. Eteshola, M. Elias, X. Wen,
16.74 E. Eteshola, L. Brillson, S.C. Lee: Selection and char- W. Lu, L.J. Brillson, S.C. Lee: Challenges in
acteristics of peptides that bind thermally grown optimization of nanobiotechnological devices il-
silicon dioxide films, Biomol. Eng. 22, 202–204 lustrated by partial optimization of a protein
(2005) biosensor, Proc. 2nd Int. Congr. Nanobiotechnol.
16.75 R.R. Naik, L.L. Brott, S.J. Clarson, M.O. Stone: Silica- Nanomed., NanoBio 2007 (Int. Association Nano-
precipitating peptides isolated from a combina- technology, San Jose 2007), (CD)
torial phage display peptide library, J. Nanosci. 16.91 S. Gupta, M. Elias, X. Wen, J. Shapiro, L. Brillson,
Nanotechnol. 2, 95–100 (2002) W. Lu, S.C. Lee: Detection of clinically relevant lev-
16.76 E.M. Krauland, B.R. Peelle, K.D. Wittrup, A.M. Bel- els of biological analyte under physiologic buffer
cher: Peptide tags for enhanced cellular and using planar field effect transistors, Biosens. Bio-
protein adhesion to single-crystalline sapphire, electron. 24, 505–511 (2008)
Biotechnol. Bioeng. 97, 1009–1020 (2007) 16.92 K. Decanniere, A. Desmyter, M. Lauwereys,
16.77 L. Wang, J. Xie, P.G. Schultz: Expanding the genetic M.A. Ghahroudi, S. Muyldermans, L. Wyns:
code, Annu. Rev. Biophys. Biomol. Struct. 35, 225– A single-domain antibody fragment in complex
249 (2006) with rnase a: non-canonical loop structures and
484 Part B MEMS/NEMS and BioMEMS/NEMS
nanomolar affinity using two CDR loops, Struct. mina based surfaces, J. R. Soc. Interface 6, 719–733
Fold Des. 7, 361–370 (1999) (2009)
16.93 K. Decanniere, T.R. Transue, A. Desmyter, D. Maes, 16.101 Y. Han, D. Mayer, A. Offenhausser, S. Ingebrandt:
S. Muyldermans, L. Wyns: Degenerate interfaces Surface activation of thin silicon oxides by wet
in antigen-antibody complexes, J. Mol. Biol. 313, cleaning and silanization, Thin Solid Films 510,
473–478 (2001) 175–180 (2006)
16.94 A. Desmyter, K. Decanniere, S. Muyldermans, 16.102 K. Kallury, P.M. MacDonald, M. Thompson: Ef-
L. Wyns: Antigen specificity and high affinity bind- fect of surface water and base catalysis on the
ing provided by one single loop of a camel silanization of silica by (aminopropyl)alkoxysilanes
single-domain antibody, J. Biol. Chem. 276, studied by x-ray photoelectron spectroscopy and
26285–26290 (2001) 13C cross-polarization/magic angle spinning nu-
16.95 A. Muruganandam, J. Tanha, S. Narang, D. Sta- clear magnetic resonance, Langmuir 10, 492–499
nimirovic: Selection of phage-displayed llama (1994)
single-domain antibodies that transmigrate across 16.103 J.H. Moon, J.W. Shin, S.Y. Kim, J.W. Park: Forma-
human blood-brain barrier endothelium, FASEB tion of uniform aminosilane thin layers: An imine
Journal 16, 240–242 (2002) formation to measure relative surface density of
16.96 S. Muyldermans: Single domain camel antibodies: the amine group, Langmuir 12, 4621–4624 (1996)
Current status, Mol. Biotechnol. 74, 277–302 (2001) 16.104 B. Bhushan, D.R. Tokachichu, M.T. Keener, S.C. Lee:
16.97 S. Muyldermans, M. Lauwereys: Unique single- Morphology and adhesion of biomolecules on sili-
domain antigen binding fragments derived from con based surfaces, Acta Biomater. 1, 327–341 (2005)
naturally occurring camel heavy-chain antibodies, 16.105 S.C. Lee, M.T. Keener, D.R. Tokachichu, B. Bhushan,
J. Mol. Recogn. 12, 131–140 (1999) P.D. Barnes, B.J. Cipriany, M. Gao, L.J. Brillson: Pro-
16.98 L. Riechmann, S. Muyldermans: Single domain an- tein binding on thermally grown silicon dioxide,
tibodies: comparison of camel VH and camelised J. Vac. Sci. Technol. B 23, 1856–1865 (2005)
human VH domains, J. Immunol. Meth. 231, 25–38 16.106 B. Bhushan, D.R. Tokachichu, M.T. Keener, S.C. Lee:
(1999) Nanoscale adhesion, friction and wear studies on
16.99 M.S. Hayden, L.K. Gilliland, J.A. Ledbetter: Anti- silicon based surfaces, Acta Biomater. 2, 39–49
body engineering, Curr. Opin. Immunol. 9, 201–212 (2006)
(1997) 16.107 S.C. Lee, M. Reugsegger, P.D. Barnes, B.R. Smith,
16.100 B. Bhushan, K.J. Kwak, S. Gupta, S.C. Lee: M. Ferrari: Therapeutic nanodevices. In: Springer
Part B 16
Nanoscale adhesion, friction and wear studies of Handbook of Nanotechnology, ed. by B. Bhushan
biomolecules on polymer-coated silica and alu- (Springer, Berlin, Heidelberg 2007) pp. 461–504
485
G-Protein Co
17. G-Protein Coupled Receptors: Progress in Surface
Display and Biosensor Technology
Wayne R. Leifert, Tamara H. Cooper, Kelly Bailey
Part B 17
should greatly extend the range of applications to applied.
The Superfamily of GPCRs recognition takes place, the GPCR in turn triggers
G-protein coupled receptors (GPCRs) represent a su- the activation of a heterotrimeric G-protein complex
perfamily of intramembrane proteins (polypeptides) located on the peripheral intracellular (cytoplasmic) sur-
which initiate many signal transduction pathways in face of the cell membrane (the term G-protein is used
virtually all eukaryotic cells. GPCRs are structurally since these proteins bind guanine nucleotides such as
characterized by their seven transmembrane (serpen- guanosine di- and triphosphate present in cells, as dis-
tine) spanning domains (Fig. 17.1). GPCR activation cussed in detail later). Finally, the signal transduction
can be initiated by a wide variety of extracellular cascade involves the activated G-protein altering the ac-
stimuli such as light, odorants, neurotransmitters, and tivity of some downstream effector protein(s), which
hormones. In most cases the GPCR uses a transmem- can be enzymes or ion channels located in the cell
brane signaling system which involves three separate membrane. This then leads to a change in the cellu-
components (systems). Firstly, the extracellular ligand lar concentration of cyclic adenosine monophosphate
is specifically detected by a cell-surface GPCR. Once (cAMP), calcium ions or a metabolite such as phospho-
486 Part B MEMS/NEMS and BioMEMS/NEMS
inositides within the cell, resulting in a physiological sent a significant target for medicinal pharmaceuticals.
response such as stronger and faster contraction of the Furthermore, more than 50% of all drugs currently mar-
heart. keted worldwide are directed against GPCRs [17.1],
Since many disease processes involve aberrant with this likely to increase with the recent elucidation
or altered GPCR signaling dynamics, GPCRs repre- of high-resolution structural data of the β2 -adrenergic
Table 17.1 Some examples of prescription drugs that target GPCRs for the indicated disease state
Table 17.2 A partial list of some of the known endogenous and exogenous GPCR ligands
receptor [17.2]. Table 17.1 lists some commonly pre- is this vitally important function of these cell-surface
scribed drugs acting on GPCRs. GPCRs are associated receptors, i. e., transduction of exogenous signals into
with almost every major therapeutic category or disease an intracellular response, which makes GPCRs so phys-
class, including pain, asthma, inflammation, obesity, iologically significant. Indeed, there are reported to be
cancer, as well as cardiovascular, metabolic, gastroin- ≈ 747 different human GPCRs as predicted from gene
Part B 17
testinal, and central nervous system diseases [17.3]. It sequencing analyses, 380 of which are thought to be
chemosensory receptors, whereas the remaining 367
Fig. 17.2 Future applications of GPCR platforms. The GPCRs are predicted to bind endogenous ligands such
development of functional assay platforms for integral as neurotransmitters, hormones, fatty acids, and pep-
membrane proteins, particularly those of the GPCR class, tides [17.4]. With about 230 of these GPCRs having
which are compatible with future high-throughput microar- been identified already (i. e., they have known ligands),
ray formats will offer significant opportunities in a number this currently leaves about 140 orphan GPCRs with as
of areas. It is expected that such advances in assay technol- yet undiscovered ligands. A summary of some of the
ogy will likely impact on drug discovery, diagnostics, and known GPCR ligands is presented in Table 17.2.
biosensors. There is a strong requirement for technologies
that enable screening of multiple GPCR targets simultane-
Combinatorial chemistry
ously (multiplexing). Therefore, it would be advantageous Drug discovery
Bioprospecting
in the future to design new biosensor platforms using
Cell-free
miniaturized nanotechnology approaches. Furthermore, to functional Point-of-care
Diagnostics
achieve this aim successfully, it will be an absolute require- GPCR Critical care
ment for cross-disciplinary fields of research (including assay platform
Odorant panel
biology, physics, and chemistry, as well as mathematics Biosensors
Flavor chemistry
for molecular modeling and bioinformatics) to be highly “Nano-nose”
integrated
488 Part B MEMS/NEMS and BioMEMS/NEMS
This chapter focuses firstly on possible cell-free GPCRs and G-proteins in relatively high amounts and
approaches which could be used in biosensor appli- in purified form using recombinant DNA techniques.
cations, diagnostic platforms, and for high-throughput Also, it is becoming increasingly more routine to
screening (HTS) of GPCR ligands, with particular em- produce recombinant modifications of such proteins
phasis on GPCR signaling complexes and associated using basic molecular biological approaches. These
enabling nanotechnologies (Fig. 17.2). Additionally, we modifications can include biotin tags, hexahistidine
include molecular biology approaches involving G- tags or fluorescent protein fusions which can allow
proteins and GPCRs with reference to biosensor and site-specific interaction of the recombinant protein(s)
HTS applications. One of the most important break- with appropriately derivatized biosensor surfaces such
throughs permitting these developments for GPCR and as glass or gold or the generation of a biosensor
G-protein signaling is the ability to produce these signal.
ceptor), the G-proteins Gα and Gβγ have high affinity The activated state of the Gα subunit lasts until the
for each other and remain tightly bound, forming the GTP is hydrolyzed to GDP by the intrinsic GTPase
heterotrimeric G-protein complex. In this state, guano-
sine diphosphate (GDP) is tightly bound to the Gα Fig. 17.3 Molecular switching: the regulatory cycle of
subunit associated with the Gβγ dimer. Both Gα and agonist-induced (receptor-activated) heterotrimeric G-
Gβγ subunits can bind to the GPCR. When the agonist proteins. The binding of the agonist to the unoccupied
receptor (R) causes a change in conformation, thus activat-
ing the receptor (R*), which promotes the release of GDP
GTP from the heterotrimeric G-protein complex and rapid ex-
Agonist
change with GTP into the nucleotide binding site on the Gα
α βγ subunit. In its GTP-bound state, the G-protein heterotrimer
R R* GDP +
dissociates into the Gα and Gβγ subunits, exposing new
surfaces and allowing interaction with specific downstream
GTP effectors (E). The signal is terminated by hydrolysis of
E
βγ GTP to GDP (and Pi ) by the intrinsic GTPase activity of
α the Gα subunit followed by return of the system to the basal
Pi E*
unstimulated state. Asterisk indicates activated state of re-
ceptor (R) or effector (E); Pi , inorganic phosphate; GDP,
GDP guanosine diphosphate; GTP, guanosine triphosphate
G-Protein Coupled Receptors: Progress in Surface Display and Biosensor Technology 17.2 Preparation of GPCRs and G-Proteins 489
activity of the Gα subunit. The various families of unit to another. Following the hydrolysis of GTP to
Gα subunits, i. e., Gαs , Gαi/o , Gαq/11 , and Gα12/13 , GDP on the Gα subunit, the Gα and Gβγ subunits
are all GTPases, although the intrinsic rate of GTP re-associate and are able to return to the receptor-
hydrolysis varies greatly from one type of Gα sub- associated state.
Part B 17.2
• Performs many of the post- yield
translational modifications made
to human proteins
difficult to purify, primarily due to the lipophilic (hy- DNA-based technologies. Therefore, the first step in
drophobic) nature of these receptors and the fact that generating a GPCR biosensor technology is success-
they are usually irreversibly denatured (inactivated) fully obtaining functional proteins that may also have
when they are removed from their native lipid envi- been engineered to provide new properties that enable
ronment using detergent treatment. However, partial surface attachment or a fluorescent signal.
purification of GPCRs is usually carried out in or-
der to obtain a supply of them. This results in small 17.2.1 Expression Systems
(nanometer-scale) crude membrane fragments being for Recombinant GPCRs/G-Proteins
produced. The GPCR membrane fragments will usu-
ally contain hydrophobic membrane lipids (which are A prerequisite for molecular approaches to the design
required for functionality) as well as other native, con- of cell-free GPCR assays is an expression system which
taminating proteins and can then be manipulated and produces recombinant proteins with the required activ-
immobilized by various means (discussed later) onto ity and level of expression. Expression systems utilizing
appropriate surfaces for use as biosensors. On the other either bacteria, yeast, mammalian or insect cells are de-
hand, the G-proteins, which are classified as peripheral tailed in Table 17.3. These systems are generally well
as opposed to integral membrane proteins and do not characterized and show the greatest promise in terms of
require an absolute lipid environment for activity, can their ability to produce large amounts of functional pro-
be routinely purified in relatively large amounts (mil- teins which can be utilized in GPCR biosensing assay
ligram quantities) when expressed using recombinant formats.
lanthanide binding tags [17.6] and tetracysteine mo- hematopoietic cells [17.12] and was shown to couple
tifs (TCM) [17.7]. Each of these tags are significantly to a wider range of receptors than other known alpha
smaller than GFP variants and enable fluorescence subunits, and to transduce ligand-mediated signaling
through the binding of a lanthanide such as terbium through phospholipase C (PLC), resulting in the mod-
or a fluorescent arsenic derivative such as FlAsH, re- ification of intracellular calcium concentrations [17.13–
spectively. In some cases the use of a smaller tag can 17]. Molecular biology approaches have also been uti-
prevent the loss of function of the proteins of inter- lized to increase the promiscuity of various Gα subunits
est. For example, when a TCM was used to label the by altering the sequence of amino acids within the
adenosine A2A receptor, aspects of the receptors func- protein [17.11, 18–21]. Although cell-free applications
tion that had been lost when using yellow fluorescent have not been routinely used to date, it is expected that,
protein (YFP) at the same site were restored [17.8]. within the near future, promiscuous G-proteins will be
In the future it is likely that fluorescence-based as- used in a similar manner to in whole-cell applications.
say development involving compounds such as these
will increase in efficiency and flexibility, allowing 17.3.3 Protein Engineering
such methods to be at the forefront of technologies for Surface Attachment
for determining molecular interactions using cell-free
systems. Ideally, proteins should be uniformly immobilized so
that the protein remains functional and is orientated
17.3.2 Engineering such that the required interaction can occur. To achieve
of Promiscuous Gα Proteins these ends, protein engineering can be a powerful
tool. Immobilized metal ion affinity chromatography
A major impediment to the production of homoge- (IMAC) used for protein purification has been extended
neous, cell-free, GPCR-based screening systems is the to enable functional immobilization of proteins onto
coupling between a given GPCR and a subset of Gα a surface. GPCRs and G-proteins are often fused to
subunits. For example, muscarinic receptor subtypes a 6 histidine tag [17.22–24] that has a high affinity for
M1 , M3 , and M5 typically couple to Gαq/11 , whilst nitriloacetic acid (NTA) loaded with a divalent cation,
M2 and M4 subtypes couple to Gi or Go [17.9]. Bi- often Ni2+ , on a surface. The length of the histidine
ologically, this discrimination is the basis for correct tag can also be adjusted for higher affinity to Ni2+ ,
cellular signaling but needs to be modified from the as has been demonstrated using Gαi1 [17.25]. Utilizing
in vivo situation to allow production of a generic GPCR histidine tags, it has also been possible to functionally
biosensing system. In this regard, recent attempts have reconstitute GPCRs with G-proteins on Ni-NTA beads
been made to produce promiscuous Gα subunits capa- and observe signaling upon ligand binding [17.24].
ble of transducing signals resulting from extracellular Surface immobilization has also been achieved by en-
Part B 17.4
interactions involving any GPCR [17.10, 11]. Many gineering short peptides such as the C9 peptide or Myc
of the promiscuous subunits constructed thus far are tags onto GPCRs of interest and using surfaces display-
based on variants of the human Gα16 (a member of ing the appropriate antibody to these tags to capture the
the Gαq subfamily). This protein was first isolated from receptors [17.23, 26].
17.4.1 Level 1 Biosensing – Ligand Binding magnitude as well as having good polarizing proper-
ties [17.29].
We have defined level 1 biosensing as ligand binding to
the receptor. This includes such techniques as radioli- Two-Photon Excitation Fluorescence
gand binding (not discussed here) and fluorescent (and Cross-Correlation Spectroscopy
fluorescent polarization) ligand binding assays. Ligand Two-photon excitation fluorescence cross-correlation
binding can also be detected by techniques such as spectroscopy (TPE-FCCS) is used to measure dynamic
flow cytometry, two-photon excitation cross-correlation interactions between molecules, and in the bioscience
spectroscopy (TPE-FCCS), surface plasmon resonance field has applications in monitoring DNA, protein, and
(SPR), plasmon waveguide spectroscopy, and piezo- ligand interactions. The technique allows for small mea-
electric crystal sensing. surement volumes and low sample concentrations and
This level of biosensing does not discriminate be- has increased detection specificity over classical fluo-
tween compounds which can be pharmacologically rescence techniques for monitoring molecular dynamics
defined as agonists, antagonists, partial agonists or in- in solution. TPE-FCS is an extension of fluorescence
verse agonists. Therefore its use in biosensing of the correlation spectroscopy (FCS), which analyzes minute,
activation of a signaling pathway is somewhat limited. spontaneous signal fluctuations arising from molecular
However it is still useful for some specific purposes diffusion. The term “two-photon” refers to the use of
such as screening for compounds which interact with different fluorescent molecules with distinct emission
a particular GPCR. properties, each of which can be excited by two photons
of half the energy required for a transition to the excited
Fluorescence Polarization state. Two-photon excitation spectra of many common
Polarization is a general property of most fluorescent fluorophores differ considerably from their one-photon
molecules. Polarization-based experiments are less dye spectra without a change in emission. This makes it
dependent and less susceptible to environmental inter- possible to simultaneously excite two spectrally distinct
ferences (such as pH changes) than assays based on dyes with a single infrared light source. A cross-
fluorescence intensity measurements. Fluorescence in- correlation of the two fluorophores is only generated
tensity variations due to the presence of samples which when the two detection channels measure synchronous
may be colored (e.g., in drug screening of compound li- fluorescence fluctuations, which suggests that the dif-
braries) tend to produce relatively minor interferences. ferent colored species must be spectrally linked.
The degree of polarization (or anisotropy) can be de- This technique is useful in the study of association
termined from measurements of fluorescence intensities and dissociation reactions such as that of a receptor–
parallel and perpendicular to the plane of linearly polar- ligand pair. To obtain accurate kinetic information
regarding the interaction of the human μ-opioid recep-
Part B 17.4
through a laser beam where both scattered and emit- surface-immobilized without further purification. The
ted light are measured. A benefit of this method is that selective, high-affinity interaction between biotin and
simultaneous measurements can be performed on indi- streptavidin allowed template-directed and uniform ori-
vidual particles. entation of the neurokinin-1 receptor on the support
Waller et al. [17.31] have conjugated dextran beads matrix. Additionally, the highly selective TIRF fluores-
with the cognate ligand dihydroalprenolol, which al- cence detection methodology was able to resolve the
lowed for capture of solubilized β2 -adrenergic receptors binding of fluorescently tagged agonist to as little as
(β-AR) onto this immobilized surface ligand. To mea- 1 aM of receptor molecules.
sure the specific binding of the receptor to the bead
in this flow-cytometry-based assay system, the recep- Microspotting of GPCRs on Glass
tor was expressed as a fusion protein with GFP. It The intrinsic difficulties in producing, purifying, and
was then possible to screen for ligands (either ago- manipulating membrane proteins have delayed their in-
nists or antagonists) to β-AR using a competition assay. troduction into microarray platforms. Hence there are
Another successful bead-based approach used param- no reports to date describing purified membrane pro-
agnetic beads [17.32]. In that study the authors built up tein (GPCR) microarrays and their use in functional
a surface containing the captured CCR5 receptor from screening or biosensor applications. However, as a first
a cell lysate held within a lipid bilayer. In this instance step towards such display technologies, researchers
the CCR5 receptor was not able to freely move laterally at Corning Inc. (Rochester, USA) have recently de-
in the bilayer as it was tethered via an antibody (directed scribed the fabrication of GPCR membrane arrays for
at the CCR5 receptor) conjugated to the paramagnetic the screening of GPCR ligands [17.34–37]. The array-
beads (paramagnetic proteoliposome). ing of membrane GPCRs required appropriate surface
chemistry for the immobilization of the lipid phase con-
Total Internal Reflection Fluorescence taining the GPCR of interest (Fig. 17.5). They reported
Total internal reflection fluorescence (TIRF) takes ad- that surface modification with γ -aminopropylsilane (an
vantage of refractive index differences at a solid–liquid amine-presenting surface) provided the best combi-
interface, with the solid surface being either glass or nation of properties to allow surface capture of the
plastic, e.g., cell culture containers. At a critical an- GPCR:G-protein complex from crude membrane prepa-
gle, when total internal reflection occurs, an evanescent rations, resulting in microspots of ≈ 100 μm diameter.
wave is produced in the liquid medium. This elec- Atomic force microscopy (AFM) demonstrated that the
tromagnetic field decays exponentially with increasing height of the supported lipid bilayer was ≈ 5 nm, cor-
distance from the surface. The range of this field lim- responding to GPCRs confined in a single, supported
its background fluorescence, as only fluorophores in lipid layer scaffold. Using these chemically derivatized
Part B 17.4
close proximity to the surface are excited. As such, surfaces, it was possible to demonstrate capture of the
the technique is used to examine interactions between β1 , β2 , and α2A subtypes of the adrenergic receptor,
the molecule of interest and the surface, for example, as well as neurotensin-1 receptors and D1-dopamine
receptors binding to a surface. receptors. This was achieved by using ligands with
Martinez et al. [17.33] used TIRF to demonstrate fluorescent labels covalently attached and detecting
ligand binding to the neurokinin-1 GPCR by sur- fluorescence binding to the GPCRs with a fluorescence-
face immobilization of membrane fragments containing based microarray scanner. Dose–response curves using
this receptor protein. In this study, the GPCR ex- the fluorescently labeled ligands gave 50% inhibi-
pressed as a biotinylated protein using mammalian tion concentration (IC50 ) values in the nanomolar
cells was selectively immobilized on a quartz sensor range, suggesting that the GPCR:G-protein complex
surface coated with streptavidin (streptavidin binds bi- was largely preserved and biologically intact in the mi-
otin with extremely high affinity). TIRF measurements crospot. There was no change in the performance of
were made using a fluorescence-labeled agonist (i. e., the arrays over a 60 day period, indicating good long-
the cognate agonist substance-P labeled with fluores- term stability. Although the use of glass slides for the
cein). Using this approach, it was not necessary to printing of the GPCR arrays was promoted by this
detergent-solubilize and reconstitute the neurokinin-1 research group, in some instances gold surfaces were re-
receptors, thus avoiding the deleterious effect(s) asso- quired due to nonspecific binding of fluorescent ligands.
ciated with such processes. This receptor, in the form A current limitation of this technology is the inability
of a mammalian cell membrane homogenate, was then to carry out a functional(i. e., signaling) assay, which
494 Part B MEMS/NEMS and BioMEMS/NEMS
Fluorescent-
L F* labeled ligand
Incubate wash
Fluorescent-labeled ligand
binds specifically to
GPCR immobilized on surface
L F*
Biological membrane Biological membrane
containing expressed containing expressed
GPCR GPCR
Fig. 17.5 Idealized schematic of an immobilized GPCR with associated G-proteins. The fabricated surface array is
printed on a γ -aminopropylsilane (GAPS)-presenting surface. The height of the supported lipid bilayer is ≈ 5 nm.
Fluorescently labeled (LF *) ligands (such as 4,4,-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene (BODIPY)-
tetramethylrhodamine (TMR)) will bind specifically to the GPCR (for example, a neurotensin receptor) at nanomolar
concentrations. The fluorescence is measured following an incubation/washing step to remove unbound fluorescent
ligands. When compounds of unknown activity are added to the incubation step, as in drug screening programs,
fluorescent-labeled ligand binding is blocked by agents that bind to the GPCR (for example, GPCR antagonists) (adapted
from [17.13])
would allow test compounds to be classified as agonists the reflected light. The instrument detects the change in
or antagonists. Furthermore, although there are increas- angle of the reflected light minimum. The technique can
ing numbers of commercially available fluorescently be used to study interactions between ligands, GPCRs,
labeled ligands, the need to always structurally modify and G-proteins. SPR experiments do not require a large
the ligand to accommodate some reporter moiety may amount of sample, and detection does not require fluo-
limit the implementation of the technology. Neverthe- rescent or radioisotopic labeling. A variety of available
less, the above-mentioned GPCR microarrays may have surface chemistries allows for immobilization of many
Part B 17.4
potential as functional GPCR assays when complexed types of proteins using a range of strategies.
with G-proteins and integrated with appropriate signal Ligand binding to a GPCR attached to a surface has
generation and detection methods. been reported for the chemokine CCR5 receptor using
SPR methodology [17.38]. For such display, purifica-
Surface Plasmon Resonance tion of the GPCR has not always been necessary, and
One of the most versatile techniques for measuring crude membrane preparations have either been fused
biospecific interactions in real time are biosensors based with an alkylthiol monolayer (≈ 3 nm thickness) formed
on the optical phenomenon of surface plasmon res- on a gold-coated glass surface or onto a carboxymethyl-
onance (SPR, Fig. 17.6). Surface plasmon resonance modified dextran sensor surface [17.39]. One problem
occurs when light interacts with a conducting surface of surface-based assays is orientation of the receptor
(plasmon interaction) which is positioned between two once attached to the surface. One means of overcom-
materials of different refractive index. At a specific an- ing this problem was to specifically select only those
gle the intensity of the reflected light decreases, with proteoliposomes (≈ 300 nm-diameter vesicles) in which
this angle being dependent on (among other things) the the carboxy terminus of the receptor was orientated to
refractive index of the material on the opposite side to the outside of the vesicle. This was performed using
which the light is applied. Molecules associating with conformationally dependent antibodies [17.26]. In this
or disassociating from this material (e.g., receptor to biosensor application, SPR has a distinct advantage as
surface or ligand to receptor) will change the refractive a screening tool since it can detect the cognate ligand
index of the material and can be detected by measuring without requiring fluorescent or radio labeling. This al-
G-Protein Coupled Receptors: Progress in Surface Display and Biosensor Technology 17.4 GPCR Biosensing 495
lows SPR to be used in complex fluids of natural origin tra (with both s- and p-polarized light) with increasing
and simplifies, and potentially speeds up, the develop- ligand concentration, with the results from PWR being
ment of assay technologies. compared with those obtained by traditional radioligand
binding assays. Differences in s- and p-polarized light
Plasmon-Waveguide Resonance Spectroscopy measurements demonstrated changes in receptor struc-
Plasmon-waveguide resonance (PWR) spectroscopy ture which varied depending on whether the ligand was
measures real-time binding of free molecules to immo- a full, partial or inverse agonist. Previous work using
bilized molecules such as GPCRs without the appli- PWR technology has been reported for the detection
cation of specific labels (reviewed elsewhere [17.40]). of conformational changes in a proteolipid membrane
PWR has several significant advantages compared with containing the human δ-opioid receptor following bind-
conventional surface plasmon resonance, including en- ing of nonpeptide agonists, partial agonists, antagonists,
hanced sensitivity and spectral resolution, as well as and inverse agonists [17.42]. Although the ligands in
the ability to distinguish between mass and conforma- the above study were of similar molecular weight, there
tional changes. This latter property is a consequence of were distinctly different refractive index changes in-
the use of both p- and s-polarized excitation to produce duced by ligand binding and these were too large to be
resonances. This allows for measurement of refractive accounted for by differences in mass alone. The infer-
index anisotropy, which reflects changes in mass distri-
bution and, therefore, changes in molecular orientation Fig. 17.6 Surface plasmon resonance (SPR) provides mass
and conformation. detection. Most importantly, this technique does not re-
In a recent study, ligand binding to the β2 - quire labeling of the interacting components. Since it is the
adrenergic receptor has been demonstrated using evanescent field wave and not the incident light which pen-
PWR [17.41]. Using this technique, changes in the etrates the sample, measurements can be made on turbid or
refractive index upon ligand binding to surface- even opaque samples. The detection principle of SPR relies
immobilized receptor results in a shift in the PWR spec- on electron charge density wave phenomena arising at the
tra. The authors used ligands with similar molecular surface of a metallic film when light is reflected at the film
weight in order to study structural changes in the recep- under specific conditions (surface plasmon resonance). The
tor caused by agonist, inverse agonist, and partial ago- resonance is a result of energy and momentum being trans-
nist binding. The technique was used to produce binding formed from incident photons into surface plasmons, and
curves for five ligands using shifts in the PWR spec- is sensitive to the refractive index of the medium on the
opposite side of the film from the reflected light. Quantita-
tive measurements of the binding interaction between one
Incident Reflected
light light or more molecules are dependent on the immobilization
Part B 17.4
I SPR angle of a target molecule onto the sensor chip surface. Bind-
II ing partners to the target can be captured from a complex
mixture as they pass over the chip. Interactions between
proteins, nucleic acids, lipids, carbohydrates, and even
Sensor chip
with gold film whole cells can be studied. The sensor chip consists of
and surface coating a glass surface coated with a thin layer of gold. This forms
the basis for a range of specialized surfaces designed to
optimize the binding of a variety of molecules. The gold
Flow cell layer in the sensor chip creates the physical conditions re-
quired for SPR. The upper figure shows a detector with
sensor chip. When molecules in the test solution bind to
Resonance a target molecule the mass increases; when they dissoci-
II signal
ate the mass falls. This simple principle forms the basis of
the sensorgram for continuous, real-time monitoring of the
association and dissociation of the interacting molecules
I (lower figure). The sensorgram provides quantitative in-
Time formation in real time on the specificity of binding, active
concentration of molecules in a sample, kinetics, and affin-
Sensorgram
ity. Molecules as small as 100 Da can be studied
496 Part B MEMS/NEMS and BioMEMS/NEMS
ence from this finding was that a ligand-specific con- chapter to be truly functional assays since the transducer
formation change in the receptor protein may have been G-proteins are the first differentiated site of signalling
detected. Therefore this methodology may have use as initiated from the GPCR. This therefore means that
a future biosensor, particularly with regard to GPCRs. the GPCR must be in a functional form, enabling it
to interact and activate a G-protein signaling pathway.
Piezoelectric Crystal Sensing Level 3 biosensing involves the use of nonhydrolyz-
Piezoelectric crystal sensing measures a change in mass able GTP analogs such as radiolabeled 35 Sγ GTP or
on molecule binding to the surface due to a change in fluorescent-tagged europium-GTP which bind to the
resonance frequency of the crystal. The technique has receptor-activated form of the Gα subunit targeting the
been used in an electronic nose with olfactory recep- site of guanine nucleotide exchange (GDP for GTP on
tors which are typically GPCRs [17.43], where an array the Gα subunit of the Gαβγ heterotrimer). The guanine
of six sensor elements could be used to characterize nucleotide exchange process is generally considered
each of six test compounds, emphasizing the potential the first major point of G-protein activation following
for GPCR ligand screening in the sensory area. The use GPCR stimulation (Figs. 17.1 and 17.3).
of an artificial nose (bionose) to mimic the properties of Guanine nucleotide exchange is a very early, generic
the human nose may find wide applications in the near event in the signal transduction process of GPCR acti-
future. vation and is therefore an attractive event to monitor as
it is less subject to regulation by cellular processes fur-
17.4.2 Level 2 Biosensing – ther downstream (we denote this as level 3 biosensing,
Conformational Changes in the GPCR see Fig. 17.1). The radiolabeled 35 Sγ GTP or fluores-
cent europium-GTP binding assays measure the level
Level 2 involves the detection of intrinsic conforma- of G-protein activation following agonist activation of
tional changes in the GPCR protein following agonist a GPCR by determining the binding of these nonhy-
activation and may involve the use of fluorescence- drolyzable analogs of GTP to the Gα subunit. There-
based techniques. Cell-free measurements of conforma- fore, they are defined as functional assays of GPCR ac-
tional changes in the GPCR following ligand (usually tivation. Ligand regulation of the binding of 35 Sγ GTP
agonist or partial agonist) binding have been limited to is one of the most widely used assay methods to mea-
date. sure receptor activation of heterotrimeric G-proteins, as
A good example of a level 2 cell-free assay used discussed elsewhere in detail [17.44, 45]. This method-
β2 -adrenergic receptors immobilized onto glass and ology also provides the basis for measurement of phar-
gold surfaces. In this study, the receptors were site- macological characteristics such as potency, efficacy,
specifically labeled with the fluorophore tetramethyl- and the antagonist affinity of compounds [17.45] in cell-
Part B 17.4
rhodamine-maleimide at cysteine 265 (Cys265) using free assays and artificial expression systems for GPCRs
a series of molecular biology approaches. It was (an example of typical data is shown in Fig. 17.7).
then possible to show agonist (isoproterenol)-induced However, despite the highly desirable attributes of this
conformational changes within the vicinity of the flu- methodology and its widespread use to date, ligand
orescent moiety (tetramethyl-rhodamine) at position regulation of 35 Sγ GTP binding has been largely re-
Cys265 of the recombinant β2 -adrenergic receptors. stricted to those receptors which signal through the
Moreover, the agonist-induced signal was large enough Gαi/o proteins (pertussis-toxin sensitive) and, to a lesser
to detect using a simple intensified charge-coupled de- extent, the Gαs and Gαq families of G-proteins. As
vice (ICCD) camera image. Thus, it was suggested that such, the use of these assay platforms can be prob-
the technique may be useful for drug screening with lematic for high-throughput screening as they are not
GPCR arrays. Indeed this method did not require the homogeneous (i. e., they require a separation step to
formation of lipid bilayers and did not require the use remove bound from free 35 Sγ GTP). Additionally, the
of purified G-proteins or fluorescent ligands to detect use of radioactive-based assays (including ligand bind-
receptor activation. ing assays) has led to safety, handling, waste disposal,
and cost concerns. The newly developed, fluorescence-
17.4.3 Level 3 Biosensing – GTP Binding based europium-GTP assay partly overcomes some of
the above limitations and has already been successfully
Measurements of GPCR activation further downstream used with the following GPCRs: motilin, neurotensin,
from level 2 are considered for the purposes of this muscarinic-M1 , and α2A -adrenergic receptors.
G-Protein Coupled Receptors: Progress in Surface Display and Biosensor Technology 17.4 GPCR Biosensing 497
35
receptor binding site) or inverse agonist [inhibits the in-
SγGTP bound (% of maximum) trinsic (nonagonist-stimulated) activity of the receptor
100
signaling, often observed in overexpressed receptors];
UK-14304 (agonist) however, if a functional GPCR assay is constructed in
80 which G-protein activation is an endpoint, i. e., level 4
biosensing, then it is possible to distinguish between
60 these functionally distinct ligands. For cell-free assays,
EC50
both methodologies (levels 1 and 4) are important in
40 HTS programs, for example, and may have differing
extents of applicability. Indeed, novel nanotechnol-
20 UK-14304 (agonist) + ogy approaches will be required to achieve level 4
rauwolscine (antagonist)
biosensing, including suitable surface derivatization for
0 immobilization of GPCR and G-protein complexes.
–11 –10 –9 –8 –7 –6 –5 –4 –3
[UK-14304] log (M) Level 4 biosensing encompasses those assays which
measure the final stage of activation of the G-protein
Fig. 17.7 Activation of GPCR-induced GTP binding. The heterotrimeric complex, i.e., the putative dissociation or
data show results from an experiment which was conducted rearrangement of the subunits following GPCR-induced
by incubating 20 nM purified G-proteins (Gαi1 and Gβ1 γ2 ) G-protein activation [17.46]. This level of GPCR ac-
reconstituted with 0.4 nM recombinant α2A -adrenergic tivation has currently not been investigated in great
receptor-expressing membranes (these receptors normally detail but may prove to be extremely valuable in future
bind adrenaline with high affinity). The assay also con- functional biosensor applications. Assay methodologies
tained 0.2 nM 35 Sγ GTP (a radioactive nonhydrolyzable which are examples of level 4 biosensing have been
analog of GTP). An adrenaline analog (UK-14304) was reported using surface plasmon resonance and flow
then added to the reconstituted α2A -adrenergic recep- cytometry technologies to demonstrate receptor disso-
tor membrane, at the concentrations indicated on the ciation from the G-protein complex.
x-axis (0.01 nM to 100 μM) in the presence or absence
of the α2A -adrenergic receptor antagonist, rauwolscine SPR G-Protein Dissociation
(10 μM). Following a filtration step to remove the bound Bieri et al. [17.47] used carbohydrate-specific biotiny-
35 Sγ GTP:Gα complex from unbound 35 Sγ GTP, the fil-
i1 lation chemistry to achieve appropriate orientation and
ters were subsequently counted in a scintillation counter functional immobilization of the solubilized bovine
to measure the level of radioactivity. As the concen- rhodopsin receptor, with high-contrast micropatterns
tration of the agonist (UK-14304) was increased above of the receptor being used to spatially separate pro-
1 nM (10−9 M), the characteristic sigmoidal dose–response
Part B 17.4
tein regions. This reconstituted GPCR:G-protein system
effect was seen. This result shows an increase in receptor- provided relatively stable results (over hours) with
activated binding of 35 Sγ GTP to the Gαi1 subunits as the added advantage of obtaining repeated activa-
the UK-14304 is increased in concentration, indicating tion/deactivation cycles of the GPCR:G-protein system.
functional signaling of the receptor through the G-protein Measurements were made using SPR detection of G-
complex. The concentration at which 50% (also the point protein dissociation from the receptor surface following
of inflexion) of the signaling response (effective concentra- the positioning of the biotinylated form of the rhodopsin
tion) was observed (EC50 ) was ≈ 12 nM. In the presence of receptor onto a self-assembled monolayer containing
excess α2A -adrenergic receptor antagonist (rauwolscine), streptavidin. Using this approach, G-protein activation
the signal was completely blocked at the receptor. There- could be directly monitored, giving a functional out-
fore, this type of biosensing application demonstrates put, as opposed to ligand–receptor binding interactions,
sensitivity as well as specificity which yield little information on the receptor-activated
pathway when screening agonists and antagonists. Al-
17.4.4 Level 4 Biosensing – though SPR is useful for the study of G-protein
GPCR:G-Protein Dissociation interactions, it may not be well suited to detect bind-
ing of small ligand molecules directly due to its reliance
Procedures that utilize only ligand binding (level 1) on changes in mass concentration. An advantage of re-
do not distinguish between agonist (activates recep- peated activation/deactivation cycles of GPCRs is that
tor), antagonist (blocks the action of the agonist at the different compounds can be tested serially with the
498 Part B MEMS/NEMS and BioMEMS/NEMS
same receptor preparation. The above approach appears possible to screen ligands for a known solubilized
promising for future applications of chip-based tech- GPCR, or alternatively to test which G-proteins prefer-
nologies in the area of GPCR biosensor applications. entially couple to a particular solubilized, reconstituted
GPCR. The flow cytometry system used above had
Flow Cytometry – GPCR:G-Protein Interactions a sampling rate of ≈ 50–100 samples per minute; how-
Modifying the surface of epoxy-activated dextran beads ever, flow cytometry’s greatest advantage is its ability
by forming a Ni2+ -NTA conjugate was shown to pro- to be multiplexed, where different molecular assemblies
duce beads with a surface capable of binding hexahis- can be made with one sample and yet be discriminated
tidine (his)-tagged β1 γ2 subunits (Fig. 17.8). Tethered by their unique spectral characteristics [17.31, 49, 50].
β1 γ2 subunits were then used to capture Gαs sub- In more detailed studies, the assembly and disassem-
units, which in turn were capable of binding membrane bly of the FPR and his-tagged G-proteins complexed
preparations with expressed β2 -adrenergic receptor on Ni2+ -silica particles provided insight into the acti-
containing a GFP fusion protein (see Sect. 17.3 for vation kinetics of the ternary complex (i. e., receptors
a detailed description of fusion proteins); alternatively, and heterotrimeric G-proteins) [17.49,51]. The study by
a fluorescence-labeled ligand could be detected bind- Simons et al. [17.49] extended the knowledge of ligand–
ing to the tethered β2 -adrenergic receptor, the whole GPCR interactions to involve G-protein–GPCR–ligand
complex being measured using flow cytometry. Addi- interactions assayed in a homogeneous format with
tionally, quantitative solubilization and re-assembly of a bead-based approach amenable to high-throughput
the (hexahistidine-tagged) N-formyl peptide receptor flow cytometry. Indeed, HTS and proteomic appli-
(FPR) has been demonstrated on Ni2+ -silica particles cations could easily be based on such bead arrays
using flow cytometry with dodecyl maltoside as the with potential for color-coded particles and multiplex-
detergent [17.48]. Using such approaches, it may be ing (e.g., by using quantum-dot technology [17.52]).
LF L
GPCR GPCR
GFP
βγ βγ
Part B 17.4
α α
(His)6 (His)6
Ni Ni
References
17.1 A. Wise, K. Gearing, S. Rees: Target validation of 17.8 C. Hoffmann, G. Gaietta, M. Bunemann, S.R. Adams,
G-protein coupled receptors, Drug Discov. Today 7, S. Oberdorff-Maass, B. Behr: A FlAsH-based FRET
235–246 (2002) approach to determine G protein-coupled recep-
17.2 S.G. Rasmussen, H.J. Choi, D.M. Rosenbaum, tor activation in living cells, Nat. Methods 2, 171–176
T.S. Kobilka, F.S. Thian, P.C. Edwards: Crystal (2005)
structure of the human beta(2) adrenergic G- 17.9 M.P. Caulfield, N.J. Birdsall: International union
protein-coupled receptor, Nature 450, 383–387 of pharmacology. XVII. classification of muscarinic
(2007) acetylcholine receptors, Pharmacol. Rev. 50, 279–
Part B 17
17.3 K.L. Pierce, R.T. Premont, R.J. Lefkowitz: Seven- 290 (1998)
transmembrane receptors, Nat. Rev. Mol. Cell Biol. 17.10 A.M. Liu, M.K. Ho, C.S. Wong, J.H. Chan, A.H. Pau,
3, 639–650 (2002) Y.H. Wong: Gα(16/z) chimeras efficiently link a wide
17.4 D.K. Vassilatis, J.G. Hohmann, H. Zeng, F. Li, range of G protein-coupled receptors to calcium
J.E. Ranchalis, M.T. Mortrud: The G protein-coupled mobilization, J. Biomol. Screen. 8, 39–49 (2003)
receptor repertoires of human and mouse, Proc. 17.11 A. Hazari, V. Lowes, J.H. Chan, C.S. Wong, M.K. Ho,
Natl. Acad. Sci. USA 100, 4903–4908 (2003) Y.H. Wong: Replacement of the α5 helix of Gα16
17.5 V.V. Verkhusha, K.A. Lukyanov: The molecular with gαs -specific sequences enhances promiscuity
properties and applications of anthozoa fluores- of Gα16 toward Gs-coupled receptors, Cell Signal.
cent proteins and chromoproteins, Nat. Biotech- 16, 51–62 (2004)
nol. 22, 289–296 (2004) 17.12 T.T. Amatruda III, D.A. Steele, V.Z. Slepak, M.I. Si-
17.6 T.H. Cooper, W.R. Leifert, R.V. Glatz, E.J. McMurchie: mon: Gα16 , a G protein α subunit specifically
Expression and characterisation of functional lan- expressed in hematopoietic cells, Proc. Natl. Acad.
thanide binding tags fused to a Gα-protein and Sci. USA 88, 5587–5591 (1991)
muscarinic (M2) receptor, J. Bionanosci. 2(1), 27–34 17.13 S. Offermanns, M.I. Simon: Gα15 and Gα16 couple
(2008) a wide variety of receptors to phospholipase C,
17.7 S.R. Adams, R.E. Campbell, L.A. Gross, B.R. Martin, J. Biol. Chem. 270, 15175–15180 (1995)
G.K. Walkup, Y. Yao: New biarsenical ligands and 17.14 X. Zhu, L. Birnbaumer: G protein subunits and
tetracysteine motifs for protein labeling in vitro the stimulation of phospholipase C by Gs- and
and in vivo: Synthesis and biological applications, Gi-coupled receptors: Lack of receptor selectivity
J. Am. Chem. Soc. 124, 6063–6076 (2002) of Gα(16) and evidence for a synergic interaction
500 Part B MEMS/NEMS and BioMEMS/NEMS
between Gβγ and the α subunit of a receptor ac- 17.28 C.J. Daly, J.C. McGrath: Fluorescent ligands, anti-
tivated G protein, Proc. Natl. Acad. Sci. USA 93, bodies, and proteins for the study of receptors,
2827–2831 (1996) Pharmacol. Ther. 100, 101–118 (2003)
17.15 G. Milligan, F. Marshall, S. Rees: G16 as a universal 17.29 P. Banks, M. Harvey: Considerations for using
G protein adapter: Implications for agonist screen- fluorescence polarization in the screening of G
ing strategies, Trends Pharmacol. Sci. 17, 235–237 protein-coupled receptors, J. Biomol. Screen. 7,
(1996) 111–117 (2002)
17.16 J.W. Lee, S. Joshi, J.S. Chan, Y.H. Wong: Differ- 17.30 J.L. Swift, M.C. Burger, D. Massotte, T.E. Dahms,
ential coupling of μ-, δ-, and κ-opioid receptors D.T. Cramb: Two-photon excitation fluorescence
to Gα16 -mediated stimulation of phospholipase C, cross-correlation assay for ligand-receptor bind-
J. Neurochem. 70, 2203–2211 (1998) ing: Cell membrane nanopatches containing the
17.17 E. Kostenis: Is Gα16 the optimal tool for fishing human micro-opioid receptor, Anal. Chem. 79,
ligands of orphan G-protein-coupled receptors?, 6783–6791 (2007)
Trends Pharmacol. Sci. 22, 560–564 (2001) 17.31 A. Waller, P. Simons, E.R. Prossnitz, B.S. Edwards,
17.18 B.R. Conklin, Z. Farfel, K.D. Lustig, D. Julius, L.A. Sklar: High throughput screening of G-protein
H.R. Bourne: Substitution of three amino acids coupled receptors via flow cytometry, Comb. Chem.
switches receptor specificity of Gq α to that of Gi α, High Throughput Screen. 6, 389–397 (2003)
Nature 363, 274–276 (1993) 17.32 T. Mirzabekov, H. Kontos, M. Farzan, W. Marasco,
17.19 B.R. Conklin, P. Herzmark, S. Ishida, T.A. Voyno- J. Sodroski: Paramagnetic proteoliposomes con-
Yasenetskaya, Y. Sun, Z. Farfel: Carboxyl-terminal taining a pure, native, and oriented seven-
mutations of Gq α and Gs α that alter the fidelity of transmembrane segment protein, CCR5, Nat.
receptor activation, Mol. Pharmacol. 50, 885–890 Biotechnol. 18, 649–654 (2000)
(1996) 17.33 K.L. Martinez, B.H. Meyer, R. Hovius, K. Lundstrom,
17.20 E. Kostenis, F.Y. Zeng, J. Wess: Functional char- H. Vogel: Ligand binding to G protein-coupled re-
acterization of a series of mutant G protein αq ceptors in tethered cell membranes, Langmuir 19,
subunits displaying promiscuous receptor cou- 10925–10929 (2003)
pling properties, J. Biol. Chem. 273, 17886–17892 17.34 Y. Fang, A.G. Frutos, J. Lahiri: Membrane pro-
(1998) tein microarrays, J. Am. Chem. Soc. 124, 2394–2395
17.21 S.M. Mody, M.K. Ho, S.A. Joshi, Y.H. Wong: Incor- (2002)
poration of Gα(z)-specific sequence at the carboxyl 17.35 Y. Fang, A.G. Frutos, B. Webb, Y. Hong, A. Ferrie,
terminus increases the promiscuity of Gα(16) to- F. Lai: Membrane biochips, BioTechniques Supple-
ward G(i)-coupled receptors, Mol. Pharmacol. 57, ment, 62–65 (2002)
13–23 (2000) 17.36 Y. Fang, A.G. Frutos, J. Lahiri: G-protein-coupled
17.22 C.H. Klaassen, P.H. Bovee-Geurts, G.L. Decaluwe, receptor microarrays, ChemBioChem 3, 987–991
W.J. DeGrip: Large-scale production and purifica- (2002)
tion of functional recombinant bovine rhodopsin 17.37 Y. Fang, J. Peng, A.M. Ferrie, R.S. Burkhalter: Air-
with the use of the baculovirus expression system, stable G protein-coupled receptor microarrays and
Part B 17
Biochem. J. 342, 293–300 (1999) ligand binding characteristics, Anal. Chem. 78,
17.23 D. Ott, Y. Neldner, R. Cebe, I. Dodevski, A. Pluck- 149–155 (2006)
thun: Engineering and functional immobilization 17.38 N.M. Rao, V. Silin, K.D. Ridge, J.T. Woodward,
of opioid receptors, Protein Eng. Design Sel. 18, A.L. Plant: Cell membrane hybrid bilayers con-
153–160 (2005) taining the G-protein-coupled receptor CCR5, Anal.
17.24 R.V. Glatz, W.R. Leifert, T.H. Cooper, K. Bailey, Biochem. 307, 117–130 (2002)
C.S. Barton, A.S. Martin: Molecular engineering of 17.39 O. Karlsson, L. Stefan: Flow-mediated on-surface
G protein-coupled receptors and G proteins for reconsititution of G-protein coupled receptors for
cell-free biosensing, Aust. J. Chem. 60, 309 (2007) applications in surface plasmon resonance biosen-
17.25 C.S. Barton, R.V. Glatz, A.S. Martin, L. Wani- sors, Anal. Biochem. 300, 132–138 (2002)
ganayake, E.J. McMurchie, W.R. Leifert: Interaction 17.40 G. Tollin, Z. Salamon, V.J. Hruby: Techniques:
of self-assembled monolayers incorporating NTA Plasmon-waveguide resonance (PWR) spectroscopy
disulfide with multilength histidine-tagged Gαi1 as a tool to study ligand-GPCR interactions, Trends
subunits, J. Bionanosci. 1, 22–30 (2007) Pharmacol. Sci. 24, 655–659 (2003)
17.26 P. Stenlund, G.J. Babcock, J. Sodroski, D.G. Myszka: 17.41 S. Devanathan, Z. Yao, Z. Salamon, B. Kobilka,
Capture and reconstitution of G protein-coupled G. Tollin: Plasmon-waveguide resonance studies
receptors on a biosensor surface, Anal. Biochem. of ligand binding to the human beta 2-adrenergic
316, 243–250 (2003) receptor, Biochemistry 43, 3280–3288 (2004)
17.27 J.C. Owicki: Fluorescence polarization and anisot- 17.42 I.D. Alves, S.M. Cowell, Z. Salamon, S. Devanathan,
ropy in high throughput screening: Perspectives G. Tollin, V.J. Hruby: Different structural states of
and primer, J. Biomol. Screen. 5, 297–306 (2000) the proteolipid membrane are produced by lig-
G-Protein Coupled Receptors: Progress in Surface Display and Biosensor Technology References 501
Part B 17
503
Microfluidics covers the science of fluidic behaviors reagents are used, which offers quick and low-cost Part B 18
on the micro/nanoscales and the design engineering, analysis.
simulation, and fabrication of fluidic devices for the A fluidic volume of 1 nl can be understood as the
transport, delivery, and handling of fluids on the order volume in a cube surrounded by 100 μm in each di-
of microliters or smaller volumes. It is the backbone of rection. It is much smaller than the size of a grain
biological or biomedical microelectromechanical sys- of table salt. Microfluidic devices and systems han-
tems (bioMEMS) and lab-on-a-chip concept, as most dle sample fluids in this range for various applications,
biological analyses involve fluid transport and reaction. including inkjet printing, blood analysis, biochemical
Biological or chemical reactions on the micro/nanoscale detection, chemical synthesis, drug screening/delivery,
are usually rapid since small amounts of samples and protein analysis, DNA sequencing, and so on.
504 Part B MEMS/NEMS and BioMEMS/NEMS
Microfluidic systems consist of microfluidic plat- basics of microfluidic devices and their applications
forms or devices for fluidic sampling, control, mon- to lab-on-a-chip are briefly reviewed and summarized.
itoring, transport, mixing, reaction, incubation, and Basic materials and fabrication techniques for microflu-
analysis. To construct microfluidic systems, or labs- idic devices will be introduced first and various active
on-a-chip, microfluidic devices must be functionally and passive microfluidic components will be described.
integrated on a microfluidic platform using proper Then, their applications to lab-on-a-chip, or biochemi-
micro/nanofabrication techniques. In this chapter, the cal analysis, will be discussed.
a) b) c)
Part B 18.1
Fig. 18.2a–c Isotropically etched microfluidic channels on glass substrate: (a) poorly etched channel with large under-
etching, (b) poorly etched channel with spikes, and (c) well-etched channel without any defects
506 Part B MEMS/NEMS and BioMEMS/NEMS
a)
Solvent pouring step Solvent curing Cured polymer Final polymer part
b) Pressure Pressure
Mold
Plastic
c) Molten Completely
polymer filled cavity
Molding
block
Mold
cavity Ejected plastic part
Fig. 18.3a–c Concept of polymer micro/nanofabrication techniques: (a) casting, (b) hot embossing, and (c) injection
molding
a) b) Spin coat c)
photoresist
Exposure
Develop
photoresist
Electroplate
Reactive ion
etching
Strip
photoresist
Part B 18.1
Fig. 18.4a–c Mold masters in polymer/plastic fabrication: (a) photolithography-based mold master, (b) silicon-based
mold master, and (c) mold master by electroplating
soft lithography replication, while silicon-based and the replicated elastomer is released from the mold mas-
electroplated mold masters are used for hot-embossing ter, having transferred a reverse structure of the mold
replication. For injection molding, electroplated metal- master. Patterns of a few nm can be achieved using this
lic mold masters are preferable. technique.
Casting or soft lithography [18.13,14] usually offers While casting can be carried out at room tem-
flexible access to microfluidic structures using mostly perature, hot embossing requires a slightly higher
poly(dimethyl siloxane) (PDMS) as a casting material. temperature – up to the glass-transition temperature
A mixture of the elastomer precursor and curing agent of the plastic substrate to be replicated. The hot-
is poured over a master mold structure. After curing, embossing technique has been developed by several
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.2 Active Microfluidic Devices 507
research groups [18.15–17]. A mold master is placed in ing the injection molding process, thermoplastics with
the chamber of a hot-embossing system with the plastic low or medium viscosity are desirable. The filling
substrate, then heated plates press both the plastic sub- of the mold cavity, and subsequently the microstruc-
strate and the mold master, as illustrated in Fig. 18.3b. tures, depend on the viscosity of the plastic melt,
After a certain amount of time (typically 5–20 min, injection speed, molding-block temperature, and noz-
depending on the plastic substrate), the plates are zle temperature of the injection unit. This technique
cooled down to release the replicated plastic substrate. allows very rapid replication and high-volume mass
Hot embossing offers mass production of polymer mi- production. The typical cycle time is several sec-
crostructures, as its cycle time is less than an hour. onds for most applications. However, due to the high
Injection molding is a technique to fabricate shear force on the mold master inside the mold cav-
polymer microstructures at low cost and high vol- ity during injection molding, metallic mold masters
umes [18.18]. A micromachined mold master is placed are highly recommended. Poly(methyl methacrylate)
in the molding block of the injection molding ma- (PMMA), polyethylene (PE), polystyrene (PS), poly-
chine, as illustrated in Fig. 18.3c. The plastic in granular carbonate (PC), and cyclic olefin copolymers (COC) are
form is melted and then injected into the cavity of common polymer/plastic materials for both hot emboss-
a closed mold block, where the mold master is lo- ing and injection molding replication.
cated. The molten plastic continues to flow into and All of these polymer/plastic replication techniques
fill the mold cavity until the plastic cools down to are summarized in Table 18.1. Since each technique
a highly viscous melt, and a cooled plastic part is differs from the others, fabrication techniques and ma-
ejected. In order to ensure good flow properties dur- terials have to be selected according to the application.
Part B 18.2
ment of lab-on-a-chip or micro-total analysis systems the air–liquid interface passes over the valve mech-
(mTAS). A number of different microfluidic devices anism, the operation characteristics of the valve will
have been developed with basic structures analogous differ, due to the change of surface energy over the
to macroscale fluidic devices. Such devices include channel. Similarly, passive check valves [18.20–22]
microfluidic valves, microfluidic pumps, microfluidic are dependent on the pressure of the fluid for oper-
mixers, etc. The devices listed above have been de- ation. Since the active microvalves can be triggered
veloped both as active and passive devices. While on/off depending on an external signal regardless of
passive microfluidic devices are generally easy to fab- the status of the fluid system, there has been consid-
ricate, they do not offer the same functional diversity erable research effort to develop active microfluidic
that the active microfluidic devices provide. For exam- devices. However, active devices are usually more ex-
ple, passive microvalves based on the surface tension pensive due to their desired functional and fabrication
effect [18.19] can operate a few times to hold the li- complexity.
508 Part B MEMS/NEMS and BioMEMS/NEMS
18.2.1 Microvalves
Valve diaphragm Actuator chamber
Active microvalves have been an area of intense re-
Orifice
search over the past decade, and a number of novel
design and actuation schemes have been developed. Fig. 18.5 Schematic concept of pneumatically actuated
This makes the categorization of active microvalves microvalve
a confusing enterprize. Classification schemes for active
microvalves include: Inlet Valve seat Outlet Inlet
Pneumatically actuated microvalves have also been to realize practical electrostatic microvalves. Sato and
demonstrated using polymeric substrate. Hosokawa and Shikida [18.30] have developed a novel membrane
Maeda [18.26] demonstrated a pneumatically actuated design in which the deflection propagates through
three-way microvalve system using a PDMS platform. the membrane, rather than deforming it entirely. Fig-
The microfluidic lines and the pneumatic lines are fab- ure 18.7 shows a schematic sketch of the actuation
ricated on separate layers. mechanism and valve design. The use of this S-shaped
Thermopneumatic actuation is typically performed design allows them to have relatively large gaps across
by heating a fluid (usually a gas) in a confined cavity, the two surfaces, as the electrostatic force need only
as illustrated in Fig. 18.6. The increase in tempera- be concentrated at the edges of the S-shape where the
ture leads to a rise in the pressure of the gas, and membrane is deflected. Robertson and Wise [18.31]
this pressure is used to deflect a membrane for valve have developed an array of electrostatically actuated
operation. Thermopneumatic actuation is an inher- valves using a flap design (rather than a membrane) to
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.2 Active Microfluidic Devices 509
seal the fluid flow. The demonstrated system is suitable a) Si membrane Inductor Bonding wire
for very-low-pressure gas control systems such as those
needed in a clean-room environment. Si
Wijngaart et al. [18.33] have developed a high- Ni/Fe
stroke high-pressure electrostatic actuator for valve
applications. This reference provides a good overview Glass Valve seats Glass
of the theoretical design parameters used to design and
Inlet Outlet
analyze an electrostatically actuated microvalve.
Part B 18.2
Stehr et al. [18.35] demonstrated piezoelectric actuators
for valve applications. A film of piezoelectric material
is deposited on the movable membrane, and upon ap-
plication of an electric potential, a small deformation
occurs in the piezo film that is transmitted to the valve Trace
membrane.
Other Microvalve Actuation Schemes also been widely studied for microfluidic pumping
Microvalves have most commonly been implemented applications.
with one of the actuation schemes listed above. How- In the previous section, various microvalves that
ever, these are not the only actuation schemes that are are used for microfluidic control were discussed. They
used for microvalves. Some others include the use of were presented based on the actuation schemes that they
(shape memory alloys) (SMA) [18.37], electrochem- employ. A similar classification can also be adopted
ical actuation [18.38], etc. SMA actuation schemes for micropumps. However, rather than using the same
offer the advantage of generating very large forces classification scheme, the micropumps are categorized
when the SMA material is heated to its original state. based on the type of microvalve mechanism used as part
Neagu et al. [18.38] presented an electrochemically of the pumping mechanism. Broadly, the mechanical
actuated microvalve. In their device, an electrolysis micropumps can be classified as check-valve-controlled
reaction is used to generate oxygen in a confined microvalves or diffuser pumps. Either mechanism
chamber. This chamber is sealed by a deformable can use various actuation schemes such as electro-
membrane that is deflected due to increased pressure. static, electromagnetic, piezoelectric, etc. Pneumatic
The reported microvalve has a relatively fast actua- control is another actuation scheme for micropump-
tion time and can generate very high pressures. Yoshida ing. A peristaltic micropump based on multilayer soft
et al. [18.39] present a novel approach to the mi- lithography of elastomers is a remarkable example
crovalve design: a micro-electrorheological valve (ER of pneumatically controlled micropumps. Micropumps
valve). An electrorheological fluid is loaded into the driven by direct electrical control form a separate
microchannel, and, depending on the strength of an category and are discussed following the mechanical
applied electric field, the viscosity of the ER fluid micropumps.
changes considerably. A higher viscosity is achieved
when an electric field is applied perpendicular to Micropumps Using a Check-Valve Design
the flow direction. This increased viscosity leads to Figure 18.9 shows a typical mechanical micropump
a drop in the flow rate, allowing the ER fluid to with check valves. This pump consists of an inlet and an
act as a valve. Of course, this technique is limited outlet check valve with a pumping chamber in between.
to fluids that can exhibit such properties; neverthe- A membrane is deflected upwards, and a low-pressure
less, it provides a novel idea to generate on-chip zone is created in the pumping chamber. This forces
microvalves. the inlet check valve open, and fluid is sucked into the
The ideal characteristics for a microvalve are listed pumping chamber. As the membrane returns to its orig-
by Kovacs [18.24]. However, of all the microvalves inal state and continues to travel downwards, a positive
listed above, none can satisfy all the criteria. Thus, pressure builds up, which seals the inlet valve while si-
microvalve design, fabrication, and utility are highly multaneously opening the outlet valve. The fluid is then
application-specific and most microvalves try to gener- ejected and the pump is ready for another cycle.
ate the performance characteristics that are most useful
for the intended application.
a)
18.2.2 Micropumps
Part B 18.2
Part B 18.2
pumps is based on the idea that the geometrical struc- consumption and to increase actuating ranges. An in-
ture used as an inlet valve has a preferential flow teresting micropumping device was recently reported
direction toward the pumping chamber, and the outlet using multilayer monolithic soft lithography and as-
valve structure has a preferential flow direction away sembly [18.52]. Multilayer soft lithography utilizes
from the pumping chamber. An illustrative example of sequence soft lithography cast-molding and bonding
this concept is shown in Fig. 18.10. As can be readily steps to generate a microfluidic system. In realizing
seen, these pumps are designed to work with liquids micropumps, the dead volume and sealing issues have
only. When the pump is in suction mode, the flow been the most cumbersome tasks. The reported mi-
in the inlet diffuser structure is primarily directed to- cropumping device based on multilayer soft lithography
ward the pump, and a slight back-flow occurs from shown successfully addressed the dead volume and
the outlet diffuser (acting as a nozzle) section. When sealing issues as illustrated in Fig. 18.11. There are
the pump is in pressure mode, the outlet diffuser sec- a fluidic layer and a control layer. When air pressure
512 Part B MEMS/NEMS and BioMEMS/NEMS
Part B 18.3
Microvalves are a key component of any microfluidic f) Silicone float valve
Silicone
system and are essential for fluidic sequencing opera- 1.2 mm × 1.2 mm
tions. Since most chemical and biochemical reactions
require about five to six reaction steps, passive mi- Si
crovalves with limited functionality are ideally suited
for such simple tasks. Passive microvalves can be
broadly categorized as follows: Fig. 18.13a–f Various types of passive microvalve designs:
(a) membrane type with a mesa structure; (b) cantilever or
• Silicon/polysilicon or polymer-based check valves flap valve; (c) polysilicon membrane; (d) V-groove type; (e) ti-
• Passive valves based on surface tension effects tanium/polyimide membrane; and (f) silicone float valve (af-
• Hydrogel-based biomimetic valves. ter [18.23],
c IOP Publishing Limited)
514 Part B MEMS/NEMS and BioMEMS/NEMS
Figure 18.13a shows a microvalve fabricated using Passive Valves Based on Surface Tension Effects
silicon bulk-etching techniques. A through-hole (pyra- The passive valves listed in the previous section use
midal cavity) is etched through a silicon wafer that the forced motion of the membrane or flap to control
is sandwiched between two glass wafers. The nor- the flow of fluids. These valves are prone to such prob-
mally closed valve is held in position by the spring lems as clogging and mechanical wear and tear. Passive
effect of the silicon membrane. Upon applying pres- valves based on surface tension effects, on the other
sure to the lower fluidic port, the membrane deflects hand, have no moving parts and control the fluid motion
upwards, allowing fluid flow through the check valve. based on their physical structure and the surface prop-
The same working principle is employed by the mi- erty of the substrate. Figure 18.14 shows a schematic
crovalve shown in Fig. 18.13b. However, instead of sketch of a passive microvalve on a hydrophobic sub-
using a membrane supported on all sides, a cantilever strate [18.19].
structure is used for the flap. This reduces the burst The Hagen–Poiseuille equation for laminar flow
pressure, i. e., the minimum pressure required to open governs the pressure drops in microfluidic systems with
the microvalve. Figure 18.13c shows that the mem- laminar flow. For a rectangular channel with a high
brane structure can also be realized using a polysilicon width-to-height ratio, the pressure drop is governed by
layer deposited on a bulk-etched silicon wafer. Polysil- the equation
icon processes typically allow tighter control over
12LμQ
dimensions and, consequently, offer more repeatable ΔP = , (18.1)
operating characteristics. Figure 18.13d shows the sim- wh 3
plest type of check valve where the V-groove etched where L is the length of the microchannel, μ is the dy-
in a bulk silicon substrate acts as a check valve. How- namic viscosity of the fluid, Q is the flow rate, and w
ever, the low contact area between the flaps of the and h are the width and height of the microchannel, re-
microvalve leads to nontrivial leakage rates in the spectively. Varying L or Q can control the pressure drop
forward direction. Figure 18.13e and f show check- for a given set of w and h.
valve designs that are realized using polymer/metal An abrupt change in the width of the channel
films, in addition to the traditional glass/silicon plat- causes a pressure drop at the point of restriction. For
form. This technique offers a significant advantage in a hydrophobic channel material, an abrupt decrease in
terms of biocompatibility characteristics and control- channel width causes a positive pressure drop
lable operating characteristics. The surface properties
of polymers can be easily tailored using a wide vari- 1 1 1 1
ΔP2 = 2σl cos(θc ) + − + ,
ety of techniques such as plasma treatment and surface w1 h 1 w2 h 2
adsorption [18.59]. Thus, in applications for which the (18.2)
biocompatibility requirements are very stringent, it is
preferable to have polymers as the fluid-contacting ma- where σl is the surface tension of the liquid, θc is the
terial. Furthermore, polymer properties such as stiffness contact angle, and w1 , h 1 and w2 , h 2 are the width
can also be controlled in some cases based on the com- and height of the two sections before and during the re-
position and/or processing conditions. Thus, it may be striction, respectively. Setting h as constant through the
possible to fabricate microvalves with different burst
pressures by using different processing conditions for
Part B 18.3
system, ΔP2 can be varied by adjusting the ratio of w1 low the fluid to leave the narrow channel. Thus, in this
and w2 . Ahn et al. [18.19] have proposed and imple- case, the fluid is held at the transition from the narrow
mented a novel structurally programmable microfluidic capillary to the wide outlet channel.
system (sPROMs) based on the passive microfluidic Another mechanism to implement passive valves
approach. In short, the sPROMs system consists of is the use of hydrophobic patches on a normally hy-
a network of microchannels with passive valves of the drophilic channel. Handique et al. [18.64] demonstrate
type shown in the passive valve section. If the pres- the use of this technique to implement passive valves
sure drop of the passive valves is set to be significantly for a DNA analysis system. The fluid is sucked into the
higher than the pressure drop of the microchannel net- microfluidic channels via capillary suction force. The
work, then the position of the liquid in the microchannel hydrophobic patch exerts a negative capillary pressure
network can be controlled accurately. By applying se- that stops further flow of the fluid. The use of hydropho-
quentially higher pressure pulses, the liquid is forced bic patches as passive valves is reported by Andersson
to move from one passive valve to another. Thus, the et al. [18.65].
movement of the fluid within the microfluidic channels
is programmed using the physical structure of the mi- Other Passive Microvalves
crofluidic system, and this forms the basic idea behind A novel approach to realizing passive valves that are
sPROMs. responsive to their surrounding environment is demon-
The abrupt transition from a wide channel to a nar- strated by Low et al. [18.66] and Yu et al. [18.67]. Yu
row channel can also be affected along the height of the et al. [18.67] have developed customized polymer cock-
microchannel. Furthermore, the passive valve geome- tails that are polymerized in situ around prefabricated
try shown in Fig. 18.14 is not exclusive. Puntambekar posts. The specialized polymer is selectively responsive
et al. [18.62] have demonstrated different geometries to stimuli such as pH, temperature, electric fields, light,
of passive valves, as shown in Fig. 18.15. The graph carbohydrates, and antigens.
shows that the various geometries of the passive valves Another interesting approach in developing passive
in Fig. 18.15 can act as effective passive valves without valves has been adopted by Forster et al. [18.51]. They
having an abrupt transition. This is important in order have developed the so-called no-moving-part (NMP)
to avoid the dead volume that is commonly encountered valves, which are based on a physical configuration that
across an abrupt step junction. allows a higher flow rate along one direction compared
The use of surface tension to control the operation of to the reverse direction.
passive valves is not limited to hydrophobic substrates.
Madou et al. [18.63] demonstrate a capillarity-driven 18.3.2 Passive Micromixers
stop valve on a hydrophilic substrate. On a hydrophilic
substrate, the fluid can easily wick through in the nar- The successful implementation of microfluidic systems
row region. However, at the abrupt transition to a larger for many lab-on-a-chip systems is partly owing to the
channel section, the surface tension effects will not al- significant reduction in volumes handled by such sys-
Part B 18.3
6 Valve 2 geometry
Valve 3
Valve 4 Valve 2
4 30° taper
geometry
Valve 3
2 45° taper
without
restrictor
0 channel
Valve 4
Round Fig. 18.15 Analysis of dif-
geometry ferent geometries of passive
–2
0.02 0.03 0.04 0.05 Time (s) valves (after [18.62],
c The
Royal Society of Chemistry)
516 Part B MEMS/NEMS and BioMEMS/NEMS
tems. This reduction in volume is made possible by mixing performance at low flow rates, in the range of
the use of microfabricated features and channel di- a few μl/min.
mensions ranging from a few to several hundred μm. Mitchell et al. [18.69] have demonstrated three-
Despite the advantage offered by the μm-sized chan- dimensional micromixers that can achieve better perfor-
nels, one of the significant challenges has been the mance by alternately laminating the two fluid streams
implementation of effective microfluidic mixers on the to be mixed. Beebe et al. [18.70] have created a chaotic
microscale. Mixing on the macroscale is a turbulent- mixer that has the convoluted channel along three di-
flow-regime process. However, on the microscale, mensions. This micromixer works on the principle
because of the low Reynolds numbers as a result of of forced advection resulting from repeated turns in
the small channel dimensions, most flow streams are three dimensions. Furthermore, at each turn, eddies are
laminar in nature, which does not allow for efficient generated because of the difference in flow velocities
mixing. On the other hand, diffusion is an impor- along the inner and outer radii, which enhances the
tant factor in mixing because of the short diffusion mixing.
lengths. Stroock et al. [18.71] demonstrate a passive mi-
There have been numerous attempts to realize mi- cromixer that uses chaotic mixing by superimposing
cromixers using both active and passive techniques. a transverse flow component on the axial flow. Ridges
Active micromixers rely on creating localized turbu- are fabricated at the bottom of microchannels. The flow
lence to enhance the mixing process, whereas passive resistance is lower along the ridges (peak/valley) and
micromixers usually enhance the diffusion process. The higher in the axial direction. This generates a helical
diffusion process can be modeled by the following flow pattern that is superimposed on the laminar flow.
equation The demonstrated mixer shows good mixing perfor-
mance over a wide range of flow velocities.
d2
τ= , (18.3) Hong et al. [18.68] have demonstrated a passive
D micromixer based on the Coanda effect. Their design
where τ is the mixing time, d is the distance, and D is uses the effects of diffusion mixing at low flow veloci-
the diffusion coefficient. Equation (18.3) illustrates the ties; at high flow velocities, a convective component is
diffusion-dominated mixing at the microscale. Because added perpendicular to the flow direction, allowing for
of the small diffusion lengths (d), the mixing times can rapid mixing. This mixer shows excellent mixing per-
be made very short. The simplest category of micromix- formances across a wide range of flow rates because of
ers is illustrated in Fig. 18.16. In these mixers, creating
a convoluted path increases the path length that the two
fluids share, leading to higher diffusion and more com- Fluid 1 Fluid 2
plete mixing. However, these mixers only exhibit good
Coanda
effect
Diffuser
Outlet
Coanda
effect
Part B 18.3
“Wing”
Outlet structure
Mixed fluid
Fig. 18.17 Mixing unit design for the Coanda effect mixer.
Fig. 18.16 Diffusion-enhanced mixers based on a long, convoluted The actual mixer has mixing unit pairs in series (af-
flow path ter [18.68])
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.3 Smart Passive Microfluidic Devices 517
Mostly high Fig. 18.19 Schematic sketch of the microdispenser (after [18.62],
diffusivity
c The Royal Society of Chemistry)
Diluent molecules
Fig. 18.18 Conceptual illustration of the H-sensor (af- The microdispenser works on the principle of grad-
ter [18.72]) uated volume measurement. The fluid fills up the exact
fixed volume of the reservoir, and the second passive
the dual mixing effects. Figure 18.17 shows a schematic valve at the other end of the microdispenser stops fur-
sketch of the mixer structure. ther motion. When the reservoir is filled with fluid,
The mixer works on the principle of superimposing the fluidic actuation is stopped, and, simultaneously,
a parabolic flow profile in a direction perpendicular to pneumatic actuation from the air line causes a split
the flow direction. The parabolic profile creates a Taylor in the fluid column at point A (Fig. 18.19). Thus, the
dispersion pattern across the cross section of the flow accuracy of the reservoir decides the accuracy of the
path. The dispersion is directly proportional to the flow dispensed volume. Since the device is manufactured us-
velocity, and higher flow rates generate more dispersion ing UV-LIGA lithography techniques, highly accurate
mixing. and reproducible volumes can be defined. The pre-
Brody and Yager [18.72] have used the laminar flow cisely measured volume of fluid is expelled to the right
characteristics in a microchannel to develop a diffusion- from the reservoir. The expelled fluid then starts to
based extractor. When two fluid streams, where fluid 1 fill up the measuring channel. When the fluid reaches
is loaded with particles of different diffusivity and point B (Fig. 18.19), the third microvalve holds the fluid
fluid 2 is a diluent, are forced to flow together in a mi- column.
crochannel, they form two laminar streams with little Figure 18.20 shows an actual operation sequence of
mixing. If the length that the two streams are in contact the microdispenser. Figure 18.20f shows that the dis-
is carefully adjusted, only particles with high diffusiv- pensed volume is held by passive valve 3, and at this
ity (usually small molecules) can diffuse across into the stage, the length (and hence volume) can be calculated
diluent stream, as shown in Fig. 18.18. The same idea using the on-chip scale. In experiments only, the region
can be extended to a T-filter. Weigl and Yager [18.73] in the immediate vicinity of the scale was viewed us-
have demonstrated a rapid diffusion immunoassay using ing a stereomicroscope to measure the length of the
the T-filter. fluid column. The microdispenser demonstrated above
is reported to have dispensing variation of less than 1%
Part B 18.3
18.3.3 Passive Microdispensers between multiple dispensing cycles.
The principle of the structurally programmable mi- 18.3.4 Microfluidic Multiplexer Integrated
crofluidic systems (sPROMs) was introduced earlier in with Passive Microdispenser
the passive valve section. One of the key components
of the sPROMs system is the microdispenser, which is Ahn et al. [18.19] have demonstrated the sPROMs
designed to accurately and repeatedly dispense fluidic technology to be an innovative method of controlling
volumes in the micro- to nanoliter range. This would al- liquid movement in a programmed fashion in a mi-
low the dispensing of a controlled amount of the analyte crofluidic network. By integrating this technique with
into the system that could be used for further biochem- the microdispensers, a more functionally useful mi-
ical analysis. Figure 18.19 shows a schematic sketch of crofluidic system can be realized. Figure 18.21a shows
the microdispenser design [18.62]. a schematic illustration of the microfluidic multiplexer
518 Part B MEMS/NEMS and BioMEMS/NEMS
a) b) c)
Graduated
reservoir
d) e) f)
Measurement
channel
Fig. 18.20a–f Microphotographs of the microdispenser sequence: (a) fabricated device; (b) fluid at reservoir inlet;
(c) reservoir filling; (d) reservoir filled; (e) split in liquid column due to pneumatic actuation; and (f) fluid ejected to
measurement channel and locked in by passive valve (after [18.62],
c The Royal Society of Chemistry)
with the integrated dispenser, and Fig. 18.21b shows an The microfluidic multiplexer is designed to have
actual device fabricated using rapid prototyping tech- a programmed delivery sequence, as shown in
niques [18.74]. Fig. 18.21a, where the numbers on each branch of the
The operation of the microdispenser has been ex- multiplexer indicates the filling sequence. This sequen-
plained earlier in this section. Briefly, the fluid is loaded tial filling is achieved by using different ratios of passive
in the fixed-volume metering reservoir via a syringe valves along the multiplexer section. For instance, at
pump. The fluid is locked in the reservoir by the passive the first branching point, the passive valve at the up-
valve at the outlet of the reservoir. When a higher pres- per branch offers less resistance than the passive valve
sure is applied via the air inlet line, the liquid column at the beginning of the lower branch. Thus, the dis-
is split and the fluid is dispensed into the graduating pensed fluid will first fill the top branch. After filling
channel. the top branch, the liquid encounters another passive
a) Microdispenser Multiplexer b)
Fluidic Air
Part B 18.3
4
inlet inlet
Air inlet
3 Fluidic
6 inlet
Output reservoirs
1 2 (4 × 25 nl)
5 Dispensing
reservoir
Fixed volume
metering 4
reservoir Graduating Passive 7
(100 nl) channel valves
Fig. 18.21a,b Microfluidic multiplexer with integrated dispenser: (a) schematic sketch and (b) fabricated device filled with dye
(after [18.62])
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.3 Smart Passive Microfluidic Devices 519
Part B 18.3
response times (on a macroscale) are on the order of from the small drop and add to the larger drop. The flow
several hours. However, osmotic transport scales favor- rate can be varied by changing various parameters such
ably to the microscale, and it is expected that these as the volume of the pumping drop, the surface free en-
devices will have response times on the order of sev- ergy of the liquid, or the resistance of the microchannel,
eral minutes, rather than hours. Based on this idea, etc. This pumping scheme is very easy to realize and
Nagakura et al. [18.75] are developing a miniature can be used for a wide variety of fluids.
insulin pump. Su et al. [18.76] have demonstrated a mi-
croscale osmotic actuator that is capable of developing Evaporation-Based Continuous Micropumps
pressures as high as 35 MPa. This is still a relatively Effenhauser et al. [18.78] have demonstrated a conti-
unexplored realm in bioMEMS actuation, and it has nuous-flow micropump based on a controlled evapora-
good potential for applications such as sustained drug tion approach. Their concept is based on the controlled
delivery. evaporation of a liquid through a membrane into a gas
520 Part B MEMS/NEMS and BioMEMS/NEMS
reservoir. The reservoir contains a suitable adsorption • Passive microfluidic systems with no moving parts
agent that draws out the liquid vapors and maintains are inherently more reliable because of the lack of
a low vapor pressure conducive to further evapora- mechanical wear and tear.
tion. If the liquid being pumped is replenished from • They offer very repeatable performance once the
a reservoir, capillary forces will ensure that the fluid is underlying phenomena are well understood and
continuously pumped through the microchannels as it characterized.
evaporates at the other end into the adsorption reservoir. • They are highly suited for bioMEMS applications;
Though the pump suffers from inherent disadvantages they can easily handle a limited number of microflu-
such as strong temperature dependence and operation idic manipulation sequences.
only in suction mode, it offers a very simple technique • Well suited for low-cost mass production.
for fluidic transport. • Their low cost offers the possibility of having
disposable microfluidic systems for specific appli-
18.3.6 Advantages and Disadvantages cations, such as working with blood.
of the Passive Microfluidic Approach • They can offer other interesting possibilities, such
as biomimetic responses.
This chapter has covered a number of different passive
However, like all MEMS devices, passive microflu-
microfluidic devices and systems. Passive microfluidic
idic devices or systems are not the solution to the
devices have only recently been a subject of consider-
microfluidic handling problem. Usually they are very
able research effort. One of the reasons for this interest
application-specific; they cannot be reconfigured for an-
is the long list of advantages that passive microflu-
other application easily. Other disadvantages are listed
idic devices offer. However, since most microfluidic
below:
devices are very application-specific (and even more
so for passive microfluidic devices), the advantages are • They are suited for well-understood, niche applica-
not to be considered universally applicable for all the tions for which the fluidic sequencing steps are well
devices/systems. Some of the advantages that are com- decided.
monly found are: • They are strongly dependent on variations in the
fabrication process.
• Avoiding the need for an active control system. • They are usually not very suitable for a wide range
• They are usually very easy to fabricate. of fluidic mediums.
The idea of lab-on-a-chip is basically to reduce bio- ranges) of reagents and chemicals for an application
logical or chemical laboratories to a microscale system, is very appealing. Another advantage is that lab-on-a-
hand-held size or smaller. Lab-on-a-chip systems can be chip requires very small amounts of reagents/chemicals
made out of silicon, glass, and polymeric materials, and (which enables rapid mixing and reaction) because
the typical microfluidic channel dimensions are in the biochemical reaction is mainly involved in the dif-
range of several tens to hundreds of μm. Liquid samples fusion of two chemical or biological reagents, and
or reagents can be transported through the microchan- microscale fluidics reduces diffusion time as it increases
nels from reservoirs to reactors using electrokinetic, reaction probabilities. In practical terms, reaction prod-
magnetic, or hydrodynamic pumping methods. Fluidic ucts can be produced in a matter of seconds/minutes,
motion or biochemical reactions can also be monitored whereas laboratory-scale reactions can take hours, or
using various sensors, which are often used for bio- even days. In addition, lab-on-a-chip systems minimize
chemical detection of products. harmful by-products since their volume is so small.
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.4 Lab-on-a-Chip for Biochemical Analysis 521
18.4.1 Magnetic Micro/Nano-Bead-Based Fig. 18.24 Analytical concept based on sandwich im-
Biochemical Detection System munoassay and electrochemical detection (after [18.80],
c
The Royal Society of Chemistry)
In the past few years, a large number of microfluidic
prototype devices and systems have been developed, The analytical concept is based on sandwich im-
specifically for biochemical warfare detection systems munoassay and electrochemical detection [18.83], as
and portable diagnostic applications. The bioMEMS illustrated in Fig. 18.24. Magnetic beads are used as
team at the University of Cincinnati has been work- both substrates for the antibodies and carriers for the
ing on the development of a remotely accessible target antigens. A simple concept of magnetic-bead-
generic microfluidic system for biochemical detec- based bio-sampling with an electromagnet for the case
tion and biomedical analysis, based on the concepts of sandwich immunoassay is shown in Fig. 18.25.
of surface-mountable microfluidic motherboards, sand- Antibody-coated beads are introduced on the elec-
wich immunoassays, and electrochemical detection tromagnet and separated by applying magnetic fields.
techniques [18.79, 81]. The limited goal of this work is While holding the antibody-coated beads, antigens are
to develop a generic MEMS-based microfluidic system injected into the channel. Only target antigens are im-
and to apply the fluidic system to detect biomolecules, mobilized and, thus, separated on the magnetic bead
such as specific proteins and/or antigens, in liquid sam- surface due to antibody/antigen reaction. Other antigens
ples. Figure 18.23 illustrates the schematic diagram of get washed out with the flow. Next, enzyme-labeled sec-
a generic microfluidic system for biochemical detection ondary antibodies are introduced and incubated, along
using a magnetic-bead approach for both sampling and with the immobilized antigens. The chamber is then
manipulating the target biomolecules [18.80, 82]. rinsed to remove all unbound secondary antibodies.
Part B 18.4
Biofilter Biofilter and immunosensor Flow sensor
20 mm
new immunoassay. The flow rate was set to 20 μl/min immunoassay, as a typical example of biochemical
in every step. After calibration of the electrochemi- detection and analysis, has been performed on the in-
cal immunosensor, full immunoassays were performed tegrated microfluidic biochemical analysis system that
following the sequence stated above for different anti- includes a surface-mounted biofilter and immunosensor
gen concentrations: 50, 75, 100, 250, and 500 ng/ml. on a glass microfluidic motherboard. Protein-sampling
Concentration of the primary antibody-coated mag- capability has been demonstrated by capturing target
netic beads and conjugated secondary antibody was antigens.
1.02 × 107 beads/ml and 0.7 μg/ml, respectively. Im- The methodology and system can also be ap-
munoassay results for different antigen concentrations plied to generic biomolecule detection and analysis
are shown in Fig. 18.27. systems by replacing the antibody/antigen with appro-
Immunoreactant consumed during one immunoas- priate bioreceptors/reagents, such as DNA fragments
say was 10 μl (20 μl/min × 30 s), and total assay time or oligonucleotides, for application to DNA analysis
was less than 20 min, including all incubation and de- and/or high-throughput protein analysis.
tection steps.
The integrated microfluidic biochemical detection 18.4.2 Disposable Smart Lab-on-a-Chip
system has been successfully developed and fully tested for Blood Analysis
for fast and low-volume immunoassays using magnetic
beads, which are used as both immobilization sur- One of several substrates available for biofluidic chips,
faces and biomolecule carriers. Magnetic-bead-based plastics have recently become one of the most pop-
Current (nA)
0
Part B 18.4
Antigen
–100 concentration
50 ng/ml
75 ng/ml
–150 100 ng/ml
250 ng/ml
500 ng/ml
–200 Fig. 18.27 Immunoassay
Start washing reagent results measured by ampero-
–250 metric time-based detection
0 30 60 90 120 150 method (after [18.80],
c The
Time (s)
Royal Society of Chemistry)
524 Part B MEMS/NEMS and BioMEMS/NEMS
Gel-based O2 O2
solid AgCl– 2H2O
electrolyte AgCl
Ag+
4OH –
Ag O2 + 4e–
Blood
sample Fixed-
injection volume
micro- e– Measuring e–
despenser system
Fig. 18.28 Schematic illustration of smart and disposable posable lab-on-a-chip cartridge has been fabricated
plastic lab-on-a-chip by Ahn et al. (after [18.87],
c 2004 using plastic micro-injection molding and plastic-to-
IEEE) plastic direct-bonding techniques. The biochip cartridge
consists of a fixed-volume microdispenser based on
ular and promising substrates due to their low cost, the structurally programmable microfluidic system
ease of fabrication, and favorable biochemical reli- (sPROMs) technique [18.74], an air-bursting on-chip
ability and compatibility. Plastic substrates, such as pressure source [18.85], and electrochemical biosen-
polyimide, PMMA, PDMS, polyethylene, or polycar- sors [18.86].
bonate, offer a wide range of physical and chemical A passive microfluidic dispenser measures exact
material parameters for the applications of bioflu- amounts of sample to be analyzed, and then the air-
idic chips, generally at low cost using replication bursting on-chip power source is detonated to push the
approaches. The disposable smart plastic biochip is graduated sample fluid from the dispenser reservoir.
composed of integrated modules of plastic fluidic Upon air-bursting, the graduated sample fluid travels
chips for fluid regulation, chemical and biological through the microfluidic channel into sensing reser-
sensors, and electronic controllers. As a demonstra- voirs, under which the biosensor array is located, as
tion vehicle, the biochip has the specific goal of shown in Fig. 18.29.
detecting and identifying three metabolic parameters An array of disposable biosensors consisting of
such as pO2 (partial pressure of oxygen), lactate, and an oxygen sensor, a glucose sensor and a lactate
glucose from a blood sample. The schematic con- sensor has been fabricated using screen-printing tech-
cept of the cartridge-type disposable lab-on-a-chip for nology [18.86]. Measurements from the developed
Part B 18.4
blood analysis is illustrated in Fig. 18.28. The dis- biosensor array can be done based on tiny amounts of
Fig. 18.29 Upon air-bursting, sample fluid travels through the microfluidic channel into the biosensor detection chamber. In
sequence: loading the dispenser; dispensing; multiplexing, and delivered volume to biosensor array
Microfluidic Devices and Their Applications to Lab-on-a-Chip 18.4 Lab-on-a-Chip for Biochemical Analysis 525
sample (as low as 100 nl). One of the most fundamental an oxygen semipermeable membrane because of its
sensor designs is the oxygen sensor, which is the ba- high permeability and low signal-to-noise ratio. Water
sic sensing structure for many other metabolic products molecules pass through the silicone membrane and re-
such as glucose and lactate. The principle of the oxygen constitute the gel-based electrolyte so the Cl− ions can
sensor is based on amperometric detection. Figure 18.30 move close to the anode to coalesce with Ag+ ions. The
shows a schematic representation of an oxygen sensor. number of electrons in this reaction is counted by the
When the diffusion profile for oxygen from the sample measuring system.
to the electrode surface is saturated, a constant oxygen For the glucose sensor, additional layers – a glucose
gradient profile is generated. Under these circumstances semipermeable membrane (polyurethane) and immobi-
the detection current is proportional only to the oxygen lized glucose oxidase (GOD) in a polyacrylamide gel
concentration in the sample. for the glucose sensor – allow direct conversion of the
The gel-based electrolyte is essential for the ion- oxygen sensor into a glucose sensor. A similar modi-
exchange reactions at the anode of the electrochemical fication is made for the lactate sensor by replacing the
pair. The oxygen semipermeable membrane ensures that immobilized glucose oxidase with lactate oxidase. The
mainly oxygen molecules permeate through this layer glucose molecules will pass through the semiperme-
and that the electrochemical cell is not exposed to other able layer and be oxidized immediately. The oxygen
ions. A silicone layer was spin-coated and utilized as sensor will measure hydrogen peroxide, which is a by-
250
80
y = 0.2314x + 5.9103
y = 31.993x – 4.632
R2 = 0.9982
Part B 18.4
200 R2 = 0.9848
60
150
40
100
20
50
0 0
0 100 200 300 400 0 2 4 6 8
Glucose concentration (mg/l) Lactate concentration (mEq/l)
Fig. 18.32a,b Measurement results from the biochip cartridge and analyzer: (a) glucose level and (b) lactate level
(after [18.87],
c 2004 IEEE)
526 Part B MEMS/NEMS and BioMEMS/NEMS
contacts
components of the polymer lab-on-a-chip system are
RE the magnetic separator and the IDA electrochemical
WE1 sensor.
WE2 Magnetic separation is achieved with electroplated
CE 500 µm
permalloy microarray on a polymer substrate where
IDA sensor Magnetic bead magnetic field is applied by external permanent magnets
separator
as shown in Fig. 18.34. External permanent magnets
Fig. 18.34a,b Magnetic bead separator and disposable lab-on-a- (NdFeB) are located at both sides of the magnetic sep-
chip: (a) schematic illustration of magnetic bead separator with arator creating the focused magnetic flux along with
electroplated permalloy microarray and (b) photograph of a fab- the permalloy microarray and trapping magnetic par-
ricated lab-on-a-chip (after [18.88], c The Royal Society of ticles on the magnetic array. The IDA electrochemical
Chemistry) sensor is similar to that of in previous Sect. 18.4.1,
Microfluidic Devices and Their Applications to Lab-on-a-Chip References 527
7 60 1.00 µg/ml
0.75 µg/ml
6 50 0.50 µg/ml
0.25 µg/ml
5 0.10 µg/ml
40 0.05 µg/ml
Buffer
4
30
3
20
2
10
1
0
0
10 20 30 40 50 60 0 20 40 60 80
Time (s) Time (s)
Fig. 18.35 Separation efficiency of magnetic beads. Time Fig. 18.36 Immunoassay result for the various mouse IgG
period for flowing magnetic beads varied from 10 to 60 s at concentrations detected in the disposable lab-on-a-chip
flow rate of 20 μl/min (after [18.88],
c The Royal Society system for magnetic immunoassay (after [18.88], c The
of Chemistry) Royal Society of Chemistry)
which has two working electrodes for signal ampli- ciency was obtained by measuring the signal from the
fication by redox cycling [18.87]. All three modules bead-conjugated alkaline phosphatase and comparing it
are assembled using a room temperature UV bonding with a calibration curve as shown in Fig. 18.35.
technique [18.89] and a new metal pattern embedding After verification of each components of the poly-
technique [18.90]. mer lab-on-a-chip, a full immunoassay was carried
The polymer lab-on-a-chip for magnetic bead-based out with the lab-on-a-chip system. Same sequences
immunoassay was characterized to verify its func- described in Sect. 18.4.1 were applied. Figure 18.36
tionality of sampling and detection. The steps are shows the immunoassay results for different antigen
described in Sect. 18.4.1 using mouse IgG as target concentration. A low detection limit of 16.4 ng/ml was
antigen. The dynamic range of the IDA electrochem- achieved using the described method with a disposable
ical sensor was investigated by measuring the signal polymer lab-on-a-chip.
from known concentrations of bead-conjugated alkaline The immunoassay results have proved that the dis-
phosphatase (AP) before the magnetic immunoassay. posable polymer lab-on-a-chip system promises a po-
Bead-conjugated alkaline phosphatase was externally tential in fast and small volume biochemical detection
prepared at known concentrations. The separation effi- and analysis.
References
Part B 18
18.1 L. Ristic, H. Hughes, F. Shemansky: Bulk micro- in deep silicon trench etching with profile
machining technology. In: Sensor Technology and control, J. Micromech. Microeng. 5, 115–120
Devices, ed. by L. Ristic (Artech House, Norwood (1995)
1994) pp. 49–93 18.4 Z.L. Zhang, N.C. MacDonald: A RIE process for
18.2 M. Esashi, M. Takinami, Y. Wakabayashi, K. Mi- submicron silicon electromechanical structures,
nami: High-rate directional deep dry etching for J. Micromech. Microeng. 2, 31–38 (1992)
bulk silicon micromachining, J. Micromech. Micro- 18.5 C. Linder, T. Tschan, N.F. de Rooij: Deep dry etch-
eng. 5, 5–10 (1995) ing techniques as a new IC compatible tool for
18.3 H. Jansen, M. de Boer, R. Legtenberg, M. El- silicon micromachining, Technical Digest 6th Int.
wenspoek: The black silicon method: A universal Conf. Solid-State Sens. Actuators (Transducers ’91),
method for determining the parameter set- San Francisco June 24–27, 1991 (IEEE, Piscataway
ting of a fluorine-based reactive ion etcher 1991) 524–527
528 Part B MEMS/NEMS and BioMEMS/NEMS
18.6 D.J. Harrison, A. Manz, Z. Fan, H. Ludi, H.M. Wid- 18.19 C.H. Ahn, A. Puntambekar, S.M. Lee, H.J. Cho, C.-
mer: Capillary electrophoresis and sample injection C. Hong: Structurally programmable microfluidic
systems integrated on a planar glass chip, Anal. systems, Proc. Micro-Total Anal. Syst. 2000, En-
Chem. 64, 1926–1932 (1992) schede 2000, ed. by A. van den Berg, W. Olthius, P.
18.7 A. Manz, D.J. Harrison, E.M.J. Verpoorte, J.C. Fet- Bergveld (Kluwer, Dordrecht 2000) 205–208
tinger, A. Paulus, H. Ludi, H.M. Widmer: Planar 18.20 W. Schomburg, B. Scherrer: 3.5 mm thin valves in
chips technology for miniaturization and inte- titanium membranes, J. Micromech. Microeng. 2,
gration of separation technique into monitoring 184–186 (1992)
systems, J. Chromatogr. 593, 253–258 (1992) 18.21 A. Ilzofer, B. Ritter, C. Tsakmasis: Development of
18.8 T.R. Dietrich, W. Ehrfeld, M. Lacher, M. Krämer, passive microvalves by the finite element method,
B. Speit: Fabrication technologies for microsystems J. Micromech. Microeng. 5, 226–230 (1995)
utilizing photoetchable glass, Microelectron. Eng. 18.22 B. Paul, T. Terharr: Comparison of two passive
30, 497–504 (1996) microvalve designs for microlamination architec-
18.9 S.C. Jacobson, R. Hergenroder, L.B. Koutny, tures, J. Micromech. Microeng. 10, 15–20 (2000)
J.M. Ramsey: Open channel electrochromatogra- 18.23 S. Shoji, M. Esashi: Microflow device and systems,
phy on a microchip, Anal. Chem. 66, 2369–2373 J. Micromech. Microeng. 4, 157–171 (1994)
(1994) 18.24 G.T.A. Kovacs: Micromachined Transducers Source-
18.10 E.W. Becker, W. Ehrfeld, P. Hagmann, A. Maner, book (McGraw–Hill, New York 1998)
D. Munchmeyer: Fabrication of microstructures 18.25 W. Schomburg, J. Fahrenberg, D. Maas, R. Rapp:
with high aspect ratios and great structural Active valves and pumps for microfluidics, J. Mi-
heights by synchrotron radiation lithography, gal- cromech. Microeng. 3, 216–218 (1993)
vanoforming, and plastic moulding (LIGA process), 18.26 K. Hosokawa, R. Maeda: A pneumatically-actuated
Microelectron. Eng. 4, 35–36 (1986) three way microvalve fabricated with poly-
18.11 M.G. Allen: Polyimide-based process for the fabri- dimethylsiloxane using the membrane transfer
cation of thick electroplated microstructures, Proc. technique, J. Micromech. Microeng. 10, 415–420
7th Int. Conf Solid-State Sens. Actuators (Transduc- (2000)
ers ’93), Yokohama June 7–10, 1993 (IEE of Japan, 18.27 J. Fahrenberg, W. Bier, D. Maas, W. Menz,
Tokyo 1993) 60–65 R. Ruprecht, W. Schomburg: A microvalve system
18.12 C.H. Ahn, M.G. Allen: A fully integrated sur- fabricated by thermoplastic molding, J. Mi-
face micromachined magnetic microactuator with cromech. Microeng. 5, 169–171 (1995)
a multilevel meander magnetic core, IEEE/ASME J. 18.28 C. Goll, W. Bacher, B. Bustgens, D. Maas, W. Menz,
Microelectromech. Syst. (MEMS) 2, 15–22 (1993) W. Schomburg: Microvalves with bistable buck-
18.13 C.S. Effenhauser, G.J.M. Bruin, A. Paulus, M. Ehrat: led diaphragms, J. Micromech. Microeng. 6, 77–79
Integrated capillary electrophoresis on flexible sil- (1996)
icone microdevices: Analysis of DNA restriction 18.29 A. Ruzzu, J. Fahrenberg, M. Heckele, T. Schaller:
fragments and detection of single DNA molecules Multi-functional valve components fabricated by
on microchips, Anal. Chem. 69, 3451–3457 (1997) combination of LIGA process and high precision
18.14 D.C. Duffy, J.C. McDonald, O.J.A. Schueller, mechanical engineering, Microsyst. Technol. 4,
G.M. Whitesides: Rapid prototyping of microfluidic 128–131 (1998)
systems in poly(dimethylsiloxane), Anal. Chem. 70, 18.30 K. Sato, M. Shikida: An electrostatically actuated
4974–4984 (1998) gas valve with S-shaped film element, J. Mi-
18.15 L. Martynova, L. Locascio, M. Gaitan, G.W. Kramer, cromech. Microeng. 4, 205–209 (1994)
R.G. Christensen, W.A. MacCrehan: Fabrication of 18.31 J. Robertson, K.D. Wise: A low pressure micro-
plastic microfluid channels by imprinting methods, machined flow modulator, Sens. Actuators A 71,
Part B 18
18.35 M. Stehr, S. Messner, H. Sandmaier, R. Zangerle: 18.51 F. Forster, R. Bardell, M. Afromowitz, N. Sharma:
The VAMP – A new device for handling liquids and Design, fabrication and testing of fixed-valve mi-
gases, Sens. Actuators A 57, 153–157 (1996) cropumps, Proc. ASME Fluids Engineering Division,
18.36 W. Jackson, H. Tran, M. O’Brien, E. Rabinovich, ASME Int. Mech. Eng. Congr. Expo. 234, 39–44 (1995)
G. Lopez: Rapid prototyping of active microfluidic 18.52 T. Thorsen, S.J. Maerkl, S.R. Quake: Monolithic mi-
components based on magnetically modified elas- crofabricated valves and pumps by multilayer soft
tomeric materials, J. Vac. Sci. Technol. B 19, 596 lithography, Science 288, 113–116 (2000)
(2001) 18.53 T. Thorsen, S.J. Maerkl, S.R. Quake: Microflu-
18.37 M. Kohl, K. Skrobanek, S. Miyazaki: Development of idic large-scale integration, Science 298, 580–584
stress-optimized shape memory microvalves, Sens. (2002)
Actuators A 72, 243–250 (1999) 18.54 J.W. Hong, V. Studer, G. Hang, W.F. Anderson,
18.38 C. Neagu, J. Gardeniers, M. Elwenspoek, J. Kelly: An S.R. Quake: A nanoliter-scale nucleic acid proces-
electrochemical active valve, Electrochem. Acta 42, sor with parallel architecture, Nature Biotechnol.
3367–3373 (1997) 22, 435–439 (2004)
18.39 K. Yoshida, M. Kikuchi, J. Park, S. Yokota: Fabrica- 18.55 J.C. Love, J.R. Anderson, G.M. Whitesides: Fabrica-
tion of micro-electro-rheological valves (ER valves) tion of three-dimensional microfluidic systems by
by micromachining and experiments, Sens. Actua- soft lithography, MRS Bulletin 7, 523–528 (2001)
tors A 95, 227–233 (2002) 18.56 G. Fuhr, T. Schnelle, B. Wagner: Travelling
18.40 N. Jeon, D. Chiu, C. Wargo, H. Wu, I. Choi, J. An- wave driven microfabricated electrohydrodynamic
derson, G.M. Whitesides: Design and fabrication pumps for liquids, J. Micromech. Microeng. 4, 217–
of integrated passive valves and pumps for flex- 226 (1994)
ible polymer 3-dimensional microfluidic systems, 18.57 J. Jang, S. Lee: Theoretical and experimental study
Biomed. Microdevices 4, 117–121 (2002) of MHD micropump, Sens. Actuators A 80, 84–89
18.41 M. Koch, N. Harris, A. Evans, N. White, (2000)
A. Brunnschweiler: A novel micropump design with 18.58 A. Lemoff, A.P. Lee: An AC magnetohydrodynamic
thick-film piezoelectric actuation, Meas. Sci. Tech- micropump, Sens. Actuators B 63, 178–185 (2000)
nol. 8, 49–57 (1997) 18.59 J. DeSimone, G. York, J. McGrath: Synthe-
18.42 L. Cao, S. Mantell, D. Polla: Design and simulation sis and bulk, surface and microlithographic
of an implantable medical drug delivery system characterization of poly(1-butene sulfone)-g-
using microelectromechanical systems technology, poly(dimethylsiloxane), Macromolecules 24, 5330–
Sens. Actuators A 94, 117–125 (2001) 5339 (1994)
18.43 J. Park, K. Yoshida, S. Yokota: Resonantly driven 18.60 A. Wego, L. Pagel: A self-filling micropump based
piezoelectric micropump fabrication of a microp- on PCB technology, Sens. Actuators A 88, 220–226
ump having high power density, Mechatronics 9, (2001)
687–702 (1999) 18.61 A. Terray, J. Oakey, D. Marr: Fabrication of colloidal
18.44 M. Koch, A. Evans, A. Brunnschweiler: The dy- structures for microfluidic applications, Appl. Phys.
namic micropump driven with a screen printed Lett. 81, 1555–1557 (2002)
PZT actuator, J. Micromech. Microeng. 8, 119–122 18.62 A. Puntambekar, J.-W. Choi, C.H. Ahn, S. Kim,
(1998) V.B. Makhijani: Fixed-volume metering microdis-
18.45 D. Xu, L. Wang, G. Ding, Y. Zhou, A. Yu, B. Cai: penser module, Lab Chip 2, 213–218 (2002)
Characteristics and fabrication of NiTi/Si diaphragm 18.63 M. Madou, Y. Lu, S. Lai, C. Koh, L. Lee, B. Wenner:
micropump, Sens. Actuators A 93, 87–92 (2001) A novel design on a CD disc for 2-point calibra-
18.46 E. Makino, T. Mitsuya, T. Shibata: Fabrication of TiNi tion measurement, Sens. Actuators A 91, 301–306
shape memory micropump, Sens. Actuators A 88, (2001)
Part B 18
256–262 (2001) 18.64 K. Handique, D. Burke, C. Mastrangelo, C. Burns:
18.47 H. Chou, M. Unger, S. Quake: A microfabricated ro- On-chip thermopneumatic pressure for discrete
tary pump, Biomed. Microdevices 3, 323–330 (2001) drop pumping, Anal. Chem. 73, 1831–1838 (2001)
18.48 R. Zengerle, J. Ulrich, S. Kulge, M. Richter, 18.65 H. Andersson, W. Wijngaart, P. Griss, F. Niklaus,
A. Richter: A bidirectional silicon micropump, Sens. G. Stemme: Hydrophobic valves of plasma de-
Actuators A 50, 81–86 (1995) posited octafluorocyclobutane in DRIE channels,
18.49 M. Carrozza, N. Croce, B. Magnani, P. Dario: Sens. Actuators B 75, 136–141 (2001)
A piezoelectric driven stereolithography-fabricated 18.66 L. Low, S. Seetharaman, K. He, M. Madou: Mi-
micropump, J. Micromech. Microeng. 5, 177–179 croactuators toward microvalves for responsive
(1995) controlled drug delivery, Sens. Actuators B 67, 149–
18.50 H. Andersson, W. Wijngaart, P. Nilsson, P. Enoks- 160 (2000)
son, G. Stemme: A valveless diffuser micropump 18.67 Q. Yu, J. Bauer, J. Moore, D. Beebe: Responsive
for microfluidic analytical systems, Sens. Actuators biomimetic hydrogel valve for microfluidics, Appl.
B 72, 259–265 (2001) Phys. Lett. 78, 2589–2591 (2001)
530 Part B MEMS/NEMS and BioMEMS/NEMS
18.68 C.-C. Hong, J.-W. Choi, C.H. Ahn: A novel in- H.T. Henderson, C.H. Ahn: An integrated mi-
plane passive microfluidic mixer with modified crofluidic biochemical detection system for protein
Tesla structures, Lab. Chip 4, 109–113 (2004) analysis with magnetic bead-based sampling ca-
18.69 M. Mitchell, V. Spikmans, F. Bessoth, A. Manz, A. pabilities, Lab Chip 2, 27–30 (2002)
de Mello: Towards organic synthesis in microfluidic 18.81 C.H. Ahn, H.T. Henderson, W.R. Heineman,
devices: Multicomponent reactions for the con- H.B. Halsall: Development of a generic microfluidic
struction of compound libraries, Proc. Micro-Total system for electrochemical immunoassay-based
Anal. Syst. 2000, Enschede 2000, ed. by A. van den remote bio/chemical sensors, Proc. Micro-Total
Berg, W. Olthius, P. Bergveld (Kluwer, Dordrecht Anal. Syst. ’98, Banff, Canada 1998, ed. by D.J.
2000) 463–465 Harrison, A. van den Berg (Kluwer, Dordrecht 1998)
18.70 D. Beebe, R. Adrian, M. Olsen, M. Stremler, H. Aref, pp. 225–230
B. Jo: Passive mixing in microchannels: Fabrication 18.82 J.-W. Choi, C.H. Ahn, S. Bhansali, H.T. Hen-
and flow experiments, Mec. Ind. 2, 343–348 (2001) derson: A new magnetic bead-based, filterless
18.71 A. Stroock, S. Dertinger, A. Ajdari, I. Mezic, H. Stone, bio-separator with planar electromagnet surfaces
G.M. Whitesides: Chaotic mixer for microchannels, for integrated bio-detection systems, Sens. Actua-
Science 295, 647–651 (2002) tors B 68, 34–39 (2000)
18.72 J. Brody, P. Yager: Diffusion-based extraction in 18.83 O. Niwa, Y. Xu, H.B. Halsall, W.R. Heineman: Small-
a microfabricated device, Sens. Actuators A 54, volume voltammetric detection of 4-aminophenol
704–708 (1996) with interdigitated array electrodes and its appli-
18.73 B. Weigl, P. Yager: Microfluidic diffusion-based cation to electrochemical enzyme immunoassay,
separation and detection, Science 283, 346–347 Anal. Chem. 65, 1559–1563 (1993)
(1999) 18.84 K.W. Oh, A. Han, S. Bhansali, H.T. Henderson,
18.74 A. Puntambekar, J.-W. Choi, C.H. Ahn, S. Kim, C.H. Ahn: A low-temperature bonding technique
S. Bayyuk, V.B. Makhijani: An air-driven fluidic using spin-on fluorocarbon polymers to assemble
multiplexer integrated with microdispensers, Proc. microsystems, J. Micromech. Microeng. 12, 187–191
μTAS 2001 Symp., Monterey 2001, ed. by J.M. Ram- (2002)
sey, A. van den Berg (Kluwer, Dordrecht 2001) 18.85 C.-C. Hong, J.-W. Choi, C.H. Ahn: An on-chip air-
78–80 bursting detonator for driving fluids on disposable
18.75 T. Nagakura, K. Ishihara, T. Furukawa, K. Masuda, lab-on-a-chip systems, J. Micromech. Microeng.
T. Tsuda: Auto-regulated osmotic pump for insulin 17, 410–417 (2007)
therapy by sensing glucose concentration with- 18.86 C. Gao, J.-W. Choi, M. Dutta, S. Chilukuru,
out energy supply, Sens. Actuators B 34, 229–233 J.H. Nevin, J.Y. Lee, M.G. Bissell, C.H. Ahn: A fully
(1996) integrated biosensor array for measurement of
18.76 Y. Su, L. Lin, A. Pisano: Water-powered osmotic mi- metabolic parameters in human blood, Proc. 2nd
croactuator, Proc. of the 14th IEEE MEMS Workshop Annu. Int. IEEE–EMBS Spec. Top. Conf. Microtech-
(MEMS 2001), Interlaken 2001 (IEEE, Piscataway) nol. Med. Biol., Madison May 2–4, 2002, ed. by A.
pp. 393–396 Dittmar, D. Beebe (IEEE, New York 2002) pp. 223–226
18.77 G.M. Walker, D.J. Beebe: A passive pumping 18.87 C.H. Ahn, J.-W. Choi, G. Beaucage, J.H. Nevin, J.-
method for microfluidic devices, Lab Chip 2, 131–134 B. Lee, A. Puntambekar, J.Y. Lee: Disposable Smart
(2002) Lab on a Chip for Point-of-Care Clinical Diagnostics,
18.78 C. Effenhauser, H. Harttig, P. Kramer: An Proc. IEEE (2004) pp. 154–173
evaporation-based disposable micropump concept 18.88 J. Do, C.H. Ahn: A polymer lab-on-a-chip for mag-
for continuous monitoring applications, Biomed. netic immunoassay with on-chip sampling and
Microdevices 4, 27–32 (2002) detection capabilities, Lab. Chip 8, 542–549 (2008)
Part B 18
18.79 J.-W. Choi, K.W. Oh, A. Han, N. Okulan, 18.89 J. Han, S.H. Lee, A. Puntambekar, S. Murugesan,
C.A. Wijayawardhana, C. Lannes, S. Bhansali, J.-W. Choi, G. Beaucage, C.H. Ahn: UV adhesive
K.T. Schlueter, W.R. Heineman, H.B. Halsall, bonding techniques at room temperature for plas-
J.H. Nevin, A.J. Helmicki, H.T. Henderson, C.H. Ahn: tic lab-on-a-chip, Proc. 7th Int. Conf. Micro Total
Development and characterization of microflu- Anal. Syst. (μ-TAS), Squaw Valley 2003, ed. by M.A.
idic devices and systems for magnetic bead-based Northrup, K.F. Jensen, D.J. Harrison (Transducers
biochemical detection, Biomed. Microdevices 3, Research Foundation, Cleveland 2003) pp. 1113–1116
191–200 (2001) 18.90 J. Do, J.-W. Choi, C.H. Ahn: Low-cost magnetic in-
18.80 J.-W. Choi, K.W. Oh, J.H. Thomas, W.R. Heine- terdigitated array on a plastic wafer, IEEE Trans.
man, H.B. Halsall, J.H. Nevin, A.J. Helmicki, Magn. 40, 3009–3011 (2004 )
531
Centrifuge-B
19. Centrifuge-Based Fluidic Platforms
Jim V. Zoval, Guangyao Jia, Horacio Kido, Jitae Kim, Nahui Kim, Marc J. Madou
Once it became apparent that individual chemical or bi- envisioned that instrumentation would reduce matrix
ological sensors used in complex samples would not complexities through filtration, separation, and con-
Part B 19
attain the hoped for sensitivity or selectivity, wide centration of the target compound while at the same
commercial use became severely hampered and sensor time ameliorating selectivity and sensitivity of the over-
arrays and sensor instrumentation were proposed in- all system by frequent recalibration and washing of
stead. It was projected that by using orthogonal sensor the sensors. With microfluidics, the miniaturization
array elements (e.g., in electronic noses and tongues) se- of analytical equipment may potentially alleviate the
lectivity would be improved dramatically [19.1]. It was shortcomings associated with large and expensive in-
532 Part B MEMS/NEMS and BioMEMS/NEMS
strumentation through the reduction in reagent volumes, putting a damper on the original enthusiasm for this
favorable scaling properties of several important instru- sensor approach. In the case of miniaturization of in-
ment processes (basic theory of hydrodynamics and strumentation through the application of microfluidics,
diffusion predicts faster heating and cooling and more progress was made in the development of platforms for
efficient chromatographic and electrophoretic separa- high-throughput screening (HTS) as evidenced by new
tions in miniaturized equipment) and batch fabrication products introduced by, for example, Caliper and Tecan
which may enable low-cost disposable instruments Boston [19.3, 4]. In contrast, progress with miniatur-
to be used once and then thrown away to prevent ized analytical equipment remains limited; platforms
sample contamination [19.2]. Micromachining (micro- have been developed for a limited amount of human
electromechanical systems (MEMS)) might also allow and veterinary diagnostic tests that do not require com-
cofabrication of many integrated functional instrument plex fluidic design, see for example Abaxis [19.5]. In
blocks. Tasks that are now performed in a series of this review paper we are, in a narrow sense, summa-
conventional bench top instruments could then be com- rizing the state of the art of compact disc (CD)-based
bined into one unit, reducing labor and minimizing the microfluidics and in a broader sense we are comparing
risk of sample contamination. the technical barriers involved in applying microfluidics
Today it appears that sensor-array development in to sensing and diagnostic as opposed to applying such
electronic noses and tongues has slowed down because techniques to HTS. It will quickly become apparent that
of the lack of highly stable chemical and biologi- the former poses the more severe technical challenges
cal sensors: too frequent recalibration of the sensors and as a result the promise of lab-on-a-chip has not been
and relearning of the pattern recognition software is fulfilled yet.
General remarks Inexpensive CD drive, Standard technique Least mature of the Mixing difficult
mixing is easy, Difficult to four techniques High-voltage source
most samples possible miniaturize and Might be too is dangerous and many
(including cells) multiplex expensive parameters influence
Better for diagnostics Better for smallest propulsion, better for
samples smallest samples (HTS)
Centrifuge-Based Fluidic Platforms 19.1 Why Centripetal Force for Fluid Propulsion? 533
ics, direct-current (DC) and alternating-current (AC) flow channels can simply be laid on top of the acous-
electrokinetics, centrifuge, etc. In Table 19.1 we com- tic pump network in the reader instrument. The method
pare four of the more important and promising fluid is considerably more complex to implement than elec-
propulsion means [19.6]. The pressure that mechani- troosmosis (see next paragraph) but the insensitivity of
cal pumps have to generate to propel fluids through acoustic streaming to the chemical nature of the fluids
capillaries is higher the narrower the conduit. Pres- inside the fluidic channels and its ability to mix fluids
sure and centripetal force are both volume-dependent make it a potentially viable approach. A typical flow
forces, which scale as L 3 (in this case L is the charac- rate measured for water in a small metal pipe lying
teristic length corresponding to the capillary diameter). on a piezoelectric plate is 0.02 cm3 /s at 40 V, peak to
Piezoelectric, electroosmotic, electrowetting and elec- peak [19.7]. Today acoustic streaming as a propulsion
trohydrodynamic (EHD) pumping (the latter two are not mechanism remains in the research stage.
shown in Table 19.1) all scale as surface forces (L 2 ), Electro-osmotic pumping (DC electrokinetics) in
which represent more favorable scaling behavior in the a capillary does not involve any moving parts and
microdomain (propulsion forces scaling with a lower is easily implemented. All that is needed is a metal
power of the critical dimension become more attrac- electrode in some type of a reservoir at each end of
tive in the microdomain) and lend themselves better to a small flow channel. Typical electroosmotic flow ve-
pumping in smaller and longer channels. In principle, locities are on the order of 1 mm/s with a 1200 V/cm
this should make pressure- and centrifuge-based sys- applied electric field. For example, in free-flow capil-
tems less favorable but other factors turn out to be more lary electrophoresis work by Jorgenson and Guthrie,
decisive; despite better scaling of the nonmechanical electroosmotic flow of 1.7 mm/s was reported [19.8].
pumping approaches in Table 19.1, almost all biotech- This is fast enough for most analytical purposes. Har-
nology equipment today remain based on traditional rison et al. achieved electroosmotic pumping with flow
external syringe or peristaltic pumps. The advantages rates up to 1 cm/s in 20 μm capillaries that were micro-
of these approach are that they rely on well-developed, machined in glass [19.9]. They also demonstrated the
commercially available components and that a very injection, mixing and reaction of fluids in a manifold of
wide range of flow rates is attainable. Although inte- micromachined flow channels without the use of valves.
grated micromachined pumps based on two one-way The key aspect for tight valving of liquids at intersecting
valves may achieve precise flow control on the order of capillaries in such a manifold is the suppression of con-
1 μl/min with fast response, high sensitivity, and negli- vective and diffusion effects. The authors demonstrated
gible dead volume, these pumps generate only modest that these effects can be controlled by the appropri-
flow rates and low pressures, and consume a large ate application of voltages to the intersecting channels
amount of chip area and considerable power. simultaneously. Some disadvantages of electroosmosis
Acoustic streaming is a constant (DC) fluid motion are the high voltage required (1–30 kV power supply)
induced by an oscillating sound field at a solid/fluid and direct electrical–fluid contact with resulting sensi-
boundary. A disposable fluidic manifold with capillary tivity of flow rate to the charge of the capillary wall
LabCD reader
LabCD disc
Spectrophotometric
Analysis optics
read cuvette
Part B 19.1
Fluidics manifold
and the ionic strength and pH of the solution. It is mented. A whole range of fluidic functions including
consequently more difficult to make it into a generic valving, decanting, calibration, mixing, metering, sam-
propulsion method. For example, liquids with high ionic ple splitting, and separation can be implemented on
strength cause excessive Joule heating; it is therefore this platform and analytical measurements may be
difficult or impossible to pump biological fluids such as electrochemical, fluorescent or absorption based, and
blood and urine. informatics embedded on the same disc could provide
Using a rotating disc, centrifugal pumping pro- test-specific information.
vides flow rates ranging from less than 10 nL/s to An important deciding factor in choosing a fluidic
greater than 100 μL/s depending on disc geometry, system is the ease of implementing valves; the method
rotational rate (revolutions per minute (RPM)), and that solves the valving issue most elegantly (traditional
fluid properties (Fig. 19.1) [19.10]. Pumping is rela- pumps) is already commercially accepted, even if it is
tively insensitive to physicochemical properties such as not the most easily scaled method. In traditional pumps
pH, ionic strength, or chemical composition (in con- two one-way valves form a barrier for both liquids and
trast to AC and DC electrokinetic means of pumping). vapors. In the case of the microcentrifuges, valving is
Aqueous solutions, solvents (e.g., dimethyl sulfoxide accomplished by varying the rotation speed and capil-
(DMSO)), surfactants, and biological fluids (blood, lary diameter. Thus, no real physical valve is required to
milk, and urine) have all been pumped successfully. stop water flow, but as in the case of acoustic and elec-
Fluid gating, as we will describe in more detail fur- trokinetic pumping, there is no simple means to stop
ther below, is accomplished using capillary valves in vapors from spreading over the whole fluidic platform.
which capillary forces pin fluids at an enlargement in If the liquids need to be stored for a long time, the
a channel until rotationally induced pressure is suffi- valves, which are often disposable in sensing and di-
cient to overcome the capillary pressure (at the so-called agnostics applications, must be barriers to both liquid
burst frequency) or by hydrophobic methods. Since the and vapor. Some initial attempts at implementing vapor
types and the amounts of fluids one can pump on a cen- barriers on CDs will be reported in this review.
trifugal platform spans a greater dynamic range than From the preceding comparison of fluidic propul-
for electrokinetic and acoustic pumps, this approach sion methods for sensing and diagnostic applications,
seems more amenable to sample preparation tasks centrifugation in fluidic channels and reservoirs crafted
than electrokinetic and acoustic approaches. Moreover in a CD-like plastic substrate as shown in Fig. 19.1 con-
miniaturization and multiplexing are quite easily imple- stitute an attractive fluidic platform.
measured flow rates of water, plasma, bovine blood, terfacial area where diffusion occurs, which increases
three concentrations of hematocrit, urine, dimethyl sul- the mean values of the diffusion gradients that drive
foxide (DMSO), and polymerase chain reaction (PCR) the diffusion process; one may call this process an en-
products and report that centrifugal pumping is rela- hanced diffusional process. In addition to the simple and
tively insensitive to such physicochemical properties as enhanced diffusional processes, one can create turbu-
ionic strength, pH, conductivity, and the presence of lence on the CD by emptying two narrow streams to be
various analytes, noting good agreement between exper- mixed into a common chamber. The streams violently
iment and theory for all the liquids. splash against a common chamber wall, causing their
effective mixing (no continuity of the liquid columns
19.2.2 Some Simple Fluidic Functions is required on the CD-based system, in contrast to the
Demonstrated on a CD case of electrokinetics platforms where a broken liquid
column would cause a voltage overload).
Fluid Mixing
In the work by Madou and Kellogg [19.11] and Duffy Valving
et al. [19.10], different means to mix liquids were Valving is an important function in any type of fluidic
designed, implemented, and tested. Observations of platform. Both hydrophobic and capillary valves have
flow velocities in narrow channels on the CD enabled been integrated into the CD platform [19.10,11,13–23].
Reynolds numbers (Re) calculations that established Hydrophobic valves feature an abrupt decrease in the
that the flow remained laminar in all cases. Even in hydrophobic channel cross section, i. e., a hydropho-
the largest fluidic channels tested Re was smaller than bic surface prevents further fluid flow (Fig. 19.2b–d).
100, well below the transition regime from laminar to In contrast, in capillary valves (Fig. 19.2e), liquid flow
turbulent flow (Re ≈ 2300) [19.12]. The laminar flow is stopped by a capillary pressure barrier at junctions
condition necessitates mixing by simple diffusion or by where the channel diameter suddenly expands.
creating special features on the CD that enable advec-
tion or turbulence. In one scenario, fluidic diffusional Hydrophobic Valving. The pressure drop in a chan-
mixing was implemented by emptying two microflu- nel with laminar flow is given by the Hagen–Poiseuille
idic channels together into a single long meandering equation [19.12]
fluidic channel. Proper design of channel length and 12LμQ
reagent reservoirs allowed for stoichiometric mixing ΔP = , (19.3)
wh 3
in the meandering channel by maintaining equal flow
rates of the two streams joining in the mixing chan- where L is the microchannel length, μ is the dynamic
viscosity, Q is the flow rate, and w and h are the channel
Part B 19.2
where σl is the liquid’s surface tension, θc is the contact pressure at the capillary barrier expressed in terms of
angle, w1 and h 1 are the width and height of the channel the interfacial free energy [19.16] is given by
before the restriction, and w2 and h 2 are the width and
Pcb = 4γal sin θc /Dh , (19.5)
height after the restriction, respectively. In hydropho-
bic valving, in order for liquid to move beyond these where γal is the surface energy per unit area of the
pressure barriers, the CD must be rotated above a criti- liquid–air interface, θc is the equilibrium contact an-
cal speed, at which point the centripetal forces exerted gle, and Dh is the hydraulic diameter. Assuming low
on the liquid column overcome the pressure needed to liquid velocities, the flow dynamics may be modeled
move past the valve. by balancing the centripetal force and the capillary bar-
Ekstrand et al. [19.13] used hydrophobic valving on rier pressure (19.5). The liquid pressure at the meniscus,
a CD to control discrete sample volumes in the nanoliter from the centripetal force acting on the liquid, can be
range with centripetal force. Capillary forces draw li- described as
quid into the fluidic channel until there is a change in the Pm = ρω2rΔr , (19.6)
surface properties at the hydrophobic valve region. The
valving was implemented as described schematically in where ρ is the density of the liquid, ω is the angular
Fig. 19.2c. Tiensuu et al. [19.14] introduced localized velocity, r is the average distance from the liquid el-
hydrophobic areas in CD microfluidic channels by ink- ement to the center of the CD, and Δr is the radial
jet printing of hydrophobic polymers onto hydrophilic length of the liquid sample (Fig. 19.2a,e). Liquid will
channels. In this work, hydrophobic lines were printed not pass a capillary valve as long as the pressure at the
onto the bottom wall of channels with both unstructured meniscus Pm is less than or equal to the capillary bar-
(Fig. 19.2c) and structured (Fig. 19.2d) vertical chan- rier pressure Pcb . Zeng and coworkers [19.16] named
nel walls. Several channel width-to-depth ratios were the point at which Pm equals Pcb , the critical burst con-
investigated. The CDs were made by injection mold- dition and the rotational frequency at which it occurs
ing of polycarbonate and were subsequently rendered they called the burst frequency. Experimental values of
hydrophilic by oxygen plasma treatment. Ink-jet print- critical burst frequencies versus channel geometry, for
ing was used for the introduction of the hydrophobic rectangular cross sections over a range of channel sizes,
polymeric material at the valve position. The parts were show good agreement with simulation over the entire
capped with polydimethylsiloxane (PDMS) to form the range of diameters studied. Since these simulations did
fourth wall of the channel. In testing of unstructured not assume an axisymmetric capillary with a circular
channels (without the sawtooth pattern) there were no contact line and a diameter Dh , the meniscus contact
valve failures for 300 and 500 μm wide channels but line may be a complex shape. Burst frequencies were
some failures for the 100 μm channels, however, in shown to be cross-section dependent for equal hydraulic
structured vertical walls (with sawtooth patterns), there diameters. The theoretical burst-frequency equation was
were no valve failures. The authors attribute the better modified as follows to account for variation of the chan-
results of the structured vertical walls to both the fa- nel cross section
vorable distribution of hydrophobic polymer within the ρω2rΔr < 4γal sin θc /(Dh )n , (19.7)
channel and the sharper sidewall geometry to be wetted
(the side walls are hydrophilic since the printed hy- where n = 1.08 for an equilateral triangular cross sec-
drophobic material is only on the bottom of the channel) tion and n = 1.14 for a rectangular cross section. For
compared to the unstructured vertical channel walls. pipe flow (circular cross section) an additional term is
used in the burst-frequency expression
Capillary Valving. Capillary valves have been imple- ρω2rΔr < 4γal sin θc /Dh
mented frequently on CD fluidic platforms [19.10, 11,
+ γal sin θc (1/Dh − 1/D0 ) , (19.8)
15–18, 21, 22]. The physical principle involved is based
on the surface tension, which develops when the cross where the empirically determined constant D0 =
Part B 19.2
section of a hydrophilic capillary expands abruptly as 40 μm. The physical reason for the additional pipe-flow
illustrated in Fig. 19.2e. As shown in this figure, a cap- term, used to get a fit to the simulation results, is not
illary channel connects two reservoirs, and the top well understood at this time.
reservoir (the one closest to the center of the CD) and Duffy et al. [19.10] modeled capillary valving by
the connecting capillary is filled with liquid. For capil- balancing the pressure induced by the centripetal force
laries with axisymmetric cross sections, the maximum (ρω2rΔr) at the exit of the capillary with the pressure
Centrifuge-Based Fluidic Platforms 19.2 Compact Disc or Microcentrifuge Fluidics 537
inside the liquid droplet being formed at the capillary outlet scaled by a factor a (for nonspherical droplet
outlet and the pressure required to wet the chamber be- shapes) and the second term on the right b represent-
yond the valve. The pressure inside a droplet is given by ing the pressure required to wet the chamber beyond
the Young–Laplace equation [19.24] the valve. The b term depends on the geometry of the
ΔP = γ (1/R1 + 1/R2 ) , (19.9) chamber to be filled and the wettability of its walls.
A plot of the centripetal pressure (ρω2crΔr) at which
where γ is the surface tension of the liquid and R1 and the burst occurs verses 1/Dh was linear, as expected
R2 are the meniscus radii of curvature in the x- and y- from (19.11), with a 4.3% coefficient of variation. The
dimensions of the capillary cross section. In the case authors note a potential limitation with capillary valves
of small circular capillary cross sections with spherical due to the fact that liquids with low surface tension tend
droplet shapes, R1 = R2 ∼ = channel cross-section radius to wet the walls of the chamber at the capillary valve
and (19.9) can be rewritten as opening, resulting in the inability to gate the flow. The
ΔP = 4γ /Dh . (19.10) b term in (19.11) is beneficial in gating flow unless the
On this basis Duffy et al. [19.10] derived a simplified surface walls at the abrupt enlargement of the capillary
expression for the critical burst frequency (ωc ) as valve are so hydrophilic that the liquid is drawn past the
valve and into the reservoir.
ρω2crΔr = a(4γ /Dh ) + b , (19.11) Badr et al. [19.17] and Johnson et al. [19.18] have
with the first term on the right representing the pressure designed a CD to sequentially valve fluids through
inside the liquid droplet being formed at the capillary a monotonic increase of rotational rate with progres-
sively higher burst frequencies. The CD, shown in
Fig. 19.3, was designed to carry out an assay for ions
CD based on an optode-based detection scheme. The CD
center design employed five serial capillary valves opening at
different times as actuated by rotational speed. Results
showed good agreement between the observed and the
7 calculated burst frequencies (see later).
It is very important to realize that the valves we
mentioned thus far constitute liquid barriers and that
6 they are not barriers for vapors. Vapor barriers must
5 be implemented in any fluidic platform where reagents
need to be stored for long periods of time. This is
4 especially important for a disposable diagnostic assay
platform. A multimonth, perhaps multiyear, shelf life
3 would require vapor locks in order to prevent reagent
solutions from drying or liquid evaporation and conden-
1 sation in undesirable areas of the fluidic pathway. Tecan
Boston have investigated vapor-resistant valves made of
2 wax that was melted to actuate valve opening [19.25].
ibrant, respectively, reservoirs (2) and (4) contain wash cols, which has been added, for example, to the fluidic
solutions, and reservoir (5) contains the sample. Upon in- design in the Gyrolab MALDI SP1 CD [19.20]. In
creasing rotation rates, calibrant 1, wash 1, calibrant 2, this CD, developed for matrix assisted laser desorption
rinse 2, and then sample were serially gated into the op- ionization (MALDI) sample preparation, a common
tical detection chamber. Absorption of the calibrants and distribution channel feeds several parallel individ-
sample was measured ual sample-preparation fluidic structures (Fig. 19.4).
538 Part B MEMS/NEMS and BioMEMS/NEMS
19.3 CD Applications
19.3.1 Two-Point Calibration of an ionophore, a chromoionophore, and a lipophilic an-
Optode-Based Detection System ionic additive, were cast, with a spin-on device, onto
a support layer and then immobilized on the disc. With
A CD based system with ion-selective optode detection this system, it is possible to deliver calibrant solutions,
and a two-point-calibration structure for the accurate washing buffers, and unknown solutions (e.g., saliva,
detection of a wide variety of ions has been devel- blood, urine, etc.) to the measuring chamber where
oped [19.15, 17, 18]. The microfluidic architecture, the optode membrane is located. Absorbance measure-
Part B 19.3
depicted in Fig. 19.3, is comprised of channels, five ments on a potassium optode indicate that optodes
solution reservoirs, a chamber for colorimetric measure- immobilized on the platform exhibit the theoretical ab-
ment of the optode membrane, and a waste reservoir, sorbance response. Samples of unknown concentration
all manufactured onto a poly(methyl methacrylate) can be quantified to within 3% error by fitting the re-
disc. Ion-selective optode membranes, composed of sponse curve for a given optode membrane using an
plasticized poly(vinyl chloride) impregnated with an acid (for measuring the signal for a fully protonated
Centrifuge-Based Fluidic Platforms 19.3 CD Applications 539
chromoionophore), a base (for fully deprotonated chro- is performed after the substrate is introduced into the de-
moionophore), and two standard solutions. Further, the tection reservoir (reservoir 2). Endpoint measurements
ability to measure ion concentrations employing one (completion of enzyme–substrate reaction) were made
standard solution in conjunction with an acid and base, and compared to conventional microtiter plate methods
and with two standards alone were studied to delineate using similar protocols. The CD ELISA platform was
whether the current architecture could be simplified. Fi- shown to have advantages such as lower reagent con-
nally, the efficacy of incorporating washing steps into sumption, and shorter assay times, explained in terms of
the calibration protocol was investigated. larger surface-to-volume ratios, which favor diffusion-
This work was further extended to include anion- limited processes. Since the reagents were all loaded
selective optodes and fluorescence rather than ab- into the CD at the same time, there was no need for
sorbance detection [19.17]. Furthermore, in addition to manual operator interventions in between fluidic assay
employing a standard excitation source where a fiber steps. The consistent control and repeatability of li-
optic probe is coupled to a lamp, laser diodes were eval- quid propulsion removes experimental errors associated
uated as excitation sources to enhance the fluorescence with inconsistent manual pipetting methods, for exam-
signal. ple, rinsing/washing can be carried out not only with
equal volumes, but equal flow conditions.
19.3.2 CD Platform for Enzyme-Linked
Immunosorbant Assays (ELISA) 19.3.3 Multiple Parallel Assays
The automation of immunoassays on microfluidic plat- The ability to obtain simultaneous and identical flow
forms presents multiple challenges because of the high rates, incubation times, mixing dynamics, and detec-
number of fluidic processes and the many different li- tion makes the CD an attractive platform for multiple
quid reagents involved. Often there is also the need parallel assays. Kellogg et al. [19.22] have reported on
for highly accurate quantitative results at extremely low a CD system that performs multiple (48) enzymatic as-
concentration and care must be taken to prevent nonspe- says simultaneously by combining centrifugal pumping
cific binding of reporter enzymes and to deliver well- in microfluidic channels with capillary valving and col-
defined volumes of reagents consistently. An enzyme- orimetric detection. The investigation of multiplexed
linked immunosorbant assay (i. e., ELISA) is one of parallel enzyme inhibitor assays are needed for high-
the most common immunoassay methods and is of- throughput screening in diagnostics and in screening of
ten carried out in microtiter plates using labor-intensive drug libraries. For example, enzymatic dephosphoryla-
manual pipetting techniques. Recently, Lai et al. [19.21] tion of colorless p-nitrophenol phosphate by alkaline
have implemented an automated enzyme-linked im- phosphatase results in the formation of the yellow-
munosorbant assay on the CD platform. This group used colored p-nitrophenol and inhibition of this reaction
a five-step flow sequence in the same CD design illus- may be quantified by light absorption measurement.
trated in Fig. 19.3. A capture antibody (anti-rat IgG) Theophylline, a known inhibitor of the reaction, was
was applied to the detection reservoir (reservoir 2 in used as the model inhibitory compound in Kellogg
Fig. 19.3) by adsorption to the PMMA CD surface, then et al.’s [19.22] feasibility study. A single assay element
the surface was blocked to prevent nonspecific bind- on the CD contains three reservoirs: one for the en-
ing. Antigen/sample (rat IgG), wash solution, second zyme, one for the inhibitor, and one for the substrate.
antibody, and substrate solutions were loaded into reser- Rotation of the CD allows the enzyme and inhibitor to
voirs 3–7 (Fig. 19.3) respectively. Using capillary valv- pass capillary valves, mix in a meandering 100 μm-wide
ing techniques, the sample and reagents were pumped, channel, and then move to a point where flow is stopped
one at a time, through the detection chamber. First, the by another capillary valve. A further increase in the ro-
sample was introduced for antibody antigen binding tational speed allows the enzyme/inhibitor mixture and
(reservoir 3), then a wash solution (reservoir 4), then an substrate to pass through the next set of capillary valves
Part B 19.3
enzyme-labeled secondary antibody (reservoir 5), then where they are mixed in a second meandering channel
another wash solution (reservoir 6), and finally the sub- and emptied into an on-disc planar cuvette. The CD is
strate was added (reservoir 7). The U-shaped bend in the slowed and absorption through each of the 48 parallel
fluidic path allows the solutions to incubate in the cap- assay cuvettes is measured by reflectance, all in a period
ture zone/detection chamber until the next solution is of 60 s, the entire fluidic process including measure-
released into the chamber. Detection of the fluorescence ment took about 3 min. The CDs were fabricated using
540 Part B MEMS/NEMS and BioMEMS/NEMS
PDMS replication techniques [19.26], with the addi- 19.3.4 Cellular-Based Assays
tion of a white pigment to the PDMS polymerization on CD Platform
for enhanced reflectivity in the colorimetric measure-
ments. The flow rates and meandering channel widths Cell-based assays are often used in drug screen-
were selected such that the diffusion rate would allow ing [19.27] and rely on labor-intensive microtiter plate
90% mixing of the solutions. technologies. Microtiter plate methods may be diffi-
The variation in performance between the individ- cult to automate without the use of large and expensive
ual fluidic CD structures was quantified by carrying liquid-handling systems and they present problems
out the same assay 45 times simultaneously on a CD. with evaporation when scaled down to small volumes.
The background-corrected absorbance was measured Thomas et al. [19.23] have reported on a CD platform-
and the coefficient of variation in the assay was ≈ 3.2%. based automated adherent cell system. This adherent
When the experiment was repeated on different discs cell assay involved introducing the compounds to be
the coefficient of variation was 3–3.5%. Furthermore, screened to a cell culture, then determining if the cells
variation of absorption across a single cuvette was less were killed (a cell viability assay).
than 1%, confirming complete mixing. In experiments Reagents for cell growth, rinsing and viability stain-
to show enzyme inhibition, 45 simultaneous reactions ing were serially loaded into an annular, common
were carried out on the CD using fixed concentra- distribution chamber and centripetal force was used
tions of enzyme and substrate and 15 concentrations of for reagent loading, exchange, and rinsing of the cell
theophylline in triplicate and a complete isotherm was growth chamber (Fig. 19.5). Individual inlets were used
generated for the inhibition of alkaline phosphatase. for the addition of compounds to be screened. The
The three remaining structures were used for calibration plastic channels (Fig. 19.5b) were capped with a poly-
with known concentrations of p-nitrophenol. A dose dimethylsiloxane (PDMS) sheet capable of fast gas
response was seen over three logs of theophylline con- transport in and out of the culture reservoirs.
centration in the range of 0.1–100 mM. The authors HeLA, L929, CHO-M1, and MRC-5 cell lines were
concluded that a large number of identical assays, with cultivated on the CD device. Cell viability assays were
applications in rapid, high-throughput screening, can be performed, on the CD, by removing the growth medium
carried out on the CD platform simultaneously because from the cells, washing the cells with PBS, and intro-
of the symmetric force acting on the fluids in high- ducing a solution of the fluorescence assay reagents
quality identical microfluidic structures and that detec- into the growth chamber. The LIVE/DEAD Viability
tion was simplified by rotating all the reaction mixtures Assay (Molecular Probes, Inc., Eugene) uses a mix-
under a fixed detector. In later work [19.22], the same ture of calcein green-fluorescent nucleic-acid stain and
group has extended the number of assays to 96 per CD the red-fluorescent nucleic-acid stain, ethidium. The
and has investigated fluorescent enzymatic assays. assay performance is based on the differing abili-
ties of the stains to penetrate healthy bacterial cells.
The calcein green-fluorescent dye will label all cells,
a) b)
live or dead. The red-fluorescent ethidium stain will
only label cells with damaged membranes. The red
stain causes a reduction in the green stain fluores-
cence when both dyes are present. When the appropriate
mixture of green and red stains is used, cells with in-
3
tact membranes will have a green fluorescence and
1 cells with damaged membranes will have a red fluo-
2
rescence. The background remains almost completely
nonfluorescent (Fig. 19.6). All liquid transfers were
carried out using centripetal force from CD rotation
Part B 19.3
a) b) c)
Fig. 19.6a–c L929 fibroblasts cultured for 48 hours in CD growth chambers. (a) Phase contrast (scale bar 100 μm),
(b) epifluorescence image of calcein-stained viable cells, (c) epifluorescence image of ethidium-stained nonviable cells
(after [19.23])
polycarbonate disc with predrilled holes for sample in- a similar approach used to demonstrate multiple enzy-
troduction. The microfluidic pattern was made using matic assays on CD [19.10].
a photolithographic pattern on the negative photoresist. The dyes and sample were introduced into reser-
The fluidic system was capped with a polycarbonate voir chambers using a pipette. The dyes fill the chamber
disc that had been laminated with an optical-quality stopping at a capillary valve (valve 1 in Fig. 19.7).
pressure-sensitive adhesive (3M 8142, 3M, Minneapo- Similarly, the sample containing cells was introduced
542 Part B MEMS/NEMS and BioMEMS/NEMS
19.3.5 Integrated
Nucleic-Acid Sample Preparation
and PCR Amplification
m k j 1 cm
spindle of a rotary motor, with the correct angular align-
ment, which is connected by a slip ring to stationary n
power supplies and a temperature controller. Thermo-
couples are used for closed-loop temperature control
and air sockets are used as insulators to isolate heat- n
ing to reservoirs of interest. The thermoelectric at the l
PCR chamber both heats and cools and since the PCR p o
reaction chamber is thin, 0.5 mm, fast thermocycling is
achieved. Slew rates of ±2 ◦ C /s with fluid volumes of q
25 μL and thermal gradients across the liquid of 0.5 ◦ C
are reported. It is important to note here that the PCR
chambers were not sealed; vapor generated inside the the sample-preparation process, without sample loss or
PCR chamber condensed on the cooler surfaces of the contamination, has been enabled on the CD platform by
connecting microfluidic chamber and, since the CD is the Gyrolab MALDI SP1 CD and the Gyrolab Worksta-
rotating, the condensed drops are centrifuged back into tion (Gyros AB, Sweden) [19.20].
the hot PCR chamber. This microcondensation appara- The Gyrolab MALDI SP1 sample-preparation CD
tus is unique for the centrifugal CD platform. Details will process up to 96 samples simultaneously using sep-
of the experimental parameters used can be found in arate microfluidic structures. Protein digest from gels
the original reference [19.22], but to summarize, sam- or solutions are concentrated, desalted, and eluted with
ple preparation and PCR amplification for two types of matrix onto a MALDI target area. The CD is then
samples, whole blood and E. coli, were demonstrated transferred to a MALDI instrument for analysis without
on the CD platform and shown to be comparable to the need for further transfer to a separate target plate.
conventional methods. The CD fluidic structure contains functions for com-
mon reagent distribution, volume definition (metering),
19.3.6 Sample Preparation valving, reverse phase column (RPC) for concentration
for MALDI MS Analysis and desalting, washing, and target areas for external
calibrants. Figure 19.11 shows the Gyrolab MALDI
MALDI MS peptide mapping is a commonly used SP1 sample preparation CD. The CDs are loaded with
method for protein identification. Correct identification reagents and processed in a completely automated,
and highly sensitive MS analysis require careful sam- custom workstation capable of holding up to five mi-
Part B 19.3
ple preparation. Manual sample preparation is quite crotiter plates containing samples and reagents and up
tedious, time-consuming, and can introduce errors to five CD microlaboratories. The reagents are taken
common to multistep pipetting. MALDI MS sample from the microplates to the CD inlets using a preci-
preparation protocols employ a protein digest followed sion robotic arm fitted with multiple needles, the liquid
by sample concentration, purification, and recrystal- is drawn into specific inlets by capillary forces, and
lization with minimal loss of protein. Automation of then the needles are cleaned by rinsing at a wash sta-
544 Part B MEMS/NEMS and BioMEMS/NEMS
it reaches a hydrophobic valve and the CD is rotated zero, an edge of a pit gives a value of one. The data
to clear the excess liquid in the distribution channel. is retrieved at a constant acquisition rate and the serial
Not until the rotational velocity is further increased is values (0/1) are converted to data of different kinds for
the defined wash volume able to pass through the hy- various applications (music, data, etc.). In addition to
drophobic valve and into the RPC column (SOURCE15 the code generated by the spacing of the pits, optical sig-
RPC). The peptides are eluted from the column and nals necessary for focusing, laser tracking of the spiral
Centrifuge-Based Fluidic Platforms 19.3 CD Applications 545
track, and radial position determination of the read head forward primers and specific reverse primers resulting
are monitored and used in feedback loops for proper CD in amplicons of different length for various species,
operation. The laser is scanned in a radial direction to- were used for verification of PCR on external gels)
ward the outer diameter of the disc with an elaborate were incubated on arrays with species-specific capture
servo that maintains both lateral tracking and vertical probes. Removal of one of the species in the sample re-
focusing. sulted in no probes present on that specific array spot,
Researchers [19.29,30] have taken advantage of this verifying the specificity of the assay.
low-cost high-resolution optical platform in analytical Alexandre et al. [19.30] at Advanced Array Technol-
DNA array applications. Barathur et al. [19.29] from ogy (Namur, Belgium), utilize the inner diameter area
Burstein Technologies (Irvine), for example, have mod- of a CD and standard servo optics for numerical in-
ified the normal CD drive for use as a sophisticated formation and operational control and employ a second
laser-scanning microscope for analysis of a Bio Com- scanning laser system to image DNA arrays on transpar-
pact Disk assay, where all analysis is carried out in ent surfaces at the outer perimeter of a CD. The second
microfluidic chambers on the CD. The assay is carried laser system, consisting of a laser-diode module that il-
out concurrently with the normal optical scanning ca- luminates a 50 μm spot on the CD surface, is scanned
pabilities of a regular CD drive. The authors report on radially at a constant linear velocity of 20 mm/min
the application of this device for DNA microspot-array while the CD is rotating. Each CD contains 15 arrays
hybridization assays and comment on its use in other arranged in a single ring on the CD perimeter that ex-
diagnostic and clinical research applications. For the tends in the radial direction for 15 mm. The arrays are
DNA spot-array application, arrays of captures probes rectangular and consist of four rows and 11 columns
for specific DNA sequences are immobilized on the of 300 μm spots. The normal CD servo optics are lo-
surface of the CD in microfluidic chambers. Sample cated below the disc and the added imaging optics are
preparation and multiplexed PCR, using biotinylated above the disc. A photodiode head follows the imaging
primers, are carried out off-disc, then the biotinylated laser and the refracted light intensity is stored digitally
amplicons are introduced into the array chamber and at a high sampling rate. An image of each array on
hybridization occurs if amplicons with the correct se- the disc is reconstructed by deconvolution of the light-
quence are present. Hybridization detection is achieved intensity data. The entire CD can be scanned in less than
by monitoring the optical signal from the CD photode- one minute, producing a total of 6 MB of information.
tector, while the CD is rotating. To generate an optical Sample preparation and PCR amplification was carried
signal when hybridization has occurred a reporter is out off-disc. Specific DNA capture probes were spot-
used, for the Bio Compact Disk assay the reporter ted on the surface of the CD using a custom arrayer
is a streptavidin-labeled microsphere that will bind that transfers the probes from a multiwell plate onto the
only to the array spots which have successfully cap- surface of up to 12 discs using a robotic arm. Biotiny-
tured biotinylated amplicons. The unbound microsphere lated amplicons are introduced onto the array chambers
reporters are removed from the array using simple cen- (one chamber for each array) and hybridization occurs
trifugation and no further rinsing is needed. As the laser if amplicons with the correct sequence are present. In
is scanned across the CD surface, the microparticle scat- order to get an optical signal that can be detected, after
ters light that would have normally been reflected to a rinse step, a solution of streptavidin-labeled colloidal
the photodetector resulting in less light on the detec- gold particles is applied to the array followed by a Silver
tor (bright-field microscopy) and a distinctive electronic Blue solution (AAT, Namur, Belgium). The silver solu-
signal is generated. The electronic signal-intensity data tion causes silver metal to grow on the gold particles,
can be stored in memory then deconvoluted into an im- thereby making the hybridization-positive microarray
age. A 1 cm2 microarray can be scanned in 20–30 s spots refractive to the incident laser light. Results
with a data-reduction time of 5 min and custom algo- were shown for the detection of the five most com-
rithms that perform the interpretations in real time. Data mon species of Staphylococci and an antibiotic-resistant
Part B 19.3
was shown for identification of three different species strain. The fem A and mec A genes of the various species
of the Brucella coccobacilli on the CD platform. Hu- of Staphylococci were amplified by primers common
man infection occurs by transmission from animals by to all Staphylococci species then hybridized to a mi-
ingestion of infected food products, contact with an croarray containing spots with probes specific for the
infected animal, or inhalation of aerosols. Multiplex different Staphylococci species. The array also included
PCR-amplified DNA from all three species (common a capture probe for the genus Staphylococci and a probe
546 Part B MEMS/NEMS and BioMEMS/NEMS
for the mec A gene that is associated with methicillin re- A number of investigators have attempted to adapt this
sistance of the Staphylococci species. The results were technology to rapidly detect infectious agents in clinical
digitized and quantified with software that is part of specimens for diagnostic purposes [19.31–35]. How-
the custom Bio-CD workstation. Signal-to-noise ratios ever, such systems are still in their infancy and most
were above 50 for all positive signals. of them require technologically complex biochips with
integrated heating/cooling systems [19.31, 32, 36]. The
19.3.8 Microarray Hybridization Madou group at UCI together with the Bergeron group
for Molecular Diagnosis at Laval University have reported [19.37] a CD-based
of Infectious Diseases microfluidic platform for DNA microarray analysis
of infectious disease, presenting an elegant solution
In recent years, microarrays have become important to automate and speed up microarray hybridization.
tools for nucleic-acid analysis and gene-expression pro- Staphiloccocal-specific oligonucleotides were used as
filing. The expression of thousands of genes can be capture probes immobilized in 4 × 5 arrays of 125 μm
monitored in a single experiment using this technology. spots on a standard 3 × 1 in glass slide. The layout
of the array is shown in Fig. 19.12a. A flow cell is
2 designed to realize the self-contained hybridization pro-
a) 4 3
cess in the CD platform. As shown in Fig. 19.12b, the
1
flow cell consists of a hybridization column 1, aligned
with the DNA microarray on the glass slide, a sam-
ple chamber 2, and a rinsing chambers 3 and 4. The
reagent chambers are connected to the hybridization
column with a microchannel which is 50 μm in width
and 25 μm in depth. The flow cell is aligned with
b) and adhered to the glass slide to form a DNA hy-
3600 bridization detection unit, up to five of which can be
mounted into the CD platform fabricated from acrylic
400 plastic using computer numerical control (CNC) ma-
chining (Fig. 19.12c). The reagents are positioned to be
pumped through the hybridization column by centrifu-
c)
gal force in a sequence beginning with chamber 2 up
to chamber 4 and this flow sequence is achieved by
manipulating the balance between the capillary force
and centrifugal pressure. The sample (chamber 2) is
released first and flows over the 140 nl hybridization
chamber (chamber 1) where the oligonucleotide capture
probes are spotted onto the glass support. The rinsing the higher diffusion coefficient of the smaller oligonu-
buffers (chamber 3 and 4) are then released sequentially cleotide molecules.
at a higher angular velocity and are used to wash the To be used for clinical applications, in addition to
nonspecifically bound targets following the hybridiza- being rapid and inexpensive, a molecular test should
tion process. be sensitive and specific. In 5 min of hybridization,
This custom microarray hybridization microfluidic the CD system showed a detection limit of 500 amol
platform is easy to use, automated, and rapid. It uses of amplified target. This result is comparable with
standard glass slides which are compatible with com- results obtained with more complex microfluidic de-
mercial arrayers and standard commercial scanners vices [19.31,43]. One system using chemiluminescence
found in most academic departments. In this remov- shows a detection limit of 250 amol, but requires a 3 h
able microfluidic system, the hybridization chamber is hybridization time [19.44]. In order to detect a signifi-
composed of a low-cost elastomeric material, PDMS cant fluorescent signal, an amplification step is required
using standard moulding methods [19.26], engrafted with microarray technology. The CD microfluidic sys-
with a microfluidic network. This elastomeric material tem reported allows detection of amplicons amplified
reversibly sticks to the glass slide without any adhe- from 10 bacterial genome copies, which is at least
sives or chemical reactions, forming the microfluidic 1000 times more sensitive than results obtained by
unit. Placed onto a plastic compact disc-like support, other groups showing microarray hybridization using
the microfluidic units are spun at different speeds to microfluidic devices [19.45].
control fluid movements. To simplify hybridization ex- In terms of specificity, the CD system was able to
periments using this device, buffer compositions and discriminate four different Staphylococcus species us-
capture probe sequences were optimized to be compat- ing a post-PCR hybridization protocol of only 15 min.
ible with room-temperature hybridizations to prevent The S. aureus probe designed with only one mismatch
the need for a heating device. Furthermore, this mi- in the S. epidermidis amplicon sequence, did not show
crofluidic system allows one to drastically reduce the any significant cross-hybridization. This clearly demon-
volume of reagents needed for microarray hybridiza- strates the possibility to discriminate one SNP using the
tions and does not require a PCR amplicon purification CD system at room temperature and with only 10 μl
step, which may be time-consuming. of washing and rinsing buffer. This SNP discrimination
In a passive hybridization system, a hybridiza- capacity will allow rapid identification of bacteria and
tion event requiring collision between a capture probe their antibiotic-resistance genes.
and the analyte relies solely on diffusion. In such
systems, sensitivity is increased by using longer hy- 19.3.9 Cell Lysis on CD
bridization periods [19.38, 39]. One advantage of flow
through hybridization is that the probability of colli- There are many types of cell-lysis methods used today
sion between the probe and the analyte is increased that are based on mechanical [19.46], physicochemi-
by the much shorter diffusion distance allowed by cal [19.47], chemical [19.48] and enzymatic [19.48]
the shallow hybridization chamber, thereby accelerat- principles. The most commonly used methods in bi-
ing the hybridization kinetics [19.39–41]. In the study it ology research labs rely on chemical and enzymatic
was shown that for the same concentration of 15-mer principles. The main drawbacks of those procedures in-
oligonucleotides or 368 bp amplicons, a five-minute clude intensive labor, adulteration of cell lysate, and
flow through hybridization increased the kinetics of the need for additional purification steps. In order to
hybridization respectively by a factor of 2.5 and 7.5, minimize the required steps for cell lysis, a rapid and
respectively, in comparison with the passive hybridiza- reagent-less cell-lysis method would be very useful.
tion. These results are in line with a previous study. Recently, cell lysis has been demonstrated by Kim
Using a microfluidic system, Chung et al. have shown et al. [19.49] on a microfluidic CD platform. In this
a sixfold rate increase between flow-through hybridiza- purely mechanical lysis method, spherical particles
Part B 19.3
tion versus passive hybridization. However, this system (beads) in a lysis chamber microfabricated in a CD
required a 30 min hybridization step [19.42]. Interest- caused disruption of mammalian (CHO-K1), bacterial
ingly, the difference between passive and flow-through Escherichia coli, and yeast (Saccharomyces cerevisiae)
hybridization was about three times more important cells. Investigators took advantage of interactions be-
for the amplicons compared to the shorter 15 mer tween beads and cells generated in rimming flow [19.50,
oligonucleotides [19.38]. This could be explained by 51] established inside a partially filled annular chamber
548 Part B MEMS/NEMS and BioMEMS/NEMS
used the multiplexed lysis design shown in Fig. 19.14. tivation and completely sequenced genome. Moreover,
Bead–cell interactions for lysing arise while the beads one can observe its transparent body with a microscope.
and cells are pushed back and forth (by switching the The main fabrication material of microfluidic plat-
CD rotational direction) through a continuously nar- form is polydimethylsiloxane (PDMS), which is highly
rowing chamber wall. This phenomenon is called the permeable to gases (a requirement for any aerobic
keystone effect. There are two interaction forces associ- culture), a chemically inert surface and optically trans-
Centrifuge-Based Fluidic Platforms 19.4 Conclusion 549
19.4 Conclusion
In comparing miniaturized centrifugal fluidic platforms to perform identical volume additions, to establish iden-
to other available microfluidic propulsion methods we tical incubation times, mixing dynamics, and detection
have demonstrated how CD-based centrifugal methods in a multitude of parallel CD assay elements makes
are advantageous in many analytical situations because the CD an attractive platform for multiple parallel as-
of their versatility in handling a wide variety of sam- says. The platform has been commercialized by Tecan
ple types, ability to gate the flow of liquids (valving), Boston for high-throughput screening (HTS) [19.4],
simple rotational motor requirements, ease and eco- by Gyros AB for sample-preparation techniques for
Part B 19.4
nomic fabrication methods, and the large range of flow MALDI [19.20] and by Abaxis (in a somewhat larger
rates attainable. Most analytical functions required for and less-integrated rotor format compared to the CD
a lab-on-a-disc, including metering, dilution, mixing, format) for human and veterinary diagnostic blood
calibration, separation, etc., have all been successfully analysis [19.5]. The Abaxis system for human and vet-
demonstrated in the laboratory. Moreover, the possibil- erinary medicine uses only dry reagents, but for many
ity of maintaining simultaneous and identical flow rates, diagnostic assays, requiring more fluidic steps, there
550 Part B MEMS/NEMS and BioMEMS/NEMS
are severe limitations in progressing toward the lab- The CD platform is easily adapted to optical
on-a-disc goal, as liquid storage on the disc becomes detection methods because it is manufactured with
necessary. In high-throughput screening (HTS) applica- high-optical-quality plastics, enabling absorption, fluo-
tions, the CD platform is being coupled to automated rescence, and microscopy techniques. Additionally, the
liquid-reagent loading systems and no liquids/reagents technology developed by the optical disc industry is be-
need to be stored on the disc. The latter makes the ing used to image the CD at micrometer resolution and
commercial introduction of the CD platform for HTS move to DVD and high-density (HD) DVD will allow
somewhat simpler [19.4,20,30]. There is an urgent need submicrometer resolution. The latter evolution will con-
though for the development of methods for long-term tinue to open up new applications for the CD-based fluid
reagent storage that incorporates both liquid and va- platform. Whereas today the CD fluidic platform may
por barriers to enable the introduction of lab-on-a-disc be considered a smart microcentrifuge, we believe that
platforms for a wide variety of fast diagnostic tests. in the future the integration of fluidics and informatics
One possible solution to this problem involves the use on DVDs and HD DVDs may lead to a merging of in-
of lyophilized reagents with common hydration reser- formatics and fluidics on the same disc. One can then
voir feeds, but the issue in this situation becomes the envision making very sharp images of the bacteria un-
speed of the test as the time required to redissolve the der test and correlate both test and images with library
lyophilized reagents is often substantial. data on the disc.
References
19.1 Alpha-MOS: http://www.alpha-mos.com (Alpha- 19.12 G.T.A. Kovacs: Micromachined Transducers Source-
MOS, Hillsborough 2008) Example for commercial book (Dordrecht/WCB/McGraw-Hill, Boston 1998)
electronic noses and tongues pp. 787–793
19.2 A. Manz, E. Verpoorte, C.S. Effenhauser, N. Burggraf, 19.13 G. Ekstrand, C. Holmquist, A. Edman Örlefors,
D.E. Raymond, D.J. Harrison, H.M. Widmer: Minia- B. Hellman, A. Larsson, P. Anderson: Microfluidics
turization of separation techniques using planar in a rotating CD. In: Micro Total Analysis Sys-
chip technology, HRC J. High Resolut. Chromatogr. tems 2000, ed. by A. van den Berg, W. Olthuis,
16, 433–436 (1993) P. Bergveld (Kluwer, Dordrecht 2000) pp. 311–314
19.3 Caliper Life Science: http://www.caliperLS.com 19.14 A.-L. Tiensuu, O. Öhman, L. Lundbladh, O. Larsson:
(Caliper Life Science, Hopkinton 2008) Hydrophobic valves by ink-jet printing on plastic
19.4 Tecan: Look for LabCD-ADMET System (Tecan, Boston CDs with integrated microfluidics. In: Micro Total
2008), http://www.tecan-us.com/us-index.htm Analysis Systems 2000, ed. by A. van den Berg,
19.5 Abaxis: http://www.abaxis.com (Abaxis, Union City W. Olthuis, P. Bergveld (Kluwer, Dordrecht 2000)
2008) pp. 575–578
19.6 M.J. Madou: Fundamentals of Microfabrication, 19.15 M.J. Madou, Y. Lu, S. Lai, J. Lee, S. Daunert: A cen-
2nd edn. (CRC, Boca Raton 2002) trifugal microfluidic platform – A comparison. In:
19.7 S. Miyazaki, T. Kawai, M. Araragi: A piezo-electric Micro Total Analysis Systems 2000, ed. by A. van
pump driven by a flexural progressive wave, Proc. den Berg, W. Olthuis, P. Bergveld (Kluwer, Dor-
IEEE Micro Electro Mech. Syst. (MEMS ’91) (Nara 1991) drecht 2000) pp. 565–570
pp. 283–288 19.16 J. Zeng, D. Banerjee, M. Deshpande, J.R. Gilbert,
19.8 J.W. Jorgenson, E.J. Guthrie: Liquid chromatogra- D.C. Duffy, G.J. Kellogg: Design analysis of capillary
phy in open-tubular columns, J. Chromatogr. 255, burst valves in centrifugal microfluidics. In: Micro
335–348 (1983) Total Analysis Systems 2000, ed. by A. van den Berg,
19.9 D.J. Harrison, Z. Fan, K. Fluri, K. Seiler: Integrated W. Olthuis, P. Bergveld (Kluwer, Dordrecht 2000)
electrophoresis systems for biochemical analyses, pp. 579–582
Solid State Sens. Actuator Workshop, Tech. Dig. 19.17 I.H.A. Badr, R.D. Johnson, M.J. Madou, L.G. Bach-
(Hilton Head Island 1994) pp. 21–24 as: Fluorescent ion-selective optode membranes
Part B 19
19.10 D.C. Duffy, H.L. Gills, J. Lin, N.F. Sheppard, G.J. Kel- incorporated onto a centrifugal microfluidics plat-
logg: Microfabicated centrifugal microfluidic sys- form, Anal. Chem. 74(21), 5569–5575 (2002)
tems: Characterization and multiple enzymatic 19.18 R.D. Johnson, I.H.A. Badr, G. Barrett, S. Lai, Y. Lu,
assays, Anal. Chem. 71(20), 4669–4678 (1999) M.J. Madou, L.G. Bachas: Development of a fully
19.11 M.J. Madou, G.J. Kellogg: A centrifuge-based mi- integrated analysis system for ions based on ion-
crofluidic platform for diagnostics, LabCD 3259, selective optodes and centrifugal microfluidics,
80–93 (1998) Anal. Chem. 73(16), 3940–3946 (2001)
Centrifuge-Based Fluidic Platforms References 551
for the detection of low-abundant DNA mutations, microchannel biochip, Anal. Chem. 73, 5777–5783
Anal. Chem. 75, 1130–1140 (2003) (2001)
19.32 V. Mikhailovich, S. Lapa, D. Gryadunov, A. Sobolev, 19.45 R. Lenigk, R.H. Liu, M. Athavale, Z. Chen, D. Ganser,
B. Strizhkov, N. Chernyh, O. Skotnikova, O. Ir- J. Yang, C. Rauch, Y. Liu, B. Chan, H. Yu, M. Ray,
tuganova, A. Moroz, V. Litvinov, M. Vladimirskii, R. Marrero, P. Grodzinski: Plastic biochannel hy-
M. Perelman, L. Chernousova, V. Erokhin, A. Zase- bridization devices: A new concept for microfluidic
datelev, A. Mirzabekov: Identification of rifampin- DNA arrays, Anal. Biochem. 311, 40–49 (2002)
552 Part B MEMS/NEMS and BioMEMS/NEMS
19.46 D. Di Carlo, K.-H. Jeong, L.P. Lee: Reagentless 19.52 J. Kim, H. Kido, J.V. Zoval, D. Gagné, R. Peytavi,
mechanical cell lysis by nanoscale barbs in mi- F.J. Picard, M. Bastien, M. Boissinot, M.G. Berg-
crochannels for sample preparation, Lab Chip 3, eron, M.J. Madou: Rapid and automated sample
287–291 (2003) preparation for nucleic acid extraction on a mi-
19.47 S.W. Lee, Y.-C. Tai: A micro cell lysis device, Sens. crofluidic CD, Int. Symp. LifeChips (2006), Poster
Actuators A 73, 74–79 (1999) Abstracts
19.48 J. Sambrook, D.W. Russell: Molecular Cloning (CSHL, 19.53 N. Kim, C.M. Dempsey, J.V. Zoval, J. Sze,
Cold Spring Harbor 2001) M.J. Madou: Automated microfluidic compact disc
19.49 J. Kim, S.H. Jang, G. Jia, J.V. Zoval, N.A. Da Silva, (CD) cultivation system of C. elegans, Sens. Actua-
M.J. Madou: Cell lysis on a microfluidic CD (compact tors B 122(2), 511–518 (2007)
disc), Lab Chip 4, 516–522 (2004) 19.54 E. Le Bourg: A review of the effects of microgravity
19.50 K.K.J. Ruschak, L.E. Scriven: Rimming flow of liquid and of hypergravity on aging and longevity, Exp.
in a rotating horizontal cylinder, J. Fluid Mech. 76, Gerontol. 34, 319–336 (1999)
113–125 (1976) 19.55 H. Wu, T.W. Odom, D.T. Chiu, G.M. Whitesides:
19.51 S.T. Thoroddsen, L. Mahadevan: Experimental Fabrication of complex three-dimensional mi-
study of coating flows in a partially-filled horizon- crochannel systems in PDMS, J. Am. Chem. Soc. 125,
tally rotating cylinder, Exp. Fluids 23, 1–13 (1997) 554–559 (2003)
Part B 19
553
Micro-/Nanod
20. Micro-/Nanodroplets in Microfluidic Devices
Part B 20
Yung-Chieh Tan, Shia-Yen Teh, Abraham P. Lee
Droplet microfluidics, also termed digital microfluidics, terface, lipid self-assembly at the oil–water interface,
has been the focus of much interest due to its po- and ion exchange at pH-sensitive interfaces, all add to
tential for manipulating small quantities of reagent the benefits of droplet-based systems. These advantages
volumes and controlling complex reaction processes. have transformed the application of droplet technology,
Conventional microfluidics, or analog microfluidics, is making it useful not just for making bulk quantities
governed by diffusion and low-Reynolds-number lami- of particles, but also as an integrated multidisciplinary
nar flow that prevent rapid mixing of miscible liquids. tool that can be applied to fields ranging from material
When a liquid is dispersed into another immiscible li- synthesis to molecular biology [20.1].
quid as droplets, mixing is on the millisecond scale The versatility of droplet microfluidic systems lies
due to the decreased striation area between miscible essentially in the ability to selectively dial-in the fluidic
fluid interfaces. The ability to split and fuse individual volume and concentration of a single droplet. This can
droplets further simplifies the control of reagent vol- be achieved either actively or passively. Active devices
ume and concentrations. Multiphase interactions, such use dynamic modulation of forces around the droplet
as polymerization at the monomer–initiator fluid in- to achieve reconfigurable droplet flows. Passive devices
554 Part B MEMS/NEMS and BioMEMS/NEMS
use passive gradients to manipulate droplets through mechanisms for the actuation and control of droplets
Part B 20.1
a set of operations governed by channel geometries and will be introduced, and important parameters and design
predefined wetting properties. In the next two sections, criteria will be reviewed.
a) b) Ground electrode
Top plate
Wire electrode Glass substrate
Fig. 20.1 (a) The conductive path of an EWOD device. (b) A schematic of an EWOD device (after [20.2])
Micro-/Nanodroplets in Microfluidic Devices 20.1 Active or Programmable Droplet Systems 555
have suggested that this change in contact angle is re- threshold voltage, the droplet transport speed increases
Part B 20.1
quired for droplet transport, it has been suggested by with the driving voltage. A transport speed of 250 mm/s
Zeng and Korsmeyer [20.26] that the change in contact with an alternating-current (AC) voltage of > 150 V has
angle merely reflects the difference in interfacial energy been reported by Cho et al. [20.18], and the transport of
and is not required for droplet transport. droplet volume as small as 5 nl has been reported by Lee
For droplet-based EWOD, the device consists of et al. [20.11].
a top ground electrode layered on glass and a bot- Similarly, more complex droplet manipulations can
tom layer of control electrodes underneath an insulating be achieved through the simultaneous operation of mul-
layer of dielectric material. A thin hydrophobic layer is tiple droplet control sequences. Droplets can be fused
coated on the surface of the electrodes and the insulat- by transporting two droplets toward the same elec-
ing material. The hydrophobic coating acts to prevent trode, or a single droplet can be split into smaller
droplets from spreading into the channel and is not droplets by simultaneously transporting the two halves
considered to be insulative. A typical EWOD setup is of a droplet in different directions. Droplet operations
shown in Fig. 20.1 [20.2]. can also be combined to form sequential fuse and split
During operation, the activation of one electrode operations, as shown in Fig. 20.3. The efficiency of the
induces a local interfacial energy difference between droplet splitting process, due to the energy required
adjacent electrodes. When the droplet experiences the to overcome capillary pressure, is dependent on the
energy difference, it moves toward surface of lower geometry of the electrodes. Cho et al. [20.18] have
energy, and so through sequential activation and de- shown that smaller channel gaps and larger electrode
activation of electrodes the droplet can be transported sizes favor droplet fission. The geometric constraint
as shown in Fig. 20.2 [20.2]. Due to the localized ac- for a square electrode is d/R2 < 0.22, where d is
tivated interfacial energy difference, Cho et al. [20.18]
have shown that droplet volume large enough to cover
1.4 mm
the edge between two electrode surfaces is required to a) Time
actuate the droplet. In addition, Pollack et al. [20.13] re-
ported that a threshold voltage is required to actuate the
droplet, and it was found that the threshold voltage for
a water droplet dispersed in silicon oil is much lower
b)
than for a water droplet in air. It was also shown by Off
Moon et al. [20.7] that the threshold voltage decreases On On
with the thickness of the dielectric layer. Using barium
strontium titanate (BST) as the dielectric material, an
actuation voltage as low as 15 V can be used to transport c)
droplets [20.6]. Once the applied voltage exceeds the
a) b) d)
On Off On
e)
c) d) On Off On
the gap size and R2 is roughly half the electrode First, the surface contact required for droplet actuation
Part B 20.1
width [20.18]. limits its use for complex biological fluids; absorbance
of biomolecules onto the electrodes through electro-
20.1.3 Reagent Mixing in EWOD static interactions and passive hydrophobic absorption
eventually renders the electrode unusable after repeated
Reagent mixing in droplets is efficient in EWOD; Paik use [20.12]. Second, while it is simple to control the
et al. demonstrated that complete mixing of two 800 nl actuation of a small number of electrodes, complex log-
droplets can be achieved as rapidly as 1.7 s [20.27]. It arithms would be required to control large arrays of
was shown that the droplet mixing rate is dependent electrodes [20.15]. Third, only aqueous droplets can
on the subsequent motion of the coalesced droplet. In be actuated, due to the need for conductivity between
a linear array such as that shown in Fig. 20.4, move- the droplet and the ground electrode. Lastly, a droplet
ment of the droplet in one direction promotes mixing, volume larger than the size of the electrode is re-
while movement in the reverse direction undoes the quired for EWOD to work, which limits the fluidic
mixing, due to flow reversibility at low Reynolds num- volume that can be used. However, improvements in
ber [20.14]. The mixing rate can be further improved EWOD have addressed the first two limitations. Yoon
by increasing the rate of oscillation of the droplet be- et al. [20.12] reported that reducing the duration of
tween electrodes, increasing the number of electrodes the square-wave electric field applied improved the
for a larger transport area, and increasing the com- durability of the electrode, and Chiou et al. [20.15]
plexity of movement of the droplet through the use of demonstrated that the addition of a photoconductive
multidimensional arrays. When the aspect ratio, defined layer permits the use of light signals to control up to
by Paik et al. [20.27] as the ratio of the gap size to the 20 000 electrodes.
width of the electrodes, is 0.4, the mixing efficiency is
optimized; lower aspect ratios inhibit vertical flow and 20.1.5 Droplet Manipulation
result in longer mixing times [20.27]. via Dielectrophoresis
FDEP = 32 V εS f CM ∇ E 2 , (20.3)
t = 0.769 s t = 0.903 s
where ε d is the dielectric constant of the droplet [20.28].
When ε d > εs , positive dielectrophoresis causes the
Fig. 20.4 Time-lapse images of two-electrode mixing. The droplet to move toward the region of high field strength.
pattern created by the dyes show that the droplet can be In contrast, when ε d < εs , the droplet is displaced away
transported in the opposite direction. The arrows indicate from the high field. Since DEP is conducted through
the direction of droplet movement bulk liquid, no physical contact between the droplet
Micro-/Nanodroplets in Microfluidic Devices 20.2 Passive Droplet Control Techniques 557
Part B 20.2
a) b)
Droplet Droplet
Top view
for sequential droplet operations in one integrated de- width of the channel, the effect of the wall becomes
Part B 20.2
vice. In addition, the device does not have any moving dominant over the effect of shear stress [20.36], and
components, which makes its fabrication simple and hence the size is weakly dependent on the flow rate,
more reliable. as reported by Tan et al. [20.32] and Tice et al. [20.36,
37]. This is especially true for Ca < 1, where Ca can
20.2.1 Generation also be defined by Ca = ηU/σ , in which U = εr, and
of Monodispersed Droplets the observed droplets become confined to elongated
plugs [20.37, 38].
The key features of passive droplet-generation sys- Where Ca is less useful for predicting droplet size,
tems are that the generated droplet sizes can be much the conservation of mass flow can be used to determine
smaller than the features of the device, the generated droplet volume. Tice et al. demonstrated that the length
droplets have narrow size distributions, and the gener- of the plug can be predicted from l = p[V d /(V d + Vc )],
ation of droplet is a continuous process. Droplet size where l is the length of the plug, p is the period of break-
distributions of < 2% coefficient of variation have been up, V d is the volume flow of the dispersed phase, and
reported [20.30], and droplet sizes as small as 100 nm Vc is the volume of the continuous phase [20.37]. In
have been demonstrated [20.30, 31]. Most passive mi- addition, Tice et al. [20.37] observed different regimes
crofluidic droplet-generating devices utilize either the of droplet formation. Stable droplet formation occurs
shear stress created at the immiscible flow interface or within a limited regime. This change in behavior for dif-
the pressure gradient created at a junction of narrow ferent regimes was also observed by Anna et al. [20.31]
pores to initiate the continuous break-up of droplets. and Dreyfus et al. [20.39]. The flow regime behavior ob-
The shear break-up device uses either an asymmetric or served by Dreyfus et al. indicates that, when the flow
a symmetric channel junction to introduce the immis- rate of the dispersed phase is much lower than the flow
cible fluids, and the pressure gradient break-up device rate of the continuous phase, isolated drop formation
uses either straight-through holes or microcapillary is observed, but as the flow rate of the dispersed phase
channels with flat terraces to pressurize the dispersed increases the stratified regime is observed.
phase. In the microfluidic flow, the Reynolds number Re
(inertial force/viscous force) is much less than 1, so vis- Asymmetric and Symmetric Shearing Design
cous forces dominate over inertial forces. The viscous The asymmetric shearing of immiscible fluids is
force is also weaker in magnitude when compared with achieved using a T-type intersection such as that shown
the surface tension force, and thus, regardless of the in Fig. 20.7 [20.29]. The dispersed phase is sheared at
channel geometries, in order to achieve steady gener- the junction by the continuous phase, and the droplet
ation of droplets, it is critical that the dispersed phase size and frequency of generation are controlled by the
does not wet the droplet-generating surface [20.32, 33]. flow rates of the continuous and the dispersed phases.
Due to the asymmetric nature of the droplet break-up
Shear-Induced Droplet Generation process, the reagents injected for mixing are partially
The generation of droplets in a microfluidic device is exchanged when the liquid droplets break off. This
governed by the interaction of shear stress with sur- effect is known as twirling and will be discussed in
face tension. The shear stress exerted by the continuous more detail later [20.37]. Twirling can be an advan-
phase acts to deform the liquid surface, while the in-
terfacial force at the immiscible fluid interface acts to
restore the deformation. This can be described by the
dimensionless capillary number Ca = ηεr/σ , where r
is the radius of the droplet, η is the viscosity of the
continuous phase, ε is the shear rate in the channel,
and σ is the interfacial tension at the immiscible fluid
interface [20.34]. Assuming that Ca = 1 is the critical
condition when the shear stress is large enough to break
up the liquid thread, r ∼ σ/ηε, which is the scaling of
droplet sizes observed by Thorsen et al. [20.35], cor-
roborating the observations of Nisisako et al. [20.29]. Fig. 20.7 T-junction device from Nisisako et al. [20.29],
c
As the diameter of the droplet increases to beyond the R. Soc. Chem.
Micro-/Nanodroplets in Microfluidic Devices 20.2 Passive Droplet Control Techniques 559
Part B 20.2
Thread growth Thread growth
the reagents need to be aligned to create, e.g., bicolor
polymeric beads. Furthermore, the simplicity of the de-
sign allows two or more generator units to be aligned
adjacent or opposite to each other in the same channel
for synchronized generation of droplets. This geometry
has been exploited in order to index mixing condi-
tions [20.30, 40]. 65 μm
Droplet generation from symmetric fluidic junc- 0.5 ms/frame
tions results in monodispersed droplets at controlled
frequency [20.31, 32, 41]. Figure 20.8 shows an ex-
ample of the droplet-generation process [20.32]. As
the flow passes through the junction, the narrow width
creates a maximum velocity and, after the flow has
passed the junction, the fluid velocity decreases due
to the increase in channel width. This creates a shear 65 μm
Qo = 1 μl/min Qo = 6 μl/min Qo = 10 μl/min
gradient that is maximized at the orifice, producing
shear-focusing break-up. Droplets are generated from Fig. 20.8 The shear-focusing design demonstrated by Tan et al.
the shear-focusing break-up precisely at the orifice, and [20.32],
c R. Soc. Chem.
the droplet size and generation frequency are controlled
by the relative flow rates. the edge of the liquid deforms, causing the surface ten-
sion to pinch off liquid near the edge of the terrace.
20.2.2 Devices Based When designing the channel terrace, Sugiura
on Microcapillary Arrays et al. [20.43] reported that the diameters of the droplets
generated from MC can be predicted using the terrace
Straight silicon through-holes and microcapillary ar- length L and depth H, according to the geometric pa-
rays (MC) are attractive designs for forming emulsions. rameters presented in Fig. 20.9 by
Both of these methods can be applied under conditions
6(H + 0.626)
of no external flow to produce narrow size distribu- D=
tions. Variations of < 2% have been reported for straight π
2
silicon through-holes [20.42] and ≈ 5% for MC de- L −1 L − 13.76H − 8.61
× cos
signs [20.43]. 4 L
In these devices, the production rate is rather low; L(L − 13.76H − 8.61)
Kobayashi et al. [20.41] reported a production rate of −
4
3–11 drops/s for a device based on silicon through- 1/3
−1 L − 13.76H − 8.61
holes. Furthermore, the size of the generated droplet is × sin cos .
limited by the size of the pores [20.44]. L
(20.5)
In the silicon through-hole method, the dispersed
phase is pushed through small holes etched in a silicon The effect of external flow can also reduce droplet size,
wafer. The droplet detaches when the interfacial tension as reported by Kawakatsu et al. [20.44]. The droplet size
pinches off the dispersed phase. The generated droplets decreases from 47.6–37.3 μm as the flow is increased
are monodispersed in terms of droplet size [20.42]. from 1.4 × 10−2 to 2.4 ml/min. However, for 16 and
In microcapillary array designs, a glass plate is 20 μm droplets generated with a smaller MC channel,
bonded to an etched silicon channel. The channel con- varying the flow has no effect on droplet size.
nects to a flat terrace that leads to an indented well. The
droplet-generation process is characterized by a two- 20.2.3 Double Emulsions
step process of filling the terrace with the dispersed
phase and the detachment of the liquid as it deforms Double emulsions are formed when a liquid is dis-
at the terrace. Upon increasing the pressure of the dis- persed in an immiscible fluid that is further dispersed
persed phase, the dispersed liquid is entrained on the in another immiscible fluid. Since the generated double
terrace surface. When liquid reaches the indented well, emulsion contains both organic and aqueous phases, it is
560 Part B MEMS/NEMS and BioMEMS/NEMS
tion of both the inner phase and the middle phase has
Part B 20.2
Channel Detaching
dispersed phase
been demonstrated by Okushima et al. [20.46] using se-
Terrace D
A quential break-up through two T-junctions, as shown in
Fig. 20.10. Utada et al. [20.47] also demonstrated pre-
Dispersed
cise generation of double emulsions using coaxial flow
phase φ to simultaneously break up the immiscible fluid inter-
faces between the inner and outer droplet, and the outer
(A) droplet and continuous phase, as shown in Fig. 20.11.
Both methods provide the ability to control the size of
Glass L L
plate the inner droplet, the thickness of the middle or shell
A layer, and the number of encapsulated inner droplets.
H H Ratios of shell thickness to outer drop radius as low as
Silicon plate 3% and as high as 40% have been reported [20.47].
Fig. 20.9 Model for predicting droplet diameter (af- Reagent Mixing
ter [20.43]) There are two ways to mix reagents in a passive mi-
crofluidic system. In the first technique, the reagents to
a versatile way of encapsulating and delivering polar or be mixed are introduced as adjacent laminar streams
nonpolar substances. Kawakatsu et al. [20.45] was able that are broken into single droplets using either an
to generate monodispersed double emulsions by using asymmetric or a symmetric shearing system. The
MC devices where the core droplets were generated by reagents are then mixed by diffusion or convection in-
the homogenization of water in oil. Controlled forma- duced either by the surrounding flow or the walls of the
channel. The other mixing mechanism is based on fus-
ing two droplets in the microfluidic channel, as detailed
a) b) in the next section.
Q0
100 μm Mixing in moving plugs is facilitated by recircu-
lation flow, which distributes reagents from the center
The first
Qiw to the edge of the droplet [20.48]. When the reagent
junction
60 μm ×25 μm gradient is perpendicular to the direction of transport,
recirculation is not as effective at accelerating the mix-
The second ing, as illustrated in Fig. 20.12 [20.37]. As shown in
junction Fig. 20.13, for droplets generated by a symmetric shear-
130 μm ×65 μm ing system, the reagent gradient in the laminar stream
is directly transferred into the droplet [20.49]. However,
Qaw Output for droplets generated from an asymmetric shearing sys-
tem, mixing is facilitated by an effect called twirling,
which is an eddy that transports reagents to different
c) parts of the droplet. The effect of twirling is finite and
so it increases the mixing rate for short plugs, but does
not significantly increase the mixing rate for long plugs,
as shown in Fig. 20.14 [20.37].
Mixing can be further improved through the use of
d) winding channels. As the droplet passes through these
winding channels, it is stretched, folded, and reoriented
to induce chaotic mixing inside the droplets [20.50, 51].
The time of mixing is verified experimentally to be
100 μm
aw
tmix,ca ∼ log(Pe) , (20.6)
Fig. 20.10 Generation of compound drops through a se- U
quential droplet break-up process (after [20.46],
c Am. wU
Pe = , (20.7)
Soc. Chem.) D
Micro-/Nanodroplets in Microfluidic Devices 20.2 Passive Droplet Control Techniques 561
Part B 20.2
a) Outer fluid Middle fluid through coaxial flow in a microfluidic device.
Parameters such as shell thickness, the internal
droplet number, and the sizes of the inter-
Inner fluid
nal droplets could be individually controlled.
(a) Schematic of the glass microcapillary de-
vice. (b–e) The diameter of the inner and outer
microcapillary tubes were varied from 10 to
Collection tube Injection tube 50 μm and 50 to 500 μm, respectively. This al-
b) c) d) e) f) g) lowed control over the thickness of the middle
fluid phase. (f,g) Multiple droplets of con-
trolled size can be contained within a single
droplet. (h) A large number of double emul-
60 μm 20 μm 15 μm 15 μm 125 μm 125 μm
sions containing one droplet can be generated
h) (after [20.47],
c AAAS)
50 μm
Water
Water Water
Qo = 1.2
0.14 Qw = 0.3
100 µm
Out 40 µm
PFD
0.2
Qo = 3.3
Qw = 0.3
0.3
40 µm
0.4
Water fraction
0.6 Qo = 1.8
Qw = 0.3
0.73
40 µm
a) b)
Reagent A
c) Droplet fusion
Fig. 20.16a–c Fusion mixing via a flow-rectifying design, immediately prior to fusion (a) and after fusion (b) [20.30]
and fusion within a channel expansion design (c) (after [20.53])
Micro-/Nanodroplets in Microfluidic Devices 20.2 Passive Droplet Control Techniques 563
Part B 20.2
a) w0 b) c) d) e) we
illary number as 0
2
1 e
Ccr = αε0 2/3 − 1 , (20.8)
f) g) h) i) j)
ε0
0
ε0 = , (20.9)
πw0
which shows good agreement with experimental results, k) Capillary number (ην/γ)
as shown in Fig. 20.17 [20.54]; ε0 is the initial extension Breaking
0.3
ratio of length 0 to the circumference (πw0 ), and α
is the fitting parameter, which is equal to 1 for square
channels [20.54]. 0.2
Nonbreaking
Droplets of various sizes can be created through
asymmetric break-up, such that even submicrometer- 0.1
sized droplets can be split from large droplets tens
of microns in radius [20.30]. It was shown that the 0
sizes of the split droplets depend on the size of the 0 0.5 1 1.5
Extension ( 0 /πw0)
original droplet [20.30] and the bifurcating flow dis-
tribution at the junction [20.30, 50, 54]. Using droplet Fig. 20.17a–k The parameters used to predict droplet break-up con-
break-off from unmixed reagents, by controlling the lo- ditions are shown in (a). Panels (a–j) show the time-lapse images of
cation and time of fission, the reagent concentration the break-up process. As shown in panels (e) and (j), the droplet
inside the droplet could be redistributed according to reaches maximum extension with width we and length le , obtain-
the mixed gradient at the time of break-up to produce ing values for ε of 0.95 and 1.15 respectively. (k) The predicted
arrays of split droplets with different reagent concentra- capillary number agrees with the experimental results (after [20.54])
tions [20.30].
in Fig. 20.18, a droplet at the junction will be trans-
Droplet Sorting ported toward a region of higher shear stress. Since the
In an active droplet control system, the complex shear force experienced by the droplet depends on the
droplet transport process is modulated by algorithms surface area of the droplet, passive sorting by droplet
that control the electrode switches. In a passive sys- size can also be achieved. To collect the individual
tem, transport is guided by the flow distribution in the satellite droplets efficiently, Tan et al. developed a dy-
channel, and controlled multipath transport is difficult namic flow technique that modulates the stress exerted
to achieve because individual droplets are simply dis- on the liquid thread to control the location of break-up
tributed according to the flow rates. The ability to switch
droplets between different continuous phases is use-
ful because it makes it possible to filter contaminants Q1
Complete separation
from the droplet stream, it allows the concentration of
reagents present in the continuous phase to be changed, Q1 = Q2
it means that we can organize unknown particulates by Smaller drop
size, and it allows us to set up a passive monitoring
system for variations in droplet size.
During the droplet-generation process, the contin-
PDMS
uous break-up of the neck connecting the droplet to
the liquid thread leads to the formation of small satel-
lite droplets. The presence of satellite droplets increases 40 µm Q2
the size distribution and decreases the mixing accu-
racy due to the fusion of satellite droplets with primary Fig. 20.18 When the flow rates exiting the bifurcating
droplets. Tan et al. [20.30, 55] demonstrated droplet junctions are balanced, droplets are transported toward re-
sorting in a microfluidic channel by controlling the gions of higher shear stress created by the narrow inlet
shear stress gradient at bifurcating junctions. As shown channel (after [20.30],
c R. Soc. Chem.)
564 Part B MEMS/NEMS and BioMEMS/NEMS
and the distribution of different satellite droplets in the and smaller droplets are separated into separate cham-
channel, as shown in Fig. 20.19, where larger droplets bers [20.55].
20.3 Applications
Droplet-based microfluidic systems possess tremendous droplets results in the formation of photonic balls, as
potential to improve current emulsion technologies that shown in Fig. 20.20 [20.56], and other colloid struc-
are widely used in industry to produce sol-gels, drugs, tures that show unique responses to flow and magnetic
synthetic materials, and food products. In addition, fields [20.56, 61, 62].
a wide variety of new applications can be developed By transferring laminar patterns of two reagents to
due to their precise metering capabilities, rapid and con- droplets using a symmetric shearing device, Nisisako
trollable mixing response, and automated combinatorial
capabilities.
et al. [20.49] demonstrated the production of polymeric dependent concentration gradient for the reagents inside
Part B 20.3
beads with bichromal and oriented charge polarities. the droplets. These properties can be used to screen
Similarly, as demonstrated by Millman et al. [20.63] protein crystallization conditions, as demonstrated by
using dielectrophoretic-based digital fusion of different Zheng et al. [20.77]. Volumes of < 4 nl of reagent
polymeric droplets, anisotropic particles with tailored could be used for each trial inside a 7.5 nl aqueous
properties can be synthesized. droplet, and hundreds of trials performed at rates of sev-
For biological applications, single and multiple cells eral trials per second can be achieved with computer
and organelles can be trapped inside droplets [20.66,67] control [20.77]. Subsequent crystal growth by va-
for analysis or to provide scaffolds for cell [20.52] and por diffusion from droplets generated x-ray-analyzable
tissue growth. Biocompatible and biodegradable hy- crystals, as shown in Fig. 20.22 [20.65]. To mitigate
drogels such as alginate are commonly used for cell the complexity of resolving the time-dependent screen-
encapsulation due to their relatively simple prepara- ing concentrations, an indexing stream of fluorescent
tion. Since gelation occurs immediately upon contact droplets generated at the same time as the screening
between alginate and polycations, droplet fusion tech- droplets can be used to indicate the concentration of the
niques have been implemented to encapsulate cells and protein [20.77].
polymerize particles simultaneously on a chip [20.68, Recent work has been done with droplet-based plat-
69]. Control over particle size and monodispersity is forms for biomolecule synthesis. Since droplets can
important for the use of particles in the administration be generated at micron sizes or smaller, encapsula-
and controlled release of encapsulated substances such tion of a single template copy of DNA is achievable.
as drugs, dyes, and enzymes [20.70]. Current microflu- The integration of heating elements, the ability to pre-
idic platforms allow droplets to be filled with various cisely manipulate droplet movement, and increased
hydrophilic or hydrophobic materials and the capsule mixing rates enable droplets to serve as microreac-
shell thickness to be altered to control compound re- tors for in vitro protein expression, DNA amplification,
lease rates [20.71, 72]. and other biochemical reactions. Dittrich et al. demon-
strated cell-free expression of green fluorescent protein
20.3.2 Droplets (GFP) within monodispersed water-in-oil droplets gen-
as Real-Time Chemical Processors erated in a microfluidic platform [20.79]. PCR [20.80]
and Combinatorial Synthesizers and DNA assays [20.81] in droplets recently showed
improved sensitivity and decreased reaction times, en- structures can be made into permanent microscale
Part B 20
abling higher-throughput assays. building blocks for device assembly if they are poly-
merized [20.82]. Microcomponents such as colloidal
20.3.3 Droplets microspheres can be made into pumps and valves in
as Micromechanical Components microfluidic channels [20.83]. Droplets driven by elec-
trowetting have also been used as the main driving
Since droplets can be deformed by the flow, they components for liquid micromotors [20.10] and to reg-
can be fixed into a variety of shapes and sizes that ulate the frequencies of optical-fiber devices [20.9, 83,
could be used as microcomponents; furthermore, these 84].
20.4 Conclusions
The field of droplet-based microfluidic technology is di- to be processed individually, provide accurate dispens-
verse in terms of the actuation mechanisms and methods ing of fluid volume, and improve speed of reagent
of operation used. Essentially, all droplet-based systems mixing. These factors have made droplet technology
are used to control liquid dispensing, mixing, splitting, a valuable new tool for controlling micro- and nanoint-
and localization. Microfluidic methods allow droplets eractions.
References
20.1 S.-Y. Teh, R. Lin, L.-H. Hung, A.P. Lee: Droplet 20.10 J. Lee, C. Kim: Surface-tension-driven microactu-
microfluidics, Lab Chip 8(2), 198–220 (2008) ation based on continuous electrowetting, J. Mi-
20.2 M.G. Pollack, A.D. Shenderov, R.B. Fair: Electro- croelectromech. Syst. 9(2), 171–180 (2000)
wetting-based actuation of droplets for integrated 20.11 J. Lee, H. Moon, J. Fowler, T. Schoellhammer,
microfluidics, Lab Chip 2, 96–101 (2002) C. Kim: Electrowetting and electrowetting-on-
20.3 R.H. Farahi, A. Passian, T.L. Ferrell, T. Thundat: dielectric for microscale liquid handling, Sens.
Microfluidic manipulation via Marangoni forces, Actuators A 95, 259–268 (2002)
Appl. Phys. Lett. 85(18), 4237–4239 (2004) 20.12 J. Yoon, R.L. Garrell: Preventing biomolecular ad-
20.4 B.S. Gallardo, V.K. Gupta, F.D. Eagerton, L.I. Jong, sorption in electrowetting-based biofluidic chips,
V.S. Craig, R.R. Shah, N.L. Abbott: Electrochemical Anal. Chem. 75(19), 5097–5102 (2003)
principles for active control of liquids on submil- 20.13 M.G. Pollack, R.B. Fair: Electrowetting-based
limeter scales, Science 283, 57–60 (1999) actuation of liquid droplets for microfluidic ap-
20.5 K. Hosokawa, T. Fujii, I. Endo: Handling of picoliter plications, Appl. Phys. Lett. 77(11), 1725–1726 (2000)
liquid samples in a poly(dimethysiloxane)-based 20.14 P. Paik, V.K. Pamula, M.G. Pollack, R.B. Fair:
microfluidic device, Anal. Chem. 71, 4781–4785 Electrowetting-based droplet mixers for microflu-
(1999) idic systems, Lab Chip 3, 28–33 (2003)
20.6 C. Quillet, B. Berge: Electrowetting: A recent out- 20.15 P.Y. Chiou, H. Moon, H. Toshiyoshi, C. Kim, M.C. Wu:
break, Curr. Opin. Colloid Sci. 6, 34–39 (2001) Light actuation of liquid by optoelectrowetting,
20.7 H. Moon, S.K. Cho, R.L. Garrell, C. Kim: Low voltage Sens. Actuators A 104, 222–228 (2003)
electrowetting-on-dielectric, J. Appl. Phys. 92(7), 20.16 R.A. Hayes, B.J. Freenstra: Video-speed electronic
4080–4087 (2002) paper based on electrowetting, Nature 425, 383–
20.8 J. Ding, K. Chakrabarty, R.B. Fair: Scheduling of 385 (2003)
microfluidic operations for reconfigurable two- 20.17 S.K. Cho, H. Moon, J. Fowler, C. Kim: Splitting
dimensional electrowetting arrays, IEEE Trans. a liquid droplet for electrowetting-based microflu-
Comput.-Aided Des. Integr. Circuits Syst. 20(12), idics, ASME Int. Mech. Eng. Congr. Expo. (ASME
1463–1468 (2001) International, New York 2001)
20.9 J. Hsieh, P. Mach, F. Cattaneo, S. Yang, T. Krupen- 20.18 S.K. Cho, H. Moon, C. Kim: Creating, transport-
kine, K. Baldwin, J.A. Rogers: Tunable microfluidic ing, cutting, and merging liquid droplets by
optical-fiber devices based on electrowetting electrowetting-based actuation for digital mi-
pumps and plastic microchannels, IEEE Photon. crofluidic circuits, J. Microelectromech. Syst. 12(1),
Technol. Lett. 15(1), 81–83 (2003) 70–80 (2003)
Micro-/Nanodroplets in Microfluidic Devices References 567
20.19 T.B. Jones, J.D. Fowler, Y.S. Chang, C. Kim: 20.34 B.J. Briscoe, C.J. Lawrence, W.G.P. Mietus: A review
Part B 20
Frequency-based relationship of electrowetting of immiscible fluidmixing, Adv. Colloid Interface
and dielectrophoretic liquid microactuation, Lang- Sci. 81(1), 1–17 (1999)
muir 19, 7646–7651 (2003) 20.35 T. Thorsen, R.W. Roberts, F.H. Arnold, S.R. Quake:
20.20 A.A. Darhuber, J.M. Davis, S.M. Troian, W.W. Reis- Dynamic pattern formation in a vesicle-generating
ner: Thermocapillary actuation of liquid flow on microfluidic device, Phys. Rev. Lett. 86(18), 4163–
chemically patterned surfaces, Phys. Fluids 15(5), 4166 (2001)
1295–1304 (2003) 20.36 V. Cristini, Y.C. Tan: Theory and numerical sim-
20.21 A.A. Darhuber, J.P. Valentino, S.M. Troian, S. Wag- ulation of droplet dynamics in complex flows –
ner: Thermocapillary actuation of droplets on A review, Lab Chip 4(4), 257–264 (2004)
chemically patterned surfaces by programmable 20.37 J.D. Tice, H. Song, A.D. Lyon, R.F. Ismagilov: For-
microheater arrays, J. Microelectromech. Syst. mation of droplets and mixing in multiphase
12(6), 873–879 (2003) microfluidics at low values of the Reynolds and
20.22 A.A. Darhuber, J.P. Valentino, J.M. Davis, the capillary numbers, Langmuir 19(22), 9127–9133
S.M. Troian: Microfluidic actuation by modulation (2003)
of surface stresses, Appl. Phys. Lett. 82(4), 657–659 20.38 J.D. Tice, A.D. Lyon, R.F. Ismagilov: Effects of viscos-
(2003) ity on droplet formation and mixing in microfluidic
20.23 A.A. Darhuber, J.Z. Chen, J.M. Davis, S.M. Troian: channels, Anal. Chim. Acta 507, 73–77 (2003)
A study of mixing in thermocapillary flows on mi- 20.39 R. Dreyfus, P. Tabeling, H. Willaime: Ordered and
cropatterned surfaces, Philos. Trans. R. Soc. Lond. disordered patterns in two-phase flows in mi-
Ser. A 362, 1037–1058 (2003) crochannels, Phys. Rev. Lett. 90(14), 144505 (2003)
20.24 A.A. Darhuber, S.M. Trojan: Dynamics of capillary 20.40 B. Zheng, J.D. Tice, R.F. Ismagilov: Formation of
spreading along hydrophilic microstripes, Phys. droplets of alternating composition in microfluidic
Rev. E. 64, 031603 (2001) channels and applications to indexing of concen-
20.25 M.G. Lippmann: Relations entre les phénomènes tration in droplet-based assays, Anal. Chem. 76(17),
électriques et capillaires, Anal. Chim. Phys. 5(11), 4977–4982 (2004)
494–549 (1875), in French 20.41 Q. Xu, M. Nakajima: The generation of highly
20.26 J. Zeng, T. Korsmeyer: Principles of droplet electro- monodisperse droplets through the breakup of
hydrodynamics for lab-on-a-chip, Lab Chip 4(4), hydrodynamically focused microthread in a mi-
265–277 (2004) crofluidic device, Appl. Phys. Lett. 85(17), 3726–3728
20.27 P. Paik, V.K. Pamula, R.B. Fair: Rapid droplet mix- (2004)
ers for digital microfluidic systems, Lab Chip 3, 20.42 I. Kobayashi, M. Nakajima, K. Chun, Y. Kikuchi,
253–259 (2003) H. Fujita: Silicon array of elongated through-holes
20.28 P.R.C. Gascoyne, J.V. Vykoukal, J.A. Schwartz, for monodisperse emulsion droplets, AIChE Journal
T.J. Anderson, D.M. Vykoukal, K.W. Current, C. Mc- 48(8), 1639–1644 (2002)
Conaghy, F.F. Becker, C. Andrews: Dielectropho- 20.43 S. Sugiura, M. Nakajima, M. Seki: Prediction of
resis-based programmable fluidic processors, Lab droplet diameter for microchannel emulsification,
Chip 4(4), 299–309 (2004) Langmuir 18, 3854–3859 (2002)
20.29 T. Nisisako, T. Tori, T. Higuchi: Droplet formation in 20.44 T. Kawakatsu, H. Komori, M. Nakajima, Y. Kikuchi,
a microchannel network, Lab Chip 2, 24–26 (2002) T. Yonemoto: Production of monodispersed oil-in-
20.30 Y.C. Tan, J.S. Fisher, A.I. Lee, V. Cristini, A.P. Lee: water emulsion using crossflow-type silicon mi-
Design of microfluidic channel geometries for crochannel plate, J. Chem. Eng. Jpn. 32(2), 241–244
the control of droplet volume, chemical con- (1999)
centration, and sorting, Lab Chip 4(4), 292–298 20.45 T. Kawakatsu, G. Trägårdh, C. Trägårdh: Production
(2004) of W/O/W emulsions and S/O/W pectin microcap-
20.31 S.L. Anna, N. Bontoux, H.A. Stone: Formation of sules by microchannel emulsification, Colloids Surf.
dispersions using flow focusing in microchannels, 189, 257–264 (2001)
Appl. Phys. Lett. 82(3), 364–366 (2003) 20.46 S. Okushima, T. Nisisako, T. Torii, T. Higuchi:
20.32 Y.C. Tan, V. Cristini, A.P. Lee: Monodispersed mi- Controlled production of monodisperse double
crofluidic droplet generation by shear focusing emulsions by two-step droplet breakup in mi-
microfluidic device, Sens. Actuators B 114, 350–356 crofluidic devices, Langmuir 20, 9905–9908 (2004)
(2006) 20.47 A.S. Utada, E. Lorenceau, D.R. Link, P.D. Ka-
20.33 T. Kawakatsu, G. Trägårdh, C. Trägårdh, M. Naka- plan, H.A. Stone, D.A. Weitz: Monodisperse double
jima, N. Oda, T. Yonemoto: The effect of the emulsions generated from a microcapillary device,
hydrophobicity of microchannels and components Science 308, 537–541 (2005)
in water and oil phases on droplet formation in 20.48 K. Handique, M.A. Burns: Mathematical model-
microchannel water-in-oil emulsification, Colloids ing of drop mixing in a slit-type microchannel,
Surf. 179, 29–37 (2001) J. Micromech. Microeng. 11(5), 548–554 (2001)
568 Part B MEMS/NEMS and BioMEMS/NEMS
20.49 T. Nisisako, T. Torii, T. Higuchi: Novel microreactors 20.63 J.R. Millman, K.H. Bhatt, B.G. Prevo, O.D. Velev:
Part B 20
for functional polymer beads, Chem. Eng. J. 101, Anisotropic particle synthesis in dielectrophoret-
23–29 (2004) ically controlled microdroplet reactors, Nature 4,
20.50 H. Song, M.R. Bringer, J.D. Tice, C.J. Gerdts, R.F. Is- 98–102 (2005)
magilov: Experimental test of scaling of mixing 20.64 V. Srinivasan, V.K. Pamula, R.B. Fair: An integrated
by chaotic advection in droplets moving through digital microfluidic lab-on-a-chip for clinical di-
microfluidic channels, Appl. Phys. Lett. 83(22), agnostics on human physiological fluids, Lab Chip
4664–4666 (2003) 4(4), 310–315 (2004)
20.51 H. Song, J.D. Tice, R.F. Ismagilov: A microfluidic 20.65 B. Zheng, J.D. Tice, R.F. Ismagilov: Formation of ar-
system for controlling reaction networks in time, rayed droplets by soft lithography and two phase
Angew. Chem. Int. Ed. 42(7), 768–772 (2003) fluid flow, and application in protein crystalliza-
20.52 J.M. Kohler, T. Henkel, A. Grodrian, T. Kimer, tion, Adv. Mater. 16(15), 1365–1368 (2004)
M. Roth, K. Martin, J. Metze: Digital reaction tech- 20.66 M. He, J.S. Edgar, G.D. Jeffries, R.M. Lorenz,
nology by micro segmented flow – Components, J.P. Shelby, D.T. Chiu: Selective encapsulation
concepts and applications, Chem. Eng. J. 101(1–3), of single cells and subcellular organelles into
201–216 (2004) picoliter- and femtoliter-volume droplets, Anal.
20.53 L.H. Hung, K.M. Choi, W.Y. Tseng, Y.C. Tan, Chem. 77, 1539–1544 (2005)
K.J. Shea, A.P. Lee: Alternating droplet generation 20.67 S. Sugiura, T. Oda, Y. Aoyagi, R. Matsuo, T. Enomoto,
and controlled dynamic droplet fusion in microflu- K. Matsumoto, T. Nakamura, M. Satake, A. Ochiai,
idic device for CdS nanoparticles synthesis, Lab Chip N. Ohkohchi, M. Nakajima: Microfabricated airflow
6, 1–6 (2006) nozzle for microencapsulation of living cells into
20.54 D.R. Link, S.L. Anna, D.A. Weitz, H.A. Stone: 150 micrometer microcapsules, Biomed. Microde-
Geometrically mediated breakup of drops in mi- vices 9, 91–99 (2007)
crofluidic devices, Phys. Rev. Lett. 92(5), 054503 20.68 H. Shintaku, T. Kuwabara, S. Kawano, T. Suzuki,
(2004) I. Kanno, H. Kotera: Micro cell encapsulation and
20.55 Y.C. Tan, Y.L. Ho, A.P. Lee: Microfluidic sorting of its hydrogel-beads production using microfluidic
droplets by size, Microfluid. Nanofluid. 4, 343–348 device, Microsyst. Technol. 13(8-10), 951–958 (2007)
(2008) 20.69 S. Sakai, I. Hashimoto, K. Kawakami: Agarose-
20.56 G. Yi, S. Jeon, T. Thorsen, V.N. Manoharan, gelatin conjugate for adherent cell-enclosing
S.R. Quake, D.J. Pine, S. Yang: Generation of capsules, Biotechnol. Lett. 29(5), 731–735 (2007)
uniform photonic balls by template-assisted col- 20.70 C.H. Yang, K.S. Huang, J.Y. Chang: Manufacturing
loidal crystallization, Synth. Met. 139, 803–806 monodisperse chitosan microparticles containing
(2003) ampicillin using a microchannel chip, Biomed. Mi-
20.57 S. Sugiura, T. Oda, Y. Izumida, Y. Aoyagi, M. Satake, crodevices 9(2), 253–259 (2007)
A. Ochiai, N. Ohkohchi, M. Nakajima: Size control 20.71 S. Abraham, E.H. Jeong, T. Arakawa, S. Shoji,
of calcium alginate beads containing living cells K.C. Kim, I. Kim, J.S. Go: Microfluidics assisted
using micro-nozzle array, Biomaterials 26, 3327– synthesis of well-defined spherical polymeric mi-
3331 (2005) crocapsules and their utilization as potential
20.58 L.-H. Hung, A.P. Lee: Microfluidic devices for encapsulants, Lab Chip 6(6), 752–756 (2006)
the synthesis of nanoparticles and biomaterials, 20.72 J.W. Kim, A.S. Utada, A. Fernandez-Nieves, Z.B. Hu,
J. Med. Biol. Eng. 27(1), 1–6 (2007) D.A. Weitz: Fabrication of monodisperse gel shells
20.59 E. Lorenceau, A.S. Utada, D.R. Link, G. Cristobal, and functional microgels in microfluidic de-
M. Joanicot, D.A. Weitz: Generation of polymer- vices, Angew. Chem. Int. Ed. 46(11), 1819–1822
somes from double emulsions, Langmuir 21(20), (2007)
9183–9186 (2005) 20.73 V. Srinivasan, V.K. Pamula, R.B. Fair: Droplet-
20.60 R.C. Hayward, A.S. Utada, N. Dan, D.A. Weitz: based microfluidic lab-on-a-chip for glucose
Dewetting instability during the formation of detection, Anal. Chim. Acta 507, 145–150 (2004)
polymersomes from block-copolymer-stabilized 20.74 I. Shestopalov, J.D. Tice, R.F. Ismagilov: Multi-step
double emulsions, Langmuir 22(10), 4457–4461 synthesis of nanoparticles performed on millisec-
(2006) ond time scale in a microfluidic droplet-based
20.61 G. Yi, T. Thorsen, V.N. Manoharan, M. Hwang, system, Lab Chip 4, 316–321 (2004)
S. Jeon, D.J. Pine, S.R. Quake, S. Yang: Generation 20.75 Z. Guttenberg, H. Müller, H. Habermüller, A. Geis-
of uniform colloidal assemblies in soft microfluidic bauer, J. Pipper, J. Felbel, M. Kielpinski, J. Scriba,
devices, Adv. Mater. 15(15), 1300–1304 (2003) A. Wixforth: Planar chip device for PCR and hy-
20.62 G. Yi, V.N. Manoharan, S. Klein, K.R. Brzezin- bridization with surface acoustic wave pump, Lab
ska, D.J. Pine, F.F. Lange, S. Yang: Monodisperse Chip 5, 308–317 (2004)
micrometer-scale spherical assemblies of polymer 20.76 M.A. Burns, C.H. Mastrangelo, T.S. Sammarco,
particles, Adv. Mater. 14(16), 1137–1140 (2002) F.P. Man, J.R. Webster, B.N. Johnson, B. Foerster,
Micro-/Nanodroplets in Microfluidic Devices References 569
D. Jones, Y. Fields, A.R. Kaiser, D.T. Burke: Micro- real-time, single-copy polymerase chain reaction
Part B 20
fabricated structures for integrated DNA analysis, in picoliter droplets, Anal. Chem. 79, 8471–8475
Proc. Natl. Acad. Sci. USA 93, 5556–5561 (1996) (2007)
20.77 B. Zheng, L.S. Roach, R.F. Ismagilov: Screening of 20.81 M. Srisa-Art, A.J. deMello, J.B. Edel: High-
protein crystallization conditions on a microfluidic througput DNA droplet assays using picoliter
chip using nanoliter-size droplets, J. Am. Chem. reactor volumes, Anal. Chem. 79(17), 6682–6689
Soc. 125, 11170–11171 (2003) (2007)
20.78 B. Zheng, J.D. Tice, L.S. Roach, R.F. Ismagilov: 20.82 D. Dendukuri, K. Tsoi, T.A. Hatton, P.S. Doyle:
A droplet-based, composite PDMS/glass capillary Controlled synthesis of nonspherical microparti-
microfluidic system for evaluating protein crys- cles using microfluidics, Langmuir 21, 2113–2116
tallization conditions by microbatch and vapor- (2005)
diffusion methods with on-chip x-ray diffraction, 20.83 A. Terray, J. Oakey, D.W.M. Marr: Microfluidic
Angew. Chem. Int. Ed. 43, 2508–2511 (2004) control using colloidal devices, Science 296(7),
20.79 P.S. Dittrich, M. Jahnz, P. Schwille: A new em- 1841–1844 (2005)
bedded process for compartmentalized cell-free 20.84 B.R. Acharya, T. Krupenkin, S. Ramachandran,
protein expression and on-line detection in mi- Z. Wang, C.C. Huang, J.A. Rogers: Tunable opti-
crofluidic devices, ChemBioChem. 6, 811–814 (2005) cal fiber devices based on broadband long-period
20.80 R.N. Beer, B.J. Hindson, E.K. Wheeler, S.B. Hall, gratings and pumped microfluidics, Appl. Phys.
K.A. Rose, I.M. Kennedy, B.W. Colston: On-chip, Lett. 83(24), 4912–4914 (2003)
571
Part C
Scanning Part C Scanning-Probe Microscopy
Scanning Pro
21. Scanning Probe Microscopy –
Part C 21
Principle of Operation, Instrumentation, and Probes
Bharat Bhushan, Othmar Marti
The scanning tunneling microscope (STM), developed face, whether it is electrically conductive or insulating.
by Binnig and his colleagues in 1981 at the IBM Zurich The AFM has become a popular surface profiler for
Research Laboratory in Rüschlikon (Switzerland), was topographic and normal force measurements on the
the first instrument capable of directly obtaining three- micro- to nanoscale [21.4]. AFMs modified in order
dimensional (3-D) images of solid surfaces with atomic to measure both normal and lateral forces are called
resolution [21.1]. Binnig and Rohrer received a No- lateral force microscopes (LFMs) or friction force mi-
bel Prize in Physics in 1986 for their discovery. STMs croscopes (FFMs) [21.5–11]. FFMs have been further
can only be used to study surfaces which are elec- modified to measure lateral forces in two orthogo-
trically conductive to some degree. Based on their nal directions [21.12–16]. A number of researchers
design of the STM, in 1985, Binnig et al. developed have modified and improved the original AFM and
an atomic force microscope (AFM) to measure ultra- FFM designs, and have used these improved systems
small forces (less than 1 μN) between the AFM tip to measure the adhesion and friction of solid and li-
surface and the sample surface [21.2] (also see [21.3]). quid surfaces on micro- and nanoscales [21.4, 17–30].
AFMs can be used to measure any engineering sur- AFMs have been used to study scratching and wear, and
574 Part C Scanning-Probe Microscopy
to measure elastic/plastic mechanical properties (such allow investigation and manipulation of surfaces down
as indentation hardness and the modulus of elastic- to the atomic scale. As our understanding of the un-
ity) [21.4,10,11,21,23,26–29,31–36]. AFMs have been derlying interaction mechanisms has grown, SPMs and
used to manipulate individual atoms of xenon [21.37], their derivatives have found applications in many fields
molecules [21.38], silicon surfaces [21.39] and poly- beyond basic research fields and microscopy.
mer surfaces [21.40]. STMs have been used to create Families of instruments based on STMs and
nanofeatures via localized heating or by inducing AFMs, called scanning probe microscopes (SPMs),
chemical reactions under the STM tip [21.41–43] have been developed for various applications of sci-
and through nanomachining [21.44]. AFMs have also entific and industrial interest. These include STM,
been used for nanofabrication [21.4, 10, 45–47] and AFM, FFM (or LFM), scanning electrostatic force mi-
nanomachining [21.48]. croscopy (SEFM) [21.65, 66], scanning force acoustic
STMs and AFMs are used at extreme magnifi- microscopy (SFAM) (or atomic force acoustic mi-
cations ranging from 103 to 109 in the x-, y- and croscopy (AFAM)) [21.21, 22, 36, 67–69], scanning
z-directions in order to image macro to atomic di- magnetic microscopy (SMM) (or magnetic force mi-
mensions with high resolution and for spectroscopy. croscopy (MFM)) [21.70–73], scanning near-field opti-
These instruments can be used in any environment, cal microscopy (SNOM) [21.74–77], scanning thermal
such as ambient air [21.2, 49], various gases [21.17], microscopy (SThM) [21.78–80], scanning electrochem-
liquids [21.50–52], vacuum [21.1, 53], at low tem- ical microscopy (SEcM) [21.81], scanning Kelvin probe
peratures (lower than about 100 K) [21.54–58] and at microscopy (SKPM) [21.82–86], scanning chemical
high temperatures [21.59, 60]. Imaging in liquid al- potential microscopy (SCPM) [21.79], scanning ion
lows the study of live biological samples and it also conductance microscopy (SICM) [21.87, 88] and scan-
eliminates the capillary forces that are present at the ning capacitance microscopy (SCM) [21.82, 89–91].
tip–sample interface when imaging aqueous samples When the technique is used to measure forces (as
in ambient air. Low-temperature (liquid helium tem- in AFM, FFM, SEFM, SFAM and SMM) it is also
peratures) imaging is useful when studying biological referred to as scanning force microscopy (SFM). Al-
and organic materials and low-temperature phenom- though these instruments offer atomic resolution and are
ena such as superconductivity or charge-density waves. ideal for basic research, they are also used for cutting-
Low-temperature operation is also advantageous for edge industrial applications which do not require atomic
high-sensitivity force mapping due to the reduced ther- resolution. The commercial production of SPMs started
mal vibration. They also have been used to image with the STM in 1987 and the AFM in 1989 by Digital
liquids such as liquid crystals and lubricant molecules Instruments, Inc. (Santa Barbara, USA). For compar-
on graphite surfaces [21.61–64]. While applications of isons of SPMs with other microscopes, see Table 21.1
SPM techniques initially focused on their pure imaging (Veeco Instruments, Inc., Santa Barbara, USA). Num-
capabilities, research into the physics and chemistry of bers of these instruments are equally divided between
probe–sample interactions and SPM-based quantitative the US, Japan and Europe, with the following split be-
analyses of tribological, electronic, magnetic, biologi- tween industry/university and government laboratories:
cal, and chemical surfaces have become increasingly 50/50, 70/30, and 30/70, respectively. It is clear that
popular in recent years. Nanoscale science and tech- research and industrial applications of SPMs are ex-
nology is often tied to the use of SPMs since they panding rapidly.
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.1 Scanning Tunneling Microscope 575
Part C 21.1
The principle of electron tunneling was first proposed
Constant Constant
by Giaever [21.93]. He envisioned that if a poten- current mode height mode
tial difference is applied to two metals separated by
a thin insulating film, a current will flow because of Schematic Scan Scan
the ability of electrons to penetrate a potential bar- view I I
rier. To be able to measure a tunneling current, the
two metals must be spaced no more than 10 nm apart.
Binnig et al. [21.1] introduced vacuum tunneling com-
bined with lateral scanning. The vacuum provides the
ideal barrier for tunneling. The lateral scanning allows One scan z I
one to image surfaces with exquisite resolution – lat-
erally to less than 1 nm and vertically to less than
0.1 nm – sufficient to define the position of single atoms.
The very high vertical resolution of the STM is ob- x x
tained because the tunnel current varies exponentially
Multiple scans
with the distance between the two electrodes; that is,
the metal tip and the scanned surface. Typically, the
tunneling current decreases by a factor of 2 as the
separation is increased by 0.2 nm. Very high lateral res-
olution depends upon sharp tips. Binnig et al. overcame
two key obstacles by damping external vibrations and Fig. 21.1 An STM can be operated in either the constant-
moving the tunneling probe in close proximity to the current or the constant-height mode. The images are of
sample. Their instrument is called the scanning tun- graphite in air (after [21.92])
neling microscope (STM). Today’s STMs can be used
in ambient environments for atomic-scale imaging of a metal tip can be scanned across a surface at nearly
surfaces. Excellent reviews on this subject have been constant height and constant voltage while the current
presented by Hansma and Tersoff [21.92], Sarid and is monitored, as shown in the right-hand column of
Elings [21.94], Durig et al. [21.95]; Frommer [21.96], Fig. 21.1. In this case, the feedback network responds
Güntherodt and Wiesendanger [21.97], Wiesendanger just rapidly enough to keep the average current constant.
and Güntherodt [21.98], Bonnell [21.99], Marti and The current mode is generally used for atomic-scale
Amrein [21.100], Stroscio and Kaiser [21.101], and images; this mode is not practical for rough surfaces.
Güntherodt et al. [21.102]. A three-dimensional picture [z(x, y)] of a surface con-
The principle of the STM is straightforward. sists of multiple scans [z(x)] displayed laterally to each
A sharp metal tip (one electrode of the tunnel junc- other in the y-direction. It should be noted that if differ-
tion) is brought close enough (0.3–1 nm) to the surface ent atomic species are present in a sample, the different
to be investigated (the second electrode) to make the atomic species within a sample may produce different
tunneling current measurable at a convenient operat- tunneling currents for a given bias voltage. Thus the
ing voltage (10 mV–1 V). The tunneling current in this height data may not be a direct representation of the
case varies from 0.2 to 10 nA. The tip is scanned over topography of the surface of the sample.
the surface at a distance of 0.3–1 nm, while the tun-
neling current between it and the surface is measured. 21.1.1 The STM Design of Binnig et al.
The STM can be operated in either the constant cur-
rent mode or the constant height mode (Fig. 21.1). The Figure 21.2 shows a schematic of an AFM designed
left-hand column of Fig. 21.1 shows the basic constant by Binnig and Rohrer and intended for operation in
current mode of operation. A feedback network changes ultrahigh vacuum [21.1, 103]. The metal tip was fixed
the height of the tip z to keep the current constant. The to rectangular piezodrives Px , P y , and Pz made out of
displacement of the tip, given by the voltage applied to commercial piezoceramic material for scanning. The
the piezoelectric drive, then yields a topographic map of sample is mounted via either superconducting magnetic
the surface. Alternatively, in the constant height mode, levitation or a two-stage spring system to achieve a sta-
576 Part C Scanning-Probe Microscopy
Part C 21.1
Pz –Y
Px PZT tube
Vz
scanner
–X X
δ Y
Py V
A
VT Δ d
Z I
CU d JT
Tip
Sample
Fig. 21.2 Principle of operation of the STM, from Binnig
and Rohrer [21.103] Fig. 21.3 Principle of operation of a commercial STM.
A sharp tip attached to a piezoelectric tube scanner is
ble gap width of about 0.02 nm. The tunnel current scanned on a sample
JT is a sensitive function of the gap width d where
JT ∝ VT exp(−Aφ1/2 d). Here VT is the bias voltage, first commercial STM, the Nanoscope I, in 1987. In
φ is the average barrier height (work function) and the the recent Nanoscope IV STM, intended for operation
constant A = 1.025 eV−1/2 Å−1 . With a work function in ambient air, the sample is held in position while
of a few eV, JT changes by an order of magnitude for a piezoelectric crystal in the form of a cylindrical tube
an angstrom change in d. If the current is kept con- (referred to as a PZT tube scanner) scans the sharp
stant to within, for example, 2%, then the gap d remains metallic probe over the surface in a raster pattern while
constant to within 1 pm. For operation in the constant sensing and relaying the tunneling current to the control
current mode, the control unit CU applies a voltage Vz station (Fig. 21.3). The digital signal processor (DSP)
to the piezo Pz such that JT remains constant when calculates the tip–sample separation required by sens-
scanning the tip with P y and Px over the surface. At ing the tunneling current flowing between the sample
a constant work function φ, Vz (Vx , Vy ) yields the rough- and the tip. The bias voltage applied between the sam-
ness of the surface z(x, y) directly, as illustrated by ple and the tip encourages the tunneling current to flow.
a surface step at A. Smearing the step, δ (lateral res- The DSP completes the digital feedback loop by relay-
olution) is on the order of (R)1/2 , where R is the ing the desired voltage to the piezoelectric tube. The
radius of the curvature of the tip. Thus, a lateral reso- STM can operate in either the constant height or the
lution of about 2 nm requires tip radii on the order of constant current mode, and this can be selected using
10 nm. A 1 mm diameter solid rod ground at one end at the control panel. In the constant current mode, the feed-
roughly 90◦ yields overall tip radii of only a few hun- back gains are set high, the tunneling tip closely tracks
dred nanometers, the presence of rather sharp microtips the sample surface, and the variation in the tip height
on the relatively dull end yields a lateral resolution of required to maintain constant tunneling current is meas-
about 2 nm. In situ sharpening of the tips, achieved by ured by the change in the voltage applied to the piezo
gently touching the surface, brings the resolution down tube. In the constant height mode, the feedback gains
to the 1 nm range; by applying high fields (on the order are set low, the tip remains at a nearly constant height
of 108 V/cm) for, say, half an hour, resolutions consid- as it sweeps over the sample surface, and the tunneling
erably below 1 nm can be reached. Most experiments current is imaged.
have been performed with tungsten wires either ground Physically, the Nanoscope STM consists of three
or etched to a typical radius of 0.1–10 μm. In some main parts: the head, which houses the piezoelec-
cases, in situ processing of the tips has been performed tric tube scanner which provides three-dimensional tip
to further reduce tip radii. motion and the preamplifier circuit for the tunneling
current (FET input amplifier) mounted on the top of
21.1.2 Commercial STMs the head; the base on which the sample is mounted;
and the base support, which supports the base and
There are a number of commercial STMs available on head [21.4]. The base accommodates samples which are
the market. Digital Instruments, Inc., introduced the up to 10 mm by 20 mm and 10 mm thick. Scan sizes
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.1 Scanning Tunneling Microscope 577
available for the STM are 0.7 μm (for atomic resolu- allows the tip to be engaged and withdrawn from the
Part C 21.1
tion), 12 μm, 75 μm and 125 μm square. surface automatically.
The scanning head controls the three-dimensional Samples to be imaged with the STM must be con-
motion of the tip. The removable head consists of ductive enough to allow a few nanoamperes of current
a piezo tube scanner, about 12.7 mm in diameter, to flow from the bias voltage source to the area to be
mounted into an Invar shell, which minimizes vertical scanned. In many cases, nonconductive samples can be
thermal drift because of the good thermal match be- coated with a thin layer of a conductive material to facil-
tween the piezo tube and the Invar. The piezo tube has itate imaging. The bias voltage and the tunneling current
separate electrodes for x-, y- and z-motion, which are depend on the sample. Usually they are set to a stan-
driven by separate drive circuits. The electrode con- dard value for engagement and fine tuned to enhance the
figuration (Fig. 21.3) provides x- and y-motions which quality of the image. The scan size depends on the sam-
are perpendicular to each other, it minimizes horizontal ple and the features of interest. A maximum scan rate
and vertical coupling, and it provides good sensitiv- of 122 Hz can be used. The maximum scan rate is usu-
ity. The vertical motion of the tube is controlled by ally related to the scan size. Scan rates above 10 Hz are
the Z-electrode, which is driven by the feedback loop. used for small scans (typically 60 Hz for atomic-scale
The x- and y-scanning motions are each controlled imaging with a 0.7 μm scanner). The scan rate should
by two electrodes which are driven by voltages of the be lowered for large scans, especially if the sample sur-
same magnitude but opposite signs. These electrodes faces are rough or contain large steps. Moving the tip
are called −y, −x, +y, and +x. Applying complimen-
tary voltages allows a short, stiff tube to provide a good
scan range without the need for a large voltage. The mo- Evaporated C60 film on mica
tion of the tip that arises due to external vibrations is 5 nA
proportional to the square of the ratio of vibration fre-
quency to the resonant frequency of the tube. Therefore, 2.5
2.5 nA
to minimize the tip vibrations, the resonant frequencies nA
5
of the tube are high: about 60 kHz in the vertical direc- 0 nA
tion and about 40 kHz in the horizontal direction. The 1.25
tip holder is a stainless steel tube with an inner diam- 2.5 1
eter of 300 μm when 250 μm diameter tips are used, 0.75
which is mounted in ceramic in order to minimize the 0 0.5
mass at the end of the tube. The tip is mounted either on 0 0.25
0.25 0.5
the front edge of the tube (to keep the mounting mass 0.75 1 0
1.25 nm
low and the resonant frequency high) (Fig. 21.3) or the
center of the tube for large-range scanners, namely 75
and 125 μm (to preserve the symmetry of the scanning).
0.5
0. 5 nnm
This commercial STM accepts any tip with a 250 μm
diameter shaft. The piezotube requires x–y-calibration,
which is carried out by imaging an appropriate calibra- 0.25nm
0.25 nm
tion standard. Cleaved graphite is used for heads with nm
small scan lengths while two-dimensional grids (a gold- 0.5
0 nm
plated rule) can be used for long-range heads. 3
The Invar base holds the sample in position, sup- 0.25
ports the head, and provides coarse x–y-motion for the 2
sample. A sprung-steel sample clip with two thumb
screws holds the sample in place. An x–y-translation 0 1
0
stage built into the base allows the sample to be reposi- 1
tioned under the tip. Three precision screws arranged in 2 0
3 nm
a triangular pattern support the head and provide coarse
and fine adjustment of the tip height. The base support Fig. 21.4 STM images of evaporated C60 film on gold-coated
consists of the base support ring and the motor hous- freshly cleaved mica obtained using a mechanically sheared Pt-Ir
ing. The stepper motor enclosed in the motor housing (80/20) tip in constant height mode (after [21.104])
578 Part C Scanning-Probe Microscopy
to a specific shape which is consistent from tip to available from Materials Analytical Services (Raleigh,
Part C 21.2
tip. The probes have a full cone angle of ≈ 15◦ , USA).
and a tip radius of less than 50 nm. To image very Pt alloy and W tips are very sharp and give high
deep trenches (> 0.25 μm) and nanofeatures, focused resolution, but are fragile and sometimes break when
ion beam (FIB)-milled CG probes with extremely contacting a surface. Diamond tips have been used by
sharp tips (radii < 5 nm) are used. The Pt-Ir probes Kaneko and Oguchi [21.107]. Diamond tips made con-
are coated with a nonconducting film (not shown ductive by boron ion implantation were found to be
in the figure) for electrochemistry. These probes are chip-resistant.
interaction force between tip and sample is measured or lower [21.122]. Upon increasing the force from
Part C 21.2
by monitoring the cantilever deflection. In the noncon- 10 pN, Ohnesorge and Binnig [21.122] observed that
tact (or dynamic) mode, the force gradient is obtained monoatomic steplines were slowly wiped away and
by vibrating the cantilever and measuring the shift in a perfectly ordered structure was left. This observa-
the resonant frequency of the cantilever. To obtain to- tion explains why mostly defect-free atomic resolution
pographic information, the interaction force is either has been observed with AFM. Note that for atomic-
recorded directly, or used as a control parameter for resolution measurements, the cantilever should not be
a feedback circuit that maintains the force or force so soft as to avoid jumps. Further note that performing
derivative at a constant value. Using an AFM operated measurements in the noncontact imaging mode may be
in the contact mode, topographic images with a verti- desirable for imaging with atomic resolution.
cal resolution of less than 0.1 nm (as low as 0.01 nm) The key component in an AFM is the sensor used
and a lateral resolution of about 0.2 nm have been ob- to measure the force on the tip due to its interaction
tained [21.3, 50, 110–114]. Forces of 10 nN to 1 pN are with the sample. A cantilever (with a sharp tip) with an
measurable with a displacement sensitivity of 0.01 nm. extremely low spring constant is required for high ver-
These forces are comparable to the forces associated tical and lateral resolutions at small forces (0.1 nN or
with chemical bonding, for example 0.1 μN for an ionic lower), but a high resonant frequency is desirable (about
bond and 10 pN for a hydrogen bond [21.2]. For further 10 to 100 kHz) at the same time in order to minimize
reading, see [21.94–96, 100, 102, 115–119]. the sensitivity to building vibrations, which occur at
Lateral forces applied at the tip during scan- around 100 Hz. This requires a spring with an extremely
ning in the contact mode affect roughness measure- low vertical spring constant (typically 0.05 to 1 N/m)
ments [21.120]. To minimize the effects of friction and as well as a low mass (on the order of 1 ng). Today,
other lateral forces on topography measurements in the the most advanced AFM cantilevers are microfabricated
contact mode, and to measure the topographies of soft from silicon or silicon nitride using photolithographic
surfaces, AFMs can be operated in the so-called tapping techniques. Typical lateral dimensions are on the or-
or force modulation mode [21.32, 121]. der of 100 μm, with thicknesses on the order of 1 μm.
The STM is ideal for atomic-scale imaging. To The force on the tip due to its interaction with the
obtain atomic resolution with the AFM, the spring sample is sensed by detecting the deflection of the
constant of the cantilever should be weaker than the compliant lever with a known spring constant. This can-
equivalent spring between atoms. For example, the vi- tilever deflection (displacement smaller than 0.1 nm)
bration frequencies ω of atoms bound in a molecule or has been measured by detecting a tunneling current
in a crystalline solid are typically 1013 Hz or higher. similar to that used in the STM in the pioneering
Combining this with an atomic mass m of ≈ 10−25 kg work of Binnig et al. [21.2] and later used by Giessibl
gives an interatomic spring constant k, given by ω2 m,
of around 10 N/m [21.115]. (For comparison, the spring
constant of a piece of household aluminium foil that is Electron tunneling Optical interferometry
4 mm long and 1 mm wide is about 1 N/m.) Therefore,
a cantilever beam with a spring constant of about 1 N/m
or lower is desirable. Tips must be as sharp as possible,
and tip radii of 5 to 50 nm are commonly available. STM Lever Lens
Atomic resolution cannot be achieved with these
tips at normal loads in the nN range. Atomic structures Laser beam deflection Capacitance method
at these loads have been obtained from lattice imaging
PSD
or by imaging the crystal’s periodicity. Reported data
show either perfectly ordered periodic atomic structures He-Ne
laser
or defects on a larger lateral scale, but no well-defined, Electrode
laterally resolved atomic-scale defects like those seen
in images routinely obtained with a STM. Interatomic Fig. 21.8 Schematics of the four detection systems to
forces with one or several atoms in contact are 20–40 measure cantilever deflection. In each set-up, the sample
or 50–100 pN, respectively. Thus, atomic resolution mounted on piezoelectric body is shown on the right, the
with an AFM is only possible with a sharp tip on cantilever in the middle, and the corresponding deflection
a flexible cantilever at a net repulsive force of 100 pN sensor on the left (after [21.118])
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.2 Atomic Force Microscope 581
et al. [21.56], by capacitance detection [21.123, 124], the machining of the piezo translators, causing cross-
Part C 21.2
piezoresistive detection [21.125, 126], and by four opti- talk between the Z-piezo to the x- and y-piezos, and
cal techniques, namely (1) optical interferometry [21.5, vice versa. Nonlinear distortions mainly result from the
6, 127, 128] using optical fibers [21.57, 129] (2) opti- presence of a hysteresis loop in piezoelectric ceramics.
cal polarization detection [21.72, 130], (3) laser diode They may also occur if the scan frequency approaches
feedback [21.131] and (4) optical (laser) beam de- the upper frequency limit of the x- and y-drive am-
flection [21.7, 8, 53, 111, 112]. Schematics of the four plifiers or the upper frequency limit of the feedback
more commonly used detection systems are shown in loop (z-component). In addition, electronic noise may
Fig. 21.8. The tunneling method originally used by Bin- be present in the system. The noise is removed by dig-
nig et al. [21.2] in the first version of the AFM uses ital filtering in real space [21.134] or in the spatial
a second tip to monitor the deflection of the cantilever frequency domain (Fourier space) [21.135].
with its force sensing tip. Tunneling is rather sensitive Processed data consists of many tens of thousand of
to contaminants and the interaction between the tunnel- points per plane (or data set). The outputs from the first
ing tip and the rear side of the cantilever can become STM and AFM images were recorded on an x–y-chart
comparable to the interaction between the tip and sam- recorder, with the z-value plotted against the tip position
ple. Tunneling is rarely used and is mentioned mainly in the fast scan direction. Chart recorders have slow re-
for historical reasons. Giessibl et al. [21.56] have used sponses, so computers are used to display the data these
it for a low-temperature AFM/STM design. In con- days. The data are displayed as wire mesh displays or
trast to tunneling, other deflection sensors are placed grayscale displays (with at least 64 shades of gray).
far from the cantilever, at distances of micrometers to
tens of millimeters. The optical techniques are believed 21.2.1 The AFM Design of Binnig et al.
to be more sensitive, reliable and easily implemented
detection methods than the others [21.94, 118]. The op- In the first AFM design developed by Binnig et al. [21.2],
tical beam deflection method has the largest working AFM images were obtained by measuring the force ex-
distance, is insensitive to distance changes and is ca- erted on a sharp tip created by its proximity to the
pable of measuring angular changes (friction forces); surface of a sample mounted on a 3-D piezoelectric
therefore, it is the most commonly used in commercial scanner. The tunneling current between the STM tip
SPMs. and the backside of the cantilever beam to which the tip
Almost all SPMs use piezo translators to scan the was attached was measured to obtain the normal force.
sample, or alternatively to scan the tip. An electric field This force was kept at a constant level with a feed-
applied across a piezoelectric material causes a change back mechanism. The STM tip was also mounted on
in the crystal structure, with expansion in some direc- a piezoelectric element to maintain the tunneling current
tions and contraction in others. A net change in volume at a constant level.
also occurs [21.132]. The first STM used a piezo tripod
for scanning [21.1]. The piezo tripod is one way to gen- 21.2.2 Commercial AFMs
erate three-dimensional movement of a tip attached at
its center. However, the tripod needs to be fairly large A review of early designs of AFMs has been pre-
(≈ 50 mm) to get a suitable range. Its size and asym- sented by Bhushan [21.4]. There are a number of
metric shape makes it susceptible to thermal drift. Tube commercial AFMs available on the market. Major man-
scanners are widely used in AFMs [21.133]. These pro- ufacturers of AFMs for use in ambient environments
vide ample scanning range with a small size. Electronic are: Digital Instruments, Inc., Topometrix Corp. and
control systems for AFMs are based on either analog other subsidiaries of Veeco Instruments, Inc., Molecu-
or digital feedback. Digital feedback circuits are better lar Imaging Corp. (Phoenix, USA), Quesant Instrument
suited for ultralow noise operation. Corp. (Agoura Hills, USA), Nanoscience Instruments,
Images from the AFMs need to be processed. An Inc. (Phoenix, USA), Seiko Instruments (Chiba, Japan);
ideal AFM is a noise-free device that images a sample and Olympus (Tokyo, Japan). AFM/STMs for use
with perfect tips of known shape and has a perfectly in UHV environments are manufactured by Omicron
linear scanning piezo. In reality, scanning devices are Vakuumphysik GmbH (Taunusstein, Germany).
affected by distortions and these distortions must be cor- We describe here two commercial AFMs – small-
rected for. The distortions can be linear and nonlinear. sample and large-sample AFMs – for operation in the
Linear distortions mainly result from imperfections in contact mode, produced by Digital Instruments, Inc.,
582 Part C Scanning-Probe Microscopy
a) Mirrored
prism in both constant height and constant force modes. The
DSP always adjusts the distance between the sample
AFM signal Diode laser and the tip according to the cantilever deflection error
(A + B) – (C+D) Mirror & lens signal, but if the feedback gains are low the piezo re-
mains at an almost constant height and the cantilever
B
A Cantilever deflection data is collected. With high gains, the piezo
D & substrate height changes to keep the cantilever deflection nearly
FFM C
signal Split-diode constant (so the force is constant), and the change in
(A + C) – (B +D) photo- piezo height is collected by the system.
detector Sample In the operation of a commercial small-sample AFM
(as shown in Fig. 21.9a), the sample (which is generally
z xyz
no larger than 10 mm × 10 mm) is mounted on a PZT
y PZT tube tube scanner, which consists of separate electrodes used
x scanner to precisely scan the sample in the x–y-plane in a raster
pattern and to move the sample in the vertical (z-) direc-
b) tion. A sharp tip at the free end of a flexible cantilever
Split-diode is brought into contact with the sample. Features on the
photo- sample surface cause the cantilever to deflect in the ver-
detector Laser diode,
collimator & tical and lateral directions as the sample moves under
lens the tip. A laser beam from a diode laser (5 mW max.
Adjust-
able peak output at 670 nm) is directed by a prism onto the
mirror Laser path back of a cantilever near its free end, tilted downward
Fixed at about 10◦ with respect to the horizontal plane. The
mirror reflected beam from the vertex of the cantilever is di-
rected through a mirror onto a quad photodetector (split
photodetector with four quadrants) (commonly called
a position-sensitive detector or PSD, produced by Sil-
Lens icon Detector Corp., Camarillo, USA). The difference
in signal between the top and bottom photodiodes pro-
Mirror Lens vides the AFM signal, which is a sensitive measure of
x
y the cantilever vertical deflection. The topographic fea-
Camera
objective z xyz PZT tures of the sample cause the tip to deflect in the vertical
lens tube scanner
Feedback
Cantilever holder Computer
Sample x
Motorized Laser
stage y
Photo-
Fig. 21.9a,b Principles of operation of (a) a commercial small- detector Substrate
holder
sample AFM/FFM, and (b) a large-sample AFM/FFM
Canti-
lever
piezo
with scanning lengths ranging from about 0.7 μm (for
atomic resolution) to about 125 μm [21.9,111,114,136].
The original design of these AFMs comes from Meyer Cantilever substrate
and Amer [21.53]. Basically, the AFM scans the sam- Sample
ple in a raster pattern while outputting the cantilever
deflection error signal to the control station. The can- z control xyz piezo xy control
tilever deflection (or the force) is measured using a laser
deflection technique (Fig. 21.9). The DSP in the work- Fig. 21.10 Schematic of tapping mode used for surface
station controls the z-position of the piezo based on roughness measurements
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.2 Atomic Force Microscope 583
direction as the sample is scanned under the tip. This at the tip surface during sliding. In the so-called friction
Part C 21.2
tip deflection will change the direction of the reflected mode, the sample is scanned back and forth in a direc-
laser beam, changing the intensity difference between tion orthogonal to the long axis of the cantilever beam.
the top and bottom sets of photodetectors (AFM signal). Friction force between the sample and the tip will twist
In a mode of operation called the height mode, used for the cantilever. As a result, the laser beam will be de-
topographic imaging or for any other operation in which flected out of the plane defined by the incident beam
the normal forceapplied is to be kept constant, a feed- and the beam is reflected vertically from an untwisted
back circuit is used to modulate the voltage applied to cantilever. This produces a difference in laser beam in-
the PZT scanner in order to adjust the height of the tensity between the beams received by the left-hand and
PZT, so that the cantilever vertical deflection (given by right-hand sets of quadrants of the photodetector. The
the intensity difference between the top and bottom de- intensity difference between the two sets of detectors
tector) will remain constant during scanning. The PZT (FFM signal) is directly related to the degree of twisting
height variation is thus a direct measure of the surface and hence to the magnitude of the friction force. This
roughness of the sample. method provides three-dimensional maps of the friction
In a large-sample AFM, force sensors based on op- force. One problem associated with this method is that
tical deflection methods or scanning units are mounted any misalignment between the laser beam and the pho-
on the microscope head (Fig. 21.9b). Because of the un- todetector axis introduces errors into the measurement.
wanted vibrations caused by cantilever movement, the However, by following the procedures developed by
lateral resolution of this design is somewhat poorer than Ruan and Bhushan [21.136], in which the average FFM
the design in Fig. 21.9a in which the sample is scanned signal for the sample scanned in two opposite directions
instead of the cantilever beam. The advantage of the is subtracted from the friction profiles of each of the
large-sample AFM is that large samples can be easily two scans, the misalignment effect can be eliminated.
measured. By following the friction force calibration procedures
Most AFMs can be used for topography meas- developed by Ruan and Bhushan [21.136], voltages cor-
urements in the so-called tapping mode (intermittent responding to friction forces can be converted to force
contact mode), in what is also referred to as dynamic units. The coefficient of friction is obtained from the
force microscopy. In the tapping mode, during the sur- slope of the friction force data measured as a func-
face scan, the cantilever/tip assembly is sinusoidally tion of the normal load, which typically ranges from
vibrated by a piezo mounted above it, and the os- 10 to 150 nN. This approach eliminates any contribu-
cillating tip slightly taps the surface at the resonant tions from adhesive forces [21.10]. To calculate the
frequency of the cantilever (70–400 kHz) with a con- coefficient of friction based on a single point measure-
stant (20–100 nm) amplitude of vertical oscillation, and ment, the friction force should be divided by the sum
a feedback loop keeps the average normal force constant of the normal load applied and the intrinsic adhesive
(Fig. 21.10). The oscillating amplitude is kept large force. Furthermore, it should be pointed out that the
enough that the tip does not get stuck to the sample coefficient of friction is not independent of load for
due to adhesive attraction. The tapping mode is used single-asperity contact. This is discussed in more detail
in topography measurements to minimize the effects of later.
friction and other lateral forces to measure the topogra-
phy of soft surfaces.
Topographic measurements can be made at any Fast scan direction Slow scan
direction
scanning angle. At first glance, the scanning angle may
not appear to be an important parameter. However, the
friction force between the tip and the sample will affect
the topographic measurements in a parallel scan (scan-
ning along the long axis of the cantilever). This means
that a perpendicular scan may be more desirable. Gen-
erally, one picks a scanning angle which gives the same
topographic data in both directions; this angle may be Fig. 21.11 Schematic of triangular pattern trajectory of the
slightly different to that for the perpendicular scan. AFM tip as the sample is scanned in two dimensions. Dur-
The left-hand and right-hand quadrants of the pho- ing imaging, data are only recorded during scans along the
todetector are used to measure the friction force applied solid scan lines
584 Part C Scanning-Probe Microscopy
Part C 21.2
Cantilever
Sample
AFM DVM
control switch
b) Laser diode Viewing Laser diode scan rates of less than 0.5 to 122 Hz are typically used.
y–positioner window x–positioner Higher scan rates are used for smaller scan lengths. For
AFM photo-
diode example, the scan rates in the fast and slow scan direc-
positioner tions for an area of 10 μm × 10 μm scanned at 0.5 Hz are
Preamp
FFM housing 10 μm/s and 20 nm/s, respectively.
photodiode Beam path We now describe the construction of a small-sample
positioner AFM in more detail. It consists of three main parts:
Holding arm
Cantilever
the optical head which senses the cantilever deflection;
Photo-
diode mount a PZT tube scanner which controls the scanning motion
housing Cantilever of the sample mounted on one of its ends; and the base,
which supports the scanner and head and includes cir-
Interlock cuits for the deflection signal (Fig. 21.12a). The AFM
x–y–positioning stage sensor connects directly to a control system. The optical head
consists of a laser diode stage, a photodiode stage
preamp board, the cantilever mount and its holding arm,
The tip is scanned in such a way that its trajectory on and the deflected beam reflecting mirror, which reflects
the sample forms a triangular pattern (Fig. 21.11). Scan- the deflected beam toward the photodiode (Fig. 21.12b).
ning speeds in the fast and slow scan directions depend The laser diode stage is a tilt stage used to adjust the po-
on the scan area and scan frequency. Scan sizes rang- sition of the laser beam relative to the cantilever. It con-
ing from less than 1 nm × 1 nm to 125 μm × 125 μm and sists of the laser diode, collimator, focusing lens, base-
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.2 Atomic Force Microscope 585
Part C 21.2
a) 1 0.2 nm
base of the microscope (Fig. 21.12c). The stepper mo-
tor is controlled manually with the switch on the upper
0.75 0.1 nm surface of the base and automatically by the computer
during the tip–engage and tip–withdraw processes.
0 nm The scan sizes available for these instruments are
0.5
0.7 μm, 12 μm and 125 μm. The scan rate must be de-
creased as the scan size is increased. A maximum scan
0.25 rate of 122 Hz can be used. Scan rates of about 60 Hz
should be used for small scan lengths (0.7 μm). Scan
rates of 0.5 to 2.5 Hz should be used for large scans
0 on samples with tall features. High scan rates help re-
0 0.25 0.5 0.75 1 nm
duce drift, but they can only be used on flat samples
b) 3 0.5 nm with small scan sizes. The scan rate or the scanning
speed (length/time) in the fast scan direction is equal
to twice the scan length multiplied by the scan rate in
0.25 nm
Hz, and in the slow direction it is equal to the scan
2
length multiplied by the scan rate in Hz divided by
0 nm number of data points in the transverse direction. For
example, for a scan size of 10 μm × 10 μm scanned at
1 0.5 Hz, the scan rates in the fast and slow scan direc-
tions are 10 μm/s and 20 nm/s, respectively. Normally
256 × 256 data points are taken for each image. The lat-
eral resolution at larger scans is approximately equal
0
0 1 2 3 nm
to the scan length divided by 256. The piezo tube re-
quires x–y-calibration, which is carried out by imaging
Fig. 21.13a,b Typical AFM images of freshly-cleaved an appropriate calibration standard. Cleaved graphite is
(a) highly oriented pyrolytic graphite and (b) mica surfaces used for small scan heads, while two-dimensional grids
taken using a square pyramidal Si3 N4 tip (a gold-plated rule) can be used for long-range heads.
Examples of AFM images of freshly cleaved highly The multimode AFM, when used with a grounded
Part C 21.2
oriented pyrolytic (HOP) graphite and mica surfaces are conducting tip, can be used to measure electric field
shown in Fig. 21.13 [21.50,110,114]. Images with near- gradients by oscillating the tip near its resonant fre-
atomic resolution are obtained. quency. When the lever encounters a force gradient
The force calibration mode is used to study inter- from the electric field, the effective spring constant
actions between the cantilever and the sample surface. of the cantilever is altered, changing its resonant fre-
In the force calibration mode, the x- and y-voltages quency. Depending on which side of the resonance
applied to the piezo tube are held at zero and a saw- curve is chosen, the oscillation amplitude of the can-
tooth voltage is applied to the z-electrode of the piezo tilever increases or decreases due to the shift in the
tube (Fig. 21.14a). At the start of the force measure- resonant frequency. By recording the amplitude of the
ment the cantilever is in its rest position. By changing cantilever, an image revealing the strength of the electric
the applied voltage, the sample can be moved up and field gradient is obtained.
down relative to the stationary cantilever tip. As the In the magnetic force microscope (MFM), used with
piezo moves the sample up and down, the cantilever de- a magnetically coated tip, static cantilever deflection is
flection signal from the photodiode is monitored. The detected when a magnetic field exerts a force on the
force–distance curve, a plot of the cantilever tip deflec- tip, and MFM images of magnetic materials can be
tion signal as a function of the voltage applied to the obtained. MFM sensitivity can be enhanced by oscil-
piezo tube, is obtained. Figure 21.14b shows the typi- lating the cantilever near its resonant frequency. When
cal features of a force–distance curve. The arrowheads the tip encounters a magnetic force gradient, the effec-
indicate the direction of piezo travel. As the piezo ex- tive spring constant (and hence the resonant frequency)
tends, it approaches the tip, which is in mid-air at this is shifted. By driving the cantilever above or below the
point and hence shows no deflection. This is indicated resonant frequency, the oscillation amplitude varies as
by the flat portion of the curve. As the tip approaches the resonance shifts. An image of the magnetic field gra-
the sample to within a few nanometers (point A), an dient is obtained by recording the oscillation amplitude
attractive force kicks in between the atoms of the tip as the tip is scanned over the sample.
surface and the atoms of the surface of the sample. Topographic information is separated from the elec-
The tip is pulled towards the sample and contact oc- tric field gradient and magnetic field images using the
curs at point B on the graph. From this point on, the so-called lift mode. In lift mode, measurements are
tip is in contact with the surface, and as the piezo taken in two passes over each scan line. In the first pass,
extends further, the tip gets deflected further. This is topographical information is recorded in the standard
represented by the sloped portion of the curve. As the tapping mode, where the oscillating cantilever lightly
piezo retracts, the tip moves beyond the zero deflec- taps the surface. In the second pass, the tip is lifted
tion (flat) line due to attractive forces (van der Waals to a user-selected separation (typically 20–200 nm) be-
forces and long-range meniscus forces), into the ad- tween the tip and local surface topography. By using
hesive regime. At point C in the graph, the tip snaps stored topographical data instead of standard feedback,
free of the adhesive forces, and is again in free air. the tip–sample separation can be kept constant. In this
The horizontal distance between points B and C along way, the cantilever amplitude can be used to mea-
the retrace line gives the distance moved by the tip in sure electric field force gradients or relatively weak but
the adhesive regime. Multiplying this distance by the long-range magnetic forces without being influenced by
stiffness of the cantilever gives the adhesive force. In- topographic features. Two passes are made for every
cidentally, the horizontal shift between the loading and scan line, producing separate topographic and magnetic
unloading curves results from the hysteresis in the PZT force images.
tube [21.4].
Electrochemical AFM
Multimode Capabilities This option allows one to perform electrochemical
The multimode AFM can be used for topography mea- reactions on the AFM. The technique involves a poten-
surements in the contact mode and tapping mode, tiostat, a fluid cell with a transparent cantilever holder
described earlier, and for measurements of lateral (fric- and electrodes, and the software required to operate the
tion) force, electric force gradients and magnetic force potentiostat and display the results of the electrochemi-
gradients. cal reaction.
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.2 Atomic Force Microscope 587
21.2.3 AFM Probe Construction cantilevers with reduced lateral rigidity are preferred.
Part C 21.2
A sharp protruding tip must be present at the end of the
Various probes (cantilevers and tips) are used for cantilever to provide a well-defined interaction with the
AFM studies. The cantilever stylus used in the AFM sample over a small area. The tip radius should be much
should meet the following criteria: (1) low normal smaller than the radii of the corrugations in the sample
spring constant (stiffness); (2) high resonant frequency; in order for these to be measured accurately. The lat-
(3) high cantilever quality factor Q; (4) high lateral eral spring constant depends critically on the tip length.
spring constant (stiffness); (5) short cantilever length; Additionally, the tip should be centered at the free end.
(6) incorporation of components (such as mirror) for de- In the past, cantilevers have been cut by hand from
flection sensing; and (7) a sharp protruding tip [21.137]. thin metal foils or formed from fine wires. Tips for these
In order to register a measurable deflection with small cantilevers were prepared by attaching diamond frag-
forces, the cantilever must flex with a relatively low ments to the ends of the cantilevers by hand, or in the
force (on the order of few nN), requiring vertical spring case of wire cantilevers, electrochemically etching the
constants of 10−2 to 102 N/m for atomic resolution in wire to a sharp point. Several cantilever geometries for
the contact profiling mode. The data rate or imaging wire cantilevers have been used. The simplest geome-
rate in the AFM is limited by the mechanical resonant try is the L-shaped cantilever, which is usually made by
frequency of the cantilever. To achieve a large imaging bending a wire at a 90◦ angle. Other geometries include
bandwidth, the AFM cantilever should have a resonant single-V and double-V geometries, with a sharp tip at-
frequency of more than about 10 kHz (30–100 kHz is tached at the apex of the V, and double-X configuration
preferable), which makes the cantilever the least sen- with a sharp tip attached at the intersection [21.31,138].
sitive part of the system. Fast imaging rates are not These cantilevers can be constructed with high vertical
just a matter of convenience, since the effects of ther- spring constants. For example, a double-cross cantilever
mal drifts are more pronounced with slow scanning with an effective spring constant of 250 N/m was used
speeds. The combined requirements of a low spring by Burnham and Colton [21.31]. The small size and low
constant and a high resonant frequency are met by re- mass needed in the AFM make hand fabrication of the
ducing the mass of the cantilever. The quality factor Q cantilever a difficult process with poor reproducibility.
(= ωR /(c/m), where ωR is the resonant frequency of the Conventional microfabrication techniques are ideal for
damped oscillator, c is the damping constant and m is constructing planar thin-film structures which have sub-
the mass of the oscillator) should have a high value for micron lateral dimensions. The triangular (V-shaped)
some applications. For example, resonance curve detec- cantilevers have improved (higher) lateral spring con-
tion is a sensitive modulation technique for measuring stants in comparison to rectangular cantilevers. In terms
small force gradients in noncontact imaging. Increas- of spring constants, the triangular cantilevers are ap-
ing the Q increases the sensitivity of the measurements. proximately equivalent to two rectangular cantilevers
Mechanical Q values of 100–1000 are typical. In con- placed in parallel [21.137]. Although the macroscopic
tact modes, the Q value is of less importance. A high radius of a photolithographically patterned corner is
lateral cantilever spring constant is desirable in order seldom much less than about 50 nm, microscopic asper-
to reduce the effect of lateral forces in the AFM, as ities on the etched surface provide tips with near-atomic
frictional forces can cause appreciable lateral bending dimensions.
of the cantilever. Lateral bending results in erroneous Cantilevers have been used from a whole range
topography measurements. For friction measurements, of materials. Cantilevers made of Si3 N4 , Si, and dia-
c) Diamond tip
bonded with epoxy
Table 21.3 Measured vertical spring constants and natural Table 21.4 Vertical (k z ), lateral (k y ), and torsional (k yT ) spring con-
Part C 21.2
frequencies of triangular (V-shaped) cantilevers made of stants of rectangular cantilevers made of Si (IBM) and PECVD
PECVD Si3 N4 (data provided by Digital Instruments, Inc.) Si3 N4 (source: Veeco Instruments, Inc.)
a)
100 nm
200 nm
21.2.4 Friction Measurement Methods angle corresponds to the sample scan in the y-direction,
Part C 21.2
and the 90◦ scanning angle corresponds to the sample
The two methods for performing friction measure- scan perpendicular to this axis in the x–y-plane (along
ments that are based on the work by Ruan and x-axis). If both the y- and −y-directions are scanned,
Bhushan [21.136] are now described in more detail we call this a parallel scan. Similarly, a perpendicu-
(also see [21.8]). The scanning angle is defined as the lar scan means that both the x- and −x-directions are
angle relative to the y-axis in Fig. 21.19a. This is also scanned. The direction of sample travel for each of these
the long axis of the cantilever. The zero-degree scanning two methods is illustrated in Fig. 21.19b.
Using method 1 (height mode with parallel scans)
in addition to topographic imaging, it is also possible
a) to measure friction force when the sample scanning di-
Cantilever rection is parallel to the y-direction (parallel scan). If
Laser y
substrate there was no friction force between the tip and the mov-
beam spot
ing sample, the topographic feature would be the only
x factor that would cause the cantilever to be deflected
Flexible cantilever
vertically. However, friction force does exist on all sur-
Sample traveling Sample traveling faces that are in contact where one of the surfaces is
direction in method 1 direction in method 2 moving relative to the other. The friction force between
b) the sample and the tip will also cause the cantilever to
Method 1 be deflected. We assume that the normal force between
Cantilever
Photodetector normal direction
the sample and the tip is W0 when the sample is sta-
tionary (W0 is typically 10 to 200 nN), and the friction
T Incident beam force between the sample and the tip is Wf as the sam-
L R Cantilever
ple is scanned by the tip. The direction of the friction
substrate force (Wf ) is reversed as the scanning direction of the
sample is reversed from the positive (y) to the negative
B
(−y) direction (Wf(y) = −Wf(−y) ).
When the vertical cantilever deflection is set at
Reflected
beam Tip Cantilever a constant level, it is the total force (normal force and
friction force) applied to the cantilever that keeps the
Traveling direction of the sample (y)
a)
Method 2 L L
l l
Wf P Wf P
T W0 W0 –ΔW1
Sliding direction y
L R of the sample
Tip b)
B
Twisted cantilever
Traveling direction of the sample (x) Wf P Wf P
W0 W0 + ΔW2
Fig. 21.19 (a) Schematic defining the x- and y-directions
relative to the cantilever, and showing the direction of sam- Sliding direction y
ple travel in two different measurement methods discussed of the sample
in the text. (b) Schematic of deformation of the tip and Fig. 21.20 (a) Schematic showing an additional bending
cantilever shown as a result of sliding in the x- and y- of the cantilever due to friction force when the sample is
directions. A twist is introduced to the cantilever if the scanned in the y- or −y-directions (left). (b) This effect
scanning is performed in the x-direction ((b), lower part) can be canceled out by adjusting the piezo height using
(after [21.136]) a feedback circuit (right) (after [21.136])
592 Part C Scanning-Probe Microscopy
cantilever deflection at this level. Since the friction force There are adhesive and interatomic attractive forces
Part C 21.2
is directed in the opposite direction to the direction between the cantilever tip and the sample at all times.
of travel of the sample, the normal force will have to The adhesive force can be due to water from the capil-
be adjusted accordingly when the sample reverses its lary condensation and other contaminants present at the
traveling direction, so that the total deflection of the surface, which form meniscus bridges [21.4, 143, 144]
cantilever will remain the same. We can calculate the and the interatomic attractive force includes van der
difference in the normal force between the two direc- Waals attractions [21.18]. If these forces (and the ef-
tions of travel for a given friction force Wf . First, since fect of indentation too, which is usually small for rigid
the deflection is constant, the total moment applied to samples) can be neglected, the normal force W0 is then
the cantilever is constant. If we take the reference point equal to the initial cantilever deflection H0 multiplied
to be the point where the cantilever joins the cantilever by the spring constant of the cantilever. (ΔW1 + ΔW2 )
holder (substrate), point P in Fig. 21.20, we have the can be derived by multiplying the same spring constant
following relationship by the change in height of the piezo tube between the
two traveling directions (y- and −y-directions) of the
(W0 − ΔW1 )L + Wf sample. This height difference is denoted as (ΔH1 +
= (W0 + ΔW2 )L − Wf (21.1) ΔH2 ), shown schematically in Fig. 21.21. Thus, (21.4)
can be rewritten as
or
Wf (ΔH1 + ΔH2 ) L
μ= = . (21.5)
(ΔW1 + ΔW2 )L = 2Wf . (21.2) W0 H0 2
Since the vertical position of the piezo tube is af-
Thus
fected by the topographic profile of the sample surface
Wf = (ΔW1 + ΔW2 )L/(2) , (21.3)
in addition to the friction force being applied at the
tip, this difference must be found point-by-point at
where ΔW1 and ΔW2 are the absolute values of the the same location on the sample surface, as shown in
changes in normal force when the sample is traveling Fig. 21.21. Subtraction of point-by-point measurements
in the −y- and y-directions, respectively, as shown in may introduce errors, particularly for rough samples.
Fig. 21.20; L is the length of the cantilever; is the ver- We will come back to this point later. In addition, pre-
tical distance between the end of the tip and point P. cise measurements of L and (which should include the
The coefficient of friction (μ) between the tip and the cantilever angle) are also required.
sample is then given as If the adhesive force between the tip and the sam-
ple is large enough that it cannot be neglected, it should
Wf (ΔW1 + ΔW2 ) L be included in the calculation. However, determinations
μ= = . (21.4)
W0 W0 2 of this force can involve large uncertainties, which is
introduced into (21.5). An alternative approach is to
make the measurements at different normal loads and
PZT height H
to use Δ(H0 ) and Δ(ΔH1 + ΔH2 ) in (21.5). Another
(Δ W1 + Δ W2) = k(Δ H1 + ΔH2)
comment on (21.5) is that, since only the ratio between
(ΔH1 + ΔH2 ) and H0 enters this equation, the vertical
position of the piezo tube H0 and the difference in posi-
Δ H2 tion (ΔH1 + ΔH2 ) can be in volts as long as the vertical
H0 (ΔH1 + Δ H2) travel of the piezo tube and the voltage applied to have
Δ H1 a linear relationship. However, if there is a large non-
linearity between the piezo tube traveling distance and
the applied voltage, this nonlinearity must be included
in the calculation.
It should also be pointed out that (21.4) and (21.5)
Sliding distance y
are derived under the assumption that the friction force
Fig. 21.21 Schematic illustration of the height difference Wf is the same for the two scanning directions of the
for the piezoelectric tube scanner as the sample is scanned sample. This is an approximation, since the normal
in the y- and −y-directions force is slightly different for the two scans and the
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.2 Atomic Force Microscope 593
Part C 21.2
Path of the laser beam
on the photodetector the cantilever tip with the sample. The left and right
T detectors can be balanced beforehand by adjusting the
positions of the detectors so that the intensity differ-
ence between these two detectors is zero (FFM signal is
“Height” zero). Once the tip is engaged with the sample, this sig-
– AFM signal
nal is no longer zero, even if the sample is not moving in
L R the x–y-plane with no friction force applied. This would
be a detrimental effect. It has to be understood and elim-
“Aux”
inated from the data acquisition before any quantitative
– FFM measurement of friction force is made.
signal One of the reasons for this observation is as follows.
B The detectors may not have been properly aligned with
respect to the laser beam. To be precise, the vertical
Vertical axis of axis of the detector assembly (the line joining T–B in
the photodetector
Fig. 21.22) is not in the plane defined by the incident
Fig. 21.22 The trajectory of the laser beam on the pho- laser beam and the beam reflected from the untwisted
todetectors as the cantilever is vertically deflected (with cantilever (we call this plane the beam plane). When the
no torsional motion) with respect to the laser beam for cantilever vertical deflection changes due to a change
a misaligned photodetector. For a change of normal force in the normal force applied (without the sample being
(vertical deflection of the cantilever), the laser beam is scanned in the x–y-plane), the laser beam will be re-
projected to a different position on the detector. Due to flected up and down and form a projected trajectory on
a misalignment, the projected trajectory of the laser beam the detector. (Note that this trajectory is in the defined
on the detector is not parallel with the detector vertical axis beam plane.) If this trajectory is not coincident with
(the line T–B) (after [21.136]) the vertical axis of the detector, the laser beam will not
evenly bisect the left and right quadrants of the detec-
friction may be direction-dependent. However, this dif- tors, even under the condition of no torsional motion of
ference is much smaller than W0 itself. We can ignore the cantilever (Fig. 21.22). Thus, when the laser beam
the second-order correction. is reflected up and down due a change in the normal
Method 2 (aux mode with perpendicular scan) force, the intensity difference between the left and right
of measuring friction was suggested by Meyer and detectors will also change. In other words, the FFM sig-
Amer [21.8]. The sample is scanned perpendicular to nal will change as the normal force applied to the tip is
the long axis of the cantilever beam (along the x- or changed, even if the tip is not experiencing any friction
−x-direction in Fig. 21.19a) and the outputs from the force. This (FFM) signal is unrelated to friction force
two horizontal quadrants of the photodiode detector are or to the actual twisting of the cantilever. We will call
measured. In this arrangement, as the sample moves un- this part of the FFM signal FFMF , and the part which is
der the tip, the friction force will cause the cantilever to truly related to friction force FFMT .
twist. Therefore, the light intensity between the left and The FFMF signal can be eliminated. One way of do-
right (L and R in Fig. 21.19b, right) detectors will be dif- ing this is as follows. First the sample is scanned in both
ferent. The differential signal between the left and right the x- and the −x-directions and the FFM signals for
detectors is denoted the FFM signal [(L − R)/(L + R)]. scans in each direction are recorded. Since the friction
This signal can be related to the degree of twisting, and force reverses its direction of action when the scanning
hence to the magnitude of friction force. Again, because direction is reversed from the x- to the −x-direction, the
possible errors in measurements of the normal force due FFMT signal will change signs as the scanning direction
to the presence of adhesive force at the tip–sample in- of the sample is reversed (FFMT (x) = −FFMT (−x)).
terface, the slope of the friction data (FFM signal versus Hence the FFMT signal will be canceled out if we take
normal load) needs to be measured for an accurate value the sum of the FFM signals for the two scans. The aver-
of the coefficient of friction. age value of the two scans will be related to FFMF due
While friction force contributes to the FFM sig- to the misalignment,
nal, friction force may not be the only contributing
factor in commercial FFM instruments (for example, FFM(x) + FFM(−x) = 2FFMF . (21.6)
594 Part C Scanning-Probe Microscopy
This value can therefore be subtracted from the original to calculate the absolute values of normal and fric-
Part C 21.2
FFM signals of each of these two scans to obtain the tion forces in Newtons using the measured AFM and
true FFM signal (FFMT ). Or, alternately, by taking the FFMT voltage signals, it is necessary to first have an
difference of the two FFM signals, one gets the FFMT accurate value of the spring constant of the cantilever
value directly (kc ). The spring constant can be calculated using the
geometry and the physical properties of the cantilever
FFM(x) − FFM(−x) = FFMT (x) − FFMT (−x) material [21.8, 94, 137]. However, the properties of the
= 2FFMT (x) . (21.7) PECVD Si3 N4 (used to fabricate cantilevers) can be
different from those of the bulk material. For exam-
Ruan and Bhushan [21.136] have shown that the er- ple, using ultrasonics, we found the Young’s modulus
ror signal (FFMF ) can be very large compared to the of the cantilever beam to be about 238 ± 18 GPa, which
friction signal FFMT , so correction is required. is less than that of bulk Si3 N4 (310 GPa). Furthermore,
Now we compare the two methods. The method of the thickness of the beam is nonuniform and difficult to
using the height mode and parallel scanning (method 1) measure precisely. Since the stiffness of a beam goes as
is very simple to use. Technically, this method can the cube of thickness, minor errors in precise measure-
provide 3-D friction profiles and the corresponding to- ments of thickness can introduce substantial stiffness
pographic profiles. However, there are some problems errors. Thus one should measure the spring constant of
with this method. Under most circumstances, the piezo the cantilever experimentally. Cleveland et al. [21.145]
scanner displays hysteresis when the traveling direction measured normal spring constants by measuring reso-
of the sample is reversed. Therefore, the measured sur- nant frequencies of beams.
face topographic profiles will be shifted relative to each For normal spring constant measurement, Ruan and
other along the y-axis for the two opposite (y and −y) Bhushan [21.136] used a stainless steel spring sheet of
scans. This would make it difficult to measure the local known stiffness (width = 1.35 mm, thickness = 15 μm,
difference in height of the piezo tube for the two scans. free hanging length = 5.2 mm). One end of the spring
However, the average difference in height between the was attached to the sample holder and the other end was
two scans and hence the average friction can still be made to contact with the cantilever tip during the mea-
measured. The measurement of average friction can surement (Fig. 21.23). They measured the piezo travel
serve as an internal means of friction force calibration. for a given cantilever deflection. For a rigid sample
Method 2 is a more desirable approach. The subtraction (such as diamond), the piezo travel Z t (measured from
of the FFMF signal from FFM for the two scans does the point where the tip touches the sample) should equal
not introduce any error into local friction force data. An
ideal approach when using this method would be to add
the average values of the two profiles in order to get the a) Sample traveling Cantilever kc
distance
error component (FFMF ) and then subtract this compo-
nent from either profile to get true friction profiles in
Zt
either directions. By performing measurements at vari-
ous loads, we can get the average value of the coefficient Rigid sample
of friction which then can be used to convert the fric-
tion profile to the coefficient of friction profile. Thus,
any directionality and local variations in friction can be
easily measured. In this method, since topography data b) Sample traveling
distance Cantilever kc
are not affected by friction, accurate topography data
can be measured simultaneously with friction data and
a better localized relationship between the two can be Z t' Zt
established.
Flexible spring ks
Sample
21.2.5 Normal Force and Friction Force holder PZT tube scanner
Calibrations of Cantilever Beams
Fig. 21.23a,b Illustration showing the deflection of the
Based on Ruan and Bhushan [21.136], we now discuss cantilever as it is pushed by (a) a rigid sample, (b) a flexible
normal force and friction force calibrations. In order spring sheet (after [21.136])
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.3 AFM Instrumentation and Analyses 595
the cantilever deflection. To maintain the cantilever de- the cantilever [21.53,144]. One would need information
Part C 21.3
flection at the same level using a flexible spring sheet, about the detector such as its quantum efficiency, laser
the new piezo travel Z t would need to be different from power, gain and so on in order to be able convert the
Z t . The difference between Z t and Z t corresponds to signal into the degree of twisting. Generally speaking,
the deflection of the spring sheet. If the spring con- this procedure can not be accomplished without having
stant of the spring sheet is ks , the spring constant of the some detailed information about the instrument. This
cantilever kc can be calculated by information is not usually provided by the manufac-
(Z t − Z t )ks = Z t kc turer. Even if this information is readily available, errors
may still occur when using this approach because there
or
will always be variations as a result of the instrumental
kc = ks (Z t − Z t )/Z t . (21.8) set-up. For example, it has been noticed that the meas-
The spring constant of the spring sheet (ks ) used in this ured FFMT signal varies for the same sample when dif-
study is calculated to be 1.54 N/m. For the wide-legged ferent AFM microscopes from the same manufacturer
cantilever used in our study (length = 115 μm, base are used. This means that one can not calibrate the in-
width = 122 μm, leg width = 21 μm and thickness = strument experimentally using this calculation. O’Shea
0.6 μm), kc was measured to be 0.40 N/m instead of the et al. [21.144] did perform a calibration procedure in
0.58 N/m reported by its manufacturer – Digital Instru- which the torsional signal was measured as the sample
ments, Inc. To relate the photodiode detector output to was displaced a known distance laterally while ensuring
the cantilever deflection in nanometers, they used the that the tip did not slide over the surface. However, it is
same rigid sample to push against the AFM tip. Since difficult to verify that tip sliding does not occur.
the cantilever vertical deflection equals the sample trav- A new method of calibration is therefore required.
eling distance measured from the point where the tip There is a simpler, more direct way of doing this. The
touches the sample for a rigid sample, the photodiode first method described above (method 1) of measuring
output observed as the tip is pushed by the sample can friction can provide an absolute value of the coefficient
be converted directly to the cantilever deflection. For of friction directly. It can therefore be used as an internal
these measurements, they found the conversion factor calibration technique for data obtained using method 2.
to be 20 nm/V. Or, for a polished sample, which introduces the least er-
The normal force applied to the tip can be calcu- ror into friction measurements taken using method 1,
lated by multiplying the cantilever vertical deflection method 1 can be used to calibrate the friction force for
by the cantilever spring constant for samples that have method 2. Then this calibration can be used for mea-
very small adhesion with the tip. If the adhesive force surements taken using method 2. In method 1, the length
between the sample and the tip is large, it should be of the cantilever required can be measured using an op-
included in the normal force calculation. This is par- tical microscope; the length of the tip can be measured
ticularly important in atomic-scale force measurements, using a scanning electron microscope. The relative an-
because the typical normal force that is measured in gle between the cantilever and the horizontal sample
this region is in the range of a few hundreds of nN to surface can be measured directly. This enables the co-
a few mN. The adhesive force could be comparable to efficient of friction to be measured with few unknown
the applied force. parameters. The friction force can then be calculated
The conversion of friction signal (from FFMT ) to by multiplying the coefficient of friction by the normal
friction force is not as straightforward. For example, one load. The FFMT signal obtained using method 2 is then
can calculate the degree of twisting for a given friction converted into the friction force. For their instrument,
force using the geometry and the physical properties of they found the conversion to be 8.6 nN/V.
Stiffness and Resonances Ebh 3
of Lumped Mass Systems ω0 = . (21.12)
4L 3 m eff
All of the building blocks of an AFM, including the
body of the microscope itself and the force-measuring The effective mass can be calculated using Raleigh’s
cantilevers, are mechanical resonators. These reso- method. The general formula using Raleigh’s method
nances can be excited either by the surroundings or by for the kinetic energy T of a bar is
the rapid movement of the tip or the sample. To avoid
problems due to building- or air-induced oscillations, it L 2
1 m ∂z(x)
is of paramount importance to optimize the design of T= dx . (21.13)
the AFM for high resonant frequencies. This usually 2 L ∂t
0
means decreasing the size of the microscope [21.146].
By using cube-like or sphere-like structures for the For the case of a uniform beam with a constant cross
microscope, one can considerably increase the lowest section and length L, one obtains for the deflection
eigenfrequency. The fundamental natural frequency ω0 z(x) = z max 1 − (3x/2L) + (x 3 /2L 3 ) . Inserting z max
of any spring is given by into (21.13) and solving the integral gives
1 k L 3 2
ω0 = , (21.9) 1 m ∂z max (x) 3x x
2π m eff T= 1− + dx
2 L ∂t 2L L3
where k is the spring constant (stiffness) in the normal 0
direction and m eff is the effective mass. The spring con- 1
= m eff (z max t)2 ,
stant k of a cantilever beam with uniform cross section 2
(Fig. 21.24) is given by [21.147] which gives
3E I
k= 3 , (21.10) 9
L m eff = m. (21.14)
where E is the Young’s modulus of the material, L is 20
the length of the beam and I is the moment of inertia Substituting (21.14) into (21.12) and noting that
of the cross section. For a rectangular cross section with m = ρLbh, where ρ is the mass density, one obtains the
a width b (perpendicular to the deflection) and a height following expression
h one obtains the following expression for I
√
bh 3 5 E h
I= . (21.11) ω0 = . (21.15)
12 3 ρ L2
Stiffness and Resonances of Cantilevers and it will twist. The stiffness in the lateral (y) direc-
Part C 21.3
Cantilevers are mechanical devices specially shaped to tion k y can be calculated with (21.23) by exchanging b
measure tiny forces. The analysis given in the previ- and h
ous section is applicable. However, to better understand Eh b 3
the intricacies of force detection systems, we will dis- ky = . (21.25)
4 L
cuss the example of a cantilever beam with uniform Therefore, the bending stiffness in the lateral direction
cross section (Fig. 21.24). The bending of a beam due is larger than the stiffness for bending in the normal di-
to a normal load on the beam is governed by the Euler rection by (b/h)2 . The twisting or torsion on the other
equation [21.147] hand is more complicated to handle. For a wide, thin
d2 z cantilever (b h) we obtain torsional stiffness along
M = E I (x) , (21.16) y-axis k yT
dx 2
where M is the bending moment acting on the beam Gbh 3
k yT = , (21.26)
cross section. I (x) is the moment of inertia of the cross 3L2
section with respect to the neutral axis, defined by where G is the modulus of rigidity (= E/2(1 + ν); ν is
Poisson’s ratio). The ratio of the torsional stiffness to
I (x) = z 2 dy dz . (21.17) the lateral bending stiffness is
z y k yT 1 b 2
= , (21.27)
For a normal force Fz acting at the tip, ky 2 hL
M(x) = (L − x) Fz (21.18) where we assume ν = 0.333. We see that thin, wide can-
tilevers with long tips favor torsion while cantilevers
since the moment must vanish at the endpoint of the with square cross sections and short tips favor bend-
cantilever. Integrating (21.16) for a normal force Fz act- ing. Finally, we calculate the ratio between the torsional
ing at the tip and observing that E I is a constant for stiffness and the normal bending stiffness,
beams with a uniform cross section, one gets 2
k yT L
L 3 x
2 x
=2 . (21.28)
z(x) = 3− Fz . (21.19)
kz
6E I L L Equations (21.26) to (21.28) hold in the case where the
The slope of the beam is cantilever tip is exactly in the middle axis of the can-
Lx x
tilever. Triangular cantilevers and cantilevers with tips
z (x) = 2− Fz . (21.20) which are not on the middle axis can be dealt with by
2E I L
finite element methods.
From (21.19) and (21.20), at the end of the cantilever
The third possible deflection mode is the one from
(for x = L), for a rectangular beam, and by using an
the force on the end of the tip along the cantilever axis,
expression for I in (21.11), one gets
3 Fx (Fig. 21.24). The bending moment at the free end of
4 L the cantilever is equal to Fx . This leads to the following
z(L) = Fz , (21.21)
modification of (21.18) for forces Fz and Fx
Eb h
3z
M (x) = (L − x) Fz + Fx . (21.29)
z (L) = . (21.22)
2 L Integration of (21.16) now leads to
Now, the stiffness in the normal (z) direction k z is 1 2 x
z(x) = Lx 1 − Fz + x 2 Fx (21.30)
Fz Eb h 3 2E I 3L
kz = = . (21.23) and
z(L) 4 L
1 Lx x
and the change in angular orientation of the end of can- z (x) = 2− Fz + xFx . (21.31)
EI 2 L
tilever beam is Evaluating (21.30) and (21.31) at the end of the can-
2
3z 6 L tilever, we get the deflection and the tilt
Δα = = Fz . (21.24)
2L Ebh h L2 L
z (L) = Fz − Fx ,
Now we ask what will, to a first-order approxima- EI 3 2
tion, happen if we apply a lateral force Fy to the end of L L
z (L) = Fz + Fx . (21.32)
the tip (Fig. 21.24). The cantilever will bend sideways EI 2
598 Part C Scanning-Probe Microscopy
From these equations, one gets where Q is the quality factor of the cantilever, described
Part C 21.3
Part C 21.3
present in the heterodyne detection scheme shown in
Ω02
Δz (Ω) = Δz 0 , (21.40) Fig. 21.26. Light incident from the left with a frequency
2 Ω 2 Ω 2 ω is split into a reference path (upper path in Fig. 21.26)
Ω 2 − Ω02 + Q 2 0
and a measurement path. Light in the measurement path
is shifted in frequency to ω1 = ω + Δω and focused
where Δz 0 is the constant drive amplitude and Ω0
onto the cantilever. The cantilever oscillates at the fre-
the resonant frequency of the cantilever. The resonant
quency Ω, as in the homodyne detection scheme. The
frequency of the cantilever is given by the effective po-
reflected light A (t) is collimated by the same lens and
tential
interferes on the photodiode with the reference light
∂2U 1 Ar (t). The fluctuating term of the intensity is given by
Ω0 = k+ 2 , (21.41)
∂z m eff
2A (t) Ar (t)
where U is the interaction potential between the tip
4πδ
and the sample. Equation (21.41) shows that an attrac- = A,0 Ar,0 sin (ω + Δω) t +
λ
tive potential decreases Ω0 . The change in Ω0 in turn
results in a change in Δz (21.40). The movement of 4πΔz
+ sin (Ωt) sin (ωt) , (21.44)
the cantilever changes the path difference in the inter- λ
ferometer. The light reflected from the cantilever with
where the variables are defined as in (21.42). Setting the
amplitude A,0 and the reference light with amplitude
path difference sin(4πδ/λ) ≈ 0 and taking the time aver-
Ar,0 interfere on the detector. The detected intensity
age, omitting small quantities and linearizing functions
I (t) = [A (t) + Ar (t)]2 consists of two constant terms with small arguments, we get
and a fluctuating term
2A (t) Ar (t) 2A (t) Ar (t) T
4πδ 4πΔz 4πδ 4πΔz
= A,0 Ar,0 sin ωt+ + sin(Ωt) sin(ωt) . ∝ cos Δωt + + sin (Ωt)
λ λ λ λ
(21.42) 4πδ 4πΔz
= cos Δωt + cos sin (Ωt)
λ λ
Here ω is the frequency of the light, λ is the wavelength
4πδ 4πΔz
of the light, δ is the path difference in the interferometer, − sin Δωt + sin sin (Ωt)
and Δz is the instantaneous amplitude of the cantilever, λ λ
given according to (21.40) and (21.41) as a function
of Ω, k, and U. The time average of (21.42) then be- ω0
comes
4πδ 4πΔz
2A (t) Ar (t) T ∝ cos + sin (Ωt) Light Beam Sample
λ λ splitter Scan
ω0 ω1
piezo
4πδ 4πΔz
≈ cos − sin sin (Ωt)
λ λ
Beam Modulator Lens Cantilever
4πδ 4πΔz
≈ cos − sin (Ωt) . (21.43) splitter ω0 ω1
λ λ Cantilever
Here all small quantities have been omitted and func- drive piezo
Photodiode
tions with small arguments have been linearized. The
amplitude of Δz can be recovered with a lock-in Fig. 21.26 Principle of a heterodyne interferometric AFM. Light
technique. However, (21.43) shows that the measured with frequency ω0 is split into a reference path (upper path) and
amplitude is also a function of the path difference δ in a measurement path. The light in the measurement path is frequency
the interferometer. Hence, this path difference δ must shifted to ω1 by an acousto-optical modulator (or an electro-optical
be very stable. The best sensitivity is obtained when modulator). The light reflected from the oscillating cantilever inter-
sin(4δ/λ) ≈ 0. feres with the reference beam on the detector
600 Part C Scanning-Probe Microscopy
4πδ 4πΔz
Part C 21.3
Part C 21.3
Calcite Wollaston
disturbances like air currents have.
45°
Sarid [21.116] has given values for the sensitiv-
ities of different interferometeric detection systems.
Table 21.5 presents a summary of his results.
2-segment
Input beam diode Optical Lever
The most common cantilever deflection detection sys-
Fig. 21.28 Principle of Nomarski AFM. The circularly tem is the optical lever [21.53, 111]. This method,
polarized input beam is deflected to the left by a nonpolar- depicted in Fig. 21.29, employs the same technique as
izing beam splitter. The light is focused onto a cantilever. light beam deflection galvanometers. A fairly well colli-
The calcite crystal between the lens and the cantilever mated light beam is reflected off a mirror and projected
splits the circular polarized light into two spatially sepa- to a receiving target. Any change in the angular posi-
rated beams with orthogonal polarizations. The two light tion of the mirror will change the position where the
beams reflected from the lever are superimposed by the cal- light ray hits the target. Galvanometers use optical path
cite crystal and collected by the lens. The resulting beam lengths of several meters and scales projected onto the
is again circularly polarized. A Wollaston prism produces target wall are also used to monitor changes in position.
two interfering beams with a π/2 phase shift between In an AFM using the optical lever method, a pho-
them. The minimal path difference accounts for the excel- todiode segmented into two (or four) closely spaced
lent stability of this microscope devices detects the orientation of the end of the can-
tilever. Initially, the light ray is set to hit the photodiodes
select the ordinary ray or the extraordinary ray or one
can get any mixture of the two rays. A detailed de-
scription of birefringence can be found in textbooks Parallel D
plate Detector
(e.g., [21.150]). A calcite crystal deflects the extraor- Laser Δx
dinary ray at an angle of 6◦ within the crystal. Any
separation can be set by choosing a suitable length for
the calcite crystal.
The focus of one light ray is positioned near the free
end of the cantilever while the other is placed close
to the clamped end. Both arms of the interferometer Δz
L
pass through the same space, except for the distance be-
tween the calcite crystal and the lever. The closer the Fig. 21.29 Set-up for an optical lever detection microscope
Table 21.5 Noise in interferometers. F is the finesse of the cavity in the homodyne interferometer, Pi the incident power,
Pd is the power on the detector, η is the sensitivity of the photodetector and RIN is the relative intensity noise of the
laser. PR and PS are the power in the reference and sample beam in the heterodyne interferometer. P is the power in
the Nomarski interferometer, δθ is the phase difference between the reference and the probe beam in the Nomarski
interferometer. B is the bandwidth, e is the electron charge, λ is the wavelength of the laser, k the cantilever stiffness, ω0
is the resonant frequency of the cantilever, Q is the quality factor of the cantilever, T is the temperature, and δi is the
variation in current i
Homodyne Heterodyne Nomarski
interferometer, interferometer interferometer
fiber-optic interferometer
1 2 2 2
1 2 2
Laser noise δi 2 L η F Pi RIN η2 PR2 + PS2 RIN η P δθ
4 16
16π 2 2 2 2 4kB TBQ 4π 2 2 2 4kB TBQ π 2 2 2 4kB TBQ
Thermal noise δi 2 T η F Pi η Pd η P
λ2 ω0 k λ2 ω0 k λ2 ω0 k
1
Shot noise δi 2 S 4eηPd B 2eη (PR + PS ) B eηPB
2
602 Part C Scanning-Probe Microscopy
in the middle of the two subdiodes. Any deflection of of the quadrant detector and the cantilever. This rela-
Part C 21.3
the cantilever will cause an imbalance of the number of tion is true if the light spot is smaller than the quadrant
photons reaching the two halves. Hence the electrical detector. If it is greater, the difference current ΔI be-
currents in the photodiodes will be unbalanced too. The comes smaller with increasing distance. In reality, the
difference signal is further amplified and is the input light beam has a Gaussian intensity profile. For small
signal to the feedback loop. Unlike the interferometeric movements Δx (compared to the diameter of the light
AFMs, where a modulation technique is often necessary spot at the quadrant detector), (21.49) still holds. Larger
to get a sufficient signal-to-noise ratio, most AFMs em- movements Δx, however, will introduce a nonlinear re-
ploying the optical lever method are operated in a static sponse. If the AFM is operated in a constant force mode,
mode. AFMs based on the optical lever method are uni- only small movements Δx of the light spot will occur.
versally used. It is the simplest method for constructing The feedback loop will cancel out all other movements.
an optical readout and it can be confined in volumes that The scanning of a sample with an AFM can
are smaller than 5 cm in side length. twist the microfabricated cantilevers because of lateral
The optical lever detection system is a simple yet forces [21.5, 7, 8] and affect the images [21.120]. When
elegant way to detect normal and lateral force signals the tip is subjected to lateral forces, it will twist the
simultaneously [21.7, 8, 53, 111]. It has the additional cantilever and the light beam reflected from the end
advantage that it is a remote detection system. of the cantilever will be deflected perpendicular to the
ordinary deflection direction. For many investigations
Implementations. Light from a laser diode or from this influence of lateral forces is unwanted. The de-
a super luminescent diode is focused on the end of the sign of the triangular cantilevers stems from the desire
cantilever. The reflected light is directed onto a quad- to minimize the torsion effects. However, lateral forces
rant diode that measures the direction of the light beam. open up a new dimension in force measurements. They
A Gaussian light beam far from its waist is character- allow, for instance, two materials to be distinguished
ized by an opening angle β. The deflection of the light because of their different friction coefficients, or ad-
beam by the cantilever surface tilted by an angle α is 2α. hesion energies to be determined. To measure lateral
The intensity on the detector then shifts to the side by forces, the original optical lever AFM must be modi-
the product of 2α and the separation between the detec- fied. The only modification compared with Fig. 21.29
tor and the cantilever. The readout electronics calculates is the use of a quadrant detector photodiode instead of
the difference in the photocurrents. The photocurrents, a two-segment photodiode and the necessary readout
in turn, are proportional to the intensity incident on the electronics (Fig. 21.9a). The electronics calculates the
diode. following signals
The output signal is hence proportional to the
change in intensity on the segments Unormal force = α Iupper left + Iupper right
α − Ilower left + Ilower right ,
Isig ∝ 4 Itot . (21.47)
β Ulateral force = β Iupper left + Ilower left
For the sake of simplicity, we assume that the light beam
− Iupper right + Ilower right .
is of uniform intensity with its cross section increasing (21.50)
in proportion to the distance between the cantilever and
the quadrant detector. The movement of the center of The calculation of the lateral force as a function of
the light beam is then given by the deflection angle does not have a simple solution
D for cross sections other than circles. An approximate
ΔxDet = Δz . (21.48) formula for the angle of twist for rectangular beams
L
is [21.151]
The photocurrent generated in a photodiode is propor-
tional to the number of incoming photons hitting it. If Mt L
the light beam contains a total number of N0 photons, θ= , (21.51)
βGb3 h
then the change in difference current becomes
where Mt = Fy is the external twisting moment due
Δ (IR − IL ) = ΔI = const Δz D N0 . (21.49)
to lateral force Fy and β a constant determined by the
Combining (21.48) and (21.49), one obtains that the value of h/b. For the equation to hold, h has to be larger
difference current ΔI is independent of the separation than b.
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.3 AFM Instrumentation and Analyses 603
Inserting the values for a typical microfabricated where is the Planck constant (= 6.626 × 10−34 J s).
Part C 21.3
cantilever with integrated tips At very low temperatures and for high-frequency can-
tilevers this could become the dominant noise source.
b = 6 × 10−7 m , A second noise source is the shot noise of the light. The
h = 10−5 m , shot noise is related to the particle number. We can cal-
L = 10−4 m , culate the number of photons incident on the detector
using
= 3.3 × 10−6 m ,
Iτ Iλ I [W]
G = 5 × 1010 Pa , n= = = 1.8 × 109 , (21.57)
ω 2π B c B[Hz]
β = 0.333 (21.52)
where I is the intensity of the light, τ the measurement
into (21.51) we obtain the relation time, B = 1/τ the bandwidth, and c the speed of light.
The shot noise is proportional to the square root of the
Fy = 1.1 × 10−4 N × θ . (21.53) number of particles. Equating the shot noise signal with
Typical lateral forces are of the order of 10−10 N. the signal resulting from the deflection of the cantilever
one obtains
Sensitivity. The sensitivity of this set-up has been L B [kHz]
calculated in various papers [21.116, 148, 152]. Assum- Δz shot = 68 [fm] , (21.58)
ing a Gaussian beam, the resulting output signal as w I [mW]
a function of the deflection angle is dispersion-like. where w is the diameter of the focal spot. Typical AFM
Equation (21.47) shows that the sensitivity can be in- set-ups have a shot noise of 2 pm. The thermal noise
creased by increasing the intensity of the light beam Itot can be calculated from the equipartition principle. The
or by decreasing the divergence of the laser beam. The amplitude at the resonant frequency is
upper bound of the intensity of the light Itot is given by
saturation effects on the photodiode. If we decrease the B
Δz therm = 129 pm . (21.59)
divergence of a laser beam we automatically increase k N/m ω0 Q
the beam waist. If the beam waist becomes larger than
the width of the cantilever we start to get diffraction. A typical value is 16 pm. Upon touching the surface, the
Diffraction sets a lower bound on the divergence angle. cantilever increases its resonant frequency by a factor of
Hence one can calculate the optimal beam waist wopt 4.39. This results in a new thermal noise amplitude of
and the optimal divergence angle β [21.148, 152] 3.2 pm for the cantilever in contact with the sample.
ΔR
= Πt δt + Πl δl . (21.61)
R0 Fig. 21.32a,b Measuring the capacitance. (a) Low pass fil-
The maximum values of the stress components are Πt = ter, (b) capacitive divider. C (left) and C2 (right) are the
− 64.0 × 10−11 m2 /N and Πl = − 71.4 × 10−11 m2 /N capacitances under test
for a resistor oriented along the (110) direction in sil-
icon [21.126]. In the resistor arrangement of Fig. 21.30, parallel plates form a simple capacitor (Fig. 21.31, up-
two of the resistors are subject to the longitudinal piezo- per left), with capacitance
resistive effect and two of them are subject to the
transversal piezo-resistive effect. The sensitivity of that εε0 A
C= , (21.64)
set-up is about four times that of a single resistor, with x
the advantage that temperature effects cancel to first
order. The resistance change is then calculated as where A is the area of the plates, assumed equal, and x is
the separation. Alternatively one can consider a sphere
ΔR 3Eh 6L versus an infinite plane (Fig. 21.31, lower left). Here the
= Π 2 Δz = Π 2 Fz , (21.62)
R0 2L bh capacitance is [21.116]
where Π = 67.7 × 10−11 m2 /N is the averaged piezo- ∞
resistive coefficient. Plugging in typical values for sinh (α)
C = 4πε0 R (21.65)
the dimensions (Fig. 21.24) (L = 100 μm, b = 10 μm, sinh (nα)
n =2
h = 1 μm), one obtains
ΔR 4 × 10−5 where R is the radius of the sphere, and α is defined by
= Fz . (21.63) ⎛ ⎞
R0 nN
z z 2 z
The sensitivity can be tailored by optimizing the dimen- α = ln ⎝1 + + +2 ⎠ . (21.66)
sions of the cantilever. R R2 R
Capacitance Detection One has to bear in mind that the capacitance of a parallel
The capacitance of an arrangement of conductors plate capacitor is a nonlinear function of the separation.
depends on the geometry. Generally speaking, the ca- One can circumvent this problem using a voltage di-
pacitance increases for decreasing separations. Two vider. Figure 21.32a shows a low-pass filter. The output
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.3 AFM Instrumentation and Analyses 605
Part C 21.3
1 pacitance C1 (Fig. 21.33).
jωC 1 Another idea is to keep the distance constant and
Uout = U≈ = U≈
R + jωC
1 jωC R + 1 to change the relative overlap of the plates (Fig. 21.31,
U≈ right side). The capacitance of the moving center plate
∼
= . (21.67) versus the stationary outer plates becomes
jωC R
Here C is given by (21.64), ω is the excitation frequency εε0 bx
C = Cs + 2 , (21.71)
and j is the imaginary unit. The approximate relation at s
the end is true when ωC R 1. This is equivalent to the where the variables are defined in Fig. 21.31. The stray
statement that C is fed by a current source, since R must capacitance comprises all effects, including the ca-
be large in this set-up. Plugging (21.64) into (21.67) and pacitance of the fringe fields. When the length x is
neglecting the phase information, one obtains comparable to the width b of the plates, one can safely
U≈ x assume that the stray capacitance is constant and inde-
Uout = , (21.68) pendent of x. The main disadvantage of this set-up is
ωRεε0 A
that it is not as easily incorporated into a microfabri-
which is linear in the displacement x. cated device as the others.
Figure 21.32b shows a capacitive divider. Again the
output voltage Uout is given by Sensitivity. The capacitance itself is not a measure of
C1 C1 the sensitivity, but its derivative is indicative of the sig-
Uout = U≈ = U≈ εε0 A
. (21.69)
nals one can expect. Using the situation described in
C2 + C1 + C1
x
Fig. 21.31 (upper left) and in (21.64), one obtains for
If there is a stray capacitance Cs then (21.69) is modi- the parallel plate capacitor
fied as
dC εε0 A
C1 =− 2 . (21.72)
Uout = U≈ εε A . (21.70) dx x
x + Cs + C1
0
Alternatively, one could build an oscillator with this This is the frequency shift for a deflection of 1 Å. The
Part C 21.3
capacitance and measure the frequency. RC-oscillators calculation shows that this is a measurable quantity. The
typically have an oscillation frequency of quality also indicates that there is no physical reason
1 x why this scheme should not work.
f res ∝ = . (21.75)
RC Rεε0 A
21.3.3 Combinations
Again the resistance R must be of the order of
for 3-D Force Measurements
1 GΩ when stray capacitances Cs are neglected. How-
ever Cs is of the order of 1 pF. Therefore one gets Three-dimensional force measurements are essential if
R = 10 MΩ. Using these values, the sensitivity becomes one wants to know all of the details of the interaction
C dx 0.1 Hz between the tip and the cantilever. The straightfor-
d f res = ≈ dx . (21.76)
R (C + Cs ) x
2
Å ward attempt to measure three forces is complicated,
The bad thing is that the stray capacitances have made since force sensors such as interferometers or capaci-
the signal nonlinear again. The linearized set-up in tive sensors need a minimal detection volume, which is
Fig. 21.31 has a sensitivity of often too large. The second problem is that the force-
sensing tip has to be held in some way. This implies
dC εε0 b that one of the three Cartesian axes is stiffer than the
=2 . (21.77)
dx s others.
Substituting typical values (b = 10 μm, s = 1 μm), one However, by combining different sensors it is pos-
gets dC/ dx = 1.8 × 10−10 F/m. It is noteworthy that the sible to achieve this goal. Straight cantilevers are
sensitivity remains constant for scaled devices. employed for these measurements, because they can
be handled analytically. The key observation is that
Implementations. Capacitance readout can be achieved the optical lever method does not determine the po-
in different ways [21.123, 124]. All include an alternat- sition of the end of the cantilever. It measures the
ing current or voltage with frequencies in the 100 kHz to orientation. In the previous sections, one has always
100 MHz range. One possibility is to build a tuned cir- made use of the fact that, for a force along one of
cuit with the capacitance of the cantilever determining the orthogonal symmetry directions at the end of the
the frequency. The resonance frequency of a high- cantilever (normal force, lateral force, force along the
quality Q tuned circuit is cantilever beam axis), there is a one-to-one corre-
ω0 = (LC)−1/2 , (21.78)
spondence of the tilt angle and the deflection. The
problem is that the force along the cantilever beam
where L is the inductance of the circuit. The capaci- axis and the normal force create a deflection in the
tance C includes not only the sensor capacitance but same direction. Hence, what is called the normal
also the capacitance of the leads. The precision of a fre- force component is actually a mixture of two forces.
quency measurement is mainly determined by the ratio The deflection of the cantilever is the third quan-
of L and C tity, which is not considered in most of the AFMs.
1/2 A fiber-optic interferometer in parallel with the op-
L 1
Q= . (21.79) tical lever measures the deflection. Three measured
C R
quantities then allow the separation of the three or-
Here R symbolizes the losses in the circuit. The thonormal force directions, as is evident from (21.27)
higher the quality, the more precise the frequency mea- and (21.33) [21.12–16].
surement. For instance, a frequency of 100 MHz and Alternatively, one can put the fast scanning di-
a capacitance of 1 pF gives an inductance of 250 μH. rection along the axis of the cantilever. Forward and
The quality then becomes 2.5 × 108 . This value is an backward scans then exert opposite forces Fx . If the
upper limit, since losses are usually too high. piezo movement is linearized, both force components
Using a value of dC/ dx = 31 fF/μm, one gets in AFM based on optical lever detection can be de-
ΔC/Å = 3.1 aF/Å. With a capacitance of 1 pF, one gets termined. In this case, the normal force is simply
Δω 1 ΔC the average of the forces in the forward and back-
= ,
ω 2 C ward direction. The force Fx is the difference in
1 3.1aF the forces measured in the forward and backward
Δω = 100 MHz × = 155 Hz . (21.80)
2 1 pF directions.
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.3 AFM Instrumentation and Analyses 607
Part C 21.3
V
Almost all SPMs use piezo translators to scan the tip or ΔL = L (E · n) d31 = L d31 , (21.81)
t
the sample. Even the first STM [21.1, 103] and some of
its predecessors [21.153, 154] used them. Other mater- where d31 is the transverse piezoelectric constant, V is
ials or set-ups for nanopositioning have been proposed, the applied voltage, t is the thickness of the piezo slab
but they have not been successful [21.155, 156]. or the distance between the electrodes where the voltage
is applied, L is the free length of the piezo slab, and n is
Piezo Tubes
the direction of polarization. Piezo translators based on
A popular solution is tube scanners (Fig. 21.34). They the transverse piezoelectric effect have a wide range of
are now widely used in SPMs due to their simplicity and sensitivities, limited mainly by mechanical stability and
their small size [21.133,157]. The outer electrode is seg- breakdown voltage.
mented into four equal sectors of 90◦ . Opposite sectors
Scan Range
are driven by signals of the same magnitude, but oppo-
site sign. This gives, through bending, two-dimensional The scanning range of a piezotube is difficult to cal-
movement on (approximately) a sphere. The inner elec- culate [21.157–159]. The bending of the tube depends
trode is normally driven by the z-signal. It is possible, on the electric fields and the nonuniform strain induced.
however, to use only the outer electrodes for scanning A finite element calculation where the piezo tube was
and for the z-movement. The main drawback of apply- divided into 218 identical elements was used [21.158]
ing the z-signal to the outer electrodes is that the applied to calculate the deflection. On each node, the mechan-
voltage is the sum of both the x- or y-movements and ical stress, the stiffness, the strain and the piezoelectric
the z-movement. Hence a larger scan size effectively stress were calculated when a voltage was applied on
reduces the available range for the z-control. one electrode. The results were found to be linear on
the first iteration and higher order corrections were very
Piezo Effect
small even for large electrode voltages. It was found
An electric field applied across a piezoelectric material that, to first order, the x- and z-movement of the tube
causes a change in the crystal structure, with expansion could be reasonably well approximated by assuming
in some directions and contraction in others. Also, a net that the piezo tube is a segment of a torus. Using this
volume change occurs [21.132]. Many SPMs use the model, one obtains
transverse piezo electric effect, where the applied elec- L2
tric field E is perpendicular to the expansion/contraction dx = (V+ − V− ) |d31 | , (21.82)
2td
L
dz = (V+ + V− − 2Vz ) |d31 | , (21.83)
2t
where |d31 | is the coefficient of the transversal piezo-
electric effect, L is the tube’s free length, t is the tube’s
wall thickness, d is the tube’s diameter, V+ is the volt-
age on the positive outer electrode, while V− is the
voltage of the opposite quadrant negative electrode and
Vz is the voltage of the inner electrode.
–y +y The cantilever or sample mounted on the piezotube
has an additional lateral movement because the point of
+x
measurement is not in the endplane of the piezotube.
z inner The additional lateral displacement of the end of the tip
electrode is sin ϕ ≈ ϕ, where is the tip length and ϕ is the
deflection angle of the end surface. Assuming that the
Fig. 21.34 Schematic drawing of a piezoelectric tube scan- sample or cantilever is always perpendicular to the end
ner. The piezo ceramic is molded into a tube form. The of the walls of the tube, and calculating with the torus
outer electrode is separated into four segments and con- model, one gets for the angle
nected to the scanning voltage. The z-voltage is applied to L 2dx
the inner electrode ϕ= = , (21.84)
R L
608 Part C Scanning-Probe Microscopy
where R is the radius of curvature of the piezo tube. Us- equations demonstrate that the true x-travel is small at
Part C 21.3
ing the result of (21.84), one obtains for the additional the beginning of the scan and becomes larger towards
x-movement the end. Therefore, images are stretched at the begin-
2 dx ning and compressed at the end.
dxadd = ϕ = Similar equations hold for the slow scan direction.
L
L The coefficients, however, are different. The combined
= (V+ − V− ) |d31 | (21.85) action causes a greatly distorted image. This distortion
td
and for the additional z-movement due to the x-move- can be calculated. The data acquisition systems record
ment the signal as a function of V . However the data is meas-
ured as a function of x. Therefore we have to distribute
ϕ2 2 ( dx)2 the x-values evenly across the image. This can be done
dz add = − cos ϕ = =
2 L2 by inverting an approximation of (21.87). First we write
L 2
= (V+ − V− )2 |d31 |2 2 2 . (21.86) B
2t d x = AV 1 − V . (21.89)
A
Carr [21.158] assumed for his finite element calcula-
tions that the top of the tube was completely free to For B
A we can approximate
move and, as a consequence, the top surface was dis- x
torted, leading to a deflection angle that was about half V= . (21.90)
A
that of the geometrical model. Depending on the attach-
We now substitute (21.90) into the nonlinear term of
ment of the sample or the cantilever, this distortion may
(21.89). This gives
be smaller, leading to a deflection angle in-between that
of the geometrical model and the one from the finite Bx
element calculation. x = AV 1 + 2 ,
A
x 1 x Bx
Nonlinearities and Creep V= ≈ 1 − . (21.91)
A (1 + Bx/A2 ) A A2
Piezo materials with a high conversion ratio (a large
d31 or small electrode separations with large scanning Hence an equation of the type
ranges) are hampered by substantial hysteresis result-
xtrue = x (α − βx/xmax )
ing in a deviation from linearity by more than 10%.
The sensitivity of the piezo ceramic material (mechani- with 1 = α − β (21.92)
cal displacement divided by driving voltage) decreases takes out the distortion of an image. α and β are depen-
with reduced scanning range, whereas the hysteresis is dent on the scan range, the scan speed and on the scan
reduced. Careful selection of the material used for the history, and have to be determined with exactly the same
piezo scanners, the design of the scanners, and of the settings as for the measurement. xmax is the maximal
operating conditions is necessary to obtain optimum scanning range. The condition for α and β guarantees
performance. that the image is transformed onto itself.
Similar equations to the empirical one shown above
Passive Linearization: Calculation. The analysis of (21.92) can be derived by analyzing the movements of
images affected by piezo nonlinearities [21.160–163] domain walls in piezo ceramics.
shows that the dominant term is
x = AV + BV 2 , (21.87) Passive Linearization: Measuring the Position. An al-
ternative strategy is to measure the positions of the piezo
where x is the excursion of the piezo, V is the applied
translators. Several possibilities exist.
voltage and A and B are two coefficients describing
the sensitivity of the material. Equation (21.87) holds 1. The interferometers described above can be used
for scanning from V = 0 to large V . For the reverse to measure the elongation of the piezo elongation.
direction, the equation becomes The fiber-optic interferometer is especially easy to
implement. The coherence length of the laser only
x = ÃV − B̃ (V − Vmax )2 , (21.88)
limits the measurement range. However, the sig-
where à and B̃ are the coefficients for the back scan and nal is of a periodic nature. Hence direct use of the
Vmax is the applied voltage at the turning point. Both signal in a feedback circuit for the position is not
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes 21.3 AFM Instrumentation and Analyses 609
possible. However, as a measurement tool and, es- all of the systems described above, with the exception of
Part C 21.3
pecially, as a calibration tool, the interferometer is the interferometers, are suitable. The most common so-
without competition. The wavelength of the light, lutions include the strain gauge approach, capacitance
for instance that in a He-Ne laser, is so well de- measurement or the LVDT, which are all electronic so-
fined that the precision of the other components lutions. Optical detection systems have the disadvantage
determines the error of the calibration or measure- that the intensity enters into the calibration.
ment.
2. The movement of the light spot on the quad- Alternative Scanning Systems
rant detector can be used to measure the position The first STMs were based on piezo tripods [21.1]. The
of a piezo [21.164]. The output current changes piezo tripod (Fig. 21.35) is an intuitive way to gener-
by 0.5 A/cm × P(W)/R(cm). Typical values (P = ate the three-dimensional movement of a tip attached to
1 mW, R = 0.001 cm) give0.5 A/cm. The noise its center. However, to get a suitable stability and scan-
limit is typically 0.15 nm × Δ f (Hz)/H(W/cm2 ). ning range, the tripod needs to be fairly large (about
Again this means that the laser beam above would 50 mm). Some instruments use piezo stacks instead of
have a 0.1 nm noise limitation for a bandwidth of monolithic piezoactuators. They are arranged in a tri-
21 Hz. The advantage of this method is that, in prin- pod. Piezo stacks are thin layers of piezoactive materials
ciple, one can linearize two axes with only one glued together to form a device with up to 200 μm of
detector. actuation range. Preloading with a suitable metal casing
3. A knife-edge blocking part of a light beam incident reduces the nonlinearity.
on a photodiode can be used to measure the posi- If one tries to construct a homebuilt scanning
tion of the piezo. This technique, commonly used in system, the use of linearized scanning tables is recom-
optical shear force detection [21.75, 165], has a sen- mended. They are built around solid state joints and
sitivity of better than 0.1 nm. actuated by piezo stacks. The joints guarantee that the
4. The capacitive detection [21.166, 167] of the can- movement is parallel with little deviation from the pre-
tilever deflection can be applied to the measurement defined scanning plane. Due to the construction it is
of the piezo elongation. Equations (21.64) to (21.79) easy to add measurement devices such as capacitive
apply to the problem. This technique is used in some sensors, LVDTs or strain gauges, which are essential for
commercial instruments. The difficulties lie in the a closed loop linearization. Two-dimensional tables can
avoidance of fringe effects at the borders of the two be bought from several manufacturers. They have lin-
plates. While conceptually simple, one needs the lat- earities of better than 0.1% and a noise level of 10−4 to
est technology in surface preparation to get a decent 10−5 for the maximal scanning range.
linearity. The electronic circuits used for the readout
are often proprietary. Control Systems
5. Linear variable differential transformers (LVDT) Basics. The electronics and software play an important
are a convenient way to measure positions down role in the optimal performance of an SPM. Control
to 1 nm. They can be used together with a solid electronics and software are supplied with commer-
state joint set-up, as often used for large scan range cial SPMs. Electronic control systems can use either
stages. Unlike capacitive detection, there are few analog or digital feedback. While digital feedback of-
difficulties in implementation. The sensors and the
detection circuits LVDTs are available commer-
cially.
6. A popular measurement technique is the use of
strain gauges. They are especially sensitive when z
mounted on a solid state joint where the curvature
is maximal. The resolution depends mainly on the
x y
induced curvature. A precision of 1 nm is attainable.
The signals are low – a Wheatstone bridge is needed
for the readout.
Part C 21.3
y to external parameter
of the tip on the sample surface, together with the
High voltage
x-channel. amplifier y
y
determined with a fit. This function is appropriate for of the height along each scan line from each point in
Part C 21
almost planar data, where the nonlinearity of the piezos that scan line. This brings each scan line to the same
caused the distortion. height. Next, a first-order plane fit is applied in the fast
In the image processing software from Digital In- scan direction. The plane-fit operation is used to elimi-
struments, up to three operations are performed on the nate bow and slope in the fast scan direction. The plane
raw data. First, a zero-order flatten is applied. The flat- fit operation calculates a best fit plane for the image
ten operation is used to eliminate image bow in the slow and subtracts it from the image. This plane has a con-
scan direction (caused by a physical bow in the instru- stant nonzero slope in the fast scan direction. In some
ment itself), slope in the slow scan direction, and bands cases a higher order polynomial plane may be required.
in the image (caused by differences in the scan height Depending upon the quality of the raw data, the flatten-
from one scan line to the next). The flattening opera- ing operation and/or the plane fit operation may not be
tion takes each scan line and subtracts the average value required at all.
References
21.1 G. Binnig, H. Rohrer, C. Gerber, E. Weibel: Surface 21.13 S. Fujisawa, E. Kishi, Y. Sugawara, S. Morita: Fluc-
studies by scanning tunneling microscopy, Phys. tuation in 2-dimensional stick-slip phenomenon
Rev. Lett. 49, 57–61 (1982) observed with 2-dimensional frictional force mi-
21.2 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi- croscope, Jpn. J. Appl. Phys. 33, 3752–3755
croscope, Phys. Rev. Lett. 56, 930–933 (1986) (1994)
21.3 G. Binnig, C. Gerber, E. Stoll, T.R. Albrecht, 21.14 S. Grafstrom, J. Ackermann, T. Hagen, R. Neumann,
C.F. Quate: Atomic resolution with atomic force O. Probst: Analysis of lateral force effects on the
microscope, Europhys. Lett. 3, 1281–1286 (1987) topography in scanning force microscopy, J. Vac.
21.4 B. Bhushan: Handbook of Micro-/Nanotribology, Sci. Technol. B 12, 1559–1564 (1994)
2nd edn. (CRC, Boca Raton 1999) 21.15 R.M. Overney, H. Takano, M. Fujihira, W. Paulus,
21.5 C.M. Mate, G.M. McClelland, R. Erlandsson, S. Chi- H. Ringsdorf: Anisotropy in friction and molecular
ang: Atomic-scale friction of a tungsten tip on stick-slip motion, Phys. Rev. Lett. 72, 3546–3549
a graphite surface, Phys. Rev. Lett. 59, 1942–1945 (1994)
(1987) 21.16 R.J. Warmack, X.Y. Zheng, T. Thundat, D.P. Allison:
21.6 R. Erlandsson, G.M. McClelland, C.M. Mate, S. Chi- Friction effects in the deflection of atomic force
ang: Atomic force microscopy using optical inter- microscope cantilevers, Rev. Sci. Instrum. 65, 394–
ferometry, J. Vac. Sci. Technol. A 6, 266–270 (1988) 399 (1994)
21.7 O. Marti, J. Colchero, J. Mlynek: Combined scanning 21.17 N.A. Burnham, D.D. Domiguez, R.L. Mowery,
force and friction microscopy of mica, Nanotech- R.J. Colton: Probing the surface forces of mono-
nology 1, 141–144 (1990) layer films with an atomic force microscope, Phys.
21.8 G. Meyer, N.M. Amer: Simultaneous measurement Rev. Lett. 64, 1931–1934 (1990)
of lateral and normal forces with an optical-beam- 21.18 N.A. Burham, R.J. Colton, H.M. Pollock: Interpreta-
deflection atomic force microscope, Appl. Phys. tion issues in force microscopy, J. Vac. Sci. Technol.
Lett. 57, 2089–2091 (1990) A 9, 2548–2556 (1991)
21.9 B. Bhushan, J. Ruan: Atomic-scale friction meas- 21.19 C.D. Frisbie, L.F. Rozsnyai, A. Noy, M.S. Wrighton,
urements using friction force microscopy: Part II – C.M. Lieber: Functional group imaging by chemical
Application to magnetic media, ASME J. Tribol. 116, force microscopy, Science 265, 2071–2074 (1994)
389–396 (1994) 21.20 V.N. Koinkar, B. Bhushan: Microtribological stud-
21.10 B. Bhushan, V.N. Koinkar, J. Ruan: Microtribology ies of unlubricated and lubricated surfaces using
of magnetic media, Proc. Inst. Mech. Eng. Part J atomic force/friction force microscopy, J. Vac. Sci.
208, 17–29 (1994) Technol. A 14, 2378–2391 (1996)
21.11 B. Bhushan, J.N. Israelachvili, U. Landman: Nano- 21.21 V. Scherer, B. Bhushan, U. Rabe, W. Arnold: Lo-
tribology: Friction, wear, and lubrication at the cal elasticity and lubrication measurements using
atomic scale, Nature 374, 607–616 (1995) atomic force and friction force microscopy at ultra-
21.12 S. Fujisawa, M. Ohta, T. Konishi, Y. Sugawara, sonic frequencies, IEEE Trans. Magn. 33, 4077–4079
S. Morita: Difference between the forces measured (1997)
by an optical lever deflection and by an optical in- 21.22 V. Scherer, W. Arnold, B. Bhushan: Lateral force mi-
terferometer in an atomic force microscope, Rev. croscopy using acoustic friction force microscopy,
Sci. Instrum. 65, 644–647 (1994) Surf. Interface Anal. 27, 578–587 (1999)
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes References 613
21.23 B. Bhushan, S. Sundararajan: Micro-/nanoscale 21.39 I.W. Lyo, P. Avouris: Field-induced nanometer-to-
Part C 21
friction and wear mechanisms of thin films us- atomic-scale manipulation of silicon surfaces with
ing atomic force and friction force microscopy, Acta the STM, Science 253, 173–176 (1991)
Mater. 46, 3793–3804 (1998) 21.40 O.M. Leung, M.C. Goh: Orientation ordering of
21.24 U. Krotil, T. Stifter, H. Waschipky, K. Weishaupt, polymers by atomic force microscope tip-surface
S. Hild, O. Marti: Pulse force mode: A new method interactions, Science 225, 64–66 (1992)
for the investigation of surface properties, Surf. 21.41 D.W. Abraham, H.J. Mamin, E. Ganz, J. Clark: Sur-
Interface Anal. 27, 336–340 (1999) face modification with the scanning tunneling
21.25 B. Bhushan, C. Dandavate: Thin-film friction and microscope, IBM J. Res. Dev. 30, 492–499 (1986)
adhesion studies using atomic force microscopy, 21.42 R.M. Silver, E.E. Ehrichs, A.L. de Lozanne: Direct
J. Appl. Phys. 87, 1201–1210 (2000) writing of submicron metallic features with a scan-
21.26 B. Bhushan: Micro-/Nanotribology and Its Appli- ning tunnelling microscope, Appl. Phys. Lett. 51,
cations (Kluwer, Dordrecht 1997) 247–249 (1987)
21.27 B. Bhushan: Principles and Applications of Tribol- 21.43 A. Kobayashi, F. Grey, R.S. Williams, M. Ano: For-
ogy (Wiley, New York 1999) mation of nanometer-scale grooves in silicon with
21.28 B. Bhushan: Modern Tribology Handbook – Vol. 1: a scanning tunneling microscope, Science 259,
Principles of Tribology (CRC, Boca Raton 2001) 1724–1726 (1993)
21.29 B. Bhushan: Introduction to Tribology (Wiley, New 21.44 B. Parkinson: Layer-by-layer nanometer scale
York 2002) etching of two-dimensional substrates using the
21.30 M. Reinstädtler, U. Rabe, V. Scherer, U. Hart- scanning tunneling microscopy, J. Am. Chem. Soc.
mann, A. Goldade, B. Bhushan, W. Arnold: On the 112, 7498–7502 (1990)
nanoscale measurement of friction using atomic 21.45 A. Majumdar, P.I. Oden, J.P. Carrejo, L.A. Nagahara,
force microscope cantilever torsional resonances, J.J. Graham, J. Alexander: Nanometer-scale lithog-
Appl. Phys. Lett. 82, 2604–2606 (2003) raphy using the atomic force microscope, Appl.
21.31 N.A. Burnham, R.J. Colton: Measuring the nanome- Phys. Lett. 61, 2293–2295 (1992)
chanical properties and surface forces of materials 21.46 B. Bhushan: Micro-/nanotribology and its applica-
using an atomic force microscope, J. Vac. Sci. Tech- tions to magnetic storage devices and MEMS, Tribol.
nol. A 7, 2906–2913 (1989) Int. 28, 85–96 (1995)
21.32 P. Maivald, H.J. Butt, S.A.C. Gould, C.B. Prater, 21.47 L. Tsau, D. Wang, K.L. Wang: Nanometer scale pat-
B. Drake, J.A. Gurley, V.B. Elings, P.K. Hansma: Us- terning of silicon(100) surface by an atomic force
ing force modulation to image surface elasticities microscope operating in air, Appl. Phys. Lett. 64,
with the atomic force microscope, Nanotechnology 2133–2135 (1994)
2, 103–106 (1991) 21.48 E. Delawski, B.A. Parkinson: Layer-by-layer etch-
21.33 B. Bhushan, A.V. Kulkarni, W. Bonin, J.T. Wyrobek: ing of two-dimensional metal chalcogenides with
Nano/picoindentation measurements using cap- the atomic force microscope, J. Am. Chem. Soc. 114,
acitive transducer in atomic force microscopy, 1661–1667 (1992)
Philos. Mag. A 74, 1117–1128 (1996) 21.49 B. Bhushan, G.S. Blackman: Atomic force mi-
21.34 B. Bhushan, V.N. Koinkar: Nanoindentation hard- croscopy of magnetic rigid disks and sliders and
ness measurements using atomic force microscopy, its applications to tribology, ASME J. Tribol. 113,
Appl. Phys. Lett. 75, 5741–5746 (1994) 452–458 (1991)
21.35 D. DeVecchio, B. Bhushan: Localized surface 21.50 O. Marti, B. Drake, P.K. Hansma: Atomic force
elasticity measurements using an atomic force microscopy of liquid-covered surfaces: atomic res-
microscope, Rev. Sci. Instrum. 68, 4498–4505 olution images, Appl. Phys. Lett. 51, 484–486 (1987)
(1997) 21.51 B. Drake, C.B. Prater, A.L. Weisenhorn, S.A.C. Gould,
21.36 S. Amelio, A.V. Goldade, U. Rabe, V. Scherer, T.R. Albrecht, C.F. Quate, D.S. Cannell, H.G. Hansma,
B. Bhushan, W. Arnold: Measurements of mechan- P.K. Hansma: Imaging crystals, polymers and pro-
ical properties of ultra-thin diamond-like carbon cesses in water with the atomic force microscope,
coatings using atomic force acoustic microscopy, Science 243, 1586–1589 (1989)
Thin Solid Films 392, 75–84 (2001) 21.52 M. Binggeli, R. Christoph, H.E. Hintermann,
21.37 D.M. Eigler, E.K. Schweizer: Positioning single J. Colchero, O. Marti: Friction force measurements
atoms with a scanning tunnelling microscope, Na- on potential controlled graphite in an electrolytic
ture 344, 524–528 (1990) environment, Nanotechnology 4, 59–63 (1993)
21.38 A.L. Weisenhorn, J.E. MacDougall, J.A.C. Gould, 21.53 G. Meyer, N.M. Amer: Novel optical approach to
S.D. Cox, W.S. Wise, J. Massie, P. Maivald, V.B. El- atomic force microscopy, Appl. Phys. Lett. 53, 1045–
ings, G.D. Stucky, P.K. Hansma: Imaging and 1047 (1988)
manipulating of molecules on a zeolite surface 21.54 J.H. Coombs, J.B. Pethica: Properties of vacuum
with an atomic force microscope, Science 247, tunneling currents: Anomalous barrier heights, IBM
1330–1333 (1990) J. Res. Dev. 30, 455–459 (1986)
614 Part C Scanning-Probe Microscopy
21.55 M.D. Kirk, T. Albrecht, C.F. Quate: Low-temperature 21.70 Y. Martin, H.K. Wickramasinghe: Magnetic imaging
Part C 21
atomic force microscopy, Rev. Sci. Instrum. 59, by force microscopy with 1000 Å resolution, Appl.
833–835 (1988) Phys. Lett. 50, 1455–1457 (1987)
21.56 F.J. Giessibl, C. Gerber, G. Binnig: A low- 21.71 D. Rugar, H.J. Mamin, P. Güthner, S.E. Lambert,
temperature atomic force/scanning tunneling mi- J.E. Stern, I. McFadyen, T. Yogi: Magnetic force mi-
croscope for ultrahigh vacuum, J. Vac. Sci. Technol. croscopy – General principles and application to
B 9, 984–988 (1991) longitudinal recording media, J. Appl. Phys. 63,
21.57 T.R. Albrecht, P. Grutter, D. Rugar, D.P.E. Smith: 1169–1183 (1990)
Low temperature force microscope with all-fiber 21.72 C. Schönenberger, S.F. Alvarado: Understanding
interferometer, Ultramicroscopy 42–44, 1638–1646 magnetic force microscopy, Z. Phys. B 80, 373–383
(1992) (1990)
21.58 H.J. Hug, A. Moser, T. Jung, O. Fritz, A. Wadas, 21.73 U. Hartmann: Magnetic force microscopy, Annu.
I. Parashikor, H.J. Güntherodt: Low temperature Rev. Mater. Sci. 29, 53–87 (1999)
magnetic force microscopy, Rev. Sci. Instrum. 64, 21.74 D.W. Pohl, W. Denk, M. Lanz: Optical stethoscopy-
2920–2925 (1993) image recording with resolution lambda/20, Appl.
21.59 C. Basire, D.A. Ivanov: Evolution of the lamellar Phys. Lett. 44, 651–653 (1984)
structure during crystallization of a semicrystalline- 21.75 E. Betzig, J.K. Troutman, T.D. Harris, J.S. Weiner,
amorphous polymer blend: Time-resolved hot- R.L. Kostelak: Breaking the diffraction barrier – op-
stage SPM study, Phys. Rev. Lett. 85, 5587–5590 tical microscopy on a nanometric scale, Science 251,
(2000) 1468–1470 (1991)
21.60 H. Liu, B. Bhushan: Investigation of nanotribolog- 21.76 E. Betzig, P.L. Finn, J.S. Weiner: Combined shear
ical properties of self-assembled monolayers with force and near-field scanning optical microscopy,
alkyl and biphenyl spacer chains, Ultramicroscopy Appl. Phys. Lett. 60, 2484 (1992)
91, 185–202 (2002) 21.77 P.F. Barbara, D.M. Adams, D.B. O’Connor: Char-
21.61 J. Foster, J. Frommer: Imaging of liquid crystal us- acterization of organic thin film materials with
ing a tunneling microscope, Nature 333, 542–547 near-field scanning optical microscopy (NSOM),
(1988) Annu. Rev. Mater. Sci. 29, 433–469 (1999)
21.62 D. Smith, H. Horber, C. Gerber, G. Binnig: Smectic 21.78 C.C. Williams, H.K. Wickramasinghe: Scanning
liquid crystal monolayers on graphite observed by thermal profiler, Appl. Phys. Lett. 49, 1587–1589
scanning tunneling microscopy, Science 245, 43–45 (1986)
(1989) 21.79 C.C. Williams, H.K. Wickramasinghe: Microscopy of
21.63 D. Smith, J. Horber, G. Binnig, H. Nejoh: chemical-potential variations on an atomic scale,
Structure, registry and imaging mechanism of Nature 344, 317–319 (1990)
alkylcyanobiphenyl molecules by tunnelling mi- 21.80 A. Majumdar: Scanning thermal microscopy, Annu.
croscopy, Nature 344, 641–644 (1990) Rev. Mater. Sci. 29, 505–585 (1999)
21.64 Y. Andoh, S. Oguchi, R. Kaneko, T. Miyamoto: Eval- 21.81 O.E. Husser, D.H. Craston, A.J. Bard: Scanning
uation of very thin lubricant films, J. Phys. D 25, electrochemical microscopy – High resolution de-
A71–A75 (1992) position and etching of materials, J. Electrochem.
21.65 Y. Martin, C.C. Williams, H.K. Wickramasinghe: Soc. 136, 3222–3229 (1989)
Atomic force microscope-force mapping and pro- 21.82 Y. Martin, D.W. Abraham, H.K. Wickramasinghe:
filing on a sub 100 scale, J. Appl. Phys. 61, High-resolution capacitance measurement and
4723–4729 (1987) potentiometry by force microscopy, Appl. Phys.
21.66 J.E. Stern, B.D. Terris, H.J. Mamin, D. Rugar: Depo- Lett. 52, 1103–1105 (1988)
sition and imaging of localized charge on insulator 21.83 M. Nonnenmacher, M.P. O’Boyle, H.K. Wickramas-
surfaces using a force microscope, Appl. Phys. Lett. inghe: Kelvin probe force microscopy, Appl. Phys.
53, 2717–2719 (1988) Lett. 58, 2921–2923 (1991)
21.67 K. Yamanaka, H. Ogisco, O. Kolosov: Ultrasonic force 21.84 J.M.R. Weaver, D.W. Abraham: High resolution
microscopy for nanometer resolution subsurface atomic force microscopy potentiometry, J. Vac. Sci.
imaging, Appl. Phys. Lett. 64, 178–180 (1994) Technol. B 9, 1559–1561 (1991)
21.68 K. Yamanaka, E. Tomita: Lateral force modulation 21.85 D. DeVecchio, B. Bhushan: Use of a nanoscale
atomic force microscope for selective imaging of Kelvin probe for detecting wear precursors, Rev.
friction forces, Jpn. J. Appl. Phys. 34, 2879–2882 Sci. Instrum. 69, 3618–3624 (1998)
(1995) 21.86 B. Bhushan, A.V. Goldade: Measurements and
21.69 U. Rabe, K. Janser, W. Arnold: Vibrations of free and analysis of surface potential change during wear
surface-coupled atomic force microscope: Theory of single-crystal silicon (100) at ultralow loads us-
and experiment, Rev. Sci. Instrum. 67, 3281–3293 ing Kelvin probe microscopy, Appl. Surf. Sci. 157,
(1996) 373–381 (2000)
Scanning Probe Microscopy – Principle of Operation, Instrumentation, and Probes References 615
21.87 P.K. Hansma, B. Drake, O. Marti, S.A.C. Gould, tips for scanning tunneling microscopy, J. Vac. Sci.
Part C 21
C.B. Prater: The scanning ion-conductance micro- Technol. A 8, 3570–3575 (1990)
scope, Science 243, 641–643 (1989) 21.107 R. Kaneko, S. Oguchi: Ion-implanted diamond tip
21.88 C.B. Prater, P.K. Hansma, M. Tortonese, C.F. Quate: for a scanning tunneling microscope, Jpn. J. Appl.
Improved scanning ion-conductance microscope Phys. 28, 1854–1855 (1990)
using microfabricated probes, Rev. Sci. Instrum. 21.108 F.J. Giessibl: Atomic resolution of the silicon(111)–
62, 2634–2638 (1991) (7×7) surface by atomic force microscopy, Science
21.89 J. Matey, J. Blanc: Scanning capacitance mi- 267, 68–71 (1995)
croscopy, J. Appl. Phys. 57, 1437–1444 (1985) 21.109 B. Anczykowski, D. Krüger, K.L. Babcock, H. Fuchs:
21.90 C.C. Williams: Two-dimensional dopant profiling Basic properties of dynamic force spectroscopy with
by scanning capacitance microscopy, Annu. Rev. the scanning force microscope in experiment and
Mater. Sci. 29, 471–504 (1999) simulation, Ultramicroscopy 66, 251–259 (1996)
21.91 D.T. Lee, J.P. Pelz, B. Bhushan: Instrumentation 21.110 T.R. Albrecht, C.F. Quate: Atomic resolution imag-
for direct, low frequency scanning capacitance mi- ing of a nonconductor by atomic force microscopy,
croscopy, and analysis of position dependent stray J. Appl. Phys. 62, 2599–2602 (1987)
capacitance, Rev. Sci. Instrum. 73, 3523–3533 (2002) 21.111 S. Alexander, L. Hellemans, O. Marti, J. Schneir,
21.92 P.K. Hansma, J. Tersoff: Scanning tunneling mi- V. Elings, P.K. Hansma: An atomic-resolution
croscopy, J. Appl. Phys. 61, R1–R23 (1987) atomic-force microscope implemented using an
21.93 I. Giaever: Energy gap in superconductors meas- optical lever, J. Appl. Phys. 65, 164–167 (1989)
ured by electron tunneling, Phys. Rev. Lett. 5, 21.112 G. Meyer, N.M. Amer: Optical-beam-deflection
147–148 (1960) atomic force microscopy: The NaCl(001) surface,
21.94 D. Sarid, V. Elings: Review of scanning force mi- Appl. Phys. Lett. 56, 2100–2101 (1990)
croscopy, J. Vac. Sci. Technol. B 9, 431–437 (1991) 21.113 A.L. Weisenhorn, M. Egger, F. Ohnesorge,
21.95 U. Durig, O. Zuger, A. Stalder: Interaction force S.A.C. Gould, S.P. Heyn, H.G. Hansma, R.L. Sin-
detection in scanning probe microscopy: Methods sheimer, H.E. Gaub, P.K. Hansma: Molecular
and applications, J. Appl. Phys. 72, 1778–1797 (1992) resolution images of Langmuir–Blodgett films and
21.96 J. Frommer: Scanning tunneling microscopy and DNA by atomic force microscopy, Langmuir 7, 8–12
atomic force microscopy in organic chemistry, (1991)
Angew. Chem. Int. Ed. 31, 1298–1328 (1992) 21.114 J. Ruan, B. Bhushan: Atomic-scale and microscale
21.97 H.J. Güntherodt, R. Wiesendanger (Eds.): Scanning friction of graphite and diamond using friction
Tunneling Microscopy I: General Principles and Ap- force microscopy, J. Appl. Phys. 76, 5022–5035
plications to Clean and Adsorbate-Covered Surfaces (1994)
(Springer, Berlin, Heidelberg 1992) 21.115 D. Rugar, P.K. Hansma: Atomic force microscopy,
21.98 R. Wiesendanger, H.J. Güntherodt (Eds.): Scan- Phys. Today 43, 23–30 (1990)
ning Tunneling Microscopy II: Further Applications 21.116 D. Sarid: Scanning Force Microscopy (Oxford Univ.
and Related Scanning Techniques (Springer, Berlin, Press, Oxford 1991)
Heidelberg 1992) 21.117 G. Binnig: Force microscopy, Ultramicroscopy 42–
21.99 D.A. Bonnell (Ed.): Scanning Tunneling Microscopy 44, 7–15 (1992)
and Spectroscopy – Theory, Techniques, and Appli- 21.118 E. Meyer: Atomic force microscopy, Surf. Sci. 41,
cations (VCH, New York 1993) 3–49 (1992)
21.100 O. Marti, M. Amrein (Eds.): STM and SFM in Biology 21.119 H.K. Wickramasinghe: Progress in scanning probe
(Academic, San Diego 1993) microscopy, Acta Mater. 48, 347–358 (2000)
21.101 J.A. Stroscio, W.J. Kaiser (Eds.): Scanning Tunneling 21.120 A.J. den Boef: The influence of lateral forces in
Microscopy (Academic, Boston 1993) scanning force microscopy, Rev. Sci. Instrum. 62,
21.102 H.J. Güntherodt, D. Anselmetti, E. Meyer (Eds.): 88–92 (1991)
Forces in Scanning Probe Methods (Kluwer, Dor- 21.121 M. Radmacher, R.W. Tillman, M. Fritz, H.E. Gaub:
drecht 1995) From molecules to cells: Imaging soft samples with
21.103 G. Binnig, H. Rohrer: Scanning tunnelling mi- the atomic force microscope, Science 257, 1900–
croscopy, Surf. Sci. 126, 236–244 (1983) 1905 (1992)
21.104 B. Bhushan, J. Ruan, B.K. Gupta: A scanning tun- 21.122 F. Ohnesorge, G. Binnig: True atomic resolution
nelling microscopy study of fullerene films, J. Phys. by atomic force microscopy through repulsive and
D 26, 1319–1322 (1993) attractive forces, Science 260, 1451–1456 (1993)
21.105 R.L. Nicolaides, W.E. Yong, W.F. Packard, H.A. Zhou: 21.123 G. Neubauer, S.R. Coben, G.M. McClelland,
Scanning tunneling microscope tip structures, D. Horne, C.M. Mate: Force microscopy with a bidi-
J. Vac. Sci. Technol. A 6, 445–447 (1988) rectional capacitance sensor, Rev. Sci. Instrum. 61,
21.106 J.P. Ibe, P.P. Bey, S.L. Brandon, R.A. Brizzolara, 2296–2308 (1990)
N.A. Burnham, D.P. DiLella, K.P. Lee, C.R.K. Mar- 21.124 T. Goddenhenrich, H. Lemke, U. Hartmann, C. Hei-
rian, R.J. Colton: On the electrochemical etching of den: Force microscope with capacitive displace-
616 Part C Scanning-Probe Microscopy
ment detection, J. Vac. Sci. Technol. A 8, 383–387 nanotube AFM probes, Prog. Biophys. Mol. Biol. 77,
Part C 21
21.161 E.P. Stoll: Restoration of STM images distorted by 21.165 R. Toledo-Crow, P.C. Yang, Y. Chen, M. Vaez-
Part C 21
time-dependent piezo driver aftereffects, Ultrami- Iravani: Near-field differential scanning optical
croscopy 42–44, 1585–1589 (1991) microscope with atomic force regulation, Appl.
21.162 R. Durselen, U. Grunewald, W. Preuss: Calibration Phys. Lett. 60, 2957–2959 (1992)
and applications of a high precision piezo scanner 21.166 J.E. Griffith, G.L. Miller, C.A. Green: A scanning
for nanometrology, Scanning 17, 91–96 (1995) tunneling microscope with a capacitance-based
21.163 J. Fu: In situ testing and calibrating of Z-piezo of position monitor, J. Vac. Sci. Technol. B 8, 2023–
an atomic force microscope, Rev. Sci. Instrum. 66, 2027 (1990)
3785–3788 (1995) 21.167 A.E. Holman, C.D. Laman, P.M.L.O. Scholte,
21.164 R.C. Barrett, C.F. Quate: Optical scan-correction W.C. Heerens, F. Tuinstra: A calibrated scanning
system applied to atomic force microscopy, Rev. tunneling microscope equipped with capacitive
Sci. Instrum. 62, 1393 (1991) sensors, Rev. Sci. Instrum. 67, 2274–2280 (1996)
619
General and
22. General and Special Probes
in Scanning Microscopies
Jason Hafner, Edin (I-Chen) Chen, Ratnesh Lal, Sungho Jin
Part C 22
22.1 Atomic Force Microscopy........................ 620
Scanning probe microscopy (SPM) provides
22.1.1 Principles of Operation ............... 620
nanometer-scale mapping of numerous sample 22.1.2 Standard Probe Tips.................... 621
properties in essentially any environment. This 22.1.3 Probe Tip Performance ................ 622
unique combination of high resolution and broad 22.1.4 Oxide-Sharpened Tips................. 623
applicability has led to the application of SPM to 22.1.5 Focused Ion Beam Tips................ 624
many areas of science and technology, especially 22.1.6 Electron-Beam Deposition Tips .... 624
those interested in the structure and properties of 22.1.7 Single- and Multiwalled
materials at the nanometer scale. SPM images are Carbon Nanotube Tips ................. 624
generated through measurements of a tip–sample 22.1.8 Bent Carbon Nanotube Tips ......... 628
interaction. A well-characterized tip is the key el- 22.1.9 Low-Stiffness Cantilevers
ement to data interpretation and is typically the with Carbon Nanotube Tips ......... 629
limiting factor. 22.1.10 Conductive Probe Tips ................. 630
Commercially available atomic force microscopy 22.2 Scanning Tunneling Microscopy ............. 630
(AFM) tips, integrated with force-sensing can- 22.2.1 Mechanically Cut STM Tips............ 630
tilevers, are microfabricated from silicon and 22.2.2 Electrochemically Etched STM Tips 631
silicon nitride by lithographic and anisotropic
etching techniques. The performance of these tips References .................................................. 631
can be characterized by imaging nanometer-scale
standards of known dimension, and the resolution
is found to roughly correspond to the tip radius (EBD), in which a carbon spike is deposited onto
of curvature, the tip aspect ratio, and the sam- the tip apex from the background gases in an elec-
ple height. Although silicon and silicon nitride tips tron microscope. Finally, carbon nanotubes have
have a somewhat large radius of curvature, low been employed as AFM tips. Their nanometer-
aspect ratio, and limited lifetime due to wear, the scale diameter, long length, high stiffness, and
widespread use of AFM today is due in large part to elastic buckling properties make them possibly
the broad availability of these tips. In some spe- the ultimate tip material for AFM. Nanotubes can
cial cases, small asperities on the tip can provide be manually attached to silicon or silicon ni-
resolution much higher than the tip radius of cur- tride AFM tips or grown onto tips by chemical
vature for low-Z samples such as crystal surfaces vapor deposition (CVD), which should soon make
and ordered protein arrays. them widely available. In scanning tunneling
Several strategies have been developed to microscopy (STM), the electron tunneling signal
improve AFM tip performance. Oxide sharpen- decays exponentially with tip–sample separation,
ing improves tip sharpness and enhances tip so that in principle only the last few atoms con-
asperities. For high-aspect-ratio samples such as tribute to the signal. STM tips are, therefore, not
integrated circuits, silicon AFM tips can be modified as sensitive to the nanoscale tip geometry and
by focused ion beam (FIB) milling. FIB tips reach can be made by simple mechanical cutting or
3◦ cone angles over lengths of several microns and electrochemical etching of metal wires. In choos-
can be fabricated at arbitrary angles. ing tip materials, one prefers hard, stiff metals
Other high resolution and high-aspect-ratio that will not oxidize or corrode in the imaging
tips are produced by electron-beam deposition environment.
620 Part C Scanning-Probe Microscopy
Part C 22.1
created by gluing diamond fragments to the foil can-
tilevers or electrochemically etching the wires to a sharp
point. Since these methods were labor intensive and not
highly reproducible, they were not amenable to large-
scale production. To address this problem, and the need
for smaller cantilevers with higher resonant frequencies,
Fig. 22.2 An illustration of the optical beam deflection sys- batch fabrication techniques were developed (Fig. 22.3).
tem that detects cantilever motion in the AFM. The voltage Building on existing methods to batch-fabricate Si3 N4
signal VA − VB is proportional to the deflection cantilevers, Albrecht et al. [22.2] etched an array of
small square openings in an SiO2 mask layer over
proportional to the difference signal VA − VB . Sub- a (100) silicon surface. The exposed square (100) re-
angstrom deflections can be detected, and therefore gions were etched with KOH, an anisotropic etchant
forces down to tens of piconewtons can be measured. that terminates at the (111) planes, thus creating pyra-
A more recently developed method of cantilever de- midal etch pits in the silicon surface. The etch pit mask
flection measurement is through a piezoelectric layer was then removed and another was applied to define
on the cantilever that registers a voltage upon deflec- the cantilever shapes with the pyramidal etch pits at the
tion [22.1]. end. The Si wafer was then coated with a low-stress
A piezoelectric scanner rasters the sample under the Si3 N4 layer by low-pressure chemical vapor deposi-
tip while the forces are measured through deflections tion (LPCVD). The Si3 N4 fills the etch pit, using it as
of the cantilever. To achieve more controlled imaging a mold to create a pyramidal tip. The silicon was later
conditions, a feedback loop monitors the tip–sample removed by etching to free the cantilevers and tips. Fur-
force and adjusts the sample z-position to hold the ther steps resulting in the attachment of the cantilever
force constant. The topographic image of the sample to a macroscopic piece of glass are not described here.
is then taken from the sample z-position data. The The resulting pyramidal tips were highly symmetric and
mode described is called the contact mode, in which had a tip radius of < 30 nm, as determined by scanning
the tip is deflected by the sample due to repulsive electron microscopy (SEM). This procedure has likely
forces, or contact. It is generally only used for flat sam- not changed significantly, since commercially available
ples that can withstand lateral forces during scanning. Si3 N4 tips are still specified to have a radius of curvature
To minimize lateral forces and sample damage, two of 30 nm.
alternating-current (AC) modes have been developed. In
these, the cantilever is driven into AC oscillation near its
Si3N4 tip fabrication
resonant frequency (tens to hundreds of kHz) with de-
sired amplitudes. When the tip approaches the sample, Apply Anisotropic Remove Deposit
mask etch mask Si3N4
the oscillation is damped, and the reduced amplitude is
the feedback signal, rather than the direct-current (DC)
Si Remove Si
deflection. Again, topography is taken from the vary-
ing Z-position of the sample required to keep the tip
oscillation amplitude constant. The two AC modes dif- Si tip fabrication
fer only in the nature of the interaction. In intermittent Si Apply Under Remove
cantilever mask etch mask
contact mode, also called tapping mode, the tip con-
tacts the sample on each cycle, so the amplitude is Si
reduced by ionic repulsion as in contact mode. In non-
contact mode, long-range van der Waals forces reduce Fig. 22.3 A schematic overview of the fabrication of Si and
the amplitude by effectively shifting the spring con- Si3 N4 tip fabrication as described in the text
622 Part C Scanning-Probe Microscopy
Wolter et al. [22.3] developed methods to batch- a mask for deposited material. Commercially available
fabricate single-crystal Si cantilevers with integrated silicon probes are made by similar refined techniques
tips. Microfabricated Si cantilevers were first prepared and provide a typical radius of curvature of < 10 nm.
using previously described methods, and a small mask
was formed at the end of the cantilever. The Si around 22.1.3 Probe Tip Performance
the mask was etched by KOH, so that the mask was
undercut. This resulted in a pyramidal silicon tip un- In atomic force microscopy the question of resolution
Part C 22.1
der the mask, which was then removed. Again, this can be a rather complicated issue. As an initial ap-
partial description of the full procedure only describes proximation, resolution is often considered strictly in
tip fabrication. With some refinements the silicon tips geometrical terms that assume rigid tip–sample contact.
were made in high yield with radii of curvature of less The topographical image of a feature is broadened or
than 10 nm. Si tips are sharper than Si3 N4 tips, be- narrowed by the size of the probe tip, so the resolu-
cause they are directly formed by the anisotropic etch tion is approximately the width of the tip. Therefore, the
in single-crystal Si, rather than using an etch pit as resolution of AFM with standard commercially avail-
able tips is on the order of 5–10 nm. Bustamante and
Keller [22.4] carried the geometrical model further by
Scan direction Scan direction
drawing an analogy to resolution in optical systems.
Consider two sharp spikes separated by a distance d to
Tip Tip be point objects imaged by AFM (Fig. 22.4). Assume
the tip has a parabolic shape with an end radius R.
Δz Δh The tip-broadened image of these spikes will appear
R R Δz as inverted parabolas. There will be a small depression
between the images of depth Δz. The two spikes are
considered resolved if Δz is larger than the instrumental
noise in the z-direction. Defined in this manner, the res-
d d olution d, the minimum separation at which the spikes
are resolved, is
Sample spikes Inverted √
tip surfaces d = 2 2R(Δz) , (22.1)
Part C 22.1
(Fig. 22.4). Noncontact AFM is sensitive to the gradient
of long-range forces, so the van der Waals force gradi- Support
ent was calculated as a function of position for the tip
at height h above the surface. If the resolution d is de- Fig. 22.5 A tip model to explain the high resolution ob-
fined as the full-width at half-maximum of this curve, tained on ordered samples in contact mode (after [22.6])
the resolution is
confirmed by independent electron microscopy images
d = 0.8h . (22.3)
of the tip. Another model, considered especially appli-
This shows that, even for an ideal geometry, the resolu- cable to atomic resolution on crystal surfaces, assumes
tion is fundamentally limited in noncontact mode by the that the tip is in contact with a region of the sample
tip–sample separation. Under UHV conditions, the tip– much larger than the observed resolution, and that force
sample separation can be made very small, so atomic components matching the periodicity of the sample are
resolution is possible on flat, crystalline surfaces. Un- transmitted to the tip, resulting in an averaged image
der ambient conditions, however, the separation must of the periodic lattice. Regardless of the mechanism,
be larger to keep the tip from being trapped in the ambi- the structures determined are accurate and make this
ent water layer on the surface. This larger separation can a highly valuable method for membrane proteins. How-
lead to a point where further improvements in tip sharp- ever, this level of resolution should not be expected for
ness do not improve resolution. It has been found that most biological systems.
imaging 5 nm gold nanoparticles in noncontact mode
with carbon nanotube tips of 2 nm diameter leads to par- 22.1.4 Oxide-Sharpened Tips
ticle widths of 12 nm, larger than the 7 nm width one
would expect assuming rigid contact [22.8]. However, Both Si and Si3 N4 tips with increased aspect ratio
in tapping-mode operation, the geometrical definition and reduced tip radius can be fabricated through ox-
of resolution is relevant, since the tip and sample come ide sharpening of the tip. If a pyramidal or cone-shaped
into rigid contact. When imaging 5 nm gold particles silicon tip is thermally oxidized to SiO2 at low tempera-
with 2 nm carbon nanotube tips in tapping mode, the ture (< 1050 ◦ C), Si–SiO2 stress formation reduces the
expected 7 nm particle width is obtained [22.9]. oxidation rate at regions of high curvature. The result
The above descriptions of AFM resolution can- is a sharper, higher-aspect-ratio cone of silicon at the
not explain the subnanometer resolution achieved high-curvature tip apex inside the outer pyramidal layer
on crystal surfaces [22.10] and ordered arrays of
biomolecules [22.11] in contact mode with commer-
cially available probe tips. Such tips have nominal radii
of curvature ranging from 5 to 30 nm, an order of mag-
nitude larger than the resolution achieved. A detailed
model to explain the high resolution on ordered mem-
brane proteins has been put forth by [22.6]. In this
model, the larger part of the silicon nitride tip apex bal-
ances the tip–sample interaction through electrostatic
forces, while a very small tip asperity interacts with 20 nm
the sample to provide contrast (Fig. 22.5). This model
is supported by measurements at varying salt concen- Fig. 22.6 Oxide sharpening of silicon tips. The left image
trations to vary the electrostatic interaction strength and shows a sharpened core of silicon in an outer layer of SiO2 .
the observation of defects in the ordered samples. How- The right image is a higher magnification view of such a tip
ever, the existence of such asperities has never been after the SiO2 is removed (after [22.7])
624 Part C Scanning-Probe Microscopy
Part C 22.1
direct-current plasma-enhanced CVD (DC-PECVD) of-
5 nm
ten exhibit a stacked cone structure (also known as
herringbone- or bamboo-like structures, a cross-section 100 Å
of which is illustrated in Fig. 22.9b). Herringbone- Fig. 22.8a,b The structure of carbon nanotubes. (a) TEM image of
like CNTs are also called carbon nanofibers (CNFs) SWNTs (after [22.16]). (b) TEM image of MWNTs (after [22.17])
since they are not made of perfect graphene tube
cylinders.
a) Thermal CVD b) DC plasma CVD
Carbon Nanotube Probes
by Attachment Approaches
CNTs have been attached onto AFM cantilever pyra-
mid tips by various approaches. The first CNT AFM
probes [22.18] were fabricated by techniques devel-
oped for assembling single-nanotube field-emission
tips [22.19]. This process, illustrated in Fig. 22.10, used
a purified MWNT material synthesized by the carbon
arc procedure. The raw material must contain at least
a few percent of long nanotubes (> 10 μm), purified by Fig. 22.9a,b Schematic structures of (a) tubelike carbon
oxidation to ≈ 1% of its original mass. A torn edge nanotubes and (b) stacked-cone nanotubes
of the purified material was attached to a microma-
nipulator by carbon tape and viewed under an optical a)
microscope. Individual nanotubes and nanotube bundles
were visible as filaments under dark-field illumina-
tion. A commercially available AFM tip was attached
to another micromanipulator opposing the nanotube
material. Glue was applied to the tip apex from high-
vacuum carbon tape supporting the nanotube material.
Nanotubes were then manually attached to the tip apex
by micromanipulation. As assembled, MWNT tips were
often too long for imaging due to thermal vibration 50 × 0.6
b)
during their use as AFM probes. Nanotubes tips were
shortened by applying 10 V pulses to the tip while
near a sputtered niobium surface. This process etched
100 nm lengths of nanotubes per pulse.
Since the nanotube orientation cannot be well con-
trolled during manual attachment processes, the transfer
procedure from the nanotube probe cartridge to the Si
tips was operated under an electric field [22.20]. When
applying a low voltage, the nanotube is attracted to the Fig. 22.10a,b Schematic drawing of the setup for manual assembly
cantilever tip and aligned with the apex of the tip. This of carbon nanotube tips (a) and (b) optical microscopy images of
approach provides better control of the orientation of the assembly process (the cantilever was drawn in for clarity)
nanotube probes because of the electric-field alignment
and electrostatic attraction of nanotube probes. When to induce an arc discharge in which the nanotube is en-
the nanotube is suitably aligned, the voltage is increased ergetically disassociated and the formation of a carbide
626 Part C Scanning-Probe Microscopy
may occur at the contact site. Thus, the nanotube can be 1. The alignment of the nanotubes at the tip
attached to the cantilever free from the cartridge. The 2. The number of nanotubes that grow at the tip
mechanical attachment method has also been carried 3. The length of the nanotube tip.
out in a SEM rather than an optical microscope [22.21].
This process allows selecting a single nanotube and at- Li et al. [22.28] found that nanotubes grow per-
taching it to a specific site on the Si tip. This approach pendicular to a porous surface containing embedded
eliminates the need for pulse-etching, since short nano- catalyst. This approach was exploited to fabricate nano-
Part C 22.1
tubes can be attached to the tip, and the glue can be tube tips by CVD [22.29] with the proper alignment, as
applied by EBD. illustrated in Fig. 22.11. A flattened area of ≈ 1–5 μm2
A method to attach CNTs onto AFM tips using was created on Si tips by scanning in contact mode at
magnetic-field alignment has been developed [22.23]. high load (1 μN) on a hard, synthetic diamond surface.
The experimental apparatus is designed to introduce The tip was then anodized in HF to create 100 nm-
a magnetic field onto a single AFM probe and a nano- diameter pores in this flat surface [22.30]. It is important
tube suspension. With this apparatus, the anisotropic to anodize only the last 20–40 μm of the cantilever,
properties of the CNT cause the nanotubes that come which includes the tip, so that the rest of the cantilever
into contact with the probe tip to be preferentially ori- is still reflective for use in the AFM. This was achieved
ented parallel to the tip direction and hence protrude by anodizing the tip in a small drop of HF under the
down from the end. Another attachment method based view of an optical microscope. Next, iron was elec-
on liquid deposition of CNTs onto AFM probes is the trochemically deposited into the pores to form catalyst
dielectrophoresis process [22.24, 25]. A Si AFM probe particles [22.31]. Tips prepared in this way were heated
and a metal plate are used as electrodes to apply the in low concentrations of ethylene at 800 ◦ C, which is
AC electric field. A charge-coupled device (CCD) con- known to favor the growth of thin nanotubes [22.26].
nected to a computer could be used to monitor the When imaged by SEM, nanotubes were found to grow
process. With in situ observation using the CCD im- perpendicular to the surface from the pores as desired,
age, the counterelectrode was slowly moved close to and TEM revealed that the nanotubes were thin, in-
the AFM probe until the suspension surface touches its dividual, multiwalled nanotubes with typical radii of
apex. The electrode was then gradually withdrawn until 3–5 nm.
a CNT tip with the desired length was assembled. In this These pore-growth CVD nanotube tips were typ-
dielectrophoresis process, the length of the CNT probe ically several micrometers in length – too long for
is controlled by the distance that the counterelectrode is imaging – and were pulse-etched to usable length of
translated under the AC field. < 500 nm. The tips exhibited elastic buckling behav-
ior and were very robust in imaging. The pore-growth
Nanotube Probe Synthesis by Thermal CVD method demonstrated the potential of CVD to simplify
The mechanical attachment approaches are tedious and the fabrication of nanotube tips, although there were
time consuming since nanotube tips are made indi-
vidually. So, these methods cannot be applied for
mass production. The problems of manual assembly a) b)
of nanotube probes discussed above can largely be
solved by directly growing nanotubes onto AFM tips
by metal-catalyzed chemical vapor deposition (CVD). FIB
Nanometer-scale metal catalyst particles are heated in ‘Gaussian’
FIB
a gas mixture containing hydrocarbon or CO. The gas
molecules dissociate on the metal surface, and car-
bon is adsorbed into the catalyst particle. When this CNT
carbon precipitates, it nucleates a nanotube of similar
diameter to the catalyst particle. Therefore, CVD al- ‘Truncated
Gaussian’
lows control over nanotube size and structure, including
the production of SWNTs [22.26] with radii as low as
3.5 Å [22.27]. Fig. 22.11a,b Schematics for two experimental setup con-
Several key issues must be addressed to grow nano- ditions using a focused ion beam for (a) aligning a nano-
tube AFM tips by CVD: tube tip and (b) bending the tip (after [22.22])
General and Special Probes in Scanning Microscopies 22.1 Atomic Force Microscopy 627
still limitations. In particular, the porous layer was dif- attached to the tip. This assembly process happened au-
ficult to prepare and rather fragile. tomatically when imaging in tapping mode; no special
An alternative approach to CVD fabrication of tip manipulation was required.
nanotube tips involves direct growth of SWNTs on the
surface of a pyramidal AFM tip [22.32, 33]. In this PECVD-Grown Nanotube Probe
surface-growth approach, an alumina/iron/molybdenum The attachment methods are time consuming and often
powdered catalyst known to produce SWNT [22.26] result in nonreproducible CNT configuration and place-
Part C 22.1
was dispersed in ethanol at 1 mg/ml. Silicon tips were ment. While thermal CVD approaches can potentially
dipped in this solution and allowed to dry, leaving lead to wafer-scale production of AFM tips, the number,
a sparse layer of ≈ 100 nm catalyst clusters on the orientation, and length of CNTs are difficult to control.
tip. When CVD conditions were applied, single-walled At the end of the fabrication, these processes usually re-
nanotubes grew along the silicon tip surface. At a pyra- quire a one-at-a-time manipulation approach to remove
mid edge, nanotubes can either bend to align with the extra CNTs and/or to shorten the remaining CNTs for
edge or protrude from the surface. If the energy required SPM applications.
to bend the tube and follow the edge is less than the at- The key process for CVD-grown CNT probe fab-
tractive nanotube surface energy, then the nanotube will rication is catalyst patterning, which determines the
follow the pyramid edge to the apex. Therefore, nano- position, number, and diameter of the probe. Elec-
tubes were effectively steered toward the tip apex by trophoretically deposited or spin-coated colloidal cata-
the pyramid edges. At the apex, the nanotube protruded lyst particles on Si pyramid tips cannot provide reliable
from the tip, since the energetic cost of bending around control of the position and number of catalyst par-
the sharp silicon tip was too high. The high aspect ratio ticles. MWNT probes on tipless cantilevers have been
at the oxide-sharpened silicon tip apex was critical for fabricated based on conventional Si fabrication pro-
good nanotube alignment. These surface-growth nano- cess in which the catalyst pattering was proceeded
tube tips exhibit a high aspect ratio and high-resolution by typical e-beam lithography and lift-off of spin-
imaging, as well as elastic buckling. This method has coated poly(methyl methacrylate) (PMMA) layer, and
been expanded to include wafer-scale production of plasma-enhanced chemical vapor deposition (PECVD)
nanotube tips with high yields [22.34], yet one obsta- was used for CNT growth [22.35, 36]. The fabrication
cle remains to the mass production of nanotube probe method described in [22.35] allows CNT tips to be
tips. Nanotubes protruding from the tip are several mi- grown directly on silicon cantilevers at the wafer scale.
crometers long, and since they are so thin, they must be CNT tip locations and diameters are defined by e-beam
etched to less than 100 nm. lithography. CNT length and orientation are controlled
by the growth conditions of the PECVD method. There-
Hybrid Nanotube Tip Fabrication: Pick-Up Tips fore, there is no need to shorten the CNT after the
Another method of creating nanotube tips is something growth. In PECVD, an electric field is present in the
of a hybrid between assembly and CVD. The motiva- plasma discharge to direct the nanotubes to grow par-
tion was to create AFM probes that have an individual allel to the electric field. A tilted probe is desirable as
SWNT at the tip to achieve the ultimate imaging reso- it compensates for the operating tilt angle of the AFM
lution. In order to synthesize isolated SWNT, isolated cantilever so that the probe itself is close to vertical for
catalyst particles were formed by dipping a silicon stable imaging.
wafer in an isopropyl alcohol solution of Fe(NO3 )3 . A spin-coated PMMA layer cannot be uniformly
When heated in a CVD furnace, the iron became mobile conformal on the relatively small piece of tipless
and aggregated to form isolated iron particles. By con- cantilevers or on the Si pyramid tip. For e-beam
trolling the reaction time, the SWNT lengths were kept lithography-based processes, the patterned catalyst dots
shorter than their typical separation, so that the nano- either have to be formed before the fabrication of
tubes never had a chance to form bundles. In the pick-up the cantilevers (although then a protection layer is
tip method, these isolated SWNT substrates were im- needed) [22.35] or lithography steps have to be ap-
aged by AFM with silicon tips in air [22.9]. When the plied twice to remove extra catalyst on commercial
tip encountered a vertical SWNT, the oscillation am- tipless-cantilever chips [22.36]. Therefore, the electron-
plitude was damped, so the AFM pulled the sample beam-induced deposition (EBID) technique has been
away from the tip. This procedure pulled the SWNT into developed to make catalyst patterns for CNT probe fab-
contact with the tip along its length, so that it became rication [22.37,38]. EBID is a simple and fast technique
628 Part C Scanning-Probe Microscopy
Part C 22.1
electric-field angles during the growth process. 1 µm 1 µm
a) NH3/C2H2
1 µm 1 µm 250 nm
Fig. 22.16a–c Carbon nanotube bending using tilted bias electric
field during plasma enhanced CVD growth. The nanotube tip can be
bent (a) either slightly, (b) by ≈ 45◦ , or (c) by ≈ 90◦ using various
electric field angles during the growth process (after [22.40, 41])
200 µm
controlled. The plasma-induced stresses and damages
introduced during PECVD growth of nanotubes, how-
b) H2/C2H2
ever, result in severely bent cantilevers when a thin,
low-stiffness cantilever is utilized as the substrate. If the
bend is sufficiently large, the AFM laser spot focused at
their end will be deflected off of the position-sensitive
detector, rendering the cantilevers unusable for AFM
measurements.
200 µm An in situ process to control the deflection of
cantilever beams during CNT growth has been demon-
strated by introducing hydrogen gas into the (acety-
c) NH3/H2/C2H2
lene + ammonia) feed gas and adjusting the ammonia-
to-hydrogen flow ratio [22.42]. The total flow rate of
NH3 and H2 was kept constant during growth, while
the gas mix ratio (R), defined as NH3 /(NH3 + H2 ),
was varied in the range 0 ≤ R ≤ 1. Figure 22.17 shows
comparative, cross-sectional cantilever images for three
200 µm different CNT growth conditions using different feed
gas compositions. A large upward or downward bend-
Fig. 22.17a–c Optical microscope images of cantilevers ing of the cantilever is observed for R = 1 and R = 0,
bending after plasma treatments with C2 H2 gas and respectively. By employing a particular gas ratio of
(a) NH3 gas (R = 1). (b) H2 gas (R = 0). (c) Mixed R = 0.5, a nearly flat cantilever beam can be obtained
NH3 /H2 gas (R = 0.5) (after [22.42]) after PECVD growth of a CNT probe.
630 Part C Scanning-Probe Microscopy
vature in the range of 0.1–1 μm (Fig. 22.19a). They sizes (Fig. 22.19), although the exact atomic structure
are not useful for imaging samples with surface rough- of the tip apex is still not well controlled. The param-
ness above a few nanometers. However, on atomically eters of electrochemical etching depend greatly on the
flat samples, mechanically cut tips can achieve atomic tip material and the desired tip shape. The following
resolution due to the nature of the tunneling signal, is an entirely general description. A fine metal wire
which drops exponentially with tip–sample separation. (0.1–1 mm diameter) of the tip material is immersed in
Since mechanically cut tips contain many small asperi- an appropriate electrochemical etchant solution. A bias
Part C 22
ties on the larger tip structure, atomic resolution is easily voltage of 1–10 V is applied between the tip and
achieved as long as one atom of the tip is just a few a counterelectrode such that the tip is etched. Due to
angstroms lower than all of the others. the enhanced etch rate at the electrolyte–air interface,
a neck is formed in the wire. This neck is eventually
22.2.2 Electrochemically Etched STM Tips etched thin enough that it cannot support the weight
of the part of the wire suspended in the solution, and
For samples with more than a few nanometers of sur- it breaks to form a sharp tip. The widely varying pa-
face roughness, the tip structure in the nanometer size rameters and methods will be not be covered in detail
range becomes an issue. Electrochemical etching can here, but many recipes can be found in the literature for
provide tips with reproducible and desirable shapes and common tip materials [22.47–51].
References
22.1 R. Linnemann, T. Gotszalk, I.W. Rangelow, P. Du- 22.10 F. Ohnesorge, G. Binnig: True atomic resolution
mania, E. Oesterschulze: Atomic force microscopy by atomic force microscopy through repulsive and
and lateral force microscopy using piezoresistive attractive forces, Science 260, 1451–1456 (1993)
cantilevers, J. Vac. Sci. Technol. B 14(2), 856–860 22.11 D.J. Müller, D. Fotiadis, A. Engel: Mapping flexible
(1996) protein domains at subnanometer resolution with
22.2 T.R. Albrecht, S. Akamine, T.E. Carver, C.F. Quate: the atomic force microscope, FEBS Letters 430(1/2),
Microfabrication of cantilever styli for the atomic 105–111 (1998), Special Issue SI
force microscope, J. Vac. Sci. Technol. A 8(4), 3386– 22.12 S. Akamine, R.C. Barrett, C.F. Quate: Improved
3396 (1990) atomic force microscope images using microcan-
22.3 O. Wolter, T. Bayer, J. Greschner: Micromachined tilevers with sharp tips, Appl. Phys. Lett. 57(3),
silicon sensors for scanning force microscopy, 316–318 (1990)
J. Vac. Sci. Technol. B 9(2), 1353–1357 (1991) 22.13 D.J. Keller, C. Chih-Chung: Imaging steep, high
22.4 C. Bustamante, D. Keller: Scanning force mi- structures by scanning force microscopy with elec-
croscopy in biology, Phys. Today 48(12), 32–38 tron beam deposited tips, Surf. Sci. 268, 333–339
(1995) (1992)
22.5 J. Vesenka, S. Manne, R. Giberson, T. Marsh, 22.14 T. Ichihashi, S. Matsui: In situ observation on elec-
E. Henderson: Colloidal gold particles as an tron beam induced chemical vapor deposition by
incompressible atomic force microscope imag- transmission electron microscopy, J. Vac. Sci. Tech-
ing standard for assessing the compressibility of nol. B 6(6), 1869–1872 (1988)
biomolecules, Biophys. J. 65, 992–997 (1993) 22.15 K.I. Schiffmann: Investigation of fabrication pa-
22.6 D.J. Müller, D. Fotiadis, S. Scheuring, S.A. Müller, rameters for the electron-beam-induced deposi-
A. Engel: Electrostatically balanced subnanometer tion of contamination tips used in atomic force
imaging of biological specimens by atomic force microscopy, Nanotechnology 4, 163–169 (1993)
microscope, Biophys. J. 76(2), 1101–1111 (1999) 22.16 D.S. Bethune, C.H. Kiang, M.S. de Vries, G. Gorman,
22.7 R.B. Marcus, T.S. Ravi, T. Gmitter, K. Chin, D.J. Liu, R. Savoy, J. Vazquez, R. Beyers: Cobalt-catalysed
W. Orvis, D.R. Ciarlo, C.E. Hunt, J. Trujillo: Formation growth of carbon nanotubes with single-atomic-
of silicon tips with < 1 nm radius, Appl. Phys. Lett. layer walls, Nature 363(6430), 605–607 (1993)
56(3), 236–238 (1990) 22.17 E.T. Thostenson, Z. Ren, T.W. Chou: Advances in the
22.8 J.H. Hafner, C.L. Cheung, C.M. Lieber: unpublished science and technology of carbon nanotubes and
results (2001) their composites: A review, Compos. Sci. Technol.
22.9 J.H. Hafner, C.L. Cheung, T.H. Oosterkamp, 61(13), 1899–1912 (2001)
C.M. Lieber: High-yield assembly of individual 22.18 H.J. Dai, J.H. Hafner, A.G. Rinzler, D.T. Colbert,
single-walled carbon nanotube tips for scanning R.E. Smalley: Nanotubes as nanoprobes in scan-
probe microscopies, J. Phys. Chem. B 105(4), 743– ning probe microscopy, Nature 384(6605), 147–150
746 (2001) (1996)
632 Part C Scanning-Probe Microscopy
22.19 A.G. Rinzler, Y.H. Hafner, P. Nikolaev, L. Lou, 22.33 E.B. Cooper, S.R. Manalis, H. Fang, H. Dai,
S.G. Kim, D. Tomanek, D.T. Colbert, R.E. Smalley: K. Matsumoto, S.C. Minne, T. Hunt, C.F. Quate:
Unraveling nanotubes: Field emission from atomic Terabit-per-square-inch data storage with the
wire, Science 269, 1550 (1995) atomic force microscope, Appl. Phys. Lett. 75(22),
22.20 R. Stevens, C. Nguyen, A. Cassell, L. Delzeit, 3566–3568 (1999)
M. Meyyappan, J. Han: Improved fabrication ap- 22.34 E. Yenilmez, Q. Wang, R.J. Chen, D. Wang, H. Dai:
proach for carbon nanotube probe devices, Appl. Wafer scale production of carbon nanotube scan-
Phys. Lett. 77, 3453–3455 (2000) ning probe tips for atomic force microscopy, Appl.
Part C 22
22.21 H. Nishijima, S. Kamo, S. Akita, Y. Nakayama, Phys. Lett. 80(12), 2225–2227 (2002)
K.I. Hohmura, S.H. Yoshimura, K. Takeyasu: 22.35 Q. Ye, A.M. Cassell, H.B. Liu, K.J. Chao, J. Han,
Carbon-nanotube tips for scanning probe mi- M. Meyyappan: Large-scale fabrication of carbon
croscopy: Preparation by a controlled process and nanotube probe tips for atomic force microscopy
observation of deoxyribonucleic acid, Appl. Phys. critical dimension imaging applications, Nano Lett.
Lett. 74, 4061–4063 (1999) 4, 1301–1308 (2004)
22.22 B.C. Park, K.Y. Jung, W.Y. Song, O. Beom-Hoan, 22.36 H. Cui, S.V. Kalinin, X. Yang, D.H. Lowndes: Growth
S.J. Ahn: Bending of a carbon nanotube in vacuum of carbon nanofibers on tipless cantilevers for high
using a focused ion beam, Adv. Mater. 18, 95–98 resolution topography and magnetic force imag-
(2006) ing, Nano Lett. 4, 2157–2161 (2004)
22.23 A. Hall, W.G. Matthews, R. Superfine, M.R. Falvo, 22.37 I.-C. Chen, L.-H. Chen, X.-R. Ye, C. Daraio, S. Jin,
S. Washburna: Simple and efficient method for car- C.A. Orme, A. Quist, R. Lal: Extremely sharp car-
bon nanotube attachment to scanning probes and bon nanocone probes for atomic force microscopy
other substrates, Appl. Phys. Lett. 82, 2506–2508 imaging, Appl. Phys. Lett. 88, 153102 (2006)
(2003) 22.38 I.-C. Chen, L.-H. Chen, C.A. Orme, A. Quist,
22.24 J. Tang, G. Yang, Q. Zhang, A. Parhat, B. Maynor, R. Lal, S. Jin: Fabrication of high-aspect-ratio car-
J. Liu, L.C. Qin, O. Zhou: Rapid and reproducible bon nanocone probes by electron beam induced
fabrication of carbon nanotube AFM probes by di- deposition patterning, Nanotechnology 17, 4322
electrophoresis, Nano Lett. 5, 11–14 (2005) (2006)
22.25 J.-E. Kim, J.-K. Park, C.-S. Han: Use of di- 22.39 Z.F. Deng, E. Yenilmez, A. Reilein, J. Leu, H. Dai,
electrophoresis in the fabrication of an atomic K.A. Moler: Nanotube manipulation with focused
force microscope tip with a carbon nanotube: ion beam, Appl. Phys. Lett. 88, 023119 (2006)
Experimental investigation, Nanotechnology 17, 22.40 J.F. AuBuchon, L.-H. Chen, S. Jin: Control of carbon
2937–2941 (2006) capping for regrowth of aligned carbon nanotubes,
22.26 J.H. Hafner, M.J. Bronikowski, B.R. Azamian, J. Phys. Chem. B 109, 6044–6048 (2005)
P. Nikolaev, A.G. Rinzler, D.T. Colbert, K.A. Smith, 22.41 J.F. AuBuchon, L.-H. Chen, A.I. Gapin, S. Jin:
R.E. Smalley: Catalytic growth of single-wall car- electric-field-guided growth of carbon nanotubes
bon nanotubes from metal particles, Chem. Phys. during DC plasma-enhanced CVD, Chem. Vap. De-
Lett. 296(1/2), 195–202 (1998) pos. 12(6), 370–374 (2006)
22.27 P. Nikolaev, M.J. Bronikowski, R.K. Bradley, 22.42 I.-C. Chen, L.-H. Chen, C.A. Orme, S. Jin: Control
F. Rohmund, D.T. Colbert, K.A. Smith, R.E. Smal- of curvature in highly compliant probe cantilevers
ley: Gas-phase catalytic growth of single-walled during carbon nanotube growth, Nano Lett. 7(10),
carbon nanotubes from carbon monoxide, Chem. 3035–3040 (2007)
Phys. Lett. 313(1/2), 91–97 (1999) 22.43 A. Quist, I. Doudevski, H. Lin, R. Azimova, D. Ng,
22.28 W.Z. Li, S.S. Xie, L.X. Qian, B.H. Chang, B.S. Zou, B. Frangione, B. Kagan, J. Ghiso, R. Lal: Amyloid ion
W.Y. Zhou, R.A. Zhao, G. Wang: Large-scale synthe- channels: A common structural link for protein-
sis of aligned carbon nanotubes, Science 274(5293), misfolding disease, Proc. Natl. Acad. Sci. USA 102,
1701–1703 (1996) 10427 (2005)
22.29 J.H. Hafner, C.L. Cheung, C.M. Lieber: Growth 22.44 A.P. Quist, A. Chand, S. Ramachandran, C. Daraio,
of nanotubes for probe microscopy tips, Nature S. Jin, R. Lal: AFM imaging and electrical record-
398(6730), 761–762 (1999) ing of lipid bilayers supported over microfabricated
22.30 V. Lehmann: The physics of macroporous silicon silicon chip nanopores: A lab on-chip system for
formation, Thin Solid Films 255, 1–4 (1995) lipid membrane and ion channels, Langmuir 23(3),
22.31 F. Ronkel, J.W. Schultze, R. Arensfischer: Electrical 1375 (2007)
contact to porous silicon by electrodeposition of 22.45 J. Thimm, A. Mechler, H. Lin, S.K. Rhee, R. Lal: Cal-
iron, Thin Solid Films 276(1–2), 40–43 (1996) cium dependent open-closed conformations and
22.32 J.H. Hafner, C.L. Cheung, C.M. Lieber: Direct growth interfacial energy maps of reconstituted individual
of single-walled carbon nanotube scanning probe hemichannels, J. Biol. Chem. 280, 10646 (2005)
microscopy tips, J. Am. Chem. Soc. 121(41), 9750– 22.46 A. Stemmer, A. Hefti, U. Aebi, A. Engel: Scanning
9751 (1999) tunneling and transmission electron microscopy on
General and Special Probes in Scanning Microscopies References 633
identical areas of biological specimens, Ultrami- 22.49 J.P. Ibe, P.P. Bey, S.L. Brandow, R.A. Brizzolara,
croscopy 30(3), 263 (1989) N.A. Burnham, D.P. DiLella, K.P. Lee, C.R.K. Mar-
22.47 J.H. Hafner, C.L. Cheung, A.T. Woolley, C.M. Lieber: rian, R.J. Colton: On the electrochemical etching of
Structural and functional imaging with carbon tips for scanning tunneling microscopy, J. Vac. Sci.
nanotube AFM probes, Prog. Biophys. Mol. Biol. Technol. A 8, 3570–3575 (1990)
77(1), 73–110 (2001) 22.50 L. Libioulle, Y. Houbion, J.-M. Gilles: Very sharp
22.48 R. Nicolaides, L. Yong, W.E. Packard, W.F. Zhou, platinum tips for scanning tunneling microscopy,
H.A. Blackstead, K.K. Chin, J.D. Dow, J.K. Furdyna, Rev. Sci. Instrum. 66(1), 97–100 (1995)
Part C 22
M.H. Wei, R.C.J. Jaklevic, W. Kaiser, A.R. Pelton, 22.51 A.J. Nam, A. Teren, T.A. Lusby, A.J. Melmed: Benign
M.V. Zeller, J. Bellina Jr.: Scanning tunneling mi- making of sharp tips for STM and FIM: Pt, Ir, Au,
croscope tip structures, J. Vac. Sci. Technol. A 6(2), Pd, and Rh, J. Vac. Sci. Technol. B 13(4), 1556–1559
445–447 (1988) (1995)
635
Noncontact A
23. Noncontact Atomic Force Microscopy
and Related Topics
Franz J. Giessibl, Yasuhiro Sugawara, Seizo Morita, Hirotaka Hosoi, Kazuhisa Sueoka, Koichi Mukasa,
Akira Sasahara, Hiroshi Onishi
Part C 23
for future bottom-up processes. In Sect. 23.1, the (FM-AFM) .................................... 639
principles of AFM such as its operating modes and 23.1.7 Relation Between Frequency Shift
the NC-AFM frequency-modulation method are and Forces .................................. 640
fully explained. Then, in Sect. 23.2, applications 23.1.8 Noise in Frequency Modulation
of NC-AFM to semiconductors, which make clear AFM: Generic Calculation ............... 641
its potential in terms of spatial resolution and 23.1.9 Conclusion .................................. 641
function, are introduced. Next, in Sect. 23.3, ap-
plications of NC-AFM to insulators such as alkali
23.2 Applications to Semiconductors ............. 641
23.2.1 Si(111)-(7 × 7) Surface ...................... 642
halides, fluorides and transition-metal oxides are
23.2.2 Si(100)-(2 × 1) and Si(100)-(2 × 1):H
introduced. Lastly, in Sect. 23.4, applications of
Monohydride Surfaces .................. 643
NC-AFM to molecules such as carboxylate (RCOO− ) 23.2.3 Metal Deposited Si Surface ............ 645
with R = H, CH3 , C(CH3 )3 and CF3 are introduced.
Thus, NC-AFM can observe atoms and molecules on 23.3 Applications to Insulators...................... 647
various kinds of surfaces such as semiconductors, 23.3.1 Alkali Halides, Fluorides
insulators and metal oxides with atomic or mo- and Metal Oxides ......................... 647
lecular resolution. These sections are essential to 23.3.2 Atomically Resolved Imaging
understand the state of the art and future possi- of a NiO(001) Surface ..................... 652
bilities for NC-AFM, which is the second generation
23.4 Applications to Molecules ...................... 654
of atom/molecule technology. 23.4.1 Why Molecules
and Which Molecules? .................. 654
23.1 Atomic Force Microscopy (AFM) ............... 636 23.4.2 Mechanism of Molecular Imaging ... 654
23.1.1 Imaging Signal in AFM .................. 636 23.4.3 Perspectives ................................ 657
23.1.2 Experimental Measurement
and Noise ................................... 637 References .................................................. 658
The scanning tunneling microscope (STM) is an atomic a mechanical method that can even deal with insula-
tool based on an electric method that measures the tor surfaces. Since the invention of noncontact AFM
tunneling current between a conductive tip and a con- (NC-AFM) in 1995, the NC-AFM and NC-AFM-based
ductive surface. It can electrically observe individual methods have rapidly developed into powerful surface
atoms/molecules. It can characterize or analyze the tools on the atomic/molecular scales, because NC-AFM
electronic nature around surface atoms/molecules. In has the following characteristics: (1) it has true atomic
addition, it can manipulate individual atoms/molecules. resolution, (2) it can measure atomic force (so-called
Hence, the STM is the first generation of atom/molecule atomic force spectroscopy), (3) it can observe even in-
technology. On the other hand, the atomic force mi- sulators, and (4) it can measure mechanical responses
croscopy (AFM) is a unique atomic tool based on such as elastic deformation. Thus, NC-AFM is the sec-
636 Part C Scanning-Probe Microscopy
ond generation of atom/molecule technology. Scanning tions to semiconductors will be presented. Next, in
probe microscopy (SPM) such as STM and NC-AFM Sect. 23.3, applications to insulators will be described.
is the basic technology for nanotechnology and also for And, in Sect. 23.4, applications to molecules will be
future bottom-up processes. introduced. These sections are essential to under-
In Sect. 23.1, the principles of NC-AFM will standing the state of the art and future possibilities
be fully introduced. Then, in Sect. 23.2, applica- for NC-AFM.
(STM) [23.3]. The STM is covered in several books next to a flat surface (z is the distance between the plane
and review articles, e.g. [23.4–9]. Early in the devel- connecting the centers of the surface atoms and the cen-
opment of STM it became evident that relatively strong ter of the closest tip atom) the van-der-Waals potential
forces act between a tip in close proximity to a sam- is given by [23.13]
ple. It was found that these forces could be put to good
AH
use in the atomic force microscope (AFM). Detailed VvdW = − . (23.1)
information about the noncontact AFM can be found 6z
in [23.10–12]. The Hamaker constant AH depends on the type of
materials (atomic polarizability and density) of the
23.1.1 Imaging Signal in AFM tip and sample and is of the order of 1 eV for most
solids [23.13].
Figure 23.1 shows a sharp tip close to a sample. The po- When the tip and sample are both conductive and
tential energy between the tip and the sample Vts causes have an electrostatic potential difference U = 0, elec-
a z-component of the tip–sample force Fts = −∂Vts /∂z. trostatic forces are important. For a spherical tip with
Depending on the mode of operation, the AFM uses Fts , radius R, the force is given by [23.14]
or some entity derived from Fts , as the imaging signal.
πε0 RU 2
Unlike the tunneling current, which has a very Felectrostatic = − . (23.2)
strong distance dependence, Fts has long- and short- z
range contributions. We can classify the contributions Chemical forces are more complicated. Empirical
by their range and strength. In vacuum, there are van- model potentials for chemical bonds are the Morse po-
der-Waals, electrostatic and magnetic forces with a long tential (see e.g. [23.13])
range (up to 100 nm) and short-range chemical forces
(fractions of nm). VMorse = −E bond 2 e−κ(z−σ) − e−2κ(z−σ) (23.3)
Tip
Sample
Part C 23.1
0.1
microfabricated silicon cantilevers were first created in 0.01 0.1 1 10 100 1000 10 000
the group of Quate [23.15–17] and at IBM [23.18]. f (Hz)
The cantilever is characterized by its spring con- Fig. 23.3 Schematic view of 1/ f noise apparent in force detectors.
stant k, eigenfrequency f 0 and quality factor Q. Static AFMs operate in a frequency range from 0.01 Hz to a few
For a rectangular cantilever with dimensions w, hundred Hz, while dynamic AFMs operate at frequencies around
t and L (Fig. 23.2), the spring constant k is given 10 kHz to a few hundred kHz. The noise of the cantilever deflec-
by [23.6] tion sensor is characterized by the 1/ f corner frequency f c and the
E Y wt 3 constant deflection noise density n q for the frequency range where
k= , (23.5) white noise dominates
4L 3
where E Y is the Young’s modulus. The eigenfrequency was metalized, and a tunneling tip was brought close
f 0 is given by [23.6] to it to measure the deflection [23.2]. Today’s designs
t E use optical (interferometer, beam-bounce) or electrical
f 0 = 0.162 2 , (23.6) methods (piezoresistive, piezoelectric) to measure the
L ρ
cantilever deflection. A discussion of the various tech-
where ρ is the mass density of the cantilever mater- niques can be found in [23.19], descriptions of piezore-
ial. The Q-factor depends on the damping mechanisms sistive detection schemes are found in [23.17, 20] and
present in the cantilever. For micromachined cantilevers piezoelectric methods are explained in [23.21–24].
operated in air, Q is typically a few hundred, while Q The quality of the cantilever deflection measure-
can reach hundreds of thousands in vacuum. ment can be expressed in a schematic plot of the
In the first AFM, the deflection of the cantilever was deflection noise density versus frequency as in Fig. 23.3.
measured with an STM; the back side of the cantilever The noise density has a 1/ f dependence for low fre-
quency and merges into a constant noise density (white
noise) above the 1/ f corner frequency.
[110]
the tip and sample, the cantilever should be much softer AFM [23.31], the actuator is driven by a fixed amplitude
than the bonds between the bulk atoms in the tip and Adrive at a fixed frequency f drive where f drive is close to
sample. Interatomic force constants in solids are in the f 0 . When the tip approaches the sample, elastic and in-
range 10–100 N/m; in biological samples, they can be elastic interactions cause a change in both the amplitude
as small as 0.1 N/m. Thus, typical values for k in the and the phase (relative to the driving signal) of the can-
static mode are 0.01–5 N/m. tilever. These changes are used as the feedback signal.
Even though it has been demonstrated that atomic While the AM mode was initially used in a noncon-
resolution is possible with static AFM, the method can tant mode, it was later implemented very successfully at
only be applied in certain cases. The detrimental ef- a closer distance range in ambient conditions involving
fects of 1/ f -noise can be limited by working at low repulsive tip–sample interactions.
temperatures [23.25], where the coefficients of ther- The change in amplitude in AM mode does not oc-
mal expansion are very small or by building the AFM cur instantaneously with a change in the tip–sample
using a material with a low thermal-expansion coeffi- interaction, but on a timescale of τAM ≈ 2Q/ f 0 and the
Part C 23.1
cient [23.26]. The long-range attractive forces have to AM mode is slow with high-Q cantilevers. However,
be canceled by immersing the tip and sample in a li- the use of high Q-factors reduces noise. Albrecht et al.
quid [23.26] or by partly compensating the attractive found a way to combine the benefits of high Q and high
force by pulling at the cantilever after jump-to-contact speed by introducing the frequency-modulation (FM)
has occurred [23.27]. Jarvis et al. have canceled the mode [23.32], where the change in the eigenfrequency
long-range attractive force with an electromagnetic settles on a timescale of τFM ≈ 1/ f 0 .
force applied to the cantilever [23.28]. Even with these Using the FM mode, the resolution was improved
restrictions, static AFM does not produce atomic reso- dramatically and finally atomic resolution [23.33, 34]
lution on reactive surfaces like silicon, as the chemical was obtained by reducing the tip–sample distance and
bonding of the AFM tip and sample poses an unsur- working in vacuum. For atomic studies in vacuum, the
mountable problem [23.29, 30]. FM mode (Sect. 23.1.6) is now the preferred AFM tech-
nique. However, atomic resolution in vacuum can also
23.1.4 Dynamic AFM Operating Mode be obtained with the AM mode, as demonstrated by
Erlandsson et al. [23.35].
In the dynamic operation modes, the cantilever is
deliberately vibrated. There are two basic methods 23.1.5 The Four Additional Challenges
of dynamic operation: amplitude-modulation (AM) Faced by AFM
and frequency-modulation (FM) operation. In AM-
Some of the inherent AFM challenges are apparent by
comparing the tunneling current and tip–sample force
Fts (z) (nN) It (z) (nA)
4 as a function of distance (Fig. 23.4).
Short-range force (Morse potential) The tunneling current is a monotonic function of the
3 Long-range (vdW) force tip–sample distance and has a very sharp distance de-
2
Total force pendence. In contrast, the tip–sample force has long-
Tunneling current and short-range components and is not monotonic.
1
Jump-to-Contact and Other Instabilities
0
If the tip is mounted on a soft cantilever, the initially
–1 attractive tip–sample forces can cause a sudden jump-
to-contact when approaching the tip to the sample. This
–2 instability occurs in the quasistatic mode if [23.36, 37]
–3 2
∂ Vts
k < max − 2 = ktsmax . (23.7)
–4 ∂z
0 5 10 15 z (Å) 20
Jump-to-contact can be avoided even for soft cantilevers
Fig. 23.4 Plot of the tunneling current It and force Fts (typical val- by oscillating at a large enough amplitude A [23.38]
ues) as a function of the distance z between the front atom and
surface atom layer k A > max (−Fts ) . (23.8)
Noncontact Atomic Force Microscopy and Related Topics 23.1 Atomic Force Microscopy (AFM) 639
Part C 23.1
Waals and chemical forces in vacuum. All of these
force types except for the chemical forces have strong
long-range components which conceal the atomic force
components. For imaging by AFM with atomic res-
olution, it is desirable to filter out the long-range
force contributions and only measure the force com-
ponents which vary on the atomic scale. While there
is no way to discriminate between long- and short-
range forces in static AFM, it is possible to enhance
the short-range contributions in dynamic AFM by
proper choice of the oscillation amplitude A of the
cantilever. 10 nm
the STM [23.39] and, in 1994, this surface was im- tion q (t) of the tip of the cantilever: it oscillates with an
aged by AFM with true atomic resolution for the first amplitude A at a distance q(t) from a sample.
time [23.33] (Fig. 23.6).
The initial parameters which provided true atomic Generic Calculation
resolution (see caption of Fig. 23.6) were found empir- The Hamiltonian of the cantilever is
ically. Surprisingly, the amplitude necessary to obtain p2 kq 2
good results was very large compared to atomic dimen- H= + + Vts (q) (23.10)
2m ∗ 2
sions. It turned out later that the amplitudes had to be so ∗
where p = m dq / dt. The unperturbed motion is given
large to fulfill the stability criteria listed in Sect. 23.1.5.
by
Cantilevers with k ≈ 2000 N/m can be operated with
amplitudes in the Å-range [23.24]. q (t) = A cos(2π f 0 t) (23.11)
and the frequency is
23.1.7 Relation Between Frequency Shift
1 k
and Forces f0 = .
Part C 23.1
(23.12)
2π m ∗
The cantilever (spring constant k, effective mass m ∗ ) is If the force gradient kts = −∂Fts /∂z = ∂ Vts /∂z 2 is
2
a macroscopic object and its motion can be described constant during the oscillation cycle, the calculation of
by classical mechanics. Figure 23.7 shows the deflec- the frequency shift is trivial
f0
Δ f = kts . (23.13)
2k
q(t) q'(t) However, in classic FM-AFM kts varies over orders
of magnitude during one oscillation cycle and a pertur-
d+2A A bation approach, as shown below, has to be employed
Cantilever
for the calculation of the frequency shift.
d+A 0
Hamilton–Jacobi Method
d –A The first derivation of the frequency shift in FM-AFM
d
0
Sample was achieved in 1997 [23.38] using canonical perturba-
tion theory [23.40]. The result of this calculation is
f0
Fig. 23.7 Schematic view of an oscillating cantilever and
Δ f = − 2 Fts q
definition of geometric terms kA
f0
1/
f0
kts (N/m), weight function w(z,A) (0.1/nm) =− 2 Fts (d + A + q (t))q (t) dt . (23.14)
75 kA
0
Tip–sample force gradient The applicability of first-order perturbation theory
50 w (z, A = 1 Å) is justified because, in FM-AFM, E is typically in
w (z, A = 5 Å) the range of several keV, while Vts is of the order of
a few eV. Dürig [23.41] found a generalized algo-
25 w (z, A = 10 Å)
rithm that even allows one to reconstruct the tip–sample
potential if not only the frequency shift, but the higher
0 harmonics of the cantilever oscillation are known.
A Descriptive Expression for Frequency Shifts
– 25 as a Function of the Tip–Sample Forces
With integration by parts, the complicated expression
(23.14) is transformed into a very simple expression that
– 50 resembles (23.13) [23.42]
2 3 4 5 6 7 8 9 10 11 12 13
z (Å)
A A2 − q 2
f0
Fig. 23.8 The tip–sample force gradient kts and weight function for Δf = kts (z − q ) π 2 dq . (23.15)
2k 2 kA
the calculation of the frequency shift −A
Noncontact Atomic Force Microscopy and Related Topics 23.2 Applications to Semiconductors 641
Part C 23.2
tude confirms an empirical conjecture: small amplitudes
increase the sensitivity to short-range forces. Adjust- surement is given by the noise of the frequency measurement δΔf
ing the amplitude in FM-AFM is comparable to tuning divided by the slope of the frequency shift curve
an optical spectrometer to a passing wavelength. When
short-range interactions are to be probed, the ampli- noise frequency measurement and a steep slope of the
tude should be in the range of the short-range forces. frequency-shift curve.
While using amplitudes in the Å-range has been elusive The frequency noise δΔ f is typically inversely pro-
with conventional cantilevers because of the instabil- portional to the cantilever amplitude A [23.32, 43]. The
ity problems described in Sect. 23.1.5, cantilevers with derivative of the frequency shift with distance is con-
a stiffness of the order of 1000 N/m like those intro- stant for A λ where λ is the range of the tip–sample
duced in [23.23] are well suited for small-amplitude interaction and proportional to A−1.5 for A λ [23.38].
operation. Thus, minimal noise occurs if [23.44]
Figure 23.9 shows a typical curve of frequency Dynamic force microscopy, and in particular frequency-
shift versus distance. Because the distance between the modulation atomic force microscopy has matured into
tip and sample is measured indirectly through the fre- a viable technique that allows true atomic resolution of
quency shift, it is clearly evident from Fig. 23.9 that conducting and insulating surfaces and spectroscopic
the noise in the frequency measurement δΔ f trans- measurements on individual atoms [23.10, 45]. Even
lates into vertical noise δz and is given by the ratio true atomic resolution in lateral force microscopy is now
between δΔ f and the slope of the frequency shift curve possible [23.46]. Challenges remain in the chemical
Δ f (z) (23.16). Low vertical noise is obtained for a low- composition and structural arrangement of the AFM tip.
a) b) c)
a 3 Hz
15 Hz
46.6 Å
Fig. 23.10a–c Noncontact-mode AFM images of a Si(111)-(7 × 7) reconstructed surface obtained using the Si tips
(a) without and (b) with a dangling bond. The scan area is 99 Å × 99 Å. (c) The cross-sectional profiles along the long
Part C 23.2
diagonal of the 7 × 7 unit cell indicated by the white lines in (a) and (b)
Si(111)-(7 × 7) [23.34, 35, 45, 47], InP(110) [23.48] and (type I) was obtained using the Si tip without dan-
Si(100)-(2 × 1) [23.34] surfaces have been successively gling, which is covered with an inert oxide layer.
resolved with true atomic resolution. Furthermore, ther- Figure 23.10b (type II) was obtained using the Si tip
mally induced motion of atoms or atomic-scale point with a dangling bond, on which the Si atoms were de-
defects on a InP(110) surface have been observed at posited due the mechanical soft contact between the tip
room temperature [23.48]. In this section we will de- and the Si surface. The variable frequency shift mode
scribe typical results of atomically resolved noncontact was used. We can see not only adatoms and corner
AFM imaging of semiconductor surfaces. holes but also missing adatoms described by the dimer–
adatom–stacking (DAS) fault model. We can see that
23.2.1 Si(111)-(7 × 7) Surface the image contrast in Fig. 23.10b is clearly stronger than
that in Fig. 23.10a.
Figure 23.10 shows the atomic-resolution images of Interestingly, by using the Si tip with a dangling
the Si(111)-(7 × 7) surface [23.49]. Here, Fig. 23.10a bond, we observed contrast between inequivalent halves
and between inequivalent adatoms of the 7 × 7 unit cell.
Namely, as shown in Fig. 23.11a, the faulted halves
a)
(surrounded with a solid line) are brighter than the un-
faulted halves (surrounded with a broken line). Here, the
positions of the faulted and unfaulted halves were deter-
mined from the step direction. From the cross-sectional
profile along the long diagonal of the 7 × 7 unit cell
in Fig. 23.11b, the heights of the corner adatoms are
slightly higher than those of the adjacent center adatoms
in the faulted and unfaulted halves of the unit cell. The
measured corrugation are in the following decreasing
order: Co-F > Ce-F > Co-U > Ce-U, where Co-F and
Ce-F indicate the corner and center adatoms in faulted
halves, and Co-U and Ce-U indicate the corner and cen-
b) ter adatoms in unfaulted halves, respectively. Averaging
1.7 Å 1.4 Å over several units, the corrugation height differences are
estimated to be 0.25 Å, 0.15 Å and 0.05 Å for Co-F,
23.3 Å Ce-F and Co-U, respectively, with respect to to Ce-U.
Fig. 23.11 (a) Noncontact mode AFM image with contrast of in- This tendency, that the heights of the corner adatoms
equivalent adatoms and (b) a cross-sectional profile indicated by the are higher than those of the center adatoms, is consis-
white line. The halves of the 7 × 7 unit cell surrounded by the solid tent with the experimental results using a silicon tip
line and broken line correspond to the faulted and unfaulted halves, [23.47], although they could not determine the faulted
respectively. The scan area is 89 Å × 89 Å and unfaulted halves of the unit cell in the measured
Noncontact Atomic Force Microscopy and Related Topics 23.2 Applications to Semiconductors 643
AFM images. However, this tendency is completely the AFM data, while the amount of charge of adatom
contrary to the experimental results using a tungsten and the chemical reactivity of adatoms can explain the
tip [23.35]. This difference may originate from the dif- our data. The contrast due to the amount of charge of
ference between the tip materials, which seems to affect adatom means that the AFM image is originated from
the interaction between the tip and the reactive sample the difference of the vdW or electrostatic physical inter-
surface. Another possibility is that the tip is in con- actions between the tip and the valence electrons at the
tact with the surface during the small fraction of the adatoms. The contrast due to the chemical reactivity of
oscillating cycle in their experiments [23.35]. adatoms means that the AFM image is originated from
We consider that the contrast between inequivalent the difference of covalent bonding chemical interaction
adatoms is not caused by tip artifacts for the follow- between the atoms at the tip apex and dangling bond
ing reasons: (1) each adatom, corner hole and defect of adatoms. Thus, we can see there are two possible
was clearly observed, (2) the apparent heights of the interactions which explain the strong contrast between
adatoms are the same whether they are located adjacent inequivalent adatoms of 7 × 7 unit cell observed using
Part C 23.2
to defects or not, and (3) the same contrast in several the Si tip with dangling bond.
images for the different tips has been observed. The weak-contrast image in Fig. 23.10a is due to
It should be noted that the corrugation amplitude vdW and/or electrostatic force interactions. On the
of adatoms ≈ 1.4 Å in Fig. 23.11b is higher than that other hand, the strong-contrast images in Figs. 23.10b
of 0.8–1.0 Å obtained with the STM, although the depth and 23.11a are due to a covalent bonding formation be-
of the corner holes obtained with noncontact AFM is al- tween the AFM tip with Si atoms and Si adatoms. These
most the same as that observed with STM. Moreover, in results indicate the capability of the noncontact-mode
noncontact-mode AFM images, the corrugation ampli- AFM to image the variation in chemical reactivity of Si
tude of adatoms was frequently larger than the depth of adatoms. In the future, by controlling an atomic species
the corner holes. The origin of such large corrugation at the tip apex, the study of chemical reactivity on an
of adatoms may be due to the effect of the chemical atomic scale will be possible using noncontact AFM.
interaction, but is not yet clear.
The atom positions, surface energies, dynamic prop- 23.2.2 Si(100)-(2 × 1) and Si(100)-(2 × 1):H
erties and chemical reactivities on the Si(111)-(7 × 7) Monohydride Surfaces
reconstructed surface have been extensively investi-
gated theoretically and experimentally. From these In order to investigate the imaging mechanism of the
investigations, the possible origins of the contrast be- noncontact AFM, a comparative study between a reac-
tween inequivalent adatoms in AFM images are the tive surface and an insensitive surface using the same tip
followings: the true atomic heights that correspond to is very useful. Si(100)-(2 × 1):H monohydride surface is
the adatom core positions, the stiffness (spring constant) a Si(100)-(2 × 1) reconstructed surface that is terminated
of interatomic bonding with the adatoms corresponding by a hydrogen atom. It does not reconstruct as metal
to the frequencies of the surface mode, the charge on is deposited on the semiconductor surface. The surface
the adatom, and the chemical reactivity of the adatoms. structure hardly changes. Thus, the Si(100)-(2 × 1):H
Table 23.1 summarizes the decreasing orders of the monohydride surface is one of most useful surface for
inequivalent adatoms for individual property. From Ta- a model system to investigate the imaging mechanism,
ble 23.1, we can see that the calculated adatom heights experimentally and theoretically. Furthermore, whether
and the stiffness of interatomic bonding cannot explain the interaction between a very small atom such as hy-
Table 23.1 Comparison between the adatom heights observed in an AFM image and the variety of properties for inequiv-
alent adatoms
Decreasing order Agreement
AFM image Co-F > Ce-F > Co-U > Ce-U –
Calculated height Co-F > Co-U > Ce-F > Ce-U ×
Stiffness of interatomic bonding Ce-U > Co-U > Ce-F > Co-F ×
Amount of charge of adatom Co-F > Ce-F > Co-U > Ce-U
Experimental chemical reactivity Co-F > Ce-F > Co-U > Ce-U
a) a)
A
3.5 ± 0.
3. 0.1 Å
3.2 ± 0.1 Å
3.
b) b)
9 Hz 6 Hz
Part C 23.2
46 Å 34 Å
Fig. 23.12 (a) Noncontact AFM image of a Si(001)(2 × 1) Fig. 23.13 (a) Noncontact AFM image of Si(001)-(2 × 1):H
reconstructed surface. The scan area was 69 × 46 Å. One surface. The scan area was 69 × 46 Å. One 2 × 1 unit cell
2 × 1 unit cell is outlined with a box. White rows are su- is outlined with a box. White rows are superimposed to
perimposed to show the bright spots arrangement. The show the bright spots arrangement. The distance between
distance between the bright spots on the dimer row is the bright spots on the dimer row is 3.5 ± 0.1 Å. (b) Cross-
3.2 ± 0.1 Å. On the white arc, the alternative bright spots sectional profile indicated by the white dotted line
are shown. (b) Cross-sectional profile indicated by the
white dotted line The bright spots in Fig. 23.12 do not merely image
the silicon-atom site, because the distance between the
drogen and a tip apex is observable with noncontact bright spots forming the dimer structure of Fig. 23.12a,
AFM is interested. Here, we show noncontact AFM 3.2 ± 0.1 Å, is lager than the distance between silicon
images measured on a Si(100)-(2 × 1) reconstructed sur- atoms of every dimer structure model. (The maximum
face with a dangling bond and on a Si(100)-(2 × 1):H is the distance between the upper silicones in an asym-
monohydride surface on which the dangling bond is metric dimer structure 2.9 Å.) This seems to be due to
terminated by a hydrogen atom [23.50]. the contribution to the imaging of the chemical bonding
Figure 23.12a shows the atomic-resolution image of interaction between the dangling bond from the apex of
the Si(100)-(2 × 1) reconstructed surface. Pairs of bright the silicon tip and the dangling bond on the Si(100)-
spots arranged in rows with a 2 × 1 symmetry were (2 × 1) reconstructed surface. Namely, the chemical
observed with clear contrast. Missing pairs of bright bonding interaction operates strongly, with strong direc-
spots were also observed, as indicated by arrows. Fur- tion dependence, between the dangling bond pointing
thermore, the pairs of bright spots are shown by the out of the silicon dimer structure on the Si(100)-(2 × 1)
white dashed arc and appear to be the stabilize-buckled reconstructed surface and the dangling bond pointing
asymmetric dimer structure. Furthermore, the distance out of the apex of the silicon tip; a dimer structure is
between the pairs of bright spots is 3.2 ± 0.1 Å. obtained with a larger separation than between silicones
Figure 23.13a shows the atomic-resolution im- on the surface.
age of the Si(100)-(2 × 1):H monohydride surface. The bright spots in Fig. 23.13 seem to be located at
Pairs of bright spots arranged in rows were ob- hydrogen atom sites on the Si(100)-(2 × 1):H monohy-
served. Missing paired bright spots as well as those dride surface, because the distance between the bright
paired in rows and single bright spots were ob- spots forming the dimer structure (3.5 ± 0.1 Å) approx-
served, as indicated by arrows. Furthermore, the imately agrees with the distance between the hydrogens,
distance between paired bright spots is 3.5 ± 0.1 Å. i. e., 3.52 Å. Thus, the noncontact AFM atomically re-
This distance of 3.5 ± 0.1 Å is 0.2 Å larger than solved the individual hydrogen atoms on the topmost
that of the Si(100)-(2 × 1) reconstructed surface. layer. On this surface, the dangling bond is terminated
Namely, it is found that the distance between by a hydrogen atom, and the hydrogen atom on the
bright spots increases in size due to the hydrogen topmost layer does not have chemical reactivity. There-
termination. fore, the interaction between the hydrogen atom on
Noncontact Atomic Force Microscopy and Related Topics 23.2 Applications to Semiconductors 645
the topmost layer and the apex of the silicon tip does
not contribute to the chemical bonding interaction with Top view
strong direction dependence as on the silicon surface,
Si
and the bright spots in the noncontact AFM image cor-
respond to the hydrogen atom sites on the topmost layer. 2.31 Å
Part C 23.2
Ag (hereafter referred to as 3-Ag) surface was
used. √
For the 3-Ag surface, the honeycomb-chained 3.43 Å
trimer (HCT) model has been accepted as the appro-
priate model. As shown in Fig. 23.14, this structure
contains a Si trimer in the second layer, 0.75 Å below [112]
the Ag trimer in the topmost layer. The topmost Ag
atoms and lower Si atoms form covalent bonds. The
interatomic distances between the nearest-neighbor Ag Side view
atoms forming the Ag trimer and between the lower Si
atoms forming the Si trimer are 3.43 and 2.31 Å, re- 0.75 Å
spectively. The apexes of the Si trimers and Ag trimers
face the [112̄] direction and the direction tilted a little to
the [1̄1̄2] direction, respectively.
In Fig. 23.15, we show the noncontact AFM images
measured using a normal Si tip at a frequency shift of [112]
(a) −37 Hz, (b) −43 Hz and (c) −51 Hz, respectively.
These frequency shifts correspond to tip–sample dis-
tances of about 0–3 Å. We defined the zero position √ √
of the tip–sample distance, i. e., the contact point, as Fig. 23.14 HCT model for the structure of the Si(111)-( 3 × 3)-
the point at which the vibration amplitude
√ √began to de- Ag surface. Black closed circle, gray closed circle, open circle, and
crease. The rhombus indicates the 3 × 3 unit cell. closed circle with horizontal line indicate Ag atom at the topmost
When the tip approached the surface, the contrast of layer, Si atom at the second layer, Si atom at the third layer, and
the noncontact AFM images become strong and the Si
√ atom√ at the fourth layer, respectively. The rhombus indicates the
pattern changed remarkably. That is, by approaching 3 × 3 unit cell. The thick, large, solid triangle indicates an Ag
the tip toward the sample surface, the hexagonal pat- trimer. The thin, small, solid triangle indicates a Si trimer
tern, the trefoil-like pattern composed of three dark
lines, and the triangle pattern can be observed sequen- spots is 3.9 ± 0.2Å. In Fig. 23.15c, the distance between
tially. In Fig. 23.15a, the distance between the bright the bright spots is 3.0 ± 0.2 Å, and the direction of the
a) b) c)
Fig. 23.15a–c Noncontact AFM images ob-
tained at frequency shifts of (a) −37 Hz,√
(b) −43 Hz, and (c) −51 Hz on a Si(111)-( 3 ×
√
≈ 3.
3.9 Å ≈ 3.
3.0 Å
3)-Ag surface. This distance dependence
was obtained with a Si tip. The scan √
area √
is
– – –
[11 ]
[112 [11 ]
[112 [11 ]
[112 38 Å × 34 Å. A rhombus indicates the 3 × 3
unit cell
646 Part C Scanning-Probe Microscopy
apex of all the triangles composed of three bright spots model within the experimental error. Furthermore, the
is [112̄]. direction of the apex of the triangles composed of
In Fig. 23.16, we show the noncontact AFM im- three bright spots also agrees with the direction of the
ages measured by using Ag-absorbed tip at a frequency apex of the Ag trimer, i. e., tilted [1̄1̄2], in the HCT
shift of (a) − 4.4 Hz, (b) − 6.9 Hz and (c) − 9.4 Hz,
Part C 23.2
the center of the Ag trimer. As a result, bright contrast apex of the Si tip and a Si−Ag covalent bond on the
is obtained at the center of the Ag trimer with the STM. surface. Hence, the individual Ag atoms will not be re-
Finally, we consider the origin of the atomic- solved and the image pattern will change depending on
resolution
√ imaging of the individual Ag atoms on the the tip–sample distance. On the other hand, as shown
3-Ag surface. Here, we discuss the difference be- in Fig. 23.17b, by using the Ag-adsorbed tip, the dan-
tween the force interactions when using the Si tip and gling bond localized out of topmost Si atom on the apex
the Ag-adsorbed tip. As shown in Fig. 23.17a, when us- of the Si tip is terminated by the adsorbed Ag atom.
ing the Si tip, there is a dangling bond pointing out As a result, even at very close tip–sample distances,
of the topmost Si atom on the apex of the Si tip. As the force interaction is dominated by physical bond-
a result, the force interaction is dominated by physi- ing interactions such as the vdW force. Namely, if the
cal bonding interactions, such as the Coulomb force, Ag-adsorbed tip approaches the surface, the vdW force
far from the surface and by chemical bonding inter- acts between the Ag atom on the apex of the tip and
action very close to the surface. Namely, if a reactive the Ag or Si atom√on the surface. Ag atoms in the top-
Part C 23.3
Si tip with a dangling bond approaches a surface, at most layer of the 3-Ag surface are located higher than
distances far from the surface the Coulomb force acts the Si atoms in the lower layer. Hence, the individual
between the electron localized on the dangling bond Ag atoms (or their nearly true topography) will be re-
pointing out of the topmost Si atom on the apex of the solved, and the image pattern will not change even at
tip, and the positive charge distributed around the center very small tip–sample distances. It should be empha-
of the Ag trimer. At distances very close to the sur- sized that there is a possibility to identify or recognize
face, the chemical bonding interaction will occur due atomic species on a sample surface using noncontact
to the onset of orbital hybridization between the dan- AFM if we can control the atomic species at the tip
gling bond pointing out of the topmost Si atom on the apex.
formation on an ionic surface [23.57]. A significant part the atomic corrugation varies strongly between the po-
of the contrast is due to the displacement of ions in sitions of the ions imaged as depressions. This variation
the force field, not only enhancing the atomic corruga- in the corrugations corresponds to the constituents of
tions, but also contributing to the electrostatic potential the crystal, i. e. the Cl and Br ions, and it is concluded
by forming dipoles at the surface. The experimentally that the tip apex is negatively charged. Moreover, the
observed atomic corrugation height is determined by deep depressions can be assigned to Br ions by com-
the interplay of the long- and short-range forces. In the paring the number with the relative density of anions.
case of NaCl, it has been experimentally demonstrated The difference between Cl and Br anions with differ-
that a blunter tip produces a lager corrugation when ent masses is enhanced in the damping signal measured
the tip–sample distance is shorter [23.54]. This result simultaneously with the topographic image [23.59].
shows that the increased long-range forces induced by The damping is recorded as an increase in the exci-
a blunter tip allow for more stable imaging closer to tation amplitude necessary to maintain the oscillation
the surface. The stronger electrostatic short-range in- amplitude of the cantilever in the constant-amplitude
teraction and lager ion displacement produce a more mode [23.56]. Although the dissipation phenomena on
pronounced atomic corrugation. At steps and kinks on an atomic scale are a subject under discussion, any dis-
an NaCl thin film on Cu(111), the corrugation ampli- sipative interaction must generally induce energy losses
tude of atoms with low coordination number has been in the cantilever oscillation [23.60, 61]. The measure-
observed to increase by a factor of up to two more than ment of energy dissipation has the potential to enable
that of atomically flat terraces [23.56]. The low coordi- chemical discrimination on an atomic scale. Recently,
nation number of the ions results in an enhancement of a new procedure for species recognition on a alkali
the electrostatic potential over the site and an increase in halide surface was proposed [23.62]. This method is
the displacement induced by the interaction with the tip. based on a comparison between theoretical results and
Theoretical study predicts that the image contrast the site-specific measurement of frequency versus dis-
depends on the chemical species at the apex of the tip. tance. The differences in the force curves measured at
Bennewitz et al. [23.56] have performed the calcula- the typical sites, such as protrusion, depression, and
tions using an MgO tip terminated by oxygen and an their bridge position, are compared to the correspond-
Mg ion. The magnitude of the atomic contrast for the ing differences obtained from atomistic simulation. The
Mg-terminated tip shows a slight increase in compari- polarity of the tip apex can be determined, leading to the
son with an oxygen-terminated tip. The atomic contrast identification of the surface species. This method is ap-
with the oxygen-terminated tip is dominated by the at- plicable to highly symmetric surfaces and is useful for
tractive electrostatic interaction between the oxygen on determining the sign of the tip polarity.
the tip apex and the Na ion, but the Mg-terminated tip
attractively interacts with the Cl ion. In other words, Fluorides
these results demonstrated that the species of the ion im- Fluorides are important materials for the progress of
aged as the bright protrusions depends on the polarity of an atomic-scale-resolution NC-AFM imaging of insula-
the tip apex. tors. There are reports in the literature of surface images
These theoretical results emphasized the importance for single-crystal BaF2 , SrF2 [23.63], CaF2 [23.64–66]
of the atomic species at the tip apex for the alkali halide and a CaF bilayer on Si(111) [23.67]. Surfaces of
Noncontact Atomic Force Microscopy and Related Topics 23.3 Applications to Insulators 649
Part C 23.3
loop gain, can be categorized into two contrast patterns.
In the first of these the ions appear as triangles and in 6
the second they have the appearance of circles, similar 4
to the contrast obtained in a topographic image. The- 2
oretical studies demonstrated that these two different 0
contrast patterns could be explained as a result of imag- 0 0.5 1 1.5 2
ing with tips of different polarity [23.68–70]. When c) f)
imaging with a positively charged (cation-terminated)
10
tip, the triangular pattern appears. In this case, the con-
trast is dominated by the strong short-range electrostatic 8
attraction between the positive tip and the negative 6
F ions. The cross section along the [121] direction of 4
the triangular image shows two maxima: one is a larger 2
peak over the F(I) ions located in the topmost layer 0
and the other is a smaller peak at the position of the 0 0.5 1 1.5 2
Distance (nm)
F(III) ions in the third layer. The minima appear at
the position of the Ca ions in the second layer. When Fig. 23.18 (a)–(c) CaF2 (111) surface images obtained by using the
imaging with a negatively charged (anion-terminated) constant-height mode. From (a) to (c) the frequency shift was low-
tip, the spherical image contrast appears and the main ered. The white lines represent the positions of the cross section.
periodicity is created by the Ca ions between the top- (d)–(f) The cross section extracted from the Fourier-filtered im-
most and the third F ion layers. In the cross section ages of (a)–(c). The white and black arrows represent the scanning
along the [121] direction, the large maxima correspond direction. The images and the cross sections are from [23.68]
to the Ca sites because of the strong attraction of the
negative tip and the minima appear at the sites of maxi- theoretical works. As the tip approaches, the amplitude
mum repulsion over the F(I) ions. At a position between of the shoulder increases until it is equal to that of the
two F ions, there are smaller maxima. This reflects the main peak, and this feature gives rise to the honeycomb
weaker repulsion over the F(III) ion sites compared to pattern image, as shown in Fig. 23.18c. Moreover, the-
the protruding F(I) ion sites and a slower decay in the oretical results predict that the image contrast changes
contrast on one side of the Ca ions. again when the tip apex is in close proximity to sur-
The triangular pattern obtained with a positively face. Recently, Giessibl and Reichling [23.71] achieved
charged tip appears at relatively large tip–sample dis- atomic imaging in the repulsive region and proved ex-
tance, as shown in Fig. 23.18a. The cross section along perimentally the predicted change of the image contrast.
the [121] direction, experiment results and theoretical As described here, there is good correspondence in the
studies both demonstrate the large-peak and small- distance dependency of the image obtained by experi-
shoulder characteristic for the triangular pattern image mental and theoretical investigations.
(Fig. 23.18d). When the tip approaches the surface more From detailed theoretical analysis of the electro-
closely, the triangular pattern of the experimental im- static potential [23.72], it was suggested that the change
ages is more vivid (Fig. 23.18b), as predicted in the in displacement of the ions due to the proximity of the
650 Part C Scanning-Probe Microscopy
tip plays an important role in the formation of the im- tal and theoretical studies will help to advance surface
age contrast. Such a drastic change in image contrast, imaging of insulators by NC-AFM.
depending on both the polarity of the terminated tip
atom and on the tip–sample distance, is inherent to the Metal Oxides
fluoride (111) surface, and this image-contrast feature Most of the metal oxides that have attracted strong in-
cannot be seen on the (001) surface of alkali halides terest for their technological importance are insulating.
with a simple crystal structure. Therefore, in the case of atomically resolved imaging
The results of careful experiments show another of metal oxide surfaces by STM, efforts to increase
feature: that the cross sections taken along the three the conductivity of the sample are needed, such as, the
equivalent [121] directions do not yield identical re- introduction of anions or cations defects, doping with
sults [23.68]. It is thought that this can be attributed other atoms and surface observations during heating of
to the asymmetry of the nanocluster at the tip apex, the sample. However, in principle, NC-AFM provides
which leads to different interactions in the equivalent the possibility of observing nonconductive metal oxides
Part C 23.3
directions. A better understanding of the asymmetric without these efforts. In cases where the conductivity
image contrast may require more complicated model- of the metal oxides is high enough for a tunneling cur-
ing of the tip structure. In fact, it should be mentioned rent to flow, it should be noted that most surface images
that perfect tips on an atomic scale can occasionally be obtained by NC-AFM and STM are not identical.
obtained. These tips do yield identical results in for- Since the first report of atomically resolved im-
ward and backward scanning, and cross sections in the ages on a TiO2 (110) surface with oxygen point
three equivalent directions taken with this tip are almost defects [23.75], they have also been reported on ru-
identical [23.74]. tile TiO2 (100) [23.76–78], anatase TiO2 (001) thin film
The fluoride (111) surface is an excellent standard on SrTiO3 (100) [23.79] and on LaAO3 (001) [23.80],
surface for calibrating tips on an atomic scale. The SnO2 (110) [23.81], NiO(001) [23.82, 83], SrTiO3 (100)
polarity of the tip-terminated atom can be determined [23.84], CeO2 (111) [23.85] and MoO3 (010) [23.86]
from the image contrast pattern (spherical or triangu- surfaces. Also, Barth and Reichling have succeeded
lar pattern). The irregularities in the tip structure can in obtaining atomically resolved NC-AFM images of
be detected, since the surface structure is highly sym- a clean α-Al2 O3 (0001) surface [23.73] and of a UHV
metric. Therefore, once such a tip has been prepared, cleaved MgO(001) [23.87] surface, which are impossi-
it can be used as a calibrated tip for imaging unknown ble to investigate using STM. In this section we describe
surfaces. typical results of the imaging of metal oxides by NC-
The polarity and shape of the tip apex play an im- AFM.
portant role in interpreting NC-AFM images of alkali The α-Al2 O3 (0001) surface exists in several ordered
halide and fluorides surfaces. It is expected that the phases that can reversibly be transformed into each
achievement of good correlation between experimen- other by thermal treatments and oxygen exposure. It
a) b) c) d)
+9°
3 nm [1120] 1 nm
is√known √ that the high-temperature phase has a large which are not observed by STM, are arranged along
( 31 × 31)R ± 9◦ unit cell. However, the details of the the [001] and [010] directions in the NC-AFM image.
atomic structure of this surface have not been revealed, A theoretical simulation of the NC-AFM image using
and two models have been proposed. Barth and Reich- first-principles calculations shows that the bright and
ling [23.73] have directly observed this reconstructed α- dark spots correspond to Sr and oxygen atoms, respec-
Al2 O3 (0001) surface by NC-AFM. They confirmed that tively. It has been proposed that the structural model
the dominant contrast of the low-magnification image of the reconstructed surface consists of an ordered Sr
corresponds
√ √ to a rhombic grid representing a unit cell adatom located at the oxygen fourfold site on the TiO2 -
of ( 31 × 31)R + 9◦ , as shown in Fig. 23.19a. Also, terminated layer (Fig. 23.20c).
more details of the atomic structures were determined Because STM images are related to the spatial dis-
from the higher-magnification image (Fig. 23.19b), tribution of the wave functions near the Fermi level,
which was taken at a reduced tip–sample distance. In atoms without a local density of states near the Fermi
this atomically resolved image, it was revealed that each level are generally invisible even on conductive mater-
Part C 23.3
side of the rhombus is intersected by ten atomic rows, ials. On the other hand, the NC-AFM image reflects
and that a hexagonal arrangement of atoms exists in the the strength of the tip–sample interaction force originat-
center of the rhombi (Fig. 23.19c). This feature agrees ing from chemical, electrostatic and other interactions.
with the proposed surface structure that predicts order in Therefore, even STM and NC-AFM images obtained
the center of the hexagonal surface domains and disor- using an identical tip and sample may not be identical
der at the domain boundaries. Their result is an excellent generally. The simultaneous imaging of a metal ox-
demonstration of the capabilities of the NC-AFM for ide surface enables the investigation of a more detailed
the atomic-scale surface investigation of insulators. √ surface structure. The images of a TiO2 (110) surface si-
√ The atomic structure of the SrTiO3 (100)-( 5 multaneously obtained with STM and NC-AFM [23.78]
× 5)R26.6◦ surface, as well as that of Al2 O3 (0001) are a typical example. The STM image shows that the
can be determined on the basis of the results of NC- dangling-bond states at the tip apex overlap with the
AFM imaging [23.84]. SrTiO3 is one of the perovskite dangling bonds of the 3d states protruding from the Ti
oxides, and its (100) surface exhibits the many differ- atom, while the NC-AFM primarily imaged the upper-
ent
√ kinds
√ of reconstructed structures. In the case of the most oxygen atom.
( 5 × 5)R26.6◦ reconstruction, the oxygen vacancy– Recently, calculations of the interaction of a Si
Ti3+ –oxygen model (where the terminated surface is tip with metal oxides surfaces, such as Al2 O3 (0001),
TiO2 and the observed spots are related to oxygen TiO2 (110), and MgO(001), were reported [23.88, 89].
vacancies) was proposed from the results of STM imag- Previous simulations of AFM imaging of alkali halides
ing. As shown in Fig. 23.20, Kubo and Nozoye [23.84] and fluorides assume that the tip would be oxides or
have performed measurements using both STM and contaminated and hence have been performed with
NC-AFM, and have found that the size of the bright a model of ionic oxide tips. In the case of imaging
spots as observed by NC-AFM is always smaller than a metal oxide surface, pure Si tips are appropriate for
that for STM measurement, and that the dark spots, a more realistic tip model because the tip is sputtered for
a) b) c)
a b
[010] Sr atom
Ti atom
a b 2 nm c [001]
10 nm O atom
Fig. √ (a) STM and (b) NC-AFM images of a SrTiO3 (100) surface. (c) A proposed model of the SrTiO3 (100)-
√ 23.20
( 5 × 5)R26.6◦ surface reconstruction. The images and the schematic representations are from [23.84]
652 Part C Scanning-Probe Microscopy
cleaning in many experiments. The results of ab initio force interaction between a magnetic tip and a sample
calculations for a Si tip with a dangling bond demon- will be described.
strate that the balance between polarization of the tip
and covalent bonding between the tip and the surface Theoretical Studies of the Exchange Force
should determine the tip–surface force. The interaction In the system of a magnetic tip and sample, the inter-
force can be related to the nature of the surface elec- action detected by NC-AFM includes the short-range
tronic structure. For wide-gap insulators with a large magnetic interaction in addition to the long-range mag-
valence-band offset that prevents significant electron- netic dipole interaction. The energy of the short-range
density transfer between the tip and the sample, the interaction depends on the electron spin states of the
force is dominated by polarization of the tip. When atoms on the apex of the tip and the sample surface,
the gap is narrow, the charge transfer increase and co- and the energy difference between spin alignments (par-
valent bonding dominates the tip–sample interaction. allel or antiparallel) is referred to as the exchange
The forces over anions (oxygen ions) in the surface interaction energy. Therefore, the short-range magnetic
Part C 23.3
are larger than over cations (metal ions), as they play interaction leads to the atomic-scale magnetic contrast,
a more significant role in charge transfer. This implies depending on the local energy difference between spin
that a pure Si tip would always show the brightest con- alignments.
trast over the highest anions in the surface. In addition, In the past, extensive theoretical studies on the
Foster et al. [23.88] suggested the method of using ap- short-range magnetic interaction between a ferro-
plied voltage, which controls the charge transfer, during magnetic tip and a ferromagnetic sample have been
an AFM measurement to define the nature of tip apex. performed by a simple calculation [23.92], a tight-
The collaboration between experimental and theo- binding approximation [23.93] and first-principles cal-
retical studies has made great progress in interpreting culations [23.94]. In the calculations performed by
the imaging mechanism for binary insulators surface Nakamura et al. [23.94], three-atomic-layer Fe(001)
and reveals that a well-defined tip with atomic resolu- films are used as a model for the tip and sample. The
tion is preferable for imaging a surface. As described exchange force is defined as the difference between
previously, a method for the evaluation of the nature the forces in each spin configuration of the tip and
of the tip has been developed. However, the most de- sample (parallel and antiparallel). The result of this
sirable solution would be the development of suitable calculation demonstrates that the amplitude of the ex-
techniques for well-defined tip preparation and a few change force is measurable for AFM (about 0.1 nN).
attempts at controlled production of Si tips have been Also, they forecasted that the discrimination of the
reported [23.24, 90, 91]. exchange force would enable direct imaging of the
magnetic moments on an atomic scale. Foster and
23.3.2 Atomically Resolved Imaging Shluger [23.95] have theoretically investigated the in-
of a NiO(001) Surface teraction between a spin-polarized H atom and a Ni
atom on a NiO(001) surface. They demonstrated that
The transition metal oxides, such as NiO, CoO, and the difference in magnitude in the exchange interac-
FeO, feature the simultaneous existence of an energy tion between opposite-spin Ni ions in a NiO surface
gap and unpaired electrons, which gives rise to a va- could be sufficient to be measured in a low-temperature
riety of magnetic property. Such magnetic insulators NC-AFM experiment. Recently, first-principles calcu-
are widely used for the exchange biasing for mag- lation of the interaction of a ferromagnetic Fe tip with
netic and spintronic devices. NC-AFM enables direct an NiO surface has demonstrated that it should be feasi-
surface imaging of magnetic insulators on an atomic ble to measure the difference in exchange force between
scale. The forces detected by NC-AFM originate from opposite-spin Ni ions [23.96].
several kinds of interaction between the surface and
the tip, including magnetic interactions in some cases. Atomically Resolved Imaging Using Noncoated
Theoretical studies predict that short-range magnetic and Fe-Coated Si Tips
interactions such as the exchange interaction should en- The detection of the exchange interaction is a chal-
able the NC-AFM to image magnetic moments on an lenging task for NC-AFM applications. An antiferro-
atomic scale. In this section, we will describe imaging magnetic insulator NiO single crystal that has regularly
of the antiferromagnetic NiO(001) surface using a ferro- aligned atom sites with alternating electron spin states
magnetic tip. Also, theoretical studies of the exchange is one of the best candidates to prove the feasibility of
Noncontact Atomic Force Microscopy and Related Topics 23.3 Applications to Insulators 653
Part C 23.3
the atom site. From the results of first-principles cal-
[010]
culations [23.94], the contribution of the short-range
magnetic interaction to the measured corrugation am-
plitude is expected to be about a few percent of the
total interaction. Discrimination of such small perturba-
[100]
tions is therefore needed. In order to reduce the noise,
b) Height (pm) the corrugation amplitude was added on the basis of
the periodicity of the NC-AFM image. In addition, the
50 topographical asymmetry, which is the index character-
40 izing the difference in atomic corrugation amplitude,
30
has been defined [23.101]. The result shows that the
20
10
value of the topographical asymmetry calculated from
0 the image obtained with an Fe-coated Si tip depends
0 1 2 3 4 on the direction of summing of the corrugation am-
Distance along [100] direction (nm)
plitude, and that the dependency corresponds to the
Fig. 23.21 (a) Atomically resolved image obtained with an antiferromagnetic spin ordering of the NiO(001) sur-
Fe-coated tip. (b) Shows the cross sections of the middle face [23.101, 102]. Therefore, this result implies that
part in (a). Their corrugations are about 30 pm the dependency of the topographical asymmetry origi-
nates in the short-range magnetic interaction. However,
detecting the exchange force for the following reason. in some cases the topographic asymmetry with uncoated
NiO has an antiferromagnetic AF2 structure as the most Si tips has a finite value [23.103]. The possibility that
stable below the Néel temperature of 525 K. This well- the asymmetry includes the influence of the structure
defined magnetic structure, in which Ni atoms on the of tip apex and of the relative orientation between the
(001) surface are arranged in a checkerboard pattern, surface and tip cannot be excluded. In addition, it is
leads to the simple interpretation of an image containing suggested that the absence of unambiguous exchange
the atomic-scale contrast originating in the exchange contrast is due to the fact that surface ion instabilities
force. In addition, a clean surface can easily be prepared occur at tip–sample distances that are small enough for
by cleaving. a magnetic interact [23.100]. Another possibility is that
Figure 23.21a shows an atomically resolved image the magnetic properties of the tips are not yet fully con-
of a NiO(001) surface with a ferromagnetic Fe-coated trolled because the topographic asymmetries obtained
tip [23.97]. The bright protrusions correspond to atoms by Fe- and Ni-coated tips show no significant differ-
spaced about 0.42 nm apart, consistent with the ex- ence [23.103]. In any cases, a careful comparison is
pected periodic arrangement of the NiO(001) surface. needed to evaluate the exchange interaction included in
The corrugation amplitude is typically 30 pm, which an atomic image.
is comparable to the value previously reported [23.82, From the aforementioned theoretical works, it is
83, 98–100], as shown in Fig. 23.21b. The atomic- presumed that a metallic tip has the capability to im-
resolution image (Fig. 23.21b), in which there is one age an oxygen atom as a bright protrusion. Recently,
maximum and one minimum within the unit cell, re- the magnetic properties of the NiO(001) surface were
654 Part C Scanning-Probe Microscopy
investigated by first-principles electronic-structure cal- The measurements presented here demonstrate the
culations [23.104]. It was shown that the surface oxygen feasibility of imaging magnetic structures on an atomic
has finite spin magnetic moment, which originates from scale by NC-AFM. In order to realize explicit de-
symmetry breaking. We must take into account the pos- tection of exchange force, further experiments and
sibility that a metal atom at the ferromagnetic tip apex a theoretical study are required. In particular, the de-
may interact with a Ni atom on the second layer through velopment of a tip with well-defined atomic structure
a magnetic interaction mediated by the electrons in an and magnetic properties is essential for exchange force
oxygen atom on the surface. microscopy.
Access to individual molecules has not been a trivial are presented in Figs. 23.22 and 23.23 to demonstrate
task even for noncontact atomic force microscopy (NC- the current stage of achievement. The proceedings of
AFM). The force pulling the tip into the surface is less the annual NC-AFM conference represent a conve-
sensitive to the gap width (r), especially when chem- nient opportunity for us to update the list of molecules
ically stable molecules cover the surface. The attractive imaged.
potential between two stable molecules is shallow and
exhibits r −6 decay [23.13]. 23.4.2 Mechanism of Molecular Imaging
High-resolution topography of formate (HCOO− )
[23.109] was first reported in 1997 as a molecu- A systematic study of carboxylates (RCOO− ) with
lar adsorbate. The number of imaged molecules is R = H, CH3 , C(CH3 )3 , C≡CH, and CF3 revealed
now increasing because of the technological impor- that the van der Waals force is responsible for the
tance of molecular interfaces. To date, the following molecule-dependent microscope topography despite its
studies on molecular topography have been published: long-range (r −6 ) nature. Carboxylates adsorbed on the
C60 [23.105, 110], DNAs [23.111, 112], adenine and (110) surface of rutile TiO2 have been extensively
thymine [23.113], alkanethiols [23.113,114], a perylene studied as a prototype for organic materials interfaced
derivative (PTCDA) [23.115], a metal porphyrin (Cu- with an inorganic metal oxide [23.127]. A carboxylic
TBPP) [23.116], glycine sulfate [23.117], polypropy- acid molecule (RCOOH) dissociates on this surface
lene [23.118], vinylidene fluoride [23.119], and a series to a carboxylate (RCOO− ) and a proton (H+ ) at
of carboxylates (RCOO− ) [23.120–126]. Two of these room temperature, as illustrated in Fig. 23.24. The pair
Noncontact Atomic Force Microscopy and Related Topics 23.4 Applications to Molecules 655
Part C 23.4
groups can be quantitatively compared while minimiz-
ing tip-dependent artifacts.
Figure 23.25 presents the observed constant
frequency-shift topography of four carboxylates termi-
nated by different alkyl groups. On the formate-covered
surface of panel (a), individual formates (R = H) were
resolved as protrusions of uniform brightness. The
Height (pm)
dark holes represent unoccupied surface sites. The 0.5
cross section in the lower panel shows that the ac-
curacy of the height measurement was 0.01 nm or
better. Brighter particles appeared in the image when
the formate monolayer was exposed to acetic acid 0 27
(CH3 COOH) as shown in panel (b). Some formates Distance (nm)
a) b) HH
[110] H F F F
C C
[001] H 0.11
0.11 0.15 0.13 0.15
C 0.06 0.46 C 0.46 C
0.38 nm O 126° O O O O O
0.21
Ti Ti Ti Ti Ti Ti
0.6 nm
H
H 0.11
H3C H C H C
0.65 nm CH3 0.12
0.15 C C
0.15 0.64 0.14
0.58
C C
O O O O
Ti Ti Ti Ti
Pivalate Propiolate
Fig. 23.24a,b The carboxylates and TiO2 substrate. (a) Top and side view of the ball model. Small shaded and large
shaded balls represent Ti and O atoms in the substrate. Protons yielded in the dissociation reaction are not shown.
(b) Atomic geometry of formate, acetate, pivalate, propiolate, and trifluoroacetate adsorbed on the TiO2 (110) surface.
The O−Ti distance and O−C−O angle of the formate were determined in the quantitative analysis using photoelectron
diffraction [23.128]
656 Part C Scanning-Probe Microscopy
a) [001] b) c) d)
2 nm 0.20 nm
Part C 23.4
0.11 nm
0.06 nm
0.02 nm
Fig. 23.25a–d The constant frequency-shift topography of carboxylate monolayers prepared on the TiO2 (110) surface
based on [23.121, 123, 125]. Image size: 10 × 10 nm2 . (a) Pure formate monolayer; (b) formate–acetate mixed layer;
(c) formate–pivalate mixed layer; (d) formate–propiolate mixed layer. Cross sections determined on the lines are shown
in the lower panel
were exchanged with acetates (R = CH3 ) impinging H atom terminating the triple-bonded carbon chain in
from the gas phase [23.129]. Because the number of the propiolate is at 0.64 nm. Figure 23.26 summarizes
brighter spots increased with exposure time to acetic the observed image heights relative to the formate,
acid, the brighter particle was assigned to the ac- as a function of the physical height of the topmost
etate [23.121]. Twenty-nine acetates and 188 formates H atoms given in the model. The straight line fitted the
were identified in the topography. An isolated acetate four observations [23.122]. When the horizontal axis
and its surrounding formates exhibited an image height was scaled with other properties (molecular weight, the
difference of 0.06 nm. Pivalate is terminated by bulky number of atoms in a molecule, or the number of elec-
R = (CH3 )3 . Nine bright pivalates were surrounded by trons in valence states), the correlation became poor.
formates of ordinary brightness in the image of panel
(c) [23.123]. The image height difference of an iso-
Microscope topography relative to formate (nm)
lated pivalate over the formates was 0.11 nm. Propiolate
with C≡CH is a needle-like adsorbate of single-atom
0.2
diameter. That molecule exhibited in panel (d) a mi-
croscope topography 0.20 nm higher than that of the
formate [23.125].
The image topography of formate, acetate, pivalate, 0.1
and propiolate followed the order of the size of the
alkyl groups. Their physical topography can be assumed
based on the C−C and C−H bond lengths in the corre- 0
sponding RCOOH molecules in the gas phase [23.130],
and is illustrated in Fig. 23.24. The top hydrogen atom 0.3 0.4 0.5 0.6 0.7
Physical topography relative to Ti-plane (nm)
of the formate is located 0.38 nm above the surface
plane containing the Ti atom pair, while three equiva- Fig. 23.26 The constant frequency-shift topography of the
lent hydrogen atoms of the acetate are more elevated alkyl-substituted carboxylates as a function of their phys-
at 0.46 nm. The uppermost H atoms in the pivalate ical topography given in the model of Fig. 23.3 based
are raised by 0.58 nm relative to the Ti plane. The on [23.123]
Noncontact Atomic Force Microscopy and Related Topics 23.4 Applications to Molecules 657
On the other hand, if the tip apex traced the con- topography of Fig. 23.25d. A calculation that does not
tour of a molecule composed of hard-sphere atoms, include quantum chemical treatment is expected to
the image topography would reproduce the physical work, unless the tip approaches the surface too closely,
topography in a one-to-one ratio, as shown by the bro- or the molecule possesses a dangling bond.
ken line in Fig. 23.26. However, the slope of the fitted In addition to the contribution of the dispersion
line was 0.7. A slope of less than unity is interpreted force, the permanent dipole moment of molecules may
as the long-range nature of the tip–molecule force. perturb the microscope topography through electrostatic
The observable frequency shift reflects the sum of the coupling with the tip. Its possible role was demon-
forces between the tip apex and individual molecules. strated by imaging a fluorine-substituted acetate. The
When the tip passes above a tall molecule embedded strongly polarized C−F bonds were expected to perturb
in short molecules, it is pulled up to compensate for the electrostatic field over the molecule. The constant
the increased force originating from the tall molecule. frequency-shift topography of acetate (R = CH3 ) and
Forces between the lifted tip and the short molecules trifluoroacetate (R = CF3 ) was indeed sensitive to the
Part C 23.4
are reduced due to the increased tip–surface distance. fluorine substitution. The acetate was observed to be
Feedback regulation pushes down the probe to restore 0.05 nm higher than the trifluoroacetate [23.122], al-
the lost forces. though the F atoms in the trifluoroacetate as well as the
This picture predicts that microscope topography is H atoms in the acetate were lifted by 0.46 nm from the
sensitive to the lateral distribution of the molecules, and surface plane, as illustrated in Fig. 23.24.
that was in fact the case. Two-dimensionally clustered
acetates exhibited enhanced image height over an iso- 23.4.3 Perspectives
lated acetate [23.121]. The tip–molecule force therefore
remained nonzero at distances over the lateral separa- The experimental results summarized in this section
tion of the carboxylates on this surface (0.59–0.65 nm). prove the feasibility of using NC-AFM to identify
Chemical bond interactions cannot be important across individual molecules. A systematic study on the con-
such a wide tip–molecule gap, whereas atom-scale im- stant frequency-shift topography of carboxylates with
ages of Si(111)(7 × 7) are interpreted with the fractional R = CH3 , C(CH3 )3 , C≡CH, and CF3 has revealed the
formation of tip–surface chemical bonds [23.24,45,49]. mechanism behind the high-resolution imaging of the
Instead, the attractive component of the van der Waals chemically stable molecules. The dispersion force is
force is probable responsible for the observed molecule- primarily responsible for the molecule-dependent to-
dependent topography. The absence of the tip–surface pography. The permanent dipole moment of the imaged
chemical bond is reasonable on the carboxylate-covered molecule, if it exists, perturbs the topography through
surface terminated with stable C−H bonds. the electrostatic coupling with the tip. A tiny calculation
The attractive component of the van der Waals containing empirical force fields works when simulat-
force contains electrostatic terms caused by permanent- ing the microscope topography.
dipole/permanent-dipole coupling, permanent-dipole/ These results make us optimistic about analyz-
induced-dipole coupling, and induced-dipole/induced- ing physical and chemical properties of nanoscale
dipole coupling (dispersion force). The four car- supramolecular assemblies constructed on a solid sur-
boxylates examined are equivalent in terms of their face. If the accuracy of topographic measurement is
permanent electric dipole, because the alkyl groups are developed by one more order of magnitude, which
nonpolar. The image contrast of one carboxylate rela- is not an unrealistic target, it may be possible to
tive to another is thus ascribed to the dispersion force identify structural isomers, chiral isomers, and confor-
and/or the force created by the coupling between the mational isomers of a molecule. Kelvin probe force
permanent dipole on the tip and the induced dipole microscopy (KPFM), an extension of NC-AFM, pro-
on the molecule. If we further assume that the Si vides a nanoscale analysis of molecular electronic
tip used exhibits the smallest permanent dipole, the properties [23.118,119]. Force spectroscopy with chem-
dispersion force remains dominant to create the NC- ically modified tips seems promising for the detection
AFM topography dependent on the nonpolar groups of of a selected chemical force. Operation in a liquid at-
atoms. A numerical simulation based on this assump- mosphere [23.131] is required for the observation of
tion [23.125] successfully reproduced the propiolate biochemical materials in their natural environment.
658 Part C Scanning-Probe Microscopy
References
23.1 G. Binnig: Atomic force microscope, method for microscope in ultrahigh vacuum, Rev. Sci. Instrum.
imaging surfaces with atomic resolution, US Patent 65, 1923–1929 (1994)
4724318 (1986) 23.21 P. Güthner, U.C. Fischer, K. Dransfeld: Scanning
23.2 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi- near-field acoustic microscopy, Appl. Phys. B 48,
croscope, Phys. Rev. Lett. 56, 930–933 (1986) 89–92 (1989)
23.3 G. Binnig, H. Rohrer, C. Gerber, E. Weibel: Surface 23.22 K. Karrai, R.D. Grober: Piezoelectric tip–sample
studies by scanning tunneling microscopy, Phys. distance control for near field optical microscopes,
Rev. Lett. 49, 57–61 (1982) Appl. Phys. Lett. 66, 1842–1844 (1995)
23.4 G. Binnig, H. Rohrer: The scanning tunneling mi- 23.23 F.J. Giessibl: High-speed force sensor for force mi-
croscope, Sci. Am. 253, 50–56 (1985) croscopy and profilometry utilizing a quartz tuning
23.5 G. Binnig, H. Rohrer: In touch with atoms, Rev. fork, Appl. Phys. Lett. 73, 3956–3958 (1998)
Mod. Phys. 71, S320–S330 (1999) 23.24 F.J. Giessibl, S. Hembacher, H. Bielefeldt,
23.6 C.J. Chen: Introduction to Scanning Tunneling Mi- J. Mannhart: Subatomic features on the silicon
Part C 23
croscopy (Oxford Univ. Press, Oxford 1993) (111)-(7×7) surface observed by atomic force mi-
23.7 H.-J. Güntherodt, R. Wiesendanger (Eds.): Scan- croscopy, Science 289, 422–425 (2000)
ning Tunneling Microscopy I–III (Springer, Berlin, 23.25 F. Giessibl, C. Gerber, G. Binnig: A low-temperature
Heidelberg 1991) atomic force/scanning tunneling microscope for
23.8 J.A. Stroscio, W.J. Kaiser (Eds.): Scanning Tunneling ultrahigh vacuum, J. Vac. Sci. Technol. B 9, 984–
Microscopy (Academic, Boston 1993) 988 (1991)
23.9 R. Wiesendanger: Scanning Probe Microscopy and 23.26 F. Ohnesorge, G. Binnig: True atomic resolution
Spectroscopy: Methods and Applications (Cam- by atomic force microscopy through repulsive and
bridge Univ. Press, Cambridge 1994) attractive forces, Science 260, 1451–1456 (1993)
23.10 S. Morita, R. Wiesendanger, E. Meyer (Eds.): Non- 23.27 F.J. Giessibl, G. Binnig: True atomic resolution on
contact Atomic Force Microscopy (Springer, Berlin, KBr with a low-temperature atomic force micro-
Heidelberg 2002) scope in ultrahigh vacuum, Ultramicroscopy 42–44,
23.11 R. Garcia, R. Perez: Dynamic atomic force mi- 281–286 (1992)
croscopy methods, Surf. Sci. Rep. 47, 197–301 (2002) 23.28 S.P. Jarvis, H. Yamada, H. Tokumoto, J.B. Peth-
23.12 F.J. Giessibl: Advances in atomic force microscopy, ica: Direct mechanical measurement of interatomic
Rev. Mod. Phys. 75, 949–983 (2003) potentials, Nature 384, 247–249 (1996)
23.13 J. Israelachvili: Intermolecular and Surface Forces, 23.29 L. Howald, R. Lüthi, E. Meyer, P. Güthner, H.-
2nd edn. (Academic, London 1991) J. Güntherodt: Scanning force microscopy on the
23.14 L. Olsson, N. Lin, V. Yakimov, R. Erlandsson: Si(111)7×7 surface reconstruction, Z. Phys. B 93, 267–
A method for in situ characterization of tip shape 268 (1994)
in AC-mode atomic force microscopy using elec- 23.30 L. Howald, R. Lüthi, E. Meyer, H.-J. Güntherodt:
trostatic interaction, J. Appl. Phys. 84, 4060–4064 Atomic-force microscopy on the Si(111)7×7 surface,
(1998) Phys. Rev. B 51, 5484–5487 (1995)
23.15 S. Akamine, R.C. Barrett, C.F. Quate: Improved 23.31 Y. Martin, C.C. Williams, H.K. Wickramasinghe:
atomic force microscopy images using cantilevers Atomic force microscope – force mapping and pro-
with sharp tips, Appl. Phys. Lett. 57, 316–318 filing on a sub 100-Å scale, J. Appl. Phys. 61,
(1990) 4723–4729 (1987)
23.16 T.R. Albrecht, S. Akamine, T.E. Carver, C.F. Quate: 23.32 T.R. Albrecht, P. Grütter, H.K. Horne, D. Ru-
Microfabrication of cantilever styli for the atomic gar: Frequency modulation detection using high-Q
force microscope, J. Vac. Sci. Technol. A 8, 3386– cantilevers for enhanced force microscope sensi-
3396 (1990) tivity, J. Appl. Phys. 69, 668–673 (1991)
23.17 M. Tortonese, R.C. Barrett, C. Quate: Atomic res- 23.33 F.J. Giessibl: Atomic resolution of the silicon (111)-
olution with an atomic force microscope using (7×7) surface by atomic force microscopy, Science
piezoresistive detection, Appl. Phys. Lett. 62, 834– 267, 68–71 (1995)
836 (1993) 23.34 S. Kitamura, M. Iwatsuki: Observation of sili-
23.18 O. Wolter, T. Bayer, J. Greschner: Micromachined con surfaces using ultrahigh-vacuum noncontact
silicon sensors for scanning force microscopy, atomic force microscopy, Jpn. J. Appl. Phys. 35,
J. Vac. Sci. Technol. 9, 1353–1357 (1991) 668–L671 (1995)
23.19 D. Sarid: Scanning Force Microscopy, 2nd edn. (Ox- 23.35 R. Erlandsson, L. Olsson, P. Mårtensson: Inequiva-
ford Univ. Press, New York 1994) lent atoms and imaging mechanisms in AC-mode
23.20 F.J. Giessibl, B.M. Trafas: Piezoresistive cantilevers atomic-force microscopy of Si(111)7×7, Phys. Rev. B
utilized for scanning tunneling and scanning force 54, R8309–R8312 (1996)
Noncontact Atomic Force Microscopy and Related Topics References 659
23.36 N. Burnham, R.J. Colton: Measuring the nanome- 23.52 K. Yokoyama, T. Ochi, Y. Sugawara, √ S. Morita:
√
chanical and surface forces of materials using an Atomically resolved Ag imaging on Si(111) 3× 3-Ag
atomic force microscope, J. Vac. Sci. Technol. A 7, surface with noncontact atomic force microscope,
2906–2913 (1989) Phys. Rev. Lett. 83, 5023–5026 (1999)
23.37 D. Tabor, R.H.S. Winterton: Direct measurement of 23.53 M. Bammerlin, R. Lüthi, E. Meyer, A. Baratoff, J. Lü,
normal and related van der Waals forces, Proc. R. M. Guggisberg, C. Gerber, L. Howald, H.-J. Gün-
Soc. Lond. A 312, 435 (1969) therodt: True atomic resolution on the surface of
23.38 F.J. Giessibl: Forces and frequency shifts in atomic an insulator via ultrahigh vacuum dynamic force
resolution dynamic force microscopy, Phys. Rev. B microscopy, Probe Microsc. J. 1, 3–7 (1997)
56, 16011–16015 (1997) 23.54 M. Bammerlin, R. Lüthi, E. Meyer, A. Baratoff, J. Lü,
23.39 G. Binnig, H. Rohrer, C. Gerber, E. Weibel: 7 × 7 re- M. Guggisberg, C. Loppacher, C. Gerber, H.-J. Gün-
construction on Si(111) resolved in real space, Phys. therodt: Dynamic SFM with true atomic resolution
Rev. Lett. 50, 120–123 (1983) on alkali halide surfaces, Appl. Phys. A 66, S293–
23.40 H. Goldstein: Classical Mechanics (Addison Wesley, S294 (1998)
Reading 1980) 23.55 R. Hoffmann, M.A. Lantz, H.J. Hug, P.J.A. van
Part C 23
23.41 U. Dürig: Interaction sensing in dynamic force mi- Schendel, P. Kappenberger, S. Martin, A. Baratoff,
croscopy, New J. Phys. 2, 5.1–5.12 (2000) H.-J. Güntherodt: Atomic resolution imaging and
23.42 F.J. Giessibl: A direct method to calculate tip– force versus distance measurements on KBr(001)
sample forces from frequency shifts in frequency- using low temperature scanning force microscopy,
modulation atomic force microscopy, Appl. Phys. Appl. Surf. Sci. 188, 238–244 (2002)
Lett. 78, 123–125 (2001) 23.56 R. Bennewitz, A.S. Foster, L.N. Kantotovich,
23.43 U. Dürig, H.P. Steinauer, N. Blanc: Dynamic force M. Bammerlin, C. Loppacher, S. Schär, M. Guggis-
microscopy by means of the phase-controlled os- berg, E. Meyer, A.L. Shluger: Atomically resolved
cillator method, J. Appl. Phys. 82, 3641–3651 (1997) edges and kinks of NaCl islands on Cu(111): Ex-
23.44 F.J. Giessibl, H. Bielefeldt, S. Hembacher, periment and theory, Phys. Rev. B 62, 2074–2084
J. Mannhart: Calculation of the optimal imaging (2000)
parameters for frequency modulation atomic force 23.57 A.I. Livshits, A.L. Shluger, A.L. Rohl, A.S. Foster:
microscopy, Appl. Surf. Sci. 140, 352–357 (1999) Model of noncontact scanning force microscopy on
23.45 M.A. Lantz, H.J. Hug, R. Hoffmann, P.J.A. van ionic surfaces, Phys. Rev. 59, 2436–2448 (1999)
Schendel, P. Kappenberger, S. Martin, A. Baratoff, 23.58 R. Bennewitz, O. Pfeiffer, S. Schär, V. Barwich,
H.-J. Güntherodt: Quantitative measurement of E. Meyer, L.N. Kantorovich: Atomic corrugation in
short-range chemical bonding forces, Science 291, nc-AFM of alkali halides, Appl. Surf. Sci. 188, 232–
2580–2583 (2001) 237 (2002)
23.46 F.J. Giessibl, M. Herz, J. Mannhart: Friction traced 23.59 R. Bennewitz, S. Schär, E. Gnecco, O. Pfeiffer,
to the single atom, Proc. Natl. Acad. Sci. USA 99, M. Bammerlin, E. Meyer: Atomic structure of alkali
12006–12010 (2002) halide surfaces, Appl. Phys. A 78, 837–841 (2004)
23.47 N. Nakagiri, M. Suzuki, K. Oguchi, H. Sugimura: Site 23.60 M. Gauthier, L. Kantrovich, M. Tsukada: Theory
discrimination of adatoms in Si(111)-7 × 7 by non- of energy dissipation into surface viblationsed.
contact atomic force microscopy, Surf. Sci. Lett. 373, In: Noncontact Atomic Force Microscopy, ed. by
L329–L332 (1997) S. Morita, R. Wiesendanger, E. Meyer (Springer,
23.48 Y. Sugawara, M. Ohta, H. Ueyama, S. Morita: De- Berlin, Heidelberg 2002) pp. 371–394
fect motion on an InP(110) surface observed with 23.61 H.J. Hug, A. Baratoff: Measurement of dissipation
noncontact atomic force microscopy, Science 270, induced by tip–sample interactions. In: Noncon-
1646–1648 (1995) tact Atomic Force Microscopy, ed. by S. Morita,
23.49 T. Uchihashi, Y. Sugawara, T. Tsukamoto, M. Ohta, R. Wiesendanger, E. Meyer (Springer, Berlin, Hei-
S. Morita: Role of a covalent bonding interaction delberg 2002) pp. 395–431
in noncontact-mode atomic-force microscopy on 23.62 R. Hoffmann, L.N. Kantorovich, A. Baratoff,
Si(111)7 × 7, Phys. Rev. B 56, 9834–9840 (1997) H.J. Hug, H.-J. Güntherodt: Sublattice identifica-
23.50 K. Yokoyama, T. Ochi, A. Yoshimoto, Y. Sugawara, tion in scanning force microscopy on alkali halide
S. Morita: Atomic resolution imaging on Si(100)2 × surfaces, Phys. Rev. B 92, 146103–1–146103–4 (2004)
1 and Si(100)2 × 1-H surfaces using a non-contact 23.63 C. Barth, M. Reichling: Resolving ions and vacan-
atomic force microscope, Jpn. J. Appl. Phys. 39, cies at step edges on insulating surfaces, Surf. Sci.
L113–L115 (2000) 470, L99–L103 (2000)
23.51 Y. Sugawara, T. Minobe, S. Orisaka, T. Uchihashi, 23.64 R. Bennewitz, M. Reichling, E. Matthias: Force
T. Tsukamoto, S. Morita:
√ √ Non-contact AFM images microscopy of cleaved and electron-irradiated
measured on Si(111) 3× 3-Ag and Ag(111) surfaces, CaF2 (111) surfaces in ultra-high vacuum, Surf. Sci.
Surf. Interface Anal. 27, 456–461 (1999) 387, 69–77 (1997)
660 Part C Scanning-Probe Microscopy
23.65 M. Reichling, C. Barth: Scanning force imaging of TiO2 (001)-(1 × 4) thin films studied by NC-AFM and
atomic size defects on the CaF2 (111) surface, Phys. STM, J. Phys. Chem. B 106, 8211–8222 (2002)
Rev. Lett. 83, 768–771 (1999) 23.80 A. Sasahara, T.C. Droubay, S.A. Chambers, H. Uet-
23.66 M. Reichling, M. Huisinga, S. Gogoll, C. Barth: suka, H. Onishi: Topography of anatase TiO2 film
Degradation of the CaF2 (111) surface by air exposure, synthesized on LaAlO3 (001), Nanotechnology 16,
Surf. Sci. 439, 181–190 (1999) S18–S21 (2005)
23.67 A. Klust, T. Ohta, A.A. Bostwick, Q. Yu, F.S. Ohuchi, 23.81 C.L. Pang, S.A. Haycock, H. Raza, P.J. Møller,
M.A. Olmstead: Atomically resolved imaging of G. Thornton: Structures of the 4 × 1 and 1 × 2 recon-
a CaF bilayer on Si(111): Subsurface atoms and the structions of SnO2 (110), Phys. Rev. B 62, R7775–R7778
image contrast in scanning force microscopy, Phys. (2000)
Rev. B 69, 035405–1–035405–5 (2004) 23.82 H. Hosoi, K. Sueoka, K. Hayakawa, K. Mukasa:
23.68 C. Barth, A.S. Foster, M. Reichling, A.L. Shluger: Atomic resolved imaging of cleaved NiO(100) sur-
Contrast formation in atomic resolution scanning faces by NC-AFM, Appl. Surf. Sci. 157, 218–221 (2000)
force microscopy of CaF2 (111): Experiment and the- 23.83 W. Allers, S. Langkat, R. Wiesendanger: Dynamic
ory, J. Phys. Condens. Matter 13, 2061–2079 (2001) low-temperature scanning force microscopy on
Part C 23
23.69 A.S. Foster, C. Barth, A.L. Shulger, M. Reichling: nickel oxide (001), Appl. Phys. A 72, S27–S30 (2001)
Unambiguous interpretation of atomically resolved 23.84 T.√Kubo,
√ H. Nozoye: Surface Structure of SrTiO3(100)-
force microscopy images of an insulator, Phys. Rev. ( 5× 5)-R 26.6◦ , Phys. Rev. Lett. 86, 1801–1804
Lett. 86, 2373–2376 (2001) (2001)
23.70 A.S. Foster, A.L. Rohl, A.L. Shluger: Imaging prob- 23.85 K. Fukui, Y. Namai, Y. Iwasawa: Imaging of sur-
lems on insulators: What can be learnt from face oxygen atoms and their defect structures on
NC-AFM modeling on CaF2 ?, Appl. Phys. A 72, S31– CeO2 (111) by noncontact atomic force microscopy,
S34 (2001) Appl. Surf. Sci. 188, 252–256 (2002)
23.71 F.J. Giessibl, M. Reichling: Investigating atomic de- 23.86 S. Suzuki, Y. Ohminami, T. Tsutsumi, M.M. Shoaib,
tails of the CaF2 (111) surface with a qPlus sensor, M. Ichikawa, K. Asakura: The first observation of an
Nanotechnology 16, S118–S124 (2005) atomic scale noncontact AFM image of MoO3 (010),
23.72 A.S. Foster, C. Barth, A.L. Shluger, R.M. Niemi- Chem. Lett. 32, 1098–1099 (2003)
nen, M. Reichling: Role of tip structure and surface 23.87 C. Barth, C.R. Henry: Atomic resolution imaging
relaxation in atomic resolution dynamic force mi- of the (001) surface of UHV cleaved MgO by dy-
croscopy: CaF2 (111) as a reference surface, Phys. Rev. namic scanning force microscopy, Phys. Rev. Lett.
B 66, 235417–1–235417–10 (2002) 91, 196102–1–196102–4 (2003)
23.73 C. Barth, M. Reichling: Imaging the atomic 23.88 A.S. Foster, A.Y. Gal, J.M. Airaksinen, O.H. Pakari-
arrangements on the high-temperature recon- nen, Y.J. Lee, J.D. Gale, A.L. Shluger, R.M. Niem-
structed α-Al2 O3 surface, Nature 414, 54–57 (2001) inen: Towards chemical identification in atomic-
23.74 M. Reichling, C. Barth: Atomically resolution imag- resolution noncontact AFM imaging with sili-
ing on fluorides. In: Noncontact Atomic Force con tips, Phys. Rev. B 68, 195420–1–195420–8
Microscopy, ed. by S. Morita, R. Wiesendan- (2003)
ger, E. Meyer (Springer, Berlin, Heidelberg 2002) 23.89 A.S. Foster, A.Y. Gal, J.D. Gale, Y.J. Lee, R.M. Niem-
pp. 109–123 inen, A.L. Shluger: Interaction of silicon dangling
23.75 K. Fukui, H. Ohnishi, Y. Iwasawa: Atom-resolved bonds with insulating surfaces, Phys. Rev. Lett. 92,
image of the TiO2 (110) surface by noncontact atomic 036101–1–036101–4 (2004)
force microscopy, Phys. Rev. Lett. 79, 4202–4205 23.90 T. Eguchi, Y. Hasegawa: High resolution atomic
(1997) force microscopic imaging of the Si(111)-(7×7) sur-
23.76 H. Raza, C.L. Pang, S.A. Haycock, G. Thornton: face: Contribution of short-range force to the
Non-contact atomic force microscopy imaging of images, Phys. Rev. Lett. 89, 266105–1–266105–4
TiO2 (100) surfaces, Appl. Surf. Sci. 140, 271–275 (2002)
(1999) 23.91 T. Arai, M. Tomitori: A Si nanopillar grown on a Si
23.77 C.L. Pang, H. Raza, S.A. Haycock, G. Thornton: tip by atomic force microscopy in ultrahigh vacuum
Imaging reconstructed TiO2 (100) surfaces with non- for a high-quality scanning probe, Appl. Phys. Lett.
contact atomic force microscopy, Appl. Surf. Sci. 86, 073110–1–073110–3 (2005)
157, 223–238 (2000) 23.92 K. Mukasa, H. Hasegawa, Y. Tazuke, K. Sueoka,
23.78 M. Ashino, T. Uchihashi, K. Yokoyama, Y. Sugawara, M. Sasaki, K. Hayakawa: Exchange interaction be-
S. Morita, M. Ishikawa: STM and atomic-resolution tween magnetic moments of ferromagnetic sample
noncontact AFM of an oxygen-deficient TiO2 (110) and tip: Possibility of atomic-resolution images
surface, Phys. Rev. B 61, 13955–13959 (2000) of exchange interactions using exchange force
23.79 R.E. Tanner, A. Sasahara, Y. Liang, E.I. Altmann, microscopy, Jpn. J. Appl. Phys. 33, 2692–2695
H. Onishi: Formic acid adsorption on anatase (1994)
Noncontact Atomic Force Microscopy and Related Topics References 661
23.93 H. Ness, F. Gautier: Theoretical study of the interac- 23.106 R.M. Overney, E. Meyer, J. Frommer, D. Brodbeck,
tion between a magnetic nanotip and a magnetic R. Lüthi, L. Howald, H.-J. Güntherodt, M. Fujihira,
surface, Phys. Rev. B 52, 7352–7362 (1995) H. Takano, Y. Gotoh: Friction measurements on
23.94 K. Nakamura, H. Hasegawa, T. Oguchi, K. Sueoka, phase-separated thin films with amodified atomic
K. Hayakawa, K. Mukasa: First-principles calcula- force microscope, Nature 359, 133–135 (1992)
tion of the exchange interaction and the exchange 23.107 D. Frisbie, L.F. Rozsnyai, A. Noy, M.S. Wrighton,
force between magnetic Fe films, Phys. Rev. B 56, C.M. Lieber: Functional group imaging by chemical
3218–3221 (1997) force microscopy, Science 265, 2071–2074 (1994)
23.95 A.S. Foster, A.L. Shluger: Spin-contrast in non- 23.108 E. Meyer, L. Howald, R.M. Overney, H. Heinzel-
contact SFM on oxide surfaces: Theoretical mod- mann, J. Frommer, H.-J. Güntherodt, T. Wagner,
eling of NiO(001) surface, Surf. Sci. 490, 211–219 H. Schier, S. Roth: Molecular-resolution images of
(2001) Langmuir–Blodgett films using atomic force mi-
23.96 T. Oguchi, H. Momida: Electronic structure and croscopy, Nature 349, 398–400 (1992)
magnetism of antiferromagnetic oxide surface – 23.109 K. Fukui, H. Onishi, Y. Iwasawa: Imaging of individ-
First-principles calculations, J. Surf. Sci. Soc. Jpn. ual formate ions adsorbed on TiO2 (110) surface by
Part C 23
26, 138–143 (2005) non-contact atomic force microscopy, Chem. Phys.
23.97 H. Hosoi, M. Kimura, K. Sueoka, K. Hayakawa, Lett. 280, 296–301 (1997)
K. Mukasa: Non-contact atomic force microscopy of 23.110 K. Kobayashi, H. Yamada, T. Horiuchi, K. Mat-
an antiferromagnetic NiO(100) surface using a fer- sushige: Investigations of C60 molecules deposited
romagnetic tip, Appl. Phys. A 72, S23–S26 (2001) on Si(111) by noncontact atomic force microscopy,
23.98 H. Hölscher, S.M. Langkat, A. Schwarz, R. Wiesen- Appl. Surf. Sci. 140, 281–286 (1999)
danger: Measurement of three-dimensional force 23.111 T. Uchihashi, M. Tanigawa, M. Ashino, Y. Sugawara,
fields with atomic resolution using dynamic force K. Yokoyama, S. Morita, M. Ishikawa: Identifica-
spectroscopy, Appl. Phys. Lett. 81, 4428–4430 tion of B-form DNA in an ultrahigh vacuum by
(2002) noncontact-mode atomic force microscopy, Lang-
23.99 S.M. Langkat, H. Hölscher, A. Schwarz, R. Wiesen- muir 16, 1349–1353 (2000)
danger: Determination of site specific interaction 23.112 Y. Maeda, T. Matsumoto, T. Kawai: Observation of
forces between an iron coated tip and the NiO(001) single- and double-strand DNA using non-contact
surface by force field spectroscopy, Surf. Sci. 527, atomic force microscopy, Appl. Surf. Sci. 140, 400–
12–20 (2003) 405 (1999)
23.100 R. Hoffmann, M.A. Lantz, H.J. Hug, P.J.A. van 23.113 T. Uchihashi, T. Ishida, M. Komiyama, M. Ashino,
Schendel, P. Kappenberger, S. Martin, A. Baratoff, Y. Sugawara, W. Mizutani, K. Yokoyama, S. Morita,
H.-J. Güntherodt: Atomic resolution imaging H. Tokumoto, M. Ishikawa: High-resolution imag-
and frequency versus distance measurement on ing of organic monolayers using noncontact AFM,
NiO(001) using low-temperature scanning force Appl. Surf. Sci. 157, 244–250 (2000)
microscopy, Phys. Rev. B 67, 085402–1–085402–6 23.114 T. Fukuma, K. Kobayashi, T. Horiuchi, H. Ya-
(2003) mada, K. Matsushige: Alkanethiol self-assembled
23.101 H. Hosoi, K. Sueoka, K. Hayakawa, K. Mukasa: monolayers on Au(111) surfaces investigated by
Atomically resolved imaging of a NiO(001) surface. non-contact AFM, Appl. Phys. A 72, S109–S112 (2001)
In: Noncontact Atomic Force Microscopy, ed. by 23.115 B. Gotsmann, C. Schmidt, C. Seidel, H. Fuchs:
S. Morita, R. Wiesendanger, E. Meyer (Springer, Molecular resolution of an organic monolayer by
Berlin, Heidelberg 2002) pp. 125–134 dynamic AFM, Eur. Phys. J. B 4, 267–268 (1998)
23.102 K. Sueoka, A. Subagyo, H. Hosoi, K. Mukasa: Mag- 23.116 C. Loppacher, M. Bammerlin, M. Guggisberg,
netic imaging with scanning force microscopy, E. Meyer, H.-J. Güntherodt, R. Lüthi, R. Schlit-
Nanotechnology 15, S691–S698 (2004) tler, J.K. Gimzewski: Forces with submolecular
23.103 H. Hosoi, K. Sueoka, K. Mukasa: Investigations on resolution between the probing tip and Cu-TBPP
the topographic asymmetry of non-contact atomic molecules on Cu(100) observed with a com-
force microscopy images of NiO(001) surface ob- bined AFM/STM, Appl. Phys. A 72, S105–S108
served with a ferromagnetic tip, Nanotechnology (2001)
15, 505–509 (2004) 23.117 L.M. Eng, M. Bammerlin, C. Loppacher, M. Guggis-
23.104 H. Momida, T. Oguchi: First-principles studies of berg, R. Bennewitz, R. Lüthi, E. Meyer, H.-J. Gün-
antiferromagnetic MnO and NiO surfaces, J. Phys. therodt: Surface morphology, chemical contrast,
Soc. Jpn. 72, 588–593 (2003) and ferroelectric domains in TGS bulk single crystals
23.105 K. Kobayashi, H. Yamada, T. Horiuchi, K. Mat- differentiated with UHV non-contact force mi-
sushige: Structures and electrical properties of croscopy, Appl. Surf. Sci. 140, 253–258 (1999)
fullerene thin films on Si(111)-7 × 7 surface investi- 23.118 S. Kitamura, K. Suzuki, M. Iwatsuki: High resolu-
gated by noncontact atomic force microscopy, Jpn. tion imaging of contact potential difference using
J. Appl. Phys. 39, 3821–3829 (2000) a novel ultrahigh vacuum non-contact atomic
662 Part C Scanning-Probe Microscopy
force microscope technique, Appl. Surf. Sci. 140, contact atomic force microscopy, Appl. Surf. Sci.
265–270 (1999) 188, 265–271 (2002)
23.119 H. Yamada, T. Fukuma, K. Umeda, K. Kobayashi, 23.126 H. Onishi, A. Sasahara, H. Uetsuka, T. Ishibashi:
K. Matsushige: Local structures and electrical prop- Molecule-dependent topography determined by
erties of organic molecular films investigated by noncontact atomic force microscopy: Carboxy-
non-contact atomic force microscopy, Appl. Surf. lates on TiO2 (110), Appl. Surf. Sci. 188, 257–264
Sci. 188, 391–398 (2000) (2002)
23.120 K. Fukui, Y. Iwasawa: Fluctuation of acetate ions 23.127 H. Onishi: Carboxylates adsorbed on TiO2 (110). In:
in the (2 × 1)-acetate overlayer on TiO2 (110)-(1 × 1) Chemistry of Nano-molecular Systems, ed. by
observed by noncontact atomic force microscopy, T. Nakamura (Springer, Berlin, Heidelberg 2002)
Surf. Sci. 464, L719–L726 (2000) pp. 75–89
23.121 A. Sasahara, H. Uetsuka, H. Onishi: Singlemolecule 23.128 S. Thevuthasan, G.S. Herman, Y.J. Kim, S.A. Cham-
analysis by non-contact atomic force microscopy, bers, C.H.F. Peden, Z. Wang, R.X. Ynzunza,
J. Phys. Chem. B 105, 1–4 (2001) E.D. Tober, J. Morais, C.S. Fadley: The structure
23.122 A. Sasahara, H. Uetsuka, H. Onishi: NC-AFM topog- of formate on TiO2 (110) by scanned-energy and
Part C 23
raphy of HCOO and CH3 COO molecules co-adsorbed scanned-angle photoelectron diffraction, Surf. Sci.
on TiO2 (110), Appl. Phys. A 72, S101–S103 (2001) 401, 261–268 (1998)
23.123 A. Sasahara, H. Uetsuka, H. Onishi: Image topogra- 23.129 H. Uetsuka, A. Sasahara, A. Yamakata, H. Onishi:
phy of alkyl-substituted carboxylates observed by Microscopic identification of a bimolecular reaction
noncontact atomic force microscopy, Surf. Sci. 481, intermediate, J. Phys. Chem. B 106, 11549–11552
L437–L442 (2001) (2002)
23.124 A. Sasahara, H. Uetsuka, H. Onishi: Noncontact 23.130 D.R. Lide: Handbook of Chemistry and Physics, 81st
atomic force microscope topography dependent on edn. (CRC, Boca Raton 2000)
permanent dipole of individual molecules, Phys. 23.131 K. Kobayashi, H. Yamada, K. Matsushige: Dy-
Rev. B 64, 121406 (2001) namic force microscopy using FM detection in
23.125 A. Sasahara, H. Uetsuka, T. Ishibashi, H. Onishi: various environments, Appl. Surf. Sci. 188, 430–434
A needle-like organic molecule imaged by non- (2002)
663
Low-Tempera
24. Low-Temperature Scanning Probe Microscopy
This chapter is dedicated to scanning probe mi- achieved since the invention of the method
croscopy (SPM) operated at cryogenic temperatures, about 30 years ago. We first focus on the scan-
where the more fundamental aspects of phenom- ning tunneling microscope, giving examples of
ena important in the field of nanotechnology atomic manipulation and the analysis of electronic
can be investigated with high sensitivity under properties in different material arrangements. Af-
well-defined conditions. In general, scanning terwards, we describe results obtained by scanning
probe techniques allow the measurement of phys- force microscopy, showing atomic-scale imaging
ical properties down to the nanometer scale. on insulators, as well as force spectroscopy analy-
Some techniques, such as scanning tunneling mi- sis. Finally, the magnetic force microscope, which
croscopy and scanning force microscopy, even go images domain patterns in ferromagnets and vor-
Part C 24
down to the atomic scale. Various properties are tex patterns in superconductors, is discussed.
accessible. Most importantly, one can image the Although this list is far from complete, we feel that
arrangement of atoms on conducting surfaces by it gives an adequate impression of the fascinating
scanning tunneling microscopy and on insulating possibilities of low-temperature scanning probe
substrates by scanning force microscopy. However, instruments.
the arrangement of electrons (scanning tunnel- In this chapter low temperatures are defined as
ing spectroscopy), the force interaction between lower than about 100 K and are normally achieved
different atoms (scanning force spectroscopy), by cooling with liquid nitrogen or liquid helium.
magnetic domains (magnetic force microscopy), Applications in which SPMs are operated close to
the local capacitance (scanning capacitance mi- 0 ◦ C are not covered in this chapter.
croscopy), the local temperature (scanning thermo
microscopy), and local light-induced excitations 24.1 Microscope Operation
(scanning near-field microscopy) can also be at Low Temperatures ............................ 664
measured with high spatial resolution. In addition, 24.1.1 Drift ......................................... 664
some techniques even allow the manipulation of 24.1.2 Noise ........................................ 665
atomic configurations. 24.1.3 Stability .................................... 665
Probably the most important advantage of 24.1.4 Piezo Relaxation and Hysteresis ... 665
the low-temperature operation of scanning probe 24.2 Instrumentation ................................... 666
techniques is that they lead to a significantly better 24.2.1 A Simple Design
signal-to-noise ratio than measuring at room tem- for a Variable-Temperature STM ... 666
perature. This is why many researchers work be- 24.2.2 A Low-Temperature SFM
low 100 K. However, there are also physical reasons Based on a Bath Cryostat ............ 668
to use low-temperature equipment. For example, 24.3 Scanning Tunneling Microscopy
the manipulation of atoms or scanning tunnel- and Spectroscopy.................................. 669
ing spectroscopy with high energy resolution can 24.3.1 Atomic Manipulation .................. 669
only be realized at low temperatures. Moreover, 24.3.2 Imaging Atomic Motion............... 671
some physical effects such as superconductivity 24.3.3 Detecting Light
from Single Atoms and Molecules . 672
or the Kondo effect are restricted to low temper-
24.3.4 High-Resolution Spectroscopy ..... 673
atures. Here, we describe the design criteria of
24.3.5 Imaging Electronic Wavefunctions 679
low-temperature scanning probe equipment and 24.3.6 Imaging Spin Polarization:
summarize some of the most spectacular results Nanomagnetism ........................ 686
664 Part C Scanning-Probe Microscopy
24.4 Scanning Force Microscopy 24.4.4 Electrostatic Force Microscopy ...... 695
and Spectroscopy.................................. 688 24.4.5 Magnetic Force Microscopy .......... 696
24.4.1 Atomic-Scale Imaging................. 689 24.4.6 Magnetic Exchange Force
24.4.2 Force Spectroscopy ..................... 692 Microscopy ................................ 698
24.4.3 Atomic Manipulation .................. 695 References .................................................. 700
Nearly three decades ago, the first design of an ex- Another advantage is their ability to work within
perimental setup was presented where a sharp tip was a wide temperature range. A microscope operation at
systematically scanned over a sample surface in order higher temperatures is chosen to study surface diffu-
to obtain local information on the tip–sample interac- sion, surface reactivity, surface reconstructions that only
tion down to the atomic scale. This original instrument manifest at elevated temperatures, high-temperature
used the tunneling current between a conducting tip and phase transitions, or to simulate conditions as they
a conducting sample as a feedback signal and was thus occur, e.g., in engines, catalytic converters or reac-
named the scanning tunneling microscope [24.1]. Soon tors. Ultimately, the upper limit for the operation of
after this historic breakthrough, it became widely rec- an SPM is determined by the stability of the sample,
ognized that virtually any type of tip–sample interaction but thermal drift, which limits the ability to move the
could be used to obtain local information on the sample tip in a controlled manner over the sample, as well
by applying the same general principle, provided that as the depolarization temperature of the piezoelectric
Part C 24.1
the selected interaction was reasonably short-ranged. positioning elements might further restrict successful
Thus, a whole variety of new methods has been in- measurements.
troduced, which are denoted collectively as scanning On the other hand, low-temperature (LT) applica-
probe methods. An overview is given, e.g., by Wiesen- tion of SPMs is much more widespread than operation
danger [24.2]. at high temperatures. Essentially five reasons make
The various methods, especially the above men- researchers adapt their experimental setups to low-
tioned scanning tunneling microscopy (STM) and temperature compatibility. These are: (1) the reduced
scanning force microscopy (SFM) – which is often fur- thermal drift, (2) lower noise levels, (3) enhanced sta-
ther classified into subdisciplines such as topography- bility of tip and sample, (4) the reduction in piezo
reflecting atomic force microscopy (AFM), magnetic hysteresis/creep, and (5) probably the most obvious, the
force microscopy (MFM) or electrostatic force mi- fact that many physical effects are restricted to low tem-
croscopy (EFM) – have been established as standard perature. Reasons 1–4 only apply unconditionally if the
methods for surface characterization on the nanometer whole microscope body is kept at low temperature (typ-
scale. The reason is that they feature extremely high ically in or attached to a bath cryostat, see Sect. 24.2).
resolution (often down to the atomic scale for STM Setups in which only the sample is cooled may show
and AFM), despite a principally simple, compact, and considerably less favorable operating characteristics.
comparatively inexpensive design. As a result of 1–4, ultrahigh resolution and long-term
A side-effect of the simple working principle and stability can be achieved on a level that significantly ex-
the compact design of many scanning probe micro- ceeds what can be accomplished at room temperature
scopes (SPMs) is that they can be adapted to different even under the most favorable circumstances. Typical
environments such as air, all kinds of gaseous at- examples of effect 5 are superconductivity [24.3] and
mospheres, liquids or vacuum with reasonable effort. the Kondo effect [24.4].
ature, typical values for solids are on the order of Another, even more striking, example is the spec-
(1–50) × 10−6 K−1 . If the temperature could be kept troscopic resolution in scanning tunneling spectroscopy
precisely constant, any thermal drift would vanish, (STS). This depends linearly on the temperature [24.2]
regardless of the absolute temperature of the sys- and is consequently reduced even more at LT than the
tem. The close coupling of the microscope to a large thermal noise in AFM. This provides the opportunity
temperature bath that maintains a constant tempera- to study structures or physical effects not accessible at
ture ensures a significant reduction in thermal drift room temperature such as spin and Landau levels in
and allows for distortion-free long-term measurements. semiconductors [24.9].
Microscopes that are efficiently attached to suffi- Finally, it might be worth mentioning that the en-
ciently large bath cryostats, therefore, show a one- hanced stiffness of most materials at low temperatures
to two-order-of-magnitude increase in thermal stability (increased Young’s modulus) leads to a reduced cou-
compared with nonstabilized setups operated at room pling to external noise. Even though this effect is
temperature. considered small [24.6], it should not be ignored.
A second effect also helps suppress thermally in-
duced drift of the probing tip relative to a specific 24.1.3 Stability
location on the sample surface. The thermal expan-
sion coefficients at liquid-helium temperatures are two There are two major stability issues that considerably
or more orders of magnitude smaller than at room improve at low temperature. First, low temperatures
temperature. Consequently, the thermal drift during close to the temperature of liquid helium inhibit most of
low-temperature operation decreases accordingly. the thermally activated diffusion processes. As a con-
Part C 24.1
For some specific scanning probe methods, there sequence, the sample surfaces show a significantly
may be additional ways in which a change in increased long-term stability, since defect motion or
temperature can affect the quality of the data. In adatom diffusion is massively suppressed. Most strik-
frequency-modulation SFM (FM-SFM), for example, ingly, even single xenon atoms deposited on suitable
the measurement principle relies on the accurate de- substrates can be successfully imaged [24.10, 11] or
termination of the eigenfrequency of the cantilever, even manipulated [24.12]. In the same way, low tem-
which is determined by its spring constant and its effec- peratures also stabilize the atomic configuration at the
tive mass. However, the spring constant changes with tip end by preventing sudden jumps of the most loosely
temperature due to both thermal expansion (i. e., the bound, foremost tip atom(s). Secondly, the large cryo-
resulting change in the cantilever dimensions) and the stat that usually surrounds the microscope acts as an
variation of the Young’s modulus with temperature. As- effective cryo-pump. Thus samples can be kept clean for
suming drift rates of about 2 mK/min, as is typical for several weeks, which is a multiple of the corresponding
room-temperature measurements, this effect might have time at room temperature (about 3–4 h).
a significant influence on the obtained data.
24.1.4 Piezo Relaxation and Hysteresis
24.1.2 Noise
The last important benefit from low-temperature oper-
The theoretically achievable resolution in SPM often in- ation of SPMs is that artifacts from the response of the
creases with decreasing temperature due to a decrease piezoelectric scanners are substantially reduced. After
in thermally induced noise. An example is the thermal applying a voltage ramp to one electrode of a piezo-
noise in SFM, which is proportional to the square root electric scanner, its immediate initial deflection, l0 , is
of the temperature [24.5, 6]. Lowering the temperature followed by a much slower relaxation, Δl, with a log-
from T = 300 K to T = 10 K thus results in a reduction arithmic time dependence. This effect, known as piezo
of the thermal frequency noise by more than a factor of relaxation or creep, diminishes substantially at low tem-
five. Graphite, e.g., has been imaged with atomic reso- peratures, typically by a factor of ten or more. As
lution only at low temperatures due to its extremely low a consequence, piezo nonlinearities and piezo hysteresis
corrugation, which was below the room-temperature decrease accordingly. Additional information is given
noise level [24.7, 8]. by Hug et al. [24.13].
666 Part C Scanning-Probe Microscopy
24.2 Instrumentation
The two main design criteria for all vacuum-based scan- will, somewhat arbitrarily, give two examples at differ-
ning probe microscope systems are: (1) to provide an ent levels of complexity that might serve as illustrative
efficient decoupling of the microscope from the vac- model designs.
uum system and other sources of external vibrations,
and (2) to avoid most internal noise sources through the 24.2.1 A Simple Design
high mechanical rigidity of the microscope body itself. for a Variable-Temperature STM
In vacuum systems designed for low-temperature ap-
plications, a significant degree of complexity is added, A simple design for a variable-temperature STM system
since, on the one hand, close thermal contact of the is presented in Fig. 24.1; similar systems are also of-
SPM and cryogen is necessary to ensure the (approxi- fered by Omicron (Germany) or Jeol (Japan). It should
mately) drift-free conditions described above, while, on give an impression of what the minimum requirements
the other hand, good vibration isolation (both from the are, if samples are to be investigated successfully at
outside world, as well as from the boiling or flowing low temperatures. It features a single ultrahigh-vacuum
cryogen) has to be maintained. (UHV) chamber that houses the microscope in its cen-
Plenty of microscope designs have been presented ter. The general idea to keep the setup simple is that
in the last 10–15 years, predominantly in the field of only the sample is cooled, by means of a flow cryostat
STM. Due to the variety of the different approaches, we that ends in the small liquid-nitrogen (LN) reservoir.
Part C 24.2
Sample holder
on holders dock
Tip transfer
holders on
storage elevator
Sample holder Copper braid
on storage
elevator LN
reservoir
Thermo-
Scan couple
head
Fig. 24.1 One-chamber UHV system with variable-temperature STM based on a flow cryostat design. (
c RHK Technology, USA)
Low-Temperature Scanning Probe Microscopy 24.2 Instrumentation 667
This reservoir is connected to the sample holder with ple holder (see Fig. 24.2). The special design of the
copper braids. The role of the copper braids is to at- scan head (see [24.14] for details) allows not only
tach the LN reservoir thermally to the sample located on flexible positioning of the tip on any desired loca-
the sample holder in an effective manner, while vibra- tion on the sample surface but also compensates to
tions due to the flow of the cryogen should be blocked a certain degree for the thermal drift that inevitably
as much as possible. In this way, a sample temperature occurs in such a design due to temperature gradi-
of about 100 K is reached. Alternatively, with liquid- ents.
helium operation, a base temperature of below 30 K In fact, thermal drift is often much more prominent
can be achieved, while a heater that is integrated into in LT-SPM designs, where only the sample is cooled,
the sample stage enables high-temperature operation up than in room-temperature designs. Therefore, to bene-
to 1000 K. fit fully from the high-stability conditions described in
A typical experiment would run as follows. First, the introduction, it is mandatory to keep the whole mi-
the sample is brought into the system by placing it in croscope at the exact same temperature. This is mostly
the so-called load-lock. This small part of the cham- realized by using bath cryostats, which add a certain
ber can be separated from the rest of the system by degree of complexity.
a valve, so that the main part of the system can re-
main under vacuum at all times (i. e., even if the
Electrical and fiber Motor chain
load-lock is opened to introduce the sample). After vac- feedthroughs drive
uum is reestablished, the sample is transferred to the
main chamber using the transfer arm. A linear-motion Main
Part C 24.2
chamber
feedthrough enables the storage of sample holders or, Vertical
alternatively, specialized holders that carry replacement Getter chain transfer
pumps
tips for the STM. Extending the transfer arm further, Analysis Preparation
the sample can be placed on the sample stage and chamber chamber
subsequently cooled down to the desired temperature.
The scan head, which carries the STM tip, is then
lowered with the scan-head manipulator onto the sam-
Getter
pump
Turbo
Scan head pump Turbo
manipulator pump
Table
Scan Copper
head braids
Sample Tip Pneumatic
holder leg
Dewar
Heating/
cooling Copper
stage cone
Cryostat
SFM
Sand
24.2.2 A Low-Temperature SFM In this design, tip and sample are exchanged at room
Based on a Bath Cryostat temperature in the main chamber. After the transfer
into the cryostat, the SFM can be cooled by either li-
As an example of an LT-SPM setup based on a bath quid nitrogen or liquid helium, reaching temperatures
cryostat, let us take a closer look at the LT-SFM system down to 10 K. An all-fiber interferometer as the de-
sketched in Fig. 24.3, which has been used to acquire tection mechanism for the cantilever deflection ensures
the images on graphite, xenon, NiO, and InAs presented high resolution, while simultaneously allowing the con-
in Sect. 24.4. The force microscope is built into a UHV struction of a comparatively small, rigid, and symmetric
system that comprises three vacuum chambers: one for microscope.
cantilever and sample preparation, which also serves Figure 24.4 highlights the layout of the SFM body
as a transfer chamber, one for analysis purposes, and itself. Along with the careful choice of materials, the
a main chamber that houses the microscope. A specially symmetric design eliminates most of the problems with
designed vertical transfer mechanism based on a dou- drift inside the microscope encountered when cooling
ble chain allows the lowering of the microscope into or warming it up. The microscope body has an overall
a UHV-compatible bath cryostat attached underneath cylindrical shape with a height of 13 cm and a diameter
the main chamber. To damp the system, it is mounted of 6 cm and exact mirror symmetry along the cantilever
on a table carried by pneumatic damping legs, which, in axis. The main body is made of a single block of macor,
turn, stand on a separate foundation to decouple it from a machinable glass ceramic, which ensures a rigid and
building vibrations. The cryostat and dewar are sepa- stable design. For most of the metallic parts titanium
rated from the rest of the UHV system by a bellow. In was used, which has a temperature coefficient similar
Part C 24.2
addition, the dewar is surrounded by sand for acoustic to macor. The controlled but stable accomplishment of
isolation. movements, such as coarse approach and lateral posi-
a) Fiber b)
Macor body
Fiber approach
Piezo tube
Cantilever stage
Sample
Sample approach
Sapphire prism
Fig. 24.4a,b The scanning force microscope incorporated into the system presented in Fig. 24.3. (a) Section along plane
of symmetry. (b) Photo from the front
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 669
tioning in other microscope designs, is a difficult task at approach and the sample approach labels in Fig. 24.4);
low temperatures. The present design uses a special type it is described in detail in [24.15]. More information
of piezo motor that moves a sapphire prism (see the fiber regarding this design is given in [24.16].
Part C 24.3
move atoms on a surface [24.12], cut molecules into atom to a desired position, and finally to reduce the
pieces [24.17], reform bonds [24.18], charge individual current again in order to decouple the atom and tip.
atoms [24.19], and, consequently, establish new struc- The first demonstration of this technique was performed
tures on the nanometer scale. Also, the detection of by Eigler and Schweizer [24.12], who used Xe atoms
light resulting from tunneling into a particular mol- on a Ni(110) surface to write the three letters “IBM”
ecule [24.20,21], the visualization of thermally induced (their employer) on the atomic scale (Fig. 24.5a). Nowa-
atomic movements [24.22], and the detection of hys- days, many laboratories are able to move different
teresis curves of individual atoms [24.23] require LT kinds of atoms and molecules on different surfaces with
instrumentation. high precision. An example featuring CO molecules
Second, the spectroscopic resolution in STS de- on Cu(110) is shown in Fig. 24.5b–g. Even more com-
pends linearly on temperature and is, therefore, consid- plex structures than the “2000”, such as cascades of CO
erably reduced at LT. This provides the opportunity to molecules that by mutual repulsive interaction mimic
study physical effects inaccessible at room temperature. different kinds of logic gates, have been assembled and
Examples are the resolution of spin and Landau levels in their functionality tested [24.36]. Although these de-
semiconductors [24.9], or the investigation of lifetime- vices are slow and restricted to low temperature, they
broadening effects on the nanometer scale [24.24]. Also nicely demonstrate the high degree of control achieved
the imaging of distinct electronic wavefunctions in real on the atomic scale.
space requires LT-STM [24.25]. More recently, vibra- The basic modes of controlled motion of atoms and
tional levels, spin-flip excitations, and phonons have molecules by the tip are pushing, pulling, and slid-
been detected with high spatial resolution at LT us- ing. The selection of the particular mode depends on
ing the additional inelastic tunneling channel [24.26– the tunneling current, i. e., the distance between tip
28]. and molecule, as well as on the particular molecule–
Third, many physical effects, in particular, effects substrate combination [24.37]. It has been shown
guided by electronic correlations, are restricted to experimentally that the potential landscape for the ad-
low temperature. Typical examples are superconductiv- sorbate movement is modified by the presence of the
ity [24.3], the Kondo effect [24.4], and many of the tip [24.38, 39] and that excitations induced by the
electron phases found in semiconductors [24.29]. Here, tunneling current can trigger atomic or molecular mo-
LT-STM provides the possibility to study electronic ef- tion [24.40,41]. Other sources of motion are the electric
fects on a local scale, and intensive work has been done field between tip and molecule or electromigration
in this field, the most elaborate with respect to high- caused by the high current density [24.35]. The re-
temperature superconductivity [24.30–32]. quired lateral tip force for atomic motion has been
670 Part C Scanning-Probe Microscopy
a) h)
i)
j)
b) c) d)
k)
Part C 24.3
l)
e) f) g)
m)
Fig. 24.5 (a) STM image of single Xe atoms positioned on a Ni(110) surface in order to realize the letters “IBM”
on the atomic scale ( c D. Eigler, Almaden); (b–f) STM images recorded after different positioning processes of CO
molecules on a Cu(110) surface; (g) final artwork greeting the new millennium on the atomic scale ((b–g) c G. Meyer,
Zürich). (h–m) Synthesis of biphenyl from two iodobenzene molecules on Cu(111): First, iodine is abstracted from both
molecules (i,j); then the iodine between the two phenyl groups is removed from the step (k), and finally one of the phenyls
is slid along the Cu step (l) until it reacts with the other phenyl (m); the line drawings symbolize the actual status of the
molecules ((h–m) c S. W. Hla and K. H. Rieder, Berlin)
measured for typical adsorbate–substrate combinations pulse is applied at distances of 100 nm away from
to be ≈ 0.1 nN [24.42]. Other types of manipulation on the molecule [24.49]. Also, association of individual
the atomic scale are feasible. Some of them require a se- molecules [24.18, 50–52] can require voltage pulses in
lective inelastic tunneling into vibrational or rotational order to overcome local energy barriers. The process of
modes of the molecules [24.43]. This leads to con- controlled bond formation can even be used for dop-
trolled desorption [24.44], diffusion [24.45], molecular ing of single C60 molecules by up to four potassium
rotation [24.46, 47], conformational change [24.48] atoms [24.53]. As an example of controlled manipu-
or even controlled pick-up of molecules by the lation, Fig. 24.5h–m shows the production of biphenyl
tip [24.18]. Dissociation can be achieved by volt- from two iodobenzene molecules [24.54]. The iodine
age pulses [24.17] inducing local heating, even if the is abstracted by voltage pulses (Fig. 24.5i,j), then the
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 671
iodine is moved to the terrace by the pulling mode molecules carry larger ones [24.61] or, very interest-
(Fig. 24.5k,l), and finally the two phenyl parts are slid ingly, the influence of quantum tunneling [24.62]. The
along the step edge until they are close enough to latter is deduced from the Arrhenius plot of hopping
react (Fig. 24.5m). The chemical identification of the rates of H and D on Cu(001), as shown in Fig. 24.6e.
product is not deduced straightforwardly and partly
requires detailed vibrational STM spectroscopy (see be-
low and [24.55]). a) b) c)
(001)
Finally, also the charge state of a single atom or –
(001)
molecule can be manipulated, tested, and read out.
A Au atom has been switched reversibly between two
charge states using an insulating thin film as the sub-
strate [24.19]. In addition, the carrier capture rate of
a single impurity level within the bandgap of a semi-
d) 211 193 178 166 T (K)
conductor has been quantified [24.56], and the point ln h 2 ln D
conductance of a single atom has been measured and (s–1) D (cm2 s–1)
– 34
turned out to be a reproducible quantity [24.57]. These h
0
promising results might trigger a novel electronic field
of manipulation of matter on the atomic scale, which – 36
–2
is tightly related to the currently very popular field of
molecular electronics. – 38
Part C 24.3
–4
1000
include the movement of surface vacancies [24.64] or
0 2
Ag4
bulk interstitials close to the surface [24.65], the Brow-
2.2 V
–4
nian motion of vacancy islands [24.66] as well as
0 500
laser-induced diffusion distinct from thermally excited
2 diffusion [24.67].
2.1 V
2.0 V
0 2
Ag3 0 24.3.3 Detecting Light
–2 700 800 900 1000 from Single Atoms and Molecules
0 Wavelength (nm)
e) Photon Photon
counts energy (eV) It had already been realized in 1988 that STM experi-
2
1.78 1.55 1.38 1.24 ments are accompanied by light emission [24.68]. The
0 2
Ag2 fact that molecular resolution in the light intensity was
–2 achieved at LT (Fig. 24.7a,b) [24.20] raised the hope
0 2000 of performing quasi-optical experiments on the molecu-
lar scale. Meanwhile, it is clear that the basic emission
2
1500 process observed on metals is the decay of a local plas-
0
Ag1 2 mon induced in the area around the tip by inelastic
–2 tunneling processes [24.69, 70]. Thus, the molecular
0 1000 resolution is basically a change in the plasmon environ-
1
NiAl 2
ment, largely given by the increased height of the tip
4 3 with respect to the surface above the molecule [24.71].
0.5 500
0 NiAl 4 However, the electron can, in principle, also decay via
Oxide
single-particle excitations. Indeed, signatures of single-
–4
0 0 particle levels have been observed for a Na monolayer
0 1 2 3 700 800 900 1000 on Cu(111) [24.72] as well as for Ag adatom chains on
Bias voltage (V) Wavelength (nm)
NiAl(110) [24.21]. As shown in Fig. 24.7c, the peaks
of differential photon yield dY/ dV as a function of ap-
The hopping rate of H levels off at about 65 K, while plied bias V are at identical voltages to the peaks in
the hopping rate of the heavier D atom goes down dI/ dV intensity. This is evidence that the density of
to nearly zero, as expected from thermally induced states of the Ag adsorbates is responsible for the ra-
hopping. Quantum tunneling has surprisingly also been diative decay. Photon emission spectra displaying much
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 673
more details could be detected by depositing the adsor- that it is simply proportional to the LDOS at the posi-
bates of interest on a thin insulating film [24.73, 74]. tion of the tip [24.78]. Therefore, as long as the decay
Figure 24.7d shows spectra of ZnEtiol deposited on length is spatially constant, one measures the LDOS
a 0.5 nm-thick Al2 O3 layer on NiAl(110). Importantly, at the surface (24.1). Note that the contributing states
the peaks within the light spectra do not shift with ap- are not only surface states, but also bulk states. How-
plied voltage, ruling out that they are due to a plasmon ever, surface states usually dominate if present. Chen
mode induced by the tip. As shown in Fig. 24.7e, the has shown that higher orbital tip states lead to the so-
photon spectra show distinct variations by changing the called derivation rule [24.79]: pz -type tip states de-
position within the molecule, demonstrating that atomi- tect d(LDOS)/ dz, dz 2 -states detect d2 (LDOS)/ dz 2 ,
cally resolved maps of the excitation probability can be and so on. As long as the decay into vacuum is ex-
measured by STM. ponential and spatially constant, this leads only to an
Meanwhile, external laser light has also been cou- additional, spatially constant factor in dI/ dV . Thus, it
pled to the tunneling contact between the STM tip and is still the LDOS that is measured (24.1). The require-
a molecule deposited on an insulating film. A magne- ment of a spatially constant decay is usually fulfilled on
sium porphine molecule positioned below the tip could larger length scales, but not on the atomic scale [24.79].
be charged reversibly either by increasing the voltage There, states located close to the atoms show a stronger
of the tip or by increasing the photon energy of the decay into vacuum than the less localized states in
laser. This indicates selective absorption of light en- the interstitial region. This effect can lead to STS
ergy by the molecule leading to population of a novel corrugations that are larger than the real LDOS corru-
charge level by tunneling electrons [24.75], a result gations [24.80].
Part C 24.3
that raises the hope that STM can probe photochem- The voltage dependence of dI/ dV is sensitive
istry on the atomic scale. STM-induced light has also to a changing decay length with V , which increases
been detected from semiconductors [24.76], including with V . This influence can be reduced at higher V
heterostructures [24.77]. This light is again caused by by displaying dI/ dV/(I/V ) [24.81]. Additionally,
single-particle relaxation of injected electrons, but with-
out contrast on the atomic scale.
Differential conductance (arb. units)
24.3.4 High-Resolution Spectroscopy 8.6 K
7K
One of the most important modes of LT-STM is 5K
STS, which detects the differential conductivity dI/ dV 3K
as a function of the applied voltage V and the
position (x, y). The dI/ dV signal is basically propor-
tional to the local density of states (LDOS) of the 1.6 K
sample, the sum over squared single-particle wavefunc-
tions Ψi [24.2]
dI
(V, x, y) ∝ L DOS(E, x, y)
dV
= |Ψi (E, x, y)|2 , (24.1)
ΔE 380 mK
suring the decay of standing electron waves scattered it exhibits a low effective mass m eff /m e = 0.023 and
from a step edge as a function of energy [24.85]. a high g-factor of 14 in the bulk conduction band. The
Figure 24.9c,d shows the resulting oscillating dI/ dV values in metals are m eff /m e ≈ 1 and g ≈ 2, resulting in
signal measured for two different energies. To deduce energy splittings of only 1.25 and 1.2 meV at B = 10 T.
the coherence length L Φ , which is inversely propor- This is obviously lower than the typical lifetime broad-
tional to the lifetime τΦ , one has to consider that the enings discussed in the previous section and also close
finite energy resolution ΔE in the experiment also to ΔE = 1.1 meV achievable at T = 4 K.
leads to a decay of the standing wave away from the Fortunately, the electron density in doped semicon-
step edge. The dotted fit line using L Φ = ∞ indicates ductors is much lower, and thus the lifetime increases
this effect and, more importantly, shows a discrepancy significantly. Figure 24.10a shows a set of spectroscopy
from the measured curve. Only including a finite coher- curves obtained on InAs(110) in different magnetic
ence length of 6.2 nm results in good agreement, which fields [24.9]. Above E F , oscillations with increasing in-
in turn determines L Φ and thus τΦ , as displayed in tensity and energy distance are observed. They show the
Fig. 24.9c. The found 1/E 2 dependence of τΦ points to separation expected from Landau quantization. In turn,
a dominating influence of electron–electron interactions they can be used to deduce m eff from the peak separa-
at higher energies in the surface band. tion (Fig. 24.10b). An increase of m eff with increasing E
has been found, as expected from theory. Also, at high
Landau and Spin Levels fields, spin quantization is observed (Fig. 24.10c). It is
Moreover, the increased energy resolution at LT allows larger than expected from the bare g-factor due to con-
the resolution of electronic states that are not resolvable tributions from exchange enhancement [24.87].
Part C 24.3
at room temperature (RT); for example, Landau and
spin quantization appearing in a magnetic field B have Atomic Energy Levels
been probed on InAs(110) [24.9,86]. The corresponding Another opportunity at LT is to study electronic states
quantization energies are given by E Landau = eB/m eff and resonances of single adatoms. A complicated
and E spin = gμB. Thus InAs is a good choice, since resonance is the Kondo resonance described below.
a) b) c)
dI/dV (arb. units) meff /me dI/dV (arb. units)
6T
EBCBM EF 0.06 LL1 Exp.
Fit
5T 0.05
LL2
4T
0.04
3T 0.03
2T
1T 2.5–6 T
0.02 Tsui (1971)
k·p theory
0.01
0 50 100 150 200 250 300
E – EF (meV)
– 40 0 40 80 120 0 10 20 30 40 50 60 70
Sample voltage (mV) Sample voltage (mV)
Fig. 24.10 (a) dI/ dV curves of n-InAs(110) at different magnetic fields, as indicated; E BCBM marks the bulk conduction
band minimum; oscillations above E BCBM are caused by Landau quantization; the double peaks at B = 6 T are caused
by spin quantization. (b) Effective-mass data deduced from the distance of adjacent Landau peaks ΔE according to
ΔE = h eB/m eff (open symbols); filled symbols are data from planar tunnel junctions (Tsui), the solid line is a mean-
square fit of the data and the dashed line is the expected effective mass of InAs according to k · p theory. (c) Magnification
of a dI/ dV curve at B = 6 T, exhibiting spin splitting; the Gaussian curves marked by arrows are the fitted spin levels
676 Part C Scanning-Probe Microscopy
b) c)
Vibrational Levels Top I (nA)
As discussed with respect to light emission in STM, in- 0.15 V pulse
elastic tunneling processes contribute to the tunneling 40
current. The coupling of electronic states to vibrational
levels is one source of inelastic tunneling [24.26]. It 30
Part C 24.3
graphite [24.97], and on metal surfaces [24.28]. One plementary novel approach to inelastic effects might be
finds distinct dependencies of excitation probability on the recently developed radiofrequency STM [24.101],
the position of the STM tip with respect to the investi- which could give access to low-energy excitations, such
gated structure. as GHz spin-wave modes in nanostructures, which are
First indications for the d2 I/ dV 2 (V )-based detec- not resolvable by d2 I/ dV 2 at LT.
tion of extended magnons [24.98] and plasmons [24.97]
have also been published. Kondo Resonance
The inelastic tunneling current has, moreover, been A rather intricate interaction effect is the Kondo ef-
used to study single spin-flip excitations in magnetic fect. It results from a second-order scattering process
field for atoms and atomic assemblies deposited on between itinerate states and a localized state [24.102].
a thin insulator. The excitation probability was high The two states exchange some degree of freedom back
enough to observe the spin flip even as a step in dI/ dV and forth, leading to a divergence of the scattering prob-
instead of as a peak in d2 I/ dV 2 . Figure 24.12a shows ability at the Fermi level of the itinerate state. Due to the
the dI/ dV curves recorded above a single Mn atom divergence, the effect strongly modifies sample proper-
on Al2 O3 /NiAl(110) at different B fields. The lin- ties. For example, it leads to an unexpected increase in
ear shift of the step with B field is obvious, and the resistance with decreasing temperature for metals con-
step voltage can be fitted by eV = gμB B with μB be- taining magnetic impurities [24.4]. Here, the exchanged
ing the Bohr magneton and a reasonable g-factor of degree of freedom is the spin. A spectroscopic signature
g ≈ 2 [24.27]. Figure 24.12b shows the dI/ dV spectra of the Kondo effect is a narrow peak in the DOS at the
obtained on a Mn dimer embedded in CuN/Cu(100). Fermi level, continuously disappearing above a char-
Part C 24.3
A step is already visible at B = 0 T, splitting into acteristic temperature (the Kondo temperature). STS
three steps at higher field. This result can be ex- provides the opportunity to study this effect on the local
plained straightforwardly, as sketched in the inset, by scale [24.103, 104].
a singlet–triplet transition of the combined two spins Figure 24.13a–d shows an example of Co clusters
(coupled by an exchange energy of J ≈ 6 meV). Investi- deposited on a carbon nanotube [24.105]. While only
gating longer chains revealed an even–odd asymmetry, a small dip at the Fermi level, probably caused by curva-
i. e., chains consisting of 2, 4, 6, . . . atoms exhibit ture influences on the π-orbitals, is observed without Co
a singlet–triplet transition, while chains of 1, 3, 5, . . . (Fig. 24.13b) [24.106], a strong peak is found around
atoms exhibit a transition from S = 5/2 to S = 3/2. a Co cluster deposited on top of the tube (Co cluster
This indicates antiferromagnetic coupling within the is marked in Fig. 24.13a). The peak is slightly shifted
chain [24.99]. Figure 24.12c shows spectra of a sin- with respect to V = 0 mV due to the so-called Fano
gle Fe atom embedded within CuN. The spectrum resonance [24.107], which results from interference of
reveals several steps already at B = 0 T, showing that the tunneling processes into the localized Co level and
different spin orientations Sz must exhibit different the itinerant nanotube levels. The resonance disappears
energies due to magnetic anisotropy. In order to deter- within several nanometers of the cluster, as shown in
mine the anisotropy, the step energies and intensities Fig. 24.13d.
were measured at different magnetic fields applied The Kondo effect has also been detected for
in three different directions. Amazingly, the results different magnetic atoms deposited on noble-metal sur-
could be fitted completely by a single model with faces [24.103, 104]. There, it disappears at about 1 nm
an out-of-plane anisotropy of D = −1.55 meV and an from the magnetic impurity, and the effect of the Fano
in-plane anisotropy of E = 0.31 meV. Therefore, one resonance is more pronounced, contributing to dips
has to assume five different spin states of the Fe in dI/ dV (V ) curves instead of peaks. Detailed in-
being mixtures of the five Sz states of a total Fe vestigations show that the d-level occupation of the
spin of |S| = 2. The excellent fit is shown for en- adsorbate [24.108] as well as the surface charge den-
ergy and intensity of a particular B-field direction in sity [24.109, 110] matter for the Kondo temperature.
Fig. 24.12d,e [24.100]. Exchange interaction between adsorbates tunable by
The different experiments of inelastic tunneling their mutual distance can be used to tune the Kondo
demonstrate that details of atomic excitations in a solid temperature [24.111] or even to destroy the Kondo
environment can be probed by LT-STM, even if they resonance completely [24.112]. Meanwhile, magnetic
are not of primary electronic origin. This might be molecules have also been shown to exhibit Kondo res-
a highly productive method in the near future. A com- onances. This increases the tunability of the Kondo
678 Part C Scanning-Probe Microscopy
8 0→4
0.2 7
7T
6
5T
0.15 3T 5 0→3
4 0→2
1T
0.1 0T 3
2
0.05 1
0→1
0
–8 –6 –4 –2 0 2 4 6 8 0 1 2 3 4 5 6 7 8
Voltage (mV) Magnetic field (T)
e) Fe atom in CuN/Cu(100)
Normalized intensity
1
0→2
0.8
0.6
0.4
0.2
0→3
0 0→1
0 1 2 3 4 5 6 7 8
Magnetic field (T)
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 679
Fig. 24.12 (a) dI/ dV curves taken above a single Mn atom deposited on Al2 O3 /NiAl(110) at different magnetic fields
as indicated; (b) dI/ dV curves taken above a Mn dimer deposited onto CuN/Cu(100) at different magnetic fields as
indicated; inset shows the three possible spin-flip transitions between singlet and triplet; (c) dI/ dV curves taken above
a single Fe atom deposited onto CuN/Cu(100) at different magnetic fields as indicated; transitions are marked by arrows;
(d,e) energy and intensity of the steps in dI/ dV measured with magnetic field along the direction of the N rows of the
CuN surface (symbols) in comparison with calculated results (lines) ( c A. Heinrich, C. F. Hirjibehedin, Almaden)
effect, e.g., by the selection of adequate ligands sur- reappeared in the opposite focus, but not away from
rounding the localized spins [24.113, 114], by distant the focus (Fig. 24.13e, top). This shows amazingly that
association of other molecules [24.115] or by confor- complex local effects such as the Kondo resonance can
mational changes within the molecule [24.116]. be wave-guided to remote points.
A fascinating experiment has been performed by
Manoharan et al. [24.117], who used manipulation to 24.3.5 Imaging Electronic Wavefunctions
form an elliptic cage for the surface states of Cu(111)
(Fig. 24.13e, bottom). This cage was constructed to Bloch Waves
have a quantized level at E F . Then, a cobalt atom Since STS measures the sum of squared wavefunc-
was placed in one focus of the elliptic cage, producing tions (24.1), it is an obvious task to measure the local
a Kondo resonance. Surprisingly, the same resonance appearance of the most simple wavefunctions in solids,
namely Bloch waves. The atomically periodic part of
the Bloch wave is always measured if atomic resolution
Part C 24.3
a) c)
is achieved (inset of Fig. 24.15a). However, the long-
range wavy part requires the presence of scatterers. The
Co
electron wave impinges on the scatterer and is reflected,
leading to self-interference. In other words, the phase of
the Bloch wave becomes fixed by the scatterer.
Such self-interference patterns were first found on
graphite(0001) [24.118] and later on noble-metal sur-
b) dI/dV (arb. units) d) dI/dV (arb. units)
faces, where adsorbates or step edges scatter the surface
SWNT states (Fig. 24.14a) [24.25]. Fourier transforms of the
Co real-space images reveal the k-space distribution of the
corresponding states [24.119], which may include addi-
tional contributions besides the surface state [24.120].
Using particular geometries such as so-called quantum
corrals, the Bloch waves can be confined (Fig. 24.14b).
–0.2 0 0.2 –0.2 0 0.2
Bias voltage (V) Bias voltage (V)
Depending on the geometry of the corral, the result state
looks rather complex, but it can usually be reproduced
e)
Fig. 24.13 (a) STM image of a Co cluster on a single-wall
carbon nanotube (SWNT). (b) dI/ dV curves taken directly
above the Co cluster (Co) and far away from the Co clus-
ter (SWNT); the arrow marks the Kondo peak. (c) STM
image of another Co cluster on a SWNT with symbols
marking the positions where the dI/ dV curves displayed
in (d) are taken. (d) dI/ dV curves taken at the positions
marked in (c) ((a–d) c C. Lieber, Cambridge). (e) Lower
part: STM image of a quantum corral of elliptic shape
made from Co atoms on Cu(111); one Co atom is placed
at one of the foci of the ellipse. Upper part: map of the
strength of the Kondo signal in the corral; note that there is
also a Kondo signal at the focus that is not covered by a Co
atom ((e) c D. Eigler, Almaden)
680 Part C Scanning-Probe Microscopy
Part C 24.3
Fig. 24.15 (a) dI/ dV image of InAs(110) at V = 50 mV, by recalling that the electron tries to move in a cy-
B = 0 T; circular wave patterns corresponding to standing clotron circle, which becomes a cycloid path along an
Bloch waves around each sulphur donor are visible; in- equipotential line within an inhomogeneous electro-
set shows a magnification revealing the atomically periodic static potential [24.138].
part of the Bloch wave. (b) Same as (a), but at B = 6 T; the The same effect has been found in two-dimensional
stripe structures are drift states. (c) dI/ dV image of a 2-D (2-D) electron systems (2-DES) of InAs at the same
electron system on InAs(110) induced by the deposition of large B-field (Fig. 24.15d) [24.139]. However the scat-
Fe, B = 0 T. (d) Same as (c) but at B = 6 T; note that the tering states at B = 0 T are much more complex in 2-D
contrast in (a) is increased by a factor of ten with respect (Fig. 24.15c) [24.140]. The reason is the tendency of
to (b–d) a 2-DES to exhibit closed scattering paths [24.141].
Consequently, the self-interference does not result from
More localized structures, where a Bloch wave scattering at individual scatterers, but from complicated
description is not appropriate, have been imaged, self-interference paths involving many scatterers. Nev-
too. Examples are the highest occupied molecular ertheless, the wavefunction pattern can be reproduced
orbital (HOMO) and lowest unoccupied molecular by including these effects within the calculations.
orbital (LUMO) of pentacene molecules deposited Reducing the dimensionality to one dimension (1-
on NaCl/Cu(100) (Fig. 24.14k,l) [24.51], the differ- D) leads again to complicated self-interference patterns
ent molecular states of C60 on Ag(110) (Fig. 24.3m–o) due to the interaction of the electrons with sev-
[24.130], the anisotropic states of Mn acceptors in eral impurities [24.142, 143]. For InAs, they can be
a semiconducting host [24.131, 132], and the hy- reproduced by single-particle calculations. However,
bridized states developing within short monoatomic experiments imaging self-interference patterns close to
Au chains, which develop particular states at the end the end of a C-nanotube are interpreted as indications
of the chains [24.133, 134]. Using pairs of remote of spin-charge separation, a genuine property of 1-D
Mn acceptors, even symmetric and antisymmetric pair electrons not feasible within the single-particle descrip-
wavefunctions have been imaged in real space [24.135]. tion [24.144].
The central requirements for a detailed imaging of
wavefunctions are LT for an appropriate energetic dis- Charge Density Waves, Jahn–Teller Distortion
tinction of an individual state, adequate decoupling of Another interaction modifying the LDOS is the
the state from the substrate in order to decrease lifetime- electron–phonon interaction. Phonons scatter electrons
induced broadening effects, and, partly, the selection between different Fermi points. If the wavevectors
of a system with an increased Bohr radius in order to connecting Fermi points exhibit a preferential orien-
682 Part C Scanning-Probe Microscopy
b
b a a
B
TCNQ TTF
0 1 2 3 4 5 6 7 8
A Distance (nm) B
e) f) g) h)
Part C 24.3
Fig. 24.16 (a) STM image of the ab-plane of the organic quasi-1-D conductor tetrathiafulvalene tetracyanoquinodimethane (TTF-
TCNQ), T = 300 K; while the TCNQ chains are conducting, the TTF chains are insulating. (b) Stick-and-ball model of the
ab-plane of TTF-TCNQ. (c) STM image taken at T = 61 K; the additional modulation due to the Peierls transition is visible in the
profile along line AB shown in (d); the brown triangles mark the atomic periodicity and the black triangles the expected CDW
periodicity ((a–d) c M. Kageshima, Kanagawa). (e–h) Low-voltage STM images of the two-dimensional CDW system 1 T-TaS2
at T = 242 K (e), 298 K (f), 349 K (g), and 357 K (h). A long-range, hexagonal modulation is visible besides the atomic spots; its
periodicity is highlighted by large white dots in (e); the additional modulation obviously weakens with increasing T , but is still
apparent in (f) and (g), as evidenced in the lower-magnification images in the insets ((e–h)
c C. Lieber, Cambridge)
tation, a so-called Peierls instability occurs [24.145]. CDWs have also been found on monolayers
The corresponding phonon energy goes to zero, the of adsorbates such as a monolayer of Pb on
atoms are slightly displaced with the periodicity of the Ge(111) [24.150]. These authors performed a nice
corresponding wavevector, and a charge density wave temperature-dependent study revealing that the CDW
(CDW) with the same periodicity appears. Essentially, is nucleated by scattering states around defects, as
the CDW increases the overlap of the electronic states one might expect [24.151]. Some of the transitions
with the phonon by phase-fixing with respect to the have been interpreted as more complex Mott–Hubbard
atomic lattice. The Peierls transition naturally occurs in transitions caused primarily by electron–electron inter-
one-dimensional (1-D) systems, where only two Fermi actions [24.152]. One-dimensional systems have also
points are present and hence preferential orientation is been prepared on surfaces showing Peierls transi-
pathological. It can also occur in 2-D systems if large tions [24.153, 154]. Finally, the energy gap occurring at
parts of the Fermi line run in parallel. the transition has been studied by measuring dI/ dV (V )
STS studies of CDWs are numerous (e.g., [24.146, curves [24.155].
147]). Examples of a 1-D CDW on a quasi-1-D bulk ma- A more local crystallographic distortion due to
terial and of a 2-D CDW are shown in Fig. 24.16a–d and electron–lattice interactions is the Jahn–Teller effect.
Fig. 24.16e–h, respectively [24.148, 149]. In contrast Here, symmetry breaking by elastic deformation can
to usual scattering states, where LDOS corrugations lead to the lifting of degeneracies close to the Fermi
are only found close to the scatterer, the corrugations level. This results in an energy gain due to the low-
of CDWs are continuous across the surface. Heating ering of the energy of the occupied levels. By tuning
the substrate toward the transition temperature leads to the Fermi level of an adsorbate layer to a degen-
a melting of the CDW lattice, as shown in Fig. 24.16f–h. eracy via doping, such a Jahn–Teller deformation
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 683
has been induced on a surface and visualized by tronic coherence length. Thus, STS probes a different
STM [24.156]. property of the vortex than the usual magnetic imaging
techniques (see Sect. 24.4.4). Surprisingly, first mea-
Superconductors surements of the vortices on NbSe2 revealed vortices
An intriguing effect resulting from electron–phonon in- shaped as a sixfold star [24.162] (Fig. 24.17c). With in-
teraction is superconductivity. Here, the attractive part creasing voltage inside the gap, the orientation of the
of the electron–phonon interaction leads to the cou- star rotates by 30◦ (Fig. 24.17d,e). The shape of these
pling of electronic states with opposite wavevector stars could finally be reproduced by theory, assuming
and mostly opposite spin [24.157]. Since the resulting an anisotropic pairing of electrons in the superconduc-
Cooper pairs are bosons, they can condense at LT, form- tor (Fig. 24.17f–h) [24.163]. Additionally, bound states
ing a coherent many-particle phase, which can carry inside the vortex core, which result from confine-
current without resistance. Interestingly, defect scatter- ment by the surrounding superconducting material, are
ing does not influence the condensate if the coupling found [24.162]. Further experiments investigated the ar-
along the Fermi surface is homogeneous (s-wave su- rangement of the vortex lattice, including transitions
perconductor). The reason is that the symmetry of the between hexagonal and quadratic lattices [24.164], the
scattering of the two components of a Cooper pair effec- influence of pinning centers [24.165], and the vortex
tively leads to a scattering from one Cooper pair state to motion induced by current [24.166].
another without affecting the condensate. This is differ- A central topic is still the understanding of high-
ent if the scatterer is magnetic, since the different spin temperature superconductors (HTCS). An almost ac-
components of the pair are scattered differently, lead- cepted property of HTCS is their d-wave pairing
Part C 24.3
ing to an effective pair breaking, which is visible as symmetry, which is partly combined with other con-
a single-particle excitation within the superconducting tributions [24.167]. The corresponding k-dependent
gap. On a local scale, this effect was first demonstrated gap (where k is the reciprocal lattice vector) can be
by putting Mn, Gd, and Ag atoms on a Nb(110) sur- measured indirectly by STS using a Fourier trans-
face [24.158]. While the nonmagnetic Ag does not formation of the LDOS(x, y) determined at different
modify the gap shown in Fig. 24.17a, it is modified in energies [24.168]. This shows that LDOS modulations
an asymmetric fashion close to Mn or Gd adsorbates, as in HTCS are dominated by simple self-interference
shown in Fig. 24.17b. The asymmetry of the additional patterns of the Bloch-like quasiparticles [24.169]. How-
intensity is caused by the breaking of the particle–hole ever, scattering can also lead to pair breaking (in
symmetry due to the exchange interaction between the contrast to s-wave superconductors), since the Cooper-
localized Mn state and the itinerate Nb states. pair density vanishes in certain directions. Indeed,
Another important local effect is caused by the rel- scattering states (bound states in the gap) around
atively large coherence length of the condensate. At nonmagnetic Zn impurities have been observed in
a material interface, the condensate wavefunction can- Bi2 Sr2 CaCu2 O8+δ (BSCCO) (Fig. 24.17i,j) [24.170].
not stop abruptly, but overlaps into the surrounding They reveal a d-like symmetry, but not the one expected
material (proximity effect). Consequently, a supercon- from simple Cooper-pair scattering. Other effects such
ducting gap can be measured in areas of nonsupercon- as magnetic polarization in the environment probably
ducting material. Several studies have shown this effect have to be taken into account [24.171]. An interesting
on the local scale using metals and doped semiconduc- topic is the importance of inhomogeneities in HTCS
tors as surrounding materials [24.159, 160]. materials. Evidence for inhomogeneities has indeed
While the classical type I superconductors are ideal been found in underdoped materials, where puddles of
diamagnets, the so-called type II superconductors can the superconducting phase identified by the coherence
contain magnetic flux. The flux forms vortices, each peaks around the gap are shown to be embedded in
containing one flux quantum. These vortices are ac- nonsuperconducting areas [24.30].
companied by the disappearance of the superconducting In addition, temperature-dependent measurements
gap and, therefore, can be probed by STS [24.161]. of the gap size development at each spatial position
LDOS maps measured inside the gap lead to bright exhibit a percolation-type behavior above Tc [24.32].
features in the area of the vortex core. Importantly, This stresses the importance of inhomogeneities, but
the length scale of these features is different from the the observed percolation temperature being higher than
length scale of the magnetic flux due to the difference Tc shows that Tc is not caused by percolation of su-
between the London penetration depth and the elec- perconducting puddles only. On the other hand, it
684 Part C Scanning-Probe Microscopy
0.8 0.2
0.4 0
Bare Nb
BCS fit
0 – 0.2
– 0.008 – 0.004 0 0.004 0.008 –8 –4 0 4 8
Voltage (V) Voltage (mV)
c) d) e) f) g) h)
Part C 24.3
i) j) dI/dV (ns)
3.5
3
(Δ)
2.5 34.5 mV
43.5 mV
2
55.5 mV
1.5 67.5 mV
118.5 mV
1
0.5
0
–100 0 100
Voltage (mV)
k) l)
2 pA 0.49 ns
0 pA 0.02 ns
100 Å
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 685
Fig. 24.17 (a) dI/ dV curve of Nb(110) at T = 3.8 K (symbols) in comparison with a BCS fit of the superconducting
gap of Nb (line). (b) Difference between the dI/ dV curve taken directly above a Mn atom on Nb(110) and the dI/ dV
curve taken above clean Nb(110) (symbols) in comparison with a fit using the Bogulubov–de Gennes equations (line) ( c
D. Eigler, Almaden (a,b)). (c–e) dI/ dV images of a vortex core in the type II superconductor 2H-NbSe2 at 0 mV (c),
0.24 mV (d), and 0.48 mV (e) ((c–e) c H. F. Hess). (f–h) Corresponding calculated LDOS images within the Eilenberger
framework ((f–h) c K. Machida, Okayama). (i) Overlap of an STM image at V = −100 mV (background 2-D image)
and a dI/ dV image at V = 0 mV (overlapped 3-D image) of optimally doped Bi2 Sr2 CaCu2 O8+δ containing 0.6% Zn
impurities. The STM image shows the atomic structure of the cleavage plane, while the dI/ dV image shows a bound state
within the superconducting gap, which is located around a single Zn impurity. The fourfold symmetry of the bound state
reflects the d-like symmetry of the superconducting pairing function; (j) dI/ dV spectra of Bi2 Sr2 CaCu2 O8+δ measured at
different positions of the surface at T = 4.2 K; the phonon peaks are marked by arrows, and the determined local gap size
Δ is indicated; note that the strength of the phonon peak increases with the strength of the coherence peaks surrounding
the gap; (k) LDOS in the vortex core of slightly overdoped Bi2 Sr2 CaCu2 O8+δ , B = 5 T; the dI/ dV image taken at
B = 5 T is integrated over V = 1–12 mV, and the corresponding dI/ dV image at B = 0 T is subtracted to highlight
the LDOS induced by the magnetic field. The checkerboard pattern within the seven vortex cores exhibits a periodicity,
which is fourfold with respect to the atomic lattice shown in (i) and is thus assumed to be a CDW; (l) STM image of
cleaved Ca1.9 Na0.1 CuO2 Cl2 at T = 0.1 K, i. e., within the superconducting phase of the material; a checkerboard pattern
with fourfold periodicity is visible on top of the atomic resolution (
c S. Davis, Cornell and S. Uchida, Tokyo (i–l))
was found that for overdoped and optimally doped tiferromagnetic order and superconductivity in HTCS
Part C 24.3
samples the gap develops continuously across Tc , show- materials [24.176]. Since the spin density wave is ac-
ing a universal relation between the local gap size companied by a charge density wave of half wavelength,
Δ(T = 0) (measured at low temperature) and the lo- it can be probed by STS [24.177]. Indeed, a checker-
cal critical temperature Tp (at which the gap completely board pattern of the right periodicity has been found
disappears): 2Δ(T = 0)/kB Tp ≈ 8. The latter result is in and around vortex cores in BSCCO (Fig. 24.17k).
evidence that the so-called pseudogap phase is a phase Similar checkerboards, which do not show any E(k)
with incoherent Cooper pairs. The results are less clear dispersion, have also been found in the underdoped
in the underdoped region, where probably two gaps pseudogap phase at temperatures higher than the super-
complicate the analysis. Below Tc , it turns out that the conducting transition temperature [24.178] or at dopant
strength of the coherence peak is anticorrelated to the densities lower than the critical doping [24.179]. De-
local oxygen acceptor density [24.169] and, in addi- pending on the sample, the patterns can be either
tion, correlated to the energy of an inelastic phonon homogeneous or inhomogeneous and exhibit slightly
excitation peak in dI/ dV spectra [24.31]. Figure 24.17j different periodicities. However, the fact that the pat-
shows corresponding spectra taken at different posi- tern persists within the superconducting phase as shown
tions, where the coherence peaks and the nearby phonon in Fig. 24.17l, at least for Na-CCOC, indicates that the
peaks marked by arrows are clearly visible. The phonon corresponding phase can coexist with superconductiv-
origin of the peak has been proven by the isotope ity. This raises the question of whether spin density
effect, similar to Fig. 24.11a. The strong intensity of waves are the central opponent to HTCS. Interestingly,
the phonon side-peak as well as the correlation of its a checkerboard pattern of similar periodicity, but with-
strength with the coherence peak intensity points to- out long-range order, is also found, if one displays
wards an important role of electron–phonon coupling the particle–hole asymmetry of dI/ dV (V ) intensity
for the pairing mechanism. However, since the gap in underdoped samples at low temperature [24.180].
size does not scale with the strength of the phonon Since the observed asymmetry is known to be caused
peak [24.172], other contributions must be involved too. by the lifting of the correlation gap with doping, the
Of course, vortices have also been investigated for checkerboard pattern might be directly linked to the cor-
HTCS [24.173]. Bound states are found, but at energies responding localized holes in the CuO planes appearing
that are in disagreement with simple models, assum- at low doping. Although a comprehensive model for
ing a Bardeen–Cooper–Schrieffer (BCS)-like d-wave HTCS materials is still lacking, STS contributes signif-
superconductor [24.174, 175]. Theory predicts, instead, icantly to disentangling this puzzle.
that the bound states are magnetic-field-induced spin Even more complex superconductors are based on
density waves, stressing the competition between an- heavy fermions, where superconductivity is known to
686 Part C Scanning-Probe Microscopy
coexist with ferromagnetism. First attempts to obtain Different tips have to be prepared to detect different
information about these materials by STM have been spin orientations. Moreover, the stray magnetic field of
made using very low temperature (190 mK). They ex- the tip can perturb the spin orientation of the sample.
hibit indeed spatial fluctuations of the superconducting To avoid this, a technique using antiferromagnetic Cr
gap [24.181, 182]. However, the key issue for these ma- as a tip coating material has been developed [24.188].
terials is still the preparation of high-quality surfaces This avoids stray fields, but still provides a preferen-
similar to the HTCS materials, where cleavage was ex- tial spin orientation of the few atoms at the tip apex that
tremely advantageous to obtain high-quality data. dominate the tunneling current. Depending on the thick-
Notice that all the measurements described above ness of the Cr coating, spin orientations perpendicular
have probed the superconducting phase only indirectly or parallel to the sample surface, implying correspond-
by measuring the quasiparticle LDOS. The supercon- ing sensitivities to the spin directions of the sample, are
ducting condensate itself could principally also be achieved.
probed directly using Cooper-pair tunneling between SP-STM has been used to image the evolution of
a superconducting tip and a superconducting sam- magnetic domains with increasing B field (Fig. 24.18a–
ple. A proof of principle of this detection scheme d) [24.189], the antiferromagnetic order of a Mn
has indeed been given at low tunneling resistance monolayer on W(110) [24.190], as well as of a Fe
(R ≈ 50 kΩ) [24.183], but meaningful spatially re- monolayer on W(100) (Fig. 24.18e) [24.191], and the
solved data are still lacking. out-of-plane orientation of a magnetic vortex core in
the center of a nanomagnet exhibiting the flux closure
Complex Systems (Manganites) configuration [24.192].
Part C 24.3
Complex phase diagrams are not restricted to HTCS In addition, more complex atomic spin structures
materials (cuprates). They exist with similar complexity showing chiral or noncollinear arrangements have been
for other doped oxides such as manganites. Only a few identified [24.193–195]. Even the spin orientation of
studies of these materials have been performed by STS. a single adatom could be detected, if the adatom is
Some of them show the inhomogeneous evolution of placed either directly on a ferromagnetic island [24.196]
metallic and insulating phases across a metal–insulator or close to a ferromagnetic stripe [24.23]. In the lat-
transition [24.184, 185]. Within layered materials, such ter case, hysteresis curves of the ferromagnetic adatoms
a phase separation has been found to be absent [24.186]. could be measured, as shown in Fig. 24.18f–h. It was
This experiment performed on LaSrMnO revealed, in found that the adatoms couple either ferromagnetically
addition, a peculiar atomic structure, which appears (Fig. 24.18g) or antiferromagnetically (Fig. 24.18h) to
only locally. It has been attributed to the observation the close-by magnetic stripe; i. e., the hysteresis is either
of a local polaron bound to a defect. Since inhomo- in phase or out of phase with the hysteresis of the stripe.
geneities seem to be crucial also in these materials, This behavior, depending on adatom–stripe distance in
a local method such as STS might continue to be an oscillating fashion, directly visualizes the famous
important for the understanding of their complex prop- Ruderman–Kittel–Kasuya–Yoshida (RKKY) interac-
erties. tion [24.23].
An interesting possibility of SP-STM is the observa-
24.3.6 Imaging Spin Polarization: tion of magnetodynamics on the nanoscale. Nanoscale
Nanomagnetism ferromagnetic islands become unstable at a certain
temperature, the so-called superparamagnetic transition
Conventional STS couples to the LDOS, i. e., the charge temperature. Above this temperature, the direction of
distribution of the electronic states. Since electrons also magnetization switches back and forth due to ther-
have spin, it is desirable to also probe the spin distribu- mal excitations. This switching results in a stripe-like
tion of the states. This can be achieved by spin-polarized contrast in SP-STM images, as visible in the inset of
STM (SP-STM) using a tunneling tip covered by a fer- Fig. 24.18i. The island appears dark during the time
romagnetic material [24.187]. The coating acts as a spin when the orientation of the island spin is opposite to
filter or, more precisely, the tunneling current depends the orientation of the tip spin, and switches to bright
on the relative angle αij between the spins of the tip and when the island spin orientation changes. By observing
the sample according to cos(αij ). Consequently, a par- the switching as a function of time on different islands
ticular tip is not sensitive to spin orientations of the at different temperatures the energy barriers of indi-
sample perpendicular to the spin orientation of the tip. vidual islands can be determined [24.197]. Even more
Low-Temperature Scanning Probe Microscopy 24.3 Scanning Tunneling Microscopy and Spectroscopy 687
50 nm 2 nm
f) i) Incidence (%)
6
2000 nA
Pt(111)
4 800 nA
5 nm
1.8 nm
State “0” State “1”
Part C 24.3
Co stripe
2
2 nA
importantly, the preferential orientation during switch- observed asymmetry in the peak intensity increases with
ing can be tuned by the tunneling current. This is visible current, providing evidence that current-induced mag-
in Fig. 24.18i, which shows the measured orientational netization switching is possible even on the atomic
probability at different tunneling currents [24.198]. The scale.
consist of a cantilever with a tip at its free end that oscillating tip (see Chaps. 23 and 26).
is brought close to the sample surface. The cantilever Whatever method is used, the resolution of
can be mounted parallel or perpendicular to the surface a cantilever-based force detection is fundamentally lim-
(general aspects of force probe designs are described in ited by its intrinsic thermomechanical noise. If the
Chap. 22). Basically, two methods exist to detect forces cantilever is in thermal equilibrium at a temperature T ,
with cantilever-based probes: the static and the dynamic the equipartition theorem predicts a thermally induced
mode (see Chap. 21). They can be used to generate root-mean-square (RMS) motion of the cantilever in
a laterally resolved image (microscopy mode) or deter- the z direction of z RMS = (kB T/ceff )1/2 , where kB is
mine its distance dependence (spectroscopy mode). One the Boltzmann constant and ceff = cz + ∂Fts /∂z. Note
can argue about this terminology, since spectroscopy is that usually dFts / dz cz in the contact mode and
usually related to energies and not to distance depen- dFts / dz < cz in the noncontact mode. Evidently, this
dencies. Nevertheless, we will use it throughout the text, fundamentally limits the force resolution in the static
because it avoids lengthy paraphrases and is established mode, particularly if operated in the noncontact mode.
in this sense throughout the literature. Of course, the same is true for the different dynamic
In the static mode, a force that acts on the tip modes, because the thermal energy kB T excites the
bends the cantilever. By measuring its deflection Δz eigenfrequency f 0 of the cantilever. Thermal noise is
the tip–sample force Fts can be directly calculated white noise, i. e., its spectral density is flat. However,
from Hooke’s law: Fts = cz Δz, where cz denotes the if the cantilever transfer function is taken into account,
spring constant of the cantilever. In the various dynamic one can see that the thermal energy mainly excites f 0 .
modes, the cantilever is oscillated with amplitude A This explains the term “thermo” in thermomechanical
at or near its eigenfrequency f 0 , but in some appli- noise, but what is the “mechanical” part?
cations also off-resonance. At ambient pressures or in A more detailed analysis reveals that the thermally
liquids, amplitude modulation (AM-SFM) is used to induced cantilever motion is given by
detect amplitude changes or the phase shift between
the driving force and cantilever oscillation. In vac- 2kB TB
uum, the frequency shift Δ f of the cantilever due to z RMS = , (24.2)
πcz f 0 Q
a tip–sample interaction is measured by the frequency-
modulation technique (FM-SFM). The nomenclature is where B is the measurement bandwidth and Q is the
not standardized. Terms such as tapping mode or in- quality factor of the cantilever. Analogous expressions
termittent contact mode are used instead of AM-SFM, can be obtained for all quantities measured in dynamic
and NC-AFM (noncontact atomic force microscopy) or modes, because the deflection noise translates, e.g., into
Low-Temperature Scanning Probe Microscopy 24.4 Scanning Force Microscopy and Spectroscopy 689
frequency noise [24.5]. Note that f 0 and cz are corre- flat surface. The latter condition can only be fulfilled in
lated with each other via 2π f 0 = (cz /m eff )1/2 , where the dynamic mode, where the additional restoring force
the effective mass m eff depends on the geometry, den- cz A at the lower turnaround point prevents the jump-
sity, and elasticity of the material. The Q-factor of the to-contact. As described in Chap. 23, by preventing
cantilever is related to the external damping of the can- the so-called jump-to-contact, true atomic resolution is
tilever motion in a medium and to the intrinsic damping nowadays routinely obtained in vacuum by FM-AFM.
within the material. This is the “mechanical” part of the The nature of the short-range tip–sample interaction
fundamental cantilever noise. during imaging with atomic resolution has been stud-
It is possible to operate a low-temperature force mi- ied experimentally as well as theoretically. Si(111)-(7 ×
croscope directly immersed in the cryogen [24.199,200] 7) was the first surface on which true atomic resolu-
or in the cooling gas [24.201], whereby the cooling is tion was achieved [24.205], and several studies have
simple and very effective. However, it is evident from been performed at low temperatures on this well-known
(24.2) that the smallest fundamental noise is achievable material [24.206–208]. First-principles simulations per-
in vacuum, where the Q-factors are more than 100 times formed on semiconductors with a silicon tip revealed
larger than in air, and at low temperatures. that chemical interactions, i. e., a significant charge re-
The best force resolution up to now, which is distribution between the dangling bonds of the tip and
better than 1 × 10−18 N/Hz1/2 , has been achieved by sample, dominate the atomic-scale contrast [24.209–
Mamin et al. [24.202] in vacuum at a temperature be- 211]. On V–III semiconductors, it was found that only
low 300 mK. Due to the reduced thermal noise and the one atomic species, the group V atoms, is imaged as
lower thermal drift, which results in a higher stability protrusions with a silicon tip [24.210, 211]. Further-
Part C 24.4
of the tip–sample gap and a better signal-to-noise ratio, more, these simulations revealed that the sample, as
the highest resolution is possible at low temperatures in well as the tip atoms, are noticeably displaced from
ultrahigh vacuum with FM-SFM. A vertical RMS noise their equilibrium position due to the interaction forces.
below 2 pm [24.203, 204] and a force resolution below At low temperatures, both aspects could be observed
1 aN [24.202] have been reported. with silicon tips on indium arsenide [24.203, 212]. On
Besides the reduced noise, the application of weakly interacting surfaces the short-range interatomic
force detection at low temperatures is motivated by van der Waals force has been believed responsible for
the increased stability and the possibility to observe the atomic-scale contrast [24.213–215].
phenomena that appear below a certain critical temper-
ature Tc , as outlined on page 664. The experiments, Chemical Sensitivity of Force Microscopy
which have been performed at low temperatures until The (110) surface of the III–V semiconductor in-
now, were motivated by at least one of these reasons dium arsenide exhibits both atomic species in the top
and can be roughly divided into four groups: layer (see Fig. 24.19a). Therefore, this sample is well
suited to study the chemical sensitivity of force mi-
(i) Atomic-scale imaging
croscopy [24.203]. In Fig. 24.19b, the usually observed
(ii) Force spectroscopy
atomic-scale contrast on InAs(110) is displayed. As pre-
(iii)Investigation of quantum phenomena by measuring
dicted, the arsenic atoms, which are shifted by 80 pm
electrostatic forces
above the indium layer due to the (1 × 1) relaxation,
(iv)Utilizing magnetic probes to study ferromagnets,
are imaged as protrusions. While this general appear-
superconductors, and single spins
ance was similar for most tips, two other distinctively
In the following, we describe some exemplary re- different contrasts were also observed: a second protru-
sults. sion (Fig. 24.19c) and a sharp depression (Fig. 24.19d).
The arrangement of these two features corresponds well
24.4.1 Atomic-Scale Imaging to the zigzag configuration of the indium and arsenic
atoms along the [11̄0] direction. A sound explanation
In a simplified picture, the dimensions of the tip end would be as follows: the contrast usually obtained with
and its distance to the surface limit the lateral resolution one feature per surface unit cell corresponds to a silicon-
of force microscopy, since it is a near-field technique. terminated tip, as predicted by simulations. A different
Consequently, atomic resolution requires a stable single atomic species at the tip apex, however, can result in
atom at the tip apex that has to be brought within a dis- a very different charge redistribution. Since the atomic-
tance of some tenths of a nanometer of an atomically scale contrast is due to a chemical interaction, the two
690 Part C Scanning-Probe Microscopy
-
a) [110] b) c) d)
In 1st layer
In 2nd layer
-
As 1st layer [110]
- 1 nm -
[001] As 2nd layer [001]
Fig. 24.19a–d The structure of InAs(110) as seen from above (a) and three FM-AFM images of this surface obtained with
different tips at 14 K (b–d). In (b), only the arsenic atoms are imaged as protrusions, as predicted for a silicon tip. The two
features in (c) and (d) corresponds to the zigzag arrangement of the indium and arsenic atoms. Since force microscopy is
sensitive to short-range chemical forces, the appearance of the indium atoms can be associated with a chemically different
tip apex
other contrasts would then correspond to a tip that has other arsenic atoms are also pulled, but they are less dis-
been accidentally contaminated with sample material placed, because they have three bonds to the bulk, while
(an arsenic- or indium-terminated tip apex). Never- the two arsenic atoms in the neighborhood of an indium
theless, this explanation has not yet been verified by vacancy have only two bonds. This direct experimen-
Part C 24.4
simulations for this material. tal proof of the presence of tip-induced relaxations is
also relevant for STM measurements, because the tip– of the atoms, i. e., the minima correspond to the posi-
sample distances are similar during atomic-resolution tions of the carbon atoms. This can be related to the
imaging. Moreover, the result demonstrates that FM-
AFM can probe elastic properties on an atomic level.
a) Corrugation a. u.
Imaging of Weakly Interacting
van der Waals Surfaces 3
For weakly interacting van der Waals surfaces, much
2
smaller atomic corrugation amplitudes are expected
compared with strongly interacting surfaces of semi-
conductors. A typical example is graphite, a layered 1
material, where the carbon atoms are covalently bonded
H H
and arranged in a honeycomb structure within the
(0001) plane. Individual graphene layers stick together A B
by van der Waals forces. Due to the ABA stacking,
three distinctive sites exist on the (0001) surface: car- 426 pm
bon atoms with (A-type) and without (B-type) neighbor
b) Corrugation a. u.
in the next graphite layer and the hollow site (H-site) in
the hexagon center. In static contact force microscopy 3
as well as in STM the contrast usually exhibits a trigo- 3
Part C 24.4
1 2
nal symmetry with a periodicity of 246 pm, where A-
and B-site carbon atoms could not be distinguished.
However, in high-resolution FM-AFM images acquired 2 1
at low temperatures, a large maximum and two differ-
ent minima have been resolved, as demonstrated by the
profiles along the three equivalent [1-100] directions in H H
Fig. 24.21a. A simulation using the Lennard-Jones (LJ) C C
potential, given by the short-range interatomic van der
Waals force, reproduced these three features very well 426 pm
(dotted line). Therefore, the large maximum could be c) Corrugation a. u.
assigned to the H-site, while the two different minima
represent A- and B-type carbon atoms [24.214].
Compared with graphite, the carbon atoms in 2
a single-walled carbon nanotube (SWNT), which con-
sists of a single rolled-up graphene layer, are indis- 1 Xe Xe
tinguishable. For the first time Ashino et al. [24.215]
successfully imaged the curved surface of a SWNT
2 1
with atomic resolution. Note that, for geometric rea-
sons, atomic resolution is only achieved on the top
(see Fig. 24.21b). Indeed, as shown in Fig. 24.21b, all Xe Xe Xe
profiles between two hollow sites across two neighbor-
900 pm
ing carbon atoms are symmetric [24.217]. Particularly,
curves 1 and 2 exhibit two minima of equal depth, Fig. 24.21a–c FM-AFM images of (a) graphite(0001),
as predicted by theory (cf., dotted line). The assump- (b) a single-walled carbon nanotube (SWNT), and
tion used in the simulation (dotted lines in the profiles (c) Xe(111) recorded at 22 K. On the right side, line sec-
of Fig. 24.21) that interatomic van der Waals forces tions taken from the experimental data (solid lines) are
are responsible for the atomic-scale contrast has been compared with simulations (dotted lines). A- and B-type
supported by a quantitative evaluation of force spec- carbon atoms, as well as the hollow site (H-site) on
troscopy data obtained on SWNTs [24.215]. graphite can be distinguished, but are imaged with inverted
Interestingly, the image contrast on graphite and contrast, i. e., the carbon sites are displayed as minima.
SWNTs is inverted with respect to the arrangement Such an inversion does not occur on Xe(111)
692 Part C Scanning-Probe Microscopy
small carbon–carbon distance of only 142 pm, which order of 0.1 nm. On the other hand, stable imaging
is in fact the smallest interatomic distance that has without a jump-to-contact is only possible as long as
been resolved with FM-AFM so far. The van der Waals the restoring force cz A at the lower turnaround point
radius of the front tip atom, (e.g., 210 pm for sili- of each cycle is larger than the maximal attractive
con) has a radius that is significantly larger than the tip–sample force. Therefore, reducing the desired am-
intercarbon distance. Therefore, next-nearest-neighbor plitude by a factor of 100 requires a 100 times larger
interactions become important and result in contrast in- spring constant. Indeed, Hembacher et al. [24.219]
version [24.217]. could demonstrate atomic resolution with small am-
While experiments on graphite and SWNTs ba- plitudes (about 0.25 nm) and large spring constants
sically take advantage of the increased stability and (about 1800 N/m) utilizing a qPlus sensor [24.220].
signal-to-noise ratio at low temperatures, solid xenon Figure 24.22 shows a constant-height image of graphite
(melting temperature Tm = 161 K) can only be observed recorded at 4.9 K within the repulsive regime. Note
at sufficient low temperatures [24.8]. In addition, xenon that, compared with Fig. 24.21a,b, the contrast is in-
is a pure van der Waals crystal and, since it is an insu- verted, i. e., the carbon atoms appear as maxima.
lator, FM-AFM is the only real-space method available This is expected, because the imaging interaction is
today that allows the study of solid xenon on the atomic switched from attractive to repulsive regime [24.213,
scale. 217].
Allers et al. [24.8] adsorbed a well-ordered xenon
film on cold graphite(0001) (T < 55 K) and studied it 24.4.2 Force Spectroscopy
subsequently at 22 K by FM-AFM (Fig. 24.21c). The
Part C 24.4
sixfold symmetry and the distance between the pro- A wealth of information about the nature of the tip–
trusions corresponds well with the nearest-neighbor sample interaction can be obtained by measuring its
distance in the close-packed (111) plane of bulk xenon, distance dependence. This is usually done by record-
which crystallizes in a face-centered cubic structure. ing the measured quantity (deflection, frequency shift,
A comparison between experiment and simulation con- amplitude change, phase shift) and applying an appro-
firmed that the protrusions correspond to the position of priate voltage ramp to the z-electrode of the scanner
the xenon atoms [24.214]. However, the simulated cor- piezo, while the z-feedback is switched off. Accord-
rugation amplitudes do not fit as well as for graphite ing to (24.2), low temperatures and high Q-factors
(see sections in Fig. 24.21c). A possible reason is that (vacuum) considerably increase the force resolution. In
tip-induced relaxations, which were not considered in the static mode, long-range forces and contact forces
the simulations, are more important for this pure van can be examined. Force measurements at small tip–
der Waals crystal xenon than they are for graphite, be- sample distances are inhibited by the jump-to-contact
cause in-plane graphite exhibits strong covalent bonds. phenomenon: If the force gradient ∂Fts /∂z becomes
Nevertheless, the results demonstrated for the first time larger than the spring constant cz , the cantilever can-
that a weakly bonded van der Waals crystal could be not resist the attractive tip–sample forces and the tip
imaged nondestructively on the atomic scale. Note that snaps onto the surface. Sufficiently large spring con-
on Xe(111) no contrast inversion exists, presumably be- stants prevent this effect, but reduce the force resolution.
cause the separation between Xe sites is about 450 pm, In the dynamic modes, the jump-to-contact can be
i. e., twice as large as the van der Waals radius of a sili- avoided due to the additional restoring force (cz A) at
con atom at the tip end. the lower turnaround point. The highest sensitivity can
be achieved in vacuum by using the FM technique, i. e.,
Atomic Resolution by recording Δ f (z) curves. An alternative FM spec-
Using Small Oscillation Amplitudes troscopy method, the recording of Δ f (A) curves, has
All the examples above described used spring con- been suggested by Hölscher et al. [24.221]. Note that,
stants and amplitudes on the order of 40 N/m and if the amplitude is much larger than the characteris-
10 nm, respectively, to obtain atomic resolution. How- tic decay length of the tip–sample force, the frequency
ever, Giessibl et al. [24.218] pointed out that the shift cannot simply be converted into force gradients
optimal amplitude should be on the order of the char- by using ∂Fts /∂z = 2cz Δ f/ f 0 [24.222]. Several meth-
acteristic decay length λ of the relevant tip–sample ods have been published to convert Δ f (z) data into
interaction. For short-range interactions, which are re- the tip–sample potential Vts (z) and tip–sample force
sponsible for the atomic-scale contrast, λ is on the Fts (z) [24.223–226].
Low-Temperature Scanning Probe Microscopy 24.4 Scanning Force Microscopy and Spectroscopy 693
Part C 24.4
reconstructed from the Δ f (z) curve recorded above an
adatom and represents the total force. After removing –60
the long-range contribution from the data, the much
steeper brown line is obtained, which corresponds to the –80 3 nm
short-range force between the adatom and the atom at
the tip apex. The measured maximum attractive force 0 0.5 1 1.5 2
( − 2.1 nN) agrees well with that obtained from first- z-displacement (nm)
principles calculations ( − 2.25 nN). b) Force (nN)
0
Short-range
force Fsr
–1
–2
Total
force Ftot
1.49 –3
0 0.4 0.6 0.8 1 1.2
z-displacement (nm)
Fig. 24.23a,b FM force spectroscopy on specific atomic
0.86
sites at 7.2 K. In (a), an FM-SFM image of the
Si(111)-(7 × 7) surface is displayed together with two
Fig. 24.22 Constant-height FM-AFM image of graphite Δ f (z) curves, which have been recorded at the positions
(0001) recorded at 4.9 K using a small amplitude ( A = indicated by the arrows, i. e., above the corner hole (black)
0.25 nm) and a large spring constant (cz = 1800 N/m). and above an adatom (brown). In (b), the total force above
As in Fig. 24.20a, A- and B-site carbon atoms can be an adatom (black line) has been recovered from the Δ f (z)
distinguished. However, they appear as maxima, because curve. After subtraction of the long-range part, the short-
imaging has been performed in the repulsive regime (c range force can be determined (brown line) (courtesy of
F. J. Giessibl [24.219]) H. J. Hug; cf. [24.227])
694 Part C Scanning-Probe Microscopy
Three-Dimensional Force Field Spectroscopy reactivity. A first step in this direction has been ac-
Further progress with the FM technique has been made complished on SWNTs. Its structural unit, a hexagonal
by Hölscher et al. [24.231]. They acquired a complete carbon ring, is common to all aromatic molecules. Like
3-D force field on NiO(001) with atomic resolution (3- the constant frequency-shift image of an SWNT shown
D force field spectroscopy). In Fig. 24.24, the atomically in Fig. 24.21b the force map shows clear differences
resolved FM-AFM image of NiO(001) is shown to- between hollow sites and carbon sites [24.215]. Analyz-
gether with the coordinate system used and the tip to ing site-specific individual force curves extracted from
illustrate the measurement principle. NiO(001) crystal- the 3-D data revealed a maximum attractive force of
lizes in the rock-salt structure. The distance between ≈ −0.106 nN above H-sites and ≈ −0.075 nN above
the protrusions corresponds to the lattice constant of carbon sites. Since the attraction is one order of mag-
417 pm, i. e., only one type of atom (most likely the oxy- nitude weaker than on Si(111)-(7 × 7) (Fig. 24.23b), it
gen) is imaged as a protrusion. In an area of 1 nm × 1 nm, has been inferred that the short-range interatomic van
32 × 32 individual Δ f (z) curves have been recorded der Waals force and not a chemical force is responsible
at every (x, y) image point and converted into Fts (z) for atomic-scale contrast formation on such nonreac-
curves. The Δ f (x, y, z) data set is thereby converted tive surfaces. It is worth mentioning that 3-D force field
into the 3-D force field Fts (x, y, z). Figure 24.24, where spectroscopy data have been acquired at room tempera-
a specific x–z-plane is displayed, demonstrates that ture as well [24.232, 233].
atomic resolution is achieved. It represents a 2-D cut Apart from calculating the vertical tip–sample force
Fts (x, y = const, z) along the [100] direction (corre- Schwarz et al. [24.234] demonstrated that it is also pos-
sponding to the shaded slice marked in Fig. 24.24). sible to obtain the lateral tip–sample force from 3-D
Part C 24.4
Since a large number of curves have been recorded, data sets. First, the tip–sample potential Vts (x, y, z) has
Langkat et al. [24.228] could evaluate the whole data to be determined. Then the lateral force components can
set by standard statistical means to extract the long- be calculated by taking the derivative with respect to the
and short-range forces. A possible future application x- and y-coordinate, respectively. This technique has
of 3-D force field spectroscopy could be to map the been employed to determine the lateral force needed
short-range forces of complex molecules with function- to move an atom sideways by Ternes et al. [24.42].
alized tips in order to resolve locally their chemical
Lateral force Fx (nN)
a) b) z z = 160 pm
z
z = 170 pm
0.4 0.2 z = 195 pm
–0.2 nN z = 245 pm
0.3
0.1
x 0.2
0
0.1
–1 nN
x
0 0.2 0.4 0.6 0.8 1 –0.1
y x
Fig. 24.24a,b Principle of the 3-D force field spectroscopy method –0.2
(a) and a 2-D cut through the 3-D force field Fts (x, y, z) recorded at
14 K (b). At all 32 × 32 image points of the 1 nm × 1 nm scan area
on NiO(001), a Δ f (z) curve has been recorded. The Δ f (x, y, z) –600 –400 –200 0 200 400 600
data set obtained is then converted into the 3-D tip–sample force x (pm)
field Fts (x, y, z). The shaded slice Fts (x, y = const, z) in (a) cor- Fig. 24.25 Lateral force curves recorded at constant tip–
responds to a cut along the [100] direction and demonstrates that sample separation z. At the lowest separation a discon-
atomic resolution has been obtained, because the distance between tinuity appears, which marks the jump of the Co atom
the protrusions corresponds well to the lattice constant of nickel form one site to the next as indicated in the inset (.
c
oxide M. Ternes [24.42]).
Low-Temperature Scanning Probe Microscopy 24.4 Scanning Force Microscopy and Spectroscopy 695
They recorded constant-height data at different tip– write nanostructures in a bottom-up process with single
sample distances to obtain first Vts (x, z) and then the atoms [24.237].
lateral force Fx (x, z) = d/ dxVts (x, z). Four curves of
the whole data set are displayed in Fig. 24.25. The dis- 24.4.4 Electrostatic Force Microscopy
continuity of the lateral force at the lowest adjusted
tip–sample distance (z = 160 pm) indicates the jump of Electrostatic forces are readily detectable by a force mi-
the Co atom from one hollow site to the next on Pt(111), croscope, because the tip and sample can be regarded
cf., inset. It takes place at about 210 pN. as two electrodes of a capacitor. If they are electrically
connected via their back sides and have different work
24.4.3 Atomic Manipulation functions, electrons will flow between the tip and sam-
ple until their Fermi levels are equalized. As a result,
Nowadays, atomic-scale manipulation is routinely per- an electric field, and consequently an attractive elec-
formed using an STM tip (see Sect. 24.3.1). In most trostatic force, exists between them at zero bias. This
of these experiments an adsorbate is dragged with the contact potential difference can be balanced by applying
tip by using an attractive force between the foremost an appropriate bias voltage. It has been demonstrated
tip apex atoms and the adsorbate. By adjusting a large that individual doping atoms in semiconducting mater-
or a small tip–surface distance via the tunneling re- ials can be detected by electrostatic interactions due
sistance, it is possible to switch between imaging and to the local variation of the surface potential around
manipulation. Recently, it has been demonstrated that them [24.238, 239].
controlled manipulation of individual atoms is also
Part C 24.4
possible in the dynamic mode of atomic force mi- Detection of Edge Channels
croscopy, i. e., FM-AFM. Vertical manipulation was in the Quantum Hall Regime
demonstrated by pressing the tip in a controlled manner At low temperatures, electrostatic force microscopy has
into a Si(111)-(7 × 7) surface [24.236]. The strong repul- been used to measure the electrostatic potential in the
sion leads to the removal of the selected silicon atom. quantum Hall regime of a two-dimensional electron gas
The process could be traced by recording the frequency (2-DEG) buried in epitaxially grown GaAs/AlGaAs
shift and the damping signal during the approach. For heterostructures [24.240–243]. In the 2-DEG, electrons
lateral manipulation a rubbing technique has been uti- can move freely in the x–y-plane, but they cannot
lized [24.235], where the slow scan axis is halted move in z-direction. Electrical transport properties of
above a selected atom, while the tip–surface distance a 2-DEG are very different compared with normal
is gradually reduced until the selected atom hops to metallic conduction. Particularly, the Hall resistance
a new stable position. Figure 24.26 shows a Ge adatom RH = h/ne2 (where h represents Planck’s constant, e is
on Ge(111)-c(2 × 8) that was moved during scanning the electron charge, and n = 1, 2, . . .) is quantized in
in two steps from its original position (Fig. 24.26a) the quantum Hall regime, i. e., at sufficiently low tem-
to its final position (Fig. 24.26c). In fact, manipula- peratures (T < 4 K) and high magnetic fields (up to
tion by FM-AFM is reproducible and fast enough to 20 T). Under these conditions, theoretical calculations
a) b) c)
Fig. 24.26a–c Consecutively recorded FM-AFM images showing the tip-induced manipulation of a Ge adatom on
Ge(111)-c(2 × 8) at 80 K. Scanning was performed from bottom to top (
c N. Oyabu [24.235])
696 Part C Scanning-Probe Microscopy
predict the existence of edge channels in a Hall bar. where h represents Planck’s constant and e the electron
A Hall bar is a strip conductor that is contacted in a spe- charge).
cific way to allow longitudinal and transversal transport
measurements in a perpendicular magnetic field. The 24.4.5 Magnetic Force Microscopy
current is not evenly distributed over the cross-section
of the bar, but passes mainly along rather thin paths To detect magnetostatic tip–sample interactions with
close to the edges. This prediction has been verified by magnetic force microscopy (MFM), a ferromagnetic
measuring profiles of the electrostatic potential across probe has to be used. Such probes are readily prepared
a Hall bar in different perpendicular external magnetic by evaporating a thin magnetic layer, e.g., 10 nm iron,
fields [24.240–242]. onto the tip. Due to the in-plane shape anisotropy of thin
Figure 24.27a shows the experimental setup used films, the magnetization of such tips lies predominantly
to observe these edge channels on top of a Hall bar along the tip axis, i. e., perpendicular to the surface.
with a force microscope. The tip is positioned above Since magnetostatic interactions are long range, they
the surface of a Hall bar under which the 2-DEG is can be separated from the topography by scanning at
buried. The direction of the magnetic field is oriented a certain constant height (typically around 20 nm) above
perpendicular to the 2-DEG. Note that, although the the surface, where the z-component of the sample stray
2-DEG is located several tens of nanometers below field is probed (Fig. 24.28a). Therefore, MFM is al-
the surface, its influence on the electrostatic surface ways operated in noncontact mode. The signal from the
potential can be detected. In Fig. 24.27b, the results cantilever is directly recorded while the z-feedback is
of scans perpendicular to the Hall bar are plotted switched off. MFM can be operated in the static mode or
Part C 24.4
against the magnitude of the external magnetic field. in the dynamic modes (AM-MFM at ambient pressures
The value of the electrostatic potential is grey-coded and FM-MFM in vacuum). A lateral resolution below
in arbitrary units. In certain field ranges, the potential 50 nm can be routinely obtained.
changes linearly across the Hall bar, while in other field
ranges the potential drop is confined to the edges of Observation of Domain Patterns
the Hall bar. The predicted edge channels can explain MFM is widely used to visualize domain patterns of
this behavior. The periodicity of the phenomenon is ferromagnetic materials. At low temperatures, Moloni
related to the filling factor ν, i. e., the number of Lan- et al. [24.244] observed the domain structure of
dau levels that are filled with electrons (Sect. 24.3.4). magnetite below its Verwey transition temperature
Its value depends on 1/B and is proportional to the (TV = 122 K), but most of the work has concentrated
electron concentration n e in the 2-DEG (ν = n e h/eB, on thin films of La1−x Cax MnO3 [24.245–247]. Below
a) b) c)
205 mT 400 mT
d) e) f)
800 mT 360 mT 0 mT
Fig. 24.28a–f Principle of MFM operation (a) and field-dependent domain structure of a ferromagnetic thin film
Part C 24.4
(b–f) recorded at 5.2 K with FM-MFM. All images were recorded on the same 4 μm × 4 μm scan area. The
La0.7 Ca0.3 MnO3 /LaAlO3 system exhibits a substrate-induced out-of-plane anisotropy. Bright and dark areas are vis-
ible and correspond to attractive and repulsive magnetostatic interactions, respectively. The series shows how the domain
pattern evolves along the major hysteresis loop, i. e., from zero field to saturation at 600 mT and back to zero field
TV , the conductivity decreases by two orders of magni- drical domains first nucleate and then start to grow. At
tude and a small structural distortion is observed. The zero field, the maze-type domain pattern has evolved
domain structure of this mixed-valence manganite is of again. Such data sets can be used to analyze domain nu-
great interest, because its resistivity strongly depends cleation and the domain growth mode. Moreover, due
on the external magnetic field, i. e., it exhibits a large to the negligible drift, domain structure and surface
colossal-magnetoresistive effect. To investigate the field morphology can be directly compared, because every
dependence of the domain patterns under ambient con- MFM can be used as a regular topography-imaging
ditions, electromagnets have to be used. They can cause force microscope.
severe thermal drift problems due to Joule heating of
the coils by large currents. Flux densities on the order Detection of Individual Vortices
of 100 mT can be achieved. In contrast, much larger in Superconductors
flux densities (more than 10 T) can be rather easily pro- Numerous low-temperature MFM experiments have
duced by implementing a superconducting magnet in been performed on superconductors [24.248–255].
low-temperature setups. Using such a design, Liebmann Some basic features of superconductors have been men-
et al. [24.247] recorded the domain structure along the tioned already in Sect. 24.3.5. The main difference of
major hysteresis loop of La0.7 Ca0.3 MnO3 epitaxially STM/STS compared to MFM is its high sensitivity to
grown on LaAlO3 (Fig. 24.28b–f). The film geometry the electronic properties of the surface. Therefore, care-
(with thickness of 100 nm) favors an in-plane mag- ful sample preparation is a prerequisite. This is not
netization, but the lattice mismatch with the substrate so important for MFM experiments, since the tip is
induces an out-of-plane anisotropy. Thereby, an irregu- scanned at a certain distance above the surface.
lar pattern of strip domains appears at zero field. If the Superconductors can be divided into two classes
external magnetic field is increased, the domains with with respect to their behavior in an external magnetic
antiparallel orientation shrink and finally disappear in field. For type I superconductors, all magnetic flux is
saturation (Fig. 24.28b,c). The residual contrast in satu- entirely excluded below their critical temperature Tc
ration (Fig. 24.28d) reflects topographic features. If the (Meissner effect), while for type II superconductors,
field is decreased after saturation (Fig. 24.28e,f), cylin- cylindrical inclusions (vortices) of normal material exist
698 Part C Scanning-Probe Microscopy
in a superconducting matrix (vortex state). The radius superconducting material. Pi et al. [24.255] slowly in-
of the vortex core, where the Cooper-pair density de- creased the external magnetic field until the vortex front
creases to zero, is on the order of the coherence length ξ. approaching from the edge reached the scanning area.
Since the superconducting gap vanishes in the core, they If the vortex configuration is dominated by the
can be detected by STS (see Sect. 24.3.5). Addition- pinning of vortices at randomly distributed structural
ally, each vortex contains one magnetic quantum flux defects, no Abrikosov lattice emerges. The influence of
Φ = h/2e (where h represents Planck’s constant and pinning centers can be studied easily by MFM, because
e the electron charge). Circular supercurrents around every MFM can be used to scan the topography in its
the core screen the magnetic field associated with AFM mode. This has been done for natural growth de-
a vortex; their radius is given by the London penetra- fects by Moser et al. [24.251] on YBa2 Cu3 O7 and for
tion depth λ of the material. This magnetic field of YBa2 Cu3 O7 and niobium thin films, respectively, by
the vortices can be detected by MFM. Investigations Volodin et al. [24.254]. Roseman and Grütter [24.256]
have been performed on the two most popular copper investigated the formation of vortices in the presence
oxide high-Tc superconductors, YBa2 Cu3 O7 [24.248, of an artificial structure on niobium films, while Pi
249, 251] and Bi2 Sr2 CaCu2 O8 [24.249, 255], on the et al. [24.255] produced columnar defects by heavy-
only elemental conventional type II superconductor ion bombardment in a Bi2 Sr2 CaCu2 O8 single crystal to
Nb [24.252, 253] and on the layered compound crystal study the strong pinning at these defects.
NbSe2 [24.250, 252]. Figure 24.29 demonstrates that MFM is sensitive
Most often, vortices have been generated by cool- to the polarity of vortices. In Fig. 24.29a, six vortices
ing the sample from the normal state to below Tc in have been produced in a niobium film by field cool-
Part C 24.4
an external magnetic field. After such a field-cooling ing in + 0.5 mT. The external magnetic field and tip
procedure, the most energetically favorable vortex ar- magnetization are parallel, and therefore the tip–vortex
rangement is a regular triangular Abrikosov lattice. interaction is attractive (bright contrast). To remove the
Volodin et al. [24.250] were able to observe such an vortices, the niobium was heated above Tc (≈ 9 K).
Abrikosov lattice on NbSe2 . The intervortex distance d Thereafter, vortices of opposite polarity were produced
is related to the external field during B cool down via by field-cooling in − 0.5 mT, which appear dark in
d = (4/3)1/4 (Φ/B)1/2 . Another way to introduce vor- Fig. 24.29b. The vortices are probably bound to strong
tices into a type II superconductor is vortex penetration pinning sites, because the vortex positions are identical
from the edge by applying a magnetic field at tempera- in both images of Fig. 24.29. By imaging the vortices at
tures below Tc . According to the Bean model, a vortex different scanning heights, Roseman et al. [24.253] tried
density gradient exists under such conditions within the to extract values for the London penetration depth from
the scan-height dependence of their profiles. While
good qualitative agreement with theoretical predictions
a) b)
has been found, the absolute values do not agree with
published literature values. The disagreement was at-
tributed to the convolution between the tip and vortex
stray fields. Better values might be obtained with cali-
brated tips.
Magnetic tip
[010]
[110]
[100] 1 nm 1 nm Spin
Resonant
Fig. 24.30 (a) Pure chemical contrast on NiO(001) ob- slice
tained with AFM using a nonmagnetic tip. Oxygen and z
y
nickel atoms are represented as maxima and minima, re-
spectively, forming the (1 × 1) surface unit cell (black
square). Arrows indicate the main crystallographic direc- x
tions. (b) Additional modulation on neighboring nickel Fig. 24.31 MRFM setup. The cantilever with the magnetic
rows along the [110] direction (see arrows) due to the mag- tip oscillates parallel to the surface. Only electron spins
netic exchange interaction obtained with MExFM using within a hemispherical slice, where the stray field of the
a magnetic tip. The (2 × 1) structure (black rectangle in the tip plus the external field matches the condition for mag-
inset) represents the magnetic surface unit cell. The inset is netic resonance, can contribute to the MRFM signal due to
Part C 24.4
tiled together from the averaged magnetic unit cell calcu- cyclic spin inversion (c D. Rugar [24.262])
lated from the raw data, whereby the signal-to-noise ratio
is significantly increased Moreover, a comparison between simulation and ex-
perimental data revealed complex interplay between
tions. They found a magnetic exchange force between chemical and magnetic interaction, which results in the
the magnetic moments of a single iron atom (the tip) and observed atomic-scale contrast.
nickel surface atoms of more than 0.1 nN at tip–sample Even more ambitious is the proposed detection of
distances below 0.5 nm. Recently, Kaiser et al. [24.260] individual nuclear spins by magnetic resonance force
were able to prove the feasibility of magnetic exchange microscopy (MRFM) using a magnetic tip [24.263,
force microscopy (MExFM) on NiO(001). The superex- 264]. Conventionally, nuclear spins are investigated by
change between neighboring {111} planes via bridging nuclear magnetic resonance (NMR), a spectroscopic
oxygen atoms results in a row-wise antiferromagnetic technique to obtain microscopic chemical and physical
configuration of magnetic moments on the (001) sur- information about molecules. An important applica-
face. Hence the magnetic surface unit cell is twice as tion of NMR for medical diagnostics of the inside of
large as the chemical surface unit cell. Figure 24.30a humans is magnetic resonance imaging (MRI). This
shows the atomic-scale contrast due to a pure chemical tomographic imaging technique uses the NMR signal
interaction. Maxima and minima correspond to the oxy- from thin slices through the body to reconstruct its
gen and nickel atoms, respectively. Their arrangement three-dimensional structure. Currently, at least 1012 nu-
represents the (1 × 1) surface unit cell. Figure 24.30b clear spins must be present in a given volume to obtain
exhibits an additional modulation on chemically and a significant MRI signal. The ultimate goal of MRFM
structurally equivalent rows of nickel atoms (the min- is to combine aspects of force microscopy with MRI
ima). The structure corresponds to the (2 × 1) magnetic to achieve true 3-D imaging with atomic resolution and
surface unit cell. Since the spin-carrying nickel 3d states elemental selectivity.
are highly localized, the magnetic contrast only be- The experimental setup is sketched in Fig. 24.31.
comes significant at very small tip–sample distances. An oscillating cantilever with a magnetic tip at its end
More recently Schmidt et al. [24.261] were able to per- points toward the surface. The spherical resonant slice
form MExFM with much better signal-to-noise ratio on within the sample represents those points where the
an itinerant metallic system: the antiferromagnetic iron stray field from the tip and the external field match the
monolayer on W(001). Density-functional theory per- condition for magnetic resonance. The cyclic spin flip
formed with a realistic tip model indicated significant causes a slight shift of the cantilever frequency due to
relaxations of tip and sample atoms during imaging. the magnetic force exerted by the spin on the tip. Since
700 Part C Scanning-Probe Microscopy
the forces are extremely small, very low temperatures ferromagnetic resonance [24.266] experiments of spin
are required. ensembles with micrometer resolution. Moreover, in
To date, no individual nuclear spins have been SiO2 the magnetic moment of a single electron, which
detected by MRFM. However, the design of ultrasen- is three orders of magnitude larger than the nuclear
sitive cantilevers has made considerable progress, and magnetic moment, could be detected [24.262] using
the detection of forces below 1 × 10−18 N has been the setup shown in Fig. 24.31 at 1.6 K. This major
achieved [24.202]. Therefore, it has become possible breakthrough demonstrates the capability of force mi-
to perform nuclear magnetic resonance [24.265], and croscopy to detect single spins.
References
24.1 G. Binnig, H. Rohrer, C. Gerber, E. Weibel: Surface 24.13 H. Hug, B. Stiefel, P.J.A. van Schendel, A. Moser,
studies by scanning tunneling microscopy, Phys. S. Martin, H.-J. Güntherodt: A low temperature ul-
Rev. Lett. 49, 57–61 (1982) trahigh vacuum scanning force microscope, Rev.
24.2 R. Wiesendanger: Scanning Probe Microscopy and Sci. Instrum. 70, 3627–3640 (1999)
Spectroscopy (Cambridge Univ. Press, Cambridge 24.14 S. Behler, M.K. Rose, D.F. Ogletree, F. Salmeron:
1994) Method to characterize the vibrational response of
24.3 M. Tinkham: Introduction to Superconductivity a beetle type scanning tunneling microscope, Rev.
(McGraw–Hill, New York 1996) Sci. Instrum. 68, 124–128 (1997)
24.4 J. Kondo: Theory of dilute magnetic alloys, Solid 24.15 C. Wittneven, R. Dombrowski, S.H. Pan, R. Wiesen-
Part C 24
24.24 J. Kliewer, R. Berndt, E.V. Chulkov, V.M. Silkin, 24.40 J.A. Stroscio, F. Tavazza, J.N. Crain, R.J. Celotta,
P.M. Echenique, S. Crampin: Dimensionality ef- A.M. Chaka: Electronically induced atom motion
fects in the lifetime of surface states, Science 288, in engineered CoCun nanostructures, Science 313,
1399–1401 (2000) 948–951 (2006)
24.25 M.F. Crommie, C.P. Lutz, D.M. Eigler: Imaging 24.41 M. Lastapis, M. Martin, D. Riedel, L. Hellner,
standing waves in a two-dimensional electron gas, G. Comtet, G. Dujardin: Picometer-scale electronic
Nature 363, 524–527 (1993) control of molecular dynamics inside a single mol-
24.26 B.C. Stipe, M.A. Rezaei, W. Ho: Single-molecule ecule, Science 308, 1000–1003 (2005)
vibrational spectroscopy and microscopy, Science 24.42 M. Ternes, C.P. Lutz, C.F. Hirjibehedin, F.J. Giessibl,
280, 1732–1735 (1998) A.J. Heinrich: The force needed to move an atom
24.27 A.J. Heinrich, J.A. Gupta, C.P. Lutz, D. Eigler: Single- on a surface, Science 319, 1066–1069 (2008)
atom spin-flip spectroscopy, Science 306, 466–469 24.43 J.I. Pascual, N. Lorente, Z. Song, H. Conrad,
(2004) H.P. Rust: Selectivity in vibrationally mediated
24.28 H. Gawronski, M. Mehlhorn, K. Morgenstern: Imag- single-molecule chemistry, Nature 423, 525–528
ing phonon excitation with atomic resolution, (2003)
Science 319, 930–933 (2008) 24.44 T.C. Shen, C. Wang, G.C. Abeln, J.R. Tucker,
24.29 C.W.J. Beenakker, H. van Houten: Quantum trans- J.W. Lyding, P. Avouris, R.E. Walkup: Atomic-
port in semiconductor nanostructures, Solid State scale desorption through electronic and vibrational
Phys. 44, 1–228 (1991) excitation mechanisms, Science 268, 1590–1592
24.30 K.M. Lang, V. Madhavan, J.E. Hoffman, E.W. Hud- (1995)
son, H. Eisaki, S. Uchida, J.C. Davis: Imaging the 24.45 T. Komeda, Y. Kim, M. Kawai, B.N.J. Persson,
granular structure of high-T c superconductivity in H. Ueba: Lateral hopping of molecules induced
Part C 24
underdoped Bi2 Sr2 CaCu2 O8+δ , Nature 415, 412–416 by excitations of internal vibration mode, Science
(2002) 295, 2055–2058 (2002)
24.31 J. Lee, K. Fujita, K. McElroy, J.A. Slezak, M. Wang, 24.46 Y.W. Mo: Reversible rotation of antimony dimers on
Y. Aiura, H. Bando, M. Ishikado, T. Masui, J.X. Zhu, the silicon(001) surface with a scanning tunneling
A.V. Balatsky, H. Eisaki, S. Uchida, J.C. Davis: microscope, Science 261, 886–888 (1993)
Interplay of electron-lattice interactions and su- 24.47 B.C. Stipe, M.A. Rezaei, W. Ho: Inducing and view-
perconductivity in Bi2 Sr2 CaCu2 O8+δ , Nature 442, ing the rotational motion of a single molecule,
546–550 (2006) Science 279, 1907–1909 (1998)
24.32 K.K. Gomes, A.N. Pasupathy, A. Pushp, S. Ono, 24.48 P. Liljeroth, J. Repp, G. Meyer: Current-induced hy-
Y. Ando, A. Yazdani: Visualizing pair formation drogen tautomerization and conductance switch-
on the atomic scale in the high-T c super- ing of napthalocyanine molecules, Science 317,
conductor Bi2 Sr2 CaCu2 O8+δ , Nature 447, 569–572 1203–1206 (2007)
(2007) 24.49 P. Maksymovych, D.B. Dougherty, X.-Y. Zhu,
24.33 R.S. Becker, J.A. Golovchenko, B.S. Swartzentruber: J.T. Yates Jr.: Nonlocal dissociative chemistry of
Atomic-scale surface modifications using a tunnel- adsorbed molecules induced by localized electron
ing microscope, Nature 325, 419–421 (1987) injection into metal surfaces, Phys. Rev. Lett. 99,
24.34 P.G. Piva, G.A. DiLabio, J.L. Pitters, J. Zikovsky, 016101–1–016101–4 (2007)
M. Rezeq, S. Dogel, W.A. Hofer, R.A. Wolkow: 24.50 G.V. Nazin, X.H. Qiu, W. Ho: Visualization and
Field regulation of single-molecule conductivity spectroscopy of a metal-molecule-metal bridge,
by a charged surface atom, Nature 435, 658–661 Science 302, 77–81 (2003)
(2005) 24.51 J. Repp, G. Meyer, S. Paavilainen, F.E. Olsson,
24.35 J.A. Stroscio, D.M. Eigler: Atomic and molecular M. Persson: Imaging bond formation between
manipulation with the scanning tunneling micro- a gold atom and pentacene on an insulating sur-
scope, Science 254, 1319–1326 (1991) face, Science 312, 1196–1199 (2006)
24.36 A.J. Heinrich, J.A. Gupta, C.P. Lutz, D. Eigler: Mol- 24.52 S. Katano, Y. Kim, M. Hori, M. Trenary, M. Kawai:
ecule cascades, Science 298, 1381–1387 (2002) Reversible control of hydrogenation of a single
24.37 L. Bartels, G. Meyer, K.H. Rieder: Basic steps of lat- molecule, Science 316, 1883–1886 (2007)
eral manipulation of single atoms and diatomic 24.53 R. Yamachika, M. Grobis, A. Wachowiak, M.F. Crom-
clusters with a scanning tunneling microscope, mie: Controlled atomic doping of a single C60
Phys. Rev. Lett. 79, 697–700 (1997) molecule, Science 304, 281–284 (2004)
24.38 J.A. Stroscio, R.J. Celotta: Controlling the dynam- 24.54 S.W. Hla, L. Bartels, G. Meyer, K.H. Rieder: Inducing
ics of a single atom in lateral atom manipulation, all steps of a chemical reaction with the scanning
Science 306, 242–247 (2004) tunneling microscope tip: Towards single molecule
24.39 J.J. Schulz, R. Koch, K.H. Rieder: New mechanism engineering, Phys. Rev. Lett. 85, 2777–2780 (2000)
for single atom manipulation, Phys. Rev. Lett. 84, 24.55 Y. Kim, T. Komeda, M. Kawai: Single-molecule
4597–4600 (2000) reaction and characterization by vibrational ex-
702 Part C Scanning-Probe Microscopy
citation, Phys. Rev. Lett. 89, 126104–1–126104–4 24.70 P. Johansson, R. Monreal, P. Apell: Theory for light
(2002) emission from a scanning tunneling microscope,
24.56 M. Berthe, R. Stiufiuc, B. Grandidier, D. Deresmes, Phys. Rev. B 42, 9210–9213 (1990)
C. Delerue, D. Stievenard: Probing the carrier cap- 24.71 J. Aizpurua, G. Hoffmann, S.P. Apell, R. Berndt:
ture rate of a single quantum level, Science 319, Electromagnetic coupling on an atomic scale, Phys.
436–438 (2008) Rev. Lett. 89, 156803–1–156803–4 (2002)
24.57 N. Neel, J. Kröger, L. Limot, K. Palotas, W.A. Hofer, 24.72 G. Hoffmann, J. Kliewer, R. Berndt: Lumines-
R. Berndt: Conductance and Kondo effect in a con- cence from metallic quantum wells in a scanning
trolled single-atom contact, Phys. Rev. Lett. 98, tunneling microscope, Phys. Rev. Lett. 78, 176803–
016801–1–016801–4 (2007) 1–176803–4 (2001)
24.58 F. Rosei, M. Schunack, P. Jiang, A. Gour- 24.73 X.H. Qiu, G.V. Nazin, W. Ho: Vibrationally resolved
don, E. Laegsgaard, I. Stensgaard, C. Joachim, fluorescence excited with submolecular precission,
F. Besenbacher: Organic molecules acting as tem- Science 299, 542–546 (2003)
plates on metal surfaces, Science 296, 328–331 24.74 E. Cavar, M.C. Blüm, M. Pivetta, F. Patthey,
(2002) M. Chergui, W.D. Schneider: Fluorescence and
24.59 E. Ganz, S.K. Theiss, I.S. Hwang, J. Golovchenko: phosphorescence from individual C60 molecules
Direct measurement of diffusion by hot tunnel- excited by local electron tunneling, Phys. Rev. Lett.
ing microscopy: Activations energy, anisotropy, 95, 196102–1–196102–4 (2005)
and long jumps, Phys. Rev. Lett. 68, 1567–1570 24.75 S.W. Wu, N. Ogawa, W. Ho: Atomic-scale coupling
(1992) of photons to single-molecule junctions, Science
24.60 M. Schunack, T.R. Linderoth, F. Rosei, E. Laegs- 312, 1362–1365 (2006)
gaard, I. Stensgaard, F. Besenbacher: Long jumps 24.76 A. Downes, M.E. Welland: Photon emission from
Part C 24
in the surface diffusion of large molecules, Phys. Si(111)-(7 × 7) induced by scanning tunneling mi-
Rev. Lett. 88, 156102–1–156102–4 (2002) croscopy: atomic scale and material contrast, Phys.
24.61 K.L. Wong, G. Pawin, K.Y. Kwon, X. Lin, T. Jiao, Rev. Lett. 81, 1857–1860 (1998)
U. Solanki, R.H.J. Fawcett, L. Bartels, S. Stolbov, 24.77 M. Kemerink, K. Sauthoff, P.M. Koenraad, J.W. Ge-
T.S. Rahman: A molecule carrier, Science 315, 1391– ritsen, H. van Kempen, J.H. Wolter: Opti-
1393 (2007) cal detection of ballistic electrons injected by
24.62 L.J. Lauhon, W. Ho: Direct observation of the quan- a scanning–tunneling microscope, Phys. Rev. Lett.
tum tunneling of single hydrogen atoms with 86, 2404–2407 (2001)
a scanning tunneling microscope, Phys. Rev. Lett. 24.78 J. Tersoff, D.R. Hamann: Theory and application
85, 4566–4569 (2000) for the scanning tunneling microscope, Phys. Rev.
24.63 F. Ronci, S. Colonna, A. Cricenti, G. LeLay: Evi- Lett. 50, 1998–2001 (1983)
dence of Sn adatoms quantum tunneling at the 24.79 C.J. Chen: Introduction to Scanning Tunneling Mi-
α-Sn/Si(111) surface, Phys. Rev. Lett. 99, 166103–1– croscopy (Oxford Univ. Press, Oxford 1993)
166103–4 (2007) 24.80 J. Winterlin, J. Wiechers, H. Brune, T. Gritsch,
24.64 N. Kitamura, M. Lagally, M.B. Webb: Real-time ob- H. Hofer, R.J. Behm: Atomic-resolution imaging of
servation of vacancy diffusion on Si(001)-(2 × 1) by close-packed metal surfaces by scanning tunneling
scanning tunneling microscopy, Phys. Rev. Lett. 71, microscopy, Phys. Rev. Lett. 62, 59–62 (1989)
2082–2085 (1993) 24.81 J.A. Stroscio, R.M. Feenstra, A.P. Fein: Electronic
24.65 M. Morgenstern, T. Michely, G. Comsa: Onset of structure of the Si(111) 2 × 1 surface by scanning-
interstitial diffusion determined by scanning tun- tunneling microscopy, Phys. Rev. Lett. 57, 2579–
neling microscopy, Phys. Rev. Lett. 79, 1305–1308 2582 (1986)
(1997) 24.82 A.L. Vázquez de Parga, O.S. Hernan, R. Miranda,
24.66 K. Morgenstern, G. Rosenfeld, B. Poelsema, A. Levy Yeyati, N. Mingo, A. Martín-Rodero, F. Flo-
G. Comsa: Brownian motion of vacancy islands on res: Electron resonances in sharp tips and their
Ag(111), Phys. Rev. Lett. 74, 2058–2061 (1995) role in tunneling spectroscopy, Phys. Rev. Lett. 80,
24.67 L. Bartels, F. Wang, D. Möller, E. Knoesel, T.F. Heinz: 357–360 (1998)
Real-space observation of molecular motion in- 24.83 S.H. Pan, E.W. Hudson, J.C. Davis: Vacuum tun-
duced by femtosecond laser pulses, Science 305, neling of superconducting quasiparticles from
648–651 (2004) atomically sharp scanning tunneling microscope
24.68 B. Reihl, J.H. Coombs, J.K. Gimzewski: Local in- tips, Appl. Phys. Lett. 73, 2992–2994 (1998)
verse photoemission with the scanning tunneling 24.84 J.T. Li, W.D. Schneider, R. Berndt, O.R. Bryant,
microscope, Surf. Sci. 211/212, 156–164 (1989) S. Crampin: Surface-state lifetime measured by
24.69 R. Berndt, J.K. Gimzewski, P. Johansson: Inelas- scanning tunneling spectroscopy, Phys. Rev. Lett.
tic tunneling excitation of tip-induced plasmon 81, 4464–4467 (1998)
modes on noble-metal surfaces, Phys. Rev. Lett. 24.85 L. Bürgi, O. Jeandupeux, H. Brune, K. Kern: Probing
67, 3796–3799 (1991) hot-electron dynamics with a cold scanning tun-
Low-Temperature Scanning Probe Microscopy References 703
neling microscope, Phys. Rev. Lett. 82, 4516–4519 24.99 C.F. Hirjibehedin, C.P. Lutz, A.J. Heinrich: Spin cou-
(1999) pling in engineered atomic structures, Science 312,
24.86 J.W.G. Wildoer, C.J.P.M. Harmans, H. van Kem- 1021–1024 (2006)
pen: Observation of Landau levels at the InAs(110) 24.100 C.F. Hirjibehedin, C.Y. Lin, A.F. Otte, M. Ternes,
surface by scanning tunneling spectroscopy, Phys. C.P. Lutz, B.A. Jones, A.J. Heinrich: Large magnetic
Rev. B 55, R16013–R16016 (1997) anisotropy of a single atomic spin embedded in
24.87 M. Morgenstern, V. Gudmundsson, C. Wittneven, a surface molecular network, Science 317, 1199–1203
R. Dombrowski, R. Wiesendanger: Nonlocality of (2007)
the exchange interaction probed by scanning 24.101 U. Kemiktarak, T. Ndukum, K.C. Schwab, K.L. Ekinci:
tunneling spectroscopy, Phys. Rev. B 63, 201301(R)- Radio-frequency scanning tunneling microscopy,
1–201301(R)-4 (2001) Nature 450, 85–89 (2007)
24.88 F.E. Olsson, M. Persson, A.G. Borisov, J.P. Gauyacq, 24.102 A. Hewson: From the Kondo Effect to Heavy
J. Lagoute, S. Fölsch: Localization of the Cu(111) sur- Fermions (Cambridge Univ. Press, Cambridge 1993)
face state by single Cu adatoms, Phys. Rev. Lett. 24.103 V. Madhavan, W. Chen, T. Jamneala, M.F. Crom-
93, 206803–1–206803–4 (2004) mie, N.S. Wingreen: Tunneling into a sin-
24.89 L. Limot, E. Pehlke, J. Kröger, R. Berndt: Surface- gle magnetic atom: Spectroscopic evidence of
state localization at adatoms, Phys. Rev. Lett. 94, the Kondo resonance, Science 280, 567–569
036805–1–036805–4 (2005) (1998)
24.90 N. Nilius, T.M. Wallis, M. Persson, W. Ho: Distance 24.104 J. Li, W.D. Schneider, R. Berndt, B. Delley: Kondo
dependence of the interaction between single scattering observed at a single magnetic impurity,
atoms: Gold dimers on NiAl(110), Phys. Rev. Lett. Phys. Rev. Lett. 80, 2893–2896 (1998)
90, 196103–1–196103–4 (2003) 24.105 T.W. Odom, J.L. Huang, C.L. Cheung, C.M. Lieber:
Part C 24
24.91 H.J. Lee, W. Ho, M. Persson: Spin splitting of s and Magnetic clusters on single-walled carbon nano-
p states in single atoms and magnetic coupling in tubes: the Kondo effect in a one-dimensional host,
dimers on a surface, Phys. Rev. Lett. 92, 186802– Science 290, 1549–1552 (2000)
1–186802–4 (2004) 24.106 M. Ouyang, J.L. Huang, C.L. Cheung, C.M. Lieber:
24.92 M.V. Grishin, F.I. Dalidchik, S.A. Kovalevskii, Energy gaps in metallic single-walled carbon
N.N. Kolchenko, B.R. Shub: Isotope effect in the nanotubes, Science 292, 702–705 (2001)
vibrational spectra of water measured in experi- 24.107 U. Fano: Effects of configuration interaction on
ments with a scanning tunneling microscope, JETP intensities and phase shifts, Phys. Rev. 124, 1866–
Lett. 66, 37–40 (1997) 1878 (1961)
24.93 Y. Sainoo, Y. Kim, T. Okawa, T. Komeda, 24.108 P. Wahl, L. Diekhöner, M.A. Schneider, L. Vitali,
H. Shigekawa, M. Kawai: Excitation of molecular G. Wittich, K. Kern: Kondo temperature of magnetic
vibrational modes with inelastic scanning tun- impurities at surfaces, Phys. Rev. Lett. 93, 176603–
neling microscopy: Examination through action 1–176603–4 (2004)
spectra of cis-2-butene on Pd(110), Phys. Rev. Lett. 24.109 Y.S. Fu, S.H. Ji, X. Chen, X.C. Ma, R. Wu, C.C. Wang,
95, 246102–1–246102–4 (2005) W.H. Duan, X.H. Qiu, B. Sun, P. Zhang, J.F. Jia,
24.94 X.H. Qiu, G.V. Nazin, W. Ho: Vibronic states in single Q.K. Xue: Manipulating the Kondo resonance
molecule electron transport, Phys. Rev. Lett. 92, through quantum size effects, Phys. Rev. Lett. 99,
206102–1–206102–4 (2004) 256601–1–256601–4 (2007)
24.95 L. Vitali, M. Burghard, M.A. Schneider, L. Liu, 24.110 J. Henzl, K. Morgenstern: Contribution of the sur-
S.Y. Wu, C.S. Jayanthi, K. Kern: Phonon spectro- face state to the observation of the surface Kondo
microscopy of carbon nanostructures with atomic resonance, Phys. Rev. Lett. 98, 266601–1–266601–
resolution, Phys. Rev. Lett. 93, 136103–1–136103–4 4 (2007)
(2004) 24.111 P. Wahl, P. Simon, L. Diekhöner, V.S. Stepanyuk,
24.96 B.J. LeRoy, S.G. Lemay, J. Kong, C. Dekker: Elec- P. Bruno, M.A. Schneider, K. Kern: Exchange in-
trical generation and absorption of phonons teraction between single magnetic adatoms, Phys.
in carbon nanotubes, Nature 432, 371–374 Rev. Lett. 98, 056601–1–056601–4 (2007)
(2004) 24.112 T. Jamneala, V. Madhavan, M.F. Crommie: Kondo
24.97 L. Vitali, M.A. Schneider, K. Kern, L. Wirtz, A. Rubio: response of a single antiferromagnetic chromium
Phonon and plasmon excitation in inelastic scan- trimer, Phys. Rev. Lett. 87, 256804–1–256804–4
ning tunneling spectroscopy of graphite, Phys. Rev. (2001)
B 69, 121414–1–121414–4 (2004) 24.113 P. Wahl, L. Diekhöner, G. Wittich, L. Vitali,
24.98 T. Balashov, A.F. Takacz, W. Wulfhekel, J. Kirschner: M.A. Schneider, K. Kern: Kondo effect of molecu-
Magnon excitation with spin-polarized scanning lar complexes at surfaces: Ligand control of the
tunneling microscopy, Phys. Rev. Lett. 97, 187201– local spin coupling, Phys. Rev. Lett. 95, 166601–1–
1–187201–4 (2006) 166601–4 (2005)
704 Part C Scanning-Probe Microscopy
24.114 A. Zhao, Q. Li, L. Chen, H. Xiang, W. Wang, S. Pan, quantized energy levels in carbon nanotubes, Na-
B. Wang, X. Xiao, J. Yang, J.G. Hou, Q. Zhu: Con- ture 283, 52–55 (1999)
trolling the Kondo effect of an adsorbed magnetic 24.128 S.G. Lemay, J.W. Jannsen, M. van den Hout,
ion through its chemical bonding, Science 309, M. Mooij, M.J. Bronikowski, P.A. Willis, R.E. Smal-
1542–1544 (2005) ley, L.P. Kouwenhoven, C. Dekker: Two-dimen-
24.115 V. Iancu, A. Deshpande, S.W. Hla: Manipulation of sional imaging of electronic wavefunctions in
the Kondo effect via two-dimensional molecular carbon nanotubes, Nature 412, 617–620 (2001)
assembly, Phys. Rev. Lett. 97, 266603–1–266603–4 24.129 C.R. Moon, L.S. Matos, B.K. Foster, G. Zeltzer, W. Ko,
(2006) H.C. Manoharan: Quantum phase extraction in
24.116 L. Gao, W. Ji, Y.B. Hu, Z.H. Cheng, Z.T. Deng, Q. Liu, isospectral electronic nanostructures, Science 319,
N. Jiang, X. Lin, W. Guo, S.X. Du, W.A. Hofer, 782–787 (2008)
X.C. Xie, H.J. Gao: Site-specific Kondo effect at am- 24.130 X. Lu, M. Grobis, K.H. Khoo, S.G. Louie, M.F. Crom-
bient temperatures in iron-based molecules, Phys. mie: Spatially mapping the spectral density of a
Rev. Lett. 99, 106402–1–106402–4 (2007) single C60 molecule, Phys. Rev. Lett. 90, 096802–
24.117 H.C. Manoharan, C.P. Lutz, D.M. Eigler: Quantum 1–096802–4 (2003)
mirages formed by coherent projection of elec- 24.131 A.M. Yakunin, A.Y. Silov, P.M. Koenraad, J.H. Wolter,
tronic structure, Nature 403, 512–515 (2000) W. van Roy, J. de Boeck, J.M. Tang, M.E. Flatte: Spa-
24.118 H.A. Mizes, J.S. Foster: Long-range electronic tial structure of an individual Mn acceptor in GaAs,
perturbations caused by defects using scanning Phys. Rev. Lett. 92, 216806–1–216806–4 (200405)
tunneling microscopy, Science 244, 559–562 (1989) 24.132 F. Marczinowski, J. Wiebe, J.M. Tang, M.E. Flatte,
24.119 P.T. Sprunger, L. Petersen, E.W. Plummer, E. Laegs- F. Meier, M. Morgenstern, R. Wiesendanger: Lo-
gaard, F. Besenbacher: Giant Friedel oscillations cal electronic structure near Mn acceptors in InAs:
Part C 24
on beryllium(0001) surface, Science 275, 1764–1767 Surface-induced symmetry breaking and coupling
(1997) to host states, Phys. Rev. Lett. 99, 157202–1–
24.120 P. Hofmann, B.G. Briner, M. Doering, H.P. Rust, 157202–4 (2007)
E.W. Plummer, A.M. Bradshaw: Anisotropic two- 24.133 N. Nilius, T.M. Wallis, W. Ho: Development of
dimensional Friedel oscillations, Phys. Rev. Lett. one-dimensional band structure in artificial gold
79, 265–268 (1997) chains, Science 297, 1853–1856 (2002)
24.121 E.J. Heller, M.F. Crommie, C.P. Lutz, D.M. Eig- 24.134 J.N. Crain, D.T. Pierce: End states in one-
ler: Scattering and adsorption of surface electron dimensional atom chains, Science 307, 703–706
waves in quantum corrals, Nature 369, 464–466 (2005)
(1994) 24.135 D. Kitchen, A. Richardella, J.M. Tang, M.E. Flatte,
24.122 M.C.M.M. van der Wielen, A.J.A. van Roij, H. van A. Yazdani: Atom-by-atom substitution of Mn in
Kempen: Direct observation of Friedel oscillations GaAs and visualization of their hole-mediated in-
around incorporated SiGa dopants in GaAs by low- teractions, Nature 442, 436–439 (2006)
temperature scanning tunneling microscopy, Phys. 24.136 C. Wittneven, R. Dombrowski, M. Morgenstern,
Rev. Lett. 76, 1075–1078 (1996) R. Wiesendanger: Scattering states of ionized
24.123 O. Millo, D. Katz, Y.W. Cao, U. Banin: Imaging and dopants probed by low temperature scanning tun-
spectroscopy of artificial-atom states in core/shell neling spectroscopy, Phys. Rev. Lett. 81, 5616–5619
nanocrystal quantum dots, Phys. Rev. Lett. 86, (1998)
5751–5754 (2001) 24.137 D. Haude, M. Morgenstern, I. Meinel, R. Wiesen-
24.124 T. Maltezopoulos, A. Bolz, C. Meyer, C. Heyn, danger: Local density of states of a three-
W. Hansen, M. Morgenstern, R. Wiesendanger: dimensional conductor in the extreme quantum
Wave-function mapping of InAs qunatum dots by limit, Phys. Rev. Lett. 86, 1582–1585 (2001)
scanning tunneling spectroscopy, Phys. Rev. Lett. 24.138 R. Joynt, R.E. Prange: Conditions for the quantum
91, 196804–1–196804–4 (2003) Hall effect, Phys. Rev. B 29, 3303–3317 (1984)
24.125 R. Temirov, S. Soubatch, A. Luican, F.S. Tautz: Free- 24.139 M. Morgenstern, J. Klijn, R. Wiesendanger: Real
electron-like dispersion in an organic monolayer space observation of drift states in a two-
film on a metal substrate, Nature 444, 350–353 dimensional electron system at high magnetic
(2006) fields, Phys. Rev. Lett. 90, 056804–1–056804–4
24.126 K. Suzuki, K. Kanisawa, C. Janer, S. Perraud, (2003)
K. Takashina, T. Fujisawa, Y. Hirayama: Spatial 24.140 M. Morgenstern, J. Klijn, C. Meyer, M. Getzlaff,
imaging of two-dimensional electronic states in R. Adelung, R.A. Römer, K. Rossnagel, L. Kipp,
semiconductor quantum wells, Phys. Rev. Lett. 98, M. Skibowski, R. Wiesendanger: Direct comparison
136802–1–136802–4 (2007) between potential landscape and local density of
24.127 L.C. Venema, J.W.G. Wildoer, J.W. Janssen, states in a disordered two-dimensional electron
S.J. Tans, L.J.T. Tuinstra, L.P. Kouwenhoven, system, Phys. Rev. Lett. 89, 136806–1–136806–4
C. Dekker: Imaging electron wave functions of (2002)
Low-Temperature Scanning Probe Microscopy References 705
24.141 E. Abrahams, P.W. Anderson, D.C. Licciardello, 24.154 K. Swamy, A. Menzel, R. Beer, E. Bertel: Charge-
T.V. Ramakrishnan: Scaling theory of local- density waves in self-assembled halogen-bridged
ization: absence of quantum diffusion in metal chains, Phys. Rev. Lett. 86, 1299–1302 (2001)
two dimensions, Phys. Rev. Lett. 42, 673–676 24.155 J.J. Kim, W. Yamaguchi, T. Hasegawa, K. Kitazawa:
(1979) Observation of Mott localization gap using low
24.142 C. Meyer, J. Klijn, M. Morgenstern, R. Wiesendan- temperature scanning tunneling spectroscopy in
ger: Direct measurement of the local density of commensurate 1 T-TaSe2 , Phys. Rev. Lett. 73, 2103–
states of a disordered one-dimensional conductor, 2106 (1994)
Phys. Rev. Lett. 91, 076803–1–076803–4 (2003) 24.156 A. Wachowiak, R. Yamachika, K.H. Khoo, Y. Wang,
24.143 N. Oncel, A. van Houselt, J. Huijben, A.S. Hallbäck, M. Grobis, D.H. Lee, S.G. Louie, M.F. Crommie: Visu-
O. Gurlu, H.J.W. Zandvliet, B. Poelsema: Quantum alization of the molecular Jahn–Teller effect in an
confinement between self-organized Pt nanowires insulating K4 C60 monolayer, Science 310, 468–470
on Ge(001), Phys. Rev. Lett. 95, 116801–1–116801–4 (2005)
(2005) 24.157 J. Bardeen, L.N. Cooper, J.R. Schrieffer: Theory of
24.144 J. Lee, S. Eggert, H. Kim, S.J. Kahng, H. Shinohara, superconductivity, Phys. Rev. 108, 1175–1204 (1957)
Y. Kuk: Real space imaging of one-dimensional 24.158 A. Yazdani, B.A. Jones, C.P. Lutz, M.F. Crommie,
standing waves: Direct evidence for a Luttinger D.M. Eigler: Probing the local effects of magnetic
liquid, Phys. Rev. Lett. 93, 166403–1–166403–4 impurities on superconductivity, Science 275, 1767–
(2004) 1770 (1997)
24.145 R.E. Peierls: Quantum Theory of Solids (Clarendon, 24.159 S.H. Tessmer, M.B. Tarlie, D.J. van Harlingen,
Oxford 1955) D.L. Maslov, P.M. Goldbart: Probing the super-
24.146 C.G. Slough, W.W. McNairy, R.V. Coleman, B. Drake, conducting proximity effect in NbSe2 by scanning
Part C 24
P.K. Hansma: Charge-density waves studied with tunneling micrsocopy, Phys. Rev. Lett. 77, 924–927
the use of a scanning tunneling microscope, Phys. (1996)
Rev. B 34, 994–1005 (1986) 24.160 K. Inoue, H. Takayanagi: Local tunneling spec-
24.147 X.L. Wu, C.M. Lieber: Hexagonal domain-like troscopy of Nb/InAs/Nb superconducting proximity
charge-density wave of TaS2 determined by scan- system with a scanning tunneling microscope,
ning tunneling microscopy, Science 243, 1703–1705 Phys. Rev. B 43, 6214–6215 (1991)
(1989) 24.161 H.F. Hess, R.B. Robinson, R.C. Dynes, J.M. Valles,
24.148 T. Nishiguchi, M. Kageshima, N. Ara-Kato, J.V. Waszczak: Scanning-tunneling-microscope
A. Kawazu: Behaviour of charge density waves in observation of the Abrikosov flux lattice and the
a one-dimensional organic conductor visualized density of states near and inside a fluxoid, Phys.
by scanning tunneling microscopy, Phys. Rev. Lett. Rev. Lett. 62, 214–217 (1989)
81, 3187–3190 (1998) 24.162 H.F. Hess, R.B. Robinson, J.V. Waszczak: Vortex-
24.149 X.L. Wu, C.M. Lieber: Direct observation of growth core structure observed with a scanning tunneling
and melting of the hexagonal-domain charge- microscope, Phys. Rev. Lett. 64, 2711–2714 (1990)
density-wave phase in 1 T-TaS2 by scanning 24.163 N. Hayashi, M. Ichioka, K. Machida: Star-shaped
tunneling microscopy, Phys. Rev. Lett. 64, 1150– local density of states around vortices in a type-
1153 (1990) II superconductor, Phys. Rev. Lett. 77, 4074–4077
24.150 J.M. Carpinelli, H.H. Weitering, E.W. Plum- (1996)
mer, R. Stumpf: Direct observation of a sur- 24.164 H. Sakata, M. Oosawa, K. Matsuba, N. Nishida:
face charge density wave, Nature 381, 398–400 Imaging of vortex lattice transition in YNi2 B2 C by
(1996) scanning tunneling spectroscopy, Phys. Rev. Lett.
24.151 H.H. Weitering, J.M. Carpinelli, A.V. Melechenko, 84, 1583–1586 (2000)
J. Zhang, M. Bartkowiak, E.W. Plummer: Defect- 24.165 S. Behler, S.H. Pan, P. Jess, A. Baratoff, H.-J. Gün-
mediated condensation of a charge density wave, therodt, F. Levy, G. Wirth, J. Wiesner: Vortex
Science 285, 2107–2110 (1999) pinning in ion-irradiated NbSe2 studied by scan-
24.152 S. Modesti, L. Petaccia, G. Ceballos, I. Vobornik, ning tunneling microscopy, Phys. Rev. Lett. 72,
G. Panaccione, G. Rossi, L. Ottaviano, R. Lar- 1750–1753 (1994)
ciprete, S. Lizzit, A.√ Goldoni:
√ Insulating ground 24.166 R. Berthe, U. Hartmann, C. Heiden: Influence of
state of Sn/Si(111)-( 3× 3)R30◦ , Phys. Rev. Lett. a transport current on the Abrikosov flux lat-
98, 126401–1–126401–4 (2007) tice observed with a low-temperature scanning
24.153 H.W. Yeom, S. Takeda, E. Rotenberg, I. Matsuda, tunneling microscope, Ultramicroscopy 42-44,
K. Horikoshi, J. Schäfer, C.M. Lee, S.D. Kevan, 696–698 (1992)
T. Ohta, T. Nagao, S. Hasegawa: Instability and 24.167 N.C. Yeh, C.T. Chen, G. Hammerl, J. Mannhart,
charge density wave of metallic quantum chains A. Schmehl, C.W. Schneider, R.R. Schulz, S. Tajima,
on a silicon surface, Phys. Rev. Lett. 82, 4898–4901 K. Yoshida, D. Garrigus, M. Strasik: Evidence of
(1999) doping-dependent pairing symmetry in cuprate
706 Part C Scanning-Probe Microscopy
superconductors, Phys. Rev. Lett. 87, 087003–1– intrinsic bond-centered electronic glass with uni-
087003–4 (2001) directional domains in underdoped cuprates,
24.168 K. McElroy, R.W. Simmonds, J.E. Hoffman, D.H. Lee, Science 315, 1380–1385 (2007)
J. Orenstein, H. Eisaki, S. Uchida, J.C. Davis: 24.181 M. Crespo, H. Suderow, S. Vieira, S. Bud’ko,
Relating atomic-scale electronic phenomena to P.C. Canfield: Local superconducting density of
wave-like quasiparticle states in supercon- states of ErNi2 B2 C, Phys. Rev. Lett. 96, 027003–1–
ducting Bi2 Sr2 CaCu2 O8+δ , Nature 422, 592–596 027003–4 (2006)
(2003) 24.182 H. Suderow, S. Vieira, J.D. Strand, S. Bud’ko,
24.169 K. McElroy, J. Lee, J.A. Slezak, D.H. Lee, H. Eisaki, P.C. Canfield: Very-low-temperature tunneling
S. Uchida, J.C. Davis: Atomic-scale sources and spectroscopy in the heavy-fermion superconductor
mechanism of nanoscale electronic disorder in PrOs4 Sb12 , Phys. Rev. B 69, 060504–1–060504–4
Bi2 Sr2 CaCu2 O8+δ , Science 309, 1048–1052 (2005) (2004)
24.170 S.H. Pan, E.W. Hudson, K.M. Lang, H. Eisaki, 24.183 O. Naaman, W. Teizer, R.C. Dynes: Fluctuation
S. Uchida, J.C. Davis: Imaging the effects of dominated Josephson tunneling with a scanning
individual zinc impurity atoms on superconduc- tunneling microscope, Phys. Rev. Lett. 87, 097004–
tivity in Bi2 Sr2 CaCu2 O8+δ , Nature 403, 746–750 1–097004–4 (2001)
(2000) 24.184 M. Fäth, S. Freisem, A.A. Menovsky, Y. Tomioka,
24.171 A. Polkovnikov, S. Sachdev, M. Vojta: Impurity in J. Aaarts, J.A. Mydosh: Spatially inhomogeneous
a d-wave superconductor: Kondo effect and STM metal–insulator transition in doped manganites,
spectra, Phys. Rev. Lett. 86, 296–299 (2001) Science 285, 1540–1542 (1999)
24.172 A.N. Pasupathy, A. Pushp, K.K. Gomes, C.V. Parker, 24.185 C. Renner, G. Aeppli, B.G. Kim, Y.A. Soh,
J. Wen, Z. Xu, G. Gu, S. Ono, Y. Ando, A. Yaz- S.W. Cheong: Atomic-scale images of charge or-
Part C 24
dani: Electronic origin of the inhomogeneous dering in a mixed-valence manganite, Nature 416,
pairing interaction in the high-T c superconductor 518–521 (2000)
Bi2 Sr2 CaCu2 O8+δ , Science 320, 196–201 (2008) 24.186 H.M. Ronnov, C. Renner, G. Aeppli, T. Kimura,
24.173 I. Maggio-Aprile, C. Renner, E. Erb, E. Walker, Y. Tokura: Polarons and confinement of electronic
Ø. Fischer: Direct vortex lattice imaging and tun- motion to two dimensions in a layered manganite,
neling spectroscopy of flux lines on YBa2 Cu3 O7−δ , Nature 440, 1025–1028 (2006)
Phys. Rev. Lett. 75, 2754–2757 (1995) 24.187 M. Bode, M. Getzlaff, R. Wiesendanger: Spin-
24.174 C. Renner, B. Revaz, K. Kadowaki, I. Maggio-Aprile, polarized vacuum tunneling into the exchange-
Ø. Fischer: Observation of the low temperature split surface state of Gd(0001), Phys. Rev. Lett. 81,
pseudogap in the vortex cores of Bi2 Sr2 CaCu2 O8+δ , 4256–4259 (1998)
Phys. Rev. Lett. 80, 3606–3609 (1998) 24.188 A. Kubetzka, M. Bode, O. Pietzsch, R. Wiesendan-
24.175 S.H. Pan, E.W. Hudson, A.K. Gupta, K.W. Ng, ger: Spin-polarized scanning tunneling microscopy
H. Eisaki, S. Uchida, J.C. Davis: STM studies with antiferromagnetic probe tips, Phys. Rev. Lett.
of the electronic structure of vortex cores in 88, 057201–1–057201–4 (2002)
Bi2 Sr2 CaCu2 O8+δ , Phys. Rev. Lett. 85, 1536–1539 24.189 O. Pietzsch, A. Kubetzka, M. Bode, R. Wiesen-
(2000) danger: Observation of magnetic hysteresis at the
24.176 D.P. Arovas, A.J. Berlinsky, C. Kallin, S.C. Zhang: Su- nanometer scale by spin-polarized scanning tun-
perconducting vortex with antiferromagnetic core, neling spectroscopy, Science 292, 2053–2056 (2001)
Phys. Rev. Lett. 79, 2871–2874 (1997) 24.190 S. Heinze, M. Bode, A. Kubetzka, O. Pietzsch, X. Xie,
24.177 J.E. Hoffmann, E.W. Hudson, K.M. Lang, V. Mad- S. Blügel, R. Wiesendanger: Real-space imaging
havan, H. Eisaki, S. Uchida, J.C. Davis: A four unit of two-dimensional antiferromagnetism on the
cell periodic pattern of quasi-particle states sur- atomic scale, Science 288, 1805–1808 (2000)
rounding vortex cores in Bi2 Sr2 CaCu2 O8+δ , Science 24.191 A. Kubetzka, P. Ferriani, M. Bode, S. Heinze,
295, 466–469 (2002) G. Bihlmayer, K. von Bergmann, O. Pietzsch,
24.178 M. Vershinin, S. Misra, S. Ono, Y. Abe, Y. Ando, S. Blügel, R. Wiesendanger: Revealing antiferro-
A. Yazdani: Local ordering in the pseudogap state magnetic order of the Fe monolayer on W(001):
of the high-T c superconductor Bi2 Sr2 CaCu2 O8+δ , Spin-polarized scanning tunneling microscopy and
Science 303, 1995–1998 (2004) first-principles calculations, Phys. Rev. Lett. 94,
24.179 T. Hanaguri, C. Lupien, Y. Kohsaka, D.H. Lee, 087204–1–087204–4 (2005)
M. Azuma, M. Takano, H. Takagi, J.C. Davis: A 24.192 A. Wachowiak, J. Wiebe, M. Bode, O. Pietzsch,
‘checkerborad’ electronic crystal state in lightly M. Morgenstern, R. Wiesendanger: Internal spin-
hole-doped Ca2−x Nax CuO2 Cl2 , Nature 430, 1001– structure of magnetic vortex cores observed by
1005 (2004) spin-polarized scanning tunneling microscopy,
24.180 Y. Kohsaka, C. Taylor, K. Fujita, A. Schmidt, Science 298, 577–580 (2002)
C. Lupien, T. Hanaguri, M. Azuma, M. Takano, 24.193 M. Bode, M. Heide, K. von Bergmann, P. Ferriani,
H. Eisaki, H. Takagi, S. Uchida, J.C. Davis: An S. Heinze, G. Bihlmeyer, A. Kubetzka, O. Pietzsch,
Low-Temperature Scanning Probe Microscopy References 707
S. Blügel, R. Wiesendanger: Chiral magnetic order 24.208 N. Suehira, Y. Sugawara, S. Morita: Artifact and fact
at surfaces driven by inversion asymmetry, Nature of Si(111)-(7 × 7) surface images observed with a low
447, 190–193 (2007) temperature noncontact atomic force microscope
24.194 K. von Bergmann, S. Heinze, M. Bode, E.Y. Vedme- (LT-NC-AFM), Jpn. J. Appl. Phys. 40, 292–294 (2001)
denko, G. Bihlmayer, S. Blügel, R. Wiesendanger: 24.209 R. Pérez, M.C. Payne, I. Štich, K. Terakura: Role
Observation of a complex nanoscale magnetic of covalent tip–surface interactions in noncontact
structure in a hexagonal Fe monolayer, Phys. Rev. atomic force microscopy on reactive surfaces, Phys.
Lett. 96, 167203–1–167203–4 (2006) Rev. Lett. 78, 678–681 (1997)
24.195 C.L. Gao, U. Schlickum, W. Wulfhekel, J. Kirschner: 24.210 S.H. Ke, T. Uda, R. Pérez, I. Štich, K. Terakura: First
Mapping the surface spin structure of large unit principles investigation of tip–surface interaction
cells: Reconstructed Mn films on Fe(001), Phys. Rev. on GaAs(110): Implication for atomic force and tun-
Lett. 98, 107203–1–107203–4 (2007) neling microscopies, Phys. Rev. B 60, 11631–11638
24.196 Y. Yayon, V.W. Brar, L. Senapati, S.C. Erwin, (1999)
M.F. Crommie: Observing spin polarization of in- 24.211 J. Tobik, I. Štich, R. Pérez, K. Terakura: Simulation of
dividual magnetic adatoms, Phys. Rev. Lett. 99, tip–surface interactions in atomic force microscopy
067202–1–067202–4 (2007) of an InP(110) surface with a Si tip, Phys. Rev. B 60,
24.197 M. Bode, O. Pietzsch, A. Kubetzka, R. Wiesendan- 11639–11644 (1999)
ger: Shape-dependent thermal switching behavior 24.212 A. Schwarz, W. Allers, U.D. Schwarz, R. Wiesen-
of superparamagnetic nanoislands, Phys. Rev. Lett. danger: Simultaneous imaging of the In and As
92, 067201–1–067201–4 (2004) sublattice on InAs(110)-(1 × 1) with dynamic scan-
24.198 S. Krause, L. Berbil-Bautista, G. Herzog, M. Bode, ning force microscopy, Appl. Surf. Sci. 140, 293–297
R. Wiesendanger: Current-induced magnetization (1999)
Part C 24
switching with a spin-polarized scanning tunnel- 24.213 H. Hölscher, W. Allers, U.D. Schwarz, A. Schwarz,
ing microscope, Science 317, 1537–1540 (2007) R. Wiesendanger: Interpretation of ‘true atomic
24.199 M.D. Kirk, T.R. Albrecht, C.F. Quate: Low- resolution’ images of graphite (0001) in noncon-
temperature atomic force microscopy, Rev. Sci. tact atomic force microscopy, Phys. Rev. B 62,
Instrum. 59, 833–835 (1988) 6967–6970 (2000)
24.200 D. Pelekhov, J. Becker, J.G. Nunes: Atomic force 24.214 H. Hölscher, W. Allers, U.D. Schwarz, A. Schwarz,
microscope for operation in high magnetic fields R. Wiesendanger: Simulation of NC-AFM images of
at millikelvin temperatures, Rev. Sci. Instrum. 70, xenon(111), Appl. Phys. A 72, S35–S38 (2001)
114–120 (1999) 24.215 M. Ashino, A. Schwarz, T. Behnke, R. Wiesendan-
24.201 J. Mou, Y. Jie, Z. Shao: An optical detection low ger: Atomic-resolution dynamic force microscopy
temperature atomic force microscope at ambient and spectroscopy of a single-walled carbon nano-
pressure for biological research, Rev. Sci. Instrum. tube: characterization of interatomic van der Waals
64, 1483–1488 (1993) forces, Phys. Rev. Lett. 93, 136101–1–136101–4 (2004)
24.202 H.J. Mamin, D. Rugar: Sub-attonewton force de- 24.216 G. Schwarz, A. Kley, J. Neugebauer, M. Scheffler:
tection at millikelvin temperatures, Appl. Phys. Electronic and structural properties of vacancies on
Lett. 79, 3358–3360 (2001) and below the GaP(110) surface, Phys. Rev. B 58,
24.203 A. Schwarz, W. Allers, U.D. Schwarz, R. Wiesendan- 1392–1499 (1998)
ger: Dynamic mode scanning force microscopy of 24.217 M. Ashino, A. Schwarz, H. Hölscher, U.D. Schwarz,
n-InAs(110)-(1 × 1) at low temperatures, Phys. Rev. R. Wiesendanger: Interpretation of the atomic
B 61, 2837–2845 (2000) scale contrast obtained on graphite and single-
24.204 W. Allers, S. Langkat, R. Wiesendanger: Dynamic walled carbon nanotubes in the dynamic mode
low-temperature scanning force microscopy on of atomic force microscopy, Nanotechnology 16,
nickel oxide(001), Appl. Phys. A 72, S27–S30 (2001) 134–137 (2005)
24.205 F.J. Giessibl: Atomic resolution of the silicon(111)- 24.218 F.J. Giessibl, H. Bielefeldt, S. Hembacher,
(7 × 7) surface by atomic force microscopy, Science J. Mannhart: Calculation of the optimal imaging
267, 68–71 (1995) parameters for frequency modulation atomic force
24.206 M.A. Lantz, H.J. Hug, P.J.A. van Schendel, R. Hoff- microscopy, Appl. Surf. Sci. 140, 352–357 (1999)
mann, S. Martin, A. Baratoff, A. Abdurixit, 24.219 S. Hembacher, F.J. Giessibl, J. Mannhart,
H.-J. Güntherodt: Low temperature scanning force C.F. Quate: Local spectroscopy and atomic imag-
microscopy of the Si(111)-(7 × 7) surface, Phys. Rev. ing of tunneling current, forces, and dissipation on
Lett. 84, 2642–2645 (2000) graphite, Phys. Rev. Lett. 94, 056101–1–056101–4
24.207 K. Suzuki, H. Iwatsuki, S. Kitamura, C.B. Mooney: (2005)
Development of low temperature ultrahigh vac- 24.220 F.J. Giessibl: High-speed force sensor for force mi-
uum force microscope/scanning tunneling micro- croscopy and profilometry utilizing a quartz tuning
scope, Jpn. J. Appl. Phys. 39, 3750–3752 (2000) fork, Appl. Phys. Lett. 73, 3956–3958 (1998)
708 Part C Scanning-Probe Microscopy
24.221 H. Hölscher, W. Allers, U.D. Schwarz, A. Schwarz, 24.235 N. Oyabu, Y. Sugimoto, M. Abe, O. Custance,
R. Wiesendanger: Determination of tip–sample in- S. Morita: Lateral manipulation of single atoms
teraction potentials by dynamic force spectroscopy, at semiconductor surfaces using atomic force mi-
Phys. Rev. Lett. 83, 4780–4783 (1999) croscopy, Nanotechnology 16, 112–117 (2005)
24.222 H. Hölscher, U.D. Schwarz, R. Wiesendanger: 24.236 N. Oyabu, O. Custance, I. Yi, Y. Sugawara, S. Morita:
Calculation of the frequency shift in dynamic Mechanical vertical manipulation of selected single
force microscopy, Appl. Surf. Sci. 140, 344–351 atoms by soft nanoindentation using near con-
(1999) tact atomic force microscopy, Phys. Rev. Lett. 90,
24.223 B. Gotsman, B. Anczykowski, C. Seidel, H. Fuchs: 176102–1–176102–4 (2004)
Determination of tip–sample interaction forces 24.237 Y. Sugimoto, M. Abe, S. Hirayama, N. Oyabu, O. Cus-
from measured dynamic force spectroscopy curves, tance, S. Morita: Atom inlays performed at room
Appl. Surf. Sci. 140, 314–319 (1999) temperature using atomic force microscopy, Nat.
24.224 U. Dürig: Extracting interaction forces and comple- Mater. 4, 156–160 (2005)
mentary observables in dynamic probe microscopy, 24.238 C. Sommerhalter, T.W. Matthes, T. Glatzel, A. Jäger-
Appl. Phys. Lett. 76, 1203–1205 (2000) Waldau, M.C. Lux-Steiner: High-sensitivity quan-
24.225 F.J. Giessibl: A direct method to calculate tip– titative Kelvin probe microscopy by noncontact
sample forces from frequency shifts in frequency- ultra-high-vacuum atomic force microscopy, Appl.
modulation atomic force microscopy, Appl. Phys. Phys. Lett. 75, 286–288 (1999)
Lett. 78, 123–125 (2001) 24.239 A. Schwarz, W. Allers, U.D. Schwarz, R. Wiesendan-
24.226 J.E. Sader, S.P. Jarvis: Accurate formulas for inter- ger: Dynamic mode scanning force microscopy of
action force and energy in frequency modulation n-InAs(110)-(1 × 1) at low temperatures, Phys. Rev.
force spectroscopy, Appl. Phys. Lett. 84, 1801–1803 B 62, 13617–13622 (2000)
Part C 24
hari: Observation of single vortices condensed into bium thin film, J. Appl. Phys. 89, 6787–6789
a vortex-glass phase by magnetic force microscopy, (2001)
Phys. Rev. Lett. 74, 1847–1850 (1995) 24.257 A. Schwarz, R. Wiesendanger: Magnetic sensitive
24.249 C.W. Yuan, Z. Zheng, A.L. de Lozanne, M. Tortonese, force microscopy, Nano Today 3, 28–39 (2008)
D.A. Rudman, J.N. Eckstein: Vortex images in thin 24.258 R. Wiesendanger, D. Bürgler, G. Tarrach, A. Wadas,
films of YBa2 Cu3 O7−x and Bi2 Sr2 Ca1 Cu2 O8−x obtained D. Brodbeck, H.J. Güntherodt, G. Güntherodt,
by low-temperature magnetic force microscopy, R.J. Gambio, R. Ruf: Vacuum tunneling of spin-
J. Vac. Sci. Technol. B 14, 1210–1213 (1996) polarized electrons detected by scanning tunneling
24.250 A. Volodin, K. Temst, C. van Haesendonck, microscopy, J. Vac. Sci. Technol. B 9, 519–524 (1991)
Y. Bruynseraede: Observation of the Abrikosov 24.259 H. Momida, T. Oguchi: First-principles study on
vortex lattice in NbSe2 with magnetic force exchange force image of NiO(001) surface using
microscopy, Appl. Phys. Lett. 73, 1134–1136 a ferromagnetic Fe probe, Surf. Sci. 590, 42–50
(1998) (2005)
24.251 A. Moser, H.J. Hug, B. Stiefel, H.J. Güntherodt: 24.260 U. Kaiser, A. Schwarz, R. Wiesendanger: Magnetic
Low temperature magnetic force microscopy on exchange force microscopy with atomic resolution,
YBa2 Cu3 O7−δ thin films, J. Magn. Magn. Mater. 190, Nature 446, 522–525 (2007)
114–123 (1998) 24.261 R. Schmidt, C. Lazo, H. Hölscher, U.H. Pi, V. Caciuc,
24.252 A. Volodin, K. Temst, C. van Haesendonck, A. Schwarz, R. Wiesendanger, S. Heinze: Probing
Y. Bruynseraede: Imaging of vortices in con- the magnetic exchange forces of iron on the atomic
ventional superconductors by magnetic force scale, Nano Lett. 9, 200–204 (2008)
microscopy images, Physica C 332, 156–159 (2000) 24.262 D. Rugar, R. Budakian, H.J. Mamin, B.W. Chui: Sin-
24.253 M. Roseman, P. Grütter: Estimating the magnetic gle spin detection by magnetic resonance force
Part C 24
penetration depth using constant-height mag- microscopy, Nature 430, 329–332 (2004)
netic force microscopy images of vortices, New J. 24.263 J.A. Sidles, J.L. Garbini, G.P. Drobny: The theory
Phys. 3, 24.1–24.8 (2001) of oscillator-coupled magnetic resonance with po-
24.254 A. Volodin, K. Temst, C. van Haesendonck, tential applications to molecular imaging, Rev. Sci.
Y. Bruynseraede, M.I. Montero, I.K. Schuller: Mag- Instrum. 63, 3881–3899 (1992)
netic force microscopy of vortices in thin niobium 24.264 J.A. Sidles, J.L. Garbini, K.J. Bruland, D. Rugar,
films: Correlation between the vortex distribution O. Züger, S. Hoen, C.S. Yannoni: Magnetic res-
and the thickness-dependent film morphology, onance force microscopy, Rev. Mod. Phys. 67,
Europhys. Lett. 58, 582–588 (2002) 249–265 (1995)
24.255 U.H. Pi, T.W. Noh, Z.G. Khim, U. Kaiser, M. Lieb- 24.265 D. Rugar, O. Züger, S. Hoen, C.S. Yannoni, H.M. Vi-
mann, A. Schwarz, R. Wiesendanger: Vortex eth, R.D. Kendrick: Force detection of nuclear
dynamics in Bi2 Sr2 CaCu2 O8 single crystal with low magnetic resonance, Science 264, 1560–1563 (1994)
density columnar defects studied by magnetic force 24.266 Z. Zhang, P.C. Hammel, P.E. Wigen: Observation of
microscopy, J. Low Temp. Phys. 131, 993–1002 (2003) ferromagnetic resonance in a microscopic sample
24.256 M. Roseman, P. Grütter, A. Badia, V. Metlushko: using magnetic resonance force microscopy, Appl.
Flux lattice imaging of a patterned nio- Phys. Lett. 68, 2005–2007 (1996)
711
Higher Harm
25. Higher Harmonics and Time-Varying Forces
in Dynamic Force Microscopy
Ozgur Sahin, Calvin F. Quate, Olav Solgaard, Franz J. Giessibl
In atomic force microscopy, a force-sensing can- graphy of the surface is then obtained by recording
tilever probes a sample and thereby creates the feedback signal.
a topographic image of its surface. The sim- Tapping-mode AFM has the potential to mea-
plest implementation uses the static deflection of sure much more than simply the topography of
the cantilever to probe the forces. More recently, a surface, however. As can be seen from Fig. 25.1,
dynamic operation modes have been introduced, the tip–sample interaction forces as the AFM tip
which either work at a constant oscillation fre- approaches, interacts with, and retracts from the
quency and sense the amplitude variations caused surface has a complex time dependence. This time
by tip–sample forces (amplitude modulation or dependence reflects the attractive and repulsive
tapping mode) or operate at constant amplitude forces that act between the tip and the sample,
and varying frequency (frequency modulation and contains information about the chemical and
mode). Here, we report about new operational physical properties of the sample.
concepts capturing the higher harmonics in either In the remaining sections of this chapter we
amplitude modulation or frequency modula- describe methods that enable measurements of
tion mode. Higher-harmonic detection in atomic the time-varying tip–sample force waveforms in
force microscopy allows us to measure time- tapping-mode AFM. We first present a simple
varying tip–sample forces that contain detailed model to calculate time-varying tip–sample force
Part C 25
information about the material characteristics waveforms and show how these forces depend
of the sample, while higher-harmonic detection on sample properties. Then we will present two
in small-amplitude frequency modulation mode strategies to engineer the force-sensing cantilever
allows a significant improvement in spatial res- to measure the tip–sample force waveform and its
olution, in particular when operating in vacuum frequency components. As application examples
at low temperatures. The most widely used mode we present: (1) time-varying force measurements
of operation of atomic force microscopy (AFM) that allow quantitative comparisons of material
is tapping mode, because in this mode lateral stiffness, and (2) observation of the glass transition
tip–sample interaction forces are minimized. The of polymer blends with nanometer-scale lateral
gentle interaction between the AFM tip and the resolution.
sample under test reduces wear on the sample and After the discussion of time-varying force
localizes the deformations to give nanometer, or measurements in standard AFM tapping mode, we
even molecular, resolution [25.1, 2]. introduce higher-harmonic imaging in AFM with
In tapping mode, the AFM cantilever is vibrated small vibration amplitudes. In small-amplitude
at resonance in the vicinity of the sample so AFM imaging, the tip is in the force field of the
that the tip makes contact with the sample once sample during most of its vibration cycle. Relatively
during each cycle. The tip–sample forces reduce low-order harmonics of the tip–sample force
the vibration amplitude of the cantilever. The then contain information about the higher-order
vibrating cantilever is scanned across the surface gradients of the tip–sample interaction force field.
while a feedback mechanism adjusts the height These low-order harmonics in small-amplitude
of the cantilever base to maintain the vibration dynamic AFM imaging can be measured directly,
amplitude at a constant setpoint value. The topo- yielding excellent spatial resolution.
712 Part C Scanning-Probe Microscopy
its fundamental resonant frequency. Under typical op- Because the tip–sample force Fts is a periodic wave-
erating conditions, the periodic driving force results in form we can expand it as a Fourier series as follows:
∞
a) Distance (nm) Fts (t) = an cos(nωt) + bn sin(nωt) ,
100 n=0
n = 0, 1, 2, . . . . (25.1)
50
Here, the frequency ω, is the driving frequency, which
is chosen close to the fundamental resonance frequency
0
0 180 360 540 720 of the cantilever. The coefficients an and bn are given as
b) Force (nN)
2π/ω
50 ω
an = Fts cos(nωt) dt , (25.2a)
π
0 0
2π/ω
–50 ω
0 180 360 540 720 bn = Fts sin(nωt) dt . (25.2b)
π
Phase of tip position (deg) 0
Fig. 25.1a,b Calculated tip–sample distance (a) and tip– The k-th harmonic force can be written as
sample interaction forces (b) over two cycles of cantilever
oscillation. Negative displacements correspond to sample Fts n cos(nωt + θn ) = an cos(nωt) + bn sin(nωt) .
indentation. Attractive (negative) and repulsive (positive) (25.3)
forces appear during the tip–sample interaction. The mag-
nitude and duration of these forces depend on the physical Here, Fts n = an2 + b2n and θn are the magnitude and
properties of the sample phase of the n-th harmonic force, respectively. The
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy 25.1 Modeling of the Tip–Sample Interaction Forces 713
phase θn of a higher harmonic is defined relative to a ref- sample interaction forces by the equations below
erence harmonic at the same frequency that is in phase HR σ 2 1 σ 8
with the fundamental displacement; i. e., if we repre- f ts (r) = 2 − + , (25.4a)
sent the tip displacement with As cos(ωt), the reference 6σ r 30 r
signal is cos(nωt). 4 √
f ts (d) = E Rd 3/2 . (25.4b)
The tip–sample forces can be seen as a superpo- 3
sition of harmonic forces, each at an integer multiple Here, r is the tip–sample separation, and d is the sample
of the driving frequency. Later we will show how we indentation. H, R, σ, and E are the Hamaker con-
can measure these higher harmonics by tuning a higher- stant, tip radius, typical atomic distance for the tip and
order resonance of the cantilever to be an integer the sample, and the reduced Young’s modulus for the
multiple of its fundamental resonance frequency (i. e., tip and the sample, respectively. Tip–sample forces are
nω = ωk , where ωk is the resonance frequency of the k- governed by (25.4a) when the tip is away from the
th flexural mode of the cantilever). The higher-harmonic sample and by (25.4b) when the tip is indenting the
force will then drive the higher-order resonance, and sample. The reduced Young’s modulus E of the sam-
the deflection of the cantilever in the higher-order reso- ples is the main factor that determines the tip–sample
nance is a measure of the higher-harmonic force. contact duration. In Fig. 25.2, E values for the three
samples are chosen to give contact durations of 5%,
25.1.2 Frequency Spectrum 10%, and 15% of the period on the stiff, medium,
of the Tip–Sample Force and compliant samples, respectively. Attractive forces
on all the samples are assumed to be equal, giving
Figure 25.2 shows the calculated periodic tip–sample equal amounts of energy dissipation at contact. For
forces and their respective harmonic force components the Hamaker constants we used Ha = 10 × 10−20 J for
Fts n for stiff, medium, and compliant samples. De- approach and Hr = 30 × 10−20 J for retraction. The pa-
tails of the tip–sample force calculations can be found rameters R and σ are chosen to be 10 and 0.1 nm,
in [25.4]. The harmonic force components are calcu- respectively. These values result in energy dissipation
lated for a cantilever with spring constant K 1 = 10, of ≈ 30 eV per contact. The parameters other than E are
Part C 25.1
quality factor Q 1 = 100, free amplitude A0 = 100 nm, chosen to be the same on all three materials in order to
and setpoint amplitude As = 80 nm. We model the tip– simplify the analysis.
0 2
–100 0
–2 0 2 0 5 10 15 20 25
b)
50
4
0
2
–50
–100 0
–2 0 2 0 5 10 15 20 25
c)
50
4 Fig. 25.2a–c Interaction forces be-
0 tween the tip and the sample for
2 three different samples: (a) stiff,
–50
(b) medium, and (c) compliant. The
–100 0 amplitudes of the harmonics for the
–2 0 2 0 5 10 15 20 25 three tip–sample forces are shown on
Phase of tip position (rad) Index of the harmonic
the right
714 Part C Scanning-Probe Microscopy
For the specific example shown in Fig. 25.2, har- ping force on the surface. The average tapping force
monics above the tenth are strongly dependent on the mainly depends on the cantilever, the drive amplitude,
stiffness of the sample. The tip–sample force is ap- and the setpoint amplitude, which are all equal for the
proximately a periodic clipped sine wave, so the pulse three samples. This analysis shows that only the har-
width, or contact duration, determines the harmonic monics that are sufficiently high to have periods that
content. Shorter-duration contacts generate larger am- are comparable to, or shorter than, the contact du-
plitudes at higher harmonics. The contact duration ration contain significant information on the stiffness
increases for more compliant samples, resulting in of the samples. This information can be recovered in
smaller magnitudes at higher harmonics. These cal- part by measuring the amplitude of higher-harmonic
culations show that the first harmonics for each of vibrations of the cantilever [25.4–7]. However, low
the three cases have the same magnitude. This is be- signal-to-noise ratios of high-frequency vibrations of
cause the magnitude of the first harmonic [n = 1 in the cantilevers generally limit their practical applica-
(25.1–25.2b)] is approximately equal to the average tap- tions.
25.2.1 Response of the Cantilever face. With tip–sample interaction as the driving force,
to High-Frequency Forces
Free end
We have discussed the origin of higher-harmonic forces
and how they depend on the physical properties of
samples. In an experiment we can only measure the
higher-harmonic forces through their effect on can-
Fixed end
tilever motion. So, it is necessary to understand how
the cantilever responds to higher-harmonic force com- z
ponents. To obtain good signal-to-noise ratios for the
measurements, the cantilever must have a good re-
sponse to higher harmonics.
In determining the response of the cantilever to
forces at different frequencies we need to go beyond
simple harmonic oscillator models [25.8] and model the z
cantilever as a continuum mechanical system [25.9].
The AFM cantilever is fixed at the base and free to
move at the tip end. The external forces acting on the
cantilever are the drive force at the base and the tip–
sample forces at the tip. The motion of the cantilever
is governed by the Euler–Bernoulli equation. The solu- z
tion of this equation for a rectangular cantilever can be
found in [25.10]. The Euler–Bernoulli equation is linear
in time, so we can describe the cantilever response in Fig. 25.3 Mode shapes of a rectangular cantilever fixed at
terms of the eigenmodes of the cantilever. Each of these one end and free at the other end
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy 25.2 Enhancing the Cantilever Response 715
the motion of the cantilever can be expressed as a su- harmonics, which contain much of the information
perposition of the responses of the eigenmodes. The on the stiffness of the sample, yield relatively low
response of the cantilever y(t) to an external harmonic cantilever responses, limiting our ability to measure
force applied to the tip at the free end can be approxi- the higher harmonics. This situation is less severe
mated as in liquid environments, where higher-harmonics-based
∞
F 4 force measurements have been demonstrated by several
y(t) = eiωt . (25.5) groups [25.11–13].
M ω − ω2 + iωω
k=1 k
2
Qk
k
Here, y(t) is the displacement of the tip (free end) of 25.2.2 Improving Cantilever Response
the cantilever at time t, and F and ω are the magni- by Tuning Flexural Resonance
tude and frequency of the harmonic force acting on Frequencies
the tip, respectively. The parameters ωk and Q k are
the resonance frequency and quality factor of the k-th Figure 25.4 shows that harmonics close to a resonance
eigenmode. frequency will have larger deflections. A correctly de-
Figure 25.4 shows the frequency response of the signed cantilever geometry can tune a higher-order
optical-lever signal for a rectangular cantilever. These flexural resonance frequency to an integer multiple of
calculations are based on (25.5) while taking into the drive so that the corresponding harmonic is en-
account that the optical-lever signals in the position- hanced by the resonance peak. This can be done by
sensitive detector are proportional to the slopes of appropriately removing mass from regions where the
the free end of the cantilever. The peaks in the re- cantilever has high mechanical stress in that particular
sponse curves are the resonances of each flexural mode [25.14]. Recently, it was suggested that placing
vibration mode. The frequency axis is normalized to a concentrated mass on the cantilever could also result
the first resonance frequency, which equals the driving in a good match between higher harmonics and flexural
frequency in tapping-mode AFM. The higher harmon- resonances [25.15]. Engineering the cantilever geome-
ics, marked with circles on the frequency response try to improve force sensitivity of higher-order flexural
curve, are therefore located at integer multiples of the modes in scanning capacitance microscopy and Kelvin
Part C 25.2
first resonance frequency. The cantilever responds to probe microscopy has also been demonstrated [25.16,
each force harmonic with a displacement given by 17].
this frequency response. We see that the higher-order
Free end
Magnitude (a.u.)
102
Fixed end
100 z
10–2
z
10– 4
0 2 4 6 8 10 12 14 16 18 20
Normalized frequency (Hz)
z
Fig. 25.4 Calculated frequency response of a rectangular
cantilever. The frequency axis is normalized to the first
resonance frequency. The magnitudes represent the optical
signal at the position-sensitive detector. This optical sig- Fig. 25.5 Mode shapes of a rectangular cantilever with
nal is proportional to the slope of the cantilever at the laser a notch. The notch is located at the high bending region
spot. The circles are located at the integer multiples of the in the third flexural mode shape and is ≈ 1/3 of the total
first resonance frequency length away from the free end
716 Part C Scanning-Probe Microscopy
responds to a highly curved region of the third mode, neighbors. These are the harmonics that are closest to
but not to highly curved regions of the first two modes. the resonance frequencies of the harmonic cantilever.
Therefore, the effect of the notch in reducing the elas- Especially the 16th harmonic has a much higher sig-
tic energy is more prominent in the third mode. Highly nal level relative to its neighbors. This is because
curved regions of a mode are also highly displaced, so the frequency of this particular harmonic matches the
third resonance frequency of the harmonic cantilever.
Such cantilevers can be fabricated with conventional
Deflection signal (dBm)
10
silicon-based microfabrication techniques. A more de-
0
tailed discussion on the fabrication of the cantilever in
–10 Fig. 25.6 is given in [25.14].
–20
–30 25.2.3 Recovering the Time-Varying
–40 Tip–Sample Forces
–50 with Torsional Vibrations
–60
–70 AFM cantilevers have a second type of vibration modes,
–80 called torsional modes, in addition to the flexural modes
–90 discussed above. Vibrations in these modes result in
0 100 200 300 400 500 600 700
angular deflections of the cantilever. These modes are
Vibration frequency (kHz)
excited as a result of torque acting on the cantilever.
Fig. 25.7 Vibration spectrum of a harmonic cantilever in tapping- The tip of a typical cantilever is located on the longitu-
mode AFM. The cantilever is driven at its fundamental resonance dinal axis, preventing tip–sample forces from creating
frequency (37.4 kHz), and higher-harmonic generation is observed. torque on the cantilever when tapping on a sample.
The second (240 kHz) and third (598 kHz) harmonics coincide with In this section we will describe a class of cantilevers,
higher resonances and have relatively large signal power called torsional harmonic cantilevers, that enables the
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy 25.2 Enhancing the Cantilever Response 717
excitation of torsional vibration modes [25.18, 19]. Tor- The torsional harmonic cantilever has a tip that is
sional vibration modes are very sensitive to tip–sample offset from the long axis of the cantilever. An example
forces and allow simultaneous measurement of a large of such a cantilever is shown in Fig. 25.8a. When a tor-
number of higher harmonics so that the tip–sample sional harmonic cantilever is vibrated in tapping mode,
forces can be recreated with high temporal resolu- tip–sample interaction forces generate torque around
tion. We will begin with a theoretical discussion of the the long axis of the cantilever and excite the torsional
torsional response of an AFM cantilever with an off- modes (Fig. 25.8b). The overall motion of the cantilever
set tip and then show experimental results from the is a combination of flexural and torsional vibrations.
vibration measurements of a torsional harmonic can- The vibration at the fundamental flexural resonance
tilever. frequency is still the dominant component. The mo-
tion of the cantilever is detected with a laser beam
reflected from the backside of the cantilever falling onto
a) a four-quadrant position-sensitive diode (Fig. 25.8c).
The difference in optical powers in the upper and lower
halves is proportional to longitudinal (flexural-mode)
deflection and the difference in left and right halves is
proportional to torsional deflection.
When the tip interacts with the surface as it ap-
proaches and retracts, the torsional vibration mode acts
as a force sensor that measures the force acting on
the tip. The torsional resonance frequency is much
higher than the first flexural resonance frequency, so
the torsional mode responds to the variations in the
b) tip–sample force over a wide frequency range. Fig-
ure 25.9 shows the calculated frequency response of
the torsional and flexural modes of the torsional har-
Part C 25.2
Magnitude
c) Q1 Q2 102
100
10–2
Q3 Q4 10–4
0 2 4 6 8 10 12 14 16 18 20
Normalized frequency
Fig. 25.8 (a) Scanning electron micrograph image of a tor-
sional harmonic cantilever. The cantilever is nominally Fig. 25.9 Calculated frequency response of a rectangular
300 μm long, 22 μm thick, and 302 μm wide. The tip is cantilever in the torsional and flexural modes. The fre-
offset 15 μm from the centerline of the cantilever. (b) Sim- quency axis is normalized to the first flexural resonance
ulated first torsional mode shape of a rectangular cantilever frequency. The magnitudes represent the optical signal at
fixed at the base. (c) Illustration of the laser spot on the the position-sensitive detector. These optical signals are
four-quadrant position-sensitive photodetector. The optical proportional to the slope of the cantilever at the laser
power difference (Q1 + Q2) − (Q3 + Q4) is proportional to spot. The circles are located at the integer multiples of the
vertical cantilever deflection, and the optical power dif- first resonance frequency. Note that the torsional response
ference (Q1 + Q3) − (Q2 + Q4) is proportional to torsional (solid curve) is much higher than the flexural response
angle (dashed curve) at higher harmonics
718 Part C Scanning-Probe Microscopy
monic cantilever. These curves correspond to the lateral The higher harmonics in the flexural vibration
and vertical deflection signals at the position-sensitive spectrum have signal-to-noise ratios that are too low
detector. for practical measurements. In the torsional vibration
At frequencies below the first flexural resonance fre- spectrum in Fig. 25.10b, on the other hand, the signal-
quency, the flexural response is much higher than the to-noise ratios are sufficient for practical measurements
torsional response. This is because the effective spring for the first 19 harmonics. The signal levels around
constant of the first flexural mode is much smaller than the 16th harmonic increase due to the first torsional
that of the first torsional mode. On the other hand, at resonance of the cantilever located at 16.2 times the
higher frequencies where the higher harmonics of the drive frequency. The vibration spectrum in Fig. 25.10b
tip–sample forces are located, the torsional response is shows that the torsional vibrations provide good sig-
larger than the flexural response. nal levels up to the 19th harmonic of the first flexural
Figure 25.10a,b shows the flexural and torsional resonance frequency. This means that this particular
vibration spectra of a torsional harmonic cantilever torsional-mode force sensor can resolve tip–sample
while tapping on a polystyrene sample. The cantilever forces with a temporal resolution roughly 20 times
is driven near the first flexural resonance frequency shorter than the fundamental flexural oscillation pe-
(52.5 kHz) by a piezoelectric element from the base. riod.
The free vibration amplitude and setpoint amplitude are
100 and 60 nm, respectively. While tapping on the sur- 25.2.4 Time-Varying Tip–Sample Force
face the cantilever simultaneously moves in flexural and Measurements
torsional modes, as shown by the vibration spectra. The
dominant component of the cantilever motion is the first The mechanical bandwidth of the torsional mode de-
peak in the flexural vibration spectrum. This is the mo- termines the response to variations in the tip–sample
tion at the drive frequency. The other harmonics are forces as the tip vibrates. This bandwidth is deter-
generated by the tip–sample interaction. mined by the first torsional resonance frequency, which
is 16.2 times the drive frequency for this cantilever.
In general, it is possible to measure the first few har-
a) Detector signal (dBm)
Part C 25.2
Part C 25.2
to achieve an approximate noise bandwidth of 500 Hz. Fig. 25.11a–c Vibration signals from flexural and torsional motions,
The measured vibration signals coming from both flex- and tip–sample forces. (a) The signals at the four-quadrant photode-
ural and torsional modes over one period is shown tector for vertical and torsional displacements. The solid curve is the
in Fig. 25.11a. The effect of nonlinear mechanical gain torsional signal. We multiplied the torsional signal by a factor of 10
due to the torsional frequency response is removed to view the two curves clearly in one graph. (b) The torsional vibra-
digitally with the aid of (25.6). The resulting wave- tion signal after being divided by the torsional frequency response.
form is given in Fig. 25.11b. In this waveform the Except for the pulse located between the 300th and 400th time steps,
tip–sample forces are not zero even when the tip is the tip–sample forces should have been close to zero, because the
away from the surface. This error is due to the non- tip is far away from the surface at those times. The measured sig-
linearity of the detection circuit and cross-talk from nals when not in contact come from cross-talk from the flexural
the large flexural signal that produces additional sig- deflection signal. The dashed curve estimates the error introduced
nals at the first few harmonics. Those components are by these sources. When it is subtracted from the solid curve we get
removed with a signal-processing procedure that as- the time-varying forces plotted in (c)
sumes that the tip–sample interaction forces are zero
when the cantilever is away from the surface, and sub- tip–sample forces are indicated in newtons. Conver-
tracts any additional signal at the first few harmonics sions of measured voltages into force units are done
that results in a nonzero tip–sample force. In Fig. 25.11b by comparing the time-average forces as measured by
the computed correction signal that arises from cross- flexural and torsional deflections. If the spring constant
talk and nonlinearities is also given. Notice that the of the fundamental flexural mode has been calibrated
two curves overlap during the times when the tip is with conventional methods [25.21], then the torsional
not in contact. Once this additional component is sub- mode can be calibrated against the flexural spring
tracted we get the corrected tip–sample force waveform constant. In practice, measurements of time-average
in Fig. 25.11c. quantities are subject to drifts and misalignments of
It is important to note that, while the measure- the position sensors. To overcome this difficulty, bifur-
ments of torsional vibrations are in units of volts, the cations in force–distance curves in tapping mode can
720 Part C Scanning-Probe Microscopy
be used [25.22]. Rather than comparing the absolute average forces. Therefore, calibration is performed free
values, this method compares relative changes in time- from drift-related errors.
by the stiffness of the sample. Larger loading forces To achieve absolute measurements, we have cho-
are required to produce a given depth of indentation sen graphite as our reference material for measurements
into stiffer materials. Therefore, the force increases of the indentation forces for high-density and low-
faster on graphite than on the polyethylene samples density polyethylene. To compare the forces curves
in Fig. 25.10b. The indentation forces do not follow obtained on the three samples, we need to eliminate
a linear relationship. An approximate model predicts the contribution of attractive forces because they are
that the repulsive forces vary with a power-law rela- not governed by the elastic properties of the mater-
tion [25.24] ials. In the graphs of Fig. 25.12b, the points with the
highest attractive forces are marked with crosses. Tip–
Frep = γ d n . (25.7)
sample mechanical contact is broken at approximately
Here, Frep is the repulsive tip–sample force, γ is a con- these points as the tip is retracted. There are no repul-
stant that depends on the reduced Young’s modulus and sive forces at these points, so the net negative force
tip diameter, d is the indentation depth, and n is a num- is due to the attractive forces. To the left of these
ber that depends on the tip and surface geometry. For crosses, the tip–sample forces increase as the tip in-
a spherical tip and flat surface, n = 1.5 [25.25]. If the dents the samples. The insert in Fig. 25.12b shows
same tip is used, the value of n is expected to be the a plot of the indentation forces for the first 3 nm to
same for experiments on different materials. The dif- the left of the crosses. We choose the points of max-
ferences will arise only in the multiplicative term γ imum attractive forces as the origin of the graphs
because the reduced Young’s modulus values are dif- because this is where the indentation forces and in-
ferent. dentation depths are all zero. By scaling the force on
The curves in Fig. 25.12b show double values for high-density polyethylene by a factor of 3.0 and the
forces mainly at positive separations. This is because forces on low-density polyethylene by a factor of 27.0,
tip–sample forces are different during the approach they both match the reference forces on graphite. The
and retraction of the tip. The upper branches of the matching of these curves supports the assumption that
curves show the approach of the tip, whereas the lower the functional dependency of forces with respect to
branches show retraction. Capillary forces are larger in indentations is similar. Graphite has an approximate
Part C 25.3
retraction due to attractive van der Waals forces that elastic modulus of 5 GPa, so our measured values for
pull the surface and raise it above its equilibrium level the elastic moduli of high-density polyethylene and
during retraction, resulting in attractive forces at larger low-density polyethylene are 1.7 GPa and 180 MPa, re-
tip–sample separations. There is also the possibility of spectively. High-density polyethylene is more ordered
forming a liquid neck between the tip and the sample and is expected to be significantly stiffer than low-
during retraction, but the samples used for the mea- density polyethylene. Unfortunately, there is a relatively
surements are hydrophobic, so this is not likely to be large spread in published values for the elastic modulus
the origin of the hysteresis observed in the force curves for graphite, so a well-calibrated reference material is
of Fig. 25.12. still needed for accurate measurements.
These quantitative comparisons and absolute mea-
25.3.2 Quantitative Comparison surements are made without knowledge of many
of Material Properties parameters of the AFM experiment, such as cantilever
spring constant, tip geometry, drive force, setpoint am-
The ratio of the Young’s modulus values of different plitude, photodetector gain, position of the laser spot,
materials can be quantitatively measured by compar- and tip offset distance. This is possible because the cal-
ing the tip–sample force curves for those materials, culations for quantitative measurements use only the
because for a given depth of indentation, the ratio of voltages at the output of the photodetector, together
forces is equal the ratio of the respective γ coefficients. with the flexural and torsional resonance frequencies.
For these quantitative measurements we do not need Because we take the ratio of force values in the force–
to assume a power-law functional dependency as in distance curves, the ratio of measured voltages is all
(25.7). The approximation that the ratio of the Young’s that is required, while knowledge of the relationship
modulus values equals the ratio of measured forces at between measured voltages and actual forces is not
a given depth of indentation is valid as long as the needed. This technique eliminates all the instrument
functional dependency is the same for the two mater- variables that are difficult to measure and control in
ials. AFM experiments.
722 Part C Scanning-Probe Microscopy
25.3.3 Imaging the Glass Transition and studied the glass transition of the two compo-
of a Binary Polymer Blend Film nents, polystyrene (PS) and poly(methyl methacrylate)
(PMMA). These components with different glass-
According to the theoretical analysis carried out in the transition temperatures (around 100 ◦ C and 130 ◦ C,
first section, the magnitudes of the higher harmonics respectively) form submicrometer domains within the
depend on the stiffness of the samples. If carefully film. As the temperature is elevated, polystyrene goes
chosen, the amplitude of a particular harmonic will through the glass transition before PMMA. The re-
monotonically increase with increasing stiffness of the gions of rubbery phase will be less stiff than the glassy
samples. We are interested in mapping mechanical regions, so that we can observe the glass transitions
property variations, so imaging the amplitude of a single of individual components of the composite polymer
higher harmonic is the simplest solution. The torsional film. For the measurements a torsional harmonic can-
harmonic cantilever provides the first 20 torsional har- tilever with a torsional resonance frequency about 11
monics with good signal levels, so we can measure times higher than the fundamental is used. The verti-
any one of these harmonics with a lock-in amplifier cal spring constant and quality factor of this cantilever
to produce the corresponding harmonic image of the were 6 N/m and 91, respectively. The free amplitude
surface. and setpoint amplitude are chosen as 100 and 70 nm,
By recording the torsional harmonic amplitudes respectively.
while scanning the surface in tapping mode we have Images of the topography, phase, and tenth torsional
generated the harmonic force images of an ultrathin harmonic at temperatures between 85 ◦ C and 215 ◦ C
(about 50 nm) binary polymer film on a silicon substrate are presented in Fig. 25.13. The two components of
85°C
Part C 25.3
10 nm 10° 10 V
115°C
10 nm 10° 10 V
145°C
10 nm 10° 10 V
160°C
10 nm 20° 10 V
175°C
Fig. 25.13 Topography, phase and
tenth-harmonic images of a thin poly-
50 nm 20° 10 V
mer film composed of PS and PMMA.
190°C The circular features are PMMA,
and the matrix is PS. Brighter color
represents larger height, phase or har-
50 nm 90° 10 V
monic amplitude. The scan area is
215°C 2.5 × 5 mm2 . The color bar below the
panels corresponds to signal ranges
indicated on the upper left corner of
each panel (after [25.18])
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy 25.3 Application Examples 723
the polymer blend are easily distinguishable in the to- ture and magnitude of the changes cannot be extracted
pography images up to 145 ◦ C because of the height from the topography images.
differences. The higher regions are PMMA domains In the same experiment, phase images distinguish
and lower regions are composed of polystyrene. Above the two material components but show a gradually in-
190 ◦ C the boundaries of the two material components creasing contrast up to 215 ◦ C. The phase image is
are not clear anymore. The topographical images be- primarily determined by energy dissipation in the tip–
come blurry at elevated temperatures because the glass sample contact. There are several mechanisms that are
transition is accompanied by morphological changes involved in tip–sample energy dissipation. In the present
due to increased molecular mobility. However, the na- case, the viscous response of the sample at elevated
30 30
160°C
160°C
20 20 175°C
Part C 25.3
190°C
10 175°C 10
190°C
215 °C
0 0
–10 –8 –6 –4 –2 0 –10 –8 –6 –4 –2 0
Indentation depth (nm) Indentation depth (nm)
c) Effective modulus (Pa)
10 G
PPMA
PS
1G
100 M
10 M
0 50 100 150 200 250
Temperature (°C)
Fig. 25.14a–c Indentation force curves obtained from time-varying force measurements on polystyrene regions (a) and
on PMMA regions (b). Effective Young’s modulus values corresponding to each curve is plotted in (c) (after [25.18])
724 Part C Scanning-Probe Microscopy
temperatures is likely to play an important role. In 25.3.4 Detailed Analysis with Time-Varying
addition, on compliant samples, attractive forces be- Nanomechanical Forces
tween the tip and the sample can pull the sample
and raise it above its equilibrium level. The contact The changes in the polymer components were studied in
is eventually broken as the tip retracts and the en- greater detail by measuring the time-varying forces and
ergy stored in the pulled sample is dissipated. These generating force–distance curves as previously illus-
two mechanisms will result in increased energy dis- trated in Fig. 25.12b. In Fig. 25.14 we plot the unloading
sipation at higher temperatures. However, once both portions of the force–distance curves as obtained from
components are in the rubbery phase, one would expect time-varying forces on polystyrene and PMMA regions
a reduction in phase contrast. The measurements show at each temperature. On both samples, the rate of in-
that phase contrast is even larger at elevated tempera- crease in the repulsive forces decreases with increasing
tures. It is therefore difficult to identify and quantify temperatures. We used the Derjaguin–Muller–Toporov
the changes in the material properties with tempera- (DMT) contact mechanics model [25.25] to calculate
ture. the elastic modulus of these samples from the curves
Simultaneously recorded harmonic images show recorded at each temperature. The resulting values are
contrast between the two material components first in- plotted in Fig. 25.14c. We see that, at low temperatures,
creasing and then decreasing with temperature. The the effective elastic modulus values of PS and PMMA
contrast increases dramatically between 145 ◦ C and regions are 2.3 and 3.7 GPa, respectively. These values
160 ◦ C and decreases between 190 ◦ C and 215 ◦ C. are only slightly reduced at temperatures up to 145 ◦ C.
When PS goes through its glass transition, the harmonic However, the effective elastic modulus of both materials
amplitude corresponding to PS regions reduces. PMMA dramatically reduce around 160 ◦ C and 190 ◦ C. Modu-
goes through its glass transition at a higher temperature, lus values of both materials reduce more than an order
which is accompanied by a reduction in the harmonic of magnitude, identifying the glass transition. From
amplitude measured on PMMA. As a result the con- these plots we estimate the apparent glass-transition
trast in stiffness first increases and then decreases as the temperatures of PS and PMMA to be 160 ◦ C and
polymer blend is heated. This result suggests that the 180 ◦ C, respectively. The relatively high glass-transition
Part C 25.4
harmonic images can provide a qualitative explanation temperatures measured in these experiments are ex-
for the origin of the changes on the surface. A more pected consequences of the frequency dependence of
detailed understanding and even quantification of the the glass transition. Our technique measures the me-
changes can be gained by using the full spectrum of chanical properties at the drive frequency of 50 kHz.
available harmonic signals, i. e., by analyzing the force– Glass transitions of bulk materials are determined either
distance curves derived from time-varying tip–sample through thermal measurements or dynamic mechanical
forces. measurements with frequencies below 100 Hz [25.26].
AFM, the full tip–sample potential can be immediately can be expressed by a convolution of the n-th force
recovered over the z-range covered by the oscillating gradient dFtsn / dz n with a bell-shaped weight function
cantilever if the amplitudes and phases of all higher har- (1 − u 2 )n−1/2 . Figure 25.15 shows three snapshots of
monics are available. Dürig has expressed the higher the oscillating cantilever in the upper turnaround point,
harmonics as a Tchebyshev expansion of the tip–sample the neutral position, and the closest sample approach.
force [25.29]. Sahin et al. [25.4] and de Lozanne [25.30] The graph on the left of the figure shows various weight
expand the temporal dependence of the tip–sample functions for deriving the frequency shift and higher
force in a Fourier series with base frequency f and harmonics from the force gradient and higher-order
express the higher harmonics as the response of the gradients. The semicircular weight function is convo-
cantilever to an excitation at frequency n × f . Mathe- luted with the force gradient, yielding the frequency
matically, these two notions are of course equivalent. shift [25.28]. The bell-shaped weight functions are con-
We start with Dürig’s formula for the amplitude of the voluted with higher-order force gradients to derive the
n-th harmonic higher harmonics (25.10). The weight functions have
1 their maxima at u = 0, i. e., a distance of A further
2 1 Tn (u) away from the surface than the minimal tip–sample dis-
an = Fts (z + Au) √ du , (25.9)
πk 1 − n 2 1 − u2 tance z min .
−1 If the amplitude A is small enough such that higher-
where Tn (u) is the n-th Tchebyshev polynomial of the order force gradients still have a reasonable magnitude
first kind. We can show at distance z min + A, the higher harmonics are not just
by applying integration by parts
n-times ( gh = − g h + gh) that proportional to the frequency shift Δ f , but contain in-
formation about higher-order force gradients.
2 1 1 The benefit of higher-order force gradient maps
an = An
πk 1 − n 2 (2n + 1) . . . 3 · 1 is demonstrated by an elementary but instructive
1 example illuminating the contrast achievable by tip–
dFtsn (z + Au)
× (1 − u 2 )n−1/2 du , (25.10) sample interaction potential, force, force gradient,
dz n and higher-order gradients shown in Fig. 25.16. Fig-
−1
Part C 25.4
ure 25.16a shows a model of the charge distribution
because the n-th integral of Tn (u)/(1 − u 2 )1/2 is (1 − where the AFM tip is replaced by a single electron
u 2 )n−1/2 /[(2n + 1) . . . 3 · 1]. Thus, the n-th harmonic with charge −e. The sample ion probed by the tip is
modeled by a central ion with charge +9e, surrounded
by eight electrons with charge −e that point towards
z
the corners of a cubic lattice. This charge distribution
for the sample ion is motivated by the charge density
zmin + 2A calculations by Posternak et al. [25.31] and Mattheis
and Hamann [25.32] for the (100) surface of tungsten.
a10 zmin + A While tungsten (001) is a very special surface, it is
a3 reasonable to assume that most surface atoms do not
zmin have a spherically symmetric charge distribution as free
Δf
atoms with filled shells, but local charge maxima that
w reflect the chemical bonding symmetry. Figure 25.16b
Fig. 25.15 Cantilever in three phases of its oscillation cy- shows the potential energy as a function of lateral dis-
cle: minimal distance z min , average distance z min + A, and placements in x and y for a constant height z. It is
maximal distance z min + 2A. The frequency shift is cal- interesting to note that both the energy and the force
culated by the convolution of the force gradient with shown in Fig. 25.16c are almost symmetric with respect
a semispherical weight function indicated by Δ f , the to rotations around the z-axis. However, the symmetry
second-harmonic amplitude a2 is calculated by convolut- of the underlying charge distribution becomes more and
ing the second-order force gradient by a weight function more apparent with increasing order of the force gradi-
(1 − u 2 )3/2 , and the n-th harmonic an is computed by con- ent, as shown in Fig. 25.16d–h. In theory, it is possible
voluting the n-th-order force gradient by (1 − u 2 )n−1/2 . to record (x, y) maps of the tip–sample force for various
The left part shows the weight functions for Δ f (dark z-values and calculate gradients and higher-order gradi-
brown), a3 (brown) and a10 (light brown) ents from these maps later. However, experimental noise
726 Part C Scanning-Probe Microscopy
a) b) c) d)
∂ Fts
(N/m)
Vts (eV) Fts (nN) ∂z
–6 –2 0
–7 –6 –100
–8
–9 –10 –200
–10 –14 –300
–11
–18 –400
–150 –150 –150 –150 –150 –150
–100 –100 –100 –100 –100 –100
50 –50 50 –50 50 –50
0 0 0 0 0 0
50 50 50 50 50 50
100 100 100 100 100 100
y (pm) 150 150 x (pm) y (pm) 150 150 x (pm) y (pm) 150 150 x (pm)
e) f) g) h)
∂ 2Fts 12 ∂ 3Fts 23 ∂ 4Fts 34 ∂ 5Fts 45
(10 N/m2) (10 N/m3) (10 N/m4) (10 N/m5)
∂z 2 ∂z 3 ∂z 4 ∂z 5
2
0 1 1
0 0
–4 0
–2 –1
–8 –4 –2 –1
–12 –3
–6 –2
–4
–16
–150 –150 –150 –150 –150 –150 –150
–100 –100 –100 –100 –100 –100 –100 –100
50 –50 50 –50 50 –50 50 –50
0 0 0 0 0 0 0 0
50 50 50 50 50 50 50 50
100 100 100 100 100 100 100 100
y (pm) 150 150 x (pm) y (pm) 150 150 x (pm) y (pm) 150 150 x (pm) y (pm) 150 150 x (pm)
Fig. 25.16 (a) Simple electrostatic model for the tip–sample interaction in AFM: the tip is a test charge with −e, the sample
Part C 25.4
atom that is probed consists of a central ion with charge +9e and the valence charge distribution is modeled by eight surrounding
charges −e oriented towards the corners of a cube. The tip atom is located 132 pm above the center of the sample ion, and the point
charges are located at a distance of 55 pm from the center of the sample ion. (b) Tip–sample interaction potential for a distance of
132 pm as a function of the lateral positions x and y. (c) Tip–sample force, (d) force gradient, (e) second derivative of force versus
z, (f) third derivative, (g) fourth derivative, and (h) fifth derivative of force versus z
would render these maps totally useless for obtaining the base frequency, both the deflection signal and the
reliable data for higher-order gradients. Instead, a direct electrical signal look like the solid curve in Fig. 25.17.
method that couples higher-order gradients to experi- However, in piezoelectric detectors such as the qPlus
mental observables is needed. The higher harmonics sensor [25.33], a charge accumulates at the electrodes
described in (25.10) are perfect for this purpose. How- of the cantilever that is proportional to the deflection
ever, the magnitude of higher harmonics is typically ([25.33, Eq. 2]). When the sensor oscillates, an alter-
very small, therefore they have to be enhanced either nating current results that is proportional to deflection
by resonant enhancement as described in the previous times frequency. Typically, this alternating current is
sections or by using small amplitudes and enhancing transformed to a voltage by a transimpedance ampli-
the magnitude of higher harmonics by other means as fier [25.33]. Because of the proportionality between
outlined below. current and deflection times frequency, higher harmon-
The mechanical bending of AFM cantilevers is ics generate greater signal strength in piezoelectric
transformed into an electrical signal by a deflection sensors. The dashed curve in Fig. 25.17 shows the
sensor. Optical and piezoresistive deflection sensors current that is generated by a qPlus sensor oscillat-
generate an output signal that is proportional to the ing at a base frequency and the fifth harmonic with
cantilever deflection. For example, if the mechanical a relative amplitude of 4%. Due to the enhancement
deflection is composed of a sinusoidal oscillation with of the sensitivity for higher harmonics, the current at
frequency f and amplitude A plus an oscillation at frequency 5 × f already amounts to 20% of the base
5 × f with an amplitude of 4% of the amplitude A at frequency.
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy 25.4 Higher-Harmonics with Small Amplitudes 727
q´/A, I / I0 Proposed
1.5 Current Higher harmonics atom orientation
q´
1 I
0.5
–1
–1.5
0 0.5 1 1.5 2
t×f
Fig. 25.17 Deflection of a cantilever versus time (solid)
for a cantilever according to q (t) = A[cos(2π ft) + 200 pm 200 pm
0.04 cos(2π5 ft)] and current (dashed) I (t) = −const A
[sin(2π ft) + 0.2 sin(2π5 ft)] produced by a piezoelectric
force detector such as the qPlus sensor [25.33]. Because
piezoelectric detection is proportional to frequency, higher
harmonics produce more current than the oscillation at the
base frequency f
200 pm 200 pm
In principle, one can pick an individual higher har-
monic by analyzing the deflection of the cantilever Fig. 25.18 Simultaneous images of tunneling current
with a lock-in amplifier that is triggered at the base (left column) and higher harmonics (center column) in
Part C 25.4
frequency f and set to the n-th harmonic of f . Of a constant-height measurement. A graphite sample was im-
course, one could also apply a battery of lock-in aged by a tungsten tip in a 4 K STM/AFM in ultrahigh
amplifiers and record as many higher harmonics as vacuum [25.34] for details. The highly resolved details
practical for full potential recovery, as proposed by in the higher-harmonic images are caused by the elec-
Dürig [25.29]. Because the higher force gradient im- tronic structure of the tungsten tip atom. Because tungsten
ages in Fig. 25.16e–h are very similar, we can just crystallizes with body-centered cubic (bcc) symmetry, the
acquire the root-mean-square (RMS) sum of all higher high-symmetry configurations are the twofold [110] sym-
harmonics by routing the deflection signal through metry shown in the first row, the threefold [111] symmetry
a high-pass filter that is set to pass all frequencies above shown in the second row, and the fourfold [001] sym-
f followed by an RMS-to-direct current converter. metry shown in the third row. The right column shows
While this high-pass technique does not allow imme- the Wigner–Seitz unit cell of bcc materials such as tung-
diate full potential recovery as proposed in [25.19], sten. Typical acquisition speed for each set of images is
it has a good signal-to-noise ratio because it uses all 0.5 lines/s at 256 × 256 pixels, i. e., a typical time of 10 min
the higher harmonics and it is very simple to im- per image
plement and operate. This technique has been used
in [25.34]. column maps the tunneling current, and the center col-
umn shows the intensity of the higher-harmonics data,
25.4.2 Application Examples clearly showing the enhanced spatial resolution. The
intensity of the higher harmonics can not only be mon-
The enhanced resolution power that should be avail- itored in constant-height mode or when the feedback
able by higher-harmonic AFM with small amplitude uses a tunneling current or frequency shift for distance
is best demonstrated by a direct comparison. Fig- regulation, but can also be used for z-feedback. Fig-
ure 25.18 shows simultaneous constant-height data ure 25.19 shows an image of a Si(111)-(7 × 7) surface
where a graphite sample was imaged by a tungsten that has been acquired by higher-harmonic AFM. The
tip mounted on an oscillating qPlus sensor. The left deflection signal of the cantilever was fed into a lock-
728 Part C Scanning-Probe Microscopy
10 nm
spatial resolution. The signal levels are small, therefore
the acquisition speed is very low, and low-temperature
operation is helpful to reduce thermal drift to low values
in amplifier, and the second harmonic was used as the such that it does not harm the spatial resolution. Possi-
signal. bly, a combination of small-amplitude higher-harmonic
AFM with the resonance-enhancement technique de-
25.4.3 Conclusions scribed in the previous section might allow to retain the
high-resolution capability of small-amplitude higher-
Higher-harmonic AFM with small amplitudes is an in- harmonic AFM while increasing the possible scanning
teresting AFM mode because it allows greatly increased speed.
References
25.1 Q. Zhong, D. Inniss, K. Kjoller, V.B. Elings: Fractured 25.9 R.W. Stark, W.M. Heckl: Fourier transformed atomic
polymer/silica fiber surface studied by tapping force microscopy: Tapping mode atomic force mi-
Part C 25
mode atomic force microscopy, Surf. Sci. 280, croscopy beyond the Hookian approximation, Surf.
L688–L692 (1993) Sci. 457, 219–228 (2000)
25.2 D. Klinov, S. Magonov: True molecular resolution in 25.10 U. Rabe, K. Janser, W. Arnold: Vibrations of free
tapping mode atomic force microscopy, Appl. Phys. and surface-coupled atomic force microscope can-
Lett. 84, 2697–2699 (2004) tilevers: Theory and experiment, Rev. Sci. Instrum.
25.3 M.V. Salapaka, D.J. Chen, J.P. Cleveland: Linear- 67, 3281–3293 (1996)
ity of amplitude and phase in tapping-mode 25.11 J. Legleiter, M. Park, B. Cusick, T. Kowalewski:
atomic force microscopy, Phys. Rev. B 61, 1106–1115 Scanning probe acceleration microscopy (SPAM) in
(2000) fluids: Mapping mechanical properties of surfaces
25.4 O. Sahin, A. Atalar, C.F. Quate, O. Solgaard: Reso- at the nanoscale, Proc. Natl. Acad. Sci. USA 103,
nant harmonic response in tapping-mode atomic 4813 (2006)
force microscopy, Phys. Rev. B 69, 5416–5424 25.12 J. Preiner, J.L. Tang, V. Pastushenko, P. Hinter-
(2004) dorfer: Higher harmonic atomic force microscopy:
25.5 R. Hillenbrand, M. Stark, R. Guckenberger: Imaging of biological membranes in liquid, Phys.
Higherharmonics generation in tapping-mode Rev. Lett. 99, 046102 (2007)
atomic force microscopy: Insights into tip–sample 25.13 S. Basak, A. Raman: Dynamics of tapping mode
interaction, Appl. Phys. Lett. 76, 3478–3480 atomic force microscopy in liquids: Theory and
(2000) experiments, Appl. Phys. Lett. 91, 064107 (2007)
25.6 R.W. Stark, W.M. Heckl: Higher harmonics imaging 25.14 O. Sahin, G. Yaralioglu, R. Grow, S.F. Zappe, A. Ata-
in tapping-mode atomic-force microscopy, Rev. lar, C. Quate, O. Solgaard: High resolution imaging
Sci. Instrum. 74, 5111–5114 (2003) of elastic properties using harmonic cantilevers,
25.7 S. Crittenden, A. Raman, R. Reifenberger: Probing Sens. Actuators A 114, 183–190 (2004)
attractive forces at the nanoscale using higher- 25.15 H. Li, Y. Chen, L. Dai: Concentrated-mass cantilever
harmonic dynamic force microscopy, Phys. Rev. B enhances multiple harmonics in tapping-mode
72(13), 235422 (2005) atomic force microscopy, Appl. Phys. Lett. 92,
25.8 A.S. Paulo, R. Garcia: Unifying theory of tapping 151903 (2008)
mode atomic force microscope, Phys. Rev. B 66, 25.16 S. Sadewasser, G. Villanueva, J.A. Plaza: Modified
041406–041409(R) (2002) atomic force microscopy cantilever design to facili-
Higher Harmonics and Time-Varying Forces in Dynamic Force Microscopy References 729
tate access of higher modes of oscillation, Rev. Sci. 25.26 I.M. Ward, J. Sweeney: An Introduction to the
Instrum. 77, 073703 (2006) Mechanical Properties of Solid Polymers (Wiley,
25.17 K. Kimura, K. Kobayashi, K. Matsushige, H. Ya- Chichester 2004)
mada: Improving sensitivity in electrostatic force 25.27 U. Dürig: Relations between interaction force,
detection utilizing cantilever with tailored reso- frequency shift in large-amplitude dynamic
nance modes, Appl. Phys. Lett. 90, 053113 (2007) force microscopy, Appl. Phys. Lett. 75, 433–435
25.18 O. Sahin, S. Magonov, C. Su, C.F. Quate, O. Sol- (2004)
gaard: An atomic force microscope tip designed 25.28 S. Morita, F.J. Giessibl, Y. Sugawara, H. Hosoi,
to measure time-varying nanomechanical forces, K. Mukasa, A. Sasahara, H. Onishi: Noncontact
Nat. Nanotechnol. 2, 507 (2007) atomic force microscopy and related topics. In:
25.19 O. Sahin: Time-varying tip-sample force mea- Springer Handbook of Nanotechnology, 3rd edn.,
surements and steady-state dynamics in tapping- ed. by B. Bhushan (Springer, Berlin Heidelberg
mode atomic force microscopy, Phys. Rev. B 77, 2010) Chap. 23
115405 (2008) 25.29 U. Dürig: Interaction sensing in dynamic force mi-
25.20 M. Stark, R.W. Stark, W.M. Heckl, R. Guckenberger: croscopy, New J. Phys. 2, 5.1–5.12 (2000)
Inverting dynamic force microscopy: From signals 25.30 A. de Lozanne: Music of the spheres at the atomic
to time resolved forces, Proc. Natl. Acad. Sci. USA scale, Science 305, 348 (2004)
99, 8473–8478 (2002) 25.31 M. Posternak, H. Krakauer, A.J. Freeman,
25.21 J.L. Hutter, J. Bechhoefer: Calibration of atomic- D.D. Koelling: Self-consistent electronic struc-
force microscope tips, Rev. Sci. Instrum. 64, 1868 ture of surfaces: Surface states, surface reso-
(1993) nances on W(001), Phys. Rev. B 21, 5601–5612
25.22 O. Sahin: Harnessing bifurcations in tapping-mode (1980)
atomic force microscopy to calibrate time-varying 25.32 F. Mattheiss, D.R. Hamann: Electronic structure of
tip-sample force measurements, Rev. Sci. Instrum. the tungsten (001) surface, Phys. Rev. B 20, 5372–
78, 103707 (2007) 5381 (1984)
25.23 L. Zitzler, S. Herminghaus, F. Mugele: Capillary 25.33 F.J. Giessibl: Atomic resolution on Si(111)-(7×7) by
forces in tapping-mode atomic force microscopy, noncontact atomic force microscopy with a force
Phys. Rev. B 66, 155436–155443 (2002) sensor based on a quartz tuning fork, Appl. Phys.
25.24 I.N. Sneddon: The relation between load and pen- Lett. 76, 1470–1472 (2000)
Part C 25
etration in the axisymmetric Boussinesq problem 25.34 S. Hembacher, F.J. Giessibl, J. Mannhart: Force mi-
for a punch of arbitrary profile, Int. J. Eng. Sci. 3, croscopy with light-atom probes, Science 305, 380
47–57 (1965) (2004)
25.25 J.N. Isrelachvili: Intermolecular and Surface Forces 25.35 F.J. Giessibl: Higher-harmonic atomic force mi-
(Academic, London 2003) croscopy, Surf. Interface Anal. 38, 1696–1701 (2006)
731
Dynamic Mod
26. Dynamic Modes of Atomic Force Microscopy
Part C 26
separates the purely attractive force interaction quality, or Q-factor, and extend the regime of sta-
regime from the attractive–repulsive regime. Sec- ble operation. This so-called Q-control module
ond, the self-excitation mode is derived and allows one to increase as well as decrease the Q-
its experimental realization is outlined. While factor. Apart from the advantages of dynamic force
the tapping mode is primarily used for imag- microscopy as a nondestructive, high-resolution
ing in air and liquid, the self-excitation mode imaging method, it can also be used to obtain in-
is typically used under ultrahigh vacuum (UHV) formation about energy-dissipation phenomena
conditions for atomic-resolution imaging. In par- at the nanometer scale. This measurement chan-
ticular, we explain the influence of different forces nel can provide crucial information on electric
on spectroscopy curves obtained in dynamic force and magnetic surface properties. Even atomic-
microscopy. A quantitative link between the ex- resolution imaging has been obtained in the
perimental spectroscopy curves and the interaction dissipation mode. Therefore, in the last section,
forces is established. the quantitative relation between the experi-
Force microscopy in air suffers from small mental measurement channels and the dissipated
quality factors of the force sensor (i. e., the power is derived.
732 Part C Scanning-Probe Microscopy
Since van der Waals forces range over typical distances However, we have neglected one important issue
of 10 nm, the vibration amplitude should be less than for the operation of the AFM thus far: the mechan-
1 nm. Furthermore, we require that the force gradient ical stability of the measurement. In static AFM,
∂Fts /∂z does not vary significantly over one oscillation the tip is slowly approached toward the surface. The
cycle. We can view the AFM setup as a coupling of force between the tip and the surface will always be
two springs (Fig. 26.1). Whereas the cantilever is rep- counteracted by the restoring force of the cantilever.
resented by a spring with spring constant k, the force Figure 26.2 shows a typical force–distance curve. Upon
interaction between the tip and the surface can be mod- approach of the tip toward the sample, the negative at-
eled by a second spring. The derivative of the force with tractive forces, representing van der Waals or chemical
respect to the tip–sample distance is the force gradient interaction forces, increase until a maximum is reached.
and represents the spring constant kts of the interaction This turnaround point is due to the onset of repulsive
spring. This spring constant kts is constant only with forces caused by Coulomb repulsion, which will start
respect to one oscillation cycle, but varies with the av- to dominate upon further approach. The spring con-
erage tip–sample distance as the probe is approached to stant of the cantilever is represented by the slope of the
the sample. The two springs are effectively coupled in straight line. The position of the z-transducer (typically
parallel, since sample and tip support are rigidly con- a piezoelectric element), which moves the probe, is at
nected for a given value of z 0 . Therefore, we can write the intersection of the line with the horizontal axis. The
for the total spring constant of the AFM system position of the tip, shifted from the probe’s base due to
the lever bending, can be found at the intersection of
∂Fts
ktotal = k + kts = k − . (26.2) the cantilever line with the force curve. Hence, the to-
∂z
From the simple harmonic oscillator (neglecting any Force
damping effects) we find that the resonant frequency ω
of the system is shifted by Δω from the free resonant Static mode
frequency ω0 due to the force interaction Dynamic mode
k k + ∂F
∂z
ts
total
ω2 = (ω0 + Δω)2 = ∗ = . (26.3)
m m∗ Probe–sample
distance z0
Herem∗ represents the effective mass of the can- B
tilever. A detailed analysis of how m ∗ is related to the A
geometry and total mass of the cantilever can be found
in the literature [26.3]. In the approximation that Δω is
Part C 26.1
much smaller than ω0 , we can write C
Δω ∼ 1 ∂Fts
=− . (26.4) D
ω0 2k ∂z Oscillation amplitude
Therefore, we find that the frequency shift of the
cantilever resonance is proportional to the force gradi- Fig. 26.2 Force–distance curve of a typical tip–sample in-
ent of the tip–sample interaction. teraction. In static-mode AFM the tip would follow the
Although the above consideration is based on force curve until point B is reached. If the slope of the
a highly simplified model, it shows qualitatively that in force curve becomes larger than the spring constant of the
dynamic force microscopy we will find that the oscilla- cantilever (dashed line) the tip will suddenly jump to po-
tion frequency depends on the force gradient, whereas sition C. Upon retraction a different path will be followed
static force microscopy measures the force itself. In along D and A again. In dynamic AFM the cantilever os-
principle, we can calculate the force curve from the cillates with amplitude. Although the equilibrium position
force gradient and vice versa (neglecting a constant of the oscillation is far from the surface, the tip will experi-
offset). It seems, therefore, that the two methods are ence the maximum attractive force at point D during some
equivalent, and our choice will depend on whether we parts of the oscillation cycle. However, the total force is al-
can measure the beam deflection or the frequency shift ways pointing away from the surface, therefore avoiding
with better precision at the cost of technical effort. an instability
734 Part C Scanning-Probe Microscopy
tal force is zero, i. e., the cantilever is in its equilibrium Consider the first solution, which seems simpler at
position (note that the spring constant line here shows first glance. Chemical bonding forces extend over a dis-
attractive forces, although in reality the forces are re- tance range of about 0.1 nm. Typical binding energies
pulsive, i. e., pulling the tip back from the surface). As of a couple of eV will lead to adhesion forces on the or-
soon as position A in Fig. 26.2 is reached, we find two der of some nN. Force gradients will, therefore, reach
possible intersection points, and upon further approach values of some 10 N/m. A spring for stable force mea-
there are even three force equilibrium points. However, surements will have to be as stiff as 100 N/m to ensure
between points A and B the tip is at a local energy min- that no instability occurs (a safety factor of ten seems
imum and, therefore, will still follow the force curve. to be a minimum requirement, since usually one cannot
However, at point B, when the adhesion force upon be sure a priori that only one atom will dominate the in-
further approach would become larger than the spring teraction). In order to measure the nN interaction force,
restoring force, the tip will suddenly jump to point C. a static cantilever deflection of 0.01 nm has to be de-
We can then probe the predominantly repulsive force tected. With standard beam deflection AFM setups this
interaction by further reducing the tip–sample distance. becomes a challenging task.
When retracting the tip, we will pass point C, because This problem was solved by using an in situ optical
the tip is still in a local energy minimum. Only at posi- interferometer measuring the beam deflection at liquid-
tion D will the tip jump suddenly to point A again, since nitrogen temperature in a UHV environment [26.4, 5].
the restoring force now exceeds the adhesion. From In order to ensure that the force gradients are smaller
Fig. 26.2 we can see that the sudden instability will hap- than the lever spring constant (50 N/m), the tips were
pen at exactly the point where the slope of the adhesion fabricated to terminate in only three atoms, thereby
force exceeds the slope of the spring constant. There- minimizing the total force interaction. The field ion mi-
fore, if the negative force gradient of the tip–sample croscope (FIM) is a tool which allows scanning probe
interaction will at any point exceed the spring constant, microscopy (SPM) tips to be engineered down to atomic
a mechanical instability occurs. Mathematically speak- dimensions. This technique not only allows imaging
ing, we demand that for a stable measurement of the tip apex with atomic precision, but also can
be used to manipulate the tip atoms by field evapora-
∂Fts tion [26.6], as shown in Fig. 26.3. Atomic interaction
− < k , for all points z . (26.5)
∂z z
Part C 26.1
tions, amplitudes of 10–100 nm are used to stay on the sections is a measure of the bending.
safe side. This means that the oscillation amplitude is In order to enhance sensitivity, several groups have
much larger than the force interaction range. The above adopted an interferometer system to measure the can-
simplification, that the force gradient remains constant tilever deflection. A thorough comparison of different
within one oscillation cycle, does not hold anymore. measurement methods with analysis of sensitivity and
Measurement stability is gained at the cost of a sim- noise level is given in reference [26.3].
ple quantitative analysis of the experiments. In fact, The cantilever is mounted on a device that al-
dynamic AFM was first used to obtain atomic resolu- lows the beam to be oscillated. Typically a piezo
tion images of clean surfaces [26.8], and it took another element directly underneath the cantilever beam serves
6 years [26.1] before quantitative measurements of sin- this purpose. The reflected laser beam is analyzed
gle bond forces were obtained. for oscillation amplitude, frequency, and phase differ-
The technical realization of dynamic-mode AFMs ence. Depending on the mode of operation, a feedback
is based on the same key components as a static AFM mechanism will adjust oscillation parameters and/or
setup. The most common principle is the method of tip–sample distance during the scanning. The setup can
laser deflection sensing (Fig. 26.4). A laser beam is fo- be operated in air, UHV, and even fluids. This allows
cused on the back side of a microfabricated cantilever. measurement of a wide range of surface properties from
The reflected laser spot is detected with a position- atomic-resolution imaging [26.8] up to studying biolog-
sensitive diode (PSD). This photodiode is sectioned ical processes in liquid [26.9, 10].
736 Part C Scanning-Probe Microscopy
solution for amplitude and phase of the oscillation as ω∗0 . As we will discuss later, it is also possible to en-
a function of the excitation frequency ω: hance Q by using a special excitation method called
Q-control.
Ad Qω20
A0 = 2 , (26.12) In the case that the excitation frequency is equal to
ω2 ω20 + Q 2 ω20 − ω2 the resonant frequency of the undamped cantilever ω =
ω0 , we find the useful relation
ωω0
ϕ = arctan . (26.13) A0 = Q Ad , for ω = ω0 . (26.15)
Q ω20 − ω2
Amplitude and phase diagrams are depicted in Since ω∗0 ≈ ω0 for most cases, we find that (26.15) holds
Fig. 26.5. As can be seen from (26.12), the amplitude true for exciting the cantilever at its resonance. From
will reach its maximum at a frequency different from a similar argument, the phase becomes approximately
ω0 if Q has a finite value. The damping term of the har- 90◦ for the resonance case. We also see that, in order
monic oscillator causes the resonant frequency to shift to reach vibration amplitudes of some 10 nm, the ex-
from ω0 to ω∗0 citation only has to be as small as 1 pm, for typical
cantilevers operated in vacuum.
1 So far we have not considered an additional force
∗
ω0 = ω0 1 − . (26.14) term, describing the interaction between the probing tip
2Q 2
and the sample. For typical, large vibration amplitudes
The shift is negligible for Q-factors of 100 and of 10–100 nm the tip experiences a whole range of force
above, which is the case for most applications in vac- interactions during one single oscillation cycle, rather
uum or air. However, for measurements in liquids, Q than one defined tip–sample force. How this problem
can be smaller than 10 and ω0 differs significantly from can be attacked will be shown in the next paragraphs.
Part C 26.3
a viable direction toward achieving this task. However gerprint of the tip–sample interactions.
interpretation of the measurements generally appears to In dynamic AFM the situation is rather complex.
be more difficult. Different operational modes are em- Any of the internal parameters can be used for feedback
ployed in dynamic AFM, and the following paragraphs of the tip–sample distance z 0 . However, we already re-
are intended to distinguish these modes and categorize alized that, in general, the tip–sample forces could only
them in a systematic way. be fully assessed by measuring all three parameters.
The oscillation trajectory of a dynamically driven Therefore, dynamic AFM images are difficult to inter-
cantilever is determined by three parameters: the am- pret. A solution to this problem is to establish additional
plitude, the phase, and the frequency. Tip–sample feedback loops, which keep the internal parameters con-
interactions can influence all three parameters, in the stant by adjusting the external variables. In the simplest
following, termed the internal parameters. The oscilla- setup, the excitation frequency and the excitation am-
tion is driven externally, with excitation amplitude Ad plitude are set to predefined values. This is the so-called
and excitation frequency ω. These variables will be amplitude-modulation (AM) mode or tapping mode. As
referred to as the external parameters. The external stated before, in principle, any of the internal parame-
parameters are set by the experimentalist, whereas ters can be used for feedback to the tip–sample distance
the internal parameters are measured and contain the – in AM mode the amplitude signal is used. A certain
crucial information about the force interaction. In scan- amplitude (smaller than the free oscillation amplitude)
ning probe applications, it is common to control the at a frequency close to the resonance of the cantilever
probe–surface distance z 0 in order to keep an internal is chosen, the tip is approached toward the surface un-
738 Part C Scanning-Probe Microscopy
der investigation, and the approach is stopped as soon as Since the phase remains at a fixed value, the oscillat-
the setpoint amplitude is reached. The oscillation phase ing system is much better defined than before, and the
is usually recorded during the scan; however, the shift degrees of freedom for the oscillation are reduced. To
of the resonant frequency of the cantilever cannot be di- even reduce the last degree of freedom an additional
rectly accessed, since this degree of freedom is blocked feedback loop can be incorporated to keep the oscil-
by the external excitation at a fixed frequency. It turns lation amplitude A constant by varying the excitation
out that this mode is simple to operate from a technical amplitude Ad . Now, all internal parameters have a fixed
perspective. Therefore, it is one of the most commonly relation to the external excitation variables, the system
used modes in dynamic AFM operated in air, and even is well defined, and all parameters can be assessed dur-
in liquid. The strength of this mode is the easy and ing the measurement.
reliable high-resolution imaging of a large variety of In the following section we want to discuss the two
surfaces. most popular operational modes, tapping mode and self-
It is interesting to discuss the AM mode in the sit- excitation mode, in more detail.
uation that the external excitation frequency is much
lower than the resonant frequency [26.12, 13]. This 26.3.1 Amplitude-Modulation/
results in a quasistatic measurement, although a dy- Tapping-Mode AFM
namic oscillation force is applied, and therefore this
mode can be viewed as a hybrid between static and In tapping mode, or AM-AFM, the cantilever is excited
dynamic AFM. Unfortunately, it has the drawbacks of externally at a constant frequency close to its reso-
the static mode, namely that stiff spring constants must nance. Oscillation amplitude and phase during approach
be used and therefore the sensitivity of the deflection of tip and sample serve as the experimental observa-
measurement must be very good, typically employing tion channels. Figure 26.6 shows a diagram of a typical
a high-resolution interferometer. Still, it has the advan- tapping-mode AFM setup. The oscillation of the can-
tage of the static measurement in terms of quantitative
interpretation, since in the regime of small amplitudes
(< 0.1 nm) direct interpretation of the experiments is Detector Lock-in amplifier
possible. In particular, the force gradient at tip–sample Mirror Phase
Detector Amplitude
distance z 0 is given by the change of the amplitude A signal
and the phase angle ϕ Laser
∂Fts A0
= k 1 − cos ϕ . (26.16)
∂z
z0 A Reference
signal
Part C 26.3
tilever is detected with the photodiode, whose output related to the cooling process during the polymer prepa-
signal is analyzed with a lock-in amplifier to obtain am- ration.
plitude and phase information. The amplitude is then Imaging in liquids opens up the possibility of
compared with the setpoint, and the resulting difference the investigation of biological samples in their nat-
or error signal is fed into the proportional–integral– ural environment. For example Möller et al. [26.17]
differential (PID) controller, which adjusts the z-piezo, have obtained high-resolution images of the topogra-
i. e., the probe–sample distance, accordingly. The exter- phy of hexagonally packed intermediate (HPI) layer
nal modulation unit supplies the signal for the excitation of Deinococcus radiodurans with tapping-mode AFM.
piezo, and at the same time the oscillation signal serves Another interesting example is the imaging of DNA
as the reference for the lock-in amplifier. As shown by in liquid, as shown in Fig. 26.8. Jiao et al. [26.10]
the following applications the tapping mode is typically measured the time evolution of a single DNA strand in-
used to measure surface topography and other material teracting with a molecule as shown by a sequence of
parameters on the nanometer scale. The tapping mode images acquired in liquid over a time period of several
is mostly used in ambient conditions and in liquids. minutes.
High-resolution imaging has been extensively per- For a quantitative interpretation of tip–sample
formed in the area of materials science. Due to its forces one has to consider that during one oscillation
technical relevance the investigation of polymers has cycle with amplitudes of 10–100 nm the tip–sample in-
been the focus of many studies (see, e.g., a recent review teraction will range over a wide distribution of forces,
about AFM imaging on polymers by Magonov [26.16]). including attractive as well as repulsive forces. We
In Fig. 26.7 the topography of a diblock copolymer will, therefore, measure a convolution of the force–
(BC0.26 -3A0.53 F8 H10 ) at different magnifications is distance curve with the oscillation trajectory. This
shown [26.15]. On the large scan (Fig. 26.7a) the large- complicates the interpretation of AM-AFM measure-
scale structure of the microphase-separated polystyrene ments appreciably.
(PS) cylinders (within a polyisoprene (PI) matrix) lying At the same time, the resonant frequency of the
parallel to the substrate can be seen. In the high- cantilever will change due to the appearing force gra-
resolution image (Fig. 26.7b) a surface substructure of dients, as could already be seen in the simplified model
regular domes can be seen, which were found to be in (26.4). If the cantilever is excited exactly at its reso-
a) µm b) nm
1.5 500
Part C 26.3
1
250
0.5
0 0
0 0.5 1 1.5 0 250 500
µm nm
Fig. 26.7a,b Tapping-mode images of BC0.26 –3A0.53 F8 H10 at (a) low resolution and (b) high resolution. The height
scale is 10 nm (after [26.15],
c American Chemical Society, 2001)
740 Part C Scanning-Probe Microscopy
nant frequency before, it will be excited off resonance dered from indenting further into the sample, resulting
after interaction forces are encountered. This, in turn, in smaller oscillation amplitudes.
changes the amplitude and phase in (26.12) and (26.13), However, in order to gain some qualitative insight
which serve as the measurement signals. Consequently, into the complex relationship between forces and oscil-
a different amplitude will cause a change in the en- lation parameters, we resort to numerical simulations.
countered effective force. We can see already from Anczykowski et al. [26.18, 19] have calculated the os-
this simple gedanken experiment that the interpreta- cillation trajectory of the cantilever under the influence
tion of the measured phase and amplitude curves is not of a given force model. van der Waals interactions
straightforward. were considered the only effective attractive forces,
The qualitative behavior for amplitude versus z 0 - and the total interaction resembled a Lennard–Jones-
position curves is depicted in Fig. 26.9. At large type potential. Mechanical relaxations of the tip and
distances, where the forces between tip and sample are sample surface were treated in the limits of contin-
negligible, the cantilever oscillates with its free oscil- uum theory with the numerical Muller–Yushchenko–
lation amplitude. Upon approach of the probe toward Derjaguin/Burgess–Hughes–White MYD/BHW [26.20,
the surface the interaction forces cause the amplitude 21] approach, which allows the simulations to be com-
to change, typically resulting in an amplitude that gets pared with experiments.
smaller with continuously decreased tip–sample dis- The cantilever trajectory was analyzed by numeri-
tance. This is expected, since the force–distance curve cally solving the differential equation (26.7) extended
will eventually reach the repulsive part and the tip is hin- by the tip–sample force. The results of the simulation
for the amplitude and phase of the tip oscillation as
a function of z-position of the probe are presented in
a) 4'16'' b) 6'13''
Fig. 26.10. One has to keep in mind that the z-position
of the probe is not equivalent to the real tip–sample dis-
tance at equilibrium position, since the cantilever might
bend statically due to the interaction forces. The be-
havior of the cantilever can be subdivided into three
different regimes. We distinguish the cases in which the
beam is oscillated below its resonant frequency ω0 , ex-
actly at ω0 , and above ω0 . In the following, we will refer
c) 12'49'' d) 14'22''
Amplitude
Part C 26.3
Setpoint
a) ω < ω0 b) ω = ω0 c) ω > ω0
Amplitude (nm) Amplitude (nm) Amplitude (nm)
25 25 25
dret
20 20 20
dapp
15 15 15
dret 25 dret
10 dapp 10 24 10
dret dapp
5 5 23 dapp 5
0 0 22 24 26 0
Phase (deg) Phase (deg) Phase (deg)
0 0 0
45 45 45
90 90 90
dret
dapp
135 135 dret 135
dapp
dret
180 180 180 dapp
Force (nN) Force (nN) Force (nN)
75 75 75 dapp dret
50 50 dapp 50
dret
dapp
25 25 25
dret
0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
z-position (nm) z-position (nm) z-position (nm)
Fig. 26.10a–c Amplitude and phase diagrams with excitation frequency: (a) below, (b) exactly at, and (c) above the
resonant frequency for tapping-mode AFM from numerical simulations. Additionally, the bottom diagrams show the
interaction forces at the point of closest tip–sample distance, i. e., the lower turnaround point of the oscillation
Part C 26.3
to ω0 as the resonant frequency, although the correct which again gives rise to a stronger attractive force. The
resonant frequency is ω∗0 if taking into account the finite system becomes unstable until the point z 0 = dapp is
Q-value. reached, where repulsive forces stop the self-enhancing
Clearly, Fig. 26.10 exhibits more features than were instability. This can be clearly observed in Fig. 26.10a.
anticipated from the initial, simple arguments. Ampli- Large parts of the force–distance curve cannot be meas-
tude and phase seem to change rather abruptly at certain ured due to this instability.
points when the z 0 -position is decreased. Additionally, In the second case, where the excitation equals
we find hysteresis between approach and retraction. the free resonant frequency, only a small discontinu-
As an example, let us start by discussing the dis- ity is observed upon reduction of the z-position. Here,
continuous features in the AFM spectroscopy curves of a shift of the resonant frequency toward smaller val-
the first case, where the excitation frequency is smaller ues, induced by the attractive force interaction, will
than ω0 . Consider the oscillation amplitude as a func- reduce the oscillation amplitude. The distance between
tion of excitation frequency in Fig. 26.5. Upon approach tip and sample is, therefore, reduced as well, and
of probe and sample, attractive forces will lower the the self-amplifying effect with the sudden instability
effective resonant frequency of the oscillator. There- does not occur as long as repulsive forces are not
fore, the excitation frequency will now be closer to the encountered. However, at closer tip–sample distances,
resonant frequency, causing the vibration amplitude to repulsive forces will cause the resonant frequency to
increase. This, in turn, reduces the tip–sample distance, shift again toward higher values, increasing the ampli-
742 Part C Scanning-Probe Microscopy
45 45 45
90 dret 90 90
dapp
135 135 dret 135
dapp dret
180 180 180 dapp
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
z-position (nm) z-position (nm) z-position (nm)
Fig. 26.11a–c Amplitude and phase diagrams with excitation frequency: (a) below, (b) exactly at, and (c) above the
resonant frequency for tapping-mode AFM from experiments with a Si cantilever on a Si wafer in air
tude with decreasing tip–sample distance. Therefore, important factors necessary for an appropriate descrip-
a self-enhancing instability will also occur in this case, tion of the experimental situation.
but at the crossover from purely attractive forces to the However, what is the reason for this unexpected be-
regime where repulsive forces occur. Correspondingly, havior? We have to turn to the numerical simulations
a small kink in the amplitude curve can be observed again, where we have access to all physical parameters,
in Fig. 26.10b. An even clearer indication of this effect in order to understand the underlying processes. The
is manifested by the sudden change in the phase signal lower part of Fig. 26.10 also shows the interaction force
at dapp . between the tip and the sample at the point of closest
In the last case, with ω > ω0 , the effect of ampli-
Part C 26.3
Part C 26.3
the phase is a clear indicator of the mode switch. On the jump into the intermittent contact mode, and the previ-
other hand, if point dapp is never reached during the ap- ously explained hysteresis will tend to keep the system
proach, the system will stay in the net-attractive regime in this mode. It is, therefore, of great importance to tune
and no hysteresis is observed, i. e., the system remains the oscillation parameters in such a way that the AFM
stable. stays in the net-attractive regime [26.24]. A concept
In conclusion, we find that, although a qualita- that achieves this task is the Q-control system, which
tive interpretation of the interaction forces is possible will be discussed in some detail in the forthcoming
from the amplitude and phase curves, they do not give paragraphs.
direct quantitative knowledge of tip–sample force inter- A last word concerning the overlap of simula-
actions. However, it is a very useful tool for imaging tion and experimental data: Whereas the qualitative
nanometer-sized structures in a wide variety of setups, agreement down to the detailed shape of hysteresis
in air or even in liquid. We find that two distinct modes and instabilities is rather striking, we still find some
exist for the externally excited oscillation – the net- quantitative discrepancies between the positions of the
attractive and the intermittent contact mode – which instabilities dapp and dret . This is probably due to
describe what kind of forces govern the tip–sample in- the simplified force model, which only takes into ac-
teraction. The phase can be used as an indicator of the count van der Waals and repulsive forces. Especially
current mode of the system. at ambient conditions, an omnipresent water menis-
In particular, it can be easily seen that, if the free cus between tip and sample will give rise to much
resonant frequency of the cantilever is higher than the stronger attractive and also dissipative forces than con-
744 Part C Scanning-Probe Microscopy
sidered in the model. A very interesting feature is As shown by Hölscher [26.25] the tip–sample force
that the simulated phase curves in the intermittent can be calculated with the help of the integral equation
contact regime tend to have a steeper slope in the
A
D+2
simulation than in the experiments (Fig. 26.13). We ∂ κ(z)
will show later that this effect is a fingerprint of Fts (D) = − √ dz , (26.17a)
∂D z−D
an effect that had not been included in the above D
simulation at all: dissipative processes during the os-
where
cillation, giving rise to an additional loss of oscillation
energy. k A3/2 Ad cos ϕ ω20 − ω2
In the above paragraphs, we have outlined the in- κ= √ − . (26.17b)
2 A ω2
fluence of the tip–sample interaction on the cantilever
oscillation, calculated the maximum tip–sample interac- It is now straightforward to recover the tip–sample
tion forces based on the assumption of a specific model force using (26.17a,b) from a spectroscopy experiment,
force, and subsequently discussed possible routes for i. e., an experiment where the amplitude and the phase
image optimization. However, in practical imaging, the are continuously measured as a function of the actual
tip–sample interaction is not known a priori. In con- nearest tip–sample distance D = z 0 − A at a fixed loca-
trast, the ability to measure the continuous tip–sample tion above the sample surface. With this input, one first
interaction force as a function of both the tip–sample calculates κ as a function of D. In a second step, the
distance as well as the lateral location (e.g., in order to tip–sample force is computed by solving the integral in
identify different bond strengths on chemically inhomo- (26.17a) numerically.
geneous surfaces) would add a tool of great value to the A verification of the algorithm is shown in
force-microscopist’s toolbox. Fig. 26.14, which presents computer simulations of the
Surprisingly, despite the more than 15 years dur- method by calculating numerical solutions of the equa-
ing which the amplitude-modulation technique has tion of motion. Figure 26.14a,b shows the resulting
been used, it was only recently that two solutions to curves of amplitude and phase versus distance during
this inversion problem have been suggested [26.25, approach, respectively. The subsequent reconstruction
26]. As already discussed in the previous paragraphs, of the tip–sample interaction based on the data provided
conventional force–distance curves suffer from a jump- by the curves of amplitude and phase versus distance is
to-contact due to attractive surface forces. As a result, presented in Fig. 26.14d. The assumed tip–sample force
the most interesting range of the tip–sample force, the and energy dissipation are plotted by solid lines, while
last few nanometers above the surface, is left out, and the reconstructed data is indicated by symbols; the ex-
conventional force–distance curves thus mainly serve to cellent agreement demonstrates the reliability of the
Part C 26.3
–120 –4
6 8 10 12 –0.5 0 0.5 1 1.5 2
Cantilever position (nm) Tip–sample distance (nm)
26.3.2 Self-Excitation Modes The only degree of freedom that the oscillation system
still has that can react to the tip–sample forces is the
Despite the wide range of technical applications change of the resonant frequency. This shift of the fre-
of the AM mode of dynamic AFM, it has been quency is detected and used as the setpoint signal for
found unsuitable for measurements in an environ- surface scans. Therefore, this mode is also called the
ment extremely useful for scientific research: vacuum frequency-modulation (FM) mode.
or ultrahigh vacuum (UHV) with pressures reaching
10−10 mbar. The STM has already shown how much
Frequency
insight can be gained from experiments under those Detector demodulator
conditions. Mirror Detector signal
Consider (26.11) from the previous section. The
time constant τ for the amplitude to adjust to a dif- Laser
ferent tip–sample force scales with 1/Q. In vacuum G
Amplitude
Part C 26.3
applications, the Q-factor of the cantilever is on the control
order of 10 000, which means that τ is in the range
of some 10 ms. This time constant is clearly too long Variable
for a scan of at least 100 × 100 data points. On the Φ
phase shifter
other hand, the resonant frequency of the system will
Piezo Excitation signal
react instantaneously to tip–sample forces. This has
led Albrecht et al. [26.11] to use a modified excitation Cantilever
scheme. + tip
Sample
The system is always oscillated at its resonant fre-
quency. This is achieved by feeding back the oscillation
z-signal Error signal
signal from the cantilever into the excitation piezo el- Setpoint
ement. Figure 26.15 pictures the method in a block
x,y,z-scanner PID controller
diagram. The signal from the PSD is phase-shifted by
90◦ (and, therefore, always exciting in resonance) and Fig. 26.15 Dynamic AFM operated in the self-excitation mode,
used as the excitation signal of the cantilever. An ad- where the oscillation signal is directly fed back to the excitation
ditional feedback loop adjusts the excitation amplitude piezo. The detector signal is amplified with the variable gain G and
in such a way that the oscillation amplitude remains phase-shifted by phase φ. The frequency demodulator detects the
constant. This ensures that the tip–sample distance is frequency shift due to tip–sample interactions, which serves as the
not influenced by changes in the oscillation amplitude. control signal for the probe–sample distance
746 Part C Scanning-Probe Microscopy
Let us take a look at the sensitivity of the dynamic force gradient, i. e., the measurement sensitivity, is in-
AFM. If electronic noise, laser noise, and thermal drift versely proportional to the square root of the Q-factor
can be neglected, the main noise contribution will come of the cantilever. This means that it should be possible
from thermal excitations of the cantilever. A detailed to achieve very high-resolution imaging under vacuum
analysis of a dynamic system yields for the minimum conditions where the Q-factor is very high.
detectable force gradient the relation [26.11] A breakthrough in high-resolution AFM imaging
was the atomic resolution imaging of the Si(111)-(7 × 7)
∂F 4kkB TB surface reconstruction by Giessibl [26.8] under UHV
= . (26.18)
∂z min ω0 Q z 2osc conditions. Moreover, Sugawara et al. [26.28] observed
the motion of single atomic defects on InP with true
Here, B is thebandwidth
of the measurement, T the atomic resolution. However, imaging on conducting
temperature, and z 2osc is the mean-square amplitude of or semiconducting surfaces is also possible with the
the oscillation. Please note that this sensitivity limit was scanning tunneling microscope (STM) and these first
deliberately calculated for the FM mode. A similar anal- noncontact atomic force microscopy (NC-AFM) images
ysis of the AM mode, however, yields virtually the same provided little new information on surface properties.
result [26.27]. We find that the minimum detectable The true potential of NC-AFM lies in the imaging of
nonconducting surface with atomic precision, which
was first demonstrated by Bammerlin et al. [26.29]
a) on NaCl. A long-standing question about the surface
reconstruction of the technological relevant material
aluminum oxide could be answered by Barth and Re-
ichling [26.30], who imaged the atomic structure of the
high-temperature phase of α-Al2 O3 (0001).
The high-resolution capabilities of noncontact
atomic force microscopy are nicely demonstrated by
the images shown in Fig. 26.16. Allers et al. [26.31]
imaged steps and defects on the insulator nickel oxide
with atomic resolution. Recently, Kaiser et al. [26.32]
succeeded in imaging the antiferromagnetic structure
of NiO(001). Nowadays, true atomic resolution is rou-
tinely obtained by various research groups (for an
overview, see [26.33–36]).
Part C 26.3
Part C 26.3
dependent on the shape of the force curve F(z) but in-
dependent of the external parameters of the oscillation.
In the approximation that the vibration amplitude The function γ (z) is also referred to as the normalized
is much larger than the range of the tip–sample forces, frequency shift [26.7], a very useful parameter, which
(26.19) can be simplified to allows us to compare measurements independent of res-
onant frequency, amplitude, and spring constant of the
cantilever.
√
∞
2 3/2 F(x) The dependence of the frequency shift on the vibra-
keff (z) = A √ dx . (26.20) tion amplitude is an especially useful relation, since this
π x−z
z parameter can be easily varied during one experiment.
A nice example is depicted in Fig. 26.18, where fre-
quency shift curves for different amplitudes were found
This effective force gradient can now be used in
to collapse into one curve in the γ (z)-diagram [26.41].
(26.4), the relation between frequency shift and force
This relationship has been nicely exploited for
gradient. We find
the calibration of the vibration amplitude by Guggis-
berg [26.42], which is a problem often encountered
∞ in dynamic AFM operation and worthy of discussion.
f0 F(x)
Δf = √ √ dx . (26.21) One approaches tip and sample and records curves of
2πk A3/2 x−z frequency shift versus distance, which show a repro-
z
748 Part C Scanning-Probe Microscopy
0.4 0.4
D
Sample
0 0
180 Å 180 Å
126 Å 126 Å
–0.4 90 Å –0.4 90 Å
72 Å 72 Å
54 Å 54 Å
–0.8 –0.8
–10 –5 0 Z0 5 10 15 –10 –5 0 Z0 5 10 15
Distance D (Å) Distance D (Å)
Fig. 26.18 (a) Frequency-shift curves for different oscillation amplitudes for a silicon tip on a graphite surface in UHV,
(b) γ curves calculated from the Δ f curves in (a) (after [26.41],
c The American Physical Society)
with the z-distances. Alternatively, if the force law is not known be-
For the often encountered force contributions from forehand, the experimentalist wants to analyze the
electrostatic, van der Waals, and chemical binding experimental frequency-shift data curves and extract the
forces the frequency shift has been calculated from the force or energy potential curves. We therefore have to
force laws. In the approximation that the tip radius R invert the integral in (26.21) to find the tip–sample in-
is larger than the tip–sample distance z, an electro- teraction potential Vts from the γ (z)-curves [26.40]
static potential V will yield a normalized frequency shift
√
γ (x)
∞
of [26.43]
Vts (z) = 2 √ dx . (26.26)
πε0 RV 2 −1/2 x−z
γ (z) = √ z . (26.23) z
2 Using this method, quantitative force curves were
For van der Waals forces with Hamaker constant H extracted from Δ f spectroscopy measurements on dif-
and also with R larger than z we find accordingly ferent, atomically resolved sites of the Si(111)-(7 × 7)
HR −3/2 reconstruction [26.1]. Comparison with theoretical
γ (z) = √ z . (26.24) molecular dynamics (MD) simulations showed good
12 2 quantitative agreement with theory and confirmed the
Finally, short-range chemical forces represented by assumption that force interactions were governed by
the Morse potential (with the parameters binding en- a single atom at the tip apex. Our initially formulated
ergy U0 , decay length λ, and equilibrium distance z equ ) goal seems to be achieved: With FM-AFM we have
Dynamic Modes of Atomic Force Microscopy 26.3 Dynamic AFM Operational Modes 749
Part C 26.3
dimensional force field between tip and sample can At first glance, quantitative interpretation of the ob-
be recovered with atomic resolution. Figure 26.19b tained frequency spectra seems more complicated, since
shows a cut through the force field as a two-
dimensional map. The 3-D force technique has been
applied also to a NaCl(100) surface, where not only a) b) Potential energy (eV)
0.04
conservative but also the dissipative tip–sample in-
teraction could be measured in full space [26.49]. 0.02
On the one hand, the forces were measured in the 0
attractive as well as repulsive regime, allowing for Epot
–0.02
the determination of the local minima in the corre- z
sponding potential energy curves in Fig. 26.20. This 0 0.5 1 1.5 2
x
x-position (nm)
information is directly related to the atomic en-
ergy barriers responsible for a multitude of dynamic Fig. 26.20 (a) Three-dimensional representation of the interaction
phenomena in surface science, such as diffusion, energy map determine from 3-D force spectroscopy experiments
faceting, and crystalline growth. The direct compari- on a NaCl(100) crystal surface. The circular depressions repre-
son of conservative with the simultaneously acquired sent the local energy minima. (b) Potential energy profile obtained
dissipative processes furthermore allowed the deter- from (a) by collecting the minimum-energy values along the x-
mination of atomic-scale mechanical relaxation pro- axis. This curve thus directly reveals the potential energy barrier
cesses. of ΔE = 48 meV separating the local energy minima
750 Part C Scanning-Probe Microscopy
the amplitude as well as the tip–sample distance is al- If it is possible to measure atomic-scale forces with
tered during the measurement. However, it was found the NC-AFM, it should vice versa also be possible to
by Hölscher et al. [26.51] that for the CE mode in the exert forces with similar precision. In fact, the new
large-amplitude approximation the distance and the am- and exciting field of nanomanipulation would be driven
plitude channel can be decoupled by calculating the to a whole new dimension if defined forces could be
effective tip–sample distance from the piezo-controlled reliably applied to single atoms or molecules. In this
tip–sample distance z 0 and the change in the amplitude respect, Loppacher et al. [26.55] managed to push
with distance A(z) : D(z 0 ) = z 0 − A(z 0 ). As a result, different parts of an isolated Cu-tetra-3,5 di-tertiary-
(26.22) can then be directly used to calculate the nor- butyl-phenyl porphyrin (Cu-TBBP) molecule, which is
malized frequency shift γ (D) and consequently the known to possess four rotatable legs. They measured
force curve can be obtained from (26.26). This con- the force–distance curves while pushing one of the legs
cept has been verified in experiments by Schirmeisen with the AFM tip. From the force curves they were able
et al. [26.52] through a direct comparison of spec- to determine the energy which was dissipated during the
troscopy curves acquired in the CE mode and CA mode. switching process of the molecule. The manipulation
Until now, we have always associated the self- of single silicon atoms with NC-AFM was demon-
excitation scheme with vacuum applications. Although strated by Oyabu et al. [26.56], who removed single
it is difficult to operate the FM-AFM in constant- atoms from a Si(111)-7 × 7 surface with the AFM tip
amplitude mode in air, since large dissipative effects and could subsequently deposit atoms from the tip on
make it difficult to ensure a constant amplitude, it is in- the surface again. This technique was further improved
deed possible to use the constant-excitation FM-AFM in by Sugimoto et al. [26.57], who wrote artificial atomic
air or even in liquid [26.51, 53, 54]. Interestingly, a low- structures with single Sn atoms. The possibility to exert
budget construction set (employing a tuning-fork force and measure forces simultaneously during single atom
sensor) for a CE-mode dynamic AFM setup has been or molecule manipulation is an exciting new application
published on the internet (http://sxm4.uni-muenster.de). of high-resolution NC-AFM experiments.
26.4 Q-Control
We have already discussed the virtues of a high Q value square root of Q. In vacuum, Q mainly represents the
for high-sensitivity measurements: The minimum de- internal dissipation of the cantilever during oscillation,
tectable force gradient was inversely proportional to the an internal damping factor. Low damping is obtained by
Part C 26.4
The working principle of the feedback loop can be This means, that the additional feedback force sig-
understood by analyzing the equation of motion of the nal G eiφ z(t) is proportional to the velocity of the
modified dynamic system cantilever, just like the damping term in the equation
of motion. We can define an effective damping constant
m ∗ z̈(t) + αż(t) + kz(t) − Fts [z 0 + z(t)] αeff , which combines the two terms
= Fext cos(ωt) + G eiφ z(t) . (26.27) m ∗ z̈(t) + αeff ż(t) + kz(t) − Fts [z 0 + z(t)]
= Fext cos(ωt) ,
This ansatz takes into account the feedback of the
1 π
detector signal through a phase shifter, amplifier, and with αeff = α ∓ G , for φ = ± . (26.29)
adder as an additional force, which is linked to the ω 2
cantilever deflection z(t) through the gain G and the Equation (26.28) shows that the damping of the
phase shift eiφ . We assume that the oscillation can be oscillator can be enhanced or weakened by choos-
described by a harmonic oscillation trajectory. With ing φ = + π2 or φ = − π2 , respectively. The feedback
a phase shift of φ = ±π/2 we find loop therefore allows us to vary the effective qual-
ity factor Q eff = mω0 /αeff of the complete dynamic
1 system. Hence, this system was termed Q-control. Fig-
e±iπ/2 z(t) = ± ż(t) . (26.28) ure 26.22 shows experimental data regarding the effect
ω
of Q-control on the amplitude and phase as a function
of the external excitation frequency [26.60]. In this ex-
Amplitude (arb. units)
ample, Q-control was able to increase the Q-value by
a factor of > 40.
0.6 The effect of improved image contrast is demon-
strated in Fig. 26.23. Here, a computer hard disk was
analyzed with a magnetic tip in tapping mode, where the
0.4 magnetic contrast is observed in the phase image. The
upper part shows the magnetic data structures recorded
in standard mode, whereas in the lower part of the image
0.2
Q-control feedback was activated, giving rise to an im-
Q-control on proved signal, i. e., magnetic contrast. A more detailed
Q-control off Q = 19.379 analysis of measurements on a magnetic tape shows that
Q = 499 the signal amplitude (upper diagrams in Fig. 26.24) was
0 increased by a factor of 12.4 by the Q-control feedback.
Part C 26.4
Phase (deg) The lower image shows a noise analysis of the signal,
indicating an improvement of the signal-to-noise ratio
0
by a factor of 2.3.
It might be interesting to note that Q-control can
Q-control on
Q-control off Q = 19.379 also be applied in FM mode, which might be coun-
Q = 499 terintuitive at first sight. However, it has been shown
90 by Ebeling et al. [26.61, 62] that the increase of the
Q-factor in liquids helps to increase the imaging fea-
tures of the FM mode in liquids.
The diagrams represent measurements in air with
an AFM operated in AM mode. Only then can we
180 make a distinction between excitation and vibration fre-
quency, since in the FM mode these two frequencies are
259.5 260 260.5 261 261.5
equal by definition. Although the relation between sen-
Frequency (kHz)
sitivity and Q-factor in (26.17a) is the same for AM
Fig. 26.22 Amplitude and phase diagrams measured in air and FM mode, it must be critically investigated to see
with a Si cantilever far away from the sample. The quality whether the enhanced quality factor by Q-control can
factor can be increased from 450 to 20 000 by using the be inserted into the equation for FM-mode AFM. In
Q-control feedback method vacuum applications, Q is already very high, which
752 Part C Scanning-Probe Microscopy
0.25
0
0 0.25 0.5 0.75 1
Q-control off Q-control on
60
41
50
40 3.7 46
40
39 30
20
38
3 3
2 2
0.9%
2.1%
1 1
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
x-position (nm) x-position (nm)
Fig. 26.23 Enhancement of the contrast in the phase channel due to Q-control on a magnetic hard disk measured with
a magnetic tip in tapping-mode AFM in air. Scan size 5 × 5 μm, phase range 10 (www.nanoanalytics.com) (MFM –
Part C 26.4
Q-control off
makes it unnecessary to operate an additional Q-control
module.
As stated before, we can also use Q-control to
enhance the damping in the oscillating system. This
would decrease the sensitivity of the system. However,
on the other hand, the response time of the amplitude
change is decreased as well. For tapping-mode applica-
tions, where high-speed scanning is the goal, Q-control
was able to reduce the scan speed limiting relaxation
time [26.63].
A large quality factor Q does not only have the
virtue of increasing the force sensitivity of the in-
Part C 26.4
h ((nm))
Height Q-control off H i h ((nm))
Height Q-control on
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
x-position (nm) x-position (nm)
Fig. 26.26 AFM images of DNA on mica scanned in buffer solution (600 × 600 nm2 ). Each scan line was scanned twice –
in standard tapping mode during the first scan of the line (left data) and with Q-control being activated by a trigger signal
during the subsequent scan of the same line (right data). This interleave technique allows direct comparison of the results
of the two modes obtained on the same surface area while minimizing drift effects. Cross-sections of the topographic
data reveal that the observed DNA height is significantly higher in the case of imaging under Q-control ( c D. Ebeling,
University of Münster)
754 Part C Scanning-Probe Microscopy
operation of the AFM is possible in the net-attractive material during imaging. It was then possible to observe
regime. However, as explained before, below a cer- a DNA diameter close to the theoretical value with the
tain tip–sample separation dapp , the system switches Q-control feedback [26.68].
suddenly into intermittent contact mode, where sur- The same technique has been successfully em-
face modifications are more likely due to the onset ployed to minimize the interaction forces during
of strong repulsive forces. The steeper the amplitude scanning in liquids. This is of special relevance for
curve, the larger the regime of stable, net-attractive imaging delicate biological samples in environments
AFM operation. Looking at Fig. 26.22 we find that the such as water or buffer solution. When the AFM probe
slope of the amplitude curve is governed by the quality is submerged in a liquid medium, the oscillation of the
factor Q. A high Q, therefore, facilitates stable op- AFM cantilever is strongly affected by hydrodynamic
eration of the AM-AFM in the net-attractive regime. damping. This typically leads to quality factors < 10
(A more detailed discussion about this topic can be and accordingly to a loss in force sensitivity. However,
found in [26.64].) the Q-control technique allows the effective quality
An example can be seen in Fig. 26.25, which shows factor to be increased by about three orders of magni-
a surface scan of an ultrathin organic film acquired tude in liquids. Figure 26.26 shows results of scanning
in tapping mode under ambient conditions. First, the DNA structures on a mica substrate under buffer so-
inner square was scanned without the Q enhance- lution [26.69]. Comparison of the topographic data
ment, and then a wider surface area was scanned obtained in standard tapping mode and under Q-control,
with applied Q-control. The high quality factor pro- in particular the difference in the observed DNA height,
vides a larger parameter space for operating the AFM indicates that the imaging forces were successfully re-
in the net-attractive regime, allowing good resolution duced by employing Q-control.
of the delicate organic surface structure. Without the In conclusion, we have shown that, by applying an
Q-control the surface structures are deformed and even additional feedback circuit to the dynamic AFM system,
destroyed due to the strong repulsive tip–sample inter- it is possible to influence the quality factor Q of the
actions [26.65–67]. This also allowed imaging of DNA oscillator system. High-resolution, high-speed, or low-
structures without predominantly depressing the soft force scanning is then possible.
ploited to image the doping level of semiconductors. was successfully imaged in this mode. The step edges
Energy dissipation has also been observed in imaging of monatomic NaCl islands on single-crystalline cop-
magnetic materials. Liu and Grütter [26.73] found that per have also rendered atomic-resolution contrast in the
energy dissipation due to magnetic interactions was en- dissipation channel [26.84]. The dissipation processes
hanced at the boundaries of magnetic domains, which discussed so far are mostly in the configuration in which
was attributed to domain wall oscillations. Even a sim- the tip is oscillated perpendicular to the surface. Friction
ple system such as two clean metal surfaces which are is usually referred to as the energy loss due to lateral
moved in close proximity can give rise to frictional movement of solid bodies in contact. It is interesting
forces. Stipe et al. [26.74] have measured the energy to note in this context that Israelachivili [26.85] has
dissipation due to fluctuating electromagnetic fields be- pointed out a quantitative relationship between lateral
tween two closely spaced gold surfaces, which was later and vertical (with respect to the surface) dissipation.
interpreted by Volokitin and Persson [26.75] in terms of He states that the hysteresis in vertical force–distance
van der Waals friction. curves should equal the energy loss in lateral friction.
However, also in the absence of external electro- An experimental confirmation of this conjecture at the
magnetic fields, energy dissipation was observed in molecular level is still lacking.
close proximity of tip and sample, within 1 nm. Clearly, Physical interpretation of energy-dissipation pro-
mechanical surface relaxations must give rise to en- cesses at the atomic scale seems to be a daunting task at
ergy losses. One could model the AFM tip as a small this point. Notwithstanding, we can find a quantitative
hammer, hitting the surface at high frequency, possibly relation between the energy loss per oscillation cycle
resulting in phonon excitations. From a continuum- and the experimental parameters in dynamic AFM, as
mechanics point of view, we assume that the mechanical will be shown in the following section.
relaxation of the surface is not only governed by elastic In static AFM it was found that permanent changes
responses. Viscoelastic effects of soft surfaces will also of the sample surface by indentations can cause hystere-
render a significant contribution to energy dissipation. sis between approach and retraction. The area between
The whole area of phase imaging in tapping mode is the approach and retraction curves in a force–distance
concerned with those effects [26.76–79]. diagram represents the lost or dissipated energy caused
In the atomistic view, the last tip atom can be by the irreversible change of the surface structure. In
envisaged to change position while experiencing the dynamic-mode AFM, the oscillation parameters such as
tip–sample force field. A strictly reversible change of amplitude, frequency, and phase must contain the infor-
position would not result in a loss of energy. Still, it mation about the dissipated energy per cycle. So far, we
has been pointed out by Sasaki and Tsukada [26.80] have resorted to a treatment of the equation of motion
that a change in atom position would result in a change of the cantilever vibration in order to find a quantitative
Part C 26.5
in the force interaction itself. Therefore, it is pos- correlation between forces and the experimental param-
sible that the tip atom changes position at different eters. For the dissipation it is useful to treat the system
tip–surface distances during approach and retraction, from the energy-conservation point of view.
effectively causing atomic-scale hysteresis to develop. Assuming that a dynamic system is in equilibrium,
Hoffmann et al. [26.13] and Hembacher et al. [26.81] the average energy input must equal the average energy
have measured the short-range energy dissipation for output or dissipation. Applying this rule to an AFM run-
different combinations of tip and surface materials in ning in dynamic mode means that the average power
UHV. For atomic-resolution experiments at low temper- fed into the cantilever oscillation by an external driver,
atures on graphite [26.81] it was found that the energy denoted by P̄in , must equal the average power dissi-
dissipation is a step-like function. A similar shape of pated by the motion of the cantilever beam P̄0 and by
dissipation curves was found in a theoretical analysis by tip–sample interaction P̄tip
Kantorovich and Trevethan [26.82], where the energy
P̄in = P̄0 + P̄tip . (26.30)
dissipation was directly associated with atomic instabil-
ities at the sample surface. The term P̄tip is what we are interested in, since it
The dissipation channel has also been used to im- gives us a direct physical quantity to characterize the
age surfaces with atomic resolution [26.83]. Instead of tip–sample interaction. Therefore, we have first to cal-
feeding back the distance on the frequency shift, the culate and then measure the two other terms in (26.30)
excitation amplitude in FM mode has been used as in order to determine the power dissipated when the tip
the control signal. The Si(111)-(7 × 7) reconstruction periodically probes the sample surface. This requires an
756 Part C Scanning-Probe Microscopy
Part C 26.5
the same time the oscillation amplitude A is kept con-
stant by adjusting the drive amplitude Ad . By measuring tent contact (ϕ < 90◦ ) interaction between the tip and
these quantities, one can apply (26.38) to determine the the sample [26.19,60,93,94]. For example, consider the
average power dissipation related to tip–sample inter- phase shift in tapping mode as a function of z-position
action. In spectroscopy applications usually Ad (z) is (Fig. 26.12). If phase measurements are performed close
not measured directly, but a signal G(z) proportional to to the point where the oscillation switches from the
Ad (z) is acquired, representing the gain factor applied net-attractive to the intermittent contact regime, a large
to the excitation piezo. With the help of (26.15) we can contrast in the phase channel is observed. However, this
write contrast is not due to dissipative processes. Only a vari-
A0 G(z) ation of the phase signal within the intermittent contact
Ad (z) = , (26.39) regime will give information about the tip–sample dis-
QG 0
sipative processes.
where A0 and G 0 are the amplitude and gain at large An example of dissipation measurement is de-
tip–sample distances where the tip–sample interactions picted in Fig. 26.28. The surface of a polymer blend
are negligible. was imaged in air, simultaneously acquiring the to-
Now let us consider the tapping-mode AFM. In this pography and dissipation. The dissipation on the
case the cantilever is driven at a fixed frequency and softer polyurethane matrix is significantly larger than
with constant drive amplitude, while the oscillation am- on the embedded, mechanically stiffer polypropylene
plitude and phase shift may change when the probing particles.
758 Part C Scanning-Probe Microscopy
26.6 Conclusions
Dynamic force microscopy is a powerful tool, which namic AFM; energy dissipation also influences the
is capable of imaging surfaces with atomic precision. cantilever oscillation and can therefore serve as a new
It also allows us to look at surface dynamics and can information channel.
operate in vacuum, air or even liquid. However, the Open questions are still concerned with the ex-
oscillating cantilever system introduces a level of com- act geometric and chemical identity of the probing
plexity which disallows straightforward interpretation tip, which significantly influences the imaging and
of acquired images. An exception is the self-excitation spectroscopic results. Using predefined tips such as
mode, where tip–sample forces can be successfully ex- single-walled nanotubes or using atomic-resolution
tracted from spectroscopic experiments. However, not techniques such as field ion microscopy to image the
only conservative forces can be investigated with dy- tip itself are possible approaches addressing this issue.
References
26.1 M.A. Lantz, H.J. Hug, R. Hoffmann, P.J.A. van 26.12 S.P. Jarvis, M.A. Lantz, U. Dürig, H. Tokumoto: Off
Schendel, P. Kappenberger, S. Martin, A. Baratoff, resonance ac mode force spectroscopy and imaging
H.-J. Güntherodt: Quantitative measurement of with an atomic force microscope, Appl. Surf. Sci.
short-range chemical bonding forces, Science 291, 140, 309–313 (1999)
2580–2583 (2001) 26.13 P.M. Hoffmann, S. Jeffery, J.B. Pethica, H.Ö. Özer,
26.2 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi- A. Oral: Energy dissipation in atomic force mi-
croscope, Phys. Rev. Lett. 56, 930–933 (1986) croscopy and atomic loss processes, Phys. Rev. Lett.
26.3 O. Marti: AFM Instrumentation and Tips. In: Hand- 87, 265502–265505 (2001)
book of Micro/Nanotribology, 2nd edn., ed. by 26.14 H. Hölscher, B. Gotsmann, W. Allers, U.D. Schwarz,
B. Bushan (CRC, Boca Raton 1999) pp. 81–144 H. Fuchs, R. Wiesendanger: Comment on damp-
26.4 G. Cross, A. Schirmeisen, A. Stalder, P. Grütter, ing mechanism in dynamic force microscopy, Phys.
M. Tschudy, U. Dürig: Adhesion interaction be- Rev. Lett. 88, 019601 (2002)
tween atomically defined tip and sample, Phys. 26.15 E. Sivaniah, J. Genzer, G.H. Fredrickson, E.J. Kra-
Rev. Lett. 80, 4685–4688 (1998) mer, M. Xiang, X. Li, C. Ober, S. Magonov: Periodic
26.5 A. Schirmeisen, G. Cross, A. Stalder, P. Grütter, surface topology of three-arm semifluorinated
U. Dürig: Metallic adhesion and tunneling at the alkane monodendron diblock copolymers, Lang-
atomic scale, New J. Phys. 2, 1–29 (2000) muir 17, 4342–4346 (2001)
26.6 A. Schirmeisen: Metallic adhesion and tunneling 26.16 S.N. Magonov: Visualization of polymer structures
Part C 26
at the atomic scale. Ph.D. Thesis (McGill University, with atomic force microscopy. In: Applied Scan-
Montréal 1999) pp. 29–38 ning Probe Methods, ed. by H. Fuchs, M. Hosaka,
26.7 F.J. Giessibl: Forces and frequency shifts in atom- B. Bhushan (Springer, Berlin 2004) pp. 207–
icresolution dynamic-force microscopy, Phys. Rev. 250
B 56, 16010–16015 (1997) 26.17 C. Möller, M. Allen, V. Elings, A. Engel, D.J. Müller:
26.8 F.J. Giessibl: Atomic resolution of the silicon (111)- Tapping-mode atomic force microscopy produces
(7 × 7) surface by atomic force microscopy, Science faithful high-resolution images of protein sur-
267, 68–71 (1995) faces, Biophys. J. 77, 1150–1158 (1999)
26.9 M. Bezanilla, B. Drake, E. Nudler, M. Kashlev, 26.18 B. Anczykowski, D. Krüger, H. Fuchs: Cantilever
P.K. Hansma, H.G. Hansma: Motion and enzymatic dynamics in quasinoncontact force microscopy:
degradation of DNA in the atomic forcemicroscope, Spectroscopic aspects, Phys. Rev. B 53, 15485–
Biophys. J. 67, 2454–2459 (1994) 15488 (1996)
26.10 Y. Jiao, D.I. Cherny, G. Heim, T.M. Jovin, T.E. Schäf- 26.19 B. Anczykowski, D. Krüger, K.L. Babcock, H. Fuchs:
fer: Dynamic interactions of p53 with DNA in Basic properties of dynamic force spectroscopy with
solution by time-lapse atomic force microscopy, the scanning force microscope in experiment and
J. Mol. Biol. 314, 233–243 (2001) simulation, Ultramicroscopy 66, 251–259 (1996)
26.11 T.R. Albrecht, P. Grütter, D. Horne, D. Rugar: 26.20 V.M. Muller, V.S. Yushchenko, B.V. Derjaguin: On
Frequency modulation detection using high-Q the influence of molecular forces on the deforma-
cantilevers for enhanced force microscopy sensi- tion of an elastic sphere and its sticking to a rigid
tivity, J. Appl. Phys. 69, 668–673 (1991) plane, J. Colloid Interface Sci. 77, 91–101 (1980)
Dynamic Modes of Atomic Force Microscopy References 759
26.21 B.D. Hughes, L.R. White: ‘Soft’ contact problems in 26.38 B. Gotsmann, H. Fuchs: Dynamic force spectroscopy
linear elasticity, Q. J. Mech. Appl. Math. 32, 445–471 of conservative and dissipative forces in an Al-
(1979) Au(111) tip–sample system, Phys. Rev. Lett. 86,
26.22 P. Gleyzes, P.K. Kuo, A.C. Boccara: Bistable behavior 2597–2600 (2001)
of a vibrating tip near a solid surface, Appl. Phys. 26.39 H. Hölscher, W. Allers, U.D. Schwarz, A. Schwarz,
Lett. 58, 2989–2991 (1991) R. Wiesendanger: Determination of tip–sample in-
26.23 A. San Paulo, R. Garcia: High-resolution imag- teraction potentials by dynamic force spectroscopy,
ing of antibodies by tapping-mode atomic force Phys. Rev. Lett. 83, 4780–4783 (1999)
microscopy: Attractive and repulsive tip–sample 26.40 U. Dürig: Relations between interaction force and
interaction regimes, Biophys. J. 78, 1599–1605 frequency shift in large-amplitude dynamic force
(2000) microscopy, Appl. Phys. Lett. 75, 433–435 (1999)
26.24 D. Krüger, B. Anczykowski, H. Fuchs: Physical prop- 26.41 H. Hölscher, A. Schwarz, W. Allers, U.D. Schwarz,
erties of dynamic force microscopies in contact and R. Wiesendanger: Quantitative analysis of dyn-
noncontact operation, Ann. Phys. 6, 341–363 (1997) amicforce-spectroscopy data on graphite (0001) in
26.25 H. Hölscher: Quantitative measurement of tip– the contact and noncontact regime, Phys. Rev. B
sample interactions in amplitude modulation 61, 12678–12681 (2000)
atomic force microscopy, Appl. Phys. Lett. 89, 26.42 M. Guggisberg: Lokale Messung von atomaren
123109 (2006) Kräften. Ph.D. Thesis (University of Basel, Basel
26.26 M. Lee, W. Jhe: General theory of amplitude- 2000) pp. 9–11, in German
modulation atomic force microscopy, Phys. Rev. 26.43 M. Guggisberg, M. Bammerlin, E. Meyer, H.-J. Gün-
Lett. 97, 036104 (2006) therodt: Separation of interactions by noncontact
26.27 Y. Martin, C.C. Williams, H.K. Wickramasinghe: force microscopy, Phys. Rev. B 61, 11151–11155
Atomic force microscope – force mapping and pro- (2000)
filing on a sub 100-Å scale, J. Appl. Phys. 61, 26.44 U. Dürig: Extracting interaction forces and comple-
4723–4729 (1987) mentary observables in dynamic probemicroscopy,
26.28 Y. Sugawara, M. Otha, H. Ueyama, S. Morita: De- Appl. Phys. Lett. 76, 1203–1205 (2000)
fect motion on an InP(110) surface observed with 26.45 F.J. Giessibl: A direct method to calculate tip–
noncontact atomic force microscopy, Science 270, Sample forces from frequency shifts in frequen-
1646–1648 (1995) cymodulation atomic force microscopy, Appl. Phys.
26.29 M. Bammerlin, R. Lüthi, E. Meyer, A. Baratoff, Lett. 78, 123–125 (2001)
J. Lue, M. Guggisberg, C. Gerber, L. Howald, H.- 26.46 J.E. Sader, S.P. Jarvis: Accurate formulas for inter-
J. Güntherodt: True atomic resolution on the action force and energy in frequency modulation
surface of an insulator via ultrahigh vacuum dy- force spectroscopy, Appl. Phys. Lett. 84, 1801–1803
namic dynamic force microscopy, Probe Microsc. 1, (2004)
3–9 (1996) 26.47 O. Pfeiffer: Quantitative dynamische Kraft- und
26.30 C. Barth, M. Reichling: Imaging the atomic ar- Dissipationsmikroskopie auf molekularer Skala.
rangement on the high-temperature reconstructed Ph.D. Thesis (Universität Basel, Basel 2004), , in
Part C 26
α-Al2 O3 (0001) Surface, Nature 414, 54–57 (2001) German
26.31 W. Allers, S. Langkat, R. Wiesendanger: Dynamic 26.48 H. Hölscher, S.M. Langkat, A. Schwarz, R. Wiesen-
low-temperature scanning force microscopy on danger: Measurement of three-dimensional force
nickel oxyde(001), Appl. Phys. A [Suppl.] 72, S27–S30 fields with atomic resolution using dynamic force
(2001) spectroscopy, Appl. Phys. Lett. 81, 4428 (2002)
26.32 U. Kaiser, A. Schwarz, R. Wiesendanger: Magnetic 26.49 A. Schirmeisen, D. Weiner, H. Fuchs: Single-atom
exchange force microscopy with atomic resolution, contact mechanics: From atomic scale energy bar-
Nature 446, 522–525 (2007) rier to mechanical relaxation hysteresis, Phys. Rev.
26.33 S. Morita, R. Wiesendanger, E. Meyer: Noncontact Lett. 97, 136101 (2006)
Atomic Force Microscopy (Springer, Berlin Heidel- 26.50 T. Uchihashi, T. Ishida, M. Komiyama, M. Ashino,
berg 2002) Y. Sugawara, W. Mizutani, K. Yokoyama, S. Morita,
26.34 R. García, R. Pérez: Dynamic atomic force mi- H. Tokumoto, M. Ishikawa: High-resolution imag-
croscopy methods, Surf. Sci. Rep. 47, 197–301 (2002) ing of organic monolayers using noncontact AFM,
26.35 F.J. Giessibl: Advances in atomic force microscopy, Appl. Surf. Sci. 157, 244–250 (2000)
Rev. Mod. Phys. 75, 949–983 (2003) 26.51 H. Hölscher, B. Gotsmann, A. Schirmeisen: Dynamic
26.36 H. Hölscher, A. Schirmeisen: Dynamic force mi- force spectroscopy using the frequency modulation
croscopy and spectroscopy, Adv. Imaging Electron technique with constant excitation, Phys. Rev. B
Phys. 135, 41–101 (2005) 68, 153401-1–153401-4 (2003)
26.37 H. Hölscher, U.D. Schwarz, R. Wiesendanger: Cal- 26.52 A. Schirmeisen, H. Hölscher, B. Anczykowski,
culation of the frequency shift in dynamic force D. Weiner, M.M. Schäfer, H. Fuchs: Dynamic force
microscopy, Appl. Surf. Sci. 140, 344–351 (1999) spectroscopy using the constant-excitation and
760 Part C Scanning-Probe Microscopy
constant-amplitude modes, Nanotechnology 16, structures in monolayers, Adv. Mater. 12, 25–30
13–17 (2005) (2000)
26.53 T. Uchihashi, M.J. Higgins, S. Yasuda, S.P. Jarvis, 26.66 S. Gao, L.F. Chi, S. Lenhert, B. Anczykowski,
S. Akita, Y. Nakayama, J.E. Sader: Quantitative force C. Niemeyer, M. Adler, H. Fuchs: High-quality
measurements in liquid using frequency modula- mapping of DNA–protein complexes by dynamic
tion atomic force microscopy, Appl. Phys. Lett. 85, scanning forcemicroscopy, ChemPhysChem 6, 384–
3575 (2004) 388 (2001)
26.54 J.-E. Schmutz, H. Hölscher, D. Ebeling, M.M. Schä- 26.67 B. Zou, M. Wang, D. Qiu, X. Zhang, L.F. Chi,
fer, B. Ancyzkowski: Mapping the tip–sample H. Fuchs: Confined supramolecular nanostructures
interactions on DPPC and DNA by dynamic force of mesogen-bearing amphiphiles, Chem. Com-
spectroscopy under ambient conditions, Ultrami- mun. 9, 1008–1009 (2002)
croscopy 107, 875–881 (2007) 26.68 B. Pignataro, L.F. Chi, S. Gao, B. Anczykowski,
26.55 C. Loppacher, M. Guggisberg, O. Pfeiffer, E. C. Niemeyer, M. Adler, H. Fuchs: Dynamic scanning
Meyer, M. Bammerlin, R. Lüthi, R. Schlittler, J.K. force microscopy study of self-assembled DNA–
Gimzewski, H. Tang, C. Joachim: Direct determi- protein nanostructures, Appl. Phys. A 74, 447–452
nation of the energy required to operate a single (2002)
molecule switch, Phys. Rev. Lett. 90, 066107-1– 26.69 D. Ebeling, H. Hölscher, H. Fuchs, B. Anczykowski,
066107-4 (2003) U.D. Schwarz: Imaging of biomaterials in liquids:
26.56 N. Oyabu, O. Custance, I. Yi, Y. Sugawara, S. Morita: A comparison between conventional and Q-
Mechanical vertical manipulation of selected single controlled amplitude modulation (‘tapping mode’)
atoms by soft nanoindentation using near contact atomic force microscopy, Nanotechnology 17, S221–
atomic force microscopy, Phys. Rev. Lett. 90, 176102 S226 (2005)
(2003) 26.70 U. Dürig: Interaction sensing in dynamic force mi-
26.57 Y. Sugimoto, M. Abe, S. Hirayama, N. Oyabu, O. Cus- croscopy, New J. Phys. 2, 1–5 (2000)
tance, S. Morita: Atom inlays performed at room 26.71 H. Hölscher, B. Gotsmann, W. Allers, U.D. Schwarz,
temperature using atomic force microscopy, Nat. H. Fuchs, R. Wiesendanger: Measurement of con-
Mater. 4, 156–159 (2005) servative and dissipative tip–sample interaction
26.58 J. Mertz, O. Marti, J. Mlynek: Regulation of a micro- forces with a dynamic force microscope using the
cantilever response by force feedback, Appl. Phys. frequency modulation technique, Phys. Rev. B 64,
Lett. 62, 2344–2346 (1993) 075402 (2001)
26.59 D. Rugar, P. Grütter: Mechanical parametric ampli- 26.72 T.D. Stowe, T.W. Kenny, D.J. Thomson, D. Rugar:
fication and thermomechanical noise squeezing, Silicon dopant imaging by dissipation force mi-
Phys. Rev. Lett. 67, 699–702 (1991) croscopy, Appl. Phys. Lett. 75, 2785–2787 (1999)
26.60 B. Anczykowski, J.P. Cleveland, D. Krüger, V.B. E- 26.73 Y. Liu, P. Grütter: Magnetic dissipation force mi-
lings, H. Fuchs: Analysis of the interaction croscopy studies of magnetic materials, J. Appl.
mechanisms in dynamic mode SFM by means of Phys. 83, 7333–7338 (1998)
experimental data and computer simulation, Appl. 26.74 B.C. Stipe, H.J. Mamin, T.D. Stowe, T.W. Kenny, D.
Part C 26
Phys. A 66, 885 (1998) Rugar: Noncontact friction and force fluctuations
26.61 D. Ebeling, H. Hölscher, B. Anczykowski: In- between closely spaced bodies, Phys. Rev. Lett. 87,
creasing the Q-factor in the constant-excitation 96801-1–96801-4 (2001)
mode of frequency-modulation atomic force mi- 26.75 A.I. Volokitin, B.N.J. Persson: Resonant photon
croscopy in liquid, Appl. Phys. Lett. 89, 203511 tunneling enhancement of the van der Waals fric-
(2006) tion, Phys. Rev. Lett. 91, 106101-1–106101-4 (2003)
26.62 D. Ebeling, H. Hölscher: Analysis of the constant- 26.76 J. Tamayo, R. Garcia: Effects of elastic and inelastic
excitation mode in frequency-modulation atomic interactions on phase contrast images in tapping-
force microscopy with active Q-control applied in mode scanning force microscopy, Appl. Phys. Lett.
ambient conditions and liquids, J. Appl. Phys. 102, 71, 2394–2396 (1997)
114310 (2007) 26.77 J.P. Cleveland, B. Anczykowski, A.E. Schmid,
26.63 T. Sulchek, G.G. Yaralioglu, C.F. Quate, S.C. Minne: V.B. Elings: Energy dissipation in tapping-mode
Characterization and optimisation of scan speed atomic force microscopy, Appl. Phys. Lett. 72, 2613–
for tapping-mode atomic force microscopy, Rev. 2615 (1998)
Sci. Instrum. 73, 2928–2936 (2002) 26.78 R. García, J. Tamayo, M. Calleja, F. García: Phase
26.64 H. Hölscher, U.D. Schwarz: Theory of amplitude contrast in tapping-mode scanning force mi-
modulation atomic force microscopy with and croscopy, Appl. Phys. A 66, S309–S312 (1998)
without Q-control, Int. J. Nonlinear Mech. 42, 608– 26.79 B. Anczykowski, B. Gotsmann, H. Fuchs, J.P. Cleve-
625 (2007) land, V.B. Elings: How to measure energy dissi-
26.65 L.F. Chi, S. Jacobi, B. Anczykowski, M. Overs, pation in dynamic mode atomic force microscopy,
H.-J. Schäfer, H. Fuchs: Supermolecular periodic Appl. Surf. Sci. 140, 376–382 (1999)
Dynamic Modes of Atomic Force Microscopy References 761
26.80 N. Sasaki, M. Tsukada: Effect of microscopic non- point-mass models, Appl. Phys. Lett. 80, 1646–
conservative process on noncontact atomic force 1648 (2002)
microscopy, Jpn. J. Appl. Phys. 39, 1334 (2000) 26.88 J. Tamayo, R. García: Relationship between phase
26.81 S. Hembacher, F.J. Giessibl, J. Mannhart, C.F. shift and energy dissipation in tapping-mode
Quate: Local spectroscopy and atomic imaging scanning force microscopy, Appl. Phys. Lett. 73,
of tunneling current, forces, and dissipation on 2926–2928 (1998)
graphite, Phys. Rev. Lett. 94, 056101-1–056101-4 26.89 R. García, J. Tamayo, A. San Paulo: Phase contrast
(2005) and surface energy hysteresis in tapping mode
26.82 L.N. Kantorovich, T. Trevethan: General theory of scanning force microcopy, Surf. Interface Anal.
microscopic dynamical response in surface probe 27(5/6), 312–316 (1999)
microscopy: From imaging to dissipation, Phys. 26.90 S.N. Magonov, V.B. Elings, M.H. Whangbo: Phase
Rev. Lett. 93, 236102-1–236102-4 (2004) imaging and stiffness in tapping-mode atomic
26.83 R. Lüthi, E. Meyer, M. Bammerlin, A. Baratoff, force microscopy, Surf. Sci. 375, 385–391 (1997)
L. Howald, C. Gerber, H.-J. Güntherodt: Ultrahigh 26.91 J.P. Pickering, G.J. Vancso: Apparent contrast re-
vacuum atomic force microscopy: True atomic res- versal in tapping mode atomic force microscope
olution, Surf. Rev. Lett. 4, 1025–1029 (1997) images on films of polystyrene-b-polyisoprene-b-
26.84 R. Bennewitz, A.S. Foster, L.N. Kantorovich, polystyrene, Polym. Bull. 40, 549–554 (1998)
M. Bammerlin, C. Loppacher, S. Schär, M. Guggis- 26.92 X. Chen, S.L. McGurk, M.C. Davies, C.J. Roberts,
berg, E. Meyer, A.L. Shluger: Atomically resolved K.M. Shakesheff, S.J.B. Tendler, P.M. Williams,
edges and kinks of NaCl islands on Cu(111): Ex- J. Davies, A.C. Dwakes, A. Domb: Chemical
periment and theory, Phys. Rev. B 62, 2074–2084 and morphological analysis of surface enrich-
(2000) ment in a biodegradable polymer blend by
26.85 J. Israelachvili: Intermolecular and Surface Forces phase-detection imaging atomic force microscopy,
(Academic, London 1992) Macromolecules 31, 2278–2283 (1998)
26.86 U. Rabe, J. Turner, W. Arnold: Analysis of the high- 26.93 A. Kühle, A.H. Sørensen, J. Bohr: Role of attrac-
frequency response of atomic force microscope tive forces in tapping tip force microscopy, J. Appl.
cantilevers, Appl. Phys. A 66, 277 (1998) Phys. 81, 6562–6569 (1997)
26.87 T.R. Rodriguez, R. García: Tip motion in am- 26.94 A. Kühle, A.H. Sørensen, J.B. Zandbergen, J. Bohr:
plitude modulation (tapping-mode) atomic-force Contrast artifacts in tapping tip atomic force mi-
microscopy: Comparison between continuous and croscopy, Appl. Phys. A 66, 329–332 (1998)
Part C 26
763
Molecular Re
27. Molecular Recognition Force Microscopy:
From Molecular Bonds to Complex Energy Landscapes
Peter Hinterdorfer, Andreas Ebner, Hermann Gruber, Ruti Kapon, Ziv Reich
Part C 27
interaction bond breaks at a critical (unbinding) of filamentous proteins. Recognition imaging,
force. Such experiments allow for estimation of developed by combing dynamic force microscopy
affinity, rate constants, and structural data of the with force spectroscopy, allows for localization
binding pocket. Comparing them with values ob- of receptor sites on surfaces with nanometer
tained from ensemble-average techniques and positional accuracy.
Molecular recognition plays a pivotal role in nature. pathways all rely critically on specific recognition. In
Signaling cascades, enzymatic activity, genome repli- fact, every process which requires molecules to interact
cation and transcription, cohesion of cellular structures, with each other in a specific manner requires that they
interaction of antigens and antibodies, and metabolic be able to recognize each other.
764 Part C Scanning-Probe Microscopy
Molecular recognition studies emphasize specific erativity of bond formation, is the key to understanding
interactions between receptors and their cognitive lig- receptor–ligand recognition.
ands. Despite a growing body of literature on the For this, experiments at the single-molecule level,
structure and function of receptor–ligand complexes, it and on time scales typical for receptor–ligand complex
is still not possible to predict reaction kinetics or ener- formation and dissociation, are required. The methodol-
getics for any given complex formation, even when the ogy described in this chapter for investigating molecular
structures are known. Additional insights, in particular dynamics of receptor–ligand interactions, molecular
into the molecular dynamics within the complex dur- recognition force microscopy (MRFM) [27.1–3], is
ing the association and dissociation process, are needed. based on atomic force microscope (AFM) technol-
The high-end strategy is to probe the energy landscape ogy [27.4]. The ability of the AFM [27.4] to measure
that underlies the interactions between molecules whose ultralow forces at high lateral resolution together with
structures are known with atomic resolution. its unique capability to operate in an aqueous and phys-
Receptor–ligand complexes are usually formed by iological environment opens the possibility of studying
a few, noncovalent weak interactions between contact- biological recognition processes in vivo. The interac-
ing chemical groups in complementary determining tion between a receptor and a ligand complex is studied
regions, supported by framework residues providing by exerting a force on the complex and following the
structurally conserved scaffolding. Both the comple- dissociation process over time. Dynamic aspects of
mentary determining regions and the framework have recognition are addressed in force spectroscopy (FS) ex-
a considerable amount of plasticity and flexibility, periments, where distinct force–loading rate profiles are
allowing for conformational movements during associ- used to provide insight into the energy landscape un-
ation and dissociation. In addition to knowledge about derlying the reaction. It is also possible to investigate
structure, energies, and kinetic constants, information the force–time behavior to unravel changes of confor-
about these movements is required to understand the mation which occur during the dissociation process. It
recognition process. Deeper insight into the nature of will be shown that MRFM is a versatile tool to ex-
these movements as well as the spatiotemporal action plore kinetic and structural details of receptor–ligand
of the many weak interactions, in particular the coop- recognition.
most reactive sites (Fig. 27.1), allowed measurement of nanotubes attached to the tips of gold-coated Si
single receptor–ligand unbinding forces. Nevertheless, cantilevers.
parallel breakage of multiple bonds was predominately In a number of laboratories, a distensible and flex-
observed with this configuration. ible linker was used to distance the ligand molecule
To measure interactions between isolated recep- from the tip surface (e.g., [27.3, 13]) (Fig. 27.2). At
tor–ligand pairs, strictly defined conditions need to be a given low number of spacer molecules per tip, the
fulfilled. Covalently coupling ligands to gold-coated tip ligand can freely orient and diffuse within a cer-
surfaces via freely accessible SH groups guarantees tain volume, provided by the length of the tether,
sufficiently stable attachment because these bonds are to achieve unconstrained binding to its receptor. The
about ten times stronger than typical ligand–receptor unbinding process occurs with little torque and the lig-
interactions [27.5]. This chemistry has been used to and molecule escapes the danger of being squeezed
detect the forces between complementary deoxyribonu- between the tip and the surface. This approach also
cleic acid (DNA) strands [27.1] as well as between opens the possibility of site-directed coupling for a de-
isolated nucleotides [27.6]. Self-assembled monolayers fined orientation of the ligand relative to the receptor
of dithio-bis(succinimidylundecanoate) were formed at receptor–ligand unbinding. As a cross-linking el-
to enable covalent coupling of biomolecules via ement, polyethylene glycol (PEG), a water-soluble
amines [27.7] and were used to study the binding nontoxic polymer with a wide range of applications
strength between cell adhesion proteoglycans [27.8] and in surface technology and clinical research, was often
between biotin-directed immunoglobulin G (IgG) anti- used [27.17]. PEG is known to prevent surface ad-
bodies and biotin [27.9]. Vectorial orientation of Fab sorption of proteins and lipid structures and therefore
molecules on gold tips was achieved by site-directed appears ideally suited for this purpose. Glutaralde-
chemical binding via their SH groups [27.10], without hyde [27.12] and DNA [27.1] were also successfully
the need for additional linkers. To this end, antibodies applied as molecular spacers in recognition force stud-
were digested with papain and subsequently purified ies. Cross-linker lengths, ideally arriving at a good
to generate Fab fragments with freely accessible SH compromise between high tip molecule mobility and
groups in the hinge region. narrow lateral resolution of the target recognition site,
Gold surfaces exhibit a unique and selective affin- varied from 2 to 100 nm.
ity for thiols, although the adhesion strength of the
resulting bonds is comparatively weak [27.5]. Since
all commercially available AFM tips are etched from Cantilever
silicon nitride or silicon oxide material, deposition of NH2
NH2
NH2
a gold layer onto the tip surface is required prior to us- NH2
ing this chemistry. Therefore, designing a sensor with NHS
covalent attachments of biomolecules to the silicon
surface may be more straightforward. Amine function-
alization procedures, a strategy widely used in surface PEG
biochemistry, were applied using ethanolamine [27.3,
11] and various silanization methods [27.12–15], as PDP
a first step in thoroughly developed surface anchor-
Part C 27.1
Cys
ing protocols suitable for single-molecule experiments.
Since the amine surface density determines, to a large Lig
extent, the number of ligands on the tip which can
specifically bind to the receptors on the surface, it has Fig. 27.2 Linkage of ligands to AFM tips. Ligands
to be sufficiently low to guarantee single-molecular were covalently coupled to AFM tips via a het-
recognition events [27.3, 11]. Typically, these densities erobifunctional polyethylene glycol (PEG) derivative
are kept between 200 and 500 molecules/μm2 , which of 8 nm length. Silicon tips were first functionalized
for AFM tips with radii of ≈ 5–20 nm, amounts to with ethanolamine (NH2 −C2 H4 OH · HCl). Then, the N-
about one molecule per effective tip area. A strik- hydroxy-succinimide (NHS)-end of the PEG linker was
ing example of a minimally ligated tip was given covalently bound to amines on the tip surface before lig-
by Wong et al. [27.16], who derivatized a few car- ands were attached to the pyridyldithiopropionate (PDP)
boxyl groups present at the open end of carbon end via a free thiol or cysteine
766 Part C Scanning-Probe Microscopy
For coupling to the tip surface and to the ligand, the are commercially available (Pierce, Rockford, IL) to
cross-linker typically carries two different functional generate active antibody fragments with free cysteines.
ends, e.g., an amine reactive N-hydroxysuccinimidyl Half-antibodies are produced by cleaving the two disul-
(NHS) group on one end, and a thiol reactive 2- fide bonds in the central region of the heavy chain using
pyridyldithiopropionyl group (PDP) [27.18, 19] on the 2-mercaptoethylamine HCl [27.20], and Fab fragments
other (Fig. 27.2). This sulfur chemistry is highly advan- are generated from papain digestion [27.10]. The most
tageous, since it is very reactive and readily enables elegant methods are to introduce a cysteine into the pri-
site-directed coupling. However, free thiols are hardly mary sequence of proteins or to append a thiol group
available on native ligands and must often be added by to the end of a DNA strand [27.21], allowing for well-
chemical derivatization. defined sequence-specific coupling of the ligand to the
Different strategies have been used to achieve cross-linker.
this goal. Lysine residues were derivatized with the An attractive alternative for covalent coupling is
short heterobifunctional linker N-succinimidyl-3-(S- provided by the widely used nitrilotriacetate (NTA)-
acetylthio)propionate (SATP) [27.18]. Subsequent de- His6 system. The strength of binding in this system,
protection with NH2 OH led to reactive SH groups. which is routinely used in chromatographic and biosen-
Alternatively, lysins can be directly coupled via alde- sor matrices, is significantly larger than that between
hyde groups [27.15]. The direct coupling of proteins via most ligand–receptor pairs [27.22–24]. For receptor–
an NHS–PEG–aldehyde linker allows binding via lysine ligand interactions with very high unbinding force,
groups without prederivatization. Nevertheless, since NTA can be substituted with a recently developed
both ends are reactive against amino groups, loop for- Tris-NTA linker [27.25, 26]. Since a His6 tag can be
mation can occur between adjacent NH2 groups on the readily introduced in recombinant proteins, a cross-
tip. The probability for this side-effect is significantly linker containing an (Tris-)NTA residue is ideally suited
lowered (1) by the much higher amino reactivity of the for coupling proteins to the AFM tip. This generic,
NHS ester in comparison with the aldehyde function site-specific coupling strategy also allows rigorous and
and (2) by high linker concentration. Another disad- ready control of binding specificity by using Ni2+ as
vantage of the latter two methods is that it does not a molecular switch of the NTA–His6 bond. A detailed
allow for site-specific coupling of the cross-linker, since description of actual coupling strategies can by found
lysine residues are quite abundant. Several protocols in [27.26].
Freshly cleaved muscovite mica is a perfectly pure ing is however quite limited and generally offers no
and atomically flat surface and, therefore, ideally suited means to orient the molecules over the surface in
for MRFM studies. The strong negative charge of mica a desirable direction. Immobilization by covalent at-
also accomplishes very tight electrostatic binding of tachment must therefore be frequently explored. When
various biomolecules; for example, lysozyme [27.20] glass, silicon or mica are used as probe surfaces,
and avidin [27.27] strongly adhere to mica at pH < 8. immobilization is essentially the same as described
In such cases, simple adsorption of the receptors from above for tip functionalization. The number of reac-
solution is sufficient, since attachment is strong enough tive SiOH groups of the chemically relatively inert
to withstand pulling. Nucleic acids can also be firmly mica can be optionally increased by water plasma treat-
bound to mica through mediatory divalent cations such ment [27.32]. As with tips, cross-linkers are also often
as Zn2+ , Ni2+ or Mg2+ [27.28]. The strongly acidic used to provide receptors with motional freedom and
sarcoplasmic domain of the skeletal muscle calcium re- to prevent surface-induced protein denaturation [27.3].
Molecular Recognition Force Microscopy 27.3 Single-Molecule Recognition Force Detection 767
Part C 27.3
able. For adherent cells, the easiest approach is to grow penetrate into the lipid bilayer of the cell membrane.
the cells directly on glass or other surfaces suitable Such linkers provide sufficiently strong fixation without
for MRFM [27.36]. Firm immobilization of non- and interference with membrane proteins [27.37].
between single molecular pairs can only be estimated particles with nanometer positional accuracy. Using this
because the net force applied to the particles can only technique, forces in the range of 10−13 –10−10 N can be
be approximated and the number of bonds per particle measured accurately. Optical tweezers have been used
is unknown. extensively to measure the force-generating properties
SFA measures the forces between two surfaces to of various molecular motors at the single-molecule
which different interacting molecules are attached, us- level [27.44–46] and to obtain force–extension profiles
ing a cantilever spring as force probe and interferometry of single DNA [27.47] or protein [27.48] molecules.
for detection. The technique, which has a distance res- Defined, force-controlled twisting of DNA using ro-
olution of ≈ 1 Å, allows the measurement of adhesive tating magnetically manipulated particles gave further
and compressive forces and rapid transient effects to insights into DNA’s viscoelastic properties [27.49,
be followed in real time. However, the force sensi- 50].
tivity of the technique (≈ 10 nN) does not allow for AFM has successfully been used to measure the
single-molecule measurements of noncovalent interac- interaction forces between various single-molecular
tion forces. pairs [27.1–3]. In these measurements, one of the bind-
The biomembrane force probe (BFP) technique uses ing partners is immobilized onto a tip mounted at the
pressurized membrane capsules rather than mechanical end of a flexible cantilever that functions as a force
springs as a force transducer (Fig. 27.4; see, for ex- transducer and the other is immobilized over a hard sur-
ample, [27.41]). To form the transducer, a red blood face such as mica or glass. The tip is initially brought
cell or a lipid bilayer vesicle is pressurized into the to, and subsequently retracted from the surface, and the
tip of a glass micropipette. The spring constant of interaction (unbinding) force is measured by following
the capsule can then be varied over several orders of the cantilever deflection, which is monitored by mea-
magnitude by suction. This simple but highly effec- suring the reflection of a laser beam focused on the
tive configuration enables the measurement of forces back of the cantilever using a split photodiode. Ap-
ranging from 0.1–1000 pN with a force resolution proach and retract traces obtained from the unbinding
of about 1 pN, allowing probing of single-molecular of a single-molecular pair is shown in Fig. 27.5 [27.3].
bonds. In this experiment, the binding partners were immobi-
In optical tweezers (OT), small dielectric par- lized onto their respective surfaces through a distensible
ticles (beads) are manipulated by electromagnetic PEG tether.
traps [27.42, 43]. Three-dimensional light intensity gra- Cantilever deflection, Δx, relates directly to the
dients of a focused laser beam are used to pull or push force F acting on it through Hook’s law F = kΔx,
where k is the spring constant of the cantilever. Dur-
ing most of the approach phase (trace, and points 1–5),
when the tip and the surface are sufficiently far away
from each other (1–4), cantilever deflection remains
zero because the molecules are still unbound from
each other. Upon contact (4) the cantilever bends up-
wards (4–5) due to a repulsive force that increases
linearly as the tip is pushed further into the surface.
If the cycle was futile, and no binding had occurred,
Part C 27.3
Molecular recognition is mediated by a multitude of The weak bonds that govern molecular cohesion are
noncovalent interactions whose energy is only slightly believed to be formed in a spatially and temporar-
higher than that of thermal energy. Due to the power- ily correlated fashion. Protein binding often involves
law dependence of these interactions on distance, structural rearrangements that can be either localized
the attractive forces between noncovalently interacting or global. These rearrangements often bear functional
molecules are extremely short-ranged. A close geo- significance by modulating the activity of the inter-
metrical and chemical fit within the binding interface actants. Signaling pathways, enzyme activity, and the
is therefore a prerequisite for productive association. activation and inactivation of genes all depend on
770 Part C Scanning-Probe Microscopy
measured bond strengths. On the time scale of AFM contributions arising from thermal activation manifest
experiments (milliseconds to seconds), thermal im- themselves mostly near the point of detachment. There-
pulses govern the unbinding process. In the thermal fore, the change of force with time or the loading rate
activation model, the lifetime of a molecular com- r (= d f/ dt) can be derived from the product of the
plex in solution is described by a Boltzmann ansatz, pulling velocity and the effective spring constant at the
τ(0) = τosc exp[E b /(kB T )] [27.56], where τosc is the in- end of the force curve, just before unbinding occurs.
verse of the natural oscillation frequency and E b is the The dependence of the rupture force on the load-
height of the energy barrier for dissociation. This gives ing rate (force spectrum) in the thermally activated
a simple Arrhenius dependency of dissociation rate on regime was first derived by Evans and Ritchie [27.57]
barrier height. and described further by Strunz et al. [27.54]. Forced
A force acting on a complex deforms the interaction dissociation of receptor–ligand complexes using AFM
free energy landscape and lowers barriers for dissocia- or BFP can often be regarded as an irreversible pro-
Molecular Recognition Force Microscopy 27.5 Recognition Force Spectroscopy: From Isolated Molecules to Biological Membranes 771
cess because the molecules are kept away from each istic barrier length scale and signifying that a crossover
other after unbinding occurs (rebinding can be safely between barriers has occurred.
neglected when measurements are made with soft Dynamic force spectroscopy (DFS) exploits the de-
springs). Rupture itself is a stochastic process, and the pendence of bond strength on the loading rate to obtain
likelihood of bond survival is expressed in the mas- detailed insights into intra- and intermolecular interac-
ter equation as a time-dependent probability N(t) to tions. By measuring bond strength over a broad range
be in the bound state under a steady ramp of force, of loading rates, length scales and relative heights of
namely dN(t)/ dt = −koff (rt)N(t) [27.54]. This results energy barriers traversing the free energy surface can
in a distribution of unbinding forces P(F) parame- be readily obtained. The lifetime of a bond at any
terized by the loading rate [27.54, 57, 58]. The most given force is likewise contained in the complete force
probable force for unbinding f ∗ , given by the maximum distribution [27.3]. Finally, one may attempt to ex-
of the distribution,
−1 relates to the loading rate through tract dissociation rate constants by extrapolation to zero
f ∗ = f β ln rkoff / f β , where the force scale f β is set force [27.59]. However, the application of force acts to
by the ratio of thermal energy to xβ [27.54, 57]. Thus, select the dissociation path. Since the kinetics of re-
the unbinding force scales linearly with the logarithm of actions is pathway dependent, such a selection implies
the loading rate. For a single barrier, this would give rise that kinetic parameters extracted from force–probe ex-
to a simple, linear force spectrum f ∗ versus log(r). In periments may differ from those obtained from assays
cases where the escape path is traversed by several bar- conducted in the absence of external force. For ex-
riers, the curve will follow a sequence of linear regimes, tremely fast complexation/decomplexation kinetics the
each marking a particular barrier [27.41, 57, 58]. Tran- forces can be independent of the loading rate, indicating
sition from one regime to the other is associated with that the experiments were carried out under thermody-
an abrupt change of slope determined by the character- namic equilibrium [27.60].
Part C 27.5
vealed that different unbinding forces can be obtained ing the elastic properties of the molecule. Hydrogen
for the same pulling velocity if the spring constant of bonds between nucleotides have been probed for all
the cantilever is varied [27.1], consistent with the afore- 16 combinations of the four DNA bases [27.6]. Di-
mentioned dependency of bond strength on the loading rectional hydrogen-bonding interactions were measured
rate. The interaction force between several biotin ana- only when complementary bases were present on tip
logues and avidin or streptavidin [27.61] and between and probe surface, indicating that AFM can be used to
biotin and a set of streptavidin mutants [27.62] was follow specific pairing of DNA strands.
investigated and found to generally correlate with the Strunz et al. [27.14] measured the forces required
equilibrium binding enthalpy and the enthalpic activa- to separate individual double-stranded DNA molecules
tion barrier. No correlation with the equilibrium free of 10, 20, and 30 base pairs (Fig. 27.7). The parameters
energy of binding or the activation free energy barrier describing the energy landscape, i. e., the distance from
to dissociation was observed, suggesting that internal the energy barrier to the minimum energy along the
772 Part C Scanning-Probe Microscopy
0.1 40
30
0.05 20
10
0 0
0 20 40 60 80 100 10–10 10–5 10 0 10 5
Force (pN) Velocity (nm/s)
Fig. 27.7a,b Dependence of the unbinding force between DNA single-strand duplexes on the retract velocity. In addition
to the expected logarithmic behavior on the loading rate, the unbinding force scales with the length of the strands,
increasing from the 10- to 20- to 30-base-pair duplexes (after [27.14])
separation path and the logarithm of the thermal dissoci- Besides measurements of interaction forces, single-
ation rate, were found to be proportional to the number molecule force spectroscopy also allows estimation of
of base pairs of the DNA duplex. Such scaling suggests association and dissociation rate constants, notwith-
that unbinding proceeds in a highly cooperative manner standing the concern stated above [27.3, 11, 23, 59, 64,
characterized by one length scale and one time scale. 65], and measurement of structural parameters of the
Studying the dependence of rupture forces on temper- binding pocket [27.3, 11, 14, 64, 65]. Quantification of
ature, it was proposed by Schumakovitch et al. [27.63] the association rate constant kon requires determination
that entropic contributions play an important role in the of the interaction time needed for half-maximal prob-
unbinding of complementary DNA strands [27.63]. ability of binding (t1/2 ). This can be obtained from
Prevalent as it is, molecular recognition has mostly experiments where the encounter time between recep-
been discussed in the context of interactions between tor and ligand is varied over a broad range [27.64].
antibodies and antigens. To maximize motional freedom Given that the concentration of ligand molecules on the
and to overcome problems associated with misorienta- tip available for interaction with the surface-bound re-
tion and steric hindrance, antibodies and antigens were ceptors ceff is known, the association rate constant can
−1 −1
immobilized onto the AFM tip and probe surface via be derived from kon = t0.5 ceff . Determination of the ef-
flexible molecular spacers [27.3, 9, 12, 13]. By optimiz- fective ligand concentration requires knowledge of the
ing the antibody density over the AFM tip [27.3, 11], effective volume Veff explored by the tip-tethered ligand
Part C 27.5
the interaction between individual antibody–antigen which, in turn, depends on the tether length. There-
pairs could be examined. Binding of antigen to the fore, only order-of-magnitude estimates of kon can be
two Fab fragments of the antibody was shown to oc- obtained from such measurements [27.64].
cur independently and with equal probability. Single Additional information about the unbinding process
antibody–antigen recognition events were also recorded is contained in the distributions of the unbinding forces.
with tip-bound antigens interacting with intact antibod- Concomitant with the shift of maxima to higher unbind-
ies [27.9,12] or with single-chain Fv fragments [27.13]. ing forces, increasing the loading rate also leads to an
The latter study also showed that an Fv mutant whose increase in the width σ of the distributions [27.23, 41],
affinity to the antigen was attenuated by about 10-fold indicating that at lower loading rates the system ad-
dissociated from the antigen under applied forces that justs closer to equilibrium. The lifetime τ( f ) of a bond
were 20% lower than those required to unbind the wild- under an applied force was estimated by the time the
type (Fv) antibody. cantilever spends in the force window spanned by the
Molecular Recognition Force Microscopy 27.5 Recognition Force Spectroscopy: From Isolated Molecules to Biological Membranes 773
standard deviation of the most probable force for un- et al. also suggested a method to analyze force spec-
binding [27.3]. In the case of Ni2+ -His6 , the lifetime tra which also does not assume stationarity of the
of the complex decreased from 17 to 2.5 ms when the energy barrier [27.71]. In their treatment, they find
force was increased from 150 to 194 pN [27.23]. The a (ln r)2/3 dependence of the mean force of disso-
data fit well to Bell’s model, confirming the predicted ciation, where r is the loading rate. They also find
exponential dependence of bond lifetime on the applied the distribution of unbinding forces to be asymmet-
force, and yielded an estimated lifetime at zero force of ric, as indeed observed many times. Evstigneev and
about 15 s. A more direct measurement of τ is afforded Reimann [27.72] suggest that the practice of fitting this
by force-clamp experiments in which the applied force asymmetric distribution with a Gaussian one in order
is kept constant by a feedback loop. This configura- to extract the mean rupture leads to the latter’s over-
tion was first adapted for use with AFM by Oberhauser estimation and consequently to an overestimate of the
et al. [27.66], who employed it to study the force de- force-free dissociation rate. They suggest an optimized
pendence of the unfolding probability of the I27 and statistical data analysis which overcomes this limitation
I28 modules of cardiac titin as well as of the full-length by combining data at many pulling rates into a sin-
protein [27.66]. gle distribution of the probability of rupture versus
However, as discussed above, in most experiments force.
the applied force is not constant but varies with time, The force spectra may also be used to derive the
and the measured bond strength depends on the loading dissociation rate constant koff by extrapolation to zero
rate [27.55, 57, 67]. In accordance with this, exper- force [27.59, 64, 65]. As mentioned above, values de-
imentally measured unbinding forces do not assume rived in this manner may differ from those obtained
unitary values but rather vary with both pulling veloc- from bulk measurements because only a subset of dis-
ity [27.59,64] and cantilever spring constant [27.1]. The sociation pathways defined by the force is sampled.
slopes of the force versus loading rate curves contain in- Nevertheless, a simple correlation between unbind-
formation about the length scale xβ of prominent energy ing forces and thermal dissociation rates was obtained
barriers along the force-driven dissociation pathway, for a set consisting of nine different Fv fragments
which may be related to the depth of the binding pocket constructed from point mutations of three unrelated
of the interaction [27.64]. The predicted logarithmic de- antifluorescein antibodies [27.65, 70]. This correlation,
pendence of the unbinding force on the loading rate which implies a close similarity between the force- and
holds well when the barriers are stationary with force, as thermally driven pathways explored during dissocia-
confirmed by a large number of unbinding and unfold- tion, was probably due to the highly rigid nature of the
ing experiments [27.14, 23, 41, 59, 64, 65, 68]. However, interaction, which proceeds in a lock-and-key fashion.
if the position of the transition state is expected to vary The force spectra obtained for the different constructs
along the reaction coordinate with the force, as for ex- exhibited a single linear regime, indicating that in all
ample when the curvature at the top of the barrier is cases unbinding was governed by a single prominent en-
small, the strict logarithmic dependence gives way to ergy barrier (Fig. 27.8). Interestingly, the position of the
more complex forms. energy barrier along the forced-dissociation pathway
Schleirf and Rief [27.69] used a Kramers diffusion was found to be proportional to the height of the bar-
model to calculate the probability force distributions rier and, thus, most likely includes contributions arising
when the barrier cannot be assumed to be stationary. from elastic stretching of the antibodies during the un-
Part C 27.5
Notably, although the position of the transition state binding process.
predicted by the Bell model was smaller than that A good correspondence between dissociation rates
predicted by the Kramer analysis by 6 Å, the most derived from mechanical unbinding experiments and
probable unfolding forces showed an almost perfect from bulk assays was also reported by Neuert
logarithmic dependence on the pulling velocity, indi- et al. [27.70]. In this case, the experimental system
cating that great care should be taken before the linear consisted of digoxigenin and its specific antibody. This
theory of DFS is applied. Failure to fit force distri- pair is used as a noncovalent coupler in various appli-
butions at high loading rates using a Bell model was cations, including forced-unbinding experiments. The
also reported by Neuert et al. [27.70] for the interac- force spectra obtained for the complex suggested that
tion of digoxigenin and antidigoxigenin. Poor matches the unbinding path is traversed by two activation energy
were observed in the crossover region between the two barriers located at xβ = 0.35 nm and xβ = 1.15 nm. Lin-
linear regimes of the force spectrum as well. Klafter ear fit of the low-force regime revealed a dissociation
774 Part C Scanning-Probe Microscopy
F (pN)
27.5.2 Complex Bonds
and Energy Landscapes
160 The energy landscapes that describe proteins are gener-
FITC-E2 w.t.
ally not smooth. Rather, they are traversed by multiple
koff (Solution) = 0.0044
120 energy barriers of various heights that render them
koff (Solution) = 0.062 highly corrugated or rugged. All these barriers affect the
kinetics and conformational dynamics of proteins and
80 any one of them may govern interaction lifetime and
4D5-Flu strength on certain time scales. Dynamic force spec-
40 troscopy provides an excellent tool to detect energy
barriers which are difficult or impossible to detect by
conventional, near-equilibrium assays and to probe the
0 free energy surface of proteins and protein complexes.
0.01 0.1 1 10 100 1000 104
r (pN /s)
It also provides a natural means to study interactions
which are normally subjected to varying mechanical
Fig. 27.8 The dependence of the unbinding force on the loads [27.59, 64, 73–75].
loading rate for two antifluorescein antibodies. For both A beautiful demonstration of the ability of dy-
FITC-E2 w.t. and 4D5-Flu a strictly single-exponential namic force spectroscopy to reveal hidden barriers was
dependence was found in the range accessed, indicating provided by Merkel et al. [27.41], who used BFP to
that only a single energy barrier was probed. The same probe bond formation between biotin and streptavidin
energy barrier dominates dissociation without forces ap- or avidin over a broad range of loading rates. In con-
plied because extrapolation to zero force matches kinetic trast to early studies which reported fixed values of
off-rates determined in solution (indicated by the arrow) bond strength [27.61, 62], a continuous spectrum of
(after [27.65]) unbinding forces ranging from 5 to 170 pN was ob-
tained (Fig. 27.9). Concomitantly, interaction lifetime
rate at zero force of 0.015 s−1 , in close agreement with decreased from about 1 min to 0.001 s, revealing the
the 0.023 s−1 value obtained from bulk measurements reciprocal relation between bond strength and lifetime
made on antidigoxigenin Fv fragments. expected for thermally activated kinetics under a ris-
a) Frequency b) c)
Force (pN) x ts
0.050
200
AFM
Force (pN) Streptavidin E (x) x
300
150 Avidin
0.025 200
100
100
Part C 27.5
100
50
102 –( f cos θ) x
4 0
10
10–2 100 102 104 106
Loading rate Loading rate (pN /s)
(pN / s) 106
Fig. 27.9a–c Unbinding force distributions and energy landscape of a complex molecular bond. (a) Force histograms of single
biotin–streptavidin bonds recorded at different loading rates. The shift in peak location and the increase in width with increasing
loading rate is clearly demonstrated. (b) Dynamic force spectra for biotin–streptavidin (circles) and biotin–avidin (triangles). The
slopes of the linear regimes mark distinct activation barriers along the direction of force. (c) Conceptual energy landscape traversed
along a reaction coordinate under force. The external force f adds a mechanical potential that tilts the energy landscape and lowers
the barriers. The inner barrier starts to dominate when the outer has fallen below it due to the applied force (after [27.41])
Molecular Recognition Force Microscopy 27.5 Recognition Force Spectroscopy: From Isolated Molecules to Biological Membranes 775
ing force. Most notably, depending on the loading rate, found by investigating the dependence of the adhe-
unbinding kinetics was dominated by different activa- sion probability between the two molecules on the
tion energy barriers positioned along the force-driven velocity of the AFM probe. Counterintuitively and
unbinding pathway. Barriers emerged sequentially, with in contrast to experiments with avidin–biotin [27.61],
the outermost barrier appearing first, each giving rise antibody–antigen [27.3], or cell adhesion proteogly-
to a distinct linear regime in the force spectrum. Going cans [27.8], the adhesion probability between P-selectin
from one linear regime to the next was associated with and PSGL-1 was found to increase with increasing ve-
an abrupt change in slope, indicating that a crossover locities [27.59]. This unexpected dependency explains
between an outer to (more) inner barrier had occurred. the increase in leukocyte tethering probability with in-
The position of two of the three barriers identified in the creased shear flow observed in rolling experiments.
force spectra was consistent with the location of promi- Since the adhesion probability approached 1, it was
nent transition states revealed by molecular dynamics concluded that binding occurs instantaneously as the
simulations [27.55, 67]. However, as was mentioned tip reaches the surface and, thus, proceeds with a very
earlier, unbinding is not necessarily confined to a sin- fast on-rate. The complex also exhibited a fast forced
gle, well-defined path, and may take different routes off-rate. Such a fast-on/fast-off kinetics is probably im-
even when directed by an external force. Molecular portant for the ability of leukocytes to bind and detach
dynamics simulations of force-driven unbinding of an rapidly from the endothelial cell surface. Likewise, the
antibody–antigen complex characterized by a highly long contour length of the complex together with its
flexible binding pocket revealed a large heterogeneity high elasticity reduces the mechanical loading on the
of enforced dissociation pathways [27.76]. complex upon binding and allows leukocyte rolling
The rolling of leukocytes on activated endothe- even at high shear rates.
lium is a first step in the emergence of leukocytes out Evans et al. [27.73] used BPF to study the inter-
of the blood stream into sites of inflammation. This action between PSGL-1 and another member of the
rolling, which occurs under hydrodynamic shear forces, selectin family, L-selectin. The force spectra, obtained
is mediated by selectins, a family of extended, calcium- over a range of loading rates extending from 10 to
dependent lectin receptors present on the surface of 100 000 pN/s, revealed two prominent energy barriers
endothelial cells. To fulfill their function, selectins and along the unbinding pathway: an outer barrier, probably
their ligands exhibit a unique combination of mechani- constituted by an array of hydrogen bonds, that im-
cal properties: they associate rapidly and avidly and can peded dissociation under slow detachment, and an inner,
tether cells over very long distances by their long, exten- Ca2+ -dependent barrier that dominated dissociation un-
sible structure. In addition, complexes formed between der rapid detachment. The observed hierarchy of inner
selectins and their ligands can withstand high tensile and outer activation barriers was proposed to be impor-
forces and dissociate in a controllable manner, which al- tant for multibond recruitment during selectin-mediated
lows them to maintain rolling without being pulled out function.
of the cell membrane. Using force-clamp AFM [27.66], bond lifetimes
Fritz et al. [27.59] used dynamic force spectroscopy were directly measured in dependence on a constantly
to study the interaction between P-selectin and its applied force. For this, lifetime–force relations of
leukocyte-expressed surface ligand P-selectin glycopro- P-selectin complexed to two forms of P-selectin gly-
tein ligand-1 (PSGL-1). Modeling both intermolecular coprotein ligand 1 (PSGL-1) and to G1, a blocking
Part C 27.5
and intramolecular forces, as well as adhesion proba- monoclonal antibody against P-selectin, respectively,
bility, they were able to obtain detailed information on were determined [27.75]. Both monomeric (sPSGL-1)
rupture forces, elasticity, and the kinetics of the inter- and dimeric PSGL-1 exhibited a biphasic relation-
action. Complexes were able to withstand forces up to ship between lifetime and force in their interaction
165 pN and exhibited a chain-like elasticity with a mo- to P-selectin (Fig. 27.10a,b). The bond lifetimes ini-
lecular spring constant of 5.3 pN/nm and a persistence tially increased, indicating the presence of catch bonds.
length of 0.35 nm. Rupture forces and the lifetime of After reaching a maximum, the lifetimes decreased
the complexes exhibited the predicted logarithmic de- with force, indicating a catch bond. In contrast, the
pendence on the loading rate. P-selectin/G1 bond lifetimes decreased exponentially
An important characteristics of the interaction with force (Fig. 27.10c), displaying typical slip bond
between P-selectin and PSGL-1, which is highly rel- characteristics that are well described by the single-
evant to the biological function of the complex, was energy-barrier Bell model. The curves of lifetime
776 Part C Scanning-Probe Microscopy
a) Time (s) Mean lifetime b) Time (s) Mean lifetime c) Time (s) Mean lifetime
1.5 1.5 2
sPSGL-1 PSGL-1 G1
1.2 1.2
1.5
0.9 0.9
1
0.6 0.6
0.5
0.3 0.3
0 0 0
0 20 40 60 0 20 40 60 0 20 40 60
Force f (pN) Force f (pN) Force f (pN)
Fig. 27.10a–c Lifetimes of bonds of single-molecular complexes, depending on a constantly applied force. (a) sPSGL-
1/P-selectin: catch bond and slip bond. (b) PSGL-1/P-selectin: catch bond and slip bond. (c) G1/P-selectin: slip bond only
(after [27.75])
against force for the two forms of PSGL1-1 had sim- centration in the body. Binding also revealed a strong
ilar biphasic shapes (Fig. 27.10a,b), but the PSGL-1 dependence on calcium concentration, giving rise to an
curve (Fig. 27.10b) was shifted relative to the sPSGL- unusually high Hill coefficient of ≈ 5. This steep depen-
1 curve (Fig. 27.10a), approximately doubling the force dency suggests that local changes of free extracellular
and the lifetime. These data suggest that sPSGL-1 forms Ca2+ in the narrow intercellular space may facilitate
monomeric bonds with P-selectin, whereas PSGL-1 rapid remodeling of intercellular adhesion and perme-
forms dimeric bonds with P-selectin. In agreement with ability.
the studies describes above, it was concluded that the Odorico et al. [27.77] used DFS to explore the
use of force-induced switching from catch to slip bonds energy landscape underlying the interaction between
might be physiologically relevant for the tethering and a chelated uranyl compound and a monoclonal antibody
rolling process of leukocytes on selectins [27.75]. raised against the uranyl-dicarboxy-phenanthroline
Baumgartner et al. [27.64] used AFM to probe complex. To isolate contributions of the uranyl moiety
specific trans-interaction forces and conformational to the binding interaction, measurements were per-
changes of recombinant vascular endothelial (VE)- formed with and without the ion in the chelating ligand.
cadherin strand dimers. VE-cadherins are cell-surface In the presence of uranyl, the force spectra contained
proteins that mediate the adhesion of cells in the vas- two linear regimes, suggesting the presence of at least
cular endothelium through Ca2+ -dependent homophilic two major energy barriers along the unbinding pathway.
interactions of their N-terminal extracellular domains. To relate the experimental data to molecular events, the
Acting as such they play an important role in the authors constructed a model with a variable fragment of
regulation of intercellular adhesion and communica- the antibody and used computational graphics to dock
tion in the inner surface of blood vessels. Unlike the chelated uranyl ion into the binding pocket. The
Part C 27.5
selectin-mediated adhesion, association between trans- analysis suggested that the inner barrier (xβ = 0.5 Å)
interacting VE dimers was slow and independent of reflects the rupture of coordination bonds between the
probe velocity, and complexes were ruptured at rela- uranium atom and an Asp residue, whereas the outer
tively low forces. These differences were attributed to barrier (xβ = 3.9 Å) amounts to the detachment of the
the fact that, as opposed to selectins, cadherins mediate entire ligand from the Ab binding site.
adhesion between resting cells. Mechanical stress on the Nevo et al. [27.78, 79] used single-molecule force
junctions is thus less intense and high-affinity binding spectroscopy to discriminate between alternative mech-
is not required to establish and maintain intercellu- anisms of protein activation (Fig. 27.11). The activation
lar adhesion. Determination of Ca2+ dependency of of proteins by other proteins, protein domains or small
recognition events between tip- and surface-bound VE- ligands is a central process in biology, e.g., in signalling
cadherins revealed a surprisingly high K D (1.15 mM), pathways and enzyme activity. Moreover, activation
which is very close to the free extracellular Ca2+ con- and deactivation of genes both depend on the switch-
Molecular Recognition Force Microscopy 27.5 Recognition Force Spectroscopy: From Isolated Molecules to Biological Membranes 777
ing of proteins between alternative functional states. with GDP, Ran associated weakly with impβ to form
Two general mechanisms have been proposed. The a single bound state characterized by unimodal distri-
induced-fit model assigns changes in protein activity butions of small unbinding forces (Fig. 27.11a, dotted
to conformational changes triggered by effector bind- line). Addition of Ran BP1 resulted in a marked shift of
ing. The population-shift model, on the other hand, the distribution to higher unbinding forces (Fig. 27.11b,
ascribes these changes to a redistribution of preexisting dashed to solid line). These results were interpreted
conformational isomers. According to this model, also to be consistent with an induced-fit mechanism where
known as the preequilibrium or conformational selec- binding of RanBP1 induces a conformational change in
tion model, protein structure is regarded as an ensemble the complex, which, in turn, strengthens the interaction
of conformations existing in equilibrium. The ligand between impβ and Ran(GDP). In contrast, association
binds to one of these conformations, i. e., the one to of RanGTP with impβ was found to lead to alterna-
which it is most complementary, thus shifting the equi- tive bound states of relatively low and high adhesion
librium in favor of this conformation. Discrimination strength represented by partially overlapping force dis-
between the two models of activation requires that the tributions (Fig. 27.11a, solid line). When RanBP1 was
distribution of conformational isomers in the ensem- added to the solution, the higher-strength population,
ble is known. Such information, however, is very hard which predominated the ensemble in the absence of
to obtain from conventional bulk methods because of the effector (Fig. 27.11c, dashed lines), was diminished,
ensemble averaging. and the lower-strength conformation became corre-
Using AFM, Nevo and coworkers measured the spondingly more populated (Fig. 27.11c, solid line).
unbinding forces of two related protein complexes in The means of the distributions, however, remain un-
the absence or presence of a common effector. The changed, indicating that the strength of the interaction
complexes consisted of the nuclear transport receptor in the two states of the complex had not been altered
importin β(impβ) and the small GTPase Ran. The dif- by the effector. These data fit a dynamic population-
ference between them was the nucleotide-bound state shift mechanism in which RanBP1 binds selectively
of Ran, which was either guanosine diphosphate (GDP) to the lower-strength conformation of RanGTP–impβ,
or guanosine-5’-triphosphate (GTP). The effector mol- changing the properties and function of the complex by
ecule was the Ran-binding protein RanBP1. Loaded shifting the equilibrium between its two states.
80 150
RanGppNHp –RanBP1
+RanBP1 120
60
90
60
RanGDP 40
2000 –RanBP1 +RanBP1
nm/s 30
Part C 27.5
20 0
0 50 100 150 200 250 102 103 104 0 103 104 105
Force (pN) Loading rate (pN/s) Loading rate (pN/s)
Fig. 27.11a–c Protein activation revealed by force spectroscopy. Ran and importin β (impβ) were immobilized onto
the AFM cantilevered tip and mica, respectively, and the interaction force was measured at different loading rates in
the absence or presence of RanBP1, which was added as a mobile substrate to the solution in the AFM liquid cell.
Unbinding force distributions obtained for impβ–Ran complexes at pulling velocity of 2000 nm/s. Association of impβ
with Ran loaded with GDP (a) or with nonhydrolyzable GTP analogue (GppNHp) (b) gives rise to uni- or bimodal force
distributions, respectively, reflecting the presence of one or two bound states. (b–c) Force spectra obtained for complexes
of impβ with RanGDP or with RanGppNHp, in the absence (dashed lines) or presence (solid lines) of RanBP1. The
results indicate that activation of impβ–RanGDP and imp–RanGTP complexes by RanBP1 proceeds through induced-fit
and dynamic population-shift mechanisms, respectively (see text for details) (after [27.78, 79])
778 Part C Scanning-Probe Microscopy
The complex between impβ and RanGTP was also consistent within each of the two regimes, averaging
used in studies aimed to measure the energy landscape at 7.5 and ≈ 5.5kB T along the outer and inner barri-
roughness of proteins. The roughness of the energy ers of the transition, respectively. The difference was
landscapes that describe proteins has numerous effects attributed to contributions from the intermediate state
on their folding and binding as well as on their behav- of the reaction, which is suppressed (along with the
ior at equilibrium, since undulations in the free energy outer barrier) at high loading rates. The origin of rough-
surface can attenuate diffusion dramatically. Thus, to ness was attributed to competition of solvent water
understand how proteins fold, bind, and function, one molecules with some of the hydrogen bonds that sta-
needs to know not only the energy of their initial and fi- bilize the complex and to the aforementioned 3–4 loop
nal states, but also the roughness of the energy surface of streptavidin, which is highly flexible and, therefore,
that connects them. However, for a long time, knowl- may induce the formation of multiple conformational
edge of protein energy landscape roughness came solely substates in the complex. It was also proposed by the
from theory and simulations of small model proteins authors that the large roughness detected in the energy
Adopting Zwanzig’s theory of diffusion in rough landscape of streptavidin–biotin is a significant contrib-
potentials [27.80], Hyeon and Thirumalai [27.81] pro- utor to the unusually slow dissociation kinetics of the
posed that the energy landscape roughness of proteins complex and may account for the discrepancies in the
can be measured from single-molecule mechanical unbinding forces measured for this pair.
unfolding experiments conducted at different temper-
atures. In particular, their simulations showed that at 27.5.3 Live Cells and Membranes
a constant loading rate the most probable force for
unfolding increases because of roughness that acts to Thus far, there have been only a few attempts to ap-
attenuate diffusion. Because this effect is tempera- ply recognition force spectroscopy to cells. In one
ture dependent, an overall energy scale of roughness, of the early studies, Lehenkari and Horton [27.84]
ε, can be derived from plots of force versus loading measured the unbinding forces between integrin recep-
rate acquired at two arbitrary temperatures. Extending tors present on the surface of intact cells and several
this theory to the case of unbinding, and perform- RGD-containing (Arg–Gly–Asp) ligands. The unbind-
ing single-molecule force spectroscopy measurements, ing forces measured were found to be cell and amino
Nevo et al. [27.82] extracted the overall energy scale acid sequence specific, and sensitive to pH and the diva-
of roughness ε for RanGTP–impβ. The results yielded lent cation composition of the cellular culture medium.
ε > 5kB T , indicating a bumpy energy surface, which is In contrast to short linear RGD hexapeptides, larger
consistent with the unusually high structural flexibility peptides and proteins containing the RGD sequence
of impβ and its ability to interact with different, struc- showed different binding affinities, demonstrating that
turally distinct ligands in a highly specific manner. This the context of the RGD motif within a protein has
mechanistic principle may also be applicable to other a considerable influence upon its interaction with the
proteins whose function demands highly specific and receptor. In another study, Chen and Moy [27.85] used
regulated interactions with multiple ligands. AFM to measure the adhesive strength between con-
More recently, the same type of analysis using three canavalin A (Con A) coupled to an AFM tip and
temperatures and pulling speeds in the range of 100 Con A receptors on the surface of NIH3T3 fibrob-
to 38 000 nm/s, was applied to derive ε for the well- lasts. Cross-linking of receptors with either glutaralde-
Part C 27.5
for the occurrence of HSP60 on the surface of stressed ers [27.74, 88]. The experimental system consisted of
HUVECs, but not on unstressed HUVECs. LFA-1-expressing Jurkat T-cells attached to the end of
The sidedness and accessibility of protein epitopes the AFM cantilever and surface-immobilized ICAM-1
of the Na2+ /d-glucose cotransporter 1 (SGLT1) was or -2. For both ICAM-1 and ICAM-2, the force spec-
probed in intact brush border membranes by a tip-bound tra exhibited fast and slow loading regimes, amounting
antibody directed against an amino acid sequence close to a sharp, inner energy barrier (xβ ≈ 0.56 Å and 1.5 Å,
to the glucose binding site [27.38]. Binding of glucose for complexes formed with ICAM-1 and ICAM-2) and
and transmembrane transport altered both the binding a shallow, outer barrier (xβ ≈ 3.6 Å and 4.9 Å), respec-
probability and the most probable unbinding force, sug- tively. Addition of Mg2+ led to an increase of the
gesting changes in the orientation and conformation rupture forces measured in the slow loading regime,
of the transporter. These studies were extended to live indicating an increment of the outer barrier in the pres-
SGLT1-transfected CHO cells [27.87]. Using AFM tips ence of the divalent cation. Comparison between the
carrying the substrate 1-β-thio-d-glucose, direct evi- force spectra obtained for the complexes formed be-
dence could be obtained that, in the presence of sodium, tween LFA-1 and ICAM-1 or ICAM-2 indicated that, in
a sugar binding site appears on the SGLT1 surface. the fast loading regime, the rupture of LFA-1–ICAM-1
It was shown that this binding site accepts the sugar depends more steeply on the loading rate than that of
residue of the glucoside phlorizin, free d-glucose and d- LFA-1–ICAM-2. The difference in dynamic strength
galactose, but not free l-glucose. The data indicate the between the two interactions was attributed to the pres-
importance of stereoselectivity for sugar binding and ence of wider barriers in the LFA-1–ICAM-2 complex,
transport. which render the interaction more receptive to the ap-
Studies on the interaction between leukocyte plied load. The enhanced sensitivity of complexes with
function-associated antigen-1 (LFA-1) and its cog- ICAM-2 to pulling forces was proposed to be important
nate ligand, intercellular adhesion molecules 1 and for the ability of ICAM-2 to carry out routine immune
2 (ICAM-1 and ICAM-2), which play a crucial role surveillance, which might otherwise be impeded due to
in leukocyte adhesion, revealed two prominent barri- frequent adhesion events.
Part C 27.6
sample elasticity were extracted from the local force damping. In addition, the ligand needs to be detached
ramps. This strategy was also used to map binding sites from the receptor in each retract–approach cycle, neces-
on cells [27.90, 91] and to differentiate between red sitating large working amplitudes (50 nm). Therefore,
blood cells of different blood groups (A and 0) us- the surface-bound receptor is inaccessible to the tip-
ing AFM tips functionalized with a group A-specific immobilized ligand on the tip during most of the time of
lectin [27.92]. the experiment. This problem, however, should be over-
Identification and localization of single antigenic come with the use of small cantilevers [27.53], which
sites was achieved by recording force signals during the should increase the speed for force mapping because the
scanning of an AFM tip coated with antibodies along hydrodynamic forces are significantly reduced and the
a single line across a surface immobilized with a low resonance frequency is higher than that of commercially
density of antigens [27.3, 11]. Using this method, anti- available cantilevers. Short cantilevers were recently
gens could be localized over the surface with positional applied to follow the association and dissociation of in-
780 Part C Scanning-Probe Microscopy
dividual chaperonin proteins, GroES to GroEL, in real each sinusoidal cantilever deflection period were de-
time using dynamic force microscopy topography imag- picted in a peak detector, filtered, and amplified.
ing [27.94]. Direct-current (DC) offset signals were used to com-
An imaging method for mapping antigenic sites on pensate for the thermal drifts of the cantilever. Uup
surfaces was developed [27.20] by combining molecu- and Udown were fed into the AFM controller, with
lar recognition force spectroscopy [27.3] with dynamic Udown driving the feedback loop to record the height
force microscopy (DFM) [27.28,96]. In DFM, the AFM (i. e., topography) image and Uup providing the data for
tip is oscillated across a surface and the amplitude re- constructing the recognition image (Fig. 27.12a). Since
duction arising from tip–surface interactions is held we used cantilevers with low Q-factor (≈ 1 in liquid)
constant by a feedback loop that lifts or lowers the tip driven at frequencies below resonance, the two types of
according to the detected amplitude signal. Since the information were independent. In this way, topography
tip contacts the surface only intermittently, this tech- and recognition image were recorded simultaneously
nique provides very gentle tip–surface interactions and and independently.
the specific interaction of the antibody on the tip with The circuit was applied to mica containing singly
the antigen on the surface can be used to localize anti- distributed avidin molecules using a biotinylated AFM
genic sites for recording recognition images. The AFM tip [27.95]. The sample was imaged with an antibody-
tip is magnetically coated and oscillated by an alter- containing tip, yielding the topography (Fig. 27.12b, left
nating magnetic field at very small amplitudes while image) and the recognition image (Fig. 27.12b, right
being scanned along the surface. Since the oscillation image) at the same time. The tip oscillation ampli-
frequency is more than a hundred times faster than typi- tude (5 nm) was chosen to be slightly smaller than
cal frequencies in conventional force mapping, the data the extended cross-linker length (8 nm), so that both
acquisition rate is much higher. This method was re- the antibody remained bound while passing a bind-
cently extended to yield fast, simultaneous acquisition ing site and the reduction of the upwards deflection
of two independent maps, i. e., a topography image and was of sufficient significance compared with the ther-
a lateral map of recognition sites, recorded with nm res- mal noise. Since the spring constant of the polymeric
olution at experimental times equivalent to normal AFM cross-linker increases nonlinearly with the tip–surface
imaging [27.95, 97, 98]. distance (Fig. 27.5), the binding force is only sensed
Topography and recognition images were simul- close to full extension of the cross-linker (given at
taneously obtained (TREC imaging) using a special the maxima of the oscillation period). Therefore, the
electronic circuit (PicoTrec, Agilent, Chandler, AZ) recognition signals were well separated from the to-
(Fig. 27.12a). Maxima (Uup ) and minima (Udown ) of pographic signals arising from the surface, in both
a) b)
Recognition
image
Pico
TREC
Topography
image
Part C 27.6
Oscillation
100 nm 100 nm
Scan direction 0 1 2 3 4 nm 0 1 2 3 4 nm
Fig. 27.12a,b Simultaneous topography and recognition (TREC) imaging. (a) Principle: the cantilever oscillation is split
into lower and upper parts, resulting in simultaneously acquired topography and recognition images. (b) Avidin was
electrostatically adsorbed to mica and imaged with a biotin-tethered tip. Good correlation between topography (left
image, bright spots) and recognition (right image, dark spots) was found (solid circles). Topographical spots without
recognition denote structures lacking specific interaction (dashed circle). Scan size was 500 nm (after [27.95])
Molecular Recognition Force Microscopy References 781
space (Δz ≈ 5 nm) and time (half-oscillation period methodology, topography and recognition images can
≈ 0.1 ms). be obtained at the same time and distinct receptor sites
The bright dots with 2–3 nm height and 15–20 nm in the recognition image can be assigned to structures
diameter visible in the topography image (Fig. 27.12b, from the topography image. This method is applicable
left image) represent single avidin molecules stably to any ligand, and therefore it should prove possible
adsorbed onto the flat mica surface. The recogni- to recognize many types of proteins or protein lay-
tion image shows black dots at positions of avidin ers and carry out epitope mapping on the nm scale on
molecules (Fig. 27.12b, right image) because the os- membranes, cells, and complex biological structures.
cillation maxima are lowered due to the physical In a striking recent example, histone proteins H3 were
avid–biotin connection established during recognition. identified and localized in a complex chromatin prepa-
That the lateral positions of the avidin molecules ob- ration [27.98].
tained in the topography image are spatially correlated Recently, TREC imaging was applied to gently
with the recognition signals of the recognition image fixed microvascular endothelial cells from mouse my-
is indicated by solid circles in the images (Fig. 27.12). ocardium (MyEnd) in order to visualize binding sites
Recognition between the antibody on the tip and the of VE-cadherin, known to play a crucial role in ho-
avidin on the surface took place for almost all avidin mophilic cell adhesion [27.99]. TREC images were
molecules, most likely because avidin contains four bi- acquired with AFM tips coated with a recombinant
otin binding sites, two on either side. Thus, one would VE-cadherin. The recognition images revealed promi-
assume to have always binding epitopes oriented away nent, irregularly shaped dark spots (domains) with
from the mica surface and accessible to the biotinylated size from 30 to 250 nm. The domains enriched in
tip, resulting in a high binding efficiency. Structures VE-cadherins molecules were found to be collocated
observed in the topography image and not detected in with the cytoskeleton filaments supporting the an-
the recognition image were very rare (dotted circle in chorage of VE-cadherins to F-actin. Compared with
Fig. 27.12b). conventional techniques such as immunochemistry or
It is important to note that topography and recog- single-molecule optical microscopy, TREC represents
nition images were recorded at speeds typical for an alternative method to quickly obtain the local distri-
standard AFM imaging and were therefore consider- bution of receptors on cell surface with unprecedented
ably faster than conventional force mapping. With this lateral resolution of several nm.
Part C 27
the proof-of-principle stage of the pioneering experi- dynamics of biological processes and interactions.
References
27.1 G.U. Lee, D.A. Kidwell, R.J. Colton: Sensing discrete ization of individual antibody-antigen recognition
streptavidin-biotin interactions with atomic force events by atomic force microscopy, Proc. Natl.
microscopy, Langmuir 10, 354–357 (1994) Acad. Sci. USA 93, 3477–3481 (1996)
27.2 E.L. Florin, V.T. Moy, H.E. Gaub: Adhesion forces 27.4 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi-
between individual ligand receptor pairs, Science croscope, Phys. Rev. Lett. 56, 930–933 (1986)
264, 415–417 (1994) 27.5 M. Grandbois, W. Dettmann, M. Benoit, H.E. Gaub:
27.3 P. Hinterdorfer, W. Baumgartner, H.J. Gruber, How strong is a covalent bond, Science 283, 1727–
K. Schilcher, H. Schindler: Detection and local- 1730 (1999)
782 Part C Scanning-Probe Microscopy
27.6 T. Boland, B.D. Ratner: Direct measurement of hy- of amines with thiols, Bioconjug. Chem. 6, 242–248
drogen bonding in DNA nucleotide bases by atomic (1995)
force microscopy, Proc. Natl. Acad. Sci. USA 92, 27.19 A.S.M. Kamruzzahan, A. Ebner, L. Wildling, F. Kien-
5297–5301 (1995) berger, C.K. Riener, C.D. Hahn, P.D. Pollheimer,
27.7 P. Wagner, M. Hegner, P. Kernen, F. Zaugg, P. Winklehner, M. Holzl, B. Lackner, D.M. Schorkl,
G. Semenza: Covalent immobilization of native P. Hinterdorfer, H.J. Gruber: Antibody linking
biomolecules onto Au(111) via N-hydroxysuccinim- to atomic force microscope tips via disulfide
ide ester functionalized self assembled monolayers bond formation, Bioconjug. Chem. 17(6), 1473–1481
for scanning probe microscopy, Biophys. J. 70, (2006)
2052–2066 (1996) 27.20 A. Raab, W. Han, D. Badt, S.J. Smith-Gill,
27.8 U. Dammer, O. Popescu, P. Wagner, D. Anselmetti, S.M. Lindsay, H. Schindler, P. Hinterdorfer: Anti-
H.-J. Güntherodt, G.M. Misevic: Binding strength body recognition imaging by force microscopy, Nat.
between cell adhesion proteoglycans measured Biotech. 17, 902–905 (1999)
by atomic force microscopy, Science 267, 1173–1175 27.21 G.U. Lee, A.C. Chrisey, J.C. Colton: Direct mea-
(1995) surement of the forces between complementary
27.9 U. Dammer, M. Hegner, D. Anselmetti, P. Wag- strands of DNA, Science 266, 771–773 (1994)
ner, M. Dreier, W. Huber, H.-J. Güntherodt: Specific 27.22 M. Conti, G. Falini, B. Samori: How strong is the
antigen/antibody interactions measured by force coordination bond between a histidine tag and Ni-
microscopy, Biophys. J. 70, 2437–2441 (1996) nitriloacetate? An experiment of mechanochem-
27.10 Y. Harada, M. Kuroda, A. Ishida: Specific and quan- istry on single molecules, Angew. Chem. 112,
tized antibody-antigen interaction by atomic force 221–224 (2000)
microscopy, Langmuir 16, 708–715 (2000) 27.23 F. Kienberger, G. Kada, H.J. Gruber, V.P. Pas-
27.11 P. Hinterdorfer, K. Schilcher, W. Baumgartner, tushenko, C. Riener, M. Trieb, H.-G. Knaus,
H.J. Gruber, H. Schindler: A mechanistic study of H. Schindler, P. Hinterdorfer: Recognition force
the dissociation of individual antibody-antigen spectroscopy studies of the NTA-His6 bond, Single
pairs by atomic force microscopy, Nanobiology 4, Mol. 1, 59–65 (2000)
39–50 (1998) 27.24 L. Schmitt, M. Ludwig, H.E. Gaub, R. Tampe:
27.12 S. Allen, X. Chen, J. Davies, M.C. Davies, A metal-chelating microscopy tip as a new toolbox
A.C. Dawkes, J.C. Edwards, C.J. Roberts, J. Sefton, for single-molecule experiments by atomic force
S.J.B. Tendler, P.M. Williams: Spatial mapping of microscopy, Biophys. J. 78, 3275–3285 (2000)
specific molecular recognition sites by atomic force 27.25 S. Lata, A. Reichel, R. Brock, R. Tampe, J. Piehler:
microscopy, Biochemistry 36, 7457–7463 (1997) High-affinity adaptors for switchable recognition
27.13 R. Ros, F. Schwesinger, D. Anselmetti, M. Kubon, of histidine-tagged proteins, J. Am. Chem. Soc. 127,
R. Schäfer, A. Plückthun, L. Tiefenauer: Antigen 10205–10215 (2005)
binding forces of individually addressed single- 27.26 A. Ebner, L. Wildling, R. Zhu, C. Rankl, T. Hasel-
chain Fv antibody molecules, Proc. Natl. Acad. Sci. grübler, P. Hinterdorfer, H.J. Gruber: Function-
USA 95, 7402–7405 (1998) alization of probe tips and supports for single
27.14 T. Strunz, K. Oroszlan, R. Schäfer, H.-J. Gün- molecule force microscopy, Top. Curr. Chem. 285,
therodt: Dynamic force spectroscopy of single DNA 29–76 (2008)
molecules, Proc. Natl. Acad. Sci. USA 96, 11277–11282 27.27 C. Yuan, A. Chen, P. Kolb, V.T. Moy: Energy land-
(1999) scape of avidin-biotin complexes measured burey
27.15 A. Ebner, P. Hinterdorfer, H.J. Gruber: Comparison atomic force microscopy, Biochemistry 39, 10219–
of different aminofunctionalization strategies for 10223 (2000)
attachment of single antibodies to AFM cantilevers, 27.28 W. Han, S.M. Lindsay, M. Dlakic, R.E. Harrington:
Part C 27
Ultramicroscopy 107, 922–927 (2007) Kinked DNA, Nature 386, 563 (1997)
27.16 S.S. Wong, E. Joselevich, A.T. Woolley, C.L. Cheung, 27.29 G. Kada, L. Blaney, L.H. Jeyakumar, F. Kien-
C.M. Lieber: Covalently functionalyzed nanotubes berger, V.P. Pastushenko, S. Fleischer, H. Schindler,
as nanometre-sized probes in chemistry and biol- F.A. Lai, P. Hinterdorfer: Recognition force mi-
ogy, Nature 394, 52–55 (1998) croscopy/spectroscopy of ion channels: Applica-
27.17 P. Hinterdorfer, F. Kienberger, A. Raab, H.J. Gruber, tions to the skeletal muscle Ca2+ release channel
W. Baumgartner, G. Kada, C. Riener, S. Wielert- (RYR1), Ultramicroscopy 86, 129–137 (2001)
Badt, C. Borken, H. Schindler: Poly(ethylene glycol): 27.30 D.J. Müller, W. Baumeister, A. Engel: Controlled un-
An ideal spacer for molecular recognition force mi- zipping of a bacterial surface layer atomic force
croscopy/spectroscopy, Single Mol. 1, 99–103 (2000) microscopy, Proc. Natl. Acad. Sci. USA 96, 13170–
27.18 T. Haselgrübler, A. Amerstorfer, H. Schindler, 13174 (1999)
H.J. Gruber: Synthesis and applications of a new 27.31 F. Oesterhelt, D. Oesterhelt, M. Pfeiffer, A. Engle,
poly(ethylene glycol) derivative for the crosslinking H.E. Gaub, D.J. Müller: Unfolding pathways of in-
Molecular Recognition Force Microscopy References 783
dividual bacteriorhodopsins, Science 288, 143–146 stranded and single-stranded DNA molecules,
(2000) Science 271, 795–799 (1996)
27.32 E. Kiss, C.-G. Gölander: Chemical derivatization of 27.48 M.S.Z. Kellermayer, S.B. Smith, H.L. Granzier,
muscovite mica surfaces, Colloids Surf. 49, 335–342 C. Bustamante: Folding-unfolding transitions in
(1990) single titin molecules characterized with laser
27.33 S. Karrasch, M. Dolder, F. Schabert, J. Ramsden, tweezers, Sience 276, 1112–1216 (1997)
A. Engel: Covalent binding of biological samples 27.49 T.R. Strick, J.F. Allemend, D. Bensimon, A. Ben-
to solid supports for scanning probe microscopy in simon, V. Croquette: The elasticity of a single
buffer solution, Biophys. J. 65, 2437–2446 (1993) supercoiled DNA molecule, Biophys. J. 271, 1835–
27.34 N.H. Thomson, B.L. Smith, N. Almqvist, L. Schmitt, 1837 (1996)
M. Kashlev, E.T. Kool, P.K. Hansma: Oriented, active 27.50 T. Lionnet, D. Joubaud, R. Lavery, D. Bensimon,
escherichia coli RNA polymerase: An atomic force V. Croquette: Wringing out DNA, Phys. Rev. Lett.
microscopy study, Biophys. J. 76, 1024–1033 (1999) 96, 178102 (2006)
27.35 G. Kada, C.K. Riener, P. Hinterdorfer, F. Kienberger, 27.51 F. Kienberger, V.P. Pastushenko, G. Kada, H.J. Gru-
C.M. Stroh, H.J. Gruber: Dithio-phospholipids for ber, C. Riener, H. Schindler, P. Hinterdorfer: Static
biospecific immobilization of proteins on gold sur- and dynamical properties of single poly(ethylene
faces, Single Mol. 3, 119–125 (2002) glycol) molecules investigated by force spec-
27.36 C. LeGrimellec, E. Lesniewska, M.C. Giocondi, troscopy, Single Mol. 1, 123–128 (2000)
E. Finot, V. Vie, J.P. Goudonnet: Imaging of the 27.52 S. Liang, D. Medich, D.M. Czajkowsky, S. Sheng,
surface of living cells by low-force contact-mode J.-Y. Yuan, Z. Shao: Thermal noise reduction of
atomic force microscopy, Biophys. J. 75(2), 695–703 mechanical oscillators by actively controlled exter-
(1998) nal dissipative forces, Ultramicroscopy 84, 119–125
27.37 K. Schilcher, P. Hinterdorfer, H.J. Gruber, (2000)
H. Schindler: A non-invasive method for the tight 27.53 M.B. Viani, T.E. Schäffer, A. Chand, M. Rief,
anchoring of cells for scanning force microscopy, H.E. Gaub, P.K. Hansma: Small cantilevers for force
Cell. Biol. Int. 21, 769–778 (1997) spectroscopy of single molecules, J. Appl. Phys. 86,
27.38 S. Wielert-Badt, P. Hinterdorfer, H.J. Gruber, J.- 2258–2262 (1999)
T. Lin, D. Badt, H. Schindler, R.K.-H. Kinne: Single 27.54 T. Strunz, K. Oroszlan, I. Schumakovitch, H.-
molecule recognition of protein binding epitopes J. Güntherodt, M. Hegner: Model energy land-
in brush border membranes by force microscopy, scapes and the force-induced dissociation of
Biophys. J. 82, 2767–2774 (2002) ligand-receptor bonds, Biophys. J. 79, 1206–1212
27.39 P. Bongrand, C. Capo, J.-L. Mege, A.-M. Benoliel: (2000)
Use of hydrodynamic flows to study cell adhesion. 27.55 H. Grubmüller, B. Heymann, P. Tavan: Ligand
In: Physical Basis of Cell Adhesion, ed. by P. Bon- binding: Molecular mechanics calculation of the
grand (CRC Press, Boca Raton 1988) pp. 125–156 streptavidin-biotin rupture force, Science 271, 997–
27.40 J.N. Israelachvili: Intermolecular and Surface 999 (1996)
Forces, 2nd edn. (Academic, London New York 1991) 27.56 G.I. Bell: Models for the specific adhesion of cells
27.41 R. Merkel, P. Nassoy, A. Leung, K. Ritchie, E. Evans: to cells, Science 200, 618–627 (1978)
Energy landscapes of receptor-ligand bonds ex- 27.57 E. Evans, K. Ritchie: Dynamic strength of molecular
plored by dynamic force spectroscopy, Nature 397, adhesion bonds, Biophys. J. 72, 1541–1555 (1997)
50–53 (1999) 27.58 E. Evans, K. Ritchie: Strength of a weak bond-
27.42 A. Askin: Optical trapping and manipulation of connecting flexible polymer chains, Biophys. J. 76,
neutral particles using lasers, Proc. Natl. Acad. Sci. 2439–2447 (1999)
USA 94, 4853–4860 (1997) 27.59 J. Fritz, A.G. Katopidis, F. Kolbinger, D. Ansel-
Part C 27
27.43 K.C. Neuman, S.M. Block: Optical trapping, Rev. Sci. metti: Force-mediated kinetics of single P-
Instrum. 75, 2787–2809 (2004) selectin/ligand complexes observed by atomic force
27.44 K. Svoboda, C.F. Schmidt, B.J. Schnapp, S.M. Block: microscopy, Proc. Natl. Acad. Sci. USA 95, 12283–
Direct observation of kinesin stepping by optical 12288 (1998)
trapping interferometry, Nature 365, 721–727 (1993) 27.60 T. Auletta, M.R. de Jong, A. Mulder, F.C.J.M. van
27.45 S.M. Block, C.L. Asbury, J.W. Shaevitz, M.J. Lang: Veggel, J. Huskens, D.N. Reinhoudt, S. Zou, S. Za-
Probing the kinesin reaction cycle with a 2D optical potocny, H. Schönherr, G.J. Vancso, L. Kuipers:
force clamp, Proc. Natl. Acad. Sci. USA 100, 2351– β-cyclodextrin host-guest complexes probed un-
2356 (2003) der thermodynamic equilibrium: Thermodynamics
27.46 A.E.M. Clemen, M. Vilfan, J. Jaud, J. Zhang, M. Bar- and force spectroscopy, J. Am. Chem. Soc. 126,
mann, M. Rief: Force-dependent stepping kinetics 1577–1584 (2004)
of myosin-V, Biophys. J. 88, 4402–4410 (2005) 27.61 V.T. Moy, E.-L. Florin, H.E. Gaub: Adhesive forces
27.47 S. Smith, Y. Cui, C. Bustamante: Overstretching between ligand and receptor measured by AFM,
B-DNA: The elastic response of individual double- Science 266, 257–259 (1994)
784 Part C Scanning-Probe Microscopy
27.62 A. Chilkoti, T. Boland, B. Ratner, P.S. Stayton: bonds involving cell adhesion molecules, Nature
The relationship between ligand-binding ther- 423, 190–193 (2003)
modynamics and protein-ligand interaction forces 27.76 B. Heymann, H. Grubmüller: Molecular dynamics
measured by atomic force microscopy, Biophys. J. force probe simulations of antibody/antigen un-
69, 2125–2130 (1995) binding: Entropic control and non additivity of
27.63 I. Schumakovitch, W. Grange, T. Strunz, P. Bertonci- unbinding forces, Biophys. J. 81, 1295–1313 (2001)
ni, H.-J. Güntherodt, M. Hegner: Temperature 27.77 M. Odorico, J.M. Teulon, T. Bessou, C. Vidaud,
dependence of unbinding forces between com- L. Bellanger, S.W. Chen, E. Quemeneur, P. Parot,
plementary DNA strands, Biophys. J. 82, 517–521 J.L. Pellequer: Energy landscape of chelated uranyl:
(2002) Antibody interactions by dynamic force spec-
27.64 W. Baumgartner, P. Hinterdorfer, W. Ness, A. Raab, troscopy, Biophys. J. 93, 645 (2007)
D. Vestweber, H. Schindler, D. Drenckhahn: 27.78 R. Nevo, C. Stroh, F. Kienberger, D. Kaftan,
Cadherin interaction probed by atomic force mi- V. Brumfeld, M. Elbaum, Z. Reich, P. Hinterdor-
croscopy, Proc. Natl. Acad. Sci. USA 8, 4005–4010 fer: A molecular switch between two bound states
(2000) in the RanGTP-importinβ1 interaction, Nat. Struct.
27.65 F. Schwesinger, R. Ros, T. Strunz, D. Ansel- Mol. Biol. 10, 553–557 (2003)
metti, H.-J. Güntherodt, A. Honegger, L. Jermutus, 27.79 R. Nevo, V. Brumfeld, M. Elbaum, P. Hinter-
L. Tiefenauer, A. Plückthun: Unbinding forces of dorfer, Z. Reich: Direct discrimination between
single antibody-antigen complexes correlate with models of protein activation by single-molecule
their thermal dissociation rates, Proc. Natl. Acad. force measurements, Biophys. J. 87, 2630–2634
Sci. USA 29, 9972–9977 (2000) (2004)
27.66 A.F. Oberhauser, P.K. Hansma, M. Carrion-Vazquez, 27.80 R. Zwanzig: Diffusion in a rough potential, Proc.
J.M. Fernandez: Stepwise unfolding of titin under Natl. Acad. Sci. USA 85, 2029–2030 (1988)
force-clamp atomic force microscopy, Proc. Natl. 27.81 C.B. Hyeon, D. Thirumalai: Can energy landscape
Acad. Sci. USA 16, 468–472 (2000) roughness of proteins and RNA be measured by us-
27.67 S. Izraelev, S. Stepaniants, M. Balsera, Y. Oono, ing mechanical unfolding experiments?, Proc. Natl.
K. Schulten: Molecular dynamics study of unbind- Acad. Sci. USA 100, 10249–10253 (2003)
ing of the avidin-biotin complex, Biophys. J. 72, 27.82 R. Nevo, V. Brumfeld, R. Kapon, P. Hinterdorfer,
1568–1581 (1997) Z. Reich: Direct measurement of protein energy
27.68 M. Rief, F. Oesterhelt, B. Heyman, H.E. Gaub: Sin- landscape roughness, EMBO Reports 6, 482–486
gle molecule force spectroscopy on polysaccharides (2005)
by atomic force microscopy, Science 275, 1295–1297 27.83 F. Rico, V.T. Moy: Energy landscape roughness of
(1997) the streptavidin-biotin interaction, J. Mol. Recog-
27.69 M. Schlierf, M. Rief: Single-molecule unfolding nit. 20, 495–501 (2007)
force distributions reveal a funnel-shaped energy 27.84 P.P. Lehenkari, M.A. Horton: Single integrin mol-
landscape, Biophys. J. 90, L33 (2006) ecule adhesion forces in intact cells measured by
27.70 G. Neuert, C. Albrecht, E. Pamir, H.D. Gaub: atomic force microscopy, Biochem. Biophys. Res.
Dynamic force spectroscopy of the digoxigenin- Commun. 259, 645–650 (1999)
antibody complex, FEBS Letters 580, 505–509 27.85 A. Chen, V.T. Moy: Cross-linking of cell surface
(2006) receptors enhances cooperativity of molecular ad-
27.71 O.K. Dudko, A.E. Filippov, J. Klafter, M. Urback: hesion, Biophys. J. 78, 2814–2820 (2000)
Beyond the conventional description of dynamic 27.86 G. Pfister, C.M. Stroh, H. Perschinka, M. Kind,
force spectroscopy of adhesion bonds, Proc. Natl. M. Knoflach, P. Hinterdorfer, G. Wick: Detection
Acad. Sci. USA 100, 11378–11381 (2003) of HSP60 on the membrane surface of stressed
Part C 27
27.72 M. Evstigneev, P. Reimann: Dynamic force spec- human endothelial cells by atomic force and
troscopy: Optimized data analysis, Phys. Rev. E 68, confocal microscopy, J. Cell Sci. 118, 1587–1594
045103(R) (2003) (2005)
27.73 E. Evans, E. Leung, D. Hammer, S. Simon: 27.87 T. Puntheeranurak, L. Wildling, H.J. Gruber,
Chemically distinct transition states govern rapid R.K.H. Kinne, P. Hinterdorfer: Ligands on the
dissociation of single L-selectin bonds under force, string: single molecule studies on the interaction
Proc. Natl. Acad. Sci. USA 98, 3784–3789 (2001) of antibodies and substrates with the surface of
27.74 X. Zhang, E. Woijcikiewicz, V.T. Moy: Force spec- the Na+ -glucose cotransporter SGLT1 in living cells,
troscopy of the leukocyte function-associated J. Cell Sci. 119, 2960–2967 (2006)
antigen-1/intercellular adhesion molecule-1 inter- 27.88 E.P. Wojcikiewicz, M.H. Abdulreda, X. Zhang,
action, Biophys. J. 83, 2270–2279 (2002) V.T. Moy: Force spectroscopy of LFA-1 and its lig-
27.75 B.T. Marshall, M. Long, J.W. Piper, T. Yago, ands, ICAM-1 and ICAM-2, Biomacromolecules 7,
R.P. McEver, Z. Zhu: Direct observation of catch 3188 (2006)
Molecular Recognition Force Microscopy References 785
27.89 M. Ludwig, W. Dettmann, H.E. Gaub: Atomic force protein interactions in real time, Nat. Struct. Biol.
microscopy imaging contrast based on molecuar 7, 644–647 (2000)
recognition, Biophys. J. 72, 445–448 (1997) 27.95 A. Ebner, F. Kienberger, G. Kada, C.M. Stroh,
27.90 P.P. Lehenkari, G.T. Charras, G.T. Nykänen, M. Geretschläger, A.S.M. Kamruzzahan, L. Wildling,
M.A. Horton: Adapting force microscopy for cell W.T. Johnson, B. Ashcroft, J. Nelson, S.M. Lindsay,
biology, Ultramicroscopy 82, 289–295 (2000) H.J. Gruber, P. Hinterdorfer: Localization of sin-
27.91 N. Almqvist, R. Bhatia, G. Primbs, N. Desai, gle avidin biotin interactions using simultaneous
S. Banerjee, R. Lal: Elasticity and adhesion force topography and molecular recognition imaging,
mapping reveals real-time clustering of growth ChemPhysChem 6, 897–900 (2005)
factor receptors and associated changes in lo- 27.96 W. Han, S.M. Lindsay, T. Jing: A magnetically driven
cal cellular rheological properties, Biophys. J. 86, oscillating probe microscope for operation in li-
1753–1762 (2004) quid, Appl. Phys. Lett. 69, 1–3 (1996)
27.92 M. Grandbois, M. Beyer, M. Rief, H. Clausen- 27.97 C.M. Stroh, A. Ebner, M. Geretschläger, G. Freuden-
Schaumann, H.E. Gaub: Affinity imaging of red thaler, F. Kienberger, A.S.M. Kamruzzahan,
blood cells using an atomic force microscope, S.J. Smith-Gill, H.J. Gruber, P. Hinterdorfer: Simul-
J. Histochem. Cytochem. 48, 719–724 (2000) taneous topography and recognition imaging using
27.93 O.H. Willemsen, M.M.E. Snel, K.O. van der Werf, force microscopy, Biophys. J. 87, 1981–1990 (2004)
B.G. de Grooth, J. Greve, P. Hinterdorfer, H.J. Gru- 27.98 C. Stroh, H. Wang, R. Bash, B. Ashcroft, J. Nelson,
ber, H. Schindler, Y. van Kyook, C.G. Figdor: H.J. Gruber, D. Lohr, S.M. Lindsay, P. Hinterdorfer:
Simultaneous height and adhesion imaging of Single-molecule recognition imaging microscope,
antibody antigen interactions by atomic force mi- Proc. Natl. Acad. Sci. USA 101, 12503–12507 (2004)
croscopy, Biophys. J. 57, 2220–2228 (1998) 27.99 L. Chtcheglova, J. Waschke, L. Wildling, D. Drenck-
27.94 B.V. Viani, L.I. Pietrasanta, J.B. Thompson, hahn, P. Hinterdorfer: Nano-scale dynamic recog-
A. Chand, I.C. Gebeshuber, J.H. Kindt, M. Richter, nition imaging on vascular endothelial cells,
H.G. Hansma, P.K. Hansma: Probing protein- Biophys. J. 9(3), L11–L13 (2007)
Part C 27
787
Part D
Bio-/Nan Part D Bio-/Nanotribology and Bio-/Nanomechanics
Nanotribolog
28. Nanotribology, Nanomechanics,
Part D 28
and Materials Characterization
Bharat Bhushan
The mechanisms and dynamics of the interactions of been used for mechanical and electrical characteriza-
two contacting solids during relative motion, ranging tion, in situ characterization of local deformation, and
from the atomic to microscale, need to be under- other nanomechanics studies.
stood in order to develop a fundamental understanding Nanotribological and nanomechanics studies are
of adhesion, friction, wear, indentation, and lubri- needed to develop a fundamental understanding of
cation processes. For most solid–solid interfaces of interfacial phenomena on a small scale and to study in-
technological relevance, contact occurs at multiple as- terfacial phenomena in nanostructures used in magnetic
perities. Consequently the importance of investigating storage devices, nanotechnology, and other applica-
single-asperity contacts in studies of the fundamen- tions [28.4–20]. Friction and wear of lightly loaded
tal micro/nanomechanical and micro/nanotribological micro/nanocomponents are highly dependent on the
properties of surfaces and interfaces has long been rec- surface interactions (few atomic layers). These struc-
ognized. The recent emergence and proliferation of tures are generally coated with molecularly thin films.
proximal probes, in particular scanning probe micro- Nanotribological and nanomechanics studies are also
scopies (the scanning tunneling microscope and the valuable in the fundamental understanding of interfacial
atomic force microscope), surface force apparatus, and phenomena in macrostructures, and provide a bridge
computational techniques for simulating tip–surface between science and engineering.
interactions and interfacial properties have allowed The surface force apparatus (SFA), the scanning
systematic investigations of interfacial problems with tunneling microscopes (STM), and atomic force and
high resolution as well as ways and means for mod- friction force microscopes (AFM and FFM) are widely
ifying and manipulating nanoscale structures. These used in nanotribological and nanomechanics studies.
advances have led to the appearance of the new field Typical operating parameters are compared in Ta-
of nanotribology, which pertains to experimental and ble 28.1. The SFA was developed in 1968 and is
theoretical investigations of interfacial processes on commonly employed to study both static and dynamic
scales ranging from the atomic and molecular to the properties of molecularly thin films sandwiched be-
microscale, occurring during adhesion, friction, scratch- tween two molecularly smooth surfaces. The STM,
ing, wear, indentation, and thin-film lubrication at developed in 1981, allows imaging of electrically con-
sliding surfaces [28.1–14]. Proximal probes have also ducting surfaces with atomic resolution, and has been
Table 28.1 Comparison of typical operating parameters in SFA, STM, and AFM/FFM used for micro/nanotribological studies
Operating parameter SFA STM a AFM/FFM
Radius of mating surface/tip ≈ 10 mmb 5 –100 nm 5 –100 nm
Radius of contact area 10–40 μm N/A 0.05– 0.5 nm
Normal load 10–100 mN N/A < 0.1 –500 nN
Sliding velocity 0.001–100 μm/s 0.02–200 μm/s 0.02– 200 μm/s
(scan size ≈ 1 nm × 1 nm (scan size ≈ 1 nm × 1 nm
to 125 μm × 125 μm; to 125 μm × 125 μm;
scan rate < 1 –122 Hz) scan rate < 1 –122 Hz)
Sample limitations Typically atomically smooth, Electrically conducting None of the above
optically transparent mica; samples
opaque ceramic,
smooth surfaces can also be used
a Can be used for atomic-scale imaging
b Since stresses scale inverse of tip radius, SFA can provide low stress measurement capabilities
Nanotribology, Nanomechanics, and Materials Characterization 28.1 Description of AFM/FFM and Various Measurement Techniques 791
Part D 28.1
bricant molecules. The introduction of the AFM in
1985 provided a method for measuring ultrasmall forces
between a probe tip and an engineering (electrically
conducting or insulating) surface, and has been used Engineering interface Scanning probe microscope tip
on a surface
for morphological and surface roughness measurements
Simulation of a single-asperity
of surfaces on the nanoscale, as well as for adhe- contact
sion measurements. Subsequent modifications of the
AFM led to the development of the FFM, designed Fig. 28.1 Schematics of an engineering interface and scanning
for atomic- and microscale studies of friction. This probe microscope tip in contact with an engineering interface
instrument measures forces in the scanning direction.
The AFM is also being used for various investiga- tween surfaces and the way in which these are modified
tions including scratching, wear, indentation, detection by the presence of a thin liquid or a polymer film. The
of transfer of material, boundary lubrication, and fab- frictional properties of such systems have been studied
rication and machining [28.14, 21–33]. Meanwhile, by moving the surfaces laterally, and such experiments
significant progress in understanding the fundamen- have provided insights into the molecular-scale opera-
tal nature of bonding and interactions in materials, tion of lubricants such as thin liquid or polymer films.
combined with advances in computer-based model- Complementary to these studies are those in which the
ing and simulation methods, has allowed theoretical AFM tip is used to simulate a single-asperity contact
studies of complex interfacial phenomena with high res- with a solid or lubricated surface (Fig. 28.1). These
olution in space and time. Such simulations provide experiments have demonstrated that the relationship
insights into atomic-scale energetics, structure, dynam- between friction and surface roughness is not always
ics, thermodynamics, transport, and rheological aspects simple or obvious. AFM studies have also revealed
of tribological processes. much about the nanoscale nature of intimate contact
The nature of interactions between two surfaces during wear, indentation, and lubrication.
brought close together, and those between two surfaces In this chapter, we present a review of sig-
in contact as they are separated, have been studied ex- nificant aspects of nanotribological, nanomechanical,
perimentally with the surface force apparatus. This has and materials characterization studies conducted using
led to a basic understanding of the normal forces be- AFM/FFM.
microscope (LFM), designed for atomic-scale and mi- 41,60–63], and fabrication/machining [28.5,14,41]. An
Part D 28.1
croscale studies of friction [28.4–6, 8, 9, 14, 36–50] and oscillating cantilever is used for localized surface elas-
lubrication [28.20, 51–55]. This instrument measures ticity and viscoelastic mapping, referred to as dynamic
lateral or friction forces (in the plane of sample sur- AFM [28.48, 64–72]. In situ surface characterization of
face and in the scanning direction). By using a standard local deformation of materials and thin coatings has
or a sharp diamond tip mounted on a stiff cantilever been carried out by imaging the sample surfaces us-
beam, AFM is used in investigations of scratching and ing an AFM during tensile deformation using a tensile
wear [28.7,10,14,41,56–59], indentation [28.10,14,17, stage [28.73–75].
AFM signal Diode laser Surface height imaging down to atomic resolution of
(A+B) – (C+D) Mirror and lens electrically conducting surfaces is carried out using
an STM. An AFM is also used for surface height
B imaging and roughness characterization down to the
A Cantilever
D and substrate nanoscale. Commercial AFM/FFMs are routinely used
FFM C
signal
for simultaneous measurements of surface roughness
Split-diode and friction force [28.5, 13]. These instruments are
(A +C) – (B+D) photo-
detector Sample available for measurement of both small and large sam-
ples. In a small-sample AFM (Fig. 28.2a), the sample,
z x–y–z generally no larger than 10mm × 10 mm, is mounted on
PZT tube a piezoelectric crystal in the form of a cylindrical tube
y
x scanner (referred to as a PZT tube scanner) which consists of
separate electrodes to scan the sample precisely in the
b) x–y plane in a raster pattern and to move the sam-
Split-diode ple in the vertical (z) direction. A sharp tip at the free
photo- end of a flexible cantilever is brought into contact with
detector Laser diode,
collimator,
the sample. Normal and frictional forces being applied
and lens at the tip–sample interface are measured using a laser
Adjust-
able beam deflection technique. A laser beam from a diode
mirror Laser path laser is directed by a prism onto the back of a cantilever
Fixed near its free end, tilted downward at ≈ 10◦ with re-
Mirror spect to the horizontal plane. The beam reflected from
the vertex of the cantilever is directed through a mir-
ror onto a quad photodetector (a split photodetector
with four quadrants). The differential signal from the
Lens top and bottom photodiodes provides the AFM signal,
which is a sensitive measure of the cantilever vertical
Mirror Lens deflection. Topographic features of the sample cause
y x the tip to deflect in the vertical direction as the sam-
Camera x–y–z
objective z PZT tube ple is scanned under the tip. This tip deflection will
lens scanner change the direction of the reflected laser beam, chang-
ing the intensity difference between the top and bottom
sets of photodetectors (AFM signal). In the AFM op-
Cantilever holder erating mode called the height mode, for topographic
Sample x
Motorized y imaging or for any other operation in which the ap-
stage
plied normal force is to be kept constant, a feedback
Fig. 28.2a,b Schematics (a) of a commercial small-sample atomic circuit is used to modulate the voltage applied to the
force microscope/friction force microscope (AFM/FFM), and (b) of PZT scanner to adjust the height of the PZT, so that
a large-sample AFM/FFM the cantilever vertical deflection (given by the intensity
Nanotribology, Nanomechanics, and Materials Characterization 28.1 Description of AFM/FFM and Various Measurement Techniques 793
Part D 28.1
Tapping-mode phase imaging
remain constant during scanning. The PZT height vari-
ation is thus a direct measure of the surface roughness
of the sample.
Extender Controller Phase
In a large-sample AFM, both force sensors using op- electronics data
tical deflection method and scanning unit are mounted
on the microscope head (Fig. 28.2b). Because of vibra-
tions added by cantilever movement, lateral resolution Laser Computer
Photo- Substrate
of this design can be somewhat poorer than the design in detector Canti- holder
Fig. 28.2 in which the sample is scanned instead of can- lever Height
tilever beam. The advantage of the large-sample AFM piezo data
is that large samples can be measured readily.
Most AFMs can be used for surface roughness mea- Cantilever
surements in the so-called tapping mode (intermittent Material 1
substrate
contact mode), also referred to as dynamic (atomic)
force microscopy. In the tapping mode, during scanning Sample Material 2
over the surface, the cantilever–tip assembly with a nor-
z control x–y–z piezo x–y control
mal stiffness of 20–100 N/m (Digital Instrument (DI)
tapping mode etched Si probe or TESP) is sinusoidally
vibrated at its resonant frequency (350–400 kHz) by Cantilever Cantilever
a piezo mounted above it, and the oscillating tip slightly response in free air
taps the surface. The piezo is adjusted using feed-
back control in the z-direction to maintain a constant
(20–100 nm) oscillating amplitude (setpoint) and con-
stant average normal force (Fig. 28.3 [28.5, 13]). The phase angle
feedback signal to the z-direction sample piezo (to keep Viscoelastic material Nearly elastic material
the setpoint constant) is a measure of surface roughness.
The cantilever–tip assembly is vibrated at some ampli- AFM setting
tude, here referred to as the free amplitude, before the definitions Before
tip engages the sample. The tip engages the sample at engagement
some setpoint, which may be thought of as the ampli-
2 × free
tude of the cantilever as influenced by contact with the amplitude
sample. The setpoint is defined as a ratio of the vibration
amplitude after engagement to the vibration amplitude
in free air before engagement. A lower setpoint gives
a reduced amplitude and closer mean tip–sample dis-
tance. The amplitude should be kept large enough that During
the tip does not get stuck to the sample because of adhe- engagement
sive attractions. Also the oscillating amplitude applies
less average (normal) load as compared with the contact 2 × set-point
mode and reduces sample damage. The tapping mode is
used in topography measurements to minimize effects
of friction and other lateral forces and to measure the
topography of soft surfaces. Fig. 28.3 Schematic of tapping mode used to obtain height
For measurement of friction force at the tip surface and phase data and definitions of free amplitude and set-
during sliding, left-hand and right-hand sets of quad- point. During scanning, the cantilever is vibrated at its
rants of the photodetector are used. In the so-called resonant frequency and the sample x–y–z piezo is ad-
friction mode, the sample is scanned back and forth justed by feedback control in the z-direction to maintain
in a direction orthogonal to the long axis of the can- a constant setpoint. The computer records height (which is
tilever beam. A friction force between the sample and a measure of surface roughness) and phase angle (which is
the tip will produce a twisting of the cantilever. As a function of the viscoelastic properties of the sample) data
794 Part D Bio-/Nanotribology and Bio-/Nanomechanics
a) b) graphs of a square-pyramidal
plasma-enhanced chemical
vapor deposition (PECVD)
Si3 N4 tip with a trian-
gular cantilever beam,
200 nm a square-pyramidal etched
single-crystal silicon tip
with a rectangular sili-
c)
2 µm con cantilever beam, and
a three-sided pyramidal
natural diamond tip with
a square stainless-steel
cantilever beam. (b) SEM
micrograph of a multi-
Si3N4 – 0.05 µm walled carbon nanotube
(MWNT) physically attached
on a single-crystal silicon
square-pyramidal tip, and
(c) optical micrographs of
5 µm a commercial Si3 N4 tip and
two modified tips showing
SiO2 spheres mounted over
the sharp tip, at the end of the
SiO2 –3.8 µm triangular Si3 N4 cantilever
beams (radii of the tips are
given in the figure)
25 µm
10 µm
SiO2 –14.5 µm
a result, the laser beam will be reflected out of the alignment effect is eliminated. This method provides
plane defined by the incident beam and the beam re- three-dimensional maps of friction force. By following
flected vertically from an untwisted cantilever. This the friction force calibration procedures developed by
produces an intensity difference of the laser beam re- Ruan and Bhushan [28.38], voltages corresponding to
ceived in the left-hand and right-hand sets of quadrants friction forces can be converted to force units [28.76].
of the photodetector. The intensity difference between The coefficient of friction is obtained from the slope
the two sets of detectors (the FFM signal) is directly of friction force data measured as a function of nor-
related to the degree of twisting and hence to the mag- mal loads typically ranging from 10 to 150 nN. This
nitude of the friction force. One problem associated approach eliminates any contributions due to the adhe-
with this method is that any misalignment between the sive forces [28.41]. For calculation of the coefficient of
laser beam and the photodetector axis would introduce friction based on a single point measurement, friction
error in the measurement. However, by following the force should be divided by the sum of applied normal
procedures developed by Ruan and Bhushan [28.38], load and intrinsic adhesive force. Furthermore it should
in which the average FFM signal for the sample be pointed out that, for a single-asperity contact, the co-
scanned in two opposite directions is subtracted from efficient of friction is not independent of load (see the
the friction profiles of each of the two scans, the mis- discussion later).
Nanotribology, Nanomechanics, and Materials Characterization 28.1 Description of AFM/FFM and Various Measurement Techniques 795
Part D 28.1
Fast scan direction Slow scan
direction 20 nm/s, respectively.
a feedback loop is used to adjust the DC bias on the tip to the center of the tensile specimen. One end of the
Part D 28.1
to exactly cancel the electric field, and thus the vibra- sample was mounted on the slider via a force sensor
tions of the cantilever. The required bias voltage follows to monitor the tensile load. The samples were stretched
the localized potential of the surface. The surface poten- for various strains using a stepper motor and the same
tial is obtained by reversing the sign of the bias potential control area at different strains was imaged. In order to
provided by the electronics [28.80,81]. Surface and sub- better locate the control area for imaging, a set of four
surface changes of structure and/or chemistry can cause markers was created at the corners of a 30 μm × 30 μm
changes in the measured potential of a surface. Thus, square at the center of the sample by scratching the sam-
mapping of the surface potential after sliding can be ple with a sharp silicon tip. The scratching depth was
used for detecting wear precursors and studying the controlled such that it did not affect the cracking behav-
early stages of localized wear. ior of the coating. A minimum displacement of 1.6 μm
could be obtained. This corresponded to a strain incre-
28.1.5 In Situ Characterization ment of 8 × 10−3 % for a sample length of 38 mm. The
of Local Deformation Studies maximum travel was about 100 mm. The resolution of
the force sensor was 10 mN with a capacity of 45 N.
In situ characterization of local deformation of mater- During stretching, a stress–strain curve was obtained
ials can be carried out by performing tensile, bending during the experiment to study any correlation between
or compression experiments inside an AFM and by the degree of plastic strain and propensity for cracking.
observing nanoscale changes during the deformation
experiment [28.17]. In these experiments, small defor- 28.1.6 Nanoindentation Measurements
mation stages are used to deform the samples inside an
AFM. In tensile testing of the polymeric films carried For nanoindentation hardness measurements the scan
out by Bobji and Bhushan [28.73, 74] and Tambe and size is set to zero, and then a normal load is applied to
Bhushan [28.75] a tensile stage was used (Fig. 28.8). make the indents using the diamond tip (Sect. 28.1.5).
The stage with a left–right combination lead screw During this procedure, the tip is continuously pressed
(which helps to move the slider in the opposite direc- against the sample surface for about 2 s at various in-
tion) was used to stretch the sample to minimize the dentation loads. The sample surface is scanned before
movement of the scanning area, which was kept close and after the scratching, wear or indentation to obtain
Stepper motor
AFM tip
Stage
Force
sensor
Slider
Left–right
lead screw
Support
Stepper
Sample motor
controller
z
Signal A/D
y x conditioner board PC
Base plate
Fig. 28.8 Schematic of the tensile stage to conduct in situ tensile testing of polymeric films in an AFM
798 Part D Bio-/Nanotribology and Bio-/Nanomechanics
the initial and final surface topography, at a low normal of elasticity is obtained from the slope of the unloading
Part D 28.1
Part D 28.1
a) Force modulation phase imaging b) Atomic force acoustic microscopy
Height data AFM controller Laser diode
Stiff material Compliant material records amplitude (which is a function of the elastic
stiffness of the material). Contact analyses can be used
AFM setting definitions 2 × force modulation
to obtain a quantitative measure of localized elasticity of
amplitude soft surfaces [28.66]. Etched single-crystal silicon can-
tilevers with integrated tips (DI force modulation etched
Si probe or FESP) with a radius of 25–50 nm, a stiffness
of 1–5 N/m, and a natural frequency of 60–100 kHz
are commonly used for the measurements. Scanning is
Lift scan height
normally set to a rate of 0.5 Hz along the fast axis.
In the AFAM technique [28.67, 68, 83], the
cantilever–tip assembly is moved either in the normal
or lateral mode, and the contact stiffness is evaluated
by comparing the resonant frequency of the cantilever
800 Part D Bio-/Nanotribology and Bio-/Nanomechanics
in contact with the sample surface with those of the during sample engagement. As illustrated in Fig. 28.3,
Part D 28.1
free vibrations of the cantilever. Several free resonant the phase angle lag (at least partially) is a function
frequencies are measured. Based on the shift of the of the viscoelastic properties of the sample material.
measured frequencies, the contact stiffness is deter- A range of tapping amplitudes and setpoints can be
mined by solving the characteristic equation for the tip used for measurements. Commercially an etched single-
vibrating in contact with the sample surface. The elastic crystal silicon tip (DI TESP) used for tapping mode,
modulus is calculated from contact stiffness using Hertz with a radius of 5–10 nm, a stiffness of 20–100 N/m,
analysis for a spherical tip indenting a plane. Contact and a natural frequency of 350–400 kHz, is normally
stiffness is equal to 8 × contact radius × reduced shear used. Scanning is normally set to a rate of 1 Hz along
modulus in shear mode. the fast axis.
In the lateral mode using the AFAM technique, the In the TR mode, a tip is vibrated in the torsional
sample is glued onto cylindrical pieces of aluminum mode at high frequency at the resonant frequency of
which serve as ultrasonic delay lines coupled to an ul- the cantilever beam. An etched single-crystal silicon
trasonic shear wave transducer (Fig. 28.10b) [28.46, 67, cantilever with integrated tip (DI FESP) with a radius
68]. The transducer is driven with frequency sweeps of ≈ 5–10 nm, normal stiffness of 1–5 N/m, torsional
to generate in-plane lateral sample surface vibrations. stiffness of ≈ 30 times normal stiffness, and torsional
These couple to the cantilever via the tip–sample con- natural frequency of 800 kHz is normally used. A ma-
tact. To measure torsional vibrations of the cantilever jor difference between the TM and TR modes is the
at frequencies up to 3 MHz, the original electronic cir- directionality of the applied oscillation – a normal
cuit of the lateral channel of the AFM (using a low-pass (compressive) amplitude exerted for the TM and a tor-
filter with limited bandwidth to a few hundred kHz) sional amplitude for the TR mode. The TR mode is
was replaced by a high-speed scheme which bypasses expected to provide good contrast in the tribological
the low-pass filter. The high-frequency signal was fed and mechanical properties of the near-surface region
to a lock-in amplifier, digitized using a fast analog-to- as compared with the TM. Two of the reasons are as
digital (A/D) card, and fed into a broadband amplifier follows:
followed by a root-mean-square (RMS)-to-DC con-
verter, and read by a computer. Etched single-crystal 1. In the TM, the interaction is dominated by the ver-
silicon cantilevers (normal stiffness of 3.8–40 N/m) in- tical properties of the sample, so the tip spends
tegrated tips are used. a small fraction of its time in the near-field inter-
action with the sample. Furthermore, the distance
Viscoelastic Mapping between the tip and the sample changes during the
Another form of dynamic force microscopy, phase- measurements, which changes interaction time and
contrast microscopy, is used to detect the contrast in forces, and affects measured data. In the TR mode,
viscoelastic (viscous energy dissipation) properties of the distance remains nearly constant.
different materials across the surface [28.65, 69–72, 84, 2. The lateral stiffness of a cantilever is typically about
85]. In these techniques, both deflection amplitude and two orders of magnitude larger than the normal
phase angle contrasts are measured, which are mea- (flexural) stiffness. Therefore, in the TM, if the
sures of the relative stiffness and viscoelastic properties, sample is relatively rigid, much of the deformation
respectively. Two phase measurement techniques – tap- occurs in the cantilever beam, whereas in the TR
ping mode and torsional resonance (TR) mode – have mode, much of the deformation occurs in the sam-
been developed. We describe them next. ple. A few comments on the special applications of
In the tapping mode (TM) technique, as de- the TR mode are made next. Since most of the de-
scribed earlier, the cantilever–tip assembly is sinu- formation occurs in the sample, the TR mode can
soidally vibrated at its resonant frequency, and the be used to measure stiff and hard samples. Fur-
sample x–y–z piezo is adjusted using feedback con- thermore, properties of thin films can be measured
trol in the z-direction to maintain a constant setpoint more readily with the TR mode. For both the TM
(Fig. 28.3) [28.69, 70]. The feedback signal to the z- and TR modes, if the cantilever is driven to vibrate
direction sample piezo (to keep the setpoint constant) is at frequencies above resonance, it would have less
a measure of surface roughness. The extender electron- motion (high apparent stiffness), leading to higher
ics is used to measure the phase angle lag between the sample deformation and better contrast. It should
cantilever piezo drive signal and the cantilever response be further noted that the TM exerts a compressive
Nanotribology, Nanomechanics, and Materials Characterization 28.1 Description of AFM/FFM and Various Measurement Techniques 801
force, whereas the TR mode exerts a torsional force, hesion problem present in mode I and reveals true
Part D 28.1
therefore normal and shear properties are measured surface properties.
in the TM and TR modes, respectively. Song and Bhushan [28.86] presented a forced tor-
sional vibration model for a tip–cantilever assembly
In the TR mode, the torsional vibration of the can- under viscoelastic tip–sample interaction. This model
tilever beam is achieved using a specially designed provides the relationship of torsional amplitude and
cantilever holder. It is equipped with a piezo system phase shift with lateral contact stiffness and viscosity
mounted in a cantilever holder, in which two piezos vi- which can be used to extract in-plane interfacial me-
brate out of phase with respect to each other. A tuning chanical properties.
process prior to scanning is used to select the torsional Various operating modes of AFM used for sur-
vibration frequency. The piezo system excites torsional face roughness, localized surface elasticity, viscoelastic
vibration at the cantilever’s resonant frequency. The tor- mapping, and friction force measurements (to be dis-
sional vibration amplitude of the tip (TR amplitude) cussed later) are summarized in Table 28.2.
is detected by the lateral segments of the split-diode
photodetector (Fig. 28.11) [28.71]. The TR mode mea-
a) TR-mode imaging
sures surface roughness and phase angle as follows.
Feedback loop Controller
During the measurement, the cantilever–tip assembly is electronics
first vibrated at its resonance at some amplitude depen-
dent upon the excitation voltage, before the tip engages Detector Diode
electronics laser
the sample. Next, the tip engages the sample at some
setpoint. A feedback system coupled to a piezo stage
is used to keep a constant TR amplitude during scan- Piezo
ning. This is done by controlling the vertical position
of the sample using a piezo moving in the z-direction, Split-diode Cantilever
which changes the degree of tip interaction. The dis- photodetector Tip
placement of the sample z piezo gives a roughness Sample
image of the sample. A phase-angle image can be ob-
tained by measuring the phase lag of the cantilever z
vibration response in the torsional mode during engage-
ment with respect to the cantilever vibration response
in free air before engagement. The control feedback of Scanner x
the TR mode is similar to that of tapping, except that y
the torsional resonance amplitude replaces the flexural
resonance amplitude [28.71]. b) Phase angle definition
Chen and Bhushan [28.72] used a variation to the Cantilever in free air
approach just described (referred to as mode I here). Cantilever response during engagement
They performed measurements at constant normal can-
tilever deflection (constant load) (mode II) instead of
using the constant setpoint in the Kasai et al. [28.71]
approach. Their approach overcomes the meniscus ad- Phase angle
Viscoelastic material Nearly elastic material
Fig. 28.11a,b Schematic of torsional resonance mode
shown at the top. Two examples of the phase-angle re- AFM setting definition
sponse are shown in the middle. One is for materials Static Dynamic
exhibiting viscoelastic (a) and the other nearly elastic prop- Contact TM TR mode
erties (b). Three AFM settings are compared at the bottom: 2 × setpoint 2 × setpoint
≈10–100 nm ≈ 0.3–2 nm
contact, tapping mode (TM), and TR modes. The TR mode
is a dynamic approach with a laterally vibrating cantilever
tip that can interact with the surface more intensively than
other modes. Therefore, more detailed near-surface infor-
Vertical force Lateral force
mation is available
802 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Table 28.2 Summary of various operating modes of AFM for surface roughness, stiffness, phase angle, and friction
Part D 28.2
28.1.8 Boundary Lubrication Measurements 130 nN. The samples are generally scanned at a rate
of 0.5 Hz, resulting in a scanning speed of 2 μm/s. Ve-
To study nanoscale boundary lubrication properties, locity effects on friction are studied by changing the
adhesive forces are measured in the force calibration scan frequency from 0.1 to 60 Hz while the scan size
mode, as previously described. The adhesive forces is maintained at 2 × 2 μm2 , which allows velocity to
are also calculated from the horizontal intercept of vary from 0.4 to 240 μm/s. To study durability prop-
friction versus normal load curves at a zero value erties, the friction force and coefficient of friction are
of friction force. For friction measurements, the sam- monitored during scanning at normal load of 70 nN and
ples are typically scanned using a Si3 N4 tip over an scanning speed of 0.8 μm/s, for a desired number of
area of 2 × 2 μm2 at normal load ranging from 5 to cycles [28.51, 52, 54].
Part D 28.2
Surface height image (nA) 5 a) 1 0.2nm
0.3
0.75 0.1V
nm
0.5
3 0
0.5 0V
0.25
2
1 0.25
0
0
1
2
3 0
nm
0 0.25 0.5 0.75 1 nm
Friction
b) Sliding direction
Topography
Friction
2
1 nm
4
x 2 nm/div
6 z 0.5 nm/div
8 nm Bucky balls
a)
(raw data) of a 1 × 1 nm2 area of freshly cleaved HOPG,
showing atomic-scale variation of the friction force. High
points are shows by lighter color. Also shown is a line plot
of the friction force profile along the line indicated by ar-
rows. The normal load was 25 nN and the cantilever normal
stiffness was 0.4 N/m [28.39]. (b) Schematic of a model
for a tip atom sliding on an atomically flat periodic surface.
The schematic shows the tip jumping from one potential
minimum to another, resulting in stick–slip behavior
B B
also shows a line plot of the friction force profile along
some crystallographic direction. The actual shape of the
friction profile depends upon the spatial location of the
axis of tip motion. Note that a portion of the atomic-
scale lateral force is conservative. Mate et al. [28.36]
Lateral force (nN) and Ruan and Bhushan [28.39] reported that the average
1
friction force increased linearly with normal load and
was reversible with load. Friction profiles were similar
Average
during sliding of the tip in either direction.
During scanning, the tip moves discontinuously
over the sample surface and jumps with discrete steps
0 from one potential minimum (well) to the next. This
leads to a sawtooth-like pattern for the lateral motion
(force) with periodicity of the lattice constant. This mo-
tion is called stick–slip movement of the tip [28.6, 11,
29, 36, 39]. The observed friction force includes two
components – conservative and periodic, and noncon-
–1
0 0.25 0.5 0.75 1 servative and constant. If the relative motion of the
B–B nm sample and tip were simply that of two rigid collections
b) Equilibrium of atoms, the effective force would be a conservative
position before force oscillating about zero. Slow reversible elastic de-
sliding begins Stick Slip Slip event is
AFM tip– a dissipative formation would also contribute to conservative force.
cantilever process The origin of the nonconservative direction-dependent
model
force component could be phonon generation, viscous
Periodic dissipation or plastic deformation.
Sample interaction
surface potential Stick–slip on the atomic scale, discussed above, is
Atomic lattice Direction of motion the result of the energy barrier required to be overcome
constant a of sample surface for jumping over the atomic corrugations on the sam-
ple surface. It corresponds to the energy required for
Sawtooth pattern of friction force arising from atomic scale the jump of the tip from a stable equilibrium position
stick–slip on the surface into a neighboring position. The perfect
Friction force atomic regularity of the surface guarantees the periodic-
ity of the lateral force signal, independent of the actual
atomic structure of the tip apex. A few atoms (based on
a
the magnitude of the friction force, < 10) on a tip slid-
ing over an array of atoms on the sample are expected
Distance to go through the stick–slip. For simplicity, Fig. 28.14b
shows a simplified model for one atom on a tip with
after two-dimensional (2-D) spectrum filtering with a one-dimensional spring–mass system. As the sample
high-frequency noise truncated [28.39]. Figure 28.14a surface slides against the AFM tip, the tip remains stuck
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 805
initially until it can overcome the energy (potential) maps of Al2 O3 -TiC (70–30 wt%). TiC grains have
Part D 28.2
barrier, which is illustrated by a sinusoidal interaction a Knoop hardness of ≈ 2800 kg/mm2 and Al2 O3 has
potential as experienced by the tip. After some motion, 2100 kg/mm2 , therefore TiC grains do not polish as
there is enough energy stored in the spring, which leads much and therefore have a slightly higher elevation
to slip into the neighboring stable equilibrium position. (≈ 2–3 nm higher than that of Al2 O3 grains). TiC
During the slip and before attaining stable equilibrium, grains exhibit higher friction force than Al2 O3 grains.
stored energy is converted into vibrational energy of The coefficients of friction of TiC and Al2 O3 grains
the surface atoms in the range of 1013 Hz (phonon gen- are 0.034 and 0.026, respectively, and the coefficient
eration) and decays within the range of 10−11 s into of friction of the Al2 O3 -TiC composite is 0.03. Lo-
heat. (A wave of atoms vibrating in concert are termed cal variation in friction force also arises from the
a phonon.) The stick–slip phenomenon, resulting from scratches present on the Al2 O3 -TiC surface. Meyer
irreversible atomic jumps, can be modeled theoreti- et al. [28.90] also used FFM to measure structural vari-
cally with classical mechanical models [28.88, 89]. The ations of organic mono- and multilayer films. All of
Tomanek–Zhong–Thomas model [28.89] is the starting these measurements suggest that the FFM can be used
point for determining friction force during atomic-scale for structural mapping of the surfaces. FFM measure-
stick–slip. The AFM model describes the total poten- ments can also be used to map chemical variations,
tial as the sum of the potential acting on the tip due as indicated by the use of the FFM with a modified
to interaction with the sample and the elastic energy probe tip to map the spatial arrangement of chem-
stored in the cantilever. Thermally activated stick–slip
behavior can explain the velocity effects on friction, to
be presented later.
Finally, based on Fig. 28.13a, the atomic-scale fric- a) Surface height
tion force of HOPG exhibited the same periodicity as Height
that of the corresponding topography, but the peaks (nm)
in friction and those in topography are displaced rel- 2
ative to each other (Fig. 28.13b). A Fourier expansion 1 µm
of the interatomic potential was used by Ruan and 0.8
0
Bhushan [28.39] to calculate the conservative inter- 0.6
atomic forces between atoms of the FFM tip and those
0.4
of the graphite surface. Maxima in the interatomic –2
forces in the normal and lateral directions do not oc- 0 0.2
0.2 0.4
cur at the same location, which explains the observed 0.6 0
0.8 1 µm
shift between the peaks in the lateral force and those in
the corresponding topography. b) Friction force
Friction
28.2.2 Microscale Friction force
(nN)
Local variations in the microscale friction of cleaved 4
graphite are observed (Fig. 28.15). Microscale friction 1 µm
is defined as the friction measured with a scan size 0
0.8
equal to or larger than 1 μm × 1 μm. These arise from 0.6
structural changes that occur during the cleaving pro- 0.4
cess [28.40]. The cleaved HOPG surface is largely –4
0 0.2
atomically smooth but exhibits line-shaped regions in 0.2 0.4 0
which the coefficient of friction is more than an order 0.6 0.8 1 µm
of magnitude larger. Transmission electron microscopy
indicates that the line-shaped regions consist of graphite Fig. 28.15 (a) Surface roughness and (b) friction force
planes of different orientation, as well as of amor- maps at normal load of 42 nN for a freshly cleaved HOPG
phous carbon. Differences in friction have also been surface against an Si3 N4 FFM tip. Friction in the line-
observed for multiphase ceramic materials [28.57]. Fig- shaped region is over an order of magnitude larger than
ure 28.16 shows surface roughness and friction force in the smooth areas (after [28.39])
806 Part D Bio-/Nanotribology and Bio-/Nanomechanics
0 Surface
2.5
height
nm
40 nm
500
400
0 20
0 2.5 5 µm 300
Friction force 200
5 50 nN –0 100
0 100 200 300 0
400 500 nm
25 nN Surface
slope
0 nN nm
2.5
0.75 nm
500
400
0
300
200
– 0.75 100
0 0 100 200
0 2.5 5 µm 300 0
400 500 nm
Fig. 28.16 Gray-scale surface roughness (σ = 0.80 nm) Friction
and friction force maps (mean = 7.0 nN, σ = 0.90 nN) for force
Al2 O3 -TiC (70–30 wt%) at normal load of 138 nN (af-
ter [28.57]) nN
15 nm
500
ical functional groups in mixed organic monolayer 10
400
films [28.91]. Here, sample regions that had stronger 5
interactions with the functionalized probe tip exhibited 300
0 200
larger friction.
Local variations in the microscale friction of nom- –5 100
0 100
inally rough surfaces of homogeneous material can 200 300 0
be significant, and are seen to depend on the local 400 500 nm
surface slope rather than the surface height distribu-
tion (Fig. 28.17). This dependence was first reported Fig. 28.17 Surface roughness map (σ = 4.4 nm), surface
by Bhushan and Ruan [28.37], Bhushan et al. [28.41], slope map taken in the sample sliding direction (the
and Bhushan [28.78] and later discussed in more de- horizontal axis; mean = 0.023, σ = 0.197), and friction
tail by Koinkar and Bhushan [28.92] and Sundararajan force map (mean = 6.2 nN, σ = 2.1 nN) for a lubricated
and Bhushan [28.93]. In order to elegantly show any thin-film magnetic rigid disk for normal load of 160 nN
correlation between local values of friction and sur- (after [28.41])
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 807
Part D 28.2
a strong correlation between the surface slopes and fric- a)
tion forces. For example, in Fig. 28.18, the friction force Surface height
is high locally at the edge of the grids and pits with
nm
a positive slope and is low at the edges with negative 200 µm
slope. 5
We now examine the mechanism of microscale fric-
tion, which may explain the resemblance between the 100
slope of surface roughness maps and the correspond- 2.5
ing friction force maps [28.5, 6, 13, 39–41, 49, 92, 93].
There are three dominant mechanisms of friction: adhe- 0
0
sive, ratchet, and plowing [28.11, 17]. To first order, we 2.5 0
may assume these to be additive. The adhesive mecha- 5 µm
nism cannot explain the local variation in friction. Next
Surface slope
we consider the ratchet mechanism. We consider a small
tip sliding over an asperity making an angle θ with the
horizontal plane (Fig. 28.19). The normal (to the gen-
eral surface) force W applied by the tip to the sample 6 µm
surface is constant. The friction force F on the sample 5
would be a constant for a smooth surface if the fric- 3
tion mechanism does not change. For a rough surface
shown in Fig. 28.19, if the adhesive mechanism does not 2.5
change during sliding, the local value of the coefficient 0
of friction remains constant, 0
2.5 0
μ0 = S/N , (28.1) 5 µm
where S is the local friction force and N is the local Friction force
normal force. However, the friction and normal forces
are measured with respect to global horizontal and nor-
mal axes, respectively. The measured local coefficient V µm
5
1
Fig. 28.18a,b Surface roughness map, surface slope map
taken in the sample sliding direction (the horizontal axis), 0.5
and friction force map for (a) a gold-coated ruler (with 2.5
somewhat rectangular grids with a pitch of 1 μm and a rul-
0
ing step height of about 70 nm) at normal load of 25 nN, 0
and (b) a silicon grid (with 5 μm square pits of depth 2.5 0
180 nm and pitch 10 μm) (after [28.93]) 5 µm
b)
10
Surface height (nm) Surface slope Friction force (V)
250 2 0.35
High
friction
5
0 0 0
Low
friction
0 –250 –2 –0.35
0 5 10 0 10 0 10 0 10
Scan distance (µm) Scan distance (µm) Scan distance (µm)
808 Part D Bio-/Nanotribology and Bio-/Nanomechanics
W
N believed to be the dominant mechanism for the local
AFM tip
variations in the friction force map. With the tip slid-
ing over the leading (ascending) edge of an asperity,
the surface slope is positive; it is negative during sliding
F over the trailing (descending) edge of an asperity. Thus,
measured friction is high at the leading edge of asperi-
S
ties and low at the trailing edge. In addition to the slope
Sample effect, the collision of the tip when encountering an as-
θ
perity with a positive slope produces additional torsion
Sample sliding
direction of the cantilever beam leading to higher measured fric-
tion force. When encountering an asperity with the same
Fig. 28.19 Schematic illustration showing the effect of an negative slope, however, there is no collision effect and
asperity (making an angle θ with the horizontal plane) on hence no effect on torsion. This effect also contributes
the surface in contact with the tip on local friction in the to the difference in friction forces when the tip scans up
presence of the adhesive friction mechanism. W and F are and down on the same topography feature. The ratchet
the normal and friction forces, respectively, and S and N mechanism and the collision effects thus semiquanti-
are the force components along and perpendicular to the tatively explain the correlation between the slopes of
local surface of the sample at the contact point, respectively the roughness maps and friction force maps observed in
Figs. 28.17 and 28.18. We note that, in the ratchet mech-
of friction μ1 in the ascending part is anism, the FFM tip is assumed to be small compared
F (μ0 + tan θ) with the size of asperities. This is valid since the typical
μ1 = = ∝ μ0 + tan θ , radius of curvature of the tips is ≈ 10–50 nm. The radii
W (1 − μ0 tan θ) of curvature of the asperities of the samples measured
for small μ0 tan θ , (28.2) here (the asperities that produce most of the friction
indicating that in the ascending part of the asperity one variation) are found to be typically ≈ 100–200 nm,
may simply add the friction force and the asperity slope which is larger than that of the FFM tip [28.94]. It is im-
to one another. Similarly, on the right-hand side (de- portant to note that the measured local values of friction
scending part) of the asperity, and normal forces are measured with respect to global
(and not local) horizontal and vertical axes, which are
(μ0 − tan θ) believed to be relevant in applications.
μ2 = ∝ μ0 − tan θ ,
(1 + μ0 tan θ)
for small μ0 tan θ . (28.3) 28.2.3 Directionality Effect on Microfriction
For a symmetrical asperity, the average coefficient of During friction measurements, the friction force data
friction experienced by the FFM tip traveling across the from both the forward (trace) and backward (retrace)
whole asperity is scans are useful in understanding the origins of the ob-
(μ1 + μ2 ) served friction forces. Magnitudes of material-induced
μave = effects are independent of the scanning direction
2
whereas topography-induced effects are different be-
(1 + tan2 θ)
= μ0 ∝ μ0 (1 + tan2 θ) , tween forward and backward scanning directions. Since
1 − μ20 tan2 θ the sign of the friction force changes as the scanning
for small μ0 tan θ . (28.4) direction is reversed (because of the reversal of torque
applied to the end of the tip), addition of the friction
Finally, we consider the plowing component of friction force data of the forward and backward scan eliminates
with the tip sliding in either direction, which is [28.11, the material-induced effects while topography-induced
17] effects remain. Subtraction of the data between forward
μp ∝ tan θ . (28.5) and backward scans does not eliminate either effect
(Fig. 28.20) [28.93].
Because in FFM measurements we notice little damage Owing to the reversal of the sign of the retrace (R)
of the sample surface, the contribution from plowing friction force with respect to the trace (T) data, the
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 809
friction force variations due to topography are in the slopes are virtually identical, therefore the tip shape
Part D 28.2
same direction (peaks in the trace correspond to peaks asymmetry should not have much effect.
in the retrace). However, the magnitudes of the peaks Figure 28.21 shows surface height and friction force
in the trace and retrace at a given location are differ- data for a gold ruler and a silicon grid in the trace
ent. The increase in the friction force experienced by and retrace directions. Subtraction of the two sets of
the tip when scanning up a sharp change in topogra- friction data yields a residual peak because of the dif-
phy is greater than the decrease in the friction force ferences in the magnitudes of the friction forces in the
experienced when scanning down the same topography two directions. This effect is observed at all locations of
change, partly because of the collision effects discussed significant changes in topography.
earlier. Asperities on engineering surfaces are asymmet- In order to facilitate comparison of the effect of
rical, which also affects the magnitude of the friction directionality on friction, it is important to take into ac-
force in the two directions. Asymmetry in the tip shape count the change of sign of the surface slope and friction
may also have an effect on the directionality of fric- force in the trace and retrace directions. Figure 28.22
tion. We will note later that the magnitude of the surface shows surface height, surface slope, and friction force
data for two samples in the trace and retrace directions.
The correlations between the surface slope and friction
T B forces are clear. The third column in the figure shows
A
the retrace slope and friction data with an inverted
sign (−retrace). Now we can compare trace data with
µ1 µ2 > µ1
−retrace data. It is clear that the friction experienced
by the tip is dependent upon the scanning direction
Surface slope because of the surface topography. In addition to the ef-
fect of topographical changes discussed earlier, during
T
surface-finishing processes, material can be transferred
preferentially onto one side of the asperities, which
Friction force also causes asymmetry and direction dependence. Re-
duction of local variations and in the directionality
T of friction properties requires careful optimization of
surface roughness distributions and surface-finishing
R processes.
The directionality as a result of the effect of sur-
Surface slope face asperities will also be manifested in macroscopic
friction data; i. e., the coefficient of friction may be
R different in one sliding direction than the other. The
asymmetrical shape of the asperities accentuates this
R effect. Frictional directionality can also exist in mater-
ials with particles having a preferred orientation. The
directionality effect in friction on a macroscale is ob-
served in some magnetic tapes. In a macroscale test,
a 12.7 mm-wide polymeric magnetic tape was wrapped
Friction force over an aluminum drum and slid in a reciprocating mo-
tion with a normal load of 0.5 N and a sliding speed of
≈ 60 mm/s [28.4]. The coefficient of friction as a func-
T–R
tion of sliding distance in either direction is shown
in Fig. 28.23. We note that the coefficient of friction
Fig. 28.20 Schematic of friction forces expected when on a macroscale for this tape is different in different
a tip traverses a sample composed of different materials directions. Directionality in friction is sometimes ob-
with sharp changes in topography. A schematic of the sur- served on the macroscale; on the microscale this is the
face slope is also shown norm [28.5,15]. On the macroscale, the effect of surface
810 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Part D 28.2
0
0 nm
Surface
height –250
– 50 0 10
Friction force (V)
0.35 0.35
High
A friction
0V B 0
Friction Low
force (T) friction T
–0.35 –0.35
0 10
Friction force (V)
0.35 0.35
Low
friction
0V 0
Friction R
force (R) High
friction
–0.35 –0.35
0 10
Friction force (V)
0.35
0.35
0V 0
Friction T–R
force (T–R)
0 1.5 –0.35 –0.35
µm 0 0.75 1.5 0 10
Scan distance (µm) Scan distance (µm)
Fig. 28.21 (a) Gray-scale images and two-dimensional profiles of surface height and friction forces across a single ruling
of the gold-coated ruler, and (b) two-dimensional profiles of surface height and friction forces across a silicon grid pit.
Friction force data in trace and retrace directions, and subtracted force data are presented
asperities is usually averaged out over a large number of cantilever twist is dependent on the sliding direction be-
contacting asperities. cause of the local surface slope. Hence it is difficult
to separate friction-induced from roughness-induced
28.2.4 Surface-Roughness-Independent cantilever twist in the image. To obtain roughness-
Microscale Friction independent friction, lateral or torsional modulation
techniques are used, in which the tip is oscillated in-
As just reported, the friction contrast in conven- plane with a small amplitude at a constant normal
tional friction measurements is based on interactions load, and change in the shape and magnitude of the
dependent upon interfacial material properties super- cantilever resonance is used as a measure of the fric-
imposed by roughness-induced lateral forces, and the tion force [28.44–49, 95]. These techniques also allow
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 811
Part D 28.2
a) Trace Retrace
Surface
height
– Retrace
Surface
slope
Friction
force
0 5 µm 0 5 µm 0 5 µm
–250
Surface slope – Retrace
–2
Friction force (V)
0.35
High Low
friction friction
0
Low High
friction friction
–0.35
0 10 0 10 0 10
Scan distance (µm) Scan distance (µm) Scan distance (µm)
Fig. 28.22 (a) Gray-scale images of surface heights, surface slopes, and friction forces for scans across a gold-coated
ruling, and (b) two-dimensional profiles of surface heights, surface slopes, and friction forces for scans across a silicon
grid pit. Arrows indicate the tip sliding direction (after [28.93])
812 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Coefficient of friction
lubricant film. Figure 28.25a shows the amplitude of
0.4 Forward the cantilever torsional vibration as a function of fre-
Backward
0.3 quency on a bare silicon sample. The frequency sweep
0.2
was adjusted such that a contact resonant frequency was
covered. The different curves correspond to different
0.1 excitation voltages applied to the shear wave transducer.
0 At low amplitudes, the shape of the resonance curve
0 25 50 75 100 125 is Lorentzian. Above a critical excitation amplitude of
Number of drum passes
the transducer (excitation voltage = 4 V, corresponding
Fig. 28.23 Coefficient of macroscale friction as a function to ≈ 0.2 nm lateral surface amplitude as measured by
of drum passes for a polymeric magnetic tape sliding over interferometry), the resonance curve flattens out, and
an aluminum drum in a reciprocating mode in both di- the frequency range of the flattened part increases fur-
rections. Normal load = 0.5 N over 12.7 mm-wide tape, ther with the excitation amplitude. Here, the static force
sliding speed = 60 mm/s (after [28.78]) applied was 47 nN and the adhesion force was 15 nN.
The resonance behavior of the tip–cantilever system in
measurements over a very small region (a few nm to
a few μm).
Scherer et al. [28.45] and Reinstädtler et al. [28.46, TR amplitude versus frequency
47] used the lateral mode for friction measurements TR amplitude (V)
(Fig. 28.10b) whereas Bhushan and Kasai [28.49] used Material with two stiffness regions
the TR mode for these measurements (Fig. 28.11). Be- with no slip
fore engagement, the cantilever is driven into torsional
motion of the cantilever–tip assembly with a given nor-
mal vibration amplitude (the vibration amplitude in
free air). After engagement, the vibration amplitude de-
creases due to the interaction between the tip and the
sample, the vibration frequency increases, and phase
shift occurs. During scanning, the normal load is kept
constant, and the vibration amplitude of the cantilever
is measured at the contact frequency.
As mentioned earlier, the shift in contact reso-
TR amplitude (V)
nant frequency in both the lateral and TR modes is
a measure of contact stiffness, as shown schematically Material with uniform stiffness
in Fig. 28.24. At an excitation voltage above a cer- with slip (plateau)
tain value, as a result of microslip at the interface,
a flattening of the resonant frequency spectra occurs
(Fig. 28.22). At low excitation voltage, the AFM tip
sticks to the sample surface and follows the motion Excitation
like an elastic contact with viscous damping, in which voltage
case the resonance curve is Lorentzian with a well-
defined maximum. The excitation voltage should be
high enough to initiate microslip. The maximum tor-
sional amplitude at a given resonance frequency is
Frequency (kHz)
a function of the friction force and sample stiffness,
so the technique is not valid for inhomogeneous sam- Fig. 28.24 Schematic showing frequency profiles of the
ples. If the torsional stiffness of the cantilever is very TR amplitude for materials with two phases and a single
high compared with the sample stiffness, the technique phase. The maximum TR amplitude at the contact reso-
should work. nant frequency of the resonance curve with a flattened top,
Reinstädtler et al. [28.46] performed lateral-mode resulting from slip, can be used for friction force measure-
experiments on bare Si and Si lubricated with 5 nm- ment
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 813
Part D 28.2
a) Torsional amplitude θ (arb. units) a) TR-mode surface height Contact-mode surface height
nm TR amplitude = 0.3 nm nm
60 60
6 Si Excitation
voltage (V) –60 –60
5 0.5 10 10
200 nm 200 nm
2
4
4
6
3 5 5
8
2 10
0 0
1 0
0 5 10 µm
0
0 5 10 µm
TR-mode friction force Contact-mode friction force
0 Normal load = 100 nN Normal load = 50 nN
230 240 250 (V) Trace (V) Trace
1.05 0.5
Frequency (kHz) 0.25
0.8 0
10 10
b) Effect of load c) Effect of lubricant film
θ (arb. units) θ (arb. units) Low High
23 nN Si
Si Si 5
2 70 nN 5
+ 5 nm
95 nN Z-DOL
1 High Low
1 0 0
0 5 10 µm 0 5 10 µm
(V) Retrace (V) Retrace
1.05 0
–0.25
0 0
210 230 250 210 230 250 0.8
10
–0.5
10
Frequency (kHz) Frequency (kHz)
Low Low
contact with the lubricated silicon sample (Fig. 28.25b) High Low
was similar to that with the bare silicon sample. By in-
creasing the static load, the critical amplitude for the 0
0 5 10 µm
0
0 5 10 µm
appearance of the flattening increases. Deviations from (V)
1.05
Trace–retrace (V)
0.25
Trace–retrace
loads < 95 nN. As shown in Fig. 28.25c, the resonance a silicon ruler obtained using the TR-mode and contact-
Part D 28.2
curve obtained at the same normal load of 70 nN and mode techniques. A comparison is made between the
the same excitation voltage (7 V) is more flattened on TR-mode and contact-mode friction force maps. For
the lubricated sample than on the bare silicon, which led easy comparison, the line scan profiles near the cen-
us to conclude that the critical amplitude is lower on the tral area are shown on top of the gray scale maps. The
lubricated sample than on the bare sample. These exper- vertical scales of the friction force profiles in the two
iments clearly demonstrate that torsional vibration of an graphs are selected to cover the same range of fric-
AFM cantilever at ultrasonic frequencies leads to stick– tion force so that direct comparison can be made, i. e.,
slip phenomena and sliding friction. Above a critical 0.25 V at full scale for the TR mode corresponds to
vibration amplitude, sliding friction sets in. 0.5 V for the contact mode in these measurements. As
Bhushan and Kasai [28.49] performed friction mea- expected, for the trace scan, small downward peaks in
surements on a silicon ruler and demonstrated that the TR-mode map and large upward and downward
friction data in TR mode is essentially independent of peaks in the contact-mode map are observed. The po-
surface roughness and sliding direction. Figure 28.26a sitions of these peaks coincide with those of the surface
shows surface height and friction force maps on slope; therefore, the peaks in the friction signals are
0 0
0
0 1.5 µm 3 –60 –60
TR amplitude (V) Friction force (V)
1.05 0.25 Trace
0
Trace
Retrace Retrace
0.8 –0.25
TR amplitude (V) Friction force (V)
0.13 0.5
Trace–retrace Trace–retrace
0 0.25
–0.13 0
TR amplitude (V) Friction force (V)
1.05 0.25
Trace–retrace Trace–retrace
0.8 –0.25
0 3 0 3
Scan distance (µm) Scan distance (µm)
attributed to a topography-induced effect. For the re- oriented such that the scanning axis is perpendicular
Part D 28.2
trace scan, the peak pattern for the TR mode stays to the long axis of the AFM cantilever (which corre-
similar, but for the contact mode the pattern becomes sponds to the 90◦ scan angle mode of the commercial
reversed. AFM). The displacement is monitored using an inte-
The subtraction image for the TR mode shows
almost flat contrast, since the trace and retrace fric-
tion data profiles are almost identical. For the contact a) Linear optical Position Slider Glued friction
encoder photosensor bar
mode, the subtraction image shows that the topography-
induced contribution still exists. As stated earlier, the
addition image of the TR mode and the addition image
of the contact mode enhance the topography-induced
effect, as observed in the figure.
A closer look at the silicon ruler images at one
pitch was taken, and the associated images are shown Piezo
in Fig. 28.26b. The surface-height profiles in the TR actuator
mode and contact mode are somewhat different. The TR Mounting
mode shows sharper edges than those in contact mode. platform
The ratios of the change in amplitude at the steps to the Excitation
Friction tip electrodes
change in the mean amplitude in the TR mode and in
the contact mode are a measure of topography effects. Ultrahigh-velocity stage (up to 200 mm/s)
The ratio in the contact mode (≈ 85%) is about seven
times larger than that in the TR mode (≈ 12%). b) Flexure
design
28.2.5 Velocity Dependence Integrated Piezo crystal
of Micro/Nanoscale Friction capacitive
sensor (target Vi/p
AFM/FFM experiments can generally be conducted at and probe)
relative velocities as high as ≈ 100–250 μm/s. To sim-
ulate applications, it is of interest to conduct friction
experiments at higher velocities (up to 1 m/s). Fur-
Stage motion
thermore, high-velocity experiments would be useful to
High-velocity stage (up to 10 mm/s)
study the velocity dependence of friction and wear. One
approach has been to mount samples on a shear wave
c) Vertical deflection
transducer (an ultrasonic transducer) and then drive it at feedback
very high frequencies (in the MHz range), as reported
earlier (Fig. 28.10) [28.44–48, 95, 97]. The coefficient
of friction on the nanoscale is estimated based on the Optical detection
contact resonant frequency and requires the solution of system
Laser
the characteristic equations for the tip vibrating in con-
tact with the sample surface. The approach is complex AFM controller
and depends upon various assumptions.
An alternative approach is to utilize piezo stages High-speed A/D AFM
tip
with large amplitude (≈ 10–100 μm) and relatively data board
low resonance frequency (a few kHz) and measure Sample
the friction force on the microscale directly using the
FFM signal without any analysis with the assump- Single-axis piezo stage
PC
tions used in the previous approaches based on shear
wave transducers. A commercial AFM setup modified Fig. 28.27a–c Schematics of (a) an ultrahigh-velocity piezo stage
with this approach can yield sliding velocities up to and (b) a high-velocity piezo stage, and (c) a block diagram of the
200 mm/s [28.50, 96]. In the high-velocity piezo stage high-speed data collection and processing system used for friction
shown in Fig. 28.27a, the single-axis piezo stage is force measurement (after [28.50, 96])
816 Part D Bio-/Nanotribology and Bio-/Nanomechanics
grated capacitive feedback sensor, located diametrically Bhushan [28.96, 102]. The friction force as a function
Part D 28.2
opposite to the piezo crystal. The capacitive change, of velocity for Si(100) and DLC (deposited by filtered
corresponding to the stage displacement, gives a mea- cathodic arc) is shown in Fig. 28.28 on a logarithm ve-
sure of the amount of displacement and can be used locity scale (middle column). The solid lines in the
as feedback to the piezo controller for better guidance figure represent the results for a scan length of 1000 μm
and tracking accuracy during scanning. The closed-loop with velocity ranging from 1000 to 2 × 105 μm/s using
position control of the piezoelectric-driven stages us- the ultrahigh-velocity stage. The dotted lines represent
ing capacitive feedback sensors provides linearity of results for a 25 μm scan length with velocity ranging
motion better than 0.01% with nanometer resolution from 5 to 500 μm/s using the high-velocity stage. To
and stable drift-free motion [28.50]. In the ultrahigh- show the friction force dependence on velocity in the
velocity piezo stage shown in Fig. 28.27a, a rectangular lower range clearly, the test results with velocity varying
monolithic piezoceramic plate (the stator) with two ex- from 5 to 500 μm/s for 25 μm are shown on a magnified
citation electrodes is resonated using a 12 V power scale in the left column of Fig. 28.28.
supply. Depending on the desired direction of the mo- On the Si(100) sample, the friction force decreased
tion, the left or right electrode is excited to produce with velocity at low velocities (v < 10 μm/s) and then
high-frequency eigenmode oscillations up to 200 kHz. increased linearly with log(v) for the 25 μm scan length.
Simultaneous eigenmodes result in quasielliptical mo- For the 1000 μm scan length, the friction force in-
tion. An alumina friction tip (pusher) attached to the creased linearly with log(v) when the velocity was
plate pushes a slider with a glued friction bar which < 2 × 104 μm/s. When the velocity was > 2 × 104 μm/s,
rests on a set of bearings. Through its contact with the the friction force increased linearly with velocity. For
friction bar, the piezoceramic plate provides microim- DLC, the friction force increased linearly with log(v)
pulses and drives the slider forward or backward. While from 5 to 500 μm/s for the 25 μm scan length. For
the longitudinal oscillation component provides the en- the 1000 μm scan length, the friction force increased
ergy as the driving force, the transverse component with velocity until about 2 × 104 μm/s, where the fric-
serves to change the pressure of the friction tip against tion force reaches a maximum, after which the friction
the friction bar. The transverse oscillation energy de- force decreased with velocity.
termines the maximum frictional force and hence the For different samples, the change in the friction
holding and driving force of the stage. An optical po- force with velocity involves different mechanisms due
sition reference photosensor is located approximately to the sample surface conditions. The silicon surface
in the middle of the range of travel and is used to ref- is hydrophilic whereas the DLC surface is nearly hy-
erence the absolute position of the stage within 1 μm drophobic. Under ambient conditions, a thin water film
repeatability. During motion, the increments of the lin- is condensed on a hydrophilic sample surface. On a hy-
ear scale from a home (reference) position point are drophobic surface, with high contact angle, it is difficult
converted to determine position using a linear optical for a water film to form on the sample surface, and the
encoder. effect of the water film on the adhesive force and friction
A block diagram of the high-speed data collec- force can be neglected.
tion and processing system used for the friction force On the silicon surface, when the velocity is
measurement is shown in Fig. 28.27b. During the ex- < 10 μm/s, the friction force decreased with velocity.
periments, the AFM cantilever is held stationary by This can be explained as follows. The water meniscus
maintaining a scan size of zero. The mounted sample is bridges develop as a function of time around the tip
scanned below the AFM tip by moving stages, and the until reaching the equilibrium condition, being the dom-
normal and torsional deflections of the tip are recorded inant contributor to the friction force [28.5,6,11,13,19].
by a photodiode detector. The raw deflection signals The motion of the tip results in continuous breaking and
from the optical detection system are directly routed reforming of the meniscus bridges. As the tip sliding
to a high-speed data-acquisition A/D board. Raw fric- velocity exceeds a critical velocity (10 μm/s), there is
tion data is acquired at a high sampling rate of up to not sufficient time for the menisci to reform, and the
80 kilosamples/s. meniscus force will not play a dominant role any more.
The velocity dependence of friction for Si(100), Between 10 and 2 × 104 μm/s, the friction increases
diamond-like carbon (DLC), self-assembled monolayer, linearly with log(v) for both 25 and 1000 μm scan
and perfluoropolyether lubricant films has been studied lengths. This logarithmic dependence can be explained
by Tambe and Bhushan [28.50, 98–101] and Tao and by atomic-scale stick–slip [28.99, 102]. At velocity
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 817
Part D 28.2
a) Friction force on log scale (low velocity) b) Friction force on log scale
Normal load = 100 nN
Friction force (nN) Friction force (nN)
10 60
Si(100) Si(100)
0 0
Friction force (nN) Friction force (nN)
10 60
DLC DLC
Phase transformation/
7.5 tip jump
40
20 Atomic-scale
2.5 stick–slip dominates
0 0 0 0
10 101 102 103 10 101 102 103 104 105 106
Velocity (µm/s) Velocity (µm/s)
c) Friction force on linear scale
Fig. 28.28a–c Friction force as a function of sliding veloc-
Friction force (nN) ity obtained with a 25 μm scan length using a high-velocity
60
Si(100) stage (dotted line) and with a 1000 μm scan length using an
ultrahigh-velocity stage (solid line). In (a) and (b), veloc-
ity is plotted on a logarithmic scale. (a) Lower range of
40
the velocity (1–500 μm/s). (c) Data at the higher range of
velocity on a linear scale (after [28.102])
For velocities < 2 × 104 μm/s, the damping part dissipation due to surface variation at relatively low ve-
Part D 28.2
(ηẋt ) in (28.6) is comparatively low, and atomic-scale locities. Thus the linear relationship between friction
stick–slip is dominant. To investigate the stick–slip, and log(v) could be maintained. When the velocity in-
Tomlinson [28.88] assumed a periodic surface with po- creases above a certain value, the tip would lose direct
tential contact with the sample surface and shear the water film.
At velocities > 2 × 104 μm/s, the asperity deformation
2πx
V (x) = V0 1 − cos , (28.7) from the high-velocity impact could be another mecha-
a
nism, as proposed by Tambe and Bhushan [28.99].
where V0 is the surface barrier potential height and a is For the DLC film, since the surface is nearly hy-
the lattice constant of the surface. Then the force F in drophobic, a uniform water film would not form on
(28.6) can be expressed as the surface. When sliding at a velocity lower than
1000 μm/s, the friction force increased linearly with
2πV0 2πx
F = V (x) = sin . (28.8) log(v), which could also be explained by atomic-scale
a a stick–slip. At velocities > 1000 μm/s, the friction force
Based on the Tomlinson model, and taking into ac- increased with velocity until the local maximum at the
count the effect of thermal activation, or the elastic velocity of 2 × 104 μm/s, then decreased with veloc-
energy stored in the cantilever during sliding, Gnecco ity. The decreasing trend in friction at higher velocities
et al. [28.104] derived the relationship between the fric- could be due to tip jump during sliding, as illustrated
tion force and velocity, which is expressed as in Fig. 28.29b. Tip jump results in the reduction of the
lateral force during sliding. Variation of friction force
Fstick–slip = F0 + c ln v , (28.9) with distance, indicative of tip jump, was observed
from the lateral force signal (not shown). When damp- tact pressures in the range of 1.8–4.4 GPa for DLC for
Part D 28.2
ing is low and velocity is high, the tip could jump normal loads of 10–150 nN [28.109]. A combination
several periodical cycles or several peaks [28.106]. At of the high contact pressures that are encountered on
a given low damping coefficient, the slip results in the nanoscale and the high frictional energy dissipation
a low transient lateral force, as discussed by Fusco and arising from the asperity impacts at the tip–sample in-
Fasolino [28.106]. Thus the average lateral force (fric- terface due to the high sliding velocities accelerates the
tion force) over the scan length is low. The tip jump phase-transition process whereby a low-shear-strength
could also cause high-velocity impact of asperities on graphite-like layer is formed at the sliding interface.
the DLC surface, resulting in the phase transformation Similar to friction mapping, one way of exploring
of DLC from sp3 to sp2 , as explained by Tambe and the broader wear patterns is to construct wear mecha-
Bhushan [28.99]. The layer of sp2 phase can act as nism maps that summarize data and models for wear,
a lubricant and reduce the interfacial friction. thereby showing mechanisms for any given set of con-
ditions [28.108, 110–112]. Wear of sliding surfaces
28.2.6 Nanoscale Friction can occur by one or more wear mechanisms, includ-
and Wear Mapping ing adhesive, abrasive, fatigue, impact, corrosive, and
fretting [28.6, 11]. Tambe and Bhushan [28.109, 112]
Contrary to classical friction laws postulated by Amon- performed AFM experiments to develop nanoscale wear
tons and Coulomb centuries ago, nanoscale friction maps. Figure 28.31 shows a nanowear map generated
force is found to be strongly dependent on the normal for a DLC sample by simultaneously varying the nor-
load and sliding velocity. Many materials, coatings, and mal load and sliding velocity over the entire scan area.
lubricants that have wide applications show reversals in The wear map was generated for a normal load range of
friction behavior corresponding to transitions between 0–1000 nN and sliding velocity range of 0–2.5 mm/s.
different friction mechanisms [28.50, 98–100, 108]. Wear debris, believed to result from the sp3 -to-sp2 DLC
Most of the analytical models developed for explain- phase transition, was seen to form only for a high value
ing nanoscale friction behavior have remained limited of the product of sliding velocity and normal load,
in their focus and have left investigators short-handed i. e., only beyond a certain threshold of friction energy
when trying to explain friction behavior spanning dissipation [28.109, 112]. Hence the wear region ex-
multiple regimes. Nanoscale friction maps provide fun- hibits a transition line, indicating that for low velocities
damental insights into friction behavior. They help to and low normal loads there is no phase transforma-
identify and classify the dominant friction mechanisms tion. For clarity, the wear mark corners are indicated
as well as to determine the critical operating parameters
that influence transitions between different mecha-
Normal load (nN)
nisms [28.99, 100]. Figure 28.30 shows a nanoscale
24
friction map for DLC with friction mapped as a func-
tion of normal load and sliding velocity [28.107]. The 3
contours represent lines of constant friction force. The 20
friction force is seen to increase with normal load as
well as velocity. The increase in friction force with ve- 16 2.5
locity is the result of atomic-scale stick–slip. This is
a result of thermal activation of the irreversible jumps
of the AFM tip that arise from overcoming the energy 12
2
barrier between adjacent atomic positions, as described
earlier. The concentric contour lines corresponding to 8 1.5
constant friction force predict a peak point, i.e., a point 1
where the friction force reaches a maxima and beyond 4
which any further increase in normal load or sliding
velocity results in a decrease in friction force. This 101 102 103
Velocity (µm/s)
characteristic behavior for DLC is the result of phase
transformation of DLC into a graphite-like phase by the Fig. 28.30 Contour map showing the dependence of fric-
sp3 -to-sp2 phase transition, as described earlier. During tion force on normal load and sliding velocity for DLC
the AFM experiments, the Si3 N4 tip gives rise to con- (after [28.107])
820 Part D Bio-/Nanotribology and Bio-/Nanomechanics
5 nm
2 µm in Wet Environments
Experimental Observations
Wear mark Relative humidity affects adhesion and friction for dry
corners and lubricated surfaces [28.18, 43, 113]. Figure 28.32
0 shows the variation of single-point adhesive force mea-
surements as a function of tip radius on a Si(100)
sample for several humidities. The adhesive force data
are also plotted as a function of relative humidity for
several tip radii. The general trend at humidities up to
the ambient is that a 50 nm-radius Si3 N4 tip exhibits
a lower adhesive force compared with other microtips of
larger radii; in the latter case, values are similar. Thus,
Increasing for the microtips there is no appreciable variation in
normal load Region of highest wear
0 – 1000 nN adhesive force with tip radius at given humidity up to
Wear mark ambient. The adhesive force increases with the relative
boundary humidity for all tips.
(dashed line)
Sources of adhesive force between a tip and a sam-
ple surface are van der Waals attraction and meniscus
formation [28.6, 11, 18]. The relative magnitudes of the
Phase forces from these two sources are dependent upon var-
trans-
formation ious factors, including the distance between the tip and
boundary the sample surface, their surface roughness, their hy-
drophobicity, and the relative humidity [28.114]. For
Increasing velocity 0 – 2.5 mm/s
most rough surfaces, the meniscus contribution domi-
nates at moderate to high humidities, due to capillary
Fig. 28.31 Nanowear map (AFM image and schematic) il- condensation of water vapor from the environment. If
lustrating the effect of sliding velocity and normal load enough liquid is present to form a meniscus bridge,
on the wear of DLC resulting from phase transformation. the meniscus force should increase with increasing tip
Curved area shows debris lining and is indicative of the radius (proportional to the tip radius for a spherical
minimum frictional energy needed for phase transforma- tip). In addition, an increase in tip radius results in in-
tion. For clarity, the wear mark corners are indicated by creased contact area, leading to higher values of the van
white dots in the AFM image and the various zones of inter- der Waals forces. However, if nanoasperities on the tip
est over the entire wear mark are schematically illustrated and sample are considered, then the number of con-
(after [28.112]) tacting and near-contacting asperities forming meniscus
bridges increases with increasing humidity, leading to
by white dots in the AFM image (top) and the two an increase in the meniscus forces. These explain the
zones of interest over the entire wear mark are illus- trends observed in Fig. 28.32. From the data, the tip
trated schematically in Fig. 28.31a (top). radius has little effect on the adhesive forces at low hu-
Nanoscale friction and wear mapping are novel midities but increases with tip radius at high humidity.
techniques for investigating friction and wear behavior The adhesive force also increases with increasing hu-
on the nanoscale over a range of operating parameters. midity for all tips. This observation suggests that the
By simultaneously varying the sliding velocity and nor- thickness of the liquid film at low humidity is insuffi-
mal load over a large range of values, nanoscale friction cient to form continuous meniscus bridges to affect the
and wear behavior can be mapped, and the transitions adhesive forces in the case of all tips.
between different wear mechanisms can be investigated. Figure 28.32 also shows the variation in the coef-
These maps help identify and demarcate critical op- ficient of friction as a function of tip radius at a given
erating parameters for different wear mechanisms and humidity, and as a function of relative humidity for
are very important tools in the process of design and a given tip radius for Si(100). It can be observed that
selection of materials/coatings. for 0% relative humidity (RH), the coefficient of fric-
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 821
Part D 28.2
Adhesive force (nN) Adhesive force (nN)
250 0% RH 250 0.05 µm (Si3N4))
4
15% 3.8
200 45% 200 6.9
65% 9.5
14.5
150 150
100 100
50 50
0 0
0 4 8 12 16 0 25 50 75 100
Tip radius (µm) Relative humidity (%)
0.1 0.1
0.05 0.05
0 0
0 4 8 12 16 0 25 50 75 100
Tip radius (µm) Relative humidity (%)
Fig. 28.32 Adhesive force and coefficient of friction as a function of tip radius at several humidities and as a function of
relative humidity at several tip radii on Si(100) (after [28.43])
tion is about the same for all the tip radii except the cus bridges and higher friction (larger shear force). The
largest one, which shows a higher value. At all other same trend is expected for microtips beyond 65% RH.
humidities, the trend consistently shows that the coef- This is attributed to the fact that, at higher humidity,
ficient of friction increases with tip radius. An increase the adsorbed water film on the surface acts as a lubri-
in friction with tip radius at low to moderate humidi- cant between the two surfaces. Thus the interface is
ties arises from increased contact area (higher van der changed at higher humidities, resulting in lower shear
Waals forces) and the higher values of the shear forces strength and hence lower friction force and coefficient
required for the larger contact area. At high humidities, of friction.
similar to the adhesive force data, an increase with tip
radius occurs because of both contact area and meniscus Adhesion and Friction Force Expressions
effects. Although AFM/FFM measurements are able to for a Single-Asperity Contact
measure the combined effect of the contribution of van We now obtain the expressions for the adhesive force
der Waals and meniscus forces towards friction force or and coefficient of friction for a single-asperity contact
adhesive force, it is difficult to measure their individual with a meniscus formed at the interface (Fig. 28.33). For
contributions separately. It can be seen that, for all tips, a spherical asperity of radius R in contact with a flat,
the coefficient of friction increases with humidity up to smooth surface with composite modulus of elasticity E ∗
about ambient, beyond which it starts to decrease. The and in the presence of a liquid with a concave meniscus,
initial increase in the coefficient of friction with humid- the attractive meniscus force (adhesive force), desig-
ity arises from the fact that the thickness of the water nated as Fm or Wad , is given by [28.7, 11]
film increases with increasing humidity, which results
in a larger number of nanoasperities forming menis- Wad = 2π Rγ (cos θ1 + cos θ2 ) , (28.11)
822 Part D Bio-/Nanotribology and Bio-/Nanomechanics
and
Part D 28.2
Fm
3.2τ
Sphere μe ≈ 1 ,
σp 2
E∗ Rp + Wad
W
R
θ2
where σp and Rp are the standard deviation of the sum-
θ1
Flat mit heights and the average summit radius, respectively.
Liquid with surface tension γ Note that the coefficient of friction depends upon the
Fm = 2πRγ (cosθ1 + cosθ2)
surface roughness. In the plastic contact regime, the
expression for μp in (28.14) does not change.
Fig. 28.33 Meniscus formation from a liquid condensate The sources of the adhesive force in a wet contact
at the interface for a sphere in contact with a plane surface in AFM experiments performed in an ambient environ-
ment include mainly attractive meniscus force due to
where γ is the surface tension of the liquid and θ1 and θ2 capillary condensation of water vapor from the environ-
are the contact angles of the liquid with surfaces 1 and ment. The meniscus force for a single contact increases
2, respectively. For an elastic contact for both extrinsic with an increase in tip radius. A sharp AFM tip in con-
(W) and intrinsic (Wad ) normal load, the friction force tact with a smooth surface at low loads (on the order
is given by of a few nN) for most materials can be simulated as
2 a single-asperity contact. At higher loads, for rough and
3(W + Wad )R 3 soft surfaces, multiple contacts would occur. Further-
Fe = πτ , (28.12)
4E ∗ more, at low loads (nN range) for most materials the
where W is the external load and τ is the average shear local deformation would be primarily elastic. Assum-
strength of the contacts. (Surface energy effects are not ing that the shear strength of contacts does not change,
considered here.) Note that adhesive force increases lin- the adhesive force for smooth and hard surfaces at low
early with increasing tip radius, and the friction force normal load (on the order of a few nN) (for a single-
increases with tip radius as R2/3 and with normal load asperity contact in the elastic contact regime) would
as (W + Wad )2/3 . Experimental data in support of the increase with increasing tip radius, and the coefficient
W 2/3 dependence on the friction force can be found in of friction would decrease with increasing total nor-
various references [28.115]. The coefficient of friction mal load as (W + Wad )−1/3 and would increase with
μe is obtained from (28.12) as increasing tip radius as R2/3 . In this case, the Amon-
tons law of friction, which states that the coefficient
2 of friction is independent of normal load and indepen-
Fe 3R 3 1
μe = = πτ . dent of apparent area of contact, does not hold. For
(W + Wad ) 4E ∗ (W + Wad ) 3
1
a single-asperity plastic contact and multiple-asperity
(28.13) plastic contacts, neither the normal load nor the tip ra-
In the plastic contact regime [28.7], the coefficient of dius comes into play in the calculation of the coefficient
friction μp is obtained as of friction. In the case of multiple-asperity contacts,
the number of contacts increases with increasing nor-
Fp τ
μp = = , (28.14) mal load; therefore the adhesive force increases with
(W + Wad ) Hs increasing load.
where Hs is the hardness of the softer material. Note In the data presented earlier in this section, the effect
that, in the plastic contact regime, the coefficient of of tip radius and humidity on the adhesive forces and
friction is independent of the external load, adhesive coefficient of friction is investigated for experiments
contributions, and surface geometry. with Si(100) surface at loads in the range 10–100 nN.
For comparison, for multiple-asperity contacts in The multiple-asperity elastic-contact regime is relevant
the elastic contact regime, the total adhesive force Wad for this study involving large tip radii. An increase
is the summation of the adhesive forces at n individual in humidity generally results in an increase in the
contacts, number of meniscus bridges, which would increase
n the adhesive force. As suggested earlier, this increase
Wad = (Wad )i (28.15) in humidity may also decrease the shear strength
i=1 of contacts. A combination of an increase in adhe-
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 823
sive force and a decrease in shear strength would the volume of the tip and sample. It is given as [28.114]
Part D 28.2
affect the coefficient of friction. An increase in tip
c A B
radius would increase the meniscus force (adhesive V (D) = − + , (28.16)
force). A substantial increase in the tip radius may 12 D 210D7
also increase the interatomic forces. These effects in- where c is the width of the parabolic tip (= the diam-
fluence the coefficient of friction with increasing tip eter in the case of a spherical tip), and A and B are
radius. two potential parameters, where A is Hamaker constant.
Equation (28.16) provides expressions for attractive
28.2.8 Separation Distance Dependence of and repulsive parts. The calculations were made for
Meniscus and van der Waals Forces Lennard–Jones force (total) and van der Waals force (at-
tractive part) for two Hamaker constants: 0.04 × 10−19 J
When two surfaces are in close proximity, sources of (representative of polymers) and 3.0 × 10−19 J (repre-
adhesive forces are weak van der Waals attraction and sentative of ceramics), and meniscus force for a water
meniscus formation. The relative magnitudes of the film (γ = 72.5 N/m). Figure 28.34 shows various
forces from these two sources are dependent upon var- forces as a function of separation distance. The effect
ious factors, including the interplanar separation, their of two relative humidities and three tip radii, which af-
surface roughness, their hydrophobicity, and the rela-
tive humidity (liquid volume) [28.114]. The meniscus
contribution dominates at moderate to high humidities, a) F (nN)
0
whereas van der Waals forces dominate at asperities
a few nm apart. In some micro/nanocomponents, it is –2
important to know the relative contribution of these two –4
p/p0 = 0.1 p/p0 = 0.9
sources as a function of interplanar separation in order –6
to design an interface for low adhesion. For example,
–8
if two ultrasmooth surfaces come into close proximity,
with an interplanar separation on the order of 1 nm, van –10 Lennard–Jones force
der Waals forces may dominate, and their magnitude –12 van der Waals force
may be reduced by creating bumps on one of the inter- Meniscus force
–14
γl = 72.5 N/m, θ1 = θ2 = 0°, R = 20 nm
faces. This analysis is also of interest in AFM studies to –16
understand the distance dependence of adhesive forces 0 1 2 3 4 5
as the tip goes in and out of contact. D (nm)
b) F (nN)
Stifter et al. [28.114] modeled the contact of 0
a parabolic-shaped tip and a flat, smooth sample sur-
face. The tip may represent a surface asperity on –10
an interface or an AFM tip in an AFM experiment. –20
They calculated van der Waals and meniscus forces R = 20 nm
as a function of various parameters, namely tip ge- –30 R = 50 nm
ometry, tip–sample starting distance, relative humidity, –40 R = 80 nm
surface tension, and contact angle. They compared the Lennard–Jones force
–50 van der Waals force
meniscus forces with van der Waals forces to under-
Meniscus force
stand their relative importance under various operating –60
γl = 72.5 N/m, p/p0 = 0.1, θ1 = θ2 = 0°
conditions. –70
The interaction force between the tip and sample un- 0 0.5 1 1.5 2 2.5
der dry conditions is the Lennard–Jones force derived D (nm)
from the Lennard–Jones potential. The Lennard–Jones Fig. 28.34a,b Relative contribution of meniscus, van der
potential is composed of two interactions – the van Waals, and Lennard–Jones forces (F) as a function of
der Waals attraction and Pauli repulsion. van der Waals separation distance (D) and at (a) two values of relative
forces are significant because they are always present. humidity ( p/ p0 ) for tip radius of 20 nm and Hamaker
For a parabolic tip above a half-plane with a separation constants of 0.04 × 10−19 and 3.0 × 10−19 J, and (b) three
D between the tip and plane, the Lennard–Jones poten- tip radii (R) and Hamaker constant of 3.0 × 10−19 J (af-
tial is obtained by integrating the atomic potential over ter [28.114])
824 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Table 28.3 Micro- and nanoscale values of adhesive force and coefficient of friction in micro- and nanoscale measure-
Part D 28.2
fect meniscus forces, was also studied. The two dashed croscale [28.38, 98, 116, 124]. Adhesive force and
curves indicate the spread of possible van der Waals coefficient of friction values on the nanoscale are about
forces for the two Hamaker constants. The figure shows half to one order of magnitude lower than that on the
that meniscus forces exhibit weaker distance depen- microscale. Scale dependence is clearly observed in
dence. The meniscus forces can be stronger or weaker this data. As a further evidence of scale dependence,
than the van der Waals forces for distances smaller than Table 28.4 shows the coefficient of friction measured
≈ 0.5 nm. For longer distances, the meniscus forces are for Si(100), HOPG, natural diamond, and DLC on the
stronger than the van der Waals forces. van der Waals nanoscale and microscale. It is clearly observed that
forces must be considered for a tip–sample distance up friction values are scale dependent.
to a few nm (D < 5 nm). The meniscus forces operate To estimate the scale length, the apparent contact
up to breakage of the meniscus in the range from 5 to radius at test loads was calculated and is presented in
20 nm [28.114]. the table. Mean apparent pressures are also calculated
and presented. For nanoscale AFM experiments, it is as-
28.2.9 Scale Dependence in Friction sumed that an AFM tip coming into contact with a flat
surface represents a single-asperity elastic contact, and
Table 28.3 presents adhesive force and coefficient Hertz analysis was used for the calculations. In the mi-
of friction data obtained on the nanoscale and mi- croscale experiments, a ball coming into contact with
Table 28.4 Micro- and nanoscale values of the coefficient of friction, typical physical properties of specimen, and calculated ap-
parent contact radii and apparent contact pressures at loads used in micro- and nanoscale measurements. For calculation purposes
it is assumed that contacts on micro- and nanoscale are single-asperity elastic contacts (after [28.123])
Coefficient Apparent contact radius Mean apparent pressure
of friction at test load for at test load for
Sample Micro- Nano- Elastic Poisson’s Hardness Microscale Nano- Microscale Nanoscale
scale scale modulus ratio (GPa) (μm) scale (GPa) (GPa)
(GPa) (upper limit) (nm) (lower limit)
Si(100) 0.47 a 0.06 c 130 e,f 0.28 f 9–10 e,f 0.8–2.2 a 1.6–3.4 c 0.05–0.13 a 1.3–2.8 c
wafer
Graphite 0.1 b 0.006 c 9–15 g − 0.01 j 62 b 3.4–7.4 c 0.082 b 0.27–0.58 c
(HOPG) (9) (0.25)
Natural 0.2 b 0.05 c 1140 h 0.07 h 80–104 g,h 21 b 1.1–2.5 c 0.74 b 2.5–5.3 c
diamond
DLC film 0.19 a 0.03 d 280 i 0.25 i 20–30 i 0.7–2.0 a 1.3–2.9 d 0.06–0.16 a 1.8–3.8 d
a 500 μm-radius Si(100) ball at 100–2000 μN and 720 μm/s in dry air [28.116]
b 3 mm-radius Si3 N4 ball (elastic modulus 310 GPa, Poisson’s ratio 0.22 [28.117]) at 1 N and 800 μm/s [28.38]
c 50 nm-radius Si N tip at load range from 10–100 nN and 0.5 nm/s, in dry air [28.38]
3 4
d 50 nm-radius Si N tip at load range from 10–100 nN in dry air [28.116]
3 4
e [28.118], f [28.119], g [28.117], h [28.120], i [28.121], j [28.122]
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 825
a flat surface represents multiple-asperity contacts due as bulk liquid and in the form of annular-shaped cap-
Part D 28.2
to the roughness, and the contact pressure of the asper- illary condensate in the contact zone. A quantitative
ity contacts is higher than the apparent pressure. For the theory of scale effects in friction should consider the
calculation of a characteristic scale length for multiple- effect of scale on physical properties relevant to various
asperity contacts, which is equal to the apparent length contributions.
of contact, Hertz analysis was also used. This analysis According to the adhesion and deformation model
provides an upper limit on apparent radius and lower of friction, the coefficient of dry friction μ is the sum of
limit on the mean contact pressure. an adhesion component μa and a deformation (plowing)
There are several factors responsible for the dif- component μd . The latter, in the presence of particles, is
ferences in the coefficients of friction at the micro- the sum of an asperity-summit deformation component
and nanoscale. Among these are the contributions from μds and a particle-deformation component μdp , so that
wear and contaminant particles, the transition from the total coefficient of friction is [28.125]
elasticity to plasticity, and the meniscus effect. The Fa + Fds + Fdp
contribution of wear and contaminant particles is more μ = μa + μds + μdp =
significant at the macro/microscale because of the larger W
Ara τa + Ads τds + Adp τdp
number of trapped particles, referred to as the third- = ,
body contribution. It can be argued that for nanoscale W
(28.17)
AFM experiments the asperity contacts are predom-
inantly elastic (with average real pressure being less where W is the normal load, F is the friction force, and
than the hardness of the softer material), and adhe- Ara , Ads , and Adp are the real areas of contact during
sion is the main contributor to the friction, whereas adhesion, two-body deformation, and with particles, re-
for microscale experiments the asperity contacts are spectively; τ is the shear strength. The subscripts “a,”
predominantly plastic, and deformation is an impor- “ds,” and “dp” correspond to adhesion, summit defor-
tant factor. It will be shown later that hardness has mation, and particle deformation, respectively.
a scale effect; it increases with decreasing scale and The adhesional component of friction depends on
is responsible for less deformation on a smaller scale. the real area of contact and adhesion shear strength.
The meniscus effect results in an increase of friction The real area of contact is scale dependent due to
with increasing tip radius (Fig. 28.32). Therefore, the the scale dependence of the surface roughness (for
third-body contribution, the scale-dependent hardness, elastic and plastic contacts) and due to the scale de-
and other properties transition from elastic contacts in pendence of hardness (for plastic contacts) [28.125].
nanoscale contacts to plastic deformation in microscale We limit the analysis here to multiple-asperity contacts.
contacts, and the meniscus contribution plays an impor- For this case, the scale L is defined as the appar-
tant role [28.123, 125, 126]. ent size of the contact between the two bodies. (For
Friction is a complex phenomenon, which involves
asperity interactions involving adhesion and deforma-
tion (plowing). Adhesion and plastic deformation imply Solid–solid contact
W W
energy dissipation, which is responsible for friction
(Fig. 28.35) [28.6, 11]. A contact between two bod- 1 1
ies takes place on high asperities, and the real area of
contact (Ar ) is a small fraction of the apparent area 3
of contact. During the contact of two asperities, a lat-
2 2
eral force may be required for asperities of a given
slope to climb against each other. This mechanism Two-body contact Three-body contact
is known as the ratchet mechanism, and it also con-
Plowing during sliding
tributes to the friction. Wear and contaminant particles
present at the interface, referred as the third body,
also contribute to the friction (Fig. 28.35). In addi-
tion, during contact, even at low humidity, a meniscus
is formed (Fig. 28.33). Generally any liquid that wets
or has a small contact angle on surfaces will con- Fig. 28.35 Schematic of two-body and three-body dry
dense from vapor into cracks and pores on surfaces contacts of rough surfaces
826 Part D Bio-/Nanotribology and Bio-/Nanomechanics
completeness, for single-asperity contact, the scale is a random Gaussian surface [28.125]
Part D 28.2
where μae0 and μap0 are the values of the coefficient of μae /μae0 Elastic
friction at the macroscale (L ≥ L c ), m and n are in- 1
dices that characterize the scale dependence of surface Ls /L lc = 1000
parameters, a0 is the macroscale value of the mean con-
Ls /Llc = 1
tact radius, L c is the long-wavelength limit for scale
dependence of the contact parameters, s and d are Ls /L lc = 0
material-specific characteristic length parameters, and 0
L s and L d are length parameters related to s and d . 0 0.5 1
L /L lc
The scale dependence of the adhesional component of μap /μap0 Plastic
the coefficient of friction is presented in Fig. 28.36, 2
based on (28.18) and (28.19). Ld /Ls = 0.25 L d /Llc
1
Based on the assumption that multiple asperities of 1000
two rough surfaces in contact have a conical shape, the 1
two-body deformation component of friction can be de-
termined as [28.6, 11] Ld /Ls = 5
0
2 tan θr 0 0.5 1
μds = , (28.20) L /L lc
π
Fig. 28.36 Normalized results for the adhesional compo-
where θr is the roughness angle (or attack angle) of nent of the coefficient of friction, as a function of L/L c for
a conical asperity. Mechanical properties affect the real multiple-asperity contact. Data are presented for m = 0.5,
area of contact and shear strength, and these cancel out n = 0.2. For multiple-asperity plastic contact, data are pre-
in (28.16) [28.125]. Based on a statistical analysis of sented for two values of L d /L c (after [28.125])
Nanotribology, Nanomechanics, and Materials Characterization 28.2 Surface Imaging, Friction, and Adhesion 827
Part D 28.2
Asperities plowing contribution
μds /μds0
5 d 2 1 + 2 d
d
m = 0.5 μdp = μdp0 n tr , (28.22)
2 d
n = 0.2 d02
1+
d0
where d is the mean particle diameter, d0 is the
2.5
macroscale value of the mean particle diameter, n tr
is the number of trapped particles divided by the to-
tal number of particles, and μdp0 is the macroscale
0 (L → ∞, n tr → 1) value of the third-body deformation
0 0.5 1 component of the coefficient of friction. The scale de-
L/L lc
Three-body plowing contribution pendence of μdp is shown in Fig. 28.37 (bottom curve)
ntr, μdp /μdp0 Log normal distribution based on (28.22). Based on scale effect predictions
1 presented in Figs. 28.36 and 28.37, the trends in the
experimental results presented in Table 28.3 can be ex-
Fraction of trapped particles plained.
0.5 The scale dependence of meniscus effects in fric-
Coefficient of friction tion, wear, and interface temperature can be analyzed in
ln(dln) = 2, a similar way [28.126].
σln = 1, ld /σln = 1
0 0 To demonstrate the load dependence of friction at
10 102 104 106 the nano/microscale, the coefficient of friction as a func-
L/α (nm)
tion of normal load is presented in Fig. 28.39. The
Fig. 28.37 Normalized results for the two-body deforma- coefficient of friction was measured by Bhushan and
tion component of the coefficient of friction, and the Kulkarni [28.42] for a Si3 N4 tip versus Si, SiO2 , and
number of trapped particles divided by the total number natural diamond using an AFM. They reported that,
of particles and three-body deformation component of the for low loads, the coefficient of friction is independent
coefficient of friction, normalized by the macroscale value of load and then increases with increasing load after
for the log-normal distribution of debris size, where α is a certain load. It is noted that the critical load values
the probability of a particle in the border zone leaving for Si and SiO2 correspond to stresses equal to their
the contact region. Various constants given in the figure
correspond to the log-normal distribution (after [28.125])
Coefficient of friction
0.2 Si(111)
SiO2
d Natural diamond
0.15
0.1
Contact region
d/2
0.05
0
Border region Corner 0 10 20 30 40 50
Normal load (µN)
hardness values, which suggests that the transition to at higher loads for Si and SiO2 approach the macroscale
Part D 28.3
10 nN
nm
nm 20
50 4
400
10 3
25 300
2
200 0 1
0 0
100 1
0 2 0
100 3
200 4 µm
300 0
400 nm Fig. 28.41a,b Surface plots of (a) Si(111) scratched for ten
100 nN cycles at various loads and scanning velocity of 2 μm/s.
Note that x- and y-axes are in μm and z-axis is in nm,
nm and (b) Si(100) scratched in one unidirectional scan cycle
50 at normal force of 80 μN and different scanning veloci-
400 ties
25 300
tape surface. Thus, deformation and movement of the
200 soft materials on a nanoscale can be observed.
0 100
0
100
28.3.2 Microscale Scratching
200 0
300
400 nm
The AFM can be used to investigate how surface
Fig. 28.40 Surface roughness maps of a polymeric mag- materials can be moved or removed on micro- to
netic tape at applied normal loads of 10 and 100 nN. nanoscales, for example, in scratching and wear [28.5,
Location of the change in surface topography as a result 29] (where these things are undesirable) and nanofab-
of nanowear is indicated by arrows (after [28.37]) rication/nanomachining (where they are desirable).
Nanotribology, Nanomechanics, and Materials Characterization 28.3 Wear, Scratching, Local Deformation 829
Figure 28.41a shows microscratches made on Si(111) AFM in tapping mode to study failure mechanisms. Fig-
Part D 28.3
at various loads with a scanning velocity of 2 μm/s af- ure 28.42 shows data from a scratch test on Si(100)
ter ten cycles [28.41]. As expected, the scratch depth with scratch length of 25 μm and scratching velocity of
increases linearly with load. Such microscratching mea- 0.5 μm/s. At the beginning of the scratch, the coeffi-
surements can be used to study failure mechanisms on cient of friction is 0.04, a typical value for silicon. At
the microscale and to evaluate the mechanical integrity about 35 μN (indicated by the arrow in the figure), there
(scratch resistance) of ultrathin films at low loads. is a sharp increase in the coefficient of friction, which
To study the effect of scanning velocity, unidi- indicates the critical load. Beyond the critical load, the
rectional scratches 5 μm in length were generated at coefficient of friction continues to increase steadily. In
scanning velocities ranging from 1 to 100 μm/s at vari- the postscratch image, we note that, at the critical load,
ous normal loads ranging from 40 to 140 μN. No effect a clear groove starts to form. This implies that Si(100)
of scanning velocity was observed for a given nor- was damaged by plowing at the critical load, associ-
mal load. For representative scratch profiles at 80 μN ated with plastic flow of the material. At and after
(Fig. 28.41b). This may be because of a small effect of the critical load, small and uniform debris is observed,
frictional heating with the change in scanning velocity and the amount of debris increases with increasing
used here. Furthermore, for a small change in inter- normal load. Sundararajan and Bhushan [28.59] have
face temperature, there is a large underlying volume to also used this technique to measure the scratch resis-
dissipate the heat generated during scratching. tance of diamond-like carbon coatings with thickness
Scratching can be performed under ramped load- of 3.5–20 nm.
ing to determine the scratch resistance of materials and
coatings. The coefficient of friction is measured dur- 28.3.3 Microscale Wear
ing scratching, and the load at which the coefficient of
friction increases rapidly is known as the critical load, By scanning the sample in two dimensions with the
which is a measure of scratch resistance. In addition, AFM, wear scars are generated on the surface. Fig-
postscratch imaging can be performed in situ with the ure 28.43 shows the effect of normal load on wear
100 2 0.4
75 1.5 0.3
50 1 0.2
25 0.5 0.1
0 0 0
0 5 10 15 20 25 0 25 50 75 100 125
Distance (µm) Normal load (µN)
0 nm 20 nm
c)
2 µm
Fig. 28.42 (a) Applied normal load and friction signal measured during a microscratch experiment on Si(100) as a func-
tion of scratch distance, and (b) friction data plotted in the form of the coefficient of friction as a function of normal load.
(c) AFM surface height image of a scratch obtained in tapping mode (after [28.59])
830 Part D Bio-/Nanotribology and Bio-/Nanomechanics
2
0 1
0
1
2 0
3
4 µm
Fig. 28.44 (a) Typical gray-scale and (b) inverted AFM
images of a wear mark created using a diamond tip at
normal load of 40 μN and one scan cycle on Si(100) sur-
face
Part D 28.3
shown in Fig. 28.46a,b. The bend contours are observed Tip a) 40 µN b) 40 µN
sliding Outside wear mark
to pass through the wear mark in the micrograph. The direction
bend contours around and inside the wear mark are in-
dicative of a strain field, which in the absence of applied
stresses can be interpreted as plastic deformation and/or
elastic residual stresses. Often, localized plastic defor-
mation during loading would lead to residual stresses
during unloading; therefore, bend contours reflect a mix
of elastic and plastic strains. The wear debris is ob- 1 µm
served outside the wear mark. The enlarged view of the
wear debris in Fig. 28.46c shows that much of the de-
bris is ribbon-like, indicating that material is removed c) 40 µN d) 40 µN
by a cutting process via plastic deformation, which is Wear
ear debris Wear
ear debris
consistent with the SEM observations. The diffraction
pattern from inside the wear mark is similar to that of
virgin silicon, showing no evidence of any phase trans-
formation (amorphization) during wear. A selected-area
diffraction pattern of the wear debris shows some dif-
fuse rings, which indicates the existence of amorphous
material in the wear debris, confirmed as silicon ox-
ide products from chemical analysis. It is known that 200 nm
plastic deformation occurs by generation and propaga-
tion of dislocations. No dislocation activity or cracking Fig. 28.46a–d Bright-field TEM micrographs (a) and diffraction
was observed at 40 μN. However, dislocation arrays patterns (b), of the wear mark (a,b) and wear debris (c,d) on Si(100)
could be observed at 80 μN. Figure 28.47 shows TEM produced at normal load of 40 μN and one scan cycle. Bend con-
micrographs of the worn region at 80 μN; for better ob- tours around and inside wear mark are observed
servation of the worn surface, wear debris was moved
out of the wear mark by using AFM with a large-area plastic deformation. This corroborates the observations
scan at 300 nN after the wear test. The existence of dis- made in scratch tests at ramped load in the previous sec-
location arrays confirms that material removal occurs by tion. It is concluded that the material on the microscale
at high loads is removed by plastic deformation with
a small contribution from elastic fracture [28.129].
a) 80 µN b) 80 µN
To understand wear mechanisms, evolution of wear
can be studied using AFM. Figure 28.48 shows the evo-
lution of wear marks of a DLC-coated disk sample. The
data illustrate how the microwear profile for a load of
20 μN develops as a function of the number of scanning
cycles [28.41]. Wear is not uniform, but is initiated at
the nanoscratches. Surface defects (with high surface
energy) present at the nanoscratches act as initiation
200 nm
20
sites for wear. Coating deposition also may not be uni-
1 µm
form on and near nanoscratches, which may lead to
coating delamination. Thus, scratch-free surfaces will
Tip Tip
sliding sliding
be relatively resistant to wear.
direction direction Wear precursors (precursors to measurable wear)
can be studied by making surface potential measure-
Fig. 28.47 (a) Bright-field and (b) weak-beam TEM mi- ments [28.79–81]. The contact potential difference, or
crographs of a wear mark produced in Si(100) at normal simply the surface potential between two surfaces, de-
load of 80 μN and one scan cycle, showing bend contours pends on a variety of parameters such as the electronic
and dislocations (after [28.129]) work function, adsorption, and oxide layers. The sur-
832 Part D Bio-/Nanotribology and Bio-/Nanomechanics
2
0 1
0
1 20 µN
2 0
3
4 µm 10 cycles
nm
1000
4
500 3
10 µm 10 µm
0 100 nm 0 200 mV 2
0 1
b)
0
1 20 µN
2 0
3
4 µm 15 cycles
nm
1000
4
500 3
2
0 1
5 µm 5 µm 0
1 20 µN
0 25 nm 0 150 mV 2 0
3
4 µm 20 cycles
Fig. 28.49 (a) Surface height and change in surface potential maps
of wear regions generated at 1 μN (a) and 9 μN (b) on a single-
crystal aluminum sample showing bright contrast in the surface 9 μN, are shown in Fig. 28.49a. (Note that the sign of
potential map on the worn regions. (b) Close-up of the upper (low- the change in surface potential is reversed here from
load) wear region (after [28.79]) that in [28.79].) It is evident that both abraded regions
Nanotribology, Nanomechanics, and Materials Characterization 28.3 Wear, Scratching, Local Deformation 833
Part D 28.3
Strain
different strains (after [28.73]) 0.83%
coating with and without a thin diamond-like carbon newer cracks keep nucleating and propagate with re-
Part D 28.3
(DLC) overcoat on a polymeric substrate, all of which duced crack spacing. At 3.75% strain, another crack
had a particulate back-coating [28.15]. They also stud- can be seen nucleating. This crack continues to grow
ied the polyethylene terephthalate (PET) substrate with parallel to the first one. When the tape is unloaded af-
6 μm thickness. They reported that cracking of the coat- ter stretching up to a strain of ≈ 2%, i. e., within the
ings started at ≈ 1% strain for all tapes, much before the elastic limit of the substrate, the cracks close perfectly,
substrate started to yield at ≈ 2% strain. Figure 28.50 and it is impossible to determine the difference from the
shows topographical images of the MP tape at different unstrained tape.
strains. At 0.83% strain, a crack can be seen, origi- Figure 28.51 shows topographical images of the
nating at the marked point. As the tape is stretched three magnetic tapes and the PET substrate after being
further along this direction, as shown in Fig. 28.50, strained to 3.75%, which is well beyond the elastic limit
the crack propagates along the shorter boundary of the of the substrate. The MP tape develops numerous short
ellipsoidal particle. However, the general direction of cracks perpendicular to the direction of loading. In tapes
the crack propagation remains perpendicular to the di- with metallic coating, the cracks extend throughout the
rection of stretching. The length, width, and depth of tape width. In the ME tape with the DLC coating, there
the cracks increase with strain, and at the same time is a bulge in the coating around the primary cracks
that are initiated when the substrate is still elastic, like
crack A in the figure. The white band on the right-hand
Stress MP tape Crack width Crack spacing
(MPa) side of the figure is the bulge of another crack. Sec-
(µm) (µm)
150 0.6 25 ondary cracks, such as B and C, are generated at higher
strains and are straighter compared with the primary
cracks. In ME tape with a Co-O film on a PET substrate,
20
with a thickness ratio of 0.03, both with and without
100 Width 0.4 DLC coating, no difference is observed in the rate of
15 growth between primary and secondary cracks. Failure
is cohesive with no bulging of the coating. This seems to
10 suggest that the DLC coating has residual stresses that
50 0.2
relax when the coating cracks, causing delamination.
Spacing 5 Since the stresses are already relaxed, the secondary
crack does not result in delamination. The presence of
0 0 0 the residual stress is confirmed by the fact that a free-
0 2 4 6 8 10 standing ME tape curls up (in a cylindrical form with
Strain (%) its axis perpendicular to the tape length) with a radius
of curvature of ≈ 6 mm, whereas the ME tape without
Stress ME tape Crack width Crack spacing
(MPa) (µm) (µm) the DLC does not curl. The magnetic coating side of the
(67) PET substrate is much smoother at smaller scan lengths.
150 0.6 25
However, in 20 μm scans it has a lot of bulges, which
20
appear as white spots in the figure. These spots change
shape even while scanning the samples in tapping mode
100 0.4 at very low contact forces.
15
The variation of average crack width and average
Width crack spacing with strain is plotted in Fig. 28.52. The
10 crack width is measured at a spot along a given crack
50 0.2
over a distance of 1 μm in the 5 μm scan image at differ-
Spacing 5 ent strains. The crack spacing is obtained by averaging
the intercrack distance measured in five separate 50 μm
0 0 0 scans at each strain. It can be seen that the cracks nucle-
0 2 4 6 8 10 ate at a strain of about 0.7–1.0%, well within the elastic
Strain (%)
limit of the substrate. There is a definite change in the
Fig. 28.52 Variation of stress, crack width, and crack spacing with slope of the load–displacement curve at the strain where
strain in two magnetic tapes (after [28.73]) cracks nucleate, and the slope after that is closer to the
Nanotribology, Nanomechanics, and Materials Characterization 28.3 Wear, Scratching, Local Deformation 835
Part D 28.3
Maximum stress (MPa)
40
ing mean stress. This is consistent with the literature,
ME tape and is because for lower mean stress the corresponding
stress amplitude is relatively high and this causes fail-
ure. The endurance limit is found to be almost the same
35 for all three mean stresses. In the case of ME tape with-
σmean = 30.6 MPa
out DLC as well, the critical number of cycles is found
to be in the same range.
30 24.5 MPa In situ surface characterization of unstretched and
stretched films has been used to measure the Poisson’s
21.4 MPa ratio of polymeric thin films by Bhushan et al. [28.130].
Uniaxial tension is applied by the tensile stage. Surface
25 0
10 101 102 103 104 105 height profiles obtained from the AFM images of un-
Number of cycles stretched and stretched samples are used to monitor the
Maximum stress (MPa)
40
ME tape without DLC a) Si(100)
(µm)
10 nm
1.5
35
5 nm
0 nm
25 0
10 101 102 103 104 105 0.5
Number of cycles
Fig. 28.53 S–N curve for two magnetic tapes with maxi-
mum stress plotted on the ordinate and number of cycles
0
to failure on the abscissa. The data points marked with 0 0.5 1 1.5 (µm)
arrows indicate tests for which no failure (cracking) was b)
observed in the scan area, even after a large number of (µm)
1
cycles (10 000) 10 nm
changes in displacements of the polymer films in the geometry – can be controlled precisely to control the
Part D 28.4
longitudinal and lateral directions simultaneously. depth and length of the devices.
Nanofabrication using mechanical scratching has
28.3.5 Nanofabrication/Nanomachining several advantages over other techniques. Better control
over the applied normal load, scan size, and scanning
An AFM can be used for nanofabrication/nanomachin- speed can be used for nanofabrication of devices. Us-
ing by extending the microscale scratching opera- ing the technique, nanofabrication can be performed
tion [28.5,14,41,78]. Figure 28.54 shows two examples on any engineering surface. Use of chemical etching
of nanofabrication. The patterns were created on or reactions is not required, and this dry nanofabrica-
a single-crystal Si(100) wafer by scratching the sample tion process can be used where the use of chemicals
surface with a diamond tip at specified locations and and electric field is prohibited. One disadvantage of
scratching angles. Each line is scribed manually at nor- this technique is the formation of debris during scratch-
mal load of 15 μN and writing speed of 0.5 μm/s. The ing. At light loads, debris formation is not a problem
separation between lines is ≈ 50 nm, and the variation compared with during high-load scratching. However,
in line width is due to the tip asymmetry. Nanofabrica- debris can be easily removed from the scan area at light
tion parameters – normal load, scanning speed, and tip loads during scanning.
28.4 Indentation
Mechanical properties on relevant scales are needed Detection of the transfer of material on a nanoscale
for the analysis of friction and wear mechanisms. is possible with the AFM. Indentation of C60 -rich
Mechanical properties, such as hardness and Young’s fullerene films with an AFM tip has been shown [28.60]
modulus of elasticity, can be determined on micro- to result in the transfer of fullerene molecules to the
to picoscales using the AFM [28.37, 41, 56, 62] and AFM tip, as indicated by discontinuities in the can-
a depth-sensing indentation system used in conjunction tilever deflection as a function of sample traveling
with an AFM [28.42, 131–133]. distance in subsequent indentation studies.
Indentability on the scale of subnanometers of soft The indentation hardness of surface films with indenta-
samples can be studied in the force calibration mode tion depth as small as ≈ 1 nm can be measured using
(Fig. 28.6) by monitoring the slope of cantilever de- an AFM [28.14, 61, 62]. Figure 28.56 shows gray scale
flection as a function of sample traveling distance after
the tip is engaged and the sample is pushed against
Tip deflection (6 nm/div)
the tip. For a rigid sample, cantilever deflection equals
the sample traveling distance, but the former quantity Retracting
is smaller if the tip indents the sample. In an exam- Extending
ple for a polymeric magnetic tape shown in Fig. 28.55,
the line in the left portion of the figure is curved with
a slope of < 1 shortly after the sample touches the
tip, which suggests that the tip has indented the sam-
ple [28.37]. Later, the slope is unity, suggesting that A
C D
the tip no longer indents the sample. This observa-
tion indicates that the tape surface is soft (polymer B
rich) locally but hard (as a result of magnetic par- Z position (15 nm/div)
ticles) underneath. Since the curves in extending and
retracting modes are identical, the indentation is elas- Fig. 28.55 Tip deflection (normal load) as a function of
tic up to the maximum load of ≈ 22 nN used in the z (separation distance) for a polymeric magnetic tape (af-
measurements. ter [28.37])
Nanotribology, Nanomechanics, and Materials Characterization 28.4 Indentation 837
Part D 28.4
60 µN increased, the indents become clearer, and indentation
1 nm depth increases. For the case of hardness measurements
40 at shallow depths on the same order as variations in sur-
nm 500 face roughness, it is desirable to subtract the original
20 20 nm (unindented) map from the indent map for an accurate
250
measurement of the indentation size and depth [28.41].
10 nm To make accurate measurements of hardness at
0
0 shallow depths, a depth-sensing nano/picoindentation
250 0 system (Fig. 28.9) is used [28.61]. Figure 28.57 shows
500 nm 0 nm
65 µN, load–displacement curves at different peak loads for
2.5 nm, Si(100). Loading/unloading curves often exhibit sharp
16.6 GPa discontinuities, particularly at high loads. Discontinu-
40 ities, also referred to as pop-ins, occurring during the
nm 500 initial loading part of the curve, mark a sharp transition
20 20 nm from pure elastic loading to plastic deformation of the
250
specimen surface, thus corresponding to an initial yield
10 nm point. The sharp discontinuities in the unloading part
0
0 of the curves are believed to be due to the formation
250 0 of lateral cracks which form at the base of the median
500 nm 0 nm
crack, which results in the surface of the specimen being
70 µN, thrusted upward. Load–displacement data at residual
3 nm,
15.8 GPa depths as low as ≈ 1 nm can be obtained. The inden-
tation hardness of surface films has been measured for
40
nm 500 various materials at a range of loads, including Si(100)
20 nm
up to a peak load of 500 μN and Al(100) up to a peak
20
load of 2000 μN by Bhushan et al. [28.61] and Kulkarni
250 and Bhushan [28.131–133]. The hardnesses of single-
0 10 nm
crystal silicon and single-crystal aluminum at shallow
0
250 0 depths on the order of a few nm (i.e., on the nanoscale)
500 nm 0 nm
100 µN,
7 nm, Load (µN)
11.7 GPa
350
40 60
nm 50 Si(100)
500 300
40
20 20 nm
250 30
250 20
10 nm 200 10
0
0 0
250 0 0 1 2 3 4 5 6
500 nm 0 nm 150
Hardness (GPa)
15 a lower probability of encountering material defects.
Bhushan and Koinkar [28.56] have used AFM mea-
10 Si(100) surements to show that ion implantation of silicon
5 surfaces increases their hardness and thus their wear
resistance. Formation of surface alloy films with im-
0
0 5 10 15 20 25 proved mechanical properties by ion implantation is
Residual depth (nm) of growing technological importance as a means of
Hardness (GPa) improving the mechanical properties of materials. Hard-
0.75 ness of 20 nm-thick DLC films have been measured by
Al(100)
0.5 Kulkarni and Bhushan [28.133].
The creep and strain-rate effects (viscoelastic ef-
0.25
fects) of ceramics can be studied using a depth-sensing
0 indentation system. Bhushan et al. [28.61] and Kulkarni
0 100 200 300
and Bhushan [28.131–133] have reported that ceramics
Residual depth (nm)
(single-crystal silicon and diamond-like carbon) ex-
Fig. 28.58 Indentation hardness as a function of residual hibit significant plasticity and creep on a nanoscale.
indentation depth for Si(100) (after [28.61]) and Al(100) Figure 28.59a shows load–displacement curves for
(after [28.131]) single-crystal silicon at various peak loads held for
180 s. To demonstrate the creep effects, the load–
are found to be higher than at depths on the order of displacement curves for 500 μN peak load held for 0
a few hundred nm (i.e., on the microscale) (Fig. 28.58). and 30 s are also shown in the inset. Note that signif-
Microhardness has also been reported to be higher than icant creep occurs at room temperature. Nanoindenter
that on the millimeter scale by several investigators. The experiments conducted by Li et al. [28.136] exhibited
data reported to date show that hardness exhibits scale significant creep only at high temperatures (greater than
(size) effects. or equal to 0.25 times the melting point of silicon).
During loading, generation and propagation of dis- The mechanism of dislocation glide plasticity is be-
locations is responsible for plastic deformation. A strain lieved to dominate the indentation creep process on the
gradient plasticity theory has been developed for macroscale. To study the strain-rate sensitivity of sili-
micro/nanoscale deformations, based on randomly cre- con, data at two different (constant) rates of loading are
ated, statistically stored, and geometrically necessary presented in Fig. 28.59b. Note that a change in the load-
dislocations [28.134, 135]. Large strain gradients in- ing rate by a factor of about five results in a significant
herent to small indentations lead to accumulation of change in the load–displacement data. The viscoelastic
geometrically necessary dislocations located in a cer- effects observed here for silicon at ambient tempera-
tain subsurface volume for strain compatibility reasons, ture could arise from the size effects mentioned earlier.
which cause enhanced hardening. The large strain gra- Most likely, creep and strain rate experiments are being
dients in small indentations require these dislocations conducted on the hydrated films present on the silicon
to account for the large slope at the indented surface. surface in the ambient environment, and these films are
These are a function of strain gradient, whereas statisti- expected to be viscoelastic.
cally, stored dislocations are a function of strain. Based
on this theory, scale-dependent hardness is given as 28.4.3 Localized Surface Elasticity
and Viscoelasticity Mapping
d
H = H0 1 + , (28.23) The Young’s modulus of elasticity can be calculated
a
from the slope of the indentation curve during unload-
where H0 is the hardness in the absence of strain ing. However, these measurements provide a single-
gradient or macrohardness, d is the material-specific point measurement. By using the force modulation
characteristic length parameter, and a is the contact ra- technique, it is possible to obtain localized elasticity
dius. In addition to the role of strain gradient plasticity maps of soft and compliant materials of near-surface re-
theory, an increase in hardness with decreasing indenta- gions with nanoscale lateral resolution. This technique
tion depth can possibly be rationalized on the basis that, has been successfully used for polymeric magnetic
Nanotribology, Nanomechanics, and Materials Characterization 28.4 Indentation 839
Part D 28.4
a) Load (µN) b) Load (µN)
500 200
0 0
0 15 30 45 60 75 90 0 10 20 30 40 50 60
Displacement (nm) Displacement (nm)
Fig. 28.59 (a) Creep behavior and (b) strain-rate sensitivity of Si(100) (after [28.61])
tapes, which consist of magnetic and nonmagnetic ce- Surface height (nm) Elasticity (nm)
ramic particles in a polymeric matrix. Elasticity maps of 300 300
a tape can be used to identify the relative distribution of
hard magnetic and nonmagnetic ceramic particles on the
tape surface, which has an effect on friction and stiction 200 200
at the head–tape interface [28.15]. Figure 28.60 shows
surface height and elasticity maps on a polymeric mag-
netic tape [28.66]. The elasticity image reveals sharp 100 100
variations in surface elasticity due to the composite na-
ture of the film. As can be clearly seen, regions of high
elasticity do not always correspond to high or low to- 0 0
pography. Based on a Hertzian elastic-contact analysis, 0 100 200 300 0 100 200 300
nm nm
the static indentation depth of these samples during the
force modulation scan is estimated to be about 1 nm. We 0 nm 25 nm Compliant Stiff
conclude that the observed contrast is influenced most
strongly by material properties in the top few nanome- Fig. 28.60 Surface height and elasticity maps on a polymeric mag-
ters, independent of the composite structure beneath the netic tape (σ = 6.7 nm and P–V = 32 nm; σ and P–V refer to the
surface layer. standard deviation of surface height and the peak-to-valley dis-
By using phase-contrast microscopy, it is possible to tance, respectively). The gray scale on the elasticity map is arbitrary
obtain phase-contrast maps or the contrast in viscoelas- (after [28.66])
tic properties of near-surface regions with nanoscale
lateral resolution. This technique has been successfully images, the distribution of magnetic particles can be
used for polymeric films and magnetic tapes that consist clearly seen, and with better contrast than in the TR
of ceramic particles in a polymeric matrix [28.69–72]. surface height image. MP tape samples show a granu-
Figure 28.61 shows typical surface height, TR am- lar structure with elliptically shaped magnetic particle
plitude, and TR phase-angle images for a MP tape aggregates (50–100 nm in diameter). Studies by Scott
using TR mode II, described earlier. The TR ampli- and Bhushan [28.69], Bhushan and Qi [28.70], and Ka-
tude image provides contrast in lateral stiffness, and the sai et al. [28.71] have indicated that the phase shift
TR phase-angle image provides contrast in viscoelas- can be related to the energy dissipation through the
tic properties. In the TR amplitude and phase-angle viscoelastic deformation process between the tip and
840 Part D Bio-/Nanotribology and Bio-/Nanomechanics
0 0 0
20 nm 1V 180 °
0 1 µm 0 1 µm 0 1 µm
Fig. 28.61 Images of an MP tape obtained with TR mode II (constant deflection). TR mode II amplitude and phase-angle
images have the largest contrast among tapping, TR mode I, and TR mode II techniques (after [28.72])
the sample. Recent theoretical analysis has established phase-angle images is due to the in-plane (lateral) het-
a quantitative correlation between the lateral surface erogeneity of the surface. Based on the TR amplitude
properties (stiffness and viscoelasticity) of materials and phase-angle images, mapping of the lateral sur-
and the amplitude/phase-angle shift in TR measure- face properties (lateral stiffness and viscoelasticity) of
ments [28.86]. The contrast in the TR amplitude and materials can be obtained.
Part D 28.5
a) Adhesive force (nN)
solid-like phase of the Z-DOL (fully bonded) film re-
100
duces the adhesive force as compared with Si(100), Force 22 °C, RH 45–55%
because of the absence of mobile liquid. The schematic calibration plot
in Fig. 28.65b (bottom) shows the relative size and Friction
sources of the meniscus. It is well known that the native 75 force plot
oxide layer (SiO2 ) on the top of Si(100) wafer ex-
hibits hydrophilic properties, and some water molecules
can be adsorbed on this surface. The condensed water 50
will form a meniscus as the tip approaches the sam-
ple surface. The larger adhesive force in Z-15 is not
only caused by the Z-15 meniscus alone; the nonpo- 25
larized Z-15 liquid does not have good wettability and
strong bonding with Si(100). Consequently, in the am-
bient environment, condensed water molecules from the
environment will permeate through the liquid Z-15 lu- 0
Si(100) Z-15 Z-DOL
bricant film and compete with the lubricant molecules (fully bonded)
present on the substrate. The interaction of the liquid lu- b)
bricant with the substrate is weakened, and a boundary
Z-15
layer of the liquid lubricant forms puddles [28.51, 52].
Z-DOL O Si O
This dewetting allows water molecules to be adsorbed H2O
O
onto the Si(100) surface along with Z-15 molecules, and
both of them can form meniscus as the tip approaches
the surface. Thus the dewetting of liquid Z-15 film re-
Si(100) Z-15 Z-DOL
sults in a higher adhesive force and poorer lubrication
(fully bonded)
performance. In addition, the Z-15 film is soft compared
with the solid Si(100) surface, and penetration of the tip Fig. 28.62a,b Summary of the adhesive forces of Si(100) and Z-
into the film occurs when pushing the tip down. This 15 and Z-DOL (fully bonded) films measured by force calibration
results in a large area of the tip being wetted by the li- plots and plots of friction force versus normal load in ambient air
quid to form the meniscus at the tip–liquid (mixture of (a). (b) Schematic showing the effect of meniscus formed between
Z-15 and water) interface. It should also be noted that the AFM tip and the surface sample on the adhesive and friction
Z-15 has a higher viscosity compared with water; there- forces (after [28.54])
fore Z-15 film provides greater resistance to motion
and higher coefficient of friction. In the case of Z-DOL bonded); it reduced slightly only at very high velocity.
(fully bonded) film, both of the active groups of Z-DOL Figure 28.63 also indicates that the adhesive force of
molecules are mostly bonded onto the Si(100) substrate, Si(100) is increased when the velocity is > 10 μm/s.
thus the Z-DOL (fully bonded) film has low free sur- The adhesive force of Z-15 is reduced dramatically with
face energy and cannot be readily displaced by water a velocity increase up to 20 μm/s, after which it is re-
molecules or readily adsorb water molecules. Thus, the duced slightly, and the adhesive force of Z-DOL (fully
use of Z-DOL (fully bonded) can reduce the adhesive bonded) is also decreased at high velocity. In the tested
force. range of velocity, only the coefficient of friction of
To study the effect of velocity on friction and ad- Si(100) decreases with velocity, while the coefficients
hesion, the variation of friction force, adhesive force, of friction of Z-15 and Z-DOL (fully bonded) remain
and coefficient of friction of Si(100), Z-15, and Z-DOL almost constant. This implies that the friction mecha-
(fully bonded) as a function of velocity is summarized nisms of Z-15 and Z-DOL (fully bonded) do not change
in Fig. 28.63 [28.54]. The results indicates that, for sili- with velocity.
con wafer, the friction force decreases logarithmically The mechanisms of the effect of velocity on ad-
with increasing velocity. For Z-15, the friction force hesion and friction can be explained based on the
decreases with increasing velocity up to 10 μm/s, af- schematics shown in Fig. 28.63b (right) [28.54]. For
ter which it remains almost constant. The velocity has Si(100), tribochemical reaction plays a major role. Al-
a very small effect on the friction force of Z-DOL (fully though, at high velocity, the meniscus is broken and
842 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Part D 28.5
15
Si(100)
Si(100)
10
Z-15
Z-DOL (fully bonded)
5
0
Adhesive force (nN) Z-15
125
Z-15 From friction force plot
100
Z-DOL O Si O
O
75
Si(100)
50 Z-DOL (fully bonded)
0.4 µm/s 240 µm/s
25 Z-DOL (fully bonded) Increasing velocity
Part D 28.5
a) Friction force (nN) b)
25
70 nN, 2 µm/s, 22 °C H2O
20
15 Si(100)
Z-15
Thermally treated Si(100)
10
Si(100)
Z-15
5
Z-DOL (fully bonded)
0
Adhesive force (nN)
200 Z-15
From friction force plot
175 Z-15
150
O Si O Z-DOL
125 O
100
Si(100)
75
Z-DOL (fully bonded)
50
Z-DOL 0% 70%
25 (fully bonded) Increasing relative humidity
Thermally treated Si(100)
0
Fig. 28.64a,b Influence of relative humidity on the friction force,
Coefficient of friction
0.15 adhesive force, and coefficient of friction of Si(100) and Z-15 and
Z-DOL (fully bonded) films at 70 nN, 2 μm/s, and in 22 ◦ C air (a).
Thermally treated Si(100)
(b) Schematic showing the change of meniscus while increasing the
relative humidity. In this figure, the thermally treated Si(100) repre-
0.1 sents the Si(100) wafer baked at 150 ◦ C for 1 h in an oven (in order
to remove adsorbed water) just before it was placed in the 0% RH
chamber (after [28.54])
Si(100) Z-15
0.05 To study the effect of relative humidity on friction
and adhesion, the variation of friction force, adhesive
Z-DOL (fully bonded) force, and coefficient of friction of Si(100), Z-15, and
Z-DOL (fully bonded) as a function of relative humid-
0 ity is shown in Fig. 28.64 [28.54], showing that, for
0 20 40 60 80
Relative humidity (%)
Si(100) and Z-15 film, the friction force increases with
relative humidity up to 45% and then shows a slight
decrease with further increase in relative humidity. Z-
case of samples with mobile films such as condensed DOL (fully bonded) has a smaller friction force than
water and Z-15 films, alignment of liquid molecules Si(100) and Z-15 over the whole testing range, and its
(shear thinning) is responsible for the drop in fric- friction force shows a relative apparent increase when
tion force with increasing scanning velocity. This could the relative humidity is higher than 45%. For Si(100),
be another reason for the decrease in friction force Z-15, and Z-DOL (fully bonded), the adhesive forces
with velocity for the Si(100) and Z-15 film in this increase with relative humidity, and their coefficients of
study. friction increase with relative humidity up to 45%, af-
844 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Part D 28.5
15 Z-15
Si(100)
10
Z-15
Si(100)
5
Z-DOL (fully bonded)
0
Adhesive force (nN) Z-15
125
From force calibration plot
100
Z-15 O Si O
Z-DOL
O
75
50
Si(100) Z-DOL (fully bonded)
25 °C 125 °C
25 Increasing temperature
Z-DOL (fully bonded)
Fig. 28.65a,b The influence of temperature on the fric-
0 tion force, adhesive force, and coefficient of friction of
Coefficient of friction Si(100) and Z-15 and Z-DOL (fully bonded) films at
0.15 70 nN, 2 μm/s, and in RH 40–50% air (a). (b) Schematic
showing that, at high temperature, desorption of water de-
creases the adhesive forces. The reduced viscosity of Z-15
leads to the decrease of the coefficient of friction. High
0.1 Z-15 temperature facilitates the orientation of molecules in Z-
DOL (fully bonded) film, which results in lower coefficient
of friction (after [28.54])
Si(100)
0.05
adsorbed water, which is responsible for the smaller ef-
Z-DOL (fully bonded) fect with increasing relative humidity. However, if the
Si(100) wafer is thermally treated by baking at 150 ◦ C
0 for 1 h, a larger effect is observed.
0 50 100 150
The schematic in Fig. 28.64b (right) shows that,
Temperature (°C)
for Si(100), because of its high free surface energy, it
can adsorb more water molecules with increasing rel-
ter which they decrease with further increase of relative ative humidity [28.54]. As discussed earlier, for Z-15
humidity. It is also observed that the effect of humidity film in the humid environment, the water condensed
on Si(100) really depends on the history of the Si(100) from the humid environment competes with the lubri-
sample. As the surface of the Si(100) wafer readily ad- cant film present on the sample surface, and interaction
sorbs water in air, without any pretreatment the Si(100) of the liquid lubricant film with the silicon substrate is
used in our study almost reaches its saturated stage of weakened and a boundary layer of the liquid lubricant
Nanotribology, Nanomechanics, and Materials Characterization 28.5 Boundary Lubrication 845
forms puddles. This dewetting allows water molecules then approach higher, stable values. This is believed
Part D 28.5
to be adsorbed on the Si(100) substrate mixed with Z-15 to be caused by the attachment of Z-15 molecules to
molecules [28.51,52]. Obviously, more water molecules the tip. After several scans, the molecular interaction
can be adsorbed on the Z-15 surface with increasing rel- reaches an equilibrium, and after that the friction force
ative humidity. The greater amount of adsorbed water and coefficient of friction remain constant. In the case
molecules in the case of Si(100), along with the lu- of Z-DOL (fully bonded) film, the friction force and
bricant molecules in the case of the Z-15 film, form coefficient of friction start out low and remain low dur-
a larger meniscus, which leads to an increase of fric-
tion force, adhesive force, and coefficient of friction for
Si(100) and Z-15 with humidity, although at very high Friction force (nN) Molecularly thick Z-15 film
humidity of 70% large quantities of adsorbed water can Lack of meniscus
form a continuous water layer that separate the tip and reformation decreases
friction with
sample surface and acts as a kind of lubricant, which increasing velocity
causes a decrease in the friction force and coefficient of
friction. For Z-DOL (fully bonded) film, because of its
hydrophobic surface properties, water molecules can be
adsorbed at humidity above 45%, causing an increase in Reaches equilibrium
the adhesive force and friction force.
To study the effect of temperature on friction and
adhesion, the variation of friction force, adhesive force,
and coefficient of friction of Si(100), Z-15, and Z-
Log velocity
DOL (fully bonded) with temperature is summarized Friction force (nN)
in Fig. 28.65 [28.54]. The results shows that increasing
temperature causes a decrease of friction force, adhesive
force, and coefficient of friction for Si(100), Z-15, and
Z-DOL (fully bonded). The schematic in Fig. 28.65b
(right) indicates that, at high temperature, desorption of Thick H2O
water leads to decrease of the friction force, adhesive film serves
forces, and coefficient of friction for all of the samples. as a lubricant
For the Z-15 film, the reduction of viscosity at high tem- Meniscus formation
perature also contributes to the decrease of friction force increases friction with
and coefficient of friction. In the case of Z-DOL (fully increasing RH
bonded) film, molecules are easily oriented at high tem-
perature, which may be partly responsible for the low Relative humidity
friction force and coefficient of friction. Friction force (nN)
To summarize, the influence of velocity, relative hu-
midity, and temperature on the friction force of mobile Desorption of water and
decrease of viscosity decrease
Z-15 film is presented in Fig. 28.66 [28.54]. The chang- friction with increase of
ing trends are also addressed in this figure. temperature
To study the durability of lubricant films at the
nanoscale, the friction of Si(100), Z-15, and Z-DOL
(fully bonded) as a function of the number of scan-
ning cycles is shown in Fig. 28.67 [28.54]. As observed
earlier, the friction force for Z-15 is higher than that
for Si(100), with the lowest values for Z-DOL (fully
bonded). During cycling, the friction force and coef-
Temperature
ficient of friction for Si(100) show a slight decrease
during the first few cycles, then remain constant. This is Fig. 28.66 Schematic showing the change of friction force
related to the removal of the native oxide. In the case of of molecularly thick Z-15 films with log velocity, relative
the Z-15 film, the friction force and coefficient of fric- humidity, and temperature. The changing trends are also
tion show an increase during the first few cycles and addressed in this figure (after [28.54])
846 Part D Bio-/Nanotribology and Bio-/Nanomechanics
Part D 28.5
a) Hexadecane thiol CH3 CH3
(HDT)
Alkyl
–(CH2)n –
S S α1 α2
Au(111) Substrate
40
20
0
Au HDT BPT BPTC 0 µm 6 0 µm 6
Coefficient of friction 0 10 nm Stiff Soft
0.08
b) Surface height Friction force
0.06
0.04
0.02
0
Au HDT BPT BPTC
0 µm 10 0 µm 10
Materials
Fig. 28.68 (a) Schematics of structures of hexadecane 0 10 nm 0 4.5 nN
thiol and biphenylthiol SAMs on Au(111) substrates, and Fig. 28.70 (a) AFM gray-scale surface height and stiffness
(b) adhesive force and coefficient of friction of Au(111) images, and (b) AFM gray-scale surface height and friction
substrate and various SAMs force images of micropatterned BDCS (after [28.147])
848 Part D Bio-/Nanotribology and Bio-/Nanomechanics
brought into contact with the tip and then pulled away
Part D 28.6
Adhesive force Adhesive force
at a velocity of 400 nm/s. The zero tip–sample sep- 3.5 nm uniform Z-DOL/ Si(100) 2 –10 nm nonuniform Z-DOL/ Si(100)
aration is defined to be the position where the force 30 30
µm µm
Force (nN)
200
Tip
150
H
100 Conditioner
h
Hair surface
50 0 0
µm 30 0 µm 30
Virgin hair
0
60 20 nN 60 20 nN
28.6 Conclusion
For most solid–solid interfaces of technological rele- such contact. However, asperities come in all shapes
vance, contact occurs at multiple asperities. A sharp and sizes. The effect of the radius of a single asper-
AFM/FFM tip sliding on a surface simulates just one ity (tip) on the friction/adhesion performance can be
850 Part D Bio-/Nanotribology and Bio-/Nanomechanics
studied using tips of different radii. AFM/FFM is used corresponds to the hardness of the softer of the two
Part D 28.6
ing sliding, and surface properties play an important provide insights into the failure mechanisms of mater-
Part D 28
role in the adhesion, friction, and durability of these ials. Coefficients of friction, wear rates, and mechanical
films. Sliding velocity, relative humidity, and tempera- properties such as hardness have been found to be
ture affect adhesion and friction. For SAMs, the friction different on the nanoscale than on the macroscale;
mechanism is explained by a so-called molecular spring generally, coefficients of friction and wear rates on
model. Films with high-compliance long carbon chains micro- and nanoscales are smaller, whereas hardness is
exhibit low friction and wear. Also perfluoroalkylsilane greater. Therefore, micro/nanotribological studies may
SAMs on Si appear to be more hydrophobic with lower help to define the regimes for ultralow friction and near-
adhesion than alkylsilane SAMs on Si. zero wear. These studies also provide insight into the
Investigations of adhesion, friction, wear, scratch- atomic origins of adhesion, friction, wear, and lubrica-
ing, and indentation on the nanoscale using AFM can tion mechanisms.
References
28.1 I.L. Singer, H.M. Pollock (Eds.): Fundamentals of 28.17 B. Bhushan: Wear and mechanical characterisation
Friction: Macroscopic and Microscopic Processes, on micro- to picoscales using AFM, Int. Mater. Rev.
NATO ASI Ser. E, Vol. 220 (Kluwer, Dordrecht 1992) 44, 105–117 (1999)
28.2 H.J. Güntherodt, D. Anselmetti: Forces in Scanning 28.18 B. Bhushan: Adhesion and stiction: Mechanisms,
Probe Methods, NATO ASI Ser. E, Vol. 286 (Kluwer, measurement techniques, and methods for reduc-
Dordrecht 1995) tion (invited), J. Vac. Sci. Technol. B 21, 2262–2296
28.3 B.N.J. Persson, E. Tosatti: Physics of Sliding Friction, (2003)
NATO ASI Ser. E, Vol. 311 (Kluwer, Dordrecht 1996) 28.19 B. Bhushan: Nanotribology, nanomechanics and
28.4 B. Bhushan: Micro/Nanotribology and its Applica- nanomaterials characterization, Philos. Trans. R.
tions, NATO ASI Ser. E, Vol. 330 (Kluwer, Dordrecht Soc. A 366, 1351–1381 (2008)
1997) 28.20 B. Bhushan, A.V. Kulkarni, V.N. Koinkar, M. Boehm,
28.5 B. Bhushan: Handbook of Micro/Nanotribology, L. Odoni, C. Martelet, M. Belin: Microtribo-
2nd edn. (CRC, Boca Raton 1999) logical characterization of self-assembled and
28.6 B. Bhushan: Principles and Applications of Tribol- Langmuir–Blodgett monolayers by atomic and fric-
ogy (Wiley, New York 1999) tion force microscopy, Langmuir 11, 3189–3198
28.7 B. Bhushan: Nanoscale tribophysics and tribome- (1995)
chanics, Wear 225–229, 465–492 (1999) 28.21 B. Bhushan, H. Fuchs, S. Hosaka (Eds.): Applied
28.8 B. Bhushan: Modern Tribology Handbook, Vol. 1: Scanning Probe Methods (Springer, Berlin Heidel-
Principles of Tribology (CRC, Boca Raton 2001) berg 2004)
28.9 B. Bhushan: Fundamentals of Tribology and Bridg- 28.22 B. Bhushan, H. Fuchs, S. Kawata (Eds.): Applied
ing the Gap Between the Macro- and Micro/ Scanning Probe Methods V – Scanning Probe Mi-
Nanoscales, NATO Sci. Ser. II, Vol. 10 (Kluwer, Dor- croscopy Techniques (Springer, Berlin Heidelberg
drecht 2001) 2007)
28.10 B. Bhushan: Nano- to microscale wear and me- 28.23 B. Bhushan, H. Fuchs, M. Tomitori (Eds.): Applied
chanical characterization studies using scanning Scanning Probe Methods VIII – Scanning Probe Mi-
probe microscopy, Wear 251, 1105–1123 (2001) croscopy Techniques (Springer, Berlin Heidelberg
28.11 B. Bhushan: Introduction to Tribology (Wiley, New 2008)
York 2002) 28.24 B. Bhushan, H. Fuchs, M. Tomitori (Eds.): Applied
28.12 B. Bhushan: Nanotribology and nanomechanics, Scanning Probe Methods IX – Characterization
Wear 259, 1507–1531 (2005) (Springer, Berlin Heidelberg 2008)
28.13 B. Bhushan: Nanotribology and Nanomechanics – 28.25 B. Bhushan, H. Fuchs, M. Tomitori (Eds.): Applied
An Introduction, 2nd edn. (Springer, Berlin Heidel- Scanning Probe Methods X – Biomimetics and In-
berg 2008) dustrial Applications (Springer, Berlin Heidelberg
28.14 B. Bhushan, J.N. Israelachvili, U. Landman: Nano- 2008)
tribology: Friction, wear and lubrication at the 28.26 B. Bhushan, H. Fuchs (Eds.): Applied Scanning
atomic scale, Nature 374, 607–616 (1995) Probe Methods II – Scanning Probe Microscopy
28.15 B. Bhushan: Tribology and Mechanics of Magnetic Techniques (Springer, Berlin Heidelberg 2006)
Storage Devices, 2nd edn. (Springer, New York 1996) 28.27 B. Bhushan, H. Fuchs (Eds.): Applied Scanning
28.16 B. Bhushan: Tribology Issues and Opportunities in Probe Methods III – Characterization (Springer,
MEMS (Kluwer, Dordrecht 1998) Berlin Heidelberg 2006)
852 Part D Bio-/Nanotribology and Bio-/Nanomechanics
28.28 B. Bhushan, H. Fuchs (Eds.): Applied Scan- 28.45 V. Scherer, W. Arnold, B. Bhushan: Lateral force mi-
Part D 28
ning Probe Methods IV – Industrial Applications croscopy using acoustic friction force microscopy,
(Springer, Berlin Heidelberg 2006) Surf. Interface Anal. 27, 578–587 (1999)
28.29 B. Bhushan, H. Fuchs (Eds.): Applied Scanning 28.46 M. Reinstädtler, U. Rabe, V. Scherer, U. Hart-
Probe Methods VII – Biomimetics and Industrial mann, A. Goldade, B. Bhushan, W. Arnold: On the
Applications (Springer, Berlin Heidelberg 2007) nanoscale measurement of friction using atomic-
28.30 B. Bhushan, H. Fuchs (Eds.): Applied Scanning force microscope cantilever torsional resonances,
Probe Methods XI – Scanning Probe Microscopy Appl. Phys. Lett. 82, 2604–2606 (2003)
Techniques (Springer, Berlin Heidelberg 2009) 28.47 M. Reinstädtler, U. Rabe, A. Goldade, B. Bhushan,
28.31 B. Bhushan, H. Fuchs (Eds.): Applied Scanning W. Arnold: Investigating ultra-thin lubricant lay-
Probe Methods XII – Characterization (Springer, ers using resonant friction force microscopy, Tribol.
Berlin Heidelberg 2009) Int. 38, 533–541 (2005)
28.32 B. Bhushan, H. Fuchs (Eds.): Applied Scanning 28.48 M. Reinstädtler, T. Kasai, U. Rabe, B. Bhushan,
Probe Methods XIII – Biomimetics and Industrial W. Arnold: Imaging and measurement of elastic-
Applications (Springer, Berlin Heidelberg 2009) ity and friction using the TR mode, J. Phys. D 38,
28.33 B. Bhushan, S. Kawata (Eds.): Applied Scanning R269–R282 (2005)
Probe Methods VI – Characterization (Springer, 28.49 B. Bhushan, T. Kasai: A surface topography in-
Berlin Heidelberg 2007) dependent friction measurement technique using
28.34 G. Binnig, C.F. Quate, C. Gerber: Atomic force mi- torsional resonance mode in an AFM, Nanotech-
croscopy, Phys. Rev. Lett. 56, 930–933 (1986) nology 15, 923–935 (2004)
28.35 G. Binnig, C. Gerber, E. Stoll, T.R. Albrecht, 28.50 N.S. Tambe, B. Bhushan: A new atomic force mi-
C.F. Quate: Atomic resolution with atomic force croscopy based technique for studying nanoscale
microscope, Europhys. Lett. 3, 1281–1286 (1987) friction at high sliding velocities, J. Phys. D: Appl.
28.36 C.M. Mate, G.M. McClelland, R. Erlandsson, S. Chi- Phys. 38, 764–773 (2005)
ang: Atomic-scale friction of a tungsten tip on 28.51 V.N. Koinkar, B. Bhushan: Micro/nanoscale studies
a graphite surface, Phys. Rev. Lett. 59, 1942–1945 of boundary layers of liquid lubricants formagnetic
(1987) disks, J. Appl. Phys. 79, 8071–8075 (1996)
28.37 B. Bhushan, J. Ruan: Atomic-scale friction mea- 28.52 V.N. Koinkar, B. Bhushan: Microtribological stud-
surements using friction force microscopy: Part II – ies of unlubricated and lubricated surfaces using
Application to magnetic media, ASME J. Tribol. 116, atomic force/friction force microscopy, J. Vac. Sci.
389–396 (1994) Technol. 14, 2378–2391 (1996)
28.38 J. Ruan, B. Bhushan: Atomic-scale friction mea- 28.53 B. Bhushan, H. Liu: Nanotribological properties
surements using friction force microscopy: Part and mechanisms of alkylthiol and biphenyl thiol
I – General principles and new measure- self-assembled monolayers studied by AFM, Phys.
ment techniques, ASME J. Tribol. 116, 378–388 Rev. B 63, 245412-1–245412-11 (2001)
(1994) 28.54 H. Liu, B. Bhushan: Nanotribological character-
28.39 J. Ruan, B. Bhushan: Atomic-scale and microscale ization of molecularly thick lubricant films for
friction of graphite and diamond using friction applications to MEMS/NEMS by AFM, Ultrami-
force microscopy, J. Appl. Phys. 76, 5022–5035 croscopy 97, 321–340 (2003)
(1994) 28.55 B. Bhushan, T. Kasai, G. Kulik, L. Barbieri, P. Hoff-
28.40 J. Ruan, B. Bhushan: Frictional behavior of highly mann: AFM study of perfluorosilane and alkylsilane
oriented pyrolytic graphite, J. Appl. Phys. 76, 8117– self-assembled monolayers for antistiction in
8120 (1994) MEMS/NEMS, Ultramicroscopy 105, 176–188 (2005)
28.41 B. Bhushan, V.N. Koinkar, J. Ruan: Microtribology 28.56 B. Bhushan, V.N. Koinkar: Tribological studies of
of magnetic media, Proc. Inst. Mech. Eng. J: J. Eng. silicon for magnetic recording applications, J. Appl.
Tribol. 208, 17–29 (1994) Phys. 75, 5741–5746 (1994)
28.42 B. Bhushan, A.V. Kulkarni: Effect of normal load on 28.57 V.N. Koinkar, B. Bhushan: Microtribological studies
microscale friction measurements, Thin Solid Films of Al2 O3 -TiC, polycrystalline and single-crystal Mn-
278, 49–56 (1996) Zn ferrite and SiC head slider materials, Wear 202,
28.43 B. Bhushan, S. Sundararajan: Micro/nanoscale fric- 110–122 (1996)
tion and wear mechanisms of thin films using 28.58 V.N. Koinkar, B. Bhushan: Microtribological prop-
atomic force and friction force microscopy, Acta erties of hard amorphous carbon protective
Mater. 46, 3793–3804 (1998) coatings for thin film magnetic disks and heads,
28.44 V. Scherer, W. Arnold, B. Bhushan: Active friction J. Eng. Tribol. 211, 365–372 (1997)
control using ultrasonic vibration. In: Tribology Is- 28.59 S. Sundararajan, B. Bhushan: Development of
sues and Opportunities in MEMS, ed. by B. Bhushan a continuous microscratch technique in an atomic
(Kluwer, Dordrecht 1998) pp. 463–469 force microscope and its application to study
Nanotribology, Nanomechanics, and Materials Characterization References 853
scratch resistance of ultra-thin hard amorphous 28.74 M.S. Bobji, B. Bhushan: In-situ microscopic surface
Part D 28
carbon coatings, J. Mater. Res. 16, 75–84 (2001) characterization studies of polymeric thin films
28.60 J. Ruan, B. Bhushan: Nanoindentation studies during tensile deformation using atomic force mi-
of fullerene films using atomic force microscopy, croscopy, J. Mater. Res. 16, 844–855 (2001)
J. Mater. Res. 8, 3019–3022 (1993) 28.75 N. Tambe, B. Bhushan: In situ study of nano-
28.61 B. Bhushan, A.V. Kulkarni, W. Bonin, J.T. Wyrobek: cracking of multilayered magnetic tapes under
Nano/picoindentation measurement using a ca- monotonic and fatigue loading using an AFM, Ul-
pacitance transducer system in atomic force tramicroscopy 100, 359–373 (2004)
microscopy, Philos. Mag. 74, 1117–1128 (1996) 28.76 M. Palacio, B. Bhushan: Normal and lateral force
28.62 B. Bhushan, V.N. Koinkar: Nanoindentation hard- calibration techniques for AFM cantilevers, (sub-
ness measurements using atomic force microscopy, mitted for publication) UPDATE WITH JOURNAL
Appl. Phys. Lett. 64, 1653–1655 (1994) 28.77 B. Bhushan, T. Kasai, C.V. Nguyen, M. Meyyap-
28.63 B. Bhushan, X. Li: Nanomechanical characterisa- pan: Multiwalled carbon nanotube AFM probes for
tion of solid surfaces and thin films, Int. Mater. surface characterization of micro/nanostructures,
Rev. 48, 125–164 (2003) Microsyst. Technol. 10, 633–639 (2004)
28.64 P. Maivald, H.J. Butt, S.A.C. Gould, C.B. Prater, 28.78 B. Bhushan: Micro/nanotribology and Its applica-
B. Drake, J.A. Gurley, V.B. Elings, P.K. Hansma: Us- tions to magnetic storage devices and MEMS, Tribol.
ing force modulation to image surface elasticities Int. 28, 85–95 (1995)
with the atomic force microscope, Nanotechnology 28.79 D. DeVecchio, B. Bhushan: Use of a nanoscale
2, 103–106 (1991) Kelvin probe for detecting wear precursors, Rev.
28.65 B. Anczykowski, D. Kruger, K.L. Babcock, H. Fuchs: Sci. Instrum. 69, 3618–3624 (1998)
Basic properties of dynamic force microscopy with 28.80 B. Bhushan, A.V. Goldade: Measurements and
the scanning force microscope in experiment and analysis of surface potential change during wear
simulation, Ultramicroscopy 66, 251–259 (1996) of single crystal silicon (100) at ultralow loads us-
28.66 D. DeVecchio, B. Bhushan: Localized surface ing Kelvin probe microscopy, Appl. Surf. Sci. 157,
elasticity measurements using an atomic force 373–381 (2000)
microscope, Rev. Sci. Instrum. 68, 4498–4505 28.81 B. Bhushan, A.V. Goldade: Kelvin probe microscopy
(1997) measurements of surface potential change under
28.67 V. Scherer, B. Bhushan, U. Rabe, W. Arnold: Lo- wear at low loads, Wear 244, 104–117 (2000)
cal elasticity and lubrication measurements using 28.82 H.U. Krotil, T. Stifter, H. Waschipky, K. Weishaupt,
atomic force and friction force microscopy at ultra- S. Hild, O. Marti: Pulse force mode: A new method
sonic frequencies, IEEE Trans. Magn. 33, 4077–4079 for the investigation of surface properties, Surf.
(1997) Interface Anal. 27, 336–340 (1999)
28.68 S. Amelio, A.V. Goldade, U. Rabe, V. Scherer, 28.83 U. Rabe, K. Janser, W. Arnold: Vibrations of free
B. Bhushan, W. Arnold: Measurements of elas- and surface-coupled atomic force microscope can-
tic properties of ultra-thin diamond-like carbon tilevers: Theory and experiment, Rev. Sci. Instrum.
coatings using atomic force acoustic microscopy, 67, 3281–3293 (1996)
Thin Solid Films 392, 75–84 (2001) 28.84 J. Tamayo, R. Garcia: Deformation, contact time,
28.69 W.W. Scott, B. Bhushan: Use of phase imaging in and phase contrast in tapping mode scanning force
atomic force microscopy for measurement of vis- microscopy, Langmuir 12, 4430–4435 (1996)
coelastic contrast in polymer nanocomposites and 28.85 R. Garcia, J. Tamayo, M. Calleja, F. Garcia: Phase
molecularly-thick lubricant films, Ultramicroscopy contrast in tapping-mode scanning force mi-
97, 151–169 (2003) croscopy, Appl. Phys. A 66, 309–312 (1998)
28.70 B. Bhushan, J. Qi: Phase contrast imaging of 28.86 Y. Song, B. Bhushan: Quantitative extraction of in-
nanocomposites and molecularly thick lubricant plane surface properties using torsional resonance
films in magnetic media, Nanotechnology 14, 886– mode in atomic force microscopy, J. Appl. Phys. 87,
895 (2003) 083533 (2005)
28.71 T. Kasai, B. Bhushan, L. Huang, C. Su: Topography 28.87 B. Bhushan, J. Ruan, B.K. Gupta: A scanning tun-
and phase imaging using the torsional resonance nelling microscopy study of fullerene films, J. Phys.
mode, Nanotechnology 15, 731–742 (2004) D: Appl. Phys. 26, 1319–1322 (1993)
28.72 N. Chen, B. Bhushan: Morphological, nanome- 28.88 G.A. Tomlinson: A molecular theory of friction, Phi-
chanical and cellular structural characterization of los. Mag. Ser. 7, 905–939 (1929)
human hair and conditioner distribution using tor- 28.89 D. Tomanek, W. Zhong, H. Thomas: Calculation of
sional resonance mode in an AFM, J. Microsc. 220, an atomically modulated friction force in atomic
96–112 (2005) forcemicroscopy, Europhys. Lett. 15, 887–892 (1991)
28.73 M.S. Bobji, B. Bhushan: Atomic force microscopic 28.90 E. Meyer, R. Overney, R. Lüthi, D. Brodbeck,
study of the micro-cracking of magnetic thin films L. Howald, J. Frommer, H.J. Güntherodt, O. Wolter,
under tension, Scr. Mater. 44, 37–42 (2001) M. Fujihira, T. Takano, Y. Gotoh: Friction force mi-
854 Part D Bio-/Nanotribology and Bio-/Nanomechanics
croscopy of mixed Langmuir–Blodgett films, Thin 28.105 J.S. Helman, W. Baltensperger, J.A. Holyst: Simple-
Part D 28
Solid Films 220, 132–137 (1992) model for dry friction, Phys. Rev. B 49, 3831–3838
28.91 C.D. Frisbie, L.F. Rozsnyai, A. Noy, M.S. Wrighton, (1994)
C.M. Lieber: Functional group imaging by chemical 28.106 C. Fusco, A. Fasolino: Velocity dependence of
force microscopy, Science 265, 2071–2074 (1994) atomic-scale friction: A comparative study of the
28.92 V.N. Koinkar, B. Bhushan: Effect of scan size and one- and two-dimensional Tomlinson model,
surface roughness on microscale friction measure- Phys. Rev. B 71, 045413 (2005)
ments, J. Appl. Phys. 81, 2472–2479 (1997) 28.107 N.S. Tambe, B. Bhushan: Nanoscale friction map-
28.93 S. Sundararajan, B. Bhushan: Topography-induced ping, Appl. Phys. Lett. 86, 193102-1–193102-3 (2005)
contributions to friction forces measured using 28.108 N.S. Tambe, B. Bhushan: Nanoscale friction and
an atomic force/friction force microscope, J. Appl. wear maps, Philos. Trans. R. Soc. A 366, 1405–1424
Phys. 88, 4825–4831 (2000) (2008)
28.94 B. Bhushan, G.S. Blackman: Atomic force mi- 28.109 N.S. Tambe, B. Bhushan: Nanoscale frictioninduced
croscopy of magnetic rigid disks and sliders and phase transformation of diamond-like carbon, Scr.
its applications to tribology, ASME J. Tribol. 113, Mater. 52, 751–755 (2005)
452–458 (1991) 28.110 S.C. Lim, M.F. Ashby: Wear mechanism maps, Acta
28.95 K. Yamanaka, E. Tomita: Lateral force modulation Metall. 35, 1–24 (1987)
atomic force microscope for selective imaging of 28.111 S.C. Lim, M.F. Ashby, J.H. Brunton: Wear-rate tran-
friction forces, Jpn. J. Appl. Phys. 34, 2879–2882 sitions and their relationship to wear mechanisms,
(1995) Acta Metall. 35, 1343–1348 (1987)
28.96 Z. Tao, B. Bhushan: A new technique for study- 28.112 N.S. Tambe, B. Bhushan: Nanowear mapping:
ing nanoscale friction at sliding velocities up to A novel atomic force microscopy based approach
200 mm/s using atomic force microscope, Rev. Sci. for studying nanoscale wear at high sliding veloc-
Instrum. 71, 103705 (2006) ities, Tribol. Lett. 20, 83–90 (2005)
28.97 O. Marti, H.-U. Krotil: Dynamic friction measure- 28.113 B. Bhushan, C. Dandavate: Thin-film friction and
ment with the scanning force microscope. In: adhesion studies using atomic force microscop,
Fundamentals of Tribology and Bridging the Gap J. Appl. Phys. 87, 1201–1210 (2000)
Between the Macro- and Micro/Nanoscales, NATO 28.114 T. Stifter, O. Marti, B. Bhushan: Theoretical investi-
Sci. Ser. II, Vol. 10, ed. by B. Bhushan (Kluwer, gation of the distance dependence of capillary and
Dordrecht 2001) pp. 121–135 van der Waals forces in scanning probe microscopy,
28.98 N.S. Tambe, B. Bhushan: Scale dependence of Phys. Rev. B 62, 13667–13673 (2000)
micro/nano-friction and adhesion of MEMS/NEMS 28.115 U.D. Schwarz, O. Zwörner, P. Köster, R. Wiesen-
materials, coatings and lubricants, Nanotechnol- danger: Friction force spectroscopy in the
ogy 15, 1561–1570 (2004) lowload regime with well-defined tips. In: Mi-
28.99 N.S. Tambe, B. Bhushan: Friction model for the cro/Nanotribology and Its Applications, ed. by
velocity dependence of nanoscale friction, Nano- B. Bhushan (Kluwer, Dordrecht 1997) pp. 233–
technology 16, 2309–2324 (2005) 238
28.100 N.S. Tambe, B. Bhushan: Durability studies of 28.116 B. Bhushan, H. Liu, S.M. Hsu: Adhesion and friction
micro/nanoelectromechanical system materials, studies of silicon and hydrophobic and low fric-
coatings, and lubricants at high sliding velocities tion films and investigation of scale effects, ASME
(up to 10 mm/s) using a modified atomic force mi- J. Tribol. 126, 583–590 (2004)
croscope, J. Vac. Sci. Technol. A 23, 830–835 (2005) 28.117 B. Bhushan, B.K. Gupta: Handbook of Tribol-
28.101 N.S. Tambe, B. Bhushan: Identifying materials with ogy: Materials, Coatings and Surface Treatments
low friction and adhesion for nanotechnology ap- (McGraw-Hill, New York 1991), reprinted by Krieger,
plications, Appl. Phys. Lett 86, 061906-1–061906-3 Malabar (1997)
(2005) 28.118 B. Bhushan, S. Venkatesan: Mechanical and tri-
28.102 Z. Tao, B. Bhushan: Velocity dependence and rest bological properties of silicon for micromechanical
time effect in nanoscale friction of ultrathin films applications: A review, Adv. Info. Storage Syst. 5,
at high sliding velocities, J. Vac. Sci. Technol. A 25, 211–239 (1993)
1267–1274 (2007) 28.119 Anonymous: Properties of Silicon, EMIS Data Re-
28.103 O. Zwörner, H. Hölscher, U.D. Schwarz, R. Wiesen- views Series No. 4. INSPEC, Institution of Electrical
danger: The velocity dependence of frictional Engineers, London. See also Anonymous, MEMS
forces in point-contact friction, Appl. Phys. A 66, Materials Database,
S263–S267 (1998) http://www.memsnet.org/material/ (2002)
28.104 E. Gnecco, R. Bennewitz, T. Gyalog, C. Loppacher, 28.120 J.E. Field (Ed.): The Properties of Natural and Syn-
M. Bammerlin, E. Meyer, H.J. Güntherodt: Velocity thetic Diamond (Academic, London 1992)
dependence of atomic friction, Phys. Rev. Lett. 84, 28.121 B. Bhushan: Chemical, mechanical and tribological
1172–1175 (2000) characterization of ultra-thin and hard amorphous
Nanotribology, Nanomechanics, and Materials Characterization References 855
carbon coatings as thin as 3.5 nm: Recent develop- 28.138 Z. Tao, B. Bhushan: Bonding, degradation, and
Part D 28
ments, Diam. Relat. Mater. 8, 1985–2015 (1999) environmental effects on novel perfluoropolyether
28.122 Anonymous: The Industrial Graphite Engineering lubricants, Wear 259, 1352–1361 (2005)
Handbook (National Carbon Company, New York 28.139 B. Bhushan, M. Cichomski, Z. Tao, N.T. Tran,
1959) T. Ethen, C. Merton, R.E. Jewett: Nanotribological
28.123 M. Nosonovsky, B. Bhushan: Scale effects in dry characterization and lubricant degradation studies
friction during multiple-asperity contact, ASME J. of metal-film magnetic tapes using novel lubri-
Tribol. 127, 37–46 (2005) cants, ASME J. Tribol. 129, 621–627 (2007)
28.124 H. Liu, B. Bhushan: Adhesion and fric- 28.140 M. Palacio, B. Bhushan: Surface potential and
tion studies of microelectromechanical sys- resistance measurements for detecting wear of
tems/nanoelectromechanical systems materials chemically-bonded and unbonded molecularly-
using a novel microtriboapparatus, J. Vac. Sci. thick perfluoropolyether lubricant films using
Technol. A 21, 1528–1538 (2003) atomic force microscopy, J. Colloid Interface Sci.
28.125 B. Bhushan, M. Nosonovsky: Comprehensive model 315, 261–269 (2007)
for scale effects in friction due to adhesion and 28.141 M. Palacio, B. Bhushan: Wear detection of can-
two- and three-body deformation (plowing), Acta didate MEMS/NEMS lubricant films using atomic
Mater. 52, 2461–2474 (2004) force microscopy-based surface potential mea-
28.126 B. Bhushan, M. Nosonovsky: Scale effects in dry surements, Scr. Mater. 57, 821–824 (2007)
and wet friction, wear, and interface temperature, 28.142 B. Bhushan, M. Palacio, B. Kinzig: AFM-based
Nanotechnology 15, 749–761 (2004) nanotribological and electrical characterization of
28.127 B. Bhushan, M. Nosonovsky: Scale effects in friction ultrathin wear-resistant ionic liquid films, J. Col-
using strain gradient plasticity and dislocationas- loid Interface Sci. 317, 275–287 (2008)
sisted sliding (microslip), Acta Mater. 51, 4331–4345 28.143 M. Palacio, B. Bhushan: Ultrathin wear-resistant
(2003) ionic liquid films for novel MEMS/NEMS applica-
28.128 V.N. Koinkar, B. Bhushan: Scanning and trans- tions, Adv. Mater. 20, 1194–1198 (2008)
mission electron microscopies of single-crystal 28.144 M. Palacio, B. Bhushan: Molecularly thick dica-
silicon microworn/machined using atomic force tionic ionic liquid films for nanolubrication, J. Vac.
microscopy, J. Mater. Res. 12, 3219–3224 (1997) Sci. Technol. A 27(4), 986–995 (2009)
28.129 X. Zhao, B. Bhushan: Material removal mecha- 28.145 B. Bhushan, D. Hansford, K.K. Lee: Surface modifi-
nism of single-crystal silicon on nanoscale and at cation of silicon and polymethylsiloxane surfaces
ultralow loads, Wear 223, 66–78 (1998) with vapor-phase-deposited ultrathin fluorosi-
28.130 B. Bhushan, P.S. Mokashi, T. Ma: A new technique lane films for biomedical nanodevices, J. Vac. Sci.
to measure Poisson’s ratio of ultrathin polymeric Technol. A 24, 1197–1202 (2006)
films using atomic force microscopy, Rev. Sci. In- 28.146 H. Liu, B. Bhushan, W. Eck, V. Stadtler: Investiga-
strum. 74, 1043–1047 (2003) tion of the adhesion, friction, and wear properties
28.131 A.V. Kulkarni, B. Bhushan: Nanoscale mechani- of biphenyl thiol self-assembled monolayers by
cal property measurements using modified atomic atomic force microscopy, J. Vac. Sci. Technol. A 19,
forcemicroscopy, Thin Solid Films 290/291, 206–210 1234–1240 (2001)
(1996) 28.147 H. Liu, B. Bhushan: Investigation of nanotribolog-
28.132 A.V. Kulkarni, B. Bhushan: Nano/picoindentation ical properties of self-assembled monolayers with
measurements on single-crystal aluminum using alkyl and biphenyl spacer chains, Ultramicroscopy
modified atomic force microscopy, Mater. Lett. 29, 91, 185–202 (2002)
221–227 (1996) 28.148 T. Kasai, B. Bhushan, G. Kulik, L. Barbieri,
28.133 A.V. Kulkarni, B. Bhushan: Nanoindentation mea- P. Hoffman: Micro-/nanotribological study of per-
surement of amorphous carbon coatings, J. Mater. fluorosilane SAMs for antistiction and low wear,
Res. 12, 2707–2714 (1997) J. Vac. Sci. Technol. B 23, 995–1003 (2005)
28.134 N.A. Fleck, G.M. Muller, M.F. Ashby, J.W. Hutchin- 28.149 K.K. Lee, B. Bhushan, D. Hansford: Nanotribologi-
son: Strain gradient plasticity: Theory and experi- cal characterization of fluoropolymer thin films for
ment, Acta Metall. Mater. 42, 475–487 (1994) biomedical micro-/nanoelectromechemical sys-
28.135 W.D. Nix, H. Gao: Indentation size effects in tems applications, J. Vac. Sci. Technol. A 23,
crystalline materials: A law for strain gradient plas- 804–810 (2005)
ticity, J. Mech. Phys. Solids 46, 411–425 (1998) 28.150 N.S. Tambe, B. Bhushan: Nanotribological charac-
28.136 W.B. Li, J.L. Henshall, R.M. Hooper, K.E. Easterling: terization of self assembled monolayers deposited
The mechanism of indentation creep, Acta Metall. on silicon and aluminum substrates, Nanotechnol-
Mater. 39, 3099–3110 (1991) ogy 16, 1549–1558 (2005)
28.137 F.P. Bowden, D. Tabor: The Friction and Lubrication 28.151 Z. Tao, B. Bhushan: Degradation mechanisms and
of Solids (Clarendon, Oxford 1950) environmental effects on perfluoropolyether, self-
856 Part D Bio-/Nanotribology and Bio-/Nanomechanics
assembled monolayers, and diamondlike carbon 28.156 E. Hoque, J.A. DeRose, B. Bhushan, H.J. Math-