Encyclopedia of Nanotechnology
Bharat Bhushan Editor
Encyclopedia of Nanotechnology
With 1976 Figures and 124 Tables
Editor Professor Bharat Bhushan Ohio Eminent Scholar and The Howard D. Winbigler Professor, Director, Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics (NLB2) Ohio State University 201 W. 19th Avenue Columbus, Ohio, 43210-1142 USA
ISBN 978-90-481-9750-7 ISBN 978-90-481-9751-4 (eBook) DOI 10.1007/978-90-481-9751-4 ISBN 978-90-481-9752-1 (print and electronic bundle) Springer Dordrecht Heidelberg New York London Library of Congress Control Number: 2012940716 # Springer ScienceþBusiness Media B.V. 2012 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer is part of Springer ScienceþBusiness Media (www.springer.com)
Preface
On December 29, 1959, at the California Institute of Technology, Nobel Laureate Richard P. Feynman gave a speech at the Annual meeting of the American Physical Society that has become one of the twentieth-century classic science lectures, titled “There’s Plenty of Room at the Bottom.” He presented a technological vision of extreme miniaturization in 1959, several years before the word “chip” became part of the lexicon. He spoke about the problem of manipulating and controlling things on a small scale. Extrapolating from known physical laws, Feynman envisioned a technology using the ultimate toolbox of nature, building nanoobjects atom by atom or molecule by molecule. Since the 1980s, many inventions and discoveries in the fabrication of nanoobjects have been testament to his vision. In recognition of this reality, National Science and Technology Council (NSTC) of the White House created the Interagency Working Group on Nanoscience, Engineering and Technology (IWGN) in 1998. In a January 2000 speech at the same institute, former President W. J. Clinton spoke about the exciting promise of “nanotechnology” and the importance of expanding research in nanoscale science and technology, more broadly. Later that month, he announced in his State of the Union Address an ambitious $497 million federal, multiagency National Nanotechnology Initiative (NNI) in the fiscal year 2001 budget, and made the NNI a top science and technology priority. The objective of this initiative was to form a broad-based coalition in which the academe, the private sector, and local, state, and federal governments work together to push the envelope of nanoscience and nanoengineering to reap nanotechnology’s potential social and economic benefits. The funding in the USA has continued to increase. In January 2003, the US senate introduced a bill to establish a National Nanotechnology Program. On December 3, 2003, President George W. Bush signed into law the 21st Century Nanotechnology Research and Development Act. The legislation put into law programs and activities supported by the (NNI). The bill gave nanotechnology a permanent home in the federal government and authorized $3.7 billion to be spent in the 4-year period beginning in October 2005, for nanotechnology initiatives at five federal agencies. The funds have provided grants to researchers, coordinated R&D across five federal agencies (National Science Foundation (NSF), Department of Energy (DOE), NASA, National Institute of Standards and Technology (NIST), and Environmental Protection Agency (EPA)), established interdisciplinary research centers, and accelerated technology transfer into the private sector. In addition, the Department of 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 for more than one-third of the federal budget for nanotechnology.
v
vi
The European Union (EU) made nanosciences and nanotechnologies a priority in the Sixth Framework Program (FP6) in 2002 for the period 2003–2006. They had dedicated small funds in FP4 and FP5 before. FP6 was tailored to help better structure European research and to cope with the strategic objectives set out in Lisbon in 2000. Japan identified nanotechnology as one of its main research priorities in 2001. The funding levels increased sharply from $400 million in 2001 to around $950 million in 2004. In 2003, South Korea embarked upon a 10-year program with around $2 billion of public funding, and Taiwan has committed around $600 million of public funding over 6 years. Singapore and China are also investing on a large scale. Russia is well funded as well. Nanotechnology literally means any technology done on a nanoscale that has applications in the real world. Nanotechnology encompasses production and application of physical, chemical, and biological systems at scales, ranging from individual atoms or molecules to submicron dimensions, as well as the integration of the resulting nanostructures into larger systems. Nanotechnology is likely to have a profound impact on our economy and society in the early twenty-first century, comparable to that of semiconductor technology, information technology, or cellular and molecular biology. Science and technology research in nanotechnology is leading to breakthroughs in areas such as materials and manufacturing, nanoelectronics, medicine and healthcare, energy, biotechnology, information technology, and national security. It is widely felt that nanotechnology will be the next industrial revolution. There is an increasing need for a multidisciplinary, system-oriented approach toward designing and manufacturing micro/nanodevices which function reliably. This can only be achieved through the cross-fertilization of ideas from different disciplines and the systematic flow of information and people among research groups. Reliability is a critical technology for many micro- and nanosystems and nanostructured materials. The first edition of a broad-based Handbook of Nanotechnology from Springer was published in April 2004, the second edition in 2007, and the third edition in 2010. It presents an overview of nanomaterial synthesis, micro/nanofabrication, micro- and nanocomponents and systems, scanning probe microscopy, reliability issues (including nanotribology and nanomechanics) for nanotechnology, and various industrial including biomedical applications. The field of nanotechnology is getting a strong foothold. It attracts people from various disciplines including science and engineering. Given the explosive growth in nanoscience and nanotechnology, this Encyclopedia of Nanotechnology is being launched with essays written by experts in the field from academia and industry. The objective of this encyclopedia is to introduce a large number of terms, devices, and processes. For each entry, a brief description is provided by experts in the field. The entries have been written by a large number of internationally recognized experts in the field, from academia, national research labs, and industry. The Encyclopedia of Nanotechnology is expected to provide a comprehensive and multidisciplinary reference to the many fields relevant to the general field of nanotechnology. The encyclopedia focuses on engineering and applications with some coverage of the science of nanotechnology. It aims to be a comprehensive and genuinely international reference work and is aimed at graduate students, researchers, and practitioners. The development of the encyclopedia was undertaken by an Editorial Board, who were responsible for the focus and quality of contributions, and an Advisory Board, who were responsible for advising about the selection of topics.
Preface
Preface
vii
The print version of the encyclopedia contains four volumes with a total of 325 entries and about 3068 pages. The encyclopedia is also available online. Editors expect to update it periodically. The editor-in-chief and all the editors thank a large number of authors for making contributions to this major reference work. We also thank the referees who meticulously read the entries and made their recommendations. Powell, Ohio USA May 2012
Bharat Bhushan
Biography
Bharat Bhushan Ohio Eminent Scholar The Howard D. Winbigler Professor, Director Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics (NLB2) Ohio State University 201 W. 19th Avenue Columbus, Ohio, 43210-1142 USA
[email protected] Dr. Bharat Bhushan received an M.S. in mechanical engineering from the Massachusetts Institute of Technology in 1971; an M.S. in mechanics and a Ph.D. in mechanical engineering from the University of Colorado at Boulder in 1973 and 1976, respectively; an MBA from Rensselaer Polytechnic Institute at Troy, NY, in 1980; a Doctor Technicae from the University of Trondheim at Trondheim, Norway, in 1990; a Doctor of Technical Sciences from the Warsaw University of Technology at Warsaw, Poland, in 1996; and Doctor Honouris Causa from the National Academy of Sciences at Gomel, Belarus, in 2000 and University of Kragujevac, Serbia, in 2011. He is a registered professional engineer. He is presently an Ohio Eminent Scholar and The Howard D. Winbigler Professor in the College of Engineering and the Director of the Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics (NLB2) at the Ohio State University, Columbus, Ohio. His research interests include fundamental studies with a focus on scanning probe techniques in the interdisciplinary areas of bio/nanotribology, bio/nanomechanics, and bio/nanomaterials characterization and applications to bio/nanotechnology and biomimetics. He is an internationally recognized expert of bio/nanotribology and bio/nanomechanics using scanning probe microscopy and is one of the most prolific authors. He is considered by some a pioneer of the tribology and mechanics of magnetic storage devices. He has authored 8 scientific books, approximately 90 handbook chapters, 700 scientific papers (h-index – 57; ISI Highly Cited in Materials Science, since 2007; ISI Top 5% Cited Authors for Journals in Chemistry since 2011), and 60 technical reports. He has also edited more than 50 books and holds 17 US and foreign patents. He is coeditor of Springer NanoScience and Technology Series and coeditor of Microsystem Technologies. He has given more than 400 invited presentations on six continents and more than 200 keynote/plenary addresses at major international conferences. ix
x
Dr. Bhushan is an accomplished organizer. He organized the 1st Symposium on Tribology and Mechanics of Magnetic Storage Systems in 1984 and the 1st International Symposium on Advances in Information Storage Systems in 1990, both of which are now held annually. He is the founder of an ASME Information Storage and Processing Systems Division founded in 1993 and served as the founding chair during the period 1993–1998. His biography has been listed in over two dozen Who’s Who books including Who’s Who in the World and has received more than two dozen awards for his contributions to science and technology from professional societies, industry, and US government agencies. He is also the recipient of various international fellowships including the Alexander von Humboldt Research Prize for Senior Scientists, Max Planck Foundation Research Award for Outstanding Foreign Scientists, and the Fulbright Senior Scholar Award. He is a foreign member of the International Academy of Engineering (Russia), Byelorussian Academy of Engineering and Technology, and the Academy of Triboengineering of Ukraine; an honorary member of the Society of Tribologists of Belarus; a fellow of ASME, IEEE, STLE, and the New York Academy of Sciences; and a member of ASEE, Sigma Xi, and Tau Beta Pi. Dr. Bhushan has previously worked for Mechanical Technology Inc., Latham, NY; SKF Industries Inc., King of Prussia, PA; IBM, Tucson, AZ; and IBM Almaden Research Center, San Jose, CA. He has held visiting professorship at University of California at Berkeley; University of Cambridge, UK; Technical University Vienna, Austria; University of Paris, Orsay; ETH, Zurich; and EPFL, Lausanne. He is currently a visiting professor at KFUPM, Saudi Arabia; Harbin Inst., China; University of Kragujevac, Serbia, and University of Southampton, UK.
Biography
Section Editors
Bharat Bhushan Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA Donald Brenner Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA Jason V. Clark Department of Mechanical Engineering, Purdue University, W. Lafayette, IN, USA Robert J. Davis The Ohio State University, Columbus, OH, USA Maarten deBoer Mechanical Pittsburgh, PA, USA
Engineering,
Carnegie-Mellon
University,
Paolo Decuzzi Department of Translational Imaging and Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA Lixin Dong Electrical & Computer Engineering, Michigan State University, East Lansing, MI, USA Walter Federle Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, UK Emmanuel Flahaut Cirimat/Lcmie, Umr Cnrs 5085 Baˆtiment, Universite Paul Sabatier, France Enrico Gnecco Campus Universitario de Cantoblanco, IMDEA Nanociencia, Madrid, Spain Jian He Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Michael Heller Department of Bioengineering, University of California at San Diego, La Jolla, CA, USA David Holmes London Centre for Nanotechnology, University College London, London, UK Ali Khademhosseini Harvard Medical School, Cambridge, MA, USA Philippe Lutz The Universite´ de Franche-Comte´, Besanc¸on, France Marc Madou Department of Mechanical and Aerospace Engineering, University of California at Irvine, Irvine, CA, USA
xi
xii
Section Editors
Ernest Mendoza Nanomaterials Aplicats Centre de Recerca en Nanoenginyeria, Universitat Polite`cnica de Catalunya, Barcelona, Spain Mohammad Mofrad Director, Molecular Cell Biomechanics Lab, University of California Berkeley, Berkeley, CA, USA Michael Nosonovsky Department of Mechanical Engineering, University of Wisconsin at Milwaukee, Milwaukee, WI, USA Gianluca (“Gian”) Piazza Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, USA Fabrizio Pirri Dipartimento di Scienza dei Materiali e Ing. Chimica Laboratorio Materiali e Microsistemi / LATEMAR, Torino, Italy Shaurya Prakash Department of Mechanical Engineering, The Ohio State University, Columbus, OH, USA Themistoklis Prodromakis Institute of Biomedical Engineering, Imperial College, London, UK Yabing Qi Energy Materials and Surface Sciences (EMSS) Unit, Okinawa Institute of Science and Technology Graduate University (OIST), Okinawa, Japan Apparao M. Rao Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Maxim Sukharev Department of Applied Sciences and Mathematics, Arizona State University, Polytechnic Campus, Mesa, AZ, USA Srinivas Tadigadapa Department of Electrical Engineering, The Pennsylvania State University, University Park, PA, USA Lorenzo Valdevit Department of Mechanical and Aerospace Engineering, University of California, Irvine, Irvine, CA, USA Chunlei (Peggy) Wang Mechanical and International University, Miami, FL, USA
Materials
Engineering,
Florida
Steve Wereley Department of Mechanical Engineering, Purdue University, West Lafayette, IN, USA Huikai Xie Biophotonics & Microsystems Lab Department of Electrical & Computer Engineering, University of Florida, Gainesville, FL, USA Yi Zhao Department of Biomedical Engineering, The Ohio State University, Columbus, OH, USA
Advisory Board
Eduard Arzt Scientific Director and CEO, Leibniz Institute for New Materials, Saarbruecken, Germany Hans-J€ urgen Butt Max-Planck-Institute for Polymer Research, Mainz, Germany Nico de Rooij EPFL STI IMT & CSEM SA, Neuchaˆtel, Switzerland Gary K. Fedder Department of Electrical and Computer Engineering, and The Robotics Institute, Pittsburgh, PA, USA Peter Fratzl Max Planck Institute of Colloids and Interfaces, Department of Biomaterials, Potsdam, Germany Mauro Ferrari Texas Health Science Center, Houston, TX, USA Harald Fuchs Physikalisches Institut, University of Muenster, Muenster, Germany Christoph Gerber University of Basel, NCCR National Center of Competence for Nanoscience, Institute of Physics, Basel, Switzerland Christofer Hierold Micro and Nanosystems, Department of Mechanical and Process Engineering, Zurich, Switzerland Marc Madou University of California at Irvine, Department of Mechanical and Aerospace Engineering, Irvine, CA, USA Ernst Meyer University of Basel, National Competence Center for Research in Nanoscale Science (NCCR), Institut fuer Physik, Basel, Switzerland Bernd Michel Fraunhofer MicroMat. Center, Berlin, Germany Patrick Pons LAAS-CNRS, Toulouse Cedex 4, France Miquel Salmeron Director of the Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA Thomas Schimmel Karlsruhe Institute of Technology, Institute for Applied Physics, Karlsruhe, Germany Roland Wiesendanger University of Hamburg, Institute of Applied Physics, Hamburg, Germany
xiii
Contributors
Ahmad Nabil Abbas Department of Electrical and Computer Engineering, Florida International University, Miami, FL, USA Patrick Abgrall Formulaction, L’Union, France Angelo Accardo Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy Soft Matter Structures Group ID13 – MICROFOCUS Beamline, European Synchrotron Radiation Facility, Grenoble Cedex, France Wafa Achouak iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France Laboratoire d’Ecologie Microbienne de la Rhizosphe`re et d’Environnements Extreˆme, UMR 6191 CNRS–CEA–Aix–Marseille Universite´ de la Me´diterrane´e, St Paul lez Durance, France Ranjeet Agarwala Department of Mechanical & Aerospace Engineering, North Carolina State University, Raleigh, NC, USA Alessandro Alabastri Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Muhammad A. Alam School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN, USA Tuncay Alan Mechanical and Aerospace Engineering Department, Monash University, Victoria, Australia Antonio Aliano Department of Physics, Politecnico di Torino, Torino, Italy Wafa’ T. Al-Jamal Nanomedicine Laboratory, Centre for Drug Delivery Research, The School of Pharmacy, University of London, London, UK Paolo Allia Materials Science and Chemical Engineering Department, Politecnico di Torino, Torino, Italy N. R. Aluru Department of Mechanical Science and Engineering, Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana – Champaign, Urbana, IL, USA Giampiero Amato Quantum Research Laboratory, Electromagnetism Division, Istituto Nazionale di Ricerca Metrologica, Torino, Italy xv
xvi
Contributors
Benoy Anand Department of Physics, Sri Sathya Sai Institute of Higher Learning, Prashanthinilayam, Andhra Pradesh, India Francesco De Angelis Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Jose´ V. Anguita Nano Electronics Center, Advanced Technology Institute, University of Surrey, Guildford, Surrey, UK Wadih Arap David H. Koch Center, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Departments of Genitourinary Medical Oncology and Cancer Biology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Walter Arnold Department of Material Science and Technology, Saarland University, Saarbr€ ucken, Germany Physikalisches Institut, G€ ottingen University, G€ottingen, Germany Christopher Arntsen Department of Chemistry and Biochemistry, UCLA, Los Angeles, CA, USA Eduard Arzt INM – Leibniz Institute for New Materials and Saarland University, Saarbr€ ucken, Germany Burcu Aslan Department of Experimental Therapeutics, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Elena Astanina University of Torino, Department of Oncological Sciences, Candiolo, Torino, Italy Orlando Auciello Materials Science Division, Argonne National Laboratory, Argonne, IL, USA Me´lanie Auffan CEREGE, Aix–en–Provence, France
UMR
6635
CNRS/Aix–Marseille
Universite´,
iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France Thomas Bachmann Institute for Fluid Mechanics and Aerodynamics, Technische Universit€at Darmstadt, Darmstadt, Germany Armelle Baeza-Squiban Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France Darren M. Bagnall Nano Group, Electronics and Computer Science, University of Southampton, Highfield, Southampton, UK Xuedong Bai Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing, P. R. China Anisullah Baig Department of Electrical and Computer Engineering, University of California, Davis, USA
Contributors
xvii
David J. Bakewell Department of Electrical Engineering and Electronics, University of Liverpool, Liverpool, UK Mirko Ballarini DIFIS – Dipartimento di Fisica, Politecnico di Torino, Torino, Italy Bele´n Ballesteros CIN2 (ICN-CSIC), Catalan Institute of Nanotechnology, Bellaterra, Barcelona, Spain Antoine Barbier CEA-Saclay, DSM/IRAMIS/SPCSI, Gif-sur-Yvette, France Robert Barchfeld Department of Electrical and Computer Engineering, University of California, Davis, USA Yoseph Bar-Cohen Jet Propulsion Laboratory (JPL), California Institute of Technology, Pasadena, CA, USA W. Jon. P. Barnes Centre for Cell Engineering, University of Glasgow, Glasgow, Scotland, UK Larry R. Barnett Department of Electrical and Computer Engineering, University of California, Davis, USA Friedrich G. Barth Life Sciences, Department of Neurobiology, University of Vienna, Vienna, Austria Michael H. Bartl Department of Chemistry, University of Utah, Salt Lake City, UT, USA R. Baskaran Department of Electrical Engineering, University of Washington, Seattle, WA, USA Components Research, Intel Corporation, USA Werner Baumgartner Department of Cellular Neurobionics, Institut f€ur Biologie II RWTH-Aachen University, Aachen, Germany Pierre Becker Laboratoire de Biologie des Organismes Marins et Biomime´tisme, Universite´ de Mons – UMONS, Mons, Belgium Novid Beheshti Swedish Biomimetics 3000 ® Ltd, Birmingham, UK Rachid Belkhou Synchrotron SOLEIL, L’Orme des Merisiers, Gif-sur-Yvette, France Laila Benameur Laboratoire de Bioge´notoxicologie et Mutagene`se Environnementale (EA 1784/FR CNRS 3098 ECCOREV), Aix-Marseille Universite´, Marseille Cedex 5, France R€ udiger Berger Max Planck Institute for Polymer Research, Mainz, Germany Magnus Berggren Department of Science and Technology, Link€opings University, Norrk€ oping, Sweden Pierre Berini School of Information Technology and Engineering (SITE), University of Ottawa, Ottawa, ON, Canada Shekhar Bhansali Department of Electrical and Computer Engineering, Florida International University, Miami, FL, USA
xviii
Vikram Bhatia Science and Technology, Corning Incorporated SP-PR-02-1, Corning, NY, USA Rustom B. Bhiladvala Department of Mechanical Engineering, University of Victoria, Victoria, BC, Canada Bharat Bhushan Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA Stefano Bianco Center for Space Human Robotics, Fondazione Istituto Italiano di Tecnologia, Torino, Italy Ion Bita Qualcomm MEMS Technologies, Inc., San Jose, CA, USA Martin G. Blaber Department of Chemistry and International Institute for Nanotechnology, Northwestern University, Evanston, IL, USA Luca Boarino Quantum Research Laboratory, Electromagnetism Division, Istituto Nazionale di Ricerca Metrologica, Torino, Italy Jorge Boczkowski INSERM U955 Eq04, Cre´teil, France University Paris Est Val de Marne (UPEC), Cre´teil, France Stuart A. Boden Nano Group, Electronics and Computer Science, University of Southampton, Highfield, Southampton, UK Peter Bøggild DTU Nanotech – Department of Micro- and Nanotechnology, Technical University of Denmark, Lyngby, Denmark K. F. B€ ohringer Department of Electrical Engineering, University of Washington, Seattle, WA, USA Maarten P. de Boer Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, USA Ardemis A. Boghossian Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Paul W. Bohn Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, IN, USA Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, IN, USA Sonja Boland Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France Robert D. Bolskar TDA Research, Inc., Wheat Ridge, CO, USA Alexander Booth Energy and Resources Research Institute, University of Leeds, Leeds, West Yorkshire, UK Garry J. Bordonaro Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NY, USA Edward Bormashenko Laboratory of Polymers, Applied Physics Department, Ariel University Center of Samaria, Ariel, Israel
Contributors
Contributors
xix
Ce´line Botta CEREGE, UMR 6635, CNRS, Aix-Marseille Universite´, Aix en Provence, Cedex 04, France Alain Botta Laboratoire de Bioge´notoxicologie et Mutagene`se Environnementale (EA 1784/FR CNRS 3098 ECCOREV), Aix-Marseille Universite´, Marseille Cedex 5, France Jean-Yves Bottero CEREGE, UMR 6635 CNRS/Aix–Marseille Universite´, Aix–en–Provence, France iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix– En–Provence, Cedex 4, France Alessia Bottos University of Torino, Department of Oncological Sciences, Candiolo, Torino, Italy Floriane Bourdiol EcoLab – Laboratoire d’e´cologie fonctionnelle ´ et environnement, Universite de Toulouse, INP, UPS, Castanet Tolosan, France CNRS UMR 5245, EcoLab, Castanet Tolosan, France Institut Carnot Cirimat, Universite´ de Toulouse, UPS, INP, Toulouse cedex 9, France Olivier Bourgeois Institut Ne´el CNRS-UJF, Grenoble, France Herbert Bousack Peter Gr€unberg Institut, Forschungszentrum J€ulich GmbH, J€ ulich, Germany Thomas Braschler Laboratory of stem cell dynamics, SV-EPFL, Lausanne, Switzerland Graham Bratzel Laboratory for Atomistic and Molecular Mechanics, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Donald W. Brenner Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA Victor M. Bright Department of Mechanical Engineering, University of Colorado, Boulder, CO, USA Lawrence F. Bronk David H. Koch Center, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Joseph J. Brown Department of Mechanical Engineering, University of Colorado, Boulder, CO, USA Dorothea Br€ uggemann The Naughton Institute, School of Physics, Trinity College Dublin, CRANN, Dublin, Ireland Markus J. Buehler Laboratory for Atomistic and Molecular Mechanics, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Sven Burger Zuse Institute Berlin, Berlin, Germany
xx
Contributors
Federico Bussolino University of Torino, Department of Oncological Sciences, Candiolo, Torino, Italy Donald P. Butler The Department of Electrical Engineering, The University of Texas at Arlington, Arlington, TX, USA Hans-J€ urgen Butt Max Planck Institute for Polymer Research, Mainz, Germany Javier Calvo Fuentes NANOGAP SUB-NM-POWDER S.A., Milladoiro – Ames (A Corun˜a), Spain Mary Cano-Sarabia CIN2 (ICN-CSIC), Catalan Institute of Nanotechnology, Bellaterra, Barcelona, Spain Andre´s Cantarero Materials Science Institute, University of Valencia, Valencia, Spain Francesca Carpino Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, IN, USA Marie Carrie`re Laboratoire Le´sions des Acides Nucle´iques, Commissariat a` l’Energie Atomique, SCIB, UMR-E 3 CEA/UJF-Grenoble 1, INAC, Grenoble, Cedex, France Je´roˆme Casas Institut de Recherche en Biologie de l’Insecte, IRBI UMR CNRS 6035, Universite´ of Tours, Tours, France Zeynep Celik-Butler The Department of Electrical Engineering, The University of Texas at Arlington, Arlington, TX, USA Frederik Ceyssens Department ESAT-MICAS, KULeuven, Leuven, Belgium Nicolas Chaillet FEMTO-ST/UFC-ENSMM-UTBM-CNRS, Besanc¸on, France Audrey M. Chamoire Department of Mechanical and Aerospace Engineering, The Ohio State University, E443 Scott Laboratory, Columbus, OH, USA Peggy Chan Micro/Nanophysics Melbourne, VIC, Australia
Research
Laboratory,
RMIT
University,
Munish Chanana Departamento de Quı´mica Fı´sica, Universidade de Vigo, Vigo, Spain Pen-Shan Chao Chung-Shan Institute of Science and Technology, Taoyuan, Taiwan ROC Satish C. Chaparala Science and Technology, Corning Incorporated SP-PR-02-1, Corning, NY, USA Wei Chen Department of Mechanical and Materials Engineering, Florida International University, Miami, FL, USA Jian Chen Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada Department of Chemistry and Biochemistry, University of Wisconsin-Milwaukee, Milwaukee, WI, USA Angelica Chiodoni Center for Space Human Robotics, Fondazione Istituto Italiano di Tecnologia, Torino, Italy
Contributors
xxi
Alessandro Chiolerio Physics Department, Politecnico di Torino, Torino, Italy Yoon-Kyoung Cho School of Nano-Bioscience and Chemical Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan, Republic of Korea Jong Hyun Choi School of Mechanical Engineering, Birck Nanotechnology Center, Bindley Bioscience Center, Purdue University, West Lafayette, IN, USA Michael Chu Advanced Pharmaceutics and Drug Delivery Laboratory, University of Toronto, Toronto, ON, Canada Han-Sheng Chuang Department of Biomedical Engineering, National Cheng Kung University, Tainan, Taiwan Medical Device Innovation Center, National Cheng Kung University, Taiwan Giancarlo Cicero Materials Science and Chemical Engineering Department, Politecnico di Torino, Torino, Italy C. Cierpka Institute of Fluid Mechanics and Aerodynamics, Universit€at der Bundeswehr M€ unchen, Neubiberg, Germany Dominique Collard LIMMS/CNRS-IIS (UMI 2820), Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Lucio Colombi Ciacchi Hybrid Materials Interfaces Group, Faculty of Production Engineering and Bremen Center for Computational Materials Science, University of Bremen, Bremen, Germany Maria Laura Coluccio Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy A. T. Conlisk Department of Mechanical Engineering, The Ohio State University, Columbus, OH, USA Andrew Copestake Swedish Biomimetics 3000 ® Ltd, Southampton, UK Miguel A. Correa-Duarte Departamento de Quı´mica Fı´sica, Universidade de Vigo, Vigo, Spain Giovanni Costantini Department of Chemistry, The University of Warwick, Coventry, UK Claire Coutris Department of Plant and Environmental Sciences, Norwegian ˚ s, Norway University of Life Sciences, A Eduardo Cruz-Silva Department of Polymer Science and Engineering, University of Massachusetts Amherst, Amherst, MA, USA Steven Curley Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Department of Mechanical Engineering and Materials Science, Rice University, Houston, TX, USA
xxii
Christian Dahmen Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany Lu Dai The Key Laboratory of Remodeling-related Cardiovascular Diseases, Capital Medical University, Ministry of Education, Beijing, China Gobind Das Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Bakul C. Dave Department of Chemistry and Biochemistry, Southern Illinois University Carbondale, Carbondale, IL, USA Enrica De Rosa Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA Paolo Decuzzi Dept of Translational Imaging, and Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA Christian L. Degen Department of Physics, ETH Zurich, Zurich, Switzerland Ada Della Pia Department of Chemistry, The University of Warwick, Coventry, UK Gregory Denbeaux College of Nanoscale Science and Engineering, University at Albany, Albany, NY, USA Parag B. Deotare Electrical Engineering, Harvard School of Engineering and Applied Sciences, Cambridge, MA, USA Emiliano Descrovi DISMIC – Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, Torino, Italy Joseph M. DeSimone Department of Chemistry, University of North Carolina, Chapel Hill, NC, USA Department of Pharmacology, Eshelman School of Pharmacy, University of North Carolina, Chapel Hill, NC, USA Carolina Center of Cancer Nanotechnology Excellence, University of North Carolina, Chapel Hill, NC, USA Institute for Advanced Materials, University of North Carolina, Chapel Hill, NC, USA Institute for Nanomedicine, University of North Carolina, Chapel Hill, NC, USA Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC, USA Department of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh, NC, USA Sloan–Kettering Institute for Cancer Research, Memorial Sloan–Kettering Cancer Center, New York, NY, USA Hans Deyhle Biomaterials Science Center (BMC), University of Basel, Basel, Switzerland Charles L. Dezelah IV Picosun USA, LLC, Detroit, MI, USA Nathan Doble The New England College of Optometry, Boston, MA, USA
Contributors
Contributors
xxiii
Mitchel J. Doktycz Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Calvin Domier Department of Electrical and Computer Engineering, University of California, Davis, USA Lixin Dong Electrical and Computer Engineering, Michigan State University, East Lansing, MI, USA Avinash M. Dongare Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA Emmanuel M. Drakakis Department of Bioengineering, The Sir Leon Bagrit Centre, Imperial College London, London, UK Wouter H. P. Driessen David H. Koch Center, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Carlos Drummond Centre de Recherche Paul Pascal, CNRS–Universite´ Bordeaux 1, Pessac, France Jie Du Beijing Institute of Heart Lung and Blood Vessel Diseases, Beijing Anzhen Hospital, Beijing, China Jean-Marie Dupret Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France Julianna K. Edwards David H. Koch Center, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Volkmar Eichhorn Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany Masayoshi Esashi The World Premier International Research Center Initiative for Atom Molecule Materials, Tohoku University, Aramaki, Aoba-ku Sendai, Japan Mikael Evander Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden Enzo Di Fabrizio Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy Yubo Fan Center for Bioengineering and Informatics, Department of Systems Medicine and Bioengineering, The Methodist Hospital Research Institute, Weill Cornell Medical College, Houston, TX, USA Zheng Fan Electrical and Computer Engineering, Michigan State University, East Lansing, MI, USA Sergej Fatikow Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany
xxiv
Henry O. Fatoyinbo Centre for Biomedical Engineering, University of Surrey, Guildford, Surrey, UK Joseph Fernandez-Moure The Methodist Hospital Research Institute, Houston, TX, USA Mauro Ferrari Department of NanoMedicine, The Methodist Hospital Research Institute, Houston, TX, USA Benjamin M. Finio School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Emmanuel Flahaut Institut Carnot Cirimat, Universite´ de Toulouse, UPS, INP, Toulouse cedex 9, France CNRS, Institut Carnot Cirimat, Toulouse, France Patrick Flammang Laboratoire de Biologie des Organismes Marins et Biomime´tisme, Universite´ de Mons – UMONS, Mons, Belgium Richard G. Forbes Advanced Technology Institute, Faculty of Engineering and Physical Sciences, University of Surrey, Guildford, UK Isabelle Fourquaux CMEAB, Centre de Microscopie Electronique Applique´e a` la Biologie, Universite´ Paul Sabatier, Faculte´ de Me´decine Rangueil, Toulouse cedex 4, France Marco Francardi Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy International School for Advanced Studies (SISSA), Edificio Q1 Trieste, Italy Francesca Frascella DISMIC – Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, Torino, Italy Roger H. French Department of Materials Science and Engineering, Case Western Reserve University, Cleveland, OH, USA James Friend Micro/Nanophysics Research Laboratory, RMIT University, Melbourne, VIC, Australia Hiroyuki Fujita Center for International Research on MicroMechatronics (CIRMM), Institute of Industrial Science, The University of Tokyo, Meguro-ku, Tokyo, Japan Kenji Fukuzawa Department of Micro System Engineering, Nagoya University, Nagoya, Chikusa-ku, Japan Diana Gamzina Department of Electrical and Computer Engineering, University of California, Davis, USA Xuefeng Gao Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences, Suzhou, PR China Pablo Garcı´a-Sa´nchez Departamento de Electro´nica y Electromagnetismo, Universidad de Sevilla, Sevilla, Spain Jean-Luc Garden Institut Ne´el CNRS-UJF, Grenoble, France
Contributors
Contributors
xxv
Paolo Gasco Nanovector srl, Torino, Italy Laury Gauthier EcoLab – Laboratoire d’e´cologie fonctionnelle et environnement, Universite´ de Toulouse, INP, UPS, Castanet Tolosan, France CNRS UMR 5245, EcoLab, Castanet Tolosan, France Shady Gawad MEAS Switzerland, Bevaix, Switzerland Denis Gebauer Department of Chemistry, Physical Chemistry, University of Konstanz, Konstanz, Germany Ille C. Gebeshuber Institute of Microengineering and Nanoelectronics (IMEN), Universiti Kebangsaan Malaysia, Bangi, Selangor, Malaysia Institute of Applied Physics, Vienna University of Technology, Vienna, Austria Francesco Gentile Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy Claudio Gerbaldi Center for Space Human Robotics, Fondazione Istituto Italiano di Tecnologia, Torino, Italy Amitabha Ghosh Bengal Engineering & Science University, Howrah, India Ranajay Ghosh Department of Mechanical, Aerospace and Nuclear Engineering, Rensselaer Polytechnic Institute, Troy, NY, USA Larry R. Gibson II Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, IN, USA Jason P. Gleghorn Department of Chemical and Biological Engineering, Princeton University, Princeton, NJ, USA Biana Godin Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA Irene Gonza´lez-Valls Laboratory of Nanostructured Materials for Photovoltaic Energy, Escola Tecnica Superior d Enginyeria (ETSE), Centre d’Investigacio´ en Nanocie`ncia i Nanotecnologı´a (CIN2, CSIC), Bellaterra (Barcelona), Spain Ashwini Gopal Department of Biomedical Engineering, The University of Texas at Austin, Austin, TX, USA Claudia R. Gordijo Advanced Pharmaceutics and Drug Delivery Laboratory, University of Toronto, Toronto, ON, Canada Yann Le Gorrec FEMTO-ST/UFC-ENSMM-UTBM-CNRS, Besanc¸on, France Alok Govil Qualcomm MEMS Technologies, Inc., San Jose, CA, USA Paul Graham Centre for Computational Neuroscience and Robotics, University of Sussex, Brighton, UK Dmitri K. Gramotnev Nanophotonics Pty Ltd, Brisbane, QLD, Australia
xxvi
Contributors
Nicolas G. Green School of Electronics and Computer Science, University of Southampton, Highfield, Southampton, UK Robert J. Greenberg Second Sight Medical Products (SSMP), Sylmar, CA, USA Julia R. Greer Division of Engineering and Applied Sciences, California Institute of Technology, Pasadena, CA, USA Dane A. Grismer Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, IN, USA Petra Gruber Transarch - Biomimetics and Transdisciplinary Architecture, Vienna, Austria Rina Guadagnini Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France Vladimir Gubala Biomedical Diagnostics Institute, Dublin City University, Glasnevin, Dublin, Ireland Pablo Gurman Materials Science Division, Argonne National Laboratory, Argonne, IL, USA Evgeni Gusev Qualcomm MEMS Technologies, Inc., San Jose, CA, USA MEMS Research and Innovation Center, Qualcomm MEMS Technologies, Inc., San Jose, CA, USA Maria Laura Habegger Department of Integrative Biology, University of South Florida, Tampa, FL, USA Yassine Haddab FEMTO-ST/UFC-ENSMM-UTBM-CNRS, Besanc¸on, France Neal A. Hall Electrical and Computer Engineering, University of Texas at Austin, Austin, TX, USA Moon-Ho Ham Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Hee Dong Han Gynecologic Oncology, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Center for RNA Interference and Non–coding RNA, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Xiaodong Han Institute of Microstructure and Property of Advanced Materials, Beijing University of Technology, Beijing, People’s Republic of China Aeraj ul Haque Biodetection Technologies Section, Energy Systems Division, Argonne National Laboratory, Lemont, IL, USA Nadine Harris Department of Chemistry and International Institute for Nanotechnology, Northwestern University, Evanston, IL, USA Judith A. Harrison Department of Chemistry, United States Naval Academy, Annapolis, MD, USA Achim Hartschuh Department Chemie Universit€at M€ unchen, Munich, Germany
and
CeNS,
Ludwig-Maximilians-
Contributors
xxvii
Jian He Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Martin Hegner The Naughton Institute, School of Physics, Trinity College Dublin, CRANN, Dublin, Ireland Michael G. Helander Department of Materials Science and Engineering, University of Toronto, Toronto, Ontario, Canada Michael J. Heller Department of Nanoengineering, University of California San Diego, La Jolla, CA, USA Department of Bioengineering, University of California San Diego, La Jolla, CA, USA Simon J. Henley Nano Electronics Center, Advanced Technology Institute, University of Surrey, Guildford, Surrey, UK Elise Hennebert Laboratoire de Biologie des Organismes Marins et Biomime´tisme, Universite´ de Mons – UMONS, Mons, Belgium Joseph P. Heremans Department of Mechanical and Aerospace Engineering, The Ohio State University, E443 Scott Laboratory, Columbus, OH, USA Simone Hieber Biomaterials Science Center (BMC), University of Basel, Basel, Switzerland Dale Hitchcock Department of Physics and Astronomy, Clemson University, Clemson, SC, USA David Holmes London Centre for Nanotechnology, University College London, London, UK Hendrik H€ olscher Karlsruher Institut f€ur Technologie (KIT), Institut f€ur Mikrostrukturtechnik, Karlsruhe, Germany J. H. Hoo Department of Electrical Engineering, University of Washington, Seattle, WA, USA Bart W. Hoogenboom London Centre for Nanotechnology and Department of Physics and Astronomy, University College London, London, UK Kazunori Hoshino Department of Biomedical Engineering, The University of Texas at Austin, Austin, TX, USA Larry L. Howell Department of Mechanical Engineering, Brigham Young University, Provo, UT, USA Hou-Jun Hsu C.C.P. Contact Probes CO., LTD, New Taipei City, Taiwan ROC Jung-Tang Huang National Taipei University of Technology, Taipei, Taiwan ROC Michael P. Hughes Centre for Biomedical Engineering, University of Surrey, Guildford, Surrey, UK Shelby B. Hutchens California Institute of Technology MC 309-81, Pasadena, CA, USA John W. Hutchinson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA
xxviii
Gilgueng Hwang Laboratoire de Photonique et de Nanostructures (LPN-CNRS), Site Alcatel de Marcoussis, Route de Nozay, Marcoussis, France Hyundoo Hwang School of Nano-Bioscience and Chemical Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan, Republic of Korea Barbara Imhof LIQUIFER Systems Group, Austria Hiromi Inada Hitachi High-Technologies America, Pleasanton, CA, USA M. Saif Islam Electrical and Computer Engineering, University of California Davis Integrated Nanodevices & Nanosystems Lab, Davis, CA, USA Mitsumasa Iwamoto Department of Physical Electronics, Tokyo Institute of Technology, Meguro-ku, Tokyo, Japan Esmaiel Jabbari Biomimetic Materials and Tissue Engineering Laboratory, Department of Chemical Engineering, Swearingen Engineering Center, Rm 2C11, University of South Carolina, Columbia, SC, USA Laurent Jalabert LIMMS/CNRS-IIS (UMI 2820), Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Dongchan Jang Department of Applied Physics and Materials Science, California Institute of Technology, Pasadena, CA, USA Daniel Jasper Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany Debdeep Jena Department of Electrical Engineering, University of Notre Dame, Notre Dame, IN, USA Taeksoo Ji School of Electronics and Computer Engineering, Chonnam National University, Gwangju, Korea Lixin Jia The Key Laboratory of Remodeling-related Cardiovascular Diseases, Capital Medical University, Ministry of Education, Beijing, China Lei Jiang Center of Molecular Sciences, Institute of Chemistry Chinese Academy of Sciences, Beijing, People’s Republic of China Dilip S. Joag Centre for Advanced Studies in Materials Science and Condensed Matter Physics, Department of Physics, University of Pune, Pune, Maharashtra, India ˚ s, Norway Erik J. Joner Bioforsk Soil and Environment, A Suhas S. Joshi Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai, Maharastra, India Gabriela Juarez-Martinez Centeo Biosciences Limited, Dumbarton, UK Soyoun Jung Samsung Mobile Display Co., LTD, Young-in, Korea C. J. K€ ahler Institute of Fluid Mechanics and Aerodynamics, Universit€at der Bundeswehr M€unchen, Neubiberg, Germany
Contributors
Contributors
xxix
Sergei V. Kalinin Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Ping Kao The Pennsylvania State University, University Park, PA, USA Swastik Kar Department of Physics, Northeastern University, Boston, MA, USA Mustafa Karabiyik Department of Electrical and Computer Engineering, Florida International University, Miami, FL, USA Sinan Karaveli School of Engineering, Brown University, Providence, RI, USA David Karig Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Michael Karpelson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Andreas G. Katsiamis Toumaz Technology Limited, Abingdon, UK Christine D. Keating Department of Chemistry, Penn State University, University Park, PA, USA Pamela L. Keating Department of Chemistry, United States Naval Academy, Annapolis, MD, USA John B. Ketterson Department of Physics and Astronomy, Northwestern University, Evanston, IL, USA S. M. Khaled Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA Arash Kheyraddini Mousavi Department of Mechanical Engineering, University of New Mexico, Albuquerque, NM, USA Andrei L. Kholkin Center for Research in Ceramic and Composite Materias (CICECO) & DECV, University of Aveiro, Aveiro, Portugal Seonghwan Kim Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB, Canada Seong H. Kim Department of Chemical Engineering, Pennsylvania State University, University Park, PA, USA Moon Suk Kim Department of Molecular Science and Technology, Ajou University, Suwon, South Korea CJ Kim Mechanical and Aerospace Engineering Department, University of California, Los Angeles (UCLA), Los Angeles, CA, USA Bongsang Kim Advanced MEMS, Sandia National Laboratories, Albuquerque, NM, USA Pilhan Kim Graduate School of Nanoscience and Technology, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea M. Todd Knippenberg Department of Chemistry, High Point University, High Point, NC, USA
xxx
Yee Kan Koh Department of Mechanical Engineering, National University of Singapore room: E2 #02-29, Singapore, Singapore Helmut Kohl Physikalisches Institut, Westf€alische Wilhelms-Universit€at M€unster, Wilhelm-Klemm-Straße 10, M€ unster, Germany Mathias Kolle Harvard School of Engineering and Applied Sciences, Cambridge, MA, USA Susan K€ oppen Hybrid Materials Interfaces Group, Faculty of Production Engineering and Bremen Center for Computational Materials Science, University of Bremen, Bremen, Germany Kostas Kostarelos Nanomedicine Laboratory, Centre for Drug Delivery Research, The School of Pharmacy, University of London, London, UK Roman Krahne Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Gijs Krijnen Transducers Science & Technology group, MESA + Research Institute for Nanotechnology, University of Twente, Enschede, The Netherlands Rajaram Krishnan Biological Dynamics, Inc., University of California San Diego, San Diego, CA, USA Florian Krohs Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany Elmar Kroner INM – Leibniz Institute for New Materials, Saarbr€ucken, Germany Tom N. Krupenkin Department of Mechanical Engineering, The University of Wisconsin-Madison, Madison, WI, USA Satish Kumar G. W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USA Aloke Kumar Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Momoko Kumemura LIMMS/CNRS-IIS (UMI 2820), Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Harry Kwok Department of Electrical and Computer Engineering, University of Victoria, Victoria, Canada Jae-Sung Kwon School of Mechanical Engineering and Birck Nanotechnology Center, Purdue University, West Lafayette, IN, USA Je´roˆme Labille CEREGE, UMR 6635, CNRS, Aix-Marseille Universite´, Aix en Provence, Cedex 04, France Jean Christophe Lacroix Interfaces, Traitements, Organisation et Dynamique des Systemes Universite Paris 7-Denis Diderot, Paris Cedex 13, France Nicolas Lafitte LIMMS/CNRS-IIS (UMI 2820), Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Jao van de Lagemaat National Renewable Energy Laboratory, Golden, CO, USA Renewable and Sustainable Energy Institute, Boulder, CO, USA
Contributors
Contributors
xxxi
Akhlesh Lakhtakia Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA, USA Pe´rine Landois Institut Carnot Cirimat, Universite´ de Toulouse, UPS, INP, Toulouse cedex 9, France Amy Lang Department of Aerospace Engineering & Mechanics, University of Alabama, Tuscaloosa, AL, USA Sophie Lanone Inserm U955, E´quipe 4, Universite´ Paris Est Val de Marne (UPEC), Cre´teil, France Hoˆpital Intercommunal de Cre´teil, Service de pneumologie et pathologie professionnelle, Cre´teil, France Gregory M. Lanza C-TRAIN Labs, Washington University, St. Louis, MO, USA Lars Uno Larsson Swedish Biomimetics 3000 ® AB, Stockholm, Sweden Camille Larue Laboratoire Le´sions des Acides Nucle´iques, Commissariat a` l’Energie Atomique, SCIB, UMR-E 3 CEA/UJF-Grenoble 1, INAC, Grenoble, Cedex, France Michael J. Laudenslager Department of Materials Science and Engineering, University of Florida, Gainesville, FL, USA Thomas Laurell Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden M. Laver Laboratory for Neutron Scattering, Paul Scherrer Institut, Villigen, Switzerland Materials Research Division, Risø DTU, Technical University of Denmark, Roskilde, Denmark Nano–Science Center, Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark Department of Materials Science and Engineering, University of Maryland, Maryland, USA Falk Lederer Institute of Condensed Matter Theory and Solid State Optics, Abbe Center of Photonics, Friedrich-Schiller-Universit€at Jena, Jena, Germany Kuo-Yu Lee National Taipei University of Technology, Taipei, Taiwan ROC David W. Lee Department of Biological Sciences, Florida International University Modesto Maidique Campus, Miami, FL, USA Andreas Lenshof Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden Donald J. Leo Mechanical Engineering, Center for Intelligent Material Systems and Structures, Virginia Tech, Virginia Polytechnic Institute and State University, Arlington, VA, USA Zayd Chad Leseman Department of Mechanical Engineering, University of New Mexico, Albuquerque, NM, USA
xxxii
Nastassja A. Lewinski Department of Bioengineering, Rice University, Houston, TX, USA Mo Li Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, USA Jason Li Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada Wenzhi Li Department of Physics, Florida International University, Miami, FL, USA Chen Li Department of Chemistry and Biochemistry, University of Bern, Bern, Switzerland Chun Li Department of Experimental Diagnostic Imaging-Unit 59, The University of Texas MD Anderson Cancer Center, Houston, TX, USA King C. Li Department of Bioengineering, Rice university, Houston, TX, USA Weicong Li Department of Electrical and Computer Engineering, University of Victoria, Victoria, Canada Meng Lian Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Carlo Liberale Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Ling Lin Beijing National Laboratory for Molecular Sciences (BNLMS), Key Laboratory of Organic Solids, Institute of Chemistry Chinese Academy of Sciences, Beijing, People’s Republic of China Chung-Yi Lin FormFactor, Inc., Taiwan, Hsinchu, Taiwan ROC Lih Y. Lin Department of Electrical Engineering, University of Washington, Seattle, WA, USA Mo´nica Lira-Cantu´ Laboratory of Nanostructured Materials for Photovoltaic Energy, Escola Tecnica Superior d Enginyeria (ETSE), Centre d’Investigacio´ en Nanocie`ncia i Nanotecnologı´a (CIN2, CSIC), Bellaterra (Barcelona), Spain Shawn Litster Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA, USA Ying Liu School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA, USA Chang Liu Tech Institute, Northwestern University, ME/EECS, Room L288, Evanston, IL, USA Gang Logan Liu Micro and Nanotechnology Laboratory, Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA Matthew T. Lloyd National Renewable Energy Laboratory MS 3211/SERF W100-43, Golden, CO, USA
Contributors
Contributors
xxxiii
Sarah B. Lockwood Department of Chemistry and Biochemistry, Southern Illinois University Carbondale, Carbondale, IL, USA V. J. Logeeswaran Electrical and Computer Engineering, University of California Davis Integrated Nanodevices & Nanosystems Lab, Davis, CA, USA Mariangela Lombardi IIT - Italian Institute of Technology @ POLITO - Centre for Space Human Robotics, Torino, Italy Marko Loncar Electrical Engineering, Harvard School of Engineering and Applied Sciences, Cambridge, MA, USA Kenneth A. Lopata William R. Wiley Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA, USA Gabriel Lopez-Berestein Department of Experimental Therapeutics, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Cancer Biology, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Center for RNA Interference and Non–coding RNA, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA The Department of Nanomedicine and Bioengineering, UTHealth, Houston, TX, USA Alejandro Lopez-Bezanilla National Center for Computational Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Jun Lou Department of Mechanical Engineering and Materials Science, Rice University 223 MEB, Houston, TX, USA M. Arturo Lo´pez-Quintela Laboratory of Magnetism and Nanotechnology, Institute for Technological Research, University of Santiago de Compostela, Santiago de Compostela, Spain Yang Lu Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Zheng-Hong Lu Department of Physics, Yunnan University, Yunnan, Kunming, PR China Department of Materials Science and Engineering, University of Toronto, Toronto, Ontario, Canada Michael S.-C. Lu Department of Electrical Engineering, Institute of Electronics Engineering, and Institute of NanoEngineering and MicroSystems, National Tsing Hua University, Hsinchu, Taiwan, Republic of China Wei Lu Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI, USA Jia Grace Lu Departments of Physics and Electrophysics, University of Southern California Office: SSC 215B, Los Angeles, CA, USA Vanni Lughi DI3 – Department of Industrial Engineering and Information Technology, University of Trieste, Trieste, Italy
xxxiv
Neville C. Luhmann Jr. Department of Electrical and Computer Engineering, University of California, Davis, USA Lorenzo Lunelli Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation and CNR-IBF, Povo, TN, Italy Richard F. Lyon Google Inc, Santa Clara, CA, USA Kuo-Sheng Ma IsaCal Technology, Inc., Riverside, CA, USA Marc Madou Department of Mechanical and Aerospace Engineering & Biomedical Engineering, University of California at Irvine, Irvine, CA, USA Daniele Malleo Fluxion Biosciences, South San Francisco, CA, USA Supone Manakasettharn Department of Mechanical Engineering, The University of Wisconsin-Madison, Madison, WI, USA Richard P. Mann Centre for Interdisciplinary Mathematics, Uppsala University, Uppsala, Sweden Liberato Manna Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Shengcheng Mao Institute of Microstructure and Property of Advanced Materials, Beijing University of Technology, Beijing, People’s Republic of China Francelyne Marano Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France Sylvain Martel NanoRobotics Laboratory, Department of Computer and Software Engineering, and Institute of Biomedical Engineering, E´cole Polytechnique de Montre´al (EPM), Montre´al, QC, Canada Pascal Martin University Paris 7-Denis Diderot, ITODYS, Nanoelectrochemistry Group, UMR CNRS 7086, Batiment Lavoisier, Paris cedex 13, France Paola Martino Politronica inkjet printing technologies S.r.l., Torino, Italy Armand Masion CEREGE, UMR 6635 CNRS/Aix–Marseille Universite´, Aix–en–Provence, France iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix– En–Provence, Cedex 4, France Daniel Maspoch CIN2 (ICN-CSIC), Catalan Institute of Nanotechnology, Bellaterra, Barcelona, Spain Cintia Mateo Departamento de Quı´mica Fı´sica, Universidade de Vigo, Vigo, Spain Shinji Matsui Laboratory of Advanced Science and Technology for Industry, University of Hyogo, Hyogo, Japan Theresa S. Mayer Department of Electrical Engineering and Materials Science and Engineering, The Pennsylvania State University, University Park, PA, USA Jeffrey S. Mayer Department of Electrical Engineering, Penn State University, University Park, PA, USA
Contributors
Contributors
xxxv
Chimaobi Mbanaso College of Nanoscale Science and Engineering, University at Albany, Albany, NY, USA Eva McGuire Department of Materials, Imperial College London, London, UK Andy C. McIntosh Energy and Resources Research Institute, University of Leeds, Leeds, West Yorkshire, UK Federico Mecarini Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Ernest Mendoza Centre de Recerca en Nanoenginyeria, Universitat Polite`cnica de Catalunya, Barcelona, Spain Christoph Menzel Institute of Condensed Matter Theory and Solid State Optics, Abbe Center of Photonics, Friedrich-Schiller-Universit€at Jena, Jena, Germany Timothy J. Merkel Department of Chemistry, University of North Carolina, Chapel Hill, NC, USA Vincent Meunier Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute, Troy, NY, USA Paul T. Mikulski Department of Physics, United States Naval Academy, Annapolis, MD, USA Hwall Min The Pennsylvania State University, University Park, PA, USA Rodolfo Miranda Dep. Fı´sica de la Materia Condensada, Universidad Auto´noma de Madrid and Instituto Madrilen˜o de Estudios Avanzados en Nanociencia (IMDEA-Nanociencia), Madrid, Spain Sushanta K. Mitra Micro and Nano-scale Transport Laboratory, Department of Mechanical Engineering, University of Alberta, Edmonton, AB, Canada Cristian Mocuta Synchrotron SOLEIL, L’Orme des Merisiers, Gif-sur-Yvette, France Mohammad R. K. Mofrad Molecular Cell Biomechanics Lab, Department of Bioengineering, University of California, Berkeley, CA, USA Seyed Moein Moghimi Centre for Pharmaceutical Nanotechnology and Nanotoxicology, Department of Pharmaceutics and Analytical Chemistry, University of Copenhagen, Copenhagen, Denmark Farghalli A. Mohamed Department of Chemical Engineering and Materials Science, University of California, Irvine The Henry Samueli School of Engineering, Irvine, CA, USA Nancy A. Monteiro-Riviere Center for Chemical Toxicological Research and Pharmacokinetics, North Carolina State University, Raleigh, NC, USA Philip Motta Department of Integrative Biology, University of South Florida, Tampa, FL, USA Florence Mouchet EcoLab – Laboratoire d’e´cologie fonctionnelle environnement, Universite´ de Toulouse, INP, UPS, Castanet Tolosan, France CNRS UMR 5245, EcoLab, Castanet Tolosan, France
et
xxxvi
Prachya Mruetusatorn Department of Electrical Engineering and Computer Science, University of Tennessee Knoxville, Knoxville, TN, USA Oakridge National Laboratory, Oak Ridge, TN, USA Weiqiang Mu Department of Physics and Astronomy, Northwestern University, Evanston, IL, USA Stefan M€ uhlig Institute of Condensed Matter Theory and Solid State Optics, Abbe Center of Photonics, Friedrich-Schiller-Universit€at Jena, Jena, Germany Partha P. Mukherjee Computer Science and Mathematics Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Bert M€ uller Biomaterials Science Center (BMC), University of Basel, Basel, Switzerland Claudia Musicanti Nanovector srl, Torino, Italy Jit Muthuswamy School of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ, USA Shrikant C. Nagpure Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA Vishal V. R. Nandigana Department of Mechanical Science and Engineering, Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana – Champaign, Urbana, IL, USA Avinash P. Nayak Electrical and Computer Engineering, University of California Davis Integrated Nanodevices & Nanosystems Lab, Davis, CA, USA Suresh Neethirajan School of Engineering, University of Guelph, Guelph, ON, Canada Celeste M. Nelson Department of Chemical and Biological Engineering, Princeton University, Princeton, NJ, USA Department of Molecular Biology, Princeton University, Princeton, NJ, USA Bradley J. Nelson Institute of Robotics and Intelligent Systems, ETH Zurich, Zurich, Switzerland Gilbert Daniel Nessim Chemistry department, Bar-Ilan Institute of Nanotechnology and Advanced Materials (BINA), Bar-Ilan University, Ramat Gan, Israel Daniel Neuhauser Department of Chemistry and Biochemistry, UCLA, Los Angeles, CA, USA J. Tanner Nevill Fluxion Biosciences, South San Francisco, CA, USA Nam-Trung Nguyen School of Mechanical and Aerospace Engineering, Nanyang Technological Univeristy, Singapore, Singapore Hossein Nili School of Electronics and Computer Science, University of Southampton, Highfield, Southampton, UK Nano Research Group, University of Southampton, Highfield, Southampton, UK
Contributors
Contributors
xxxvii
Vincent Niviere GDRI ICEINT: International Center for the Environmental Implications of Nanotechnology, CNRS–CEA, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France Laboratoire de Chimie et Biologie des Me´taux, UMR 5249, iRTSV–CEA Bat. K’, 17 avenue des Martyrs, Grenoble Cedex 9, France Michael Nosonovsky Department of Mechanical Engineering, University of Wisconsin-Milwaukee, Milwaukee, WI, USA Thomas Nowotny School of Informatics, University of Sussex, Falmer, Brighton, UK Seajin Oh Department of Mechanical and Aerospace Engineering & Biomedical Engineering, University of California at Irvine, Irvine, CA, USA Murat Okandan Advanced MEMS and Novel Silicon Technologies, Sandia National Laboratories, Albuquerque, NM, USA Brian E. O’Neill Department of Radiology Research, The Methodist Hospital Research Institute, Houston, TX, USA Takahito Ono Department of Mechanical Systems and Design, Graduate School of Engineering, Tohoku University, Aramaki, Aoba-ku Sendai, Japan Clifford W. Padgett Chemistry & Physics, Armstrong Atlantic State University, Savannah, GA, USA Christine Paille`s CEREGE, UMR 6635 CNRS/Aix–Marseille Universite´, Aix– en–Provence, France iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix– En–Provence, Cedex 4, France Nezih Pala Department of Electrical and Computer Engineering, Florida International University, Miami, FL, USA Manuel L. B. Palacio Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA Jeong Young Park Graduate School of EEWS (WCU), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea K. S. Park Department of Electrical Engineering, University of Washington, Seattle, WA, USA Kinam Park Departments of Biomedical Engineering and Pharmaceutics, Purdue University, West Lafayette, IN, USA Woo-Tae Park Seoul National University of Science and Technology, Seoul, Korea Andrew R. Parker Department of Zoology, The Natural History Museum, London, UK Green Templeton College, University of Oxford, Oxford, UK Alessandro Parodi Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA
xxxviii
Renata Pasqualini David H. Koch Center, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Laura Pasquardini Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation, Povo, TN, Italy Melissa A. Pasquinelli Fiber and Polymer Science, Textile Engineering, Chemistry and Science, North Carolina State University, Raleigh, NC, USA Siddhartha Pathak California Institute of Technology MC 309-81, Pasadena, CA, USA Cecilia Pederzolli Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation, Povo, TN, Italy Natalia Pelinovskaya CEREGE, UMR 6635, CNRS, Aix-Marseille Universite´, Aix en Provence, Cedex 04, France Ne´stor O. Pe´rez-Arancibia School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Dimitrios Peroulis School of Electrical and Computer Engineering, Birck Nanotechnology Center, Purdue University, West Lafayette, IN, USA Vinh-Nguyen Phan School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore, Singapore Reji Philip Light and Matter Physics Group, Raman Research Institute, Sadashivanagar, Bangalore, India Andrew Philippides Centre for Computational Neuroscience and Robotics, University of Sussex, Brighton, UK Gianluca Piazza Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, USA Remigio Picone Dana-Farber Cancer Institute – Harvard Medical School, David Pellman Lab – Department of Pediatric Oncology, Boston, MA, USA Ilya V. Pobelov Department of Chemistry and Biochemistry, University of Bern, Bern, Switzerland Ryan M. Pocratsky Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, USA Ramakrishna Podila Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Martino Poggio Department of Physics, University of Basel, Basel, Switzerland R. G. Polcawich US Army Research Laboratory RDRL-SER-L, Adelphi, MD, USA Jan Pomplun Zuse Institute Berlin, Berlin, Germany Alexandra Porter Department of Materials, Imperial College London, London, UK Cristina Potrich Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation and CNR-IBF, Povo, TN, Italy
Contributors
Contributors
xxxix
Siavash Pourkamali Department of Electrical and Computer Engineering, University of Denver, Denver, CO, USA Shaurya Prakash Department of Mechanical and Aerospace Engineering, The Ohio State University, Columbus, OH, USA Luigi Preziosi Dipartimento di Matematica, Politecnico di Torino, Torino, Italy Luca Primo University of Torino, Department of Clinical and Biological Sciences, Candiolo, Italy R. M. Proie US Army Research Laboratory RDRL-SER-E, Adelphi, MD, USA Olivier Proux GDRI ICEINT: International Center for the Environmental Implications of Nanotechnology, CNRS–CEA, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France OSUG, Universite´ Joseph Fourier BP 53, Grenoble, France Pascal Puech CEMES, Toulouse Cedex 4, France Robert Puers Department ESAT-MICAS, KULeuven, Leuven, Belgium J. S. Pulskamp US Army Research Laboratory RDRL-SER-L, Adelphi, MD, USA Yongfen Qi The Key Laboratory of Remodeling-related Cardiovascular Diseases, Capital Medical University, Ministry of Education, Beijing, China Qiquan Qiao Department of Electrical Engineering and Computer Science, South Dakota State University, Brookings, USA Aisha Qi Micro/Nanophysics Research Laboratory, RMIT University, Melbourne, VIC, Australia Yabing Qi Energy Materials and Surface Sciences (EMSS) Unit, Okinawa Institute of Science and Technology, Kunigami-gun, Okinawa, Japan Hongwei Qu Department of Electrical and Computer Engineering, Oakland University, Rochester, MI, USA Marzia Quaglio Center for Space Human Robotics, Fondazione Istituto Italiano di Tecnologia, Torino, Italy Regina Ragan The Henry Samueli School of Engineering, Chemical Engineering and Materials Science University of California, Irvine, Irvine, CA, USA Melur K. Ramasubramanian Department of Mechanical & Aerospace Engineering, North Carolina State University, Raleigh, NC, USA Antonio Ramos Departamento de Electro´nica y Electromagnetismo, Universidad de Sevilla, Sevilla, Spain Hyacinthe Randriamahazaka University Paris 7-Denis Diderot, ITODYS, Nanoelectrochemistry Group, UMR CNRS 7086, Batiment Lavoisier, Paris cedex 13, France Apparao M. Rao Department of Physics and Astronomy, Clemson University, Clemson, SC, USA
xl
Center for Optical Materials Science & Engineering Technologies, Clemson University, Clemson, SC, USA E. Reina-Romo School of Engineering, University of Seville, Seville, Spain Ste´phane Re´gnier Institut des Syste`mes Intelligents et de Robotique, Universite´ Pierre et Marie Curie, CNRS UMR7222, Paris, France Philippe Renaud Microsystems Laboratory, Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland Tian-Ling Ren Institute of Microelectronics, Tsinghua University, Beijing, China Scott T. Retterer Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Department of Electrical Engineering and Computer Science, University of Tennessee Knoxville, Knoxville, TN, USA Roberto de la Rica MESA + Institute for Nanotechnology, University of Twente, Enschede, The Netherlands Agneta Richter-Dahlfors Swedish Medical Nanoscience Center, Department of Neuroscience, Karolinska Institutet, Stockholm, Sweden Miche`le Riesen Department of Genetics Evolution and Environment, Institute of Healthy Ageing, University College London, London, UK Matteo Rinaldi Department of Electrical and Computer Engineering, Northeastern University, Boston, MA, USA Aditi Risbud Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, CA, USA Jose´ Rivas Laboratory of Magnetism and Nanotechnology, Institute for Technological Research, University of Santiago de Compostela, Santiago de Compostela, Spain INL – International Iberian Nanotechnology Laboratory, Braga, Portugal Paola Rivolo Dipartimento di Scienza dei Materiali e Ingegneria Chimica – Politecnico di Torino, Torino, Italy Stephan Roche CIN2 (ICN–CSIC), Catalan Institute of Nanotechnology, Universidad Auto´noma de Barcelona, Bellaterra (Barcelona), Spain Institucio´ Catalana de Recerca i Estudis Avanc¸ats (ICREA), Barcelona, Spain Carsten Rockstuhl Institute of Condensed Matter Theory and Solid State Optics, Abbe Center of Photonics, Friedrich-Schiller-Universit€at Jena, Jena, Germany Fernando Rodrigues-Lima Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France
Contributors
Contributors
xli
Brian J. Rodriguez Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Belfield, Dublin 4, Ireland Je´roˆme Rose CEREGE UMR 6635– CNRS–Universite´ Paul Ce´zanne Aix–Marseille III, Aix–Marseille Universite´, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France GDRI ICEINT: International Center for the Environmental Implications of Nanotechnology, CNRS–CEA, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France Yitzhak Rosen Superior NanoBiosystems LLC, Washington, DC, USA M. Rossi Institute of Fluid Mechanics and Aerodynamics, Universit€at der Bundeswehr M€ unchen, Neubiberg, Germany Marina Ruths Department of Chemistry, University of Massachusetts Lowell, Lowell, MA, USA Kathleen E. Ryan Department of Chemistry, United States Naval Academy, Annapolis, MD, USA Malgorzata J. Rybak-Smith Department of Pharmacology, University of Oxford, Oxford, UK V. Sai Muthukumar Department of Physics, Sri Sathya Sai Institute of Higher Learning, Prashanthinilayam, Andhra Pradesh, India Vero´nica Salgueirino Departamento de Fı´sica Aplicada, Universidade de Vigo, Vigo, Spain Meghan E. Samberg Center for Chemical Toxicological Research and Pharmacokinetics, North Carolina State University, Raleigh, NC, USA Florence Sanchez Department of Civil and Environmental Engineering, Vanderbilt University, Nashville, USA Catherine Santaella iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France Laboratoire d’Ecologie Microbienne de la Rhizosphe`re et d’Environnements Extreˆme, UMR 6191 CNRS–CEA–Aix–Marseille Universite´ de la Me´diterrane´e, St Paul lez Durance, France J. A. Sanz-Herrera School of Engineering, University of Seville, Seville, Spain Stephen Andrew Sarles Mechanical Aerospace and Biomedical Engineering, University of Tennessee, Knoxville, TN, USA J. David Schall Department of Mechanical Engineering, Oakland University, Rochester, MI, USA George C. Schatz Department of Chemistry and International Institute for Nanotechnology, Northwestern University, Evanston, IL, USA Andre´ Schirmeisen Institute of Applied Physics, Justus-Liebig-University Giessen, Giessen, Germany
xlii
Frank Schmidt Zuse Institute Berlin, Berlin, Germany Helmut Schmitz Institute of Zoology, University of Bonn Poppelsdorfer Schloss, Bonn, Germany Scott R. Schricker College of Dentistry, The Ohio State University, Columbus, OH, USA Georg Schulz Biomaterials Science Center (BMC), University of Basel, Basel, Switzerland Udo D. Schwarz Department of Mechanical Engineering, Yale University, New Haven, USA Praveen Kumar Sekhar Electrical Engineering, School of Engineering and Computer Science, Washington State University Vancouver, Vancouver, WA, USA Rita E. Serda Department of NanoMedicine, The Methodist Hospital Research Institute, Houston, TX, USA Nika Shakiba Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada Karthik Shankar Department of Electrical and Computer Engineering, W2-083 ECERF, University of Alberta, Edmonton, AB, Canada Yunfeng Shi Department of Materials Science and Engineering, MRC RM114, Rensselaer Polytechnic Institute, Troy, NY, USA Li Shi Department of Mechanical Engineering, The University of Texas at Austin, Austin, TX, USA Youngmin Shin Department of Electrical and Computer Engineering, University of California, Davis, USA Wolfgang M. Sigmund Department of Materials Science and Engineering, University of Florida, Gainesville, FL, USA Department of Energy Engineering, Hanyang University, Seoul, Republic of Korea S. Ravi P. Silva Nano Electronics Center, Advanced Technology Institute, University of Surrey, Guildford, Surrey, UK Nipun Sinha Department of Mechanical Engineering and Applied Mechanics, Penn Micro and Nano Systems (PMaNS) Lab, University of Pennsylvania, Philadelphia, PA, USA S. Siva Sankara Sai Department of Physics, Sri Sathya Sai Institute of Higher Learning, Prashanthinilayam, Andhra Pradesh, India Dunja Skoko The Naughton Institute, School of Physics, Trinity College Dublin, CRANN, Dublin, Ireland Craig Snoeyink Birck Nanotechnology Center, Mechanical Engineering, Purdue University, West Lafayette, IN, USA Konstantin Sobolev Department of Civil Engineering and Mechanics, University of Wisconsin-Milwaukee, Milwaukee, WI, USA
Contributors
Contributors
xliii
Helmut Soltner Zentralabteilung Technologie, Forschungszentrum J€ulich GmbH, J€ ulich, Germany Thomas Søndergaard Department of Physics and Nanotechnology, Aalborg University, Aalborg Øst, Denmark Youngjun Song Department of Electrical and Computer Engineering, University of California San Diego, San Diego, CA, USA Anil K. Sood Gynecologic Oncology, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Cancer Biology, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA Center for RNA Interference and Non–coding RNA, M.D. Anderson Cancer Center, The University of Texas, Houston, TX, USA The Department of Nanomedicine and Bioengineering, UTHealth, Houston, TX, USA Pratheev S. Sreetharan School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Bernadeta Srijanto Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Tomasz Stapinski Department of Electronics, AGH University of Science and Technology, Krakow, Poland Ullrich Steiner Department of Physics, Cavendish Laboratories, University of Cambridge, Cambridge, UK Michael S. Strano Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA Arunkumar Subramanian Department of Mechanical and Nuclear Engineering, Virginia Commonwealth University, Richmond, VA, USA Maxim Sukharev Department of Applied Sciences and Mathematics, Arizona State University, Mesa, AZ, USA Bobby G. Sumpter Computer Science and Mathematics Division and Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA Yu Sun Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering and Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada Tao Sun Research Laboratory of Electronics, Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, USA Vishnu-Baba Sundaresan Mechanical and Nuclear Engineering, Virginia Commonwealth University, Richmond, VA, USA Srinivas Tadigadapa Department of Electrical Engineering, The Pennsylvania State University, University Park, PA, USA
xliv
Saikat Talapatra Department of Physics, Southern Illinois University Carbondale, Carbondale, IL, USA Qingyuan Tan Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada Hiroto Tanaka School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Xinyong Tao College of Chemical Engineering and Materials Science, Zhejiang University of Technology, Hangzhou, China Ennio Tasciotti Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USA J. Ashley Taylor Department of Mechanical Engineering, The University of Wisconsin-Madison, Madison, WI, USA Raviraj Thakur School of Mechanical Engineering and Birck Nanotechnology Center, Purdue University, West Lafayette, IN, USA Antoine Thill iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France Laboratoire Interdisciplinaire sur l’Organisation Nanome´trique et Supramole´culaire, UMR 3299 CEA/CNRS SIS2M, Gif–surYvette, France Alain Thie´ry iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France IMEP, UMR 6116 CNRS/IRD, Aix–Marseille Universite´, Marseille, Cedex 03, France Thomas Thundat Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB, Canada Katarzyna Tkacz–Smiech Faculty of Materials Science and Ceramics, AGH University of Science and Technology, Krakow, Poland Steve To Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada Gerard Tobias Institut de Cie`ncia de Materials de Barcelona (ICMAB-CSIC), Bellaterra, Barcelona, Spain Andrea Toma Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Katja Tonisch Institut f€ ur Mikro- und Nanotechnologien, Technische Universit€at Ilmenau Fachgebiet Nanotechnologie, Ilmenau, Germany Elka Touitou Institute of Drug Research, School of Pharmacy, The Hebrew University of Jerusalem, Jerusalem, Israel Lesa A. Tran Department of Chemistry and the Richard E. Smalley Institute for Nanoscale Science and Technology, Rice University, Houston, TX, USA
Contributors
Contributors
xlv
Alexander A. Trusov Department of Mechanical and Aerospace Engineering, The University of California, Irvine, CA, USA Soichiro Tsuda Exploratory Research for Advanced Technology, Japan Science and Technology Agency, Osaka, Japan Lorenzo Valdevit Department of Mechanical and Aerospace Engineering, University of California, Irvine, CA, USA Ana Valero Microsystems Laboratory, Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland Pablo Varona Dpto. de Ingenieria Informatica, Universidad Auto´noma de Madrid, Madrid, Spain Amadeo L. Va´zquez de Parga Dep. Fı´sica de la Materia Condensada, Universidad Auto´noma de Madrid and Instituto Madrilen˜o de Estudios Avanzados en Nanociencia (IMDEA-Nanociencia), Madrid, Spain K. Venkataramaniah Department of Physics, Sri Sathya Sai Institute of Higher Learning, Prashanthinilayam, Andhra Pradesh, India Nuria Vergara-Irigaray Department of Genetics Evolution and Environment, Institute of Healthy Ageing, University College London, London, UK Georgios Veronis Department of Electrical and Computer Engineering and Center for Computation and Technology, Louisiana State University, Baton Rouge, LA, USA Alexey N. Volkov Department of Materials Science and Engineering, University of Virginia, Charlottesville, VA, USA Frank Vollmer Laboratory of Biophotonics and Biosensing, Max Planck Institute for the Science of Light, Erlangen, Germany Fritz Vollrath Department of Zoology, University of Oxford, Oxford, UK Prashant R. Waghmare Micro and Nano-scale Transport Laboratory, Department of Mechanical Engineering, University of Alberta, Edmonton, AB, Canada Hermann Wagner Institute for Biology II, RWTH Aachen University, Aachen, Germany Richard Walker ICON plc, Marlow, Buckinghamshire, UK Thomas Wandlowski Department of Chemistry and Biochemistry, University of Bern, Bern, Switzerland Wenlong Wang Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing, P. R. China Feng-Chao Wang State Key Laboratory of Nonlinear Mechanics (LNM), Institute of Mechanics, Chinese Academy of Sciences, Beijing, China Yu-Feng Wang Institute of Microelectronics, Tsinghua University, Beijing, China Szu-Wen Wang Chemical Engineering and Materials Science, The Henry Samueli School of Engineering, University of California, Irvine, CA, USA
xlvi
Chunlei Wang Department of Mechanical and Materials Engineering, Florida International University, Miami, FL, USA Zhibin Wang Department of Materials Science and Engineering, University of Toronto, Toronto, Ontario, Canada Enge Wang International Center for Quantum Materials, School of Physics, Peking University, Beijing, China Guoxing Wang School of Microelectronics, Shanghai Jiao Tong University (SJTU), Minhang, Shanghai P R, China Zhong Lin Wang School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA, USA Reinhold Wannemacher Madrid Institute for Advanced Studies, IMDEA Nanociencia, Madrid, Spain Benjamin L. J. Webb Division of Infection & Immunity, University College London, London, UK Liu Wei CEREGE UMR 6635-CNRS, Aix-Marseille Universite´, Europoˆle de l’Arbois, Aix-en-Provence, France Michael Weigel-Jech Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany Steven T. Wereley School of Mechanical Engineering, Birck Nanotechnology Center, Purdue University, West Lafayette, IN, USA Tad S. Whiteside Savannah River National Laboratory, Aiken, SC, USA John P. Whitney School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA Samuel A. Wickline C-TRAIN Labs, Washington University, St. Louis, MO, USA Mark Wiesner iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France CEINT, Center for the Environmental Implications of NanoTechnology, Duke University, Durham, NC, USA Stuart Williams Department of Mechanical Engineering, University of Louisville, Louisville, KY, USA Kerry Allan Wilson London Centre For Nanotechnology, University College London, London, UK Lon J. Wilson Department of Chemistry and the Richard E. Smalley Institute for Nanoscale Science and Technology, Rice University, Houston, TX, USA Patrick M. Winter Department of Radiology, Imaging Research Center, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA Stephen TC Wong Center for Bioengineering and Informatics, Department of Systems Medicine and Bioengineering, The Methodist Hospital Research Institute, Weill Cornell Medical College, Houston, TX, USA
Contributors
Contributors
xlvii
Robert J. Wood School of Engineering and Applied Sciences, Wyss Institute for Biologically Inspired Engineering, Harvard University, Cambridge, MA, USA Matthew Wright NanoSight Limited, Amesbury, Wiltshire, UK Wei Wu Department of Materials Science, Fudan University, Shanghai, China Xiao Yu Wu Advanced Pharmaceutics and Drug Delivery Laboratory, University of Toronto, Toronto, ON, Canada H. Xie The State Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, China Guoqiang Xie Institute for Materials Research, Tohoku University, Sendai, Japan Huikai Xie Department of Electrical and Computer Engineering, University of Florida, Gainesville, FL, USA Tingting Xu Department of Electrical Engineering and Computer Science, South Dakota State University, Brookings, USA Didi Xu Institute of Robotics and Intelligent Systems, ETH Zurich, Zurich, Switzerland Christophe Yamahata Microsystem Lab., Ecole Polytechnique Fe´de´rale de Lausanne, Lausanne, Switzerland Keqin Yang Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Yuehai Yang Department of Physics, Florida International University, Miami, FL, USA Yi Yang Institute of Microelectronics, Tsinghua University, Beijing, China Chun Yang Division of Thermal Fluids Engineering, School of Mechanical and Aerospace Engineering, School of Chemical and Biomedical Engineering (joint appointment), Nanyang Technological University, Singapore Yoke Khin Yap Department of Physics, Michigan Technological University, Houghton, MI, USA Leslie Yeo Micro/Nanophysics Research Laboratory, RMIT University, Melbourne, VIC, Australia Zheng Yin Center for Bioengineering and Informatics, Department of Systems Medicine and Bioengineering, The Methodist Hospital Research Institute, Weill Cornell Medical College, Houston, TX, USA Yaroslava G. Yingling Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA Minami Yoda G. W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USA Sang-Hee Yoon Department of Mechanical Engineering, University of California, Berkeley, CA, USA Molecular Cell Biomechanics Lab, Department of Bioengineering, University of California, Berkeley, CA, USA
xlviii
Lidan You Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada Yanlei Yu Department of Materials Science, Fudan University, Shanghai, China Wei Yu The Key Laboratory of Remodeling-related Cardiovascular Diseases, Capital Medical University, Ministry of Education, Beijing, China Seok H. Yun Graduate School of Nanoscience and Technology, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea Wellman Center for Photomedicine, Department of Dermatology, Harvard Medical School and Massachusetts General Hospital, Boston, MA, USA The Harvard–Massachusetts Institute of Technology Division of Health Science and Technology, Cambridge, MA, USA Remo Proietti Zaccaria Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy Li Zhang Institute of Robotics and Intelligent Systems, ETH Zurich, Zurich, Switzerland Ze Zhang Institute of Microstructure and Property of Advanced Materials, Beijing University of Technology, Beijing, People’s Republic of China State Key Laboratory of Silicon Materials and Department of Materials Science and Engineering, Zhejiang University, Hangzhou, China Yi Zhang Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai, China Jin Z. Zhang Department of Chemistry and Biochemistry, University of California, Santa Cruz, CA, USA Xiaobin Zhang Department of Materials Science and Engineering, Zhejiang University, Hangzhou, China John Xiaojing Zhang Department of Biomedical Engineering, The University of Texas at Austin, Austin, TX, USA Jianqiang Zhao Department of Materials Science, Fudan University, Shanghai, China Ya-Pu Zhao State Key Laboratory of Nonlinear Mechanics (LNM), Institute of Mechanics, Chinese Academy of Sciences, Beijing, China Jinfeng Zhao Department of Electrical and Computer Engineering, University of California, Davis, USA Leonid V. Zhigilei Department of Materials Science and Engineering, University of Virginia, Charlottesville, VA, USA Ying Zhou Department of Electrical & Computer Engineering, University of Florida, Gainesville, FL, USA
Contributors
Contributors
xlix
Menghan Zhou Department of Physics and Astronomy, Clemson University, Clemson, SC, USA Yimei Zhu Center for Functional Nanomaterials, Brookhaven National Lab, Upton, NY, USA Wenguang Zhu Department of Physics and Astronomy, The University of Tennessee, Knoxville, TN, USA Rashid Zia School of Engineering, Brown University, Providence, RI, USA
0–9
mMist ®
Definition
▶ Spray Technologies Inspired by Bombardier Beetle
A micromanipulation system is a robotic device which is used to physically interact with a sample under a microscope, where a level of precision of movement is necessary that cannot be achieved by the unaided human hand It is well known that pick-and-place is very important for 3-D microstructure fabrication since it is an indispensable step in the bottom-up building process. The main difficulty in sufficiently completing such pick-and-place manipulation at this scale lies in fabricating a very sharp end-effector that is capable of smoothly releasing microobjects deposited on the substrate. Moreover, this end-effector has to provide enough grasping force to overcome strong adhesion forces [1–3] from the substrate as well as being capable of sensing and controlling interactions with the microobjects. Furthermore, compared with the manipulation of larger microobjects under an optical microscope, visual feedback at several microns more suffers from the shorter depth of focus and the narrower field of view of lenses with high magnifications, although different schemes or algorithms have been introduced on techniques of autofocusing [4, 5] and extending focus depth [7]. Compared with vision-based automated 2-D micromanipulation, automated 3-D micromanipulation at the scale of several microns to submicron scale is more challenging because of optical microscope’s resolution limit (typically 200 nm). Moreover, additional manipulation feedback is needed that is beyond the capability of optical vision, such as in the cases of vertical contact detection along the optical axis or manipulation obstructed by opaque components. Therefore, multi-feedback is of vital
mVED (Micro Vacuum Electronic Devices) ▶ MEMS Vacuum Electronics
193-nm Lithography ▶ DUV Photolithography and Materials
248-nm Lithography ▶ DUV Photolithography and Materials
3D Micro/Nanomanipulation with Force Spectroscopy H. Xie1 and Ste´phane Re´gnier2 1 The State Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, China 2 Institut des Syste`mes Intelligents et de Robotique, Universite´ Pierre et Marie Curie, CNRS UMR7222, Paris, France
Synonyms Force spectroscopy; Nano manipulation; Nanorobotics
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
0–9
2
3D Micro/ Nanomanipulation with Force Spectroscopy, Fig. 1 (a) A photo of the AFM-FRS on the configuration for nanoscale pick-and-place manipulation. (b) A SEM image of the cantilever fabricated with a protruding tip
3D Micro/Nanomanipulation with Force Spectroscopy
a lasers
optical microscope
photodiodes
piezoscanner
b tip I & II
motorized stage
manual stage nanostage
side view
importance to achieve such accurate and stable 3-D micromanipulation at the scale of several microns to submicron scale. Pick-and-place nanomanipulation is also a promising technique in 3D nanostructure fabrication since it is an indispensable step in the bottom-up building process. It can overcome limitations of bottom-up and top-down methods of nanomanufacturing and further combine advantages of these two methods to build complex 3D nanostructures. In literature, nanostructures have been manipulated, assembled, and characterized by integrating nanomanipulators or nanogrippers into scanning electron microscopes (SEM) and transmission electron microscopes (TEM) [12–16]. Both the SEM and the TEM provide a vacuum environment where the van der Waals force is the main force to be overcome during the manipulation. 3D nanomanipulation could be also achieved with optical tweezers in liquid, where the adhesion forces are greatly reduced [17–19]. However, the pick-and-place nanomanipulation in air is still a great challenge due to the presence of strong adhesion forces, including van der Waals, electrostatic and capillary forces [20]. In this case, the main difficulties in achieving the 3D nanomanipulation are fabricating sharp end-effectors with enough grasping force, as well as capabilities of force sensing while controlling interactions between the nanoobject and the tool or the substrate. In this entry, a concept is presented to make a prototype of an AFM-based flexible robotic system (FRS) which is equipped with two collaborative cantilevers. By reconfiguring the modular hardware and software of the AFM-FRS, different configurations can be obtained. With different manipulation strategies
5 μm
and protocols, two of the configurations can be respectively used to complete pick-and-place manipulations from the microscale of several microns to the nanoscale that are still challenges. This entry is organized as follows. Section I introduces the prototype and experimental setup of the AFM-FRS. In subsection II and III, pick-and-place manipulation of microspheres and nanowires for building 3-D micro/nano structures are performed using the AFM-FRS.
AFM-Based Flexible Robotic System AFM-FRS Setup Figure 1a shows the AFM-FRS system setup, which is in the configuration for nanoscale pick-and-place. The system is equipped with an optical microscope and two sets of modules commonly used in a conventional AFM, mainly including two AFM cantilevers (namely, tip I and tip II, ATEC-FM Nanosensors, see Fig. 1b), two sets of nanopositioning devices and optical levers. The motion modules include an open-loop X-Y-Z piezoscanner (PI P-153.10H), an X-Y-Z closed-loop nanostage (MCL Nano-Bio2M on the X- and 7-axis, PI P-732. ZC on the Z-axis), an X-Y-Z motorized stage, and an X-Y-Z manual stage. Detailed specifications of the motion modules are summarized in Table 1. Figure 1c shows an optical microscope image of the collaborating tips in microsphere grasping mode. A data acquisition (NI 6289) card is used for highspeed (500800 Hz of sampling frequency for force and 600 kHz for amplitude) capture of the photodiode voltage output to estimate deflections on both tips induced by force loading or resonant oscillation.
3D Micro/Nanomanipulation with Force Spectroscopy 3D Micro/Nanomanipulation with Force Spectroscopy, Table 1 Specifications of each motion module Actuator X-Y-Z piezoscanner X-Y-Z nanostage X-Y-Z motorized stage X-Y-Z manual stage
Travel range 10 10 10(mm) 50 50 10(mm) 25 25 25(mm) 5 5 5(mm)
l
zn1 laser spot
h
z
zn1 ¼
Fz1 cos g þ Fx1 sin g Fz1 sin g þ Fx1 cos g þ kn kxz (1)
where g is the mounting angle of the cantilever, kxz ¼ 2lkn/3h is the bending stiffness due to the moment applied on the tip end, where h is the tip height. Assuming the magnitude of Fz1 and Fx1 are of the same order, contributions from the Fx1 to the normal deflection of tip I are relatively very small since kxz kn and g ¼ 5 . Therefore, the normal deflection induced from Fz1 is only considered in the following calculations of the adhesion force Fa applied on the micro/nanoobject. Thus, Fz1 can be simplified by estimating the normal voltage output DVn1 from the tip I:
ΔVn
q Fz1
tip l Ff1
Force Sensing During Pick-and-Place Figure 2 shows a schematic diagram of the nanotip gripper for the micro/nanoscale pick-and-place operation that has a clamping angle y 44 micro/nanoscale pickup manipulation, for instance, interactive forces applied on tip I include repulsive forces Fr1, friction forces Ff1, and adhesive forces Fa1. The forces applied on tip I can be resolved into two components on the X-axis and the Z-axis in the defined frame, namely, Fx1 and Fz1, respectively. Fx1 is the clamping force that holds the micro/nanoobject. Fz1 is the pickup force that balances adhesion forces from the substrate. To sense the pickup force, it is necessary to know the normal deflection on both cantilevers. The normal deflection zn1 associated with the normal voltage output of the optical lever on tip I is given by:
0–9
photodiode
g
Resolution Sub-nm 0.1 nm 50 nm 0.5 mm
A multi-thread planning and control system based on the C++ is developed for AFM image scan and two-tip coordination control during manipulation. This control system enables programming of complex tasks on the highly distributed reconfigurable system.
0–9
3
o
micro/nanoobject
Fr1
Fx1
substrate
x Fa1
Fa
v
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 2 Schematic diagram of the nanotip gripper and force simulation during the pickup manipulation
Fz1 ¼ b1 DVn1
(2)
where b1 is the normal force sensitivity of the optical lever. A similar pickup force Fz2 can be also obtained on Tip II. Before the gripper pulls off the substrate, the adhesion force Fa can be estimated as: Fa ¼ Fz1 þ Fz2 ¼ b1 DVn1 þ b2 DVn2 ðFa1 þ Fa2 Þ (3) where b2, DVn2, and Fa2 are respectively the normal force sensitivity, normal voltage output, and adhesive force on tip II. Once the gripper pulls off the substrate, e.g., in the case of nanowire/tube pick-and-place, the adhesion force Fa is estimated as: Fa ¼ Fz1 þ Fz2 ¼ b1 DVn1 þ b2 DVn2 :
(4)
Experimental Results 3-D Micromanipulation Robotic System System Configuration for 3-D Micromanipulation As the size of microobjects is reduced to several microns or submicrons, problems will arise with these conventional grippers: (1) Sticking phenomena becomes more severe due to the relatively larger contact area between the gripper and the microobject. (2) The tip diameters of the micro-fabricated clamping jaws are comparable in size to the microobjects to be grasped. Conventional grippers are not geometrically
4
3D Micro/Nanomanipulation with Force Spectroscopy
a
b y
tip I
zM motorized stage
sample platform
zn
xM nonostage
z
zp
piezo scanner
x
xp
xn
grasping point searching
x o grasping point
y
zm ym
yn
y o
tip trace
tip II yp
yM
amplitude
optical microscope
xm
Kinematic configuration
x
cantilever
o microsphere z
manual stage
tip trace contact detection
amplitude
0–9
contact
x o
Schematic diagrams
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 3 A kinematic configuration and schematic diagrams for 3-D micromanipulation at a scale of several micrometers or
grasping point searching (inset I) and contact detection (inset II) with amplitude feedback of the dithering cantilever
sharp enough to pick up microobjects of several micrometers deposited on the substrate. Fortunately, the AFM tip has a very tiny apex (typically 10 nm in radius) with respect to the size of the microobject to be manipulated. Thus, the nanotip gripper can be used to achieve pick-and-place at the scale of several microns since the contact area of the gripper-microobject is much smaller than the microobject-substrate contact. Moreover, real-time force sensing makes the manipulation more controllable. As shown in Fig. 3, the system configuration for 3-D micromanipulation is reconfigured as follows: 1. For a large manipulation travel range, the nanostage here is used to support the sample platform and transport the microobject during the manipulation. 2. Tip I, immovable during the pick-and-place micromanipulation, is fixed on the motorized stage for coarse positioning. 3. Tip II is actuated by the piezoscanner for gripper opening and closing operations. The piezoscanner is supported by the manual stage for coarse positioning. Benefiting from AFM-based accurate and stable amplitude feedback of a dithering cantilever, the grasping state can be successfully achieved by the amplitude feedback, with very weak interaction at the nano-Newton scale, protecting the fragile tips and the microobjects from damage during manipulation. As shown in inset I of Fig. 3, the dithering cantilever with its first resonant mode is used to locate
grasping points and detect contact. When approaching the microsphere with a separation between the tip and the substrate (typically 500 nm), the tip laterally sweeps the microsphere over the lower part of the microsphere. By this means, the grasping point can be accurately found by locating the minimum amplitude response of each single scan. From the scheme depicted in inset II of Fig. 3, the amplitude feedback is also used for contact detection. Tip-microsphere contact is detected as the amplitude reduces to a steady value close to zero. As shown in Fig. 4, a protocol for pick-and-place microspheres mainly consists of four steps: • System initialization and task planning: Each axis of the nanostage and the piezoscanner are set in a proper position, supplying the manipulation with enough travel range on each axis. Then the task is planned in Fig. 4a with a global view of the manipulation area that provides coarse positions of the microspheres and tips. • Making tip I-microsphere in contact: In Fig. 4b, tip I has started to approach the microsphere by moving the nanostage with amplitude feedback to search for the grasping point and detect contact. • Forming the gripper: Similarly, tip II approaches the microsphere by moving the piezoscanner. Once tip II and the microsphere are in contact, a nanotip gripper is configured in Fig. 4c for a manipulation. • Pick-and-place micromanipulation: In Fig. 4d, the microsphere is picked up, transported, and released
3D Micro/Nanomanipulation with Force Spectroscopy
a
b
5
0–9 0–9
e tip I
tip I
tip II
tip II Δyn Δxn
c
d tip I
tip II Δyn
Δyp Δxp
Δzn
tip I
tip II
Δxn
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 4 Microsphere manipulation protocol. (a) Task planning. (b) Tip I and the microsphere are in contact. (c) The nanotip
gripper is formed. (d) Pick up and release the microsphere to its target position. (e) Task descriptions of the microsphere assembly
by moving the nanostage with a proper displacement on each axis that depends on the diameter of the microsphere and its destination. The whole process of 3-D micromanipulation is monitored by real-time force sensing.
In this curve, point A represents the start of the pick-and-place, point B the pull-off location of the microsphere-substrate contact, point C nonlinear force restitution due to the tip-microsphere frictions, and point D the snap-in point between the microsphere and the substrate. The force spectroscopy curve is synthesized from force responses on tip I and tip II. Figure 6 shows an automated microassembly result consisting of two 3-D micropyramids and a 2-D pattern of a regular hexagon. The whole manipulation process was completed in 11 min, so the average manipulation time for each microsphere is about 47 s, which mainly breaks down into about 20 s for microsphere grasping including the grasping point search and contact detection processes using amplitude feedback, 1035 s for microsphere release and, the remaining time for transport. Assembly of the fourth is the key to success in building a micropyramid. During pick-and-place of the fourth microsphere, microscopic vision was firstly used for coarse positioning it target, then the normal force feedback of the gripper was used to detect the vertical contact between the fourth microsphere and other three microspheres on the base. When the contact is established, a small vertical force was applied on the fourth microsphere by moving the nanostage upward and it will adjust to contact with all the base microspheres.
3-D Microsphere Assembly
Nylon microspheres with diameter of about 3 4 mm were manipulated to build 3-D microstructures in experiments. The microspheres were deposited on a freshly cleaned glass slide and then an area of interest was selected under the optical microscope. Figure 4 shows a plan view of the selected area, in which 14 microspheres separated in a 50 mm square frame are going to be manipulated to build two 3-D micropyramids and a regular 2-D hexagon labeled by assembly sequences from 1 to 8. Each pyramid is constructed from four microspheres with two layers and the assembly sequences are shown in the bottom insets for two different arrangements of the pyramids. Figure 5 shows a result of grasping point searching, in which the dithering tip II laterally sweeps the microsphere within a range of 1.75 mm on the y-axis and with a free oscillating amplitude of about 285 nm. Ten different distances to the microsphere were tested from 100 to 10 nm with an interval of 10 nm and, consequently, the grasping point is well located with an accuracy of 10 nm. Figure 5 shows a full force spectroscopy curve during the pick-and-place of a microsphere deposited on a glass slide with an ambient temperature of 20 C and relative humidity of 40%.
3-D Nanomanipulation Robotic System System Configuration for 3-D Nanomanipulation Compared with the 3-D micromanipulation, tip alignment precision is the key factor in succeeding the
0–9 a
6
3D Micro/Nanomanipulation with Force Spectroscopy
b
350 100 nm
0
A
Pickup Release
−100
250
−200 Force [nN]
Amplitude [nm]
300
200 150
−300 −400
D C
−500
100
−600 50 0
10 nm −0.75 −0.50 −0.25
0.00 0.25 0.50 Lateral scan displacement [µm]
−700
B −1.2
0.75
−1.0
−0.8
−0.6
−0.4
−0.2
0.0
Displacement [µm]
Amplitude responses on tip II when searching the grasping points
Synthesized normal force responses from both the tips in the pick-and-place micromanipulation
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 5 Experimental results
a
3D Micro/ Nanomanipulation with Force Spectroscopy, Fig. 6 A microassembly result
c
b
10 μm
Before the microassembly
nanoscale pick-and-place. Therefore, the closed-loop nanostage is considered in this configuration for accurate tip alignment. As shown in Fig. 7, the system configuration for 3-D nanomanipulation is reconfigured as follows: 1. The nanostage is used for image scan with tip I. 2. Nanoobjects are supported and transported by the piezoscanner. 3. Tip I, fixed on the motorized stage for coarse positioning, is immovable during the pick-and-place micromanipulation. Before manipulation, cantilever I acts as an image sensor for nanoobject positioning. 4. For accurate gripper alignment between Tip I and Tip II, Tip II is fixed on the nanostage rather than the piezoscanner. Tip II is supported by the manual stage for coarse positioning.
Micropyramids were built
Enlarged image of the microassembly result under a magnification of 100×
Nanowires and nanotubes are being intensively investigated. Thus, a protocol is developed here for nanowire or nanotube pick-and-place. However, applications can easily be extended to, for example, pick-and-place of nano-rods or nanoparticles dispersed on a substrate. Once the manipulation area is selected under the optical microscope, both the tips are aligned as a quasigripper above the center of the manipulation area. Each axis of the nanostage and the piezoscanner is initialized at an appropriate position to allow for enough manipulation motion travel. In this step, tip I is used to fully scan the relevant area obtaining a topographic image that contains nanoobjects to be manipulated and the end of tip II. Figure 7a shows a simulated image that contains the topography of two nanowires and the end of tip II. The image provides the following pick-and-place with
3D Micro/Nanomanipulation with Force Spectroscopy
7
0–9 0–9
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 7 A protocol for nanowire pick-and-place (a) Image scan for task planning. (b) Tip II is in contact with the nanowire. (c) Tip I is in contact with the nanowire, forming a nanotip
gripper. (d) Pick up and release a nanowire to its target position (e) A kinematic configuration of the AFM-RFS for nanoscale pick-and-place
relative positions between tip I, tip II and the nanoobjects to be manipulated. However, after a long period image scan, relocating tip II is recommenced to eliminate system’s thermal drift. As shown in Fig. 7b, tip II approaches the nanowire to make contact by moving the X-axis of the piezoscanner. A gap (typically 20 nm above the snap-in boundary) between Tip II and the substrate should be maintained during the approach to enable a negative deflection response in the form of a tiny force applied on tip II, and hence, sensitive detection of the tip-nanowire contact. Similarly, in Fig. 7c, once tip I is in contact with the nanowire, a nanotip gripper is configured for pick-andplace manipulation of the nanowire. The nanotip gripper in this step is used to pick up, transport, and release the nanowire to its target position by moving the piezoscanner on the X, Y, or Z-axis. The displacement on each axis depends on the dimensions of the nanowire and the location of the destination. Figure 7d shows a simulated post-manipulation image, in which a nanowire crossbar is built. The complete pickand-place procedure is monitored by force sensing.
with 300 nm silicon dioxide. AFM images show that the SiNWs have a taper shape and have diameters of 25 nm (top), 200 nm (root), and lengths of about 4 7 mm. Figure 8 shows an example of the contact detection with tip II: Point A and point C are where the tip contacts with the SiNW and the Si substrate, respectively; Point B and point D are where the tip breaks the contact with the Si substrate and the SiNW, respectively. Figure 8 shows a curve of the peeling force spectroscopy on tip II for the pick-and-place manipulation of the SiNW: point A and point B are where the tip snaps in and pulls off the Si substrate, respectively. The shape of curve of the force responses on tip I are similar except for the force magnitude due to different force sensitivities on each tip and uneven grasping due to asymmetric alignment of the SiNW relative to the grasping direction. This force spectroscopy during the pickup operation shows stable grasping for further SiNW transport. Figure 9 shows an experimental result of 3-D SiNW manipulation. A prescanned image (9 9 mm) is shown in Fig. 9a, which includes the topographic image of SiNWs and the local image of tip II. A grasping location of the nanowire to be manipulated is marked with a–a, where the SiNW has a height of 160 nm. Figure 9b is a post-manipulation image. It can
Pick-and-Place Nanomanipulation
In experiments, silicon nanowires (SiNWs) were deposited on a freshly cleaned silicon wafer coated
0–9
8
3D Micro/Nanomanipulation with Force Spectroscopy
a
b −60 0
B A Force [nN]
Force [nN]
−10 −20 −30 −40 −50 0.00
−120
A
−150 −180
C Approach Retraction
−210 B
D 0.05
Pickup Release
−90
0.10 0.15 0.20 Displacement [μm]
−240
0.25
0.00
0.06
0.12 0.18 0.24 Displacement [μm]
0.30
0.36
Force detection on tip II during the pickand-place nanomanipulation of the SiNW
Contact detection on the SiNW with tip II
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 8 Contact and force detection
a
b
400 200 0
G a
a
tip II 3 iv
/d
mm 3m
iv
m/d
3D Micro/Nanomanipulation with Force Spectroscopy, Fig. 9 Pick-and-place results of the SiNWs. (a) A pre-scanned image. In which a–a and G are the grasping location and the
target position, respectively. (b) A post-manipulation image verifies that the manipulated SiNW is piled upon another nanowire deposited on the Si substrate
be seen that the SiNW has been successfully transported and piled onto another SiNW. The manipulation procedure is described as follows. Once the SiNW was reliably grasped, the piezoscanner moved down 560 nm at a velocity of 80 nm/s. In this step, the SiNW was transported a distance of 4.4 mm along the X-axis at a velocity of 120 nm/s and 0.12 mm along the Y-axis at a lower velocity of 3.3 nm/s. In the releasing step, the piezoscanner moved up at a velocity of 100 nm/s. As tip II was slightly bent upward leading to a positive response of 0.015 V, tip I and tip II were separated by moving both the nanostage and the piezoscanner on the X-axis to release the SiNW from the nanotip gripper.
Conclusion and Future Directions A flexible robotic system developed for multi-scale manipulation and assembly from nanoscale to microscale is presented. This system is based on the principle of atomic force microscopy and comprises two individually functionalized cantilevers. After reconfiguration, the robotic system could be used for pick-and-place manipulation from nanoscale to the scale of several micrometers. Flexibilities and manipulation capabilities of the developed system are validated by pick-and-place manipulation of microspheres and silicon nanowires to build three-dimensional micro/nanostructures in ambient conditions.
3D Micro/Nanomanipulation with Force Spectroscopy
Complicated micro/nano manipulation and assembly can be reliably and efficiently performed using the proposed flexible robotic system. 3-D nanomanipulation methods are indispensible for heterogeneous integration of complex nanodevices. Serial nanomanipulation systems could only enable low-volume prototyping applications. However, high-volume and high-speed nanomanipulation systems are indispensable for future nanotechnology products. Therefore, autonomous and massively parallel AFM systems are required.
Cross-References ▶ AFM ▶ Atomic Force Microscopy ▶ Force Modulation in Atomic Force Microscopy ▶ Friction Force Microscopy ▶ Manipulating ▶ Nanogrippers ▶ Nanorobotic Assembly ▶ Nanorobotics ▶ Self-assembly of Nanostructures
References 1. Menciassi, A., Eisinberg, A., Izzo, I., Dario, P.: From “macro” to “micro” manipulation: models and experiments. IEEE-ASME Trans. Mechatro. 9(2), 311–320 (2004) 2. Lambert, P., Re´gnier, S.: Surface and contact forces models within the framework of microassembly. J. Micromechatro. 3(2), 123–157 (2006) 3. Sitti, M.: Microscale and nanoscale robotics systems-characteristics, state of the art, and grand challenges. IEEE Robot. Autom. Mag. 14(1), 53–60 (2007) 4. Xie, H., Rong, W.B., Sun, L.N.: Construction and evaluation of a wavelet-based focus measure for microscopy. Imaging Microsc. Res. Tech. 70, 987–995 (2007) 5. Liu, X.Y., Wang, W.H., Sun, Y.: Dynamic evaluation of autofocusing for automated microscopic analysis of blood smear and pap smear. J. Microsc. 227, 15–223 (2007)
9
0–9
6. Xie, H., Rong, W.B., Sun, L.N., Chen, W.: Image fusion and 3-D surface reconstruction of microparts using complex valued wavelet transforms. In: IEEE International Conference on Image Process, Atlanta, pp. 2137–2140 (2006) 7. Xie, H., Vitard, J., Haliyo, S., Re´gnier, S.: Optical lever calibration in atomic force microscope with a mechanical lever. Rev. Sci. Instrum. 79, 096101 (2008) 8. Xie, H., Rakotondrabe, M., Re´gnier, S.: Characterizing piezoscanner hysteresis and creep using optical levers and a reference nanopositioning stage. Rev. Sci. Instrum. 80, 046102 (2009) 9. Vo¨geli, B., von Knel, H.: AFM-study of sticking effects for microparts handling. Wear 238, 20–24 (2000) 10. Varenberg, M., Etsion, I., Halperin, G.: An improved wedge calibration method for lateral force in atomic force microscopy. Rev. Sci. Instrum. 74, 3362–3367 (2003) 11. Xie, H., Vitard, J., Haliyo, S., Re´gnier, S.: Enhanced accuracy of force application for AFM nanomanipulation using nonlinear calibration of optical levers. IEEE Sens. J. 8(8), 1478–1485 (2008) 12. Fukuda, T., Arai, F., Dong, L.X.: Assembly of nanodevices with carbon nanotubes through nanorobotic manipulations. Proc. IEEE 91, 1803–1818 (2003) 13. Dong, L.X., Arai, F., Fukuda, T.: Electron-beam-induced deposition with carbon nanotube emitters. Appl. Phys. Lett. 81, 1919–1921 (2002) 14. Dong, L.X., Arai, F., Fukuda, T.: Destructive constructions of nanostructures with carbon nanotubes through nanorobotic manipulation. IEEE/ASME Trans. Mechatron. 9, 350–357 (2004) 15. Molhave, K., Wich, T., Kortschack, A., Boggild, P.: Pick-and-place nanomanipulation using microfabricated grippers. Nanotechnology 17, 2434–2441 (2006) 16. Dong, L.X., Tao, X.Y., Zhang, L., Nelson, B.J., Zhang, X.B.: Nanorobotic spot welding: controlled metal deposition with attogram precision from copper-filled carbon nanotubes. Nano Lett. 7, 58–63 (2007) 17. Leach, J., Sinclair, G., Jordan, P., Courtial, J., Padgett, M.J., Cooper, J., Laczik, Z.J.: 3D manipulation of particles into crystal structures using holographic optical tweezers. Opt. Exp. 12, 220–226 (2004) 18. Yu, T., Cheong, F.C., Sow, C.H.: The manipulation and assembly of CuO nanorods with line optical tweezers. Nanotechnology 15, 1732–1736 (2004) 19. Bosanac, L., Aabo, T., Bendix, P.M., Oddershede, L.B.: Efficient optical trapping and visualization of silver nanoparticles. Nano Lett. 8, 1486–1491 (2008) 20. Rollot, Y., Re´gnier, S., Guinot, J.C.: Simulation of micromanipulations: adhesion forces and specific dynamic models. Int. J. Adhes. Adhes. 19, 35–48 (1999)
0–9
A
Ab Initio DFT Simulations of Nanostructures Antonio Aliano1 and Giancarlo Cicero2 1 Department of Physics, Politecnico di Torino, Torino, Italy 2 Materials Science and Chemical Engineering Department, Politecnico di Torino, Torino, Italy
Synonyms Electronic structure calculations; First principles calculations; Mean field approaches to many-body systems
Definition Density Functional Theory (DFT) is an exact theory in which the electronic charge density is the basic quantity used to determinate the ground state properties of a many-body quantum system [1, 2]. Such a theory is an alternative formulation of the Schro¨dinger picture of quantum mechanics and represents a fundamental breakthrough in the application of computer simulation to material science and to nanosystems in particular. In the last decades, DFT-based analysis showed a quantitative predictive role and has been successfully employed in the study of molecules, crystals, and nanostructured materials. DFT calculations permit to investigate phenomena at an atomic scale hardly accessible by other techniques, emerging as a fundamental tool for complementing experimental investigations and for addressing open issues regarding
nanomaterials and novel design processes [3, 4]. The DFT methodology is an ab initio (or first principles) approach, since no parameter characterizing the Hamiltonian of the system is tuned to empirical data and simulations results are thus free from empirical hints.
Introduction and DFT Foundations The chemical and physical properties of materials both at the macro- and nano-scale strongly depend on the fundamental interactions holding together atoms and electrons present in the structure. A complete understanding of the material properties rests upon being able to accurately describe the electronic structure of the system, or, in other words, to solve the Schro¨dinger equation for the Hamiltonian characterizing the system. Unfortunately, solving this equation is not possible without making some physical assumptions that decrease the mathematical complexity of the problem. One of the most important approximations considered in the quantum treatment of materials consists of the so-called Born-Oppenheimer or adiabatic approximation: considering the three orders of magnitude difference between nuclei and electron masses (and De Broglie wavelengths), only the electrons are regarded as quantum particles and it is accepted that the motion of the nuclei does not influence the electron dynamics (nuclei are considered fixed in space). Within this hypothesis, the description of correlated nuclei and electrons can be decoupled so that the wave function C of an N-electron system depends only on the electron coordinates, namely, C (r1, r2, . . ., rN) ¼ C ({r}), and the nuclei positions enter only as parameters. The Hamiltonian operator H^ for
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
A
12
Ab Initio DFT Simulations of Nanostructures
the electrons at a fixed nuclear configuration can be written in the compact form: H^ ¼ T^ þ V^ee þ V^ext
(1)
where T^ is the kinetic energy operator, V^ee the operator corresponding to the electron–electron interactions, and V^ext the external potential operator typically defined by the Coulomb attractive potential due to nuclei. Despite the simplifications introduced, when the form of the interaction potentials are known, the ^ ¼ EC for the many-body Schro¨dinger equation HC electrons cannot be solved analytically and it is not numerically manageable because of the prohibitive number of degrees of freedom. Density Functional Theory, which was elaborated in 1960s by Hohenberg and Kohn, together with the ansatz made by Kohn and Sham has provided a way to obtain the ground state properties for real systems of many electrons, starting from the general Hamiltonian reported in (1). Theoretical Background Density Functional Theory (DFT) takes its historical root in the Thomas-Fermi model [1] and has its solid mathematical justification in the Hohenberg-Kohn theorems [1, 2]. These theorems and their corollaries prove rigorously that the electronic charge density, rðrÞ, can be used as the only basic variable to obtain the ground state properties of a system of N interacting electrons moving in an external potential vext ðrÞ. In particular, it can be demonstrated that the total energy of a quantum system is a functional of the electron density rðrÞ only: knowing the electron density explicitly implies having a complete physical description of the system and it is equivalent to know its ground state wave function (C0 ½r). The energy functional can be written as: Z E½r ¼ F½r þ
vext ðrÞrðrÞdr
(2)
with F½r ¼ T½r þ Vee ½r In (2), the functional E½r is organized in two parts: a term depending on the external potential or, in other words, by the specific material and structure under investigation, and the functional F½r which is the same for all the physical systems (atoms, molecules, solids, or nanostructures) and it is called universal
functional. In quantum mechanics the electron charge density rðrÞ is defined as: Z rðrÞ ¼ N
Z ...
jCðfrgÞj2 dr2 . . . drN
(3)
R and obeys the condition rðrÞdr ¼ N in order to represent the number of electrons per unit volume for a given state (in (3) and throughout the chapter atomic units are used). The second Hohenberg-Kohn theorem states that the exact ground state energy of a system is the global minimum value of the functional E½r and that the density that minimizes this functional corresponds to the exact ground state density r0 . In this context, if r0 is unknown, the best estimate of the ground state properties can be obtained following a variational approach by minimizing the energy functional E½r calculated with respect to a trial density. The latter condition can be expressed considering that also the wave function of the system is a functional of the density, and the total energy of the system can be written as: E½r ¼ hC½rjH^jC½ri
(4)
the ground state charge density r0 results from: r0 ¼ min E½r:
(5)
r
The Hohenberg and Kohn formulation has in principle a very general validity but it is impractical since the exact form of the universal functional F½r, is not known. A decisive step toward its practical applications was made possible thanks to the ansatz by KohnSham (KS); according to their hypothesis, the interacting electron systems can be replaced with an auxiliary system of independent electrons characterized by the same ground state charge density, r0 , as the original one. For the auxiliary system the energy functional, E½r, can be written as: E½r ¼ Ts ½r þ
1 2
Z Z
rðrÞrðr 0 Þ drdr 0 þ jr r 0 j
Z vext rðrÞdr þ Exc ½r
(6) term of noninteracting where TS ½r is the R R kinetic energy electrons, 1=2 rðrÞrðr 0 Þ=jr r 0 jdrdr 0 is a classical term representing the Coulomb repulsion between two
Ab Initio DFT Simulations of Nanostructures
13
continuous charge distributions (Hartree term), while Exc ½r is the “exchange and correlation” term in which all the many-body effects of the original interacting system are included. For the noninteracting electrons, the wave function is expressed by C ¼ cðr1 Þcðr2 Þ . . . cðrN Þ and the charge density rðrÞ becomes rðrÞ ¼
X
f jc j i i i
2
; i ¼ 1; . . . ; N
(7)
where ci are the single-particle wave functions and fi are the occupation numbers. Once the density of the auxiliary system is obtained, one knows also the ground state density of the original interacting system. The minimization of (6) with respect to r yields a set of N ground state single-particle Schro¨dinger-like equations that reads: Z 1 rðr 0 Þ dExc ½r 0 dr þ v ðrÞ þ =2 þ ext 2 jr r0 j drðrÞ cKS i
¼
KS eKS i ci ; i
(8)
¼ 1; . . . ; N
It is worth to remark that the eigenvalues (eKS i ), and the wave functions (cKS ) calculated in (8) are referred i to a Kohn-Sham system and are not those relative to the real system of interacting electrons, yet, following (7), cKS are used to obtain the correct density. The i main intricacy of (6) and (8) is related to the fact that the explicit dependence of Exc ½r on the electron density is not known, thus, even if the KS ansatz has reduced the complexity of the original many-body problem, the correct ground state is still not achievable. Following the work by KS, many approximations to Exc ½r have been proposed in literature and these are usually classified as a “Jacob’s ladder of DFT” [5] which goes from lower accuracy (lower steps) to the exact unknown form (highest step). The simplest construction of Exc ½r is represented by the Local Density Approximation (LDA) in which Exc ½r is taken so that dExc ½r=dr is equal to the exchange and correlation energy density of a homogeneous electron gas, ehom xc ðrðrÞÞ, at the density of the auxiliary system at point r of space. The quantity ehom xc ðrðrÞÞ has an analytical expression, that, once substituted in (8) allows finding the eigenfunctions of the auxiliary system. Many approximations other than LDA are available for Exc ½r; one of the most commonly used consists of
A
the Generalized Gradient Approximation (GGA) in which the exchange and correlation energy density term has an explicit dependance also on the density gradient j=rðrÞj. Several expressions for such dependence have been proposed in literature. Both LDA and GGA functionals suffer from being approximations of the true energy functional and, for this reason, they fail in describing physical phenomena in which electron correlation has an important role as for example Van der Waals forces or strongly correlated systems such as the d-shells of transition metals. To overcome the limits of KS-DFT formulation, several post-DFT methods have been developed; for example, the LDA + U (or GGA + U) approximation in which an on-site Hubbard-like correction (U) is added to the strongly correlated electrons [6], or the GW approach in which Green functions are used to evaluate selfenergy corrections on single particle eigenvalues [7]. Finally, we mention that DFT, which is a ground state theory, has been extended to a time dependent formulation (TD-DFT) to provide a solid methodology to study the excited state properties of a many-body system in the presence of an external time dependent potential [8]. Numerical Details Once the Exchange and Correlation term is made explicit, the KS wave functions cKS i (and the ground state charge density r0 ) can be obtained by solving self-consistently the set of nonlinear differential equations (8). In the iterative solution procedure, the wave function is generally represented as a linear combination of basis set functions yielding a matrix representation of the Hamiltonian. This choice transforms the solution of partial integro-differential equations into a diagonalization problem, for which many standard algorithms have been largely tested and developed. Traditionally many DFT codes, implemented to calculate the electronic structure of materials, have been developed to deal with crystalline solids, that is, with periodic arrangements of atoms in a Bravais lattice and consequently yielding vext ðrÞ appearing in (8) also periodic. In this case, according to the Bloch’s theorem [9], the KS eigenfunctions can be expressed as Bloch states denoted by a wave vector k~ within the first Brillouin Zone (BZ). A convenient basis set to expand these wave functions consists of the plane waves (PWs) basis set. Since the system is periodic, not all the PWs appear in the expansion but only those having
A
A
14
a wave vector belonging to the reciprocal lattice of the crystal. In principle one would have to include an infinite number of functions to express the solution of the KS eigenvalue equation; in practice it is enough to limit the expansion to a finite subset of the basis set space once the accuracy of the subset has been proved. In the case of nonperiodic systems, like molecules, surfaces, or nanostructures, one can still effectively adopt a finite PWs basis set and use all the established numerical strategies deviced for 3-D periodic structures, if one adopts a supercell approach [2]. With the term “supercell” one indicates a 3-D periodic cell that contains the nonperiodic (or partially periodic) structure and a vacuum region. The cell sizes along the periodic directions correspond to the lattice parameters of the structure in those directions, while in the other directions the cell size has to be large enough to accommodate the finite structure and the vacuum. Periodic boundary conditions are applied to the supercell, in this way the structure is repeated in space even if it is not fully periodic; if the vacuum size is large enough the replicas along the nonperiodic directions can be considered to be isolated. Indeed, the increment of the cell size to create a reliable system increases the computational cost, thus a compromise has to be reached to obtain the desired accuracy. One of the most used methods, employed to decrease the size of the basis set (and thus reducing the computational cost) is represented by the so-called “pseudopotential” approximation [10]. According to this approximation the electrons tightly bound to the nuclei and not involved in bond formation (core electrons) are “frozen” into the core and their effect on the other electrons (valence electrons) is included in an effective way into the external potential (vext ðrÞ). The pseudopotential represents an electrostatic potential that contains the effect of the bare ionic charge screened by the core electrons and it is characterized by pseudowavefunctions that, in the core region, are smoother than the wave functions of the bare nuclear potential and, for this reason, require less PWs to be expanded. Once the solutions of the KS equations are obtained, it is possible to calculate any physical quantity of interest for the investigated system. For the case of periodic structures many properties such as the counting of electrons in bands, total energy, density,
Ab Initio DFT Simulations of Nanostructures
etc., are obtained by summing over all the states ~ labeled by ~ k in the BZ. For a general function Fi(k) associated with a certain physical quantity, the average value Fi is defined as: 1 Fi ¼ OBZ
Z
~ k~ Fi ðkÞd
(9)
BZ
~ OBZ is the BZ where i denotes discrete states for each k, volume, and the integral is extended over the entire BZ. In computer codes, accurate integration is performed by summation over discrete set of points in the BZ. One of the methods that has been largely used to sample the BZ is the one proposed by Monkhorst and Pack [2]. This method consists of constructing a uniform grid of points indicated by three integer numbers (n1, n2, n3) that define the mesh discretization along the reciprocal space axis. If these points are chosen carefully and the grid is fine enough, the error in the estimate of Fi is small and the integral is well approximated by the summation. Finally once the ground state density of the KS system is known, according to the Hellmann-Feynman theorem [2], it is possible to calculate analytically the forces acting on each atom of the structure. Thus, if the structure is not at its equilibrium geometry, atoms can be moved and systems relaxed following the calculated forces until a selected threshold value is reached.
DFT Applications to Nanostructures In the following the application of DFT to the investigation of the structural and electronic properties of a particular type of nanostructures, namely, onedimensional nanowires (NWs) is presented. Understanding how the properties of a material change when it is reduced to a nanostructure is important for a wide range of scientific and technological applications as well as for basic science. The reduction to nanometers in the size of a semiconductor [11] demands theoretical approaches based on an atomistic description of a material; first principles–based approaches have demonstrated to be capable of exploring the electronic structure of bulk materials and small nanostructures with highly predictive character
Ab Initio DFT Simulations of Nanostructures
reaching a degree of detail that can hardly be achieved with experimental techniques. In particular, DFT simulations have played a major role in the study of electron confinement, and in determining how the band gap, the optical properties, and the electronic transport change at the nanoscale. In the last years, many different materials have been studied ranging from low (e.g., Ge, Si, InN) to wide band gap (e.g., ZnO, SiC) semiconductor, in the form of both simple and core/shell wires. Hereinafter the results for simple nanowires are discussed for a particular type of a technologically important material: Indium Nitride (InN). More in detail the effects of size reduction on the energy gap of the material are presented and the changes induced by surface termination on the electronic properties of the nanostructure are analyzed. DFT Simulations of InN Nanowires Group-III nitrides represent a material class with promising electronic and optical properties [12]. Among these, InN exhibits the narrowest band gap of about 0.67 eV at low temperatures, the lowest effective electron mass, and the highest peak drift velocity and electron mobility. InN nanowires have been proposed for applications in new generation integrated systems like quantum wire transistors with the aim of reducing the power consumption in large-scale integrated circuits. Due to their high surface to volume ratio, nanowires are also potential candidates for sensor applications and they have been proposed as key elements in new generation solar energy harvesting devices such as dye sensitized or other hybrid solar cells. A crucial role for applications of InN nanowires as novel nanoelectronic devices is played by a quasitwo-dimensional electron accumulation layer around the wire. To fully exploit the potential of InN nanowires, their physical properties have to be understood in depth. Here the LDA and LDA + U results are discussed for wurtzite InN NWs [13] of increasing diameter with both clean and hydrogenated surfaces of three sizes (see Fig. 1): NW1 has one atom ring and a diameter ˚ , NW2 has two rings and a diameter d 10.6 d 4.1 A ˚ A, while NW3 has three rings and a diameter d 17.6 ˚ . By construction, these NWs have hexagonal crossA section and present nonpolar (1 100) facets, in
15
A A
Ab Initio DFT Simulations of Nanostructures, Fig. 1 Ball and stick representation of the investigated InN ˚ ), middle – 2 rings NW nanowires: left – 1 ring NW (d 4.1 A ˚ ), right – 3 rings NW (d 17.6 A ˚ ). Large balls (d 10.6 A represent In atoms, small balls represent N atoms
agreement with experimental observation. Typically for a DFT simulation, due to the numerical approximations adopted to solve the KS equations presented previously, accuracy of the results has to be checked carefully once the following computational conditions have been chosen: • Pseudopotentials needed to describe the external potential felt by the valence electrons. • Size of the plane waves basis set employed to expand the one-electron wave function. This is expressed by an energy cutoff that corresponds to the highest kinetic energy (lowest wave vector) of the PWs included in the expansion. • Convergence criteria on the total energy of the system and on the forces acting on the atoms in the case of structural relaxation based on HellmanFeynman forces. • Accuracy of the Monkhorst-Pack grid used to sample the Brillouin Zone of the periodic structure. • Vacuum size in the supercell, in the case of structures that are nonperiodic in one or more dimensions (e.g., surfaces, nanowires, quantum dots, etc.). To simulate InN 1-D nanostructures previously mentioned, the supercell approach was used: in this case the nanostructures are periodic only along the NW axis (z), thus large supercells have to be employed to accommodate enough vacuum space surrounding the wire (i.e., ˚ ) in the directions perpendicular to the NW axis. 10 A The integration over the Brillouin Zone was performed using a (1 1 6) k-point grid. All the structures were fully relaxed until the forces acting on the atoms were less than 103 Ry/bohr. The lattice parameter along the
16
direction of the NW axis can be optimized by varying the cell size along that direction and relaxing all the atomic positions. The minimum energy structure is then obtained by fitting the total energy values with a polynomial fit. It has been found that in nanowires two main relaxation mechanisms occur: the lattice parameter along the NW axis changes with respect to the bulk value and the atoms at the NW surfaces relax. The lattice constant along the NW axis increases decreasing the NW diameter and, except for the smallest diameter NW1, the behavior appears to be linear. While the NWs elongate along their axis, the lattice contracts in the perpendicular directions. This behavior has been observed also for other more ionic compound such as ZnO [14]. After relaxation, changes of InN bonds occur at the nanowire surface: the InN ˚ (shorter than InN length bond length becomes 2.05 A ˚ in bulk of 0.1 A) and it tilts by about 9:5 , with the In atoms relaxing inward. Upon hydrogenation the In–N ˚ ), but bond length recovers its bulk value ( 2.18 A the In–N bond tilting is now opposite to the clean case ( 5 – 6 ), due to the moving out of the In atom. In Fig. 2, the band gap dependence on the inverse of the NW diameter (1/d) for NWs with clean and hydrogenated surfaces is reported. It can be noted that in both LDA and LDA + U approaches the dependence of quantum confinements is almost linear and that the inclusion of the Hubbard (U) correction determines a rigid shift of the band gap (about 0.2 eV for NWs with clean surfaces and 0.35 eV for NWs with
Ab Initio DFT Simulations of Nanostructures, Fig. 3 Electronic charge density isosurface (denoted by shaded yellow-green area) for the highest occupied state: left panel – NW3, right panel – NW3 + H. Sketched globes represent atoms
Ab Initio DFT Simulations of Nanostructures
hydrogenated surfaces) that depends only slightly on the NW diameter. The larger energy gaps observed in the case of nanowires with hydrogenated surfaces imply that quantum confinement effects are more noticeable in passivated NWs and that one might expect to see spectroscopic differences in experiments performed in different environments (e.g., solution vs ultra high vacuum). LDA and LDA + U predict similar confinement effects also in terms of charge density distribution when comparing nanowires with clean and hydrogenated surfaces. In the latter case, due to 5 4 Gap (eV)
A
3 2 rings 2
1 ring
3 rings
1 0 0 BULK
0.05
0.1
0.15 1/d (1/Å)
0.2
0.25
0.3
Ab Initio DFT Simulations of Nanostructures, Fig. 2 Band gap versus inverse diameter (1/d) of InN nanowires with clean surface indicated by solid (LDA) and dotted (LDA + U) lines, and for hydrogenated nanowires indicated by dashed (LDA) and dashed-dotted (LDA + U) lines
AC Electrokinetics of Colloidal Particles
surface dangling bonds saturation, the electronic charge density accumulates in the center of the wire, while in nanowires with clean surfaces the density is more delocalized on the surface atoms. This effect is visible in Fig. 3 where the electronic charge density isosurface of the highest occupied state for NW3 is shown for clean and hydrogenated surfaces.
Conclusion In conclusion, first principles DFT electronic structure calculations can be effectively used to systematically investigate the relative stability and electronic properties of nanostructures with clean, reconstructued, or passivated surfaces. In the case of Nanowires one can also consider the effects of the growth direction and of the diameter. Electronic properties of NWs, including band gaps, band structures, and effective masses, are usually found to depend sensitively on all the NW structural parameters. In many cases, the size dependence of the band gap depends on the growth direction of the NW, and the band gaps for a given size also depend on surface structure; moreover NWs with clean surfaces have HOMO orbitals localized on the facets and exhibit smaller band gaps. DFT calculations demonstrate the extremely rich nature of the electronic properties of nanostructures and NWs in particular and have a unique role in describing confinement effects in these small dimension systems. By changing the growth direction and NW diameter, or tuning the surface structure via chemical or physical methods, it is possible to engineer NWs with a wide range of technologically important electronic properties.
Cross-References ▶ Hybrid Solar Cells ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanostructured Materials for Sensing ▶ Optical and Electronic Properties ▶ Surface Electronic Structure ▶ Theory of Optical Metamaterials
17
A
References 1. Parr, R.G., Yang, W.: Density-Functional Theory of Atoms and Molecules. Oxford University Press, Oxford, UK (1989) 2. Martin, R.M.: Electronic Structure-Basic Theory and Practical Methods. Cambridge University Press, Cambridge, UK (2008) 3. Hafner, J., Wolverton, C., Ceder, G.: Towars computational meterial design: the impact of density functional theory on material research. MRS Bull. 31, 659 (2006) 4. Marzari, N.: Realistic modeling of nanostructures using density functional theory. MRS Bull. 31, 681 (2006) 5. Perdew, J.P., Shmidt, K.: Density Functional Theory and Its Application to Materials, p. 1. American Institute of Physics, New York (2001) 6. Anisimovy, V.I., Aryasetiawanz, F., Lichtenstein, A.I.: First-principles calculations of the electronic structure and spectra of strongly correlated systems: the LDA + U method. J. Phys. Condens. Matter 9, 768 (1997) 7. Onida, G., Reining, L., Rubio, A.: Electronic excitations: density-functional versus manybody Green’s-function approaches. Rev. Mod. Phys. 74, 601 (2002) 8. Marques, M.A.L., Ullrich, C.A., Nogueira, F., Rubio, A., Burke, K., Gross, E.K.U.: Time-Dependent Density Functional Theory. Springer, Berlin/Heidelberg (2006) 9. Ashcroft, N.W., Mermin, N.D.: Solid State Physics. Saunders College Publishing, Philadelphia (1976) 10. Bachelet, G.B., Hamann, D.R., Schl€ uter, M.: Pseudopotentials that work: from H to Pu. Phys. Rev. B 26, 4199 (1982) 11. Law, M., Goldberger, J., Yang, P.: Semiconductor mamowires and nanotubes. Annu. Rev. Mater. Res. 34, 83 (2004) 12. Wu, J.: When group-III nitrides go infrared: new properties and perspectives. Appl. Phys. Rev., J. Appl. Phys. 106, 011101 (2009) 13. Terentjevs, A., Catellani, A., Prendergast, D., Cicero, G.: Importance of on-site corrections to the electronic and structural properties of InN in crystalline solid, nonpolar surface, and nanowire forms. Phys. Rev. B 82, 165307 (2010) 14. Cicero, G., Ferretti, A., Catellani, A.: Surface induced polarity inversion in ZnO nanowires. Phys. Rev. B 80, 201304(R) (2009)
AC Electrokinetics ▶ Dielectrophoresis
AC Electrokinetics of Colloidal Particles ▶ AC Electrokinetics of Nanoparticles
A
A
18
AC Electrokinetics of Nanoparticles Hossein Nili1,2 and Nicolas G. Green2 1 School of Electronics and Computer Science, University of Southampton, Highfield, Southampton, UK 2 Nano Research Group, University of Southampton, Highfield, Southampton, UK
Synonyms AC electrokinetics of colloidal particles; AC electrokinetics of sub-micrometer particles
Definition AC electrokinetics is the name given to a group of techniques that utilize alternating (AC) electric fields to move dielectric particles in suspension. AC electrokinetics of nanoparticles refers to methods of exerting electrical force and/or torque on particles of nanometer dimensions, examples of which are viruses, macromolecules, and colloidal particles.
Introduction Dielectrics do not bear a net charge, but rather polarize when subjected to electric fields. When dielectric particles in suspension are subjected to electric fields, polarization results in charge accumulation at the particle/ electrolyte interface. Electrode polarization may also occur, giving rise to the buildup of a double layer of ions and counterions at the electrode/electrolyte interface. Particle polarization is represented by effective dipole and higher-order moments. AC electrokinetic forces and torques on particles result from interactions of an applied electric field with the effective moments. Electric field-induced fluid motion is also studied under AC electrokinetics due to the important effect it can have on particle behavior. Electric field interactions with the double layer and gradients in fluid permittivity and conductivity (resulting from localized heating of the medium) are the main factors causing fluid motion under AC electric fields. AC electrokinetic techniques are particularly fitted to lab-on-a-chip applications where multiple processes
AC Electrokinetics of Nanoparticles
involving manipulation, separation, or characterization of biological particles, mostly dielectrics, are integrated onto a single chip. The direction and magnitude of AC electrokinetic forces and torques can be easily controlled by varying electric field frequency and geometry. The techniques are advantageous over alternative means of moving particles in their noninvasive nature and easy integration onto micro-devices, not relying on any moving parts. It was long believed that for sub-micrometer particles, AC electrokinetic forces would be overwhelmed by thermal effects such as Brownian motion. The strong electric fields required to exert sufficient electrical force to dominate particle behavior at the nanometer scale were believed to generate excessive heat giving rise to strong fluid motion that would hinder AC electrokinetic interactions. Thanks to fabrication techniques that realized micro- and nano-electrode geometries, electric fields of sufficient strength to overcome Brownian motion of nanoparticles could be generated with the application of modest voltages, avoiding excessive heating of the suspension.
AC Electric Field Interactions with Nanoparticle Suspensions When an electric field is applied to dielectric particles in suspension, surface charge accumulates at interfaces between the dielectrics due to the differences in electrical properties. Since the polarizabilities of each dielectric are frequency dependent, the magnitude of the surface charge is also frequency dependent and the total (complex) permittivity of the system exhibits dispersions solely due to the polarization of the interfaces. This is referred to as the Maxwell-Wagner interfacial polarization. For a spherical particle of radius R, the effective dipole moment is given by: p ¼ 4pem R3 ½K ðoÞE
(1)
where o is the frequency of the applied electric field E, and K(o), known as the Clausius-Mossotti factor, describes a relaxation in the polarizability of the particle with a relaxation time: tMW ¼
ep þ 2em sp þ 2sm
(2)
AC Electrokinetics of Nanoparticles
19
A
sp ¼ sp;bulk þ sp;surface . The surface conductivity of the particle is given by:
E Jbulk
sp;surface ¼
2Ks R
A (3)
where Ks is the surface conductance.
Jsurface
AC Electrokinetics of Nanoparticles, Fig. 1 Schematic of the mechanism with which surface conductance affects interfacial polarization of nanoparticles in suspension. The AC electric field produces both a bulk flow of ions and a surface flow around the particle
The angular frequency oMW ¼ 2pfMW ¼ 1=tMW is often referred to as the Maxwell-Wagner relaxation frequency. During the first half of the twentieth century, dielectric spectroscopic measurements of suspensions of nanoparticles identified that the Maxwell-Wagner relaxation frequency for charged nanoparticles was higher than expected. Later, O’Konski [1] showed that the dielectric properties of nanoparticles in this frequency regime were dominated by surface conductance effects. The high value of particle conductivity determined from dielectric measurements were explained by the inclusion of a surface conductance component in the derivation of the particle’s dipole moment. The model developed by O’Konski is shown in Fig. 1. He assumed that the flux due to the transport of charge carriers, associated with the fixed charge on the particle surface, could be added to the flux due to the transport of bulk charge to and from the surface. With this assumption, he derived the potential around the spherical particle and derived the equation for the dipole moment of a dielectric sphere (1) but with the particle conductivity given by the sum of the bulk conductivity of the particle and a surface conductivity term:
AC Electrokinetic Techniques Dielectrophoresis (DEP): Figure 2 shows the underlying principle of dielectrophoresis, as the most widely used of AC electrokinetic techniques. The gradient in electric field strength gives rise to unequal forces experienced by polarization charges at the particle/electrolyte interface, hence a net force on the particle moving it toward or away from regions of high field intensity depending on whether the particle is more or less polarizable than the suspending medium. When particle dimensions are smaller than a characteristic length scale of electric field nonuniformity, higher-order terms can be neglected and the time-averaged dielectrophoretic force on a spherical particle of radius R suspended in a fluid of dielectric constant 2m is given by [2]: hFDEP i ¼ p2m R3 Re½KðoÞHjEj2
(4)
It is understood from (4) that the dielectrophoretic force scales with particle volume and also inversely with the cube of a characteristic dimension of the electrode geometry. It is due to the latter proportionality that micro- and nano-electrode geometries are capable of exerting electric fields of sufficient strength to move particles as small in volume as nanoparticles, overcoming Brownian and field-induced fluid motion. Although first observations of the effect date back to earlier times, Pohl was the first to use the term dielectrophoresis for the force exerted by a nonuniform electric field on a dielectric [3]. He observed coagulation of carbon particles from a polymer solution upon application of a DC or AC electric field. Pohl used the term “dielectrophoresis” to distinguish the effect from electrophoresis, which describes motion of charged particles under DC (and not AC) electric fields. Since its advent, dielectrophoresis has been used in a broad range of applications. Separation, as a fundamental part of many processes in micro-total-analysis
A
20
AC Electrokinetics of Nanoparticles, Fig. 2 Dielectrophoresis – Interaction of a nonuniform electric field with a dielectric particle in suspension gives rise to unequal forces experienced by polarization forces at the particle/ electrolyte interface. The result is a net dielectrophoretic force that moves the particle (a) toward or (b) away from regions of high electric field intensity depending on whether the particle is more or less polarizable than its suspending medium
AC Electrokinetics of Nanoparticles
a
b
E
+
+ − + − − +
−
−
−
−
E
+
+
+
+
+
− + + − + −
−
+ + − + + − + − +
− + − + − − + − −
induced dipole
induced dipole
DEP motion
DEP motion
systems (mTAS), has been diversely and controllably accomplished using dielectrophoresis. The dependence of the dielectric properties of particles on their structure and composition has broadened the applicability of DEP-based separation techniques. In electrode geometries with well-defined regions of electric field maxima and minima, particles of different properties can be separated into subpopulations. Figure 3 shows the separation of 280 nm tobacco mosaic viruses (TMV) from 250 nm herpes simplex viruses (HSV) in a polynomial electrode configuration [4]. As the TMV particles are more polarizable than the suspending medium, they gather at electrode edges where the electric field is strongest, while HSV particles – less polarizable than the fluid at the applied field frequency – are concentrated at the electric field minimum in the center of the electrode geometry. Dielectrophoresis has been used extensively for manipulation and characterization of nanoparticles. One of the most notable application areas is diagnostics and healthcare, toward which a huge amount of effort has been directed involving sub-micrometer biological particles such as viruses, DNA, and chromosomes [5]. Dielectrophoresis has also been used for separation of metallic from semiconducting carbon nanotubes [6], assembly of nanoparticles into microwires [7], and three-dimensional focusing of nanoparticles in microfluidic channels [8]. With the emergence of nano-electrode geometries, there is huge prospect for further applications involving
−
AC Electrokinetics of Nanoparticles, Fig. 3 Dielectrophoretic separation of nanoparticles – Applying a 6 MHz AC electric field leads to the collection of 280 nm tobacco mosaic viruses (TMV) at the electrode edges where the field is strongest, and the concentration of 250 nm herpes simplex viruses (HSV) at the center of the polynomial electrode geometry where there is a well-defined field minimum [4]
dielectrophoresis of nanoparticles including the fabrication of a new generation of electronic devices and sensors [5].
AC Electrokinetics of Nanoparticles
21
F+ +Q
E
q
F− −Q
AC Electrokinetics of Nanoparticles, Fig. 4 Principle of electro-orientation – In a uniform field E (indicated by the vector and the dotted field lines) the two charges experience equal and opposite forces resulting in a torque about the center point of the dipole
Electro-orientation: As shown in Fig. 4, when a dipole sits in a uniform field, each charge on the dipole experiences an equal and opposite force tending to align the dipole with the electric field. The effect can also be observed in nonuniform and rotating electric fields and is the basis of a phenomenon known as electro-orientation. A dielectric particle of no net charge in a uniform electric field will be subject to no net force but will experience a torque given by: G¼pE
(5)
This torque always tends to align the dipole with the electric field. However, small dipoles, such as those of nanoparticles, will not completely align with the field due to the dominating effect of Brownian motion. In typical dielectric spectroscopy measurements, the electric fields are only of sufficient magnitude to align individual molecules by fractions of degrees, but the total effect of alignments of many molecules gives rise to a large net or average alignment [9]. In AC electric fields, electro-orientation is frequency-dependent with nonspherical particles orienting along different axes in different frequency ranges. The set of cross-over frequencies is referred to
A
as the orientation spectrum. The orientation spectra of elongated biological particles such as erythrocytes and bacteria have been used to study the dielectric properties of these particles [10]. Electro-rotation (ROT): Dielectric particles subjected to rotating electric fields experience a torque based on a principle which is somewhat different from that of electro-orientation. As mentioned previously, applying an electric field to a dipole will result in the exertion of a (electro-orientational) torque that tends to align the dipole with the electric field. Alignment of the dipole with the electric field vector will not be immediate as it takes a finite amount of time for polarization charges to move toward the particle/ electrolyte interface and form a dipole. In a rotating electric field, where the field vector changes direction, this time delay gives rise to a (electro-rotational) torque, as shown in Fig. 5. The figure also shows an example electro-rotation setup, where voltages of 90o phase differences are applied to successive electrodes encircling the particle to generate a rotating electric field. The first order electro-rotational torque on a spherical particle of radius R is given by [11]: GROT ¼ 4p2m R3 Im½K ðoÞjEj2
(6)
where Im½K ðoÞ denotes the imaginary part of the Clausius-Mossotti factor. The particle will rotate with or counter to the applied field depending on whether Im½K ðoÞ is negative or positive, respectively. Accounting for viscous drag force from the suspending fluid, the rotation rate of the particle is given by [11]: O¼
2m Im½K ðoÞjEj2 2
(7)
where denotes fluid viscosity. Variations with frequency of the rotation rate are referred to as electrorotation spectra. Two important distinctions can be readily identified from a comparison of (4) and (6) for dielectrophoretic force and electro-rotational torque. Firstly, the ROT torque is proportional to the square of the electric field magnitude while the DEP force is a function of the gradient of the square of the field magnitude – and is therefore zero in uniform electric fields. Secondly, the electro-rotational torque depends on the imaginary rather than the real part of the Clausius-Mossotti
A
A
22
AC Electrokinetics of Nanoparticles
AC Electrokinetics of Nanoparticles, Fig. 5 Electro-rotation – (a) A schematic diagram of an electro-rotation setup. Four signals, successively 90o out of phase are applied to four electrodes encircling the particle. (b) Schematic diagram showing how the induced dipole moment of a particle lags behind a rotating applied electric field
±Vo cos(wt) y w
E
q p ±Vo sin(wt)
Direction of travel of field
x
F P E
+Vo sin(wt )
+Vo cos(wt )
−Vo sin(wt )
−Vo cos(wt )
AC Electrokinetics of Nanoparticles, Fig. 6 Traveling-wave dielectrophoresis – Schematic diagram of a linear traveling wave dielectrophoresis array and the consecutive phase-shifted signals required to generate the traveling electric field. Also shown are
the approximate field lines for time t ¼ 0, the electric field and the dipole moment induced in the particle together with the force on the particle
factor. As a result, particles may experience both dielectrophoresis and electro-rotation with the relationship between the two determined by the dielectric properties of the particles and their suspending media. Initially observed as a nuisance in cell electrofusion procedures, electro-rotation has since developed into a useful means of measuring dielectric properties and membrane electrical parameters of biological particles. Rotational spectra are analyzed to determine fundamental cell properties and to monitor changes in these properties upon different types of treatments [10]. Among the major works done on electro-rotation of nanoparticles are investigations by Washizu and coworkers into the torque-speed behavior of the flaggellar motor mechanism of bacterial cells [12]. Traveling-wave dielectrophoresis (twDEP): Traveling-wave DEP can be considered as the linear
analog of electro-rotation where voltages of 90o phase difference are applied to successive electrodes that are laid out as tracks, rather than being arranged in a circle. This generates an electric field wave which travels along the electrodes. The dipole induced in the particle moves with the electric field but lags behind the field, as in electro-rotation. As shown in Fig. 6, the result is the induction of a force, rather than a torque, given – for a spherical particle of radius R – by [13]: FtwDEP ¼
4p2m R3 Im½K ðoÞjEj2 l
(8)
where l is the wavelength of the traveling wave. The negative sign in (8) indicates that, as shown in Fig. 6, the twDEP force propels particles in the opposite direction to the moving field vector. For a finite twDEP
AC Electrokinetics of Nanoparticles
force to be exerted on a particle, two criteria need to be met: (a) the particle must experience a DEP force that levitates it above the electrode array, and (b) some loss mechanism needs to be present for the imaginary part of the Clausius-Mossotti factor to be nonzero. Masuda et al. [14] were the first to describe the principles of traveling-wave DEP. Since then, twDEP has been used for characterizing and separating cells and microorganisms [15]. Nanoparticle manipulation and separation has also been accomplished using the traveling electric field above interdigitated electrodes [16].
A
23
a
E −Fq
−V
E Er
Et
+V
Electrodes
b
A
Fq
Bulk fluid flow ux
AC Electrokinetic Fluid Motion AC Electro-osmosis (ACEO): AC electric fields applied to a suspension of dielectric particles can give rise to fluid motion in an effect known as AC electro-osmosis [17]. The effect arises from interaction of the electric field with charges at the double layer and is only observed if the field is nonuniform. The mechanism for AC electroosmosis is shown in Fig. 7. Applying equal and opposite voltages to successive electrodes gives rise to the electric field E with tangential component Et outside the double layer and an induced charge on each electrode. The induced charge experiences a force Fq due to the action of the tangential field, resulting in fluid flow. Figure 7a shows the system for one half-cycle of an AC field. In the other half-cycle, the sign of the potential, that of the induced charge, and the direction of the tangential field are all opposite. As a result, the direction of the force vector remains the same giving a nonzero time-averaged force and steady-state fluid flow occurs, as shown in Fig. 7b. Electrothermal fluid flow: At higher frequencies (>100 kHz), AC electroosmotic flow is negligible and the dominant fluid flow is due to electrothermal effects. This type of flow requires temperature gradients in the fluid and these can be generated both by internal and/or external sources. The internal source is Joule heating, where the electric field causes power dissipation in the fluid, and the corresponding temperature rise diffuses through the system. This gives rise to gradients in the conductivity and permittivity; the electric field acts on these gradients to give a body force on the fluid and consequently a flow. In this case, the velocity of the flow is proportional to the temperature rise in the fluid, which is in turn
Electrodes
AC Electrokinetics of Nanoparticles, Fig. 7 AC electroosmosis – (a) Schematic diagram outlining the mechanism of AC electroosmosis. (b) The interaction of the tangential field at the surface with the charge in the double layer gives rise to a surface fluid velocity and a resulting bulk flow
proportional to the conductivity of the electrolyte and the magnitude of the electric field. As a result, this type of fluid flow occurs mainly at high electrolyte conductivities. Apart from the conductivity dependence of the electrothermal effect, the magnitude of the flow is also frequency dependent. The ratio of low- to highfrequency limiting values is approximately 10:1 and the flow changes magnitude and direction at a frequency of the order of the charge relaxation frequency of the medium. The change in the velocity of the fluid as a function of the two parameters, frequency of the applied field and conductivity of the electrolyte, is complicated and is summarized in Fig. 8. In these plots, the logarithm of the magnitude of the flow velocity is indicated by a gray scale as a function of the conductivity of the electrolyte and frequency of the field, for two applied voltages. Both figures demonstrate the regions in which strong fluid flow occurs and also which regions might be “safe” from fluid motion.
AC Electrokinetic Forces on Nanoparticles As described through the different mechanisms responsible, particle behavior under AC electric fields is determined by a variety of factors. The total force on any particle is given by the sum of many forces
A
24
AC Electrokinetics of Nanoparticles
10−1
10−1
Conductivity (S m−1)
b 100
Conductivity (S m−1)
a 100
10−2
10−3
10−4 1 10
102
103
104 105 106 Frequency (Hz)
107
108
109
10−2
10−3
10−4 1 10
102
103
104 105 106 Frequency (Hz)
107
108
109
AC Electrokinetics of Nanoparticles, Fig. 8 Schematic frequency/conductivity maps of the two types of electric fielddriven fluid flow: AC electroosmosis and electrothermal. The white dotted line indicates the frequency at which electrothermal flow changes direction. The velocities are plotted on a log gray
scale and were determined for the same position (10 mm from the edge of the electrode) and applied signals: (a) 1 V and (b) 10 V on each electrode. Note that the ratio of AC electroosmotic flow is much less at 10 V
including Brownian, dielectrophoretic, and hydrodynamic forces, the latter arising from fluid motion. These forces can be comparable to, or in certain circumstances much stronger than, the dielectrophoretic force exerted on the particle. At certain combinations of medium conductivity, electric field frequency, and strength, the hydrodynamic forces can dominate particle behavior. Under other experimental conditions these forces are negligible and as a result small changes in the dielectric properties of the particles can be detected and measured by DEP. Ramos and coworkers have conducted order-ofmagnitude calculations for the forces on nanoparticles in AC electric fields [18]. It has been shown that for lowconductivity suspensions, Joule heating has little effect on system temperature and is therefore negligible. With 10 V applied to a micro-electrode system with a medium conductivity of 10mS/m, temperature rises of the order of only 1 C have been observed. In contrast, electrothermal forces have been shown to be of sufficient strength to compete with the dielectrophoretic force. However, it is important to note that the DEP force varies much more rapidly than electrothermal forces upon proximity with electrode edges. Applying 5 V to a system of parallel finger micro-electrodes and a 10 ms/mconductivity suspension of 282 nm particles has been shown to generate a fluid velocity of 5 mm/s against a DEP-induced velocity of 1.8 mm/s. Near electrode edges, DEP-induced and electrothermal velocities have
been reported to be 200 and 50 mm/s, respectively. Changes in fluid density arising from field-induced temperature gradients have been shown to be on the order of 0.01% per degree. As a result, natural convection has been found to be very small compared to electrothermal and DEP forces. Brownian motion was for long considered the biggest obstacle in AC electrokinetic motion of sub-micrometer particles. Pohl had estimated that electric field strengths required to overcome Brownian motion of nanoparticles would (a) not be realizable, and (b) lead to excessive heating, and hence electrothermal motion, of the fluid, thereby hindering DEP motion of particles [1]. Experimental findings have shown that Pohl had largely overestimated the electric field strength required to overcome thermal effects, the reason being his comparison of thermal energy with dielectrophoretic potential rather than force, the latter being the gradient of the former [19]. It has been shown that DEP forces required to overcome Brownian motion of nanoparticles are in the sub-picoNewton range and can be easily generated using micro- and nano-electrode geometries by application of voltages modest enough to avoid excessive heating of the medium. The observable deterministic force required to move a 282 nm particle has been found to be of the order of 0.01pN over a 1 s time frame of observation, requiring an electric field on the order of 100 kV/m, which can be generated by applying voltages on the order of 1 V across a 10 mm gap.
AC Electroosmosis: Basics and Lab-on-a-Chip Applications
25
A
In summary, although hydrodynamic effects can complicate AC electrokinetic motion of nanoparticles, it has been shown that the forces can be predicted and therefore controlled. Consequently, it is expected that recent technological advances in AC electrokinetics could be applied, together with fieldinduced fluid motion, to develop new methods for the characterization, manipulation, and separation of nanoparticles.
17. Ramos, A., Morgan, H., Green, N.G., Castellanos, A.: AC electric-field-induced fluid flow in microelectrodes. J. Colloid Interface Sci. 217, 420–422 (1999) 18. Ramos, A., Morgan, H., Green, N.G., Castellanos, A.: AC electrokinetics: a review of forces in microelectrode structures. J. Phys. D: Appl. Phys. 31, 2338–2353 (1998) 19. Green, N.G.: Dielectrophoresis of sub-micrometre particles. Thesis, University of Glasgow, Glasgow, UK (1998)
References
AC Electrokinetics of Sub-micrometer Particles
1. O’Konski, C.T.: Electric properties of macromolecules V: theory of ionic polarization in polyelectrolytes. J. Phys. Chem. 64, 605–619 (1960) 2. Pohl, H.A.: Dielectrophoresis: the behavior of neutral matter in nonuniform electric fields. Cambridge University Press, Cambridge (1978) 3. Pohl, H.A.: The motion and precipitation of suspensoids in divergent electric fields. J. Appl. Phys. 22, 869–871 (1951) 4. Morgan, H., Hughes, M.P., Green, N.G.: Separation of submicron bioparticles by dielectrophoresis. Biophys. J. 77, 516–525 (1999) 5. Pethig, R.: Review article – dielectrophoresis: status of the theory, technology, and applications. Biomicrofluidics 4, 022811 (2010) 6. Krupke, R., Hennrich, F., von Lohneysen, H., Kappes, M.M.: Separation of metallic from semiconducting singlewalled carbon nanotubes. Science 301, 344–347 (2003) 7. Hermanson, K.D., Lumsdon, S.O., Williams, J.P., Kaler, E.W., Velev, O.D.: Dielectrophoretic assembly of electrically functional microwires from nanoparticle suspensions. Science 238, 1082–1086 (2001) 8. Morgan, H., Holmes, D., Green, N.G.: 3D focusing of nanoparticles in microfluidic channels. IEE Proc. Nanobiotechnol. 150, 76–81 (2003) 9. Morgan, H., Green, N.G.: AC electrokinetics of colloids and nanoparticles. Research Studies Press, Hertfordshire (2003) 10. Jones, T.B.: Electromechanics of Particles. Cambridge University Press, Cambridge (1995) 11. Arnold, W.M., Zimmermann, U.: Electro-rotation – development of a technique for dielectric measurements on individual cells and particles. J. Electrostat. 21, 151–191 (1988) 12. Washizu, M., Shikida, M., Aizawa, S., Hotani, H.: Orientation and transformation of flagella in electrostatic field. IEEE Trans. IAS 28, 1194–1202 (1992) 13. Hughes, M.P.: AC electrokinetics: applications for nanotechnology. Nanotechnology 11, 124–132 (2000) 14. Masuda, S., Washizu, M., Iwadare, M.: Separation of small particles suspended in liquid by nonuniform traveling field. IEEE Trans. Ind. Appl. 23, 474–480 (1987) 15. Hagedorn, R., Fuhr, G., Muller, T., Schnelle, T., Schnakenberg, U., Wagner, B.: Design of asynchronous dielectric micromotors. J. Electrostat. 33, 159–85 (1994) 16. Li, W.H., Du, H., Chen, D.F., Shu, C.: Analysis of dielectrophoretic electrode arrays for nanoparticle manipulation. Comp. Mater. Sci. 30, 320–325 (2004)
▶ AC Electrokinetics of Nanoparticles
AC Electroosmosis: Basics and Lab-on-aChip Applications Pablo Garcı´a-Sa´nchez and Antonio Ramos Departamento de Electro´nica y Electromagnetismo, Universidad de Sevilla, Sevilla, Spain
Synonyms Induced-charge electroosmosis
Definition Microelectrode structures subjected to AC voltages can generate flow of aqueous solutions by the influence of the AC field on the charges induced by itself at the electrode–electrolyte interface, i.e., induced charges in the electrical double layer (EDL). The phenomenon is called AC Electroosmosis (ACEO) in analogy with the “classical” electroosmosis, where fluid flow is generated by the action of an applied electric field on the mobile charge of a given solid–liquid interface [1]. The main difference is that the applied field in ACEO is responsible for both inducing the charge and pulling on it. The term Induced-Charge Electrokinetics (ICEK) has been proposed to refer to all phenomena where the electric field acts on the electrical double layer induced by itself [2].
A
A
26
AC Electroosmosis: Basics and Lab-on-a-Chip Applications
Basic Mechanism E
E
F=qEt
F=qEt
V0 Electrodes
substrate
V = V0 cos (wt)
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 1 Basics of the ACEO mechanism. Electrical charge is induced at the electrode–electrolyte interface. The tangential component of the electric field acts on this charge giving rise to a surface fluid velocity
10 microns 250
15 microns 20 microns
200 Velocity (μm/s)
Possibly, the simplest system for the study of AC Electroosmosis consists of a couple of coplanar electrodes covered by an aqueous electrolyte. The electrodes are subjected to a harmonic potential difference of amplitude V0 and frequency o, VðtÞ ¼ V0 cosðotÞ. Figure 1 shows a sketch of the physical system at a given time of the AC signal cycle. The applied field attracts counterions at the solid–liquid interfaces with a certain delay and, since the electric field is nonuniform, there appears an electrical force pulling the liquid outward F ¼ qEt (Et is the component of the electric field tangential to the electrode surfaces). Note that in the other half cycle, both the tangential field and the induced charge change sign and, therefore, the force direction remains the same and the time-averaged force is nonzero. The fluid velocity generated by this mechanism is frequency dependent. At low frequencies, the electrical charges have sufficient time to completely screen the electric field at the electrodes, the electric field in the electrolyte bulk vanishes, and so does the induced velocity. In this view, the electrodes are supposed to be perfectly polarisable, i.e., no Faradaic currents occur. At high frequencies, the electrical charges have no time to follow the electric field, the induced charge at the electrode tends to zero, and, therefore, the electrical force is negligible. The theoretical expectation for the velocity versus frequency plot is a bell-shaped curve, in accordance with the experimental observations (Fig. 2).
30 microns 40 microns
150 100 50 0 1.0E–03
1.0E–02
1.0E–01
1.0E+00
wz
Simple Model for Small Voltages
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 2 Fluid velocity at several positions on the electrode surface. The velocity is plotted against the product of the angular frequency and the distance of measurement from the center of the two electrodes (Reprinted with permission from [1], # 1999 Elsevier)
The electroosmotic velocity generated at metal surfaces can be computed from the Helmholtz– Smoluchowski formula, nHS ¼ ezEt =, where e is the dielectric constant of water, z is the zeta potential of the solid–liquid interface, and is the liquid viscosity. For small voltage amplitudes, the relation between z and the charge in the EDL is linear and the formula reads vHS ¼ sq lD Et =, with ld the Debye length and sq the charge per unit area in the EDL. In ACEO, sq is the induced charge in the EDL by the AC potential. A constant (non-oscillating) term in the electrode–electrolyte surface charge is not considered here although it might exist, the so-called intrinsic
surface charge. However, in this linear model, the time-average velocity due to that charge is zero. In the system of symmetric electrodes, sq can be computed from the RC-circuit model shown in Fig. 3. In this model, the electrical current in the electrolyte is discretized in current tubes of semicircular shape and width Dz. The resistance of one of these tubes is R ¼ pz=sdDz, where z is the distance to the interelectrode gap center, s the electrolyte conductivity, and d the depth of the electrodes. The electrical current in the tubes charges the electrical double layer, represented in this model by a distributed capacitor with a capacitance per surface area that, at low
AC Electroosmosis: Basics and Lab-on-a-Chip Applications
R=
πz sdΔz
27
resistance of the current tube at z
A
η∇2v – ∇P = 0
∇2Φ = 0
A
∇• v = 0 capacitance of the EDL
υn = 0 υt = υslip
liquid
σ
electrodes
∂Φ = iωCDL(Φ – Vj) ∂n
Δz
distance z
electrodes
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 3 RC-circuit model for computing the charge density in the electrical double layer as a function of the position on the electrodes. The resistance of the liquid is modeled by the resistors. The capacitance of the electrical double layer is represented by the capacitors
amplitudes, can be estimated from the Debye–H€uckel theory as edDz=lD . The value of sq at any position can be computed from sq ¼ eVd =lD , where Vd is the voltage drop across the EDL and it is a function of the position z: V0 Vd ðzÞ ¼ 1 þ jopez=slD
(1)
and the tangential electric field is obtained from Et ¼ @Vd =@z. The time-average of the Helmholtz– Smoluchowski formula reads: 1 sq E t lD eV02 O2 ¼ hvHS i ¼ Re 2 2 8zð1 þ O2 Þ
(2)
where * indicates the complex conjugate and the nondimensional frequency O is written as O ¼ opez=2slD . Maximum velocity is predicted for O ¼ 1, i.e., omax ¼ 2slD =pez. The frequency for maximum velocity is dependent on position on the electrode z, and the electroosmotic velocity vanishes for either O > 1, as expected. The model is in qualitative agreement with experimental measurements [3]. A rigorous study of ACEO flows for small potentials can be found in ref. [4], where the electrokinetic equations are solved in the limit of thin electrical double layer, i.e., the Debye length is much smaller
υ =0 σ∂Φ/∂n = 0 υn = 0 t
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 4 Equations and boundary conditions for computing the fluid velocity generated by ACEO. Indexes n and t indicate normal and tangential component, respectively. The electric potential is obtained from Laplace equation with specific boundary conditions at the electrodes. From the solution of the potential at the level of the electrodes, the slip velocity is computed with (3) and introduced as boundary condition in solving Stokes equation for the fluid velocity
than the size of the electrodes. In this limit, the fluid velocity in the liquid can be obtained from Stokes equation with boundary condition of slip velocity on the electrodes, the electroosmotic velocity. For electrode j, subjected to a harmonic potential VðtÞ ¼ Re Vj expðiotÞ , the slip velocity is computed according to: vslip ¼
e @jF Vj j2 L 4 @x
(3)
where F is the electric potential phasor evaluated at the level of the electrodes (the electric potential in the liquid is written as fðtÞ ¼ Re½F expðiotÞ). The parameter L appears to account for the effect of a compact layer on the electrodes, acting as a capacitor in series with the Debye layer and, as a consequence, diminishing the voltage drop across the latter [5], L ¼ Cs =ðCs þ Cd Þ 1, where Cs and Cd are the capacitances of the compact and Debye layers, respectively. Thus, it is first required to solve the electric potential in the bulk of the liquid, which is solution of Laplace equation with specific boundary conditions, as shown in Fig. 4. Charge accumulation at the
A
28
AC Electroosmosis: Basics and Lab-on-a-Chip Applications
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 5 Side view of the streamlines in ACEO with two coplanar electrodes. (Left) Experiments with ac signals at 100 Hz. Fluorescent particles are used as flow tracers. (Right) Numerical Simulation (Reprinted with permission from [5], # 2002 APS)
electrode–electrolyte interface yields the following boundary condition: s
@F ¼ ioCDL ðF Vj Þ @x
(4)
where s is the liquid conductivity and the total capacitance of the double layer is given by CDL ¼ Cs Cd =ðCs þ Cd Þ. Zero normal current, @F=@n ¼ 0, is imposed at the other boundaries of the domain. Figure 5 shows a comparison of experimental streamlines obtained using fluorescent particles as tracers and streamlines computed with this model [5].
Limiting Effects The simple model presented in the previous section is rigorous for voltages of the order of kB T= e 25 mV or smaller. For these voltages, the Debye–H€uckel approximation is valid, and the capacitance of the EDL per unit area is given by e=lD . This does not hold true at higher voltages, and analytical models become more complicated, especially when the finite size effect of ions must be considered [6]. Another simplification is that the electrodes are considered to be perfectly polarizable, meaning that there is not charge transfer from the electrode to the electrolyte or vice versa. However, Faradaic currents appear for increasing voltage, and it has been found to have an important influence in the flow [7]. Finally, there is one more limitation not related to the amplitude of the applied voltage. The simple model does not consider the possible existence of an oxide layer on top of the electrode, the so-called
compact layer. Very good fit to experimental results were obtained when this layer was taken into account [4, 8]. In general, ACEO flow of electrolytes with high ionic strength (greater or equal to 100 mM) has not been reported. It seems that there are fundamental limitations around this ionic strength related to steric effects at high electrolyte concentrations as well as problems with the generation of Faradaic reactions.
Applications for the Lab-on-a-Chip Current clean-room technologies allow for the fabrication of complex microelectrode structures and their integration in microfluidic devices. Therefore, AC electrokinetic phenomena represent an opportunity for performing standard operations in Lab-on-a-Chip systems. In particular, AC Electroosmosis has been demonstrated to be useful for the following applications: ACEO Pumping and Mixing The example of two coplanar electrodes of the same size is perhaps the easiest system to illustrate the mechanism of ACEO. As a consequence of the symmetry of the system, no net pumping of fluid is produced. The flow rolls on top of the two electrodes are identical and cancel each other. However, as predicted in [9], any electrode array with some asymmetry will lead to net pumping of the electrolyte. Two different electrode arrays have been broadly explored for achieving net pumping of fluid, see Fig. 6. When pairs of asymmetric electrodes are used, the flow roll on top of the larger electrode dominates and net
AC Electroosmosis: Basics and Lab-on-a-Chip Applications
29
A
Electrodes
A –V0cos(ωt) V0cos(ωt)
0°
180° 90°
0° 270°
AC Electroosmosis: Basics and Lab-on-a-Chip Applications, Fig. 6 Microelectrode structures for pumping of electrolytes. (Left) Array of asymmetric couples of electrodes
subjected to a harmonic potential. (Right) Array of equal size electrodes subjected to a four-phase traveling-wave potential
flow is created (from left to right in the figure) [10–12]. Reference [13] demonstrates a microfluidic pump consisting of a long serpentine microchannel lined with 3D stepped electrode arrays, allowing to achieve high pressures with voltages around 1 Vrms. Because the ACEO pump employs low voltage, it has potential applications in portable and/or implantable biomedical microfluidic devices. A different strategy to break the symmetry is the application of a traveling-wave electric potential [14, 15]. In experiments, this is implemented by using an array of equal-size electrodes and applying a sinusoidal potential to each electrode but with a phase lag of 90 between neighbors. Net flow is achieved in the same direction of the traveling wave (left to right in the figure). A common feature of both arrays is that the flow direction is reversed for sufficiently high voltages [16]. This observation cannot be explained by the standard ACEO model, and limiting effects like ion crowding and Faradaic currents should be accounted for. The predominant laminar character of the flow in microfluidics makes mixing a slow process. References [17, 18] show how AC Electroosmosis can be exploited for generating micro-vortices locally and enhancing the mixing.
the generated bulk flow was able to transport DNA molecules from a large effective region to the electrode surface. The detection times for heterogeneous immunoassays can be very long if analytes have to travel only by diffusion to the device surface, where they then bind or react with surface-bound receptors. Reference [22] demonstrates that the fluid flow generated by ACEO can enhance the analyte transport and reduce the detection time. Promising results have recently been obtained by using transparent semiconductors that become conductors when light is projected on them. In this way, light patterns created at will on the substrate will function as electrodes and, for example, allow for dynamic and in situ concentrations of particles [23, 24]. Reference [23] showed the light-patterning of 31,000 microfluidic ACEO vortices on a featureless photoconductive surface which were able to concentrate and transport micro- and nanoscale particles.
Transport and Aggregation of Particles and Molecules AC Electroosmotic flows have also been used in combination with other forces (dielectrophoresis in most cases) for concentration of bio-particles. The flow is used to bring and concentrate the particles on top of the electrodes while the other forces are used for holding them [19, 20]. For example, reference [21] demonstrates ACEO for concentrating DNA molecules;
References
Cross-References ▶ Induced-Charge Electroosmosis
1. Ramos, A., Morgan, H., Green, N.G., Castellanos, A: AC electric-field-induced fluid flow in microelectrodes. J. Colloid. Interface. Sci. 217, 420–422 (1999) 2. Bazant, M.Z., Squires, T.M: Induced-charge electrokinetic phenomena: theory and microfluidic applications. Phys. Rev. Lett. 92, 066101 (2004) 3. Green, N.G., Ramos, A., Gonza´lez, A., Morgan, H., Castellanos, A: Fluid flow induced by nonuniform ac
A 4.
5.
6.
7.
8.
9. 10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
30 electric fields in electrolytes on microelectrodes. I. Experimental measurements. Phys. Rev. E. 61, 4011–4018 (2000) Gonza´lez, A., Ramos, A., Green, N.G., Castellanos, A., Morgan, H: Fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes. II. A linear double-layer analysis. Phys. Rev. E. 61, 4019–4028 (2000) Green, N.G., Ramos, A., Gonza´lez, A., Morgan, H., Castellanos, A: Fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes. III. Observations of streamlines and numerical simulation. Phys. Rev. E. 66, 026305 (2002) Bazant, M.Z., Kilic, M.S., Storey, B.D., Ajdari, A: Towards an understanding of induced-charge electrokinetics at large applied voltages in concentrated solutions. Adv. Colloid. Interface. Sci. 152, 48–88 (2009) Gonza´lez, A., Ramos, A., Garcı´a-Sa´nchez, P., Castellanos, A: Effect of the combined action of Faradaic currents and mobility differences in ac electro-osmosis. Phys. Rev. E. 81, 016320 (2010) Pascall, A.J., Squires, T.M: Induced charge electro-osmosis over controllably contaminated electrodes. Phys. Rev. Lett. 104, 088301 (2010) Ajdari, A: Pumping liquids using asymmetric electrode arrays. Phys. Rev. E. 61, R45–R48 (2000) Brown, A.B.D., Smith, C.G., Rennie, A.R: Pumping of water with ac electric fields applied to asymmetric pairs of microelectrodes. Phys. Rev. E. 63, 016305 (2000) Ramos, A., Gonza´lez, A., Castellanos, A., Green, N.G., Morgan, H: Pumping of liquids with ac voltages applied to asymmetric pairs of microelectrodes. Phys. Rev. E. 67, 056302 (2003) Studer, V., Pe´pin, A., Chen, Y., Ajdari, A: An integrated AC electrokinetic pump in a microfluidic loop for fast and tunable flow control. Analyst 129, 944–949 (2004) Huang, C., Bazant, M.Z., Thorsen, T: Ultrafast highpressure AC electro-osmotic pumps for portable biomedical microfluidics. Lab. Chip. 10, 80–85 (2010) Cahill, B.P., Heyderman, L.J., Gobrecht, J., Stemmer, A: Electro-osmotic streaming on application of traveling-wave electric fields. Phys. Rev. E. 70, 036305 (2004) Ramos, A., Morgan, H., Green, N.G., Gonza´lez, A., Castellanos, A: Pumping of liquids with traveling-wave electroosmosis. J. Appl. Phys. 97, 084906 (2005) Garcı´a-Sa´nchez, P., Ramos, A., Green, N.G., Morgan, H: Experiments on AC electrokinetic pumping of liquids using arrays of microelectrodes. IEEE Trans. Dielectr. Electr. Insul. 13, 670–677 (2006) Wang, S.H., Chen, H.P., Lee, C.Y., Yu, C.C., Chang, H.C: AC electro-osmotic mixing induced by non-contact external electrodes. Biosens. Bioelectron. 22, 563–567 (2006) Harnett, C.K., Templeton, J., Dunphy-Guzman, K.A., Sensousy, Y.M., Kanouff, M.P: Model based design of a microfluidic mixer driven by induced charge electroomosis. Lab. Chip. 8, 565–572 (2008) Wu, J., Ben, Y., Battigelli, D., Chang, H.C: Long-range AC electrosmotic trapping and detection of bioparticles. Ind. Eng. Chem. Res. 44(8), 2815–2822 (2005) Wong, P.K., Chen, C.Y., Wang, T.H., Ho, C.M: Electrokinetic bioprocessor for concentrating cells ad Molecules. Anal. Chem. 76, 6908–6914 (2004)
ac-Calorimetry 21. Lei, K.F., Cheng, H., Choy, K.Y., Chow, L: Electrokinetic DNA concentration in microsystems. Sensor. Actuat. A-Phys. 156, 381–387 (2009) 22. Hart, R., Lec, R., Noh, H: Enhancement of heterogeneous immunoassays using AC electroomosis. Sensor. Actuat. BChem. 147, 366–375 (2010) 23. Chiou, P.Y., Ohta, A.T., Jamshidi, A., Hsu, H.Y., Wu, M.C: Light-actuated AC electroosmosis for nanoparticle manipulation. J. Microelectromech. S. 17, 525–531 (2008) 24. Hwang, H., Park., J.K: Rapid and selective concentration of microparticles in an optoelectrofluidic platform. Lab. Chip. 9, 199–206 (2009)
ac-Calorimetry ▶ Nanocalorimetry
Accumulator ▶ Nanomaterials for Electrical Energy Storage Devices
Acoustic Contrast Factor Andreas Lenshof and Thomas Laurell Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden
Definition The acoustic contrast factor (F) governs whether an object will be affected by the acoustic radiation force in an acoustic standing wave field [1]. It is comprised of the density and speed of the sound of the object (rp and cp) and the surrounding medium (r0 and c0), as seen in the equation below. A positive contrast factor means that the object will move to a pressure node, while a negative factor will move the object toward a pressure antinode:
F¼
rp þ 23 ðrp r0 Þ 1 r0 c20 3 rp c2p 2rp þ r0
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
31
A
Cross-References
Overview
▶ Acoustic Trapping ▶ Acoustophoresis ▶ Integrated Micro-acoustic Devices
The synthesis of functional nanoparticles is an area of tremendous interest, particularly from the standpoint of many industry applications from catalysis, optics, and electronics to biomolecular sensing, regenerative medicine, and pharmaceutical science. In the latter, there are several challenges that current drug delivery technologies and systems face. For example, the low solubility of many drugs in biological environments prevents their absorption and distribution in vivo [1, 2]. While drug solubility limitations can be circumvented through chemical structure modification or the introduction of surfactant, the former often requires expensive and complicated procedures whereas the latter may not only increase the dosage volume but is also associated with toxicity problems [1]. In addition, the delivery of most drugs cannot be localized to target a site of interest, leading to undesirable side effects or immune responses when taken up by uninfected tissues and organs. Further, drugs are often susceptible to decomposition, enzymatic degradation, aggregation, or denaturation, thus either reducing their shelf life or their efficacy in vivo. Nanoparticle-based delivery systems, however, offer the possibility of addressing some of these issues. For example, the drug dissolution rate can be increased by reducing the particle size [1, 2]. Tumor tissues are characterized by leaky vasculature and poor lymphatic clearance; due to their subcellular dimensions, nanoparticles are able to preferentially accumulate in tumor tissue through an enhanced permeation and retention (EPR) effect [3]. As an attractive alternative to conventional invasive surgical procedures, nanoparticles have also been reported to be able to cross the blood–brain barrier to be administrated into the central nervous system [1]. The encapsulation of drugs within the polymer nanoparticle not only acts as a protective layer surrounding the drug from hostile in vivo environments but also enables the drug to diffuse out slowly over extended periods. Such controlled release as well as direct local targeting of diseased regions can also be further tuned by judicious choice of the polymeric excipient chemistry, comonomer ratios, and their degradability in certain regions. Further site-specific targeting can be achieved by functionalizing the nanoparticle surface with tissue-specific or cell-specific ligands to increase uptake into receptor expressing
References 1. Gorkov L.P.: On the forces acting on a small particle in an acoustical field in an ideal fluid. Soviet Physics Doklady (6), 9, 773–775 (1962)
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine Aisha Qi, Peggy Chan, Leslie Yeo and James Friend Micro/Nanophysics Research Laboratory, RMIT University, Melbourne, VIC, Australia
Synonyms Drug delivery and encapsulation; Nanocarriers; Nanomedicine; Polymer nanocapsules; Sound propagation in fluids; Ultrasonic atomization
Definition Sound wave propagation arising from the highfrequency acoustic irradiation of a fluid can generate considerable stresses at the free surface of the fluid leading toward its destabilization and subsequent breakup. If the fluid comprises a polymer solution, the evaporation of the solvent from the aerosols that are generated as a consequence of the atomization process leaves behind a solidified polymer core with submicron dimensions. Here, any discussion of nanoparticle synthesis due to chemical reactions driven by sound waves, i.e., sonochemistry, is omitted and the discourse is limited to the physical synthesis of such nanoparticles due to the atomization of polymer solutions into a gas, typically dry air.
A
A
32
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
cells [3]. Alternatively, it is straightforward to chemically modify the nanoparticle surface to actively target a diseased site by attaching a surface-bound ligand, e.g., monoclonal antibodies or polymeric conjugates that specifically bind to target cells [1]. There are numerous routes for nanoparticle synthesis, all of which can be broadly delineated into three general approaches. Mechanical milling techniques are widely used commercially but are typically confined to industrial applications such as ceramics and paints due to contamination issues. While the level of impurities can be reduced, for example, by carrying out the process in vacuum, this is expensive and the powders are often polydispersed. Wet chemical deposition and precipitation techniques involve phase separation and include crystallization and sol-gel processing. Polymeric nanoparticles can be synthesized using emulsion (solvent extraction/evaporation) and self- or directed-assembly methods. These are usually batch operations and mass production can lead to scale up in complexity and cost. Moreover, while these methods allow for size and size distribution reproducibility, controlling these parameters is difficult. Gas phase evaporation and condensation techniques, on the other hand, encompass a range of methods that include combustion flame pyrolysis, plasma chemical vapor deposition, and laser ablation as well as spray drying. As with all aerosol processing strategies, spray drying is typically straightforward, fast, and allows high throughput, producing dried nanoparticles by atomizing a solution into micron or submicron-sized aerosols which then rapidly pass through a drying configuration. By removing the solvent either through evaporation or extraction using a hardening agent [4–6], the material of interest, e.g., proteins, peptides, and a wide range of polymers, can solidify into dried particles usually with desired dimensions and morphologies governed by the parent aerosol [7]. Other aerosol processing techniques include hydrodynamic flow focusing and electrospraying. Hydrodynamic flow focusing generates droplets by forcing liquid through an orifice using direct pressure or a fast moving air stream [8], although this typically results in an uneven and large particle size distribution. As such, it is typically used by the food industry, for example, in the manufacture of milk powder, in which
stringent particle size requirements are not essential. Electrospraying, on the other hand, employs large voltages to stretch and pinch off aerosol droplets from a liquid meniscus at the tip of a metal capillary [9]. While this produces aerosol droplets and polymeric nanoparticles with a uniform size distribution, the throughput is generally low and the large electric fields required not only poses safety hazards but could also result in molecular lysis, although these could potentially be circumvented using high-frequency AC fields [10]. New technologies for nanoparticle synthesis and drug encapsulation, however, have not kept up with the rapid advances achieved in nanomedicine, particularly, from the standpoint of drug discovery and formulation. The rest of this entry is concerned with atomization processes via acoustic means for polymeric nanoparticle synthesis and recent advances in this field that allow it to be exploited as a powerful tool for drug delivery.
Ultrasonic Atomization Ultrasonic atomization generally refers to a method for aerosol production induced by irradiating a fluid in order to destabilize its interface with acoustic energy at driving frequencies between 20 kHz to a few MHz [11], although novel technologies such as surface acoustic waves (SAWs) allow operation at higher frequencies above 10 MHz [12]. The source of the acoustic energy is usually a piezoelectric transducer driven by an alternating electrical signal. Ultrasonic atomization therefore typically encompasses both indirect and direct vibration-induced atomization [11, 13], ultrasonic microjetting [14], and, more recently, SAW atomization [15]. Figure 1 illustrates these various conceptual strategies that are generally classified as ultrasonic atomization methods, the difference between these arising through the mechanism by which the acoustic energy is introduced or the way in which the aerosol droplets are produced. For example, bulk vibration energy is transferred indirectly from the piezoelectric transducer to the fluid drop to be atomized through a metallic horn (Fig. 1a) or directly to the sessile fluid drop placed on the piezoelectric
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 1 Different conceptual configurations for ultrasonic atomization. (a) Indirect bulk vibration driven by a piezoelectric transducer via a metallic horn on which the working fluid drop is deposited. (b) Direct bulk vibration transmitted to the fluid drop placed on the piezoelectric transducer. (c) Use of micron-sized nozzles and orifices to drive ultrasonic microjets. (d) Transmission of surface vibration energy in the form of SAWs into a fluid drop to drive interfacial destabilization
33
a
A
b Delivery of liquid to the metal “horn”
A
Reservoir
Transmitter
Piezoelectric Transducer
Piezoelectric Transducer d
Capillary wave
IDT
Streaming
c
Mesh Reservoir
Air
Piezoelectric substrate
Fluid
Piezoelectric Transducer
substrate (Fig. 1b). Nozzles and orifice plates from which fluid jets emanate that subsequently break up into aerosol droplets can also be used in an extension of the indirect method to provide finer control over the final droplet size (Fig. 1c). Alternatively, a nozzle can assume the place of the horn in the direct method in Fig. 1a, with the fluid issuing through the needle instead of placing the fluid on the horn. Nozzles or orifices however tend to clog over time and require constant cleaning; the pressure drop is also significantly increased, and hence more power is typically required to drive the atomization process. Alternatively, the surface vibration in the form of SAWs can couple acoustic energy into a fluid drop sitting on the piezoelectric substrate to effect interfacial destabilization (Fig. 1d). The operating principle and governing mechanisms that underpin ultrasonic atomization (cavitation or capillary wave destabilization) are discussed in detail in Yeo et al. [12]. The average diameter D of the aerosol droplets produced depends on the most unstable wavelength l of the capillary wave instability that is induced, specified by the Kelvin equation, which essentially arises from a dominant force balance between the stabilizing capillary stresses and the destabilizing stress imposed by the vibration:
SAW propagation
D C1 l C2
2pg rfc2
13
;
(1)
where g and r are the surface tension and density of the fluid, respectively, and fc is the capillary wave frequency. C1 and C2 are empirically determined coefficients commonly used to fit the experimental data to the predictions, and vary widely in the literature, although they are typically of order unity. It should be noted that Kelvin’s theory does not provide a way in which the capillary wave frequency is related to the forcing frequency due to the acoustic vibration, and the usual assumption made (see, for example, [15]) is that the capillary waves are excited at a subharmonic frequency that is one half of the forcing frequency, i.e., fc ¼ f/2. Recent studies have however shown that this may not always be true and that the capillary wave frequency can be predicted from a dominant balance between capillary and viscous stresses, at least for a sessile drop or thick liquid film [12, 15]:
fc
g ; mL
(2)
A
34
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
where m is the viscosity of the fluid and L is the characteristic length scale of the drop or film. A correction factor (H/L)2, wherein H is the characteristic height scale, can be included to account for the geometry of the parent drop, in particular, the axial capillary stress, when substituting Eq. 2 into Eq. 1 [15]. SAWs, generated by applying an oscillating electrical signal to the interdigital transducer electrodes patterned onto a piezoelectric substrate (Fig. 1d), the gap and width of which specifies the SAW wavelength and hence the resonant frequency (typically 10–100 MHz), have also been shown as an effective mechanism for atomizing drops. Essentially, SAWs are nanometer order amplitude ultrasonic waves that are confined to and propagate along the substrate surface in the form of a Rayleigh wave. Owing to high MHz order frequencies and substrate displacement velocities, on the order of 1 m/s in the vertical direction perpendicular to the substrate, substrate accelerations on the order of 107 m2/s arise. This, together with the strong acoustic streaming that is induced within the fluid drop placed on the substrate, leads to fast destabilization and breakup of the drop free surface to produce 1–10 mm dimension aerosol drops [15], as illustrated in Fig. 1d. One advantage of the SAW atomization over its ultrasonic counterparts is the efficient energy transfer mechanism from the substrate to the liquid – unlike bulk vibration, most of the acoustic energy of the SAW is localized within a region on the surface of the substrate about 3–4 wavelengths thick (a SAW wavelength is typically on the order of 100 mm) and is transferred to the liquid drop to drive the atomization. Consequently, it is possible to atomize fluids using the SAW at input powers of around 1 W, which is one to two orders of magnitude smaller than that required using other ultrasonic methods. Another advantage of the SAW is the high frequencies (>10 MHz) that can be assessed. The timescale associated with the period (inverse frequency) of the oscillating acoustic and electromechanical field at these frequencies is much shorter than the hydrodynamic timescale mL/g 104 s as well as the 105–106 s relaxation timescales (inverse of the strain rate) associated with shear-induced molecular lysis [12]. In addition, it is not possible to induce cavitation, which is known to cause molecular lysis, at the low powers
and high frequencies atomization [15].
associated
with
SAW
Synthesis of Polymeric Microparticles and Nanoparticles Early work to demonstrate the possibility of synthesizing polymeric particles, albeit with micron dimensions, was reported by Tsai et al. [16] and Berkland et al. [4]. The apparatus in both studies is similar, based on concept known as two-fluid atomization. A piezoelectric transducer is used to vibrate a nozzle or orifice from which a jet comprising the working fluid issues and which subsequently suffers from Rayleigh-Plateau instabilities to break up into individual aerosol droplets. In a manner similar to the Kelvin equation given in Eq. 1, the droplet diameter is controlled by the most unstable wavelength of the axisymmetric instability, which can be predicted from a dominant force balance between the inertia imposed on the jet and the capillary stresses that stabilize it, and is a function of the diameter of the undisturbed jet, which is slightly larger than the nozzle or orifice diameter. In both cases, an external annular stream surrounding the nozzle through which the liquid jet issues is employed. The annular sheath fluid consists of air in the former and an immiscible carrier liquid that does not dissolve the polymer solution in the latter. In both cases, the sheath fluid, whether air or a second immiscible liquid, is flowed much faster than the jetting fluid although the underlying reason provided for its necessity differs. In the former, the airflow is suggested to vibrate in resonance with the vibrating nozzle due to the coaxial annular arrangement, resulting in a magnification of the amplitude of the capillary waves on the liquid jet. In the latter, the interfacial shear imposed by the carrier fluid surrounding the jet is suggested to aid the primary breakup of the jet away from its parent fluid at the orifice, allowing the production of droplets one order of magnitude smaller. 40 mm diameter xanthan gum particles were synthesized in the former when driven at a fundamental resonant frequency of 54 kHz through a 0.93 mm nozzle whereas 5–500 mm diameter poly(D,L-lactic-co-glycolic acid) (PLGA) particles (Fig. 2) were synthesized in the latter when driven at frequencies between 19 and 70 kHz through 60 and
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
100 mm diameter orifices. The carrier fluid, poly(vinyl alcohol), in which the PLGA particles were collected and solidified within, however, remained on the surface of the particles, which could potentially affect physical and cellular uptake [3]. Forde et al. [17] subsequently showed the possibility of synthesizing poly(e-caprolactone) (PCL)
35
nanoparticles around 200 nm in dimension by atomizing a PCL/acetone solution using the direct method shown in Fig. 1b involving a piston-vibrated hard lead zirconate titanate piezoelectric disk between 1 and 5 MHz, as illustrated in Fig. 3. The nanoparticle diameter Dp can be estimated from volume considerations [18]: Dp D
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 2 45 mm PLGA microparticles synthesized by ultrasonically atomizing the polymer solution through a 60 mm orifice which is indirectly vibrated using a piezoelectric transducer at frequencies between 19 and 70 kHz (Reprinted with permission from Berkland et al. J. Control. Release 73, 59–74 (2001). Copyright (2001) Elsevier)
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 3 Schematic illustration of the direct ultrasonic atomization setup of Forde et al. in which a sessile drop is vibrated at 1.645 MHz or 5.345 MHz in a piston-like manner using a hard PZT disk (Reprinted with permission from Forde et al. Appl. Phys. Lett. 89, 064105 (2006), Copyright (2006) American Institute of Physics)
A
13 C ; r
(3)
wherein C denotes the initial polymer concentration, such that it is possible to tailor the nanoparticle size by tuning the aerosol size through its physical properties as well as the geometry of the parent drop, as suggested by Eq. 2. Polymeric nanoparticles have also been produced using SAW atomization. Friend et al. [19] demonstrated the synthesis of 150–200 nm clusters of 5–10 nm PCL nanoparticle aggregates, as shown in Fig. 4, the cluster size being weakly dependent on several parameters such as the surfactant used and the drying length. The grape-bunch-like cluster morphology was attributed to nonuniform solvent evaporation during in-flight drying of the droplets that drive a thermodynamic instability. This results in spinodal decomposition and phase separation, which gives rise to separate regions that are polymer-rich and solventrich. The polymer-rich regions then solidify more rapidly, creating nucleation sites that lead to the formation of a cluster of PCL molecules until a critical nucleation size is attained, estimated from classical nucleation theory to be around 10 nm and consistent
PZT element, electroded both faces
Fluid drop φ0.1 mm Fluid supply tube 8 cm Atomized particles
SDS in DI H2O
Signal generator + RF Amplifier
Syringe pump
A
A
36
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
a
15 10 5 0 10
AFM flattened agglom.
20
DLS SDS
TEM
25
AFM individual particles
Number percentage
30
50 100 Particle diameter (nm)
b 30 nm Nanoparticles
50 nm
DLS Tween-20
500
Agglomerated particle 200 nm
230 nm
0 nm
1000
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 4 The left panel shows dynamic light scattering measurements of the PCL nanoparticles produced using SAW atomization. The right panel shows (a) transmission
electron microscopy and (b) atomic force microscopy images showing 150–200 nm clusters of 5–10 nm nanoparticles (Reprinted with permission from Friend et al. Nanotechnology 19, 145301 (2008). Copyright (2008) Institute of Physics)
with the 5–10 nm nanoparticles observed. In addition to PCL, 1–10 mm dimension protein (bovine serum albumin and insulin) aerosols have been generated using the SAW, which evaporate in flight to form 50 to 100 nm nanoparticles [18].
consistent with the other studies reported below that encapsulated particles are generally of micron-order dimension due to the size of the peptides and proteins within. The frequency at which the ultrasonic atomization was carried out was also unspecified. Reported encapsulation efficiencies, obtained by dissolving the polymer particles in the original solvent (or an acetonitrile/chloroform mixture) to release the encapsulated material which is then recovered on a 200 nm cellulose filter followed by elution with phosphate buffered saline and quantitative characterization using a spectrofluorometer (protein) or through a HPLC assay (peptides), were typically low, between 10% and 35%. This was attributed to comparable extraction rates between that of the aqueous phase in which the protein or peptide resides with that of the polymer solvent through diffusion and partitioning. Consequently, the protein or peptide is removed during washing. When a different peptide which is only weakly soluble in water was encapsulated, the efficiency was observed to increase to 63% and 93%. As such, it is instructive to note here that encapsulation efficiencies are therefore likely to be higher if solvent evaporation is used as the polymer desolvation method instead of solvent extraction, since the solvent is much more likely to be removed rapidly through evaporation while in flight at a rate much faster than that of water, therefore resulting in more efficient encapsulation of the aqueous phase and the
Drug Encapsulation Early work on drug encapsulation via ultrasonic atomization was carried out by Felder et al. [5], who showed the possibility of encapsulating protein and peptides into poly(lactic acid) (PLA) and PLGA directly using ultrasonic atomization. The polymer solution was sonicated with a protein (bovine serum albumin; BSA) or peptide solution (2–10% w/w concentration) to create a stable water-in-oil emulsion, which was then atomized into a beaker in which hardening agent (octamethylcyclotetrasil, hexane, isopropyl myristate, or water) was agitated to yield 10–100 mm order microparticles. This solidification method relies on solvent extraction and hence polymer desolvation within the hardening agent as opposed to solvent evaporation inflight used in the other methods reported here. While smaller nanoparticles on the order of 100 nm diameter were also reported, it is not clear whether these contained any encapsulated material, as the authors only reported size distribution data for the naked PLA and PLGA particles. It would, however, be
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
37
A A
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 5 Confocal microscope images of the cross-sectional slice (top, bottom, and center) across a PCL microparticle showing the encapsulation of fluorescent biotin
proteins or peptides within the fast solidifying polymer shell. BSA was also encapsulated in PLGA using 100 kHz ultrasonic atomization of a similar water-in-oil emulsion in a study by Freitas et al. [6]. The solvent was evaporated and the solidified particles were collected and further desolvated in an agitated aqueous solution and subsequently recovered using a filter. To allow for aseptic encapsulation conditions, the conventional spray dryer unit was shortened by running the unit under near-vacuum conditions instead of using hot air, and the cyclone unit that is usually present in spray dryers was replaced by an aqueous collection bath; in that way, the entire unit can be placed within a laminar flow chamber. Again, the particles were in the micron dimension (13–24 mm mean diameter). As expected due to the use of solvent evaporation over solvent extraction, the BSA encapsulation efficiencies were higher, around 50–60% compared to 10–35%, which the authors claim could be improved if BSA loss in the aqueous collection solution was replaced by a fluid which is a nonsolvent for BSA. Alternatively, they suggest reducing the polymer concentration. Plasmid DNA (pDNA) and poly(ethyleneimine) (PEI) complexes were encapsulated within 10–20 mm diameter PLGA microparticles using a 40 kHz ultrasonic atomizer [20]. Solvent extraction via a poly (vinyl alcohol) hardening agent was used in this case. Encapsulation efficiencies in a wide range between
within (Reprinted with permission from Alvarez et al. Biomicrofluidics 3, 014102 (2009). Copyright (2009) American Institute of Physics)
To vacuum Water bath (~50 °C) Complementary polymer solution
Polymer solution
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 6 Schematic illustration of the SAW atomization setup involving a single atomization–evaporation– resuspension step used to deposit a single polymer layer and to resuspend it in a complementary polymer solution. The step is repeated to deposit subsequent layers for as many layers as required
20% and 90% were reported, with lower polymer concentrations and higher pDNA-PEI volume fractions giving higher efficiencies; contrary to prior claims,
38
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 7 Dynamic light scattering measurements of (a) single-layer chitosan nanoparticles, (b) bilayer chitosan and CMC nanoparticles, and (c) pDNA encapsulated chitosan nanoparticles
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine 45 40 35 30 Volume (%)
A
A
25 20
B
15 10 C
5 0
100
10
1000
10000
Size (nm) A: Chitosan nanocapsules suspended in CMC solution B: Chitosan/CMC (condensed) bilayer nanocapsules C: pDNA-encapsulated Chitosan nanocapsules suspended in CMC solution
100 Cumulative DNA release (%)
the authors however did not find significant dependence of the encapsulation efficiency on the N/P ratio (N referring to the nitrogen content and P referring to the DNA phosphate content of the pDNA-PEI complex). These values were however obtained by subtracting the mass of pDNA remaining in the hardening agent, measured using a spectrofluorometer, from that in the feedstock, which does not take into account loss of pDNA in the environment during atomization, the amount of unatomized pDNA left on the atomizer as well as the pDNA that resides on the surface of the particles instead of being encapsulated within. The authors also investigated the postatomization structural integrity of the pDNA using gel electrophoresis due to the susceptibility of pDNA to shear degradation, and found 80% retention of the pDNA in supercoiled conformation. This was attributed to cationic complexation between pDNA and PEI, which reduces its size and hence the possibility of shear-induced degradation (when naked pDNA was assessed, only 8% of this remained supercoiled). Nevertheless, PEI has been known to induce cytotoxic effects and hence its use may be limited for gene transfection and delivery. An alternative mitigation strategy involving multilayer polymer nanoparticles will be discussed in the next section. Alvarez et al. [18] later demonstrated the encapsulation of BSA in PCL particles using SAW
90 Sample 1
80 70 60
Sample 2
50 40 30 20 10 00
Sample 1: pDNA-Chi complex/CMC Sample 2: pDNA-Chi complex/CMC/Chi 100
200
300
400
500
Time (hours)
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 8 In vitro release profile showing the diffusion of pDNA out of chitosan/CMC bilayer nanoparticles (sample 1) and chitosan/CMC/chitosan trilayer nanoparticles (sample 2) acquired through fluorescence absorbance measurements of fluorescently labeled pDNA. The lines are added to aid visualization
atomization. With 10 MHz SAWs, the particle sizes had a mean of 23 mm whereas this decreased to around 6 mm with 20 MHz SAWs. Proof of encapsulation was acquired through confocal image slices of the particles showing the fluorescently tagged BSA appearing not only at the top and bottom of the cross-sectional slices across the particle but also in the particle center,
Acoustic Nanoparticle Synthesis for Applications in Nanomedicine Acoustic Nanoparticle Synthesis for Applications in Nanomedicine, Fig. 9 Confocal microscopy images of (a) COS-7 cells and (b) human mesenchymal progenitor cells (MPCs) transfected with pDNA encoded with a yellow fluorescent protein (the expression is depicted in the images in green) encapsulated in PEI/CMC bilayer nanoparticles
a
Cos-7 Cells
39
b
A MPCs
A 10 μm MPCs
10 μm
thus verifying that the BSA was entrapped within and not simply bound to the surface of the particle (Fig. 5). The encapsulation efficiency value reported was slightly different compared to other studies in that 54% of BSA from the feedstock was found to be encapsulated with its chemical structure intact (as opposed to the total BSA content encapsulated reported in previous studies). Thus, the total encapsulation efficiency value could be significantly higher (if a comparison were to be made with the values in Felder et al. [5] and Freitas et al. [6]). As discussed above, the high-frequency operation of the SAW limits the amount of shear and cavitation damage caused to the molecules, and constitutes a considerable advantage of using SAWs over conventional ultrasonic atomization.
Multilayer Nanoparticle Synthesis and Encapsulation Multilayered polymer particles offer significant advantages over a particle comprising a single layer. Layers of different polymers with varying chemistry and hence degradation profiles offer the possibility of tuning the drug release over time and in different physiological regions, therefore affording tremendous opportunities for targeted and controlled release delivery. Very recently, SAW atomization has been exploited to demonstrate the potential for rapidly synthesizing nanoparticles with alternating layers of complementary polymers of opposing charge [21]. This is done through a variation of the usual procedure of atomizing an initial polymer solution followed by evaporated-assisted solidification to produce a singlelayer polymer nanoparticle. In addition, however, the
10 μm
solidified polymer particle is collected in a solution in which the second polymer is dissolved (Fig. 6), which is then re-atomized and dried to deposit the second polymer layer over the initial polymer core. By collecting the two-polymer-layer nanoparticle in a solution comprising the polymer to form the third layer and re-atomizing, a further layer can be deposited. The atomization–evaporation–resuspension procedure is then repeated for as many times as the number of layers desired. The requirement of the polymers comprising alternating layers is that each successive polymer must be complementary to the previous polymer, i.e., they must have opposing charges. In addition, the polymer making up the subsequent layer must be soluble in a solvent that cannot dissolve the polymer making up the previous layer. Up to eight layers of alternating chitosan (or PEI) and CMC layers were synthesized; proof of the deposition of successive layers was provided by visual inspection (atomic force microscopy), charge characterization (reversal of the zeta-potential after the deposition of each successive layer), Fourier transform infrared spectrometry showing ionic complexation between the deposited layers, and fluorescence measurements of labeled polymers. In addition, pDNA was also encapsulated within the multilayer polymeric nanoparticles to demonstrate the therapeutic capability of the nanoparticles in particular for gene therapy. Figure 7 shows the size distribution of negatively charged chitosan and positively charged CMC polymer bilayer particles. The decrease in size upon deposition of the second CMC layer over the chitosan core or the encapsulated pDNA can be attributed to ionic complexation that tends to compact the particle size. Even with encapsulation, however, the size of the
A
40
particles remains in the 100–200 nm range, which is a significant advance over the micron-sized particles obtained whenever a therapeutic molecule is encapsulated (see previous section). In targeted cancer therapy, for example, the mean vascular pore size of most human tumors is around 400 nm [1], and hence extravasation is likely to be more effective with the multilayer nanoparticle drug carriers synthesized through this technique. Figure 8 shows in vitro release profiles of the pDNA, showing the possibility for slowing and hence controlling the release with the deposition of an additional layer. Good in vitro DNA transfection is also demonstrated in COS-7 and human mesenchymal progenitor cells, as seen in Fig. 9.
Cross-References ▶ Nanoencapsulation ▶ Nanomedicine ▶ Nanoparticles ▶ Polymer Coatings
References 1. Kumar, M.N.V.R. (ed.): Handbook of Particulate Drug Delivery. American Scientific, California (2008) 2. Mehnert, W., M€ader, K.M.: Solid lipid nanoparticles: production, characterization and applications. Adv. Drug. Deliv. Rev. 47, 165–196 (2001) 3. Panyam, J., Labhasetwar, V.: Biodegradable nanoparticles for drug and gene delivery to cells and tissue. Adv. Drug. Deliv. Rev. 55, 329–347 (2003) 4. Berkland, C., Kim, K., Pack, D.W.: Fabrication of PLG microspheres with precisely controlled and monodisperse size distributions. J. Control. Release 73, 59–74 (2001) 5. Felder, C., Blanco-Prieto, M., Heizmann, J., Merkle, H., Gander, B.: Ultrasonic atomization and subsequent polymer desolvation for peptide and protein microencapsulation into biodegradable polyesters. J Microencapsul. 20, 553–567 (2003) 6. Freitas, S., Merkle, H., Gander, G.: Ultrasonic atomisation into reduced pressure atmosphere – envisaging aseptic spray-drying for microencapsulation. J. Control. Release 95, 185–195 (2004) 7. Alvarez, M., Friend, J., Yeo, L.Y.: Rapid generation of protein aerosols and nanoparticles via surface acoustic wave atomization. Nanotechnology 19, 455103 (2008)
Acoustic Particle Agglomeration 8. Gan˜a´n-Calvo, A.M.: Enhanced liquid atomization: from flow-focusing to flow-blurring. Appl. Phys. Lett. 86, 214101 (2005) 9. Grace, J., Marijnissen, J.: A review of liquid atomization by electrical means. J. Aerosol. Sci. 25, 1005–1019 (1994) 10. Yeo, L.Y., Gagnon, Z., Chang, H.-C.: AC electrospray biomaterials synthesis. Biomaterials 26, 6122–6128 (2005) 11. Friend, J., Yeo, L.Y.: Microscale acoustofluidics: microfluidics driven via acoustics and ultrasonics. Rev. Mod. Phys. 83, 647–704 (2011) 12. Yeo, L.Y., Friend, J.R., McIntosh, M.P., Meeusen, E.N.T., Morton, D.A.V.: Ultrasonic nebulization platforms for pulmonary drug delivery. Expert. Opin. Drug. Deliv. 7, 663–679 (2010) 13. James, A.J., Vukasinovic, B., Smith, M.K., Glezer, A.: Vibration-induced drop atomization and bursting. J. Fluid. Mech. 476, 1–28 (2003) 14. Meacham, J.M., Varady, M.J., Degertekin, F.L., Fedorov, A.G.: Droplet formation and ejection from a micromachined ultrasonic droplet generator: visualization and scaling. Phys. Fluids 17, 100605 (2005) 15. Qi, A., Yeo, L., Friend, J.: Interfacial destabilization and atomization driven by surface acoustic waves. Phys. Fluids 20, 074103 (2008) 16. Tsai, S.C., Luu, P., Song, Y.L., Tsai, C.S., Lin, H.M.: Ultrasound-enhanced atomization of polymer solutions and applications to nanoparticles synthesis. Trans. Ultrason. Symp. 1, 687–690 (2000) 17. Forde, G., Friend, J., Williamson, T.: Straightforward biodegradable nanoparticle generation through megahertzorder ultrasonic atomization. Appl. Phys. Lett. 89, 064105 (2006) 18. Alvarez, M., Yeo, L.Y., Friend, J.R., Jamriska, M.: Rapid production of protein-loaded biodegradable microparticles using surface acoustic waves. Biomicrofluidics 3, 014102 (2009) 19. Friend, J.R., Yeo, L.Y., Arifin, D.R., Mechler, A.: Evaporative self-assembly assisted synthesis of polymeric nanoparticles by surface acoustic wave atomization. Nanotechnology 19, 145301 (2008) 20. Ho, J., Wang, H., Forde, G.: Process considerations related to the microencapsulation of plasmid DNA via ultrasonic atomization. Biotechnol. Bioeng. 101, 172–181 (2008) 21. Qi, A., Chan, P., Ho, J., Rajapaksa, A., Friend, J., Yeo, L.: Template-free synthesis and encapsulation technique for layer-by-layer polymer nanocarrier fabrication. ACS Nano, doi:10.1021/nn202833n
Acoustic Particle Agglomeration ▶ Acoustic Trapping
Acoustic Trapping
Acoustic Separation ▶ Acoustophoresis
Acoustic Trapping Mikael Evander and Thomas Laurell Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden
41
A
air and make it possible to position and levitate liquid droplets or small and light solid matter. The use of radiation forces on particles and cells in liquidsuspensions have been a more common approach than the open-air levitators however. Baker showed in 1972 that a band-formation owing to the radiation forces could be observed when subjecting erythrocytes to a 1 MHz standing wave in polystyrene containers [2]. Since then, the technique has been refined and successfully incorporated into the lab-on-a-chip field as one of several methods to control cells and particles in laminar flows.
Synonyms
Theory
Acoustic tweezers; Acoustic particle agglomeration
In order to create a standing wave, the geometry of the channel/cavity must match the actuation frequency of the transducer. To create a single pressure node, the length of the resonance cavity should be l/2, where l is the wavelength of the ultrasound in the media. To increase the trapping capacity, this can then be scaled so that the number of pressure nodes where objects can be trapped is equal to n*l/2. For water-based buffers, the sound velocity is typically 1,500 m/s. So in order to create a standing wave at 2 MHz, the resonance cavity should be 375 mm (v/2f). There are several acoustic forces acting on the objects in an acoustic standing wave, the primary radiation force, the lateral radiation force, and interparticular secondary forces. The most dominant force is the primary radiation force (Eq. 1) that is proportional to the particle volume (Vc) and is thus strongly size dependent [3]. The force also depends on the acoustic wavelength (l), the acoustic pressure amplitude (p0), and the particles’ position in the acoustic wave (x). The acoustic contrast factor (Eq. 2) is dependent on the density and compressibility of both the particle (rc, bc) and the medium (rw, bw).
Definition Acoustic trapping is the immobilization of particles and cells in the node of an ultrasonic standing wave field.
Overview Acoustic trapping has been shown to be a gentle way of performing non-contact immobilization of cells and particles in microfluidic systems. Localized ultrasonic standing waves are created in microfluidic channels, cavities, or other small, confined spaces creating pressure nodes that attract and hold particles and cells. Commonly, structures in the range of a couple of 100 mm are used, corresponding to acoustic frequencies in the MHz-range.
Background Some 30 years ago, NASA and ESA started to develop containerless processing and one of several possible techniques was acoustic trapping/levitation [1]. These open-air levitators, still in use today, are rather large instruments that create a standing wave in
2 pp0 Vc bw Fr ¼ fðb; rÞ sinð2kxÞ 2l fðb; rÞ ¼
5rc 2rw bc 2rc þ rw bw
(1)
(2)
A
A
42
Acoustic Trapping
a
b Reflector
c Reflector
Reflector
Reflector
FLAT FSEC
FPRF
FPRF
Transducer
d
FLAT FSEC
Transducer
Transducer
Transducer
Acoustic Trapping, Fig. 1 A schematic image demonstrating how different acoustic forces act on particles in a standing wave. (a) Before activating the ultrasound, particles will be evenly distributed in the channel. (b) When activating the ultrasound, the particles will be moved into the pressure node by the primary radiation force. (c) Lateral forces will then focus the particles to
the center of the sound field. Once the distance between the particles is short enough, secondary interparticular forces will create an attractive force between the particles and pull the cluster firmer together. (d) The end result is a centered, levitated cluster of particles situated in the middle of the resonance cavity (Reprinted from [5] with permission from the author)
The acoustic contrast factor can change the direction of the force so that, depending on the material parameters, an object may be pushed toward the pressure node or the pressure antinode. For most cells and particles, the force will be directed toward the pressure node however. Acoustophoresis uses this factor to enable separation between different cell types that may differ in density and/or compressibility. While the primary radiation force is the main force at play in acoustophoresis, the lateral force is what in most trapping designs keeps the particles and cells locked in the standing wave. The lateral force arises from the fact that the wave front of the sound is divergent and has reducing amplitude at the edges of the sound beam. The pressure difference between the center of the sound beam and the edges will create a component that draws objects into the center of the beams and locks them in place. By looking at a dense particle with a radius R positioned in a standing wave with a constant amplitude gradient, Gro¨schl estimated the lateral force component to [4]:
ultrasound (o), the square of the particle radius, and the difference in displacement amplitude between the center (uˆ0) and the edge of the particle (uˆ0 + uˆm). Comparing Eqs. 3 and 1 will show that the lateral force is not as size dependent as the primary radiation force since it only depends on the square of the radius. Also, the greater the amplitude difference (uˆm) between the center of the particle and the edge of the particle, the greater the trapping force. So in order to create a strong acoustic trap, a large lateral pressure gradient is needed. There are also secondary interparticular forces that arise when the incident sound wave is scattered on the objects in the standing wave. These forces only act on short distances but help to form a cluster of the objects and keep them together (Fig. 1).
FLat ¼ pro2 R2 u^0 u^m
(3)
As can be seen in Eq. 3, the lateral force depends on the medium density (r), the angular frequency of the
System Design An acoustic trap is usually designed to either use a localized pressure field or create a localized resonance. An example of a localized pressure field would be a small transducer that couples sound to a fluid only in a small portion of a channel, also called a layered
Acoustic Trapping Acoustic Trapping, Fig. 2 (a) Shows an acoustic trapping approach using a small transducer that creates a very localized pressure field in a channel. The transducer is embedded into the bottom of the fluidic channel and creates a large pressure gradient over a small area; (b) is an example of an acoustic trapping cavity. Here the entire chip is actuated and a standing wave is created only where the geometry matches the ultrasound
43
A
Top view Glass chip
Side view
a
Glass reflector/channel
Trapped objects Transducer PCB Glass chip
b Glass chip
Glass chip
Glass chip Transducer
resonator. The alternative approach is to actuate the entire chip and design a resonance cavity on the chip that matches the actuation frequency. The standing wave will then only exist where the geometry matches the standing wave criteria. Another alternative that has not been as common as the other two designs is the use of focused transducers. By focusing the ultrasound into a very narrow beam and then reflecting the wave in the focal plane, a hemispheric standing wave that can trap and hold objects can be created [6] (Fig. 2).
Designs Using Local Resonance Cavities Designs that use a transducer coupled directly or through a matching layer to the fluid typically have a larger trapping capacity than the cavity approach. However, since the cavity trap is usually created using photolithographic techniques it is easy to array and to integrate with other sample processing steps on chip. An example of a cavity array was presented by Vanherberghen et al. where a 10 10 array was used for aggregating cells for isolated cell studies using microscopy [7] (Fig. 3).
Transducer Glass chip
Designs Using a Localized Transducer Spengler and Coakley presented a trapping system that used a layered design in 2000 [8]. By using a PTFE spacer and a glass reflector, they were able to study cell aggregates in a standing wave at 1.93 MHz. This system was later redesigned to a stainless steel system using a matching layer to couple the ultrasound into the resonator and used for studies of cell membrane spreading [9]. An alternative to using a matching layer is to integrate the transducer in the bottom of a microfluidic channel. A system based around a miniature transducer that was embedded in a printed circuit board was presented by Evander et al. in 2007 [10]. A glass lid with a microfluidic channel was placed on top of the printed circuit board and a standing wave was formed above the transducer in the channel. To test the system, yeast cells were grown in the standing wave and an online viability assay was performed on trapped neural stem cells (see Fig. 4). A further development of the trapping system was presented by Hammarstro¨m et al. [11]. Instead of etched microfluidic channels, a square borosilicate capillary was coupled to an external miniature transducer through a thin glycerol layer. Using
A
A
44
Acoustic Trapping
Acoustic Trapping, Fig. 3 The 10 10 ultrasonic cavity array presented by Vanherberghen et al. [7]. In (a), human B cells are shown in the cavities without ultrasound and in (b) the cells have been aggregated using standing waves (Reproduced by permission of The Royal Society of Chemistry: http://dx.doi.org/ 10.1039/c004707d/)
Acoustic Trapping, Fig. 4 Neural stem cells, HiB5-GFP trapped in an acoustic standing wave. In (a) the cells have just been trapped in the standing wave and in (b) the cells have been levitated for 15 min after which they were perfused with
Acridine Orange to test for viability. The increase in fluorescence indicates that the cells are still viable (Reprinted and modified with permission from [10]. Copyright 2007 American Chemical Society)
Borosilicate Capillary
Ground Contact Metallic Layer
Glycerol
Epoxy PCB
Air Backing
Conductive Silver Epoxy 10 MHz PZT Transducer 1 mm
Acoustic Trapping, Fig. 5 The trapping platform presented in [11] uses a square borosilicate capillary as microfluidic channel and resonance cavity. A system with an external transducer and off-the-shelf capillaries has a lower production cost and is
also compatible with more sensitive assays where disposable channels must be used (Reproduced by permission of The Royal Society of Chemistry: http://dx.doi.org/10.1039/ c004504g)
Acoustophoresis
commercially available capillaries both lowers the production costs and allows for disposable channels for more sensitive analysis.
Cell Viability Several studies have shown that ultrasonic levitation does not seem to have any negative effects on live cells. Hultstro¨m et al. cultured cells that had been levitated for over an hour and studied the doubling times of the cells [12]. No direct or delayed damage could be detected on the cells. Bazou et al. performed a study on mouse embryonic stem cells and confirmed that they could see no changes in gene expression after ultrasonic treatments up to 1 h and the cells maintained their pluripotency [13].
Future Directions of the Field The main research focus has so far been on technology development but as the robustness of the systems is steadily increasing, more application-driven development can be expected. Further work aimed at integrating the trapping systems with other lab-on-a-chip systems can also be expected as well as an ambition to increase the particle size range that can be manipulated, in particular toward smaller objects such as bacteria.
45
A
4. Groschl, M.: Ultrasonic separation of suspended particles – part I: fundamentals. Acustica 84(3), 432–447 (1998) 5. Evander, M.: Cell and particle traping in microfluidic systems using ultrasonic standing waves. Dissertation, Department of Electrical Measurements and Industrial Engineering and Automation, Lund University, Lund (2008) 6. Wiklund, M., Nilsson, S., Hertz, H.M.: Ultrasonic trapping in capillaries for trace-amount biomedical analysis. J. Appl. Phys. 90(1), 421–426 (2001) 7. Vanherberghen, B., et al.: Ultrasound-controlled cell aggregation in a multi-well chip. Lab Chip 10(20), 2727–2732 (2010) 8. Spengler, J.F., et al.: Observation of yeast cell movement and aggregation in a small-scale MHz-ultrasonic standing wave field. Bioseparation 9(6), 329–341 (2000) 9. Coakley, W.T., et al.: Cell-cell contact and membrane spreading in an ultrasound trap. Colloids Surf. B Biointerfaces 34(4), 221–230 (2004) 10. Evander, M., et al.: Noninvasive acoustic cell trapping in a microfluidic perfusion system for online bioassays. Anal. Chem. 79(7), 2984–2991 (2007) 11. Hammarstrom, B., et al.: Non-contact acoustic cell trapping in disposable glass capillaries. Lab Chip 10(17), 2251–2257 (2010) 12. Hultstrom, J., et al.: Proliferation and viability of adherent cells manipulated by standing-wave ultrasound in a microfluidic chip. Ultrasound Med. Biol. 33(1), 145–151 (2007) 13. Bazou, D., et al.: Gene expression analysis of mouse embryonic stem cells following levitation in an ultrasound standing wave trap. Ultrasound Med. Biol. 37(2), 321–330 (2011)
Acoustic Tweezers Cross-References
▶ Acoustic Trapping
▶ Acoustic Contrast Factor ▶ Acoustic Tweezers ▶ Acoustophoresis ▶ Integrated Micro-acoustic Devices
Acoustophoresis
References 1. Lierke, E.G.: Akustische Positionierung – Ein umfassender € €uber Grundlagen und Anwendungen. Acustica Uberblick 82(2), 220–237 (1996) 2. Baker, N.V.: Segregation and sedimentation of red bloodcells in ultrasonic standing waves. Nature 239(5372), 398–399 (1972) 3. Gorkov, L.P.: On the forces acting on a small particle in an acoustic field in an ideal fluid. Sov. Phys. Doklady 6(9), 773–775 (1962)
Andreas Lenshof and Thomas Laurell Department of Measurement Technology and Industrial Electrical Engineering, Division of Nanobiotechnology, Lund University, Lund, Sweden
Synonyms Acoustic separation; Free flow aoustophoresis (FFA)
A
A
46
Acoustophoresis
Definition F¼ “Acoustophoresis” means migration with sound, i.e., “phoresis” – migration and “acousto” – sound waves are the executors of the movement. In related concepts, electric forces move particles in electrophoresis and magnetic forces in magnetophoresis [1]. Acoustophoresis is a noncontact and label-free mode of manipulating particles and cell populations and allows for implementation of several separation modes [2]. The technology is currently finding increased applications in bioanalytical and clinical applications of cell handling and manipulation.
Theory Particles in suspension exposed to an acoustic standing wave field will be affected by an acoustic radiation force [3]. The force will cause the particle to move in the sound field if the acoustic properties of the particle differ from the surrounding medium. The magnitude of the movement depends on factors, such as the size of the particle, the acoustic energy density, and the frequency of the sound wave. The direction of the particle movement is dependent on the density and speed of sound of the particle as well as the liquid media. In an acoustic standing wave generated in a liquid (e.g., water)-filled channel with wall boundaries of dense materials such as metal, silicon, or glass, a standing wave pressure maxima will form at the walls/fluid interface. If the width of the channel is matched to half a wavelength, a pressure node will form in the center of the channel. The most predominant acoustic force acting on microparticles in an acoustic standing wave is the primary axial acoustic radiation force (PRF), Eq. 1. The magnitude of the PRF is dependent on the acoustic energy density, Eac, and the radius, a, of the particle. The direction of the particle movement depends on the acoustic contrast factor F, which comprises the inherent physical properties of the cell or particle, such as the density, rp, and speed of sound, cp, relative to the properties of the surrounding medium, r0 and c0. The sign of the acoustic contrast factor defines the direction of the movement. 3 Frad y ¼ 4pka Eac Fsinð2kyÞ
(1)
rp þ 23 ðrp r0 Þ 1 r0 c20 3 rp c2p 2rp þ r0
(2)
As shown in Fig. 1, most rigid cells and particles (blue particles) have a positive contrast factor and are moved to the pressure node located in the center of the channel. Liquid vesicles or air bubbles will move to the antinodes at the wall (yellow particles). If the size of the flow channel is in micro-domain, the flow conditions is generally laminar, and the particles passing through the standing wave field will be moved to their nodal position and remain in that position throughout the flow channel even after exiting the sound field. By terminating the flow channel with a trifurcation it is possible to separate and/or concentrate the particles from the medium, see Fig. 1D. Beside the primary axial radiation force, also secondary forces act on particles in a standing wave field [4]. These interparticle forces can mostly be negligible as they are only effective when particles are very close to each other. The secondary forces assist in particles forming clusters which are less common in continuous flow systems, but are useful in stagnant systems which rely on agglomeration and sedimentation to clarify medium [5]. When designing flow channels with dimensions that match a half wavelength in the MHz frequency regime, conventional microfabrication offers simple means of fabricating acoustophoresis chips typically in glass or silicon or other high Young’s modulus materials. There are several ways of actuating an acoustophoresis chip. The most straightforward approach is to set up the standing wave between a coupling layer bonded to the transducer and the opposing channel wall, commonly referred to as a layered acoustic resonator, Fig. 2a. Another way is to place the transducer underneath the entire separation system or at any location on the chip surface where space is available. This causes the whole acoustophoresis chip to be actuated, not just the separation channel. By matching the actuation frequency to the channel dimension, it is possible to obtain a standing wave horizontally or vertically or in both directions in the microchannel, Fig. 2b. Twodimensional focusing can be obtained either using two transducers of different frequencies or using a single transducer if the cavity has height-to-width proportions, Fig. 2c, d.
Acoustophoresis
47
a
b
c
A
d
FLOW
A
Acoustophoresis, Fig. 1 Illustration of particles with different sign of the acoustic contrast factor being exposed to an acoustic stranding wave field. (a) No ultrasound active. (b) The radiation force move the particles with positive contrast factor (dark grey) to the pressure node located in the center of the
channel while the particles with negative contrast factor (light grey) move to the anti nodes at the channel side walls. (c) Particles at equilibrium states after acoustic exposure. d) The laminar flow enables separation of the two particle types using a trifurcation at the end of the channel
a
b
1D
Piezo
Piezo
c
d
2D Piezo
Acoustophoresis, Fig. 2 Two ways of actuation of the resonator system. (a) Layered resonator where the standing wave is generated in the direction of the primary direction of actuation and (b) transversal resonance. The resonance is created perpendicular to the primary direction of actuation. (c) Two-dimensional focusing using transducers of two different frequencies; (d) twodimensional focusing using a single transducer
Piezo
Applications Acoustophoretic Enrichment and Depletion of Cells and Particles Acoustophoretic enrichment/concentration of a sample is the most straightforward operation as most cells and particles easily focus in pressure nodes and thus deplete the surrounding medium of solid matter. By using flow splitters such as a trifurcation, the particle dense fraction in the central pressure node can be collected via the middle outlet of the trifurcation, Fig. 3a. Likewise, cell and particle depletion from a fluid is also obtained with the chip configuration of Fig. 3a if the fluid of interest is collected from the side branches. However, like many microfluidic devices presented in the literature, acoustophoretic systems display problems in handling fluids of high concentrations, such as
Piezo
whole blood. Plasmapheresis, i.e., the removal of the cellular content from the blood plasma, is a quite common microfluidic procedure where free plasma is desired for further diagnostic purposes. Commonly these devices are forced to work with dilute blood samples, which deteriorate the plasma composition from an analytical point of view. The acoustical chips presented earlier are no exception and the plasma separation efficiency is known to decrease significantly when the hematocrit is increased above 5–10%). Means to handle samples with higher amounts of cellular content/hematocrit can be realized by a sequential acoustophoretic enrichment and partial removal of enriched cells from the standing wave node region [6]. This requires a microchannel configuration that extends over a longer distance such that the duration in the acoustic field is sufficiently long to focus cells between each point of cell removal.
A
48
Acoustophoresis
a
b
c
Acoustophoresis, Fig. 3 (a) An acoustophoretic concentrator. The particles are concentrated in the central pressure node, while particle-depleted fluid exits to the sides. (b) Depletion of high cell or particle contents. Acoustically concentrated particles leave outlets in the bottom of the chip (black squares) thus lowering the particle load until eventually all particles can be removed through the central duct of the trifurcation outlet. This
mode of operation is suitable for plasmapheresis of whole blood. (c) Wash or fractionation of cells. Sample enters the separation channel from the side branches and becomes laminated against the side walls. Clean buffer enters through the central inlet. The acoustic radiation force moves the large cells into the clean buffer and thus completes a wash in form of a medium switch
Figure 3b shows a schematic of a chip surface conservative meander style layout, where focused blood cells are partially removed at each outlet in the center of the channel bottom indicated as dark squares. At the end of the meander structure, the concentrations of blood cells are sufficiently low for all remaining cells to be removed via the central outlet and the clean plasma fraction can be extracted from the side branches of the trifurcation.
demonstrated for red blood cell washing in glass chips [9] and in washing of microbeads used to selectively affinity capture bacteriophages in a large phage library [10] as well as extraction of phosphopeptides from protein digests using metal ion affinity-specific microbeads [8] and more recently also as a means of performing stain and wash unit operations in flow cytometry applications [11].
Acoustophoretic Cell Washing Microchip acoustophoresis can be configured to perform buffer exchange operations by adding two extra inlets on each side of an acoustophoresis channel. The sample solution is supplied via the side inlets while a clean buffer is provided via the central channel inlet, Fig. 3c. Cells/particles are acoustophoretically moved from their original sample stream along the channel side wall into the central, particle-free medium stream thus exchanging the medium the particles are suspended in [7]. The typical buffer exchange efficiency displays performance data equal or better than conventional centrifugation steps and allows continuous flow-based operation, where several sequential wash steps can be implemented depending on the requirements on the wash efficiency [8]. Acoustophoretic cell and particle washing has been
Acoustophoretic Separation/Fractionation Acoustophoretic separation can be performed in two modes where: 1. The species to be separated display different signs of the acoustic contras factor and hence gather in the acoustic pressure node and antinode, respectively, so-called binary separation. 2. The rate of acoustophoretic transport of a cell or particle into a pressure node determines the lateral position in the flow stream of a given cell or particle type as they leave the separation zone and thus the acoustophoretic mobility determines the outcome of the separation, so-called free flow acoustophoresis – FFA. An early clinical application of microchipintegrated acoustophoretic binary separation targeted blood wash in open-heart surgery. Patient bloodshed during the surgery is contaminated by lipid
Acoustophoresis
microemboli from tissue undergoing surgery and has to be removed prior to autologous retransfusion to the patient. Jo¨nsson et al. [12] demonstrated that acoustophoresis can remove the lipid microemboli from blood, utilizing the fact that erythrocytes display a positive acoustic contrast factor (+0.05) while lipid particles have negative contrast factor (0.18). Hence, the blood cells and the lipid vesicles will move to different lateral positions in the acoustic standing wave field and can thus be separated from each other as they pass through the acoustic separation channel, Fig. 1d, following the principles of binary separation. Binary separation has also been demonstrated as an efficient sample preprocessing step in milk quality control where the lipid emulsion was removed by binary acoustophoresis, enabling online Fourier transform infrared spectroscopic analysis of lactose and protein content of the milk [13]. When performing separation based on free flow acoustophoresis (FFA) the major factor that influences separation outcome is the primary axial radiation force which is dependent on the size, density and speed of sound of the particle, see Eq. 1. This commonly means that larger particles will experience a larger acoustic force than smaller-sized particles and thus size fractionation can be performed by acoustophoresis. Petersson et al. demonstrated successful FFA of suspensions with mixed particle sizes by tuning the acoustic force and the flow rate such that the largest particles reached the pressure node shortly before entering the center outlet and hence, laterally from the center to the side wall a gradient of particle sizes were located, which could be routed to individuals outlets of the acoustophoresis chip [14]. By manipulating the density of the medium through the addition of a buffer that alters the fluid density, it is also possible to improve separation conditions between species that display similar acoustophoretic mobility. Buffer density manipulation in combination with the FFA-fractionation device demonstrated different separation profiles of fractionation of erythrocytes, platelets, and leucocytes [14]. Free flow acoustophoresis has recently also been demonstrated in the preparation of peripheral blood progenitor cells from apheresis product. Platelets are an unwanted contaminant when the apheresis instrument continuously separates the buffy coat from the whole blood as they compromise the subsequent
49
A
immunomagnetic extraction of stem cells from the apheresis product. Acoustophoresis can be used to solve this problem using the same chip design as in Fig. 3c where sample enters the channel at the side inlets and clean buffer enters centrally [15]. The acoustophoretic force on the platelets is very low and thus proceeds with the laminar flow along the channel side walls while the larger leukocytes and the few remaining erythrocytes present in the apheresis product are focused in the center of the channel and exited through the central outlet as a platelet-depleted fraction [15].
Summary Microchip acoustophoresis offers simple means to noncontact handling/processing of cells in continuous flow mode. Acoustophoresis is to a large extent independent of surface charge, ionic strength, and pH variations (within physiological conditions). It is also a non-perturbing technique in the sense that cells are not experiencing any major stress when undergoing acoustophoresis as seen in several viability studies [15, 16].
Cross-References ▶ Acoustic Contrast Factor ▶ Acoustic Trapping ▶ Acoustic Tweezers ▶ Integrated Micro-acoustic Devices
References 1. Lenshof, A., Laurell, T.: Continuous separation of cells and particles in microfluidic systems. Chem. Soc. Rev. 39(3), 1203–1217 (2010) 2. Laurell, T., Petersson, F., Nilsson, A.: Chip integrated strategies for acoustic separation and manipulation of cells and particles. Chem. Soc. Rev. 36(3), 492–506 (2007) 3. Gorkov, L.P.: On the forces acting on a small particle in an acoustical field in an ideal fluid. Soviet Physics Doklady 6(9), 773–775 (1962) 4. Groschl, M.: Ultrasonic separation of suspended particles – part I: fundamentals. Acustica 84(3), 432–447 (1998) 5. Trampler, F., et al.: Acoustic cell filter for high-density perfusion culture of hybridoma cells. Biotechnology 12(3), 281–284 (1994)
A
A
50
Active Carbon Nanotube-Polymer Composites
6. Lenshof, A., et al.: Acoustic whole blood plasmapheresis chip for prostate specific antigen microarray diagnostics. Anal. Chem. 81(15), 6030–6037 (2009) 7. Petersson, F., et al.: Carrier medium exchange through ultrasonic particle switching in microfluidic channels. Anal. Chem. 77(5), 1216–1221 (2005) 8. Augustsson, P., et al.: Decomplexing biofluids using microchip based acoustophoresis. Lab Chip 9(6), 810–818 (2009) 9. Evander, M., et al.: Acoustophoresis in wet-etched glass chips. Anal. Chem. 80(13), 5178–5185 (2008) 10. Persson, J., et al.: Acoustic microfluidic chip technology to facilitate automation of phage display selection. FEBS J. 275(22), 5657–5666 (2008) 11. Lenshof, A., Warner, B., Laurell, T.: Acoustophoretic pretreatment of cell lysate prior to FACS analysis. In: Micro Total Analysis Systems 2010, Groningen (2010) 12. Jonsson, H., et al.: Particle separation using ultrasound can radically reduce embolic load to brain after cardiac surgery. Ann. Thorac. Surg. 78(5), 1572–1578 (2004) 13. Grenvall, C., et al.: Harmonic microchip acoustophoresis: A route to online raw milk sample precondition in protein and lipid content quality control. Anal. Chem. 81(15), 6195–6200 (2009) 14. Petersson, F., et al.: Free flow acoustophoresis: Microfluidic-based mode of particle and cell separation. Anal. Chem. 79(14), 5117–5123 (2007) 15. Dykes J., et al.: Efficient removal of platelets from peripheral blood progenitor cell products using a novel microchip based acoustophoretic platform. PLoS ONE, 6, e23074, (2011) 16. Hultstrom, J., et al.: Proliferation and viability of adherent cells manipulated by standing-wave ultrasound in a microfluidic chip. Ultrasound Med. Biol. 33(1), 145–151 (2007)
Introduction
Active Carbon Nanotube-Polymer Composites
Shape-Memory CNT-Polymer Composites
Jian Chen Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada Department of Chemistry and Biochemistry, University of Wisconsin-Milwaukee, Milwaukee, WI, USA
Synonyms Smart carbon nanotube-polymer composites
Definition Active carbon nanotube-polymer composites are carbon nanotube-polymer composites that display active material functions such as actuation and sensing.
Carbon nanotubes (CNTs) represent a rare class of materials, which exhibit a number of outstanding properties in a single material system, such as high aspect ratio, small diameter, light weight, high mechanical strength, high electrical and thermal conductivities, and unique optical and optoelectronic properties. CNTs are recognized as the ultimate carbon fibers for high-performance, multifunctional polymer composites, where an addition of only a small amount of CNTs, if engineered appropriately, could lead to simultaneously enhanced mechanical strength and electrical conductivity [1, 2]. While most efforts in the field of CNT-polymer composites have been focused on passive material properties such as mechanical, electrical, and thermal properties, there is growing interest in harnessing active material functions such as actuation, sensing, and power generation in designed CNT-polymer composites. The synergy between CNTs and the polymer matrix has been judiciously exploited to create highly desirable active material functions. In this entry, recent progress in active CNT-polymer composites is briefly highlighted with a focus on smart materials and infrared (IR) sensors.
Shape-memory polymers (SMPs) are polymeric smart materials that can memorize a temporary shape and are able to return from a temporary shape to their permanent shape upon exposure to an external stimulus such as heat [3]. Compared with shape-memory alloys and ceramics, SMPs offer a number of distinctive advantages, which include high recoverable strain (up to 400%), low density, ease of processing and the ability to tailor the recovery temperature, programmable and controllable recovery behavior, and low cost. Such advantages could enable a broad spectrum of demanding applications including deployable space structures, morphing wings, information storage, smart textiles, biomedical devices, and drug delivery. Although SMPs show promising shape-memory effects, several major issues remain to be addressed: (1) slow recovery speed; (2) low recovery stress; (3) lack of remote control. Vaia and coworkers demonstrate that the addition of multiwalled CNTs
Active Carbon Nanotube-Polymer Composites
(MWNTs) into a thermoplastic polyurethane matrix (Morthane) could address these issues simultaneously (Fig. 1) [4]. The low recovery speed (up to several minutes) of thermal-responsive SMPs originates from their intrinsically low thermal conductivity (< 0.3 Wm1 K1) [3]. Therefore, the heat transfer from an external heating source to the core of a SMP sample will take considerable amount of time. Vaia and coworkers show that non-radiative decay of IR photons absorbed by the nanotubes raises the internal temperature, melting strain-induced polymer crystallites (which act as physical cross-links that secure the deformed shape), and remotely trigger the release of the stored strain energy. The 1 wt% MWNT-Morthane nanocomposite displays a rapid shape-recovery within 5 s upon exposure to an IR light (Fig. 1b). Comparable effects occur for electrically induced actuation associated with Joule heating of the matrix when a current is passed through the conductive percolative network of the nanotubes within the resin in a 16.7 wt% MWNT-Morthane nanocomposite (Fig. 1d). The low recovery stress of thermal-responsive SMPs originates from their intrinsically low modulus [3]. CNTs have proven to enhance the mechanical properties, particularly modulus, of various polymers. For example, 8.5 wt% MWNTs could increase the room temperature rubbery modulus of Morthane by a factor of 5 [4]. In fact, incorporation of 5 wt% of MWNTs into polyurethane SMPs results in an increase of the recovery stress by 230% [4].
CNT-Polymer Composite Actuators Polymeric actuators are polymers that deform in response to an external stimulus such as heat, light, or electric field, and they normally recover their original shapes after the stimulus is terminated [5]. Potential applications of polymeric actuators include artificial muscles, mini- and microrobots, “smart skins,” pumps, and valves in microfluidic systems for drug delivery, ventricular assist devices for failing hearts, and sensors for mechanical strain, humidity, and gases. Researchers have recently demonstrated that the synergistic combination of CNTs and a suitable polymer matrix could lead to promising CNT-polymer composites with either new or significantly enhanced actuation properties.
51
A
Nematic liquid crystalline elastomers (LCEs) have fascinated scientists and engineers continually since 1981 when they were first synthesized by Finkelmann and coworkers [6]. LCEs bring together, as no other polymer, three important features: orientational order exhibited by the mesogenic units in amorphous soft materials; topological constraints via cross-links; and responsive molecular shape due to the strong coupling between the orientational order and the mechanical strain [7]. Acting together, they create many new physical phenomena that can lead to many potential applications. Nematic LCEs can dramatically and reversibly elongate or contract in response to temperature changes; however, this type of LCE actuator tends to have a slow response time due to the low thermal conductivity of LCEs and cannot be actuated remotely, thereby limiting its potential applications. Terentjev and coworkers have studied the rich photomechanical behavior of three types of MWNTelastomer nanocomposites irradiated with near-IR light, including polydimethylsiloxane, styrene–isoprene–styrene, and a nematic LCE with a polysiloxane backbone [8, 9]. They show that these composite materials have the novel ability to change their actuation direction, from expansive to contractive response, as greater imposed strain is applied to the sample. Unlike MWNTs, which have featureless visible/ near-IR absorption, single-walled CNTs (SWNTs) show strong and specific absorptions in the visible/ near-IR region owing to bandgap transitions [10]. Recent advances in nanotube separation make it possible to obtain enriched single-chirality SWNTs with strong and narrow optical absorptions, which may ultimately offer unique opportunities for the wavelength-selective IR actuation of LCE nanocomposites. The effective use of SWNTs in polymer-composite applications strongly depends on the ability to disperse them uniformly in a polymer matrix without destroying their integrity. Pristine SWNTs are incompatible with most solvents and polymers, which lead to poor dispersion of the nanotubes in solvents and polymer matrices. Chen and coworkers have recently developed a versatile, nondamaging chemistry platform that enables them to modify specific CNT surface properties, while preserving the CNT’s intrinsic properties [11]. They have discovered that rigid, conjugated macromolecules, for example, poly(p-phenyleneethynylene) (PPE), can be used to noncovalently functionalize and
A
A
52
Active Carbon Nanotube-Polymer Composites
Active Carbon Nanotube-Polymer Composites, Fig. 1 (a) Stretched (800%) Morthane ribbon containing 1 wt% MWNTs, tied into a loose knot and heated at 55 C. The knot closes on strain recovery. (b) Strain recovery and curling of the 1 wt% MWNT-Morthane nanocomposite ribbon upon IR irradiation within 5 s. (c) Comparison of the stress recovery before (left) and after (right) remote actuation by IR irradiation.
Neat Morthane (M) bends and does not recover. In contrast, a 1 wt% MWNT-Morthane nanocomposite (PCN) contracts on exposure to IR irradiation (arrow indicates moving direction). (d) Electrically stimulated stress recovery of a 16.7 wt% MWNT-Morthane nanocomposite (Reprinted by permission from Macmillan Publishers Ltd [4])
solubilize CNTs and disperse CNTs homogeneously in polymer matrices [2, 11]. Chen and coworkers have recently reported the reversible IR actuation behavior of a new type of nematic SWNT-LCE nanocomposite using PPE-functionalized SWNTs (PPE-SWNTs) as a filler
in a LCE matrix with side-on mesogenic units [12]. The SWNT-LCE nanocomposites have been prepared through a two-stage photopolymerization process coupled with a “hotdrawing” technique, which enable the fabrication of relatively thick films ( 160–260 mm). The excellent dispersion of
Active Carbon Nanotube-Polymer Composites
53
A
Active Carbon NanotubePolymer Composites, Fig. 2 IR actuation of a 0.2 wt % SWNT-LCE nanocomposite film: (a) before the IR was turned on; (b) after the IR was turned on for 4 s; (c) after the IR was turned on for 10 s; (d) the film returned to its original length with the IR beam turned off (Reprinted by permission from Wiley-VCH Verlag GmbH & Co. KGaA [12])
PPE-SWNTs in the LCE matrix has allowed them to observe a significant and reversible IR-induced strain ( 30%) at very low SWNT loading levels (0.1–0.2 wt %, Fig. 2). The effects of various parameters have been investigated, including the degree of pre-alignment, the loading level of the SWNTs, and the curing time. In SWNT-LCE composites, semiconducting SWNTs can efficiently absorb and transform IR light into thermal energy, thereby serving as numerous nanoscale heaters being uniformly embedded in the LCE matrix. The absorbed thermal energy then induces the LCE nematic-isotropic phase transition, which leads to the shape change of the nanocomposite film. The IR strain response of SWNT-LCE nanocomposites increases with an increasing SWNT loading level and a higher degree of hot-drawing, and decreases with longer photocuring [12]. Electric field driven pure nematic LCE actuators have not been developed because the characteristic rotation energy density is too low compared with the characteristic resistance energy density from the rubbery elastic network due to the low anisotropic dielectric permittivity of the LCEs, even under a very high electric field [13]. However, adding CNTs with high anisotropic
A
polarizability into the LCEs, aligned along the uniaxial director of a monodomain nematic elastomer network, the CNT-LCE nanocomposites below the conductivity percolation threshold can achieve an effective dielectric anisotropy many orders of magnitude higher than the LCEs alone [13]. Therefore, introducing highly polarizable anisotropic CNTs into the LCE will lead to a significantly enhanced electromechanical effect and make electrical-field driven LCE actuators possible. Terentjev and coworkers have demonstrated, for the first time, a large electromechanical response (uniaxial stress of 1 kPa in response to a constant field of 1 MV/m) in nematic LCEs filled with a very low (0.0085 wt%) concentration of MWNTs, pre-aligned along the nematic director at preparation [13]. However, the response stress in the isostrain conditions shows a linear dependence on the applied field E, instead of E4 that would be expected from a CNT rotationinduced electromechanical response, a puzzling finding to the researchers as stated in the original paper [13]. Park and coworkers have investigated the electromechanical properties of a 0.05 wt% SWNT/LaRCEAP (Langley Research Center-ElectroActive Polyimide) composite, which forms an intrinsic
A
54
Active Carbon Nanotube-Polymer Composites
a
V = 0 V ; T* = 22 C
e
t=6s 5 mm
5 mm
b
f
t=9s
g
t = 14 s
V = 400 V ; T* = 67 ± 3 C 74 C 22 C
c
V = 700 V ; T* = 136 ± 6 C 151 C
h
t = 16 s
22 C
d
V = 925 V ; T* = 238 ± 7 C
i
V=0V
312 C 22 C
Active Carbon Nanotube-Polymer Composites, Fig. 3 (a–d) Deflection of a 0.5 vol% SWNT-CP2 beam (L ¼ 4 cm; d ¼ 140 mm; w ¼ 1.9 mm) as a function of applied voltage and temperature. For T* < Tg (220 C), thermal expansion induces a positive deflection in the beam due to a buckling instability (b, c). At T* Tg, mechanical softening promotes a deflection inversion as gravity becomes the dominant force acting on the beam (d). Sample temperature was determined at each voltage using the thermal images shown inset; (e–i) Deflection of
a 0.5 vol% SWNT-CP2 beam as a function of time at Vth (voltage where T* is first observed to exceed Tg). The sample initially buckles as a result of thermal expansion (e); however, as the sample temperature continues to increase, the sample kinks (f, g), and then eventually sags under its own weight at Tg (h). Once the voltage is removed, the beam reversibly actuates back to its original form (i) (Reprinted by permission from Wiley-VCH Verlag GmbH & Co. KGaA [15])
unimorph during the fabrication process to actuate without the need for additional inactive layers [14]. The 0.05 wt% SWNT/LaRC-EAP exhibits a large strain ( 2.6%) at a low driving voltage (40 nm). (c) Intensity profile along
the arrow shown in a and b. Arrows indicate that the position of the second order “green” minimum coincides with the second order “blue” maximum. From this, it can be deduced that the central dark areas of most epithelial cells are zero order minima. (d) Reconstruction of the fluid film thickness along the arrow show in a and b (Redrawn from [5])
frogs did not completely overcome the allometry problem, for they fell from the rotation platform at smaller angles than younger frogs. In ecological terms, this disadvantage is less than one might expect, in that frogs can use their digits to grasp objects such as twigs, as described above. Indeed, there is a reasonable correlation between adult size and degree of arboreality in that the largest species are often found high in the canopy, while smaller species are most commonly found in shrubs no more than a meter or so above the ground. Finally, the way in which forces scale with size can provide insights into the underlying mechanisms of adhesion. Classic formulae for wet adhesion, based on flat metal disks (e.g., Eq. 4), show capillarity forces scaling with toe pad area and have
therefore been used as evidence for the importance of capillarity in tree frog adhesion. However, at the small values of h measured by interference reflection microscopy, they predict adhesive forces far greater than toe pads produce, even when the wedge of fluid around the pad perimeter is taken into account. Indeed, the softness of the pad material makes this model inappropriate for tree frogs. Equation 2 predicts forces in the right range but scaling with length and not area. Peeling models, applicable to the forces that prevent peeling in adhesive tape, are also worthy of consideration because of the way in which toe pads detach (see below). However, they also have force scaling with length, the forces being concentrated in a very narrow zone along the peeling line and thus dependent
A
78
Adhesion in Wet Environments: Frogs
a
1000
10
1
0.1 10
100
Adhesive force (mN)
1000 Toe-pad area (mm2)
Mass (g)
c
b 100
100
10
1 10
100
SVL (mm)
10
1 10
SVL (mm)
e
10
Fall angle (⬚)
Force/unit area (mN mm−2)
d
100
1
0.1 10
100 SVL (mm)
100 SVL (mm)
180 135 90 45 0 10
100 SVL (mm)
Adhesion in Wet Environments: Frogs, Fig. 8 Allometric relationships of 13 hylid species (open circles) and two nonhylids (filled triangles) plotted against snout-vent length (SVL). (a, b) Morphometric relationships between frog mass, toe pad area, and length (log-log plots); mass / SVL2.65, significantly different from expected value of 3.0 (p < 0.01), and area / SVL1.85, not significantly different from expected value
of 2.0; (c, d) Adhesive force and force per unit area plotted against length (log-log plots); force / SVL2.24, not significantly different from 2.0, the value expected from area scaling, and unit force / SVL0.46, a significant positive relationship (p < 0.05); (e) fall angle against length (log-linear plot). The regression line has a slope of 68, a significant negative relationship (p < 0.01) (From [20])
on the width of the tape. Possibly, the solution lies in the recent analysis of Butt et al. [7] for adhesion using the model illustrated in Fig. 1. This analysis predicts a change from length to area scaling with increasing R and decreasing Eeff. A common misconception is that the subdivision of the surface by the channels between the cells increases adhesion according to the principle of contact splitting. This theory states that adhesive force is proportional to the length of the contact; therefore, by splitting up the contact zone into many small areas of contact, the total adhesive force can be increased in direct proportion to the density of these small areas [22]. This principle clearly applies to wet adhesion. However, as the major force component of wet adhesion is capillarity, it is necessary that an air-water interface (meniscus) should surround each small area of contact. This is true of the hairy pads of insects, but is not applicable to tree frogs, where the meniscus surrounds the whole toe pad, not the individual epithelial cells.
In recent years, the development of a technique to measure adhesion and friction forces from single toe pads (Fig. 9) has resulted in a number of interesting findings. The frog is held in a container with holes in its base through which individual digits can protrude. Toe pad forces are measured by a two-axis force plate, manipulated so that it makes contact with the protruding toe. As well as recording forces, the setup has a feedback facility so that, for instance, load can be kept constant while recording friction forces. This technique has provided good evidence for the presence of static friction in tree frog toe pads [5]. This was demonstrated both by the build-up of friction force at the onset of sliding and also from the presence of a remaining shear force 2 min after the sliding motion had stopped (Fig. 10). Since static friction is expected to be very small in a completely fluidfilled joint, this is strongly indicative of direct contact of the nanopillars with the substrate, in accord with the findings of interference reflection microscopy
Adhesion in Wet Environments: Frogs
79
A
computer friction force adhesion force
A
frog container control of motors micromanipulator
force plate
strain gauges
computer xz translational stage microscope
light source
video images
Adhesion in Wet Environments: Frogs, Fig. 9 Experimental setup for measuring adhesion and friction forces from single toe pads of tree frogs. Frogs are held stationary in a small chamber (foam padding not shown), with a toe pad protruding through a hole in its base. Using a micromanipulator, a toe pad is brought into contact with a glass force plate attached to a 2D bending beam force transducer for measuring friction (shear) and
adhesion (normal) forces. A computer-controlled XZ translational stage moves the frog in relation to the bending beam. Normal forces can be controlled by a feedback mechanism when required. Contact area is imaged from below using reflected light (Developed from a setup for measuring adhesion and friction forces from insects devised by Drs Drechsler and Federle at the University of W€ urzburg, Germany)
2 min
10 8 Friction force (mN)
camera
‘sliding’
6 4
‘remaining’
2
‘onset’
0 −2 0
10
20
30
40
140
150
Time (s)
Adhesion in Wet Environments: Frogs, Fig. 10 Shear force measurement in single toe pads of White’s tree frog, Litoria caerulea. The recording shows the result of a 20-s sliding movement toward the body (500 mm s–1), followed by a 2-min period when no movement occurred. The arrows indicate the force when the pad began to slide (‘onset’), the force when sliding ended (‘sliding’), and the force when the experiment was terminated after the 2-min rest (‘remaining’) (Redrawn from [5])
(see above). By making measurements from pads of different sizes, one can also do scaling experiments with single pads. Such experiments demonstrate that friction forces are independent of load but correlate well with the area of contact. This is contrary to what one would expect of classic (Coulomb) friction but in line with rubber friction. This conclusion fits well with the results of indentation experiments (see above), since the low elastic modulus of tree frog toe pads allows close contact between pad and substrate. Adhesion forces of single pads scale with pad area when pads are pulled off the force plate at a low angle (20–50 ), in line with whole animal measurements. However, they scale with length when the pull-off is at right angles to the surface. This has significance for toe pad detachment mechanisms (see below). Finally, such single pad experiments provide evidence for the dominant role of capillarity in tree frog adhesion since toe pad adhesive forces fall to low levels when the meniscus surrounding the pad is removed by submerging the whole pad in a drop of water. Preliminary results from studies on torrent frogs
A
80
Adhesion in Wet Environments: Frogs
6
Contact area (mm2)
5 4 3 2 1 0
0
0.1
0.2
0.3 0.4 Time (seconds)
0.5
0.6
0.7
Adhesion in Wet Environments: Frogs, Fig. 11 Toe pad detachment by peeling during walking on a level surface in the tree frog, Litoria caerulea. The graph shows the pad contact area during a complete step, while the images are single frames from a 250-frames-per-sec video, taken by a camera on the underside of the glass plate on which the frog walked. The glass plate is
illuminated from the side by arrays of light-emitting diodes that direct the light within the thickness of the plate, the light only escaping (and therefore visible to the camera) when an object, here the toe pad, contacts the glass (Unpublished data kindly provided by Diana Samuel, University of Glasgow)
show that they adhere particularly well in wet and flooded conditions, but the underlying mechanisms remain to be determined.
vertical surfaces in a head-down orientation. Indeed, frogs on a rotating vertical surface have been observed to adjust their orientations back toward the vertical whenever their deviation from the vertical reached 85 21 . During forward locomotion, peeling seems to occur as a natural consequence of the way in which toes are lifted off surfaces from the rear forward, while during backward locomotion, it is an active process involving the distal tendons of the toes. The process is seen most clearly when the pad is viewed from below using high-speed video imaging (Fig. 11).
Toe Pad Detachment During locomotion, toe pads are detached by peeling, a mechanism that reduces the forces required to detach their feet from the substrate every time they wish to jump or take a step to very low levels [11]. During forward walking, climbing or jumping, toe pads are peeled from the rear forward, i.e., in a proximal to distal direction. When frogs are induced to walk backward, peeling occurs in the opposite direction, from the front of the pad rearward in a distal to proximal direction. Use of a force platform to measure directly the forces generated by the feet during locomotion shows that during forward locomotion, this peeling is not accompanied by any detectable detachment forces. Such forces of detachment are seen, however, during backward walking and when belly skin inadvertently comes into contact with the force platform. That peeling occurs automatically during forward locomotion is supported both by observations of peeling in single toe pads of anesthetized frogs and by the inability of frogs to adhere to
Behavioral Strategies of Adhering Frogs Interestingly, peeling has consequences for frog behavior. This is because pads can produce strong adhesion and friction when the angle of force application is low but peel easily off the surface when the pull-off angle approaches 90 . On vertical and overhanging surfaces, frogs must therefore adjust their posture so that gravity acts on the pads at angles where they can withstand the force and not fall from the surface. On a rotation platform, the first response to the changing angle occurs at angles between 60 and 100 , heavier frogs tending to respond first. Any
Adhesion in Wet Environments: Frogs
81
A
a b a
A
b
gravity
Both toes and legs are extended to reduce angles a and b and thus reduce the chances of peeling a, leg/substrate angle; b, toe/substrate angle
b
Adhesion in Wet Environments: Frogs, Fig. 12 (a) Videoimage of ventral view of a tree frog (Hypsiboas boans) clinging to an overhanging translucent surface, with both fore and hind
limbs stretched out sideways. Only areas in contact with surface are in sharp focus. Scale: large squares are 10 mm across (From [30]). (b) Explanatory model
frog not already in a head-up position will turn to face head-up so that gravity will not be acting to pull the pads off the surface. At still higher angles, both tree and torrent frogs spread their legs out sideways (Fig. 12). This is an adaptation to prevent peeling since, as the figure shows, spread-out limbs have a much smaller pad-substrate angle than limbs held close to the body. Gravity, of course, will tend to pull the limbs inward, but this is resisted by pad friction forces. In this way, friction and adhesion both play roles in helping frogs to maintain their grip under challenging conditions. The term “frictional adhesion” has been coined to describe adhesion in geckos [23]. It is unclear whether this term should also be applied to tree frog adhesion, but there is certainly interplay between friction and adhesion forces in frog toe pads too.
Planck Institute for Polymer Research in Mainz, Germany, is shown in Fig. 13b, c. However, the role of surface patterning in increasing adhesion is well established. Adhesive tape detaches by the spreading of cracks into the adhesive from the point of peeling. When all the energy can be concentrated at a single crack, then peeling occurs readily, but micropatterning can increase the force required to produce peeling by up to a factor of 3 by the principle of contact separation. Cracks form wherever there is a groove in the pattern so that the energy is spread between many cracks, resulting in an increased force being required to produce separation [13]. Recently, this principle has been taken a step further, and the role of subsurface structures (air- or oil-filled microchannels) has been investigated. These have crack-arresting properties just like the transverse grooves of a patterned surface, but the effect is much more dramatic, for adhesion can be increased up to 30-fold. The adhesive itself remains elastic, which renders it reusable with no reduction in adhesive efficiency [24]. Turning from patterned dry adhesion to wet adhesion, all published work is insect- rather than frog-inspired. Of particular interest is a study of the frictional properties of a structure like that in Fig. 13a. Although the addition of fluid lowered friction as would be expected, the micropatterned surface was able to generate significantly higher friction forces
Biomimetic Advances As far as tree frog adhesion is concerned, the process of micro- or nanofabrication of toe pad replicas, where the effect of the individual design parameters can be analyzed and quantified precisely, is in its infancy. An early attempt at copying the hexagonal cell structure of tree frog toe pads in polydimethylsiloxane (PDMS) is illustrated in Fig. 13a, whereas a recent mimic of the structure of a tree frog’s toe pad epithelium that replicates both nano- and microstructures, from del Campo’s research group at the Max
A
82
Adhesion in Wet Environments: Frogs, Fig. 13 (a) Polydimethylsiloxane (PDMS) replica of a tree frog toe pad, copying the hexagonal pattern of epithelial cells (Image: Centre for Cell Engineering, University of Glasgow); (b, c) hierarchical PDMS replica of both microand nanostructuring on a tree frog toe pad. The PDMS was made hydrophilic by treatment in a plasma chamber and has an elastic modulus in the range 1–5 MPa (Image: Max Planck Institute for Polymer Research, Mainz)
Adhesion in Wet Environments: Frogs
a
during sliding than an otherwise identical smooth one [25]. In another study, the same authors studied surfaces with a mushroom-shaped fibrillar structure not unlike the shape of the nanopillars seen on tree frog toe pads. Surfaces with such structures adhered over 20 times better than flat ones underwater, possibly due to a suction effect [26].
b
c
Cross-References ▶ Atomic Force Microscopy ▶ Bioadhesion ▶ Biomimetics ▶ Gecko Adhesion ▶ Optical Tweezers ▶ Scanning Electron Microscopy
Examples of Possible Applications Since adhesive tapes inspired by the dry adhesive mechanisms of gecko toe pad setae are now reaching a stage where commercialization is imminent, it is appropriate to consider whether the rather different wet adhesion mechanism of tree and torrent frogs might also have biomimetic relevance. Improved wet weather tires [14, 27, 28], nonslip footwear, plasters for surgery able to adhere to tissue, and holding devices for neurosurgery or MEMS devices are obvious examples of the many uses to which these toe pad analogues might be applied. However, as biomimetics (▶ biomimetics) has as much to do with inspiration as copying, a beetleinspired device providing reversible adhesion based on capillarity is worthy of mention as it uses wet adhesion [29]. It generates an adhesive force through the formation of a large number of liquid bridges between two plates, which can be quickly made or broken by a lowvoltage pulse that drives electroosmotic flow. There is good potential for the development of a grab-andrelease device that could carry significant loads.
References 1. Gorb, S.N.: Uncovering insect stickiness: structure and properties of hairy attachment devices. Am. Entomol. 51, 31–35 (2005) 2. Gorb, S.N.: Smooth attachment devices in insects. In: Casas, J., Simpson, S.J. (eds.) Advances in Insect Physiology: Insect Mechanics and Control. Adv. Insect Physiol. 34, 81–116 (2008) 3. Autumn, K., Sitti, M., Liang, Y.C.A., Peattie, A.M., Hansen, W.R., Sponberg, S., Kenny, T.W., Fearing, R., Israelachvili, J.N., Full, R.J.: Evidence for van der Waals adhesion in gecko setae. Proc. Natl. Acad. Sci. USA 99, 12252–12256 (2002) 4. Barnes, W.J.P.: Functional morphology and design constraints of smooth adhesive pads. Mater. Res. Soc. Bull. 32, 479–485 (2007) 5. Federle, W., Barnes, W.J.P., Baumgartner, W., Drechsler, P., Smith, J.M.: Wet but not slippery: boundary friction in tree frog adhesive toe pads. J. R. Soc. Interface 3, 689–697 (2006) 6. Bhushan, B.: Introduction to Tribology. Wiley, New York (2002) 7. Butt, H.-J., Barnes, W.J.P., del Campo, A., Kappl, M.: Capillary forces between soft, elastic spheres. Soft Matter. 6, 5930–5936 (2010)
AFM in Liquids 8. Ernst, V.: The digital pads of the tree frog, Hyla cinerea. 1. The epidermis. Tissue Cell 5, 83–96 (1973) 9. Green, D.M.: Treefrog toe pads: comparative surface morphology using scanning electron microscopy. Can. J. Zool. 57, 2033–2046 (1979) 10. Emerson, S.B., Diehl, D.: Toe pad morphology and mechanisms of sticking in frogs. Biol. J. Linn. Soc. 13, 199–216 (1980) 11. Hanna, G., Barnes, W.J.P.: Adhesion and detachment of the toe pads of tree frogs. J. Exp. Biol. 155, 103–125 (1991) 12. Smith, J.M., Barnes, W.J.P., Downie, J.R., Ruxton, G.D.: Structural correlates of increased adhesive efficiency with adult size in the toe pads of hylid tree frogs. J. Comp. Physiol. A 192, 1193–1204 (2006) 13. Ghatak, A., Mahadevan, L., Chung, J.Y., Chaudhury, M.K., Shenoy, V.: Peeling from a biomimetically patterned thin elastic film. Proc. Roy. Soc. Lond. A 460, 2725–2735 (2004) 14. Persson, B.N.J.: Wet adhesion with application to tree frog adhesive toe pads and tires. J. Phys. Cond. Matter 19, 376110 (16 pp) (2007) 15. Scholz, I., Barnes, W.J.P., Smith, J.M., Baumgartner, W.: Ultrastructure and physical properties of an adhesive surface, the toe pad epithelium of the tree frog, Litoria caerulea White. J. Exp. Biol. 212, 155–162 (2009) 16. Barnes, W.J.P., Perez Goodwyn, P., Nokhbatolfoghahai, M., Gorb, S.N.: Elastic modulus of tree frog adhesive toe pads. J. Comp. Physiol. A 197, 969-978 (2011) 17. Samani, A., Zubovitz, J., Plewer, D.: Elastic moduli of normal and pathological human breast tissue: an inversiontechnique-based investigation of 169 samples. Phys. Med. Biol. 52, 1565–1576 (2007) 18. Vogel, S.: Comparative Biomechanics: Life’s Physical World. Princeton University Press, Princeton (2003) 19. Perez Goodwyn, P., Peressadko, A., Schwarz, H., Kastne, V., Gorb, S.: Material structure, stiffness, and adhesion: why attachment pads of the grasshopper (Tettigonia viridissima) adhere more strongly than those of the locust (Locusta migratoria) (Insecta: Orthoptera). J. Comp. Physiol. A 192, 1233–1243 (2006) 20. Barnes, W.J.P., Oines, C., Smith, J.M.: Whole animal measurements of shear and adhesive forces in adult tree frogs: insights into underlying mechanisms of adhesion obtained from studying the effects of size and scale. J. Comp. Physiol. A 192, 1179–1191 (2006) 21. Smith, J.M., Barnes, W.J.P., Downie, J.R., Ruxton, G.D.: Adhesion and allometry from metamorphosis to maturation in hylid tree frogs – a sticky problem. J. Zool. 270, 372–383 (2006) 22. Arzt, E., Gorb, S., Spolenak, R.: From micro to nano contacts in biological attachment devices. Proc. Natl. Acad. Sci. USA 100, 10603–10606 (2003) 23. Autumn, K., Dittmore, A., Santos, D., Spenko, M., Cutkosky, M.: Frictional adhesion: a new angle on gecko attachment. J. Exp. Biol. 209, 3569–3579 (2006) 24. Majumder, A., Ghatak, A., Sharmer, A.: Microfluidic adhesion induced by subsurface microstructures. Science 318, 258–261 (2007) 25. Varenberg, M., Gorb, S.N.: Hexagonal surface micropattern for dry and wet friction. Adv. Mater. 21, 483–486 (2009) 26. Varenberg, M., Gorb, S.N.: A beetle-inspired solution for underwater adhesion. J. R. Soc. Interface 5, 383–385 (2008)
83
A
27. Barnes, W.J.P.: Tree frogs and tire technology. Tire Technol. Int. March 42–47 (1999) 28. Barnes, W.J.P., Smith, J., Oines, C., Mundl, R.: Bionics and wet grip. Tire Technol. Int. Dec. 56–60 (2002) 29. Vogel, M.J., Steen, P.H.: Capillary-based switchable adhesion, beetle inspired. Proc. Natl. Acad. Sci. USA 107, 3377–3381 (2010) 30. Barnes, W.J.P., Pearman, J., Platter, J.: Application of peeling theory to tree frog adhesion, a biological system with biomimetic implications. Eur. Acad. Sci. E-Newsletter Sci. Technol. 1(1), 1–2 (2008)
Adsorbate Adhesion ▶ Disjoining Pressure and Capillary Adhesion
Aerosol-Assisted Chemical Vapor Deposition (AACVD) ▶ Chemical Vapor Deposition (CVD)
AFM ▶ Robot-Based Automation on the Nanoscale
AFM Force Sensors ▶ AFM Probes
AFM in Liquids Bart W. Hoogenboom London Centre for Nanotechnology and Department of Physics and Astronomy, University College London, London, UK
Synonyms Atomic force microscopy in liquids; Scanning force microscopy in liquids
A
A
84
AFM in Liquids
Definition
Mirror Laser
Atomic force microscopy (AFM) in liquids is the application of AFM in liquid environment, i.e., in which both the surface under investigation and the scanning probe are immersed in liquid.
Photodiode
Cantilever
Liquid Sample
Overview and Definitions
Tip
AFM is a microscopy technique that can provide threedimensional images of virtually any surface at nanometer-scale resolution. It relies on the force between a sharp probe and the surface, which is detected while scanning the probe over the sample. Unlike many other microscopy techniques at such a resolution, it can readily be applied in liquid environment. An atomic force microscope consists of a sharp probe (“tip”) mounted on a microfabricated cantilever beam and a mechanism (“scanner”) to scan the tip over the surface at subnanometer resolution [1], see Fig. 1. Typically, an optical detection scheme is used to detect the deflection of the cantilever. Via the spring constant of the cantilever, the cantilever deflection can be translated to a force between tip and sample. For a rectangular lever, the spring constant is given by k¼
Et3 w ; 4l3
(1)
where E is the Young’s modulus of the cantilever material (typically silicon or siliconnitride), and t, w, and l are its thickness, width, and length, respectively, which are in the micron range. Most cantilevers for AFM have spring constants between 0.01 and 100 N/m. In its most common mode of operation, the deflection of the cantilever (and thus the tip–sample force) is kept constant by adjusting the vertical position of the sample with respect to the tip. In most instruments, the cantilever deflection is recorded via the position of a laser beam that is deflected from the cantilever. Images of the surfaces are acquired by line-by-line scanning the surface and tracing its surface contours on a false-color scale. For AFM in liquid, the surface and the cantilever are immersed in liquid. Typically, the minimum liquid volume is some tens of microliters, which can be either contained in a closed liquid cell, or in the shape of
z y
Scanner
x
AFM in Liquids, Fig. 1 Schematic of AFM in liquid. A sharp tip is scanned across the sample surface while the tip–sample interaction is monitored via the deflection of the cantilever to which the tip is attached. The sample is mounted on a (usually piezoelectric) scanner for three-dimensional positioning with sub-nanometer accuracy. The sample, the tip, and the cantilever are immersed in liquid. The bending of the cantilever is usually detected via a laser beam deflected on a position-sensitive detector (4-quadrant photodiode)
a droplet that is formed by capillary forces between the sample surface and the cantilever holder. In principle, the liquid can be of arbitrary nature, though the optical deflection detection in most microscopes limit their use to liquids that are transparent for (near infra-) red light. There are many possible imaging modes of AFM and their names are rather confusing. They are divided into static mode and a variety of dynamic modes in which the cantilever is oscillating [2]. The latter are of particular importance when the lateral (drag) forces on the sample need to be minimized, e.g., for imaging DNA molecules that are loosely bound to a flat surface. The most common modes of operation in liquid are static or contact mode and amplitude-modulation or tapping mode. In contact mode, the bending or deflection of the cantilever is used to detect the tip–sample force. In tapping mode, the cantilever is vertically oscillated above the sample, and the tip–sample interactions detected via the change (usually decrease) of amplitude of the cantilever oscillation. In variations along the same theme, the cantilever may be oscillated and tip–sample interactions are detected via changes in phase or resonance frequency of the cantilever [3]. An overview of the most relevant AFM modes in liquids is given in Table 1. Confusion may arise because the
AFM in Liquids
85
AFM in Liquids, Table 1 Different modes of operation in AFM, with the parameter that is used to control the tip–sample distance while scanning over the surface Mode of operation Static Amplitude modulation (AM) Frequency modulation (FM)
Other common names Contact mode Tapping mode Intermittent contact AFM AC mode Non-contact (NC) AFM
Control parameter Static deflection Oscillation amplitude Resonance frequency
commonly used terms contact, intermittent-contact, and non-contact often – but not always – refer to a mode of operation (as referred to in Table 1), and not necessarily to the actual contact (or lack of it) between the tip and the sample. As a particular example, FM (“non-contact”) AFM in liquids is usually performed in the range of repulsive tip–sample interactions, with the tip in actual intermittent contact with the sample. In addition, if the amplitude is used to trace and control the tip–sample distance (i.e., tapping mode), the phase of the cantilever can be recorded simultaneously for phase imaging, which provides an additional means to measure tip–sample interactions. Or, in FM AFM, a frequency shift is used to trace elastic tip–sample interactions and control the tip– sample distance, while the oscillation amplitude can be kept constant by adjusting the cantilever drive signal. This cantilever drive signal then provides a simultaneous measure of the dissipative tip–sample interactions.
Physical and Chemical Principles The spatial resolution of AFM critically depends on the sharpness of the tip and on the range of the tip–sample interactions. Ideally, only the very (nanometer-scale) end of the tip interacts with the surface, thus avoiding convolution effects due to interactions between the sample and the (micron-scale) bulk of the tip. The presence of long-range tip–sample forces is therefore decremental to the spatial resolution. On the other hand, the shorter the range of the tip–sample interaction, the more likely the tip is to enter in hard contact with the sample, which may damage the tip and/or the sample.
A
By immersing the tip and sample in liquid, it is not only possible to access scientifically and technologically interesting solid–liquid interfaces (including biological samples), but also to tune the tip–sample interactions by varying the ingredients of the liquid. In particular, the electrostatic sample interaction between the tip and a flat sample can be approximated by [4]:
Fel ¼ gk st ss ekz þ g2 k s2t þ s2s e2kz ;
(2)
where gk and g2k are constants that depend on the tip geometry and the dielectric properties of the medium, st and ss are the surface charges of the tip and the sample, respectively, z is the tip–sample distance, and k1 the Debye screening length [5], 1
k
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e0 eb kB T : ¼ 2e2 I
(3)
In the latter, e0 is the permittivity of vacuum (8.854 1023 F/m), eb the permittivity of the bulk solution (or the dielectric medium), kB is the Boltzmann’s constant (1.38 1023 J/K), and T is the temperature in Kelvin. The parameter I is the ionic strength of the medium, I¼
1X 2 z ci : 2 i i
(4)
where zi is the valence of ion i and ci is its concentration. The Debye length is historically written as an inverse length (i.e., k has units of m1). In typical biological buffers, the Debye length is between 1 and 10 nm. In a chemically reactive medium, st and ss are generally not zero. For typical tip terminations such as silicon oxide or silicon nitride in water, st < 0 at neutral pH. As can be seen from Eq. 2, the strength of the electrostatic tip–sample interaction depends on st and ss. These can be changed by adding or removing solutes that react with or bind to the tip and sample surfaces. In aqueous solutions, this is most readily achieved by changing the pH (i.e., adding H3O+ or OH). Moreover, the dielectric properties – and in particular the ionic strength (Eq. 4) – of the medium can be tuned to control the range of the electrostatic tip–sample interaction. This provides a significant advantage over AFM in gaseous mediums and vacuum environment, and is one of the reasons why the first true atomic resolution (see below) AFM images were
A
a
86
AFM in Liquids
b
a Amplitude
A
Capillary
wd
c
d
+ –
b
–
Frequency
c
+ +
+
– – +
AFM in Liquids, Fig. 2 (a) Capillary forces for AFM in air arise due to the thin water layers that are usually present at both tip and sample under ambient conditions. (b) At small tip– sample distances, they pull the tip toward the surface to create a larger contact area and thus limit the lateral resolution. (c) In liquid environment, these capillary forces are absent. (d) Moreover, long-range electrostatic forces can be screened by ions in the liquid
obtained, in contact mode, in aqueous environment rather than in vacuum or air [6]. Compared to AFM in air, AFM in liquids has the additional advantage of preventing the capillary forces that arise from the humid coverage of both sample and tip under ambient conditions, as illustrated in Fig. 2. The capillary forces create a strong pull on the tip toward the sample at distances of many nanometers. On close approach, this results in a tip–sample contact area that covers many atoms at the surface, thus prohibiting high resolution. Under such conditions, AFM is likely to result in one of its most extensively documented artifacts: The apparent “atomic resolution” that results from the periodic interaction averaged over many atoms of tip and surface. This represents the atomic periodicity of the surface rather than individual atomic-scale features such as atomic defects (which are often used to demonstrate “true atomic resolution”). Apart from electrostatic and capillary forces, AFM in liquids is also (and not exclusively) sensitive to atomic bonding forces and hard-core repulsion, van der Waals interactions, dissolution/solvation forces, and general dissipative interactions. The latter are particularly important in the dynamic modes of AFM operation. To illustrate this, it is helpful
AFM in Liquids, Fig. 3 (a) Schematic resonance curve of a freely oscillating cantilever (dashed, blue curve), based on a model of a simple harmonic oscillator. In the presence of a (repulsive) elastic tip–sample force as well as dissipative tip– sample interactions, the resonance shifts to higher frequencies, broadens and decreases in amplitude (red curve). At a fixed driving frequency od, such as in tapping mode, the effects of elastic or dissipative interactions will cause the amplitude to decrease or increase, depending on the choice of od. (b) Sinusoidal oscillation of a cantilever above the surface (dashed, blue). Close to the sample, the oscillation amplitude decreases and changes phase (red). (c) For heavily damped cantilever oscillations (Q ~ 1), however, the harmonic-oscillator analysis is not valid any more. The tip–sample interaction will only affect the bottom part of the oscillation, where the tip is closest to the sample
to depict the cantilever resonance for different elastic and dissipative interactions, see Fig. 3. A repulsive (elastic) tip–sample interaction increases the effective spring constant of the cantilever and thus increases its resonance frequency, shifting the whole resonance curve to the right. Any dissipative interaction reduces the sharpness of the resonance curve and the maximum of the curve. The quality factor Q is the most common measure for this sharpness, defined as o0/Do, where o0 is the natural frequency of the resonator pffiffiffi and Do the width of the resonance curve at 1= 2 its maximum amplitude. In tapping mode, the driving frequency is usually set just below the resonance frequency, such that both elastic and dissipative interactions cause the amplitude to decrease when the tip comes into contact with the sample. In FM AFM, the shift in the resonance frequency provides a direct measure of the elastic interaction, and the conversion to a force is straightforward when the amplitude of the oscillation is (kept) constant [7]. The dissipated energy can be deduced from the
87
energy that is required to maintain a fixed amplitude. Thus, tapping mode is sensitive to a mixture of elastic and dissipative tip–sample interactions, whereas FM AFM can track elastic and dissipative interactions separately (and simultaneously). This analysis relies on the cantilever behaving as a simple harmonic oscillator, which is appropriate for
A
Frequency
>
Q 1, typical for operation in vacuum ðQ 1000Þ > and air ðQ 100Þ. However, in liquid, viscous < damping reduces Q to 10. In particular for softer < cantilevers k 1N=m, Q 1, and the contact between tip and sample will prevent the cantilever from following a harmonic, sinusoidal oscillation. The cantilever oscillation will simply be topped off at the bottom of the oscillation, at the closest tip–sample approach. In this case, significant information on the tip–sample interaction will be contained in higher harmonics of the oscillation. The interpretation of cantilever oscillation in terms of tip–sample forces thus becomes considerably more complicated. The low quality factor has another undesirable side effect for AFM operation that involves oscillating the cantilever. Cantilevers are usually driven by a small piezoelectric element that drives not only the cantilever itself, but also its (macroscopic) support chip and often the whole cantilever holder. In vacuum and air, the high Q singles out the cantilever resonance from other mechanical resonances in the instrument. In liquid, this is not the case any more, and a typical excitation spectrum contains a so-called forest of peaks, i.e., a convolution of the broad cantilever resonance with many sharp(er) mechanical resonances from the microscope, as illustrated in Fig. 4. Again, this complicates the interpretation of the cantilever behavior in terms of a simple harmonic oscillator. It also makes the choice of optimum driving frequency much less straightforward (usually, this is done by trial and error). Moreover, since this excitation spectrum depends on the macroscopic geometry of the fluid cell, it is much more dependent on drift than the cantilever alone. For these reasons, there are a number of alternative methods to only drive the microscopic cantilever (and no macroscopic parts), such as magnetic and optical actuation. In the former, the cantilever is coated with a magnetic material and driven by an AC magnetic field; in the latter, an AC-modulated actuation laser locally heats the metal-coated surface of cantilever, which acts as a bimetal and thus transduces
A
Amplitude
AFM in Liquids
AFM in Liquids, Fig. 4 Excitation spectrum such as can be obtained for a cantilever in water that is actuated by piezoelectrically driving the cantilever holder rather than the cantilever alone. Note the contrast with the ideal resonance shape in Fig. 3a
laser power to cantilever deflection. These and other methods have the disadvantage of adding complexity to the instrument and eventually to the microfabricated cantilever itself. These concerns can partly be addressed by so-called Q-control or self-oscillatory methods [8], in which the signal from the cantilever oscillation itself is phaseshifted, amplified, and added to the actuation signal. This positive feedback is very familiar in electronic oscillatory circuits. It causes any change to the oscillation to ring for a longer time, stretching its effect over many oscillations. As a result the effective Q of the cantilever oscillation is enhanced, and the cantilever is forced to follow a perfectly harmonic, sinusoidal oscillation. As for problems related to the forest of peaks in nonideal cantilever excitation, these are not resolved by Q control, since the positive feedback does not distinguish between the cantilever resonance and other mechanical resonances. Q control or selfoscillatory methods can be applied to each mode of operation (see Table 1), but for tapping mode has the side effect of slowing down the measurement: Changes in amplitude occur at a timescale of ~Q/o0, unlike changes in phase or resonance frequency, which are instantaneous. In FM AFM, the cantilever can even entirely be driven by its phase-shifted and amplified thermal noise.
Applications AFM has been used to probe a vast amount of interfaces between (hard and soft) condensed matter and liquids. It is of particular importance for the study of surfaces that are instable or show distinctly different behavior when taken out of the liquid, and for the study
A
88
of dynamic processes that depend on an exchange of material (ions, macromolecules) between the surface and the liquid. AFM in liquids has been important for the study of, among others, crystal growth/dissolution, polymer science, and the behavior of liquids in confined geometries. The vast majority of its applications, however, lie in the life sciences [9]: For most biomolecular structures, aqueous solutions represent their natural environment. Their structures and functions are strongly dependent on the presence of water, a number of ions, and other macromolecules. The structure and function of biological samples can thus be studied at (sub)nanometer resolution while they are still “alive,” and for varying liquid contents. The following represents some illustrative examples of the use of AFM in aqueous solutions on easily accessible samples, with a strong focus on biological applications. Up to now, highest resolution AFM images in liquid have been obtained on atomically flat and inorganic surfaces, with calcite and muscovite mica as most common examples. On such surfaces, the lateral reso˚ ngstro¨m and the vertical lution can be less than an A resolution a few picometers. Though initially such resolution was only obtained in contact mode [6], more recently FM AFM has become popular as a method for atomic-resolution imaging that is significantly less susceptible to thermal drift. One of the key steps toward high-resolution imaging in liquids was the development of low-noise deflection sensors [10, 11]. This has brought the measurement noise down to the thermal noise of the cantilevers, even for stiff and heavily damped cantilevers in liquid. As a result, sur˚ ngstro¨m faces can now be stably measured with A amplitudes of cantilever oscillation. This enhances the sensitivity to short-range forces and yields images such as depicted in Fig. 5. Because it is flat, hydrophilic, and easily cleaved, mica is one of the substrates of choice for adsorbing molecules in AFM experiments in aqueous solutions. AFM has been successful in imaging DNA as well as DNA–protein complexes. DNA is routinely adsorbed on mica, where it is generally assumed to adopt a two-dimensional projection of its original three-dimensional configuration [13]. The rather flat geometry required for high-resolution AFM is a natural one for proteins that are embedded in a lipid membrane. In particular when two-dimensional crystals are available, such as for bacteriorhodopsin,
AFM in Liquids 0.2 nm
2 nm 0.0 nm
AFM in Liquids, Fig. 5 Atomic-resolution image of mica in (aquaeous) buffer solution, obtained by FM AFM (Reproduced from [12], with permission)
a
b
70 nm
7 nm
AFM in Liquids, Fig. 6 (a) Tapping-mode AFM image of DNA adsorbed on mica. Image courtesy Elliot Menter. (b) Two-dimensional crystal of the membrane protein bacteriorhodopsin, obtained by FM AFM on the extracellular side of the protein. Image courtesy Carl Leung
high-resolution images can readily be obtained in the various modes of AFM operation [14]. DNA and bacteriorhodopsin represent some of the most often imaged biomolecules by AFM in liquid (Fig. 6). On a completely different scale, AFM can image objects as large as whole cells (Fig. 7). Due to thermal fluctuations, vibrations, and the overall softness of the cell, the spatial resolution is considerably lower than that obtained on flat surfaces or molecules directly adsorbed on a hard substrate. As, for cell imaging, alone, the advantage of AFM over optical techniques is therefore limited. AFM can also be used to deliberately probe sample properties other than structure [15]. Single molecules that are tethered between the tip and the substrate can be deliberately stretched, unfolded, and allowed to refold at controlled load force [16], similar to optical-tweezers experiments. Using the same force
AFM in Liquids
89
A
References
AFM in Liquids, Fig. 7 100 100 3 mm3 contact-mode AFM topograph of an osteoblast (bone cell) adsorbed on a glass coverslip. Image courtesy Guillaume Charras
spectroscopy techniques, force–distance curves can be obtained on whole cells, yielding information about their local elasticity, or – when using chemically modified tips – local binding sites for specific ligands [17]. Finally, fast AFM methods [18] have improved the image rate from the typical frame per minute up to many frames per second. This gives direct access to the kinetics of molecular-scale processes at nanometer resolution.
Cross-References ▶ AFM ▶ AFM Probes ▶ AFM, Non-Contact Mode ▶ AFM, Tapping Mode ▶ Atomic Force Microscopy ▶ Force Modulation in Atomic Force Microscopy ▶ Friction Force Microscopy ▶ Imaging Human Body Down to Molecular Level ▶ Kelvin Probe Force Microscopy ▶ Magnetic Resonance Force Microscopy ▶ Mechanical Properties of Hierarchical Protein Materials ▶ Nanomedicine ▶ Nanotechnology ▶ Nanotribology ▶ Optical Tweezers ▶ Optomechanical Resonators ▶ Surface Forces Apparatus
1. Sarid, D.: Scanning Force Microscopy with Applications to Electric, Magnetic, and Atomic Forces, 2nd edn. Oxford University Press, Oxford (1994) 2. Garcia, R., Perez, R.: Dynamic atomic force microscopy methods. Surf. Sci. Rep. 47, 197–301 (2002) 3. Giessibl, F.J.: Advances in atomic force microscopy. Rev. Mod. Phys. 75, 949–983 (2003) 4. Butt, H.-J.: Electrostatic interaction in atomic force microscopy. Biophys. J. 60, 777–785 (1991) 5. Israelachvili, J.N.: Intramolecular and Surface Forces, 3rd edn. Academic, Oxford (2011) 6. Ohnesorge, F., Binnig, G.: True atomic resolution by atomic force microscopy through repulsive and attractive forces. Science 260, 1451–1456 (1993) 7. Sader, J.E., Jarvis, S.P.: Accurate formulas for interaction force and energy in frequency modulation force spectroscopy. Appl. Phys. Lett. 84, 1801–1803 (2004) 8. Bhushan, B., Fuchs, H., Hosaka, S. (eds.): Applied Scanning Probe Methods, vol. I. Springer, Berlin/Heidelberg (2004) 9. Morris, V.J., Gunnig, A.P., Kirby, A.R.: Atomic Force Microscopy for Biologists. World Scientific, River Edge (1999) 10. Hoogenboom, B.W., Frederix, P.L.T.M., Yang, J.L., Martin, S., Pellmont, Y., Steinacher, M., Z€ach, S., Langenbach, E., Heimbeck, H.-J., Engel, A., Hug, H.J.: A Fabry-Perot interferometer for micrometer-sized cantilevers. Appl. Phys. Lett. 86, 074101 (2005) 11. Fukuma, T., Kimura, M., Kobayashi, K., Matsushige, K., Yamada, H.: Development of low noise cantilever deflection sensor for multienvironment frequency-modulation atomic force microscopy. Rev. Sci. Instrum. 76, 053704 (2005) 12. Khan, Z., Leung, C., Tahir, B., Hoogenboom, B.W.: Digitally tunable, wide-band amplitude, phase, and frequency detection for atomic-resolution scanning force microscopy. Rev. Sci. Instrum. 81, 073704 (2010) 13. Hansma, H.G.: Surface biology of DNA by atomic force microscopy. Annu. Rev. Phys. Chem. 52, 71–92 (2001) 14. M€ uller, D.J., Engel, A.: Atomic force microscopy and spectroscopy of native membrane proteins. Nat. Protoc. 2, 2191– 2197 (2007) 15. M€ uller, D.J., Dufreˆne, Y.F.: Atomic force microscopy as a multifunctional molecular toolbox in nanobiotechnology. Nat. Nanotechnol. 5, 261–269 (2008) 16. Fisher, T.E., Marszalek, P.E., Fernandez, J.M.: Stretching single molecules into novel conformations using the atomic force microscope. Nat. Struct. Biol. 7, 719–724 (2000) 17. Hinterdorfer, P., Dufreˆne, Y.F.: Detection and localization of single molecular recognition events using atomic force microscopy. Nat. Method 3, 347–355 (2006) 18. Yamamoto, D., Uchihashi, T., Kodera, N., Yamashita, H., Nishikori, S., Ogura, T., Shibata, M., Ando, T.: High-speed atomic force microscopy techniques for observing dynamic biomolecular processes. In: Walter, N.G. (ed.) Methods in Enzymology, vol. 475, pp. 541–564. Academic, Burlington (2010)
A
A
90
AFM Probes
AFM Probes Cantilever h
Kenji Fukuzawa Department of Micro System Engineering, Nagoya University, Nagoya, Chikusa-ku, Japan
w l
Synonyms
Tip
AFM force sensors; AFM tips AFM Probes, Fig. 1 Schematic of a microcantilever probe
Definition AFM probes are transducers that convert the interaction force with a sample surface into a deformation or a change of the vibrational state of the probe. Most probes consist of a sharp microtip and a force transducer. The former determines the lateral resolution of the AFM and the latter provides the force sensitivity.
Overview A typical AFM probe consists of a sharp tip and a microcantilever, which plays the role of a force transducer. In this case, the interaction force between the tip and sample deflects the cantilever. The deflection can be detected by a displacement measurement method, such as an optical lever or interference techniques. Assuming that the deflection and the spring constant of the cantilever are Dz and k, respectively, the interaction force F is given by F ¼ kDz:
(1)
In contact AFM mode, the probe height is controlled so that the probe deflection is constant, which means that the interaction force does not change during the probe scanning. Mapping the controlled height can thus provide the topography of the sample. In the early days, a metal foil with a glued microparticle or sharpened metal wire was used as an AFM probe. Here, the particle or wire acted as a tip [1]. Improving the force sensitivity and reproducibility was not easy with those probes. Then, microfabricated cantilever probes were introduced [2]. These probes are made of silicon nitride or silicon and are produced
by microfabrication techniques such as anisotropic chemical etching. Microfabricated probes enabled to improve the force sensitivity and expand the freedom of the probe design. They were also suitable for mass production. Therefore, microfabricated probes are the most common at present and various types of them are commercially available nowadays. Undoubtedly, the microfabricated probes have been playing an important role in AFM’s becoming one of the most widely used methods in nanotechnology.
Probe Design Since the specifications of the tip are rather limited by the fabrication method, the main part of the probe design is the force transducer, such as the rectangular micro cantilever sketched in Fig. 1. In addition to rectangular cantilevers, V-shaped cantilevers are widely used, especially in contact AFM mode. The force sensitivity of the probe should be high enough to detect weak interaction forces. From Eq. 1, the cantilever should convert a small force F into a large displacement Dz, which requires a small spring constant k ¼ F/Dz. This means that higher sensitivity requires softer cantilevers. In addition to the force sensitivity, the robustness against mechanical disturbances from the environment should be considered. If the vibrations induced by the disturbances exceed the deflection due to the interaction force, the AFM signal is buried in the noise. The vibrations due to disturbances are amplified when the disturbance frequency is around the resonance frequency (natural frequency) of the probe. Since the disturbance frequencies are generally low, the resonance frequency
AFM Probes
A
91
should be set at a high value, typically larger than about 10 kHz. If the force transducer of the probe is simplified as a point mass and spring system, the resonance frequency is given by
k¼
Ewh3 ; 4l3
(3)
where E is the Young’s modulus of the cantilever. It should be noted that the spring constant is proportional to (h/l)3. This means that longer and thinner cantilevers can provide higher force sensitivity. The force F to be measured depends on the interaction between the tip and sample and the deflection Dz depends on the sensitivity of the displacement measurement method. Therefore, the spring constant k should be determined by considering the typical values of F and Dz for the targeted samples and the measurement system. On the other hand, the resonance frequency of a rectangular cantilever is given by l2 h fr ¼ 4p l2
sffiffiffiffiffiffi E ; 3r
(4)
where r is the density of the cantilever and l is a constant that depends on the vibrational mode. For the first mode l is about 1.875. It should be noted that the cantilever width is not included in Eq. 4. Since fr is proportional to h/l2, higher resonance frequencies fr require shorter and thicker cantilevers. However, as described above, high sensitivity requires
Width: w = 50 μm
20
15
1
10 0.5
Resonance frequency fr⬘ kHz
where m is the mass of the transducer. The sensitivity needs low values of k whereas the robustness against the environmental disturbances requires high values of fr. This means that low values of m are required. Therefore, the force transducer of the probe should be soft and small. If a rectangle cantilever is chosen as a force transducer, it has to be thin and small. Microfabricated cantilevers meet this demand and, for this reason, they became standard probes for AFM. The design details of a rectangular cantilever are considered below. If the length, width, and thickness of the cantilever are l, w, and h, respectively, the spring constant k is given by
A
25
Length: l = 500 μm
1.5
(2)
30 Spring constant Resonance frequency
Spring constant k, N/m
rffiffiffiffi 1 k fr ¼ ; 2p m
2
5
0 0
1
2 3 Thickness h, nm
4
5
0
AFM Probes, Fig. 2 Thickness dependencies of the spring constant and resonance frequency of a microcantilever
opposite conditions. Thus, a balance between the spring constant and the resonance frequency has to be found in designing the probe. An example for contact mode cantilevers is shown in Fig. 2, where the thickness dependencies of the spring constant and resonance frequency are plotted using Eqs. 3 and 4 for a length l ¼ 500 mm and a width w ¼ 50 mm, respectively. If a spring constant of less than 1 N/m and a resonance frequency larger than 10 kHz are selected, the former curve requires a thickness of more than 1.9 mm and the latter curve requires a thickness of less than 4.0 mm. Therefore, a thickness in the overlapping range should be selected. A probe with these dimensions is not easy to fabricate by conventional machining and requires micromachining techniques. Measurements of the spring constant k and resonance frequency fr are important for converting the measured deflection signal into a force signal. The resonance frequency fr can be measured by varying the drive frequency as it is routinely done in commercial AFM equipments. However, the accurate measurement of the spring constant k is not easy. An approximate value can be calculated from the dimensions of the cantilever using Eq. 3. An experimental method using the thermal fluctuation was presented in Ref. [3].
A
92
AFM Probes
AFM Probes, Fig. 3 Fabrication process of a microfabricated probe: (a) silicon substrate, (b) formation of pit for a microtip, (c) deposition of silicon nitride layer, (d) patterning of cantilever, and (e) removal of substrate
a
d
Silicon substrate Pit
b
e
Silicon oxide layer
c
Glass base
Freestanding cantilever
Silicon nitride layer
Many types of AFMs using a vibrating cantilever have been developed, such as cyclic contact mode and non-contact mode AFMs. The probe vibrations can be induced using a quartz-oscillator. In this case, the resonance properties, especially the quality factor Q, are very important. The quality factor Q is defined as Q ¼ 1/2z, where z is the damping coefficient. A probe with a high Q is able to resonate effectively and its vibration is hard to damp. The minimum detectable force gradient in the frequency-modulation detection, which is widely used in highly sensitive AFMs such as non-contact AFMs and magnetic force microscopy, is given by sffiffiffiffiffiffiffiffiffiffiffiffiffi @F 2 kkB TB ¼ ; @z min A or Q
Cantilever shape
(5)
where B, or, and A are the measurement band width, resonance angular frequency, and vibration amplitude, respectively [4]. Higher force sensitivity requires a higher Q. Since the vibration damping of the probe is caused by the internal friction and by friction between the probe and environmental gas such as air, the quality factor Q usually depends on the probe structure and environment. Typical values of Q of silicon cantilevers are on the order of 10–100 in air whereas they can increase up to more than 10,000 in vacuum. A method to increase Q electronically was
presented, where the deflection signal is fed back to the drive signal after the phase of the deflection signal is shifted by 90 [5].
Probe Fabrication At present, most of the probes are fabricated by microfabrication techniques and various types of probes have been designed. Moreover, a considerable number of them are commercially available because microfabricated probes are suitable for mass production. The fabrication methods have also been improved, corresponding to a general improvement of the probe quality. Here, the fabrication methods for two typical types of probes are outlined. We will first address the probes made of silicon nitride. These probes are widely used in contact AFM mode. The fabrication process is shown in Fig. 3 [6]. In this method, a small pit that is formed in a silicon substrate by anisotropic wet etching is used as a mold. The silicon oxide layer is formed as an etching mask for the later substrate removal (Figs. 3a and b). A silicon nitride film is deposited onto the substrate and the cantilever is patterned by lithography. At this point, the cantilever with tip is shaped (Figs. 3c and d). By attachment of the glass base by anodic bonding and removal of the silicon substrate by wet etching,
AFM, Non-contact Mode
a freestanding cantilever is obtained (Fig. 3e). Another type of probes are silicon cantilevers [7]. Silicon cantilevers are the most widely used probes in AFM and they are operated in various modes. By using the undercut etching of the underneath of a circular mask, a tip is formed. Then the cantilever is patterned. The substrate silicon is removed by wet etching with protection of the fabricated tip and cantilever. Sharpening the tip end can be achieved by thermal oxidation followed by oxide removal [6].
Future Directions for Research In addition to standard silicon or silicon nitride cantilever probes, various types of probes have been recently introduced. In a piezoresistive probe a piezoresistive strain sensor is embedded with a silicon cantilever [8]. The embedded sensor detects the cantilever deflection and makes the detection system such as an optical lever unnecessary. This not only makes the setup compact but also provides easier operation in the special case of liquid or vacuum environments. Along the growth of microfabricated cantilever probes, probes that use a quartz-oscillator as a force transducer have also been developed. Various types of oscillators such as tuning fork or linear-extension resonators are used [9]. In many probes, a microtip is manually glued to the quartz-oscillator although the oscillator is microfabricated. Quartz-oscillator probes have advantages of high Q and large spring constant. The latter allows to avoid the jump-in of the probe. In addition, quartz-oscillator probes provide self-sensing as piezoresistive probes. As stated above, the improvement of the probes is essential for the development of AFM. Recent advances in micro/nanomachining are promoting this improvement, which is expected to open up new applications of AFM.
93
A
▶ NEMS ▶ Scanning Probe Microscopy
A References 1. Binnig, C., Quate, C.F., Gerber, Ch: Atomic force microscope. Phys. Rev. Lett. 56(9), 930–933 (1986) 2. Binnig, G., Gerber, C., Stoll, E., Albrecht, T.R., Quate, C.F.: Atomic resolution with atomic force microscope. Europhys. Lett. 3(12), 1281–1286 (1987) 3. Hutter, J.L., Bechhoefer, J.: Calibration of atomic-force microscope tips. Rev. Sci. Instrum. 64(7), 1868–1873 (1993) 4. Albrecht, T.R., Grutter, P., Horne, D., Rugar, D.: Frequency modulation detection using high-Q cantilevers for enhanced force microscope sensitivity. J. Appl. Phys. 69(2), 668–673 (1991) 5. Anczykowski, B., Cleveland, J.P., Kr€ uger, D., Elings, V., Fuchs, H.: Analysis of the interaction mechanisms in dynamic mode SFM by means of experimental data and computer simulation. Appl. Phys. A66, S885–S889 (1998) 6. Albrecht, T.R., Akamine, S., Carver, T.E., Quate, C.F.: Microfabrication of cantilever styli for the atomic force microscope. J. Vac. Sci. Technol. A8, 3386–3396 (1990) 7. Wolter, O., Bayer, Th, Greschner, J.: Micromachined silicon sensors for scanning force microscopy. J. Vac. Sci. Technol. B9, 1353–1357 (1991) 8. Tortonese, M., Barrett, R.C., Quate, C.F.: Atomic resolution with an atomic force microscope using piezoresistive detection. Appl. Phys. Lett. 62(8), 834–836 (1993) 9. Giessibl, F.J.: Atomic resolution on Si(111)-(7 7) by noncontact atomic force microscopy with a force sensor based on a quartz tuning fork. Appl. Phys. Lett. 76(11), 1470–1472 (2000)
AFM Tips ▶ AFM Probes
AFM, Non-contact Mode Hendrik Ho¨lscher Karlsruher Institut f€ur Technologie (KIT), Institut f€ur Mikrostrukturtechnik, Karlsruhe, Germany
Cross-References ▶ Atomic Force Microscopy ▶ Friction Force Microscopy ▶ MEMS ▶ Micromachining
Definition The noncontact atomic force microscopy (NC-AFM) is a specific AFM technique primarily developed for
94
AFM, Non-contact Mode, Fig. 1 The schematic set-up of a dynamic force microscope based on the frequency modulation technique often used in UHV. A significant feature is the positive feedback of the self-driven cantilever. The detector signal is amplified and phase shifted before it is used to drive the piezo. The measured quantity is the frequency shift due to the tip-sample interaction, which serves as the feedback signal for the cantileversample distance
AFM, Non-contact Mode
laser
electronics with frequency detector
photo diode
amplitude frequency amplifier phase shifter
θ0 frequency
A
z piezo
D+2A
aexc
d sample
D
PID setpoint x−y−z−scanner
application in vacuum where standard AFM cantilevers made from silicon or silicon nitride exhibit very high quality factors Q, what makes the response of the system slow if driven in AM or tapping mode (▶ AFM, Tapping Mode). The technique to oscillate the cantilever also in high Q environments is called frequencymodulation (FM) mode. In contrast to the tapping or AM mode typically applied in air or liquids, this approach features a so-called self-driven oscillator, which uses the cantilever deflection itself as drive signal, thus ensuring that the cantilever instantaneously adapts to changes in the resonance frequency. The NC-AFM technique is the method of choice to obtain true atomic resolution on non-conduction surfaces with an atomic force microscope.
Overview To obtain high resolution images with an atomic force microscope it is most important to prepare clean sample surfaces free from unwanted adsorbates. Therefore, these experiments are usually performed in ultra-high vacuum with pressures below 1 10–10 mbar. As a consequence most dynamic force microscope experiments in vacuum utilize the so-called frequency modulation (FM) detection scheme introduced by Albrecht et al. [1]. In this mode the cantilever is selfoscillated, in contrast to the AM- or tapping-mode (▶ AFM, Tapping Mode). The FM-technique enables
the imaging of single point defects on clean sample surfaces in vacuum and its resolution is comparable with the scanning tunneling microscope, while not restricted to conducting surfaces [2–6]. In the years after the invention of the FM-technique the term non-contact atomic force microscopy (NC-AFM) was established, because it is commonly believed that a repulsive, destructive contact between tip and sample is prevented by this technique. Set-Up of FM-AFM In vacuum applications, the Q-factor of silicon cantilevers is in the range of 10,000–30 000. High Q-factors, however, limit the acquisition time (bandwidth) of a dynamic force microscopy, since the oscillation amplitude of the cantilever needs a long time to adjust. This problem is avoided by the FM-detection scheme based on the specific features of a self-driven oscillator. The basic set-up of a dynamic force microscope utilizing this driving mechanism is schematically shown in Fig. 1. The movement of microfabricated cantilevers is typically measured with the laser beam deflection method or an interferometer. A self-detecting sensor like a tuning fork do not need an additional detection sensor. In any case the amplitude signal fed back into an amplifier with an automatic gain control (AGC) and is subsequently used to excite the piezo oscillating the cantilever. The time delay between the excitation signal and cantilever deflection
AFM, Non-contact Mode
is adjusted by a time (“phase”) shifter to a value of 90 , since this ensures an oscillation at resonance. Two different modes have been established: The constant amplitude-mode [1], where the oscillation amplitude A is kept at a constant value by the AGC, and the constant excitation mode [7], where the excitation amplitude is kept constant. In the following, however, only the constant amplitude mode is discussed. The key feature of the described set-up is the positive feedback-loop which oscillates the cantilever always at its resonance frequency f [8]. The reason for this behavior is that the cantilever serves as the frequency determining element. This is in contrast to an external driving of the cantilever in tapping mode by a frequency generator (▶ AFM, Tapping Mode). If the cantilever oscillates near the sample surface, the tip-sample interaction alters its resonant frequency, which is then different from the eigenfrequency f0 of the free cantilever. The actual value of the resonant frequency depends on the nearest tip-sample distance and the oscillation amplitude. The measured quantity is the frequency shift D f, which is defined as the difference between both frequencies (Df :¼ f – f0). For imaging the frequency shift D f is used to control the cantilever sample distance. Thus, the frequency shift is constant and the acquired data represents planes of constant D f, which can be related to the surface topography in many cases. Origin of the Frequency Shift Before presenting experimental results obtained in vacuum the origin of the frequency shift is analyzed in mode detail. A good insight into the cantilever dynamics is given by looking at the tip potential displayed in Fig. 2. If the cantilever is far away from the sample surface, the tip moves in a symmetric parabolic potential (dotted line), and its oscillation is harmonic. In such a case, the tip motion is sinusoidal and the resonance frequency is given by the eigenfrequency f0 of the cantilever. If, however, the cantilever approaches the sample surface, the potential – which determines the tip oscillation – is modified to an effective potential Veff (solid line) given by the sum of the parabolic potential and the tip-sample interaction potential Vts (dashed line). This effective potential differs from the original parabolic potential and shows an asymmetric shape. As a result of this modification of the tip potential the oscillation becomes anharmonic, and the resonance
95 V(z)
A
cantilever potential tip−sample potential effective potential
A
E
D
D + 2A
z
AFM, Non-contact Mode, Fig. 2 The frequency shift in dynamic force microscopy is caused by the tip-sample interaction potential (dashed line), which alters the harmonic cantilever potential (dotted line). Therefore, the tip moves in an anharmonic and asymmetric effective potential (solid line)
frequency of the cantilever depends now on the oscillation amplitude A. Since the effective potential experienced by the tip changes also with the nearest distance D, the frequency shift is a functional of both parameters () Df :¼ D f (D, A)). Figure 3 displays some experimental frequency shift versus distance curves for different oscillation amplitudes [9]. The obtained experimental frequency shift vs. distance curves show a behavior expected from the simple model explained above. All curves show a similar overall shape, but differ in magnitude in dependence of the oscillation amplitude and the nearest tip-sample distance. During the approach of the cantilever towards the sample surface, the frequency shift decreases and reaches a minimum. With a further reduction of the nearest tip-sample distance, the frequency shift increases again and becomes positive. For smaller oscillation amplitudes, the minimum of the D f (z)-curves is deeper and the slope after the minimum is steeper than for larger amplitudes, i.e., the overall effect is larger for smaller amplitudes. This can be explained by the simple potential model as well: A decrease of the amplitude A for a fixed nearest distance D moves the minimum of the effective potential closer to the sample surface. Therefore, the relative perturbation of the harmonic cantilever potential increases, which increases also the absolute value of the frequency shift.
96
AFM, Non-contact Mode 0.8
0.4
b
0.0 180 Å 126 Å 90 Å 72 Å 54 Å
–0.4
–0.8 –10
–5
0
5
10
0.10
norm. freq. shift [Nm1 / 2 x 10–15]
a
0.05
0.00 180 Å 126 Å 90 Å 72 Å 54 Å
–0.05
–0.10 –10
15
–5
distance D [Å]
c
8
tip-sample force [nN]
AFM, Non-contact Mode, Fig. 3 (a) Experimental frequency shift versus distance curves acquired with a silicon cantilever (cz ¼ 38 N/m; f0 ¼ 171 kHz) and a graphite sample for different ˚ ) in amplitudes (54–180 A UHV at low temperature (T ¼ 80 K). (b) Transformation of all frequency shift curves shown in (a) to one universal curves using Eq. 6. The normalized frequency shift g (D) is nearly identical for all amplitudes. (c) The tip-sample force calculated with the experimental data shown in (a) and (b) using the formula Eq. 7
frequency shift [kHz]
A
4
0
5
10
15
distance D [Å]
0 180 Å 126 Å 90 Å 72 Å 54 Å
–4
–8 –10
–5
0
5
10
15
distance D [Å]
Theory of FM-AFM As already described in the previous subsection it is a specific feature of the FM-modulation technique that the cantilever is “self-driven” by a positive feedback loop. Due to this experimental set-up, the corresponding equation of motion is different from the case of the externally driven cantilever discussed for the tapping-mode. The external driving term has to be replaced in order to describe the self-driving mechanism correctly. Therefore, the equation of motion is given by 2pf0 m zðtÞ _ þ cz ðzðtÞ dÞ Q þ gcz ðzðt t0 Þ dÞ |fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl}
zðtÞ ¼ d þ A cosð2pftÞ
m€ zðtÞ þ
driving
¼ Fts ½zðtÞ; zðtÞ: _
where z :¼ z(t) represents the position of the tip at the time t; cz, m, and Q are the spring constant, the effective mass, and the quality factor of the cantilever, respectively. Fts ¼ – (∂ Vts)/(∂z) is the tip-sample interaction force. The last term on the left describes the active feedback of the system by the amplification of the displacement signal by the gain factor g measured at the retarded time t – t0. The frequency shift can be calculated from the above equation of motion with the ansatz
(1)
(2)
describing the stationary solutions of Eq. 1. Again, it is assumed that the cantilever oscillations are more or less sinusoidal and develop the tip-sample force Fts
AFM, Non-contact Mode
97
into a Fourier-series. As a result the following equation for the frequency shift is obtained Df ¼
1 pcz A2
Z
dþA dA
zd ðF# þ F" Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dz (3) A2 ðz dÞ2
1 f g pcz A2 : Q f0
(4)
Since the amplitudes in FM-AFM are often considerably larger than the distance range of the tip-sample interaction, apply the “large amplitude approximation” [10, 11] can be applied. This yields the formula 1 f0 Df ¼ pffiffiffi 2p cz A3=2
Dþ2A Z
D
Fts ðzÞ pffiffiffiffiffiffiffiffiffiffiffiffi dz zD
(5)
It is interesting to note that the integral in this equation is virtually independent of the oscillation amplitude. The experimental parameters (cz, f0, and A) appear as pre-factors. Consequently, it is possible to define the normalized frequency shift [10] gðzÞ:¼
Here the approach of D€urig [11] is presented, which is based on the inversion of the integral Eq. 5. It can be transformed to 1 pffiffiffi cz A3=2 @ Z Df ðzÞ pffiffiffiffiffiffiffiffiffiffiffiffi dz; Fts ðDÞ ¼ 2 f0 @D zD
(7)
D
and the energy dissipation DE ¼
A
cz A3=2 Df ðzÞ f0
(6)
This is a very useful quantity to compare experiments obtained with different amplitudes and cantilevers. The validity of Eq. 6 is nicely demonstrated by the application of this equation to the frequency shift curves already presented in Fig. 3a. As shown in Fig. 3b all curves obtained for different amplitudes result into one universal g-curve, which depends only on the actual tip-sample distance D. These equations help to calculate the frequency shift for a given tip-sample interaction law. The inverse problem, however, is even more interesting: How can the tip-sample interaction be determined from frequency shift data? Several solutions to this question have been presented by various authors and have lead to the dynamic force spectroscopy (DFS) technique, which is a direct extension of the FM-AFM mode [12].
which allows a direct calculation of the tip-sample interaction force from the frequency shift versus distance curves. An application of this formula to the experimental frequency shift curves already presented in Fig. 3a is shown in Fig. 3c. The obtained force curves are nearly identical although obtained with different oscillation amplitudes. Since the tip-sample interactions can be measured with high resolution, dynamic force spectroscopy opens a direct way to compare experiments with theoretical models and predictions. Applications of FM-AFM The excitement about the NC-AFM technique in ultrahigh vacuum was driven by the first results of Giessibl [13] who achieved to image the true atomic structure of the Si(111)-7 7-surface with this technique in 1995. In the same year Sugawara et al. [14] observed the motion of single atomic defects on InP with true atomic resolution. However, imaging on conducting or semi-conducting surfaces is also possible with the scanning tunneling microscope (STM) and these first NC-AFM images provided no new information on surface properties. The true potential of NC-AFM lies in the imaging of non-conducting surface with atomic precision. A long-standing question about the surface reconstruction of the technological relevant material Aluminium oxide could be answered by Barth et al. [15], who imaged the atomic structure of the high temperature phase of a-Al2O3(0001). The high resolution capabilities of non-contact atomic force microscopy are nicely demonstrated by the images shown in Fig. 4. Allers et al. [16] resolved atomic steps and defects with atomic resolution on Nickel oxide. Today such a resolution is routinely obtained by various research groups (for an overview see, e.g., Refs. [2–6]). Recent efforts have also been concentrated on the analysis of functional organic molecules, since in the field of nanoelectronics it is anticipated that in particular organic molecules will
A
A
98
AFM, Non-contact Mode
˚ (b) A dopant atom is imaged as a light protrusion about 0.1 A higher as the other atoms (Images courtesy of W. Allers and S. Langkat, University of Hamburg; used with kind permission)
AFM, Non-contact Mode, Fig. 4 Imaging of a NiO(001) sample surface with a non-contact AFM. (a) Surface step and an ˚. atomic defect. The lateral distance between two atoms is 4.17 A
a z
b 0.4
D y
z–direction/nm
D+2A
0.3
–0.2 nN
0.2 0.1
–1.0 nN
0 x
0
0.2 0.4 0.6 0.8 y–direction/nm
1
AFM, Non-contact Mode, Fig. 5 (a) Principle of 3D-force spectroscopy. The cantilever oscillates near the sample surface and measure the frequency shift in a x-y-z-box. The three dimensional surface shows the topography of the sample
˚ 10 A ˚ ) obtained immediately before the (image size: 10 A recording of the spectroscopy field. (b) The reconstructed force field of NiO(001) shows atomic resolution. The data is taken along the line shown in (a)
play an important role as the fundamental building blocks of nanoscale electronic device elements. The concept of dynamic force spectroscopy can be also extended to 3D-force spectroscopy by mapping the complete force field above the sample surface [12]. Figure 5a shows a schematic of the measurement principle [17]. Frequency shift vs. distance curves are recorded on a matrix of points perpendicular to the sample surface. Using Eq. 7 the complete threedimensional force field between tip and sample can be recovered with atomic resolution. Figure 5b shows a cut through the force field as a two-dimensional map. If the NC-AFM is capable of measuring forces between single atoms with sub-nN precision, why should it not be possible to also exert forces with this technique? In fact, the new and exciting field of
nanomanipulation would be driven to a whole new dimension, if defined forces can be reliably applied to single atoms or molecules. In this respect, Loppacher et al. [18] achieved to push on different parts of an isolated Cu-TBBP molecule, which is known to possess four rotatable legs. They measured the forcedistance curves while one of the legs was pushed by the AFM tip and turned by 90 , thus being able to measure the energy which was dissipated during “switching” this molecule between different conformational states. The manipulation of single Sn-atoms with the NC-AFM was nicely demonstrated by Sugimoto et al. who manipulated single Sn-atoms on the Ge(111)-c(2 8) semiconductor surface (Fig. 6). By pushing single Sn-atoms from one lattice site to the other they finally succeeded to write the letter “Sn” with single atoms.
AFM, Tapping Mode
AFM, Non-contact Mode, Fig. 6 Final topographic NC-AFM image of the process of rearranging single Sn-atoms on a Ge (111)-c(2 8) semiconductor surface at room temperature (Reproduced from [19])
Cross-References ▶ AFM, Tapping mode ▶ Friction Force Microscopy ▶ Kelvin Probe Force Microscopy ▶ Magnetic Resonance Force Microscopy ▶ Piezoresponse Force Microscopy and Spectroscopy
References 1. Albrecht, T.R., Gr€ utter, P., Horne, D., Rugar, D.: Frequency modulation detection using high-Q cantilevers for enhanced force microscope sensitivity. J. Appl. Phys. 69, 668 (1991) 2. Morita, S., Wiesendanger, R., Meyer, E. (eds.): Noncontact Atomic Force Microscopy. Springer, Berlin (2002) 3. Garcia, R., Pe´rez, R.: Dynamic atomic force microscopy methods. Surf. Sci. Rep. 47, 197 (2002) 4. Giessibl, F.-J.: Advances in atomic force microscopy. Rev. Mod. Phys. 75, 949 (2003) 5. Meyer, E., Hug, H.J., Bennewitz, R.: Scanning Probe Microscopy - The Lab on a Tip. Springer, Berlin (2004) 6. Morita, S., Giessibl, F.J., Wiesendanger, R. (eds.): Noncontact Atomic Force Microscopy, vol. 2. Springer, Berlin (2009) 7. Ueyama, H., Sugawara, Y., Morita, S.: Stable operation mode for dynamic noncontact atomic force microscopy. Appl. Phys. A 66, S295 (1998) 8. Ho¨lscher, H., Gotsmann, B., Allers, W., Schwarz, U.D., Fuchs, H., Wiesendanger, R.: Comment on “Damping mechanism in dynamic force microscopy”. Phys. Rev. Lett. 88, 019601 (2002) 9. Ho¨lscher, H., Schwarz, A., Allers, W., Schwarz, U.D., Wiesendanger, R.: Quantitative analysis of dynamic force spectroscopy data on graphite(0001) in the contact and noncontact regime. Phys. Rev. B 61, 12678 (2000)
99
A
10. Giessibl, F.J.: Forces and frequency shifts in atomic-resolution dynamic-force microscopy. Phys. Rev. B 56, 16010 (1997) 11. D€ urig, U.: Relations between interaction force and frequency shift in large-amplitude dynamic force microscopy. Appl. Phys. Lett. 75, 433 (1999) 12. Baykara, M.Z., Schwendemann, T.C., Altman, E.I., Schwarz, U.D.: Three-dimensional atomic force microscopy taking surface imaging to the next level. Adv. Mater. 22, 2838 (2010) 13. Giessibl, F.-J.: Atomic resolution of the silicon (111)-(7 7) surface by atomic force microscopy. Science 267, 68 (1995) 14. Sugawara, Y., Otha, M., Ueyama, H., Morita, S.: Defect motion on an InP(110) surface observed with noncontact atomic force microscopy. Science 270, 1646 (1995) 15. Barth, C., Reichling, M.: Imaging the atomic arrangement on the high-temperature reconstructed a-Al2O3(0001) surface. Nature 414, 54 (2001) 16. Allers, W., Langkat, S., Wiesendanger, R.: Dynamic lowtemperature scanning force microscopy on nickel oxide (001). Appl. Phys. A 72, S27 (2001) 17. Ho¨lscher, H., Langkat, S.M., Schwarz, A., Wiesendanger, R.: Measurement of threedimensional force fields with atomic resolution using dynamic force spectroscopy. Appl. Phys. Lett. 81, 4428 (2002) 18. Loppacher, C., Guggisberg, M., Pfeiffer, O., Meyer, E., Bammerlin, M., Luthi, R., Schlittler, R., Gimzewski, J.K., Tang, H., Joachim, C.: Direct determination of the energy required to operate a single molecule switch. Phys. Rev. Lett. 90, 066107 (2003) 19. Sugimoto, Y., Abe, M., Hirayama, S., Oyabu, N., Custance, O., Morita, S.: Atom inlays performed at room temperature using atomic force microscopy. Nat. Mater. 4, 156 (2005)
AFM, Tapping Mode Hendrik Ho¨lscher Karlsruher Institut f€ur Technologie (KIT), Institut f€ur Mikrostrukturtechnik, Karlsruhe, Germany
Definition The tapping or AM-mode is the most common dynamic mode used in atomic force microscopy. In dynamic mode AFM the cantilever is oscillated with (or near) its resonance frequency near the sample surface. Using a feedback electronic the cantilever sample distance is controlled by keeping either the amplitude or the phase of the oscillating cantilever constant. Since lateral tip–sample forces are avoided by this technique the resolution is typically higher
A
100
AFM, Tapping Mode, Fig. 1 Setup of a dynamic force microscope operated in the tapping (or AM-) mode. A laser beam is deflected by the back side of the cantilever, and its deflection is detected by a split photo-diode. The cantilever vibration is caused by an external frequency generator driving an excitation piezo. A lock-in amplifier is used to compare the cantilever driving with its oscillation. The amplitude signal is held constant by a feedback loop controlling the cantilever sample distance
AFM, Tapping Mode
laser
electronics with lock−in amplifier
photo diode
phase amplitude
z piezo
D+2A
aexc function generator
d sample
D
amplitude
A
PID setpoint x−y−z−scanner
compared to the classical contact mode AFM where tip and sample are in direct mechanical contact.
Overview Since its introduction in 1986 [1], the atomic force microscope became a standard tool in nanotechnology. In early experimental setups, a sharp tip located at or near the end of a microstructured cantilever profiled the sample surface in direct mechanical contact (contact mode) to measure the force acting between tip and sample. Maps of constant tip–sample interaction force, which are usually regarded as representing the sample’s “topography,” were then recovered by keeping the deflection of the cantilever constant (▶ Friction Force Microscopy). This is achieved by means of a feedback loop that continuously adjusts the z-position of the sample during the scan process so that the output of the deflection sensor remains unchanged at a preselected set-point. Despite the widespread success of contact mode AFM in various applications, the resolution was frequently found to be limited (in particular for soft samples) by lateral forces acting between tip and sample. In order to avoid this effect, it is advantageous to vibrate the cantilever in vertical direction near the sample surface. AFM imaging with an oscillating cantilever is often denoted as dynamic force microscopy (DFM) [2].
The cantilever dynamics are governed by the tip– sample interaction as well as by its driving method. For instance, the historically oldest scheme of cantilever excitation in DFM imaging is the external driving of the cantilever at a fixed excitation frequency chosen to be exactly at, or very close to, the cantilever’s first resonance. For this driving mechanism, different detection schemes measuring either the change of the oscillation amplitude or the phase shift between driving signal and resulting cantilever motion were proposed. Over the years, the amplitude modulation (AM) or tapping mode, where the actual value of the oscillation amplitude is employed as a measure of the tip–sample distance, has been established as the most widely applied technique for ambient conditions and liquids. Experimental Setup A schematic of the experimental setup of an atomic force microscope driven in amplitude modulation mode is shown in Fig. 1. In this mode the cantilever is vibrated close to its resonant frequency near the sample surface. Due to the tip–sample interaction the resonant frequency (and consequently also amplitude A and phase f) of the cantilever changes with the cantilever sample distance d. Therefore, the amplitude as well as the phase can be used as feedback channels. A certain set-point for example, the amplitude is given, and the feedback loop will adjust the tip–sample distance such that the amplitude remains constant. The cantilever sample distance is
AFM, Tapping Mode
101
a
A
b
9 8 7 6 5 4 3 2 1 0
Phase [deg]
topogrphy [Å]
A
0
1
2
3
4
5
scan position [mm]
3 2 1 0 -1 -2 -3 -4 -5 -6
0
1
2
3
4
5
scan position [mm]
AFM, Tapping Mode, Fig. 2 (a) A dynamic force microscopy image of a monomolecular DPPC (L-a-dipalmitoyl-phophatidycholine) film adsorbed on mica. (b) The phase contrast is
different between substrate and DPPC film. Equation 6 explains how this contrast is related to the energy dissipation caused by the tip–sample contact interaction
recorded as a function of the lateral position of the tip with respect to the sample and the scanned height essentially represents the surface topography. The deflection of the cantilever is typically measured with the laser beam deflection method. During operation in conventional tapping mode, the cantilever is driven with a fixed frequency and a constant excitation amplitude using an external function generator, while the resulting oscillation amplitude and/or the phase shift are detected by a lock-in amplifier. The function generator supplies not only the signal for the dither piezo; its signal serves simultaneously as a reference for the lock-in amplifier. This setup is mostly used in air and in liquids. A typical image obtained with this experimental setup in ambient conditions is shown in Fig. 2. The
phase between excitation and oscillation can be acquired as an additional channel and gives information about the different material properties of DPPC and the mica substrate. As shown at the end of the next section, the phase signal is closely related to the energy dissipated in the tip–sample contact. Due to its technical relevance the investigation of polymers has been the focus of many studies and high-resolution imaging has been extensively performed in the area of material science. Using specific tips with additionally grown sharp spikes Klinov et al. [3] obtained molecular resolution on a polydiacetylene crystal. Imaging in liquids opens up the avenue for the investigation of biological samples in their natural environment. For example, Mo¨ller et al. [4] have obtained high-resolution images
A
102
AFM, Tapping Mode
AFM, Tapping Mode, Fig. 3 Topography of DNA adsorbed on mica imaged in buffer solution by tapping mode AFM. The right graph shows a single scan line obtained at the position marked by an arrow in the left image
height (nm)
1.0
0.5
0.0
of the topography of hexagonally packed intermediate (HPI) layer of Deinococcus radiodurans with tappingmode AFM. A typical example for the imaging of DNA in liquid solution is shown in Fig. 3. Theory of Tapping-Mode AFM Many features observed in tapping-mode AFM can be described by a simple spring-mass model which includes the tip–sample interaction force.
0
100
200 300 400 scan-position (nm)
500
tip is nearly spherical and that the noncontact forces are given by the long-range van-der-Waals forces while the contact forces are described by the well-known Hertz model. The resulting overall force law is given by ( FDMTM ðzÞ ¼
4 3E
AH R pffiffiffi 6z2 Rðz0 zÞ3=2
for for
z z0 ; z < z0 ; (2)
m€ zðtÞ þ
2pf0 m zðtÞ _ þ cz ðzðtÞ dÞ Q0
¼ ad cz cosð2pfd tÞ þ Fts ½zðtÞ; zðtÞ _ : |fflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} external driving force
(1)
tipsample force
Here, z(t) is the position of the tip at the time t; cz, m, pffiffiffiffiffiffiffiffiffiffiffiffiffi ffi and f0 ¼ ðcz =mÞ=ð2pÞ are the spring constant, the effective mass, and the eigenfrequency of the cantilever, respectively. The quality factor Q0 combines the intrinsic damping of the cantilever and all influences from surrounding media, such as air or liquid. The equilibrium position of the tip is denoted as d. The first term on the right-hand side of the equation represents the external driving force of the cantilever by the frequency generator. It is modulated with the constant excitation amplitude ad at a fixed frequency fd. The (nonlinear) tip–sample interaction force Fts is introduced by the second term. The actual tip–sample force is unknown in nearly all cases. However, in order to understand the most important effects it is often sufficient to apply the DMT-M theory [5]. In this approach it is assumed the
where AH is the Hamaker constant, E* the effective modulus, and R the tip radius. At z0 tip and sample come in contact. Figure 4 displays the resulting tip– sample force curve for this model. For the analysis of dynamic force microscopy, experiments only on the steady states with sinusoidal cantilever oscillation are of interest. Consequently, the steady-state solution is given by the ansatz
zðt 0Þ ¼ d þ A cosð2pfd t þ fÞ;
(3)
where f is the phase difference between the excitation and the oscillation of the cantilever. Here, the situation where the driving frequency is set exactly to the eigenfrequency of the cantilever (fd ¼ f0) is analyzed. With this choice, which is also very common in actual DFM experiments, defined imaging conditions are given leading to a handy formula relationship between the free oscillation amplitude A0, the actual amplitude A, and the equilibrium tip position d [6].
AFM, Tapping Mode
A
6 tip-sample force (nN)
AFM, Tapping Mode, Fig. 4 Tip–sample model force after the DMT-M model for air Eq. 2 using the typical parameters: AH ¼ 0.2 aJ, R ¼ 10 nm, z0 ¼ 0.3 nm, and E* ¼ 1 GPa. The dashed line marks the position z0 where the tip touches the surface
103
4
A
van-der-Waals force Hertz model
2 0 -2 -4 -1.5
-1.0
-0.5
0.0
z0 0.5
1.0
1.5
2.0
tip-sample distance (nm)
10
(4)
where Z 1=fd 2fd Iþ ðd; AÞ ¼ Fts ½zðtÞ; zðtÞ _ cosð2pfd t þ fÞdt cz A 0 Z dþA 1 zd ¼ ðF# þ F" Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dz: pcz A2 dA A2 ðz dÞ2
Oscillation amplitude A (nm)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi A0 ¼ A 1 þ ðQ0 Iþ ½d; AÞ2 ;
bistable regime 8
6
4 approach retraction
2
(5) 0
The function I+(d, A) depends on the actual oscillation amplitude A and cantilever–sample distance d and it is a weighted average of the tip–sample forces during approach and retraction (F# + F"). The solution of this equation allows us to study amplitude vs. distance curves as shown in Fig. 5 for a Q-factor of 300. Most noticeable, the tapping mode curve exhibits jumps between unstable branches, which occur at different locations for approach and retraction. The resulting bistable regime then causes a hysteresis between approach and retraction and divides the tip–sample interaction into two regimes. In order to identify the forces acting between tip and sample in these two regimes, the oscillation amplitude is plotted as a function of the nearest tip–sample distance in Fig. 6. In addition, the bottom graph depicts the corresponding tip–sample force (cf. Fig. 4). The origin of the nearest tip–sample position D is defined by this force curve. Since both, the amplitude curves and the tip–sample force curve, are plotted as a function of the nearest tip–sample position, it is
0
2
4 6 8 cantilever distance d (nm)
10
12
AFM, Tapping Mode, Fig. 5 Amplitude vs. distance curve for tapping-mode AFM assuming the parameters A0 ¼ 10 nm, f0 ¼ 300 kHz, and the tip–sample interaction force shown in Fig. 4. The overall amplitude decreases the sample surface distance, but instabilities (indicated by arrows) occur during approach and retraction
possible to identify the resulting maximum tip–sample interaction force for a given oscillation amplitude. During the approach of the vibrating cantilever toward the sample surface, there is discontinuity in the nearest tip–sample position D. This gap corresponds to the bi-stability and the resulting jumps in the amplitude vs. distance curve. After the jump from the attractive to the repulsive regime has occurred, the amplitude decreases continuously. The nearest tip–sample position, however, does not reduce accordingly, remaining roughly between 0.8 and 1.5 nm. As a result, larger A/A0 ratios do not necessarily result into lower tip– sample interactions, which is important to keep in
104
10 amplitude (nm)
AFM, Tapping Mode, Fig. 6 A comparison between the maximum tip–sample forces (tip–sample forces acting at the point of closest tip–sample approach/nearest tip–sample position D) experienced by conventional “tapping mode” AM-AFM assuming the same parameters as in Fig. 5. The upper graph shows the nearest tip–sample position D vs. the actual oscillation amplitude A for tapping mode. The graph at the bottom reveals the force regimes sensed by the tip. The maximal tip–sample forces in tapping mode are on the repulsive (tapping regime) as well as attractive (bistable tapping regime) part of the tip–sample force curve
AFM, Tapping Mode
8 6 approach 4
retraction
2 0 6 tip-sample force (nN)
A
tapping regime
4
bistable tapping regime
2 0 -2 -4 -1.5
-1.0
mind while adjusting imaging parameters in tapping mode. For practical applications, it is reasonable to assume that the set-point of the amplitude used for imaging has been set to a value between 90% and 10% of the free oscillation amplitude. With this condition, one can identify the accessible imaging regimes indicated by the horizontal (dashed) lines and the corresponding vertical (dotted) lines. Therefore, in tapping mode, two imaging regimes are typically accessible: the tapping regime (left) and the bistable tapping regime (middle). The bistable imaging state is only accessible during approach. Imaging in this regime is indeed possible with the limitation that the oscillating cantilever might jump into the repulsive regime [7, 8]. In the above paragraphs, the influence of the tip– sample interaction on the cantilever oscillation was analyzed, the maximum tip–sample interaction forces based on the assumption of a specific model force was calculated, and subsequently possible routes for image optimization were discussed. However, in practical imaging, the tip–sample interaction is not a priori known. Therefore, several authors [9–14] suggested solutions to this inversion problem, but some of them need further technical equipment. However, most commercial systems give access to
-0.5 -0.0 0.5 1.0 nearest tip-sample distance D (nm)
1.5
2.0
amplitude and phase vs. distance data which already allows the reconstruction of the tip–sample force curve applying numerical procedures [11, 12, 14]. The energy dissipation caused by the tip–sample interaction can be easily calculated using the conservation of energy principle [15]. DE ¼
1 f d ad þ sin f pcz A2 : Q0 f0 A
(6)
Since all parameters – except the phase f – are constant during scanning, this formula shows that the energy dissipation is roughly proportional to sin f as already mentioned in the caption of Fig. 2.
Cross-References ▶ AFM, Non-contact Mode ▶ Friction Force Microscopy ▶ Kelvin Probe Force Microscopy ▶ Magnetic Resonance Force Microscopy ▶ Piezoresponse Force Microscopy and Spectroscopy
Anamorphic Particle Tracking
References 1. Binnig, G., Quate, C.F., Gerber, Ch: Atomic force microscopy. Phys. Rev. Lett. 56, 930–933 (1986) 2. Garcia, R., Pe´rez, R.: Dynamic atomic force microscopy methods. Surf. Sci. Rep. 47, 197–301 (2002) 3. Klinov, D., Maganov, S.: True molecular resolution in tapping-mode atomic force microscopy with highresolution probes. Appl. Phys. Lett. 84, 2697–2698 (2004) 4. Mo¨ller, C., Allen, M., Elings, V., Engel, A., M€ uller, D.J.: Tapping-mode atomic force microscopy produces faithful high-resolution images of protein surfaces. Biophys. J. 77, 1150–1158 (1999) 5. Schwarz, U.D.: A generalized analytical model for the elastic deformation of an adhesive contact between a sphere and a flat surface. J. Coll. Interf. Sci. 261, 99–106 (2003) 6. Ho¨lscher, H., Schwarz, U.D.: Theory of amplitude modulation atomic force microscopy with and without Q-control. Int. J. Nonlinear Mech. 42, 608–625 (2007) 7. Paulo, A.S., Garcı´a, R.: High-resolution imaging of antibodies by tapping-mode atomic force microscopy: Attractive and repulsive tip-sample interaction regimes. Biophys. J. 78, 1599–1605 (2000) 8. Stark, R.W., Schitter, G., Stemmer, A.: Tuning the interaction forces in tapping mode atomic force microscopy. Phys. Rev. B 68, 085401 (2003) 9. Stark, M., Stark, R.W., Heckl, W.M., Guckenberger, R.: Inverting dynamic force microscopy: From signals to time-resolved interaction forces. PNAS 99, 8473–8478 (2002) 10. Legleiter, J., Park, M., Cusick, B., Kowalewski, T.: Scanning probe acceleration microscopy (SPAM) in fluids: Mapping mechanical properties of surfaces at the nanoscale. Proc. Natl. Acad. Sci. U.S.A. 103, 4813–4818 (2006) 11. Lee, M., Jhe, W.: General solution of amplitude-modulation atomic force microscopy. Phys. Rev. Lett. 97, 036104 (2006) 12. Ho¨lscher, H.: Quantitative measurement of tip-sample interactions in amplitude modulation atomic force microscopy. Appl. Phys. Lett. 89, 123109 (2006) 13. Sahin, O., Maganov, S., Chanmin, Su, Quate, C., Solgaard, O.: An atomic force microscope tip designed to measure time-varying nanomechanical forces. Nat. Nanotechnol. 2, 507–514 (2007) 14. Shuiqing, Hu, Raman, A.: Inverting amplitude and phase to reconstruct tip-sample interaction forces in tapping mode atomic force microscopy. Nanotechnology 19, 375704 (2008) 15. Cleveland, J.P., Anczykowski, B., Schmid, A.E., Elings, V.B.: Energy dissipation in tapping-mode atomic force microscopy. Appl. Phys. Lett. 72, 2613 (1998)
Aging ▶ Fate of Manufactured Nanoparticles in Aqueous Environment
105
A
ALD ▶ Atomic Layer Deposition
Alkanethiols ▶ BioPatterning
AlN ▶ Piezoelectric MEMS Switches
Alteration ▶ Fate of Manufactured Nanoparticles in Aqueous Environment
Aluminum Nitride ▶ Piezoelectric MEMS Switches
Amphibian Larvae ▶ Ecotoxicology of Carbon Nanotubes Toward Amphibian Larvae
Anamorphic Particle Tracking ▶ Astigmatic Micro Particle Imaging
A
A
106
Angiogenesis Alessia Bottos1, Elena Astanina1, Luca Primo2 and Federico Bussolino1 1 University of Torino, Department of Oncological Sciences, Candiolo, Torino, Italy 2 University of Torino, Department of Clinical and Biological Sciences, Candiolo, Italy
Definition Originally, the term “angiogenesis” was strictly confined to describe the sprouting of blood vessels from preexisting capillaries or post-capillary venules. Presently, angiogenesis refers to the molecular and cellular mechanisms leading to the formation and remodeling of vascular tree occurring during embryo development and in adult life. However, this term often includes the concept of vasculogenesis. Vasculogenesis indicates the earlier events of the formation of the cardiovascular system and is characterized by the assembly of a primitive vascular plexus constituted by angioblasts, which are believed to arise from the hemangioblast, a common precursor for endothelial and hematopoietic lineages. There are two other general concepts that strictly parallel angiogenesis. The first is arteriogenesis, which describes the growth of functional collateral arteries from preexisting arterio-arteriolar anastomoses in response to ischemic injuries. The second is lymphangiogenesis, which illustrates the formation of the lymphatic system. This forms a unidirectional network of blind-ended terminal lymphatic vessels that collect the protein-rich fluid that exudes from blood capillaries and then drain through a conduit system of collecting vessels, lymph nodes, lymphatic trunks and ducts into the venous circulation.
Introduction The notion that blood circulation is indispensable for bringing nutrients to the organs and for organism growth has been known for millennia. However, the connected concept – that growth of blood vessels may promote disease – is more recent [1, 2].
Angiogenesis
The perception that tumor growth is linked to an increased vascularization took place in the midtwentieth century. In 1971, Judah Folkman gave the first formal demonstration that tumors release soluble molecules able to recruit from surrounding tissue capillaries, which enable cancer to growth over few millions of cells, infiltrate the normal parenchyma, and colonize distant organs. These observations opened an extraordinary research area that is far to be completely exploited and is at the frontier of the molecular medicine. The greatest achievement of these studies has been made over the past years in targeting angiogenesis for human therapy. In particular, in the 2004 vascular-endothelial growth factor (VEGF)-A removal by the humanized monoclonal antibody bevacizumab and by the specific aptamer pegaptanib was, respectively, approved for the treatment of metastatic colorectal cancer and exudative retinopathy [3]. Currently more than 50 anti-angiogenic compounds are in clinical trials. However, clinical experience has shown that the antivascular effect of bevacizumab as well as of other anti-angiogenic compounds are often transient or even negligible [4, 5], indicating that new knowledge is required to better understand the complexity of vascular formation and to properly select responder patients. Nevertheless the impact of angiogenesis in medicine overcomes oncology and is extended to other relevant pathologies including ischemic diseases, chronic inflammatory diseases, metabolic disorders, obesity, some mendelian inherited diseases, reproductive disorders, and bone diseases (Table 1).
The Morpho-functional Vascular Unit Arteries, veins, and capillaries share a common structure (tunica intima) characterized by endothelial cells (EC) that line vascular lumens and are surrounded by the basement membrane (BM) constituted by extracellular matrix proteins and complex sugars (i.e., heparan sulfates, proteoglycans). The external face of tunica intima is covered by vascular smooth muscle cells and pericytes (mural cells) that form a multilayer wall of different size (thinner in the veins and ticker in the arteries), named tunica media. In capillaries, pericytes do not form a continuous layer but they are scattered and embedded in the BM. Mural cells exert key roles in multiple microvascular processes, including:
Angiogenesis
107
Angiogenesis, Table 1 Human diseases characterized by abnormal vascularization Organs Whole body Vessels
Adipose tissue Skin Eye
Lung Intestines Reproductive system Bone, joints
Diseases Cancer, autoimmune diseases, ischemic diseases, diabetes, infections Vascular malformations, cavernous hemangiomas, hemangiomas, atherosclerosis, hereditary hemorrhagic telangiectasia, Di George syndrome, glomerulonephritis, transplant arteriopathy Obesity Psoriasis, warts, blistering disease, allergic dermatitis, scar keloid, pyogenic granulomas Persistent hyperplastic vitreous syndrome, Norrin disease, exudative retinopathy, retinopathy of prematurity Primary pulmonary hypertension, asthma, nasal polyps Inflammatory bowel disease, ascites, periodontal diseases Endometriosis, ovarian cysts, ovarian hyperstimulation Arthritis, synovitis, osteomyelitis, osteophyte formation
(1) endothelial cell proliferation and differentiation, (2) contractility and tone, (3) stabilization and permeability, and (4) morphogenesis and remodeling during disease onset [6, 7]. EC thickness varies from less than 0.1 mm in capillaries and veins to 1 mm in the aorta and in arteries they are aligned in the direction of blood flow and this remodeling occurs in response to hemodynamic shear stress. On the contrary in vasculature characterized by a slow blood stream, endothelial shape may be irregular, rounded, or elliptical. Endothelium may be continuous or discontinuous (Fig. 1). Continuous endothelium, in turn, is fenestrated or nonfenestrated. Nonfenestrated continuous endothelium is found in arteries, veins, and capillaries of the brain, skin, heart, and lung. Fenestrated continuous endothelium occurs in tissues that are characterized by increased filtration or increased transendothelial transport, including exocrine and endocrine glands, gastrointestinal mucosa, and kidney. Fenestrae are transcellular pores (70 nm in diameter) that develop through the whole thickness of the cell and have a thin 5–6 nm nonmembranous diaphragm across their aperture. Discontinuous endothelium characterizes capillaries of sinusoidal vascular beds (i.e., liver and bone
A
marrow). It is characterized by large fenestrations (100–200 nm in diameter) that lack a diaphragm and contain gaps (or large holes) within individual cells. The regulation of these kinds of EC-EC interactions occurs at two types of intercellular junctions: the tight junctions (also named zona occludens), which are mainly located at the apical region of the intercellular cleft and the adherens junctions (zona adherens). Zonula occluden-1, occludin, claudin, and JAM are the major elements of tight junctions, while vascular-endothelial cadherin is a specific constituent of adherens junctions and its clustering at cell-cell contacts promotes the formation of multimolecular complexes that comprise signaling, regulatory, and scaffold proteins. In physiologic and pathological conditions, the above-described structural features of endothelium contribute to maintain the fluid and the cellular homeostasis between blood and tissues. Actually, the vascular system needs to be sufficiently permeable to allow the ready exchange of small molecules (gases, nutrients, waste products) with the tissues. Plasma proteins also need to cross the normal vascular barrier, at least in small amounts. Albumin, for example, transports fatty acids and vitamins and immunoglobulin antibodies are required for host defense. The passage of small molecules across the endothelial layer is almost passive and has to take into account three parameters: hydraulic conductivity, reflection coefficient, and diffusion. Diffusion is the most important of these for the exchange of small molecules and is driven by the molecular concentration gradient across vascular endothelium (Fick equation). On the other hand, filtration is much more important than diffusion for the flux of large molecules such as plasma proteins and is determined by the Starling equation. On structural point of view, endothelial permeability is mediated by the so-called transcellular and paracellular pathways – that is, solutes and cells can pass through (transcellular) or between (paracellular) EC. While the paracellular route restricts the passage of solutes larger than 3 nm in radius, transcellular vesicle trafficking selectively transports macromolecules such as albumin across the endothelium. Transcellular passage (transcytosis) is an energydependent mechanism that requires either cell fenestration or a complex system of transport vesicles, which includes caveolae and organelles called vesiculo-vacuolar organelles (VVO). These organelles, or similar structures, can also fuse and appear
A
A
108
Angiogenesis
Angiogenesis, Fig. 1 Morphology of capillaries. Capillaries in different anatomical districts have different morphology. Continuous nonfenestrated EC is mainly found in skin, heart, muscle, brain, and lung. It regulates the passage of molecules through transendothelial channels or by caveolae-mediated transcytosis. Continuous fenestrated ECs characterize glomerulus, endocrine glands, and intestines and demonstrate greater permeability to water and small molecules but do not allow the passage of larger macromolecules by using diaphragms as molecular filters. Discontinuous endothelium is present in liver, bone marrow, and spleen and has fenestrae (without diaphragms), gaps, and partially structured BM. The intercellular cleft represents tight junctions or adherens junctions
as channels that traverse single cells and allow the passage of leukocytes and solutes through the endothelium. Caveolae are 70-nm membrane-bound, flaskshaped vesicles that usually open to the luminal or abluminal side of the EC membrane. They are enriched in cholesterol and some, but not all, caveolae possess a thin diaphragm similar to that found in fenestrae. The density of caveolae is far greater in capillary endothelium (up to 10,000 per cell) compared with other vessels and their number is highest in continuous nonfenestrated endothelium. A notable exception is the blood-brain barrier, where caveolae are rare, which reflects the protective role of this vascular bed for the brain. VVOs are comprised of hundreds of cytoplasmic vesicles and vacuoles that together form an organelle that traverses venular endothelial cytoplasm from lumen to albumen. VVOs often extend to interendothelial cell interfaces and their individual vesicles commonly open to the inter-endothelial cell cleft. The vesicles and vacuoles comprising VVOs are linked to each other and to the luminal and abluminal plasma
membranes by stomata that are normally closed by thin diaphragms that appear similar to those found in caveolae and fenestrae. The paracellular pathway, by contrast, is mediated by the coordinated opening and closure tight junctions and adherens junctions, and is mainly operative during pathological conditions, in which this barrier is altered by the release of permeability factors (acute and chronic inflammatory injury) or by an abnormal and chaotic vascularization (i.e., tumor angiogenesis).
The Morpho-functional Vascular Unit in Pathological Settings The formation of vascular bed depends from well-regulated circuits that involve soluble molecules with inducing and inhibiting activity, mechanical forces, different cell types and features of the extracellular matrix. The final result of these dynamic interactions is the ordered distribution of capillaries (the intercapillary distance ranged from 100 to
Angiogenesis
150 mm) allowing the appropriate cell oxygenation on the basis of local pO2 and O2 coefficient diffusion [6–8]. In pathological settings (i.e., tumor, chronic inflamed tissues, wound healing, postischemic revascularization) this homeostatic balance is disturbed leading to the formation of vessels with functional and structural defects. In pathological tissues and in particular in tumors, vessels are dilated, tortuous, and show a chaotic spatial distribution. Normal vascular tree is characterized by vessels with decreasing diameter and by dichotomous branching, but tumor vessels are unorganized with trifurcations and irregular diameters. Endothelial phenotype is also modified in tumors. Large discontinuities characterize inter-endothelial junctions, as well as the increased numbers of fenestrae and VVOs and alterations or lack of BM. Similarly, mural cells are poorly distributed and show defects in association with vessel wall. The molecular and cellular mechanisms causing these abnormal vascular architectures are largely unknown, but the imbalance of pro- and anti-angiogenic associated with mechanicals stresses is believed to be instrumental in achieving these features. These structural alterations are accountable for some features of tumor microenvironment. In normal tissues permeation of hydrophilic solutes is progressively restricted with increasing molecular size, as by a sieve, with many openings 8 nm and a few 40–60 nm wide. In solid tumors vascular permeability is generally higher allowing ( 80–90 nm) or facilitating ( 10 nm) the extravasation of molecules with sizes that in general are retained. The blood flow is irregular because the flow resistance, which depends from vascular geometry and blood viscosity, is increased. Focal interruption of vessel wall may also compromise the downstream blood flow. As a result, overall perfusion rates (blood flow rate per unit volume) in tumors are lower than in healthy organs. Blood flux is erratically distributed and fluctuates with time causing a jeopardized perfusion. The increased permeability abolishes the differences between hydrostatic and colloid osmotic pressures that characterize normal tissues. As a result, tumors show interstitial hypertension and reduction of transmural pressure gradient and convection across tumor vessel walls. Consequently, the increased permeability connected with the interstitial hypertension
109
A
reduces pressure difference between up and down stream of blood vessels and leads to blood flow stasis. The final consequences of these functional abnormalities takes place at metabolic level leading to hypoxia and acidosis. Tumor vessels fail to adequately deliver oxygen and nutrient as well as to remove wastes. The imbalance of vascular network development and tumor cell proliferation results in the formation of hypovascular regions in tumors. Since tissue diffusion limit of oxygen is 80–200 mm, the regions far from blood vessels become hypoxic. Extracellular acidosis is a consequence of hypoxia and Warburg effect (cancer cells produce energy through glycolysis also in the presence of adequate amount of oxygen). When pO2 is low, the metabolic process (glycolysis followed by oxidative phosphorylation) sustaining oxidative demolition of glucose and ATP production decreases its efficiency. Actually, in the absence of oxygen, the end product of glycolysis, pyruvic acid, does not proceed toward oxidative phosphorylation but it is reduced to lactic acid. This metabolite, together carbonic acid generated by growing cells, accounts for the reduction of extracellular pH in tumors. The above-described metabolic and functional characteristics render the microenvironment hostile, leading to a genetic selection that favors the most aggressive and malignant cancer cells. Finally, the combination of acidosis (which may modify the polarity of hydrophilic compound) with the blood stasis and the increased interstitial pressure hinders the delivery and efficacy of therapeutic agents to tumors.
The Mechanisms of Angiogenesis Mesenchymal cells differentiate in angioblasts, which undergo vasculogenesis where they aggregate and form a primitive vascular network firstly located into extraembryonic tissues and then invading the embryo. As vessels begin to be remodeled, they undergo localized proliferation and regression, as well as programmed branching and migration into different regions of the body. They need to be specified into different calibers and types of vessel, including division into arteries, veins, and lymphatics, with further subdivision into large vessels, venules, arterioles, and capillaries. In addition, they need to recruit supporting mural cells and create the optimal extracellular matrix to ensure the stability of the vessels formed. Pulsatile
A
A
110
and shear stress forces mediated by heart beats are instrumental for capillary branching and pruning. The establishment of final vascular architecture is characterized by four different mechanisms, which are also present in adult life: (1) formation of primitive and immature plexus, (2) stabilization of primitive plexus, (3) remodeling toward the final architecture, and (4) vascular specialization. Molecules involved in these steps are summarized in Table 2 [9–12]. Formation of primitive and immature vascular plexus – During embryo development, the nascent capillary network is generated by combining vasculogenesis and angiogenesis. Vasculogenesis describes the de novo formation of vascular plexus. Mesenchymal cells are stimulated by soluble molecules (i.e., Indian hedgehog, basic fibroblast growth factor, bone morphogenic protein 4) released by endodermal cells to differentiate in vascular-endothelial growth receptor-2 (VEGFR-2)-positive cells. In extraembryonic yolk sac, these cells form hemangioblast, the common precursor of vascular and hematopoietic system. Primitive endothelial and hematopoietic cells coalesce to form blood islands. The outer cells of the blood islands are endothelial, whereas the inner cells give rise to hematopoietic progenitors. The endothelial-lined blood islands fuse to generate de novo a primary capillary plexus. This process is named vasculogenesis. In intra-embryonic tissues, VEGFR2-positive mesenchymal cells differentiate into angioblasts which directly aggregate into the dorsal aorta or cardinal vein, without a plexus intermediate. These large vessels then undergo connection with the primitive vascular network. Besides vasculogenesis, sprouting and intussusceptive angiogenesis take place (Fig. 2). Sprouting angiogenesis if facilitated by hypoxia, which upregulates the transcription of a large set of pro-angiogenic genes. EC loose the cell-cell contacts and vessels become leaky. The BM is digested by the release of specific proteases (metalloproteases) and a specific suppression of protease inhibitors allowing EC migration and proliferation. Intriguingly, BM cleavage can produce multiple anti-angiogenic molecules that counteract and balance the effect of angiogenic inducers thus allowing an accurate vascularization. The sprouts reorganize internally to form a vascular lumen and are finally connected to other capillary segments. VEGF signaling is critically important in this process, as it promotes endothelial cell proliferation and modulates migration.
Angiogenesis
Angiopoietin-2, in the presence of VEGF, facilitates sprouting angiogenesis because it contributes in dismantling cell contacts. The pivotal effect of VEGF in sprouting angiogenesis is regulated by Notch receptors and their Delta-like-4 (DLL4) and Jagged-1 ligands to allow a specific focalization of capillary area from which a new capillary stems. The fine-tuning activity between VEGF and DLL4 allows the selection of the tip EC, which is devoid of proliferative activity and drives the sprout by emitting filopodia. Initially, Dll4 and Notch signaling are thought to be balanced in EC until presumptive tip cells eventually increase Dll4 expression in response to VEGF signaling. Consequently, Notch is upregulated in neighboring cells, which inhibits the expression of VEGF receptors and the tip cell differentiation. Thus, Dll4-expressing tip cells react strongest to VEGF signaling and acquire a motile, invasive, and sprouting phenotype whereas tip cell behavior is suppressed in Jag1-expressing (stalk) cells, which form the base of the emerging sprout. Reduced levels of DLL4 or blocking of Notch signaling enhance the formation of tip cells, resulting in dramatically increased sprouting and chaotic angiogenesis. The principle in which a differentiating cell represses the differentiation of adjacent cells is called lateral inhibition (Fig. 3). More recently ephrins and wnt pathway has been demonstrated to modulate tip cell dynamics. Although little is known about the lumenization of blood vessels, there is strong evidence for the involvement of pinocytosis and vacuole formation. High-resolution time-lapse imaging has established that the lumen in these ECs is formed by intracellular and, subsequently, intercellular fusion of large vacuoles. Intussusception (growth within itself) is an alternative to the sprouting mode of angiogenesis. The protrusion of opposing microvascular walls into the capillary lumen creates a contact zone between ECs. The endothelial bilayer is perforated, intercellular contacts are reorganized, and a transluminal pillar with an interstitial core is formed, which is soon invaded by myofibroblasts and pericytes leading to its rapid enlargement by the deposition of collagen fibrils. When a primitive capillary plexus is generated by vasculogenesis or sprouting, intussusception is triggered and is responsible for rapid vascular growth and optimal remodeling. Stabilization of primitive plexus – Blood perfusion of nascent capillaries requires their maturation. This
Angiogenesis
111
A
Angiogenesis, Table 2 Most important molecules involved in angiogenesis Angiogenic inducers (ligand/receptor) VEGF-A/VEGFR-1 Negative regulation of embryonic angiogenesis. Later, VEGFR-1 acts as decoy receptor. Recruitment of monocytes and macrophages VEGF-A/VEGFR-2 Regulation of proliferation, permeability, migration, and survival of ECs. ECM remodeling. EC differentiation and specialization (i.e., fenestrae formation). The activity of VEGFR-2 is modulated by association with co-receptors (neuropilins, VE-cadherin, avb3 integrin). VEGF-A isoforms contribute to vascular patterning Placental growth factor/VEGFR-1 Regulation of pathological angiogenesis (i.e., tumor, collateral formation, wound healing). Contribution to hematopoiesis. Recruitment of inflammatory cells VEGF-B/VEGFR-1 Cardiac vascularization VEGF-C or VEGF-D/VEGFR-2 Control of lymphangiogenesis in physiologic and pathological settings. Modulators of vascular and/or VEGFR-3 angiogenesis Notch/Dll4 or Jagged Control of EC tip formation in sprouting angiogenesis. Differentiation and recruitment of mural cells. Venous-arterial differentiation Angiopoietins/Tie-2 Regulation of the relationships between mural cells and EC. Regulation of vessel stabilization. Accessory role in lymphangiogenesis. Activation of inflammatory cells TGFb/ALK or endoglin Production of ECM and proteases. Venous-arterial differentiation. Mural cell differentiation. Control of EC proliferation and migration PDGFs/PDGF receptors Differentiation, survival, and recruitment of mural cells Ephrin B2/EphB4 Venous-arterial differentiation. Vascular branching Slits/Robo Formation of primitive plexus. Vascular stabilization. Control of EC motility Netrins/UNC5B Negative control of sprouting angiogenesis. Regulation of vascular branching Semaphorins/plexins-Neuropilins Regulation of vascular morphogenesis. Inhibition of EC adhesion and migration. Anti-angiogenic effect in tumor angiogenesis Hepatocyte growth factor/Met Regulation of pathological angiogenesis. Collateral formation Chemokines/Chemokine receptors Members that contain the “ELR” motif are potent promoters of angiogenesis via binding and activating CXCR2 on ECs. In contrast, members, in general, those lack the ELR motif (ELR-) are potent inhibitors of angiogenesis, and bind to CXCR3 on ECs Endogenous angiogenic inhibitors Tumstatin, arresten, and canstatin They are cryptic angiostatic peptides released by collagen type IV during ECM proteolysis in sprouting angiogenesis Endostatin A cryptic angiostatic peptide released by collagen type XVIII during ECM proteolysis in sprouting angiogenesis Angiostatin A plasminogen fragment produced during sprouting angiogenesis. It Inhibits EC migration and proliferation and induces apoptosis Vasostatin Vasostatin is an N-terminal fragment of the calreticulin. It inhibits angiogenesis presumably by binding to laminin and blocking EC adhesion and proliferation Thrombospondin It inhibits EC proliferation and migration by interacting with CD36 molecule 2-methoxyestradiol It inhibits angiogenesis by reducing EC proliferation and inducing EC apoptosis Histidine-rich glycoprotein It inhibits tumor angiogenesis by preventing release of angiogenic factors from the matrix and inhibiting growth factor-induced EC migration Pigment epithelial-derived factor It is released by retinal pigment epithelium and causes EC apoptosis Maspin It is a serpin inhibitor of serine-proteases with anti-angiogenic activity in tumor PEX It is a fragment of metalloproteinase-2 that prevents its collagenolytic activity during sprouting angiogenesis Fibulin 5 It is a ECM protein that antagonizes VEGF activities and enhances that of thrombospondin Interferons (a,b,g) They inhibit angiogenesis by multiple and direct effects on EC or by modulating innate immunity
A
A
112
Angiogenesis
Angiogenesis, Fig. 2 Vasculature assembly. Vasculogenesis starts when mesenchymal cells stimulated by soluble molecule released by endoderm differentiate into EC precursors (hemangioblasts and angioblasts) and form the blood islands (left). Fusion of blood islands leads to formation of primary capillary. Sprouting angiogenesis. Blood circulation is established and primary plexi are remodeled into a hierarchical network of arteries, capillaries, and veins. Sprouting angiogenesis starts when a gradient of VEGF triggers the detachment of EC-EC contacts leading EC to proliferate and migrate.
The nascent capillaries undergo maturation characterized by reorganization of ECM and recruitment of mural cells. This step parallels the selection of the proper trajectory for the best blood supply. Intussusception. The division of vessels through the insertion of tissue pillars is an alternative mechanism that leads to the expansion of blood vessels in several organs. Little is known about the function or regulation of intussusceptive growth but the process seems to involve EC growth, motility, and remodeling of ECM
Angiogenesis, Fig. 3 Tip cell selection in sprouting angiogenesis. (a) At the beginning of VEGF stimulation ECs most probably contain equal amount of Dll4 and its receptor Notch. (b) Tip selection occurs by increasing Dll4 expression in a specific EC (red cell) stimulated by VEGF. As consequence, Notch is upregulated in neighboring cells (blue), which inhibit the
expression of VEGF receptors. (c) This modulation allows tip cells reacting strongest to VEGF signaling and acquiring sprouting phenotype whereas tip cell behavior is suppressed in Jag1-expressing (stalk) cells (blue), which form the base of the emerging sprout
Angiogenesis
step is mainly characterized by recruitment of mural cells and remodeling of extracellular matrix (ECM). Mural cells are recruited by molecules released by EC and in particular by platelet-derived growth factors (PDGF) and sphingosine-1 phosphate, which, respectively, bind the tyrosine kinase PDGF receptor and the seven spanning domains endothelial differentiation sphingolipid G-protein-coupled receptor-1. Compelling support for this hypothesis comes from mouse genetic models in which the modification of the expression of these molecules impair the correct relationships between mural cells and EC and alter production of ECM. Mural and stroma cells produce angiopoietin-1 (Ang-1), which binds the tyrosine kinase receptor, Tie2 on endothelial cells, resulting in enhanced EC-mural cell interactions leading the vessels leak-resistant. A homolog, Ang-2, appears to antagonize Ang-1, relaxing these interactions and triggering BM degradation in preparation for angiogenic sprout. Transforming-growth factor b promotes the differentiation of mesenchymal cells into mural cells and contributes to ECM production. Mutations in endoglin and activin-like kinase, two receptors of transforming-growth factor b, cause hereditary hemorrhagic telangiectasia, a disease that is characterized by capillary dilatation, microaneurysms, and bleeding. Interactions between ECs and their surrounding BM and ECM are critical for vascular maturation and occur through specialized adhesive receptors named integrins. The importance of these connections are underscored by the vascular defects seen in mice deficient in some types of integrins and various BM constituents such as type IV collagen, fibronectin, and laminins. Interestingly, the adhesion of EC to BM through integrins mediates anti-apoptotic signals that are instrumental to avoid vessel regression. Actually, anti-integrin molecules, including specific antibodies, have anti-angiogenic effects. Besides an adhesive role, integrins act as co-receptors of several tyrosine kinase receptors involved in angiogenesis modulating their activities. Following BM assembly, vessels are stabilized by cross-linking of ECM components and associated proteins by transglutaminase-2-mediated lysine bonds. The ECM cross-linking activity leads to inhibition of angiogenesis by increasing ECM stiffness and deposition. Remodeling toward the final architecture – The transition from the plexus to the vascular tree is characterized by regression and pruning of some vascular
113
A
segments, new branching, remodeling, and navigation to build the most appropriate network to distribute oxygen, nutrients, and wash catabolites off. This step is regulated by the combined activities of physical forces, axon guidance cues, and ECM features. Using in vivo molecular imaging associated with genetic and vascular grafting studies it has been established that physical cues drive guidance of lumenized vessel sprouts. ECs are subjected to tension created by cellcell and cell-ECM interactions or by blood flow (shear stress) and pressure (circumferential wall stress). Interestingly, the absence of blood flow or sufficient tensions promotes vascular regression. Physical stimuli must be sensed by cells and transmitted through intracellular transduction pathways to the nucleus, resulting in gene expression. However, how cells sense biomechanical stimuli is still under debate. Several hypotheses experimentally supported have been proposed. They include the capacity of heterotypic (integrinmediated cell-ECM) and homotypic (VE-cadherin mediated EC-EC interaction) adhesive receptors to link the internal cytoskeleton and provide mechanical coupling across the cell surface, including ion- channel opening [13]. As above mentioned, proteases released by angiogenic EC modify the structure of ECM during the formation of primitive plexus. Furthermore, this mechanism occurs also during the remodeling phase leading to pro- or anti-angiogenic activities. This stimulatory effect may be due to decreasing the density of extracellular matrix proteins, and/or by exposing cryptic binding sites within matrix molecules that promote migration. In contrast, the action of proteases can be anti-angiogenic, due to their ability to generate fragments of matrix molecules that have pro-apoptotic properties not possessed by the intact molecule. Therefore the characteristics of surrounding ECM may favor the progression or the regression of primitive capillary network. On the other hand ECM elements, and in particular fibronectin, collagens, laminins, and heparan sulfates, participate in vascular stabilization and patterning as demonstrated by specific vascular defects (increase in lumen size, edema, leakage, vessel rupture) in mouse genetic models. ECM also affects the availability of angiogenic regulators. For instance, different isoforms of VEGF-A, characterized by increased affinity to ECM, form specific gradients required for vessel patterning. Similarly, fibroblast growth factor-2 may be
A
A
114
sequestered by ECM and released to cell surface by proteolysis. Generally, ECM and proteolytic activities adapt the strength and the duration of angiogenic signals and train ECs to initiate fine-tuned and specific signal genetic programs regulating vascular patterning. During the remodeling step, vessels select specific trajectories leading to an optimal spatial distribution instrumental for the molecular exchanger in the tissues. This strategy is shared with the nervous system, in which axon guidance molecules are responsible for guiding the navigating axon toward its target area. Indeed ECs and neurons share some fundamental mechanisms during the formation of their networks. On an anatomical point of view, in several tissues (skin, muscles, intestine) nerves and larger blood vessel align into two parallel structures and genetic studies indicate that peripheral nerves in the dermis provide a spatial template for the growth and differentiation of arteries. During axon navigation, growth cones project numerous filopodia as shown in endothelial tip cells that actively extend and retract in response to extracellular cues. Four ligand-receptor pairs of guidance cues are known to play attractive of repulsive effects in nervous and vascular systems: ephrins/Eph, semaphorins/plexins-neuropilins, slit/Robo, and netrins/UNC-DCC. Many studies performed in genetic modified mice or zebrafish clearly indicate that these molecules mainly act in defining the final shape and architecture of vessels in body tissues. Vascular specialization – The paradigmatic example of this concept is the differentiation of arteries and veins. During vascular development cardiac work presses high-pressure blood flow through large diameter arterial vessels, which acquire an extensive supporting system of mural cells. Blood comes back to the heart with low pressure through veins, which develop venous valves to prevent blood from flowing back. Therefore hemodynamic forces have a crucial role in defining arterial and venous circulation. These mechanisms are strictly intermingled with early genetic determinant. The landmark of arterial–venous cell fate is the expression of ephrinB2 and EphB4, respectively, in plasma membrane of arterial and venous ECs. The interactions between these two markers are essential for proper vascular development. However, the expression of these molecules is differently controlled. VEGF upregulates the Notch pathway that through hesr1/hesr2 transcription factors leads to arterial specification via EphrinB2 activation. In veins,
Angiogenesis
COUP-TFII is required for strong expression of EphB4 and venous differentiation. Therefore it seems that molecular differences between arteries and veins are predetermined but the final fate is contributed by epigenetic and physical cues indicating a prominent plasticity of the system. The vascular architecture has definite features in different organs to respond to specific needs in terms of nutrient requirements and transport [14]. For instance, the filtration barrier of glomerular endothelial cells in kidney results from the presence of continuous fenestrations allowing bulk fluid and anionically charged molecules to be filtered and facilitates movement of cationically charged molecules. The sinusoidal hepatic system is characterized by discontinuous endothelium with sieve-like pores ( 100 nm diameter) and nondiaphragmatic fenestrae, which regulate the permeability and the blood flow in hepatic sinusoids. Endocrine organs have fenestrated capillaries that contain pores sometimes lined with thin diaphragms. These structures are thought to allow increased transport of long-range hormones between extravascular cells and blood. In contrast to the fenestrated vascular beds present in high-transport organs, brain capillaries are linked by extensive tight junctions that restrict transendothelial movement. These differences are furtherer envisaged by specific patterns of gene expression orchestrated by mechanisms largely unexplored. It is emerging the existence of angiogenic inducers specific for each organ. They include members of Wnt family and endocrine glandVEGF.
The Mechanisms of Pathological Angiogenesis In adult life, physiologic angiogenesis employs the same mechanisms illustrated above and occurs in endometrium during female cycle, in corpus luteum during reproductive cycle, in hypertrophic glands (i.e., breast during lactation), and in exercise-trained skeletal muscles. Several diseases are connected with and sustained by an abnormal angiogenesis. In some pathological settings, including tumors and chronic inflammatory diseases, abnormal angiogenesis results from an altered homeostasis between angiogenic inducers and inhibitors driven and sustained by specific pathogenic events. On the contrary, ischemic
Angiogenesis
115
A A
Angiogenesis, Fig. 4 Homeostatic hypothesis in pathological angiogenesis. During embryonic development and in some process in adult life (i.e., female cycle, reproduction, lactation, muscle exercise), tissues express balanced amounts of activators and inhibitors of angiogenesis, which allow the formation of regular vessels. In other adult tissues, angiogenesis is repressed by an increased amount of inhibitors. Vascular regeneration may
occur in ischemic tissues but usually it is insufficient to recover tissue oxygenation. On the other hand, tumors, obesity, chronic inflammatory diseases, and retinopathies are characterized by a prominent but chaotic angiogenesis. These different scenarios are associated with unbalanced ratios between activators and inhibitors
diseases are characterized by attempts to revascularize the necrotic area through angiogenesis and arteriogenesis (Fig. 4) [2, 15, 16 ]. Tumor angiogenesis – Oncogene-mediated cell transformation is necessary but not sufficient to promote metastatic and aggressive cancer. Tumor cell proliferation alone, in the absence of angiogenesis, can give rise to dormant, microscopic tumors of 1 mm3 or less, but these in situ cancers are harmless to the host. Oncogene somatic mutations deeply modify the transcription profile of transformed cells, including the upregulation of VEGF and other growth factors often associated with the down-modulation of inhibitors of vessel formation (i.e., thrombospondin). Therefore the accumulation of genetic mutation along the tumor span-life progressively modifies the angiogenic homeostasis leading to the formation of a chaotic vasculature (see above). However the process of tumor angiogenesis is not limited to the altered production of factors in tumor cells and other mechanisms may occur. The angiogenic process may be partially sustained by EC precursors mobilized by bone marrow and migrated to the angiogenic site. Bone marrow participates to the angiogenic process by producing
myeloid cells that differently modulate angiogenesis. For example, macrophages have long been characterized as a highly plastic cell type capable of tumorsuppressive or tumor-promoting effects, depending on the polarization state (M1 vs. M2). Tie2-expressing monocytes have also been reported to promote tumor growth through secretion of angiogenic factors, as also demonstrated for neutrophils and platelets. Interestingly the accumulation of CD11b+Gr1+ cells in mice has been associated with the refractoriness to antiVEGF therapies. At least in their early phase, some tumors do not mediate sprouting angiogenesis but they utilize preexisting vessels. For instance, brain tumors are highly vascularized and vessels have a similar phenotype than normal brain. Vessels are surrounded, co-opted by tumor cells, and no sprouts are observed (co-option mechanism). Later, when the tumor mass collapses normal capillaries, hypoxia takes places and activates a VEGF-mediated sprouting angiogenesis. In some tumors and in particular in colon-rectal cancers, the capillary wall is often constituted by EC and tumor cells (mosaic vessel). This scenario may be induced by the endothelial migration during rapid
A
116
vessel growth without sufficient proliferation to complete the endothelial lining, leaving cancer cells exposed to the lumen. Second, ECs could be shed from the vessel lining, leaving exposed tumor cells. Third, migrating tumor cells could invade the vessel wall, displacing endothelial cells from the lining. Finally vascular mimicry describes the potential of tumor cells (i.e., in melanomas) or cancer stem cells (at least in glioblastoma) to differentiate in EC with a mixed phenotype and be able to form functional vessels. Despite a significant body of literature, vascular mimicry is not yet a universally accepted mechanism, as the concept has been challenged by several investigators. Arteriogenesis – The development of an arterial circulation circumventing the occlusion of a large artery from preexistent small arterioles is named arteriogenesis and occurs exclusively in adult injured tissues. It shares some features with angiogenesis, but the pathways leading to it are different. Furthermore arteriogenesis is potentially able to fully replace an occluded artery whereas angiogenesis cannot. As demonstrated in embryonic arterial differentiation, a pivotal role in arteriogenesis is played by physical forces and in particular by shear stress. It is proportional to blood flow velocity and inversely related to the cube of the collateral vessel radius Therefore, since growth increases the collateral vessel radius, shear stress falls quickly. This may be the reason why arteriogenesis arrests without a completed vascularization of ischemic tissue. Collateral vessels of an occluded artery show activated ECs, able to release molecules involved in the recruitment of circulating monocytes, in vascular proliferation and remodeling (proteases, Notch ligands, integrins, VEGF, angiopoietins, fibroblast growth factors) and in control of vascular tone (nitric oxide). Furthermore these collaterals are leaky. A similar activation is observed in smooth muscle cells, which undergo a massive proliferation. Therefore it seems to be conceivable to hypothesize that high fluid shear stress starts at least two signaling mechanisms: one that attracts bone marrow-derived cells for remodeling and another that causes endothelial and smooth muscle cells to enter the cell cycle, leading to proliferation. Acknowledgments Some of the concepts here reported originate from experimental papers supported by “Associazione Italiana per la Ricerca sul Cancro” (AIRC).
Angiogenesis
Cross-Reference ▶ Biosensors ▶ Cell Adhesion ▶ Intravital Microscopy Analysis ▶ Nanomedicine ▶ Nanotechnology in Cardiovascular Diseases ▶ Synthetic Biology
References 1. Carmeliet, P.: Angiogenesis in health and disease. Nat. Med. 9, 653–660 (2003) 2. Folkman, J.: Angiogenesis: an organizing principle for drug discovery? Nat. Rev. Drug Discov. 6, 273–286 (2007) 3. Ferrara, N., Kerbel, R.S.: Angiogenesis as a therapeutic target. Nature 438, 967–974 (2005) 4. Bergers, G., Hanahan, D.: Modes of resistance to anti-angiogenic therapy. Nat. Rev. Cancer 8, 592–603 (2008) 5. Jain, R.K.: Lessons from multidisciplinary translational trials on anti-angiogenic therapy of cancer. Nat. Rev. Cancer 8, 309–316 (2008) 6. Aird, W.C.: Phenotypic heterogeneity of the endothelium: II. Representative vascular beds. Circ. Res. 100, 174–190 (2007) 7. Aird, W.C.: Phenotypic heterogeneity of the endothelium: I. Structure, function, and mechanisms. Circ. Res. 100, 158–173 (2007) 8. Fukumura, D., Jain, R.K.: Tumor microvasculature and microenvironment: targets for anti-angiogenesis and normalization. Microvasc. Res. 74, 72–84 (2007) 9. Jain, R.K.: Molecular regulation of vessel maturation. Nat. Med. 9, 685–693 (2003) 10. Serini, G., Napione, L., Arese, M., Bussolino, F.: Besides adhesion: new perspectives of integrin functions in angiogenesis. Cardiovasc. Res. 78, 213–222 (2008) 11. Roca, C., Adams, R.H.: Regulation of vascular morphogenesis by Notch signaling. Genes Dev. 21, 2511–2524 (2007) 12. Coultas, L., Chawengsaksophak, K., Rossant, J.: Endothelial cells and VEGF in vascular development. Nature 438, 937–945 (2005) 13. Hoffman, B.D., Grashoff, C., Schwartz, M.A.: Dynamic molecular processes mediate cellular mechanotransduction. Nature 475(7356), 316–23 (2011) 14. Aitsebaomo, J., Portbury, A.L., Schisler, J.C., Patterson, C.: Brothers and sisters: molecular insights into arterial-venous heterogeneity. Circ. Res. 103, 929–939 (2008) 15. Schaper, W.: Collateral circulation: past and present. Basic Res. Cardiol. 104, 5–21 (2009) 16. Coffelt, S.B., Lewis, C.E., Naldini, L., Brown, J.M., Ferrara, N., De Palma, M.: Elusive identities and overlapping phenotypes of proangiogenic myeloid cells in tumors. Am. J. Pathol. 176, 1564–1576 (2010)
Antifogging Properties in Mosquito Eyes
Animal Coloration
117
a
b
▶ Structural Color in Animals
▶ Biomimetics of Optical Nanostructures
Anodic Arc Deposition ▶ Physical Vapor Deposition
Antifogging Properties in Mosquito Eyes Xuefeng Gao Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences, Suzhou, PR China
Definition Fogging occurs when vapor condenses into droplets on condensation nuclei (e.g., dust) or exposed objects (e.g., glass and plastics) with diameters larger than 190 nm, namely, half of the shortest wavelength of visible light. These droplets can firmly stick to the surface of optical transparent materials so as to reduce the visibility by light scattering and reflection, which are undesired. Antifogging is to avoid condensed water droplets on a surface. Antifogging properties in mosquito eyes present a novel surface nonstick superhydrophobicity for micrometer-size water droplets [1].
Overview Surface wettability is a very common but important surface property closely related to the chemical
A
q
q
Animal Reflectors and Antireflectors
A
Antifogging Properties in Mosquito Eyes, Fig. 1 Contact angle and wetting states. (a) A hydrophilic surface with a water contact angle less than 90º. (b) A hydrophobic surface with a water contact angle higher than 90º
composition and microstructures of material surfaces. According to the affinity of water to material chemicals, the wettability may be simply divided into the hydrophobicity and hydrophilicity. The intrinsic contact angle y of a water droplet on a smooth solid surface is given by the classical Young’s equation: cos y ¼
gSV gSL gLV
(1)
where gSL, gSV, gLV are the interfacial free energies per unit area of the solid–liquid, solid–gas, and liquid–gas interfaces, respectively. The higher surface energy of solid materials means a lower contact angle with water, and vice versa. Generally, one defines that a surface with y < 90 is called a hydrophilic surface; conversely, a surface with y > 90 is called the hydrophobic (Fig. 1). Note that glass and PMMA are two kinds of most widely used optical transparent materials, both of which are hydrophilic with water contact angles of about 30 and 68 , respectively. Water vapor easily condenses in the form of tiny droplets on any flat surface made of hydrophobic and hydrophilic materials once its temperature is lower than the dew point. Strictly speaking, the water condensed on exposed objects such as plastic and glass should be called dew although the formation of fog and dew virtually follows the same thermodynamic principles. However, for material sciences and engineering, the measures of preventing the moisture condensation on optical transparent materials are customarily called antifogging while those aiming at other opaque materials such as metals and paints are called antidewing. It is known that the ordinary wettability makes the condensed water droplets firmly stick to the surfaces once contacting, which generates light scattering and reflection so as to greatly blur the visual field. It often occurs that the eyeglasses fog up when one comes inside on a cold day. This is because the eyeglasses have been
A
118
cooled outside down to a temperature that is below the dew point and the moisture in the inside air will come in contact with the cooled eyeglasses and condense into tiny droplets. Similar phenomena can frequently occur on the aluminum heat exchange fins in air conditioners, resulting in higher energy consumption and mildew wind. Besides, other side effects caused by moisture condensation are often neglected, for example, the corrosion of metal, the swelling and peeling of paints, and the attachment of dust particles. Up to now, most antifogging coatings reported are highly hydrophilic, aiming at protecting the optical transparency. A known example is the use of photocatalytic TiO2 nanoparticle coatings that become superhydrophilic under UV irradiation [2]. Besides, a three-dimensional capillary effect was adopted to achieve superhydrophilic properties by constructing porous nanostructures from layer-by-layer assembled nanoparticles [3]. The key to these two wet-style antifogging strategies is to make these condensed droplets rapidly spread into a uniform thin film, which can avoid the light scattering and reflection. However, many other problems caused by adsorption of moisture are hard to solve by the superhydrophilic approach due to its inherently wet nature. Thus, a drystyle antifogging technique that can prevent the condensed drops from sticking to the material surface is highly desired, which may be developed into multifunctional coatings such as antifogging, antirust, and antifouling.
Basic Principles Fogging of material surfaces generally forms via the impact of fog drops suspending in air or the direct condensation of moisture. The antifogging property can be achieved by changing the water affinity of the material surfaces toward the two extremities, as shown in Fig. 2. One is the superhydrophilic approach, with a water contact angle close to 0 [2, 3]. The other is the nonstick superhydrophobic approach, with a contact angle close to 180 and extremely low adhesion for micrometer-size water droplets [1]. The first strategy is now well understood. Thus, we mainly intend to introduce the possibility of the superhydrophobic strategy in this text. With the advance of nanofabrication techniques, one can easily construct micro- and nanostructures on
Antifogging Properties in Mosquito Eyes
b
a
q
q
Antifogging Properties in Mosquito Eyes, Fig. 2 (a) A superhydrophilic surface with a water contact angle close to zero. (b) A superhydrophobic surface with a water contact angle larger than 150
the surfaces of different materials, which provide the possibility to study and modulate surface wettability toward the nonstick superhydrophobicity. The real material surfaces are actually not ideally smooth at atomic and molecular scale, which means that the classic Young’s equation is not suitable in analyzing their water contact angles. Thus, Wenzel [4] proposed a theoretical model describing the apparent contact angle yW on a rough surface and modified Young’s equation as follows: cos yW ¼
r ðgSV gSL Þ ¼ r cos y gLV
(2)
where r is a roughness factor, defined as the ratio of the actual area of a rough surface to the geometric projected area. Apparently, this factor is always larger than unity and the hydrophobicity of a rough surface can be augmented by the increase of the solid–liquid contact area. However, such mechanism brings about very strong water adhesion, as shown in Fig. 3a. Thus, Wenzel model may be effective in the case of superhydrophobic surfaces with microscopic large interspaces or hollows enabling the water penetration but inapplicable to the development of nonstick and antifogging materials. Air is an effective hydrophobic medium with a water contact angle 180 [5]. As the air phase can be stably trapped into the microscopic hollows, a rough hydrophobic surface may be considered a composite surface composed of air and solid. Accordingly, the contact angle yC of a water droplet on the composite surface can be given by Cassie and Baxter’s equation [5] as follows: cos yC ¼ f cos y þ ð1 f Þ cos 180 ¼ f ðcos y þ 1Þ 1
(3)
Antifogging Properties in Mosquito Eyes
a
119
A
b Liquid
Liquid Gas
A
Gas
Antifogging Properties in Mosquito Eyes, Fig. 3 Two types of superhydrophobic states. (a) Wenzel’s model with water trapping into the intervals of structures on surface. (b) Cassie’s model without any water trapping
where f and (1f ) represent the area fraction of solid and trapped air, respectively; y is the intrinsic contact angle of solid materials. Clearly, the hydrophobicity of a rough surface can be amplified by the decrease of the solid–liquid contact area (Fig. 3b). It has been reported that the surface adhesion of a superhydrophobic surface is governed by van der Waals attraction at the solid–liquid interfaces and the capillary force caused by the curving of the liquid–air interfaces [6]. Thus, it is theoretically possible that the nonstick superhydrophobic property can be obtained by fine surface nanoengineering for the antifogging purpose.
Key Research Findings Lotus leaves are a sort of classic examples of natural nonstick superhydrophobic with a water contact angle >160 and a sliding angle > > > = dh < x2 1
þ 1 2 1 > dt P > B > ; :a1 1 b1 exp l2n nt > n¼1
(11) where h i 2 ðfÞ4 4exp f3 t i a1 ¼ h 2 ðfÞ4 6exp f3 t and f ¼ ln B. A similar approach can be followed as explained in the previous section and further the governing equation for capillary transport can be derived with the transient velocity profile provided in Eq. 11. Moreover, the difference in the penetration depth with both approaches, i.e., with the steady state and transient velocity profile, under different operating conditions can be compared. Figure 2 shows the difference in the
penetration depth in such cases where the difference in the penetration depth is more at the beginning of the filling process, as shown in the inset of Fig. 2. This difference in the penetration depth decreases as the flow progresses along the microchannel where the flow becomes a fully developed flow. This can also be explained with the help of boundary layer theory which is the effect of fluid viscosity. The boundary layer thickness increases as the viscosity of fluid increases because of the retardation of flow due to increase in the viscosity, whereas in the case of the fluid density, the effect is opposite to viscosity. Therefore, the difference in penetration depth with the high density fluid (Bo ¼ 0.01) is higher than the difference with the high viscous fluid. It is evident from the analysis that the transience effect in the analysis has a significant impact on the filling process prediction, particularly at the beginning of the filling process. At microscale, such difference needs to be accounted prior to the design. As discussed earlier, the pressure force at the entrance of the microchannel is determined with the help of the pressure field at the microchannel entrance. Several researchers [3–8] have adopted the pressure field expression with an equivalent radius assumption. Levin et al. [6] developed an entrance pressure field expression for circular capillary, assuming
Capillary Flow
379
Capillary Flow, Fig. 3 The fluid volume from infinite reservoir considered as control volume for pressure field expression analysis in the case of rectangular microchannel. The arrow shows the direction of the fluid flow from the reservoir into the microchannel [6]
2B
l inlet plane
Microchanne
C
z=0 lc
Oc
Os rs
rc
C
a hemispherical control volume as a separate control volume at the entrance which is responsible for a sink flow at the entrance of capillary and the pressure field. Moreover, a similar expression for rectangular capillaries is extended with an equivalent radius assumption. In such cases, the radius of circular capillary is replaced by the equivalent radius of projected area at the entrance of the channel. This is not a realistic representation for noncircular capillaries particularly for high aspect ratio microchannels where it is not appropriate to consider the hemispherical control volume for the sink flow or pressure field at the microchannel entrance. In the case of such geometries, the control volume needs to be considered as a combination of semicylinder and hemisphere as shown in Fig. 3. The detailed derivation of the pressure field expression with this control volume can be seen in [6] which is: (
4g þ 3ð1 gÞ pð0; tÞ ¼ patm rB 24 (
)
) 2 1 2 6 2 R1 d h þ þ þ 1 ln p 2p p2 10 p B dt2
4ð1 gÞ 6 4g þ 3ð1 gÞ þr p2 5 6
) 2 ð2 gÞ ð1 gÞ dh 4m 2 2p p dt B
ð1 gÞ dh ð2 gÞ þ p dt
(12)
where R1 represents the radial distance far away from the control volume in the reservoir, where the sink action, i.e., entrance pressure force, disappears. One can re-derive the governing Eq. 4, using pressure field
expression presented in the Eq. 12, and determine the effect of such a pressure field on the analysis. Figure 4 shows the comparison of variations in the penetration depth with recently proposed pressure and with equivalent radius field expressions. The approximated pressure field overpredicts the penetration depth. The difference in the penetration depth with the proposed pressure field is significant, which shows that it is important to consider the proposed pressure field for a rectangular microchannel rather than an approximated pressure field. The transport with a capillary action is the balance among surface, viscous, and other body forces which retard the flow as it progresses. Hence, the capillary flow always attains a steady state which is generally termed as an equilibrium penetration depth in the literature. If the length of the channel is longer than that of the equilibrium penetration depth, then flow front cannot reach the outlet and, therefore, the assistance to the capillary flow is attempted in such cases. Passive or nonmechanical pumping approaches combined with the capillary flow serve this enhancement. The scaling analysis suggests that the gravity force is less dominant at microscale [12], but several researchers have demonstrated that gravity can be used as an assistance to the capillary flow [13–15]. Generally, the capillary flow analysis is performed with an assumption of infinite reservoir. Hence, the reservoir effect and the gravitational force from the reservoir are generally neglected in the analysis. To accommodate the entrance effect of finite size reservoir at the inlet of the microchannel in the theoretical modeling, the entrance pressure field is developed for the arrangements shown in Fig. 5. The rectangular microchannel with rectangular reservoir on the top of the microchannel is considered and the pressure field with the gravity and reservoir effect is developed in the flow [16]. Moreover, this pressure field
C
380
Capillary Flow
12 Bo = 0.0076; Oh = 0.0075
2B
2W 2
2
Reservoir
8 z 6 y
H x
4 2
h
Capillary flow front
Pressure field with equivalent radius Proposed pressure filed
0 0
10
20
30
40
50
Time(t*)
Capillary Flow, Fig. 4 The comparison of variations in the penetration depth with equivalent radius and recently proposed pressure field expressions. Figure 4 shows the comparison of penetration depth for g ¼ 0:9 with the corresponding difference in the penetration depth [6]
is used to obtain the governing equation for capillary flow under the influence of gravity head from the reservoir. The reservoir with three different levels of fluid in the reservoirs (H*), namely, 10, 50, and 100, is considered for the analysis. Figure 6 shows the variations in the penetration depth (h*) with different operating conditions. This analysis represents the interplay between the surface tension force and gravity head from the reservoir. The capillary flow takes place in the channel which remains same for all three cases, whereas the level of the fluid from the reservoir increases from case I to case III. Thus, for three cases, the capillary effect is the same but the gravity head is different. At the beginning of the transport, the fluid from the reservoir offers less inertia to the fluid transport and the capillary force dominates over the gravity from the reservoir. Therefore, at the beginning of the transport, the penetration depth with a lower reservoir fluid level (H* ¼ 10.0) is higher than the other two penetration depths as shown in the inset I. Similarly, the penetration depth with the highest reservoir fluid level (H* ¼ 100.0) has the lowest penetration depth as compared to others. Moreover, as the fluid progresses in the microchannel, the momentum from the reservoir fluid assists the capillary flow and the gravitational force,
2B
Capillary
Penetration depth(h*)
10
2W1
1
Capillary Flow, Fig. 5 Schematic of a gravity-assisted capillary flow in a vertically oriented capillary of width 2B1 and depth 2W1. The additional gravitational head from the fluid in a finite reservoir of size (2B2 2W2) is assisting the capillary flow [16]
due to which the fluid from the reservoir becomes dominant over the capillary force within the microchannel. This results in transcendence among the penetration depths with a different gravity head. The penetration depth with the highest gravity head (H* ¼ 100) surpasses the penetration depth with gravity head H* ¼ 50 and H ¼ 10* in inset I and II two, respectively. This can be attributed to as an interplay between the surface tension force, i.e., the capillarity and gravitational force from reservoir. In the case of microfluidic applications, the sizes of reservoir and microchannel are comparable to each other. Hence one cannot neglect the effect of the reservoir in such cases, particularly if it is surface tension–driven pumping. Further, one can assist the capillary flow with an appropriate arrangement of reservoir. There are always certain limitations to the autonomous pumping which make them inadequate in long microchannels. Hence, it is important to enhance the pumping ability by other means. Further enhancement in the capillary flow can be achieved by coupling the capillary flow with the electroosmotic flow which is one of the electrokinetic pumping mechanisms. In most of the cases, the inner wall of a microchannel always has surface charges due to different mechanisms like ionization, dissociation of ions, isomorphic substitution, etc., [17]. These surface charges
Capillary Flow 120
90 Flow front penetration (h*)
Capillary Flow, Fig. 6 Transient response of a flow front transport for different gravitational heads in the reservoir with Bo ¼ 0.0055, Oh ¼ 0.0084, B1/W1 ¼ 0.05 B2/W2 ¼ 0.2. I. Flow front penetration rate for H* ¼ 100 surpasses the penetration rate for H* ¼ 50. II: Flow front penetration rate for H* ¼ 100 surpasses the penetration rate for H* ¼ 10. III: Flow front penetration rate for H ¼ 50 surpasses the penetration rate for H* ¼ 10 [16]
C
381
H* = 10.0 H* = 50.0
74
H* = 100.0
72
Bo = 0.0055 Oh = 0.0084;
B1/W1 = 0.05; B2/W2 = 0.2;
70
C
68
60
III
66 12
44
13
14
15
16
60 40 56
36
30 32
52
I
28 2.5
2.0
0 0
14
distribute ions of the electrolytes in a specific pattern when brought into contact with an electrolyte which is generally termed as the formation of electrical double layer (EDL). After applying the electric field across the channel, the movement of the ions takes place, which results in the movement of the fluid due to an electric field [18]. The electrolyte solution is transported with the capillary action; one can further assist the capillary flow. An additional body force due to electroosmotism is added to Eq. 2, which accommodates the additional effect of electroosmosis. Further one can analyze the interplay between the capillarity and electroosmotisms as presented in the recent studies [19]. Figure 7 shows the variation in the penetration depth of the capillary flow under the influence of electroosmotism. Through a nondimensional analysis, a new nondimensional number is proposed, i.e., Eo which represents the ratio between the surface tension force and electroosmotic force. The direction of the electroosmotic flow can be reversed by changing the electric field direction. Hence, negative and positive Eo numbers are observed in the analysis. The negative Eo numbers represent the change in the direction of the electric field as compared to positive Eo numbers. The pure capillary flow can be seen as Eo ¼ 0.
3.0
3.5
28
II 48 4.0 4
Time(t*)
5
42
6
7
8
56
9
10
70
The variation in the penetration depth under three different operating conditions is shown in Fig. 7. As observed in the pure capillary case (Eo ¼ 0), the penetration depth attains the equilibrium penetration depth, whereas in the case of Eo numbers, the equilibrium penetration depth increases with increment in the magnitude of Eo numbers. This represents that in the case of electroosmotic flow with Eo number, the capillary flow is assisted by electroosmotism. In this analysis, the nondimensional length of the microchannel (L*) is considered as 300 and with –Eo ¼ 0.01 the entire filling of the microchannel is observed. In the case of positive Eo numbers, it is observed that the electroosmotism acts in a opposite direction of the capillary flow. Hence, it retards the flow and this can be observed by the decrement in the equilibrium penetration depth with the increment in the positive Eo numbers. For the enhancement in the capillary flow, a transport with additional gravity head and electroosmotic forces are considered. A generalized theoretical modeling for a gravity-assisted capillary flow with reservoir effects and electroosmotically assisted capillary flow is reported in brief. It is observed that even though the scaling among forces suggests that the
382 300 Oh = 0.0070
E0 = −0.005
Bo = 0.0075
E0 = −0.001
E0 = −0.01
γ = 0.006
250
ε* = 0.07 L* = 300
200
E0 = 0.001
E0 = 0.0
300
Bκ = 4.0 E0 = 0.005 E0 = 0.01
150
100
50
250 Flow front penetration (h*)
Capillary Flow, Fig. 7 Variation in the penetration depth for vertically oriented channel with water as electrolyte, where B ¼ 100 mm, W ¼ 400 mm, L ¼ 75 mm, z ¼ 75 mV and constant contact angle is 27 . The inset shows the variation in the electric field within the electrolyte as the flow front progresses under different applied voltages [19]
Capillary Flow
Penetration depth (h*)
C
E0 = −0.5
E0 = −0.1 E0 = −0.05
200
150
100
50
0 0
50
100 150 Time(t*)
0 0
200
gravitation force is negligible at microscale, the reported analysis infers that with a finite reservoir, an added advantage due to gravity can be a useful tool to transport the fluid at microscale. This added force for the capillary transport can be utilized without any additional burden in the design of the LOC device. The electroosmotically assisted capillary flow model suggests that in a combined flow the electrokinetic parameters have an important influence on the capillary flow. Such electrokinetic flow approaches can be coupled to enhance the capillary flow transport in the microchannel. The wetting properties of the fluid decide the capability of pumping with a capillary flow. Therefore, it is important to know the precise magnitude of wetting properties like contact angle and surface tension of the working fluid. The microfluidics has become a promising option for biomedical application and inclusion of biomolecules is an unavoidable part in such applications. In most cases, the biomolecules are attached with the microbeads and transported to the desired locations. It is evident from the experimental analysis that the inclusion of microbeads changes the wetting behavior drastically [20]. Therefore, it is necessary to consider the effect of microbeads in
400
Time(t*)
600
800
200
250
1000
the fluid for the analysis. This can be done by considering the following expressions for surface tension and contact angle: density and viscosity which are functions of volume fraction of microbeads. Such correlations of the surface tension and contact angle are provided in [20]. Such expressions for the variation in the contact angle and surface tension with the volume fraction can be readily used in modeling transport processes of microbead suspensions in micro-capillaries, used in the microfluidic devices. In passive pumping, particularly with the capillary flow, different aspects due to microscale effects like aspect ratio–dependent velocity profile, contact angle at four walls, fluid-air interface dynamics in the case of suspension flow, etc., need to be investigated in detail. Theoretically the concept of electroosmotically assisted capillary flow has been presented but the experimental demonstration of such phenomena is also an interesting area of research. Moreover, wetting of biomolecule suspensions under transient effects instead of the steady state is also needed to be studied. The experimental study of the flow behind the front and at the entrance of the microchannels is also an interesting study to perform.
Capillary Origami
Cross-References ▶ AC Electroosmosis: Basics and Lab-on-a-Chip Applications ▶ Electrowetting ▶ Micro/Nano Flow Characterization Techniques ▶ Micropumps ▶ Surface Tension Effects of Nanostructures ▶ Wetting Transitions
References 1. Saha, A., Mitra, S.K.: Numerical study of capillary flow in microchannels with alternate hydrophilic-hydrophobic bottom wall. J. Fluid Eng. Trans. ASME 131, 061202 (2009) 2. Saha, A., Mitra, S.: Effect of dynamic contact angle in a volume of fluid (VOF) model for a microfluidic capillary flow. J. Colloid Interface Sci. 339, 461–480 (2009) 3. Xiao, Y., Yang, F., Pitchumani, R.: A generalized flow analysis of capillary flows in channels. J. Colloid Interface Sci. 298, 880–888 (2006) 4. Washburn, E.: The dynamics of capillary flow. Phys. Rev. 17, 273 (1921) 5. Chakraborty, S.: Electroosmotically driven capillary transport of typical non-Newtonian biofluid in rectangular microchannels. Anal. Chim. Acta 605, 175–184 (2007) 6. Waghmare, P.R., Mitra, S.K.: A comprehensive theoretical model of capillary transport in rectangular microchannels. Microfluid. Nanofluid. (2011). doi:10.1007/s10404-0110848-8 7. Levin, S., Reed, P., Watson, J.: A theory of the rate of rise a liquid in a capillary. In: Kerker, M. (ed.) Colloid and Interface Science, p. 403. Academic, New York (1976) 8. Marwadi, A., Xiao, Y., Pitchumani, R.: Theoretical analysis of capillary-driven nanoparticulate slurry flow during a micromold filling process. Int. J. Multiph. Flow 34, 227 (2008) 9. Dreyer, M., Delgado, A., Rath, H.: Fluid motion in capillary vanes under reduced gravity. Microgravity Sci. Technol. 4, 203 (1993) 10. Bhattacharya, S., Gurung, D.: Derivation of governing equation describing time-dependent penetration length in channel flows driven by non-mechanical forces. Anal. Chim. Acta 666, 51–54 (2010) 11. Keh, H., Tseng, H.: Transient electrokinetic flow in fine capillaries. J. Colloid Interface Sci. 242, 450 (2001) 12. Nguyen, N., Werely, S.: Fundamentals and Applications of Microfluidics. Artech House, New York (2003) 13. Yamada, H., Yoshida, Y., Terada, N., Hagihara, T., Teasawa, A.: Fabrication of gravity-driven microfluidic device. Rev. Sci. Instrum. 79, 124301 (2008) 14. Jong, W.R., Kuo, T.H., Ho, S.W., Chiu, H.H., Peng, S.H.: Flows in rectangular microchannels driven by capillary force and gravity. Int. Commun. Heat Mass Transf. 34, 186–196 (2007) 15. Kung, C., Chui, C., Chen, C., Chang, C., Chu, C.: Blood flow driven by surface tension in a microchannel. Microfluid Nanofluid 6, 693 (2009)
383
C
16. Waghmare, P.R., Mitra, S.K.: Finite reservoir effect on capillary flow of microbead suspension in rectangular microchannels. J. Colloid Interface Sci. 351(2), 561–569 (2010) 17. Hunter, R.: Zeat Potential in Colloid Science, Principle and Applications, Principle and Applications. Academic, London (1981) 18. Israelachvili, J.N.: Intermolecular and Surface Forces. Academic, London (1998) 19. Waghmare, P.R., Mitra, S.K.: Modeling of combined electroosmotic and capillary flow in microchannels. Anal. Chim. Acta 663, 117–126 (2010) 20. Waghmare, P.R., Mitra, S.K.: Contact angle hysteresis of microbead suspensions. Langmuir 26, 17082–17089 (2010)
Capillary Origami Supone Manakasettharn, J. Ashley Taylor and Tom N. Krupenkin Department of Mechanical Engineering, The University of Wisconsin-Madison, Madison, WI, USA
Synonyms Capillarity induced folding; Elasto-capillary folding; Surface tension–powered self-assembly
Definition Capillary origami is folding of an elastic planar structure into a three-dimensional (3D) structure by capillary action between a liquid droplet/bubble and a structure surface.
Why Capillary Origami? The fabrication of 3D structures is one of the major challenges for micro- and nano-fabrication. Folding of an elastic planar structure after patterning and release is one technique to fabricate a 3D structure using self-assembly. The term origami is taken from the Japanese art of paper folding; while the actuation of the folding is accomplished by using capillary forces of a fluid droplet, hence the technique has been termed capillary origami. The combination of the folding process with capillary forces has resulted in a new technique for micro- and nano-fabrication.
C
384
History The term capillary origami was first introduced in 2007 by Charlotte Py et al. to describe the folding of a polydimethylsiloxane (PDMS) sheet into a 3D structure by using capillary forces created by a water droplet [1]. As early as 1993, Syms and Yeatman demonstrated that 3D structures could be fabricated by folding surfaces using capillary forces produced by molten solder [2]. Later Richard R. A. Syms introduced the term surface tension–powered self-assembly to describe the technique [3, 4]. Both of these techniques are quite similar in that 3D structures can be produced by folding elastic thin films. Both use capillary forces for self-assembly. In the first example, various liquids such as water are used, while for the second study molten metals such as solder were used, which then solidified fixing the 3D microstructures.
Principles At the macroscale level, the influence of capillary forces is negligible compared to other forces such as gravity, electrostatic, or magnetic. Because capillary forces scale linearly with the characteristic size of the system, at sub-millimeter dimensions capillary forces begin to dominate since the majority of other forces decrease much more rapidly than the first power of the length. For example, a human cannot walk on water because capillary forces produced at the water surface are much smaller than the gravitational force acting on a human, which scales as the cube of the length. On the other hand, the much smaller water strider can easily walk on water because capillary forces are large enough to balance the gravitational force produced by the water strider. For capillary origami, capillary forces need to be large enough to counteract the weight of the liquid droplet and the structural forces of the planar layer. In terms of energy, for capillary origami one needs to consider the interplay of three different energies: capillary energy, bending energy, and gravitational potential energy. For a two-dimensional (2D) model, the capillary energy per unit length of the interface (2D analog of the surface energy) is defined as Ec ¼ L0 g, where L0 is the length of the interfacial surface of the fluid and g is the surface tension [5]. The bending LB energy per unit length is approximately Eb ¼ 2R 2 ,
Capillary Origami 1,000
No folding
100
Lcrit / Lec
C
Folding 10
No folding 1 0.01
0.1
1
10
Lec / Lc
Capillary Origami, Fig. 1 Folding criteria plotted from (1) assuming complete circular folding and neglecting the effect of gravity
where L is the length of the structure and R is the radius Eh3 of curvature [6]. B ¼ 12ð1v 2 Þ is the bending rigidity of the structure, where E is Young’s modulus, h is thickness of the layer, and v is Poisson’s ratio. If one only considers the mass of the fluid, assuming that it is much larger than the mass of the structure, then the gravitational potential energy per unit length is Eg ¼ rSgz, where r is the density, S is the surface area, g is the constant of gravity, and z is the height of the center of mass. By neglecting the effect of gravity and assuming complete circular folding, de Langre et al. [7] derived simplified criteria for folding considering the interplay between capillary and bending energies, which are expressed as pffiffiffi pffiffiffi Lc Lcrit 2p < < 2 2p Lec Lec
(1)
where Lcrit is the critical qffiffiffiffi length of a structure for g folding to occur, L ¼ c qffiffiffi rg is the capillary length [5], B and Lec ¼ g is the elasto-capillary length [8]. The simplified criteria for folding derived from (1) can be plotted as shown in Fig. 1. To fold a structure pffiffiffiffiffiffi Brg requires LLecc < 2 or Lc > L2ec or g > 2 indicating that the capillary length must be larger than half of the elasto-capillary length so that the capillary effect can overcome bending rigidity of the structure. pffiffiffi The other requirement for folding is LLcrit > 2p ffi 4:44 or ec Lcrit > 4:44Lec confirming that the length of the structure should also be long enough for a liquid droplet to
Capillary Origami
C
385
1 mm
1 mm
500 μm
Capillary Origami, Fig. 2 Capillary origami 3D structures of a pyramid, a cube, and a quasi-sphere obtained by folding triangle-, cross-, and flower-shaped PDMS sheets, respectively,
actuated with a water droplet (Reprinted with permission from [10]. Copyright 2007, American Institute of Physics)
Capillary Origami, Fig. 3 Capillary origami 3D structures formed from triangle- and flower-shaped templates using soap bubbles (scale bar: 2 cm) (Images reprinted from [11] with permission)
wet the surface to produce sufficient capillary forces to fold the structure. For the 3D structures the folding criteria become more complex. In particular in 3D the critical length also depends on the shape of the initial template such that Lcrit ffi 7Lec for squares and Lcrit ffi 12Lec for triangles [9]. Figure 2 shows examples of capillary origami structures of a pyramid, a cube, and a quasi-sphere obtained from folding triangle-, cross-, and flower-shaped PDMS sheets, respectively [10]. Besides using liquid droplets, capillary origami structures can be constructed by using soap bubbles as shown in Fig. 3. The weight of a soap bubble is much less than that of a liquid droplet especially for
large droplets capable of covering centimeter size structures when gravitational forces become significant. A soap bubble was shown to fold a centimetersize elastic structure, which cannot be accomplished using a liquid droplet [11]. Petals of a flower also can be folded into a structure similar to capillary origami when submerged in water as shown in Fig. 4. The folding of the flower in water is accomplished by the interplay of elastic, capillary, and hydrostatic forces. During submersion, hydrostatic pressure pushes against the back of petals, and surface tension prevents water from penetrating through the spacing between petals resulting in trapping an air
C
C
386
Capillary Origami
Capillary Origami, Fig. 4 The folding of an artificial flower when submerged in water (Reprinted with permission from [12]. Copyright 2009, American Institute of Physics)
a
b
50 μm
c
d
Capillary Origami, Fig. 5 (a) Schematics of initial templates. (b)–(d) SEM images of 3D microstructures after folding (scale bar: 50 mm) (Reprinted with permission from [13]. Copyright 2010, American Institute of Physics)
bubble inside a flower. The inside of the folded flower remains dry protected by the air bubble [12].
Applications Capillary origami has been used to fabricate a number of 3D microstructures. Figure 5 illustrates the self-assembly of structures with various geometries. The initial planar templates are shown in Fig. 5(a), and the folded final 3D microstructures are shown in Fig. 5(b)–(d). The initial planar templates with lengths ranging from 50 to100 mm and a thickness of 1 mm were fabricated from silicon nitride thin films deposited and patterned by using standard micromachining processing typically used for integrated circuit and MEMS fabrication. Water droplets then were deposited on the templates to fold 3D microstructures [13]. Figure 6 shows another example of microfabrication of a quasi-spherical silicon solar cell based
on capillary origami. After fabrication by conventional micromachining processing, the initial flower-shaped silicon template was folded into a sphere using a water droplet. Unlike, conventional flat solar cells, this spherical solar cell enhanced light trapping and served as a passive tracking optical device, absorbing light from a wide range of incident angles [14]. Structures formed by capillary origami also can be actuated by using electrostatic fields to reversibly fold and unfold then. For this application we need to take into account the interplay of capillary, elastic, and electrostatic forces. As shown in Fig. 7, an electric field was applied between the droplet and the substrate. When the voltage was increased, the electrostatic force increased eventually overcoming capillary forces resulting in unfolding of the PDMS sheet. When the voltage was decreased below a certain threshold, the electrostatic force was no longer strong enough to prevent capillary forces from again folding the elastic sheet [15].
Capillary Origami Cr/Au
387
C
+ n+ p
Etch undercuts
Fold into a sphere
SO
I wa
ter
C
Ag wire
Capillary Origami, Fig. 6 Three schematics from left to right showing steps to fabricate a spherical-shaped silicon solar cell. The image at the far right shows the final spherical-shaped silicon solar cell (Images reprinted from [14] with permission)
Electrode Droplet
0V
Ui
+ +
++ + +
+ + +
700 V
200 V
Elastic sheet E
Isolating layer Counter electrode
Capillary Origami, Fig. 7 Capillary origami controlled by an electric field. The schematic of the experimental setup is shown at the far left and three images to the right show the results of
increasing voltage from 0 V to 700 V and decreasing voltage from 700 V to 200 V. Images from [15] – reproduced by permission of The Royal Society of Chemistry
Capillary origami is a simple and inexpensive method to fabricate 3D structures at the sub-millimeter scale. By using capillary forces, intricate and delicate 3D thin-film structures can easily be fabricated, which would be difficult to obtain by other means. More applications exploiting the advantages of capillary origami itself or in combination with electric fields can readily be envisioned. Ultimately one expects to see more commercial products based on this versatile technique.
3. Syms, R.R.A.: Surface tension powered self-assembly of 3-D micro-optomechanical structures. J. Microelectromech. Syst. 8, 448–455 (1999) 4. Syms, R.R.A., Yeatman, E.M., Bright, V.M., Whitesides, G.M.: Surface tension-powered self-assembly of microstructures – the state-of-the-art. J. Microelectromech. Syst. 12, 387–417 (2003) 5. Berthier, J.: Microdrops and digital microfluids. William Andrew Pub, New York (2008) 6. Timoshenko, S., Woinowsky-Krieger, S.: Theory of plates and shells, 2nd edn. McGraw-Hill, New York (1959) 7. de Langre, E., Baroud, C.N., Reverdy, P.: Energy criteria for elasto-capillary wrapping. J. Fluids Struct. 26, 205–217 (2010) 8. Bico, J., Roman, B., Moulin, L., Boudaoud, A.: Adhesion: elastocapillary coalescence in wet hair. Nature 432, 690 (2004) 9. Py, C., Reverdy, P., Doppler, L., Bico, J., Roman, B., Baroud, C.N.: Capillarity induced folding of elastic sheets. Eur. Phys. J. Spec. Top. 166, 67–71 (2009) 10. Py, C., Reverdy, P., Doppler, L., Bico, J., Roman, B., Baroud, C.: Capillary origami. Phys. Fluids 19, 091104 (2007) 11. Roman, J., Bico, J.: Elasto-capillarity: deforming an elastic structure with a liquid droplet. J. Phys. Condens. Matter 22, 493101 (2010) 12. Jung, S., Reis, P.M., James, J., Clanet, C., Bush, J.W.M.: Capillary origami in nature. Phys. Fluids 21, 091110 (2009) 13. van Honschoten, J.W., Berenschot, J.W., Ondarcuhu, T., Sanders, R.G.P., Sundaram, J., Elwenspoek, M., Tas, N.R.: Elastocapillary fabrication of three-dimensional microstructures. Appl. Phys. Lett. 97, 014103 (2010)
Cross-References ▶ Self-assembly ▶ Surface Tension Effects of Nanostructures
References 1. Py, C., Reverdy, P., Doppler, L., Bico, J., Roman, B., Baroud, C.N.: Capillary origami: spontaneous wrapping of a droplet with an elastic sheet. Phys. Rev. Lett. 98, 156103 (2007) 2. Syms, R.R.A., Yeatman, E.M.: Self-assembly of threedimensional microstructures using rotation by surface tension forces. Electron. Lett. 29, 662–664 (1993)
C
388
14. Guo, X., Li, H., Yeop Ahn, B., Duoss, E.B., Jimmy Hsia, K., Lewis, J.A., Nuzzo, R.G.: Two- and three-dimensional folding of thin film single-crystalline silicon for photovoltaic power applications. Proc. Natl Acad. Sci. U.S.A. 106, 20149–20154 (2009) 15. Pineirua, M., Bico, J., Roman, B.: Capillary origami controlled by an electric field. Soft Matter. 6, 4491–4496 (2010)
Carbon Nanotube Materials ▶ Computational Study of Nanomaterials: From Large-Scale Atomistic Simulations to Mesoscopic Modeling
Carbon Nanotube-Metal Contact Wenguang Zhu Department of Physics and Astronomy, The University of Tennessee, Knoxville, TN, USA
Synonyms Carbon nanotube-metal interface
Definition Carbon nanotube-metal contacts are widely present in many carbon nanotube-based nanodevices, and their electronic structures may significantly influence the operation and performance of carbon nanotube-based nanodevices.
Overview Carbon nanotubes (CNTs) are quasi-one-dimensional materials with remarkable mechanical and electronic properties promising a wide range of applications from field-effect transistors (FETs) and chemical sensors to photodetectors and electroluminescent light emitters. In most of these CNT-based nanodevices, metals are present as electrodes in contact with the CNTs. Many factors including the CNT-metal contact geometry,
Carbon Nanotube Materials
microscopic atomic details at the interface, and the resulting electronic structure can play a significant role in determining the functionality and performance of the devices. For instance, it has been demonstrated that an individual semiconducting CNT can operate either as a conventional FET or an unconventional Schottky barrier transistor, depending on the properties of the metal-CNT contact. In general, the electrical transport characteristics of the CNT-metal systems are sensitive to the choice of metal element as the electrode.
CNT-Metal Contact Geometry There are two types of interface geometries of CNTmetal contacts, i.e., end contact and side contact [1, 2]. The end-contact geometry refers to the cases where metals are merely in contact with the open ends of one-dimensional CNTs, as illustrated in Fig. 1a. This contact geometry can be naturally achieved in the catalytic CVD growth of CNTs, where CNTs sprout from catalytic metal particles with the CNT axis normal to the metal surface. Figure 1b shows a sample experimental image of an end contact between a single-wall CNT and two Co tips in an in situ electron microscopy setup. The side-contact geometry refers to the cases where metals are in contact with the sidewall of CNTs, as illustrated in Fig. 1c. This contact geometry occurs when a CNT lays on the surface of a flat metal substrate. In most of CNT-based nanodevices, such as CNT FETs, metal strips are deposited from above to cover the CNTs laying on the surface as to build electrodes, fully covering sections of the CNTs, as shown in Fig. 1d. Among these two contact geometries, the side-contact geometry is more technologically relevant to CNT-based nanodevices.
Bonding and Wetting Properties of Metals on CNTs In the end-contact geometry, metals form strong covalent bonds with carbon atoms at the open ends of CNTs [3]. The bonding energy can be as high as, e.g., 7.6 eV for a single bond at a CNT–Co contact, according to density functional calculations. Due to the strong covalent nature of the bonding, large mismatch-induced strains or high tensile strength can be built up at the interface.
Carbon Nanotube-Metal Contact
a
b
c
d
Carbon Nanotube-Metal Contact, Fig. 1 A schematic illustration of (a) an end contact and (c) a side contact between a CNT and a metal (From Palacios, J.J., Pe´rez-Jime´nez, A.J., Louis, E., SanFabia´n, E., Verge´s, J.A.: Phys. Rev. Lett. 90, 106801 (2003), Fig. 1). (b) A CNT forming end contacts with Co tips (From Rodr´iguez-Manzo, J.A. et al.: Small, 5, 2710–2715 (2009), Fig. 1). (d) A CNT forming side contacts on gold electrodes (From Anantram, M.P., Le´onard F.: Rep. Prog. Phys. 69, 507–561 (2006), Fig. 24)
389
C
In the side-contact geometry, metals and CNTs form much weaker bonds due to the nearly chemically inert side walls of CNTs [3]. Single-wall CNTs are built up of a cylindrically closed sheet of graphene, in which carbon atoms arranged in a honeycomb structure form very stable sp2-hybridized covalent bonds with the pz-orbitals of carbon extending normal to the sidewalls. The interaction between metals and CNTs in the side-contact geometry is determined by the hybridization between the carbon pz-orbitals and the unbonded orbitals of the metals. Alkali and simple metals have binding energy around 1.5 eV per atom. Some transition metal atoms with unpaired d electrons, such as Sc, Ti, Co, Ni, Pd, Pt, form strong bonds with a binding energy around 2.0 eV per atom, whereas the transition metals with fully occupied d orbitals such as Cu, Au, Ag, and Zn have relatively weak binding with a binding energy less than 1.0 eV per atom. On the other hand, the binding energy of metal on CNTs also depends on the radius of CNTs. In general, the larger is the radius, the weaker the binding energy. The wettability of metals on CNTs is critical to the electrical transport properties at CNT-metal contacts. In addition, CNTs can be used as templates to produce metallic nanowires with controllable radius by continuously coating the sidewalls of CNTs with metals. Experiments using different techniques such as electron beam evaporation, sputtering, and electrochemical approaches have achieved continuous coating of Ti and quasi-continuous coating of Ni and Pd on CNTs [3, 4]. Such metallic nanowires are ideal to be used as conducting interconnects in nanodevices. Metals such as Au, Al, Fe, Pb form isolated discrete clusters rather than a uniform coating layer on the surface of CNTs. Figure 2 shows sample TEM images of Ti, Ni, Pd, Au, and Fe coatings on CNTs [4]. The correlation between the wettability of these metals and their binding energies on CNTs is clear, i.e., metals with relatively strong binding energies with CNTs tend to form uniform coatings.
Electronic Structures of CNT-Metal Contacts The electronic structure of CNT-metal contacts has a significant impact on the operation and performance of CNT-based nanodevices [2]. Due to the one-dimensional nature of CNTs and their special
C
C
390
Carbon Nanotube-Metal Contact, Fig. 2 TEM images of (a) Ti, (b) Ni, (c) Pd, (d) Au, (e) Al, and (f) Fe coatings on carbon nanotubes
contact geometries, CNT-metal contacts exhibit some unusual features when compared to traditional planar contacts. For metallic CNTs, as in contact with metals, ohmic contacts are normally formed at the interface, where no interface potential barrier exists, and the contact resistance is primarily determined by the wettability of the
Carbon Nanotube-Metal Contact
metal and the local atomic bonding and orbital hybridization at the interface [2]. Palladium is found to be optimal as electrodes to make ohmic contacts with metallic CNTs. Semiconducting CNTs form either ohmic contacts or Schottky barriers at the interface with metals [2]. Figure 3 schematically illustrates the energy levels for an ohmic and a Schottky contact between a metal and a semiconducting CNT. A distinctive feature of CNTmetal contacts from traditional planar metal/semiconductor interfaces is that the height of the Schottky barrier formed at CNT-metal contacts strongly depends on the work function of the metal for a given semiconducting CNT [5, 6]. In general, at traditional planar metal/semiconductor interfaces, the Schottky barrier height shows very weak dependence on the metal work function due to the so-called Fermi-level pinning effect [7]. The strong dependence of the Schottky barrier on the metal work function in CNTmetal contacts is attributed to the reduced dimensionality of CNTs, which entirely changes the scaling of charge screening at the interface, making the depletion region decay rapidly in a direction normal to the interface and thus significantly weakening the Fermi-level pinning effect [5]. Experimental and theoretical work has shown that the interface Schottky barrier regions are much thinner in one dimension than those in three dimensions. In this case, charge carrier tunneling through the Schottky barriers becomes important. Because of the involvement of tunneling and thermionic emission in the carrier transport at the interfaces, the dependence of the on-current of CNT transistors on the Schottky barrier becomes very strong. Figure 4 shows experimental CNT-FET on-current and Schottky barrier height as a function of the CNT diameter for three different metal electrodes, Pd, Ti, and Al [6]. When the metal work functions are in the valence or conduction band of the semiconducting CNTs, ohmic contacts will likely be formed at the interface. Experimental measurements have shown that certain metals with high work functions, such as Pd, can produce nearly ohmic contacts with semiconducting CNTs. Ohmic contacts are more desirable in devices where contact resistance needs to be minimized. In addition to the metal work function, other factors, such as the contact geometry and the chemical bonding at the interface also play important roles in the transport properties of CNT-metal contacts.
Carbon Nanotube-Metal Interface
C
391
Carbon Nanotube-Metal Contact, Fig. 3 A schematic illustration of the energy levels for an ohmic and a Schottky contact between a metal and a semiconducting CNT
Ec
EF Ec
Ev Ev
metal
metal
Ohmic
Schottky
10−5 Pd contacts 10−6
0.0
Ti contacts AI contacts
10−8
0.4 10
10−9 Ion (A)
Carbon Nanotube-Metal Contact, Fig. 4 Experimental CNTFET on-current (left axis) and computed Schottky barrier height (right axis) as a function of the CNT diameter for three different metal electrodes, Pd, Ti, and Al
0.2
10−10
1
0.6
10−6
10
10−11
−5
2 +
10
3
−7
Schottky Barrier (eV)
Ion (A)
10
−7
0.8
−8
0.5
1.0
1.5 d (nm)
2.0
2.5
1.0
−12
10
0.6
Cross-References ▶ Carbon Nanotubes for Chip Interconnections ▶ Carbon-Nanotubes ▶ CMOS-CNT Integration
References 1. Banhart, F.: Interactions between metals and carbon nanotubes: at the interface between old and new materials. Nanoscale 1, 201–213 (2009) 2. Anantram, M.P., Le´onard, F.: Physics of carbon nanotube electronic devices. Rep. Prog. Phys. 69, 507–561 (2006) 3. Ciraci, S., Dag, S., Yildirim, T., G€ ulseren, O., Senger, R.T.: Functionalized carbon nanotubes and device applications. J. Phys. Condens. Matter 16, R901–R960 (2004)
0.8
1.0
1.2
1.4
Diameter (nm)
4. Zhang, Y., Franklin, N.W., Chen, R.J., Dai, H.J.: Metal coating on suspended carbon nanotubes and its implication to metal-tube interaction. Chem. Phys. Lett. 331, 35–41 (2000) 5. Le´onard, F., Tersoff, J.: Role of fermi-level pinning in nanotube schottky diodes. Phys. Rev. Lett. 84, 4693–4696 (2000) 6. Chen, Z.H., Appenzeller, J., Knoch, J., Lin, Y.-M., Avouris, P.: The role of metal nanotube contact in the performance of carbon nanotube field-effect transistors. Nano Lett. 5, 1497–1502 (2005) 7. Tung, R.T.: Recent advances in Schottky barrier concepts. Mater. Sci. Eng. R 35, 1–138 (2001)
Carbon Nanotube-Metal Interface ▶ Carbon Nanotube-Metal Contact
C
C
392
Carbon Nanotubes 10
Carbon Nanotubes
CPU Transistor Count
109
2x every 2 years
1 Microns
▶ Ecotoxicology of Carbon Nanotubes Toward Amphibian Larvae ▶ Computational Study of Nanomaterials: From LargeScale Atomistic Simulations to Mesoscopic Modeling
107
0.1 Feature Size 0.7x every 2 years
Carbon Nanotubes (CNTs)
0.01 1970
1980
1990
2000
65 nm 45 nm 32 nm
2010
105
103 2020
▶ Chemical Vapor Deposition (CVD) ▶ Physical Vapor Deposition
Carbon Nanotubes for Chip Interconnections, Fig. 1 Moore’s Law: Transistor count has doubled while feature size has decreased by 0.7X every 2 years (Figure reprinted with permission from Kuhn [1])
Carbon Nanotubes for Chip Interconnections
Motivation
Gilbert Daniel Nessim Chemistry department, Bar-Ilan Institute of Nanotechnology and Advanced Materials (BINA), Bar-Ilan University, Ramat Gan, Israel
Synonyms Carbon nanotubes for interconnects in integrated circuits; Carbon nanotubes for interconnects in microprocessors
Definition Chip interconnections electrically connect various devices in a microprocessor. Today’s established technology for interconnects is based on copper. However, it may be technically challenging to extend copper use to future interconnects in microprocessors with smaller lithographic dimensions due to materials properties limitations. Carbon nanotubes are currently investigated as a potential replacement for future integrated circuits (microprocessors). Although carbon nanotubes are a clear winner against copper in terms of materials properties, multiple fabrication challenges need to be overcome for carbon nanotubes to enter the semiconductor fab and replace copper for chip interconnections.
Following over 40 years of successful fulfillment of Moore’s law, stating that the number of transistors in a chip doubles every 2 years, we have already moved from microelectronics to nanoelectronics [1]. Although the “end of scaling” has been predicted many times in the past, enormous technical challenges, especially quantum mechanical issues and billiondollar lithography investments, are a serious threat to further miniaturization (Fig. 1). Today’s latest processors are manufactured using the 32-nm technology. To move toward the 22-nm node and beyond, issues such as lithographic limitations, leaking currents in ultra-thin dielectrics (only a few monolayers thick), insufficient power and thermal dissipation, and interconnect reliability must be resolved [1]. At the transistor level, the performance is negatively affected by increased off-state currents due to short channel effects, increased gate leakage due to tunneling through nanometer-thin dielectric layers, and increased overall gate capacitance due to decreasing gate pitch. Although quantum mechanical tunneling and leakage currents may eventually stop further scaling, efficient heat removal from a chip is currently the biggest obstacle. In this respect, the many kilometers of copper interconnects present in today’s chips are the main culprit for heat generation. For instance, in 2004, Magen et al. [2] showed that for a microprocessor fabricated with the 0.13-mm-node technology consisting of 77 million transistors, interconnects consumed more than 50% of the total dynamic power.
Carbon Nanotubes for Chip Interconnections
Given the increased length of interconnects, their reduced cross section, and the increased current densities circulating into the interconnects of our latest chips, the problem has been further exacerbated. Additionally, copper interconnects are a major contributor to the total resistance-capacitance (RC) delay of the chip, can fail by electromigration, and need a liner to avoid diffusion into the silicon. Bottom line, the interconnect issue is so serious that the International Semiconductor Roadmap [3] (ITRS, an expert team assessing the semiconductor industry’s future technology requirements for the next 15 years) indicates copper interconnects as a possible dealbreaker to further miniaturization for IC nodes beyond 22 nm. Many technology options are currently under investigation to replace copper for interconnects. Among them, we can mention other metals (mainly silicides), wireless, plasmonics, and optical interconnects. Most notably, there has been an intense research effort on new nanotechnology materials such as carbon nanotubes, which, at the theoretical level, could solve all the above technical issues suffered by copper. The plan of this section is to first introduce the reader to copper interconnects’ fabrication and limitations. Next, we will compare copper to carbon nanotubes (CNTs) and detail possible models for implementation. An important paragraph will focus on the state-of-the-art of CNT fabrication, prior to concluding on the outstanding issues and outlook for future CNT-based chip interconnections.
Background on Copper Interconnects and Dual-Damascene Process In 1997, IBM introduced the revolutionary “dualdamascene” process to fabricate copper interconnects and to replace aluminum interconnects, the industry standard at the time. Compared to aluminum, copper presents two major advantages: (1) 50% lower resistivity (Cu 1.75 mm cm versus Al 3.3 mm cm) and (2) higher current densities before failure by electromigration (up to 5 106 A/cm2) [4, 5]. Although, as a material, copper was a clear winner against aluminum, fabrication challenges delayed its introduction. Historically, we may be at a similar juncture with carbon nanotubes compared to copper as we were with copper compared to aluminum in 1997: in spite of their superior materials properties,
393
C
mainly fabrication issues are now preventing the introduction of nanotubes in the semiconductor industry to replace copper interconnects. Copper diffuses into silicon, generating mid-gap states that significantly lower the minority carrier lifetime and which lead to leakage in diodes and bipolar transistors. Copper also diffuses through SiO2 and low-k dielectrics, and therefore requires complete encapsulation in diffusion barriers. Since no dry etches were known for copper, IBM’s bold innovation of polishing using chemical mechanical polishing (CMP), after electroplating the copper, was significantly at odds with the technological processes at that time in semiconductor fabrication. The copper dual-damascene process consists of the following steps: • Develop a pattern for wires or vias by patterned etching of the dielectric. • Deposit a barrier layer (usually Ta) to prevent copper diffusion into silicon. • Deposit a copper seed layer. • Fill the vias with copper using electrodeposition. • Remove excess copper using CMP. • Repeat the process to lay the alternating layers of wires and vias which will form the complete wiring system of the chip (Fig. 2). Typical microprocessor design follows a “reverse scaling” metallization scheme with multiple layers of interconnects labeled as local, intermediate, and global interconnects, with increasing width. Very thin local interconnects locally connect gates and transistors within a functional block and are usually found in the lower two metal layers. The wider and taller intermediate interconnects have lower resistance and provide clock and signal within a functional block up to 4 mm. Global interconnects are found at the top metal layers and provide power to all functions in addition to connecting functional blocks through clock and signal. They are usually longer than 4 mm (up to half of the chip perimeter) and exhibit very low resistance to minimize RC delay and voltage drop. Below are a typical cross section of an I.C. chip and a possible implementation using CNTs (Figs. 3 and 4).
Limitations of Copper Interconnections Copper interconnects have efficiently scaled down to the current 32-nm-node microprocessors, although this
C
C
394
Carbon Nanotubes for Chip Interconnections, Fig. 2 Dual-damascene process of copper filling an interconnect via (Figure reprinted with permission from Jackson et al. [21])
Carbon Nanotubes for Chip Interconnections SiN 6. Treach and via etch
1. SiN deposition Metal 1 Contacts
Devices
2. Oxide dielectric deposition
7. Barrier/ seed deposition
8. Copper till
3. Via patterning
4. Partial via etch
9. Copper CMP and SiN cap layer
5. Treach Patterning
Passivation Dielectric Etch Stop Layer
Wire Via
Dielectric Capping Layer Copper Conductor with Barrier/Nucleation Layer
Global
Carbon Nanotubes for Chip Interconnections, Fig. 3 Typical cross sections of hierarchical scaling in current microprocessor (Figure reprinted with permission from the Semiconductor Industry Association [22])
Intermediate Metal 1
has required many technological advances to allow ever-shrinking copper cross sections to carry increasing currents without failure. However, we may be very close to smashing against a technical wall because of materials failure and related fabrication issues. Alternative materials or technologies would require many changes in semiconductor fabrication and
Pre-Metal Dielectric Tungsten Contact Plug
Metal 1 Pitch
massive investments; thus the large semiconductor companies are doing the impossible to extend copper application to future nodes. It is clear that only when up against an insurmountable technical wall will the semiconductor industry switch to a new technology. Electrical resistance is a major issue now that copper interconnect cross sections are comparable to
Carbon Nanotubes for Chip Interconnections
395
C
Carbon Nanotubes for Chip Interconnections, Table 1 Selected critical parameters for copper use as interconnects in future IC nodes (Data from the Semiconductor Industry Association [22]) Year of production (Estimated) MPU/ASIC metal 1 ½ pitch (nm) (contacted) Total interconnect length (m/cm2) – Metal 1 and 5 intermediate levels, active wiring only Barrier/cladding thickness (for Cu intermediate wiring) (nm) Interconnect RC delay [ps] for a 1-mm Cu intermediate wire, assumes no scattering and an effective r of 2.2 mO-cm
Carbon Nanotubes for Chip Interconnections, Fig. 4 Schematic view of possible implementation of carbon nanotube via interconnects in lieu of copper (Figure reprinted with permission from Awano et al. [7])
the mean free path of electrons in copper (40 nm in Cu at room temperature). Grain boundary and surface scattering are significant contributors to the increased resistance, especially now that we have reached nanoscale dimensions. At the microstructural level, the grain boundaries play an important role, hence, among other fabrication concerns, controlling the copper grain size during electrodeposition has allowed to limit the grain boundary scattering impact thus far. The steep rise in interconnect resistance for smaller IC nodes is a major source of RC delays and directly affects the chip reliability by increasing the risk of electromigration failure, a major issue for further downscaling. Electromigration is the transport of material caused by the gradual movement of the copper ions due to the momentum transfer between conducting electrons and diffusing metal atoms, which occurs for high current densities, which can create voids leading to open circuits. Given that downscaling leads to a reduction of the interconnects’ cross section, the problem is amplified at subsequently smaller nodes. To compound the issue, the need for a resistive diffusion barrier layer, also called a liner (usually Ta), to avoid copper diffusion into silicon, further reduces the available conductive copper cross section, thus increasing the risk of electromigration failure, especially as the operating temperature rises.
2010 2015 2020 45 25 14 2,222 4,000 7,143
3.3
1.9
1.1
1,132 3,128 9,206
In addition to the increased resistance and the electromigration failure risk, many other aspects of the dual-damascene process are becoming potential sources of failure as the node shrinks. Among the many integration concerns, we can mention materials issues such as interface adhesion between the different materials (copper, low-k dielectrics, etc.), liner effectiveness, metal voids, CMP interface defects, etc. Concurrently, there is a long list of process-related issues such as the need for etch/strip/ clean processes (to avoid damage to low-k dielectric materials), atomic layer deposition (ALD) processes to deposit liners, copper plating and CMP techniques, etc. A few interesting numerical estimates taken from the 2009 projections from ITRS, [3] provide the reader with the magnitude of the technical challenge to extend copper interconnect technology (Table 1). As already mentioned, many alternative technologies are currently being investigated for replacement of copper as interconnect material that would require significant chip redesigns and new fabrication technologies. Some examples include optical interconnects, radio frequency (RF) interconnects, plasmonics, and 3-D interconnects (probably still copper). The interested reader can find more details on these alternative technologies in the review paper from Havemann et al. [6] (now a little dated) or in the latest ITRS report on interconnects [3].
The Case for Carbon Nanotube Interconnects An interesting solution, which has been the subject of intense research in recent years, is to replace copper
C
C
396
Carbon Nanotubes for Chip Interconnections, Fig. 5 Graphical representations of ideal graphene sheet, SWCNT, MWCNT (Figure reprinted with permission from Graham et al. [22])
with carbon nanotubes. If a reliable and repeatable fabrication process consistent with Complementary Metal Oxide Semiconductor (CMOS) technology requirements could be developed, integration into existing chip architectures may not require significant process redesign (Fig. 5). Carbon nanotubes, which can be visualized as rolled sheets of graphene, have been widely investigated as a promising new material for many electrical device applications [5, 7] (e.g., transistor (CNT-FET), interconnects) as they exhibit exceptional electrical, thermal, and mechanical properties [8]. When comparing materials properties, CNTs are a clear winner against copper. Studies show that CNTs are stable for current densities up to 109 A/cm2, two orders of magnitude higher than copper. CNTs can exhibit multichannel ballistic conduction over distances of microns. Because of their higher chemical stability relative to copper, diffusion barriers (liners) are not needed for CNTs, thus allowing a larger conductive cross section compared to copper for the same technology node. Additionally, their mechanical tensile strength (100 times that of steel) and their high thermal conductivity (comparable to diamond) give CNTs an edge compared to copper. Finally, growing CNTs in high aspect ratio vias could allow the design of chips with higher interlayer spacing to reduce overall RC losses and to decrease chip-layer energy dissipation [9]. Before examining possible models of CNT-based interconnect architectures, it is important to clearly understand CNTs’ electrical properties, which represent the most critical material limitation to resolve with respect to copper. The electronic band structures of single-wall CNTs (SWCNTs) and of graphene are very similar. For graphene and metallic SWCNTs, the valence band and the conduction band
Carbon Nanotubes for Chip Interconnections
touch at specific points in the reciprocal space. For semiconducting SWCNTs, the conduction band and the valence band do not touch. Semiconducting SWCNTs have been extensively studied as channels in transistor devices while metallic SWCNTs have been considered for applications such as IC interconnects and field emission. The resistance of a CNT contacted at both ends is the sum of three resistances [5, 10]: RCNT ¼ RQ þ RL þ RCONTACT where RQ is the quantum resistance, RL is the scattering resistance, and RCONTACT is the contact resistance. We will now discuss these three resistances. An ideal (defect-free) metallic SWCNT electrically contacted at both ends, in the absence of scattering or contact resistance, exhibits a resistance R ¼ 2 RQ 13kO as a SWCNT has two conduction channels. The quantum resistance RQ ¼ 6:5kO is due to the mismatch between the number of conduction channels in the nanotube and the macroscopic metallic contacts. The one-dimensional confinement of electrons, combined with the requirement for energy and momentum conservation, leads to ballistic conduction over distances in the order of a micron. The scattering resistance is due to impurities or nanotube defects that reduce the electron mean free path, and depends on the length l of the SWCNT: 1 l RL ¼ 2 RQ l0 For defect-free SWCNT lengths below a micron, we can neglect the scattering resistance The contact resistance, which results from connecting the SWCNT to a contact (usually metallic), depends strongly on the material in contact with the nanotube, and on the difference between their work functions. The work functions of multiwall CNTs (MWCNTs) and SWCNTs have been estimated to be 4.95 and 5.10 eV, respectively [10]. Palladium has been found to be one of the materials minimizing the contact resistance, better than titanium or platinum contacts (which exhibited nonohmic behavior when in contact with CNTs) [10]. For interconnect applications, most often bundles of SWCNTs are considered. It is important to note that the coupling between adjacent SWCNTs is negligible
Carbon Nanotubes for Chip Interconnections Carbon Nanotubes for Chip Interconnections, Fig. 6 Equivalent circuit model of metallic SWCNT used in HSPICE simulations (Figure reprinted with permission from Naeemi and Meindl [14])
RC1
397 dx
RQ /2
RQ /2
RC2 (RQ / l0 ) = 216KΩ /μm Rshunt = (RQ / l0 )dx Re = (RQ / le )dx
RV lM .dx lk .dx Re
since, for defect-free SWCNTs, the electrons would rather travel along the SWCNT axis (ballistic path) than across SWCNTs because of the large inter-CNT tunneling resistance (2–140 MO) [10]. Thus, the resistance of a bundle of SWCNTs can be viewed as a parallel circuit of the resistances of the individual SWCNTs. If we have n SWCNTs, the resistance of the bundle will be:
bundle
¼
RSCWNT n
The above overview related to SWCNTs. The electrical properties of MWCNTs have not been as extensively studied because of the additional complexities arising from their structure, as every shell has different electronic characteristics and chirality, in addition to interactions between the shells [11]. Geometrically, the interwall distance in a MWCNT is 0.34 nm, the same as the spacing between graphene sheets in graphite. What still has to be clarified is how the conductivity of a MWCNT varies with the number of walls. Initially, it was thought that the conductance of a MWCNT occurred only through the most external wall, which seems to be the case at low bias and temperatures, where electronic transport is dominated by outer-shell conduction. However, theoretical models and experimental results indicate that shell-to-shell interactions can significantly lower the resistance of MWCNTs with many walls [5, 10]. One view is that the conductivity of a MWCNT with n walls is simply n times the conductivity of a SWCNT. Li et al. [12] experimentally measured an electrical resistance of only 34.4O for a large MWCNT with outer diameter of 100 nm and inner diameter of 50 nm (¼ > 74 walls). This value was much lower than
RV = (dV (x)/ I0) CQ .dx
Rshunt
RSCWNT
C
CE .dx
I0 = 25 μA
the one that could be calculated assuming all walls participated separately in the electrical conduction (i.e., calculated as the parallel of the resistances for each wall) showing that interwall coupling contributes to additional channels of conductance. Naeemi et al. [13] also assumed intercoupling between CNT walls in their models to increase the channels of conduction with increasing number of walls. However, their conductance was lower compared to that measured experimentally by Li’s team. Bottom line: The conductivity of MWCNTs increases with the number of walls but the exact relationship has not yet been exactly clarified.
Models of CNTs as Interconnects Various studies investigated replacing copper interconnects with bundles of CNTs (SWCNTs or MWCNTs) or with one large MWCNT. A first approach consists of using densely packed SWCNTs. Modeling SWCNTs as equivalent electrical circuits and using SPICE simulations, Naeemi et al. [14] showed that a target density of SWCNTs of at least 3.3 1013 CNTs/cm2 was required. Currently achieved maximum densities for CNTs in vias barely reach 1012 CNTs/cm2, which is still an order of magnitude smaller than required. Furthermore, since statistically only one-third of the SWCNTs grown are metallic (the other two-third are semiconducting), the conduction of the bundle will only occur in the metallic SWCNTs (Fig. 6). Naeemi et al. [14] also compared SWCNT bundles to copper as local, intermediate, and global interconnects. They showed that, in SWCNT bundles, resistance and kinetic inductance decreased linearly with the number of nanotubes in the bundle, while magnetic inductance changed very slowly. The resistance of
C
C
398
Carbon Nanotubes for Chip Interconnections le = 1.6 μm
scopper-Bulk scopper@22 nm-Node
0.10 VAB = 0 VAB = 0.05 V VAB = 0.2 V VAB = 0.5 V
0.01
0.10
1.00
10.00
100.00
SWNT–Bundle Length, l (μm)
Conductivity, σ(μΩ–cm)−1
Conductivity, σ (μΩ–cm)−1
1.00
1.00
Cu Wire, W = 100 nm Cu Wire, W = 50 nm Cu Wire, W = 20 nm
SWCN, D=1 nm
0.10
l = l0b/a MWCN, D = 10 nm MWCN, D = 20 nm MWCN, D = 50 nm MWCN, D = 100 nm
0.01 0.1
1.0
100.0 10.0 Length, l (μm)
1000.0
Carbon Nanotubes for Chip Interconnections, Fig. 7 Conductivity of densely packed SWCNT bundles versus length for various bias voltages (Figure reproduced with permission from Naeemi and Meindl [14])
Carbon Nanotubes for Chip Interconnections, Fig. 8 Conductivity of MWCNTs with various diameters compared to Cu wires and dense bundles of SWCNTs (Figure reproduced with permission from Naeemi and Meindl [13])
a bundle of SWCNTs with sufficient metallic nanotubes was smaller than the resistance of copper wires, while capacitance was comparable. SWCNT bundles also fared better compared to copper in reducing power dissipation, delay, and crosstalk. For local interconnects, they quantified the improvements as 50% reduction in capacitance, 48% reduction of capacitance coupling between adjacent lines, and 20% reduction in delay. For intermediate interconnects, the improvements were more marked, especially in terms of improved conductivities. For global interconnects, dense SWCNT bundles proved critical to improve bandwidth density (Fig. 7). Using MWCNTs, which are all electrically conductive as they exhibit multiple channels of conduction (compared to only one-third metallic SWCNTs), could lower the resistivity of the bundle, although fewer of them can be packed in the same space because they usually have larger diameters (but also require a lower packing density compared to SWCNTs). In a different modeling study, Naeemi et al. [13] explored the suitability of MWCNTs as replacement for copper interconnects. They concluded that for long lengths (over 100 mm), MWCNTs have conductivities many times that of copper and even of SWCNT bundles. However, for short lengths (less than 10 mm), dense SWCNT bundles can exhibit a conductivity that is twice that of MWCNT bundles. Thus, for via applications, they recommended using dense bundles of SWCNTs or,
alternatively, bundles of MWCNTs with small diameter (i.e., with few walls) (Fig. 8). Using an individual MWCNT with large diameter could offer high conductivity due to the participation of multiple walls to significantly increase the channels for conduction. However, as previously mentioned, the exact relationship between the number of walls and the conductance has yet to be clarified. In conclusion, the choice of nanotubes may differ depending on the type of interconnect. For instance, dense bundles of SWCNTs or MWCNTs with few walls may be more suitable for small-section vertical vias, while dense bundles of larger MWCNTs may be more appropriate for long-range interconnects. The option of using a large MWCNT which fills all the space available needs to be further investigated. It is also plausible that hybrid systems of copper/SWCNTs/ MWCNTs may be the best solution; for instance, small-section vertical vias may be replaced by dense SWCNT bundles, while larger long-range horizontal interconnects may still use copper, dense bundles of MWCNTs, or even metal-CNT composites.
Practical Implementation: Fabrication State of the Art and Outstanding Issues Reliable and repeatable high-yield CNT fabrication compatible with CMOS standards is the main
Carbon Nanotubes for Chip Interconnections Carbon Nanotubes for Chip Interconnections, Fig. 9 Pictorials comparing the “grow-in-place” and “grow-then-place” techniques (Reproduced with permission from Professor Carl V. Thompson [18])
399 Grow-In -Place C2H2
C
Grow-Then-Place C2H2
C
C2H2
C2H2
Chemically-directed Field-directed Catalyst: Ni, Co, Fe ...
bottleneck in replacing copper in chip interconnections. Although hundreds of research teams have focused their efforts on nanotube growth and thousands of papers detailing growth recipes have been published, surprisingly, very few have focused on the growth on conductive substrates at CMOS-compatible processing temperature [7, 10, 15, 16]. It is still a challenge to reliably and consistently synthesize CNTs on conductive layers at temperatures below 400–450 C, the maximum temperature allowed in CMOS fabrication to avoid disrupting previous diffusion patterns. Furthermore, it is still difficult to precisely control CNT diameter and height, although chemical vapor deposition (CVD) from thin films of controlled thickness [17] or from nanoparticles [16] of controlled size has shown encouraging results. To utilize carbon nanotubes in industrial applications, two main approaches have been considered: “grow-in-place” and “grow-then-place” [18] (Fig. 9). Grow-then-place: This technique consists of first preparing nanotubes and subsequently transferring them to a substrate. Arc discharge and laser ablation are the main techniques used to synthesize free-standing nanotubes. The nanotubes may be subsequently selected (e.g., separating SWCNTs or metallic SWCNTs) and purified prior to use. To transfer them to another substrate, CNTs are usually functionalized in a way that they will attach to pre-patterned areas of the substrate which will attract functionalized CNTs. An interesting technique for interconnect vias is based on using electrophoresis to push CNTs dispersed in a liquid solution into a matrix with pits (e.g., porous alumina matrix).
The advantages of this method are that it places no restrictions on the process or temperature used for CNT synthesis and allows to pretreat the CNTs (e.g., select, purify, functionalize). The major drawback is that no successful and repeatable technique to transfer the CNTs to the substrate has been developed to date. The challenge of resolving this issue appears too high to make this technique a candidate for the CMOS industry. However, free-standing, purified, CNTs are manufactured by many companies and sold for other applications (e.g., CNT-polymer composites). Grow-in-place: this technique usually consists of preparing the sample with a catalyst present in the locations where the nanotubes will be synthesized. For instance, a thin catalyst film can be deposited using e-beam evaporation or sputtering; alternatively, nanoparticles can be deposited on a substrate. Synthesis is usually performed using thermal or assisted (e.g., plasma) CVD. This method has several advantages: (1) good control of nanotube position (CNTs will grow where there are catalysts), (2) proven recipes to obtain crystalline CNTs (at least on insulating substrates), (3) proven capabilities to obtain carpets of vertically aligned CNTs, (4) physical contact with the substrate, (5) electrical contact with the substrate, and (6) CVD techniques are commonplace in the CMOS industry. The major drawbacks are that (1) the processing temperature should be below 400–450 C (CMOScompatibility), thus putting serious limits on the synthesis method and (2) the CNTs should be directly synthesized on the substrate of choice, usually
C
400
Carbon Nanotubes for Chip Interconnections
h c
CNTs
b a
Via hole
Ta barrier
1 μm
Co catalyst particles
SiO2
g
TiN
cap
Cu Si substrate
d
Spin on Glass (SOG)
e
f
Planarization by CMP
Ta barrier Cu
Ti contact layer
5 nm
Carbon Nanotubes for Chip Interconnections, Fig. 10 Process to synthesize CNTs into pits, SEM cross section, and TEM showing crystalline MWCNT (Reprinted with permission from Yokoyama et al. [23])
a metallic layer to provide electrical contact. Although growing dense carpets of crystalline CNTs on insulating substrates such as alumina or silicon oxide has been achieved by many, CNT growth on metallic layers still remains a serious challenge. Interactions between the catalyst and the metallic substrate (e.g., alloying) are the major impediments for the successful growth of dense carpets of CNTs on metallic layers. In addition to interesting results obtained from university research, good progress on the growth and characterization of carbon nanotubes for interconnects has been achieved by industrial laboratories, initially by Infineon and now by the Fujitsu laboratories. In 2002, Kreupl et al. [19] of Infineon showed that bundles of CNTs could be grown in pits of defined geometry. More recently, Awano et al. [7] of Fujitsu grew bundles of MWCNTs in a 160 nm via at 450 C and measured an electrical resistance of 34O (the CNT density observed was 3 1011 CNTs/cm2). This follows a previous result obtained earlier by the same team where they grew MWCNTs into a 2 mm via at temperatures close to 400 C with the lowest resistance measured of 0.6O after CMP and annealing in a hydrogen atmosphere [5] (Fig. 10).
Kreupl et al. [19] succeeded in growing a single MWCNT into a 25-nm hole and measured a high resistance of 20–30 kO. Given the difficulty of growing a single large MWCNT and the difficult task of making a precise electrical measurement, there may be room for further improvement if we could grow an individual, crystalline (defect-free) MWCNT with the maximum number of walls for a given external diameter, thus maximizing the number of channels of conduction. Most of the effort on CNT synthesis to replace copper interconnects has focused on vertical growth of dense carpets of CNTs, which can be achieved by a high density of active catalyst dots. In contrast there have been fewer successful reports of horizontal growth, with less spectacular results. Many techniques have been used to achieve horizontal alignment among which we can mention high gas flow rates, electric fields, and epitaxial techniques to guide horizontal alignment of the nanotubes [10]. In 2010, Yan et al. [20] obtained an interesting horizontal growth of bundled CNTs with a density of 5 1010 CNTs/cm2, which is approaching what has been achieved for vertical CNT growth (although still over an order of magnitude lower compared to the best result for vertical CNT growth) (Fig. 11).
Carbon Nanotubes for Chip Interconnections Carbon Nanotubes for Chip Interconnections, Fig. 11 Scanning electron microscope images of dense carpets of horizontally aligned CNTs grown using CVD (Figure reproduced from Yan et al. [20])
401
a
b
C 200 nm
Although the experimental results obtained are encouraging, there are still numerous challenges that need to be resolved for CNTs to enter the semiconductor fab: 1. Increase the CNT areal density by one or two orders of magnitude. For SWCNTs, assuming all of them are metallic, a packing density of 1013–1014 CNTs/cm2 is required to compete with copper in terms of resistance, while for MWCNTs, the required packing density is lower and depends on the number of channels of conduction (i.e., number of walls). This will require, among other considerations, adequate catalyst and underlayer materials choice and deposition, possible surface pretreatment (e.g., plasma, reduction, and etching), maximum nucleation of active catalyst dots, and optimizing the CNT growth process. Although CNT areal density is an important issue, it may not be the dealbreaker. 2. Minimize the contact resistance between the CNTs and the substrate. To achieve maximum conductivity, the choice of the appropriate underlayer is critical; specific metals (e.g., Pd) and possibly silicides are good candidates. For MWCNTs, it is also important to ensure electrical contact with all the walls. 3. When using SWCNTs, synthesize only metallic SWCNTs (on average one-third of the SWCNTs grown) which are the ones participating in the electrical conduction. This is closely linked to the issue of chirality control, for which no solution has been proposed to date. Selective catalyst choice may provide an alternative avenue to synthesize a higher fraction of metallic SWCNTs. 4. Control growth direction of CNTs. This is especially challenging for horizontal interconnects where the directionality and the packing density achieved are still lagging compared to vertical
C
2 μm
growth of CNTs, despite some interesting progress in this area [10, 20]. 5. Synthesize crystalline, defect-free CNTs to ensure maximum electrical conductivity in the nanotube. This is challenging, especially when combined with the requirement of growing CNTs at low temperature to achieve CMOS compatibility. 6. Synthesize CNTs at temperatures below 400– 450 C to ensure CMOS compatibility. 7. Repeatably yield the same CNT structures when the same process conditions are applied. This is a major issue since important variations in the structure and shape of the CNTs grown have been experimentally observed.
Conclusions and Outlook for CNTs as Chip Interconnections In the past decade, the synthesis of CNTs and the understanding of their growth mechanisms have massively improved. However, for CNTs to enter the CMOS fab and replace copper, significant challenges still need to be resolved. In my opinion, the most significant challenge to overcome is developing a reliable and repeatable fabrication process consistent with CMOS conditions. To achieve that tall order, we need to improve our understanding of the CNT growth mechanisms. Although many simulation models and many experimentally-based insights have been achieved [10], there are still many questions related to CNT growth mechanisms that have not been fully answered. For instance: – Which precursor gases favor CNT growth and which gases hinder CNT growth? What is the role of the gases in the resulting level of crystallinity of
C
402
the CNTs grown? Could we pretreat the gases to improve the CNT yield or structure? – What is the exact role of the catalyst? How does its materials properties and its lattice structure influence the resulting CNTs grown (in shape and structure)? – What is the role of the underlayer (layer below the catalyst) and its interactions with the catalyst? Why is it so challenging to grow CNTs on metallic layers? In addition to improving our mechanistic understanding of CNT growth, I believe that a parallel effort focused on developing better reactors is needed. Most researchers use standard CVD-based systems that were designed for a general purpose. A customized reactor, where the same growth conditions can be repeatably achieved with very small variations could provide the repeatability in results (CNT structures) that has eluded us so far. Carbon nanotubes are already becoming a manufacturing reality in mechanical engineering applications (e.g., CNT-based composites) and many interesting results have been obtained to develop novel CNT structures for electrical applications. Although the jury is still out, if process repeatability could be achieved, we could hope, not only that CNTs will enter the CMOS fabs and replace copper for chip interconnections, but also that they will lead to innovative ventures requiring lower investments to develop integrated circuits with radically new architectural designs using carbon nanotubes as new building blocks.
Cross-References ▶ Carbon Nanotube-Metal Contact ▶ Carbon Nanotubes ▶ Chemical Vapor Deposition (CVD) ▶ CMOS-CNT Integration ▶ Nanotechnology ▶ Physical Vapor Deposition ▶ Synthesis of Carbon Nanotubes
References 1. Kuhn, K.J.: Moore’s Law Past 32 nm: Future Challenges in Device Scaling. Intel Publication, Hillsboro (2009) 2. Magen, N., Kolodny, A., Weiser, U.: Interconnect-power dissipation in a microprocessor. In: Proceedings of the 2004 International Workshop, Paris, 1 Jan 2004
Carbon Nanotubes for Chip Interconnections 3. ITRS. International Technology Roadmap for Semiconductors - Interconnect 2009, International Sematech, Austin 4. Goel, A.K.: High-Speed VLSI Interconnections, 2nd edn. Wiley/IEEE, Hoboken (2007) 5. Nessim, G.D.: Carbon Nanotube Synthesis for Integrated Circuit Interconnects. Massachusetts Institute of Technology, Cambridge, MA (2009) 6. Havemann, R.H., Hutchby, J.A.: High-performance interconnects: an integration overview. Proc. IEEE 89(5), 586–601 (2001) 7. Awano, Y., Sato, S., Nihei, M., Sakai, T., Ohno, Y., Mizutani, T.: Carbon nanotubes for VLSI: interconnect and transistor applications. Proc. IEEE 98(12), 2015–2031 (2010) 8. Dresselhaus, M.S., Dresselhaus, G., Avouris, P. (eds.): Carbon Nanotubes: Synthesis, Structure, Properties, and Applications. Springer, Berlin (2001) 9. Chen, F., Joshi, A., Stojanovic´, V., Chandrakasan, A.: Scaling and evaluation of carbon nanotube interconnects for VLSI applications. In: Nanonets Symposium 07, Catania, 24–26 Sept 2007 10. Nessim, G.D.: Properties, synthesis, and growth mechanisms of carbon nanotubes with special focus on thermal chemical vapor deposition. Nanoscale 2(8), 1306–1323 (2010) 11. Collins, P.G., Avouris, P.: Multishell conduction in multiwalled carbon nanotubes. Appl. Phys 74(3), 329–332 (2002) 12. Li, H.J., Lu, W.G., Li, J.J., Bai, X.D., Gu, C.Z.: Multichannel ballistic transport in multiwall carbon nanotubes. Phys. Rev. Lett. 95(8), 086601 (2005) 13. Naeemi, A., Meindl, J.D.: Compact physical models for multiwall carbon-nanotube interconnects. IEEE Electr. Device Lett. 27(5), 338–340 (2006) 14. Naeemi, A., Meindl, J.D.: Design and performance modeling for single-walled carbon nanotubes as local, semiglobal, and global interconnects in gigascale integrated systems. IEEE T Electron Dev 54(1), 26–37 (2007) 15. Nessim, G.D., Seita, M., O’Brien, K.P., Hart, A.J., Bonaparte, R.K., Mitchell, R.R., Thompson, C.V.: Low temperature synthesis of vertically aligned carbon nanotubes with ohmic contact to metallic substrates enabled by thermal decomposition of the carbon feedstock. Nano Lett. 9(10), 3398–3405 (2009) 16. Awano, Y., Sato, S., Kondo, D., Ohfuti, M., Kawabata, A., Nihei, M., Yokoyama, N.: Carbon nanotube via interconnect technologies: size-classified catalyst nanoparticles and lowresistance ohmic contact formation. Phys. Status Solidi 203(14), 3611–3616 (2006) 17. Nessim, G.D., Hart, A.J., Kim, J.S., Acquaviva, D., Oh, J.H., Morgan, C.D., Seita, M., Leib, J.S., Thompson, C.V.: Tuning of vertically-aligned carbon nanotube diameter and areal density through catalyst pre-treatment. Nano Lett. 8(11), 3587–3593 (2008) 18. Thompson, C.V.: Carbon nanotubes as interconnects: emerging technology and potential reliability issues. In: 46th International Reliability Symposium; 2008: IEEE CFP08RPS-PRT, p. 368, 2008 19. Kreupl, F., Graham, A.P., Duesberg, G.S., Steinhogl, W., Liebau, M., Unger, E., Honlein, W.: Carbon nanotubes in interconnect applications. Microelectron. Eng. 64(1–4), 399–408 (2002)
Cell Morphology 20. Yan, F., Zhang, C., Cott, D., Zhong, G., Robertson, J.: Highdensity growth of horizontally aligned carbon nanotubes for interconnects. Phys Status Solidi. 247(11–12), 2669–2672 (2010) 21. Jackson, R.L., Broadbent, E., Cacouris, T., Harrus, A., Biberger, M., Patton, E., Walsh, T.: Processing and integration of copper interconnects. In: Solid State Technology. Novellus Systems, San Jose (1998) 22. Graham, A.P., Duesberg, G.S., Hoenlein, W., Kreupl, F., Liebau, M., Martin, R., Rajasekharan, B., Pamler, W., Seidel, R., Steinhoegl, W., Unger, E.: How do carbon nanotubes fit into the semiconductor roadmap? Appl. Phys. 80, 1141–1151 (2005). Copyright 2005, Springer Berlin/ Heidelberg 23. Yokoyama, D., Iwasaki, T., Yoshida, T., Kawarada, H., Sato, S., Hyakushima, T., Nihei, M., Awano, Y.: Low temperature grown carbon nanotube interconnects using inner shells by chemical mechanical polishing. Appl. Phys. Lett. 91, 263101 (2007). Copyright 2007, American Institute of Physics
403
Catalytic Bimetallic Nanorods ▶ Molecular Modeling on Artificial Molecular Motors
Catalytic Chemical Vapor Deposition (CCVD) ▶ Chemical Vapor Deposition (CVD)
Catalytic Janus Particle ▶ Molecular Modeling on Artificial Molecular Motors
Carbon Nanotubes for Interconnects in Integrated Circuits
Cathodic Arc Deposition
▶ Carbon Nanotubes for Chip Interconnections
▶ Physical Vapor Deposition
Carbon Nanotubes for Interconnects in Microprocessors
Cavity Optomechanics
▶ Carbon Nanotubes for Chip Interconnections
Carbon Nanowalls
C
▶ Optomechanical Resonators
Cell Adhesion
▶ Chemical Vapor Deposition (CVD)
▶ Bioadhesion ▶ BioPatterning
Carbon-Nanotubes
Cell Manipulation Platform
▶ Robot-Based Automation on the Nanoscale
▶ Biological Breadboard Platform for Studies of Cellular Dynamics
Catalyst ▶ Chemical Vapor Deposition (CVD) ▶ Physical Vapor Deposition
Cell Morphology ▶ BioPatterning
C
C
404
Cell Patterning
toxicity and cell death. They are associated to the uptake of nanoparticles, their persistence at cellular level, their ability to release free radicals, and to induce an oxidative stress. The resulting activation of molecular pathways and transcription factors could lead to a pro-inflammatory response or, depending on the level of free radicals, apoptosis.
Cell Patterning ▶ BioPatterning
Cellular and Molecular Toxicity of Nanoparticles ▶ Cellular Mechanisms of Nanoparticle’s Toxicity
Cellular Electronic Energy Transfer ▶ Micro/Nano Harvesting
Transport
in
Microbial
Energy
Cellular Imaging ▶ Electrical Impedance Tomography for Single Cell Imaging
Cellular Mechanisms of Nanoparticle’s Toxicity Francelyne Marano, Rina Guadagnini, Fernando Rodrigues-Lima, Jean-Marie Dupret, Armelle BaezaSquiban and Sonja Boland Laboratory of Molecular and Cellular Responses to Xenobiotics, University Paris Diderot - Paris 7, Unit of Functional and Adaptive Biology (BFA) CNRS EAC 4413, Paris cedex 13, France
Synonyms Cellular and molecular toxicity of nanoparticles
Definition The interaction between nanoparticles and cell triggers a cascade of molecular events which could induce
Background The last five years have shown an increasing number of papers on the nanoparticle’s mechanisms of cytotoxicity. What are the reasons? It is likely that the specific useful properties which appear at nanoscale can also lead to adverse effects. This hypothesis is strongly supported by in vivo and in vitro studies which compare the toxicity of NPs with their fine counterparts of the same chemical composition. These results have clearly demonstrated a higher toxicity of particles at nanoscale than at microscale. Moreover, it appears from experimental studies that solid nano-sized particles could be translocated beyond the respiratory tract and could induce a systemic response. The interstitial translocation of a same mass of particles is higher for ultrafine than fine particles after intratracheal instillation in rats [1]. Surface area, which is strongly increased for nanoparticles compared to microparticles of same chemical composition, and surface reactivity are considered as the principal indicators of NP’s reactivity. It was shown that a toxic response could be observed even to apparently nontoxic substances when the exposure occurred in the nanometre size range. All these observations have lead to the development of a new field of toxicology, nanotoxicology [2]. However, the toxicological mechanisms which sustained the biological response are not yet clear and a matter of debates. The concerns about the toxicity of engineered nanoparticles, which are increasingly used for industrial and medical applications, came also from the knowledge on the toxicity of non-intentional atmospheric particles. Short-term epidemiological studies in Europe and North America have shown an association between cardio-respiratory morbidity and mortality and an increased concentration of atmospheric fine particles [3]. Moreover, long-term epidemiological studies have also demonstrated an association between exposure to atmospheric particles
Cellular Mechanisms of Nanoparticle’s Toxicity
(Particulate Matter or PM10 and 2.5) and increased cancer risk [3]. In parallel, in vitro and in vivo studies on fine and ultrafine airborne particles such as Diesel exhaust particles and PM2.5, gave causal explanations to these adverse health effects (reviewed in [1]). They allow to define the molecular events induced by these particles in lung cells. The major event is a pro-inflammatory response which is characterized by various cytokine releases (pro-inflammatory mediators), associated to the activation of transcription factors, and signaling pathways. This was especially demonstrated for diesel exhaust particles (DEP), a major component of urban PM in Europe. These events are mostly induced by organic components of the DEP and are probably mediated by the generation of reactive oxygen species (ROS) during the metabolism of organic compounds (for a review see [4]). These findings were used as a background for the researches on biological mechanisms induced by NPs considering that fine and ultrafine atmospheric particles have great similarities with NPs, especially Diesel exhaust particles which are of nano-size and aggregate after their release in atmosphere. It became rapidly obvious that the understanding of the cellular and molecular mechanisms leading to the biological effects of NPs is essential for the development of safe materials and accurate assays for risk assessment of engineered NPs [2] and several recent reviews were focused on demonstrated or hypothetic cellular mechanisms of these responses [4–6, 11]. The first event, when NPs enter in contact with the human body by inhalation, ingestion, dermal exposure, or intravenous application, is their interaction in the biological fluids and the cellular microenvironment with biological molecules such as proteins thus forming a protein corona [7]. Consequently, NPs do not directly interact with the cell membrane but through the protein and/or lipids of the corona. NPs-bound proteins may recognize and interact with the membraneous receptors or could bind nonspecifically cellular membranes. Whatever these interactions, they seem to play a central role which could determine further biological responses. In particular, these interactions may drive the uptake of NPs by the first target cells at the level of the biological barriers such as immune cells (macrophages, dendritic cells, and neutrophiles) or epithelial and endothelial cells. This uptake seems to be general for many NPs which are able to bind proteins at their surface and the paradigm
405
C
of “Trojan horse” was developed to explain this uptake and the further biological responses. One of the first responses is the direct or indirect production of ROS which is associated to the size, the chemical composition and the surface reactivity of the NPs. This common response occurs for a large number of NPs even with different chemical patterns and different abilities to form agglomerates, so that, the paradigm of the central role of oxidative stress was developed [5]. These authors suggested that “although not all materials have electronic configurations or surface properties to allow spontaneous ROS generation, particle interactions with cellular components are capable of generating oxidative stress.” Further activation of nuclear factors and specific genetic programs are associated to the level of ROS production leading to cell death by necrosis and apoptosis or adaptive responses such as pro-inflammatory responses, antioxidant enzyme activation, repair processes, effects on cell cycle control and proliferation. Over the last years, numerous in vitro studies have confirmed this hypothesis leading to the development of assays using the detection of ROS or oxidative stress for the screening of NPs. However, new data during the last year have pointed out other specific effects of NPs which are not related to oxidative stress. For example, NPs can interact with membrane receptors, induce their aggregation and mimic sustained physiological responses through specific signaling pathways in the target cells. This type of mechanism may contribute to the development of diseases but could also be of use to develop therapeutic strategies whereby NPs activate or block specific receptors.
Cellular Uptake of Nanoparticles and Their Fate at Cellular Level The uptake of particles by specialized immune cells in humans is a normal process which leads to their removal and contributes to the integrity of the body. However, depending on the level of the uptake, this process could induce an increasing release of inflammatory mediators and disturbance of phagocyte normal functions such as the clearance and the destruction of pathogens. One of the knowledge of the 50 last years on the effects of a sustained exposure to airborne particles, especially at occupational level, is the concept of overloading. If the mechanisms of
C
C
406
Cellular Mechanisms of Nanoparticle’s Toxicity Interactions Size
Shape Diameter Agregation/Agglomeration
Specific surface area
Interactions with Proteins + + + − − − + − + − −−− +++
Chemical composition Coating
Solubilisation Charge
Interactions with Receptors
Surface chemistry
Cellular Mechanisms of Nanoparticle’s Toxicity, Fig. 1 Different physicochemical characteristics of the nanomaterials involved in their biological activity: size, surface area, shape, bulk chemical composition, surface chemistry
including solubility as well as surface charge or coatings, interactions between particles leading to agglomeration, and aggregation as well as with proteins leading to “corona” or with receptors in the cell membrane
clearance are not sufficient to eliminate the particles and if they are persistent, the particles could accumulate in the tissues, leading to a sustained inflammation and chronic pathologies. This was demonstrated for exposure to quartz, asbestos, coal, mineral dusts but also for long time exposure to heavy PM-polluted atmospheres such as in Mexico City [1]. These questions of uptake and persistence are fundamental for risk assessment evaluation of NPs. This may explain the number of papers published recently that analyze the mechanisms of uptake, the behavior and the translocation of various NPs. So far, it appears that the response depends on several parameters. Taken together, NP’s surface and its specific chemical composition resulting from the engineering processes, the capacity of NPs to form aggregates (particle comprising strongly bonded or fused NPs) or agglomerates (collection of weakly bound NPs), the methods used for dispersion and experimental preparation determine in the NP’s ability to adsorb or not specific biological compounds, such as proteins, to form the “corona” and to interact with biological membranes [7]. The amount and the structural/functional properties of the adsorbed proteins drive the interactions of these nanomaterials with the membranes and their uptake (Fig. 1). Recent studies have clearly identified a number of serum proteins such as albumin, IgG, IgM, IgA, apolipoprotein E, cytokines, or transferrin that bind to carbon black, titanium dioxide, acrylamide or polystyrene NPs [8]. Among the identified proteins, several are ligands for cellular receptors and may contribute to the biological
effects of NPs. For example, receptor aggregation induced by NPs could lead to cell signaling: coated gold NPs were able to bind and cross-link IgE-Fc epsilon receptors leading to degranulation and consequent release of chemical mediators [9]. On another hand, integrins such as a5b3 are known to play a key role in cell signaling and their activation by extracellular ligands can modulate biological processes such as matrix remodeling, angiogenesis, tissue differentiation, and cell migration. These receptors were recently demonstrated as important membrane targets for carbon nanoparticles and their activation induced lung epithelial cell proliferation which was due at least in part to b1-integrin activation [6]. As far as, uptake process is concerned, it is likely that different cell types might have different uptake mechanisms, even for the same NPs. The possible pathways of cellular uptake were previously described by several authors (see [10]). It could occur through phagocytosis, macropinocytosis, clathrin-mediated endocytosis, non-clathrin, non-caveolae-mediated endocytosis, caveolae-mediated endocytosis, or diffusion (Fig. 2). These mechanisms have been described for different NPs and may occur for the same NP depending on the cell type, the medium, the level of aggregation. Therefore, uptake processes are considered as very complex and not easy to measure. Dawson et al. 2009 [12] have postulated that the uptake depends mostly on the size: NPs less than 100 nm can enter the cells and less than 40 nm in the nucleus. It was also suggested that the size of the NPs determine caveolin
Cellular Mechanisms of Nanoparticle’s Toxicity
407
NPs ?
Calcium Integrin Receptor Channel ?
NPs
C
?
EGFR Endocytosis ROS ↑↑↑Calcium Lysosomal damage
ROS Mitochondrial Damage
Activation of apoptosis
Activation of MAPK
(Proteases/Cathepsin release)
Activation of transcription factors
ARE DNA Damage
phase II enzymes (GST, NQO-1) antioxidant enzymes (HO-1)
Cellular Mechanisms of Nanoparticle’s Toxicity, Fig. 2 A schematic representation of NP’s triggered cellular pathways through membrane receptors, ROS production and implication of oxidative stress in these responses. NPs could induce activation of EGF or integrin receptors can lead to apoptosis, inflammation or proliferation. ROS produced by NPs in immediate cellular environment or inside the cells lead to activation of redox-dependent signaling pathways like MAPK and the activation of transcription factors for example, AP-1, NF-kB,
or Nrf2. They migrate to the nucleus and modify gene expression of cytobines, phase 2 enzymes (gluthation S transferase or GST, quinone oxydore´ductase 1 or NQO-1), and antioxydant enzymes (hemeoxygenase 1 or HO-1). Oxidative stress could also results in the damage of different organelles like the mitochondria, lysosomes and nucleus resulting apoptosis. Accumulation of high intracellular calcium levels through a direct effect on calcium channel might also act as an alternative mechanism for the induction of these mechanisms (Adapted from [11])
versus clathrin dependent uptakes [13]. However, these oversimplified scenarios are refuted by obvious discrepancies in the recent literature about the optimal size, shape, and mechanisms of internalization of NPs. The surface charge of the NPs could be an important factor for uptake since the negatively charged surface membrane could favor the positively charged NPs for higher internalization. However, negatively charged NPs were also shown to have enhanced uptake as compared to unfunctionalized NPs, perhaps by their possible interactions with proteins. Endocytosis of small NPs is energy-dependent and associated to lipid rafts, dynamin and F-actin mechanisms. Phagocytosis and macropinocytosis are mostly involved in endocytosis of large particles (more than 500 nm) and also in the uptake of the aggregates or agglomerates of NPs which could be promoted by their opsonisation in the biological fluids. Macropinocytosis (which is one kind of pinocytosis) is also an important mechanism for positively charged NPs and TiO2 or carbon black aggregates internalization [14].
The behavior of the NPs after their uptake is another important question but, surprisingly, as far as now, little is known about the intracellular fate of NPs. Most of the transmission electron microscopy (TEM) observations have shown the NPs in cytoplasm vesicles limited by membranes. These vesicles could further be transported in the cytoplasm through the microtubule network. The biopersistence of nanomaterials which are resistant to degradation in the endosomal compartment could be one of the factors of further toxicity and accumulation. However, several metal oxide NPs are toxic after dissolution in the cell. Indeed, the uptake of ZnO NPs into the lysosomal acidic medium accelerates their dissolution and the release of Zn2+ ions in the cytoplasm. Their excess could induce cytokine production and cytotoxicity and the initiation of acute inflammation at the level of the target organ such as the lung. NPs such as TiO2 or carbon black NPs were also observed only in the cytoplasm of cells [14]. Two explanations may be put forward. The first one is that
C
C
408
NPs could directly enter by diffusion through the lipid bilayer. It has been shown that cationic NPs could pass through cell membranes by generating transient holes without membrane disruption [15]. Another possible explanation could be the release of NPs after rupture of endosomal compartment. It was described that cationic NPs, after binding to lipid groups on the surface cell membrane, could be endocysed in vesicles and accumulated into the lysosomal compartment. Within, they are able to sequestrate protons which could lead to the activation of proton pump and further rupture of the ion homeostasis and lysosomal accumulation of water. The subsequent lysosomal swelling and membrane rupture leads to the cytoplasmic release of NPs [16]. In proliferating cells, these cytoplasmic NPs associated or not to microtubules, could enter in the nucleus during the mitosis, and could explain that non soluble NPs were observed in the nucleus [14]. More rarely, NPs were also observed within the mitochondrial matrix but, so far, no explanation was given to explain this organelle localisation.
The Cellular Stress Induced by Nanoparticles and Its Biological Consequences Over the last 10 years of research conducted on the mechanisms of toxicity of non- intentional as well as engineered nanoparticles, has led to the establishment of a consensus within the scientific community of toxicologists to consider the central role of oxidative stress in cellular responses to NPs leading to inflammation or apoptosis [5, 17]. The concept of oxidative stress was developed for many years to explain dysfunctions leading to pathologies. Oxidative stress could occur when reactive oxygen species (ROS) are overproduced leading to an imbalance between ROS production and antioxidant defence capacity. It could also occur when the organism shows a deficiency in antioxidant systems and, especially in antioxidant enzymatic systems (superoxide dismutase, catalase, and glutathione peroxidase). An increased concentration of ROS, exceeding the antioxidant capacity of the cells, can lead to oxidative damage at molecular or cellular level. ROS have important cellular roles either by acting as second messengers for the activation of specific pathways and gene expressions or by causing cell death. In the hierarchical oxidative stress model in
Cellular Mechanisms of Nanoparticle’s Toxicity
response to NPs, Nel et al. [5] propose that a minor level of oxidative stress leads to the activation of the antioxidant protection whereas, at a higher level, cell membrane and organelles injuries could lead to cell death by apoptosis or necrosis, but specific signaling pathways and gene expression are involved at each step. The induction of oxidative stress by several NPs is due to their ability to produce ROS (TiO2 for example) or to lead to their production. The surface properties of NPs modulate the production of ROS and the smaller they are; the higher is their surface area and their ability to react with biological components and to produce ROS. However, if this cellular induction appears to be general, all the NPs are not able to produce ROS and the cellular increase of the latter could be an indirect effect of the uptake. ROS interact nonspecifically with biological compounds; however, some macromolecules are more sensitive such as the unsaturated lipids, the amino acids with a sulphydril group (SH) and guanine sites in nucleic acids. When lipid bilayer is attacked by ROS, cascade peroxidation occurs leading to the disorganization of the membranes and of their functions (exchange, barriers, and information). The most sensitive proteins contain methionine or cysteine residues, especially in their active site and their oxidation could lead to modify their activity and even to their inactivation. The adaptive cellular responses to NPs are associated to the modulation of different redox-sensitive cellular pathways. Tyrosine kinases and serine/threonine kinase such as mitogen-activated protein kinases or MAP kinases were especially studied (ERK, p38, and JNK) in association with several transcription factors such as NFkB. The free radical can degrade the NFkB inhibitor IkB by the activation of the cascades leading to its proteolysis. The activation of NFkB induces its translocation within the nucleus and its link to consensus sequences in the promoter of numerous genes leading to their transcription. It is also true for other transcriptions factors such as AP1 and NrF2. The latter plays an essential role in the Antioxidant Response Element (ARE) mediated expression of phase 2 enzymes such as NQO1 (NADPH quinone oxidoreductase-1) and antioxidant enzymes such as heme-oxygenase-1 (HO-1). Indeed, HO-1 was found to be activated by CeO2 NPs exposure of human bronchial cells via the p38-Nrf-2 signaling pathway. The ability of NPs to interact with these
Cellular Mechanisms of Nanoparticle’s Toxicity
signaling pathways could partially explain their cytotoxicity. Recently, TiO2 and SiO2 NPs were demonstrated in vitro and in vivo to induce the release of IL1b and IL1a, two potent mediators of innate immunity, via the activation of inflammasome, a large multiprotein complex containing caspase 1 which cleaves pro IL1a and b in their active forms. These results lead to consider that these NPs could induce a potent inflammatory response. However, the mechanisms leading to this activation are not yet clear. Another important target of ROS produced by NPs is DNA. Oxidative damage of DNA could generate intra-chain adducts and strand breakage. The bond between the base and desoxyribose could also be attacked leading to an abasic site and the attack on the sugar could create a single strand-break. The genotoxicity of NPs begins to be studied and recent reviews pointed out the possible genotoxic mechanisms. However, oxidative stress appears now not sufficient to explain all the biological effects of NPs. The role of epidermal growth factor receptor (EGFR) was investigated by the group of K. Unfried with the demonstration that carbon black NPs induce apoptosis and proliferation via specific signaling pathways both using EGFR [18]. Carbon black NPs could also impair phagosome transport and cause cytoskeletal dysfunctions with a transient increase of intracellular calcium not associated with the induction of ROS since antioxidants did not suppress the response and which could be due to a direct effect on ion channels that control the calcium homeostasis in the cell [19]. Even if all the mechanisms are not completely demonstrated, it appears now that transmembrane receptors are implicated in NP-induced cell signaling and could lead to specific biological responses to NPs.
Nanoparticles and Cell Death NPs have also been shown to induce either apoptotic or necrotic cell death in a variety of in vitro systems depending on the concentration and duration of exposure. This induction of cell death mechanism by NPs might act as basis of different pathologies and consequently it is important to understand NPs induced apoptosis pathways. Cells are able to undergo apoptosis through two major pathways, the extrinsic pathway with the activation of death receptors and the intrinsic
409
C
pathway with the central role of mitochondria, its permeabilization and the release of Cytochrome C leading to the activation of apoptosome. Recently, the permeabilization of lysosomal membrane was also shown to initiate apoptosis with the release of cathepsins and other hydrolases from the lysosomal lumen. The molecular pathways of apoptosis induction by carbon black and titanium dioxide NPs in human bronchial epithelial cells were recently studied. It was shown that the initial phase of apoptosis induction depends upon the chemical nature of the NPs. Carbon black NPs triggered the mitochondrial pathway, with the decrease of mitochondrial potential, the activation of bax ( a pro-apoptotic protein of the Bcl2 family), and the release of cytochrome C, and the production of ROS is implicated in the downstream mitochondrial events. Whereas TiO2 NPs induced lysosomal pathway with lipid peroxidation, lysosomal membrane destabilization, and catepsin B release [20], Lysosomal permeabilization has also been shown to be important in silica NPs induced apoptosis. These results point out the necessity of a careful characterization of the molecular mechanisms involved by NPs and not just describing the final outcome.
Future Directions of Research The interactions between nanomaterials and their biological target are essential to explain their biological effect and the interest of the recent researches on the cellular mechanisms induced by NPs is to take in account the specificity of the cells and of their microenvironment. The first step is the formation of the corona in biological fluids whose composition and affinity kinetics strongly depend on the characteristics of NPs and, especially, their size and surface reactivity. This coating of proteins influences the aggregation, the final size and, finally, the uptake of NPs via the interaction with the membranes, their specific receptors or lipid rafts. It could determine if the nanomaterial is bioavailable and if it induces or not adverse interactions. The central mechanism proposed to explain the biological response is the oxidative stress. However, this paradigm is debated because very similar oxidative stress effects observed in cellular models and induced by different particles could lead in vivo to different pathological effects. It is now obvious that oxidative stress is a common and nonspecific mechanism in toxicology and that the responses at the level of
C
C
410
the cell depend on the perturbation of the redox balance with a few number of induced signaling pathways. The different biological responses could depend on the tissue specificity which could lead to different diseases observed after occupational or environmental exposure to well known particles or fibers. Recent studies have also shown that NPs could develop a response without a direct contact with the cells but after an induction of secreted factors, it is the “bystander effect.” Small molecules such as purines could be increased at cytoplasmic level in response to NPs, transferred through the gap junctions within a tissue to activate specific receptors [10]. Moreover, NP-induced apoptosis was also demonstrated to be propagated through hydrogen peroxide mediated bystander killing in an in vitro model of human intestinal epithelium. These specific responses could explain the in vivo observed differences. Finally, the interactions of NPs with proteins, enzymes, cytokines, and growth factors, outside, or inside the cell lead to modify the functions of these proteins with a possible indirect pathological effect. The large variety of engineered NPs in the market and under development makes these studies very complex. However, the development of safe nanomaterials depends on better knowledge of these specific interactions
Cross-References ▶ Ecotoxicity of Inorganic Nanoparticles: From Unicellular Organisms to Invertebrates ▶ Genotoxicity of Nanoparticles ▶ In Vivo Toxicity of Carbon Nanotubes ▶ In Vivo Toxicity of Titanium Dioxide and Gold Nanoparticles ▶ Quantum-Dot Toxicity ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles ▶ Toxicology: Plants and Nanoparticles
References 1. Donaldson, K., Borm, P. (eds.): Particle Toxicology, p. 434. CRC Press, Boca Raton (2007) 2. Oberdorster, G., Oberdorster, E., Oberdorster, J.: Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles. Environ. Health Perspect. 113, 823–839 (2005)
Cellular Mechanisms of Nanoparticle’s Toxicity 3. Brunekreef, B., Holgate, S.T.: Air pollution and health. Lancet 360, 1233–1242 (2002) 4. Marano, F., Boland, S., Baeza-Squiban, A.: Particleassociated organics and proinflammatory signaling. In: Donaldson, K., Borm, P. (eds.) Particle Toxicology, pp. 211–226. CRC Press, Boca Raton (2007) 5. Nel, A., Xia, T., Madler, L., Li, N.: Toxic potential of materials at the nanolevel. Science 311, 622–627 (2006) 6. Unfried, K., Albrecht, C., Klotz, L.O., Mikecz, A.V., Grether-Beck, S., Schins, R.P.F.: Cellular responses to nanoparticles: target structures and mechanisms. Nanotoxicology 1, 52–71 (2007) 7. Nel, A.E., Madler, L., Velegol, D., Xia, T., Hoek, E.M., Somasundaran, P., Klaessig, F., Castranova, V., Thompson, M.: Understanding biophysicochemical interactions at the nano-bio interface. Nat. Mater. 8, 543–557 (2009) 8. Lynch, I., Salvati, A., Dawson, K.A.: Protein-nanoparticle interactions: what does the cell see? Nat. Nanotechnol. 4, 546–547 (2009) 9. Huang, Y.F., Liu, H., Xiong, X., Chen, Y., Tan, W.: Nanoparticle-mediated IgE-receptor aggregation and signaling in RBL mast cells. J. Am. Chem. Soc. 131, 17328–17334 (2009) 10. Bhabra, G., Sood, A., Fisher, B., Cartwright, L., Saunders, M., Evans, W.H., Surprenant, A., Lopez-Castejon, G., Mann, S., Davis, S.A., Hails, L.A., Ingham, E., Verkade, P., Lane, J., Heesom, K., Newson, R., Case, C.P.: Nanoparticles can cause DNA damage across a cellular barrier. Nat. Nanotechnol. 4, 876–883 (2009) 11. Marano, F., Hussain, S., Rodrigues-Lima, F., BaezaSquiban, A., Boland, S.: Nanoparticles: molecular target and cell signaling. Arch. Toxicol. 85, 733–741 (2011). Online May 10 12. Dawson, K.A., Salvati, A., Lynch, I.: Nanotoxicology: nanoparticles reconstruct lipids. Nat. Nanotechnol. 4, 84–85 (2009) 13. Rejman, J., Oberle, V., Zuhorn, I.S., Hoekstra, D.: Sizedependent internalization of particles via the pathways of clathrin- and caveolae-mediated endocytosis. Biochem. J. 377, 159–169 (2004) 14. Hussain, S., Boland, S., Baeza-Squiban, A., Hamel, R., Thomassen, L.C., Martens, J.A., Billon-Galland, M.A., Fleury-Feith, J., Moisan, F., Pairon, J.C., Marano, F.: Oxidative stress and proinflammatory effects of carbon black and titanium dioxide nanoparticles: role of particle surface area and internalized amount. Toxicology 260, 142–149 (2009) 15. Gratton, S.E., Ropp, P.A., Pohlhaus, P.D., Luft, J.C., Madden, V.J., Napier, M.E., Desimone, J.M.: The effect of particle design on cellular internalization pathways. Proc. Natl. Acad. Sci. U.S.A. 105, 11613–11618 (2008) 16. Xia, T., Kovochich, M., Liong, M., Zink, J.I., Nel, A.E.: Cationic polystyrene nanosphere toxicity depends on cell-specific endocytic and mitochondrial injury pathways. ACS Nano 2, 85–96 (2008) 17. Ayres, J.G., Borm, P., Cassee, F.R., Castranova, V., Donaldson, K., Ghio, A., Harrison, R.M., Hider, R., Kelly, F., Kooter, I.M., Marano, F., Maynard, R.L., Mudway, I., Nel, A., Sioutas, C., Smith, S., Baeza-Squiban, A., Cho, A., Duggan, S., Froines, J.: Evaluating the toxicity of airborne particulate matter and nanoparticles by measuring oxidative stress
Charge Transport in Self-Assembled Monolayers potential–a workshop report and consensus statement. Inhal. Toxicol. 20, 75–99 (2008) 18. Sydlik, U., Bierhals, K., Soufi, M., Abel, J., Schins, R.P., Unfried, K.: Ultrafine carbon particles induce apoptosis and proliferation in rat lung epithelial cells via specific signaling pathways both using EGF-R. Am. J. Physiol. Lung Cell. Mol. Physiol. 291, L725–L733 (2006) 19. Moller, W., Brown, D.M., Kreyling, W.G., Stone, V.: Ultrafine particles cause cytoskeletal dysfunctions in macrophages: role of intracellular calcium. Part. Fibre Toxicol. 2, 7 (2005) 20. Hussain, S., Thomassen, L.C., Feracatu, I., Borot, M.C., Andreau, K., Fleury, J., Baeza-Squiban, A., Marano, F., Boland, S.: Carbon black and titanium oxide nanoparticles elicit distinct apoptosic pathways in bronchial epithelial cells. Part. Fibre Toxicol. 7(10), 1–17 (2010). Online Apr.16
Cellular Toxicity ▶ Nanoparticle Cytotoxicity
Characterization of DNA Molecules by Mechanical Tweezers ▶ DNA Manipulation Based on Nanotweezers
Characterizations of Zinc Oxide Nanowires for Nanoelectronic Applications ▶ Fundamental Properties of Zinc Oxide Nanowires
Charge Transfer on Self-Assembled Monolayer Molecules ▶ Charge Transport in Self-Assembled Monolayers
Charge Transport in Carbon-Based Nanoscaled Materials ▶ Electronic Transport in Carbon Nanomaterials
411
C
Charge Transport in Self-Assembled Monolayers Jeong Young Park Graduate School of EEWS (WCU), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea
Synonyms Charge transfer molecules
on
self-assembled
monolayer
Definition Charge transport in self-assembled monolayers (SAMs) is the transport of an electron or a hole through an organized molecule layer which is bound to a substrate.
Overview Charge Transport Through Organic Molecules Significant studies on charge transport properties through organic molecules have been carried out in the general area of molecule-based and moleculecontrolled electronic devices, often termed “molecular electronics” [1, 2]. Self-assembled monolayers (SAMs) are composed of an organized layer of amphiphilic molecules in which one end of the molecule, the “head group,” shows a special affinity for a substrate [3]. SAMs also consist of a tail with a functional group at the terminal end, as seen in Fig. 1. Charge transport of organic molecules is usually limited by hopping processes and is therefore dominated by surface ordering. Self-assembled monolayers are a good model system of molecular electronics due to the ordered surface structure. In order to measure charge transport in a self-assembled monolayer, the substrate surface should be metallic. For example, a gold surface exhibits strong bonds with alkanethiol through S–H bonds. The other electrode should also be metallic for charge transport through the selfassembled monolayer. The measurement scheme of charge transport through a self-assembled monolayer
C
C
412
Charge Transport in Self-Assembled Monolayers
Functional group
R
R
R
R
S
S
S
S
Tail
Head group
Substrate
Charge Transport in Self-Assembled Monolayers, Fig. 1 Schematic of a self-assembled monolayer (SAMs) showing the head group that is bound to the substrate. SAMs consist of a tail with a functional group at the terminal
S
Metal electrode S S S S
S
parameter. Conformation and molecular orientation relative to the electrodes are also important. Other factors need to be considered as well, including energy positions of the highest occupied and lowest unoccupied molecular orbitals (HOMO, LUMO), electrode work function, and nature of the bonding to the electrodes. Charge transport mechanisms through selfassembled monolayers consist mainly of three processes [4]. The dominant charge transport mechanism in a molecular junction is “through-bond” (TB) tunneling, where the current follows the bond overlaps along the molecules (as illustrated in the left transport channel of Fig. 2). Another contribution involves the charge transport from electrode to electrode, in which the molecule plays the role of a dielectric medium that is called “through space” (TS), as illustrated in the right transport channel of Fig. 2. The last contribution of charge transport pathway involves a chain-to-chain coupling as illustrated in the middle of Fig. 2. As the molecular chains tilt, the decrease of the electron tunneling distance leads to a lateral hop between the neighboring molecular chains.
A
S
S
S
S
S
S
Substrate
Charge Transport in Self-Assembled Monolayers, Fig. 2 Scheme of charge transport mechanisms through selfassembled monolayers. The dominant charge transport mechanism in a molecular junction involves “through-bond” (TB) tunneling, and “through space” (TS) as illustrated in the left and right transport channel, respectively
that represents a conductor-molecule-conductor junction is shown in Fig. 2. Charge Transport Mechanism For insulating molecules, such as alkane chains, electron transport occurs via tunneling mechanisms. When such molecules are placed between electrodes, the junction resistance changes exponentially: R ¼ R0 exp(bs), with electrode separation s, where R0 is the contact resistance and b a decay parameter. In most experiments, the separation s is the length of the alkane chain. However, length is not the only important
Two Pathway Models If electron transport was determined purely by tunneling through the alkane chains, one would expect the value of b to equal zero, since the tunneling distance is the same for all tilt angles. The nonzero value of b indicates the existence of either intermolecular charge transfer or variations in the S-Au bonding as a function of tilt that affect the conductivity in an exponential way with angle. Slowinski et al. [4] proposed a two-pathway conductance model involving “through-bond” tunneling, and the “chain-to-chain” coupling. Assuming no effects due to changes in S-Au bonding, the first pathway is independent of tilt, while the second depends on the tilt angle. The tunneling current, thus, is given by It ¼ I0 expðbTB dÞ þ I0 ns exp½bTB ðd dCC tan YÞ expðbTS dCC Þ where It is the current at a specific tilt angle Y, d is the length of the molecule, ns is a statistical factor accounting for the number of pathways containing a single lateral hop as compared to those containing only through-bond hops, d is the diameter of the molecule chains, bTB and bTS are respectively through-bond and
Charge Transport in Self-Assembled Monolayers
S
S w1
S
S w1
S
413
S w1
Substrate
C
C S
S w1
S
S w2
S
S w3
Substrate
Charge Transport in Self-Assembled Monolayers, Fig. 3 Scheme of the measurement of the junction resistance for two different situations: (1) decreasing of the alkane chain (left part), and (2) the tilting of the alkane chain while
maintaining the same number of carbon atoms (right part), which will yield the resistance (1) per unit length of molecule or (2) tilting angle of the molecules, respectively
through-space decay constants. For example, in case ˚, of C16 alkanethiol molecule chains, dCC ¼ 4.3 A ˚ d ¼ 24 A, and ns ¼ 16, i.e., the number of carbon atoms in the molecule.
As one example, details on the preparation of an alkanethiol SAM will be described below. Gold substrates (200–300 nm of gold coating over 1–4 nm of chromium layer n glass) are prepared by butane flame annealing in air after cleaning in acetone, chloroform, methanol, and a piranha solution (1:3; H2O2:H2SO4). The resulting surface consisted of large grains with flat terraces of (111) orientation (sizes up to 400 nm) separated by monatomic steps. Flatness and cleanness were tested by the quality of the lattice-resolved images of the gold substrate. Two types of hexadecanethiol (C16) self-assembled monolayer can be formed on Au (111): complete monolayers of the molecules and islands of molecules covering only a fraction of the substrate. In the first case, the film was produced by immersing the substrate in 1 mM ethanolic solution of C16 for about 24 h, followed by rinsing with absolute ethanol and drying in a stream of nitrogen to remove weakly bound molecules. Incomplete monolayers in the form of islands were prepared by immersing the substrate in a 5 mM ethanolic solution of C16 for approximately 60 s, followed by rinsing. Samples consisting of islands facilitate the determination of the thickness of the molecular film relative to the surrounding exposed gold substrate. The molecular order of the islands improves with storage time at ambient conditions.
Decay Constant upon Shortening and Tilting of Molecules The junction resistance is dependent on electrode spacing for two different situations: (1) shortening of the alkane chain [5] and maintaining the same width (w) between chains and (2) tilting of the alkane chain but changing w [6, 7]. These measurements will yield the resistance per unit length of molecule or tilting angle of the molecules, respectively. The conductance decay constant b has already been measured using SAMs with different chain lengths when the separation between electrodes decreases as a function of the alkane chain length (the left image of Fig. 3). The decay constant, b, upon tilting of molecules can be measured using deformation with an AFM tip and simultaneous measurement of current (the right image of Fig. 3). This methodology will be described in the next section.
Basic Methodology Preparation of Self-Assembled Monolayer The organic molecular films on various types of substrates (conducting, semiconducting, or insulating substrates) have been prepared using techniques such as the Langmuir-Blodgett technique, dipping the substrates into solution with molecules, drop casting, or spin-coating [8].
Techniques to Measure Charge Transport in Self-Assembled Monolayers The current through a thiol SAM on a hanging Hg drop electrode can be measured in an electrochemical
C
414
solution. The current was measured as a function of the monolayer thickness that can be tuned by two methods: by changing the number of carbons in the alkane chain and therefore its length; or, expansion of the Hg drop such that the monolayer surface coverage was reduced and the molecules increased their tilt angle with respect to the surface. Slowinski et al. determined the ˚ and bTS ¼1.31/A ˚ by decay constants bTB ¼ 0.91/A both a fit to their experimental data and by independent ab initio calculations. Mercury drop expansion experiments by Slowinsky et al. have shown a dependence of the current through the alkanethiol monolayers on surface concentration, prompting the authors to suggest the existence of additional pathways for charge transfer, like chain-to-chain tunneling. Scanning tunneling microscopy and scanning tunneling spectroscopy have been used to reveal the atomic scale surface structure and charge transport properties of SAM layers [9, 10]. STM has been used to reveal various phases of surface structure and atomic scale defects, which could play a crucial role in the electrical transport. Conductance measurements were performed with a conductive-probe atomic force microscopy (CP-AFM) system. The use of AFM with conducting tips provides the ability to vary the load on the nanocontact and also opens the way for exploring electron transfer as a function of molecular deformation. A junction is fabricated by placing a conducting AFM tip in contact with a metal-supported molecular film, such as a selfassembled monolayer (SAM) on Au, as shown in Fig. 4. The normal force feedback circuit of the AFM controls the mechanical load on the nanocontact while the current–voltage (I–V) characteristics are recorded. The possibility to control the load on the contact is an unusual characteristic of this kind of junction and provides the opportunity to establish a correlation between the mechanical deformation and electronic properties of organic molecules. The normal force exerted by the cantilever was kept constant during AFM imaging, while the current between tip and sample was recorded. It is crucial to carry out the experiment in the low load regime so that there is no damage to the surface. This can be confirmed by inspection of the ˚ ngstrom depth sensitivity as well as by images with A the reproducibility of the current and adhesion measurements. If the measured conductance did not change at constant load and did not show timedependent behavior in the elastic regime, the tip
Charge Transport in Self-Assembled Monolayers
Conductive probe AFM
A
S
S
S
S
S
S
Substrate
Charge Transport in Self-Assembled Monolayers, Fig. 4 Scheme of conductance measurements of SAM with a conductive-probe atomic force microscopy (CP-AFM) system
experiences minimal changes during subsequent contact measurements.
Key Research Findings The molecular tilt induced by the pressure applied by the tip is one major factor that leads to increased film conductivity. By measuring the current between the conductive AFM tip and SAM as a function of the height of the molecules, the decay parameter (b) can be obtained [11]. Wold et al. studied the junction resistance as a function of load using AFM. The resistance was found to decrease with increasing load within two distinct power law scaling regimes [12]. Song et al. examined the dependence of the tunneling current through Au-alkanethiol-Au junctions on the tip-loading force [13]. It is found that the two-pathway model proposed by Slowinsky et al. can reasonably fit with the results, leading the authors to conclude that the tilt configuration of alkanethiol SAMs enhances the intermolecular charge transfer.
Charge Transport in Self-Assembled Monolayers Charge Transport in Self-Assembled Monolayers, Fig. 5 AFM images (200 nm 200 nm) of topography, and current images obtained simultaneously for a full monolayer of C16 on Au (111) surface. Lattice-resolved images of the film (inset in the left figure) reveal a lattice image of SAM (size: 2 nm 2 nm)
415 Topography
Current
C
Topography
Charge Transport in Self-Assembled Monolayers, Fig. 6 AFM images (500 nm 500 nm) of topographic, and current images, respectively, that were acquired simultaneously on hexadecylsilane SAM islands on silicon surface
C
Current
Silicon
Figure 5 shows topography and current images obtained simultaneously for a full monolayer of C16 on an Au (111) surface. The topographic image reveals the commonly found structure of the gold film substrate, composed of triangular-shaped terraces separated by atomic steps. Lattice-resolved images of the film (inset in the left figure) reveal a (√3 √3)-R30 periodicity of the molecules relative to the gold substrate. Qi et al. measured current–voltage (I–V) characteristics on the C16 alkanethiol sample for loads varying between 20 and 120 nN, and found that the current changes in a stepwise manner and the plateaus are associated with the discrete tilt angle of the molecules. A stepwise response of the SAM film to pressure has been observed previously in other properties such as film height and friction of alkanesilanes on mica and alkanethiols on gold. In order to measure the thickness of the selfassembled monolayer upon molecular deformation,
C16 alkylsilane
the SAM islands that partially cover the substrate can be used. The heights of the islands can be obtained from topographical AFM images, while charge transport properties of alkanesilane SAMs on silicon surface are measured using AFM with a conducting tip. In this manner, the load applied to the tip-sample contact can be varied while simultaneously measuring electric conductance. Figure 6 shows the topographic and current images, respectively, that were acquired simultaneously on hexadecylsilane islands on a silicon surface. The image size is 500 500 nm. The hexadecylsilane islands are 100–200 nm in diameter and have a height of 1.6 nm at the applied load of 0 nN (or effective total load of 20 nN). It is also clear that the current measured on the alkanesilane island is much smaller than that measured on the silicon surface. These changes were shown to correspond to the molecules adopting specific values of tilt angle relative to the surface, and explained as the result of methylene
C
416
Charge Transport in Self-Assembled Monolayers
–2 Tilting
ln [J (nA/nm2)]
–3 –4
d1 d2
–5 –6 –7 –8 10
Linear fit (b = 0.52 Å–1) Two pathway model 15
20
Height of C16 island (Å)
Charge Transport in Self-Assembled Monolayers, Fig. 7 Semilog plot of current density (nA/nm2) as a function of the height of the hexadecylsilane SAM islands on a silicon ˚ 1 was found for surface. A decay constant (b) ¼ 0.52 0.04 A the current passing through the film as a function of tip-substrate separation
groups interlocking with neighboring alkane chains. In the case of complete monolayers of alkanethiol SAM, the junction resistance (R) was measured as a function of the applied load [6]. These data were converted to current versus electrode separation by assigning each step in the current to a specific molecular tilt angle, following the sequence established in previous experiments. It was found that ln(R) increases approximately linearly with tip-surface separation, ˚ 1. Similar with an average slope b ¼ 0.57 ( 0.03) A measurement of the decay parameter upon the molecular tilts was carried out with a scanning tunneling microscope and simultaneous sensing of forces. By measuring the current as a function of applied load, ˚ 1 was a tunneling decay constant b ¼ 0.53 ( 0.02) A obtained [14]. In the case of hexadecylsilane molecules, the local conductance of hexadecylsilane SAM islands on a silicon surface was measured with conductiveprobe AFM. A semilog plot of current density (nA/nm2) was obtained as a function of the height of the hexadecylsilane SAM islands on a silicon surface, as shown in Fig. 7. A decay constant (b) ¼ 0.52
˚ 1 was found for the current passing through the 0.04 A film as a function of tip-substrate separation [7]. Figure 7 shows the best fit of the two-pathway model with the experimental current measurement as a function of the heights of molecule islands by using the fitting parameters of bTB and bTS that are 0.9 and ˚ 1, respectively. The good fit indicates that the 1.1 A
two-path tunneling model is a valid model to describe this observation. While saturated hydrocarbon chains mainly interact with each other via weak van der Waals forces, much stronger intermolecular p-p interactions can be present in organic films comprised of conjugated/hybrid molecules. This influences charge transport significantly [5, 15]. In a conductance AFM study of two SAM systems, Fang et al. revealed the role of p-p stacking on charge transport and nanotribological properties of SAM consisting of aromatic molecules [16]. The two model molecules chosen in this study are (4-mercaptophenyl) anthrylacetylene (MPAA) and (4-mercaptophenyl)-phenylacetylene (MPPA). In MPPA, the end group is a single benzene ring, while in MPAA it is changed to a three fused benzene ring structure. This structural difference induces different degrees of lattice ordering in these two molecular SAM systems. Lattice resolution is readily achieved in the MPAA SAM, but it is not possible for the MPPA SAM under the same imaging conditions, indicating the MPAA is lacking long-range order. However, it is important to note that even without long-range order, the stronger intermolecular p-p stacking in the MPAA SAM greatly facilitates charge transport, resulting in approximately one order of magnitude higher conductivity than in the MPPA SAM.
Future Directions for Research In this contribution, the basic concept of and recent progress on charge transport studies of organic SAM films formed by saturated hydrocarbon molecules and conjugated molecules has been outlined. Several techniques, including AFM, STM, and hanging Hg drop electrode, are used to elucidate the charge transport properties of SAM layers. A number of molecular scale factors such as packing density, lattice ordering, molecular deformation, grain boundaries, annealing induced morphological evolution, and phase separation play important roles in determining charge transport through SAM films. High resolution offered by scanning probe microscopy (SPM) is a key element in identifying and studying microstructures (e.g., molecular tilt, lattice ordering, defects, vacancies, grain boundaries) in organic films and their effects on electronic properties. Other advanced surface characterization techniques, such as SAM with nano-electrodes, in
Chemical Etching
combination with conductive-probe atomic force microscopy, and spectroscopic techniques such as ultraviolet photoemission spectroscopy (UPS) and inverse photoemission spectroscopy (IPES), could be promising venues to explore the correlation between microstructures and electronic properties of organic films.
Cross-References ▶ Atomic Force Microscopy ▶ Conduction Mechanisms in Organic Semiconductors ▶ Electrode–Organic Interface Physics ▶ Scanning Tunneling Microscopy ▶ Self-Assembly
References 1. Aviram, A., Ratner, M.A.: Molecular Electronics: Science and Technology. New York Academy of Sciences, New York (1998) 2. Reed, M.A., Zhou, C., Muller, C.J., Burgin, T.P., Tour, J.M.: Conductance of a molecular junction. Science 278, 252–254 (1997) 3. Ulman, A.: An Introduction to Ultrathin Organic Films from Langmuir-Blodgett to Self-Assembly. Academic, Boston (1991) 4. Slowinski, K., Chamberlain, R.V., Miller, C.J., Majda, M.: Through-bond and chain-to-chain coupling. Two pathways in electron tunneling through liquid alkanethiol monolayers on mercury electrodes. J. Am. Chem. Soc. 119, 11910–11919 (1997) 5. Salomon, A., et al.: Comparison of electronic transport measurements on organic molecules. Adv. Mater. 15, 1881–1890 (2003). doi:10.1002/adma.200306091 6. Qi, Y.B., et al.: Mechanical and charge transport properties of alkanethiol self-assembled monolayers on a Au(111) surface: The role of molecular tilt. Langmuir 24, 2219– 2223 (2008). doi:10.1021/la703147q 7. Park, J.Y., Qi, Y.B., Ashby, P.D., Hendriksen, B.L.M., Salmeron, M.: Electrical transport and mechanical properties of alkylsilane self-assembled monolayers on silicon surfaces probed by atomic force microscopy. J. Chem. Phys. 130, 114705 (2009) 8. Barrena, E., Ocal, C., Salmeron, M.: Molecular packing changes of alkanethiols monolayers on Au(111) under applied pressure. J. Chem. Phys. 113, 2413–2418 (2000) 9. Bumm, L.A., Arnold, J.J., Dunbar, T.D., Allara, D.L., Weiss, P.S.: Electron transfer through organic molecules. J. Phys. Chem. B 103, 8122–8127 (1999) 10. Xu, B.Q., Tao, N.J.J.: Measurement of single-molecule resistance by repeated formation of molecular junctions. Science 301, 1221–1223 (2003)
417
C
11. Wang, W.Y., Lee, T., Reed, M.A.: Electron tunnelling in self-assembled monolayers. Rep. Prog. Phys. 68, 523–544 (2005) 12. Wold, D.J., Haag, R., Rampi, M.A., Frisbie, C.D.: Distance dependence of electron tunneling through self-assembled monolayers measured by conducting probe atomic force microscopy: Unsaturated versus saturated molecular junctions. J. Phys. Chem. B 106, 2813–2816 (2002). doi:10.1021/jp013476t 13. Song, H., Lee, H., Lee, T.: Intermolecular chain-to-chain tunneling in metal-alkanethiol-metal junctions. J. Am. Chem. Soc. 129, 3806 (2007) 14. Park, J.Y., Qi, Y.B., Ratera, I., Salmeron, M.: Noncontact to contact tunneling microscopy in self-assembled monolayers of alkylthiols on gold. J. Chem. Phys. 128, 234701 (2008). doi:234701 10.1063/1.2938085 15. Yamamoto, S.I., Ogawa, K.: The electrical conduction of conjugated molecular CAMs studied by a conductive atomic force microscopy. Surf. Sci. 600, 4294–4300 (2006) 16. Fang, L., Park, J.Y., Ma, H., Jen, A.K.Y., Salmeron, M.: Atomic force microscopy study of the mechanical and electrical properties of monolayer films of molecules with aromatic end groups. Langmuir 23, 11522–11525 (2007)
Chem-FET ▶ Nanostructure Field Effect Transistor Biosensors
Chemical Beam Epitaxial (CBE) ▶ Physical Vapor Deposition
Chemical Blankening ▶ Chemical Milling and Photochemical Milling
Chemical Dry Etching ▶ Dry Etching
Chemical Etching ▶ Wet Etching
C
C
418
Chemical Milling and Photochemical Milling
Chemical Milling and Photochemical Milling Seajin Oh and Marc Madou Department of Mechanical and Aerospace Engineering & Biomedical Engineering, University of California at Irvine, Irvine, CA, USA
Agitator
Workpiece
Synonyms Chemical blankening; Photoetching; Photofabrication; Photomilling
Definition Photochemical milling (PCM), also known as photochemical machining, is the process of fabricating high precision metal workpieces using photographically produced masks and etchants to corrosively remove unwanted parts. This process is called wet etching in MEMS fabrication techniques and can be also applied to nonmetal materials. Wet etching, when combined with nanolithography, is a useful process to fabricate detailed nanostructures by extremely controlled removal (Fig. 1).
Overview Photochemical machining (PCM) produces threedimensional features by wet chemical etching (Fig. 2). PCM yields burr-free and stress-free metal products and allows for the machining of a wide range of materials which would not be suitable for traditional metal working techniques. PCM is also known as photoetching, photomilling, photofabrication, or chemical blankening [1]. There is a special type of photochemical milling that uses light for initiating or accelerating the wet etching process in metal or semiconductor materials. The combination of photoresists and wet etching enables the fabrication of very detailed structures with complex geometry or large arrays of variable etching profiles in thin (h
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 1 Geometry of a typical channel. In applications h 0 and the electric field is in the opposite direction if q < 0. In general, the electric field is a vector. This formula is called Coulomb’s Law and ee is called the electrical permittivity. The electrical permittivity is a transport property like the viscosity and thermal conductivity of a fluid. The electric field due to a flat wall having a surface charge density s in Coulomb on one side is directed m2 normal to the surface and has magnitude E¼
For a volume that contains a continuous distribution of charge, re, summing over all the charges using the definition of the integral, Gauss’s Law becomes
(7)
(12)
V
Using Eq. 12 and the differential form of the definition of the electrical potential, it follows that =2f ¼
re ee
(13)
This is a Poisson equation for the potential given the volume charge density. The combination of Eqs. 4 and 13 is called the Poisson-Nernst-Planck system of equations.
Electrochemistry Electrochemistry may be broadly defined as the study of the electrical properties of chemical and biological
C
C
466
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering
Computational Micro/ Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 2 The electric double layer (EDL) consists of a layer of counter ions pinned to the wall, the Stern layer, and a diffuse layer of mobile ions outside that layer. The wall is shown as being negatively charged and the z–potential is defined as the electrical potential at the Stern plane. # A.T. Conlisk used with permission
Stern plane + + + + +
+
-
+
+
+
+
+
+
+
+
+
+
+ ++ ++
+ +
+ +
z
+
+
+
+ +
+ +
+ + + + ++
f +
-
+
U +
-
+
Diffuse +layer
surface of shear
material [2]. In particular much of electrochemistry pertinent to micro and nanofluidics involves the study of the behavior of ionic solutions and the electrical double layer (EDL). Electrochemistry of electrodes is important to understand the operation of a battery. An ionic or electrolyte solution is a mixture of ions, or charged species immersed in a solvent, often water. It is the charged nature of ionic solutions that allows the fluid to move under the action of an electric field, provided by electrodes placed upstream and downstream of a channel in a nanopore membrane. The term membrane is used to mean a thin sheet of porous material that allows fluid to flow in channels that make up the porous part of the membrane. Those channels are often like those channels depicted on Fig. 1. Because the surface-to-volume ratio is so large in a nanoscale channel, the properties of the surface are extremely important. Fluid can be moved by an electric field if the surfaces of a channel are charged. If the surface is negatively charged, a surplus of positive ions will arrange themselves near the wall. This is shown in Fig. 2. It is this excess charge that allows fluid to be transported by an externally applied electric field. The nominal length scale associated with the EDL is the Debye length defined by l¼
pffiffiffiffiffiffiffiffiffiffi ee RT FI 1=2
(14)
where F is Faraday’s constant, ee is the electrical permittivity of the medium, I is the ionic strength P I ¼ i z2i ci , ci the concentrations of the electrolyte constituents at some reference location, R is the universal gas constant, zi is the valence of species i and T is the temperature. The ion distribution within the EDL can be described by using the number density, concentration, or mole fraction. Engineers usually prefer the dimensionless mole fraction whereas chemists usually use concentration or number density. There are two views of the ion distribution within the electrical double layer that are generally thought to be valid and have been verified by numerical solutions of the governing equations (see the section on Electrokinetic Phenomena below). The Gouy– Chapman [3, 4] model of the electric double layer allows counterions to collect near the surface in much greater numbers than coions. This model as numerical solutions suggest [1] occurs at higher surface charge densities. The Debye–H€uckel picture assumes that coions and counterions collect near the surface in roughly equal amounts, above and below a mean value. These pictures are depicted on Fig. 3.
Molecular Biology The nanoscale is the scale of biology since many proteins and other biomolecules have nanoscale dimensions. Many of the applications of nanofluidics such as rapid molecular analysis, drug delivery, and biochemical sensing have a biological entity as an integral part of their operation. Moreover, using nanofluidic tools, DNA sequencing is now possible. The book by Alberts [6] is a useful tool for learning molecular biology. Nucleic acids are polymers consisting of nucleotides. Those based on a sugar called ribose are called ribo nucleic acids (RNA) and those based on deoxyribose are called deoxyribonucleic acids (DNA). RNA is single stranded while DNA is usually double stranded although single-stranded DNA (ss-DNA) does exist. Nucleotides contain five-carbon sugars attached to one or more phosphate groups (a phosphorus central atom surrounded by four oxygens) and a base which can be either adenine (A), cytosine (C), guanine (G), or thymine (T). Two nucleotides connected by a hydrogen bond is called a base pair
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering a
b
g
g
467
C
C
f
f
y
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 3 (a) Debye– H€uckel [5] picture of the electric double layer. Here g denotes the cation mole fraction and f denotes the anion mole fraction. The Debye–H€uckel model assumes the cation and anion wall
(bp). Protein synthesis begins at a gene on a particular strand of a DNA molecule in a cell [6]. There are seven basic types of proteins classified according to their function although different authors use different terms to describe each class; see for example Alberts [6], panel 5–1. Enzymes are catalysts in biological reactions within the cell. For example, the immune system responds to foreign bacteria and viruses by producing antibodies that destroy or bind to the antigen, the foreign agent. The antigen is the catalyst, or reaction enhancer for inducing the immune response: the production of the antibodies. Proteins are responsible for many of the essential functions of the body, including moving material into and out of cells, regulating metabolism, managing temperature and pH, and muscle operation, among other functions. Proteins are large and complex molecules, polymers made up of a total of 20 amino acids, and held together by peptide bonds. The 20 amino acids have side chains that can be basic, acidic, polar or nonpolar. Because they are so large they cannot described easily in a single chemical formula or picture. Thus, molecular biologists depict proteins and other macromolecules in distinct levels of structure. The primary structure is the amino acid sequence, the order in which the 20 amino acids appear. The secondary structure depicts the folding properties of a protein as depicted on Fig. 4. Proteins are further described by more complex folding of the secondary structure (tertiary structure) and a quaternary structure if the protein has more than one backbone.
y
mole fractions are symmetric about a mean value which occurs for low surface charge densities. (b) Gouy–Chapman model [3, 4] of the EDL allows many more counterions than coions to collect near the charged surface and is valid at higher surface charge densities. # A.T. Conlisk used with permission
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 4 Ribbon view of the protein albumin depicting its folding pattern, the secondary structure of a protein. From the European Bioinformatics Institute, public domain, www.ebi.ac.uk
Proteins are usually negatively charged and thus nanopore membranes for rapid molecular analysis can be used to separate different types of proteins and other biomolecules based on different values of size and charge. Biomolecules are what is termed soft material in that they are porous and deform under stress. Indeed, recent measurements of the conformation of albumin show that it may take the shape of a wedge, looking like a piece of pie. With the explosive growth of computer capability, conformations of biomolecules are actually being computed using molecular simulation tools like molecular dynamics and Monte Carlo schemes.
C
468
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering
Ion channels are natural conical nanopores whose walls are made of proteins that play a crucial role in the transport of biofluids to and from cells. The basic units of all living organisms are cells. In order to keep the cells functioning properly there needs to be a continuous flux of ions in and out of the cell and the cell components. The cell and many of its components are surrounded by a plasma membrane which provides selective transfer of ions. The membrane is made up of a double layer of lipid molecules (lipid bilayer) in which proteins are embedded. Ion channels are of two categories: carrier and pore. The carrier protein channel is based on the binding of the transport ion to a larger macromolecule, which brings it through the channel. A pore ion channel is a narrow, water-filled tunnel, permeable to the few ions and molecules small enough to fit through the tunnel (approximately 10 A˚ in diameter).
Electrokinetic Phenomena As the scale of the channels in a nanopore membrane become smaller, pressure, the normal means for driving fluids through pipes and channels at macroscale (Fig. 1), becomes very difficult [1] since the pressure drop requires scales as h3 where h is the (nanoscale) channel height. Since in many applications the fluids used are electrically conducting, electric fields can be used to effectively pump fluid. Moreover electrically charged particles can move relative to the bulk fluid motion and thus species of particles can be separated. These electrokinetic phenomena are generally grouped into four classes [1]: 1. Electroosmosis (electroosmotic flow): the bulk motion of a fluid caused by an electric field 2. Electrophoresis: the motion of a charged particle in an otherwise motionless fluid or the motion of a charged particle relative to a bulk motion 3. Streaming potential or streaming current: the potential induced by a pressure gradient at zero current flow of an electrolyte mixture 4. Sedimentation potential: the electric field induced when charged particles move relative to a liquid under a gravitational or centrifugal or other force field By far the two most important of these phenomena are electroosmosis and electrophoresis and for the purposes of the theme of this entry, electroosmosis is discussed exclusively.
Cathode
Anode
Electric Double Layer
− +
− − − − − − − − − − − − − − − − − − − − + + + + + + + + + + + + + + + + + ++ + − + + + + + − + + + + + − + − − +− +
+
−
+ +
−
−
+
−
+
+
−
−
+
−
− + +
−
− + − + + + − + + + + + + + + + + − − − − − − − − − −
+ + − − +− + + + − + − + + + + + + + + + + + ++ + − − − − − − − − − − − −
Electric Double Layer
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 5 The combination of electrodes in the regions upstream and downstream of a charged channel or membrane, usually fluid reservoirs, causes electroosmotic flow. # A.T. Conlisk used with permission
The dimensionless form of the streamwise momentum equation in the fully developed flow region in the absence of a pressure gradient is e2
X @2u ¼ b zi X i @y2 i
(15)
and the Poisson equation for the potential in dimensionless form is e2
X @2f ¼ b z i Xi 2 @y i
(16)
where the partial derivatives in this one-dimensional fully developed analysis are really total derivatives, e ¼ lh and b ¼ cI where c is the total concentration including the solvent and I is the ionic strength. Here Xi is the mole fraction, but if the electrolyte concentrations are scaled on the ionic strength, P I ¼ lh i zi ci ; b ¼ 1. It is seen from Eq. 15 that the combination of the electrodes that create an electric field and the excess charge in the electrical double layers produces the electrical force that balances the viscous force causing the electrolyte to move and this is depicted on Fig. 5.
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering
a
ε = 0.01 ε = 0.05 ε = 0.1 ε = 0.5
4.5 Rescaled Mole Fractions Xi/X0i
1.4 Potential and Velocity
C
b
1.6
1.2 1 0.8 0.6 ε = 0.01 ε = 0.05 ε = 0.1 ε = 0.5
0.4 0.2 0 0
469
0.2
0.4
0.6
4 3.5
C
3 2.5 2 1.5 1 0.5
0.8
1
0
y
0.2
0.4
0.6
0.8
1
y
Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering, Fig. 6 (a) Potential, velocity, and rescaled mole fractions for a 1:1 electrolyte for various values of e. Here the dimensional potential on both walls
is z* ¼ 40 mV. In (b) the mole fractions are rescaled based on the upstream reservoir mole fractions as XX0i . The cations are i plotted in black lines and the anions are plotted in gray lines. # A.T. Conlisk used with permission
The fluid velocity satisfies the no-slip condition at the wall and the electric potential satisfies f(0) ¼ f(1) ¼ 0. Then both the equations and the boundary conditions are identical and on a dimensionless basis u(y) ¼ f (y). In reality, the potential does not vanish at the wall but if a Dirichlet boundary condition holds and the potential satisfies f ¼ z at y ¼ 0,1 and thus u ¼ f z, where z is the dimensionless z-potential at the wall. Results for the potential and velocity and the concentrations scaled on the ionic strength are presented on Fig. 6 [1]. Note that for E t (slow ambient temperature increase), the volumetric flow rate is proportional to the temperature change in the cavity. Compensation of Ambient Temperature Changes The effect of the compensation will be demonstrated for a cavity based on the data in Table 1. For the layout of the compensation, an ambient temperature increase of 10 K in 6 h, corresponding to an IR power density I0 of about 1 W/m2 for an adiabatic cavity, is assumed. A cavity without temperature compensation will build up a pressure difference of at least 900 Pa during 6 h, probably higher because of the smaller deflection of the membrane due to the nonlinearity of Eq. 2. As a consequence of this high pressure bias, the membrane will be out of the linear range (ymax/tP 1 [15] because it takes into account the membrane stress: ymax DP ¼ 64 D tP tP
R4 1 þ 0:488
ymax tP
2
(27)
The implicit Eq. 27 changes over to Eq. 2 with r ¼ 0 for ymax/tP ! 0. Figure 10 shows the comparison of the Eqs. 2 and 27. Obviously, Eq. 2 is only applicable for ymax/tP 300 mm), NPs can be considered as passive scalars advecting and diffusing in a quasi-Newtonian fluid. Under these conditions, a classical computational scheme, based on finite element methods (FEMs) and finite volume methods (FVMs), can be used quite efficiently to solve the transport problem. Recently, the isogeometric analysis (IA) has been introduced as a generalization of FEM which allows one to improve the functional representation of the geometry in the computational problem, simplifying the mesh refinement and increasing the accuracy of the solution [5]. In particular, patient-specific vascular models can be accurately constructed by identifying a set of procedures to create solid NURBS-based vascular models directly from patient-specific MRI/CT data (Fig. 3) [6]. This requires four main steps: preprocessing, where the quality of the original patient-specific data is improved
In Silico Mathematical Modeling An intravascularly injected NP navigates the circulatory system transported by the blood flow. During the journey, NPs interact with red blood cells (RBCs), the most abundant cells in the circulation; with cells of the immune system, both circulating monocytes and resident macrophages, such as the Kupffer cells in the liver and the splenic macrophages; and with endothelial cells (ECs), lining the vessel walls. Different computational tools can be employed to study the vascular transport and cell/NP interaction, depending on the complexity of the problem and desired accuracy of the solution.
I
1130
Integrated Approach for the Rational Design of Nanoparticles
NP and Cell transport in the vascular compartment
NP adhesion to the ECs
NP transcytosis
3D Design Maps for the 4S Problem
Predicting the adhesion and internalization dynamics of NPs
Organ specific accumulation of NPs
Parameters for authentically complex modeling
Refining the math tools Tumor Accumulation
RES Sequestration
1
2 5 3 4
Fluorescence
8
6 7
Cemera
Fabrication of authentically complex apparata
Monitor
IntraVital Microscopy
Integrated Approach for the Rational Design of Nanoparticles, Fig. 2 A schematic representation of the integrated approach for the rational design of nanoparticles.
In-vitro Micrfluidics Chips
Mathematical tools, in vitro experiments, and intravital microscopy are used synergistically to solve the 4S problem, presented in the quaternary diagram in the middle
Integrated Approach for the Rational Design of Nanoparticles
1131
I
Integrated Approach for the Rational Design of Nanoparticles, Fig. 3 Patient-specific modeling of the abdominal aorta. The original image is divided into 26 patches, and each color represents one different patch. The image to the right presents data on the flow field within the vessels (Adapted from [6])
I using imaging processing techniques; path extraction, where the skeleton of the arterial system is identified using Voronoi and Delaunay diagrams; control mesh construction, where hexahedral NURBS control meshes are generated by sweeping a templated quad mesh of a circle along the arterial path; solid NURBS construction, which allows one to generate a 3D exact representation of the patient-specific data, and, eventually, the reconstructed 3D geometry is used for the final numerical analysis. The IA framework has been applied successfully to solve several nonlinear and complex physical problems. It can be effectively used to model transport problems in the macrocirculation. A 3D Navier-Stokes solver for blood flow is coupled with a linear scalar advection/diffusion equation for studying the blood flow and transport of passive scalars within the authentically complex arterial vasculature. The spatial and temporal variations of the wall shear rate (WSR), wall shear stress (WSS), velocity profile, pressure field, and volumetric concentration of drug molecules and NPs can be estimated in patient-specific geometries. A classical problem solved within the IA framework is exemplified in Fig. 4, where a catheter is placed at the inlet of a coronary artery and releases drug molecules in the flow. The 3D Navier-Stokes solver allows one to quantify the magnitude of the velocity field within the vasculature in the presence of the catheter (top, right) at different time points. Using the derived flow field, the transport equation is solved for time and spatial evolution of the drug molecules released from
the catheter (bottom, right). The IA can be implemented with an efficient parallel-solution strategy. In the microcirculation (vessel diameter 2 is orders of magnitude larger than for a sphere (g ¼ 1), for a fixed V [14]. As g increases, the oblate spheroid tends in the limit to a thin disk. The large adhesion observed is explained by the larger surface area exposed (i.e., larger number of molecular bonds) to cells and lower hydrodynamic dislodging forces of a “squashed” spheroid compare with a sphere of equivalent volume. In order to analyze the interaction of NPs with RBCs, a more sophisticated computational approach is required, such as the immersed finite element method (IFEM). This technique was developed by integrating the concept of immersed boundary, finite element, and mesh-free methods. Following the IFEM framework, the immersed bodies (i.e., NPs and circulating cells) are modeled as flexible and highly deformable structures using a Lagrangian mesh, whereas the
outlet
syringe
flow deck inlet
pump gasket
vacuum petri dish
collagen on a coverslip
computer
microscope
Integrated Approach for the Rational Design of Nanoparticles, Fig. 8 Experimental setup for a microfluidic chip. The transport dynamics of NPs and cells can be studied under controlled biophysical conditions
fluid (plasma) is solved in a fixed Eulerian mesh. The cell/cell and cell/NP interactions can be modeled using a Morse potential, and Kramer’s reaction rate is employed for describing the ligand-receptor binding at the interface. Within the IFEM framework, the dynamics of immersed bodies is coupled with the surrounding incompressible viscous fluid but avoiding expensive remeshing or mesh adaptivity of the fluid domain. With this approach, the dynamics of NPs can be studied in whole blood, meaning in the presence of a large group of deformable RBCs under different flow conditions. An example is presented in Fig. 7, where NPs are dispersed within a train of RBCs moving in a capillary. The relative position of the RBCs and NPs, and their orientation can be estimated at different time points as blood flows from left to right.
Integrated Approach for the Rational Design of Nanoparticles
1135
I
Number of adhering particles 2500 2000 1500 1000 500 50 mm
10
Integrated Approach for the Rational Design of Nanoparticles, Fig. 9 Outcome of a typical MFC experiment. From left to right: bright field and fluorescence image of NPs
adhering to a layer of HUVECs, fluorescent image identifying the NPs, time-dependent adhesion of NPs under flow
I
Number of adhering nanoparticles
6 5 4
Integrated Approach for the Rational Design of Nanoparticles, Fig. 10 MFC analysis of NPs with different shapes. Discoidal NPs exhibit the higher propensity to marginate and firmly adhere to the chamber substrate under flow (Adapted from [19])
160 310 460 610 760 910 1060 Time
discoidal (r=2200 kg / m3; d=1.5 mm)
3 2 1.5 100. 70. 60. 50. 40. 30. spherical (r=2000 kg / m3; d=1.0 mm)
20. 15.
quasi-hemispherical (r=1840 kg / m3; d=1.6 mm) 5.
6.
7.
Finally, the cellular uptake of NPs is again affected by their size, shape, surface physicochemical properties, and mechanical stiffness. The size effect has been clearly shown for spherical particles, where the radius not only affects the uptake rate, in that the larger is the radius and the longer is the time for cell uptake, but it also affects the internalization mechanisms [15]. The characteristic time for NPs to be wrapped by a cell membrane can be related to their size, shape, and surface physicochemical properties. A mathematical model for receptor-mediated internalization based on an energetic analysis shows that a threshold radius
10.
15. 20. Wall Shear Rate
30.
40.
50.
exists below which particle uptake does not occur because it is energetically unfavorable. The same analysis shows that the surface physicochemical properties of the NP related to that of the cell membrane can dramatically increase or decrease the uptake rate [16]. A mathematical model has been developed to predict the rate of uptake for ellipsoidal particles as a function of their aspect ratio. Spherical or ovoidal particles can be more rapidly internalized by cells compared to elongated particles [17]. These results agree qualitatively well with experimental data [18].
I
1136
Microfluidic Chips and Intravital Microscopy Microfluidic chips (MFCs) and intravital microscopy (IVM) analysis are used to validate and refine the mathematical models. MFCs can be effectively used in conjunction with semianalytical models and IFEM for studying the NP behavior under flow. IVM is used for the final validation in vivo, for elucidating the mechanisms regulating NP dynamics, and for acquiring input parameters for the computational models. These include patient-specific data on the vascular geometry, blood flow condition, blood vessel thickness and diameter, expression of specific receptor molecules. The microfluidic chip apparatus presented in Fig. 8 can be used to estimate the rate of rolling and adhesion of nanoparticles and circulating monocytes [19]. This consists of (1) the MFC (inlet, flow deck, gasket, coverslip, and outlet), (2) a fluorescent confocal microscope, (3) a digital camera, (4) a pumping system, (5) connecting tubings, and (6) a PC. The chip is installed over the microscope stage. An external pumping system (syringe pump) is connected to the chip through tubings and connection kits. The working fluid (physiological solution with and without RBCs) is continuously infused into the chip through a main syringe pump at a constant velocity defined by the desired wall shear rate S, whereas a bolus of the sample particles is injected within a short period of time through an additional syringe. The chip is installed on the stage of the microscope, with the transparent slide upright facing the objectives. Lenses with different magnification ratios are used depending on the size of the region of interest (ROI), varying from about 100 100 mm2 for a dry 40 with digital focusing up to 2 2 mm2 with a 10 objective. Parallel plate flow chambers have been used to characterize the behavior of circulating cells, as leukocytes and platelets, and micro-/nanoparticles in vitro. By decorating the NP with fluorescent molecules, the transport along the chamber and the adhesive dynamics to the bottom cell layer can be analyzed over time. Based on the imaging analysis, the surface density of adhering NPs and the rate of adhesion and decohesion can be measured over time and compared with those derived theoretically. Images from a typical flow chamber experiment are given in Fig. 9, showing, from left to right, fluorescent beads adhering to the cellular layer (imaged in bright and fluorescent field)
Integrated Approach for the Rational Design of Nanoparticles
Integrated Approach for the Rational Design of Nanoparticles, Fig. 11 Complex vessel architecture imaged in a mouse using intravital video microscopy
and the final result with the variation of the number of beads adhering to a cell layer over time. The higher propensity of nonspherical particles to marginate and adhere more avidly to a substrate under flow has been also verified experimentally in vitro, as in Fig. 10, where the number of particles depositing over a layer of fibronectin under a laminar flow in a parallel plate flow chamber has been measured for three different particle geometries (sphere, quasihemisphere, and discoidal particles) and different shear rates S [19]. These experimental results support the notion that nonspherical particles can more effectively marginate toward the vessels and more avidly adhere under flow. Flow chambers can be heated up to 37 C or cooled down to 4 C to perform internalization experiments over time under flow or in a quiescent medium. Therefore, the same apparatus can be also used to measure the uptake rates of NPs as a function of their geometrical and physicochemical properties. Intravital microscopy (IVM) is used to collect dynamic, quantitative data in vivo with cellular and subcellular resolution. Techniques have been developed that allow the dynamics of individual cells and nanoparticles to be monitored and quantified in real time in live small animals [20]. A variety of organs can be studied acutely or longitudinally via surgical manipulations, including the aorta and arteries, the liver, spleen, and solid tumors. Nanoparticle and cell localization relative to blood vessels and specific cells of interest (as the ECs) can be visualized with the
Integrated Approach for the Rational Design of Nanoparticles
1137
I
Integrated Approach for the Rational Design of Nanoparticles, Fig. 12 Organ-specific and time-dependent accumulation of systemically injected NPs in tumor-bearing
mice. Intravital microscopy allows us to quantify the rate and level of accumulation of NPs in multiple organs (Adapted from [20])
selection of appropriate vital dyes and/or transgenic animal strains. The behavior of cells and nanoparticles within the microvasculature can be correlated to local flow conditions by simultaneously tracking fluorescently tagged red blood cells in the region of interest. Cell/particle trajectory and velocity can be quantified across multiple frames. The architecture of vascular networks can be determined using intravascularly injected fluorescent tracers, as tetramethylrhodamine isothiocyanate– labeled dextran (TRITC-dextran) with a molecular weight of 155 kDa. The tracers perfuse the vessels after tail vein injection, and images are taken at sufficiently short rates. A sample image of a vascular network from an intravital microscopy analysis is shown in Fig. 11. From the reconstructed network, the following morphological parameters can be estimated from off-line imaging analysis: volume density (volume of vessels per unit volume of tissue), vessel diameter, lineal density (length of vessels per unit volume of tissue), branch point density (number of vessel bifurcations per unit tissue volume), segment length (vessel length between branch points), and tortuosity index (segment path length/shortest distance between branch points). The permeability of the vascular walls can be assessed by injecting calibrated fluorescent nanoparticles with increasing size, ranging from 100 nm up to 1 mm. After clearance from the circulation, the extent of particle extravasation can be observed and the pore cutoff size determined as the range between the largest particle size that could extravasate and the smallest size that could not extravasate. In addition to this, the vascular permeability within a ROI can be quantified by analyzing the extravascular diffusion of fluorescent blood tracers, as fluorescein-labeled dextran. The permeability parameter
determined through the procedures described above can be used in the mathematical models. The dynamics of intravascularly injected NPs can be monitored in real time. In particular, the surface density of adhering particles can be measured within the ROI and compared with the mathematical predictions (Fig. 12) [20].
Future Directions The integrated approach for the rational design (IARD) will help in developing patient-specific therapeutic and imaging agents with superior efficiency and efficacy through the use of (1) multiscale computational tools, for the analysis of the vascular and extravascular transport of nanoparticles, cells, and molecules; (2) the fabrication of sophisticated fluidic chips, resembling the biological complexity of authentic vascular networks; and (3) in vivo imaging, for tracking the dynamics of circulating agents and their performance. Acknowledgments The author wants to acknowledge the collaboration with Dr. Thomas J.R. Hughes from the University of Texas at Austin, Dr. Yongjie Zhang from Carnegie Mellon University, and Dr. Shaolie Hossain at The Methodist Hospital Research Institute on the development of the IA for the analysis of molecule and nanoparticle transport. The author wants also to acknowledge the collaboration with Drs. Wing Kam Liu, TaeRin Lee, and Adrian Kopacz from Northwestern University on the development of the immersed finite element method for the analysis of RBC/nanoparticle interaction.
Cross-References ▶ Adhesion ▶ Nanoparticles
I
I
1138
Integrated Biosensors
References
19. Gentile, F., Chiappini, C., Bhavane, R.C., Peluccio, M., Cheng, M.C., Liu, X., Ferrari, M., Decuzzi, P.: Margination dynamics of non-spherical particles within micro-capillaries. J. Biomech. 41, 2312–2318 (2008) 20. van de Ven, A.L., Kim, P., Haley, O., Fakhoury, J.R., Adriani, G., Schmulen, J., Moloney, P., Hussain, F., Ferrari, M., Liu, X., Yun, S.H., Decuzzi, P.: Rapid tumoritropic accumulation of systemically injected plateloid particles and their biodistribution. J. Control. Release, Oct 26 (2011)
1. Ferrari, M.: Cancer nanotechnology: opportunities and challenges. Nat. Rev. Cancer 5, 161–171 (2005) 2. Peer, D., Karp, J.M., Hong, S., Farokhzad, O.C., Margalit, R., Langer, R.: Nanocarriers as an emerging platform for cancer therapy. Nat. Nanotechnol. 2, 751–760 (2007) 3. Decuzzi, P., Pasqualini, R., Arap, W., Ferrari, M.: Intravascular delivery of particulate systems: does geometry really matter? Pharm. Res. 26(1), 235–243 (2009) 4. Godin, B., Driessen, W.H., Proneth, B., Lee, S.Y., Srinivasan, S., Rumbaut, R., Arap, W., Pasqualini, R., Ferrari, M., Decuzzi, P.: An integrated approach for the rational design of nanovectors for biomedical imaging and therapy. Adv. Genet. 69, 31–64 (2010) 5. Bazilevs, Y., Calo, V.M., Zhang, Y., Hughes, T.J.R.: Isogeometric fluid–structure interaction analysis with applications to arterial blood flow. Comput. Mech. 38(4), 310–322 (2006) 6. Zhang, Y., Bazilevs, Y., Goswami, S., Bajaj, C.L., Hughes, T.J.R.: Patient-specific vascular NURBS modeling for isogeometric analysis of blood flow. Comp. Meth. Appl. Mech. Engin. 196(29–30), 2943–2959 (2007) 7. Lee, S.Y., Ferrari, M., Decuzzi, P.: Design of bio-mimetic particles with enhanced vascular interaction. J. Biomech. 42(12), 1885–1890 (2009) 8. Lee, S.Y., Ferrari, M., Decuzzi, P.: Shaping nano-/microparticles for enhanced vascular interaction in laminar flows. Nanotechnology 20(49), 495101 (2009) 9. Peskin, C.S., McQueen, D.M.: Computational biofluid dynamics. Contemp. Math. 141, 161–186 (1993) 10. Zhang, L., Gerstenberger, A., Wang, X., Liu, W.K.: Immersed finite element method. Comp. Meth. Appl. Mech. Engin. 193(21–22), 2051–2067 (2004) 11. Zhang, L.T., Wagner, G.J., Liu, W.K.: A parallelized meshfree method with boundary enrichment for large-scale cfd. J. Computat. Phys. 176, 483–506 (2002) 12. Pan, W., Fedosov, D.A., Caswell, B., Karniadakis, G.E.: Predicting dynamics and rheology of blood flow: a comparative study of multiscale and low-dimensional models of red blood cells. Microvasc. Res. 82(2), 163–170 (2012) 13. Sun, C., Munn, L.L.: Lattice Boltzmann simulation of blood flow in digitized vessel networks. Comput. Math. Appl. 55(7), 1594–1600 (2008) 14. Decuzzi, P., Ferrari, M.: The adhesive strength of nonspherical particles mediated by specific interactions. Biomaterials 27(30), 5307–5314 (2006) 15. Chithrani, B.D., Ghazani, A.A., Chan, W.C.: Determining the size and shape dependence of gold nanoparticle uptake into mammalian cells. Nano Lett. 6, 662–668 (2006) 16. Decuzzi, P., Ferrari, M.: Design maps for nanoparticles targeting the diseased microvasculature. Biomaterials 29(3), 377–384 (2008) 17. Decuzzi, P., Ferrari, M.: The receptor-mediated endocytosis of nonspherical particles. Biophys. J. 94(10), 3790–3797 (2008) 18. Champion, J.A., Mitragotri, S.: Role of target geometry in phagocytosis. Proc. Natl. Acad. Sci. U.S.A. 103(13), 4930–4934 (2006)
Integrated Biosensors ▶ CMOS MEMS Biosensors
Integrated Micro-acoustic Devices Tian-Ling Ren, Yu-Feng Wang and Yi Yang Institute of Microelectronics, Tsinghua University, Beijing, China
Synonyms Micromachined acoustic devices
Definition Integrated micro-acoustic devices are silicon-based acoustic transducers which are integrated with IC modules such as preamplifiers or analog-to-digital converters (ADCs).
Overview An integrated micro-acoustic device consists of an acoustic transducer and an integrated circuit (IC) module. The functional part is the acoustic transducer, and it can either be fabricated with the IC module on the same chip or be packaged with the IC module on the same substrate. The transducer part is fabricated on a silicon wafer by MEMS techniques. The fabrication process is usually compatible with IC fabrication. Due to the adoption of semiconductor technologies, integrated micro-acoustic devices have many
Integrated Micro-acoustic Devices
advantages such as small size, low cost, and high yield. In terms of applications, integrated micro-acoustic devices can be divided into micro-audio devices and micro-ultrasonic devices. The working frequency range of micro-audio devices is from 20 Hz to 20 kHz. Micro-audio devices include MEMS microphones, micro-receivers, and micro-speakers. Commercial MEMS microphones are very mature MEMS products and they have a wide variety of applications in mobile phones, MP3 and other consumer electronics. The working frequency range of micro-ultrasonic devices is typically greater than 20 kHz. They are also known as micromachined ultrasonic transducers (MUTs). Due to their miniature size and high performance, micro-ultrasonic devices have great potential to be used in medical imaging systems or distance measuring devices. According to the transduction principle, there are three types of integrated micro-acoustic devices: capacitive, piezoelectric, and piezoresistive. Major manufacturers producing integrated micro-acoustic devices include Analog Devices, Akustica (AKU200x), Knowles Electronics, and Memstech (MSMx).
Capacitive Micro-acoustic Devices Capacitive micro-acoustic devices are based on electrostatic principle [1–3]. The basic structure is a parallel-plate capacitor (Fig. 1). They convert sound waves into electrical signals by moving an active membrane separated from the silicon substrate by a very thin air gap. The top electrode and the highly doped substrate are conductive, and they form a parallel-plate capacitor. When sound waves hit the membrane, it vibrates, inducing an alternating current from the capacitor. When the capacitor is applied with an alternating voltage, the active membrane will move periodically by electrostatic force and transmit sound waves. According to the operation frequency, the capacitive micro-acoustic devices can be categorized as capacitive MEMS microphones and capacitive micromachined ultrasonic transducers. Capacitive MEMS Microphones Capacitive MEMS microphones are the most successful, commercial integrated micro-acoustic devices which work at the audio frequency (20 Hz–20 kHz).
1139 Top electrode
Gap
I
Membrane
Gap
Substrate (Bottom electrode)
Integrated Micro-acoustic Devices, Fig. 1 Basic structure of capacitive micro-acoustic device
The MEMS microphones are also called microphone chips or silicon microphones. The pressure-sensitive diaphragm is formed directly on a silicon chip by MEMS or CMOS techniques and is usually accompanied with IC modules such as preamplifiers or analog-to-digital converters (ADCs). Due to the adoption of semiconductor techniques, MEMS microphones have some advantages, compared with traditional ECMs (electret condenser microphones), including greater reliability, higher tolerance to mechanical vibration, smaller footprint and digital signal output. Akustica and Knowles are the first two MEMS microphone makers reporting volume shipments. In 2006, Akustica announced a singlechip digital MEMS microphone that has a downsized die that is smaller than 1 mm2 and Knowles announced a two-chip digital MEMS microphone that includes an ADC on its ASIC in the same year. Later in 2008, Analog Devices Inc. (ADI) and other companies started to enter the MEMS microphone market. A unique aspect of ADI’s design is a particle filter constructed by perforating a silicon cover over the microphone’s diaphragm. This structure can effectively prevent the dust from reaching the delicate silicon diaphragm and thus improves the performance. Due to the hot sales of smart phones and other consumer electronics, the demands of MEMS microphone increase rapidly. The Yole De´veloppement predicts that the shipment of MEMS microphones will exceed one billion units in 2013 [4]. Now the major manufacturers producing MEMS microphones are Wolfson Microelectronics (WM7xxx), Analog Devices, Akustica (AKU200x), Infineon (SMM310 product), Knowles Electronics, Memstech (MSMx), NXP Semiconductors, Sonion MEMS, AAC Acoustic Technologies, and Omron.
I
I
1140
Integrated Micro-acoustic Devices, Fig. 2 Piezoelectric micro-acoustic device of d33 mode
Integrated Micro-acoustic Devices Top electrode Bottom electrode
3
Piezoelectric layer 1 2 Substrate
Capacitive Micromachined Ultrasonic Transducers Capacitive micromachined ultrasonic transducers (cMUTs) are the kind of transducers that can convert ultrasound into electrical signal and transmit ultrasound when they are applied with an alternating voltage. cMUTs have been subject to extensive research for the last two decades. Although they were initially developed for air-coupled applications, today their main application space is medical imaging and therapy. The typical operation frequency of cMUTs used for medical imaging is from several to tens of megahertz. cMUTs, invented in the mid-1990s by the research group of Stanford University [5], have come a long way in the last two decades and recently reached the market for medical ultrasound imaging [6]. The research team at Stanford University is still one of the most active groups in the design, fabrication, and application of cMUTs [7]. In the early years of research in this field, the main focus was on the basic device fabrication and understanding device operation. Several fabrication processes based on standard surface micromachining techniques have been developed [8–12]. An alternative cMUT fabrication method based on wafer bonding was developed later in 2003 [13]. Equivalent circuit models for CMUTs have been developed to help the design of arrays for practical applications [14–16]. Finite element analysis has been used to understand transducer characteristics (especially cross talk issues) and to optimize transducer response [17–21]. Integration of ultrasonic transducer arrays and integrated circuits is highly desired in many applications, especially for transducer arrays with small elements such as in 2D arrays and arrays for use at the end of a catheter [7]. Several methods have been developed to integrate cMUT arrays with electronic circuits including monolithic integration [22–25] and
multichip integration [26, 27]. Recently, flexible cMUT arrays are successfully designed and fabricated using wafer-bonding and PDMS filling techniques [28]. Ultrasound imaging with a forward-looking 2D array and a ring array at the end of a catheter are achieved [7].
Piezoelectric Micro-acoustic Devices Piezoelectric micro-acoustic devices are based on piezoelectricity and converse piezoelectricity [29–31]. The most widely used piezoelectric materials are lead zirconate titanate (PZT), AlN, and ZnO. Most researches of the piezoelectric micro-acoustic devices focus on the supersonic range. These devices are called piezoelectric micromachined ultrasonic transducers (pMUTs). pMUTs can convert ultrasound into electric signal and will transmit ultrasound when applied with an alternating voltage. The pMUTs working at 40–80 kHz are used in positioning or distance measuring systems, and high frequency pMUTs (typically several or tens mega Hertz) are used for medical imaging systems. In terms of vibration mode, piezoelectric micro-acoustic devices can be divided into d33 mode and d31 mode. d33 Mode A piezoelectric micromachined ultrasonic transducer working at the d33 mode is an acoustic resonance device. The basic structure is shown in Fig. 2. When the sound waves hit the device, the strain of the piezoelectric layer will generate charges on the electrodes due to the piezoelectricity. According to the converse piezoelectricity, the electric field in direction “3” will cause strain in direction “3,” so the thickness of the piezoelectric layer will be changed periodically under a sine voltage signal, in other words, the layer vibrates.
Integrated Micro-acoustic Devices Integrated Micro-acoustic Devices, Fig. 3 Piezoelectric micro-acoustic device of d31 mode
I
1141 Top electrode Bottom electrode
Piezoelectric layer Supporting layer 3 Substrate 1 2
So the device will transmit sound waves under electrical stimulus. When the phase difference between the sound wave in the surface and the sound wave reflected from the bottom electrode is the multiple of 2p, the device will be in the resonance state. So the resonance frequency is decided by the thickness of the piezoelectric layer. The piezoelectric layer is the key part of pMUTs. The methods to deposit piezoelectric materials include sputter, pulse laser deposition (PLD), MOCVD, Sol–gel, and bulk ceramic bonding. The piezoelectric layers formed by these methods are either very thin (less than 10 mm) or very thick (more than 100 mm). The resonance frequency of pMUTs in d33 mode is decided by the thickness of the piezoelectric layer which means pMUTs of certain frequency bands are very difficult to fabricate because of the thickness gap between thin-film deposition and bulk ceramics. Research group of Technical University of Denmark proposed a novel method to deposit thick piezoelectric films [32]. In their work, the piezoelectric elements are formed by deposition of PZT paste into etched features of the silicon substrate such that the depth of these features determines the element thickness and hence the resonance frequency. The process leaves a nearplanar surface which is ideal for further wafer-level processing such as top electrode and interconnect formation. Using this method, they fabricated pMUTs with piezoelectric layer thicknesses ranging from 15 to 50 mm. The pulse-echo experiments on these transducers demonstrated their abilities to emit and receive ultrasound waves. d31 Mode A piezoelectric micromachined ultrasonic transducer working at the d31 mode is a mechanical resonance device. The basic structure of the piezoelectric micro-
acoustic device at d31 mode is a multilayer membrane which consists of a top electrode, a piezoelectric layer, a bottom electrode, and a supporting layer (Fig. 3). The supporting layer can be either silicon or silicon dioxide. When sound waves hit the membrane, the membrane will deform. Thus there will be charges induced on the electrodes due to piezoelectricity. According to the converse piezoelectricity, the piezoelectric layer will stretch or compress in direction “1” and direction “2” when it is exercised by an electric field in direction “3” with an opposite phase. So the multilayer membrane will bend periodically by an alternating voltage and transmit sound waves. The resonance frequency is determined by the thickness and size of the multilayer membrane. d31 mode is widely used because its design parameter is more flexible. In the early years of research in pMUTs, SiO2 or Si3N4 serves as the supporting layer. The membrane is released by bulk micromachining processing using wet etching. But it is difficult to control the thickness and residual stress of SiO2 or Si3N4, and these parameters will heavily influence the resonance frequency and the performance of the pMUTs. Researchers in the Research Triangle Park and Duke University use the device layer of SOI as the supporting layer, and the membrane is released using deep reactive ion etch (DRIE) [33]. They have successfully fabricated 2D pMUTs arrays using these techniques, and obtained B-mode images of three metal wires spaced 5 mm apart. Research group of Nanyang Technological University also uses the device layer of SOI as the supporting layer, but the piezoelectric layer is formed by a PZT/silicon waferbonding technique, and the bulk PZT is polished down to 10 mm using chemical mechanical polishing [34]. A perforated damping backplate is employed to broaden the acoustic frequency spectrum. Since the
I
I
1142
Integrated Micro-acoustic Devices
resonance of the pMUT array for medical imaging is very high (several to tens mega Hertz), it is necessary to reduce the spacing between elements in the array. Research group of Tsinghua University proposed a new fabrication process based on an SOI/Si waferbonding technique [35]. Shallow pits are etched on a Si wafer, then the device layer of an SOI wafer is bonded with the Si wafer, and after removal of the handle layer of SOI, the membrane is released. The element’s size and location are defined by the etched pitch. High element density and low cross talk are achieved by these techniques.
sc-Si [39]. They bond SOI with a Si wafer which is deposited with silicon nitride. Exfoliation of the handle layer of SOI is accomplished by heat-treatment at 550 C for 30 min. After that the device layer of the SOI which is sc-Si is transferred to the silicon nitride. The diaphragm is released from the front side through cavity etching holes using aqueous potassium hydroxide (KOH). This technique can effectively reduce the footprint. The mode shape and the resonance frequency of the microphone are measured and recorded using an optical vibrometer. For the fundamental mode, the resonance frequency is 520 kHz. The piezoresistive micro-acoustic sensors can also be used to detect sound source. Researchers in Korea Advanced Institute of Science and Technology developed a piezoresistive multicantilever microphone to locate the sound source by the ITD (interaural time difference) method [40]. In their method, the device layer of SOI wafer is served as the cantilever and boron-doped silicon is used as piezoresistor. Cantilevers with different lengths have different resonance frequencies, so the piezoresistive multicantilever microphone is a broad frequency device. Experiment results show that they can successfully detect the azimuthal angle of sound sources.
Piezoresistive Micro-acoustic Sensors
Summary
Piezoresistive micro-acoustic sensors are based on piezoresistive effect [36–38]. This kind of sensors cannot generate sound, they can only convert sound wave to electric signal. The basic structure of piezoresistive micro-acoustic sensors is shown in Fig. 4. The supporting layer can be either a diaphragm or a cantilever, and it will vibrate when it is hit by sound waves. The strain of the piezoresistive layer will cause the change of resistance due to piezoresistive effect. Polycrystalline Si (poly-Si) is the most common piezoresistor. Typically, a Wheatstone bridge configuration is adopted to convert resistance changes into voltage signals. Poly-Si piezoresistors on silicon nitride (SiN) were used in early demonstrations of piezoresistive micro-acoustic sensors. However, the gauge-factor of poly-Si is relatively low compared with single-crystal silicon (sc-Si). Research group of Hong Kong University of Science and Technology proposed a novel fabrication process to substitute poly-Si with
Capacitive micro-acoustic devices are the most mature micro-acoustic devices, mainly because of their excellent performances and ease of integration with IC modules. The capacitive MEMS microphones have already been commercialized. And the demands of MEMS microphones grow rapidly as the conventional electret microphones are substituted. Researches of piezoelectric micro-acoustic devices are mainly focused on piezoelectric micromachined ultrasonic transducers. Although a lot of prototypes with good performance and ability for medical imaging have been fabricated by researchers, it is still not commercialized yet. This is partially because the deposition of the piezoelectric materials is not so compatible with the standard IC process. Different from the capacitive and piezoelectric micro-acoustic devices which can convert sound wave and electric signal into each other, the micro-acoustic devices based on piezoresistive effect cannot generate sound wave but can only be acoustic sensors.
Piezoresistor Wire
Supporting layer
Substrate
Integrated Micro-acoustic Devices, Fig. 4 Basic structure of piezoresistive micro-acoustic device
Integrated Micro-acoustic Devices
1143
Cross-References ▶ Piezoresistivity
15.
16.
References 17. 1. Jin, X.C., Ladabaum, I., Khuri-Yakub, B.T.: The microfabrication of capacitive ultrasonic transducers. J. Microelectromech. Syst. 7(3), 295–302 (1998) 2. Jin, X.C., Ladabaum, I., Degertekin, F.L., Calmes, S., Khuri-Yakub, B.T.: Fabrication and characterization of surface micromachined capacitive ultrasonic immersion transducers. J. Microelectromech. Syst. 8(1), 100–114 (1999) 3. Knight, J., McLean, J., Degertekin, F.L.: Low temperature fabrication of immersion capacitive micromachined ultrasonic transducers on silicon and dielectric substrates. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 51(10), 1324–1333 (2004) 4. Ding, W.: MEMS Microphone, Market & Technology Report, Yole De´veloppement (2011) 5. Haller, M.I., Khuri-Yakub, B.T.: A surface micromachined electrostatic ultrasonic air transducer. In: Proceedings of the IEEE Ultrasonics Symposium, Cannes, vol. 2, pp. 1241–1244 (1994) 6. Sako, A., Ishida, K., Fukada, M., Asafusa, K., Sano, S., Izumi, M.: Development of ultrasonic transducer “Mappie” with cMUT technology, MEDIX, vol. 51, pp. 31–34 (2009) 7. Khuri-Yakub, B.T., Oralkan, O.: Capacitive micromachined ultrasonic transducers for medical imaging and therapy. J. Micromech. Microengr. 22(5), 1–11 (2011) 8. Schindel, D.W., Hutchins, D.A.: The design and characterization of micromachined air-coupled capacitance transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 42, 42–50 (1995) 9. Haller, M.I., Khuri-Yakub, B.T.: A surface micromachined electrostatic ultrasonic air transducer. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 43, 1–6 (1996) 10. Eccardt, P.C., Niederer, K., Scheiter, T., Hierhold, C.: Surface micromachined ultrasound transducers in CMOS technology. In: Proceedings of the IEEE Ultrasonics Symposium, San Antonio, vol. 2, pp. 959–962 (1996) 11. Eccardt, P.C., Niederer, K., Fischer, B.: Micromachined transducers for ultrasound applications. In: IEEE Ultrasonics Symposium. Proceedings, Toronto, vol. 2, pp. 1609–1618 (1997) 12. Cianci, E., Foglietti, V., Caliano, G., Pappalardo, M.: Micromachined capacitive ultrasonic transducers fabricated using silicon on insulator wafers. In: Proceedings of the 27th International Conference on Micro- and Nano-Engineering, Grenoble, vol. 61–62, pp. 1025–1029 (2002) 13. Huang, Y., Ergun, A.S., Haeggstrom, E., Badi, M.H., KhuriYakub, B.T.: Fabricating capacitive micromachined ultrasonic transducers with wafer-bonding technology. J. Microelectromech. Syst. 12, 128–137 (2003) 14. Ladabaum, I., Jin, X., Soh, H.T., Atalar, A., Khuri-Yakub, B.T.: Surface micromachined capacitive ultrasonic
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
I
transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 45, 678–690 (1998) Caronti, A., Caliano, G., Iula, A., Pappalardo, M.: An accurate model for capacitive micromachined ultrasonic transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 49, 159–168 (2002) Lohfink, A., Eccardt, P.C.: Linear and nonlinear equivalent circuit modeling of CMUTs. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 52, 2163–2172 (2005) Bozkurt, A., Ladabaum, I., Atalar, A., Khuri-Yakub, B.T.: Theory and analysis of electrode size optimization for capacitive microfabricated ultrasonic transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 46, 1364–1374 (1999) Wojcik, G., Mould, J., Reynolds, P., Fitzgerald, A., Wagner, P., Ladabaum, I.: Time-domain models of MUT array cross-talk in silicon substrates. In: Proceedings of the IEEE Ultrasonics Symposium, Rome, pp. 909–914 (2000) Roh, Y., Khuri-Yakub, B.T.: Finite element analysis of underwater micromachined ultrasonic transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 49, 293–298 (2002) Yaralioglu, G.G., Ergun, A.S., Khuri-Yakub, B.T.: Finiteelement analysis of capacitive micromachined ultrasonic transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 52, 2185–2198 (2005) Bayram, B., Kupnik, M., Yaralioglu, G.G., Oralkan, O., Ergun, A.S., Lin, D., Wong, S.H., Khuri-Yakub, B.T.: Finite element modeling and experimental characterization of crosstalk in 1D CMUT arrays. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 54, 418–430 (2007) Cheng, X., Lemmerhirt, D.F., Kripfgans, O.D., Zhang, M., Yang, C., Rich, C.A., Fowlkes, J.B.: CMUT-IN-CMOS ultrasonic transducer arrays with on-chip electronics. In: Proceedings of the Solid-State Sensors, Actuators and Microsystems Conference, Denver, pp. 1222–1225 (2009) Noble, R.A., Davies, R.R., King, D.O., Day, M.M., Jones, A.R.D., McIntosh, J.S., Hutchins, D.A., Saul, P.: Low temperature micromachined cMUTs with fully-integrated analogue front-end electronics. In: Proceedings of the IEEE Ultrasonics Symposium, Munich, vol. 2, pp. 1045–1050 (2002) Daft, C., Calmes, S., da Graca, D., Patel, K., Wagner, P., Ladabaum, I.: Microfabricated ultrasonic transducers monolithically integrated with high voltage electronics. In: Proceedings of the IEEE Ultrasonics Symposium, Montreal, vol. 1, pp. 493–496 (2004) Gurun, G., Qureshi, M.S., Balantekin, M., Guldiken, R., Zahorian, J., Sheng-Yu, P., Basu, A., Karaman, M., Hasler, P., Degertekin, L.: Front-end CMOS electronics for monolithic integration with CMUT arrays: circuit design and initial experimental results. In: Proceedings of the IEEE Ultrasonics Symposium, Beijing, pp. 390–393 (2008) Wygant, I., Zhuang, X., Yeh, D., Oralkan, O., Ergun, A.S., Karaman, M., Khuri-Yakub, B.T.: Integration of 2D CMUT arrays with front-end electronics for volumetric ultrasound imaging. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 55, 327–342 (2008) Wodnicki, R., Woychik, C.G., Byun, A.T., Fisher, R., Thomenius, K., Lin, D.-S., Zhuang, X., Oralkan, O., Vaithilingam, S., Khuri-Yakub, B.T.: Multi-row linear
I
I 28.
29.
30.
31.
32.
33.
34.
35.
36. 37.
38.
39.
40.
1144 cMUT array using cMUTs and multiplexing electronics. In: Proceedings of the IEEE Ultrasonics Symposium, Rome, pp. 2696–2699 (2009) Zhuang, X.F., Lin, D., Oralkan, O., Khuri-Yakub, B.T.: Fabrication of flexible transducer arrays with throughwafer electrical interconnects based on trench refilling with PDMS. J. Microelectromech. Syst. 17(2), 446–452 (2008) Lee, S.S., Ried, R.P., White, R.M.: Piezoelectric cantilever microphone and microspeaker. J. Microelectromech. Syst. 5(4), 238–242 (1996) KO, S.C., Kim, Y.C., Lee, S.S., Choi, S.H., Kim, S.R.: Piezoelectric membrane acoustic devices. In: Proceedings of the 15th IEEE International Conference on Micro Electro Mechanical Systems, Las Vegas, pp. 296–299 (2002) Zhu, Y.P., Ren, T.L., Yang, Y., Wu, X.M., Zhang, N.X., Liu, L.T., Tan, Z.M., Wang, H.N., Cai, J., Wang, S.D., Li, Z.J.: Novel ferroelectrics-based micro-acoustic devices and their ultrasonic applications. In: Proceedings of the IEEE International Electron Devices Meeting, San Francisco, pp. 51–54 (2004) Pedersen, T., Thomsen, E.V., Zawada, T., Hansen, K., Lou-Moeller, R.: New technique for fabrication of high frequency piezoelectric micromachined ultrasound transducers. In: Proceedings of the IEEE International Ultrasonics Symposium, Roma, pp. 2115–2118 (2009) Dausch, D.E., Castellucci, J.B., Chou, D.R., von Ramm, O.T.: Theory and operation of 2-D array piezoelectric micromachined ultrasound transducers. IEEE Trans. Ultrason. Ferroelectrics Freq. Contr. 55(11), 2484–2492 (2008) Wang, Z.H., Miao, J.M., Tan, C.W.: Acoustic transducers with a perforated damping backplate based on PZT/silicon wafer bonding technique. Sens. Actuator. A-Phys. 149(2), 277–283 (2009) Wang, Y.F., Ren, T.L., Yang, Y., Chen, H., Zhou, C.J., Wang, L.G., Liu, L. T.: High-density pMUT array for 3-D ultrasonic imaging based on reverse-bonding structure. In: Proceedings of the IEEE 24th International Conference on Micro Electro Mechanical Systems, Cancun, pp. 1035–1038 (2011) Burns, F.P.: Piezoresistive semiconductor microphone. J. Acoust. Soc. Am. 29(2), 248–253 (1957) Schellin, R., Hess, G.: A silicon subminiature microphone based on piezoresistive polysilicon strain-gauges. Sens. Actuator. A-Phys. 32(1–3), 555–559 (1992) Schellin, R., Mohr, R.: A monolithically-integrated transistor microphone - modeling and theoretical behavior. Sens. Actuator. A-Phys. 37–38, 666–673 (1993) Zhou, Z.J., Rufer, L., Wong, M.: Aero-acoustic microphone with layer-transferred single-crystal silicon piezoresistors. In: Proceedings of the 15th International Conference on Solid-State Sensors, Actuators and Microsystems, Denvern, pp. 1916–1919 (2009) Kim, Y.C., Lee, S.S., Kim, J.Y., Park, Y.J.: Sound source localization using piezoresistive Multi-cantilever microphones. In: Proceedings of the IEEE Sensors, Atlanta, pp. 780–783 (2007)
Integration
Integration ▶ CMOS MEMS Fabrication Technologies
Integration of Nanostructures within Microfluidic Devices Meng Lian1, Bernadeta Srijanto2, Prachya Mruetusatorn3,4, Mitchel J. Doktycz1,2 and Scott T. Retterer1,2,3 1 Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA 2 Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN, USA 3 Department of Electrical Engineering and Computer Science, University of Tennessee Knoxville, Knoxville, TN, USA 4 Oakridge National Laboratory, Oak Ridge, TN, USA
Synonyms Multiscale fluidics; Nanofabrication; Nanofluidics; Nanomaterials; Plasmonics
Definition Nanostructures encompass a broad class of materials with characteristic or functional lengths below 100 nm. For the purposes of this entry, nanostructures will be defined more broadly to include structures with functional dimensions at the submicron scale. The distinguishing characteristic of the materials discussed here is that their integration within a microfluidic platform is attained through a combination of microfabrication and material synthesis processes that allow the unique functionality of the structure or array of structures to be incorporated directly within the microfluidic system of interest. Generally, nanostructures or nanomaterials are considered to be zero dimensional (nanodots or nanoparticles) or one dimensional (nanowires and nanorods). Conventionally two-dimensional materials, such as thin-films, are not considered nanostructures, though their synthesis is fundamental to microfabrication and
Integration of Nanostructures within Microfluidic Devices
nanostructure modification. This entry includes examples of nanostructured thin films or patterned thin films into discussion as well as monolithic structures that are created using a combination of nano- and microfabrication techniques.
Concepts Nanostructured materials possess properties that are significantly different than their bulk counterparts. As a result, these materials have garnered a great deal of attention from the scientific community as well as the popular media. Increasingly, these materials are also finding their way into commercial products. Often, the material properties of these nanostructures will depend on the atomic structure of the material of interest, but are enhanced or perturbed by reductions in scale that amplify phenomena that occur at interfaces or result from material defects. Other trends in nanomaterial properties are material independent, relating to the scale of the material rather than its atomic composition. For example, the high surface area-to-volume ratio of nanostructured materials enhances the role that interfacial phenomena play in dictating the interplay between these materials and their environment. Thus, these materials are often the focus of interfacial studies that examine surface adsorption, wettability, and catalytic activity. The scale of these materials also makes them mechanically sensitive, allowing small changes in material mass to quickly impact both the static and dynamic properties of the material. One of the most captivating features of nanomaterials is that they exist and function at the same scale as the most basic elements of biological systems, providing a new approach to interfacing “hard” inorganic and “soft” biological materials. For example, nanoporous materials have pore sizes that approach the scale of biomolecules such as nucleic acids and proteins. This can facilitate both size and charge selective transport as a means of controlling or modulating chemical communication in biological systems, creating a chemical or fluid interface to biology. Because nanomaterials are smaller than most biological cells, they can also provide a model physical interface for cellular systems. The surface chemistry and physical structure of such nanostructured materials can be tuned to influence cell attachment, growth, and even differentiation. As individual entities or disordered collections of objects, nanostructures clearly exhibit unique
1145
I
properties. Additional functionality can often be achieved when these materials are arranged into ordered ensembles of interacting components. For many applications, ensembles of nanostructures have been created to simply expand the overall coverage and effective size of the nanomaterial array. In such cases the result of the ensemble is to simply scale the phenomena exhibited by any one element over a larger area. Alternatively, the ensemble can exhibit new properties that one might not predict based on the behavior of a single element. Such collective phenomena are often dependent upon the arrangement of materials within the system. Interactions, exchange or coupling of materials, can be tuned or modulated through discrete patterning and assembly techniques that allow control of relative distance and orientation of nanostructures. Because the local environment can dramatically impact such exchanges, changes in the environment can affect the collective behavior of the ensemble. Thus, changes in the collective behavior of ensembles of nanostructures may be indicative of changes in the local environment, making them ideal sensors. Because of their scale and enhanced functionality, nanostructure ensembles or arrayed nanomaterials are ideal for integration into microfluidic and Lab-on-achip (LOC) platforms. Their function can be particularly useful for platforms that are intended to interface with biological materials. As described above, these ensembles can be used as sensing elements to detect changes within the local fluid environment or within a biological system confined within the microfluidic platform. They can be used as passive elements to impart changes in the local environment, shaping fluid flow, molecular transport, reaction kinetics, and electromagnetic fields within the system. Furthermore, nanostructure ensembles can be designed to undergo directed changes, respond to changes in their environment, or be actively manipulated to change their functionality. Such alterations can cause local perturbations or lasting changes within the system. Altogether, ensembles of nanostructures provide a unique path toward integrating higher level sensing and actuation into LOC and microfluidic platforms.
Methodology One of the ongoing challenges of nanoscience is addressing the need to interface with the broad range
I
I
1146
of nanomaterials and nanoscale structures that one is capable of synthesizing. In order to take full advantage of the tremendous properties that these materials exhibit, one needs to develop technologies that allow the manipulation and interrogation of these materials at time scales that are relevant to the processes in which they are involved. Multiscale fabrication techniques combine conventional microfabrication technologies, honed through decades of integrated circuit and semiconductor manufacturing, with emerging methods for nanomaterial synthesis and assembly, to create integrated platforms in which functional nanomaterial ensembles can be created and taken advantage of. The strategies applied to the creation of nanostructures within microfluidic devices are lumped into two categories. Top-down strategies involve the step-by-step addition, patterning, and removal of bulk materials and substrate until a desired shape is achieved. Top-down techniques include microfabrication and patterning techniques conventionally used in the semiconductor industry, encompassing both surface and bulk micromachining processes. The resolution of the patterning techniques used to facilitate these processes can be used to create nanostructures that are as small as 5–10 nm, though the true limitations of such processes are dependent on structure composition. In contrast, bottom-up strategies involve the precise placement of nanoscale building blocks, template molecules, and catalysts at molecular levels to build two- and three-dimensional structures. High temperature or solution based synthesis techniques may be involved in the growth of nanostructures or nanomaterials from a patterned catalyst or initiator. The process usually involves the selfassembly of materials into energetically favorable arrangements with a very fine resolution (subnanometer range). Both covalent and non-covalent bonds have been utilized to order molecules into nanostructures. The primary challenge in bottom-up approaches is that by nature of the assembly and growth processes involved, long-range order can be difficult to achieve. Typically a combination of bottom-up and top-down approaches can be combined to integrate ensembles of nanoscale structures into functional microfluidic platforms (Fig. 1). There are three fundamental fabrication processes that are involved in the top-down approach. They are: deposition, lithography, and etching. Combination of
Integration of Nanostructures within Microfluidic Devices
these three processes in a layer-by-layer sequence allows for the creation of quasi-three-dimensional structures. As more layers are added, more complex architectures can be created. Deposition processes typically involve the creation of metal, dielectric, or semiconductor thin films using physical vapor deposition processes such as evaporation and sputtering, as well as chemical vapor deposition and oxidation. Common patterning techniques include photolithography, electron beam lithography, and X-ray lithography. These processes rely on the selective exposure and subsequent development of a typically polymeric “resist” material to create a protective layer for material etching processes. Emerging lithographic techniques include soft lithography, nanoimprinting and scanning probe lithography. Soft lithography and nanoimprinting are based on the use of a stamp made of elastomer or hard materials, such as etched silicon, to pattern a protective resist. These processes often require the use of a “master” or “mold” that is created using other lithographic techniques. Each of these techniques balances practical advantages and disadvantages, often sacrificing speed or up-front cost for higher resolution. Etching processes include wet chemical etching techniques and reactive dry processes, carried out in highly energetic plasmas. Detailed overviews of top-down manufacturing processes are provided elsewhere [1, 2]. With respect to the creation of nanoscale structures and ensembles within microfluidic devices, top-down processes can be carried out to create monolithic structures with both micro- and nanoscale features. This is accomplished using sequential electron beam and photolithographic processes to define nanoscale features and microfluidic channels in compatible resist materials respectively. Once patterned, a single directional etch process can be used to transfer this pattern into a silicon, quartz, or glass substrate. In contrast, bottom-up processes follow the approach of placing small components, at the molecular scale, into more complex assemblies. Predictable and well-controlled growth of nanostructures comprised of carbon (nanotubes, nanohorns, nanofibers) provides an example of such processes. In the case of nanofibers, for example, a catalyst material such as Ni is deposited on a surface. When heated and exposed to a carbon-rich environment, the carbon assembles into distinct cone-shaped sheets of graphene that stack to form nanofibers of carbon with a Nickel catalyst
Integration of Nanostructures within Microfluidic Devices
1147
I
a Hybrid Lithography and Etching (1) Photo- and electron beam lithographies are combined to define masking materials for both nanostructures and the microfluidic network
b Synthesis and Assembly
(2) Both “dry” and “wet” etching techniques can be used to pattern the multi-scale pattern into the substrate
(2.i) Assembly of the nanostructured substrate with a molded or etched fluidic network
(1) Nanomaterials can be deposited or patterned on a substrate. These materials may be used directly, or used to catalyze the growth of higher dimensional nanostructures (2.ii) The substrate with catalyst material may be exposed to appropriate growth conditions leading the nanostructure synthesis
c In situ synthesis (1) Nanomaterials can be deposited or patterned on a substrate. These materials may be used directly, or used to catalyze the growth of higher dimensional nanostructures
(2)
Assembly of the nanostructured substrate with a molded or etched fluidic network, and administration of growth conditions
(3) A glass or polymer top may be bonded to the etched substrate to create a sealed fluidic network
(3) An etched or molded fluidic network can be bonded to the nanostructured substrate. Alternatively a photopatternable polymer can be spin-coated, patterned capped with a glass or polymer lid
-or- Photopatternable polymer and lid
I
(3) Synthesis is followed with thorough washing and conditioning of the system for use.
Integration of Nanostructures within Microfluidic Devices, Fig. 1 Methods for integrating nanostructures into microfluidic networks
particle at the tip. When formed on a thin-film of Ni, these nanofibers grow as randomly arranged forests, derived from the dewetting of the Ni film and random formation of Ni catalyst particles during the initial heating process. Alternatively, top-down lithographic patterning processes can be used to pattern Ni into distinct shapes, that is, disks and lines, to confine the formation of nanofibers to specific positions. Ultimately it is this combination of top-down and bottom-up processes that provide the most promise for the integration of nanostructures into functional microfluidic devices. Two enabling approaches to the creation of microfluidic devices with nanostructured ensembles or elements have been the alignment and patterning of SU-8 polymer and assembly of polydimethylsiloxane (PDMS) microfuidic networks, created using soft lithographic techniques [3], to create microfluidic
networks that are aligned to nanomaterials synthesized or patterned on glass and silicon substrates. This approach requires careful alignment of microfluidic structures and precise control of nanostructure placement to achieve integration, making it a laborious and low-yield process. Nonetheless, it can provide the quickest route to creating multiscale fluidic systems. PDMS structures can be assembled directly on top of nanostructure ensembles, using an appropriate microscope, manipulator or steady pair of hands, and adequate alignment marks or guides. Alternatively, if the nanostructures are robust and can withstand a polymer spin-coating process, SU-8, or any other photopatternable polymer, can be spin-coated on the sample upon which the nanoscale structures are grown or deposited, and selectively exposed to yield a microfluidic network aligned to the existing nanostructure array.
I
1148
Applications A range of nanomaterials have been integrated into and used within microfluidic platforms to create complex sensing and actuation platforms for monitoring and modulating the environment within, and even around, the microfluidic or LOC device. One way to classify or examine these applications is to group them based on the use of nanomaterials as sensors, passive elements, or active modulators of their environment. The examples provided here are not intended to be comprehensive, but rather illustrative of the methods used to create these multiscale platforms, and the unique phenomena that can be taken advantage of in order to realize new levels of functionality within microfluidic platforms. Nanostructure Arrays as Sensing Elements The field of plasmonics, the study of charge density oscillations confined to the surface of metallic materials, has garnered a great deal of attention in recent years. Using nanostructured metal materials, including nanoparticle arrays or nanoporous thin metal films, scientists have utilized both plasmon resonance and Surface Enhanced Raman Scattering (SERS) techniques to develop highly sensitive biosensors [4]. Surface plasmon resonance results when plasmons are excited on the surface of metallic nanoparticles by an external electromagnetic field. The frequency at which adsorption, or excitation, occurs is highly dependent on particle size and distribution, particle spacing, and local refractive index. Accordingly arrays of metallic nanoparticles integrated within microfluidic devices have been used to detect the presence of specific molecules within a fluid sample. As these molecules adsorb, or are captured, onto the surfaces of arrayed nanoparticles, local changes in refractive index at the particle surface resulting in changes to the adsorption spectra of the array are indicative of the molecule’s presence. Hiep et al. [5] describe the utilization of gold-capped nanoparticles and nanopore arrays to achieve a highly sensitive platform for protein and gene sensors. Chemical modification of the arrayed nanoparticles or pores makes them sensitive to the presence of specific antigens. In this case an antibody to insulin was attached to the particles, and shifts in the absorption peak were observed as insulin, in varied concentrations, was pushed through the microfluidic device. Changes in the Local Surface Plasmon
Integration of Nanostructures within Microfluidic Devices
Resonance (LSPR) signal were recorded over time for concentrations as low as 100 ng/mL. Nanoparticle arrays were created via a random dispersion of amine functionalized silica particles onto an oppositely charged (COO terminated) gold-coated glass slide. A capping layer of gold was added to create the metallic nanoparticle arrays. Assembly of the capped nanoparticle arrays with a molded PDMS microfluidic channel allowed for facile production of the sensor platform. Many similar examples exist within the literature. Most of these utilize electron beam or colloidal patterning of thin metal films via lift-off or etching to create plasmonic materials comprised of metal nanostructure ensembles or porous metal films with discretely spaced and sized pores. Following production of the plasmonic material, SU-8 processing or assembly with PDMS channels can be used to integrate the material into a functional sensing platform. In an example that illustrates the power of combining top-down and bottom-up synthesis techniques, Kim et al. [6] describe the in situ synthesis of ZnO nanowire arrays by carrying out a controlled hydrothermal reaction within an assembled microfluidic device. In this process, zinc acetate dihydrate and ZnO thin films were used as texture seeds to catalyze ZnO nanowire growth within a microfluidic channel. Photolithography and lift-off processing were used to pattern ZnO seed films and local microheaters that helped drive localized formation of ZnO nanowires. Once patterning was completed, a PDMS channel was bonded to the substrate. Nanowire formation was demonstrated within the channels as either global or local heating was administered while flowing the nanowire precursor solution through the microfluidic device. Following in situ synthesis, the nanowire arrays were used to detect local changes in pH and physically trap particles against fluid flow. Nanostructures as Passive Elements Ensembles of nanoscale structures can be used to shape and perturb microflows, influence molecular transport and even modulate chemical reactions within microfluidic systems by simply acting as obstructions that can comprise “cages” or “sieves” within a microfluidic device. In addition to influencing physical flow within a microfluidic device, these obstructions can also be used to perturb or focus electromagnetic fields, creating field concentrations that can influence particle or cellular motion within
Integration of Nanostructures within Microfluidic Devices
the fluid flow. For example, Cao et al. [7] describe the use of nanoscale obstructions within a microfluidic device to improve the injection of biomolecules into sets of nanochannels, providing a transition from micro- to nanofluidics, preventing the accumulation and clumping of biomolecules at the nanofluidic inlets. Park et al. [8] developed three-dimensional interconnected pore networks inside microfluidic channels by combining single prism holographic and photolithography techniques. Arranged into a facecentered cubic lattice with a lattice constant of 650 nm, the pores continuously split and recombined the fluid flow to improve mixing efficiency by 84%. Hybrid lithography techniques that combine electron beam and photolithography along with anisotropic (directional) silicon etching have been used to simultaneously create arrays of nanoscale structures and microfluidic devices in silicon. These nanoscale structures include cellular-scale microreactors with nanoscale slits as well as arrays of silicon posts with controlled spacing. In both cases, modification of nanoscale structures using plasma enhanced chemical vapor deposition and vapor phase treatment with alkylsilane reagents was carried out to modulate the surface chemistry of the nanostructures and channel. Choi et al. [9] used the chemically functionalized pillar arrays to examine the effects of synthetic crowding and volume exclusion on the transport of green fluorescent protein (GFP) and “super-charged” GFP variants. Fluorescence recovery after photobleaching and computational modeling of molecular transport made it possible to quantify the effects of surface charge and molecular binding on diffusion within the nanostructured silicon post arrays. Retterer et al. [10] demonstrated the fabrication of nanoporous silicon bioreactors using hybrid lithography and etching processes, demonstrating the size selective retention of DNA and reactive enzymes within the nanostructured reactor while delivering small molecules, continuously, to the reaction system. Subsequently, Siuti et al. [11] demonstrated the use of the same reactor within a microfluidic network to carry out cell-free synthesis of GFP using cell-free extract and a GFP encoding plasmid. Protein synthesis was carried out for over 24 h under continuous flow of amino acids and ATP. Lavrik et al. [12] have used similar patterning and etching techniques to create arrays of nanopillars within microfluidic devices for use in chemical separations carried out on a microfluidic chip.
1149
I
In addition to shaping physical flow and diffusion processes, arrays of insulating nanostructures can be used to shape electromagnetic fields. Arrays of nanoscale elements can be used to “squeeze” the electric field generated through a microfluidic device into areas of high field over a short distance, creating a very high field gradient, and correspondingly high dielectrophoretic force [13]. The forces resulting from the constricted electric field can be used to concentrate bacteria for analysis or align and assemble nanostructures for further use. Nanostructures as Actuators Electrical and optical addressability of nanostructure arrays or ensembles makes it possible to monitor local changes within microfluidic systems by performing electrochemical measurements with nanofiber electrodes [14], monitoring the adsorption spectra of plasmonic materials [4, 15], or analyzing nanowire impedance [8]. While applications that utilize nanostructure ensembles as sensors are practical and appealing, just as intriguing is the possibility of using ensembles of nanostructures to actively and dynamically induce changes within the microfluidic system. Fletcher et al. [16] integrated patterned arrays of carbon nanofibers into microfluidic devices by patterning SU-8 microchannels onto arrays of nanofibers. Prior to patterning the SU-8 channels onto the device, an additional metal patterning step was used to create a set of electrodes that incorporated the carbon nanofibers. Subsequent electropolymerization of polypyrole onto the fibers allowed shrinking and swelling of the nanofiber profile via actuation of the polypyrole coating. Given the nanoscale gaps between the fibers, on-demand capture and release of fluorescent proteins within the microfluidic chip was demonstrated. Future work exploring the integration of functional polymer architectures that shrink or swell in response to changes in environmental factors such as pH or temperature may facilitate the development of smart systems that require no input from an outside user.
A Developing Field The convergence of nanoscale materials and microfluidics has created a rich field in which sensors and actuators that are created at the cellular and
I
I
1150
subcellular scale can be integrated into fluid environments that are ideal for the study and manipulation of biological systems [17–19]. This has allowed the development of biological interfaces that can be used to probe cellular systems over prolonged periods under dynamic conditions that were previously impossible to create, quantify, and visualize. This has led to new insights into biological function and the emergence of a host of techniques that promise to facilitate new levels of molecular diagnostics and therapeutic strategies that can be administered within the field and at the point-of-care. Acknowledgments The preparation of this entry was performed at the Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US DOE under Contract No. DEAC05-00OR22725.
Cross-References ▶ Biomimetic Synthesis of Nanomaterials ▶ Integration of Nanostructures within Microfluidic Devices ▶ Micro- and Nanofluidic Devices for Medical Diagnostics ▶ Nanomechanical Properties of Nanostructures ▶ Nanoscale Properties of Solid–Liquid Interfaces ▶ Nanostructured Materials for Sensing ▶ Nanostructures for Surface Functionalization and Surface Properties ▶ Self-assembly of Nanostructures
References 1. Kovacs, G.T.: Micromachined Transducers Sourcebook. McGraw-Hill, New York (1998) 2. Madou, M.: Fundamentals of Microfabrication. CRC Press, Boca Raton (2002) 3. Qin, D., Xia, Y., Whitesides, G.: Soft lithography for microand nanoscale patterning. Nat. Protoc. 5(3), 491–502 (2010) 4. Anker, J.N., Hall, W.P., Lyandres, O., Shah, N.C., Zhao, J., Van Duyne, R.P.: Biosensing with plasmonic nanosensors. Nat. Mater. 7, 442–453 (2008) 5. Hiep, H.M., Yoshikawa, H., Saito, M., Tamiya, E.: An interference localized surface plasmon resonance biosensor based on the photonic structure of Au nanoparticles and SiO2/Si multilayers. ACS Nano 3(2), 446–452 (2009) 6. Kim, J., Li, Z., Park, I.: Direct synthesis and integration of functional nanostructures in microfluidic devices. Lab Chip 11, 1946–1951 (2011)
Intelligent Drug Delivery Microchips 7. Cao, H., Tegenfeldt, J.H., Austin, R.H., Chou, S.Y.: Gradient nanostructures for interfacing microfluidics and nanofluidics. Appl. Phys. Lett. 81, 3058–3060 (2002) 8. Park, S.G., Lee, S.K., Moon, J.H., Yang, S.M.: Holographic fabrication of three-dimensional nanostructures for microfluidic passive mixing. Lab Chip 9, 3144–3150 (2009) 9. Choi, C.K., Fowlkes, J.D., Retterer, S.T., Siuti, P., Iyer, S., Doktycz, M.J.: Surface charge- and space-dependent transport of proteins in crowded environments of nanotailored posts. ACS Nano 4(6), 3345–3355 (2010) 10. Retterer, S.T., Siuti, P., Choi, C.K., Thomas, D.K., Doktycz, M.J.: Development and fabrication of nanoporous siliconbased bioreactors within a microfluidic chip. Lab Chip 10(9), 1174–1181 (2010) 11. Siuti, P., Retterer, S.T., Doktycz, M.J.: Continuous protein production in nanoporous, picoliter volume containers. Lab Chip 11, 3523–3529 (2011) 12. Lavrik, N.V., Taylor, L.C., Sepaniak, M.J.: Enclosed pillar arrays integrated on a fluidic platform for on-chip separations and analysis. Lab Chip 10(8), 1086– 1094 (2010) 13. Pohl, H.A.: Dielectrophoresis: The Behavior of Neutral Matter in Nonuniform Electric Fields. Cambridge University Press, Cambridge (1978) 14. Guillorn, M.A., McKnight, T.E., Melechko, A., Merkulov, V.I., Britt, P.F., Austin, D.W., Lowndes, D.H., Simpson, M.L.: Individually addressable vertically aligned carbon nanofiberbased electrochemical probes. J. Appl. Phys. 91, 3824–3828 (2002) 15. Malic, L., Veres, T., Tabrizian, M.: Nanostructured digital microfluidics for enhanced surface plasmon resonance imaging. Biosens. Bioelectron. 26(5), 2053–2059 (2011) 16. Fletcher, B.L., Retterer, S.T., Mcknight, T.E., Melechko, A.V., Fowlkes, J.D., Simpson, M.L., Doktycz, M.J.: Actuatable membranes based on polypyrrole-coated vertically aligned carbon nanofibers. ACS Nano 2(2), 247–254 (2008) 17. Kim, S.M., Lee, S.H., Suh, K.Y.: Cell research with physically modified microfluidic channels: a review. Lab Chip 8, 1015–1023 (2008) 18. Kumar, C. (ed.): Microfluidic Devices in Nanotechnology: Fundamental Concepts. Wiley, Hoboken (2010) 19. Kumar, C. (ed.): Microfluidic Devices in Nanotechnology: Applications. Wiley, Hoboken (2010)
Intelligent Drug Delivery Microchips ▶ Stimuli-Responsive Drug Delivery Microchips
Interface Excess Free Energy ▶ Surface Tension Effects of Nanostructures
Interfacial Investigation of Protein Films Using Acoustic Waves
Interfacial Charging Effect ▶ Maxwell–Wagner Effect
Interfacial Energy and Chemical Potential at Nanoscale ▶ Surface Energy Nanoscale
and
Chemical
Potential
at
Interfacial Investigation of Protein Films Using Acoustic Waves Hwall Min1, Ping Kao1 and Srinivas Tadigadapa2 1 The Pennsylvania State University, University Park, PA, USA 2 Department of Electrical Engineering, The Pennsylvania State University, University Park, PA, USA
Synonyms Density; Modulus; Self-assembled protein layers; Thickness; Viscosity
Definition Bulk acoustic wave quartz crystal resonator can be used for the investigation of the material properties such as the thickness, density, viscosity, and modulus of protein films adsorbed on their surface via selfassembly processes.
Overview Understanding the phenomenon of the aggregation and assembly of biological molecules such as proteins, lipids, and other biomaterials at the liquid–solid
1151
I
interface is of enormous practical consequence and is not yet well understood. For example, protein adsorption to materials is known to mediate the foreign body response to medical implants in the body. Bacterial adhesion to biomaterials has been found to be mediated by the adsorption of proteins such as fibronectin onto the material surface, creating an interface suitable for adhesion, proliferation, and biofilm formation in medical implants [1–5]. Furthermore, what appears to be driving the host response is not only the chemistry of the material, but the microstructure of individual features within the material onto which host cells can attach. Miniaturized bulk acoustic shear wave resonators exhibit high sensitivity to viscoelastic loading within a few tens of nanometers from the electrode surface and are ideally suited for in situ monitoring of the adsorption of proteins and characterization of their viscoelastic properties. While several techniques are currently available for studying biomolecular assembly processes at interfaces such as atomic force microscopy (AFM), optical reflectometry, surface plasmon resonance (SPR), and techniques based upon protein labeling (e.g., radioactive isotopes) most of these techniques are optimized to either identifying the molecular species or characterization of simple properties such as aerial mass or number density. Acoustic wave sensing techniques based on quartz crystal microbalances (QCM), in principle, are able to characterize not only the physical properties of the layer but also viscoelastic properties of the layer. As will be demonstrated in this article, the technique can be used to study adsorption of proteins on plane and textured surfaces with resolution approaching few angstroms to nanometers of the assembling layer.
Principle of Operation A typical thickness shear mode quartz crystal resonator is schematically illustrated in Fig. 1. It consists of a thin slice of single-crystal, piezoelectric quartz, with very large lateral dimensions in comparison to its thickness, which is sandwiched between two metal electrodes. Application of a sinusoidal electric field using the electrodes sets up a shear wave through the thickness of the quartz. Under the assumptions of ideal free-free boundary conditions of the electrode surfaces, the device exhibits a resonance behavior when the thickness of the quartz slab is half the wavelength
I
I
1152
Interfacial Investigation of Protein Films Using Acoustic Waves
AT-Cut Quartz
δ
Bottom Electrode Top Electrode Cross section view
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 1 Schematic drawing of a quartz crystal resonator showing the top and bottom electrodes. Application of a sinusoidal electrical stimulus across the electrodes sets-up
a shear wave across the thickness of the quartz and resonance occurs at the frequency for which the thickness of the quartz is equal to half the acoustic wavelength
of the shear acoustic wave. The fundamental resonance frequency of a quartz resonator can simply be given by
where e22(¼ 3.982 1011 F/m) is the permittivity of quartz, A is the area of the energized region of the quartz plate (electrode region), h is the thickness of the quartz plate, e26(¼ 0.095 C/m2) is the piezoelectric coefficient of quartz, mq(¼2.947 1010 N/m) is the piezoelectrically stiffened quartz elastic constant, and q(¼ 3.5 104 kg/m-s) is the viscosity of quartz. The exceptionally low phase-noise characteristically exhibited by quartz resonators allows for a precision of few parts in 1014 and is the primary reason for their widespread use in frequency control applications. The phase noise is quantitatively specified in terms of the Q-factor defined as the half-width at the full maximum of the resonance curve or in terms of an energy dissipation factor D, which is inversely proportional to the decay time constant and is also equal to the 1/Q-factor. The resonance frequency and the Q-factor of the resonance are significantly affected by any surface load placed on the quartz crystal. Generally speaking, loading on the resonator surface can be of three types: (1) pure elastic loads (rigid solids), (2) pure viscous loads (liquids), and (3) viscoelastic loads (polymers) and/or a sequential combination of any of these. The problem of interaction of the transverse shear wave with these surface loads has been treated using three fundamental approaches: (1) evaluation of electrical admittance characteristics upon mechanical perturbations on the resonators [7, 8], (2) using continuum mechanics with appropriate material properties and boundary conditions for each added material [9, 10], and (3) the energy transfer model in which the quartz resonator and the deposited film are treated as a compound resonator, and energy balance methods are used to evaluate the perturbed characteristics of the resonator [11]. These approaches are based on a one-dimensional treatment of the problem since
1 f0 ¼ 2tq
sffiffiffiffiffi mq ; rq
(1)
where tq is the thickness of the quartz slice and mq ¼ 2.947 1010 kg/m-s2 and rq ¼ 2,650 kg/m3 are the shear modulus and density of quartz, respectively. Figure 2a shows a typical resonance characteristic of a quartz resonator. Under the assumptions of (1) the quartz crystal is an infinitely large, thin plate, (2) shear wave traveling along the thickness of the quartz plate, (3) negligible material anisotropy, and (4) massless electrodes of negligible thickness, an equivalent circuit representation of the resonator can be realized and is shown in Fig. 2b. At resonance, the equivalent electrical circuit of the QCM is modeled as consisting of a capacitor C0 in parallel with the motional arm of the circuit consisting of a series circuit of Lm, Cm and Rm and is commonly referred to as the ButterworthVan Dyke (BVD) model. The values of these parameters can be derived and are given as [6]. e22 o A Static Capacitance h 9 8Ae2 > Cm ¼ 2 26 > > p hmq > > > > > 3 > rq h = Motional Arm Lm ¼ 8Ae226 > > > > > > hq p2 > > > ; Rm ¼ 2 8Ae26
C0 ¼
(2)
Interfacial Investigation of Protein Films Using Acoustic Waves
22 20
100
Phase
18
80
16
60
14
40
12
20
10
0
Rm
8
–20
6
–40
4
–60
2
–80
0 78.54
I
b
120
Impedance (Z)
Phase(°)
a
Impedance (kΩ)
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 2 (a) Typical resonance curve obtained for a micromachined quartz resonator. The thickness of the resonator is 21 mm. (b) The equivalent circuit representation of the quartz resonator at resonance shows the motional arm consisting of a series LCR circuit connected in parallel with a static capacitance C0
1153
78.56
78.58
78.6
–100 78.62
C0 Lm
Cm
Frequency (MHz)
I the lateral size of the resonators is typically much larger than the relevant dimensions such as the penetration depth of the acoustic waves in the direction perpendicular to the surface of the resonators. Elastic Loading The concept of measurement of aerial mass density using a resonating quartz crystal was presented in 1959 by Sauerbrey. For an unloaded resonator with a resonance frequency of f0(0), the decrease in frequency, Df, upon loading of a rigid mass of small thickness in comparison to the resonator thickness and under no-slip condition can be given by ! sffiffiffiffiffi! mq Dm 2f02 ð0Þ Dm 1 ¼
Df ¼ pffiffiffiffiffiffiffiffiffiffi rq mq A rq A ; t2q 2rq
(3)
where A is the area of the electrode on the quartz crystal and Eq. 1 has been used to simplify the formula. The most common commercially available quartz microbalances (QCMs) have a resonance frequency of 5–10 MHz and typically consist of 25.4–12 mm diameter and 333–167 mm thick AT-cut quartz discs with gold electrodes on each face and exhibit a mass sensitivity of 17.7–4.42 ng·cm2·Hz, respectively. From Eq. 3 it can be seen that the sensitivity to aerial mass density of the QCM can be increased by increasing its resonance frequency. Based on empirical data, the smallest frequency noise in an AT-cut quartz resonator has been approximated to be 12 1020f02.
Thus, the highest aerial mass-density resolution that can be achieved by a given QCM can be calculated by dividing this expression by the mass–sensitivity factor in Eq. 3, that is, pffiffiffiffiffiffiffiffiffiffi Dm ¼ 0:6 1020 rq mq ¼ 5:3 1015 ; A ultimate (4) where is Dm/A|ultimate is given in g/cm2. In finite electrode sized resonators, energy trapping is necessary to maintain a high Q-factor and is determined by the effectiveness of the total internal reflection of vibrations at the edge of the electrodes to confine the acoustic energy to within this region only. For the effective suppression of the spurious modes, the rule-of-thumb thickness of quartz to electrode radius ratio is given by Dfpb < 0:744 fq0
1 tq 2 ; N 2 rq
(5)
where fq0 is the expected resonance frequency of the bare quartz crystal of thickness tq before the mass loading due the electrodes, and Dfpb is the plate back frequency, that is, the frequency after the deposition of the top and bottom electrodes of radius rq, and N is the overtone number. This implies that as the quartz resonator thickness is reduced, the electrode diameter must also be reduced in order to effectively suppress spurious modes. Excessive decrease in the electrode diameter, however, will result in acoustic energy
I
1154
Interfacial Investigation of Protein Films Using Acoustic Waves
Rm
Lm
Cm
Rliq
Lliq
Liquid
Quartz C0
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 3 The equivalent model of a quartz resonator when loaded by a liquid of density rliq and viscosity Zliq. An
additional resistance and inductance term appear in the motional arm representing the effect of this load on the resonator characteristics. For a typical resonator, Rliq Rm and Lm Lliq
leakage and poor confinement of energy resulting in small Q-factors. These opposing requirements entail a design compromise. Results show that for 50–90 MHz resonators with 100 nm of gold electrodes on both faces, the optimal electrode diameter (2rq) to resonator thickness (tq) ratio is 10–20. Large deviations from this value either way result in resonators with poor characteristics.
The additional equivalent circuit elements due to the liquid loading can be written as
Lliq Rliq
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9 2rliq liq > > > > = omq rq > sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; > os Lm 2orliq liq > > > ¼ > ; Np m r
os Lm ¼ Np
(8)
q q
Viscous Loading When used for applications such as biosensing and electrochemical studies, where one of the resonator surfaces is in contact with a liquid medium, the viscous drag effects need to be considered and now determine the ultimate device performance. Behavior of quartz resonators under purely viscous loading was first reported by Kanazawa and Gordon in 1985. Under such conditions, the fundamental mode QCM frequency shift is given by 3=2
pffiffiffiffiffiffiffiffiffiffiffiffiffi f0 Df ¼ pffiffiffiffiffiffiffiffiffiffiffiffi rliq liq ; prq mq
(6)
where rliq and liq are the viscosity and density of the liquid, respectively. The change in the quality factor DQ of the resonator under viscous loading conditions can be given by rq m2q prq mq DQ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; 2 pf0 rliq liq 2pf0 q
(7)
For operation in liquids, the modified BVD model is shown in Fig. 3. Two additional terms arising from the liquid mass-loading and viscous damping appear in the motional arm of the circuit.
The expression for the ultimate aerial mass resolution of the QCM under liquid mass loading conditions is now modified and can be given as pffiffiffiffiffiffiffiffiffiffiffiffiffi rliq liq Dm 8 pffiffiffiffi ; ¼ 5:64 10 A ultimate f0
(9)
where all quantities are specified in SI units. For example, in pure water loading condition, the resolution now varies between 2.529 and 0.596 pg/cm2 for a 5 MHz to 90 MHz resonator. Clearly, the ultimate mass resolution in liquids is 102–103 times lower as compared to the case of mass resolution in air. In the case of liquid operation, it is quite clear that a high-frequency resonator is likely to provide a higher mass resolution. However, often in liquid applications, one is interested in measuring viscosity and density changes in liquid rather than adsorbed mass changes. In this case, the functional dependence of Df on the density and viscosity of the liquid, can be written as [12] Dliq Drliq þ ¼ 4 107 ; liq rliq
(10)
for the ultimate resolution of viscosity and density using a QCM. Since most liquid experiments are
Interfacial Investigation of Protein Films Using Acoustic Waves
accompanied by simultaneous changes in the density and the viscosity of the liquid, Eq. 10 gives the combined minimum resolution in both parameters. Furthermore, in liquid medium applications, the shear wave rapidly damps out as it travels through the thickness of the liquid, and consequently the QCM typically samples a layer of thickness equivalent qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi to the decay length dliq ¼ liq prliq f0 . For a commercial 5 MHz QCM, the typical decay length in water is 250 nm, which is large in comparison to the thickness of molecular or even polymer and biomolecular (e.g., proteins) films which can be in a thickness range of 10–50 nm. As a result, a commercial QCM not only samples the adsorbed film but is also significantly affected by the viscous liquid layer above it. Small changes in the viscoelastic properties of adsorbed layers are often masked by the viscous loading from the liquid overlayer. However, if the QCM thickness is decreased, its fundamental resonance frequency increases inversely as a function of the thickness of quartz, and thus the decay length in liquid also decreases consequently. For example, if the thickness of a quartz resonator is decreased from 330 mm (commercial 5 MHz resonator) to 24 mm(miniaturized 66–69 MHz resonator), the decay length in water is reduced to 68 nm. These factors are critical to the improved abilities of miniaturized devices for probing the viscoelastic properties of adsorbed layers compared to standard commercial QCMs.
Viscoelastic Loading in Liquid Ambient The QCM frequency shift resulting from the deposition of viscoelastic layer in a viscous liquid ambient can be analyzed using a continuum mechanics approach, for example, as developed by Kasemo and coworkers [10]. In order to model this situation, the QCM surface is considered to be in intimate contact with a layer of continuous viscoelastic slab with an infinitely thick Newtonian liquid overlayer on one of its surfaces. Under the assumption that the thickness of the bulk liquid layer is much larger than the decay length of the acoustic wave in the liquid, the frequency, Q-factor, and dissipation factor changes with respect to liquid loading conditions can then be written as
1155
Df
I
1 2prq tq
! 2 liq visc o2 tvisc rvisc o 2tvisc ; dliq m2visc þ o2 2visc (11) mvisc o 2tvisc 2 m þ o2 2visc visc ; DQ 2pf0 rq tq liq mvisc o 1 þ 2tvisc dliq m2visc þ o2 2visc (12) ! 2 liq 1 mvisc o DD 2tvisc ; (13) 2pf0 rq tq dliq m2visc þ o2 2visc where mvisc, visc, and rvisc are the elastic modulus, viscosity, and density of the viscoelastic layer, respectively, and tvisc is its thickness and dliq is the penetration depth of the acoustic wave in the liquid.
Basic Methodology Micromachined Quartz Resonator Arrays Figure 4a shows a schematic illustration of the fabricated resonator array with eight resonators in the array. The creation of an abrupt step in the quartz substrate for each pixel acoustically and energetically isolates each of these for the shear mode operation [13, 14]. Each resonator is individually addressed through backside (etched-side) electrodes which are extended to the rim of the sensor array chip. The top electrode is common to all the pixels of the array. In order to effectively confine the acoustic energy in each pixel, the pixel thickness to diameter of the electrodes ratio has been carefully optimized as per the conditions described in Eq. 4 [15]. The mounting and packaging of the array and the interconnection were also carefully optimized to maintain a high-quality factor of the resonators. Figure 4b shows the packaged device placed on the liquid test cell. For the fabricated 25–18 mm thick pixels, the fundamental resonance frequency is expected to be in the range of 66.72– 90 MHz. Fabricated devices were mounted in a specially designed Teflon test cell capable of liquid testing experiments.
I
I
1156
Interfacial Investigation of Protein Films Using Acoustic Waves
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 4 (a) Schematic illustration showing the topside and the bottom-side of the eight pixel resonator array used in this study. The top electrode is common to all the eight pixels
whereas the resonators are individually accessed using the independent bottom electrodes. (b) An optical photograph of a packaged sensor placed in the Teflon test cell
Key Research Findings
water-glycerol concentrations has been deduced and are listed Table 1 along with the equivalent circuit parameters deduced by nonlinear fitting.
Impedance Characteristics in Air and Liquid Loading Conditions Figure 5a shows the at-resonance real and imaginary impedance curves for one of the eight pixels of the array in air. All the fabricated pixels show similar impedance characteristics. The equivalent BVD circuit model for the quartz resonator yields the following expression for the real part of the impedance (Re Z):
Re Z ¼ where
Rm ð1 B0 Xm Þ2 þ ðB0 Rm Þ2
9 > > > > > =
> > > 1 > > ; B0 ¼ o C0 ; and Xm ¼ oLm oCm
;
(14)
The 801 data points of experimentally measured Re Z near resonance was fitted to the equivalent BVD circuit model of the resonator with C0, Cm, Lm, and Rm as parameters. Using nonlinear fitting function in Mathematica ® software, the best fit was used to obtain the values of the four equivalent circuit parameters. Figure 5b shows the experimental data and the fit obtained in air and water and the values of the resonator parameters obtained using this method. As can be seen, an excellent fit was obtained using this technique and the method was followed for all concentration of water–glycerol mixtures as well. Using this method and Eq. 8 the values of the (rliqliq)0.5 for the various
Investigation of Interfacial Layer: Protein Adsorption Studies Micromachined quartz resonators were used to measure the changes in the frequency and the Q-factor of the various pixels upon adsorption for five globular proteins as a function of concentration in the solution phase. The five proteins investigated are human serum albumin (HSA), immunoglobulin G (IgG), human fibrinogen (Fib), alpha-2-macroglobulin (AMG), and immunoglobulin M (IgM) at the fundamental and the third resonance modes. Human serum albumin (HSA), immunoglobulin G (IgG) from rabbit serum, and fibrinogen (Fib) were used as received from Sigma Aldrich and were of highest purity available (>96%). Alpha-2macroglobulin (AMG) and immunoglobulin M (IgM) were from MP Biomedicals and used without further purification. Different HSA, IgG, and fibrinogen concentration solutions were made by diluting a 10 mg/ml stock solution. AMG and IgM were made by diluting a 5 mg/ml stock solution. The 10 mM phosphate buffer solution (PBS) was made by dissolving phosphate buffered saline powder (Sigma Aldrich) into 18 MOcm DI water (Millipore Milli-Q system; Barnstead international). The micromachined quartz resonator pixels were cleaned by three cycles of exposure to ultraviolet (UV) light and ozone; each 15 min long and followed by thorough rinsing with ethanol and immersion in ethanol for 1 h.
Interfacial Investigation of Protein Films Using Acoustic Waves
b
24
8
21
4
12
0
9
4
6
–8
3
–12
0 78.57
78.58
78.59
78.6
78.61
Experimental Data (Air) Simulated Curve (Air) Experimental Data (Water) Simulated Curve (Water)
18
15
Im (Z) (kΩ)
Re (Z) (kΩ)
18
Re (Z) Im (Z)
12
Re Z (kΩ) (Air)
21
15 12
0.64 0.56 C0=1.53347 pF Cm=1.09791 fF Lm=3.73831 mH Rm=84.3483Ω
6
0.24 0.16
3
–16 78.62
0.08
0 78.2
78.3
Frequency (MHz)
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 5 (a) Real and imaginary components of the impedance of one of the pixels of the array at resonance. (b) The experimentally obtained real part of impedance and the fit
0.4 0.32
C0=1.53347 pF Cm=0.81316 fF Lm=5.05317 mH Rm=3.71869 kΩ
9
0.48
Re Z (kΩ) (Water)
a
I
1157
78.4 78.5 78.6 Frequency (MHz)
0 78.8
78.7
obtained to the data using non-linear regression to the expression in Eq. 14 is used to extract the motional circuit element parameters. The resulting simulated curve shows a very close fit to the obtained data in both air and water ambient
Interfacial Investigation of Protein Films Using Acoustic Waves, Table 1 The extracted equivalent circuit parameters of the quartz resonator in air and various liquid loading conditions. These values are used to calculate the density-viscosity product of various water-glycerol mixture concentrations and compared to the listed values in literature. The method used here has a maximum error of 28% pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi rliq liq rliq liq C (pF) C (fF) L (mH) R (O) R (O) %error 0
Air Water 10% glycerol 20% glycerol 30% glycerol 40% glycerol 50% glycerol
1.5332 1.5332 1.5332 1.5332 1.5332 1.5332 1.5332
m
1.09644 0.80954 0.77299 0.83761 0.81377 0.78477 0.78756
m
3.74238 5.07558 5.31636 4.90795 5.05300 5.24207 5.22758
m
liq
79.5620 3,725.65 3,697.47 4,392.42 5,076.26 6,282.81 8,232.40
The electrodes were exposed to 1 mM hexadecanethiol solution for 24 h. This procedure has been demonstrated to result in the formation of a hexadecanethiolate self-assembled monolayer (HD-SAM).Given the thin (2 nm), chemically attached, dense nature of the HD-SAM layer, it can be considered as a rigid extension of the electrode. The adsorbed globular protein layer attaches to the HDSAM surface via van der Waals forces with each molecule surrounded by water and in contact with the phosphate buffer saline solution. The protein layer is considered to be viscoelastic in nature whereas the surrounding PBS solution, containing only a very low-volume fraction of protein molecules, can be
3,646.08 3,617.91 4,312.86 4,996.70 6,203.25 8,152.84
Calc
1.2872 1.2195 1.5752 1.7729 2.1224 2.7988
List
1.0015 1.1575 1.3608 1.6379 2.0303 2.6103
28.52 5.36 15.75 8.24 4.53 7.22
represented as a Newtonian fluid. Furthermore, the change in frequency as well as the change in the Q-factor at the fundamental and third mode of the resonator upon adsorption for each protein at various concentrations was measured. Experimental results show a decrease in the frequency and the Q-factor at both the fundamental mode and third overtone of operation with increasing concentration for each protein. Using expressions given by Eqs. 11–13, the expected frequency and Q-factor changes can be calculated. In order to model the system, the thickness, density, elastic modulus, and viscosity of the protein layer are considered as parameters and varied to fit to the experimental data. Detailed results and analysis of
I
1158
0
0
–1
–4
b
–2
–8
–3
–12
–4
–16 st
Δf 1 mode Sigmoidal curve fit for Δf st ΔQ 1 mode Sigmoidal curve fit for ΔQ
–5 –6
ΔQ
Δf (KHz)
a
Interfacial Investigation of Protein Films Using Acoustic Waves
–20 –24
–7
–28
Dissipation Factor change (x10–5)
I
4.5 4
–9
Slope11st mode= –9.52x10 Hz
–1
3.5 3 2.5 2 –9
–1
–9
–1
Slope21st mode= –2.74x10 Hz
1.5 1
Slope31st mode= –2.36x10 Hz Data for
0.5
1st
mode
Sigmoidal curve fit for Δf
–8
0
2
4
6 8 10 12 14 16 Ln [HSA Concentration (pM)]
18
0
20
0 –7.5
–32
c
0
–6.5
2E+2 1E+2
–5.5
–4.5 –3.5 –2.5 Frequency Shift (KHz)
–1.5
–0.5
st
data for 1 mode rd
–30
–8
–60
ΔQ
Δf (KHz)
–4
Δf 3rd mode Sigmoidal curve fit for Δf
ΔQ 3rd mode
–12
–16
–90
Sigmoidal curve fit for ΔQ
0
2
4
6 8 10 12 14 16 Ln [HSA Concentration (pM)]
18
–120 20
Concentration/Δf (x10–10 M/Hz)
data for 3 mode after normalization
1E+1
5
Kfull=5.62x10 M
–1
1E+0 9
1E–1
Klow=1.22x10 M
–1
5
Kfull=7.33x10 M
–1
1E–2 9
1E–3 1E–4 0.0005
Klow=0.80x10 M
0.01
0.1
0.5
–1
2 3 5 1020
100
1000
Concentration (x10–7 M)
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 6 (a) Shows the experimentally measured frequency and quality factor shift upon adsorption of human serum albumin (HSA) in phosphate buffer solution on a hydrophobic CH3-terminated surface. The experimentally measured shifts at the fundamental mode (top) and third overtone (bottom) as a function of the concentration of HSA are shown. (b) Shows the dissipation factor (1/DQ-factor) plotted as a function of the frequency shift at the various concentrations
of HSA. Distinct regions with different slopes can be identified indicating a change in the viscoelastic behavior of the absorbed protein film. (c) Langmuir model plot of adsorbed HSA mass, represented by C/Df versus HSA concentration (C), for the entire concentration range tested. The long-dash line is a linear fit to all concentration range of data. The dotted line is a linear fit in the low concentration region of the data. The spread in the data with respect to the linear fit indicates a deviation from the fit only be seen in the low-concentration region
these measurements are presented elsewhere[16]. Here, the results are briefly summarized. Figure 6a shows a plot of the resonance frequency shift and quality factor change upon adsorption of HSA at the fundamental and the third overtone modes versus HSA solution concentration. All sets of data show a very close fit with high (>0.99) R2 value to a sigmoidal function. The observed change in the dissipation factor DD versus the change in frequency Df for various concentrations is plotted in Fig. 6b. The slopes reveal the value of ∂D/∂f in different concentration regions indicating a varying loss and rigidity of
the adsorbed film assembled on the resonator electrode at different concentrations. A low value of ∂D/∂f indicates a highly rigid, low dissipation film. In contrast, a high value of ∂D/∂f indicates a soft, dissipative film. In the low-concentration region, a small value of ∂D/∂f is obtained indicative of a low dissipation. It is reasonable to suggest that the acoustic wave is very slightly damped by protein molecules in low surface coverage region. With increasing surface coverage, the damping of acoustic wave by protein molecules continuously increases. In the high-concentration (surface coverage) region, the slope of ∂D/∂f is
Interfacial Investigation of Protein Films Using Acoustic Waves Interfacial Investigation of Protein Films Using Acoustic Waves, Table 2 Summary of the protein thickness, density and viscoelastic parameters obtained from the frequency and Q factor shift at the first and third mode of operation [16] Proteins HSA IgG Fib AMG IgM
Thickness (nm) 7.5 18 37.5 19 19.5
Density (kg/m3) 1,100 1,040 1,100 1,040 1,040
Viscosity (mPa-s) 3.5 5.5 2.7 4.8 2.3
Modulus (MPa) 1.8 6.7 0.42 1 2
markedly different from its value at low concentrations. This may be interpreted as gradual conformational changes during the self-assembly process upon reaching critical intermolecular interactions amongst the molecules in the adsorbing film. The reduction of ∂D/∂f near the saturation region implies a slightly rigid HSA film upon the completion of the adsorption process. The same experiments were repeated for other globular proteins listed and the analyzed using Eqs. 11–13. Table 2 summarizes the physical and viscoelastic properties of the various proteins that were extracted from the QCM measurements[16]. As a simple, preliminary way to model the protein adsorption trends, the proteins can be treated as spheres with a radius rv ¼ 6.72 108 MW1/3 where the volume of the core protein is considered to be proportional to their molecular weight [17]. For a core protein, including a shell of water molecules, the hydrated protein radius is 1.3 times the core protein radius. These values are in close agreement with the thickness per layer deduced from the QCM results. A linear relationship between the MW of the proteins and adsorbed capacity (density x thickness (in mg/cm2) of the protein layer) on the QCM electrode is observed for three of the five proteins, as shown in Fig. 7. Furthermore, normalization of the adsorbed capacity with respect to HSA (i.e., adsorbed capacity of given protein/adsorbed capacity of HSA) gives a simple integral number for the remaining two of the four proteins measured (see Fig. 7). HSA is known to adsorb on a surface as a single layer, which implies that the integral values obtained for two of the four proteins can be interpreted as adsorption of IgG as two layers and Fibrinogen as three layers. From the estimated value of the hydrated layer, AMG and IgM proteins are thought to adsorb as a single layer.
1159
I
The chemical and physical properties of the interface and ambient strongly influence the degree of protein adsorption at a solid interface. The relative strength of the thermodynamic adhesion of the protein to the CH3-SAM surface has been reported in the following order (Fib > Alb > IgG) [18]. Furthermore, the surface concentration of IgG adsorbed on SAMs has been shown to decrease in the order: CH3 > C6H5OH > COO > NH2 > OH [19]. Langmuir theory provides a general approach to discussing the adsorption kinetics of proteins by including the interactions between surface active sites and adsorbed species. However, extensive debates on the applicability of Langmuir model for protein adsorption can be found in literature. From the observed irreversibility of the macroscopic adsorption, as demonstrated by observations that dilution of the protein solution does not lead to a corresponding desorption of the protein films, it can be concluded that a large barrier to desorption exists. For example, addition of pure PBS solution on top of an adsorbed (saturated) protein layer for as long as 24 h showed no observable frequency and Q-factor change. Clark et al. reported similar observations where, the degree of reversibility of protein adsorption decreased with increasing contact time [20]. This desorption irreversibility has been attributed to timedependent interfacial conformational changes of the adsorbing protein molecules which result in a strong interaction with the surface. As a consequence of this irreversibility, for any given concentration of the protein solution, the adsorption process leads to saturation coverage after long enough time. However, at low concentrations of proteins, the adsorption process is extremely slow and requires very long times to approach saturation coverage – thus giving the appearance of a concentration-dependent surface coverage akin to Langmuir isotherm. Attempted modifications to Langmuir model of protein adsorption have not been very successful at explaining the kinetics of protein adsorption processes. For now, the Langmuir type of model remains the best way to empirically analyze adsorption data and has been used to interpret the results of HSA adsorption shown in Fig. 6c. It is important to realize that the high-frequency micromachined bulk acoustic resonators described here, with reduced acoustic penetration depths, are able to resolve the interfacial effects in the adsorbed films with high sensitivity.
I
I
1160
Interfacial Investigation of Protein Films Using Acoustic Waves 5
5
Adsorbed capacity for Various protein Normalized adsorbed capacity wrt HSA y=−0.01357+0.01212x
Adsorbed Capacity (μg/cm2)
4
4 Fib
3
3
2
2 AMG
IgG
IgM
1
1 HSA
0 0
200
400
600
800
Normalized Adsorbed Capacity (wrt HSA)
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 7 Absorbed capacity (density multiplied by thickness) versus protein molecular weight. Inset shows the core-shell packing construction
0 1000
Protein Molecular Weight (kDa)
Interfacial Investigation of Protein Films Using Acoustic Waves, Table 3 Summary of the protein thickness, density and viscoelastic parameters obtained from the frequency and Q factor shift at the first and third overtone of operation Surface Hydrophobic
Hydrophilic
Proteins HSA IgG Fib AMG IgM HSA IgG Fib
MW (kDa) 66.3 160 340 725 1,000 66.3 160 340
Protein diameter (nm) 7.1 9.5 12.2 15.7 17.5 7.1 9.5 12.2
In the log-log plots of C/Df versus concentration C for HSA, it is clear that only a few data points lie on a straight line. It appears that the data can be fitted by two linear regions of different slope and intercept. The calculated K value has a large variation in the two regions of concentration. To interpret these variations in the K value, the theory needs to include not only adsorption process, but also reorientation, spreading, spatial conformations, and their kinetics at the various concentrations. The DD versus Df graph, Fig. 6b, shows the dissipative effects in the adsorbing film as a function of concentration and relates more directly to reorientation, spreading, and spatial conformation effects. The presence of more than one linear region with different slopes as seen in a plot of C/Df versus concentration C and DD versus Df suggests
Thickness (nm) 7.5 18 37.5 19 19.5 7.5 30 50
Layer(s) 1 2 3 1 1 1 3 4
Thickness/layer (nm) 7.5 9 12.5 19 19.5 7.5 10 12.5
a competition between adsorption and dissipative processes which are expected to be determined by the native form of the protein. To compare the effect of surface functionalization on protein adsorption process, QCM gold electrodes were functionalized to produce a high-density carboxyl-terminated hydrophilic surface by chemisorption of 16-mercaptohexadecanoic acid. Identical frequency and Q-factor change measurements and analysis were undertaken for the adsorption of three globular proteins on the COO terminated electrode surface. The observed results show similar overall behavior for the adsorption of the globular protein layer on to COO surfaces as on hydrophobic surface. The measured change in frequency and Q-factor at the fundamental and third resonance modes were once
Interfacial Investigation of Protein Films Using Acoustic Waves 65 MHz Resonator a-s) (mP 3 sity o c 4 Vis 5 6 −15 −20 −25 −30 −35 0
Elas
5 odu lus
2
) Pa-s y (m 3 4
3 2 1
0 5 odulu
Elas
tic m
tic m
(MP
10
a)
I
2
osit Visc 5 6 4
ΔD (x10–4)
Δf (kHz)
a
s (M
Pa)
10
5 MHz Resonator
b Pa-s) sity (m 3 2 Visco 4 5 6
6 0.15 0.10 0.05 0.00
5 c mo dulus
2
0.20
−180 −190 −200 0 Elasti
Pa-s) sity (m 3 4 5
Visco
ΔD (x10–4)
Δf (Hz)
Interfacial Investigation of Protein Films Using Acoustic Waves, Fig. 8 Comparison of the frequency and dissipation factor sensitivity of (a) 65 MHz and (b) 5 MHz resonators to variations in the viscosity and modulus of the adsorbed films with 35 nm of thickness and 1,040 kg/m3 density. Clearly, the 65 MHz resonator has a better resolution over wider range of values of these properties
1161
(MPa
)
again fitted to Eqs. 11–13 with the thickness, density, elastic modulus, and viscosity of protein layer as parameters. Table 3 summarizes the physical and viscoelastic properties of the various proteins that were extracted from these measurements. However, the parameters of each protein on hydrophobic and charged hydrophilic surfaces differ due to the differences in the interfacial packing of spherical protein molecules on the two surfaces. Furthermore, normalization of the surface adsorbed capacity with respect to HSA (i.e., adsorbed capacity of given protein/adsorbed capacity of HSA) gives a simple integral number for the other two proteins on COO- SAM. The two proteins on hydrophilic surface indicate three layers for IgG, and four layers for fibrinogen on hydrophilic surface.
Future Directions for Research In conclusion, it has been demonstrated that high-frequency micromachined quartz resonators are especially suited for monitoring small changes in the viscoelastic loading in films adsorbed on their surface. Figure 8 shows the dependence of the frequency and Q-factor shift as a function of the viscosity and elastic modulus of a 35 nm thick protein film adsorbed on
I
0 5 c mod ulus (
Elasti
10
MPa)
10
Interfacial Investigation of Protein Films Using Acoustic Waves, Table 4 The sensitivities(resolution) of thickness, density, viscosity, and elastic modulus based on the measurement of five selected proteins Thickness (nm) fsample), (b) external DC voltage applied to nullify the force, and (c) external AC voltage with adjustable DC offset is applied to the tip which leads to its vibration [5]
a
b
φtip>φsample
Vdc= −ΔΦ +
AFM tip
AFM tip −
ΔΦ=(φtip−φsample)/e
force + + + + + Sample
no force Sample
c V=Vdc+Vacsin(ωt) AFM tip F =Fdc+Fω+F2ω Sample
The focus of this entry is on Kelvin probe force microscopy. Since the surface potential measurements are highly sensitive to surface charge variations, KPFM is widely used in detecting the dopant concentration and the dopant type (either p-type or n-type) on semiconductor materials. Non-semiconductor applications of KPFM include the characterization of adsorbates on Si substrates, intermetallic grain formation on metal films, nanotribology studies, and the degradation of inorganic materials for batteries, among others.
where Ftip and Fsample are work functions of the tip and the sample, respectively, and e is the magnitude of the charge of one electron. The contact potential difference provides material contrast information, because it is sensitive to adsorbed molecules on the surface or the presence of various phases within a given material. Electrostatic force is created between the tip and sample under the influence of this surface potential difference and the separation-dependent local capacitance C of the tip and sample. This force is given by 1 @C F ¼ ðDFÞ2 2 @z
Kelvin Probe Force Microscopy Principles Kelvin probe force microscopy is used to measure the surface potential with the AFM in noncontact mode, and is based on the Kelvin probe principle [3, 4, 10]. To describe the operating principle behind KPFM, consider a tip and sample interaction as shown in Fig. 1. When the tip and sample are electrically connected (Fig. 1a), electrons flow from the material with the lower work function to the material with the higher work function. Due to the difference in the work function of the electrically connected tip and the sample, an electrostatic contact potential difference (or surface potential) is created between the tip and the sample. The value of this contact potential difference (DF) is given by the following equation: DF ¼
Ftip
Fsample e
(1)
(2)
where z is the distance between the tip and sample. In KPFM, an AC voltage signal (VAC sin(ot)) with an adjustable offset DC voltage (VDC) is applied to the conducting tip (Fig. 1b and c). As shown in (Eq. 3), the electrostatic force is the sum of three components, namely, the DC force, a harmonic force with the same frequency of the AC tip bias (the first harmonic), and a harmonic force at double the frequency of the AC signal (the second harmonic): F¼
1 @C VAC 2 @C ðDF þ VDC Þ2 þ þ ðDF þ VDC ÞVAC sinðotÞ 2 @z @z 2 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} DC term
o term
1 @C VAC 2 cosð2otÞ 4 @z |fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} 2o term
(3)
Kelvin Probe Force Microscopy
K
Feedback
Laser
oto Ph ctor te de
Kelvin Probe Force Microscopy, Fig. 2 Schematic of the interleave scan method used in KPFM, where the first pass records the height profile performed in tapping mode, and the second pass in lift mode monitors the variations of electrostatic force at a constant height [12]
1175
Lock in amplifier Potential signal
X-Y-Z piezo
Computer
DC Voltage Sample
The electrostatic force is used to adjust the externally applied voltage in order to compensate the contact potential until the force between the tip and the sample is nulled. This is attained when the difference between the work functions of the tip and sample is equal to the external voltage. The presence of different materials on the sample surface leads to the formation of different material interfaces (with the tip), each having their own value of the contact potential difference. This leads to surface potential contrast [9]. In the initial implementation of KPFM, the topography and potential are measured simultaneously using two feedback circuits [13]. Subsequent modifications to the setup involve the sequential measurement of topography and potential in the so-called lift mode, which is regarded to improve on the lateral resolution of the topographic image [9]. A schematic of the twopass lift or interleave scan method used in KPFM experiments is shown in Fig. 2. On the first pass, surface height information is recorded in the standard tapping mode where the oscillating cantilever lightly taps the surface. On the second pass, the feedback loop controlling the vertical piezo is turned off, and the tip is lifted from the surface and traced following the surface profile acquired in the first pass and at constant tip– sample separation distance [7, 8]. An oscillating potential is applied to the DC-biased tip (the tip oscillation is electrically driven). The oscillating potential frequency is equal to the cantilever’s resonance frequency. An extra feedback loop is used to adjust the DC bias on the tip to nullify the first harmonic component of the electrostatic force [5, 9]. A more recent implementation of the KPFM experiment involves the detection of the surface potential
Servo controller (DC single adjusted to zero lock-in signal)
using a frequency modulation technique instead of the amplitude modulation method. The main difference between these two methods is that amplitude modulation is sensitive to the electrostatic force, while frequency modulation is sensitive to its gradient. Instead of having one feedback control/lock-in amplifier system that sequentially records the topography and surface potential, a separate lock-in amplifier is used in the frequency modulation KPFM experiment to detect surface potential. The frequency modulation method has been shown to measure a potential difference in close agreement with macroscopic measurements, and is thus regarded as a more quantitative potential measurement technique compared to amplitude modulation [16]. As mentioned earlier, KPFM is widely used in characterizing the doping of semiconductors, and it can also be utilized for non-semiconductor materials. Examples of KPFM applications are described next.
Applications of KPFM An example of the applicability of KPFM to semiconductor materials is shown in Fig. 3, which is a surface potential image of a pn junction on a silicon wafer created through ion implantation. Below the image is a corresponding cross section of the scan defined by the line A-B on the image [11]. The contrast in the surface potential image is induced by the charge carriers introduced by the ion implantation process. The brighter features correspond to areas with a higher surface potential. In this case, the surface potential in the p-type region is higher than in the n-type region.
K
K
1176
Kelvin Probe Force Microscopy
A
B
Δφ
1.2
0.8 1 μm
Kelvin Probe Force Microscopy, Fig. 3 Surface potential image of a silicon pn junction (above), with the bright and dark areas corresponding to the p-type and n-type regions, respectively. The cross section (below) refers to the scan defined by the line A-B on the image, where Df is the surface potential relative to the probe surface potential [11]
This is attributed to the bending of energy bands at the subsurface as a result of the trapping of charge carriers in surface states [11]. KPFM has been used in the characterization of perfluoroalkyl alkanes self-assembled on substrates such as graphite, mica, and silicon. As an example, Fig. 4 shows images of F14H20 adsorbates on a Si substrate [1]. Topography and surface potential images of the adsorbate layers show clusters in Fig. 4a and b. In the surface potential image, the difference between the adsorbates and silicon shows a high contrast. In Fig. 4c and d, the profiles of the white lines marked in Fig. 4a and b are displayed. The profiles indicate that there is a large surface potential drop of 0.5 V within the adsorbate clusters as a result of possible repulsion of the adsorbate composition. A correlation between the surface potential and the topography image suggests that the surface potential corresponds to the adsorbate clusters. The magnified images in Fig. 4e and f also revealed a particular shape in the cluster, spiral structures. In the example KPFM images, the capacity of the technique to detect different surface potentials on the sample is shown. Figure 5 shows a KPFM application to recognize constituent materials in a metal alloy film [14]. In this
a
Kelvin Probe Force Microscopy, Fig. 4 KPFM applied to studies of F14H20 adsorbates on Si: (a) topography, (b) surface potential image of F14H20 adsorbates on Si, (c) and (d) the cross-section profiles taken along the directions marked with the dashed white lines in the images in (a) and (b), respectively, and (e) and (f) zoomed-in topography and surface potential images on one of the locations shown in (a) and (b), respectively [1]
b
e
c
4 3 2 1 0 0 V
1
0.5
1
1.5
2
2.5 μm
800 nm
d
0.5
f
0 –0.5 0
0.5
1
1.5
2
2.5
3 μm
800 nm
Kelvin Probe Force Microscopy
K
1177 After wear test Surface height
Surface potential Uncoated Silicon
5 μm
Z-TETRAOL (Untreated)
a
K Z-TETRAOL (Fully bonded)
5 μm
b
Kelvin Probe Force Microscopy, Fig. 5 KPFM study of various intermetallics in a metal alloy: (a) surface potential image of Al/Cu film showing contrast between Al and the Al/Cu intermetallic (the dark points); the scan field is 40 mm 40 mm, (b) AFM topography image [14]
example, an alloy consisting of Al and Cu were deposited as 5 mm lines on a silicon substrate. The lines in the contact potential difference image in Fig. 5a contain features that are not present in the topographic image shown in Fig. 5b. In the potential image, the dark features contained within the lines correspond to the Al/Cu intermetallic that forms on the grain boundaries, which has a lower contact potential difference relative to the surrounding Al regions. The clear contrast between the Al/Cu and Al shows the capacity of KPFM as a technique to identify the presence of different materials on alloy surfaces at the nanoscale. KPFM can also be applied to nanotribological studies, specifically for studies involving wear initiation. In
0
10 μm
0
10 μm
0
20 nm
0
1V
Kelvin Probe Force Microscopy, Fig. 6 Surface height and surface potential maps after wear tests for an uncoated silicon wafer and two types of Z-TETRAOL-lubricated silicon wafers (untreated and fully bonded). The brighter areas correspond to higher values of both the height and surface potential
Fig. 6, surface height and surface potential data (taken after nanoscale wear tests were performed) are shown for silicon surfaces coated with the perfluoropolyether lubricant Z-TETRAOL. The lubricant is applied using two different techniques, such that the samples are either untreated or fully bonded. Data for the uncoated silicon surface subjected to the same wear test procedure is shown for reference [15].
1178
Kelvin Probe Force Microscopy
Unaged
Aged
+1.0 V
+1.0 V 1.2
20
1.1
15
1.0
Aged, fit
10
0
0.8 5μm
Aged
Unaged
5
0.9
0
Unaged +1.0 V
f (x)
K
0
5μm
0.7 0.8 0.9 1.0 1.1 1.2 Surface potential (V)
Kelvin Probe Force Microscopy, Fig. 7 Surface potential map of the unaged (left column) and the aged (middle column) LiFePO4 cathode samples with external voltage of + 1 V. The right column shows the normal probability density distribution
of the surface potential for the unaged and aged samples, indicating that the mean value of the surface potential decreases after aging
The surface height and surface potential images in Fig. 6 show contrast on both sets of images. Compared to the uncoated silicon surface, reduction in the debris is observed in the partially bonded samples. This indicates that the lubricant is able to provide wear protection. The pileup of debris observed in the untreated sample could be a combination of removed fragments of Si as well as the weakly bound lubricant. The untreated sample with its mobile lubricant fraction exhibits a lower surface potential change compared to the fully bonded sample, which only has immobile lubricant molecules. This is attributed to lubricant replenishment by the mobile lubricant molecules present on the untreated sample. In general, a considerable surface potential change is observed on the wear region when: (1) the lubricant has been fully removed from the substrate; (2) the native SiO2 layer has been abraded from the surface; (3) wear has caused subsurface structural changes; and (4) charges build up as the charges are unable to dissipate into the surrounding material [15]. Finally, the application of KPFM in studying material degradation is illustrated for the aging of lithiumion batteries. The surface potential maps of the unaged and aged LiFePO4 cathode samples are shown in Fig. 7 [12]. The maps were collected by applying +1.0 V from an external DC voltage source. On the surface potential maps for both the unaged and aged sample, a small difference in the contrast is observed, suggesting an almost uniform dissipation of charge over the surface of the samples under the externally
applied voltage. This indicates that under the external source the sample tends to charge uniformly even in an aged condition. The uniform charging is good for the cell as it avoids large local currents and subsequent damage to the cathode. The last column of Fig. 7 shows the distribution of the collected data points in unaged and aged cathodes. Then a normal probability density function is used to fit the histogram data. The unaged sample had mean values of the surface potential almost equivalent to the external applied voltage. The mean values of surface potential on the aged sample are lower than that of the unaged sample. Thus, even though the externally applied voltage was the same for the unaged and aged samples, the charge sustained on the surface of the aged cathode is less than that sustained on the unaged cathode [12]. The work function of the tip is constant in each surface potential map, since the same kind of tip is used in both samples. Therefore, Fig. 7 shows the change in the work function of the aged sample relative to the unaged sample. The decreased surface potential in the aged sample is an indication of surface modification as a result of the phase change of the cathode material upon degradation [12]).
Cross-References ▶ AFM Probes ▶ AFM, Tapping Mode ▶ Atomic Force Microscopy
Kinetic Monte Carlo Method
1179
References 1. Alexander, J., Magonova, S., Moeller, M.: Topography and surface potential in Kelvin force microscopy of perfluoroalkyl alkanes self-assemblies. J. Vac. Sci. Technol. B 27, 903–911 (2009) 2. Avila, A., Bhushan, B.: Electrical measurement techniques in atomic force microscopy. Crit. Rev. Solid State Mater. Sci. 35, 38–51 (2010) 3. Bhushan, B.: Springer Handbook of Nanotechology, 3rd edn. Springer, Heidelberg, Germany (2010) 4. Bhushan, B.: Nanotribology and Nanomechanics I – Measurement Techniques and Nanomechanics, II – Nanotribology, Biomimetics, and Industrial Applications, 3rd edn. Springer, Heidelberg (2011) 5. Bhushan, B., Goldade, A.V.: Measurements and analysis of surface potential change during wear of single-crystal silicon (100) at ultralow loads using Kelvin probe microscopy. Appl. Surf. Sci. 157(2000), 373–381 (2000) 6. Bonnell, D.: Scanning Probe Microscopy and Spectroscopy Theory, Techniques, and Applications, 2nd edn. Wiley – VCH, New York (2001) 7. DeVecchio, D., Bhushan, B.: Use of a nanoscale Kelvin probe for detecting wear precursors. Rev. Sci. Instrum. 69, 3618–3624 (1998) 8. Elings, V., Gurley, J.: US Patent No. 5,266,801 1993 9. Jacobs, H.O., Knapp, H.F., Muller, S., Stemmer, A.: Surface potential mapping: a qualitative material contrast in SPM. Ultramicroscopy 69, 39–49 (1997) 10. Kelvin, L.: Contact electricity of metals. Phil. Mag. 46, 82–120 (1898) 11. Kikukawa, A., Hosaka, S., Imura, R.: Silicon pn junction imaging and characterizations using sensitivity enhanced
12.
13.
14.
15.
16.
K
Kelvin prone force microscopy. Appl. Phys. Lett. 66, 3510–3512 (1995) Nagpure, S.C., Bhushan, B., Babu, S.S.: Surface potential measurement of aged Li-ion batteries using Kelvin probe microscopy. J Power Sources 196, 1508–1512 (2011) Nonnenmacher, M., O’Boyle, M.P., Wickramasinghe, H.K.: Kelvin probe force microscopy. Appl. Phys. Lett. 58, 2921–2923 (1991) O’Boyle, M.P., Hwang, T.T., Wickramasinghe, H.K.: Atomic force microscopy for work functions on the nanometer scale. Appl. Phys. Lett. 74, 2641–2642 (1999) Palacio, M., Bhushan, B.: Ultrathin wear-resistant ionic liquid films for novel MEMS/NEMS applications. Adv. Mater. 20, 1194–1198 (2008) Zerweck, U., Loppacher, C., Otto, T., Grafstroem, S., Eng, L.M.: Accuracy and resolution limits of Kelvin probe force microscopy. Phys. Rev. B 71, 125424 (2005)
Key Point Detection ▶ Waypoint Detection
K Kinetic Monte Carlo Method ▶ Computational Study of Nanomaterials: From Large-Scale Atomistic Simulations to Mesoscopic Modeling
L
Lab-on-a-Chip ▶ Electrical Impedance Tomography for Single Cell Imaging ▶ Micro- and Nanofluidic Devices for Medical Diagnostics
Lab-on-a-Chip (LOC) ▶ Lab-on-a-Chip for Studies in C. elegans ▶ Surface-Modified Microfluidics and Nanofluidics
Lab-on-a-Chip for Studies in C. elegans Nuria Vergara-Irigaray and Miche`le Riesen Department of Genetics Evolution and Environment, Institute of Healthy Ageing, University College London, London, UK
Overview
Synonyms Lab-on-a-chip (LOC); Microfluidic Micro-system; Micro-trap; Worm chip
a miniaturized chip format. The main characteristic of LOC devices is that they are able to handle extremely small fluid volumes down to less than pico-liters. This feature enables the analysis of very small samples in short time. LOC devices have recently been developed as micro-engineering approaches to small-size animal model research offering several advantages over conventional approaches, such as control of the microenvironment, automation, and high-throughput analysis. Caenorhabditis elegans (C. elegans) is a small, transparent, nonparasitic nematode that is used as a model organism for the study of multiple biological processes such as development, genetics, and neurobiology. This is mainly due to the fact that this animal is small in size, has a rapid life cycle, and it is easy to manipulate and cultivate in controlled laboratory conditions. In addition, there are many human disease models available in this model organism providing an in vivo environment for the study of these diseases over other approaches such as cell culture.
device;
Definitions Lab-on-a-chip (LOC) is a general term used to describe microdevices that integrate and scale down single or multiple laboratory functions and processes to
The possibility to perform laboratory operations on a small scale using miniaturized lab-on-a-chip devices has become in recent years a real tool for which biologists show an increasing interest. The use of small volume samples leads to a reduction in the time taken to perform an experiment and analyze the result; it also reduces the cost of the analysis. Together with the ability to automate sample manipulation and analysis, the use of LOC devices brings with it the potential to analyze very large populations of animals and to
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
L
Lab-on-a-Chip for Studies in C. elegans
1182 Adult 1130 μm 8 hours Young adult 920 μm
L4 635 μm
Eggs
Embryonic development 14 hours
10 hours
up to 4 months
L1 250 μm
Dauer 400 μm 13 hours Crowding Starvation High temperature
8 hours
12–14 hours
8 hours L3 500 μm
L2 370 μm
Lab-on-a-Chip for Studies in C. elegans, Fig. 1 Representation of the life cycle of C. elegans. The different stages of growth: egg, L1, L2, L3, L4, and adult are shown. Adult worms are fertile and capable of laying eggs. These are laid outside during the gastrula stage. If worms are deprived of food in larval stages or environmental conditions are not optimal
they will enter a larval stage known as dauer, in which development is arrested. When food becomes available development will continue. Numbers along the arrows indicate the length of time the animal spends at a certain stage. The length of the animal at each stage is marked underneath the stage name in micrometers (mm) (Reproduced from [28] with permission)
perform high-throughput analysis, which could not be done prior to the development of these new techniques. C. elegans has been used as a model organism for the last 40 years. The multiple characteristics that make this nematode an excellent model organism for the study of a wide range of biological processes will be discussed further in this text. In this entry, an introduction to the use of C. elegans as a model organism in a range of different biological scenarios will be provided, outlining the basic requirements of LOC devices to be used for the study of multicellular organisms and highlighting the technical challenges they have had to overcome. Finally, some approaches that so far have been of particular value for the study of C. elegans will be shown.
soil environments feeding on bacteria and fungi. It can often be found in compost heaps, fallen fruits, and some invertebrates, but hardly anything is known about its life in the wild. C. elegans is small. The length of the adult hermaphrodite worm is around 1.2–1.5 mm and the diameter 30–40 mm. It also has a very short life cycle (compared to other model organisms) of about 3 days under optimal conditions of food and temperature (Fig. 1). The life cycle of these nematodes starts with the fertilization of the eggs within the adult hermaphrodite. The fertilized eggs are laid when they are at the gastrula stage and contain about 40 cells. Embryogenesis occurs in 12–14 h after which the animal goes through four juvenile stages (L1–L4) each of which ends in a molt. Following the final molt, the animals reach adulthood at which point they become fertile producing about 300 progeny each over the course of 3 days. When environmental conditions are not optimal (e.g., high temperature, overcrowded cultures, or starvation), the L3 stage is replaced by a dauer larval stage that corresponds to the infectious stage of
Introduction to C. elegans C. elegans Physiology Caenorhabditis elegans (C. elegans) is a small, transparent, free-living nematode which lives in temperate
Lab-on-a-Chip for Studies in C. elegans Lab-on-a-Chip for Studies in C. elegans, Fig. 2 Anatomical features of an adult hermaphrodite C. elegans. Arrows point to the main features of the worm: the pharynx, intestine, vulva, and gonad
1183
L
Gonad Eggs Vulva
Anus Intestine
Intestine
Pharynx
Gonad
parasitic nematodes. The dauer of free-living soil nematodes is sometimes associated with some invertebrates. Dauer larvae are thin and are unable to feed because their mouths are closed with a cuticle but survive high temperatures and other harsh conditions much better than adult worms. Most interestingly, they can remain viable for 3 months and resume the L4 reproductive stage when they encounter more favorable environmental conditions. Adult worms have an average life span of 2–3 weeks. C. elegans displays two sexes: hermaphrodites and males. In culture, males are produced spontaneously at low frequency (about 0.1%) but their frequency and crossing rate in the wild are unknown. The hermaphrodite produces both sperm and oocytes and can therefore either self-fertilize or be fertilized by a male. The worms are cylindrical in shape. They essentially consist of an inner and outer tube separated by the pseudocoelomic space. The inner tube contains the gonad, pharynx, and intestine and the outer tube contains neurons, muscle, excretory system, hypodermis, and cuticle. Figure 2 shows a photograph of the anatomical features of a gravid adult C. elegans with the pharynx, intestine, and the reproductive system highlighted. The hermaphrodite and male worms consist of a constant number of somatic cells of 959 and 1,031, respectively. The haploid genome size is relatively small (9.7 107 base pairs) about 30 times smaller than the human genome and contains near 20,000 genes. In 1998, C. elegans became the first multicellular organism to have its genome fully sequenced. About 35% of C. elegans genes have human homologs. The genome consists of five pairs of autosomal chromosomes and a pair of sex chromosomes: XX in the hermaphrodite and XO in the male. C. elegans development has been fully characterized, meaning the complete cell lineage of the worm
has been recorded. All the cell divisions required to generate a specific group of differentiated cells has been documented for the whole animal [1]. C. elegans as a Model Organism In 1974, Sydney Brenner established C. elegans as a model organism when working on the genetics of the animal development (including neural development), behavior, and physiology [2]. Brenner, with his exhaustive labor, set up the basis for today’s worm research and as recognition for this work he was awarded the Nobel Prize in Physiology or Medicine in 2002. There are several important features that make C. elegans an attractive organism not only for the study of gene regulation and gene function, but also for the study of many other different relevant biological aspects including development, cell biology, behavior, and neurobiology. C. elegans is easy and cheap to maintain in the laboratory. It grows on agar plates feeding on Escherichia coli lawns. It can also be grown in liquid cultures in large quantities. C. elegans is a multicellular organism which goes through a complex developmental process, including embryogenesis, morphogenesis, and growth to adulthood. As mentioned before, it has a short life cycle which is an advantage for developmental studies. It is transparent; therefore, it is possible to use fluorescent makers in order to track different cell types and tissues and to perform high-resolution microscopy. C. elegans is one of the simplest organisms with a nervous system comprised of only 302 neurons. Nevertheless, it exhibits complex behavioral modalities, such as habituation, sensitization and conditioning in response to stimulus. C. elegans nervous system has been completely mapped and all neuronal connections are known. Using microfluidic platforms it has been possible to perform precise neuronal ablations and study the phenotypic and behavioral consequences of this in a high-throughput manner.
L
L
1184
In comparison to other model organisms C. elegans has a relative short lifespan representing a useful model for the study of aging. Due to its small size it is possible to perform high-throughput drug screening assays. For example, C. elegans provides a useful model for the investigation of the mode of action of anthelmintics (group of compounds to treat infections caused by parasitic nematodes). As mentioned earlier, C. elegans displays a particular mode of reproduction with self-fertilizing hermaphrodites and facultative males which makes it an interesting model for population genetic studies. Males arise infrequently in C. elegans populations, about 0.1%. This percentage can be increased through heat-shock of L4 hermaphrodites. Once males have been obtained, a population containing up to 50% males can be maintained through mating. C. elegans also has the advantage of being amenable to genetic manipulation, through mutagenesis, transgene expression, and RNA interference (RNAi) partly because of its rapid life cycle, small size, and ease of laboratory cultivation. When investigating molecular mechanisms in C. elegans either forward or reverse approaches can be taken. Reverse genetic techniques are hypothesis driven. Analysis of the phenotypes of mutants with specific mutations can be used to inform about the role of genes in a drug’s mode of action. Alternatively, RNA interference can be used to knock down the expression of specific genes allowing, for example, the study of morphological or behavioral changes associated with the gene. A number of microfluidic approaches that exploit several of these C. elegans features have been described and will be further discussed in the text.
Introduction to Worm Chip Technology Lab-on-a-chip (LOC) devices provide the ability to perform multiple laboratory operations in a miniaturized chip format. This is mainly due to the fact that they handle extremely small fluid volumes reducing the amount of sample and time needed for a particular analysis. Worm chips, as they are commonly described, are microfabricated LOC devices capable of manipulating single worms or a population of worms and their environment in a precise and automated fashion. The advantages provided by LOC devices depend very specifically on their applications. In general: (1) they handle very low fluid volumes reducing the
Lab-on-a-Chip for Studies in C. elegans
cost and sample volume needed for the analysis; (2) they are very efficient systems because they provide high control of the experimental conditions (e.g., thermal control, short diffusion distances); (3) they allow highthroughput analysis because being very compact, LOC systems are capable of massive parallelization; and (4) they can be mass produced with low fabrication costs, allowing cost-effective disposable chips. LOC devices belong to a broader group of small mechanical devices known as MEMS (MicroElectro-Mechanical-Systems). Worm chips usually integrate MEMS components to provide flow control (microchannels, pumps, mixers, and valves) and sensing capabilities to the devices (optical or electrochemical sensors) with microfluidics platforms (to deal with and manipulate tiny volumes of fluid samples) [3]. Worm Chip Materials and Fabrication Technologies Worm chips can be fabricated from many types of material including various polymers, glass, or silicon, or combinations of these materials. A broad variety of fabrication technologies are used for LOC device fabrication; the most frequent being single layer or multilayer soft lithography. Soft lithography-based microfluidic devices are typically made in polydimethylsiloxane (PDMS). This polymer offers several benefits for biological research including biocompatibility, high permeability to gas, a chemical inert surface, and optical transparency down to 300 nm. The chips normally have a closed architecture as they are sealed with a glass slide, a PDMS layer, or an agar plate. Worm chips that include MEMS are normally fabricated using standard silicon and polymer micromachining processes, which requires expertise in microfabrication and access to a clean room. Photolithography is the most common method of fabrication of microfluidic devices. Typically it consists of the transfer of a pattern into a photosensitive material by selective exposure to a radiation source such as light. Following the creation of a template (replica mold) PDMS can be cast against it to form the microfluidic devices. This process is cost effective and can be performed in any traditional laboratory setting with the mold being used repeatedly [4]. Worm Chip Technical Challenges The study of multicellular organisms using microfluidic devices presents some technical challenges that when
Lab-on-a-Chip for Studies in C. elegans
controlled allow high-throughput phenotyping, screening, imaging, and sorting of small animals in an automated manner. In particular, the organisms need to be handled efficiently and the microenvironment created for the experiment needs to be precisely controlled. Environment parameters, to be controlled, include temperature, nutrients, dissolved gas, and stimuli delivered to the animals. Worm chips have been designed for the detailed analysis of single worm phenotype. The manipulation in this type of chips involves complete or partial immobilization of the worm. The movement of the worm can be restricted physically or by applying glue, cold, or CO2. Some examples of immobilization of the worms on chip will be discussed further in section “Key Research Findings: Applications of LOC Devices to the Study of C. elegans”. Other worm chips have been designed for the analysis of large populations of animals and do not restrict worm movement. These types of chips are mainly used to study worm behavior and locomotion [5].
Key Research Findings: Applications of LOC Devices to the Study of C. elegans The development of LOC devices for the study of small animals is a new area of research still in progress. Worm chips capable of performing high-throughput studies show great potential to provide a better understanding of key aspects of the biology of nematodes. In this section, a few examples of microfluidic devices designed to study different aspects of the worm biology will be shown and some technical challenges will be highlighted. Immobilization of Single Animals: Suction Pipette, Glue, Physical Restraint (Worm Trap), Suction Microchannel (Aspiration) As discussed above, one of the challenges that LOC devices have had to overcome in order to be able to study relevant aspects of the biology of C. elegans is the total or partial immobilization of the whole animal. C. elegans is a very active animal. It constantly explores the environment while searching for food (foraging behavior). In the laboratory C. elegans is normally cultivated on agar plates where it exhibits a sinusoidal forward and backward movement of unpredictable direction. Locomotion behavior is
1185
L
often measured and quantified by researchers to describe the healthy status of the worm. Immobilization of the animal is a key element in order to get highresolution images of various anatomical features or to image cellular development and gene expression. In addition, immobilization of the worm is required to precisely deliver stimuli taking control of the microenvironment conditions for the experiment, to perform laser ablations of individual neurons for studying neural networks, axon regeneration, or monitoring neuronal responses at the presence of various stimuli. One of the first microfluidic devices developed to study biological questions using C. elegans addressed the issue of immobilization of the worm in a flowing solution; this was achieved by holding the animal by the tail using a suction pipette. Forward and backward movement was recorded in response to different chemical stimuli using an array of photocells connected to a multichannel chart recorder [6]. This allowed, for the first time, to control the timing of the stimulus presented to the worm and behavioral events were recorded with relatively high time resolution thus allowing the determination of changes in behavior. With this device the author demonstrated the mechanism of chemotaxis and described for the first time that the frequency of change of direction varied with stimulus intensity in the appropriate direction and the animal does adapt rapidly. C. elegans evaluates the quality of its environment through a complex and versatile chemosensory system. Peripheral chemo-sensation is highly divergent between mammals and C. elegans, but regulatory and behavioral mechanisms appear to be more conserved across species. C. elegans’ chemosensory behavior was tested again a few years later on a device where the worm was semi-restrained. The anterior part of the worm was glued to the substrate and the worm’s behavior was recorded and analyzed quantitatively [7]. The use of mutant worms in this study provided insight into the understanding of the neural basis of behavior and represented a new technique to use in combination with genetic or neuronal manipulations. More sophistication in the design of the LOC devices was shown by Chronis et al. In their device to measure olfactory neuron activity, the worm was physically immobilized in a narrow space (i.e., no glue or anesthetic was used) (Fig. 3). This device combined the concept of worm trap (used as well to analyze multiple worms at the same time [8]) with a
L
L
Lab-on-a-Chip for Studies in C. elegans
1186
Lab-on-a-Chip for Studies in C. elegans, Fig. 3 Lab-ona-chip device to measure olfactory neuron activity in response to chemosensory stimuli. (a) The chip consists of two channels (1 and 4) loaded with fluorescent dyes that are used to redirect the chemical stimulus (applied in channel 2) to the nose of the worm. (b, c) Higher magnification photographs of a worm trapped in the chip. The arrow points to a sensory neuron which is expressing a fluorescent marker in response to the stimulus delivered (Reproduced from [5, 9] with permission)
a
2
3
Stimulus
Dye (right)
Dye (left)
Dye (left)
Dye (right)
4
1 ON
OFF
Stimulus Off
b
Buffer
Buffer
Stimulus
2
1
ON
OFF
Stimulus On
3 4
c
microfluidic chemical delivery system and allowed to measure the activity of chemosensory neurons in response to chemosensory stimuli [9]. The same authors also designed a chip to measure neural activity in relation to behavior. In this device the worms, one at a time, were able to move forward and backward and neural activity could be measured simultaneously for the first time. The results that emerged from this study had a high biological relevance. Prior to this
work, nobody had shown neural activity in real time; this revealed important finds relating to the behavior of the olfactory neurons. Olfactory neurons are activated in the absence of the stimuli (food odor) and remain deactivated when the food odor appears. The amount of activity of the neuron depends on how strong the odor is and how long it lasts. A microsystem for high-throughput whole animal sorting and screening at subcellular resolution was
Lab-on-a-Chip for Studies in C. elegans
a
suction
inlet
C
A
F
collection
Lab-on-a-Chip for Studies in C. elegans, Fig. 4 Microfluidic worm sorter. (a) The sorter consists of control channels and valves (labeled with the letters A–F) that direct the flow of worms in different directions in the flow channels. Animals enter the chip through the inlet channel and can be continuously recirculated. The immobilization chamber captures one worm each time via suction by multiple microchannels held at low pressure and subsequently, the image acquisition and processing are performed. The worm is either collected or directed to the waste, depending on its phenotype. (b) Image of the worm sorter and a worm trapped in the system. Bright spots along the animal show mechanosensory neurons expressing a fluorescent reporter (Reproduced from [10] with permission)
L
1187
E
B D circulator
wash suction
waste/circulator
b
developed by Rohde et al. [10]. The device was capable of isolating and immobilizing single worms for the screening of phenotypic features at subcellular resolution in live animals (Fig. 4). The animals were exposed to biochemical compounds and high-resolution timelapse imaging of many individual worms on a single chip, without the need of anesthesia, was performed. Individual worms were capture in an immobilization chamber via suction by a microchannel held at a low pressure and imaged using high-resolution optics. After image acquisition and processing, the worm can be released and directed to the appropriate collection channels according to its phenotype. Zeng et al. presented in 2008 an improved immobilization system for single animals that allowed performing techniques such as femtosecond-laser manipulation (laser microsurgery) and threedimensional two-photon microscopy (which cannot be done on anesthetized animals) [11]. They described their immobilization method as fast and noninvasive
L
because it did not require the use of anesthesia or cooling and allowed high-throughput on-chip worm manipulation at subcellular level. The device combined aspiration channels to capture and align the animals in a linear position (achieving partial immobilization) and a “sealing membrane” that flexed on top of the captured animal forming a tight seal which completely restrained the motion in a linear orientation. This method allowed highly demanding subcellular resolution imaging and manipulation techniques (such as laser microsurgery for precision ablation with minimal collateral damage and 3D twophoton microscopy including two-photon excitation fluorescence or TPEF) that required long periods of stable immobilization. Immobilization of Multiple Animals: Arrays of Worm Traps, Chip-Gel Hybrid The first micro-engineered array of clamps for worms (worm trap) was designed by Hulme et al. with the
L
1188
purpose of investigating the physiology and behavior of large populations of worms. The immobilization of the worms in this device was achieved by an array of wedge-shape channels (clamps) where the worms were trapped and oriented by the direction of the microfluidic flow. The purpose of this work was to identify mutants affecting anatomical structure or body size distribution of a population of worms. One major advantage of this device was that the worms could be released after the analysis by reversing the flow direction. Therefore repeated sampling of the same population of worms at different times could be performed. The authors measured life span and brood size from the released worms [8]. This type of worm chip provides a high-throughput system for morphological analysis with potential to be used for extremely delicate operations such as microsurgery; and with the combination of high-resolution optical imaging could be used for fluorescence imaging. An improved modified version of this system was used to culture C. elegans in liquid, in separate chambers from L4 throughout the entire life span of the worm until death [12]. The device included microfluidic worm clamps to allow the temporary and reversible immobilization of each worm, making it possible to monitor changes in body size and locomotion throughout the life span of the animal. In 2008, Chung et al. presented the first microsystem with complete automation of sample handling for high-resolution microscopy phenotyping and sorting of C. elegans [13]. This device could analyze and sort several hundred worms per hour using a brief and reversible immobilization by transiently cooling to 4 C. This system in comparison with the COPAS (Complex Object Parametric Analyzer and Sorter), which is a fluorescence-activated cell sorter modified to handle small organisms, represents a much more advanced system not limited by the tissue level. The automated microsystem is capable of high-throughput high-resolution microscopy and allows sorting based on phenotype. In this case, the authors based the sorting in the expression of a reporter in neurons. Almost simultaneously another chip to perform laser ablations was also described [14]. The nanoaxotomy chip is a microfluidic device used to sever axons in C. elegans using ultrashort laser pulses with high precision and monitor axonal regeneration activity. This chip presented several advantages over the chips used previously. As mentioned before,
Lab-on-a-Chip for Studies in C. elegans
immobilization of the animal was done without the need of glue or anesthetic therefore not interfering with physiological processes. This system used a two-layer microfluidic approach with a thin membrane between the two channels. When the top layer was pressurized the membrane deflected quickly immobilizing the worm. The chip had great flexibility in terms of being able to analyze worms from L4 to adult size; furthermore, the worms did not need a recovery period after surgery, permitting immediate behavioral studies. The chip also allowed control of the sample population. In 2009, Chokshi et al. explored the use of CO2 to immobilize C. elegans on a microfluidic chip and characterized the effect of this immobilization method on worm behavior. The authors also tested mechanical immobilization through a deformable membrane of PDMS that compressed the worm to mechanically restrict its movement (compressive immobilization). They found that CO2 immobilization was effective and efficient for up to 2 h with low photobleaching, whereas the mechanical approach was best for shorter periods of time. Locomotion was measured as well as gene reporter expression [15, 16]. In 2010, Krajniak et al. presented a chip-gel hybrid microfluidic device for developmental studies. The novelty of this device was the use of a biocompatible polymer called Pluronic F127 as the immobilization agent in combination with a microfluidic platform. This polymer was reversibly transformed into a gel in a process that depends on temperature. The gel phase was sufficient to immobilize the animals. While the animals were not imaged they were cultured in individual chambers containing media with nutrients required for development. This device allowed culturing, individual tracking and repeated imaging of C. elegans at high-resolution throughout development [17]. No Immobilization: Microchamber for Small Volume Liquid Culture, CD Platform, Droplet Generator, Artificial Dirt The appearance of the first platform to maintain and monitor C. elegans in a microfluidic environment arrived almost 25 years after the first example of a microfluidic device for the study of the worm. This device was designed for the experimental study of C. elegans in space. It consisted of a chamber able to support a tiny volume of liquid culture of
Lab-on-a-Chip for Studies in C. elegans
C. elegans for long periods of time with a membrane to allow gas exchange [18]. Shadow images were acquired by a camera attached at the bottom of the microchamber. The authors measured the activity of C. elegans as a function of temperature finding that thrashing (stroke frequency) increased with temperature validating previous results obtained with manual counting. As an alternative to the immobilization of the animal with glue, other platforms for culture and behavioral observation have also been used. In 2007, Kim et al. [19] described a compact disc microfluidic system for the automatic cultivation of C. elegans. This system showed to be proficient in automated feeding, waste removal, and live-animal microscopy using centrifugal force-driven fluidics. This system could analyze and control a population of worms in one chamber (up to three generations could be monitored over time) and measure the rate of development, brood size, and the rate of population growth. As shown previously, with other microsystems, locomotion could also be measured in single worms as an indicator of the physiological state of the worm. The CD cultivation system showed the capacity to incorporate diverse molecular, biochemical, and pharmacological assays for real-time physiological and behavioral test of live animals. This system has also been used to study the effect of hypergravity and the ability of C. elegans to withstand mechanical stress. The device affords an opportunity to investigate molecular pathways that transduce mechanical forces into biological regulators and to identify genetic targets of mechanical stress in the context of the whole animal [20]. Droplet-based microfluidic systems represent another type of chips for individual worm assay [21]. These systems consist of a droplet generator and a droplet trap array, which allows the encapsulation of individual worms into parallel series of droplets of tiny volumes (Fig. 5). The authors studied worm swimming behavior in response to a neurotoxin. This work represents for the first time droplet encapsulation of small organisms. The main disadvantage is that the worms only survived for a maximum of 2 h. Artificial dirt illustrates novel experimental microfluidic devices to study C. elegans neurobiology and behavior. Artificial dirt devices mimic the natural environment where C. elegans can be found in contrast to the planar surface of an agar plate. The authors argue
1189
L
that this fails to reflect the complexity of C. elegans natural environment with complicated stimulus delivery. Artificial dirt devices are thin and transparent and consist of microchambers and agarose-free channels that mimic the moist soil environment with rapid delivery of fluid-borne stimuli. The animals are able to crawl as they would on agar plates and are compatible with optical recording. Artificial dirt devices were made to accelerate studies of the neuronal basis of behavior in C. elegans [22]. Others: Live-Animal FACS, Maze, Optofluidic On-Chip Platform A curious example of the broad applicability of worm chips is the device designed by Qin et al. to study behavioral plasticity of an animal using a microfluidic platform [23]. These authors studied the tendency of the worm to explore different environments in the presence or absence of food in one of the outlet chambers. They measured locomotion rates and compared the results obtained from mutant and wild-type worms. Live-animal FACS (la-FACS) is a FluorescenceActivated Cell Sorting (FACS) based method to sort live C. elegans larvae. Up until recently FACS only allowed sorting of C. elegans embryos [24]. The la-FACS is able to obtain large quantities of pure mutant populations. This method has a great potential for expanding the arsenal of high-throughput tools available in C. elegans [25]. There are many further examples were C. elegans has been used. For example in 2006, Heng et al. used C. elegans and a technique described as optofluidic microscopy (OFM) on chip. L1 worms were euthanized to demonstrate the feasibility of the method to perform high-resolution imaging.
Future Directions of Research The development of LOC systems over the last few years has been spectacular. The future of microfluidic devices offers a range of interesting and new opportunities in worm research. Worm chips offer several advantages over conventional approaches including the maintenance of well-controllable microenvironments, creation of reproducible experimental conditions, and automation of tedious experimental protocols. However, some improvements are still
L
L
Lab-on-a-Chip for Studies in C. elegans
1190
Lab-on-a-Chip for Studies in C. elegans, Fig. 5 Droplet-based microfluidic system for single worm analysis. (a) The device consists of a parallel series of microdroplets (droplet trap array) continuously generated by shearing a dispersed aqueous phase with a constant oil phase at the T-junction droplet generator; the inset shows a magnified view of ten traps with droplets. (b) Image of an array of droplets encapsulated with worms. The position of individual worms is indicated by white arrows (Reproduced from [21] with permission)
a
aqueous inlet waste outlet
oil inlet
droplet generator
droplet trap array
b
needed in order to overcome the challenges arising from working with small volumes and multicellular living organisms such as C. elegans. In addition to imaging different features of the worm, most of the microfluidic devices described so far have focused on measuring locomotion behavior. However, C. elegans displays many other types of behavior which have not been exploited thus far using a worm-chip platform. One example is pharyngeal pumping behavior. The C. elegans pharynx is required for the ingestion and mechanical digestion of food. There is a basement
membrane isolating the pharynx from the rest of the animal, making it an almost completely self-contained system. The pharyngeal nervous system coordinates the pharyngeal muscle cells allowing food to be sucked in, broken down, and passed into the intestine. This process is referred to as pharyngeal pumping. The pharyngeal pumping of C. elegans can either be counted optically or electrically. The method of recording pumping electrophysiologically gives what is referred to as the electropharyngeogram or EPG. This is an extracellular recording that can be performed on intact worms or on isolated
Lab-on-a-Chip for Studies in C. elegans
pharynxes. As well as producing a readout of pump rate, this technique can also be used to provide information on neuronal activity of the pharynx. The EPG trace has excitatory and inhibitory components that can be attributed to the function of specific pharyngeal neurons. Measuring EPGs is a laborious and tedious process and it takes a long time to analyze a single worm. The possibility to perform EPG-recordings in a high-throughput and automated fashion remains to be exploited using microfluidic platforms. RNAi represents another technique that has not been exploited using microfluidic devices yet. RNAi is a powerful tool to investigate gene function in C. elegans broadly used to do worm research in many different areas such as development or aging. So far nobody has used this tool on a microfluidic platform in order to perform a genetic screen. In this scenario the combination of RNAi and a microfabricated worm sorter device would result in a very powerful platform.
Further Reading Almost everything you want to know about C. elegans itself can be found online. Wormbase is a complete database about the worm with connections to data of human and other model organisms. Wormbase contains complete cell lineages as well as the genetic and physical maps and the annotated sequence of the worm. Also all web references can be found there: http://www.wormbase.org/ The latest reviews describing microfluidic devices applied for the study of C. elegans are listed next: [5, 26, 27].
Cross-References ▶ Applications of Nanofluidics ▶ MEMS Adaptive Optics ▶ MEMS Display based on Interferometric Modulation Technology ▶ MEMS on Flexible Substrates ▶ Microfabricated Probe Technology ▶ Micropumps ▶ Nanoscale Properties of Solid–Liquid Interfaces ▶ On-Chip Flow Cytometry
1191
L
▶ RNAi in Biomedicine and Drug Delivery ▶ Thermal Cancer Ablation Therapies Using Nanoparticles
References 1. Riddle, D.L., Blumenthal, T., Meyer, B.J.: C. elegans II, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor (1997) 2. Brenner, S.: The genetics of Caenorhabditis elegans. Genetics 77(1), 71–94 (1974) 3. Bhushan, B. (ed.): Springer Handbook of Nanotechnology, 3rd edn. Springer, Berlin (2010). 1964 4. Herold, K.E., Rasooly, A.: Lab-on-a-Chip Technology: Fabrication and Microfluidics, vol. 1. Caister Academic Press, Norfolk (2009) 5. Chronis, N.: Worm chips: microtools for C. elegans biology. Lab Chip 10(4), 432–437 (2010) 6. Dusenbery, D.B.: Responses of the nematode Caenorhabditis-elegans to controlled chemical-stimulation. J. Comp. Physiol. 136(4), 327–331 (1980) 7. Faumont, S., Miller, A.C., Lockery, S.R.: Chemosensory behavior of semi-restrained Caenorhabditis elegans. J. Neurobiol. 65(2), 171–178 (2005) 8. Hulme, S.E., et al.: A microfabricated array of clamps for immobilizing and imaging C.elegans. Lab Chip 7(11), 1515–1523 (2007) 9. Chronis, N., Zimmer, M., Bargmann, C.I.: Microfluidics for in vivo imaging of neuronal and behavioral activity in Caenorhabditis elegans. Nat. Methods 4(9), 727–731 (2007) 10. Rohde, C.B., et al.: Microfluidic system for on-chip high-throughput whole-animal sorting and screening at subcellular resolution. Proc. Natl. Acad. Sci. U. S. A. 104(35), 13891–13895 (2007) 11. Zeng, F., Rohde, C.B., Yanik, M.F.: Sub-cellular precision on-chip small-animal immobilization, multi-photon imaging and femtosecond-laser manipulation. Lab Chip 8(5), 653–656 (2008) 12. Hulme, S.E., et al.: Lifespan-on-a-chip: microfluidic chambers for performing lifelong observation of C. elegans. Lab Chip 10(5), 589–597 (2010) 13. Chung, K.H., Crane, M.M., Lu, H.: Automated on-chip rapid microscopy, phenotyping and sorting of C. elegans. Nat. Methods 5(7), 637–643 (2008) 14. Guo, S.X., et al.: Femtosecond laser nanoaxotomy lab-on-achip for in vivo nerve regeneration studies. Nat. Methods 5(6), 531–533 (2008) 15. Chokshi, T.V., Ben-Yakar, A., Chronis, N.: CO2 and compressive immobilization of C. elegans on-chip. Lab Chip 9(1), 151–157 (2009) 16. Chokshi, T.V., Bazopoulou, D., Chronis, N.: An automated microfluidic platform for calcium imaging of chemosensory neurons in Caenorhabditis elegans. Lab Chip 10(20), 2758–2763 (2010) 17. Krajniak, J., Lu, H.: Long-term high-resolution imaging and culture of C. elegans in chip-gel hybrid microfluidic device for developmental studies. Lab Chip 10(14), 1862–1868 (2010)
L
L
1192
18. Lange, D., et al.: A microfluidic shadow imaging system for the study of the nematode Caenorhabditis elegans in space. Sens. Actuators B-Chem. 107(2), 904–914 (2005) 19. Kim, N., et al.: Automated microfluidic compact disc (CD) cultivation system of Caenorhabditis elegans. Sens. Actuators B-Chem. 122(2), 511–518 (2007) 20. Kim, N., et al.: Gravity force transduced by the MEC-4/MEC10 DEG/ENaC channel modulates DAF-16/FoxO activity in Caenorhabditis elegans. Genetics 177(2), 835–845 (2007) 21. Shi, W.W., et al.: Droplet-based microfluidic system for individual Caenorhabditis elegans assay. Lab Chip 8(9), 1432–1435 (2008) 22. Lockery, S.R., et al.: Artificial dirt: microfluidic substrates for nematode neurobiology and behavior. J. Neurophysiol. 99(6), 3136–3143 (2008) 23. Qin, J.H., Wheeler, A.R.: Maze exploration and learning in C.elegans. Lab Chip 7(2), 186–192 (2007) 24. Stoeckius, M., et al.: Large-scale sorting of C. elegans embryos reveals the dynamics of small RNA expression. Nat. Methods 6(10), 745–751 (2009) 25. Fernandez, A.G., et al.: Automated sorting of live C. elegans using la-FACS. Nat. Methods 7(6), 417–418 (2010) 26. Crane, M.M., et al.: Microfluidics-enabled phenotyping, imaging and screening of multicellular organisms. Lab Chip 10, 1509–1517 (2010) 27. Taylor, A.M., Jeon, N.L.: Micro-scale and microfluidic devices for neurobiology. Curr. Opin. Neurobiol. 20(5), 640–647 (2010) 28. Altun, Z.F., Hall, D.H.: Introduction to C. elegans anatomy. In: Altun, Z.F., Hall, D.H. (eds.) WormAtlas. Cold Spring Harbor Laboratory Press (2009). ISBN 978-087969794-5
Lab-on-a-Chip Whole-Cell Biosensor
Laser Scanning Confocal Microscopy ▶ Confocal Laser Scanning Microscopy
Laser Tweezers ▶ Optical Tweezers
Lateral Force Microscopy (LFM) ▶ Friction Force Microscopy
Laterally Vibrating Piezoelectric Resonators Gianluca Piazza Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, USA
Synonyms
Lab-on-a-Chip Whole-Cell Biosensor ▶ Microfludic Whole-cell Biosensor
Lamb-Wave Resonators ▶ Laterally Vibrating Piezoelectric Resonators
Landmark Detection ▶ Waypoint Detection
Large Unilamellar Vesicle (LUV) ▶ Liposomes
Bulk acoustic wave MEMS resonators; Contour-mode resonators; Lamb-wave resonators
Definition A laterally vibrating piezoelectric resonator is a particular class of microelectromechanical (MEM) devices, whose resonant motion is transduced via thinfilm piezoelectrics and occurs primarily in the in-plane lateral directions of the structure (e.g., width extensional or length extensional motion). The resonant frequency is therefore set by the lateral dimensions of the device, which can operate in the few MHz to several GHz range on the same chip and is generally employed for the making of filters and oscillators.
Principle of Operation Laterally vibrating piezoelectric resonators are formed either entirely by a thin-film piezoelectric film
Laterally Vibrating Piezoelectric Resonators
1193
L
Laterally Vibrating Piezoelectric Resonators, Fig. 1 Mock-up view of laterally vibrating piezoelectric resonators formed either by (a, b) pure piezoelectrics ((a) representing a single finger and (b) the so-called multifinger embodiment) or (c, d) composite stack ((c) representing a single finger and (d) the so-called multifinger embodiment) of a piezoelectric and semiconducting or insulating material (structural layer)
fo ¼ vac =2W
Laterally Vibrating Piezoelectric Resonators, Fig. 2 Principle of transduction of a laterally vibrating piezoelectric resonator. The application of an electric field, E3, across the film thickness induces lateral expansion and contraction of the piezoelectric layer. The induced strain (s1) is approximately proportional to the electric field via an equivalent piezoelectric coefficient (d31)
sandwiched between metallic electrodes [1] or by a composite constituted by the same piezoelectric film stacked on top of another structural layer such as a semiconducting or insulating material (Fig. 1). As shown in Fig. 2, the application of a potential across the piezoelectric film induces a lateral strain in the plane of the device. When the frequency of the excitation signal coincides with the mechanical resonance of the piezoelectric/composite body, the device exhibits enhanced amplitude of vibration, which is translated to an electrical signal by the direct piezoelectric effect and generates a charge proportional to the amplitude of the displacement. The main geometry of the laterally vibrating structure is designed such that one of the in-plane lateral dimensions, W, is used to set the resonant frequency of the structure:
(1)
where vac is the acoustic velocity of the piezoelectric/ composite that forms the resonator, and depends on the square root of the ratio of the Young’s modulus and mass density of the material. This is the main important feature of laterally vibrating resonators. In fact, the ability to set frequency via the in-plane lateral geometrical variables of the structure makes it possible to lithographically define multiple frequency devices on the same substrate and effectively produce components that are capable of operating over a broad spectrum and dynamically reconfigure radio receivers. Furthermore, as with transistors, lithography scaling is accompanied by an increase in the frequency of operation of the resonator. The second in-plane lateral dimension, L, and the piezoelectric film thickness, Tp, are used to properly define the capacitance, Co, of the piezoelectric transducer and therefore its electrical impedance. Co ¼ e
WL T
(2)
In the case of a composite structure, an additional design parameter is set by the thickness of the main structural layer, Ts. This layer has generally been introduced to enhance the quality factor, Q, (an inverse measurement of the mechanical damping in a resonator) of the pure piezoelectric device. In various demonstrations, materials with intrinsically high Q have been used, such as silicon [2–6], diamond [7],
L
L
1194
and silicon carbide [8]. Although beneficial, its insertion introduces a trade-off between Q and the electromechanical coupling coefficient, kt2, a parameter that describes how effectively the input electrical energy in the resonator is converted into mechanical (i.e., motion). In the case of pure piezoelectrics, the kt2 is set exclusively by the mode of vibration and the material properties. In the case of a composite, the intrinsic electromechanical coupling is approximately scaled by the ratio of the thickness of the piezoelectric layer to the total thickness of the composite (Ts + Tp). Therefore, the selection of the thickness of the main structural layer, Ts depends on the resonator application and should be tailored to either enhance the device Q or maintain a certain kt2. The most important figure of merit (FOM) of an electromechanical resonator is in fact defined as the product of kt2 and Q (FOM ¼ kt2Q). The FOM is the main factor in determining the performance of the system in which it will be used, and affects parameters such as insertion loss and bandwidth in a filter, and power consumption and phase noise in an oscillator. Different from the pure piezoelectric device, the composite structure permits, within fabrication limits, the tuning of kt2 values and Q by varying the thickness of Ts. Various thin-film piezoelectrics have been used for making these resonators, with zinc oxide (ZnO), lead zirconate titanate (PZT), and aluminum nitride (AlN), being the most commonly available. The selection of one over the others is dictated by material properties, process compatibility, reproducibility, and specific application of interest. For example, PZT offers a very large kt2, but exhibits very low Q and therefore is exclusively used in a composite configuration with other materials such as silicon. AlN is easily sputterdeposited and has very reproducible properties with high Q and intermediate values of kt2, therefore often making it a preferred material for the fabrication of laterally vibrating resonators formed by pure piezoelectrics. The piezoelectric layer thickness can range between 100 and 3 mm. The structural layer forming the composite can be as thick as 50 mm for low frequency (10–200 MHz) devices made out of silicon, but it is generally thinner for higher frequency (0.5–3 GHz) devices and in the few mm range if made out of materials such as SiC or diamond. The in-plane dimensions of the device are generally in the 50–200 mm range for each side if arranged in a rectangular geometry.
Laterally Vibrating Piezoelectric Resonators
Laterally Vibrating Piezoelectric Resonators, Fig. 3 Equivalent electrical circuit for an electromechanical resonator
This area is kept rather constant for devices operating at different frequencies so that the device capacitance does not change significantly. At higher frequency (>200 MHz), this is attained by effectively arraying many of the basic aforementioned elements (which are called fingers in this configuration) and exciting adjacent elements 180 out of phase with respect to each other so as to induce an overtone vibration in the overall plate. In this case, the standing wave generated into the body of the resonator closely resembles a symmetric Lamb wave. From a practical standpoint, this is achieved by simply patterning small feature electrodes on top of the piezoelectric plate and applying opposite polarity voltage to the adjacent fingers (Fig. 1b or d). The previous sections intuitively describe the principle of operation of these resonators. More rigorously, an equivalent electrical model can be used to describe the device operation as a passive electronic component. The simplest equivalent circuit (Fig. 3) is formed by a capacitor representing the physical transducer in parallel with a series LCR, which models the mechanical portion of the resonator. The circuit components are completely defined by the geometry of the resonator and its electromechanical coupling and are given by the following expressions: RM ¼
1 2 oo Co p8 kt2 Q
CM ¼
8 2 k Co p2 t
LM ¼
1 o2o CM (3)
Methods of Fabrication Laterally vibrating piezoelectric resonators are generally fabricated by micromachining techniques.
Laterally Vibrating Piezoelectric Resonators
Laterally Vibrating Piezoelectric Resonators, Fig. 4 Schematic representation of the fabrication process used for the making of pure AlN resonators. (a) Bottom electrode is deposited and patterned; AlN is sputter-deposited and (b) access
Although each process has its own specific implementation, there are common steps. In particular, standard integrated circuit (IC) processes are used for the deposition and patterning of the metals that sandwich the piezoelectric film. The deposition of the piezoelectric film is one of the most challenging steps and requires the most attention since the most characteristics of the resonators are affected by the property of this film. ZnO and AlN are preferentially sputter-deposited, whereas a sol-gel technique is used for the deposition of PZT films. The as-deposited film quality is generally assessed by monitoring its orientation via x-ray diffraction measurements (note that these films are polycrystalline). A rocking curve measurement is generally performed with the same equipment to quantitatively assess the degree of orientation. A film is considered to be of good quality when the full width half maximum (FWHM) of the rocking curve measurement is less than 2 . Although higher values of the FWHM will likely not affect the piezoelectric properties of the film, they will further increase the Q of the device (intrinsic material damping). Specialized recipes have been developed for the etching of the piezoelectric layers, but ultimately use gases and equipment that are readily available in any IC fabrication facility. It is worth noting that compatibility with standard IC processes is of paramount importance for enabling large-scale manufacturing of these devices, which will be produced in the billions if commercialized for filtering or frequency setting applications. In addition, a longer term vision is to integrate these mechanical devices with electronics (mainly complementary metal oxide semiconductors [CMOS]) so that they are readily available to circuit
1195
L
to the bottom electrode is defined; (c) top electrode is deposited and patterned; (d) the main resonator body is defined by a dry etching step and the structure is released from the substrate
L Laterally Vibrating Piezoelectric Resonators, Fig. 5 Scanning electron microscope image of an AlN laterally vibrating micromechanical resonator
designers and can be broadly employed for design of low-loss components. In this case, AlN emerges as the most suitable candidate since it is very stable and is not considered to be a contaminant; it is already employed in some CMOS facilities. A simple schematic of the fabrication process for an AlN laterally vibrating resonator is shown in Fig. 4. Figure 5 is a scanning electron micrograph of the same type of AlN devices. This is an individual resonator formed by an AlN films sandwiched by three patterned metal electrodes, so as to excite an acoustic wave in the plane of the film and operate at a frequency around 200 MHz. In this case, the device release is straightforward and is generally performed by a dry fluorine etch chemistry that removes the underlying silicon. In the case of a composite structure, an additional etch step is required to define the structural layer. Bulk micromachining of silicon via a Bosch-type process, which etches through the entire substrate, has
L
1196
Laterally Vibrating Piezoelectric Resonators
Laterally Vibrating Piezoelectric Resonators, Fig. 6 Scanning electron microscope images of two types of composite resonators formed by (a) AlN on Si (courtesy of Prof. R. Abdolvand, Oklahoma State University) and (b) PZT on Si (Courtesy of Dr. R. Polcawich, Army Research Laboratory)
generally been used when Si is present. Examples of microfabricated structures are shown in Fig. 6. SiC and diamond are generally thinner and standard reactive ion etching has been employed for their patterning. In the case of a composite structure incorporating silicon, generally a silicon dioxide sacrificial layer is employed and a wet etch step is used to remove it to accomplish device release.
Main Research Accomplishment The main accomplishments in terms of device performances such as kt2, Q and frequency of operation are reported herein for the pure piezoelectrics and the composite laterally vibrating resonators. In terms of pure piezoelectrics, the reproducibility of AlN deposition and its compatibility with CMOS fabrication make it the piezoelectric of choice for laterally vibrating resonators. These devices have exhibited kt2 around 2% and quality factors as high as 4,000 in ambient conditions [1, 9]. These kt2 and Q values are sufficient to satisfy stringent power and phase noise requirements for oscillators and have proven adequate to meet the specifications for narrowband filtering. More recently the frequency of operation of these resonators has been extended around 10 GHz [10, 11] and there is no fundamental limit to operating them at higher frequencies. This implies that this technology is effectively broadband and could be employed to synthesize filters and oscillators for a wide range of communication systems that operate over different frequency bands. In terms of composite resonators, some of the best results have been attained by the integration of AlN on Si2. These devices exhibit kt2 that are generally pffiffiffiffi op [2]), then q < kd o ed =c, where kd is the wave number of the bulk wave in the adjacent dielectric medium. This means that the surface plasmon must leak into the bulk wave in the dielectric medium at an angle given by the ratio: cos(y) ¼ q/kd. This angle is determined by the Snell’s law that represents the momentum conservation in electrodynamics [5]. In this case, the plasmon is not a surface wave localized near the metal surface and guided by this surface, but is rather a leaky wave that propagates, rather than exponentially decays, in the adjacent dielectric medium. If condition (3) is not satisfied but em < 0, then q is imaginary (see Eq. 1) and the surface plasmon does not exist either. Because the wave number q for a surface plasmon is larger than the wave number kd for the bulk wave in the adjacent dielectric (see Eq. 1), the tangential component of the wave vector of the bulk wave (kd)t can never be equal to the wave vector q of the surface plasmon. Therefore, the Snell’s law (that requires the equality of the tangential components of the wave vectors of the waves involved [5]) and the momentum conservation can never be satisfied by a bulk wave and a surface plasmon (having the same frequencies) on a flat smooth metal surface. Therefore, surface plasmons on a smooth flat surface are not coupled to bulk radiation. However, surface plasmons can be generated by bulk radiation on surface nonuniformities (such as, for example, an edge of the surface – endfire excitation of surface plasmons [1]), or in periodic gratings, or in the presence of a prism coupler in the Otto or Kretschmann geometries [4]. The recent years have seen an explosion of theoretical and experimental research in the area of surface plasmons and their applications for integrated nanooptics: development of a new generation of nanooptical sensors, near-field imaging techniques with the spatial resolution as good as a few nanometers, manipulation and trapping of small nanoparticles and separate molecules, detection and imaging of separate molecules, design of nanomechanical systems, etc. These areas of application and research involving surface plasmons have started to rapidly expand when
L
it was realized that metallic nanosructures could guide and concentrate plasmon energy far beyond the diffraction limit of light and within nanoscale regions as small as a few nanometers. The first demonstrations of such unique capabilities were the original papers on sub-wavelength guiding using plasmons in metal nanorods and metallic nanoholes [6], and plasmon nanofocusing in gradually tapered metal structures [7]. In fact, the smallest dimensions of a localization region for surface plasmons are limited only by the atomic structure of matter, dissipation in the metal, and spatial dispersion, i.e., nonlocal dielectric response of materials [5, 8].
Key Research Findings Sub-wavelength Waveguiding Modern electronic is fast approaching its fundamental speed and bandwidth limits. Therefore, further increase of efficiency and speed of informationprocessing systems relies upon the development of new approaches and physical principles. One of the most feasible current options is based on using light signals and optical circuits instead of their less efficient electronic counterparts. However, one the major current problems of integrated optics is the insufficient level of miniaturization of optical components and interconnects, which is not even close to that typically achieved in electronics. Therefore, design of efficient sub-wavelength waveguides, interconnects, and optical devices with the localization of light signals within spatial regions with dimensions much smaller than the wavelength is the major goal of the modern integrated nanophotonics. As indicated above, unfortunately, dielectric waveguides do not normally allow such localization. For example, though an optical fiber can be made arbitrarily thin, this does not lead to sub-wavelength localization of the fundamental mode guided by this fiber. As the diameter of the fiber is reduced to zero, the field of the guided mode extends further into the surrounding medium (Fig. 1a), and the size of the guided mode tends to infinity. No two such fibers can be placed close to each other, as the overlapping guided modes in these fibers will produce a significant crosstalk and signal interference. A completely different situation occurs in subwavelength plasmonic waveguides, such as, for
L
L
1204
a
Light Localization for Nano-optical Devices
b d > λ0
d > λ 0p
d
a
TM plasmon
SP
εs λ
b
λp
e
es
mon
CPP
as e pl
dg We
d ∼ λ0
β
mod
)
) al (ε m
(WP
Met
d ~ λ0p
f
c WP
d < λ0
h
Slot
es
mod
εs d < λ 0p
Light Localization for Nano-optical Devices, Fig. 1 The differences between the dielectric fibers and metallic nanowires in achieving sub-wavelength waveguiding. (a) As the fiber diameter d decreases from infinity to zero, the wavelength of the guided fundamental mode l monotonically increases from the wavelength in the material of the fiber l0 to the wavelength in the dielectric medium surrounding the fiber. Simultaneously, the typical mode size reduces as the fiber diameter is reduced to about l0 and then starts to increase to infinity (as the fiber diameter tends to zero) – no sub-wavelength localization. (b) As the diameter of a cylindrical metal nanowire d decreases from infinity to zero, the wavelength of the guided fundamental TM plasmonic mode lp monotonically decreases from the plasmon wavelength on the flat metal surface l0p to zero. Simultanously, the typical mode size also monotonically decreases to zero, leading to sub-wavelength waveguiding
example, cylindrical metal nanowires (Fig. 1b). Decreasing diameter of a metal nanowire results in a monotonic decrease of the wavelength of the fundamental guided plasmon mode having the TM polarization. This leads to a monotonic increase of the mode localization or, equivalently, to a monotonic decrease of the dimensions of the localization region for the fundamental mode (these regions are shown by the dashed lines in Fig. 1). Several different guiding plasmonic structures have been proposed and analyzed in detail. These included
w
Light Localization for Nano-optical Devices, Fig. 2 Major examples of different types of plasmonic sub-wavelength waveguides: (a) metal nanostrip on a dielectric substrate; (b) free-standing sharp metal wedge surrounded by a uniform dielectric, guiding strongly localized wedge plasmons; (c) tapered section of a metal film in the shape of a metal wedge on a dielectric substrate with the permittivities of the substrate and cladding being es and ec, respectively; (d) cylindrical nanorod or cylindrical nanowire guiding strongly localized TM plasmons; (e) V-groove of a metal surface, guiding strongly localized channel plasmon-polaritons (CPP modes); and (f) slot plasmonic waveguide in a metal membrane of thin film guiding strongly localized slot modes
metal nanorods (cylindrical nanowires), nanostrips on a dielectric substrate, sharp metal wedges guiding wedge plasmons, metallic grooves guiding a special type of strongly localized plasmons called channel plasmon-polaritons (CPP modes), or nanoslots in a thin metal membrane or film [1] – see Fig. 2a–f. It is important to note that not all plasmonic modes guided by different types of metallic guiding nanostructures can experience strong sub-wavelength localization. For example, a thin metal film surrounded by a uniform dielectric can guide two different types of plasmonic modes – with symmetric and antisymmetric charge distribution across the film. If the thickness of the film is reduced to zero, it is only the plasmon with the symmetric charge distribution that experiences increased confinement to the metal film (i.e., increased localization). The other mode with the antisymmetric charge distribution across the film – the so-called longrange surface plasmon – becomes less localized and in
Light Localization for Nano-optical Devices
a
bc
b
|E| (a.u.)
0.6 0.2 0.4 0 0.2 –0.2
0 –0.2
the limiting case of zero thickness of the metal film transforms into a bulk wave propagating in the surrounding dielectric. These are the only strongly localized (“shortrange”) plasmonic modes that present a significant interest for the design of sub-wavelength waveguides, interconnects, and nano-optical devices. The indicated plasmonic structures including metal strips, nanorods, metal grooves, wedges, and nanoslots are all capable of guiding strongly localized plasmonic modes. The localization of the guided “short-range” plasmons increases with decreasing geometrical dimensions of the guiding structure, such as slot width and film thickness for a nanoslot waveguide, taper angle for a V-shaped metallic groove or a metal wedge, diameter of a nanorod, and width and thickness of a metal strip. For example, in the approximation of continuous electrodynamics, the dimensions of the guiding region for a strongly localized plasmon mode near the tip of a V-shaped metal groove or metal wedge tend to zero with decreasing taper angle. In fact, there is a critical value of the taper angle b ¼ bc:
L
1
μm
Light Localization for Nano-optical Devices, Fig. 3 Typical distributions of the magnitude of the electric field in subwavelength plasmonic waveguides: (a) fundamental mode in a 100 100 nm slot waveguide in a silver membrane in vacuum at lvac ¼ 632.8 nm (He-Ne laser); (b) fundamental mode in a 30o V-shaped 314 nm deep groove filled with vacuum on a silver surface at lvac ¼ 632.8 nm
1205
2ed =Reðem Þ; for V grooves; ðes þ ec Þ=Reðem Þ; for wedges;
(4)
below which the guided mode does not exist because of its infinite localization near the infinitely sharp tip. Here, ed is the permittivity of the dielectric medium in the metal groove (Fig. 2e), and es and ec are the permittivities of the dielectric substrate and cladding for a tapered section of a metal film in the form of a metal wedge (Fig. 2c).
0 μm
0.2
0
–0.2
0 μm
0.2
The typical distribution of the magnitude of the electric field in the fundamental plasmonic mode of a nanoslot of 100 nm width in a silver membrane of thickness 100 nm suspended in vacuum is shown in Fig. 3a for the vacuum wavelength lvac ¼ 632.8 nm (He-Ne laser) and metal permittivity em ¼ 16.2 + 0.5i. The localization of the field in this case is 150 nm, which is significantly smaller than the vacuum wavelength. Reducing the gap width and film thickness to 25 nm results in a significant reduction in the typical dimensions of the field localization region below 50 nm. The typical field distribution in the fundamental CPP mode in a V-groove waveguide of finite depth is shown in Fig. 3b, also illustrating guiding with sub-wavelength localization. Increasing localization of a guided plasmonic mode comes at a price of increased dissipative losses and thus decreased propagation distance. For example, for the 100 100 nm slot the propagation distance of the fundamental mode (Fig. 3a) appears to be 15 mm, whereas for the 25 25 nm slot it is 2.3 mm. Similar situation occurs for the other types of the plasmonic waveguides, such as V-groove and wedge waveguides [1]. Significant dissipation of the propagating signals in plasmonic sub-wavelength waveguides is a major problem of integrated plasmonics. On the other hand, the typically achievable propagation distances of around at least a few microns are sufficient for the design of interconnects between highly integrated nano-optical devices, or interconnects between the existing optical telecommunication systems and nano-electronic devices and components. However, the development of all-optical information-processing
L
L
1206
integrated circuits with high level of integration will require a resolution of the dissipation problem. One of the current approaches to the dissipation problem is to use materials with gain to compensate for the dissipation. However, the high level of dissipation of strongly localized plasmons 1,000 cm1 and higher is above what could be expected to be compensated by most gain media. Therefore, dissipation compensation or partial dissipation compensation was claimed for rather weakly localized plasmonic modes [1, 9] that cannot normally be used to design nanooptical devices and interconnects. Therefore, strong dissipation of localized plasmonic modes in subwavelength waveguides is a standing problem that currently imposes limitations upon the design of highly integrated nanophotonic circuits for efficient alloptical signal processing. Strong dissipation of plasmonic modes in metallic nanostructures is also an obstacle for the development of nanoscale optical sources. An efficient nanoscale optical source that would result in stimulated and coherent emission of surface plasmons would be a major technological step in nanophotonics. Such a source of surface plasmons is called “spaser” using the analogy with a “laser” as a light source based on stimulated emission of photons. A significant recent advancement in this direction was the first experimental observation of stimulated plasmon emission in gold nanoparticles covered in high-brightness luminescent dye doped silica shells [10]. Each of such gold-silica particles worked as a spaser nanocavity with the dimensions of 44 nm. Using the optical pumping, plasmon oscillations are stimulated in each of the goldsilica nanoparticles, and outcoupling of these oscillations into radiative photonic modes constituted a nanolaser operating in the visible range and having the cavity dimensions of 44 nm [10]. Coupling such a spaser cavity to a sub-wavelength plasmonic waveguide could lead to a plasmon source in a plasmonic information-processing circuit. Plasmon Nanofocusing A special area of interest and significant applications of strongly localized plasmons lies with plasmon nanofocusing. Nanofocusing is the process of concentration and localization of electromagnetic waves by means of tapered metallic nanostructures into regions as small as a few nanometers, i.e., far beyond the diffraction limit of light [1, 2]. Plasmon nanofocusing
Light Localization for Nano-optical Devices
occurs when a strongly localized plasmonic mode propagates along a gradual taper of a plasmonic guiding structure. For example, if a plasmon with TM polarization propagates along a tapered conical rod, it experiences a gradual increase in its localization combined with a significant local field enhancement, despite the dissipative losses in the metal [1, 2]. Disregarding the atomic structure of matter and spatial dispersion, the local field enhancement tends to infinity, and the localization region tends to zero as the infinitely sharp tip of the conical rod is approached. Realistically, the field localization region and local field enhancement are limited by the smallest practically achievable radius of the tip of the metal cone, which is typically limited by a few nanometers. Similar nanofocusing effects can be achieved in other tapered metal structures including tapered metal gaps, metal wedges, tapered metal strips on a dielectric substrate with gradually reducing width and/or thickness of the strip, etc. All of these structures are typically characterized by a significant increase of plasmon localization and enhancement of its local field near the tip of the taper. An interesting case of nanofocusing of channel plasmon-polariton modes was considered and observed experimentally in V-shaped metal grooves with gradually reducing taper angle and depth. In this case, the channel plasmon-polariton mode guided by the groove experienced gradual increase in its localization, combined with the local field enhancement, caused by the gradually changing groove parameters [1]. Nanofocusing is typically efficient if the taper angle of the metal structure is sufficiently small, such that: 1 @Q1 @z bc) in infinitely sharp metal V-grooves and wedges are exactly opposite. Optimal parameters for different nanofocusing structures were determined from the adiabatic approximation and rigorous numerical analysis. For example, if the taper angle is much smaller than bc, then adiabatic condition (5a) is satisfied, but the plasmon must propagate large distances along the taper until it reaches the tip. This means that it will experience large dissipative losses, and the local field enhancement at the tip will not be the strongest. On the other hand, if the taper angle is large, so that condition (5b) is not satisfied, then the plasmon will experience significant reflections from the taper, and only part of its energy will be able to reach the tip. This will also lead to a reduction of the achievable local field enhancement at the tip. Therefore, there is an optimal taper angle b ¼ bopt at which the local field enhancement at the tip is the strongest. For example, for a gold cone in vacuum at lvac ¼ 632.8 nm (He-Ne laser) and bopt 35o [1]. Optimal taper angles for gaps (20o) tend to be smaller than for cones, but larger than for wedges (7o). The optimal taper angle for a tapered gap or a metal wedge is close to the critical taper angle below which adiabatic nanofocusing takes place, i.e., bopt bc. This is expected, as both the angles are physically related to the same phenomenon – the appearance of significant reflections of the plasmon from the taper. Another important parameter is the optimal length of the taper Lopt. If the plasmon is generated too far from the tip, it has to propagate a large distance until it
L
reaches the tip and starts experiencing significant focusing effects. This leads to excessive dissipative losses. If the plasmon is generated too close to the tip, it does not have “an opportunity” to take all the advantage of the nanofocusing effect to enhance its field. Therefore, there is an optimal distance from the tip of a nanofocusing structure (or the optimal length of the nanofocusing structure Lopt), at which a localized plasmon mode should be generated to ensure the maximal possible local field enhancement at the tip. For example, the optimal length of a nanofocusing conical metal rod can be determined by a simple equation [11]: Lopt ¼
e1 ðed þ e1 Þ cosðb=2Þ; e2 ed Q0
(6)
where e1 and e2 are the real and imaginary parts of the metal permittivity em ¼ e1 + ie2, ed is the permittivity of the dielectric medium surrounding the metal cone, and Q0 is the real part of the plasmon wave number on a flat and smooth interface between the considered metal and dielectric medium. Equation 6 is applicable if the taper angle b is not too small: sinðb=2Þ >>
e 2 ed : e1 ðed þ e1 Þ
(7)
For example, Eq. 6 gives Lopt 10 mm for a gold cone in vacuum at lvac ¼ 632.8 nm, and this estimate is correct within a broad range of taper angles 20o < b < 160o. Typical achievable local field enhancements at the tip may vary for different nanofocusing structures. For example, conical rods represent one of the best options for plasmon nanofocusing, resulting in the largest local field enhancements up to 1,600 times for a gold rod in vacuum [1] with the localization of the enhanced field within a region of 5 nm. Tapered gaps are less efficient from this point of view and typically yield the local field enhancements of 50–100 times. Metal wedges are the least efficient nanofocusing structures compared to the gaps and conical rods with the typical enhancement factors of 10. Further enhancement of the local field at the tip can be achieved through optimizing the shape of a nanofocusing rod. For example, it has been demonstrated that if an annually converging plasmon is initially generated on a flat surface and then coupled into a conical rod (made in the form of
L
L
1208
a wizard hat), the local field enhancement at the tip could be as strong as 10,000 [12]. Importantly, such local field enhancements should lead to a Raman signal enhancements from Ramanactive molecules near the tip of the structure up to 1016, which is 2 orders of magnitude larger than what is required for the single molecule detection. Therefore, the analyzed nanofocusing structures may present a unique opportunity for the development of reliable structures for sensing, detection, and imaging of single molecules and traces of substances in the environment surrounding the nanofocusing structure [12]. Strong concentration of the local field near the tip of a nanofocusing structure provides not only the substantial field enhancements, but also enormous field gradients within the distances of about a few nanometers around the tip. This opens a unique opportunity for near-field trapping and manipulation of small nanoparticles and separate molecules with the efficiency up to 4 orders of magnitude higher than those achievable by other means [12].
Examples of Application This naturally leads to the discussion of numerous and important applications of plasmon nanfocusing. As indicated above, one of the major areas of these applications is the development of a new generation of nano-optical sensors and near-field optical imaging techniques with the spatial resolution as good as a few nanometers – far beyond the diffraction limit of light. These will likely to include a new generation of nano-optical sensors for the detection and identification of traces of explosives, drugs, and chemicals in the environment for security and environmental monitoring. Another major area of future applications of nanofocusing will involve cellular and intercellular manipulations, imaging and testing in nanobiophotonics and medicine. These might include new techniques for DNA manipulations and diagnostics, mapping of the local refractive index in a living cell, testing of macromolecules, intercellular fluids and membranes, highly targeted delivery of electromagnetic energy to nanoscale regions of a living cell with equally localized and controlled heat discharge, etc. A new generation of nonlinear devices is expected to emerge from plasmon nanofocusing structures, such as nonlinear all-optical switches, nanoscale bolometers,
Light Localization for Nano-optical Devices
active and passive optical devices, new techniques for the investigation of the local material response to highly localized heat discharge on the femtosecond scale, etc. Another important practical application of nanofocusing structures will be related to their natural capability of efficiently coupling relatively large-scale light signals from the conventional optical communication systems to nanoscale electronic chips for further information processing. This means overcoming the major problem with the efficient delivery of light energy in the visible and infrared ranges to nanoscale devices, structures, and separate molecules. A separate area of applications and further research is related to the analysis of single plasmon propagation and focusing, quantum and nanomechanical systems. A range of other nano-optical devices and techniques based on strongly localized plasmonic modes in metallic nanostructures has been considered and analyzed, including nanoscale plasmonic resonators, filters, couplers, optical adaptors, elements for integrated optical circuitry, nanoantennas, nanolasers, photodetectors and photovoltaic devices, heat-assisted magnetic recording, etc. [1, 2].
Cross-References ▶ Active Plasmonic Devices ▶ Nanostructures for Photonics ▶ Scanning Near-Field Optical Microscopy ▶ Surface Plasmon-Polariton-Based Detectors ▶ Theory of Optical Metamaterials
References 1. Gramotnev, D.K., Bozhevolnyi, S.I.: Plasmonics beyond the diffraction limit. Nat. Photonics 4, 83–91 (2010) 2. Schuller, J.A., Barnard, E.S., Cai, W., Jun, Y.C., White, J.S., Brongersma, M.L.: Plasmonics for extreme light concentration and manipulation. Nat. Mater. 9, 193–204 (2010) 3. Kittel, C.: Introduction to Solid State Physics, 4th edn. Wiley, New York (1971) 4. Raether, H.: Surface Plasmons on Smooth and Rough Surfaces and on Gratings. Springer, New York (1988) 5. Landau, L.D., Lifshitz, E.M.: Electrodynamics of Continuous Media. Pergamon, Oxford (1984) 6. Takahara, J., Yamagishi, S., Taki, H., Morimoto, A., Kobayashi, T.: Guiding of a one-dimensional optical beam with nanometer diameter. Opt. Lett. 22, 475–477 (1997) 7. Nerkararyan, K.V.: Superfocusing of a surface polariton in a wedge-like structure. Phys. Lett. A 237, 103–105 (1997)
Light-Element Nanotubes and Related Structures 8. Liebsch, A.: Screening properties of a metal surface at low frequencies and finite wave vectors. Phys. Rev. Lett. 54, 67–70 (1985) 9. Gather, M.C., Meerholz, K., Danz, N., Leosson, K.: Net optical gain in a plasmonic waveguide embedded in a fluorescent polymer. Nat. Photonics 4, 457–461 (2010) 10. Noginov, M.A., Zhu, G., Belgrave, A.M., Bakker, R., Shalaev, V.M., Narimanov, E.E., Stout, S., Herz, E., Suteewong, T., Wiesner, U.: Demonstration of a spaserbased nanolaser. Nature 460, 1110–1112 (2009) 11. Gramotnev, D.K., Vogel, M.W., Stockman, M.I.: Optimized nonadiabatic nanofocusing of plasmons by tapered metal rods. J. Appl. Phys. 104, 034311 (2008) 12. Gramotnev, D.K., Vogel, M.W.: Ultimate capabilities of sharp metal tips for plasmon nanofocusing, single molecule detection, near-field trapping and sensing. Phys. Rev. Lett. (to be published)
Light-Element Nanotubes and Related Structures Wenlong Wang1, Xuedong Bai1 and Enge Wang2 1 Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing, P. R. China 2 International Center for Quantum Materials, School of Physics, Peking University, Beijing, China
Synonyms Boron- and/or nitrogen-doped carbon nanotubes
Definition Light-element nanotubes, as specifically defined here, refer to boron (B) and/or nitrogen (N) substitutionally doped carbon (C) nanotubes (C-NTs), including both the selectively doped CBx and CNx nanotubes and the co-doped ternary system BxCyNz nanotubes. For either the CBx or CNx nanotubes, it is expected that high dopant concentrations will lead to new phases with well-defined stoichiometry; the two most eminent examples are BC3 and C3N4 nanotubes. These stoichiometric nanotubes cannot be simply treated as highly B- or N-doped carbon nanotubes as their structures and electronic properties are inherently different than the pristine C-NTs.
1209
L
Introduction The discovery of carbon nanotubes (C-NTs) in the early 1990s has evoked much interest in nanotubular materials as an important frontier area of research that is still flourishing [1–4]. Shortly after the discovery of C-NTs, it was experimentally confirmed that substitutional doping of C-NTs with B and/or N heteroatoms was feasible [5, 6]. As the nearest neighbors of C in the Periodic Table and given their size proximity to C, B, and N, probably the only elements that can be substitutionally doped in a hexagonal C network without severely distorting the lattice. Furthermore, considering the isoelectronic and isostructural relationship between a B–N pair and a C–C pair, it is expected that the simultaneous B and N doping (i.e., co-doping) of C-NTs through partially substituting for C–C units in the tube-shell lattice will lead to the formation of ternary system BxCyNz nanotubes. To some extent, the BxCyNz nanotubes can be treated as random or ordered hybrids or solid solutions of C-NTs and boron nitride (BN) nanotubes [1–11]. On the other hand, selective doping of C-NTs with B or N alone will give rise to binary CBx and CNx nanotubes, respectively [12–15]. BxCyNz, CBx, CNx nanotubes, along with pristine C and BN nanotubes, constitute the family of lightelement nanotube materials. As there have been many reviews available for C and BN nanotubes, the focus of this entry will be constrained to the doped system BxCyNz, CBx, and CNx nanotubes. Theoretical and experimental aspects concerning their crystal structures, electronic structures, electrical transport properties, state-of-the-art synthesis, characterization of bonding environments, and the atomic distribution of dopants will be surveyed.
Structure and Electronic Structure The Ternary System BxCyNz Nanotubes The existence of stable layered bulk structures is usually suggestive of the existence of corresponding stable nanotubular structures. Topologically, a single-walled carbon nanotube (C-SWNT) can be constructed by rolling a single layer of graphite (i.e., graphene) into a seamless tube. Likewise, one can view a BN-SWNT as a rolled up sheet of sp2-bonded boron nitride (h-BN). Known as the III-VI homology of
L
L
1210
C nanotubes, BN nanotubes are structurally identical to their C counterparts except that the C atoms are fully substituted by alternating B and N atoms [1, 3]. However, despite the structural similarity between C and BN nanotubes, their electronic band structures are vastly different. It is well-known that pristine C-NTs can be either semiconducting or metallic, with electronic band structures being sensitively determined by specific tube geometries, that is, diameters and chirality, whose control remains a formidable (if not impossible) challenge for all known synthetic methodologies. By contrast, as a consequence of the ionic character of B–N bond, BN nanotubes are electrical insulators with a large electronic band gap of around 5.5 eV that is almost independent of the diameter and the chirality of the BN tube [1, 3]. It is therefore inferred that the intermediate ternary system BxCyNz nanotubes will exhibit tunable semiconducting properties in between metallic or small band gap C-NTs and insulating BN-NTs [7–11]. Importantly, the electronic band structures of BxCyNz nanotubes are governed largely by the B/C/N chemical stoichiometry rather than by the specific geometry [7–11]. As the most prominent example of BxCyNz, BC2N nanotubes have been theoretically studied by Miyamoto et al. through first principles and tightbinding (TB) band structure calculations [7, 8]. They showed that BC2N nanotubes would be semiconductors and could be either p- or n-doped depending on the detailed B and N concentrations [8]. In a later study, the evolution of electronic band structures of BxCyNz nanotubes as a function of the B and N concentrations were systematically evaluated within a TB framework by Yoshioka et al. [9]. This relatively simple model gave essential qualitative features of the electronic structures of BxCyNz nanotubes with varied stoichiometry. The calculated results clearly showed the opening of a band gap for metallic (9, 0) nanotubes and the band gap increasing for a semiconducting (10, 0) nanotubes with increasing (x + y) concentration [9]. In a recent study by the present authors, Xu et al. have performed a series of ab initio band structure calculations of BxCyNz nanotubes in a framework of generalized gradient approximation (GGA) [11]. As a first step, various co-doped structures were searched, and as expected, it was found that B and N atoms have a strong preference to neighbor each other and form B–N pairs. The total energy of a unit cell containing an adjacent B–N pair is about 2.0–2.6 eV lower than those
Light-Element Nanotubes and Related Structures
with separately doped B and N atoms at isolated sites. By starting from a (5,5) metallic C-SWNT, the incorporation of one B–N pair in a unit cell (with B or N component ratio of 5%) resulted in the opening of a small band gap of about 0.1 eV at the Fermi level. With further incorporation of one more B–N pair to the unit cell (the two adjacent B–N pairs are separated by six carbon atoms), the energy gap is widened to about 0.5 eV. These results clearly revealed a transition of electronic structure of the (5, 5) nanotube from metallic to semiconducting after B–N co-doping. Other metallic SWNTs at different B–N co-doping concentrations were also calculated, and the results showed a similar band structure crossover from metallic to semiconducting, independent of the chirality of the nanotubes despite the different gap values. The gap opening is found to be fairly sensitive to B/N components and widens monotonically with increasing B/N co-doping concentration. The detailed geometry of the B/N atomic pairs may change the width of the gap slightly, but the overall trend remains the same [11]. Selectively Doped CBx and CNx Nanotubes When considering selective doping of C-NTs solely with electron-deficient B or electron-rich N, one is reminded of the substitutional doping of Silicon (Si) upon which microelectronics relies. The selective doping of Si with substitutional acceptor or donor type atoms results in a sensitive shift of the Fermi level, a concept known as the rigid-band model (i.e., shifting the Fermi level without altering the band structure). However, it is important to note that the doping levels in conventional Si technology are very low, typically on the order of parts per million (ppm). For B- or N-doped nanotubes, the rigid-band model is valid only for a very low dopant concentration [12–15]. In general, as the typical dopant concentrations are on the order of a percent and tenths of a percent, the substitutional B or N doping of nanotubes cannot be solely explained in the framework of a rigid-band model, and the modifications of band structures and density of states (DOEs) must be taken into account [12–15]. Structurally, the B dopant is expected to be incorporated into the honeycomb tube-shell lattice primarily by direct substitution of C atoms with a triplecoordinated bond [12, 13]. In the case of low B concentration, the formation of an acceptor state in the band gap just above the valence edge is expected. For instance, an early theoretical work by Yi and
Light-Element Nanotubes and Related Structures
Bernholc in 1993 has revealed an acceptor state located at 0.16 eV above the Fermi energy for a (10, 0) nanotube with 1.25 wt.% B concentration [12]. In a more recent study by Wirtz and Rubio, the authors theoretically studied the evolution of an acceptor-like level that is caused by a periodic distribution of the B dopant as a function of the B-doping ratio. They found that doping profoundly affected the physical properties [13]. Specifically, for pristine C-SWNTs, the DOS is symmetric around the band gap for the first and second van Hove Singularities (vHSs), whereas the incorporation of more B dopants led to an increasing asymmetry near the band gap due to the mixing of the p and s orbitals [13]. As far as N-doped CNx nanotubes are concerned, the situation is a little more complicated than an intuitive expectation that the incorporation of N heteroatoms will lead to n-type doping of nanotubes with localized states above the Fermi level [6]. This is the case only if a foreign N atom directly substitutes for a C atom on the honeycomb tube-shell lattice, in a bonding environment just like that of the aforementioned B dopants [6, 14, 15]. For example, a substitutional N dopant may also generate structural defects in the tube-shell lattice and lead to a rearrangement of the neighboring C atoms. In this case, it is not immediately obvious that n-type behavior will result, because the electrical behavior depends on the details of the new geometry. As a result, the asformed new tube-shell structure can be p type [6, 14, 15]. For either B- or N-doped nanotubes, it is expected that higher dopant concentrations may lead to new phases with defined stoichiometry, for example, BC3 and C3N4 [16–18]. Just like graphene sheets, the stability of the planar structures of these stoichiometric compounds could make it possible for nanotubes with the same stoichiometry to exist. Band structure calculations have revealed that individual BC3 nanotubes will act as semiconductors. Bundles of tubes are expected to exhibit metallic behavior due to intertube cohesive interactions [16]. As far as both crystallographic and electronic structures are concerned, the stoichiometric BC3 compound nanotubes cannot simply be regarded as highly B-doped nanotubes. Similarly C3N4 nanotubes, which are also theoretically predicted to be well-defined semiconductors, are inherently different from substitutionally doped CNx nanotubes [17]. To date, studies on BC3 and C3N4 nanotubes are limited to theoretical modeling. The
1211
L
experimental synthesis of these stoichiometric compound nanotubes is still elusive.
Synthesis of BxCyNz, BCx, and CNx Nanotubes Generally speaking, there are two different strategies for the production of B- and/or N-doped nanotubes, namely, direct synthesis and post-synthetic nanotubesubstitution reactions. The direct synthesis methods were in fact inherited from those that have been well established for pure C-NTs synthesis, mainly including arc-discharge, laser-ablation, and chemical vapor deposition (CVD) [19–33]. The post-synthetic nanotube-substitution reaction approach, on the other hand, refers to the use of pristine C-NTs as starting precursors that undergo carbothermal reactions with B2O3 vapor at elevated temperature (1,200–1,700 C) in N2 or NH3 atmosphere; this reaction process will induce partial substitution of the carbon lattice by foreign B and/or N atoms, thereby resulting in the formation of doped nanotubes [34]. Synthesis of B- and N-doped C-NTs in the form of multi-walled nanotubes (MWNTs) was reported by Ste´phan et al. early in 1994 using the arc-discharge method [5]. Since then, a great deal of research has been directed toward B- and/or N-doped nanotubes. However, despite the considerable progress in synthesis of doped MWNTs [1–4], substitutional doping of SWNTs turn out to be much more difficult, and it was not until very recently that reports of successful synthesis of high-quality B- and/or N-doped SWNTs have emerged [4, 10, 11, 19–33]. Typically, B- and/N-doped MWNTs suffer from large percentages of severe topological modifications and therefore usually exhibit highly defective, bamboo-like structural features. Doped SWNTs, however, normally exhibit straight, non-buckled tubular structures, close to pristine C-SWNTs [4, 10, 11, 19–33]. For SWNTs, it is likely that the B- and/or N doping introduces topological defects in the tube-shell, but not necessarily defective, bamboo-like structures. Direct Synthesis: Chemical Vapor Deposition Over the past decade, catalytic decomposition of carbon-containing precursor gases over transition metal catalyst nanoparticles in a CVD growth has proved to be an elegant strategy for the controlled synthesis of MWNTs and SWNTs. In comparison
L
L
1212
with the physical evaporation approaches such as arc-discharge and laser-ablation, CVD growth holds advantages in terms of experimental facility, potential for large-scale production at low cost, and control over tube location and orientation. CVD synthesis of B- and/or N-doped nanotubes has also been extensively exploited, but the majority of reported studies have been limited to doped MWNTs. In 2006, a breakthrough in CVD growth of ternary BxCyNzSWNTs was made by the present authors through a bias-assisted hot filament CVD (HFCVD) growth process [10]. BxCyNz-SWNTs growth by HFCVD was achieved over a high-activity powdery-type catalyst, namely, porous MgO supported Fe–Mo (denoted as Fe–Mo/MgO) with CH4, B2H6, and ethylenediamine vapor as the reactant gases and inert He as the diluting gas. Transmission electron microscopy (TEM) imaging showed that the as-grown SWNTs had a clean and smooth morphology with diameters in the range of 0.8–2.5 nm. In some cases, double-walled nanotubes (DWNTs) with slightly larger diameter coexisted with the SWNTs, but the amount was not significant. The electron energy loss spectroscopy (EELS) spectra taken from the SWNT bundles showed the core-loss K-edges of B, C, and N located at 188, 284, and 401 eV, respectively. The sharply defined 1s ! p∗ and 1s ! s∗ transition features of the C–K edge is characteristic of wellgraphitized sp2-bonding carbon networks. The B and N K-edges, though being much weaker, also showed discernible p∗ peak as well as s∗ band, indicating that the B and N atoms were in the same sp2-hybridized state as their C counterparts. Importantly, all as-grown BxCyNz-SWNTs had straight cylindrical tube walls, a morphological feature similar to that of pristine C-SWNTs. This is in sharp contrast to the previously reported defective, “bamboo-like” tube structures of doped MWNTs. Other than SWNTs and DWNTs, the HFCVD growth also produced a small amount of carbon onions and “bamboo-like” MWNTs. These byproducts were hard to remove completely but their amount could be effectively reduced by optimizing the growth parameters and the use of the Fe–Mo/ MgO catalyst [10]. In recent years, there have also been a few studies concerning the synthesis of N-doped SWNTs by conventional thermal CVD. Keskar et al. reported on the growth of isolated CNx-SWNTs on quartz and SiO2/Si substrates via a liquid injection CVD process by using
Light-Element Nanotubes and Related Structures
the mixture of xylene and acetonitrile (CH3CN, N precursor) as the feeding source [19]. The catalysts they used were nanoparticles deposited onto the SiO2/Si substrates prior to the growth. The resulting surface-grown SWNTs were characterized by resonant Raman spectroscopy. It was found that the intensity of the radial breathing mode (RBM) of the as-grown nanotubes decreased with increasing N precursor concentration in the feeding source, while the relative intensity of D/G-band increased with the appearance of the D’-band. The authors attributed these observations to the N doping in SWNTs [19]. In another CVD experiment, Villalpando-Paez et al. synthesized long strands of N-doped SWNTs through an aerosol-assisted CVD process using a mixture of ferrocene/ethanol/benzylamine, with benzylamine (C6H5CH2NH2) acting as the N precursor and ferrocene as the Fe catalyst precursor [20]. As the relative ratio of benzylamine to ethanol increased, the SWNT yield decreased gradually with the bundled SWNT strands getting shorter and thinner. At higher benzylamine concentrations, SWNT growth was completely inhibited. Neither EELS nor Energy Dispersive X-ray (EDX) spectra detected the presence of N in the as-grown SWNTs,, indicating that the N-doping ratio, if any, was rather low and likely below the detection limit of these instruments (e.g., 166 158 >150 150 169 168
Dip coating Spin coated Electrochemical deposition CVD Sol–gel
105 165 >150 159 >150
Lotus Effect, Table 2 Pros and cons of various fabrication techniques [2] Techniques Lithography Etching
Pros Accuracy, large area Fast
Deposition
Flexibility, cheap
Contact angle, 168 160 152
Technique Plasma Casting under humid environment Electro- and chemical polymerization Laser treatment Electrospining evaporation Nanoimprint Oxygen plasma etching Sol–gel Chemical etching Electrodeposition Photolithography E-beam lithography X-ray lithography Casting Plasma etching Sol–gel Replication and self-assembly Self-assembly
Cons Slow process, high cost Chemical contamination, less control Can be high temperature, less control
surfaces can help to resist destabilization by pinning the interface. It also helps in preventing the gaps between the asperities from filling with liquid, even in the case of a hydrophilic material. The effect of roughness on wetting is scale dependent, and mechanisms that lead to destabilization of a composite interface are also scale dependent. To effectively resist these scale-dependent mechanisms, it is expected that a multiscale roughness is optimum for superhydrophobicity.
Notes Transparent Reversible (electric potential)
Reversible (temperature) Hierarchical
Plant leaf replica For liquid flow Hierarchical
Hierarchical
Reversible (UV irradiation)
Ideal Surfaces with Hierarchical Structure For stability of a composite interface, Bhushan et al. [2] suggested the structure of an ideal hierarchical surface as shown in Fig. 3. The asperities should be high enough that the droplet does not touch the valleys. As an example, for a structure with circular pillars, the following relationship should hold for a composite interface. For a droplet with a radius on the order of 1 mm or larger, a value of H on the order of 30 mm, and D on the order of 15 mm, a P on the order of 130 mm is optimum. Nanoasperities can pin the liquid–air interface and thus prevent liquid from filling the valleys between asperities. They are also required to support nanodroplets, which may condense in the valleys between large asperities. Therefore, nanoasperities should have a small pitch to handle nanodroplets, less than 1 mm down to few nm radius. Values of h on the order of 10 nm and d on the order of 100 nm can be easily fabricated.
Lotus Effect
1229
L
Hierarchical structure using Lotus replica
10 μm
2 μm
0.8 μm
Hierarchical structure using micropatterned Si replica
10 μm
Lotus Effect, Fig. 4 SEM micrographs taken at 45 tilt angle (shown using three magnifications) of hierarchical structure using Lotus and micropatterned Si replicas. Nano- and
2 μm
0.8 μm
hierarchical structures were fabricated with mass 0.8 mg mm2 of Lotus wax after storage for 7 days at 50 C with ethanol vapor [3]
L Artificial Lotus-Effect Surfaces Two main requirements for a superhydrophobic surface are that the surface should be rough and that it should have a hydrophobic (low surface energy) coating. These two requirements lead to two methods of producing a superhydrophobic surface: first, it is possible to make a rough surface from an initially hydrophobic material and, second, to modify a rough hydrophilic surface by modifying surface chemistry or applying a hydrophobic material upon it. Note that roughness is usually a more important property than a low surface energy, since both moderately hydrophobic and very hydrophobic materials can exhibit similar wetting behavior when roughened [1]. In general, the same techniques that are used for micro- and nanostructure fabrication, such as lithography, etching, and deposition, have been utilized for producing superhydrophobic surfaces (Table 1). Pros and cons of these techniques are summarized in Table 2. An especially interesting development is the creation of switchable surfaces that can be turned from hydrophobic to hydrophilic by applying electric potential, heat, or ultraviolet (UV) irradiation. Another important requirement for potential applications to
optics and self-cleaning glasses is the creation of transparent superhydrophobic surfaces. In order for the surface to be transparent, roughness details should be smaller than the wavelength of visible light (about 400–700 nm) [1–3]. Fabrication using Nature’s Route – To verify the understanding of the mechanism responsible for the Lotus effect, first structures using nature’s routes were fabricated to verify that contact angle data of the fabricated structures were comparable to that of the Lotus leaves found in nature. Bhushan et al. [3] fabricated surfaces with a hierarchical structure with micropatterned epoxy replicas, and Lotus leaf microstructure, and created a second level of structuring with wax platelets and wax tubules. Platelets and tubules are the most common wax morphologies found in plant surfaces, and exist on superhydrophobic leaves, such as Lotus and Colocasia esculenta leaves. The structures developed mimic the hierarchical structures of superhydrophobic leaves. A two-step molding process was used to fabricate several structurally identical copies of the micropatterned Si surface and Lotus leaves. The technique used is a fast, precise, and low-cost molding process for biological and artificial surfaces.
1230
Static contact angle, contact angle hysteresis, tilt angle and adhesive force data Lotus wax
Static contact angle (deg)
180
Lotus replica
Micropatterned Si replica
Lotus leaf
150 120 90
Contact angle hysteresis (deg)
60 90
60
30
0 90
Tilt angle (deg)
Nanostructure was created by self-assembly of synthetic and plant waxes deposited by thermal evaporation. Hierarchical structures fabricated with microstructured Lotus leaf replicas and micropatterned Si replicas covered with a nanostructure of Lotus wax tubules are shown as Fig. 4. The grown tubules provide the desired nanostructure on microstructured surfaces. The recrystallized Lotus wax shows tubular hollow structures, with random orientation on the surfaces. Their shapes and sizes show only a few variations. The tubular diameter varied between 100 and 150 nm, and their length varied between 1,500 and 2,000 nm. To study the effect of Lotus wax tubule nanostructures on superhydrophobicity, static contact angle, contact angle hysteresis, and tilt angle were measured on flat, microstructured Lotus replica; micropatterned Si replica; and hierarchical surfaces. Hierarchical surfaces were made of the Lotus leaf replica and micropatterned Si replica with a nanostructure of wax tubules on top. Additionally, fresh Lotus leaves were investigated to compare the properties of the fabricated structures with the original biological model. Figure 5 shows that the highest static contact angles of 173 , the lowest contact angle hysteresis of 1 , and tilt angle varying between 1 and 2 were found for the hierarchical structured Si replica. The hierarchical structured Lotus leaf replica showed a static contact angle of 171 , and the same contact angle hysteresis (2 ) and tilt angles (1–2 ) as the hierarchical Si replica. Both artificial hierarchical structured surfaces show similar values as the fresh Lotus leaf surface investigated here, with static contact angle of 164 , contact angle hysteresis of 3 , and a tilt angle of 3 . However, the artificial hierarchical surfaces showed higher static contact angle and lower contact angle hysteresis. Structural differences between the original Lotus leaf and the artificial Lotus leaf produced here are limited to a difference in length of wax tubules, which are 0.5–1 mm longer in the artificial Lotus leaf. Adhesive force measured using a 15 mm radius borosilicate tip in an AFM also shows a similar trend as the wetting properties for the artificial surfaces. The adhesion force of the hierarchical surface structure was lower than that of the micro-, nanostructured, and flat surfaces because the contact between the tip and the surface was lower as a result of the contact area being reduced. However, for the fresh Lotus leaf, there is moisture within the plant material that causes softening of the leaf, and so when the tip comes into contact
Lotus Effect
60
30
0 400 Adhesive force (nN)
L
300 200 100 0
Flat
Nanostructure
Microstructure
Hierarchical sructure
Lotus Effect, Fig. 5 Bar charts showing the measured static contact angle, contact angle hysteresis and tilt angle on various structures fabricated with 0.8 mg mm2 mass of Lotus wax after storage at 50 C for 7 days with ethanol vapor. (d) Bar chart showing the adhesive forces for various structures, measured using a 15 mm radius borosilicate tip. Hierarchical structures were fabricated using Lotus and micropatterned Si replicas superimposed with nanostructures of Lotus wax tubules. The error bar represents 1 standard deviation [3]
with the leaf sample, the sample deforms, and a larger area of contact between the tip and the sample causes an increase in the adhesive force.
Lotus Effect
L
1231
Lotus Effect, Fig. 6 Bar chart showing the remaining particles after self-cleaning experiments applying droplets with nearly zero kinetic energy on various structures fabricated from Lotus wax using 1–10 and 10–15 mm SiC particles. The experiments on the surfaces with Lotus wax were carried out on stages tilted at 3 . The error bars represent 1 standard deviation [3]
Self-cleaning experiments using two different particle sizes Lotus wax Epoxy resin Microstructure (micropatterned Si replica)
Flat with thin wax layer Nanostructure
Hierarchical structure (Lotus replica)
Microstructure (Lotus replica)
Hierarchical structure (micropatterned Si replica)
100 At 3º tilt angle
Remaining particles (%)
80
60
40
20
0 1–10 μm SiC particles
10–15 μm SiC particles
L
Hierarchical structure with CNT
10 µm
2 µm
0.8 µm
Lotus Effect, Fig. 7 SEM micrographs taken at 45 tilt angle showing three magnifications of hierarchical structures fabrication with CNT-epoxy composite [7]
To test self-cleaning efficiency of hierarchical structured surfaces, silicon carbide particles in two different size ranges of 1–10 and 10–15 mm used as the contaminant were deposited. SiC particles were chosen because of their similarity in shape, size, and hydrophilicity to natural dirt contaminations. For the cleaning test, specimens with the contaminants were subjected to water droplets of approximately 2 mm diameter, using two microsyringes. Figure 6 shows that none of the investigated surfaces was fully cleaned by water rinsing. The data represent the average of five different investigated areas for each experiment. Most particles (70–80%) remained on smooth surfaces, and
50–70% of particles were found on microstructured surfaces. Most particles were removed from the hierarchical structured surfaces, but approximately 30% of particles remained. A clear difference in particle removal, independent of particle sizes, was only found in flat and nanostructured surfaces, where larger particles were removed with higher efficiency. Observations of droplet behavior during movement on the surfaces showed that droplets rolled only on the hierarchical structured surfaces. On flat, micro-, and nanostructured surfaces, the droplets first applied did not move, but the continuous application of water droplets increased the droplet volumes and led to sliding of
L
1232
Lotus Effect, Fig. 8 Surface height maps before and after wear tests with a 15 mm radius borosilicate ball at 100 nN and 10 mN for nanostructures with (a) CNTs and (b) Lotus wax using an AFM [7]
Lotus Effect Wear study using an AFM
a
Nanostructure with CNT After wear test at 100 nN (1 cycle)
Before wear test
After wear test at 10 μN (1 cycle)
3
3
3
μm
μm
μm
–3
–3
–3
100 μm
0
0 0
100 μm
0
100 μm
6 μm
Wear study using an AFM
b Before wear test
Nanostructure with Lotus wax After wear test at 100 nN (1 cycle)
After wear test at 10 μN (1 cycle)
3
3
3
μm
μm
μm
–3
–3
–3
0
these large droplets. During this, some of the particles were removed from the surfaces. However, rolling droplets on hierarchical structures did not collect the dirt particles trapped in the cavities of the microstructures. The data clearly show that hierarchical structures have superior cleaning efficiency. Fabrication of Durable Surfaces – For many commercial applications, the surfaces should have mechanical strength and chemical stability. Jung and Bhushan [7] prepared hierarchical superhydrophobic structures using carbon nanotube (CNT) – epoxy composite applied by spray method. Figure 7 shows SEM micrographs taken of the CNT-epoxy composite at
100 μm
0
100 μm
0
6 μm
0
100 μm
three magnifications. The surface shows a homogeneous distribution of CNTs. As a benchmark, structured surfaces with Lotus wax were also prepared to compare with the durability of CNT composite structures. To investigate the durability, wear tests were conducted by using an atomic force microscopy over 50 50 mm2 area at two nominal loads of 100 nN and 10 mN. Figure 8 shows surface height maps before and after wear tests. These superhydrophobic CNT composite structures showed good mechanical durability, superior to the structured surfaces with Lotus wax. These structures may be suitable for real-world applications.
Lumbricidae/Oligochaeta/Earthworms
Closure The Lotus effect provides one of the most successful examples of biomimetic surface effects. While theoretical foundations of wetting of rough surfaces have been developed decades ago, emerging applications and the need to design non-sticky surfaces has led to the intensive modeling and experimental research in the area. These findings provide new insights on the fundamental mechanisms of wetting and can lead to creation of successful nonadhesive surfaces.
Cross-References ▶ Biomimetics ▶ Rose Petal Effect ▶ Wetting Transitions
References 1. Nosonovsky, M., Bhushan, B.: Multiscale Dissipative Mechanisms and Hierarchical Surfaces: Friction, Superhydrophobicity, and Biomimetics. Springer, Heidelberg (2008) 2. Bhushan, B., Jung, Y.-C., Nosonovsky, M.: Lotus effect: surfaces with roughness-induced superhydrophobicity, selfcleaning, and low adhesion. In: Bhushan, B. (ed.) Springer Handbook of Nanotechnology, 3rd edn, pp. 1437–1524. Springer, Heidelberg (2010) 3. Bhushan, B., Jung, Y.-C., Koch, K.: Micro-, nano- and hierarchical structures for superhydrophobicity, self-cleaning and low adhesion. Phil. Trans. R. Soc. A 367, 1631–1672 (2009) 4. Bhushan, B.: Biomimetics inspired surfaces for drag reduction and oleophobicity/philicity. Beilstein J. Nanotechnol. 2, 66–84 (2011) 5. Wenzel, R.N.: Resistance of solid surfaces to wetting by water. Indust. Eng. Chem. 28, 988–994 (1936) 6. Cassie, A., Baxter, S.: Wettability of porous surfaces. Trans. Faraday Soc. 40, 546–551 (1944)
1233
L
7. Jung, Y.C., Bhushan, B.: Mechanically durable carbon nanotube-composite hierarchical structures with superhydrophobicity, self-cleaning, and low-drag. ACS Nano 3, 4155– 4163 (2009)
Low Fluid Drag Surface ▶ Shark Skin Effect
Low-Power ▶ Bio-inspired CMOS Cochlea
Low-Pressure Chemical Vapor Deposition (LPCVD) ▶ Chemical Vapor Deposition (CVD)
Lubrication ▶ Nanotribology
Lumbricidae/Oligochaeta/Earthworms ▶ Exposure and Toxicity of Metal and Oxide Nanoparticles to Earthworms
L
M
Macromolecular Crystallization Using Nano-volumes Gabriela Juarez-Martinez Centeo Biosciences Limited, Dumbarton, UK
Synonyms Biological nano-crystallization; Microarrays/ microfluidics for protein crystallization; Protein nano-crystallization
Definitions Biological macromolecules are large molecules or biopolymers which are made of smaller organic components that are linked together; they usually include proteins, nucleic acids, carbohydrates, and lipids. Crystal comes from the Greek word “krustallos” meaning “clear ice.” Crystals are geometrical solids with regular faces and with well-defined edges. They are formed by atoms, molecules, ions, or molecular assemblies that are arranged in an ordered and repeating pattern which extends over all three spatial dimensions. Crystals formed from biological macromolecules such as proteins are smaller than their inorganic and small molecule counterparts. Bio-macromolecular crystals are also very fragile as they may have between 20% and 80% solvent content; hence they are prone to dehydration which leads to cracking and destruction of the crystal. For this reason, biomacromolecular crystals need to be kept in a constant solvent saturated environment [1].
Crystallization refers to the process of forming or creating crystals, this process can be naturally or artificially induced. The term “protein crystallization” is often used instead of bio-macromolecular crystallization, and although strictly speaking it should only apply when referring to proteins. However, the crystallization techniques and methods are the same. There are many ways to crystallize a compound and the crystallization processes depend on the molecule or substrate to be crystallized. However, all crystallization process have two things in common: Firstly, all types of crystallizations are multi-parametric processes, which means that a variety of factors such as temperature, pH, purity, time, and pressure, among others, can affect the outcome of the process; secondly, they all involve three steps: nucleation, growth, and cessation of growth. These steps will be reviewed later in the chapter. Structural Biology refers to the study of tridimensional structures of biological macromolecules in order to better understand their function and interactions with other molecules at a molecular level.
Introduction To understand the importance of using nano-volumes to improve the crystallization of biological macromolecules, it is necessary to look at the historical motivation behind it. More than a century ago, scientists were fascinated with the creation of crystals of biological nature. In 1840, H€unefeld observed hemoglobin crystals in samples of almost dry blood. In 1851, Funke described a method to crystallize pure hemoglobin [2]. This was followed by the crystallization of
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
M
1236
urease by Sumner and pepsin by Northrop. They showed that urease and pepsin were pure proteins but also that they had an enzymatic activity. This was an important discovery, as at the time it was believed that proteins were incapable of catalyzing biological reactions, with proteins regarded as just carriers for the true catalytic molecules, enzymes. In 1946, the Nobel Prize in Chemistry was awarded to Sumner for the discovery that enzymes can be crystallized and to Stanley and Northrop for their preparation of enzymes and viruses in pure form [1]. It was not until 1934 that Bernal and Hodgkin produced the first X-ray diffraction pattern of a protein, pepsin. They showed that in order to obtain a good diffraction pattern, protein crystals should not be kept dry but studied surrounded by the liquid solution where they were grown from. This prompted the interest of many scientists and soon after the tridimensional structure of several proteins were reported including myoglobin by Kendrew et al.; hemoglobin by Perutz and Kendrew; and penicillin, vitamin B12, and insulin by Hodgkin [3]. The latest big push in protein crystallization came as a result of the various Genome projects carried out in the 1990s, when a large number of genes from different organisms were sequenced. Many of these genes coded for proteins of unknown function; consequently in the late 1900s and early 2000, a number of initiatives known as Structural Genomics Projects were set up across the world to determine in a highthroughput format the tridimensional structure of those proteins. The results of these projects have applications in many fields such as medicine, pharmacology, veterinary, and agriculture. Determining the tridimensional structure of unknown proteins allows scientist to better understand their function and interactions with other biological molecules at a molecular level. This is done by comparing different parts of the tridimensional structure of an unknown protein to a structural database of proteins of known function, revealing similarities from which biochemical and biological function may be deduced. Knowing the tridimensional structure of proteins of known function is also important for rational drug design, where the protein is usually involved in a disease process and can be used as a target where drug candidates can bind to inhibit, block, or modify its function. In rational drug design, scientists analyze the tridimensional structure of the target (usually
Macromolecular Crystallization Using Nano-volumes
a protein) and computer-generate a number of drug candidates (or fragments that can later be joined to form a drug candidate) that can bind to the target with high affinity. These drug candidates are then produced in the lab and tested against the target. The goal is to generate drugs that have higher affinity and selectivity for the target, therefore reducing the number of side effects [4, 5]. There are two main techniques for determining the tridimensional structure of a protein; these are nuclear magnetic resonance (NMR) and X-ray diffraction. The use of NMR is currently limited by the size of the macromolecule, (currently around 35 kDa) although advances in procedures may increase this limit. X-ray diffraction is the most common technique but requires the generation of biological crystals that are capable of diffracting at high resolution. Unfortunately, there is no way to know a priori if a biological molecule will crystallize or under what conditions. Hence, the common practice is to screen the target molecule against a large number of crystallization parameters including, but not limited to: pH, ionic strength, temperature, and chemical additives. Consequently, the crystallization of biological macromolecules has remained as the bottleneck of structural genomics.
Protein Crystallization Theory In order to create a protein crystal, it is necessary to have a highly pure and stable protein in solution and then force the protein molecules to aggregate in an ordered manner by modifying one or more of the crystallization parameters highlighted above. The formation of a protein crystal can be divided in three stages: nucleation, growth, and cessation of growth: Nucleation: This is the stage where small aggregates of protein are formed. If the aggregates have an internal order they will yield a macroscopic crystal, if not, an amorphous precipitate will be formed. The nucleation can be subclassified into a number of processes, namely: Primary nucleation: It is defined as the initial aggregation of the same type of molecules as a consequence of a crystallization process at a high degree of supersaturation. Furthermore, primary nucleation can be divided into a homogeneous or heterogeneous process. The former is produced when only the target macromolecules form
Macromolecular Crystallization Using Nano-volumes
a crystal in the bulk solution. The latter is when other molecules or the walls of the crystallization chamber act as a nucleation center. Secondary nucleation: It occurs after primary nucleation in the presence of the already formed crystals. It can be divided into attrition and catalytic mechanisms. In the former, the waves of diffusion or convection that transport the molecules in solution to the surfaces of the crystal collide directly with the crystal and can detach part of it. Thus, a new stable crystal nucleus is formed and can grow in this zone. The second type of secondary nucleation occurs when a defect on the surface of the crystal appears due to the addition of a new molecule in a disorientated way or due to the addition of an impurity. This creates a new nucleus, which may form a twin crystal [6]. Growth: Once a stable nucleus is formed, its growth is thermodynamically favored. Molecules of smaller size than the stable nucleus are transported to the surface of the stable nucleus to crystallize [6]. Cessation of growth: A crystal stops growing when the concentration of the macromolecules in solution decreases to the solubility level or when the surface of the crystal changes. These changes can be due to the addition of impurities that “poison” the surface of the crystal, such that the addition of new molecules becomes unfavorable [6].
Importance of Solubility in Protein Crystallization In order to control the crystallization of proteins, it is important to understand phase diagrams, which describe the solubility of proteins (Fig. 1). Initially the protein is in a concentrated, stable but soluble state of maximum solubility (undersaturation zone). By altering one or more crystallization parameters, it is possible to make the system shift out of equilibrium toward the supersaturation zone. Once the system arrives at a critical point in the supersaturation where it can overcome the free activation barrier, a spontaneous nucleus will form (nucleation zone). In this zone, the solid phase cannot accumulate and the system slowly returns to equilibrium, on its way it crosses a metastable zone where no more spontaneous nucleation will occur but it allows the already formed nucleus to grow. Once the system returns to equilibrium by reaching the solubility curve, the crystal will
1237
M
Macromolecular Crystallization Using Nano-volumes, Fig. 1 Schematic description of a phase diagram for the crystallization of Bio-macromolecules
not grow any further nor will it dissolve as the solid phase is in equilibrium with the saturated protein in solution [7]. Protein Crystallization Methods There are four basic methods to crystallize a protein: batch, dialysis, vapor diffusion, and liquid diffusion crystallization methods. It is important to mention that no method is considered superior to any other. The choice depends on the protein involved, its quality, and the preference of the researcher [8]. Batch crystallization method (Fig. 2a): This is the simplest and oldest method. Here the solution containing the protein of interest is mixed with the crystallizing solution that is at a concentration designed to reach the supersaturation zone instantaneously, forming a number of nuclei that can grow by returning to the metastable zone [8, 9]. Dialysis crystallization method (Fig. 2b): The solution that contains the protein is separated from the crystallizing solution by a semipermeable membrane of a specific pore size, which is smaller than the size of the protein and only allows small molecules (ions, additives) into and out of the protein compartment. This exchange of molecules allows the system to reach supersaturation and then equilibrium [8, 9]. Dialysis is a method that is suitable for independently optimizing the nucleation phase from the growth phase, as the nucleation phase requires a higher concentration of protein while the growth phase requires lower protein concentration.
M
M
1238
Macromolecular Crystallization Using Nano-volumes, Fig. 2 Shows a schematic representation of the most common protein crystallization methods: (a) batch; (b) dialysis; (c) vapor diffusion and (d) liquid diffusion crystallization methods. The schematic does not describe the dimensions of the containers as it is intended only to describe the process; however, any of these methods can be implemented in macroor microscale
Macromolecular Crystallization Using Nano-volumes Protein solution
Crystallisation solution
a
b
Porous membrane Protein solution Crystallisation solution 1/2(v)Protien solution+ 1/2(v)Crystallisation solution
c
Crystallisation solution
d Crystallisation solution Protein solution
Vapor diffusion crystallization method (Fig. 2c): This technique involves having a drop containing the protein and the crystallizing agents in a concentration lower than that required for the formation of a crystal. Then the drop is equilibrated against a reservoir containing the crystallizing agent at a higher concentration. Equilibrium is achieved by diffusion of the volatile species (water or organic solvents) out of the protein drop toward the more concentrated crystallizing agent reservoir. Consequently, the protein drop becomes more concentrated forcing the system into the supersaturation zone. Vapor diffusion can be achieved by hanging the protein drop from a sealing structure or by placing the drop on a plastic bridge [8]. Liquid diffusion crystallization method (Fig. 2d): This technique exploits the use of restricted geometries such as thin glass capillaries and microfluidic channels to force the protein solution and the crystallizing solution to mix only by diffusion. In this method, the protein solution and the crystallizing solution are placed next to each other in direct contact or separated by a physical buffer such as a permeable membrane, a gel, or by a physical barrier such as a valve. The two solutions mix into each other when the physical barrier is removed or by diffusing through the physical buffer. The liquid diffusion method has been implemented in two ways namely: free interface diffusion (also known as liquid-liquid diffusion) and counterdiffusion. Counterdiffusion (Fig. 2d): In this method, the experiment starts at high values of supersaturation.
Then when the protein solution and the crystallizing solution diffuse into each other, amorphous precipitate or microcrystals will form at the interphase. Immediately after, the concentration of protein in the vicinity of the precipitate decreases, but the molecules of the crystallizing agent will continue diffusing further into the protein solution. As the process continues, simultaneous events of diffusion and crystallization are created. As a consequence, a supersaturation gradient will be formed along the length of the glass capillary or microchannel. Typical concentration values for the protein solution vary between 4 and 20 mg mL1, the concentration for the crystallizing agent solution can be close to saturation. The length of the channel or capillary should be sufficiently long to allow a supersaturation gradient to be formed [10]. Free interface diffusion (Fig. 2d): The aim of this method is to achieve critical supersaturation for nucleation at a very slow rate in order to obtain a single nucleation event. Consequently, the concentrations of protein and crystallizing agent are lower than in the counterdiffusion method. The length of the capillaries of channels used in this method is usually shorter compared to the previous method. The free interface diffusion method can be viewed as a batch method conducted at very slow rate of mixing. This characteristic compared to the traditional batch method offers the advantage of a wider screening of crystallization conditions [10]. Unfortunately, protein crystallization is not a straightforward process, requiring the setting of
Macromolecular Crystallization Using Nano-volumes
several experiments to find clues on where the nucleation zone may lay. This is known as the screening process and consumes considerable amounts of protein. This is often problematic as many of the most interesting proteins are only available in limited quantities. Therefore, there is a demand for techniques that can obtain as much information on the crystallization conditions while consuming minimal amounts of protein. Once the nucleation zone is found, it is possible to move to the optimization process. Here the crystallization conditions found during the screening process (which do not necessarily result in a good quality crystal) are modified to obtain a large single crystal that can diffract at high resolution. In practice, the most common way of searching for adequate crystallization conditions is by mixing the protein solution with already prepared commercial crystallization kits. Crystallization kits consist of a set of solutions which screen different crystallization parameters. Crystallization assays are usually performed in commercial plastic plates known as microplates. These plastic plates come in different formats, e.g., 24, 48, 96, 384, and 1,526 well plates. The first two formats are suitable for setting crystallization conditions manually while the later ones are used for automated liquid dispensing systems. Protein solution volumes used in each experiment can vary from nanoliters to a few microliters, depending on the protein and the purpose of the experiment, e.g., screening vs. optimization.
Nano-crystallization In the 1990s, the development of automated liquid handling technology made it possible to dispense nanoliter volumes. This had a significant impact in a number of life science applications as it allowed multiple experiments to be performed in parallel while using reduced amounts of sample, which in many cases is expensive and difficult to produce or available in limited quantities. Protein crystallization samples presented an additional challenge to the automated liquid handling manufacturers as the solutions used vary greatly in terms of viscosity, vapor pressure, and surface tension. In addition, the solutions usually contain high concentrations of salts that can block the dispensing tips [11]. Despite these complications, some liquid handling robots can dispense sample
1239
M
volumes as small as 25 nL, although currently the majority of the systems dispense minimum volumes of 50 nL. The use of automated liquid dispensing systems delivering nanoliter volumes in protein crystallization allowed screening of a vast number of crystallization conditions in a fast and reliable manner. Now in a matter of minutes, hundreds of experiments could be set up in a reproducible format [11]. Also nanocrystallization may produce better ordered crystals since convection is significantly reduced. Some authors argue that this may also result in new crystal forms that are not produced in larger volumes [12]. An additional advantage is that nanoliter drops are easier to inspect using automated visualization systems as they have less depth to focus on compared to microliter drops [11]. The implementation of nano-crystallization methods has taken two different avenues namely: microarrays and microfluidic devices. The Use of Microarrays for Nano-crystallization The easiest way to work with nano-volumes in protein crystallization is by using an automated liquid dispensing system to set up crystallization trials in standard 96 well crystallization plates. However, evaporation of the sample while the system was setting up the experiment became an issue. The traditional method of sealing a crystallization experiment involved sealing the complete top of the crystallization plate with an optical clear sealing tape after the robot had finished setting up the experiment. This problem could worsen if the sealing tape was not applied correctly leaving air gaps between the microplate contents and the environment. To help solve this problem in 1992 Chayen et al. described a method now known as “batch crystallization under oil” where the mix of protein and crystallizing agent solution is dispensed inside a few microliters of inert oil to prevent evaporation [13]. In 1996 D’Arcy et al. proposed the use of mixtures of different oils to allow controlled evaporation, thus effectively implementing a vapor diffusion crystallization method under oil [14]. Another way of implementing nano-crystallization using a microarray format was reported by Bodenstaff et al. in 2002 [15]. They described the fabrication of a microarray with 70 wells made of an epoxy resin; each well had a maximum volume of 250 nL. Inside these wells Bodenstaff et al. set up batch crystallization experiments using sample volumes from 1 nL to
M
M
1240
250 nL. The dispensing of the sample was done by adapting a piezoelectric dispensing system and sealing was achieved by dispensing a layer of oil as reported previously by Chayen [13]. Bodenstaff showed that in practice, it was possible to crystallize in nanoliter volumes; however, setting up experiments in microsized arrays required a level of finesse, in practice, it is easier to set up nano-crystallization experiments in standard microplates. The Use of Microfluidic Devices for Nano-crystallization Microfluidic devices are ideal for screening a given protein against a vast number of crystallization conditions. These devices offer an additional advantage, that the mixing of the crystallization solution and the protein solution takes place by diffusion rather than convection, resulting in higher quality crystals. In one way or another, all four basic crystallization methods have be implemented in a microfluidic format. The list of reported microfluidic crystallization chips is large, as a number of the academic groups working in protein crystallization are now collaborating with groups involved in microfabrication technologies to come up with new microfluidic devices that better suit their research needs. The following is a brief review of current microfluidic devices used for protein crystallization, which is based on the crystallization method that they implement (batch, vapor diffusion, dialysis, and liquid diffusion). In some cases, a single device can be used to implement more than one crystallization method. The commercial microfluidic crystallization devices currently available to scientists are also presented. Implementation of the Liquid Diffusion Crystallization Method Using Microfluidics The first crystallization method to be implemented in a microfluidic format was the liquid diffusion method, specifically the free interface diffusion method [16]. It could be said that the traditional counterdiffusion method in glass capillaries [10] was the beginning of crystallizing using microfluidics, since the narrow capillaries offered a reduced or free convection environment depending on the diameter used. Setting up traditional liquid diffusion experiments in glass capillaries required a reasonable level of expertise and this method was not suitable for employing automated systems in the way batch and
Macromolecular Crystallization Using Nano-volumes
vapor diffusion methods are. The intrinsic characteristics of the liquid diffusion methods allowed a wider screening of crystallization conditions in a single experiment. These methods were logically much more suitable for implementing in microfluidic format, which offered a good alternative to the glass capillaries [17]. Commercial versions also allowed nonexperts to set up hundreds of experiment in a fast and easy manner. The implementation of the free interface diffusion method was presented by Hansen and Quake in 2003 [16]. Their device consisted of a glass substrate where microfluidic channels and micro-chambers were etched and multilayers of silicon elastomer provided flexible valves that opened and closed to allow the protein and crystallization solutions to mix at the required point. This device allowed the screening of 48 different crystallization solutions at 4 different concentrations (Fig. 3). In 2003 Fluidigm Inc., a company founded by Stephen Quake and Gajus Worthington, launched a commercial version of this device known as the TOPAZ chip (Fig. 4). Later an advanced version of this system was launched; here the glass substrate was replaced by a polymer. It combined free interface diffusion with controlled vapor diffusion, which was possible due to the gas-permeable material used in the construction of the TOPAZ chip. Fluidigm Inc., also developed software and peripheral instrumentation to facilitate liquid handling inside the TOPAZ chip as well as the inspection and analysis of the crystallization results. Another commercial implementation of the free interface diffusion method was introduced by Microlytic Inc., where their crystallization chip, known as the Crystal Former (Fig. 5), consists of 16 channels, each 10-mm long, allowing a volume of 150 nL per channel. Each channel has two inlets at opposite ends to allow the protein solution and the crystallization solution to be loaded. The channels are covered by a thin removable foil which acts as a seal and allows extraction of the crystals that have formed. Initially commercial microfluidic chips for protein crystallization were expensive ($600 per chip); making them only affordable to large academic institutions and pharmaceutical companies. Current prices are more accessible ($50 USD per chip) to smaller laboratories. However, price and the need for microfluidic devices that fit a specific research need prompted many academic groups to develop their own chips. A recent
Macromolecular Crystallization Using Nano-volumes Macromolecular Crystallization Using Nano-volumes, Fig. 3 Shows the free Interface diffusion (FID) microfluidic device designed by Hansen et al. in 2003. (a) Shows the loading of the protein and crystallizing solutions; (b) shows when the valve between the protein solution and crystallizing solution opens allowing the free interface diffusion to proceed; (c) and (d) are the micrographs of the (a) and (b) stages respectively; (e) shows the complete microfluidic device; and (f) shows a crystal of aquaporin obtained using this device (Figure and captions reprinted from reference 16 Copyright # 2003, with permission from Elsevier Ltd.)
a valve
b
load protein
mix
open
load precipitant
close
close
open
open
close
c
d
1mm
1mm
e
M
1241
f
M 100 μm
Macromolecular Crystallization Using Nano-volumes, Fig. 4 Shows the complete TOPAZ system which includes (a) the TOPAZ chip and (b) the peripheral instrumentation (Images courtesy of Fluidigm Inc.)
example of this is from Dhouib and coworkers who in 2008 [18] reported the design of a microfluidic chip named the CD chip to implement the counter-diffusion crystallization method. The CD chip was later enhanced in 2009 [19]. The CD chip consists of eight microchannels with a total volume of 169 nL per
channel. All eight channels converge at the protein solution inlet. On the opposite end of each channel another inlet is created to allow the loading of the crystallization agent. Each channel is 4.5-cm long which represented an advantage over other chips with shorter channels as the longer the channel, the larger
M
1242
Macromolecular Crystallization Using Nano-volumes
Macromolecular Crystallization Using Nano-volumes, Fig. 5 Shows (a) a photograph of the Crystal Former chip and (b) a schematic diagram of a side view of one of the microchannels (Images courtesy of Microlytic Inc.)
a
b 1 2 3 4 5 6 7
Side view 8 9 10 11 12 13 14 15 16
Tape seal Precipitant inlet
the gradient of crystallization conditions that can be formed, hence increasing the chances of obtaining a better quality crystal. Additionally this chip was designed to be suitable for in situ X-ray diffraction analysis without the need to remove the crystals from the chip, avoiding potential physical damage. The other advantage is that a number of labels were placed along the length of the channel to facilitate locating the exact position of a given crystal when the sample is placed in the X-ray diffractometer. This device is made of cycloolefin copolymer (COC) which also provides good optical clarity and X-ray compatibility (Fig. 6) [19]. Implementation of Batch Crystallization Method Using Microfluidics In 2004, Zheng et al. [20] reported a microfluidic platform for screening hundreds of crystallization conditions using less than 4 nL of protein solution and a total sample volume of 7.5 nL per crystallization experiment (Fig. 7). This device consisted of a PDMS block from which
Protein inlet
Crystallization Channel Peelable film for harvesting
inlets and interconnecting microchannels were fabricated. These channels converged to a single point where a thin glass capillary was inserted. The protein solution, buffer solution, and crystallization solutions were briefly allowed to flow through the channels into a stream of fluorinated oil (the carrier), thus creating a small drop containing a mixture of the aqueous solutions. The composition of the drops was changed by varying the flow rates of each of the aqueous streams. The flow was stopped once the glass capillary was full. The capillary was then disconnected from the PDMS block and the extremes sealed with wax. The crystals formed inside the droplets and no evaporation of the sample was observed even after 6 months of incubation without any humidity control. An advantage of this microfluidic chip was that if a good crystal was formed inside the glass capillary, the capillary could be mounted directly on the goniometer head of an X-ray diffractometer and a diffraction pattern of the protein crystal could be obtained.
Macromolecular Crystallization Using Nano-volumes
1243
M
Macromolecular Crystallization Using Nano-volumes, Fig. 6 Shows (a) the CD chip designed by Dhouib and coworkers; (b) the Crystal Former chip from Microlytic Inc; (c) a microfluidic chip from Greiner Bio One GmbH, and (d) a micrograph of a thaumatin crystal grown inside one of the channels of the CD chip. It also shows a label indicating in which channel and at which position in the channel the crystal is positioned (Image courtesy of Claude Sauter)
a
c buffer protein precipitant
1 cm
PDMS
glass capiilary
M
fluorinated carrier fluid
b
100 μm
d
100 μm
Macromolecular Crystallization Using Nano-volumes, Fig. 7 Shows (a) A schematic illustration of the method of forming droplets for protein crystallization trials in a PDMS/ glass capillary composite microfluidic device. (b) A micrograph showing the droplet formation. (c) A photograph of the
composite device. All inlets and channels were filled with [Fe9SCN)x](3x)+ solution for clarity. (d) A micrograph of thaumatin crystals grown inside droplets in a capillary (Figure and caption reprinted from reference 20 Copyright # 2004 with permission from Wiley-VCH Verlag GmbH & Co)
In 2008, the Accelerated Technologies Center for Gene to 3D Structures created the Microcapillary Protein Crystallization System (MPCS). This is an instrument capable of generating hundreds of crystallization drops in a microfluidic format in a fast and convenient way (Fig. 8). Each crystallization drop is separated from the other by a volume or plug of oil. The main advantage of this system is its precise and finely controlled concentration gradients that can be
performed over a series of drops, thus effectively creating a concentration screening for optimizing protein crystals [21]. The MPCS uses microfluidic chips known as Crystal Cards made of COC and containing two separated microfluidic channels, each channel with approximately 10 mL of useful volume; an independent crystallization experiment can be performed in each channel. The Crystal Cards together with the
M
1244
Macromolecular Crystallization Using Nano-volumes
Macromolecular Crystallization Using Nano-volumes, Fig. 8 Shows (a) a photograph of a Crystal Card and (b) a photograph of the MPCS system and micrographs of protein crystals grown inside the Crystal Cards (Images courtesy of Emerald Biosystems Inc.)
MPCS represented the commercial implementation of the method proposed by Zheng et al. in 2004. This also simplified the experimental process, as it was no longer required to disconnect the capillary for incubation and X-ray analysis. The Crystal Cards allow growing crystals inside the chip cards and are compatible with in situ X-ray diffraction analysis [21]. Implementation of the Vapor Diffusion Crystallization Method Using Microfluidics The microfluidic system proposed by Zheng et al. in 2004 (Fig. 7) also allowed the implementation of the vapor diffusion method where the goal is to change the concentration of the droplet containing the protein solution by extracting water molecules from it. Zheng et al. achieved this by injecting droplets containing the mixture of the protein, buffer, and crystallization solution in a stream of water permeable oil into the glass capillary. These drops were alternated with drops containing a high concentration of salt. This procedure enabled water from the protein drop to diffuse through the water permeable oil to the drop containing a high concentration of salt until the system reached osmotic equilibrium. The ability to concentrate the protein drops allowed the crystallization process to proceed and also allowed crystals to form where the original protein and crystallization agent concentrations were too low to allow the formation of crystal nuclei.
Implementation of the Dialysis Crystallization Method Using Microfluidics Dialysis is probably the most difficult method to implement in a microfluidic format as it requires the careful manipulation of the liquids inside the microfluidic chip. In dialysis, the protein solution can be concentrated or diluted by allowing aqueous solution containing higher or lower concentrations of salt to be in contact with the protein solution through a water-permeable membrane. This membrane allows the passage of water molecules but not protein molecules. In 2007, Shim et al. designed a microfluidic chip known as the Phase Chip (Fig. 9) [22], which used an advanced crystal optimization method called “seeding” combined with the dialysis method. Seeding is a crystal optimization technique that involves crushing a larger protein crystal in order to produce many microcrystals that are then used as seeds that are added to new supersaturated crystallization solutions. These seeds act as stable nucleation sites which protein molecules can use to form larger crystals under less supersaturated conditions [23]. The Phase Chip is made of PDMS and uses the method described previously by Zheng et al. to create drops of protein inside a continuous stream of oil. The Phase Chip exploits surface tension forces to guide each protein drop into wells that are located at the side of the channel, spaced 500 mm away from each other. The wells are also deeper than the flow channel. To implement the dialysis method, the bottom of each
Macromolecular Crystallization Using Nano-volumes
1245
M
Macromolecular Crystallization Using Nano-volumes, Fig. 9 Shows a composed image of the Phase Chip. The top image shows the complete phase chip showing the inlet connections, the bottom left figure shows a schematic of the microfluidic circuit with the channels arranged in an “s” shape and the protein microwells arranged on the side of the channels. The bottom right photo shows the microwells filled with a blue dye (Figure reprinted from reference 22 Copyright # 2007with permission from the American Chemical Society)
M
M
1246
well is made of a thin layer of PDMS (15-mm thick) which allows water molecules to pass but not protein molecules. On the other side of the membrane, there is a 100-nL reservoir which can be filled with dry air, a solution with a high concentration of salt or with water. If the reservoir is filled with either dry air or salt solution, the protein solution will shrink, thus concentrating the solution, as water molecules will flow into the reservoir. Alternatively, if the reservoir is filled with pure water, the protein solution will become diluted as water molecules will now flow in the other direction. The initial goal was to fill each well with protein solution drops at different concentrations. This was achieved by changing the flow rate of the different solutions injected in the chip. Then all protein drops were concentrated by introducing dry air into the reservoir that in many cases lead to the formation of many microcrystals inside the wells. These microcrystals were considered seeds that were later transformed into large crystals by introducing pure water into the reservoir. The water molecules diffused from the reservoir to the protein well and diluted the protein solution, by doing this the smallest and least stable nuclei dissolved and only the larger microcrystals grew into even larger and higher quality crystals. Another way to classify microfluidic devices for protein crystallization is by the type of mechanism used to create and mix the final crystallization solution. This classification was proposed by Liang and Ismagilov in 2010 [24] and classifies the devices as: valve-based systems, droplet-based systems, and Slip Chip systems (any combination of the above is called hybrid systems). In valve-based systems, the protein solution and the crystallization solution are placed in different chambers within the microfluidic chip and the mixing takes place by opening a valve that connects the two chambers [24]. A good example of this is the device described by Hansen [16] and commercialized by Fluidigm Inc. as shown in Fig. 4. In droplet-based systems, drops containing the protein and the crystallization agents are created by injecting these solutions into a flow of oil or other immiscible carrier fluid. Systems using this method are shown in Figs. 7 and 9. Droplet-based microfluidic systems have been useful to crystallize membrane proteins, which are traditionally difficult to work with as they are highly hydrophobic [24]. Crystallizing
Macromolecular Crystallization Using Nano-volumes
a membrane protein usually involves reconstituting it inside a lipidic mesophase, which is highly viscous and then incubating it against a set crystallizing solutions hoping to induce nucleation and crystal growth. The slip chip systems [24, 25] consist of two different plates, each containing a number of wells. One plate is loaded with various crystallizing solutions and the other is loaded with protein solutions. Then the plates slide on each other bringing both solutions into contact to form individual crystallization assays, as shown in Fig. 10. The Slip Chip represents another way of implementing the batch crystallization method and could be considered a hybrid between microarrays and microfluidic devices. The volumes of the protein solution and crystallizing solution are defined by the volumes of the corresponding wells. The wells are filled by a series of microchannels and microwells and storage under paraffin oil to prevent evaporation.
Future Trends The crystallization process has been one of the main bottlenecks in structural biology. Now with the use of methods and technologies coming from biology, physics, biochemistry, and engineering, the screening part of the process has been improved and hundreds of crystallization conditions can be screened with small amounts of sample. This will allow us to increase our chances of finding the right crystallization conditions for a given protein or at least to find initial clues on where to start the optimization phase. The challenge now is to focus on understanding the effect of alternative crystallization parameters that have not been sufficiently mastered and which may allow to control the crystallization process with more precision. These may include temperature, pressure, and electric and magnetic fields. Although a few academic groups have designed and built their own instrumentation to use these parameters to improve the quality of protein crystals, these instruments have not been available to the wider scientific community, with the exception of the TG40 System from Centeo Biosciences Ltd, which allows users to use temperature as an additional optimization parameter. The use of these parameters may also be useful for the crystallization of novel proteins coming from organisms that live in extreme environments such as deep-sea creatures.
Macromolecular Crystallization Using Nano-volumes Macromolecular Crystallization Using Nano-volumes, Fig. 10 Shows the Slip Chip. Schematics showing (a) loading of protein into a Slip Chip that has already been preloaded with precipitants and (b) slipping to combine protein and precipitants to form crystallization trials. (c) Micrographs of loading a green food dye (mimicking the protein) into a Slip Chip that has already been preloaded with colored dyes (mimicking precipitants). (d) Micrographs of the Slip Chip after slipping to combine the solutions. (e) Crystals of the photosynthetic reaction center from Blastochloris viridis obtained using this device (Figure and caption reprinted from reference 25 Copyright # 2009 with permission of the Royal Society of Chemistry)
a
1247
b
load protein
preloaded precipitants
protein
precipitants
c
M
slip
250 μm
mix protein with precipitant
250 μm
d
slip crystallization trials
slip into contact
protein 250 μm
e
M
Also with the incorporation of peripheral instrumentation it may be possible to analyze precrystallization conditions via dynamic light scattering or fluorescence techniques, allowing to know in advance if a given protein will crystallize under a given condition. Acknowledgments The author would like to thank Phillip Dobson and Philipp Steinmann for reviewing this manuscript, to Abel Moreno, Claude Sauter, and Jose Antonio Gavira for their useful discussions and comments, and to the academic groups and companies that provided many of the images for this entry.
Cross-References ▶ Prenucleation Clusters ▶ Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures ▶ Structure and Stability of Protein Materials
References 1. Giege, R., Ducruix, A.: An introduction to the crystallogenesis of biological macromolecules. In: Giege, R., Ducruix, A. (eds.) Crystallization of Nucleic Acids and Proteins, pp. 1–5. Oxford University Press, New York (1999) 2. Fruton, J.S.: Molecules of Life: Historical Essay on the Interplay of Chemistry and Biology. Wiley Interscience, New York (1972) 3. Glusker, J.P., Adams, M.J.: Dorothy crowfoot hodgkin. Phys. Today 48(5), 80 (1995). Obituraries 4. Goodwill, K.E., Tennant, M.G., Stevens, R.C.: Highthroughput x-ray crystallography for structure-based drug design. Drug Discov. Today 6(15), S113–S118 (2001) 5. Blundell, T.L., Jhoti, H., Abell, C.: High-throughput crystallography for lead discovery in drug design. Nat. Rev. Drug Discov. 1, 45–54 (2002) 6. Mersmann, A., Bartosch, K.: How to predict the metastable zone width. J. Cryst. Growth 183(1), 240–250 (1998) 7. Ries-Kautt, M., Ducruix, A.: From solution to crystals with a physic-chemical aspect. In: Giege, R., Ducruix, A. (eds.) Crystallization of Nucleic Acids and Proteins, pp. 267–312. Oxford University Press, New York (1999)
M
1248
Macroscale Platforms Capable of Biological Studies and Manipulations at Nanoscale
8. Chayen, N.E.: Recent advances in methodology for the crystallization of biological macromolecules. J. Cryst. Growth 198(199), 649–655 (1999) 9. Ducruix, A., Giege, R.: Methods in crystallization. In: Giege, R., Ducruix, A. (eds.) Crystallization of Nucleic Acids and Proteins, pp. 121–148. Oxford University Press, New York (1999) 10. Otalora, F., Gavira, J.A., Ng, J.D., et al.: Counterdiffusion methods applied to protein crystallization. Prog. Biophys. Mol. Biol. 101, 26–37 (2009) 11. Bergfors, T.: Automated liquid handling systems for submicroliter crystallization. In: Chayen, N.E. (ed.) Protein Crystallization Strategies for Structural Genomics, pp. 60–61. International University Line, La Jolla (2007) 12. Santarsiero, B.D., Yegian, D.T., Lee, C.C., et al.: An approach to rapid protein crystallization using nanodroplets. J. Appl. Crystallogr. 35, 278–281 (2002) 13. Chayen, N.E., Stewart, P.D.S., Blow, D.M.: Microbatch crystallization under oil – a new technique allowing many small-volume crystallization trials. J. Cryst. Growth 122(1–4), 176–180 (1992) 14. D’Arcy, A., Elmore, C., Stihle, M., et al.: A novel approach to crystallizing proteins under oil. J. Cryst. Growth 168(1–4), 175–180 (1996) 15. Bodenstaff, R.E., Hoedemaeker, F.J., Kuil, M.E., de Vrind, H.P.M., et al.: The prospects of protein nanocrystallography. Acta Crystallogr. D58, 1901–1906 (2002) 16. Hansen, C.L., Quake, S.R.: Microfluidics in structural biology: smaller, faster. . .better. Curr. Opin. Struct. Biol. 13, 538–544 (2003) 17. Sauter, C., Dhouib, K., Lorber, B.: From microfluidics to microfluidics for the crysallization of biological macromolecules. Cryst. Growth Des. 7(11), 2247–2250 (2007) 18. Dhouib, K., Malek, C.K., Pfleging, W., et al.: Microfluidic chips for crystallization of biomacromolecules by counterdiffusion and on-chip crystal X-ray analysis. Lab Chip 9, 1412–1421 (2009) 19. Brun, M., Dhouib, K., Morin, P., et al.: Microfluidic chips for the screening of crystallization conditions and in-situ crystallographic analyses. In: Proceedings of the 2nd European conference in microfluidics, Toulouse (2010) 20. Zheng, B., Tice, J.D., Roach, S.L., et al.: A droplet-based composite PDMS/Glass capillary microfluidic system for evaluating protein crystallization conditions by microbatch and vapor-diffusion methods with on-chip X-ray diffraction. Angew. Chem. Int. 43, 2508–2511 (2004) 21. Gerdts, C.J., Stahl, G.L., Napuli, A., et al.: Nanovolume optimization of protein crystal growth using the microcapillary protein crystallization system. J. Appl. Crystallogr. 43, 1078–1083 (2010) 22. Shim, J., Cristobal, G., Link, D.L., et al.: Using microfluidics to decouple nucleation and growth of protein crystals. Cryst. Growth Des. 7(11), 2192–2194 (2007) 23. Stura, E.A.: Seeding techniques. In: Giege, R., Ducruix, A. (eds.) Crystallization of Nucleic Acids and Proteins, pp. 177–191. Oxford University Press, New York (1999) 24. Liang, L., Ismagilov, R.F.: Protein crystallization using microfluidic technologies based on valves, droplets and slip chip. Annu. Rev. Biophys. 39, 139–158 (2010) 25. Du, W.B., Li, L., Nichols, K.P., et al.: Slip chip. Lab Chip 9, 2286–2292 (2009)
Macroscale Platforms Capable of Biological Studies and Manipulations at Nanoscale to Bio-inspired and Bacterial Nanorobotics ▶ Nanorobotics for Bioengineering
Magnetic Assembly ▶ Magnetic-Field-Based Self-Assembly
Magnetic Dipole Emitters ▶ Optical Frequency Magnetic Dipole Transitions
Magnetic Dipole Transitions ▶ Optical Frequency Magnetic Dipole Transitions
Magnetic Electromagnetic Radiation ▶ Thermal Cancer Nanoparticles
Ablation
Therapies
Using
Magnetic Nanostructures and Spintronics Alessandro Chiolerio1 and Paolo Allia2 1 Physics Department, Politecnico di Torino, Torino, Italy 2 Materials Science and Chemical Engineering Department, Politecnico di Torino, Torino, Italy
Synonyms Nanomagnetism; Spintronic devices
Magnetic Nanostructures and Spintronics
Definition Nanostructures featuring magnetic properties and devices for the elaboration and/or storage of information based on electron spin transport. The need of storing and elaborating binary information permeates everyday life. Today’s computers are mainly charge-transfer devices; this means that inside a processing unit the information chain corresponds to a physical electron (hole) drift current, characterized by diffusion lengths and flow times that strongly limit the evolution toward faster computations and smaller sizes. Codifying binary information into a spin-polarized current could allow much faster logic operations, due to the possibility of reversing spins without stopping the carrier flow. The main difficulty of implementing spin-electronic (“spintronic”) devices is that typical spin diffusion lengths are well below the hundredth of nanometer for diamagnetic metals and between some tenths of nm and some nms for ferromagnetic metals and alloys, meaning that the typical device must be nanostructured. Furthermore, since spin-flip scattering must be prevented to process uncorrupted data, both interface quality and material purity must be extremely controlled. High and ultra-high vacuum deposition equipments are needed, and perfect control of geometries down to the nm scale requires extremely sophisticated lithographic steps. Such technological issues have slowed down the challenge of bringing spintronic devices into the core of our logic devices. On the contrary, commercial data storage still relies almost entirely on magnetic nanostructures: both granular media for hard disk drives and tunneling magnetic junction-based read/write heads are worldwide diffused. Due to their technological relevance, the most important effects on which commercial spintronic devices are based will be considered: Giant Magnetoresistance, Tunneling Magnetoresistance and Spin Torque. Two more subsections are addressed to particularly hot topics: Semiconductor Spintronics and Data Storage. In all these cases, physical systems and/ or devices are authentic nanostructures, that is, artificial, nanosized structures characterized by carefully controlled geometries in spite of their reduced dimensions. On the other hand, magnetic nanoparticles, although increasingly gaining attention because of their potential applications in different
1249
M
fields (ranging from ICTs to biomedicine) are excluded from this entry because the synthesis processes, even when characterized by extremely narrow size distributions, do not allow complete control on the assembly geometries, not (yet) resulting in the fabrication of bonafide ordered nanostructures. A subsection related to magnetic anisotropies at the nanoscale represents an essential premise to the rest of the work.
Magnetic Anisotropies at the Nanoscale The magnetic properties of various classes of nanomaterials and nanostructures are strongly influenced by many types of magnetic anisotropy which determine their behavior. As a general rule, it can safely be stated that all anisotropies existing in bulk, macroscopic material play a role at the nanoscale [1]; in addition, novel types of anisotropy, typical of nanostructured materials, appear. A brief overview of the different magnetic anisotropies and their effects follows. Shape Anisotropy Shape anisotropy, emerging from the requirement that the magnetostatic energy of the magnetic system must be a minimum, effectively dictates the direction of the local magnetization vector in ultrathin films of a ferromagnetic material and in nonspherical magnetic nanoparticles. In thin films of a ferromagnetic material the direction of the magnetization vector is preferentially forced to stay parallel to the film plane by shape anisotropy. This property is widely exploited in spintronic devices. For similar reasons, in elongated (either ellipsoidic or acicular) nanoparticles, the zero-field magnetization stays aligned to the major axis. This anisotropy determines the coercivity of an assembly of randomly oriented magnetic nanoparticles, and plays, therefore, a central role in the magnetic response of particulate media for high-density magnetic recording and hybrid materials containing nanoparticles embedded in a diamagnetic, organic matrix such as a polymer. Finally, shape anisotropy is fundamental for patterned media also determining the local direction of magnetization and the magnetic domain structure within a magnetic nanodot or nanopillar, consequently influencing the magnetization reversal mechanism and its duration.
M
M
1250
Magnetocrystalline Anisotropy Magnetocrystalline anisotropy may concur in determining the local magnetization direction of magnetic nanoparticles; however, its effects often emerge only when the shape anisotropy is negligible, that is, for almost perfectly circular/spherical nanoparticles. One of the most important examples of the role played by magnetocrystalline anisotropy is provided by FePt thin films, which have been largely investigated owing to their potential as novel media for high-density perpendicular magnetic recording. In these films, shape anisotropy which would impose a parallel-tothe plane magnetization state is largely overcome by a very strong magnetocrystalline anisotropy term, such that the easy axis is pointing perpendicular to the plane. This term results from the development of a tetragonal distorted phase (L10) which is obtained by submitting cubic FePt to controlled heating. Stress (Magnetoelastic) Anisotropy In nanostructured materials obtained by means of deposition or growth techniques followed by lithographic process, the concentration of local stress fields can reach high levels. The magnetization direction of a thin ferromagnetic film or patterned medium can be affected by stress through magnetoelastic (magnetostrictive) coupling. The resulting stress anisotropy often plays a detrimental role on the magnetic response of a thin ferromagnetic film, because it acts to break the symmetry of the system, generating local disturbances in the magnetization pattern of the material and acting to reduce its performance. Surface/Interface Anisotropy In thin ferromagnetic films, specific additional forms of magnetic anisotropy exist [1]. Surface anisotropy may play a relevant role because it can be one order of magnitude larger than typical bulk anisotropy values. Surface anisotropy of transition-metal elements or alloys can be understood in terms of the different band filling of 3D states for magnetic atoms at a surface/interface, and can be either perpendicular or parallel to the film plane in dependence of the band filling level, that is, in dependence of the transition metal considered. For symmetry breaking reasons, some of the five 3D states of transition-metal atoms at a surface/interface (i.e., those whose wavefunctions extend in a direction perpendicular to the surface
Magnetic Nanostructures and Spintronics
plane) take quite different energies with respect to bulk atoms, while other 3D states (those mainly extending along the surface plane) have almost identical energies in the bulk and at the surface. Interface anisotropy is important in magnetic multilayers also, because of the hybridization of electron states of the adjacent metals (one typically being Fe, Co, Ni; the other one being either a noble metal or an early (or late) transition element). A variety of effects emerges as a consequence of such hybridization, including strong perpendicular anisotropy. Interlayer Exchange Anisotropy The transmission of a magnetic anisotropy through an interface between two magnetic materials – that is, the occurrence of a common axis of easy magnetization shared by the two materials across the interface – is a peculiar effect of layered magnetic films and plays a central role in the optimization of the properties of a number of spintronic devices, including the spin valve [2]. As known from the physics of bulk magnetic systems, long range order (ferromagnetic, antiferromagnetic, ferrimagnetic) is assured by the exchange interaction between pairs of adjacent spins. Exchange interaction, a purely quantum effect, is particularly strong between spins belonging to the nearest neighbor atoms, and goes rapidly to zero with increasing inter-spin distance. When the interface between two different magnetically ordered metals is sharp, spin information can be transmitted from one material to the other and vice versa. In particular, if one material is strongly anisotropic, it can influence the anisotropy of the adjacent layer through exchange coupling. The most interesting case from the application side arises by placing together a ferromagnet and an antiferromagnet. In this case, the antiferromagnetic layer (it can be a transition metal oxide of a transition metal, such as NiO or CoO), although acting as a definite source of magnetic anisotropy for the ferromagnetic layer, is magnetically “dead,” that is, not contributing at all to the magnetic response of the device in the range of magnetic fields of interest. Such a feature is exploited, for example, in spin valves, where the magnetization of one of the two ferromagnetic layers must remain fixed in space (“pinned”) while that of the other layer is free to reverse under the action of the same weak magnetic field. The pinning force is typically provided by the exchange anisotropy between an antiferromagnetic layer in close contact with the pinned ferromagnetic layer [3].
Magnetic Nanostructures and Spintronics
Exchange Bias Exchange bias [4] is strictly related to exchange anisotropy, even though resulting in a different variety of applications. The effect, discovered in 1956 – that is, well before the rise of interest toward nanomagnetism and spintronics – is again associated to the exchange anisotropy between an antiferromagnetic and a ferromagnetic layer. After proper thermal pretreatment (e.g., field cooling through the ordering [Ne´el] temperature of the antiferromagnet), the hysteresis loop of the FM material is permanently shifted along the field axis. This brings about a different absolute value of the material’s coercivity under decreasing and increasing fields. The mean coercivity can also increase. Possible applications of novel exchangebiased bi- or multilayers are permanent magnets and magnetic recording media. Moreover, the loop shift induced by exchange bias can act to reduce the field needed to observe giant magnetoresistance effects in spin valves or magnetic multilayers with respect to the field needed in unbiased systems.
Giant Magnetoresistance In 1975, experiments on a Fe/GeO/Co junction led to the discovery of tunneling magnetoresistance (TMR) by M. Jullie`re, which was again verified in 1995 by T. Miyazaki, N. Tezuka, and J. S. Moodera. This great delay was caused by the fact that the experiment was done at cryogenic temperatures; the interest on the phenomenon was mainly academic because commercial electronics was and is still now unable to operate at such low temperatures, where the carriers in a semiconductor are typically freezed. So in 1988, experiments on layered thin films of ferromagnets alternated to a nonmagnetic metal led to the simultaneous and independent discovery of the giant magnetoresistance (GMR) by A. Fert and P. A. Gr€ unberg, who were awarded for this discovery with the Nobel prize in 2007. They found that submitting those layered structures to a current perpendicular to the interfaces, the conduction electrons were submitted to a spin-dependent scattering whose probability depends on the relative alignment between the macroscopic magnetization of the two ferromagnetic thin films. In case they are parallel the scattering probability is low, hence the electrical resistance is low too. Typical materials used to
1251
M
Magnetic Nanostructures and Spintronics, Fig. 1 Section of the nanostructured stack of bottom spin-valve featuring an antiferromagnetic layer providing exchange bias to the pinned layer and including a cap layer to prevent oxidation
realize a GMR stack are thin films of Co/Cu, Fe/Cr, Co/Ag, Co/Ru, Co90Fe10/Cu; a typical structure is shown in Fig. 1. When applying an external magnetic field, if the two ferromagnetic films feature different coercivities, it is possible to act on the orientation of the magnetically softer one and produce a high resistance state. Different coercivities are obtained either realizing films of different thicknesses or coupling one of them, the so-called pinned layer, to an antiferromagnet by exchange bias. Between the materials used to implement exchange bias, selected ones are NiO, FeMn, Fe2O3, Mn76Ir24, and MnPt. Generally the highest values of GMR, defined as total resistance variation normalized by the resistance in case of parallel alignment times 100, are obtained for epitaxial thin films working at cryogenic temperatures: GMR between 100% and 200% were measured at temperatures lower than 5 K, while at room temperature typical values approach 50% at maximum. GMR systems are commercially used as magnetic field sensors, isolators and hard-disk read heads [5]. The typical electrical response is presented in Fig. 2.
Tunneling Magnetoresistance As already stated, TMR was observed before GMR and is still under intensive study, due to extreme
M
M
1252
Magnetic Nanostructures and Spintronics
Magnetic Nanostructures and Spintronics, Fig. 2 Typical giant magnetoresistance curve with sketches indicating the parallel/antiparallel configuration of magnetization in the free (topmost blue layer) and pinned layer (bottom blue layer close to the green antiferromagnet)
Magnetic Nanostructures and Spintronics, Fig. 3 Structure of a nanoengineered stack of a magnetic tunneling junction, comprehensive of a synthetic ferromagnetic pinned layer and a cap layer to prevent oxidation
industrial interests; devices showing TMR are called magnetic tunneling junctions (MTJs) and, if compared to more classical all-metal devices (such as GMR), are characterized by a high impedance, representing the optimal choice for electronic drivers operating at rather high frequencies. This is of particular interest for at least two applications: the elements of digital nonvolatile memories and the stray field sensors used in TMR read heads for high-density storage media. A TMR device is composed by a stack of two ferromagnets, one soft and one hard (see Fig. 3) separated by an insulator; in analogy to spin-valves they have a hysteretic three-state electrical response and could act as magnetically operated switches. Generally the materials used for soft
Magnetic Nanostructures and Spintronics, Fig. 4 Typical tunneling magnetoresistance curve with a representation of the magnetization configuration in the free layer (blue arrow) and in the pinned one (green arrow)
electrodes are FeNi, CoFeB, or a thick layer of Co, Fe, or Ni; for the hard electrodes CoFe, CoCr, or an extremely thin layer of Co, Fe, or Ni are used; the tunnel barrier may be realized in Al2O3, MgO (best performances), and many other hybrid materials such as Ta2O5, SrTiO3, and NiO [6]. Since the best electrical response is given by a single domain structure (Fig. 4), each MTJ should be nanostructured, for example, as a nanopillar with top and bottom contacts. Furthermore monocrystalline epitaxial materials, although displaying many technological problems, provide the best performances giving TMR values as high as 15,000%, versus typical TMR of sputtered devices below 1,000%. Since monocrystalline materials require extremely expensive equipments such as molecular beam epitaxy, engineers found that also sputtered materials could
Magnetic Nanostructures and Spintronics
1253
M
guarantee reasonable performances when texturized, that is, thermally treated in order to undergo solid state epitaxy, consisting in a pseudo-conformal growth at the interfaces between insulating barrier and ferromagnetic electrodes.
Spin-Torque Effects and Devices Most spintronic effects have their own opposites; this circumstance engenders a variety of novel potential applications. As known, the giant magnetoresistance effect is the result of a modulation of a spin-polarized current by spin-dependent scattering. In a spin valve, the direction taken by the magnetization vector in one ferromagnetic layer can modulate the current across the device, depending on the direction taken by the magnetization of other ferromagnetic layer, that is, the magnetization affects the current. In magnetic nanostructures, the opposite effect is observed: a strongly spin-polarized electrical current applies a torque on the magnetic moment of a ferromagnetic layer initially pointing in a different direction [7, 8]. This effect is called “spin torque” or spin-transfer switching and is exploited in various nanodevices. A typical structure resembles the spin valve: two thin magnetic layers (pinned and free) are separated by a diamagnetic spacer. When an electrical current passes through the first layer, the spin of the electrons becomes oriented in the direction of that layer. This polarized current then passes through the second layer and again the electron spins align to the direction of the magnetic moment. However, because angular momentum is conserved, the difference in the orientation of the electrons caused by the magnetic layers is seen as a torque. On the contrary, if the current is first passed through the free layer, the fixed layer works as a filter, allowing only the component of parallel magnetic moments carried by the current, and reflecting back toward the free layer the reminder; therefore, a torque again acts on the free layer. It can be easily shown that a torque can be applied to the free layer in different directions. Therefore, the torque can either force the magnetization of the free layer toward or away from the direction of the fixed layer magnetization. If the torque is large enough, it can lead to switching of the magnetic moment of the free layer. Magnetization dynamics is described by the Landau–Lifshitz–Gilbert equation,
Magnetic Nanostructures and Spintronics, Fig. 5 Structure of a spin transfer magnetic RAM, seen in section. The bit and word line connections address a single memory cell, directly switching on or off the control transistor; when it is switched on, the current emitted by the drain (D) transfers the perpendicular magnetization of the hard layer (purple) to the free one (blue), writing data. Reading is done by lowering the voltage on the gate (G) and measuring the resistance across the two bit lines, which depends on the parallel/antiparallel configuration of the MTJ electrodes
whose terms represent spin precession, magnetization damping, and spin transfer torque [9]. Depending on the direction and strength of the polarized current, the magnetization of the free layer will end either parallel to the applied field, or antiparallel to the magnetization of the fixed layer, or even will indefinitely precess around the applied field direction in steady state. Spin torque effects are being investigated in view of their potential applications as alternative methods to efficiently write data in nonvolatile MRAMs exploiting smaller currents with respect to standard writing techniques involving transient magnetic fields [10]. ST-MRAMs are characterized by notable write endurance, good compatibility with standard CMOS processing and fast operation (Fig. 5). In fact, spintransfer-driven switching of magnetization can take 1 ns or less. The same effect can also be exploited in the design of fast electronic devices such as microwave (RF) oscillators microwave oscillators, converters, microwave sources and antennas. These devices are interesting because of their small size, low power consumption and good response to frequency modulation and phase locking. However, the current density needed to achieve fast magnetization reversal is often too high to ensure a reasonable lifetime of a spin-torque device (of the order of one decade for a MRAM’s operation time).
M
M
1254
Another major issue is the control over all parameters entering the Landau–Lifshitz–Gilbert equation in order to optimize the efficiency of the device. Recently, spin torque effects and related magnetization switching have attracted much attention in connection with the development of spintronic memristors at the nanoscale. In memristors, the resistance is a function of the history of the current through and voltage across the device. Spintronic sensing elements based upon spin torque effects can be designed to explore and memorize continuously varying current and voltage applied to the device [11].
Semiconductor Spintronics Spintronics in semiconductors – as opposed to or complementary to spintronics in metals – has been thoroughly investigated since the very beginning of research on spin-polarized currents. Semiconductor materials provide a variety of quantum effects which make them the ultimate choice for applications to micro- and nanoelectronics. Semiconductor-based solid state physics can be easily applied to nanostructures which in turn provide appropriate quantum confinement effects for carriers. Moreover the fabrication technology, based on reliable, accessible techniques of materials growth and process, has reached full maturity in the past decades. From this viewpoint, applying spintronics concepts to semiconductor materials was a straightforward step; however, research in the field has faced various difficulties and drawbacks which have so far prevented widespread applications. The main reason being that standard semiconductors are diamagnetic materials without a self-sustained degree of spin polarization of the carriers, contrary to the case of ferromagnetic metals. Two main strategies have been developed in order to cope with this shortcoming. The first one is based on injection of a spin-polarized current from a ferromagnetic metal such as Fe or Co into a semiconductor [12]. Of course both materials must be in the form of thin films in (more or less) close contact. In this way, a polarized current naturally produced in the ferromagnet can be injected in the semiconductor and propagate within it, although gradually losing the polarization degree. However, it can be shown that the time needed by the spin-polarized
Magnetic Nanostructures and Spintronics
current to fully de-polarize through spin relaxation processes, although intrinsically short (being of the order of 100 ns in GaAs and up to 100 ms in GaAs quantum dots) would be enough for electrons to travel across the semiconductor over mesoscopic distances without losing spin coherence (tens to hundreds of mm), therefore allowing one to design new devices. The residual polarization degree can be experimentally checked by analyzing the light emitted in the recombination process of polarized electrons with unpolarized holes in the quantum well of a layered heterostructure (this is called the “spin-LED”). The emerging electromagnetic radiation appears to be circularly polarized. A measurement of the polarization of emitted light is therefore a measure of polarization degree of electron spins. The drawback of this approach is represented by its low efficiency, in turn determined not only by the rate of loss of spin memory within the semiconductor, but mostly by the loss of spin polarization during the injection process from the ferromagnetic metal. In fact, the Schottky barrier existing at the metal–semiconductor interface hinders the injection and reduces the efficient injection of a spin-polarized current. This difficulty can be partially – although not entirely – circumvented by adopting solutions which include, for instance, the presence of a tunnel barrier contact between the metal and the semiconductor. The second strategy consists in building novel semiconductor materials which display ferromagnetic properties while keeping their semiconducting features [13]. Adding moderate concentrations of specific dopants carrying a magnetic moment may result in obtaining the so-called dilute ferromagnetic semiconductors. Magnetic dopants exist; trivalent Mn atoms act as acceptors in GaAs-like semiconductors (i.e., those belonging to the (III,TM)V series, TM being a metal provided with intrinsic magnetic moment) and have a high permanent magnetic moment (entirely related to the spin quantum number S ¼ 5/2). The key concept is that when magnetic Mn atoms occupy substitutional sites in the GaAs structure, they induce a polarization of the free carriers of the GaAs host. These are holes, rather than electrons, owing to the acceptor nature of dopants. Therefore, a magnetization of the itinerant carriers is sustained by the magnetization of localized moments on Mn ions; the latter is in turn sustained by that of itinerant carriers, in a sort of mutually sustained
Magnetic Nanostructures and Spintronics
1255
M
the electron spin during the ballistic motion in the channel, therefore modifying the current intensity by quantum interference effects. In this interesting device, the spin flip is induced by an electrical rather than a magnetic field by thoroughly exploiting the spin– orbit interaction of ballistic electrons in the onedimensional nanoconduit.
Magnetic Nanostructures and Spintronics, Fig. 6 Sketch of a dilute magnetic semiconductor spin configuration: the blue dots represent localized moments (Mn ions), while the red arrows represent itinerant spins (holes)
exchange effect (Fig. 6). It has been shown that, as a result, the spins of both Mn ions and mobile holes display a ferromagnetic ordering below the ordering temperature, although their magnetizations are opposite (strictly speaking, the semiconductor is therefore a ferrimagnetic rather than ferromagnetic material). However, spin-polarized, mobile holes can be transferred over long distances across the semiconductor always keeping their spin memory. A strict condition for the success of this strategy is to completely avoid clustering or agglomeration of magnetic dopants: their atoms must be evenly distributed in space. As a consequence, an upper dopant concentration limit (around 5%) exists. Unfortunately, the Curie temperatures of these ferromagnetic semiconductors are still quite low (100 K). This has prevented widespread diffusion of dilute magnetic semiconductor materials so far. Finally, it should be noted that many different spintronic devices based on semiconductors have been designed and/or achieved as demonstrators. Many borrow their working principles from classical optics of polarized, coherent light, exploiting the analogies existing between coherent optics and quantum coherence of ballistic electrons moving in onedimensional semiconductor “channels” and between classical polarization of light and spin polarization. A popular spintronic device has been, for example, the spin-FET proposed by Datta and Das as early as 1990, which keeps the overall geometry and general working principle of a conventional FET, while exploiting ferromagnetic source and drain electrodes to produce a spin-polarized electron flow in the channel. The transmittance of the device is modulated by the gate voltage which may, or may not, induce a flip of
Data Storage and Patterned Media Hard disk drives (HDDs) represent one of the most diffused systems for data storage, featuring extremely low costs per bit and reasonable data transfer rates. Nanostructured magnetic materials are present both in the read head and in the disc surface. The development of magnetic field sensors from magnetoresistive systems, to devices earlier based on GMR and nowadays based on TMR allowed the continuous increment in data areal density. On the other hand, recording materials, starting from structures magnetized in the disc plane, evolved to earlier structures with perpendicular magnetization, up to today’s antiferromagnetically coupled layers that increase data stability thanks to interlayer exchange coupling [14]. For what concerns recording media (Fig. 7), generally based on sputtered CoPtCrB alloys, the main problem is data retention rate: the smaller the size of a magnetic domain, the higher the effects of the demagnetizing field, the higher the switching probability. Reducing the magnetic domain size below the superparamagnetic limit would result in a useless media containing randomly distributed information. Nowadays the commercial data density is approaching an areal density of 100 Gbit/inch2, while laboratory demonstrators are available approaching 1 Tbit/inch2, based either on patterned media (PM) or thermally assisted magnetic recording. In particular, PM provides the key to break the superparamagnetic limit, because each nanometric magnetic domain corresponds to a confined physical grain that can be kept far from others avoiding self-demagnetization. The challenge of patterning a wide surface (some square inches) with homogeneously distributed 25-nm-big elements has been accomplished by means of density multiplication, a particular process where only a small fraction of the elements is patterned thanks to an electron beam system and the full feature density is achieved thanks to a block copolymer
M
M
1256
Magnetic Nanostructures and Spintronics, Fig. 7 Hard disk media magnetic domain structure. On the left conventional media for perpendicular recording, with randomly distributed magnetic domains and irregular bit contour. On the right patterned media, where each nanostructured element may carry one single bit of information
assembly guided by the underlying chemically modified former pattern [15]. Large area–patterned nanostructures are studied also for their interesting collective behavior [16], resulting in peculiar electrical properties [17] and potentially suitable to implement full magnetic logics, particularly promising for their low electrical consumption [18].
Cross-References ▶ Chemical Vapor Deposition (CVD) ▶ Electron Beam Lithography (EBL) ▶ Nanoparticles ▶ Physical Vapor Deposition ▶ Self-Assembly of Nanostructures
References 1. Sander, D.: The magnetic anisotropy and spin reorientation of nanostructures and nanoscale films. J. Phys. Condens. Matter 16, R603–R636 (2004) 2. Nogues, J., Schuller, I.K.: Exchange bias. J. Magn. Magn. Mater. 192, 203–232 (1999) 3. Chiolerio, A., Allia, P., Chiodoni, A., Pirri, F., Celegato, F., Coisson, M.: Thermally evaporated Cu-Co top spin-valve with random exchange bias. J. Appl. Phys. 101, 123915-1–123915-6 (2007) 4. Berkowitz, A.E., Takano, K.: Exchange anisotropy – a review. J. Magn. Magn. Mater. 200, 552–570 (1999)
Magnetic Resonance Force Microscopy 5. Parkin, S.S.P.: Giant magnetoresistance in magnetic nanostructures. Annu. Rev. Mater. Sci. 25, 357–388 (1995) 6. Yuasa, S., Djayapravira, D.D.: Giant tunnel magnetoresistance in magnetic tunnel junctions with a crystalline MgO (001) barrier. J. Phys. D Appl. Phys. 40, R337–R354 (2007) 7. Slonczewski, J.C.: Current-driven excitation of magnetic multilayers. J. Magn. Magn. Mater. 159, L1–L7 (1996) 8. Ralph, D.C., Stiles, M.D.: Spin transfer torques. J. Magn. Magn. Mater. 320, 1190–1216 (2008) 9. Cros, V., Boulle, O., Grollier, J., Hamzic, A., Munoz, M., Pereira, L.G., Petroff, F.: Spin transfer torque: a new method to excite or reverse a magnetization. C.R. Physique 6, 956–965 (2005) 10. Sun, Z.: Spin angular momentum transfer in currentperpendicular nanomagnetic junctions. IBM J. Res. Dev. 50, 81–100 (2006) 11. Wang, X., Chen, Y.: Spintronic memristor devices and application. In: Proceedings of Design, Automation & Test in Europe Conference & Exhibition (DATE), IEEE Conferences, Dresden, pp. 667–672 (2010) 12. Schmidt, J.: Concepts for spin injection into semiconductors—a review. J. Phys. D Appl. Phys. 38, R107–R122 (2005) 13. Awschalom, D.D., Flatte´, M.E.: Challenges for semiconductor spintronics. Nat. Phys. 3, 153–159 (2007) 14. Dietzel, A.: Hard disk drives. In: Waser, R. (ed.) Nanoelectronics and Information Technology, 2nd edn, pp. 617–629. Wiley, Weinheim (2005) 15. Density multiplication and improved lithography by directed block copolymer assembly for patterned media at 1 Tbit/in2 and beyond, white paper WPDM08EN-01, Hitachi GST, San Jose, CA (2008) 16. Adeyeye, A.O., Singh, N.: Large area patterned magnetic nanostructures. J. Phys. D Appl. Phys. 41, 153001-1–153001-29 (2008) 17. Chiolerio, A., Allia, P., Celasco, E., Martino, P., Spizzo, F., Celegato, F.: Magnetoresistance anisotropy in a hexagonal lattice of Co antidots obtained by thermal evaporation. J. Magn. Magn. Mater. 322, 1409–1412 (2010) 18. Graziano, M., Vacca, M., Chiolerio, A., Zamboni, M.: An NCL-HDL snake-clock-based magnetic QCA architecture. IEEE Trans. Nanotech. 10, 1141–1149 (2011)
Magnetic Resonance Force Microscopy Martino Poggio1 and Christian L. Degen2 1 Department of Physics, University of Basel, Basel, Switzerland 2 Department of Physics, ETH Zurich, Zurich, Switzerland
Synonyms Force-detected magnetic resonance
Magnetic Resonance Force Microscopy
1257
Definition
a
Magnetic resonance force microscopy (MRFM) is a type of magnetic resonance imaging (MRI) now capable of acquiring three-dimensional (3D) images of tiny samples with a spatial resolution better than 10 nm. By applying the techniques of scanning probe microscopy (SPM), MRFM surpasses the sensitivity of conventional, inductive MRI by eight orders of magnitude. MRFM images the interior of nanoscale objects noninvasively and with intrinsic chemical selectivity and, in 2009, it was used to capture 3D images of individual virus particles.
cantilever
M
rf source
laser interferometer magnet
sample
b
cantilever
laser interferometer
Overview sample
MRFM relies on the mechanical measurement of the weak magnetic force between a microscopic magnet and the magnetic moments in a sample. These moments are due to either the atomic nuclei with nonzero nuclear spin or electron spins present in a sample. For a single magnetic moment m in a magnetic field B, this force can be expressed as: F ¼ Hðm BÞ:
(1)
Using a compliant cantilever, one can measure the component of F along the cantilever’s deflection direction x^: Fx ¼
@ @Bz ðm BÞ ¼ m ¼ mG; @x @x
(2)
z where m points along ^ z and G ¼ @B @x is a magnetic field gradient. First, either the sample containing nuclear or electronic moments or the nano-magnet must be fixed to the cantilever. The sample and magnet must be in close proximity, sometimes up to a few tens of nanometers from each other. A nearby radiofrequency (rf) source produces magnetic field pulses similar to those used in conventional MRI, causing the moments to periodically flip. This periodic inversion generates an oscillating magnetic force acting on the cantilever. In order to resonantly excite the cantilever, the magnetic moments can be inverted at the cantilever’s mechanical resonance frequency. The cantilever’s mechanical oscillations are then measured by an optical interferometer or beam deflection detector. The electronic
magnet rf source
Magnetic Resonance Force Microscopy, Fig. 1 Schematics of an MRFM apparatus. (a) Corresponds to the “tip-on-cantilever” arrangement, such as used in the single-electron MRFM experiment of 2004 [1]. (b) Corresponds to the “sample-on-cantilever” arrangement, like the one used for the nanoscale virus imaging experiment in 2008 [2]
signal produced by the optical detector is proportional to the cantilever oscillation amplitude, which depends on the number of moments in the imaging volume. Spatial resolution results from the fact that the nanomagnet produces a magnetic field which is a strong function of position. The magnetic resonance condition and therefore the region in which the spins periodically flip is confined to a thin, approximately hemispherical “resonant slice” that extends outward from the nano-magnet (see Figs. 1 and 4). By scanning the sample in 3D through this resonant region, a spatial map of the magnetic moment density can be made. Different types of magnetic moments (e.g., 1H,13C, 19 F, or even electrons) can be distinguished due to their different magnetic resonance frequencies, giving an additional chemical contrast. In conventional nuclear magnetic resonance detection, the sample is placed in a strong static magnetic field in order to produce a Zeeman splitting between spin states. The sample is then exposed to an rf magnetic field of a precisely defined frequency. If this frequency
M
M
1258
Magnetic Resonance Force Microscopy
matches the Zeeman splitting, then the system absorbs energy from the rf radiation resulting in transitions between the spin states. The resulting oscillations of this ensemble of magnetic moments produce a timevarying magnetic signal that can be detected with a pickup coil. The electric current induced in the coil is then amplified and converted into a signal that is proportional to the number of moments (or spins) in the sample. In MRI this signal can be reconstructed into a 3D image of the sample using spatially varying magnetic fields and Fourier transform techniques. The magnetic fields produced by nuclear moments are, however, extremely small: more than one trillion (1012) nuclear spins are typically needed to generate a detectable signal. The advantage of force-detected over inductive techniques is that much smaller samples can be measured. In the latter case, the measurement can only be sensitive if the spins significantly alter the magnetic field within the pickup coil, i.e., if the spins fill a significant fraction of the coil volume. For spin ensembles with volumes significantly smaller than (1 mm)3, it is extremely challenging to realize pickup coils small enough to ensure an adequate filling factor. As a result, even the best resolutions achieved by inductively detected MRI require sample volumes of at least (3 mm)3 [3]. Mechanical resonators, in contrast, can now be fabricated with dimensions far below a micron, such that the sample’s mass (which is the equivalent to the filling volume in a pickup coil) is always significant compared to the bare resonator mass. In addition, mechanical devices usually show resonant quality factors that surpass those of inductive circuits by orders of magnitude, resulting in a much lower baseline noise. For example, state-of-the art cantilever force transducers achieve quality factors between 104 and 107, enabling the detection of forces of aN/Hz1/2 – less than a billionth of the force needed to break a single chemical bond. In addition, scanning probe microscopy offers the stability to position and image samples with nanometer precision. The combination of these features allows mechanically detected MRI to image at resolutions that are far below 1 mm and – in principle – to aspire to atomic resolution.
small number of magnetic moments remain astoundingly small: in an achievable gradient of 106 T/m, a single 1H moment, for example, produces a force of about 1020 N. In order to detect forces due to small ensembles of moments, the most sensitive MRFM takes advantage of the thermally limited force resolution of specially made nanomechanical cantilevers. In addition, it utilizes the fact that the force of interest can be made to occur in a narrow bandwidth around the cantilever’s mechanical resonance. The sensitivity and resolution of MRFM hinge on a simple signal-to-noise ratio, which is given by the ratio of the magnetic force power exerted on the cantilever over the force noise power of the cantilever device. For small enough volumes of spins, one must measure statistical spin polarizations; therefore one is interested in force powers (or variances) rather than force amplitudes:
Basic Methodology
The use of force-detection techniques in nuclear magnetic resonance (NMR) experiments dates back to Evans in 1955 [4], and was also used in torque magnetometry measurements by Alzetta and coworkers in the
Despite the large magnetic field gradients produced by specially made magnetic tips, the forces produced by
SNR ¼ N
ðmGÞ2 : SF Df
(3)
Here, N is the number of spins in the detection volume, m is the magnetic moment of the relevant spin, G is the relevant magnetic field gradient at the position of the sample, SF is the force noise spectral density set by the fluctuations of the cantilever sensor, and Df is the bandwidth of the measurement, determined by the spin relaxation rate. This expression gives the single-shot signal-to-noise ratio of a thermally limited MRFM apparatus. The larger this signal-to-noise ratio is, the better the spin sensitivity will be. From the four parameters appearing in (3), only two can be controlled and possibly improved. On the one hand, the magnetic field gradient G can be enhanced by using higher quality magnetic tips and by bringing the sample closer to these tips. On the other hand, the force noise spectral density SF can be reduced by going to lower temperatures and by making intrinsically more sensitive mechanical transducers.
Key Findings
Magnetic Resonance Force Microscopy
First Demonstration The first MRFM apparatus operated in vacuum and at room temperature with the DPPH sample attached to the cantilever. A millimeter-sized coil produced an rf magnetic field tuned to the electron spin resonance of the DPPH (220 MHz) with a magnitude of 1 mT. By changing the strength of a polarizing magnetic field (8 mT) in time, the electron spin magnetization in the DPPH was modulated. In a magnetic field gradient of 60 T/m, produced by a nearby NdFeB permanent magnet, the sample’s oscillating magnetization resulted in a time-varying force between the sample and the magnet. This force modulation was converted into mechanical vibration by the compliant cantilever. Displacement oscillations were detected by a fiberoptic interferometer achieving a thermally limited pffiffiffiffiffiffi force sensitivity of 3 fN/ Hz. During the years following this initial demonstration of cantilever-based magnetic resonance detection, the technique has undergone a series of developments toward higher sensitivities that, as of today, is about 107 times that of the 1992 experiment (see Fig. 2). The
M
1010 magnetic sensitivity (μP / Hz1 / 2)
1960s [5]. In 1991, Sidles, independent of this very early work, realized that highly sensitive magnetic resonance detection could be achieved using microfabricated cantilevers and nanoscale ferromagnets. In these early days of scanning probe miscroscopy, he proposed the MRFM technique as a method to improve the resolution of MRI to molecular length scales [6, 7]. Subsequently, the first micrometer-scale experimental demonstration using cantilevers was realized by Rugar [8], demonstrating mechanically detected electron spin resonance in a 30-ng sample of diphenylpicrylhydrazyl (DPPH). The visionary goal of the MRFM proposal was to eventually image molecules atom by atom, so as to directly map the 3D atomic structure of macromolecules [9]. Such a “molecular structure microscope” would have a dramatic impact on modern structural biology, and would be an important tool for many future nanoscale technologies. While this ultimate goal has not been achieved to date, the technique has undergone a remarkable development into one of the most sensitive spin detection methods available to researchers today. Among the important experimental achievements are the detection of a single electronic spin [1] and the extension of the spatial resolution of MRI from several micrometers to below 10 nm [2].
1259
108
106
104 single electron spin 10
2
1
single proton spin 1995
2000
2005
2010
Year
Magnetic Resonance Force Microscopy, Fig. 2 Advances to the sensitivity in force-detected magnetic resonance over time. Remarkably, improvements have closely followed a “Moore’s law” for over a decade, with the magnetic moment sensitivity doubling roughly every 8 months. Dots are experimental values [1, 2, 8, 10–14], and dashed lines indicate sensitivities of one electron and one proton magnetic moment (mP), respectively
following touches on two important demonstrations during this development. Several review articles and book chapters have appeared in the literature that discuss the historical progress of the technique more broadly and in richer detail [15–22]. Single-Electron MRFM The measurement of a single electron spin by Rugar et al. in 2004 concluded a decade of development on the MRFM technique and stands out as one of the first single-spin measurements in solid-state physics. A variety of developments led to the exceptional measurement sensitivity required for single-spin detection. These include the operation of the apparatus at cryogenic temperatures and high vacuum, the ion-beam milling of magnetic tips in order to produce large gradients, and the fabrication of mass-loaded attonewton-sensitive cantilevers [23] (shown in Fig. 3). The thermal noise in higher order vibrational modes of mass-loaded cantilevers is suppressed compared with the noise in the higher order modes of conventional, “flat” cantilevers. Since high-frequency vibrational noise in combination with a magnetic field gradient can disturb the electron spin, the mass-loaded levers proved to be a crucial advance for singleelectron MRFM. In addition, Rugar et al. developed a sensitive interferometer employing only a few
M
M
1260
Magnetic Resonance Force Microscopy
120 μm
Magnetic Resonance Force Microscopy, Fig. 3 Image of an ultrasensitive mass-loaded Si cantilever taken from an optical microscope. This type of cantilever, which has a spring constant under 100 mN/m, has been used as a force transducer in the many of the latest MRFM experiments [23]
nanowatts of optical power for the detection of cantilever displacement [24]. This low incident laser power is crucial for achieving low cantilever temperatures and thus minimizing the effects of thermal force noise. A low-background measurement protocol called OSCAR based on the NMR technique of adiabatic rapid passage was also employed [10]. Finally, the experiment required the construction of an extremely stable measurement system capable of continuously measuring for several days in an experiment whose single-shot signal-to-noise ratio was just 0.06 [1]. Nano-MRI The detection of a single nuclear spin is far more challenging than that of single electron spin. The intrinsic magnetic moment of a nucleus is much smaller than that of an electron; as a result the force produced by a nuclear moment in (2) is proportionally smaller. A 1H nucleus (proton), for example, possess a magnetic moment that is only 1/650 of an electron spin moment. Other important nuclei, like 13C or a variety of isotopes present in semiconductors, have even weaker magnetic moments. In order to observe single nuclear spins, it is necessary to improve the state-of-the-art sensitivity by another two to three orders of magnitude. While not out of the question, this is a daunting task that requires significant advances to all aspects of the MRFM technique. Despite these challenges, significant progress has been made. In particular, one experiment that used single tobacco mosaic virus (TMV) particles as the sample shows both the feasibility of MRI imaging with nanometer resolution and the applicability of MRFM to biologically relevant samples. Figure 4 is a representation of the MRFM apparatus used in the experiment. The virus particles were transferred to the cantilever end by dipping the tip of the
Magnetic Resonance Force Microscopy, Fig. 4 Artistic view of the MRFM apparatus used for MRI of individual tobacco mosaic virus particles. Pictured is the cantilever, the laser beam used for position sensing, and the Cu microwire rf source. The inset shows a close-up representation of the gold-coated end of the cantilever with attached virus particles. On top of the microwire, is the magnetic FeCo tip with the “mushroom”shaped resonant slice hovering above
cantilever into a droplet of aqueous solution containing suspended TMV. As a result, some TMVs were attached to the gold layer previously deposited on the cantilever end. The density of TMV on the gold layer was low enough that individual particles could be isolated. Then the cantilever was mounted into the low-temperature, ultra-high-vacuum measurement system and aligned over a magnetic tip and rf field source. After applying a static magnetic field of about 3 T, resonant rf pulses were applied to the rf source in order to flip the 1H nuclear spins at the cantilever’s mechanical resonance. Finally, the end of the cantilever was mechanically scanned in 3D over the magnetic tip. Given the extended geometry of the region in which the resonant condition is met, i.e., the “resonant slice,” a spatial scan does not directly produce a map of the 1H distribution in the sample. Instead, each data point in the scan contains force signal from 1H spins at a variety of different positions. In order to reconstruct the 3D spin density (the MRI image), the force map must be deconvolved by the point spread function defined by the resonant slice. Fortunately, this point spread function can be accurately determined using a magnetostatic model based on the physical geometry of the magnetic tip and the tip magnetization. Deconvolution of the force map into the three-dimensional 1H spin density can be done in several different ways; for the
Magnetic Resonance Force Microscopy Magnetic Resonance Force Microscopy, Fig. 5 Nanoscale MRI images of tobacco mosaic virus (TMV) particles acquired by MRFM [2]. (a) The series of images to the left depicts the 3D 1H spin density of virus particles deposited on the end of the cantilever. Black represents very low or zero density of hydrogen, while white is high hydrogen density. The right side shows a representative xy-plane, with several viral fragments visible, and a cross section (xz-plane) of two virus particles that reveals an underlying molecular layer of hydrocarbons covering the cantilever surface. (b) 3D 1H spin density recorded on a different region of the same cantilever as in (a), showing an intact and several fragmented virus particles. The right side shows a representative xy-plane
1261
M
a y
490 nm 410 nm
x
6.4 nm
100 nm
20 nm
CROSS-SECTION z x
z y x 390 nm
b m
0n
39
4.3 nm
M
y x z 100 nm y x
results presented in [2] the authors applied the iterative Landweber deconvolution procedure suggested in an earlier MRFM experiment [25, 26]. This iterative algorithm starts with an initial estimate for the spin density of the object and then improves the estimate successively by minimizing the difference between the measured and predicted spin signal maps. The iterations proceed until the residual error becomes comparable with the measurement noise. The result of a representative experiment is shown in Fig. 5. Here, clear features of individual TMV particles, which are cylindrical, roughly 300-nm-long and 18 nm in diameter, are visible and can be confirmed
against a scanning electron micrograph (SEM) of the same region. Both whole virus particles and particle fragments are observed. The imaging resolution, while not fine enough to discern any internal structure of the virus particles, constitutes a 1,000-fold improvement over conventional MRI and a corresponding improvement of volume sensitivity by about 100 million.
Comparison to Other Techniques The unique position of MRFM among high-resolution microscopies becomes apparent when comparing it to
M
1262
other, more established nanoscale imaging techniques. As a genuine scanning probe method, MRFM has the potential to image matter at atomic resolution. While atomic-scale imaging is routinely achieved in scanning tunneling microscopy and atomic force microscopy, these techniques are confined to the top layer of atoms and cannot penetrate below surfaces [27, 28]. Moreover, in standard scanning probe microscopy (SPM), it is difficult and in many situations impossible to identify the chemical species being imaged. Since MRFM combines SPM and MRI, these restrictions are lifted. The 3D nature of MRI permits acquisition of subsurface images with high spatial resolution even if the probe is relatively far away. As with other magnetic resonance techniques, MRFM comes with intrinsic elemental contrast and can draw from established NMR spectroscopy procedures to perform detailed chemical analysis. In addition, MRI does not cause any radiation damage to samples, as do electron and X-ray microscopies. MRFM also distinguishes itself from superresolution optical microscopies that rely on fluorescence imaging [29]. On the one side, optical methods have the advantage of working in vivo and they have the ability to selectively target the desired parts of a cell. Fluorescent labeling is now a mature technique which is routinely used for cellular imaging. On the other side, pushing the resolution into the nanometer range is hampered by fundamental limitations, in particular, the high optical powers required and the stability of the fluorophores. Moreover, fluorescent labeling is inextricably linked with a modification of the target biomolecules, which alters the biofunctionality and limits imaging resolution to the physical size of the fluorophores. MRFM occupies a unique position among other nanoscale spin detection approaches. While single electron spin detection in solids has been shown using several techniques, these mostly rely on the indirect readout via electronic charge [30, 31] or optical transitions [32, 33]. In another approach, the magnetic orientation of single atoms has been measured via the spin-polarized current of a magnetic STM tip or using magnetic exchange force microscopy [34–36]. These tools are very valuable to study single surface atoms; however, they are ill-suited to map out subsurface spins such as paramagnetic defects. In contrast, MRFM directly measures the magnetic moment of a spin, without resorting to other degrees of
Magnetic Resonance Force Microscopy
freedom, making it a very general method. This direct measurement of magnetic moment (or magnetic stray field) could also be envisioned using other techniques, namely, SQuID microscopy [37], Hall microscopy [38], or recently introduced diamond magnetometry based on single nitrogen-vacancy centers [39–41]. So far, however, none of these methods have reached the level of sensitivity needed to detect single electron spins, or volumes of nuclear spins much less than 1 mm [42, 43]. It is certainly possible that future improvements to these methods – especially to diamond magnetometry – may result in alternative techniques for nanoscale MRI that surpass the capabilities of MRFM.
Outlook Despite the tremendous improvements made to MRFM over the last decade, several important obstacles must be overcome in order to turn the technique into a useful tool for biologists and materials scientists. Most existing MRFM instruments are technically involved prototypes; major hardware simplifications will be required for routine screening of nanoscale samples. Suitable specimen preparation methods must be developed that are compatible with the low-temperature, high-vacuum environment required for the microscope to operate at its highest sensitivity and resolution. While this is particularly challenging for biological samples, protocols exist which could be adapted to MRFM. In cryo-electron microscopy, for example, dispersed samples are vitrified to preserve their native structure by plungefreezing in liquid nitrogen [44]. As objects become smaller, isolation of samples and suppression of unwanted background signals from surrounding material will become increasingly important. The conditions under which the latest MRFM imaging experiments were carried out are remarkably similar to those prevailing in cryo-electron microscopy, the highest resolution 3D imaging technique commonly used by structural biologists today. Cryo-electron microscopy, like MRFM, operates at low temperatures and in high vacuum, requires long averaging times (on the order of days) to achieve sufficient contrast, and routinely achieves resolutions of a few nanometers [45, 46]. Unlike MRFM, however, electron microscopy suffers from fundamental
Magnetic Resonance Force Microscopy
limitations that severely restrict its applicability. Specimen damage by the high-energy electron radiation limits resolution to 5–10 nm if only a single copy of an object is available. Averaging over hundreds to thousands of copies is needed to achieve resolutions ˚ [47]. In addition, unstained images approaching 10 A have intrinsically low contrast, whereas staining comes at the expense of modifying the native structure. MRFM has the unique capability to image nanoscale objects in a noninvasive manner and to do so with intrinsic chemical selectivity. For this reason the technique has the potential to extend microscopy to the large class of structures that show disorder and therefore cannot be averaged over many copies. These structures include such prominent examples as HIV, influenza virus, and amyloid fibrils. Virtually all of these complexes are associated with important biological functions ranging from a variety of diseases to the most basic tasks within the cellular machinery. For such complexes, MRFM has the potential not only to image the 3D macromolecular arrangement, but also to selectively image specific domains in the interior through isotopic labeling. While the most exciting prospect for MRFM remains in its application to structural imaging in molecular biology, its applications are not limited to biological matter. For example, most semiconductors contain nonzero nuclear magnetic moments. Therefore, MRFM may prove useful for subsurface imaging of nanoscale electronic devices. MRFM also appears to be the only technique capable of directly measuring the dynamics of the small ensembles of nuclear spin that limit electron spin coherence in single semiconductor quantum dots. Polymer films and self-assembled monolayers – important for future molecular electronics – are another exciting target for MRFM and its capability to image chemical composition on the nanoscale. Finally, isotopically engineered materials are becoming increasingly important for tuning a variety of physical properties such as transport and spin. Researchers currently lack a general method for noninvasively imaging the isotopic composition of these materials [48–50]; MRFM techniques could fill this void. Thanks to H. J. Mamin and D. Rugar of the IBM Almaden Research Center for their many detailed comments and very helpful discussions pertaining to this text.
1263
M
Cross-References ▶ Atomic Force Microscopy ▶ Basic MEMS Actuators ▶ Imaging Human Body Down to Molecular Level
References 1. Rugar, D., et al.: Single spin detection by magnetic resonance force microscopy. Nature 430, 329 (2004) 2. Degen, C.L., Poggio, M., Mamin, H.J., Rettner, C.T., Rugar, D.: Nanoscale magnetic resonance imaging. Proc. Natl. Acad. Sci. U.S.A. 106, 1313 (2009) 3. Ciobanu, L., Seeber, D.A., Pennington, C.H.: 3D MR microscopy with resolution by 3.7 mm by 3.3 mm by 3.3 mm. J. Magn. Reson. 158, 178 (2002); Lee, S.-C., et al.: One Micrometer Resolution NMR Microscopy. J. Magn. Reson. 150, 207 (2001); Glover, P., Mansfield, P.: Limits to magnetic resonance microscopy. Rep. Prog. Phys. 65, 1489 (2002) 4. Evans, D.F.: Direct observation of static nuclear susceptibilities at room temperature. Philos. Mag. 1, 370 (1956) 5. Alzetta, G., Arimondo, E., Ascoli, C., Gozzini, A.: Paramagnetic resonance experiments at low fields with angular-momentum detection. Il Nuovo Cimento B 62, 392 (1967) 6. Sidles, J.A.: Noninductive detection of single-proton magnetic resonance. Appl. Phys. Lett. 58, 2854 (1991) 7. Sidles, J.A.: Folded Stern-Gerlach experiment as a means for detecting nuclear magnetic resonance in individual nuclei. Phys. Rev. Lett. 68, 1124 (1992) 8. Rugar, D., Yannoni, C.S., Sidles, J.A.: Mechanical detection of magnetic resonance. Nature 360, 563 (1992) 9. Sidles, J.A., Garbini, J.L., Drobny, G.P.: The theory of oscillator-coupled magnetic resonance with potential applications to molecular imaging. Rev. Sci. Instrum. 63, 3881 (1992) 10. Stipe, B.C., et al.: Magnetic dissipation and fluctuations in individual nanomagnets measured by ultrasensitive cantilever magnetometry. Phys. Rev. Lett. 86, 2874 (2001) 11. Rugar, D., et al.: Force detection of nuclear magnetic resonance. Science 264, 1560 (1994) 12. Wago, K., Z€ uger, O., Kendrick, R., Yannoni, C.S., Rugar, D.: Low-temperature magnetic resonance force detection. J. Vac. Sci. Technol. B 14, 1197 (1996) 13. Mamin, H.J., Budakian, R., Chui, B.W., Rugar, D.: Detection and manipulation of statistical polarization in small spin ensembles. Phys. Rev. Lett. 91, 207604 (2003) 14. Mamin, H.J., et al.: Isotope-selective detection and imaging of organic nanolayers. Nano Lett. 9, 3020 (2009) 15. Sidles, J.A., et al.: Magnetic resonance force microscopy. Rev. Mod. Phys. 67, 249 (1995) 16. Nestle, N., Schaff, A., Veeman, W.S.: Mechanically detected NMR, an evaluation of the applicability for chemical investigations. Prog. Nucl. Magn. Reson. Spectrosc. 38, 1 (2001)
M
M
1264
17. Suter, A.: The magnetic resonance force microscope. Prog. Nucl. Magn. Reson. Spectrosc. 45, 239 (2004) 18. Berman, G.P., Borgonovi, F., Gorshkov, V.N., Tsifrinovich, V.I.: Magnetic Resonance Force Microscopy and a Singlespin Measurement. World Scientific, Singapore-New Jersey-London-Hong Kong (2006) 19. Hammel, P.C., Pelekhov, D.V.: The magnetic resonance microscope. In: Kronm€ uller, H., Parkin, S. (eds.) Handbook of Magnetism and Advanced Magnetic Materials. Spintronics and magnetoelectronics, vol. 5. John Wiley & Sons, Chichester (2007). ISBN 978-0-470-02217-7 20. Kuehn, S., Hickman, S.A., Marohn, J.A.: Advances in mechanical detection of magnetic resonance. J. Chem. Phys. 128, 052208 (2008) 21. Barbic, M.: Magnetic resonance force microscopy, in magnetic resonance microscopy: spatially resolved NMR techniques and applications. In: Codd, S.L., Seymour, J.D. (eds.) Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany (2009) 22. Poggio, M., Degen, C.L.: Force-detected nuclear magnetic resonance: recent advances and future challenges. Nanotechnology 21, 342001 (2010) 23. Chui, B.W., et al.: Mass-loaded cantilevers with suppressed higher-order modes for magnetic resonance force microscopy. TRANSDUCERS, 12th International Conference on Solid-State Sensors, Actuators and Microsystems, Boston, vol. 2, p. 1120 (2003) 24. Mamin, H.J., Rugar, D.: Sub-attonewton force detection at millikelvin temperatures. Appl. Phys. Lett. 79, 3358 (2001) 25. Chao, S., Dougherty, W.M., Garbini, J.L., Sidles, J.A.: Nanometer-scale magnetic resonance imaging. Rev. Sci. Instr. 75, 1175 (2004) 26. Dobigeon, N., Hero, A.O., Tourneret, J.-Y.: Hierarchical bayesian sparse image reconstruction with application to MRFM. IEEE Trans. Image Process. 18, 2059 (2009) 27. Hansma, P., Tersoff, J.: Scanning tunneling microscopy. J. Appl. Phys. 61, R1–R23 (1987) 28. Giessibl, F.: Advances in atomic force microscopy. Rev. Mod. Phys. 75, 949–983 (2003) 29. Huang, B., Bates, M., Zhuang, X.: Super-resolution fluorescence microscopy. Annu. Rev. Biochem. 78, 993–1016 (2009) 30. Elzerman, J.M., et al.: Single-shot read-out of an individual electron spin in a quantum dot. Nature 430, 431 (2004) 31. Xiao, M., Martin, I., Yablonovitch, E., Jiang, H.W.: Electrical detection of the spin resonance of a single electron in a silicon field-effect transistor. Nature 430, 435 (2004) 32. Wrachtrup, J., von Borczyskowski, C., Bernard, J., Orritt, M., Brown, R.: Optical detection of magnetic resonance in a single molecule. Nature 363, 244 (1993) 33. Jelezko, F., Popa, I., Gruber, A., Tietz, C., Wrachtrup, J.: Single spin states in a defect center resolved by optical spectroscopy. Appl. Phys. Lett. 81, 2160 (2002) 34. Heinze, S., et al.: Real-space imaging of two-dimensional antiferromagnetism on the atomic scale. Science 288, 1805– 1808 (2000) 35. Durkan, C., Welland, M.E.: Electronic spin detection in molecules using scanning-tunneling- microscopy-assisted electron-spin resonance. Appl. Phys. Lett. 80, 458 (2002) 36. Kaiser, U., Schwarz, A., Wiesendanger, R.: Nature 446, 522–525 (2007) 37. Kirtley, J.R., et al.: High-resolution scanning SQUID microscope. Appl. Phys. Lett. 66, 1138 (1995)
Magnetic Self-Assembly 38. Chang, A.M., et al.: Scanning hall probe microscopy. Appl. Phys. Lett. 61, 1974 (1992) 39. Degen, C.L.: Scanning magnetic field microscope with a diamond single-spin sensor. Appl. Phys. Lett. 92, 243111 (2008) 40. Maze, J.R., et al.: Nanoscale magnetic sensing with an individual electronic spin in diamond. Nature 455, 644 (2008) 41. Balasubramanian, G., et al.: Nanoscale imaging magnetometry with diamond spins under ambient conditions. Nature 455, 648 (2008) 42. Degen, C.L.: Nanoscale magnetometry: microscopy with single spins. Nat. Nanotechnol. 3, 643 (2008) 43. Blank, A., Suhovoy, E., Halevy, R., Shtirberg, L., Harneit, W.: ESR imaging in solid phase down to sub-micron resolution: methodology and applications. Phys. Chem. Chem. Phys. 11, 6689–6699 (2009) 44. Taylor, K.A., Glaeser, R.M.: Electron diffraction of frozen, hydrated protein crystals. Science 186, 1036 (1974) 45. Lucic, V., Forster, F., Baumeister, W.: Structural studies by electron tomography: from cells to molecules. Annu. Rev. Biochem. 74, 833 (2005) 46. Subramaniam, S.: Bridging the imaging gap: visualizing subcellular architecture with electron tomography. Curr. Opin. Microbiol. 8, 316 (2005) 47. Glaeser, R.M.: Cryo-electron microscopy of biological nanostructures. Phys. Today 61, 48 (2008) 48. Shimizu, Y., Itoh, K.: Growth and characterization of shortperiod silicon isotope superlattices. Thin Solid Films 508, 160–162 (2006) 49. Shlimak, I., Safarov, V., Vagner, I.: Isotopically engineered silicon/silicon-germanium nanostructures as basic elements for a nuclear spin quantum computer. J. Phys. Cond. Matter 13, 6059 (2001) 50. Kelly, T., Miller, M.: Atom probe tomography. Rev. Sci. Instrum. 78, 031101 (2007)
Magnetic Self-Assembly ▶ Magnetic-Field-Based Self-Assembly
Magnetic-Field-Based Self-Assembly Li Zhang1, Bradley J. Nelson1 and Lixin Dong2 1 Institute of Robotics and Intelligent Systems, ETH Zurich, Zurich, Switzerland 2 Electrical and Computer Engineering, Michigan State University, East Lansing, MI, USA
Synonyms Magnetic assembly; Magnetic self-assembly
Magnetic-Field-Based Self-Assembly
Definition Magnetic-field-based self-assembly refers to the autonomous organization of discrete components to form ordered patterns, structures, devices, and systems driven by an external magnetic field or by magnetic fields from permanent dipoles incorporated into parts. The magnetic-field-based self-assembly approach can be applied for the assembly of individual magnetic components such as magnetic nanoparticles and nanowires. It can also be applied for the self-assembly of multiple components, such as components with complementary micromagnet arrays or a mixture of magnetic and nonmagnetic components.
Overview and Concepts For magnetic self-assembly, a magnetic interaction between magnetic components is required [1]. The strength of the magnetic interaction force is dependent on the distance and orientation between the magnetic components, the magnetization of the materials, and the dimensions and shapes of the magnetic components. Most research studies have used either ferromagnetic materials or superparamagnetic particles as magnetic components for magnetic assembly. Generally, ferromagnetic materials are classified as either hard magnetic or soft magnetic depending on their coercivity [2]. For hard magnetic materials that are permanently magnetized, an external magnetic field may not be required for self-assembly. The magnetization of soft-magnetic materials is highly dependent on the external magnetic field, and they tend to return to their nonmagnetized state as the field is removed because of their low remanence magnetization. When the dimensions of ferromagnetic material are reduced to a single magnetic domain, superparamagnetism occurs in which the saturation magnetization can be as high as that in a ferromagnetic state but its remanence magnetization decreases to zero. Therefore, an external field is required for the magnetic self-assembly of soft magnetic or superparamagnetic components. Furthermore, shape effect, i.e., the geometrically induced directional dependency of the magnetic properties within the magnetic component [2], significantly influences how the components magnetize and, hence, how soft-magnetic or superparamagnetic components assemble. In addition, the
1265
M
environment for self-assembly is crucial, since the components move with respect to one another [1]. One issue that can occur for micro-/nanoscale components is that van der Waals interactions and other surface forces dominate gravitational forces and other volume forces at this scale. One way to reduce these interactions is to perform self-assembly in liquid or at a liquid-air interface. Magnetic-field-based self-assembly has many advantages [3], for instance, it is compatible for both wet and dry environments; it does not require physical contact for assembly; it is scalable in a large range, i.e., from nano-/meso- to macroscale; it has robust control of angular and lateral alignment; and, in addition, a weak-strength magnetic field is considered safe for in vivo applications. Magnetic assembly has potential applications in different fields, such as microfluidics and lab on a chip [4–7], wireless biosensing [8], biomedical engineering [9], microelectronics [10–12], and microrobotics [13–15]. An overview of magnetic self-assembly including concepts and applications is shown in Fig. 1.
Key Research Findings Static Self-Assembly of Single Magnetic Components Magnetic particles and nanowires are two fundamental components for magnetic self-assembly of onedimensional (1-D) structures. Figure 2a shows a suspension of superparamagnetic beads, after a uniform magnetic field has been applied. The beads are driven to form chains and align with the external field. Selfassembly is caused by the induced magnetic moments in the beads, and the magnetic alignment of the chain resembles an elongated magnetic body aligned with the applied field [6]. In comparison to individual magnetic beads, which have no magnetic shape anisotropy, a magnetic nanowire has an “easy” magnetized axis along its long axis. When a magnetic field is applied to nanowire suspensions, they are driven to form chains and self-align their easy axis along the magnetic field, (see Fig. 2b) [16]. The results also indicate that to self-assemble magnetic micro-/nanoparticles and nanowires in fluid, a weak-strength magnetic field, i.e., in a range of milli-Tesla (mT), is sufficient, which is approximately three orders of magnitude
M
M
1266
Magnetic-Field-Based SelfAssembly, Fig. 1 Overview of magnetic self-assembly: concept and applications
Magnetic-Field-Based Self-Assembly Single component
Multi-components
Ferromagnetic materials - soft-magnets - hard-magnets - superparamagnetic beads
- Ferromagnetic materials - hybrid materials/alloy - Non-magnetic colloids - Cells
Magnetic self-assembly (static,dynamic) - ordered patterns/structures - devices/system
Micromanipulation
Functionalization
Integration
Fields of applications
MEMS/Lab on a Chip
lower than most clinical magnetic resonance imaging (MRI) systems. Therefore, the magnetic self-assembly approach is considered noninvasive for biomedical applications such as cell manipulation and biosensing. The construction of 2-D or 3-D patterns using single magnetic components is not straightforward. One method to overcome this issue is to perform the assembly in steps. Magnetic assembly of ferromagnetic nanowires to create networks comprised of crossshaped and T-shaped junctions was shown by Hangarter et al. [17]. The hierarchical 2-D patterns were obtained using two steps of magnetic alignment in which crossing angles from 30 to 90 were obtained. The magnetic self-assembly and alignment process were conducted by placing two permanent magnets around the substrate covered with droplets of the nanowire suspensions. The pattern configuration was controlled by changing the direction and the strength of the applied field in two steps. Magnetic field strengths of 28, 76, and 124 mT were applied in the experiments by adjusting the distance between the two permanent magnets [17]. It was also demonstrated that on a patterned substrate with ferromagnetic
Robotics
Biology
Microelectronics
Netted systems
electrodes, the electrodes act as micromagnet arrays which can be interconnected by magnetic nanowires due to the dipole interaction. Magnetic nanowire cross junctions were realized by a two-step alignment as shown in Fig. 2c–d. Static Self-Assembly of Multiple Components For self-assembly of more complex structures than simple periodic structures or crystals, multiple components in a mixture must come together in an organized way [1]. Unlike the magnetic self-assembly, using single components such as beads and nanowires, Yellen and coworkers reported that 3-D flower-like colloidal structures with a rotational symmetry are formed using multiple components, i.e., nonmagnetic and paramagnetic colloids mixed in a ferrofluid consisting of magnetite nanoparticles suspended in water, when a static and uniform magnetic field is applied [18]. The results showed that the paramagnetic and nonmagnetic colloids interact anti-ferromagnetically in the multiple-component suspensions, leading to the formation of ring structures as shown in Fig. 3 [18]. This is because the colloids less magnetizable than the ferrofluid exhibit
Magnetic-Field-Based Self-Assembly
1267
a
M
b 50 mm
B
B 60 µm
c
B
B
i
ii
iii
iv
d
Magnetic-Field-Based Self-Assembly, Fig. 2 Magnetic self-assembly of superparamagnetic beads and Ni nanowires. (a) Self-assembly of 1 mm diameter superparamagnetic microbeads suspension to form straight chains. The applied field strength (B) is 2.5 mT. (b) Ni nanowire chains are formed when a low strength magnetic field is applied. The applied field strength (B) is 0.8 mT. The individual nanowires have a diameter of 350 nm and a length of 12 mm (Reprinted with permission from Ref. [16]. Copyright 2002, American Institute of Physics). Insets: schematic drawings of magnetic
self-assembly of beads (a) and wires (b) when a field is applied. (c) The two-step alignment procedure with (i) alignment in a single direction, (ii) followed by a substrate rotation, (iii) applying an external magnetic field in a perpendicular direction to the assembled nanowire, and (iv) the second alignment step to generate a junction. (d) A junction of crossing nickel nanowires on a ferromagnetic electrode. The nanowires have a diameter of 200 nm. (c–d Reproduced with permission from Ref. [17]. Copyright 2007 IOP)
diamagnetic property, which implies that the particle moments prefer to align antiparallel with respect to the direction of the applied uniform field. Figure 3a and 3b show schematics of magnetic self-assembly from two and three colloidal components, respectively, which were mixed into ferrofluid. Figure 3c shows the experimental results of the fluorescence-labeled colloids, corresponding to the schematic in Fig. 3b, hierarchically self-assembling to ring-like structures with multipole symmetry configurations. Additionally, the concentration of the magnetic nanoparticles and the strength of the external magnetic field are two crucial factors for such a self-assembly process. In those tests, the applied uniform magnetic field was not higher than 100 Oe (in free air 100 Oe is equal to 10 mT). An alternative approach to construct 3-D patterns is using the origami technique to fold planar components. The magnetic assembly approach is used for the alignment of the components with a high positioning precision and efficiency. Barbasathis et al. showed
the self-alignment of thin films with nanomagnets on the surfaces [10]. The film are folded into 3-D configurations using a magnetic field. The nanomagnet arrays were prepared by lithographic patterning technique. Two types of alignment methods (form A and B) were investigated as presented in Fig. 4a–b, in which the magnets are magnetized before and after folding, respectively. There are two main advantages to perform the alignment after the folding process (i.e., form B): (1) the magnetic interaction forces are enhanced by one order of magnitude due to the applied field; (2) since an external field is “ON” during the assembly process, it can be applied for a broader range of ferromagnetic materials and the shapes of the nanomagnets. Magnetic alignment tests of folding a freestanding Si3N4 membrane on its own substrate, both are patterned with 75 nm thick cobalt nanomagnets, are conducted, and the layer-to-layer alignment accuracy of approximately 250 nm was achieved [10]. The applied field strengths were in the
M
M
1268
Magnetic-Field-Based Self-Assembly
a
Rings
Poles
Top view
Field off
Field on
Field on
b
Two tone
Flowers
c
Magnetic-Field-Based Self-Assembly, Fig. 3 (a) Illustration of magnetic assembly of two-component colloidal particle (red and gray spheres) mixed with ferrofluid (brown dots). No structures form without an external magnetic field. When a field is applied (black arrows), equatorial and polar arrangements form when the particle magnetizations, respectively, have the opposite and the same sense. Particle dipole moments are indicated by the yellow arrows. (b) Self-assembly in threecomponent suspensions is illustrated. “Two-tone” structures
consist of simultaneous polar and equatorial arrangements assembling on a single core particle, whereas “flower” structures are formed by the fractal assembly of multiple equatorial arrangements. (c) Fluorescent images of multiplecomponent particle assembly in ferrofluid corresponding to the illustrations in (b). Scale bars are 20 mm (a–c Adapted with permission from Macmillan Publishers Ltd. Ref. [18]. Copyright 2008)
range of several hundred milli-Teslas. The layer-tolayer magnetic self-assembly using permanent magnet arrays was reported by Arnold et al. to assemble different microscale components (see Fig. 4c), in which solder pads were designed for permanent bonding of the assembled parts [11]. Moreover, the results
indicate that the magnetic bonding force is enhanced by using an array of micromagnets compared to a single piece of magnet with the same volume. Spontaneous folding of flexible 2-D components to form freestanding hollow 3-D objects using magnetic self-assembly was demonstrated by Whitesides’
Magnetic-Field-Based Self-Assembly
a
1269
Form A
i
Bext
M
e
d n
ii
iii
m=2πR/6 = 12.3 mm n=2πR/2 = 36.7 mm
m
5 mm
g
f
b
Form B
i 5 mm
h m ii
Bext
i
iii n
c
Part B
5 mm
j m
South pole
k
Solder pads
North poles
M n
Part A
5 mm
Magnetic-Field-Based Self-Assembly, Fig. 4 (a–b) Two forms of the alignment method using magnetic assembly. In form A, hard magnets are magnetized in the same direction before folding. In form B, soft magnets are magnetized to saturation during the folding (Reprinted with permission from Ref. [10]. Copyright 2006, American Vacuum Society). (c) Micromagnets array for self-assembly of two complementary parts (part A and B) (Reproduced with permission from
Ref. [11]. Copyright 2007 IEEE). (d–k) Three-dimensional magnetic assembly from planar components. Three-dimensional structures (e, g, i, k) self-assembled from magnetically patterned sheets (d, f, h, j). The direction of the magnetic dipoles is indicated with white arrows (d–k Reproduced with permission from Ref. [19]. Copyright 2005 National Academy of Sciences, USA)
group. To construct a spherical shell, they investigated elastomeric sheets with three different 2-D shape designs [19]: an equatorial cut (see Fig. 4d), an orange-peel cut (see Fig. 4f), and a flower-petal cut (see Fig. 4h and j). To self-fold the sheets, different patterns of magnetic dipoles generated by permanent magnets were examined. The self-assembly processes were performed in water or at the air-water interface with some minutes of agitation. The magnetic and mechanical interactions influence the self-assembly significantly [19].
Dynamic Self-Assembly at a Liquid-Air Interface Self-assembly processes can be classified into two categories: static self-assembly and dynamic self-assembly [1]. The self-assembly approaches presented in the previous section all belong to the static self-assembly category. An assembly is defined as static when the ordered structure is in a stable equilibrium not dissipating energy [20]. Unlike the static self-assembly, in a dynamic selfassembly process, an ordered pattern/structure only occurs when it is dissipating energy [1]. Recently,
M 1
1270 2
Magnetic-Field-Based Self-Assembly 3
4
5
6
7
8
9 1 cm
10
11
12
13
14
15
16
17
18
19
Magnetic-Field-Based Self-Assembly, Fig. 5 Dynamic patterns formed by various numbers (n) of magnetic disks rotating at the ethylene glycol/water-air interface. This interface is 27 mm above the plane of the external magnet. The disks are composed of a section of polyethylene tube (white) of outer diameter 1.27 mm, filled with poly(dimethylsiloxane), PDMS, doped with 25 wt% of magnetite (black center). All disks spin around their centers at o ¼ 700 rpm, and the entire aggregate slowly (O> s2 , i.e., in a case of materials 1 and 2 showing conducting and insulating properties, respectively, E1 ¼ 0, E2 ¼ V/L with V ¼ Qs/C2 are obtained in steady state. That is, no electric field is formed in conducting material 1 and an electric field is present in material 2.
Maxwell–Wagner Effect
1279
FET System (3-Electrode System) For double-layer devices with a sandwich structure as shown in Fig. 2, the potential along the double-layer interface is constant, while it is not the case when devices with a three-electrode configuration are discussed. As illustrated in Fig. 2d, a typical example is found in field-effect transistors, where the three electrodes are Source, Drain, and Gate electrodes. In this system, the potential varies along the interface, depending on gate-source voltage, Vgs, and drain-source voltage, Vds. What kind of potential distribution should be postulated? To answer this question is difficult, but a constant-slope potential distribution along the interface would be the most simple (gradual channel approximation) and most probable one. In other words, the electric field along the interface is constant, and given by ðVds Vgs Þ=L. In that situation, the potential at x ¼ x from position of the source electrode edge is described as VðxÞ ¼ Vgs þ
x Vds Vgs : L
(6)
As the charge induced on the interface is given by Cg ðVðxÞ Vth Þ (Cg ¼ eg/d: gate capacitance per unit area, Vth: threshold voltage, eg is the dielectric constant of gate insulator, and d is the thickness of gate insulator), the charge induced at the interface between Source and Drain electrodes is given by Qint ¼ Cg Vgs Vds =2 Vth
hW V ds m h L W ¼ m Cg ðVgs Vth ÞVds Vds 2 =2 L
M
Ids ¼ Qint
(8)
Here L/W and h are the channel length/width and thickness of the channel, respectively. Equation 8 is nothing but the equation of drain-source current of field-effect transistors, which is derived under gradual channel approximation on the basis of general semiconductor device physics. Nevertheless, it should be noted that Eq. 8 is derived on the basis of classical electrostatic analysis with consideration of the Maxwell–Wagner effect. In other words, charge accumulation at semiconductor–insulator interface is a result of the Maxwell–Wagner effect. Note that in the case of constant current flowing across the FET channel, the potential distribution such as [5] rffiffiffiffiffiffiffiffiffiffiffi x VðxÞ ¼ V0 1 þ Vth L
(9)
is obtained. Here V0 is the potential below the source electrode with respect to the gate potential. Hence, the electric field along the channel is no longer constant, and dependent on the position as V0 EðxÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : 2 LðL xÞ
(10)
In this case, the drain current is described as (7)
per unit area. The accumulated charge given by Eq. 7 is of course the result of the Maxwell–Wagner effect at the gate insulator–active layer interface. In general FETs, the active layer is very conducting in comparison with the gate insulator; the relation tg >> tact thus holds. Here tg and tact are dielectric relaxation times of gate insulator and the active layer. The situation of charge accumulation is very similar to the case expressed by Eq. 3 with tg ¼ tins and tact ¼ tm . In other words, the electric field in gate insulator is much higher than the field in the active layer. The drain-source current Ids is the flow of charge Qint at a constant velocity of vð¼ mEds Þ under the average electric field along the channel between Drain and Source electrodes, Eds ¼ Vds =L. Consequently, the Ids is written as [2, 3]
Ids ¼ m
W 2 Cg ðVgs Vth ÞVds Vds 2 L 3
(11)
The deviation of the current Eq. 11 from Eq. 8 is small, but the positional dependence of the electric fields is rather different between them. Hence, it is important to measure the electric field distribution along the FET channel. As has been described above, the Maxwell–Wagner effect is actively working in field-effect transistors.
Maxwell–Wagner Effect as Interfacial Polarization Phenomena As mentioned in section Metal-Double-Layer-Metal System (2-Electrode System), charge ss is accumulated at the interface between two materials with
M
M
1280
Maxwell–Wagner Effect
different relaxation times, t1 6¼ t2 (see Eq. 2). The interfacial charge accumulation process is a macroscopic dielectric polarization process, and it is called as interfacial polarization process. This process is a single polarization process, and the charge accumulation at the interface of the double-layer system illustrated in Fig. 2c is written as ss ðtÞ ¼ ss ð1 expðt=tMW ÞÞ tMW ¼
with
C1 þ C2 G1 þ G2
(12)
Here C1 ¼ e1 S=L1 , C2 ¼ e2 S=L2 , G1 ¼ s1 S=L1 and G2 ¼ s2 S=L2 . tMW is Maxwell’s dielectric relaxation time and it differs from relaxation times of the two materials, t1 and t2 . It depends on not only material properties, dielectric constant e and conductivity s, but also on geometrical factors, film thickness L and working area S. This is the characteristic of the interfacial charging caused by the Maxwell–Wagner effect.
Probing of Interfacial Charge Caused by Maxwell–Wagner Effect As described in section Metal-Double-Layer-Metal System (2-Electrode System), accumulated charge at the interface is a source of space charge field. This results in the modification of the electric field distribution in a double-layer device in accordance with amount of accumulated charge at the interface (see Eqs. 1 and 2). This scenario is also applicable to the case of any other device systems with material–material interfaces. In that situation, probing the electric field in devices is a very important research subject. Many techniques for probing the electric field distribution have been developed. Among them are electrical measurements, i.e., charge decay method, thermally stimulated current method, and so on. However, these techniques are indirect ones and cannot directly probe carrier process. On the other hand, optical method based on optical second harmonic generation is available for directly probing carrier process related to the interfacial charging caused by the Maxwell–Wagner effect [5, 6, 7, 8]. Figure 3 portrays the experimental setup of the optical second harmonic generation method using a light source of YAG laser [8]. The attached CCD camera visualizes carrier motion in active layer of the
FET device [9], and the PMT probes the intensity of generated SHG signals. Using this technique, the electric field distributions in electronic devices are obtained, e.g., organic fieldeffect transistors, electroluminescence devices, solar cells, and others [10, 11]. As charges injected from electrodes are excess charges they act as a source of electric field. As a result, materials accepting these excess charges would be polarized. In other words, electron clouds of molecules are distorted in the presence of a space charge field arisen from injected carriers. In that situation, by probing this polarization, carrier motion in devices will be probed. The polarization to be probed is second-order nonlinear dielectric polarization that is generated in materials, due to coupling of electron clouds and electromagnetic waves. This is the electric field–induced nonlinear polarization, and this polarization is probed in the electric field–induced optical second harmonic generation (EFISHG). In more detail, EFISHG is generated due to the coupling of electrons in molecules and electromagnetic waves E(o) in the presence of electrostatic local electric field E(0). Hence the EFISHG generation is allowed from even symmetric molecular systems, such as pentacene, tetracene, phthalocyanine, etc. The EFISHG intensity is given as [10, 11, 12, 13] Ið2oÞ / jPð2oÞj2 2 . ¼ e0 wð3Þ ð0; o; oÞ .. Eð0ÞEðoÞEðoÞ ;
(13)
where P(2o) is the nonlinear polarization induced in the materials, e0 is the dielectric permittivity of vacuum, w(3) is the third order nonlinear susceptibility tensor. E(0) is the electrostatic local field in materials, and E(o) is the electric field of incident electromagnetic wave. Generally, laser beam is used as a source of E(o). The local electric field is sum of the external electric field E0 and other fields such as space charge field, inner electric field due to workfunction difference of electrodes, and so on. Hence, carrier motion is probed as propagation of nonlinear polarization in the EFISHG measurements. In the following, some examples are shown. Double-Layer Devices [14, 15, 16, 17] Figure 4 shows transient EFISHGs generated from double-layer devices by applying a step-voltage,
Maxwell–Wagner Effect Maxwell–Wagner Effect, Fig. 3 Optical setup for the SHG measurement. The sample has the FET structure
M
1281
SiO2 (500 nm)
YAG laser + 3ω generator
OPO Pentacene (100 nm)
ND Polarizer SH-cut
D
IR-cut
Bandpass
50 μm
G
S FET
i.e., ITO/Polyimide/a-NPD/Au (Fig. 4a), ITO/ Polyimide/pentacene/Au (Fig. 4b), and ITO/a-NPD/ Alq3/LiF/Al (Fig. 4c). Here pentacene and a-NPD are hole-transport layers, and Alq3 is an electron-transport layer, and polyimide is a blocking layer. The delay time (tdelay) in Fig. 4 represents the time interval between fundamental laser beam used as incident electromagnetic wave E(o), and the wave front of the applied square voltage. In the SHG experiments, the SHG wavelength was chosen so that the electric field in a-NPD and pentacene layers was probed [14, 15, 16, 17]. The black dash lines represent baselines for each device (“zero level”), with no applied voltage. At the beginning t < 107s, the three devices show similar transient characteristics. The SH intensity begins to increase at around t 107 s and then reaches some saturated value. This response time is independent of applied voltage, and reflects charging to the electrodes. This response time of tRC 107 s corresponds well with the response of the device circuit RsC* (see section Metal-Double-LayerMetal System (2-Electrode System) for C*). By contrast, the transient SHGs in the region t > 107s depends on the device and applied voltage. Figure 4a shows the transient SHG of ITO/ Polyimide/a-NPD/Au. Here polyimide layer is a blocking layer for hole and electron injection, and only hole injection from the Au electrode to the a-NPD layer is allowed. However, in actual experiments, after the charging of the electrodes, the SH intensity saturates and never changes. The results suggest that there is no charge accumulation at the interface, i.e., Qs ¼ 0 (see Eq. 1). In other words, the a-NPD layer works like
PMT Objective CCD
Polarizer
an insulating blocking layer. Figure 4b shows the transient SHG of ITO/Polyimide/pentacene/Au devices. Here polyimide is a blocking layer. In this case, hole injection from the Au electrode is expected. Actually, after charging of the electrodes, rapid relaxation of SH intensity to zero level is significant. The decrease of the SHG intensity indicates charge accumulation at the pentacene/PI interface, i.e., Qs 6¼ 0, as a result of the MW effect. Moreover, accumulated holes completely compensate the external electric field in the limit t ! 1 (see Eqs. 1 and 2). Note that the relaxation time given by Eq. 12 satisfies the relation tMW tRC (107 s) in this experiment. Figure 4c shows the transient SHG of ITO/a-NPD/Alq3/LiF/Al device. This device is a typical two-carrier transportation OLED device with hole and electron injected from opposite electrodes. After the charging of the electrodes, these carriers are conveyed into a-NPD and Alq3 layers from opposite sides, and reach at the a-NPD/Alq3 interface, where they recombine for electroluminescence. Nevertheless, from viewpoints of Maxwell–Wagner effect, electric field in this device is modified as excess charges remained at the interface. Observed SHG decaying after its saturation was due to this MW effect. In this case, the experiment shows that the relation tMW >> tRC is satisfied. Interestingly, the electroluminescence was generated at the very time when the SHG decay was started. OFET Devices [18, 19, 20] As described in section FET System (3-Electrode System), charge is accumulated at the interface. For
M
M
1282
Maxwell–Wagner Effect
In Steady State
t delay
Vext
Charging 50 ms
0
2
b
3
I (2ω) [arb. unit]
I (2ω) [arb. unit]
a tRC 1
2V 6V
1
2
tMW 2
3V 8V
1 0
tRC 2
I (2ω) [arb. unit]
c 20 10 V 10 tRC 3
5V tMW 3
0 10–9 10–8 10–7 10–6 10–5 10–4 10–3 10–2 10–1 Time [s]
Maxwell–Wagner Effect, Fig. 4 Transient SHG response from double-layer devices with applying a step-voltage: (a) aNPD/polyimide device, (b) pentacene/polyimide device, and (c) a-NPD/Alq3 device
double-layer devices illustrated in Fig. 2, it is easy to expect charge is homogeneously distributed at the interface, and as a result, constant potential is built at the interface. On the other hand, at the interface of threeelectrode devices such as FET, the distribution is dependent on configuration structure and current flowing through the system, though the Maxwell–Wagner effect suggests that charge must be accumulated at the interface. Hence, it is a fundamental task to determine experimentally charge distribution along the channel. Probing electric field distribution along the channel is of great help and EFISHG measurement is available (see Eq. 13). Interest topics here are the charge distribution in steady state and charge accumulation process along the channel. In the following, these two points are focused.
As is well known, Laplace fields are theoretically calculated if the electrode configuration and its potential distributions are known a priori. Figure 5 shows the SHG experiment employed for OFETs to determine the electric field distribution along the channel. Figure 5a and b shows the intensity profile of the fundamental laser at a focal point and the field distribution calculated using the finite element method on account of the actual electrode configuration. The separation between source and drain electrodes is L ¼ 50 mm. At off-state of the device, i.e., without no injected carriers along the channel, the potentials of the source and gate electrodes are the same, whereas potential difference between drain and gate (source) electrodes is 100 V. In this case, a Laplace field is formed in the device system, and the electric field is concentrated at the edge of drain electrodes due to edge effects. Figure 5c shows the SHG profile estimated from the convolution of the Laplace field and the intensity profile of the fundamental laser. Open circles represent the observed SHG intensity at each spot position. The theoretical curve reproduces well the experimental results, indicating actual electric field distribution in FETs is evaluated by carrying the deconvolution of the SHG profile. For simplicity, assuming the Laser beam profile as delta-function, the electric field distribution is roughly evaluated. Figure 6 shows the evaluated electric field distribution in the pentacene FET channel. At first glance, electric field is concentrated at the drain edge in the off-state (open squares). On the other hand, electric field distribution profile is getting flat in the onstate (filled squares), and it is approaching a constant electric field distribution (Eq. 6, gradual channel approximation). Transient Charge Accumulation Process [18, 19, 20]
As mentioned in section “Probing of Interfacial Change Caused by Maxwell–Wagner Effect,” using a Transient -SHG technique can directly probe carrier motion in organic materials by monitoring the propagation of induced polarization along with carrier motion. According to the Maxwell–Wagner effect, at the interface between two materials with different relaxation times, the charge Qs is accumulated. In OFETs, carriers injected from the source electrode are accumulated at the interface between activeorganic layer and gate insulator, and they are conveyed along the channel. Note that the injected
Maxwell–Wagner Effect
1
c 1.5 D
0 –40
–20
0
20
40
Position [μm]
b
drain edge
SH intensity [arb. unit]
Laser intensity [arb. Unit]
a
E [MV/cm]
Maxwell–Wagner Effect, Fig. 5 SHG experiment (a) Intensity profile of the fundamental laser at a focal point, (b) theoretical field distribution calculated using the finite element method, and (c) SHG profile estimated using (a) and (b). Open circles represent the observed SHG intensity at off-state of OFET device with applying Vdg ¼ Vsg ¼ 100 V
M
1283
S
1
0.5
1 0 0
10 20 30 40 50 60 70 80 90 Position [μm]
0
0
30
60
90
Position [μm]
6 × 104
Electric field [V/cm]
off-state on-state 4 × 104
2 × 104
0
0
10 20 30 40 Distance from drain edge [μm]
50
Maxwell–Wagner Effect, Fig. 6 Evaluated electric field distribution in the pentacene FET channel from SHG intensity; open squares: off-state, closed squares: on state. Dotted line represents the average field calculated as Vds /L, where L is the channel length
carriers are excess charges and they are acting as a source of space charge field Es. Hence actual local electric field E is a sum of the external field E0 and space charge field, i.e., E ¼ E0 + Es. Obviously, E0 is
the Laplace field and it never changes after applying voltages to the source, drain, and gate electrodes, and it is the same with that at time t ¼ 0. On the other hand, Es is the Poisson field and changing with time in accordance with carrier propagation, i.e., carrier injection, accumulation, and transport. Of course, it stabilizes in the limit t ! 1, i.e., in steady state. Carrier migration along the channel is considered as an interfacial charging process and modeled using a ladder RC circuit model. Visualization of carrier migration is of great help to verify the above carrier migration along the OFET channel. In such situation, transient EFISHG technique is available. Substituting E ¼ E0 + Es into E(0) of Eq. 13, it is found that SH intensity distribution along the channel changes with the distribution of space charge field Es originating from moving carriers. For the transient SHG measurement, time resolved microscopic–SHG (TRM-SHG) measurement is useful. This approach allows to directly probe carrier motion. Figure 7 shows the TRM-SHG experiments using top-contact pentacene FET with Au-source and drain electrodes. The samples were prepared on gate insulator surface with a 500 nm of SiO2-gate insulator. The thickness of the pentacene layer was 100 nm. Figure 7a shows the TRM-SHG
M
M
1284
Maxwell–Wagner Effect
a Source edge
D
S
0 ns
D Source edge
300 ns
S
b Source Electric field
Drain
intensity profile plotted in Fig. 7b clearly shows that holes travel in direction from the source Au electrode to the drain electrode, on satisfying the square root time dependence (see Fig. 7c). Results show that carrier transient is an interface charge propagation process, and the square root time dependence verifies the RC ladder-model.
Future Directions
0 ns 100 ns 200 ns 300 ns 0 10 20 30 40 50 Distance from source edge [μm] Distance from source edge [μm]
c
100
Cross-References
10
1 10
The Maxwell–Wagner effect is based on the classical electromagnetic theory, and accounts for charge accumulation phenomena at the interface between two materials. It can widely apply to all kinds of material interfaces, from macro- to nano-materials. Also it can apply at molecular level. Many new materials will be coming out, but physics related to charge accumulation is the same.
100 Time [ns]
1,000
Maxwell–Wagner Effect, Fig. 7 (Color online) TRM-SHG measurement. (a) SHG image from the FET channel for different delay times 0, 300 ns, and with applying positive voltage pulse to the source electrode. The area of the laser spot is indicated by the dashed circle. (b) Plots of SHG intensity profile. (c) Log-log scaled time evolution of the peak position of the SHG intensity
image obtained at various times after a positive voltage pulse of 70 V was applied to the source electrode with the gate and drain electrodes shorted and grounded. At t ¼ 0 ns, the laser pulse coincided with the rising edge of the voltage pulse, and SHG signals were found near the edge of source electrode, indicating that carrier injection was just started, and the Laplace field E0 is formed only around source electrode. Interestingly, as is clearly shown in the image, the emission band of the SHG signal gradually moved in the channel from source to drain electrode with time. Motion of the emission band from source electrode, not from drain, is the direct evidence of hole injection from the Au-source electrode. That is, pentacene FET shows a p-type behavior. The SHG
▶ Electrode–Organic Interface Physics ▶ Field Electron Emission from Nanomaterials ▶ Flexible Electronics ▶ Optical and Electronic Properties
References 1. Maxwell, J.C.: A Treatise on Electricity and Magnetism, vol. 1, 3rd edn. Dover, New York (1954) (Chap. X, p. 450) 2. Tamura, R., Lim, E., Manaka, T., Iwamoto, M.: Analysis of pentacene field effect transistor as a Maxwell-Wagner effect element. J. Appl. Phys. 100, 114515 (2006) 3. Lim, E., Manaka, T., Tamura, R., Iwamoto, M.: MaxwellWagner model analysis for the capacitance-voltage characteristics of pentacene field effect transistor. Jpn. J. Appl. Phys. Part 1 45, 3712 (2006) 4. Lim, E., Manaka, T., Iwamoto, M.: Analysis of carrier injection into a pentacene field effect transistor by optical second harmonic generation measurements. J. Appl. Phys. 101, 024515 (2007) 5. Weis, M., Lin, J., Taguchi, D., Manaka, T., Iwamoto, M.: The charge transport in organic field-effect transistor as an interface charge propagation: The Maxwell-Wagner effect model and transmission line approximation. Jpn. J. Appl. Phys. 49, 071603 (2010) 6. Yamada, D., Manaka, T., Lim, E., Tamura, R., Weis, M., Iwamoto, M.: Probing of electric field in pentacene using
Mechanical Properties of Hierarchical Protein Materials
7.
8.
9.
10.
11.
12. 13.
14.
15.
16.
17.
18.
19.
20.
microscopic optical second harmonic generation. J. Appl. Phys. 103, 084118 (2008) Weis, M., Manaka, T., Iwamoto, M.: Origin of electric field distribution in organic field-effect transistor: Experiment and analysis. J. Appl. Phys. 105, 024505 (2009) Manaka, T., Lim, E., Tamura, R., Iwamoto, M.: Direct imaging of carrier motion in organic transistors by optical second harmonic generation. Nat. Photonics 1, 581 (2007) Lim, E., Yamada, D., Weis, M., Manaka, T., Iwamoto, M.: Probing of channel region in pentacene field effect transistors by optical second harmonic generation. Chem. Phys. Lett. 477, 221 (2009) Iwamoto, M., Manaka, T., Weis, M., Taguchi, D.: Probing and modeling of interfacial carrier motion in organic devices by optical second harmonic generation. J. Vac. Sci. Technol. B 28(4), C5F12 (2010) Iwamoto, M., Manaka, T., Yamamoto, T., Lim, E.: Probing motion of electric dipoles and carriers in organic monolayers by Maxwell displacement current and optical second harmonic generation. Thin Solid Films 517, 1312 (2008) Shen, Y.R.: The Principles of Nonlinear Optics. Wiley, New York (1984) (Chap. 1) Manaka, T., Lim, E., Tamura, R., Yamada, D., Iwamoto, M.: Probing of the electric field distribution in organic field effect transistor channel by microscopic second-harmonic generation. Appl. Phys. Lett. 89, 072113 (2006) Shibata, Y., Nakao, M., Manaka, T., Lim, E., Iwamoto, M.: Probing electric field distribution in underlayer of an organic double-layer system by optical second-harmonic generation measurement. Jpn. J. Appl. Phys. 48, 021504 (2009) Zhang, L., Taguchi, D., Li, J., Manaka, T., Iwamoto, M.: Probing of interfacial charging and discharging in doublelayer devices with a polyimide blocking layer by timeresolved optical second harmonic generation. J. Appl. Phys. 108, 093707 (2010) Taguchi, D., Inoue, S., Zhang, L., Li, J., Weis, M., Manaka, T., Iwamoto, M.: Analysis of organic light-emitting diode as a Maxwell−Wagner effect element by time-resolved optical second harmonic generation measurement. J. Phys. Chem. Lett. 1, 803 (2010) Taguchi, D., Zhang, L., Li, J., Weis, M., Manaka, T., Iwamoto, M.: Analysis of carrier transients in double-layer organic light emitting diodes by electric-field-induced second-harmonic generation measurement. J. Phys. Chem. C 114, 15136 (2010) Manaka, T., Liu, F., Weis, M., Iwamoto, M.: Diffusion like electric-field migration in the channel of organic field-effect transistors. Phys. Rev. B 78, 121302R (2008) Manaka, T., Liu, F., Weis, M., Iwamoto, M.: Influence of traps on transient electric field and mobility evaluation in organic field-effect transistors. J. Appl. Phys. 107, 043712 (2010) Nakao, M., Manaka, T., Weis, M., Lim, E., Iwamoto, M.: Probing carrier injection into pentacene field effect transistor by time-resolved microscopic optical second harmonic generation measurement. J. Appl. Phys. 106, 014511 (2009)
1285
M
Mean Field Approaches to Many-Body Systems ▶ Ab Initio DFT Simulations of Nanostructures
Mechanical Behavior of Nanostructured Metals ▶ Mechanical Properties of Nanocrystalline Metals
Mechanical Characterization of Nanostructures ▶ Reliability of Nanostructures
Mechanical Properties of Biological Materials ▶ Mechanical Properties of Hierarchical Protein Materials
Mechanical Properties of Hierarchical Protein Materials Markus J. Buehler1 and Graham Bratzel1,2 1 Laboratory for Atomistic and Molecular Mechanics, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA 2 Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
Synonyms Biomolecular mechanics; Hierarchical mechanics; Materiomics; Mechanical properties of biological materials
M
M
1286
Mechanical Properties of Hierarchical Protein Materials
Definition
Whether natural composite, polymer, or elastomer, many biological materials achieve their notable mechanical properties through a balance of strong covalent bonds and weaker noncovalent bonds (e.g., hydrogen bonds) in a multilevel hierarchical structure that spans from nanoscopic to macroscopic length scales [11, 12]. Known as the universality-diversity paradigm [13] and summarized in Fig. 1, a limited number of universal protein structures or mineral components can provide a wide diversity of geometrical arrangements within the natural composites that give rise to exceptional bulk mechanical properties as compared to the material properties of the individual constituents. Certain specially constructed geometrical arrangements can enhance bulk properties much more than the rule of mixtures can predict. This puts the concept of geometry, assembly, and, more generally, the architecture of hierarchical structures at the forefront of the science of protein materials. This entry will first discuss how protein materials are limited by an intrinsic strength limit at the nanoscale that exemplifies the constraints under which many biological systems operate, how these are overcome, and how such materials exemplify a powerful paradigm of how weakness is turned into strength. Next, the multilevel structures of silk, collagen fibers, and bone are presented as specific examples of how specialized geometrical arrangement of respective protein and mineral components into a hierarchical structure can overcome the strength limit at the base level to yield highly functional macroscale materials that rival engineering composite materials.
Mechanical properties of proteins define the behavior of the deformation and failure of proteins and proteinbased biological and synthetic materials.
Overview With only 20 standard amino acids as universal building blocks, evolutionary adaptation has resulted in a multitude of polypeptides and protein structures with a wide range of properties and applications [1]. A fundamental application of protein materials is mechanical, providing static and dynamic support and defenses for the organisms they comprise. To achieve specialized mechanical properties, animal tissue shows a wide range of complex, structured composites composed of proteins (e.g., collagen, keratin, chitin) and inorganic minerals (e.g., calcite, hydroxyapatite, and aragonite) [2, 3]. Instead of summarizing primarily single-molecule protein mechanics studies [4–8], the main goal of this entry is to give an introduction to the current knowledge of the nanomechanical properties of important structural protein materials and to explain the generic role of the hierarchical structures that give rise to exceptional mechanical properties that are superior to many synthetic engineering materials. For example, shell, bone, and antler are hierarchically structured composites of collagen and minerals and have toughnesses that are about one order of magnitude greater than engineering ceramics [9]. Natural polymers and polymer composites like collagen, keratin, and silk show elastic moduli and tensile strengths larger than those of engineering polymers made from similar building blocks. For example, silks have elastic moduli between 2 and 20 GPa and strengths in the range from 0.3 to 2 GPa [9]. Of synthetic polymers, only Kevlar has a higher stiffness at 200 GPa and a strength up to 4 GPa, which it achieves through a highly oriented molecular structure which is, in fact, characteristic of biological materials. Natural elastomers, including skin, muscle, cartilage, abductin, resilin, and elastin, exhibit elastic moduli and densities similar to those of engineering elastomers. Muscle is unique in combining a low modulus on the order of an elastomer with a high strength on the order of steel [9]. Natural cellular materials like cancellous bone have low densities because of the high volume fractions of voids they contain [10].
Hierarchical Structures Overcome Strength Limits While covalent peptide bonds make up the backbone of a protein, noncovalent bonds, in particular weak bonds such as hydrogen bonds (H-bonds), are indispensable to biological function in defining the native folding and stability of protein structures. The amino acid sequence defines a protein’s secondary and higher-level structures via folding, but arrays of parallel adjacent H-bonds govern major universal secondary and tertiary structural motifs, chiefly a-helices and b-sheets, that organize the three-dimensional structure of polypeptides and thus determine their mechanical properties at a fundamental level. For example, the
Mechanical Properties of Hierarchical Protein Materials
1287
M
M
Mechanical Properties of Hierarchical Protein Materials, Fig. 1 Overview over the universality-diversity paradigm (UDP) [11, 13]. The hierarchical structure of protein materials is analogous to the construction of symphonic music. In much the same way that all music is made from a superposition of “simple” sound waves, a limited list of universal building
blocks – here, amino acids of proteins – can be combined in various ways at multiple hierarchical levels to yield a diverse groups of protein-based biomaterials. Only a small subset of this diverse group exhibit exceptional mechanical properties, e.g., an orb-weaver spider web or bone, akin to Bach or Mozart
rupture of parallel interstrand H-bonds during b-sheet failure defines the true ultimate strength of such proteins (Fig. 2a for a schematic). Atomistic simulation combined with single-molecule force microscopy studies [7] have shown that proteins rich in b-sheets exhibit higher rupture forces since they employ parallel strands with numerous H-bonds that act as strong
mechanical clamps under shear loading. At experimental pulling rates, most mechanical proteins exhibit a rupture force of a few hundred piconewtons (pN) [14], as seen in Fig. 2. An asymptotic limit at very low pulling rates can be inferred from a limit analysis for deformation through equilibrium, which suggests an upper strength limit for the unfolding b-sheet-rich
M
1288
Mechanical Properties of Hierarchical Protein Materials, Fig. 2 Strength limit of H-bond clusters under shear loading (see panel (a) for schematic of geometry before and after rupture). As shown in panel (b), at vanishing pulling rates (see inset for a detailed view of the data), clusters of H-bonds resist a force limit of 100–300 pN. At high pulling rates, considerably more force is needed to unfold and rupture H-bonds, according to experimental and computational results from fibronectin, immunoglobin, and titin deformation (for references to source data, see [18]). This data reflects the rate dependence of failure seen in many biomaterials [18]
proteins [15–17] (see results in Fig. 2b). Based on this analysis for typical single-molecule length scales at a temperature of 300 K and a vanishing pulling rate, the asymptotic force of rupture is found to be on the order of 100–300 pN [18] (Fig. 2b). This asymptotic force limit depends strongly on the H-bond dissociation energy, which can vary with solvent conditions and which explains the range of strengths obtained. The asymptotic strength limit for the unfolding b-sheet-rich proteins can be explained based on fracture analysis inspired by Griffith’s analysis that was originally applied to solids, and is here used to describe the strength of protein domains [18]. We briefly review the key concepts behind this analysis that is also shown in Fig. 3. At low force levels, the elasticity of the polypeptide chain is primarily due to entropic effects (rather than energetic effects). The Marko-Siggia Wormlike Chain (WLC) model is one of the most widely used expressions to predict the entropic elasticity of the polypeptide backbone. Indeed, comparative studies provide substantial
Mechanical Properties of Hierarchical Protein Materials
evidence that WLC theory is a good model for the behavior of individual, unconstrained polypeptide chains [19]. In the WLC theory, the force is a function of end-to-end chain length, the persistence length, the contour length, and the temperature. The asymptotic strength limit for the unfolding b-sheetrich proteins can be obtained by considering the energy dissipation during the process of first stretching of a b-strand protein (described by the WLC theory) followed by subsequent breaking of H-bonds, where the energy released must equal the energy stored in the H-bonds (Fig. 3a); where the energy released is properly normalized to describe how many H-bonds exist per unit contour length of the protein. This analysis is similar to a fracture analysis of solids where the elastic energy stored in the material (here often described using linear elastic continuum theory and not by the WLC theory as in the case of proteins) is being dissipated into the creation of new surfaces. For a given H-bond dissociation energy the force limit is independent of the number of H-bonds being sheared simultaneously, given that the number of H-bonds in a deformed cluster is larger than a critical H-bond cluster size Ncr, which can be calculated for each case [18, 20]. This is because it is found that no more than Ncr ¼ 3–4 H-bonds can rupture concurrently, thus defining a critical H-bond cluster size beyond which the strength can no longer increase and at which scale the asymptotic strength limit is reached. Therefore, while the shear strength of b-sheets with less than Ncr ¼ 3–4 H-bonds benefits from addition of H-bonds, the shear strength (the maximum force divided by the sheared area) decays for b-sheets of 4 H-bonds or longer, as seen in b, as the H-bonds along the full length of the b-sheets do not rupture concurrently since they can only rupture in groups of Ncr ¼ 3–4 H-bonds (see Fig. 3c for an illustration of this concept). Notably, following this scaling law, most protein materials are indeed found to intersperse many small b-sheets close in size to Ncr, in order to turn the weakness of the building block (H-bond) into strength by using geometry as a paradigm to achieve this goal. For example, b-sheets show a typical H-bond cluster size of 3–8 H-bonds across thousands of proteins in the Protein Data Bank, and a-helices have 3–4 H-bonds per turn [18, 20]. It is interesting to note that the scaling law of the shear strength, increasing for small values of H-bonds to a maximum value followed by a decay represents a biological material analogue of
Mechanical Properties of Hierarchical Protein Materials
1289
M
Mechanical Properties of Hierarchical Protein Materials, Fig. 3 Mechanistic view into the H-bond rupture process (a), and scaling of shear strength over size of H-bond cluster (b). (a), The area enclosed between loading and unloading curves is equivalent to the change in free energy, i.e., the adhesion energy per unit length. The increase in contour length is due to fracture, which involves breaking of H-bonds and as a result, the release of previously attached polymer length that now contributes to entropic elasticity. (b) A b-sheet of 3–4 residues in length
displays the theoretical maximum shear strength, meaning that up to 3–4 H-bonds can rupture simultaneously. The structure of H-bond clusters in a-helices are in this range, with 3.6 residues per turn; and similarly, most b-sheet protein domains feature 3–8 H-bonds in each cluster. (c) Illustration of the physical concept behind the scaling law shown in panel (b), where the breaking of H-bond clusters is confined to a critical cluster size of Ncr ¼ 3–4 (Figures adapted from [20])
the Hall–Petch relationship known from polycrystalline engineering materials (e.g., metals) [21]. The intrinsic strength limit can only be overcome and scaled up by using structural hierarchies that circumvent b-sheet saturation and other limiting factors at critical length scales. Silk, as will be discussed in the following section, is a prime example of how H-bond cooperativity within b-sheets of controlled lengths and a hierarchical structure of networks and fibrils can overcome molecular strength limits and create unusually strong and tough protein fibers at the macroscale.
pressures, and with water as the solvent. However, biomimetic reproductions of silk remain a challenge because of silk’s unique microstructural features that can only be achieved by controlled self-assembly of protein oligomers with molecular precision [25, 26]. Unlike silkworms, some spiders can use different glands to create up to seven types of silk, from the strong structural dragline to the viscoeleastic capture silk and tough egg sack casing [23]. Dragline silk, containing a high fraction of densely H-bonded domains, is used to provide the structural frame for the web and has an elastic modulus of around 10 GPa [24]. Capture silk, on the other hand, is a viscid biofilament containing cross-linked polymer networks and has an elastic modulus that is comparable to that of other elastomers [27]. The stress–strain behavior of silks is compared to KevlarTM and rubber in Fig. 4b. While rubber is extensible, and KevlarTM is stiff and strong, silks feature a combination of strength and toughness not typically found in synthetic materials. It is known that b-sheet crystals at the nanoscale play a key role in defining the mechanical properties of silk by providing stiff and
Structure and Mechanics of Exemplary Hierarchical Protein Materials Silk Silk is a hierarchically structured protein fiber (Fig. 4a) with a high tensile strength and great extensibility, making it one of the toughest and most versatile materials known [22–24]. In contrast to synthetic polymers based on petrochemicals, silk is spun into strong and totally recyclable fibers at ambient temperatures, low
M
M
1290
Mechanical Properties of Hierarchical Protein Materials
Mechanical Properties of Hierarchical Protein Materials, Fig. 4 Hierarchical structure of silk from nano to macro (a) and a summary of stress–strain plots superior strain-hardening behavior of dragline and silkworm silks (b). Rubber is extensible but not strong. KevlarTM is stiff and strong, but not tough. Silk, especially dragline silk, behave with a combination of strength and toughness not typically found in synthetic materials (Figure adapted from [31])
orderly cross-linking domains embedded in a semiamorphous matrix that consists predominantly of less orderly structures [28, 29]. These b-sheet nanocrystals, bonded by means of assemblies of H-bonds, have dimensions of a few nanometers and constitute roughly 10–15% of the silk volume. When silk fibers are stretched, the b-sheet nanocrystals reinforce the partially extended and oriented macromolecular chains by forming interlocking regions that transfer the load between chains under lateral loading, similar to their function in other mechanical proteins [30]. The hierarchical network of spider draglines, contrasting synthetic elastomers like rubber, enables quick energy absorption and efficiently suppresses vibration during an impact [31]. Rubbers, composed of random polymer chains, display an elasticity primarily due to the change in conformational entropy of these chains. In contrast, the amorphous chains in silk filaments are extended and held in partial alignment with respect to the fiber axis in its natural state, resulting in remarkably different mechanical behaviors from rubbers [29, 30]. Mesoscale simulations reveal that the
characteristic, cross-linking b-sheet nanocrystals govern the coupled effects of silk network realignment and strain-hardening behavior beyond the regime of unfolding amorphous chains [28]. Advances in nanotechnology may allow for an optimization of silk and silk-like synthetic fibers for specific applications by modifying the governing network at critical length scales. At the base of the structural hierarchy, b-sheet nanocrystal size may be tuned though genetic modification and control of the selfassembly process. Computational structure predictions of MaSp1 agglomeration in dragline silk, seen in Fig. 5a, shows that the size of the b-sheet nanocrystal may indeed be tuned by controlling the poly-alanine length [32], which in turn affects the stability and “quality” of the resulting nanostructures. A critical poly-alanine length of 3–4 residues is predicted for the amyloidization of a defined and stable b-sheet nanocrystal. A sampling of amino acid sequences of both orb-weaving and non-araneoid spiders shows that the MaSp1 and MaSp2 proteins of strong and tough dragline silk typically have poly-alanine repeats 6–8
Mechanical Properties of Hierarchical Protein Materials
1291
M
Mechanical Properties of Hierarchical Protein Materials, Fig. 5 Tuning the structure of spider silk by controlling the genetic makeup of silk proteins, used to define the structure of the material at the nanoscale (obtained using Replica Exchange Molecular Dynamics). (a) Computational structure predictions of MaSp1 agglomeration in dragline silk show that the size of the b-sheet nanocrystal may be tuned by changing the poly-alanine length [32]. (b) Percentage of b-sheet nanocrystal formation in
the poly-alanine region of the silk protein, measuring the success of the genetic signal to achieve the desired structure. The structure prediction indicates a critical poly-alanine length of 3–4 alanine residues for the establishment of a well-defined nanocrystal. The MaSp1 and MaSp2 proteins of dragline silk typically have poly-alanine repeats 6–8 residues in length, while the MiSp protein of capture silk has poly-alanine repeats of only 2–4 residues [33]
residues in length, while the MiSp protein of elastic capture silk has poly-alanine repeats of only 2–4 residues [33]. Combined with the structure predictions, this shows that the size of the b-sheet nanocrystals has a fundamental effect on the mechanical properties of the macroscopic silk threads (Fig. 5b). Further computational studies of spider silk proteins reveal that the nanoscale confinement of b-sheet nanocrystals in silks indeed has a fundamental role in achieving great stiffness, resilience, and fracture toughness at the molecular level of the structural hierarchy [28, 34], where H-bond cooperativity depends quite strongly on the size of the crystals and breaks down once b-sheet nanocrystals exceed a critical size. Larger b-sheet nanocrystals are softer and fail catastrophically at much lower forces owing to crack-like flaw formation. This catastrophic breakdown leads to rapid disintegration of silk fibers, whereas reducing the nanocrystal dimensions below 3 nm increases the ultimate strength and the modulus manyfold [34]. Furthermore, noncovalent H-bonds are able to reform during stick–slip deformation, allowing smaller crystals a self-healing ability until complete rupture occurs. It is also noted that at the macroscale, experiments have demonstrated that when the size of b-sheet nanocrystals is reduced by moderating the reeling speed or by metal infiltration, silk shows enhanced toughness and greater ultimate strength, exceeding
that of steel and other engineered materials [30]. By understanding and controlling how H-bonded silk threads can reach great strength and toughness and overcome strength limitations at the molecular scale, the behavior of a broader class of b-sheet-rich protein materials and similar synthetic composite fibers can also be tuned and adapted. Collagen-Based Protein Materials The concepts discussed above are important for many other protein materials. Collagen, for example, is present in various forms in almost all structural animal tissue and represents how the limitations of protein subunits can be overcome by the complex way they are assembled (Fig. 6a). Skin, tendon, and cartilage are all largely collagenous composites [2]. In skin, collagen is sandwiched between a basement membrane and an overlying keratinized epidermis. In tendon, collagen fibers form rope-like structures which make up 70–80% of the volume, with the remainder being a combination of non-collagenous protein, polysaccharides, and inorganic salts. In cartilage, the collagen fibers are in a proteoglycan matrix with a small fraction of elastin fibers. Bone, shell, and antler are composites of calcite, hydroxyapatite, or aragonite platelets dispersed in a helical matrix of collagen. A diverse class of materials result from the universal collagen protein.
M
M
1292
Mechanical Properties of Hierarchical Protein Materials
a
Amino acids
Collagen molecule
b
Collagen fibril
Tissue (tendon, bone)
Collagen molecule 10
Young’s modulus (GPa)
Collagen fiber
9
8 6.17 6
5.1
4
4.55
3
3
Collagen fibril
2
0.43 0.5
0.86 0.75
D M
S
M
EM M
AF
D
XS SA
M
p al
ic pt O
ic pt
tra
p
g al
tra
rin te O
ht
sc
at
te lig
lig
ht
sc
at
X-
R
rin
ay
g
0
Type of analysis
Mechanical Properties of Hierarchical Protein Materials, Fig. 6 Structure (a) and scale-dependent mechanical properties of collagen at the scale of single molecules and fibrils. (a) Hierarchical structure of collagenous tissues. The specific scales highlighted here are the collagen molecule and the collagen fibril. (b) A summary of small-strain Young’s moduli for
a single collagen molecule and a collagen fibril shows that the structure of fibrils sacrifice stiffness for increased extensibility and toughness (data taken from various experimental and computational sources, for original references please see [40–44]). The data clearly demonstrates the importance of scaledependent properties in biological protein materials
Variations in the polypeptide sequence of collagen chains gives rise to different types, termed Type I through Type XIX. Type I and Type III collagen are the most abundant, with Type I in skin and bone and Type III in blood vessel walls [2]. Three collagen chains, each of about 1,000 amino acids, arrange in the basic subunit of functional collagen, a left-handed triple-helix 300 nm long and 1.5 nm in diameter called tropocollagen. Typically, tropocollagen aggregates into a microfibril of five triple helices. About 1,000 microfibrils aggregate laterally and end to end to form a collagen fibril that displays periodic banding every 65 nm. Comparing the Young’s moduli of tropocollagen and a collagen fibril, seen in Fig. 6b, shows that the structures at distinct hierarchical levels have very different mechanical properties, where single collagen molecules are much stiffer than collagen fibrils. About 500 fibrils aggregate into a collagen fiber, which then aggregate into fiber bundles. The diameters and specific hierarchical structures of fibrils and fibers vary by function among skin, tendon, and muscle.
A bundled-fiber architecture connected by solventmediated intermolecular and intramolecular crosslinking of aldehyde groups provides stability, strength, and viscoelasticity [35, 36]. Beyond collagen fibers, a broad range of materials are available with the addition of mineral platelets within collagen. Bone and nacre belong to a family of natural ceramic composites that also includes coral, antler, and dental enamel [37, 38] that are made of mineral particles (e.g., hydroxyapatite, calcite, or aragonite) in a matrix of collagen. Their elastic moduli are lower than those of engineering ceramics, but their tensile strengths are similar, and their toughnesses are greater by a factor of ten due to an ordered, hierarchical structure [9]. The hierarchical structure of bone is outlined in Fig. 7 as an example. Bone must likewise survive cyclic compressive loading, bending, and impulses for the lifetime of large and active animals. Bone is suitable for a wide range of loading rates: the failure stress of bulk bone increases by as much as 44% when the stress rate is increased by a factor of 103 [2].
Mechanical Properties of Hierarchical Protein Materials
Mechanical Properties of Hierarchical Protein Materials, Fig. 7 The many hierarchical levels of bone allow this material to reach high levels of strength, toughness, as well as the ability to self-repair. Bone has at least seven hierarchical levels, each with a structure that contributes toward enhancing the mechanical property advantages available at that length scale [38]
In addition, since bone is continuously rebuilt, bones can adapt to creep and changes in local loading after injuries. Large bones like the femur have a structure similar to bamboo: strong and dense cortical bone tissue forms an outer sheath around a porous network of cancellous (i.e., trabecular or spongy) bone tissue with a core of marrow called the medullary cavity, and
1293
M
many reinforcing, longitudinal cortical tubes called osteons. Both cortical and cancellous bone tissue are differently organized structures of the same biomineral composite created by osteoblast cells, which secrete an amorphous protein matrix of predominantly Type I collagen. Over time, mineralization and ion deposition introduce reinforcing globular and plate-like crystals of carbonated hydroxyapatite [2]. There exist two cases of this matrix: woven bone, with a random arrangement of collagen fibers, is formed quickly in fetal bones and after fractures. Over time, woven bone is slowly converted to lamellar bone, made of ordered, concentric, parallel sheets of collagen fibers. Cancellous bone exhibits a wide range of local architecture and density, with volume fractions ranging from 0.05 to 0.6. Several factors affect mechanical property studies of cancellous bone, including donor age, sample hydration, anatomic location, and diseases such as osteoporosis, osteoarthritis, bone cancer, and osteogenesis imperfecta, but typical elastic moduli of cancellous bone are between 0.01 and 2 GPa and ultimate strengths between 0.1 and 30 MPa [39]. The rods and plates that form the individual trabeculae of cancellous bone have a uniform composition similar to cortical bone, but are slightly less mineralized, allowing plasticity during network deformation. Although the anisotropic mechanical properties of cancellous bone depend on the density of the network, the porous structure allows yielding at 1– 2% strain and can sustain large deformations up to nearly 50% strain while maintaining its load-carrying capacity [2], allowing considerable impact energy dissipation before fracture. An analysis of the molecular mechanisms of protein and mineral crystal phases under large deformation of the mineralized collagen fibrils reveal a fibrillar toughening mechanism that leads to a dramatic increase in energy dissipation compared to fibrils without a mineral phase. This fibrillar toughening mechanism increases the resistance to fracture by forming large local yield regions around crack-like defects, a mechanism that protects the integrity of the entire structure by allowing for localized failure that can later be healed. As a consequence, mineralized collagen fibrils are able to tolerate micro-cracks on the order of several hundred micrometers without causing any macroscopic failure of the tissue. At the microscale, mineralized collagen fibrils have a Young’s modulus of 6.23 GPa, yield at a tensile strain of 6.7%, and feature a fracture stress of 0.6 GPa. This shows a dramatic increase over
M
M
1294
unmineralized collagen fibrils, which only show a Young’s modulus of 4.6 GPa and fracture stress of 0.6 GPa. Other deformation and fracture mechanisms exist in bone at larger hierarchical levels, including the breaking of sacrificial bonds between discrete mineralized fibrils, the formation of microcracks ahead of the crack tip, crack bridging by collagen fibrils arranged orthogonal to the crack, and the formation of large plastic zones around crack tips that contribute to an increase in bone’s overall toughness [38].
Conclusion Nature has much to offer for the next generation of both macroscale and nanoscale materials, in particular relating to the linking of vast and seemingly separated scales (from nano to macro). The expansion of the structural design space through the concept of merging structure and material (Fig. 1) is a powerful concept that enables us to create highly functional materials through the use of relatively simple building block, a mechanism established in the universality-diversity paradigm of materials science [11, 13]. The transfer of concepts derived from biological materials, for example, mimicking of the strain-hardening behavior of spider silk, can give rise to the design of advanced functional materials with exceptional kinetic energy buffering and absorption. Developing ordered nanocomposites that are based on the structures found in bone, antler, or nacre can dramatically increase the toughness of typically brittle structures and may enable us to use abundant materials such as silica in the design of tough, elastic, and strong materials. Even the architecture of collagen’s reinforced helices, fibrils, and fibers offer much inspiration in creating advanced tensile struts from seemingly inferior materials. The key issue is to better understand the mechanisms of the origin of the mechanical properties of proteins from the bottom-up, and the identification of systematic approaches to transfer the insight from biology to engineering design through an approach referred to as materiomics [11].
Cross-References ▶ Self-assembly ▶ Spider Silk ▶ Structure and Stability of Protein Materials
Mechanical Properties of Hierarchical Protein Materials
References 1. Meyers, M.A., Chen, P.-Y., Lin, A.Y.-M., Seki, Y.: Biological materials: structure and mechanical properties. Prog. Mater. Sci. 53(1), 1–206 (2008) 2. Black, J., Hastings, G.: Handbook of Biomaterial Properties. Springer, New York (2006) 3. Wegst, U.G.K., Ashby, M.F.: The mechanical efficiency of natural materials. Philos. Mag. 84(21), 2167–2186 (2004) 4. Ackbarow, T., Sen, D., Thaulow, C., Buehler, M.J.: Alphahelical protein networks are self-protective and flaw-tolerant. PLoS ONE 4(6), e6015 (2009) 5. Kai-Nan, A., Yu-Long, S., Zong-Ping, L.: Flexibility of type I collagen and mechanical property of connective tissue. Biorheology 41(3), 239–246 (2004) 6. Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., Walter, P.: Molecular Biology of the Cell. Garland, New York (2002) 7. Gao, M., Sotomayor, M., Villa, E., Lee, E.H., Schulten, K.: Molecular mechanisms of cellular mechanics. Phys. Chem. Chem. Phys. 8(32), 3692–3706 (2006) 8. Gautieri, A., Uzel, S., Vesentini, S., Redaelli, A., Buehler, M.J.: Molecular and Mesoscale mechanisms of osteogenesis imperfecta disease in collagen fibrils. Biophys. J. 97(3), 857–865 (2009) 9. Ashby, M.F., Gibson, L.J., Wegst, U., Olive, R.: The mechanical properties of natural materials. I. Material property charts. Proc. R. Soc. Lond. A Math. Phys. Sci. 450(1938), 123–140 (1995). doi:10.1098/rspa.1995.0075 10. Gibson, L.J., Ashby, M.F.: The mechanics of threedimensional cellular materials. Proc. R. Soc. Lond. A Math. Phys. Sci. 382(1782), 43–59 (1982). doi:10.1098/ rspa.1982.0088 11. Buehler, M.J.: Tu(r)ning weakness to strength. Nano Today 5(5), 379–383 (2010). doi:10.1016/j.nantod.2010.08.001 12. Fratzl, P., Weinkamer, R.: Nature’s hierarchical materials. Prog. Mater. Sci. 52(8), 1263–1334 (2007) 13. Ackbarow, T., Buehler, M.J.: Hierarchical coexistence of universality and diversity controls robustness and multifunctionality in protein materials. J. Comput. Theor. Nanosci. 5(7), 1193–1204 (2008) 14. Isralewitz, B., Gao, M., Schulten, K.: Steered molecular dynamics and mechanical functions of proteins. Curr. Opin. Struct. Biol. 11(2), 224–230 (2001) 15. Sulkowska, J., Cieplak, M.: Mechanical stretching of proteins - a theoretical survey of the protein data bank. J. Phys. Condens. Matter 19(28), 283201 (2007) 16. Sotomayor, M., Schulten, K.: Single-molecule experiments in vitro and in silico. Science 316(5828), 1144–1148 (2007). doi:10.1126/science.1137591 17. Ackbarow, T., Chen, X., Keten, S., Buehler, M.J.: Hierarchies, multiple energy barriers, and robustness govern the fracture mechanics of a-helical and b-sheet protein domains. Proc. Natl Acad. Sci. USA 104(42), 16410– 16415 (2007). doi:10.1073/pnas.0705759104 18. Keten, S., Buehler, M.J.: Asymptotic strength limit of hydrogen-bond assemblies in proteins at vanishing pulling rates. Phys. Rev. Lett. 100(19), 198301 (2008) 19. Fisher, T.E., Oberhauser, A.F., Carrion-Vazquez, M., Marszalek, P.E., Fernandez, J.M.: The study of protein
Mechanical Properties of Nanocrystalline Metals
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35. 36.
mechanics with the atomic force microscope. Trends Biochem. Sci. 24(10), 379–384 (1999) Keten, S., Buehler, M.J.: Geometric confinement governs the rupture strength of H-bond assemblies at a critical length scale. Nano Lett. 8(2), 743–748 (2008). doi:10.1021/nl0731670 Penel, S., Morrison, R.G., Dobson, P.D., Mortishire-Smith, R.J., Doig, A.J.: Length preferences and periodicity in b-strands. Antiparallel edge b-sheets are more likely to finish in non-hydrogen bonded rings. Protein Eng. 16(12), 957–961 (2003). doi:10.1093/protein/gzg147 Hayashi, C.Y., Lewis, R.V.: Evidence from flagelliform silk cDNA for the structural basis of elasticity and modular nature of spider silks. J. Mol. Biol. 275(5), 773–784 (1998) Hu, X., Vasanthavada, K., Kohler, K., McNary, S., Moore, A., Vierra, C.: Molecular mechanisms of spider silk. Cell. Mol. Life Sci. 63(17), 1986–1999 (2006). doi:10.1007/s00018-0066090-y Denny, M.: The physical properties of Spider’s silk and their role in the design of orb-webs. J. Exp. Biol. 65(2), 483–506 (1976) Huemmerich, D., Scheibel, T., Vollrath, F., Cohen, S., Gat, U., Ittah, S.: Novel assembly properties of recombinant spider dragline silk proteins. Curr. Biol. CB 14(22), 2070–2074 (2004) Foo, C., Po, W., Bini, E., Huang, J., Lee, S.Y., Kaplan, D.L.: Solution behavior of synthetic silk peptides and modified recombinant silk proteins. Appl. Phys. A 82, 193–203 (2006) Gosline, J., Guerette, P., Ortlepp, C., Savage, K.: The mechanical design of spider silks: from fibroin sequence to mechanical function. J. Exp. Biol. 202(23), 3295–3303 (1999) Nova, A., Keten, S., Pugno, N.M., Redaelli, A., Buehler, M.J.: Molecular and nanostructural mechanisms of deformation. Strength and toughness of spider silk fibrils. Nano Lett. 10(7), 2626–2634 (2010). doi:10.1021/nl101341w Grubb, D.T., Jelinski, L.W.: Fiber morphology of spider silk: the effects of tensile deformation. Macromolecules 30(10), 2860–2867 (1997). doi:10.1021/ma961293c Gosline, J., Lillie, M., Carrington, E., Guerette, P., Ortlepp, C., Savage, K.: Elastic proteins: biological roles and mechanical properties. Philos. Trans. R. Soc. Lond. B Biol. Sci. 357(1418), 121–132 (2002). doi:10.1098/rstb.2001.1022 Du, N., Yang, Z., Liu, X.Y., Li, Y., Xu, H.Y.: Structural origin of the strain-hardening of spider silk. Adv. Funct. Mater. 21(4), 772–778 (2010). doi:10.1002/adfm.201001397 Bratzel, G.H., Buehler, M.: Sequence-structure correlations and size effects in silk nanostructure: Poly-Ala repeat of N. clavipes MaSp1 is naturally optimized. JMBBM (2011) doi:10.1016/j.jmbbm.2011.07.012 Gatesy, J., Hayashi, C., Motriuk, D., Woods, J., Lewis, R.: Extreme diversity, conservation, and convergence of spider silk fibroin sequences. Science 291(5513), 2603–2605 (2001). doi:10.1126/science.1057561 Keten, S., Buehler, M.J.: Nanostructure and molecular mechanics of spider dragline silk protein assemblies. J. R. Soc. Interface 7(53), 1709–1721 (2010). doi:10.1098/ rsif.2010.0149 Fratzl, P.: Collagen: Structure and Mechanics. Springer, New York (2008) Gautieri, A., Vesentini, S., Redaelli, A., Buehler, M.J.: Hierarchical structure and nanomechanics of collagen
1295
37.
38.
39.
40.
41.
42.
43.
44.
M
microfibrils from the atomistic scale up. Nano Lett. 11(2), 757–766 (2011). doi:10.1021/nl103943u Jackson, A.P., Vincent, J.F.V., Turner, R.M.: The mechanical design of nacre. Proc. R. Soc. Lond. Series B. Biol. Sci. 234(1277), 415–440 (1988). doi:10.1098/rspb.1988.0056 Espinosa, H., Rim, J., Barthelat, F., Buehler, M.: Merger of structure and material in nacre and bone – perspectives on de novo biomimetic materials. Prog. Mater. Sci. 54, 1059–1100 (2009) Currey, J.: Mechanical properties of vertebrate hard tissues. Proc. Inst. Mech. Eng. Part H: J. Eng. Med. 212(6), 399–411 (1998) Sasaki, N., Odajima, S.: Elongation mechanism of collagen fibrils and force-strain relations of tendon at each level of structural hierarchy. J. Biomech. 29(9), 1131–1136 (1996) Eppell, S.J., Smith, B.N., Kahn, H., Ballarini, R.: Nano measurements with micro-devices: mechanical properties of hydrated collagen fibrils. J. R. Soc. Interface 3(6), 117–121 (2006). doi:10.1098/rsif.2005.0100 Espinosa, H.D., Zhu, Y., Moldovan, N.: Design and operation of a MEMS-based nanoscale material testing system. J. Microelectromech. Syst. 16, 1219–1231 (2007) van der Rijt, J.A.J., van der Werf, K.O., Bennink, M.L., Dijkstra, P.J., Feijen, J.: Micromechanical testing of individual collagen fibrils. Macromol. Biosci. 6(9), 697–702 (2006). doi:10.1002/mabi.200600063 Aladin, D.M.K., Cheung, K.M.C., Ngan, A.H.W., Chan, D., Leung, V.Y.L., Lim, C.T., Luk, K.D.K., Lu, W.W.: Nanostructure of collagen fibrils in human nucleus pulposus and its correlation with macroscale tissue mechanics. J. Orthop. Res. 28(4), 497–502 (2009). doi:10.1002/jor.21010
Mechanical Properties of Nanocrystalline Metals Farghalli A. Mohamed Department of Chemical Engineering and Materials Science, University of California, Irvine The Henry Samueli School of Engineering, Irvine, CA, USA
Synonyms Mechanical behavior of nanostructured metals
Definition Materials are classified into four groups in terms of the value of grain size: large-grained materials, micrograined materials, ultrafine-grained (UFG) materials, and nanocrystalline (nc) materials. Large-grained materials are characterized by grain sizes larger than
M
M
1296
20 mm, micrograined materials by grain sizes in the range 1–10 mm, ultrafine-grained materials by grain sizes in the range 300–900 nm, and nc-materials by grain sizes less than 200 nm. The grain size ranges associated with the above classifications are approximate and are based on consideration of observations reported in literature. Figure 1 provides a transmission electron microscopy (TEM) micrograph for the grain size in nc-nickel. The grain size ranges from 3 to 60 nm with an average value of about 24 nm. Because of the small grain sizes characterizing ncmaterials, features such as grain boundaries, junction lines, and nodes have significant volume fractions and influence properties far more strongly than in more conventional microstructures [1, 2]. Nc-materials have novel physical, chemical, biological, and mechanical properties that may have important engineering uses. As a consequence, they are attracting wide attention in materials research. While the chemical and electromagnetic properties of nc-materials have received most of the attention, their mechanical properties are also important. It is both necessary and useful that these properties be thoroughly understood. Understanding mechanical properties is necessary since materials in almost all devices experience mechanical loads that must not only be supported but also be predicted to produce reliable designs. If the mechanical properties are unknown, even promising chemical or electromagnetic applications may be very difficult to achieve. Understanding the mechanical properties of nc-materials would be useful because of the following considerations: • It is possible to make nc-materials with exceptionally high strengths for use as hard coatings or highstrength structural elements. • The potential of using nanomaterials as machine components; if the nc-materials are used as cutting tools, localized shear should be avoided to increase the tool life. Over the past two decades, considerable progress has been made in terms of providing a better understanding of mechanical behavior in nc-materials. In particular, considerable progress has been made in areas related to the following four aspects of the mechanical behavior of nc-materials: (a) creep behavior, (b) yielding and hardness, (c) ductility, and (d) superplasticity. Before presenting key research findings in these areas, it is
Mechanical Properties of Nanocrystalline Metals
Mechanical Properties of Nanocrystalline Metals, Fig. 1 Bright field transmission electron microscopy micrograph showing the nanograin distribution in nc-nickel. The ring configuration at the upper right corner represents the diffraction pattern for the nc-grain structure
useful to provide information on some of the parameters that are important in terms of characterizing mechanical behavior in materials.
Mechanical Properties Basic Parameters Activation Volume The activation volume, v, is the rate of decrease of the activation enthalpy with respect to flow stress at constant temperature. It provides useful insight into the nature of deformation mechanisms because it is characterized by a definite value of stress dependence for each atomistic process. Measurements of the activation volume during deformation are made by (a) making either a small change in applied stress, and measuring the corresponding strain rate after the change, or a small change in strain rate and measuring the corresponding stress, and (b) applying the following equation: v ¼ kT
@ ln g_ @t
(1)
where k is Boltzmann’s constant and T is the absolute temperature, t is the shear stress, and g_ is the shear strain rate. Apparent Activation Energy When the plastic deformation of a material is controlled by a thermally activated process, the strain
Mechanical Properties of Nanocrystalline Metals
1297
rate for the i-the process may be expressed in terms of rate reaction theory as: g_i ¼ Ai ðt; TÞe
DHi ðt;TÞ RT
"
(3) t;d
where d is the grain size. Equation (3) suggests a direct method for estimating DHi . In deformation experiments, DHi is approximated by the apparent activation, Qa: " # @ ln g_ Qa DHi ¼ R 1 (4) @ T t;d
Apparent Dependence of Strain Rate on Applied Stress The apparent dependence of strain rate on applied stress is an important parameter that receives attention in studies on the mechanical behavior of materials. In investigations focusing on high temperature deformation behavior, such dependence is represented by the stress exponent, n, which is defined as: n¼
@ ln g_ @ ln t T;d
(5)
There is also another parameter that has been extensively used to express the relation between strain rate and stress, and is termed strain rate sensitivity, m. This parameter is given by: m¼
@ ln t @ ln g_ T;d
aforementioned four areas, it is useful to touch upon crystal imperfections that play key roles during plastic deformation.
(2)
where Ai is a function of stress, t, temperature, T, and substructure; R is the gas constant; and DHi is the activation enthalpy. DHi is an important parameter in the deformation equation, and it must be determined in any attempt to identify the rate-controlling process. According to (2), DHi is uniquely defined by the following expression: # @ ln g_ i DHi ¼ R 1 @ T
M
(6)
Under steady-state high deformation conditions, n ¼ 1/m. In addition to mentioning briefly the
Crystal Imperfections Metallic or ceramic crystals are not perfect. They contain imperfections [3] that include: point imperfections, line imperfections, and surface imperfections. Point imperfections include vacancies, substitutional atoms (e.g., magnesium in aluminum), and interstitial atoms (e.g., carbon in iron). Line imperfections refer to dislocations. Surface imperfections include grain boundaries. While substitutional and interstitial atoms are responsible in many ways for strengthening in materials, the creation and diffusion of vacancies provide a mode of deformation at very low stress and small grain sizes [4]. Figure 2 can illustrate this mode of deformation, known as diffusional creep [4]. According to this figure, three steps are associated with diffusional creep: (a) the creation of a vacancy in the grain volume near the grain boundary under tensile stress, (b) the migration of the vacancy to the boundary that is under compressive stress, and (c) the annihilation of the vacancy at that boundary. The flow of vacancies is matched by an equal and opposite flow of atoms resulting in a longitudinal extension and lateral contraction of the grain with time under stress. If the migration of vacancies occurs in the interior of the grain, the deformation process is referred to as Nabarro-Herring creep. If, on the other hand, the migration of vacancies takes place in the grain boundary, the process is known as Coble creep. Dislocations play a key role in plastic deformation. Their presence and movement in crystals account for the fact that the experimental yield strengths of materials are much lower than their theoretical shear strengths [3]. Their multiplication and interactions produce strain hardening while their annihilation results in crystal recovery. By using the details of dislocation theory, several investigators were able to develop a number of deformation mechanisms that have provided plausible interpretations for mechanical behavior under various conditions. Grain boundaries are regions of misorientation that separate grains. At low homologous temperatures (T/Tm < 0.2), grain boundaries act as a source for strengthening by serving as obstacles to the propagation of deformation from one grain to the next [3]. At high homologous temperatures (T/Tm > 0.4), grain
M
1298
Mechanical Properties of Nanocrystalline Metals
Mechanical Properties of Nanocrystalline Metals, Fig. 2 Stress-directed diffusion of vacancies in a polycrystalline material
σ
−σ
Va ca Mi n gr ati cy on
M
Compressive Stress (−σ)
d
σ
Tensile Stress (σ)
Vacancy flow Atom flow
a
Grain A
b Grain B Grain A θ
Grain B
− Sliding offset
w −
r
p σ
v α
σ
u
Mechanical Properties of Nanocrystalline Metals, Fig. 3 (a) Sliding offset as a result of boundary sliding. (b) Sliding displacement associated with boundary sliding
boundaries become mobile especially at low stresses, producing strain via grain boundary sliding. Figure 3 illustrates the process of boundary sliding. When enough stress is applied to a polycrystalline material, the grains slide past each other and the metal changes shape or deforms, a phenomenon known as “boundary sliding.” As a result of boundary sliding, steps (offsets) are produced on the surface of a flat specimen. Detection of these steps provides evidence for the occurrence of boundary sliding, a process that may lead to the formation of voids and cavities in the material. The formation of voids in turn results in premature failure of the material. However, in micrograined materials, the strain contributed by boundary sliding may become considerable, leading to the occurrence of micrograin superplasticity [5].
Key Research Findings Creep Behavior A polycrystalline material subjected to an applied stress or load deforms continuously. This progressive accumulation of strain with time is known as creep [4]. Creep takes place as a result of thermal activation of deformation processes. The curve in Fig. 4 illustrates the idealized shape of a creep curve. The slope of this curve, dg/dt, is referred to as the creep rate (or the strain rate). Following an initial rapid elongation of the specimen, the creep rate decreases with time (primary creep, or primary stage), then reaches a steady state (secondary creep, steady-state stage), and finally the creep rate increases with time until fracture occurs (tertiary creep or tertiary stage). Major differences
Mechanical Properties of Nanocrystalline Metals
1299 Fracture
Strain
I Primary Creep
II Secondary Creep
III Tertiary Creep
tate
dy S
Stea
γ0 Time (t)
Mechanical Properties of Nanocrystalline Fig. 4 An example for a normal creep curve
Metals,
among creep curves of various materials, however, are noted over the primary creep stage. Nc-materials offer interesting possibilities related to many structural applications. In order to explore some of these possibilities, an understanding of the deformation of these materials under creep conditions is essential. In order to develop this understanding, guiding information is needed. This guiding information can be obtained from comprehensive experimental creep measurements. During the period of 1990–2005, experimental measurements dealing with the creep behavior of ncmaterials were reported. However, these measurements not only were limited in scope but also were obtained under questionable conditions. For example, the total strain measured in creep experiments conducted on nc-materials in several investigations did not exceed in most cases 0.01. With such small strains, it is difficult to ascertain whether the steadystate creep is reached. The problem of the lack of experimental measurements that would characterize creep and deformation-induced grain size growth in nc-materials was recently addressed in a detailed investigation [6], in which the creep behavior of ncNi was studied for the first time under the conditions of large strains and constant stresses. In the investigation, creep testing was performed over more than five orders of magnitude of strain rate (109 s1 to 2 104 s1) at 393 K (0.23 Tm, where Tm is the melting point). In addition, information regarding the stress and the temperature dependence of strain rate, the activation volume, and deformation-induced grain growth was obtained. The results of the above investigation have shown that the behavior of nc-Ni exhibited a variable
M
stress exponent, an activation energy close to that expected for boundary diffusion, and an activation volume of about 19.5 b3; and that upon loading nc-Ni specimens at 393 K, there is a grain growth, and this growth is driven by a combination of stress and time. However, once steady-state creep is reached, significant grain growth ceases. Yield Strength and Hardness One of the aspects of the mechanical behavior of ncmaterials that has been extensively studied is the influence of grain size on flow shear stress, t (¼ s=2). In general, as thepgrain size, d, is refined, the strength ffiffiffi increases as 1= d according to the Hall-Petch relation [3] that can be represented by the following expression: t ¼ t0 þ c
.pffiffiffi d
(7)
where t0 is a friction stress and c is a constant. Equation (7) is also valid when the yield strength is replaced by the hardness, H (H ¼ 6 t). The mechanistic source of the Hall-Petch relation is not entirely clear, and more than one mechanism may pertain [3]. The effect of (7) in the nanoscale regime is, however, striking; if (7) holds, decreasing d from 20 to 20 nm increases the strength by a factor of 30. In general, Hall-Petch behavior for both hardness and yield strength holds to grain sizes well within the nanoscale range. However, experimental observations have suggested that when the grain size of some ncmaterials falls below a critical value in the nanoscale range, the strength may decrease, that is, nanoscale softening (an inverse Hall-Petch behavior) occurs. The observations related to nanoscale softening were initially controversial since some of them could be attributed to the presence of flaws in the material [7]. For example, the loss in strength in the nanoscale range can be the result of pores in the material whose presence leads to a decrease in both the shear stress required for deformation and the shear modulus. Despite the aforementioned controversy and others related to contaminations and grain size measurements, three pieces of information lend support to the occurrence of softening in the nanoscale range. First, it does seem reasonable that Hall-Petch behavior would be lost at the smallest grain sizes in the nanoscale region partly because it is no longer possible to
M
M
1300
Mechanical Properties of Nanocrystalline Metals
a 100
15
d (nm) 11.1
25
b 6.25
4
Ni
HV (GPa)
10
dc ∼ 8–12 nm Hall-Petch Equation
5
0 0.0
0.1
0.2
0.3 d
−1/2
0.4
0.5
0.6
(nm−1/2)
Mechanical Properties of Nanocrystalline Metals, Fig. 5 (a) Plot of hardness vs. the inverse square root of grain size for nc-nickel. (b) Data on copper plotted as yield strength as a function of the inverse square root of the grain size for Cu
maintain dislocation structures within grains, and partly because grain boundary activity assumes significance at very small grain sizes. Second, recent experimental results reported for the variation of hardness against grain size in nickel, nickel-tungsten alloys, and copper (see Fig. 5) have provided clear evidence for the occurrence of nanoscale softening [8, 9]. Finally, the results of recent molecular dynamics simulations have shown [10] the presence of a maximum in the flow stress of copper when its grain size becomes smaller than 15 nm.
Deformation Processes An understanding of the origin and nature of deformation processes in nc-materials is essential. Such understanding is important in two ways. First, when the origin of the deformation process is uncovered in sufficient detail, it should be possible to predict the mechanical behavior under a variety of conditions (stress, temperature, and grain size). Second, when the basic deformation process is known (successfully identified or developed), it is possible to introduce microstructural features that can improve the mechanical behavior of the materials in terms of ductility, toughness, etc. As a result of recent investigations on nc-materials, several deformation mechanisms have been proposed.
These mechanisms are different in terms of concept and details. The proposed mechanisms include: (a) Coble creep, (b) Triple Junction Model, (c) Atomic Scale Boundary Sliding Model, (d) Modified Coble creep (inclusion of a threshold stress), (e) Grain Boundary Sliding with Back Stress, (f) Boundary Sliding under the Condition of Nonlinear-Viscous Behavior, (g) Thermally Activated Grain Boundary Shearing Process, (h) Thermally Activated Grain Boundary Shearing Process, (i) Composite model involving amorphous boundary layer, (j) Model of strongest size, and (k) Dislocation Accommodated Boundary Sliding. These mechanisms were reviewed in reference [11]. As a result of investigations on nc-materials, extensive experimental data on the mechanical behavior of nc-materials such as nc-Ni and nc-Cu have become available [11]. The availability of the experimental data on these two metals provides an opportunity to closely examine the validity of deformation mechanisms that were proposed to account for the characteristics of deformation in nc-materials. An analysis of the data reported led to the identification of the requirements [10] that a viable deformation mechanism should meet in terms of accounting for the mechanical characteristics, which were revealed by the data. These requirements are: (a) an activation volume whose value is in the range 10–40 b3, (b) an activation energy that is close to the activation energy
Mechanical Properties of Nanocrystalline Metals
1301
a
b
1024
d (nm) 25
100
800
11.1
M 6.25
Cu
1020
10
dc = 25 nm
τ (MPa)
2 g· ( d )3 ( b ) Dgb b
600
1016 12
400
500–4040(d)−1/2
Hall-Petch Equation
108 200
104 100 100
Dislocation-accomodated boundary sliding (Model)
104
108
1012 1016 [exp( tn ) – 1] KT
1020
1024
0 0.0
0.1
0.2
0.3
0.4
−1/2 (nm−1/2)
d
Mechanical Properties of Nanocrystalline Metals, Fig. 6 (a) Plot of g_ ðd=bÞ3 b2 Dgb against expðtn=KT Þ 1 on a logarithmic scale showing the correlation between deformation data reported for nc-Ni and nc-Cu and the prediction of
the Model of Dislocation Accommodated Boundary Sliding (solid line in the plot) [6]. (b) Yield strength as a function of the inverse square root of the grain size for copper
for boundary diffusion but that decreases with increasing applied stress, (c) the magnitudes of deformation rates that cover wide ranges of temperatures, stresses, and grain sizes, (d) inverse Hall-Petch behavior (softening), and (e) limited ductility. The validity of aforementioned deformation mechanisms for nc-materials has been closely examined in the light of these requirements. The results of the examination have led to the following findings [11]: (a) Coble creep and thermally activated Grain Boundary Shearing process are ruled out as viable mechanisms not only because they predict an activation volume smaller than those experimentally reported but also because they yield unrealistic critical grain sizes for a transition from conventional Hall-Petch behavior (hardening with decreasing grain size) to inverse Hall-Petch behavior (softening with decreasing grain size). (b) Mechanisms such as modified Coble creep and thermally activated grain boundary shearing process are not applicable since they predict a very small value of the activation volume; they predict an activation energy of b3 energy compared to experimentally reported values of 20–40 b3. (c) While the composite model involving a grain boundary layer yields critical grain sizes for a transition from conventional Hall-Petch behavior (hardening with decreasing grain size) to inverse
Hall-Petch behavior (softening with decreasing grain size) that agree well with those inferred from experimental results, it fails to account for the experimental data for nc-Cu and nc-Ni. (d) The predictions of the model of dislocation accommodated boundary sliding [11] are consistent with several deformation characteristics reported for ncmaterials including the range of the activation volume, (b) the value of activation energy and its dependence on stress, (c) the trends and magnitudes of experimental data on nc-Cu and nc-Ni, and (d) the critical grain sizes for transitions from conventional Hall-Petch behavior to inverse Hall-Petch behavior in nc-Cu and nc-Ni. Figures 6a, b illustrate characteristics (c) and (d), respectively. Ductility Ductility is an important mechanical property that denotes the total elongation experienced by a tensile sample prior to fracture, ef. It is defined by the following equation [12]: ef ¼ L Lo =Lo
(8)
where L is the final length and Lo is the initial length Although nc-materials always possess higher strength than their coarse-grain counterparts, they usually exhibit low ductility. For engineering applications,
M
M
1302
Mechanical Properties of Nanocrystalline Metals
Mechanical Properties of Nanocrystalline Metals, Fig. 7 Relationship between shear stress and shear train rate for superplastic materials
Sigmoidal relationship between stress and strain rate
Region II n=2–2.5
Region II m = 0.4–0.5
n=
n
Region I n>2.5
1
1 m
Region I m < 0.4
m 1
log g·
log t Constant stress (Creep)
low ductility is undesirable for two primary reasons. First, low ductility does not permit metalworking processes to effectively produce required configurations. Second, low ductility impairs the material’s ability to absorb overload, a condition that reduces the margin of safety in design. The source of low ductility ncmaterials is not entirely clear, but it probably arises from either extrinsic flaws such as porosity or loss of work hardening due to the lack of dislocation interactions and accumulations [7]. Recently, the concept of improving ductility via enhancing dislocation interactions and accumulations has received considerable attention. This attention has led to several approaches that include the introduction of a bimodal grain size distribution, deformation twins, or dispersion of nanoparticles. The role of both bimodal grain distribution and twins in achieving high ductility can be attributed to two primary factors. First, the presence of large grains enhances strain hardening, which prevents excessively large local strains (plastic deformation becomes localized), and suppresses the propagation of cracks nucleated in the surrounding regions of the very small grains. Second, twins in the interiors of the large grains in the bimodal grain distribution can block moving dislocations, a process that results in dislocation interaction and accumulation. Dislocation interactions and accumulation in turn promote strain hardening.
Region III m < 0.4 log t
log g
Region III n >2.5
Constant strain rate
Superplasticity Before discussing superplasticity in nc-materials, it is appropriate to provide a brief review for superplastic behavior in micrograin materials. Superplasticity refers to the ability of micrograined materials (d < 10 mm, where d is the grain size) particular materials to exhibit extensive neck-free elongations (>300%) during deformation at elevated temperatures (T > 0.5 Tm, where Tm is the melting point). The behavior of micrograin superplastic alloys has been extensively studied. As a result of these investigations, three findings are well documented [5]. First, micrograin superplasticity is a diffusioncontrolled process. Second, the relationship between stress, t, and strain rate, g_ , is often sigmoidal (see Fig. 7). Under creep testing conditions, this sigmoidal relationship is manifested by the presence of three regions: region I (the low-stress region), region II (the intermediate-stress region), and region III (the high-stress region). In region II, the stress exponent, n, is about 2 and the activation energy for creep is equal to that for grain boundary diffusion, Qgb. In regions I and III, both the stress exponent, n, and the activation energy for creep, Qc, increase (Qc > Qgb). Finally, in region II, maximum ductility (>300%) occurs. Because of this characteristic, region II is often referred to as the superplastic region.
Mechanical Properties of Nanocrystalline Metals
From a practical point of view, it is very desirable to extend the superplastic region to high strain rates. As reported elsewhere [5], the superplastic region is in general observed at strain rates in the range of 105– 102 s1. These rates are too slow to be economically suitable for a variety of industrial applications. Ncmaterials appear to offer an interesting possibility to address this issue. According to information on micrograin superplasticity, as the grain size decreases from micrometer to nanometer, it is expected that the superplastic region would be transposed to high strain rates or observed at low temperatures such as room temperature, that is, high strain rate and/or low temperature superplasticity is possible. Despite this attractive possibility and its implications in terms of the potential for industrial adaptation of superplastic forming, attempts made so far to explore the occurrence of superplasticity have been unsuccessful. One possible reason for the difficulty of observing superplastic flow in nc-materials is the occurrence of grain growth: it is not possible to obtain lowtemperature superplasticity in pure metals in which grain sizes are less than 100 nm, since the advantage of observing superplastic flow at low temperatures is neutralized by the onset of significant grain growth in this temperature range. A second reason can be inferred from the rate controlling mechanism for deformation in nc-materials. Recently, a new model for deformation in nanocrystalline (nc) materials has been developed [6]. The development of the model was based on the concept that plasticity in nc-materials is the result of grain boundary sliding accommodated by the generation and motion of dislocations under local stresses, which are higher than applied stresses due to the development of stress concentrations. Specifically, it has been assumed that as a result of sliding of a group of grains, the shear stress becomes concentrated at any grain, triple point, or protrusion that obstructs motion of this group; that this high local stress can then generate dislocations in the blocking grain (or initiate voids); and that the generated dislocations move one by one to the opposite boundary where they climb to their annihilation sites (no dislocationups). For illustration, see Fig. 8. By postulating that the creep rate, g_ , is governed by the time for the climb of a dislocation along the boundary until annihilation occurs, the following rate-controlling equation was derived [6]:
M
1303
Blocking Grain τ GBS h
Mechanical Properties of Nanocrystalline Metals, Fig. 8 Schematic diagram for the model of dislocation accommodated boundary sliding, showing that as a result of boundary sliding, dislocations are generated at a triple junction and then traverse the grain to the opposite grain boundary where they climb and are annihilated
3
Dgbo Qgb b g_ ¼ 9 exp d b2 RT h t v i exp 1 kT
(9)
where b is the Burgers vector, Dgbo is the frequency factor for grain boundary diffusion, and Qgb is the activation energy for grain boundary diffusion. The stress exponent for creep, n, as estimated from (9), exhibits high and variable values; for the applicable range of stresses, n > 5. Accordingly, it is expected that ductility in nc-materials would be much lower than those characterizing micrograined superplastic alloys for which n ¼ 2, since ductility depends on 1/n (n ¼ 1/m) as shown by the following (5): ef % ¼ ½expðC=ðn 1ÞÞ 1 100
(10)
where C ¼ (n (1/n)) ln (400/n); n ¼ 1/m. As shown in reference [6], combining (9) and (10) leads to predicting that ductility not only is low and in the range 2–10%. From a microstructural point of view, the low ductility can be explained in terms of the details of the concept of dislocation accommodated boundary sliding [5]. According to the concept, dislocations are generated under the local stresses and move in the blocking grains to boundaries where they are annihilated. This scheme results in the absence of dislocation accumulation and interactions, that is,
M
M
1304
nanograins are not able to sustain arrays of dislocations. This leads to loss of work hardening, which in turn results in low ductility.
Mechanical Properties of Nanocrystalline Metals
Photodiode Laser
Future Research Creep Nc-nickel has served as a model material to study the mechanical behavior of nc-materials. However, in order to establish the generality of the results, it is necessary to investigate the deformation behavior of other nc-materials using approaches similar to those adopted in studying the creep characteristics of nc-Ni. In addition, there is a need to study the creep behavior of nc-aluminum alloys to examine their thermal stability under temperature and stress. Grain Boundary Sliding Atomistic computer simulations on nc-materials have revealed the dominance of boundary sliding with decreasing grain size at room temperature. Despite this evidence and despite the need to establish the role of boundary sliding during deformation in nc-materials, the occurrence of this process has not been fully addressed in terms of experimental measurements. It is important to experimentally examine the hypothesis that strain measured during creep in nc-metals mostly arises from sliding. Traditional techniques (surface marker and scanning electron microscopy) that were utilized in measuring sliding micrograined superplastic alloys do not have sufficient resolution. Recently, a solution for the above problem has become available: investigators have demonstrated that atomic force microscopy (AFM) provides a powerful technique for obtaining high-resolution images of the surface topography of bulk materials. AFM works by scanning a fine ceramic or semiconductor tip over a surface. The tip is positioned at the cantilever beam shaped much like a diving board. As the tip is repelled by or attracted to the surface, the cantilever beam deflects (see Fig. 9). The magnitude of the deflection is detected by a laser that reflects at an oblique angle from the very end of the cantilever. A plot of the laser deflection vs. tip
Tip (probe)
Cantilever
Piezo movement z x
Scanner & stage
y
Mechanical Properties of Nanocrystalline Metals, Fig. 9 The illustration shows how the technique of atomic force microscopy works
position on the sample surface provides the resolution of the hills and valleys that constitute the topography of the surface. On this basis, processes such as boundary sliding that are characterized by the occurrence of offsets at the surface of a flat specimen can be detected in ultrafine-grained and in nc-materials by AFM. Evidence of Dislocations Activity During Deformation Transmission electron microscopy (TEM) observations on deformed samples of nc-nickel have indicated the presence of few dislocations in grain interiors. However, the absence of evidence regarding dislocations in deformed nc-Ni is not evidence of the absence of dislocation activity during deformation. According to the model of dislocation accommodated boundary sliding [4], dislocations are generated under the local stresses and move in the blocking grains to boundaries where they are annihilated. Also, during unloading and/or subsequent specimen preparation for TEM examination, dislocations could disappear; dislocations only need to travel a very short distance of less than 100 b to be lost at nearby grain boundaries. The hypothesis to be tested is that dislocations are generated during deformation in nc-materials and that their absence in TEM micrographs are due to their annihilation at boundaries during deformation or during unloading and subsequent sample preparation. In order to test the validity of this hypothesis, it is suggested that strong obstacles such as nanoscale
MEMS Adaptive Optics
particles be introduced in the grain interior to provide effective barriers to the movement of dislocations, leading to attachment of dislocations to particles and to dislocation tangles.
Cross-References ▶ Atomic Force Microscopy ▶ Nanoparticles ▶ Nanotechnology ▶ Transmission Electron Microscopy
References 1. Gleiter, H.: Nanostructured materials: basic concepts and microstructure. Acta. Mater. 48, 1–29 (2000) 2. Suryanarayana, C.: Mechanical alloying and milling, p. 333. Marcel Dekker, New York (2004). Ch #13 3. Meyers, M., Chawla, K.: Mechanical behavior of materials, 2nd edn, pp. 345–352. Cambridge University Press, Cambridge (2009) 4. Kassner, M.E., Perez-Prade, M.-T.: Fundamental of creep in metals and alloys. Elsevier, Amsterdam (2004) 5. Mohamed, F.A.: The role of boundaries during superplastic deformation. Surf. Interface Anal. 31, 532–546 (2001) 6. Mohamed, F.A., Chauhan, M.: Interpretation of the creep behavior of nanocrystalline Ni in terms of dislocation accommodation boundary sliding. Metall. Mater. Trans. A 37A, 3555–3567 (2006) 7. Meyers, M.A., Mishra, A., Benson, D.J.: Mechanical properties of nanocrystalline materials. Prog. Mater. Sci. 51, 427–556 (2006) 8. Schuh, C.A., Nieh, T.G., Iwasaki, H.: Acta. Mater. 51, 431– 443 (2003) 9. Chokshi, A.H., Rosen, A., Karch, J., Gleiter, H.: Scripta. Metall. 23, 1679–1683 (1989) 10. Schiotz, J., Jacobsen, K.W.: Science 301, 1357–1359 (2003) 11. Mohamed, F.A., Yang, H.: Deformation mechanisms in nanocrystalline materials. Met. Mater. Trans. A 41A, 823–837 (2010) 12. Dieter, G.: Mechanical metallurgy. McGraw Hill Science Engineering, New York (1986)
1305
M
Mechanoreceptors ▶ Arthropod Strain Sensors
Medical Adhesives ▶ Bioadhesives
MEMS ▶ Nanotechnology
MEMS (Micro-electro-Mechanical Systems) ▶ CMOS MEMS Fabrication Technologies
M MEMS = Microelectromechanical Systems ▶ MEMS on Flexible Substrates
MEMS Adaptive Optics Nathan Doble The New England College of Optometry, Boston, MA, USA
Definition
Mechanical Properties of Nanomaterials ▶ Computational Study of Nanomaterials: From LargeScale Atomistic Simulations to Mesoscopic Modeling
MEMS: Micro-electro-mechanical Systems MEMS are small silicon-based mechanical devices that employ semiconductor fabrication techniques in their construction. MEMS devices may take various forms: from pressure, chemical, and vibration sensors, to optical switches and mirrors through to inkjet printer heads.
M
1306
MOEMS – Micro-Opto-Electromechanical Systems – are a subset of MEMS devices that in addition employ optical components; an example could be a photon detector mounted on a silicon chip. Note: MEMS/MOEMS devices have structures that can be measured at the micrometer level, whereas nanotechnology deals with scales at the level of an individual molecule (generally taken to be in the 1–100 nm range). Adaptive Optics Adaptive Optics (AO) are optical systems that can both measure and correct optical wavefront aberrations in real time. Real time depends on the particular AO application: in astronomy closed loop bandwidths of several hundred Hertz are required to correct the rapidly changing atmospheric turbulence. For vision science, changes in the lens/cornea and head/eye movements require a slower correction speeds (a few Hertz). Typically, an AO system has three main system components: (1) a wavefront sensor (WFS) that measures the aberration (note: there are also wavefront sensor-less AO systems), (2) a wavefront corrector – typically a deformable mirror (DM), and (3) a control computer that controls the WFS and wavefront corrector. MEMS Adaptive Optics An AO system that employs MEMS technology in the wavefront corrector
Introduction Adaptive optics (AO) is a technique that measures and corrects wavefront distortions. These distortions may be due to effects of atmospheric turbulence (astronomy or free space communications); however, they may also be biological in nature (vision science or microscopy). AO was first proposed by Babcock in 1953 for use in large, ground-based telescopes but implementation at that time was limited by the available technology. The initial development beginning in the 1970s was for military applications. Declassification in the early 1980s led to significant use in astronomy with all large terrestrial telescopes now using AO to enhance their image quality. Other scientific fields have benefited greatly from this work particularly in the fields of vision science and microscopy, both of
MEMS Adaptive Optics
which have leveraged the technology and know-how to provide AO corrected images with remarkable clarity. This entry focuses on recent developments in AO that use micro-electro-mechanical systems (MEMS) technology. The technology itself and the results in a range of research fields are described. A detailed discussion of AO principles is, however, beyond the scope of this entry and the interested reader is referred to the reference texts [1–3].
Principle and Application of Adaptive Optics (AO) A schematic AO layout is shown in Fig. 1 and generally requires the implementation of three main subsystems: The Wavefront Sensor (WFS) – This measures the optical aberrations in the incoming distorted wavefront. There are two broad WFS classes [1, 2]: (1) direct – in which the phase or optical path difference is measured, e.g., Hartmann-Shack (Fig. 2), curvature, and pyramid sensing, (2) indirect – the wavefront properties are not explicitly measured, but rather the correction signals are deduced from wholeaperture intensity measurements made at or near the focal plane. These are generally iterative in nature with examples being image sharpening and phase diversity. The Control Computer – This takes the output of the WFS and converts it to voltage commands that are sent to the wavefront corrector. The Wavefront Corrector – This device provides the conjugate correction to the measured optical distortions. The aim of any wavefront corrector is to equalize the optical path length (OPL) seen by all points on the incident wavefront, resulting in a flat wavefront after correction. The OPL is defined as the product of the refractive index, n, and the physical path length, d. Wavefront correctors can be divided into two main classes depending on how they impart the OPL change: (1) deformable mirrors (DMs) – these are by far the most common device used. As the name suggests these operate by locally deforming a mirrored surface, i.e., they change the physical path length, d, and (2) liquid crystal spatial light modulators (LC-SLMs) – these produce localized changes in their refractive index, n (d is fixed). The refractive index change is achieved through the application of a voltage that rotates the liquid crystal molecules.
MEMS Adaptive Optics
1307
M
Telescope
Dichroic Beam Splitter
Science Camera
Deformable Mirror Wavefront Sensor Actuator Control
Wavefront Reconstructor
MEMS Adaptive Optics, Fig. 1 The basic AO operating principle. In this case, the distorted wavefront from a telescope is reflected off the deformable mirror (DM). A fraction of the reflected light is redirected to the wavefront sensor (WFS) where the aberrations are measured and the appropriate control signals are determined for the DM. This continual measurement and correction of the wavefront distortion is termed closed loop, the speed of which depends on the particular application.
a
Pupil
The majority of the light is used in forming the image at the science camera, the aim being to obtain planar wavefronts that focus down close to the limits imposed by diffraction. Note: there are wavefront sensor-less AO approaches that use a measure of the light intensity at the science camera to drive the AO loop and which do not require a separate WFS arm (Reproduced from Expert Review of Medical Devices, 2, 205–216 (2005) with permission of Expert Reviews Ltd)
b Δx
Lenslet Reference Wavefront
Camera Plane
Lenslet Array
Aberrated Wavefront
MEMS Adaptive Optics, Fig. 2 Principle of the Hartmann– Shack WFS. (a) The incident pupil (gray circle) is sampled by a two-dimensional array of lenslets, the diameter and focal length of which are chosen to give the required measurement resolution and dynamic range. (b) For the reference wavefront (gray hatched line), a focal spot is formed by each lenslet on the
camera plane and is exactly centered behind each lenslet. For an aberrated wavefront (solid gray line), variations in the slope between each lenslet will cause the focal spots to be shifted from the position of the reference. By determining both the Dx and Dy shifts for each lenslet, the aberrated wavefront may be reconstructed
Consider Fig. 1 and the example of aberrations entering a telescope; the incident distorted wavefront is reflected by the DM with a fraction of the reflected light being redirected into the WFS arm where the aberrations are measured and the appropriate control
signals are determined for the DM. The DM surface profile is then updated to exactly cancel the aberrations in the wavefront. Both the DM and WFS are placed in optical planes that are conjugate to the source of the aberration, for example, in the human eye this plane is
M
M
1308
taken to be somewhere between the cornea and lens. The majority of the light is used in forming the image at the science camera, the aim being to obtain planar wavefronts that focus down close to the limits imposed by diffraction. This continual measurement and correction of the wavefront distortion is termed closed loop control. Within AO, MEMS technology has been used extensively in the construction of DMs and the next section details their design. Particular emphasis on the use of MEMS AO is given over to vision science as it was in this field that such technology was first used. Recently, MEMS AO has increased its breadth and the newer applications and results are described.
MEMS Adaptive Optics
a
Continuous Face Sheet
b
Segmented: Piston only
c
MEMS Deformable Mirrors MEMS has seen considerable research interest in the development of DMs and is a field of active research and development. The basic principle of a DM is simple; however, fabricating them with the required operating parameters has proven to be extremely challenging. Figure 3 shows the three basic DM types in cross section – light would be incident from the top of the figure. Continuous face sheet (Fig. 3a) – a continuous reflective mirror surface is deformed by an underlying two-dimensional array of actuators. Piston-only DMs (Fig. 3b) are composed of a discrete array of segments, each of which may individually move in a pure piston (vertical) motion only. Figure 3c shows a piston/tip/tilt DM, an individual segment may now also move in tip and tilt in addition to piston. All the DM types in Fig. 3 show piston-like actuators under the mirror surface(s) and an example actuation mechanism could be piezoelectric. However, there are other forms of actuation, for example, electrostatic, magnetic, thermal, or voice coil. MEMS-based DMs generally use electrostatic attraction; however, other forms have been employed. The first AO systems had physically large (>10 cm diameter), expensive (>$100 K), non-MEMS DMs that had been primarily designed for the military/ astronomy communities. The DM was required to provide a modest amount of stroke (5 mm) with quick response times (several kHz). In biomedical imaging the requirements are much different, (1) the DM diameter needs to be small (a few mm in diameter), hence minimizing the pupil magnification
Segmented: Piston/Tip/ Tilt
MEMS Adaptive Optics, Fig. 3 Deformable mirror (DM) types. (a) A continuous mirror surface is deformed by an underlying array of actuators. Many forms of actuation are possible, for example, piezoelectric, electrostatic, or magnetic. (b) The mirror surface is composed of a discrete array of segments each of which may move in a pure piston motion only. (c) Similar to (b) but each segment may now also move in tip and tilt in addition to piston
and allowing for the design of compact systems, (2) the DM must also be inexpensive particularly if AO is to transition into commercial products, and (3) these applications require slower response times (a few Hz), fewer actuators (or segments), but much more stroke (>50 mm) [4].
Structure of MEMS DMs The first MEMS mirror was the Texas Instruments digital micro-mirror device (DMD) introduced in the late 1990s which comprises up to one million mirror segments. It is used extensively in high-quality projection systems with each mirror segment representing one pixel in the image. However, it is bistable in nature with a low fill factor and a surface profile which precludes its use in AO. Around the same time, MEMS-based DMs designed specifically for AO applications were being introduced and they can be divided
MEMS Adaptive Optics
1309
a
M
b
Control electrodes Si chip
Al-coated membrane
Substrate PCB
Spacer
V1
V2
V3
Vn
Control voltages
Bias voltage Vb
MEMS Adaptive Optics, Fig. 4 Structure of the OKO Technologies’ bulk micromachined membrane DM. (a) A reflective mirrored membrane is positioned above a patterned set of bottom electrodes (V1,. . .,Vn). A voltage difference between these electrodes and the top transparent grounded
electrode causes the DM surface to deflect. (b) The packaged DM – up to 9 mm of deflection is possible at the center of the 15 mm DM aperture. The DM has 37 electrodes with operating voltages between 0 and 260 V (DM schematic and photograph provided courtesy of OKO Technologies)
into two broad classes: (1) bulk micromachined and (2) surface micromachined devices, the difference being in the way the mirrors are fabricated and assembled. The first MEMS DMs were bulk micromachined [5]; a few years later saw the introduction of the first surface micromachined devices [6].
A cross section through the OKO Technologies DM [5] is shown in Fig. 4. Up to 9 mm of mirror deflection is possible at the center of the 15 mm DM aperture. The DM has 37 electrodes with operating voltages between 0 and 260 V and has been used in a variety of AO applications.
Bulk Micromachined Membrane DMs (MMDM) In these electrostatic devices, a flexible, reflective membrane lies between a transparent top electrode and a patterned array of bottom electrodes [5]. Application of a voltage to the bottom electrodes causes the mirror surface to deflect downward. The DM has two principal components: • A reflective mirrored membrane: This is made from silicon nitride and an evaporated layer of aluminum or sometimes gold to provide the reflective layer. The membrane stress can be controlled during the nitride deposition. • The electrode array: The bottom set of electrodes are formed by patterning a layer of aluminum deposited onto a silicon wafer. Wafer bonding is then used to join these two sections; this has a particular advantage as it allows fabrication of devices with a large actuator gap and hence large dynamic range. However, membrane DMs are generally used for applications which require the correction of large, low spatial frequency aberrations as it is difficult to make devices with high actuator packing densities.
Surface Micromachined Continuous Surface MEMS Mirrors Another class of MEMS DMs is formed by surface micromachining, in which devices are built up from a series of deposited layers. A layer is deposited, etched before the next is added, much like a layer cake. This construction technique allows for devices with broad functionality. The various layers can be (1) polysilicon – structural and conductive properties, (2) oxides – sacrificial and dielectric, (3) nitride – insulator and dielectric, and (4) various metals – good conductors and optically reflective. With surface micromachining it is more challenging to achieve the necessary dynamic range for some applications; however, devices with extremely high actuator counts are possible. Instead of a continuous flexible membrane as in the MMDM case, these DMs allow for more localized deformations of the top mirrored surface. This permits the correction of higher order aberrations due to the ability to impart high spatial frequency patterns onto the DM. Figure 5a shows a cross section of a continuous surface MEMS DMs manufactured by Boston Micromachines Corporation (BMC). A top
M
M
1310
MEMS Adaptive Optics
b
a Attachment post
Mirror
Electrostatic actuator array
Silicon wafer
c d
Mirror Segment Bondsite
Electrodes Bimorph Flexure
MEMS Adaptive Optics, Fig. 5 Examples of surface micromachined MEMS DMs manufactured by Boston Micromachines Corp. (BMC) and Iris AO Inc. (a) Cross section of a continuous surface MEMS DM from BMC. A top metalcoated mirror surface is attached by silicon posts to a locally anchored, grounded top membrane. Applying individual voltages to the bottom actuator array causes the top mirror surface to deflect. The amount of deflection is proportional to the applied drive voltage. (b) Photograph of a gold-coated, 140 actuator, 5.5 mm stroke BMC MEMS DM. (c) The Iris AO
MEMS DM consist of a tiled array of hexagonal segments. Each segment is raised above a set of three diamond-shaped electrodes via the bimorph flexures. A high-quality mirror surface is bonded to the platform surface which also serves as the ground electrode. Application of voltages to the bottom three electrodes gives tip, tilt, and piston motion on a per segment basis. (d) A fully assembled MEMS DM from Iris AO Inc. consisting of a gold-coated tiled array of 163 segments (DM schematics and photographs provided courtesy of BMC and Iris AO Inc.)
metal-coated mirror surface is attached by silicon posts to a locally anchored, grounded top membrane. Applying individual voltages to the bottom actuator array causes the top mirror surface to deflect. The amount of deflection is proportional to the applied drive voltage. Figure 5b shows a photograph of a gold-coated 140 actuator, 5.5 mm stroke BMC MEMS DM. Another surface micromachined MEMS DM manufactured by Iris AO Inc. [7] consists of a tiled array of hexagonal segments (Fig. 5c). Each segment is
raised above a set of three diamond-shaped electrodes via the bimorph flexures. A high-quality mirror surface is bonded to the platform surface which also serves as the ground electrode. Application of voltages to the bottom three electrodes gives tip, tilt, and piston motion on a per segment basis. Figure 5d shows a fully assembled MEMS DM from Iris AO Inc. consisting of a gold-coated tiled array of 163 segments. The thin wires at the perimeter of the chip provide the voltage connections.
MEMS Adaptive Optics
1311
AO for Vision Science
M
Vitreous gel Iris
A research area where the acceptance and implementation of MEMS devices has been extremely rapid is in vision science. The need for low cost and physically small devices has made MEMS particularly attractive. Structure of the Eye Vision is the primary human sense with over half of the cortex being used to process the signals coming from the visual pathways. Furthermore, the human eye is the only organ allowing a direct in vivo view of the human nervous system. The eye is therefore the subject of intense interest as a means to the early detection of a host of retinal and possibly systemic diseases. The human eye is a complex optical structure (Fig. 6); incident light is focused by the lens and cornea onto the light-sensitive retinal surface. The primary sources of aberration are the cornea and lens, with the aberration magnitude increasing with larger pupil diameters. Upon reaching the retinal surface, the light traverses many layers before being incident on the light-sensitive photoreceptor cells (rods and cones) where the photons are absorbed. The macula or foveal region is where the cone photoreceptors have the smallest diameter (1.9–3.4 mm) and where one has the highest visual acuity. The signals from these photoreceptors are then processed by many intervening cell types before exiting the retina via the ganglion cells and the optic nerve. For detailed information of the physiology of the eye, the reader is referred to the references [8, 9]. The ability to visualize the individual retina layers and cells is particularly important as various retinal diseases manifest themselves in different retinal layers. Glaucoma, for example, attacks the retinal ganglion cells leading to a thinning of the nerve fiber layer. Other diseases such as retinitis pigmentosa (RP) and various cone/rod dystrophies manifest themselves in the photoreceptors. Age-related macular degeneration (AMD) first shows in the deeper retinal layers, with thickening of Bruch’s membrane and/or drusen between the retinal pigment epithelium (RPE) and the underlying membrane. Aberrations Both the lens and cornea introduce low- and high-order spatial aberrations [3], which along with temporal instabilities (accommodation, eye and head
Optic nerve Cornea Macula
Pupil
Fovea
LIGHT Lens
Retina
Iris
MEMS Adaptive Optics, Fig. 6 The structure of the human eye. The primary sources of aberration are the cornea and lens, the aberration magnitude increases with larger pupil diameters. The incident light propagates to the retinal surface where it is sampled by the rod and cone photoreceptors. The photoreceptors are a few microns in diameter and are unresolvable without the application of AO. Credit: National Eye Institute, National Institutes of Health
movements) limit the ability to obtain high-resolution, in vivo images of retinal cells. These aberrations also limit an individual’s visual perception of everyday scenes. Spectacles, contact lenses, or refractive surgery, for example, only provide a low-order, static correction of these aberrations, primarily focus and astigmatism. Theoretically, the lateral (transverse) resolution of the eye is given by (1): r ¼ 1:22 f l=nD
(1)
where r is the distance from the center of the Airy disk to the first minima, f is the focal length of the reduced eye, l is the imaging wavelength, n is the refractive index, and D is the pupil diameter. Hence, the highest resolution would be achieved using the shortest wavelength, l and the largest possible aperture, D ( f being fixed). For the human eye, with D ¼ 8 mm, f ¼ 22.2 mm, n 1.33, and imaging at l ¼ 550 nm, one could theoretically achieve a resolution on the order of 1.4 mm, comparable to the size of the smallest photoreceptors. Foveal cones have center-to-center spacing ranging from 1.9 to 3.4 mm, whereas for rods, the range is 2.2–3.0 mm. However, uncorrected aberrations limit the lateral (transverse) resolution to approximately 10 mm and hence these small structures would be unresolvable. Moreover, the aberration magnitudes change with time and hence a static correction such as spectacles or
M
M
1312
contact lenses would not be optimal. Even with static correction of the higher order aberrations, the temporal changes can reduce the retinal image contrast by 33% and the Strehl ratio (SR) by a factor of 3 [3]. The SR is defined as the ratio of the peak intensity in the aberrated point spread function (PSF) to that of the unaberrated case, a SR greater than 0.8 being considered diffraction limited. AO corrects both the low- and high-order aberrations in real time and has been applied to various ophthalmic imaging modalities: flash (flood) fundus cameras, confocal laser scanning ophthalmoscopes (cSLOs), and optical coherence tomography (OCT). To date, AO imaging has provided imagery of several important cell types, namely, cone, rod, RPE, and leukocyte cells in the intact, living human eye – in all cases without the use of any contrast-enhancing agents. Brief History of AO in the Human Eye In 1989, Dreher et al. [10] used a 13 electrode membrane DM to provide a static correction of astigmatism for use in conjunction with a scanning laser ophthalmoscope. Several years later, Liang et al. [11] implemented the Hartmann-Shack wavefront sensor (HS-WFS) for measurement of the human wavefront aberration. The same author [12] was then the first to combine a HS-WFS and a DM into one system for the active correction of the higher order ocular aberrations. Today, these systems use a variety of DM technologies with MEMS being just one subset. The next section details the application of AO to these cameras, highlighting the use of MEMS technology where appropriate. Retinal Imaging Modalities Flood-Illuminated (Flash) AO Fundus Cameras
Figure 7 shows the basic operation of a floodilluminated (flash) AO fundus camera. Such systems take short exposure (4–6 ms) images of the retinal surface; longer exposures can introduce blurring due to eye movements. Two light sources are used: a superluminescent diode (SLD) in the near IR is used to measure the ocular aberration. A small (1 mm diameter) SLD beam (shown in gray) is incident on the cornea and focuses to a 10 mm diameter spot on the retina. Small diameter beams are used to avoid blurring from the lens and cornea. The SLD light is scattered by the retina and the reflected light exits through the dilated pupil (typically 6–7 mm in
MEMS Adaptive Optics
diameter). This light is redirected by the DM and the dichroic beamsplitter into the WFS. The aberrations are measured and the DM commanded to take on the conjugate phase profile. Once the aberrations have been corrected to a suitable level (typically 1,000 Superior 0.8
Other advantages provided by flip-chip bonding include better electrical performance, shorter interconnection length, lower inductance effect, better electromagnetic shielding effect, self-alignment ability, and lower cost [8–11]. A detail comparison among wire bonding, tapeautomated bonding, and flip-chip bonding is listed in Table 1. Vertical probe cards are indispensable for testing flip chip with matrix pads due to limited space of fine pitch among pads to accommodate too many horizontal cantilever probes. At present, FormFactor Company catches the largest market share in the probe
MEMS High-Density Probe Cards
1331
M
card industry. In addition, a famous vertical probe, named as Cobra, invented by IBM was initiated in late 1970s for testing C4 (Controlled Collapse Chip Connection – which is also associated with flip-chip package) process base semiconductor devices. MEMS Technology for Matrix Pads Most of the traditional methods of vertical probe production are completed through punching or drawing; thus, it is not easy to control the radius of the curves of the probe. It is then followed up with assemble work which is particularly difficult to control its accuracy. The traditional probe card has the following disadvantages: 1. There is poor coplanarity of probes due to manual assembly, which causes defective contacts that lead to excessive contact resistance and even fail conduction. 2. It is not easy to adjust the accurate position of each probe; thus, it must go back to the factory for maintenance after every period of using time. MEMS vertical probe card employs MEMS technology with electroforming alloy technology to produce spring-like probe, which enables the probe to have greater degree of compliance, and provides adequate vertical deformation (Fig. 4). For the purpose of convenience and alignment in the probe assembly process, a guide plate is indispensable (Fig. 5). The guide plate can be fabricated by using the lithography, electroforming, and DLC coating processes to complete the definition of perfect spacing between each guide hole. The insulated guide plate is then combined with the technology of packaging to produce high precision probe card. In this way, it can reduce complexity in the assembly process with accuracy at the same time. Besides importing the technology of MEMS to produce probe card, it can also use the technology of CMOS-MEMS to design the probe’s arrangement and force sensor. It can improve the issue of accuracy control for many traditional probe cards. The standard CMOS process, such as TSMC 0.35 um or 0.18 um process can provide flexibility of designing microstructure with diversification and low development cost. The main advantages are summarized in the following points: 1. The accuracy and the coplanarity can be improved by using the technology of multiple probe packaging [12, 13].
MEMS High-Density Probe Cards, Fig. 4 Schematic of packaged with the probe photoresist
M
MEMS High-Density Probe Cards, Fig. 5 Schematic of single row of the probe assembly to the guide
2. The FEA software used to analyze design can optimize overdrive and vertical press force to reduce contact resistance and horizontal force that scrub off oxide layer on the surface of the pad. 3. The planar or 2-D manufacturing process can produce long and thin probe with different width [12, 13]. 4. Using lithography and the electroforming technology can enhance the capacity of batch production [14]. 5. The multiple probe packaging with real-time detection of force sensor can decrease the recovering or repair time of probe, and reduce cost and labor [15–18]. As the production of probe becomes smaller, it is very important to know the characteristic of the probe to find out the maximum current and resistivity that
M
1332
a probe can withstand. Based on the study [12], the result after number of measurements was that the probe maximum current was between 700 and 850 mA and the resistance was 35 mO-cm or less. This probe composes of nickel cobalt alloy, and the result of EDX quantitative ratio analysis of components was Co 28% and Ni 70%. The nano-indentation system had evenly gathered ten testing points in the testing chip, which yielded Young’s modulus of 180 GPa for nickel and cobalt alloys, and hardness of HV 450.
MEMS Probe Evolution In 1995, Mark Beiley et al. have used thin film structure with importing gas as a way for the probe to apply force, in order to achieve the concept of reduction of contact resistance [19, 20]. In 1997 and 1999, Yanwei Zhang et al. have used technology of MEMS to produce circular form of bimetallic microprobe which is working as a thermal actuator [21, 22]. In 2000, Ito Takahiro et al. have used SOI wafer technology to produce probe with arch-shaped beam [23]. In 2000, Dong-Seok Lee et al. have also used SOI chip and bulk micromachining technology to produce microprobe structure that the probe tip could move up to 80 mm. In 2001, Robert B. Marcus et al. brought up a new technology, which was to utilize the residual stress of bimetallic (Cu, Cr) that worked with sacrificial layer etching, and then, after annealing treatment, it would produce a warping cantilever [24]. In 2002, Bong-Hwan Kim et al. used monocrystalline silicon (100) to produce structure of three-dimensional microprobes [25, 26]. In 2002, Kenichi Kataoka et al. used electroforming nickel process that worked with sacrificial layer etching and utilized residual stress to form a warping cantilever. In 2003, Younghak Cho et al. used properties of anisotropic etching for KOH to etch an inverted triangle concave hole in the monocrystalline silicon, and then used deposition method to produce metal (Cr, W, Au) probe tip [27]. In 2004, K. Kataoka brought up electroformed multilayer film with three-dimensional S-shaped spring of microprobe arrangement [28]. In 2005, Si-Hyung Lee and Bruce C. Kim took advantages of two different residual stress relationships between BCB and Mo layers to produce a warp probe
MEMS High-Density Probe Cards
beam for wafer-level testing. [29]. In 2005, YoungMin Kim used KOH to etch silicon wafer and annealing process to produce microprobe [30]. In 2007, T. Namazu et al. used nickel titanium memory alloy to produce micro-actuator MEMS probe card [31]. In 2007, Seong-Hun Choe et al. used doping of P++ to silicon to make cantilever probe [32–34]. In 2008, Fei Wan et al. used wet etching and microelectroforming of nickel on a 4 in. wafer to provide 110,000 out-plane hoe-shaped probes for wafer-level IC testing [35–37]. Then all the probes were transferred onto the corresponding solder balls of the adaptor board. The contact force of the probe is 7.5 mN with deformation of 20 mm, the pitch of the probe is X-160 mm, Y-240 mm, and the contact resistance is 0.8 O. In 2008, J. Liu et al. used the technology of CMOS process, MEMS etching, and microelectroforming of nickel to produce thermoelectric actuating cantilever probe [38]. The probe is to test the memory components, and has contact resistance of 10 O and the power consumption of 3.7 mW. In 2009, Fei Wang et al. used bulk micromachining technology as cantilever arm of the probe, while using the concept of TSV to produce tip of the probe [39–41]. It was used in wafer-level IC testing, of which the displacement was about 20–33 mm, the contacting force of 50–80 mN, contact resistance of 1 O. In 2009, Bong-Hwan Kim used bulk micromachining technology and electroformed nickel and cobalt alloy material to produce MEMS probe used in testing DRAM [42–45]. The probe had contacting force of 20 mN at deformation of 50 mm and resulted in contact resistance of 3.5 O.
Summary The probe card produced by using CMOS process with MEMS process [46–49] has become indispensable testing components in the next generation of semiconductor (Fig. 6). It can be used in the practice of high frequency, wafer-level testing, and batch of production, which can reduce the cost to meet the trend. Regardless of how the semiconductor’s technology evolve, the probe card produced by CMOS process with MEMS technology can work effectively with the IC pad size and space, especially the circuits are next to the back end of the probe, so it can reduce the
MEMS High-Density Probe Cards
1333
M
MEMS High-Density Probe Cards, Fig. 6 SEM images of the CMOS-MEMS probe chip after ICP dry etching: (a) a 106-mm-tall probe tip with spiral cantilever, (b) a 106mm-tall probe tip with meander cantilever, (c) a 42.4mm-tall probe tip with spiral cantilever, and (d) the chip with a 14.8-mm-tall probe tip
transmission losses and noises. This is a technology to realize system integration on a chip (System On Chip), which cannot be reached with the traditional assembled probe. Using CMOS technology not only significantly reduces the size of the probe, but it can also connect the external components with the circuit, which are integrated in the design of CMOS chip. Thus, it can increase the convenience of wiring that is the space transformer needed by the probe card, which can be achieved using the metallic layer in the CMOS process. The process is not only simple and without traditional assemble method, but also enhances yield rate; therefore, the cost can be significantly decreased. This will be a great support for the present and future generation of testing IC. While the IC designer implements the IC’s layout, he may also provide with the circuit layout that is required for testing IC based on the technology of the built-in testing circuits on the CMOS-MEMS probe card. These can be processes together to save cost and accelerate time for product launch at market in order to achieve ultimate technology in “Using IC to test IC.” Especially when the pilot run is completed for mass production, the use of CMOS-MEMS probe card can accelerate time on product to be in the market.
References 1. Halbo, L., Ohlckers, P.: Electronic components, packaging and production (University of Oslo, Oslo, 1995) ISBN 82-992193-2-9 2. http://www.mpi.com.tw/. 2011/10/06 3. Jittinorasett, S.: UBM formation on single die/dice for flip chip applications. Thesis of Virginia Polytechnic Institute and State University, United States, pp. 5–6 (1999) 4. Lee, S., Guo, Y.X., Ong, C.K.: Electromigration effect on Cu-pillar (Sn) bumps. In: IEEE electronics package, technical conference, Singapore, 135 (2005) 5. Riley, G.A.: Introduction to Flip Chip: What, Why, How. Oct 2000. http://www.flipchips.com/tutorial01.html 2011/ 10/06 6. Riley, G.A.: Picking Your Flipchip: A Comparison of Flipchip Methods and Benefits, Electronics Manufacturing Engineering, Vol. 15, No. 3, Third Quarter, pp. 1–4 (2000) 7. http://www.esrf.fr/events/conferences/past-conferences-andworkshops/IWORID7/FinalProgramme/volpert_1.pdf 8. http://www.flipchips.com/SMErev01.htm 9. Chang, C.S., Oscilowski, A., Bracken, R.C.: Future challenges in electronics package. IEEE Circuits and Devices Magazine 14(2), 45–54 (1998) 10. Jang, S.Y., Paik, K.W.: Comparison of electroplated eutectic Bi/Sn and 9. Pb/Sn solder bumps on various UBM systems. IEEE Trans. Electron. Packag. Manuf. 24, 269 (2001) 11. Wolf, M.J., Engelmann, G., Dietrich, L., Reichl, H.: Flip chip bumping technology-status and update. J Nucl. Instrum. Method. Phys. Res. 565, 291 (2006) 12. Huang, J.T., Lee, K.Y., Wu, C.S., Lin, C.Y., Shih, S.H.: Using micro-electroforming and micro-assembly
M
M 13.
14.
15.
16.
17.
18.
19.
20.
21.
22. 23.
24.
25.
26. 27.
28.
1334
technology to fabricate vertical probe card. In: The 8th International Conference on Electronic Materials and Packaging, Hong Kong, pp. 435–439 (2006) Huang, J.T., Wu, C.S., Lee, K.Y., Hsu, H.J., Chao, P.S., Shin, S.H.: Batch mode fabrication on high coplanarity with pressure sensors by using MEMS. In: The 4th International Conference on Materials for Advanced Technologies, Singapore, pp. 84–87 (2007) Huang, J.T., Hsu, H.J., Chao, P.S., Lee, K.Y., Wu, C.S., Shih, S.H., Lin M.Z., Lee, F.Y., Lan, Z.C.: Fabrication of a MEMS-Based cobra probe. In: IEEE International Conference on Electron Devices and Solid-State Circuits, pp. 809–812 (2007) Lee, K.Y., Huang, J.T., Hsu, H.J., Chiu, M.C., Tsai, T.C., Chen, C.K.: CMOS-MEMS piezoresistive force sensor with scanning signal process circuit for vertical probe card. Sens. Actuator A 160, 22–28 (2010) Huang, J.T., Chiu, M.C., Lee, K.Y., Wu, C.S., Hsu, H.J., Chao, P.S.: Development of CMOS process compatible force sensor and its application to probe card. In: IEEE International Conference on Electron Devices and SolidState Circuits, Taiwan, pp. 817–820 (2007) Huang, J.T., Lee, K.Y., Chiu, M.C., Hsu, H.J., Shih, S.H.: The integration of CMOS process compatible force sensor and scanning signal process circuit for vertical probe card. In: International Conference on Electronic Packag, Tokyo, pp. 346–350 (2008) Lee, K.Y., Huang, J.T., Hsu, H.J., Chiu, M.C., Tsai, T.C.: CMOS-MEMS piezoresistive force sensor for vertical probe card. In: The 22nd International Microprocesses and Nanotechnology Conferences, pp. 420–421 (2008) Beiley, M., Leung, J., Wong, S.S.: A micromachined array probe card-fabrication process. IEEE Trans. Compon. Packag. Manuf. Technol. Part B 18, 179–183 (1995) Beiley, M., Leung, J.: A micromachined array probe cardcharacterization. IEEE Trans. Compon. Packag. Manuf. Technol. Part B 15, 84–191 (1995) Zhang, Y., Worsham, D., Morrow, D., Marcus, R.B.: A new MEMS wafer probe card. In: MEMS Conference, pp. 395– 399 (1997) Zhang, Y.: Thermally actuated microprobes for a new wafer probe card. J MEMS 8, 43–49 (1999) Takahiro, I., Sawada, R., Higurashi, E.: Fabrication of micro IC probe for LSI testing. Sens. Actuator. A 80, 126–131 (2000) Marcus, R.B.: A new coiled microspring contact technology. In: Electronic Components and Technical Conference, Orlando, Florida, pp. 1227–1232 (2001) Kim, B.H., Park, S.J.: Fabrication of nickel electroplated Cantilever-type MEMS probe card with Through-hole interconnection. In: Microprocesses and Nanotechnology Conference, Tokyo, pp. 170–171 (2003) Kim, B.H., Park, S.: A novel MEMS silicon probe card. In: MEMS Conference, Las Vegas, Nevada, pp. 368–371 (2002) Cho, Y., Kuki, T.: Si-based micro probe card with sharp knife-edged tips combined metal deposition. In: Actuators and Microsystems International Conference, Boston, vol 1, pp. 774–777 (2003) Kataoka, K., Itoh, T.: Multi-layer electroplated micro-spring array for MEMS probe card. In: MEMS Conference, pp. 733–736 (2004)
MEMS High-Density Probe Cards 29. Lee, S.H., Kim, B.C.: Curled micro-cantilevers using benzocyclobutene polymer and Mo for wafer level probing. Sens. Actuator. A 121, 472–479 (2005) 30. Kim, Y.M., Yoon, H.C.: Silicon Micro-probe card using porous silicon micromachining technology. J ETRI 27, 433–438 (2007) 31. Namazu, T., Tashiro, Y., Inoue, S.: Ti–Ni shape memory alloy film-actuated microstructures for a MEMS probe card. J Micromech. Microeng. 17, 154–162 (2007) 32. Choe, S.H., Fujimoto, S., Tanaka, S., Esashi, M.: A matched expansion probe card for high temperature LSI testing. In: The International Conference on Solid-state Sensors, Actuators and Microsystems, Seoul, Korea, pp. 1259–1262 (2005) 33. Choe, S.H., Tanaka, S., Esashi, M.: A matched expansion MEMS probe card with low CTE LTCC substrate. In: Proceedings. International Test Conference, Santa Clara, California, 4437618 34. Tanaka, S., Fujimoto, S., Ito, O., Choe, S.H., Esashi, M.: Fabrication of microprobes on a ultrathick glass substrate with narrow-pitch electrical feedthroughs for nextgeneration LSI burn-in tests. In: MEMS Conference, pp. 347–350 (2008) 35. Wang, F., Li, X., Feng, S.: A MEMS probe card with 2D dense-arrayed “hoe”-shaped metal tips. J Micromech. Microeng. 18, 1–8 (2008) 36. Wang, F., Li, X., Cheng, R., Yang, H., Wang, Y., Feng, S., Ge, X., Chen, T., Chen, L., Sun, L.: Two-dimensional densearrayed probe-cards with a hoe-shaped probing-tip micromachining technique. In: MEMS Conference, pp. 343–346 (2008) 37. Wang, F., Cheng, R., Li, X.: MEMS vertical probe cards with ultra densely arrayed metal probes for wafer-level IC testing. J Microelectromechan. syst. 18, 933–941 (2009) 38. Liu, J., Noman, M., Bain, J.A., Schlesinger, T.E., Fedder, G.K.: CMOS-MEMS probes for reconfigurable IC’s. In: MEMS Conference, Tucson, Arizona, pp. 517–518 (2008) 39. Wang, F., Li, X., Wang, Y., Feng, S.: Simultaneous formation of through wafer electrical interconnects and highly dense & uniform nickel tips for silicon-cantilever probecards. In: Transducers and Eurosensors – 4th International Conference on Solid-State Sensors, Actuators and Microsystems, 4300567, pp. 2051–2054 (2007) 40. Wang, F., Li, X., Feng, S.: Microcantilever probe cards with silicon and nickel composite micromachining technique for wafer-level burn-In testing. IEEE Trans. Adv. Packag 32, 468–477 (2009) 41. Wang, F., Li, X., Cheng, R., Jiang, K., Feng, S.: Silicon cantilever arrays with by-pass metal through-silicon-via (TSV) tips for micromachined IC testing probe cards. Microelectron. Eng. 86, 2211–2216 (2009) 42. Kim, B.H., Kim, J.B.: Design and fabrication of a highly manufacturable MEMS probe card for high speed testing. J Micromech. Microeng. 18, 075031 (2008) 43. Kim, B.H., Park, B.J., Kim, J.B.: Process effects of double step DRIE and Ni–Co electroplating for a trench-type cantilever probe for a fine-pitched MEMS probe card. Sens. Actuator. A: Phys. 152, 252–260 (2009) 44. Kim, B.H., Kim, J.B.: Fabrication of a high aspect ratio thick silicon wafer mold and electroplating using flipchip bonding for MEMS applications. J Micromech. Microeng. 19, 065024 (2009)
MEMS Neural Probes 45. Kim, B.H., Kim, J.B. Kim, J.H.: A highly manufacturable large area array MEMS probe card using electroplating and flipchip bonding. IEEE Trans. Ind. Electron. 56, 1079–1085 (2009) 46. Huang, J.T., Lee, K.Y., Hsu, H.J., Wu, R.G., Tsai, T.C., Chen, C.K.: Fabrication technology of microelectromechanical systems probe chip compatible with standard complementary metal-oxide-semiconductor process. J. Appl. Phys. 49, 06GN021 (2010) 47. Lee, K.Y., Huang, J.T., Hsu, H.J., Wu, R.G., Tsai, T.C.: Fabricating MEMS probe card compatible with CMOSMEMS process. In: The 22nd International Microprocesses and Nanotechnology Conference, Sapporo, pp. 88–89 (2009) 48. Lee, K.Y., Huang, J.T., Hsu, H.J., Chen, C.K., Tsai, T.C.: Fabrication technology of CMOS-MEMS probe chip compatible with electroless nickel plating process. In: The 5th International Microsystems, Packaging, Assembly and Circuits Technical Confernce, Taipei, (2010) 49. Lee, K.Y., Huang, J.T., Hsu, H.J., Chen, C.K., Tsai, T.C.: Integration via structure with electroless nickel plating process to fabricate CMOS-MEMS probe chip. In: The 23nd International Microprocesses and Nanotechnology Conference (2010)
MEMS Microphone ▶ Electrostatic MEMS Microphones
MEMS Neural Probes Jit Muthuswamy1 and Murat Okandan2 1 School of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ, USA 2 Advanced MEMS and Novel Silicon Technologies, Sandia National Laboratories, Albuquerque, NM, USA
Synonyms Brain implants; Microelectrodes; Neural interfaces; Prostheses
Definitions MEMS neural probes are microscale implants for the brain or peripheral nervous system that use microelectromechanical systems to assess function
1335
M
or stimulate activity in specific regions of the brain, ensembles of neurons, or neuronal fibers.
Brain Monitoring Technologies Hypotheses regarding brain function or dysfunction are generally verified and confirmed using accurate measurements of brain activity at different scales – from whole brain regions to single cells or neurons and subcellular structures such as synapses, ion channels, etc. Such measurements are made using a variety of different techniques, each with their own spatial and temporal resolution. The technologies are generally invasive or noninvasive; the noninvasive technologies include a whole spectrum of imaging modalities as illustrated in Fig. 1, electroencephalograms (EEG) from the scalp, etc. The invasive technologies typically involve either electrodes placed on the brain (such as in electrocorticogram or ECoG) or microelectrodes implanted in specific brain regions of interest for obtaining electrical activity from the brain. This may be in the form of field potentials from ensembles of neurons or from extracellular action potentials of single or a few neurons in the vicinity of the microelectrode. Generally, electrical activity from single neurons has been a preferred currency for testing hypotheses about brain function. Over the last 10 years, exciting applications such as cortical prostheses have also been developed using electrical activity from single neurons in motor and premotor cortices of the brain. Success in this effort will lead to a wider application of this theme in other brain prosthetic applications such as memory and cognitive prostheses. Neural probes of different kinds have been developed for a variety of applications such as electrical stimulation of specific neurons in deep brain stimulation (DBS) therapy, monitoring neurotransmitters and other neurochemicals, and delivering drugs. However, currently microelectromechanical systems (MEMS)-based technologies are primarily used for monitoring electrical activity from single neurons in mainstream research and emerging prosthetic technologies. Therefore, this chapter here will focus on neural probes used for monitoring the electrical activity of single neurons. Implantable technologies to monitor electrical activity from single neurons have existed for over six decades. The earliest single neuronal recordings
M
1336
MEMS Neural Probes, Fig. 1 Illustration of the spatiotemporal resolution of current brain monitoring technologies (Courtesy: Jack Judy, program manager, DARPA (MTO), approved for public release; distribution is unlimited)
MEMS Neural Probes 10 μs 100 μs
Temporal Resolution
M
1 ms 10 ms 100 ms 1s 10 s 100 s 10 cm
1 cm
1 mm
100 μm
10 μm
1 μm
Spatial Resolution Approved for public release; distribution is unlimited.
*Longevity of Monitoring Technique
consisted of manually fabricated glass micropipettes that tapered down to few microns at one end. These pipettes were backfilled with electrolytes and a conducting wire was inserted. The tapered end was stereotactically positioned in the brain after a craniotomy in the vicinity of neurons of interest. They were initially used to record intracellular membrane potential changes [1, 2]. Insulated tungsten or platinum microelectrodes were used for recording extracellular activity [3, 4]. One of the first attempts to use such cumbersome manual approaches to obtain long-term recordings from animals was Strumwasser’s bold effort in 1958 [5] to record activity of specific neurons in hibernating California ground squirrels. This was followed by several decades during which bundles of metal microwires in different configurations were manually assembled along with interconnects to record from ensembles of single neurons in the brain. The microwires were insulated along the length, and the insulation at the tip of the microelectrode was etched a few tens of microns to allow an electrical interface with single neurons. The length (and consequently the electrical impedance) of the uninsulated tip or the size of the recording site determined the number of neurons that would be recorded. Microelectrodes developed using MEMS-based approaches have been a relatively recent development and have significantly evolved in complexity over the last 25 years. The development of the first neural probe using microfabrication technologies in 1970 [6]
demonstrated the promise of MEMS-based systems to make neural probes more usable for uninitiated researchers. Using a manufacturing process that readily allowed batch fabrication, they dramatically improved scalability in terms of the number of neurons that could be recorded and allowed possibilities for integrating multiple functionalities. Previously, interconnects were manually soldered or glued. MEMS technology improved interconnect reliability. It also significantly reduced the overall form factor, which allowed animals to behave naturally.
General Features of Neural Probes An individual neural probe must be slender (tens to hundreds of micrometers in diameter or in crosssectional dimensions) and several millimeters long. This will allow interrogation of neurons in the cortical and deep brain structures while minimizing trauma and displacement of brain tissue during penetration and eventual implantation. Multiple probes are often integrated to form arrays of neural probes. Some studies [7, 8] have suggested that slender probes with smaller cross-sectional areas significantly decrease the reactive response of the brain tissue to the foreign body implant. However, the lower limit for slenderness appears to be that which will allow for successful penetration without bending or buckling of the probe. In addition, a smooth tapered probe is
MEMS Neural Probes
generally preferred to minimize dimpling of the meningeal layers such as the dura mater during penetration. However, the likelihood of such tissue dimpling is also dependent on the velocity of penetration and on the number of probes constituting an array. Therefore, the design of the mechanical structure of the neural probe is constrained by requirements such as minimal tissue response, adequate stiffness for penetration, minimal brain tissue dimpling during penetration, etc. One of the features of a probe used for recording from single neurons is the presence of one or more recording sites along the length of a probe. To ensure recording from one neuron or a few selected neurons, the recording sites should be of similar geometrical scale as the neuronal cell bodies, which are typically 10– 20 mm in diameter in the mammalian brain. Ion exchange across the cell membrane during a neuronal action potential results in current flow. The current flow in a volume conductor results in lines of electrical field intensities and spatial potential gradients being established in the extracellular space. Microelectrodes in the extracellular space therefore register the spatial potentials (typically less than 1 mV) at any given point. The interface between the recording site (typically made of metals like platinum, platinum/iridium, and gold and also sometimes doped polysilicon) and the surrounding ion-rich extracellular media usually determines the overall electrical impedance of the probe. This impedance is with respect to a reference electrode remotely placed on or in the brain or on the neck muscles, depending on the application and the type of recording desired. The interface impedance is largely capacitive due to the spontaneous formation of an ionic double layer on the probe recording site surface upon insertion. However, faradaic reactions take place at the recording site and give rise to a parallel pathway for charge transfer. This can be modeled as a resistive pathway in parallel to the double-layer capacitance. Given the above general mechanical and electrical features of neural probes, two broad classes of MEMS micromachining techniques have been used successfully to fabricate slender neural probes and arrays of neural probes out of silicon wafers. These are bulk micromachining and surface micromachining. Silicon has been a preferred substrate material for a variety of reasons: (a) well-established micromachining methods, tools, and recipes are readily available, (b) silicon lends itself readily to subsequent integration of circuitry and other functionalities, and (c) silicon has been shown to
1337
M
be nonreactive or inert as an implant material [9], eliciting minimal foreign body response. Directional, material, and dopant-dependent selectivities of anisotropic etching are used in both of the above classes to machine microscale mechanical structures in the form of slender neural probes. Besides these classes, MEMS neural probes have also been fabricated out of nonconductive polymers spun coat on micromachined silicon or from metal molds. A wide variety of MEMS neural probes have been developed and tested. A few representative technologies that are broadly distinct and have been tested in long-term animal experiments are described here.
Bulk Micromachining of MEMS Neural Probes The first MEMS-based neural probes were microscale shanks of silicon with metal recording sites along the length [10]. They were fabricated by bulk micromachining of silicon using anisotropic etching techniques. A boron-doped silicon substrate acted as an etch stop, and shanks of silicon 8–15 mm in thickness, 75–100 mm in width at the base that narrowed down to 25 mm, and 1.5–3 mm in length were machined in the initial versions. Each shank was capable of hosting multiple recording sites of gold with polysilicon or tantalum conductors. Currently, these probes are fabricated and distributed by Neuronexus Inc., Ann Arbor, MI. Another widely used silicon MEMS technology using bulk micromachining is a 3D array of 10 10 microelectrodes that was first reported in 1991 by the University of Utah [11]. Aluminum (5 mm thick) was deposited on the polished surface of silicon wafers (1.7 mm thick) and then patterned into squares (325-mm sides and 400-mm separation between the centers of adjacent squares). A temperature gradient between the two sides of the wafers created a thermal migration of the eutectic droplets of aluminum and silicon with silicon recrystallizing behind the eutectic droplet trail. The net result was the creation of highly conductive trails of doped (p+) silicon within the bulk silicon wafer. Chemical and mechanical polishing of the wafer resulted in pillars of p+ doped silicon (approximately 1.5 mm long) isolated from its immediate neighbors by back-to-back p-n junctions that provided the isolation. The pillar tips were finally coated with
M
M
1338
platinum and insulated with polyimide after establishing interconnects at the back end. Bulk micromachining approaches have also been used successfully with other materials such as ceramic and diamond.
Surface Micromachining of MEMS Neural Probes Surface micromachining allows the fabrication of neural probes that are generally thinner (2–5 mm thick) than typical bulk micromachined probes (>8–15 mm thick). There is no evidence yet that a reduction in one of the dimensions of the neural probes leads to an improvement in performance. However, a significant reduction in the cross-sectional area of the neural probe has been shown to significantly improve the inflammatory and foreign body reaction of the tissue immediately surrounding the neural probe [7, 8]. There are other significant and unique advantages of surface micromachining over bulk micromachining such as the ability to integrate (a) complex mechanical structures such as microactuators, gears, etc. on the same die and (b) signal conditioning and control circuitry to significantly enhance the functionality of neural probes. Complex mechanical structures such as microactuators and gears can potentially enhance the reliability and functionality of the neural probe in the development of autonomous neural probes. These are capable of adapting their position in the brain to compensate for any change in the quality of neuronal signal being recorded [12]. Such movable neural probes enhance the ability (a) to obtain well-isolated and stable single neuronal activity with high signal-tonoise ratio over longer durations, (b) to carefully and unambiguously monitor changes in the underlying neurobiology, (c) to monitor neurons and neuronal tracts of specific functional interest, and (d) potentially to enhance the reliability of brain prosthetic devices that require reliable operation over the life of a patient. Innovative approaches have been used to develop neural probes using surface micromachining process. One involved the fabrication of a 3D array of polyimide pillars (1.5 mm long) with metal (titanium on aluminum) traces for three recording sites on each pillar, as opposed to the single recording site that was found in the Utah microelectrode array [13]. Polyimide film (10 mm) was spun coat over a 5-mm-thick film of Nickel electrodeposited on silicon. Nickel acted as a backing
MEMS Neural Probes
metal for the polyimide film. After patterning the metal interconnects and recording sites on the polyimide film, a second layer of polyimide (also 10 mm) was spun coat on top. Finally, and most interestingly, an externally applied magnetic field was used to lift the nickel backing layer and, along with it, the polyimide films. Similar surface micromachining approaches have been used to fabricate slender neural probes from other polymers such as benzocyclobutene (BCB). One of the key challenges in using surface micromachining approaches to fabricate neural probes is to ensure that there will be no residual stress in the thin films that constitute the substrate of the neural probe. Simultaneously, the thin films will also need sufficient stiffness to penetrate the outer meningeal layers and the brain itself and travel through the brain tissue without significant bending, buckling, or breaking. The five-layer polysilicon SUMMiTVTM (Sandia’s ultra-planar multi-layer micromachining technology) process of Sandia National Laboratories, Albuquerque, NM, has been extremely useful in fabricating complex, surface-micromachined, movable neural probes and has been tested successfully in long-term animal experiments. Details of this process are explained in the remaining sections of this entry.
SUMMiTTM Surface Micromachining Process A brief summary of the SUMMiT process flow is presented here; more detailed process description can be found elsewhere [14]. An example of a multilayer mechanical structure that can be formed in this process is shown in Fig. 2a, a gear with close-tolerance hub and four mechanical layers. SUMMiT surface micromachining process is carried out on 150-mm, (100), 2–20-Ω cm, n-type silicon wafers. First steps are thermal oxidation (0.6 mm) and low-stress silicon nitride deposition (low-pressure chemical vapor deposition – LPCVD, 0.8 mm), which form the insulating stack for the following process steps. An etch through the insulating stack allows electrical connection between polysilicon layers and the substrate. First polysilicon layer (poly-0) deposited by LPCVD is 0.3 mm thick; this layer usually serves as a ground plane, mechanical anchor, electrostatic shield, and part of the electrical interconnect. All of the polysilicon layers in the process flow are heavily n-type doped, providing the required electrical conductivity.
MEMS Neural Probes
M
1339
a
g b
poly-3
sacox-1 poly-0 nitride oxide
silicon poly-1
c
h
poly-4 sacox-4
sacox-2
d
e
M
poly-2
i
f
sacox-3
MEMS Neural Probes, Fig. 2 Plan view and cross section of multilayer gear illustrating the process technology sequence for fabrication
The next layer is a 2-mm LPCVD sacrificial oxide layer (sacox-1). Two separate etch steps form the features for mechanical anchors, which are also electrical connections and dimples (Fig. 2b). The partial etch of oxide allows the dimples to be formed in the following polysilicon layer. This helps in preventing stiction by allowing relatively small areas to come into contact rather than the full surfaces that might be available in the structures. The following polysilicon layer (poly1) deposited by LPCVD is 1 mm thick. Mechanical features
in this layer can be defined by lithography and etching at this step or also by a following etch after the next polysilicon layer, which generates features that are a laminate of the two layers. Close tolerances needed for forming hubs, captive mechanical features, sliding seals, and pressure/flow sections are formed by deposition and patterning of a conformal, thin (0.5 mm) silicon dioxide layer (sacox-2) after an isotropic oxide etch through lithographically defined and etched regions in poly-1 (Fig. 2c–d). Poly-2 layer (1.5 mm) is then
M
1340
deposited and patterned. In areas where sacox-2 was remaining, a close-fitting and mechanically free linkage is formed between poly-1 and poly-2 layers (Fig. 2e). Sacox-3 is then deposited and chemically and mechanically polished (CMP), which removes the topography that was a result of the preceding etch and deposition steps. The final thickness of sacox-3 above poly-2 is 2 mm. The dimple and anchor etch steps prepare the dimples and mechanical connections between poly-3 and poly-2 (Fig. 2f). Poly-3 is then deposited (2.25 mm) and patterned (Fig. 2g). Then, oxide layer (sacox-4) is deposited by LPCVD and planarized with a CMP step to a thickness of 2 mm above poly-3. After the dimple and anchor etch steps, final mechanical polysilicon layer (poly-4, 2.25 mm) is deposited and patterned (Fig. 2h). An aluminum metallization layer is the last layer that is deposited and patterned on poly-4 features to enable wire-bonding and low-resistance electrical connections. The release step is performed in an HF-containing chemistry to remove all sacrificial oxide layers while preserving the polysilicon and metal features (Fig. 2i). In order to prevent stiction, a supercritical CO2 drying step is used following the release etch to displace all liquids between layers without exposure to a gas–liquid interface. The release can be performed at die level or as full wafers, depending on the design and application requirements. Further modifications of the SUMMiT process are also available, such as SwIFT [15, 16] and SFET [17, 18]. SwIFT process adds two silicon nitride layers (0.3-mm-thick nitride-1 on poly-0 and 0.8-mm-thick nitiride-2 below poly-3), a deep oxide cut (from poly3 to poly-0, filled with nitride-2), and Bosch etch through the substrate to enable the creation of insulator-lined microfluidic channels, fluidic connections, and other novel features. SFET integrates NMOS transistors in the silicon substrate, which allows preamps and addressing circuitry to be fabricated for signal conditioning and dense-matrix addressing. The neural probe in the form of a long, slender beam of doped (1021/cm3) polysilicon 7 mm long, 50 mm wide, and approximately 5 mm thick is defined in the poly-1, poly-2, poly-3, and poly-4 layers of the SUMMITVTM process. Interestingly, the nonlinear I-V characteristics of doped polysilicon has an inflection point in the same range of voltages that typically separate noise signal amplitudes (10–30 mV peak-topeak) from action-potential signal amplitudes (typically >100 mV) [19]. The above property,
MEMS Neural Probes (1)
d (2)
y x
V
(3)
Support spring
Left bank 1
Shuttle
Right bank 1
Spring anchor Right bank 2
Left bank 2
Ground Plane
MEMS Neural Probes, Fig. 3 Schematic of the comb-drive actuator and a close-up of one of the parallel plates and the associated fringe field intensity lines that make up the comb drive
y-drive
Linkarms
x-drive
MEMS Neural Probes, Fig. 4 Schematic of the x- and y-drive connected to a pin joint and a flywheel. The x- and y-drives are activated at 90 phase difference to produce a net rotational motion of the flywheel that will eventually drive a gear or the microelectrode
remarkably, results in selective suppression of noise when recording extracellular potentials from the brain [19]. The polysilicon neural probe has been demonstrated to produce significantly higher signal-to-noise ratios compared to commonly used tungsten neural probes.
MEMS Neural Probes MEMS Neural Probes, Fig. 5 (a) Schematic of a microelectrode and six electrothermal actuators associated with the microelectrode. The “Up” and “Down” actuators move the microelectrode in the center up or down, respectively. (b) SEM of one of the actuators showing the electrothermal heat strips and ratchets that move the microelectrode
1341
a
Thermal strips controlling movement
M
Recording Spring
Release/lock mechanism
Thermal strips controlling release mechanisms
Electrode
b
Microelectrode
‘Down’ actuator
MEMS Neural Probes with Electrostatic Microactuators Integrated electrostatic microactuators have been developed for moving neural probes in the brain using comb drives. Electrostatic forces due to a fringing electric field between a movable plate and stationary plate upon application of a voltage (90–110 V) between the two plates creates a movement in one of the plates. A bank of such parallel plates forming a comb drive is used to create translation (30– 40 mm) in the x-direction as shown in Fig. 3. A similar arrangement is used to create translation in the y-direction. The above two systems are driven by voltages that are 90 phase shifted, and the translations are coupled to a pin joint that rotates a flywheel as shown in Fig. 4. The direction of rotation of the flywheel is
determined by which of the two drives (x- or y-drive) is leading in phase. The flywheel is coupled to the microelectrode either directly or through gears to move the microelectrodes in steps of 1 mm over a distance of 5 mm. The above system of electrostatically actuated microelectrodes has been successfully tested in acute rodent experiments [20] but was found to be prone to mechanical failure. There were stiction issues between the parallel plates of the comb drive as well as alignment issues between the gears and the microelectrode that resulted in the microelectrode getting stuck particularly during intense behavioral routines of the rodent. Besides such serious mechanical robustness issue, the electrostatic microactuators also required high voltages (90– 110 V) and optimally designed voltage waveform shapes. Electrothermal microactuators have, on the other hand, been demonstrated to be more robust and occupy
M
M
1342
smaller footprint on the die [21, 22]. In addition, the waveforms required to drive the electrothermal actuators are 6–10 V and standard rectangular pulses that can be readily generated. At the core of electrothermal microactuators are the electrothermal strips that consist of parallel beams of doped polysilicon that expand upon the application of voltage, displacing a shuttle in the middle of the electrothermal strips as illustrated in Fig. 5. Each microelectrode was designed with one set of actuators to move the microelectrode up, one set to move the microelectrode down, two sets of microactuators to release the lock that prevents the microelectrode from moving up, and finally two sets of microactuators to release the lock that prevents the microelectrode from moving down. The above neural probes had a displacement resolution of approximately 9 mm over a total displacement capability of over 5 mm. The electrothermal microactuators and the integrated neural probes have been successfully tested in longterm rodent experiments lasting over 13 weeks [21]. However, the key challenge in the application of MEMS neural probes in brain prosthetic technologies as well as technologies to monitor single neurons for neurobiological applications has been the unreliable interface between the implanted microelectrode and the neurons in the brain tissue surrounding the microelectrode. Current technologies do not have the stability, consistency, or reliability to record from targeted single neurons for long periods of time. The MEMSbased movable microelectrodes with their flexibility to be repositioned offer a very promising approach to overcome the above challenges.
Future Directions While MEMS-based neural probes have already made a significant impact on the field of neural monitoring, the immediate interface between the neural probe and the brain tissue is a key challenge that is poorly understood. So it is important to direct future efforts toward a better understanding of the events at the interface over the lifetime of the implant. On the device side, future efforts are likely to be focused on developing technologies that will allow us to scale up the number of probes (both movable and fixed). Such high-density probes are essential for monitoring large ensembles of neurons simultaneously. Integrating wireless telemetry to the above probes will also be critical to allow
MEMS Neural Probes
monitoring from the brains of awake, behaving subjects besides enhancing the long-term reliability of the probes. Finally, integrating multiple functionalities such as the ability to perform neurochemical monitoring, deliver drugs, and stimulate the brain tissue electrically and/or optically will dramatically enhance our ability to assess brain function in a multimodal framework. Acknowledgments The authors would like to thank the Whitaker Foundation, National Institutes of Health (NIH), Arizona Biomedical Research Commission (ABRC) for supporting this research. Sandia National Laboratories is a multiprogram laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the US Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
Cross-References ▶ Basic MEMS Actuators ▶ Biosensors ▶ CMOS MEMS Biosensors ▶ Epiretinal Prosthesis ▶ MEMS Packaging ▶ Nanorobotics ▶ Nanorobotics for Bioengineering ▶ Organic Bioelectronics
References 1. Cole, K.S., Curtis, H.J.: J. Gen. Phys. 22, 649–670 (1939) 2. Ling, G., Gerard, R.W.: J. Cell. Comp. Physiol. 34, 383–396 (1949) 3. Dowben, R.M., Rose, J.E.: Science 118, 22–24 (1953) 4. Hubel, D.H.: Science 125, 549–550 (1957) 5. Strumwasser, F.: Science 127, 469–470 (1958) 6. Wise, K.D., Angell, J.B., Starr, A.: IEEE Trans. Biomed. Eng. 17, 238–247 (1970) 7. Seymour, J.P., Kipke, D.R.: Biomaterials 28, 3594–3607 (2007) 8. Stice, P., Gilletti, A., Panitch, A., Muthuswamy, J.: J. Neural. Eng. 4, 42–53 (2007) 9. Stensaas, S.S., Stensaas, L.J.: Acta Neuropathol. 41, 145–155 (1978) 10. Najafi, K., Wise, K.D., Mochizuki, T.: IEEE Trans. Electron Dev. ED-32, 1206–1211 (1985) 11. Campbell, P.K., Jones, K.E., Huber, R.J., Horch, K.W., Normann, R.A.: IEEE Trans. Biomed. Eng. 38, 758–768 (1991) 12. Muthuswamy, J., Anand, S., Sridharan, A.: Front. Neurosci. 5, 94 (2011)
MEMS on Flexible Substrates 13. Takeuchi, S., Suzuki, T., Mabuchi, K., Fujita, H.: J. Micromech. Microeng. 14, 104–107 (2004) 14. Sniegowski, J.J., de Boer, M.P.: Annu. Rev. Mater. Sci. 30, 299–333 (2000) 15. http://mems.sandia.gov/about/swift.html 16. Okandan, M., Galambos, P., Mani, S.S., Jakubczak, J.: Proceedings of SPIE 4560, p 133 (2001) 17. Okandan, M., James, C., Mani, S.S., Draper, B.L.: Proceedings of the 7th mTAS Symposium, pp 1037–1040 (2003) 18. Cowan, W.D., Resnick, P.J., Okandan, M., Spahn, O.B., Dagel, D.J., Grossetette, G.D.: Proceedings of IEEE/LEOS Conference, pp 64–65 (2006) 19. Saha, R., Jackson, N., Patel, C., Muthuswamy, J.: IEEE Trans. Neural Syst. Rehabil. Eng. 18, 489–497 (2010) 20. Muthuswamy, J., Okandan, M., Jain, T., Gilletti, A.: IEEE Trans. Biomed. Eng. 52, 1748–1755 (2005) 21. Jackson, N., Sridharan, A., Anand, S., Baker, M., Okandan, M., Muthuswamy, J.: Front. Neuroeng. 3, 10 (2010) 22. Muthuswamy, J., Okandan, M., Gilletti, A., Baker, M.S., Jain, T.: IEEE Trans. Biomed. Eng. 52, 1470–1477 (2005)
MEMS on Flexible Substrates Donald P. Butler and Zeynep Celik-Butler The Department of Electrical Engineering, The University of Texas at Arlington, Arlington, TX, USA
Synonyms MEMS ¼ microelectromechanical systems
Definition MEMS in flexible substrates can be defined as microelectromechanical devices and systems embedded in a flexible polymer or plastic film such that the device is conformal to nonplanar surfaces and objects.
Introduction Flexible electronics [1] is a rapidly emerging field that has resulted in flexible displays using organic light-emitting diodes being inserted into the marketplace in commercial products such as cell phones. Flexible circuit boards are also seeing widespread use in commercial electronics. These flexible electronic components have the advantage
1343
M
of being able to conform to nonplanar objects and surfaces. They can be bent or folded to some degree. They also provide decreased size and weight when compared to their rigid conventional electronic counterparts. The use of organic semiconductors to form transistors and diodes has so far limited the performance of the components due to the low intrinsic mobility of the materials developed so far. Current organic transistors are sufficient for the low speed switching requirement of a video display but do not possess the performance of Si or SiGe transistors that are utilized for computing, communication, and signalprocessing applications. This has resulted in hybrid systems constructed of both rigid Si components for high-performance applications and flexible organic transistors and diodes for the display. The recent development of graphene-based transistors [2] with high mobility and the continued research on other polymer materials for electronic applications may change this scenario in the near future by allowing for low-cost, high-speed electronics to be placed on or in flexible substrates. The inherent paradigm of flexible electronics would have potential beyond displays including sensing and communication electronics. Microelectromechanical (MEMS) devices exist as both sensors [3] and actuators [4]. Microelectromechanical devices were first developed by bulk micromachining where the substrate was etched to form the device structure. This process has rapidly given way to surface micromachining for many devices, where the device is formed by the subsequent deposition of layers on a substrate. Some of these layers are sacrificial layers which are removed to release parts of the device giving them the freedom to move and form the electromechanical device. Surface micromachining is independent of the choice of substrate material, hence can readily be adapted to flexible substrates made of polymers and plastics so long as low-temperature processing techniques are employed that do not exceed the glass transition temperature of the substrate. This has been a challenge in developing MEMS sensors on flexible substrates since many traditional sensor designs use polycrystalline piezoresistive materials that require high temperature processing. MEMS devices in flexible substrates have applications where the ability to conform to a nonplanar surface or the savings in size and weight are advantageous. Toward this end, MEMS sensors in
M
M
1344
MEMS on Flexible Substrates
Encapsulation
Surface with no strain R Substrate
MEMS on Flexible Substrates, Fig. 2 The structure containing the substrate, electronic layer, and the encapsulation, bent to a radius R. A similar concept will be utilized in this investigation
MEMS on Flexible Substrates, Fig. 1 Photographs of pressure/tactile sensors on flexible substrates illustrating the conformity of the devices. (Top left) A section of polyimide containing pressure sensors bent with an optical micrograph of the sensor in the background. (Top right) A section of polyimide containing pressure sensors applied to a pen. (Bottom) A closer view of the pressure sensors on the polyimide film
flexible substrates have been developed for medical and structural health-monitoring applications (Fig. 1), and also in defense. This work has been focused on MEMS sensors that measure infrared radiation, pressure/force, acceleration, strain, and other physical parameters. Traditional electronics utilize stiff and thick substrates where the films on top must conform to the substrate. Any stress that develops is mainly concentrated in the films forming the device structure on the top surface of the substrate. In the case of flexible electronics, when the substrate is bent with the circle center closer to the bottom surface, the top surface goes into tension while the bottom surface goes into compression, as shown in Fig. 2. Under compression with the circle center closer to the top surface, a film or device may buckle and delaminate from the substrate. Under tension, a film or device may tear. At the neutral plane of the stack lies a low stress plane which experiences minimal strain. The devices are best suited to be located at this low stress plane where they are supported by a substrate layer below and a superstrate or encapsulation layer above. If the
stiffness of the device layer in the middle is negligible, then the electronic layer can be made to lie on a plane of no strain as long as the following expression Ys ds2 ¼ Ye de2 is valid [5]. Here, the subscript e refers to the encapsulation while the subscript s refers to the substrate. Y is the Young’s or tensile modulus and d is the thickness of the layer. By using an encapsulation, the electronic MEMS devices will not only lie at a low stress plane but they will be protected from the environment thus removing the need for further packaging. Thus, the devices can be referred to as being self-packaged or device-level packaged [6–8] since the flexible substrate layer, encapsulation layer, and device layer are all fabricated using microfabrication techniques. This results in a MEMS device that is embedded in a flexible film or substrate. To produce a MEMS device layer in which parts of the device must be free to move, a local cavity is formed around the device by micromachining (Fig. 3). An alternative to device-level packaging the MEMS devices is to package the flexible encapsulation layer or superstrate to the substrate at the wafer level with a wafer bonder. In this entry, the reader is introduced to the development of MEMS devices on flexible substrates and their packaging through the presentation of several examples including microbolometers, pressure sensors, and accelerometers. In addition, several other examples are discussed with their relationship to developing flexible electronics that include MEMS devices.
MEMS on Flexible Substrates
1345
M
MEMS on Flexible Substrates, Fig. 3 (Left) An expanded view of a self-packaged microbolometer (not to scale: the vertical dimensions have been exaggerated and the bump bonds are placed next to the bolometer; the polyimide substrate and superstrate have been compressed for clarity).
The IR transparent aluminum oxide window and bolometer with the aluminum mirror form an optically resonant structure to maximize absorption. (Right) An optical micrograph of the completed device
Microbolometers on Flexible Substrates
by spin casting the substrate onto a Si carrier wafer and curing the polyimide at the beginning of the microfabrication process [9]. In this case, the polyimide film did not undergo distortion during the microfabrication process, allowing for precise alignment of the device features across the wafer and a corresponding high yield. Polyimide and Kapton, for that matter, are attractive materials to use due to their flexibility and relatively high glass transition temperature, up to 400 C. Polyimide is unattractive due to its moisture sensitivity, limiting its deployment in some applications. However, its material properties allowed for a good test bed to demonstrate the feasibility of the technology. New polymers are continually being developed for future application. Moreover, the evolution of high precision roll-to-roll processing equipment allows for high volume, large area fabrication of devices for commercial manufacturing [1]. The technique of utilizing a Si carrier wafer is more appropriate for the university research environment and small production lines. After spin casting and curing the polyimide substrate, a Si3N4 film was deposited to protect the substrate from future microfabrication processes and define the device plane [9]. The microbolometers were then fabricated on top of the Si3N4 film by using conventional photolithography and rf magnetron sputtering to deposit the metal electrodes and yttrium barium copper oxide (YBCO) thermometer. All sputter depositions were performed in a conventional turbo-pumped sputter system at room temperature. The highest temperature step was hard
Microbolometers are thermal optical detectors whose electrical resistance changes as a result of the heating from the absorbed optical radiation. To minimize the heat lost by the microbolometer to the substrate, the resistive thermometer is suspended above the substrate through micromachining. The detectors are also placed in vacuum to eliminate the heat loss through the surrounding atmosphere. Suspending the thermometer above the substrate also allows for the formation of an optically resonant cavity to increase absorption of the detector. Microbolometers are typically used in night vision as infrared imagers operating in the 8–14 mm atmospheric transmission window. The development of micromachined sensors on flexible substrates began with microbolometers [6, 9, 10]. There are advantages to producing curved imaging elements because the number of optical elements in the lens can be decreased, resulting in cameras that have reduced size and weight. Early work first investigated fabricating microbolometers on Kapton sheets bonded to Si carrier wafers so that conventional photolithography and semiconductor processing equipment could be used. This permitted the patterning of structures with micron precision. However thermal cycling that occurred during the microfabrication process caused distortion of the Kapton films that resulted in low device yield across the wafer [10]. The fabrication process was adapted to use polyimide (liquid Kapton, Dupont, now HD Microsystems, PI5878G) substrates that were formed
M
M
1346
MEMS on Flexible Substrates, Fig. 4 Micromachined YBCO microbolometer array fabricated on a spin-on polyimide substrate (liquid Kapton)
baking the photoresist at 105 C. This ensured that the glass transition temperature of the polyimide substrate was not exceeded. The fabrication process involved the patterning and deposition of an Al mirror that would form an optically resonant cavity with the YBCO thermometer. A photodefinable polyimide sacrificial layer was then spin coated and cured on the wafer. The polyimide sacrificial layer was designed to be approximately 2-mm thick to maximize the infrared absorption at a wavelength of 10 mm and satisfy the approximately l/4 relationship for maximum absorption in the suspended thermometer, where l is the wavelength of the incident radiation. Subsequently, Ti electrode arms were patterned and deposited, followed by Au contacts on the bond pads and to the YBCO thermometer which was deposited next. The microbolometers were patterned into 1 10 arrays of 40 mm 40 mm and 60 mm 60 mm detectors. The polyimide sacrificial layer was then removed by ashing in an oxygen plasma to suspend the microbolometers above the substrate and form the optically resonant cavity with the Al mirror. The Si3N4 film deposited initially protected the polyimide substrate from being attacked by the oxygen plasma. Some of the microbolometer die were gently peeled off the Si carrier wafer while others were left on the Si carrier wafer for characterization. An SEM micrograph of a section of a 1 10 array is shown in Fig. 4. Two microbolometer dies are applied to a fingertip in Fig. 5, demonstrating the conformal nature of the devices. The microbolometers were
MEMS on Flexible Substrates
MEMS on Flexible Substrates, Fig. 5 Microbolometer arrays fabricated on flexible polyimide substrates applied to a finger
characterized by gluing the die, both attached to and removed from the Si carrier wafer, into a conventional integrated circuit package. The packaged microbolometers were placed into an evacuated cryostat with a ZnSe window. No significant difference in the responsivity and detectivity was observed between the microbolometers that were removed from the Si carrier wafer, those that remained on the Si carrier wafer, or to ones that were fabricated directly on a Si substrate. This indicated that MEMS microbolometers can be fabricated on both rigid or flexible substrates without degradation of their performance. The infrared detectors showed broadband optical responsivity as high as 6.1 104 V/W and detectivity of 1.2 108 cm Hz1/2/W at 1 mA of current bias [9]. This work also demonstrated that, in the general case, there is no need to fabricate sensors on flexible substrates if they are going to be placed in rigid, conventional integrated circuit packages. Such packaging defeats the ability of the devices to conform to nonplanar surfaces. Therefore, as a next step, device-level packaging of the microbolometers was developed [6]. The package for the microbolometer was now created during the microfabrication process, with a local vacuum cavity surrounding the microbolometer to minimize its thermal conductance with its surroundings. A schematic and optical micrograph (top view) of the device-level vacuumpackaged microbolometer is shown in Fig. 3. This device-level package flexible device would be compatible with direct integration or bonding to
MEMS on Flexible Substrates
a flexible circuit board. In this case, the flexible die can be glued to the flexible circuit board with an appropriate adhesive with the electrical connections being made by bump bonding. Here, the fabrication proceeded along the same lines as in the previous case up through the deposition of the YBCO thermometer. The precise details can be found in Ref. [6]. After the YBCO thermometer was deposited, a second polyimide sacrificial layer was spin coated and patterned on top of the thermometer. This was followed by an aluminum oxide, Al2O3, packaging layer. Al2O3 was chosen for its high mechanical strength and optical transparency, which extends into the infrared. The Al2O3 packaging layer had trench cuts or openings to facilitate the removal of the sacrificial layers above and below the thermometer by ashing in an oxygen plasma. After the sacrificial layers were removed, the trench cuts were sealed by sputtering more Al2O3 at 10 mTorr Ar pressure, thereby creating a vacuum cavity around the microbolometer. Once sealed, a polyimide encapsulation or superstrate could be applied to place the microbolometer at a low stress plane inside the polyimide film.
Pressure/Tactile Sensors on Flexible Substrates Pressure/tactile sensors on flexible substrates have been developed for medical and structural healthmonitoring applications. Two types of MEMS pressure sensors have been investigated: absolute pressure sensors and delta pressure sensors. Absolute pressure sensors measure the external pressure relative to a sealed reference cavity while delta pressure sensors respond to an external pressure of force relative to the ambient pressure. This entry will focus upon delta pressure sensors. MEMS delta pressure sensors operate by deflecting a suspended membrane with an external pressure or force. The deflection of the membrane can be sensed capacitively or by measuring the resulting strain induced in the membrane with piezoresistors. In the latter case, the piezoresistors can be configured into a Wheatstone bridge geometry with two active and two passive piezoresistors. This configuration provides zero output under a no load condition when the resistors are balanced
1347
M
and also provides temperature drift compensation by fabricating the piezoresistors from the same material with the same temperature coefficient of resistance. A confocal microscope picture of a piezoresistive delta pressure sensor is shown in Fig. 6. In the case of pressure sensors, since the sensor is designed to measure an external pressure, the sensor must have access to the environment through a hole in the superstrate or have a thin superstrate, less than 10-mm thick, so the external pressure is transferred to the sensor. Two types of piezoresistors have been investigated, polycrystalline Si piezoresistors [11, 12] and nichrome piezoresistors [13]. In the case of polycrystalline Si piezoresistors, normally to achieve the polycrystalline lattice structure requires deposition or annealing temperatures that would be prohibitively high (>600 C) for their use with flexible polymer substrates. By utilizing aluminum-induced crystallization, Si can be crystallized after its deposition by annealing at temperatures as low as 152 C [14]. Piezoresistive Si with gauge factor ([DR/R]/average strain) in the range of 7–12 was obtained through this technique [15]. Alternatively, nichrome (nickel 80%, chromium 20%) thin films were employed as the piezoresistors [13]. The advantage of using metallic piezoresistors is their ability to be deposited at room temperature while one sacrifices the magnitude of the response due to their lower gauge factor, in the range of 1 to 3. The fabrication of the delta pressure sensors on flexible substrates proceeded along lines similar to the fabrication of the microbolometers on flexible substrates described earlier. The basic fabrication began with spin coating the polyimide substrate onto a Si carrier wafer. After curing, a protective film of Si3N4 or Al2O3 was sputtered to protect the polyimide substrate during subsequent fabrication steps. Next, the polyimide sacrificial layer was spin coated and cured on the wafer. Then, the Si3N4 or Al2O3 membrane layer was deposited. The piezoresistors were then fabricated. Here, the process differed depending upon whether nichrome [13] or aluminuminduced crystallization polysilicon [12] was used as the piezoresistor. In the case of nichrome, the few tens of nanometer thick piezoresistor was deposited by rf magnetron sputtering and patterned by photolithographic liftoff to complete the piezoresistor fabrication. For aluminum-induced crystallization of
M
M
1348
MEMS on Flexible Substrates, Fig. 6 (Left) A confocal microscope picture showing a piezoresistive delta pressure sensor in a Wheatstone bridge geometry. (Top right) A SEM micrograph of the piezoresistive delta pressure sensor in a mesa geometry. (Bottom left) A SEM micrograph of a delta pressure sensor showing the cavity underneath
MEMS on Flexible Substrates
1
20 μm
Passive R2
2 Active R1 Active R3
3 Passive R4
4
polysilicon, first a 0.5-mm-thick aluminum layer was deposited, followed by a 0.5-mm-thick amorphous Si layer. The two layers were patterned together by liftoff to define he piezoresistor geometry. Then, the wafer was annealed. During this process, the aluminum and silicon layers change positions through interdiffusion and a polycrystalline Si layer results at the bottom of the stack. The Al top layer is then removed by wet etching leaving behind an Al-doped poly-Si piezoresistor. The Al metallization layer was then deposited and patterned by liftoff to wire the active and passive piezoresistors into a Wheatstone bridge geometry. To prevent the piezoresistors from oxidizing during the removal of the polyimide sacrificial layer, a protective layer was deposited over the piezoresistor. In the case of the nichrome piezoresistors, a thin layer of Al2O3 was used. The polyimide sacrificial layer was ashed in an oxygen plasma to suspend the membrane above the substrate. In some devices, a thin polyimide layer was then bonded on top of the wafer to protect the pressure sensors. In other devices, a photo-patternable superstrate was spin coated on top of the pressure sensors and developed to open a small access hole to the sensor. At this point, the fabrication was complete and the sensor could be removed from the substrate. The sensors can be designed to have various sensitivities by varying the membrane thickness and size of the suspended region. In this case, typical dimensions were 70–150 mm across. After completion of the sensor fabrication, the sensors were tested by deflecting the
10 μm
membrane with a probe tip in a probe station or applying a known load with a load cell [13] or atomic force microscope [16].
Accelerometers on Flexible Substrates MEMS accelerometers on flexible substrates have applications in structural health monitoring of aircraft and aiding in the control of micro air vehicles [17]. The basic operating principle of a MEMS accelerometer is an under-damped to critically damped spring-mass system. The inertia of the mass causes a deflection when the mass is accelerated. By measuring the deflection, the acceleration can be determined [3]. The deflection can be measured by piezoresistive, piezoelectric, or capacitive transducers. Capacitive accelerometers, fabricated using a modified UV-LIGA process [18] with an electroplated Ni proof mass (Fig. 7), have been investigated. Both z-axis (perpendicular to substrate) and lateral axis (parallel to substrate) accelerometers have been designed to measure accelerations up to 20g where g is the acceleration due to gravity. The details of the fabrication process are given elsewhere [17]. The fabrication will be described in brief here. The basic fabrication process started with spin coating the polyimide substrate onto a Si carrier wafer followed by the sputter deposition of a Si3N4 protection layer. The accelerometers were constructed using a polyimide sacrificial layer that was patterned by
MEMS on Flexible Substrates
1349
M
MEMS on Flexible Substrates, Fig. 7 (Left) An optical micrograph of a z-axis accelerometer fabricated on a polyimide substrate. (Right) A flexible polyimide die containing z-axis and lateral axis accelerometers conformally resting on a finger
photolithography and cured. An Au seed layer was utilized for electroplating the Ni proof mass. The mold for the proof mass was defined with a 6-mm-thick negative photoresist. The low stress Ni was electroplated in a nickel sulfamate solution. After electroplating, the mold resist was stripped and the Au seed layer was wet etched. Then, the polyimide sacrificial layer was ashed with an oxygen plasma to suspend the proof mass above the substrate with Ni springs defined by the photoresist mold. The dies were then packaged for testing. Some of the dies were taken off the substrate as shown in Fig. 7. Figure 7 also depicts an optical micrograph of a z-axis accelerometer. The accelerometers were tested with a shaker that provided accelerations up to 8g. The acceleration was calibrated through a commercially available accelerometer which was mounted beside the accelerometer under test. The z-axis accelerometers displayed a sensitivity exceeding 20 fF/g while the sensitivity of the lateral axis accelerometers was more than 10 fF/g [17]. The sensitivity was compared to identical accelerometers fabricated directly on Si substrates and found to be the same within the measurement and fabrication tolerances. The noise floors due to Brownian motion of the suspended sensor was calculated to be 2.25 mV/√Hz compared to the measured noise floor of the readout circuit of 0.13 mV/√Hz indicating that the measurement system dominated the measured noise [17].
Other MEMS on Flexible Substrates At the University of Illinois at Urbana-Champaign (UIUC) work on incorporating electronic devices on
flexible substrates [19] has also turned to micromachining in some cases [20, 21]. Using micromachining, ribbons of electronic devices [20] and interconnects [21] have been suspended above the substrate to enable both flexibility and stretchability of the electronic film. This work is separated from the work described earlier in that electronic devices such as diodes and transistors are incorporated on the flexible substrate. Micromachining is used to suspend the electronic ribbons or interconnects above the substrate with periodically spaced anchor points so that when the substrate is bent or stretched, there is room for the electronic ribbons and interconnects to deform without encountering a large strain that could damage the devices. Therefore, in this case, the devices are more traditional electronic devices rather than traditional MEMS devices described earlier. The research at UIUC may form a path for the integration of MEMS devices with electronic components such as transistors. One factor not addressed by this work is the packaging of the suspended components. Along similar lines, other investigators have also reported micromachined interconnects that are suspended above the substrate [22, 23]. Using MEMS cantilever rf switches, reconfigurable frequency-selective surfaces [24] and antennas [25, 26] have been demonstrated on flexible substrates. A surface micromachined hot-wire anemometer has also been shown [27]. In addition, a variety of MEMS biomedical devices have been developed on flexible substrates including implantable moveable electrodes for the brain [28] and a polymerase chain reaction (PCR) system for DNA amplification [29]. These devices validate the attractiveness of MEMS devices on flexible substrates for
M
M
1350
a variety of applications. Yet a common set of problems always exists and is sometimes neglected such as packaging of the MEMS devices so that they have the ability to function without losing their flexibility and can be freely handled without damage. This property is more easily achieved for small-area MEMS devices but becomes more difficult as the size increases. In addition, the design process to achieve flexibility is easier for small-area devices compared to large area devices since the deflection and stress encountered by a small-area device is less than those experienced by a large area device for the same radius of curvature.
Future Directions and Conclusions This work has demonstrated MEMS sensors can be fabricated on a variety of substrates from rigid Si to flexible polyimide without losing performance. The work on flexible substrates has thus far been focused on constructing MEMS using conventional electronic materials for the insulators and metals for the wiring. The key restriction has been the necessity to use low-temperature processing to maintain the polyimide below the glass transition temperature during the fabrication process. In future, the traditional electronic materials of Al2O3, Si3N4, Al, and nichrome will give way to polymer and plastic materials that can be used for the structural MEMS materials. Work has already been started on using polymer-carbon nanocomposites for the piezoresistive transducers employed in many MEMS sensors. The compatibility with organic semiconductors and graphene transistors will pave the way for the integration of transistor readout circuitry in the near future. MEMS sensors on flexible substrates provide the ability to conform to nonplanar surfaces that are attractive in medical, robotic, and aviation applications. The resulting devices have the potential to be fabricated at reduced cost, size, and weight. Flexible electronics appears to be growing in its attractiveness for a variety of applications. Acknowledgment This material is based in part on work supported by the Air Force Office of Scientific Research, the Texas Ignition Fund, and Lockheed Martin Aeronautics Company.
MEMS on Flexible Substrates
Cross-References ▶ AFM, Tapping Mode ▶ Flexible Electronics ▶ Graphene ▶ Insect Flight and Micro Air Vehicles (MAVs) ▶ MEMS Packaging ▶ Piezoelectric Effect at Nanoscale ▶ Piezoresistivity
References 1. Wong, W.S., Salleo, A. (eds.): Flexible Electronics: Materials and Applications. Springer, New York (2010) 2. Schwierz, F.: Graphene transistors. Nat. Nanotechnol. 5, 487–496 (2010) 3. Elwenspoek, M., Wiegerink, R.: Mechanical Microsensors. Springer, New York (2001) 4. Lobontiu, N.: Dynamics of Microelectromechanical Systems (Microsystems). Springer, New York (2007) 5. Suo, Z., Ma, E.Y., Gleskova, H., Wagner, S.: Mechanics of rollable and foldable film-on-foil electronics. Appl. Phys. Lett. 74, 1177–1179 (1999) 6. Mahmood, A., Butler, D.P., Celik-Butler, Z.: Device-level vacuum-packaging scheme for microbolometers on rigid and flexible substrates. IEEE Sensor. J. 7, 1012–1019 (2007) 7. Rahman, M.S., Chitteboyina, M., Butler, D.P., Celik-Butler, Z., Pacheco, S., McBean, R.: Device-level vacuum packaged MEMS resonator. IEEE/ASME J. Microelectromech. Syst. 19, 911–917 (2010) 8. Mahmood, A., Butler, D. P., Celik-Butler, Z.: Device-level vacuum packaged micromachined infrared detectors on flexible substrates. Presented at the IEEE Sensors Conference, 31 Oct –3 Nov 2005, Irvine, CA (Proceedings) 9. Dayeh, S.A., Butler, D.P., Celik-Butler, Z.: Micromachined infrared bolometers on flexible polyimide substrates. Sensor. Actuator. A 118, 49–56 (2005) 10. Yaradanakul, A., Celik-Butler, Z., Butler, D.P.: Infrared sensing microbolometers on flexible substrates. IEEE Trans. Electron Dev. 49, 930–933 (2002) 11. Patil, S.K., Celik-Butler, Z., Butler, D.P.: Nanocrystalline piezoresistive polysilicon film by aluminum induced crystallization for pressure sensing applications. IEEE Trans.Nanotechnol. 9, 640–646 (2010) 12. Patil, S.K., Celik-Butler, Z., Butler, D.P.: Piezoresistive polysilicon film obtained by low-temperature aluminum induced crystallization. Thin Solid Films 519, 479–486 (2010) 13. Nadvi, G.S., Butler, D.P., Celik-Butler, Z., Go¨nenli, ˙I.E.: Micromachined Pressure/Force Sensors using Nickel Chromium Piezoresistors, submitted for publication 14. Radnoczi, G., Robertsson, A., Hentzell, H.T.G., Gong, S.F., Hasan, M.A.: Al induced crystallization of a-Si. J. Appl. Phys. 69, 6394–6399 (1991)
MEMS Packaging 15. Patil, S. K.: Nanocrystalline piezoresistive polysilicon film obtained by aluminum induced crystallization for pressure sensing applications. Ph.D. Dissertation, University of Texas at Arlington (2010) 16. Patil, S.K., Celik-Butler, Z., Butler, D.P.: Characterization of MEMS piezoresistive pressure sensors using AFM. Ultramicroscopy 110, 1154–1160 (2010) 17. Go¨nenli, ˙I. E., C¸elik-Butler, Z., Butler, D. P.: MEMS accelerometers on polyimide substrates, IEEE Sensors Journal 11, 2318–2326 (2011) 18. Saile, V., Wallrabe, U., Tabata, O., Korvink, J.G.: LIGA and Its Applications. Wiley-VCH, Weinheim (2009) 19. Ahn, J.-H., Kim, H.-S., Lee, K.J., Zhu, Z., Menard, E., Nuzzo, R.G., Rogers, J.A.: High-speed mechanically flexible single-crystal silicon thin-film transistors on plastic substrates. IEEE Electron Dev. Lett. 27, 460–462 (2006) 20. Sun, Y., Rogers, J.A.: Structural forms of single crystal semiconductor nanoribbons for high performance stretchable electronics. J. Mater. Chem. 17, 832–840 (2007) 21. Kim, D.-H., Choi, W.M., Ahn, J.-H., Kim, H.-S., Song, J., Huang, Y., Liu, Z., Lu, C., Koh, C.G., Rogers, J.A.: Complementary metal oxide silicon integrated circuits incorporating monolithically integrated stretchable wavy interconnects. Appl. Phys. Lett. 93, 044102 (2008) 22. Lin, K.L., Jain, K.: Design and fabrication of stretchable multilayer self-aligned interconnects for flexible electronics and large-area sensor arrays using excimer laser photoablation. IEEE Electron Dev. Lett. 30, 14–17 (2009) 23. Lin, K.L., Chae, J., Jain, K.: Design and fabrication of largearea, redundant, stretchable interconnect meshes using excimer laser photoablation and in situ masking. IEEE Trans. Adv. Pack. 33, 592–601 (2010) 24. Coutts, G.M., Mansour, R.R., Chaudhuri, S.K.: Microelectromechanical systems tunable frequency-selective surfaces and electromagnetic-bandgap structures on rigid-flex substrates. IEEE Trans. Microw. Theory Techn. 56, 1737–1746 (2008) 25. Kingsley, N., Ponchak, G.E., Papapolymerou, J.: Reconfigurable RF MEMS phased array antenna integrated within a liquid crystal polymer (LCP) system-on-Package. IEEE Trans. Antenn. Propag. 56, 108–118 (2008) 26. Kingsley, N., Anagnostou, D.E., Tentzeris, M., Papapolymerou, J.: RF MEMS sequentially reconfigurable sierpinski antenna on a flexible organic substrate with novel DC-biasing technique. J. Microelectromechan. Syst. 16, 1185–1192 (2007) 27. Buder, U., Berns, A., Petz, R., Nitsche, W., Obermeier, E.: AeroMEMS wall hot-wire anemometer on polyimide substrate featuring top side or bottom side bondpads. IEEE Sensor. J. 7, 1095–1101 (2007) 28. Jackson, N., Muthuswamy, J.: Flexible chip-scale package and interconnect for implantable MEMS Movable microelectrodes for the brain. J. Microelectromech. Syst. 18, 396–404 (2009) 29. Lee, D.-S., Park, S.H., Chung, K.H., Pyo, H.-B.: A disposable plastic-silicon micro PCR chip using flexible printed circuit board protocols and its application to genomic DNA amplification. IEEE Sensors Journal 8, 558–564 (2008)
1351
M
MEMS Packaging Bongsang Kim1 and Woo-Tae Park2 1 Advanced MEMS, Sandia National Laboratories, Albuquerque, NM, USA 2 Seoul National University of Science and Technology, Seoul, Korea
Synonyms MEMS encapsulation
Definition A method of protecting microelectromechanical system (MEMS) devices from the environment by means of physical enclosures.
Introduction Packaging is a challenging and critical step in the development of MEMS products. MEMS structures are generally very fragile, have moving structures, and must interface with the environment. The requirements for packaging are application specific and differ from each other, thus standard MEMS packaging protocols are difficult to establish, leading to an increase in product cost. For many MEMS products, the cost for packaging often dominates over the device fabrication cost. In this entry, general packaging requirements for typical MEMS devices will be discussed, followed by requirements for specific applications. Then, various state-of-the-art MEMS packaging approaches will be introduced with some examples.
General Requirements for MEMS Packaging To ensure proper operation in environments that may be encountered, packaging is required for any semiconductor electronic device. Generally, packaging is the final step of semiconductor device fabrication before final product testing. The packaging activity encompasses a wide spectrum of materials and
M
M
1352
technologies from the back-end-of-line (BEOL) of the semiconductor manufacturing process until the final product itself. The most basic function is to distribute the electrical connections from the semiconductor chip to a circuit board. This function is usually realized by wirebond or flip chip technology. Besides establishing electrical connections, other functions include: (1) mechanical protection from the environment of use or subsequent packaging process, (2) chemical protection from the environment, and (3) thermal management of the heat generated from the chip. Depending on the device and product requirement, the importance of packaging functionality listed above varies. For example, for modern central processing unit (CPU) devices, thermal management is the most critical, while mechanical protection and chemical protection are relatively less important. In the case of MEMS devices, thermal management is usually less problematic, since most applications draw relatively low current, and thus generate little heat. However, MEMS devices contain mechanical structures. Therefore, they tend to be more sensitive to mechanical perturbation from the environment. Also, as the size scales down, the surface-to-volume ratio increases, which makes MEMS devices more susceptible to surface forces that may develop. Mechanical Protection Usually, MEMS devices contain suspended microscale mechanical structures with micrometer-spaced gaps. Mechanical protection is the most obvious required functionality of MEMS packaging. Airborne particles must be kept away from structures to ensure proper function. Impact can potentially fracture fragile structures, or cause compliant structures to come into contact, leading to adhesion failure. Mechanical stress on the device structure can also jeopardize its reliability. For example, stress can lead to a change in the electrical signal in the devices using piezoresistive or piezoelectric transduction. Capacitively transduced devices can also generate unreliable output signal, if the relative positions between electrodes are shifted by stress. Therefore, one important requirement of MEMS packaging is preventing MEMS devices from exposure to direct mechanical impact or shock. For example, shock requirement becomes more important in some applications such as automotive airbag sensors (sensing range 50 g), and consumer devices that are prone to be dropped by the user.
MEMS Packaging
Chemical Protection Chemical protection must also be provided by MEMS packaging technology. Water condensation on the micromechanical structure can create permanent failure to the MEMS device. Due to their large surface-tovolume ratio, chemical reactions on the surface can completely change the device performance. For example, surface oxidation can completely change the mechanical fatigue behavior of silicon-based MEMS devices. For such reasons, from the beginning of the MEMS device development, the Hermeticity of its packaging has been a significant issue. Early studies were focused on hermeticity against humidity or moisture [1]. Later, hermeticity against atmospheric gas species has become important for many vacuum applications such as IR sensors and RF resonators, or inertial sensors where certain level of damping has to be maintained during operation [2]. Recent research works have investigated the diffusion characteristics of different gas species and diffusion pathways in MEMS packagings [3]. Cost-Effectiveness Although MEMS packages have many stringent requirements (electrical connections, mechanical and chemical protection), cost minimization is required to gain commercial success. With respect to the mass market, MEMS sensors first penetrated the automotive industry. Even though the cost of the packaging could be up to 90% of the device cost, MEMS sensors were still much more economical choice compared to the few existing solutions. In order to provide isolation from the harsh environment of automobiles (e.g., high temperature and humidity), ceramic packages that closely matched the coefficient of thermal expansion (CTE) of silicon were first used. They were later replaced by molded packages to reduce cost. Since molded packages have a different CTE from silicon, a soft and thick die-attach such as DOW RTV and gel dome coating have been used for stress isolation (i.e., to minimize the transfer of stress from the package to the device). In the consumer electronics market, the requirement for cost reduction became even higher. To bring the cost of the overall sensor to less than $1, cost reduction is needed in all areas of product manufacturing, and packaging is no exception. Today the most prevalent standard packages used are quad flat no leads (QFN), dual flat no leads (DFN), or land grid array (LGA)
MEMS Packaging
packages (testing is done in arrays before package singulation). Another important requirement for consumer electronics applications is the size of the overall package. The packages need to have the smallest possible footprint, and thin profile. The accelerometers (STMicroelectronics LIS331series) used in Apple’s Iphone 4 have dimensions of 3 mm 3 mm 1 mm. Because of this small size, the thick and soft die-attach or the gel dome coating cannot be used anymore. Although themomechanical stress or long-term reliability is not as stringent as in automotive products, stress isolation remains an issue with the thin dieattach and gel coating. Alternatively, more effort was directed toward the MEMS sensor design to compensate the stress. One example is a single anchor design, so that no thermomechancial stress is applied due to anchor movement. Advances in consumer electronics applications will be applied back to automotive applications, leading to further cost reduction. This is important as passenger safety requirements are becoming even more strict for automobiles. For example, in the USA, the National Highway Traffic Safety Administration (NHTSA) has already implemented front-seat airbag regulation, tire pressure monitoring systems, and will implement vehicle stability program regulation in 2012 for all vehicles. All of these safety regulations require MEMS sensor-based systems, and must be affordable for cars in all classes.
Various MEMS Applications and Their Packaging Requirements Inertial Sensors Accelerometers, such as shown in Fig. 1, are one of the most successful MEMS devices. This success is due to MEMS technology advances and to the NHTSA airbag safety regulation (section 208) in the 1990s. Further success will be aided by recently announced side airbag regulations beginning in 2013. Accelerometers have contributed in saving millions of lives. For consumer electronics, accelerometers were used to detect “portrait-landscape” orientation for wide-screen phones. They are currently also widely used as motion control sensors in consumer electronics. Accelerometer sensitivity can often be improved by increasing the size of released proof mass. Due to the need for a relatively large cavity, wafer bonding is
1353
M
MEMS Packaging, Fig. 1 An accelerometer that is wafer level packaged with a cap wafer bonded by glass frit bonding (left side). Afterward, the accelerometer and ASIC are packaged in a land grid array (LGA) package (right side) [4]
preferred over other wafer-level encapsulation methods. In this approach, after device release, encapsulation is accomplished by sealing the device wafer with a separate “cap” wafer. Pre-etched wafers with cavities are bonded to the device wafer immediately after the release step (see the descriptions in the next section for more details). The gas environment during the bonding step must be controlled to establish the sensor dynamics. Nitrogen is used as the main filling gas, because it is inert and the molecules are relatively large (compared to hydrogen or helium), so diffusion is slow. Pressure is carefully set to critically damp the mass. Gyroscopes have a more complex device structure than accelerometers. MEMS gyroscopes are usually designed as vibration-based structures, so high vacuum is required ( > þ H ðfc vc Þ ¼ Gc ; > > @t > > > > > < @fm þ H ðfm vm Þ ¼ Gm ; @t
> > s >v v ¼ K 1 > H ðfc T0 c Þ; c m > > jH ðfc T0 c Þj þ > > > : H ðfc T0 c þ fm T0 m Þ ¼ 0:
M
M
1444
where c refers to the cellular population, m to the extracellular matrix, (·)+ stands for the positive part of the parenthesis, K is the motility coefficient, and s measures the adhesive properties of the cell to the extracellular matrix. Of course, the equations above need to be accompanied by the constitutive equations for the excess stress tensors T0 c and T0 m , describing the response of both cells and ECM to deformation. Recently, also Lowengrub and coworkers [3] focused their attention on the adhesion properties of cellular populations introducing it as an adhesion energy. Still working in a multiphase framework, the final model consists of a fourth-order nonlinear advection-reactiondiffusion equation of Cahn–Hilliard type characterized by the presence of a term regularizing the backward heat equation term related to adhesion phenomena.
Modeling Angiogenesis When the tumor grows too much getting larger than few hundreds microns, the amount of nutrients reaching the center of the tumor is not sufficient to feed all cells, so that its central region becomes hypoxic (see Fig. 1). In this situation most cells die, but some tumor cell might have acquired the ability of producing a class of protein generally called tumor angiogenic factors that stimulate the existing capillaries to produce new capillary networks reaching the tumor. This process, called angiogenesis, is a fundamental step in tumor growth because it distinguishes the avascular phase from the vascular phase. The difference between the two phases of growth is dramatic because in the avascular phase the tumor is very small (with a radius that is smaller than a millimeter and limited by the diffusion of nutrients), stationary, and therefore harmless, while in the vascular phase the tumor has all the nutrients to grow in an unlimited way and has a highway to spread metastases throughout the body. One of the most celebrated model is the one due to Chaplain and Anderson (1998) (see [10] for a review on the topic, and the volume [11]) who had the main merit to relate a continuous model with its discrete version allowing a better visualization of the process of formation of capillary trees starting from existing vessels. In their model Chaplain and Anderson omit any birth and death process for the cells forming the capillaries focusing on the trajectories formed by the tip cells of the newborn capillaries, assuming that the endothelial cells forming the capillary walls will
Models for Tumor Growth
follow. Other key ingredients are chemotaxis toward higher densities of tumor angiogenic factors produced by the tumor, and haptotaxis toward higher concentration of extracellular matrix that is degraded by the tip cells. The model then writes as 8 @n > > þ H ½ðwc Hc þ wf Hf Þn ¼ kn H2 n; > > > @t > < @c ¼ dc nc; > @t > > > > > : @f ¼ gn df nf ; @t where n is the density of tip cells, c is the density of tumor angiogenic factor, and f is the density of fibronectin. In addition, wc and wf are motility coefficients referring respectively to chemotaxis and haptotaxis sensitivity, kn refers to random motility, dc and df are the uptake coefficients from the tip cells, and g is the production rate of fibronectin by the cells. The aim of the discrete technique then introduced develops in the framework of reinforced random walk. It consists in using the resulting coefficients of the fivepoint stencil of the standard central finite-different scheme to generate the probabilities of movement of an individual cell in response to the chemoattractant gradients and to diffusion. The extraction of a random number will decide whether the tip cell will stay still in the node or will move to a particular neighboring node rather than another. However, with respect to the standard random walk algorithm, in presence of chemotaxis or haptotaxis the motion becomes biased, because the cell has higher probabilities to move up the gradients of chemical factors. In addition to that, the discretized setup allows to include some phenomena difficult to describe using a model based on partial differential equations, e.g., capillary branching and anastomosis, i.e., the formation of capillary loops. In fact, when during their motion a capillary tip meets another capillary, then they merge to form a loop. If two sprout tips meet, then only one of the original sprouts continues to grow. As shown in Fig. 2, the capillary network built in this way looks very realistic and compares well with what is observed experimentally. A three-dimensional animation of the angiogenic process is available at the web site www.maths.dundee.ac.uk/sanderso/3d/index.html.
Models for Tumor Growth
Models for Tumor Growth, Fig. 2 Tumor-induced capillary trees as simulated by Chaplain and Anderson (1998)
Historically speaking, the interest in modeling angiogenesis was due to the interest in developing anti-angiogenic therapies aimed at destroying the disordered and leaky tumor-induced vessel network. Unfortunately, at present antiangiogenic therapies don’t look so successful as expected. However, there is now a renewed interest due to the concept of vessel remormalization proposed in order to optimize drug delivery [12].
Modeling Vascularized Tumors and Invasion Compared with the models focusing on the avascular phase, relatively fewer models were developed dealing with tumor invasion and vascular growth. Most of them have a structure similar to the ones mentioned above. For instance, some vascular growth models develop assuming a constant density of tumor cells duplicating along a blood vessel and forming what is called a tumor cord possibly surrounded by a necrotic region farther away from the blood vessel (see Chap. 3 of [13] or Chap. 7 of [6]). Others develop in the multiphase framework looking more closely at the change in metabolism occurring in some tumor cells that in hypoxic conditions acquire the ability to survive and extract energy also in absence of oxygen.
1445
M
This makes the environment more acidic and unsuitable for the host tissue that dies leaving room for tumor growth (see Chap. 4 of [13]). Other models focus on the mechanical interaction with the capillary walls, and on the fact that newborn vessels are immature and therefore subject to mechanical collapse due, for instance, to the pressure exerted by the tumor growing around them. This will cause hypoxia and the generation and remodeling of blood vessels. Others again use the discrete angiogenesis model described above to focus on the perfusion of drugs or on vessel remodeling (see Chap. 15 of [6] or [11]). Another key aspect in tumor invasion is the ability acquired by the tumor cells to produce and secrete certain factors known as matrix degrading enzymes that chop the extracellular matrix and then modify the local tissue environment making it easier for the cells to invade the surrounding tissue finding easier paths to move. This process, called metallo-proteinases, is also used by tumor cells to break through basal membranes, which otherwise would compartimentalize the tumor. A simple example of models including the action of matrix degrading enzymes is, for instance, the following (see Chap. 10 of [5] for more details): 8 @n > > þ H ðwðn; f ÞnHf Þ ¼ Dn H2 n þ G; > > > @t > < @f ¼ dmf ; > @t > > > > > : @m ¼ Dm H2 m þ mn lm; @t where the first equation describes the diffusion of tumor cells under the action of haptotaxis (second term on the left-hand side) and growth/death (second term on the right-hand side), the second equation describes the degradation of the extracellular matrix, and the third equation the diffusion of matrix degrading enzymes produced by tumor cells.
Models at the Cellular Scale Looking at the population of tumor cells at the cellular scale, it is possible to realize that the behavior of the cells is not always the same. Their evolution is characterized by the presence of switches and acquired ability that allow them to resist to particularly unfavorable conditions or by the birth of daughter cells that have a clonal advantage on their mother cells. For
M
M
1446
Models for Tumor Growth
instance, in the initial stages of tumor development a corruption in the mechanotransduction pathway leads to the existence of cells that are not so sensitive to the mechanism of contact inhibition of growth, so that they continue activating their duplication program even when they should stop proliferating. This misperception of the compression state by the abnormal cells, called cadherin switch, can generate by itself a clonal advantage on the surrounding cells, leading ultimately to the replacement and the invasion of the healthy tissue by the tumor (see Chap. 7 of [7]). After that, other landmarks characterize tumor evolution as the angiogenic switch stimulating the formation of tumor-induced capillary plexus, or the just mentioned metabolic switch from aerobic to anaerobic production of ATP, or the epithelial-mesenchymal transition, that triggers the formation of metastases. Also from the morphological viewpoint tumor cells present differences that are usually identified by histopathologists to classify tumor malignancy, e.g., presence of particular receptors or organelles, or loss of cell polarity. The mathematical models focusing on the inclusion of the inner evolution of the tumor population briefly described above needs to incorporate a new independent variable which is able to describe the state (or the states) characterizing the cells and how such a state influences their behavior. This state might refer to the aggressiveness, the activity, the maturation state, and so on (see Chap. 3 of [6], and [14, 15]). The key steps in the modeling process are then the selection of the cell populations of interest for the evolution and the identification of their specific activity/ies, their mechanisms of progression, of proliferation, and of interaction. The structure of the model then develops in the framework of kinetic theories and may be written as follows: @ @ fi ðt; uÞ þ ½ci ðt; uÞfi ðt; uÞ @t @u n Z X ¼ S i þ fi ðt; uÞ þ
"Z n X j¼1
1
Z
1
j¼1 1 1
1
1
pij ðu; wÞfj ðt; wÞ dw
ij ðv; wÞCij ðv; w; uÞ
fi ðt; vÞfj ðt; wÞ dv dw Z 1 ij ðu; wÞfj ðt; wÞ dw ; fi ðt; uÞ 1
where the second term on the left-hand side represents the intrinsic evolution operator which takes into account of the progression of the cells even in absence of interactions and proliferation/death. The second term on the right-hand side represents the net proliferation operator, i.e., stimulated mitosis minus killing due to interaction with other cells. In it ij(v, w) denotes the number of encounters per unit volume and unit time between cell pairs of the (i, j)-th populations with states v and w, respectively, and cij (v, w; u) denotes the probability of transition of the i-th cell to the state u, given its initial state v and the state w of the encountering cells belonging to the j-th population. The last two terms are related to interactions which only change state in the interacting cells, while the second one refers to proliferative/destructive interactions. Finally Si is the source/sink term which may be external (e.g., effect of chemotherapy and radiotherapy) or internal, like the net proliferation term (mitosis-apoptosis) gi(t, u)fi(t, u). Other popular models focusing on the cellular scale, in general without considering the inner evolution of cells, are based on cellular automata, cellular Potts models, and individual cell-based models [16]. In individual cell-based models, a cell is schematized by its center, then surrounded by a deformable spherical shell or by a Voronoi tessellation (see Part III of [16] for more details). If a spherical cell gets in contact with other cells, it adheres to them and as a result of the contact, the shape of the cell changes by flattening at the contact area. A similar thing occurs for the interaction with the substrate that is typically a rigid impermeable plane or with other elements of the environment, e.g., bones and vessel walls. The dynamics of each individual cell is then modeled by Langevin equations in a frictiondominated regime. Thus, in the absence of an external stimulus the cells perform a random movement. For instance, the motion of the i-th cell is modeled by X
st Cij ðwi wj Þ þ Cis wi ¼ Fdet i þ Fi ;
j
where wi are the cell velocities and Cia are related to the friction coefficients, respectively, with between cells (a ¼ j) and between cells and the substrate (a ¼ s). On the right-hand side, Fdet summarizes the i deterministic forces related to the total interaction
Models for Tumor Growth
energy between the cell pair (i, j) defined by the sum over all individual energy contributions, that depend on the distance between the cells and the radius of both cells. Thus, cell–cell contacts can equilibrate via cell displacements or changes in the cell radius R. The total interaction energy between a cell and the substrate is defined analogously. The term Fsti denotes the stochastic force with zero mean and delta-correlated autocorrelation function that models the random component of cell movement. The cellular Potts model is another modeling framework that has been used to describe several tumor growth problems, such as avascular and vascular tumor growth, angiogenesis, and cell invasion problems (see Part II of [16]). The cellular Potts model is a grid-based, stochastic Monte Carlo method in which each cell is described as an ensemble of nodes. The core driving the evolution of the system is an energy minimization philosophy. Depending on the problem, the energy functionals to be minimized may include adhesion energies between cells or with the substrate, terms mimicking energies, for instance, the energetical response of a cell to a chemical gradient, or persistence of cell motion, and possible constraints, such as on the volume or the surface area of the cells. The energy functional H ¼ Hadhesion þ Hshape þ Hchemotaxis þ Hpersistence þ . . . ::; then describes the state of the system and drives all lattice rearrangements with an algorithm of stochastic minimization (a modified Metropolis procedure). In this way, it is possible to mimic the natural behavior of migrating cells, via thermal-like membrane fluctuations and extension of pseudopodia to explore the surrounding environment. Procedurally, a lattice site is selected at random and assigned the state of one of its unlike neighbors, which has also been randomly selected. The Hamiltonian of the system is computed before and after the proposed update: If it is reduced as the result of the copy, the change is accepted, otherwise it is accepted with a probability that decreases exponentially with the energy increase that such an unfavorable choice would give. Figure 3 gives an example of a CPM describing the motion of a single cell or of a cell cluster through the mesothelial layer.
1447
M
Models for Tumor Growth, Fig. 3 Growing tumor masses (left) and invading aggregate of a liver tumor cell line (right) as simulated with a cellular Potts model
Therapy and Multiscale Models Many models based on systems of ordinary differential equations have been deduced to describe the pharmacokinetics, the pharmacodynamics, or the response of the body, possibly divided in compartments, to cancer treatments. Generally, the aim of these models is the optimization and personalization of therapeutic protocols, identifying optimal doses and optimal schedules. This is usually done using optimal control theory. Similar problems are encountered in testing immunotherapies or combined therapies and in describing the action and the use of radiotherapies [17]. Focusing here on the recent trends, nowadays there is a big interest in understanding and modeling the action of radiation and drugs at the subcellular level, which means not only how genetic changes or the expression of particular genes or gene networks determine cell behavior, how radiotherapy can induce fatal genetic damages, but also understanding and describing how the protein networks inside the tumor cells work, how the activation of specific signaling cascades govern cell motion, duplication, and death, and therefore what are the main crossroads of the networks to attack or stimulate from the therapeutical viewpoint. Eventually, from the pharmaceutical viewpoint what kind of drug should be developed because it is more likely to be successful. From the modeling viewpoint, once such networks are understood, the following step consists in inserting them into the macroscopic or the mesoscopic models building what can be called multiscale models, because they take into account of phenomena occurring at different spatial and temporal scales [18, 19]. They, in fact, nest into the macroscopic or mesoscopic
M
M
1448
representation one or more modules accounting for processes at the microscopic scale. Because of the most recent discoveries in the field of genomics, proteomics, and system biology, these descriptions are now expanding considerably. In this respect, models at the cellular scale, such as individual cell-based models or cellular Potts models, seem more flexible and suited to include subcellular mechanisms, though it is possible to nest protein networks in macroscopic models, so that they influence the coefficients of the partial differential equations. Coming back to modeling drug delivery, many of the mathematical models dealing with vascular growth were actually developed with the aim of describing the transport of chemotherapeutic drugs, their perfusion through the vessel walls, their effectiveness in killing tumor cells, or as antiangiogenic therapies [11]. This approach has been recently used also to describe the infiltration of engineered macrophages. In fact, it is known that macrophages migrate specifically toward, and localize within, hypoxic tumor regions, that are the hardest region to cure via radiotherapy. So they can be in principle used as vehicles to deliver drugs and perform antitumor activities such as inducing cell lysis or inhibiting angiogenesis. The mathematical models are generally based on advection-diffusion equations with chemotactic terms to describe the directed motion of macrophages toward hypoxic regions. Very similar models are deduced to describe and optimize drug delivery using other macromolecular carriers like liposomes or suitable nanodevices and nanoparticles. In fact, such macromolecules might more easily reach the tumor sites with respect to other parts of the body exploiting the fact that, because of the action of tumor angiogenesis factors, tumorinduced capillaries are fenestrated and therefore highly permeable. Moreover, for instance, it is possible to attach an antibody to the liposomes, so that the drug is specifically targeted to the tumor cell. The transport, extravasation, and diffusion of nanoparticle in tumors is also modeled to optimize related therapeutical actions, e.g., magnetic nanoparticles are used so that they can be trapped either to occlude capillaries or to be heated up; gold nanoshells are used to enhance laser thermal therapy [20]. As it might be easily guessed, there is not a single model to approach the same biological phenomenon. Of course, each approach has its advantages and
Models for Tumor Growth
disadvantages. For instance, models based on systems of partial differential equations are more suited and computationally cheaper to describe the behavior of tumor tissues from the macroscopic point of view, but, as discussed above, it is more difficult to account for subcellular mechanisms such as signal transduction, expression, or internalization of receptors, and so on. Conversely, individual cell-based models can focus more closely on the cellular level, but they might become computationally expensive when using a large number of cells, and inefficient in providing a general outlook on the system as a whole. Luckily, it is often not necessary to keep the same level of detail throughout the tissue. The idea behind building what can be called interfacing hybrid models is to use the modeling tools like a microscope, focusing on the cellular or subcellular level where and when needed and blurring when such a detailed description is not needed. For instance, the spatial domain can be partitioned in few time-dependent domains, and in each piece of this moving puzzle a different modeling framework can be used, so that it is possible to take advantage of the positive aspects of both frameworks. Of course, in order to do that one should compare, for the same biological phenomenon, the results obtained via different models and transfer information from one modeling framework to another. Allowing the models to properly cross talk needs the development of suitable mathematical procedures which nowadays represents a challenging research field.
Cross-References ▶ Angiogenesis ▶ Atomic Force Microscopy ▶ ECM ▶ Finite Element Methods for Computational Nano-optics
References 1. Wodarz, D., Komarova, N.: Computational Biology of Cancer: Lecture Notes and Mathematical Modeling. World Scientific, Singapore (2005) 2. Araujo, R.P., McElwain, D.L.S.: A history of the study of solid tumor growth: the contribution of mathematical modeling. Bull. Math. Biol. 66, 1039–1091 (2004)
Molecular Computing 3. Lowengrub, J.S., Frieboes, H.B., Jin, F., Chuang, Y.L., Li, X., Macklin, P., Wise, S.M., Cristini, V.: Nonlinear modeling of cancer: bridging the gap between cells and tumours. Nonlinearity 23, R1–R91 (2010) 4. Tracqui, P.: Biophysical models of tumor growth. Rep. Prog. Phys. 72, 056701 (2009) 5. Preziosi, L. (ed.): Cancer Modeling and Simulation. Chapman & Hall/CRC Press, Boca Raton (2003) 6. Bellomo, N., Chaplain, M.A.J., De Angelis, E. (eds.): Selected Topics on Cancer Modeling: Genesis – Evolution – Immune Competition – Therapy. Birkh€auser, Boston (2007) 7. Mollica, F., Preziosi, L., Rajagopal, K.R.: Modeling of Biological Materials. Birkh€auser, Boston (2007) 8. Preziosi, L., Vitale, G.: New trends in the physics and mechanics of biological systems. Lecture Notes of the Les Houches Summer School, vol. 92. Oxford University Press, Oxford (2011) 9. Preziosi, L., Vitale, G.: A multiphase model of tumour and tissue growth including cell adhesion and plastic re-organization. Math. Models Methods Appl. Sci. 21,1901–1932 (2011) 10. Mantzaris, N., Webb, S., Othmer, H.G.: Mathematical modeling of tumor-induced angiogenesis. J. Math. Biol. 49, 111–187 (2004) 11. Chaplain, M.A.J., McDougall, S.: Mathematical Modeling of Angiogenesis. Birkh€auser, Boston (2011) 12. Jain, R.K.: Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 307, 58–62 (2005) 13. Quarteroni, A., Formaggia, L., Veneziani, A.: Complex Systems in Biomedicine. Springer, Milano (2006) 14. Bellomo, N.: Modeling Complex Living Systems – Kinetic Theory and Stochastic Game Approach. Birkh€auser, Boston (2008) 15. Bellouquid, A., Delitala, M.: Mathematical Modeling of Complex Biological Systems: A Kinetic Theory Approach. Birkh€auser, Boston (2006) 16. Anderson, A.R.A., Chaplain, M.A.J., Rejniak, K.A.: SingleCell-Based Models in Biology and Medicine. Springer, Basel (2007) 17. Yakovlev, A.Y., Pavlova, L., Hanin, L.G.: Biomathematical Problems in Optimization of Cancer Radiotherapy. CRC Press, Boca Raton (1993) 18. Cristini, V., Lowengrub, J.S.: Multiscale Modeling of Cancer – An Integrated Experimental and Mathematical Modeling. Cambridge University Press, Cambridge, UK (2010) 19. Preziosi, L., Tosin, A.: Multiphase and multiscale trends in cancer modeling. Math. Model. Nat. Phenom. 4, 1–11 (2009) 20. Cristini, V., Ferrari, M., Decuzzi, P. (eds.): Nanoparticulate Delivery to Cancerous Lesions: Advances in Mathematical Modeling. Fundamental Biomedical Technologies. Springer, New York (2010)
1449
M
Modulus ▶ Interfacial Investigation of Protein Films Using Acoustic Waves
MOEMS/NOEMS ▶ Nanotechnology
Molecular Beam Epitaxy (MBE) ▶ Physical Vapor Deposition
Molecular Computation ▶ Molecular Computing
Molecular Computing Soichiro Tsuda Exploratory Research for Advanced Technology, Japan Science and Technology Agency, Osaka, Japan
Synonyms DNA computing; Molecular computation; Unconventional computing
Definition
Modification of Carbon Nanotubes ▶ Functionalization of Carbon Nanotubes
Molecular computing is a novel approach to develop computing systems and devices alternative to conventional silicon-based computing using (biological) molecules.
M
M
1450
Overview Molecular computing research aims at developing computing systems and devices that are fully or partly composed of molecular materials. It typically refers to molecular devices or systems consisting of molecular substrates (mostly biomolecules, such as protein and DNA) for information processing. Ultimate goals of molecular computing research are not to replace well-established semiconductor materials used for today’s computer systems. Molecular computer devices are generally slow, unstable, and hard to connect up together, compared to conventional semiconductor devices. It is less likely that they could be much improved in the future. Instead, a main application domain of molecular computing is the integration with biological systems, such as biological cells. Molecular computing devices can be directly interfaced with biochemical processes of living cells such as gene regulation network and intracellular signal transduction, whereas semiconductor devices are incompatible with them in terms of size and material. Ultimately, the molecular computing devices would offer more countability to the engineering of biological systems, for example, the controlled release of a therapeutic agent in the cell [1]. In addition to biocompatibility, molecular devices have several other advantages, such as massive parallelism and self-assembly. The former concerns the order in which molecules typically operate. Because molecules in general work in the order of Avogadro number, it has been suggested that theoretically molecular devices which contain a massive amount of molecules can carry out computing in truly parallel fashion, even though each molecule has a little computing power. The latter mainly concerns the free energy minimization process of molecular recognition. If two molecules are complementary, they self-assemble into a supramolecular complex. These advantages lead to DNA computing devices described below, whereas the biocompatibility of molecules bloomed into molecular logic gates, which are also covered here.
Molecular Logic Gates In living cells, there are numerous protein “logic gates” operating in order to process external and internal information. An external physical or chemical signal
Molecular Computing
arriving at cell’s surface “turns on” a receptor on the membrane, and the activated receptor releases signals to downstream messenger proteins in a cascade to activate another protein, and so forth. In this sense, a living cell can be seen as a system with a vast network of simple switches and logic gates made from molecular materials. Motivated by this view of living cells, research on molecular switch and logic gate aims at designing and developing such molecular computing elements that are controllable from the outside. A simpler but essential computational element in the electronic devices is the switching device. Tyagi and Kramer implemented a molecular switch in singlestranded DNA using fluorescence resonance energy transfer (FRET). A fluorophore is attached on one end of a single-stranded DNA and a quencher on the other end. The strand folds into a hair-pin structure because of complementary sequences in both ends, and the fluorophore and quencher come close together. If a complement single-stranded DNA string is present, they form a double-stranded DNA, which keeps the fluorophore and quencher apart. As a result, the fluorescence is activated. If the fluorescence is considered as output signals, the switch works as a one-input oneoutput logic gate, NOT gate, which inverts a binary input signal. Two-input one-output logic gates are also implemented with molecules. De Silva and coworkers developed a molecular AND gate based on photoinduced electron transfer (PET) [2]. The molecule consists of a fluorophore and two receptors (amine and benzocrown ether) connected by spacers (Fig. 1a). Amine binds a proton, and benzocrown ether targets Na+ as inputs, respectively. If both receptor sites are empty, an electron is transferred to the fluorophore from either one of receptors, and the fluorophore does not emit light. Even when one of the receptor sites is occupied by an input (H+ or Na+), an electron is transferred to the fluorophore. Light emission will be observed only when both receptor sites are occupied because electrons are used to bind input H+ and Na+ ions. Thus, the AND operation, which outputs “1” only when both inputs are “1,” is realized. Katz and coworkers employed enzyme-catalytic reactions to construct various Boolean logic gates, such as AND, OR, NOR, XOR, YES, and NOT [3]. For instance, the enzymatic XOR gate consists of two enzymes, glucose dehydrogenase (GDH) and horseradish peroxidase (HRP). GDH and HRP take glucose
Molecular Computing
1451
a
b
e− F
R1
R2
Input Glucose
R1
R2
Gluconic acid
Input H2O2
NAD+ HRP
GDH F
M
Output NADH
H2O
Molecular Computing, Fig. 1 (a) A molecular AND gate based on photoinduced electron transfer. F, R1, and R2 represent fluorophore, amine, and benzocrown ether, respectively. Dotted arrows indicate excitation and emitted lights. It uses H+ and Na+ ions as inputs and fluorescence as outputs. After [4].
(b) An enzymatic XOR gate. If both inputs (glucose and hydrogen peroxide) are either present or absent, the UV absorbance of the system will remain same as that of before input signal application. Otherwise, it decreases or increases, which is read as output “1.” After (3)
and hydrogen peroxide (H2O2) as inputs, respectively, and interact via nicotinamide adenine dinucleotide (NAD), which absorbance changes are used as output (Fig. 1b). A XOR gate outputs “1” when only one input is “1” (i.e., inputs are “0, 1” or “1, 0”). When inputs are both present or both absent (“0, 0” or “1, 1” case), the small enzyme network keeps balance of NAD, and the ratio of NAD + and NADH concentrations remains same. On the other hand, if the levels of two inputs are not equal, “0, 1” or “1, 0”, it loses the balance and the NADH concentration will decrease or increase. When the optical absorbance of the system with “0, 1” or “1, 0” input is compared to that of a control case (before input signals are applied), the absolute value of absorbance change, particularly in the ultraviolet (UV) range, is quite large, whereas that of “0, 0” or “1, 1” case is almost zero. Molecular logic gates can be implemented on various materials (cf. [3, 4]) as well as supermolecules and proteins described above. Those computing elements can be built from supermolecules, macromolecules (e.g., proteins), DNA, and RNA. In some cases, whole cells are also used. Various options are available for inputs and outputs of the systems. As to inputs, molecular logic devices have been developed using light, atomic/molecular ions, magnetic field, redox changes, chelating reactions, oligonucleotides and oligopeptides additions, etc. Output signals can be read by absorbance, fluorescence, electric current, etc. Based on those molecular systems, molecular analogs of digital computing devices can be realized: Boolean logic operations, comparator, flip-flop, arithmetic operations, half-adder, full-adder, multiplexer, keypad lock, memory unit, and automaton were developed so far.
DNA Computing DNA computing has become a vast research field which encompasses from formal language theory to practical implementations [5]. This article gives an overview of the first experimental demonstration by Adleman in 1994 [6], which well represents the core ideas of DNA computing. In the seminal paper, Adleman showed that DNA can be used to solve a type of combinatorial problems, called “directed Hamiltonian path problem (directed HPP).” This problem is defined as follows: Given that a finite number of nodes and directed connections, determine if there is a path which visits all the nodes only once (Hamiltonian path) in polynomial time. This is known as one of NP-complete problems, a class of decision problems that can be solved in polynomial time on a nondeterministic Turing machine [7]. In other words, the computational time for solving this type of problems increases exponentially as the size of the problem. Figure 2a shows a graph Adleman used for DNA computing to solve directed HPP. It has seven nodes labeled as A to G and connections between the nodes. For example, consider if there is a Hamiltonian path starting from node A (In principle, any node can be start of a Hamiltonian path). From A, possible connections are AB, AD, and AG. However, AG has no outgoing connections, and it cannot be a Hamiltonian path. Any paths after AD are not Hamiltonian paths either because they form a loop at some point in the path. By repeating this search process till it finds a path that connects all the nodes without any loop in the path, the path ABCDEFG can be determined as only one Hamiltonian path in the graph. The directed HPP can be solved by the brute-force search as above, but with
M
M
1452
Molecular Computing, Fig. 2 (a) A Hamilton path problem solved by Adleman’s DNA computer. (b) Encoding of nodes and connections into DNA strands
Molecular Computing
a
b E
Connection BC GGCTAA
Node B AATCCG
D B
Node C ATTGAC
Hybridization
A G GGC TA A
C
F
AA TCCG ATT GAC Ligation
the algorithm, the computational time exponentially increases even when the number of nodes increases linearly. Adleman tackled this problem by employing the massive parallelism and complementarity of DNA molecules. DNA (Deoxyribonucleic acid) consists of nucleotides in polymer form, and each nucleotides consists of a nitrogen-containing base, sugar, and phosphate base. Four types of nucleotides occurring in DNA are adenine (A), guanine (G), cytosine (C), and thymine (T). Adenine and thymine (A-T), and guanine and cytosine (G-C) can form a base pair with hydrogen bonds, and those pairs are said to be complementary. If a sequence of oligonucleotides (short strand of nucleotides) is complementary with another oligonucleotides, they can form a stable double-stranded DNA. Path finding of the grap by Adleman’s DNA computer was carried out as follows: 1. Encode the Graph into DNA Each node and connection was encoded as a singlestranded DNA oligonucleotide. First, each node is represented as a specific length of DNA sequence. In Fig. 2b, node B and C are represented as AATCCG and ATTGAC, respectively. Then, if there is a connection between two nodes, for example, connection BC, a DNA sequence representing the connection is designed as: half of the sequence is complementary to half of the sequence representing node B, and another half is to that of the node C sequence (In Fig. 2b, BC is encoded as GGCTAA). 2. Generate Random Paths Mix all the DNA sequences encoding nodes and connections in a test tube together with DNA
3.
4.
5.
6.
ligases. Single-stranded DNA strings self-assemble into double-stranded DNA due to the base pair complementarity. This process is called DNA Hybridization. DNA ligases repair breaks in complementary strands of DNA (Ligation). At this point, the computation of Hamiltonian path finding has effectively completed. Following steps are required to verify the answer of computation. Extract DNA Sequences Starting with Node a and Ending with Node G This step is done using polymerase chain reaction (PCR). Using complementary sequences of node A and node G as primers, PCR amplifies DNA sequences that start with node A and end with G only for easier detection of Hamiltonian path sequences. Extract Sequences with the Length of Exactly Seven Nodes Gel electrophoresis separates mixture of DNA fragments with different lengths by their sizes. By comparing with molecular weight markers, an area in the gel which contains DNA sequences with the length of exactly seven nodes can be spotted. Extract Sequences Containing All the Nodes The last verification process is done by magnetic beads. DNA fragments are first heated and denatured into single-stranded DNA sequences. Then complementary sequences of a node with magnetic beads attached bind to the single-stranded DNA and picked up by magnets. Repeat this process for each node. Check the Test Tube Check if there is any DNA sequence left in the test tube. If yes, it corresponds to a Hamiltonian path of
Molecular Computing
the graph. If not, the graph does not have any Hamiltonian paths starting from node A and ending with node G. The directed HPP is considered to be a computationally hard problem because a conventional computing system, or conceptually a Turing Machine, performs at most one action at a time step, i.e., such a system has only one single processor to process inputs. Thus, if a graph has n nodes and start–end nodes are fixed, potentially it requires (n 2)! computational steps to determine the presence or absence of a Hamiltonian path in the graph. DNA computing, on the other hand, involves a massive amount of simple processors. If a DNA computer have (n 2)! Processors, then the computation can be accomplished in a step. It has been suggested that a DNA computer could potentially execute approximately 1.2 1018 computations in a second, the energy efficiency could be up to 2 1019 computations per joule, and the memory density would be 1 nm3 per bit, all of which outperform any presentday conventional computing systems. However, several downsides of DNA computing were also pointed out [8], [1] Error rate. Computations by DNA involve a relatively large amount of errors which inherently come with the use of biological materials. They significantly reduce the reliability of computation, in particular when a number of computations are combined together [2]. Scalability. Provided that Adleman’s protocol is employed for problems with larger nodes, the amount of DNA molecules required for encoding exponentially increases, which could be over hundreds of kilograms [3]. Diffusion speed. Even if such a large amount of DNA can be is put into place, the computation relies on diffusion of DNA fragments, which does not scale up with the size of DNA. Hence, the reaction speed takes longer for a large-scale problem, but mixing is difficult because DNA molecules are fragile.
Other Molecular Computing Devices As another approach to exploit computational capabilities of DNA, Winfree proposed a experimentally implementable low-error cellular automata model that can perform Turing-universal computation using the self-assembly capability of DNA tile [9]. When carefully designed DNA strands are mixed in a solution, they self-assemble into two-dimensional shapes according to the designed sequence, called
1453
M
DNA tiles. A single-stranded DNA at the edge of a tile (“sticky end”) allows the tile to connect only to other tiles which have a complementary sticky end. This “programmability” of DNA tiles allows to self-organize a larger pattern similar to one-dimensional cellular automata (1D-CA), which development is controllable by choosing an initial pattern seed. 1D-CA is proved to have Turing universality and, in this sense, the DNAbased cellular automata system also has the computational power equivalent to a universal Turing machine. Based on this, he and his coworkers have developed tilebased DNA computing devices, such as logic gates [9], automata, gain amplifier, and neural network. A chemistry-based approach has been pursued by Adamatzky and De Lacy Costello [10]. The main focus in their approach is massively parallel nature of chemical reaction, in particular, Belousov–Zhabotinsky (BZ) reaction. This reaction is known to show excitability: If a BZ solution is stimulated (e.g., by a silver wire), the color of the solution oscillates, and propagating wave patterns are observed. Using the reaction (particularly the propagation waves and their collisions), it has been shown that Boolean logic gates, calculating Voronoi diagram, shortest path planning, robot control can be achieved.
Perspectives Although there has been a lot of research conducted on molecular computing, this field is still at an early stage of development and is unlikely that this technology would be commonly available in near future. One of the key issues for molecular computing is the application domain of the technology. It has been claimed that it could be used for life science applications, such as controlled gene regulation. However, there will be many hurdles to develop molecular computing components that work in the complex in vivo environment of living cells. Thus, numerous challenges in the molecular computing research will need to be overcome so that the molecular computing technology will come into wide use. Another issue for molecular computing devices is the development of computing concepts that make the most of molecular dynamics. Conventional computing concepts are based on the theory of Turing computability and, hence, they are irrelevant to the dynamics of materials on which computing devices are implemented.
M
M
1454
On the other hand, living organisms arguably employ radically different computing concepts that leverage the physics of biological materials. For example, selforganization, self-repair, and self-reproduction are commonly found in biological systems but are difficult to achieve with conventional computing systems. New computing paradigms, such as cellular automata, neural networks, and evolutionary computing, may provide clues to utilize the complexity of molecular dynamics for information processing. Molecular computing devices are not likely to compete with and replace conventional computing technology. Rather, they are complementary approaches to silicon-based systems, and a hybrid system which incorporates molecular devices into artificial systems would find potential application domains in, for example, medical and environmental applications. It is, thus, expected that molecular computing will be one of the important technologies, together with other technologies such as synthetic biology, in the age of biomolecular nanotechnology.
Cross-References ▶ Computational Systems Bioinformatics for RNAi ▶ Nanotechnology Applications in Polymerase Chain Reaction (PCR) ▶ Self-Assembly ▶ Synthetic Biology
References 1. Machluf, M., Orsola, A., Atala, A.: Controlled release of therapeutic agents: slow delivery and cell encapsulation. World J. Urol. 18, 80–83 (2000) 2. de Silva, A., Sandanayake, K.R.A.S.: Fluorescent pet (photoinduced electron transfer) sensors for alkali cations: optimization of sensor action by variation of structure and solvent. Tetrahedron Lett. 32(3), 421–424 (1991) 3. Katz, E., Privman, V.: Enzyme-based logic systems for information processing. Chem. Soc. Rev. 39, 1835–1857 (2010) 4. de Silva, A.P., Uchiyama, S.: Molecular logic and computing. Nat. Nanotechnol. 2, 399–410 (2007) 5. Amos, M.: Theoretical and Experimental DNA Computation. Natural Computing Series. Springer, New York (2003) 6. Adleman, L.M.: Molecular computation of solutions to combinatorial problems. Science 266(5187), 1021–1024 (1994) 7. Sipser, M.: Introduction to the Theory of Computation. Thomson Course Technology, Boston (2005)
Molecular Dynamics Method 8. Zauner, K.-P.: Molecular information technology. Crit. Rev. Solid State Mater. Sci. 30(1), 33–69 (2005) 9. Winfree, E., Yang, X., Seeman, N.C.: Universal computation via self-assembly of DNA: some theory and experiments. In: Landweber, L.F., Baum, E.B. (eds.) DNA Based Computers II: DIMACS Workshop, pp. 191–213. American Mathematical Society, Providence (1996) 10. Adamatzky, A., Costello, B.D.L., Asai, T.: ReactionDiffusion Computers. Elsevier Science, Amsterdam/Boston (2005)
Molecular Dynamics Method ▶ Computational Study of Nanomaterials: From LargeScale Atomistic Simulations to Mesoscopic Modeling
Molecular Dynamics Simulations of Interactions Between Biological Molecules and Nanomaterials ▶ Molecular Dynamics Simulations of Nano-Bio Materials
Molecular Dynamics Simulations of Nano-Bio Materials Melissa A. Pasquinelli1 and Yaroslava G. Yingling2 1 Fiber and Polymer Science, Textile Engineering, Chemistry and Science, North Carolina State University, Raleigh, NC, USA 2 Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA
Synonyms Molecular dynamics simulations of interactions between biological molecules and nanomaterials
Definitions Nano-bio materials are materials that are composed of biomolecules (protein, DNA, RNA, lipids) and nanoscale materials (nanoparticles, nanotubes, nanocrystals).
Molecular Dynamics Simulations of Nano-Bio Materials
Molecular dynamics simulation is a computer simulation technique that is based on integration of Newton’s equations of motion and reflects physical movements of atoms and molecules.
Overview Introduction Nanotechnology has received increasing public attention as more and more practical applications of nanoparticles and nanomaterials are emerging, especially in the biomedical field. For example, inorganic nanoparticles (NPs) functionalized with synthetic or biological molecules have been used in a broad range of biomedical applications, including imaging, diagnostics, and drug delivery. The advantage of using NPs in biomedical applications is that a single nanoparticle can simultaneously carry imaging probes, drug molecules, receptor-binding molecules, and antibodies on the surface, which promise the most efficient drug delivery platform and detection scheme. However, interactions of NPs with biological molecules may lead to structural and functional changes, which in turn may result in unexpected biological reactions and toxicity. For example, it has been shown that metallic NPs, quantum dots, and carbon nanotubes can translocate to the brain from different entry points, such as skin, blood, and respiratory pathways. Moreover, the properties of NPs, such as size, shape, hydrophobicity, charge density, and chemical composition, can directly influence the ability of NPs to enter the cellular environment, bind to proteins, activate receptors, and translocate and accumulate in target tissues, as depicted in Fig. 1. Biological molecules can also be used as scaffolds for precise deposition of NPs for electronic and optical devices. However, the effect of nanomaterials on the biomolecular structure and function is not yet elucidated, which limits advancements within the potential applications and the development of new ones. A better understanding of the biological effects of NPs requires knowledge of the mechanisms of interactions between NPs and biomolecules. Computer simulations can yield a wealth of insights on interfacial dynamics, structures of ligands, electrostatics, and binding of biomolecules. One of the most promising and rigorous methods for simulating complex and large-scale processes in various materials and
1455
M
biosystems is molecular dynamics (MD) simulations. MD simulations have been used for gaining comprehensive information on “nano-bio” systems, including ligand dynamics on NPs, conformational transitions of biomolecules, protein and ligand binding, enzymatic catalysis, and the dynamic, thermodynamic, and mechanical properties at different spatial and temporal resolutions [20]. Thus, MD simulations are a natural complement to experimental techniques at the nanoscale to yield insight on molecular interactions involved in nanoparticle–biomolecule binding and mechanisms of molecular recognition. Overview of Molecular Dynamics Simulations The steps of an MD simulation are given in Fig. 2. First, the initial system is built; coordinates of some specific biological molecules can be obtained from databases such as the Protein Data Bank [1], while others must be developed using in-house scripts. The initial velocities are assigned from a Boltzmann distribution at the desired temperature. The force field parameters for all of the components are assigned. The system is then evolved with Newton’s equations of motion for each Dt, which provides new positions and velocities for each atom. The potential energy of the system is calculated from force field parameters, which are comprised of both the bonded and nonbonded interactions. The bonded interactions include bond lengths, angles, and dihedrals, whereas the nonbonded terms are mainly electrostatics and van der Waals. The total potential energy can then be defined by the following equation: U¼
X bonds
k1ðr req Þ2 þ
X
k2ðf feq Þ2
angles
X Vn ð1 þ cosðn’ gÞÞ þ 2 dihedrals " #
12
6 ! X sij sij qi qj þ 4Eij þ þ ; rij rij Erij nonbonded where req and feq are equilibration structural parameters, k1, k2, and Vn are force constants, n is dihedral multiplicity, and g is phase angle for torsional angle parameters. The choice of a force field is dependent on the chemical composition of the system and the molecules
M
M
1456
Molecular Dynamics Simulations of Nano-Bio Materials
Molecular Dynamics Simulations of Nano-Bio Materials, Fig. 1 Atomistic view of the interactions of different types of nanoparticles, (a) carbon nanotubes, (b) buckyballs (fullerenes), and (c) gold nanoparticles, with biological molecules, (d) a lipid membrane, (e) a protein, and (f) nucleic acids
Build System
Dynamics
Setinitial particle positions and force field types
Molecular Dynamics Simulations of Nano-Bio Materials, Fig. 2 The main components of an MD simulation
Set Conditions Temperature,pressure, initial atom velocities
under study, and thus dictate the reliability of the results. A recent review outlines the comparison between most widely used protein force fields [2]. For example, the CHARMM27 [3] force field can be used to simulate DNA, RNA, and lipids, while for CHARMM22 [4] force field is parameterized to simulate proteins. The AMBER program contains force field parameters for DNA, RNA, and proteins (Cornell force field [5]); small molecular ligands (GAFF force field [6]); and carbohydrates (GLYCAM force field [7]). Other all-atom force fields include OPLS [8], which contains parameters for proteins, carbohydrates, and nucleic acids. United atom models such as
Evolve System
Dt
− F = d ( mv− ) /dt
Calculate new positions, velocities Calculate total energy, other properties
GROMOS [9] are also available that treat nonpolar hydrogens implicitly, which is commonly done for lipid membranes. The challenge, however, for bionano systems is that there is no generalized force field for metallic, carbon, and complex NPs and coarse-grained representations of the molecules, and thus the parameters are calculated on a case-by-case basis. In addition, all-atom MD simulations are not possible for some systems of interest and thus some level of coarse-graining is required, which necessitates a whole new set of force field parameters. An example of a commonly used coarse-grained force field for bio-nano systems is MARTINI [10].
Molecular Dynamics Simulations of Nano-Bio Materials
Solvent molecules are often included in the simulation. The most complex solvent is water, and a variety of water models have been developed; the most commonly used models are SPC, TIP3P, and TIP4P. Explicit solvent simulations are the most accurate and can provide information on specific solvent–nanoparticle or solvent–biomolecule interactions; however, they are more computationally extensive since significantly more atoms are being included in the simulations. MD simulations can be significantly accelerated by treating the solvent with continuum dielectric methods. In this case, only the intrasolute electrostatics need to be evaluated. Even though implicit MD simulations are less accurate than simulations with explicit solvent, they permit not only much longer simulations and larger molecules but also provide a variety of sampled conformations. The time length of simulations is typically based on the type of processes under study and must ensure that a sufficient amount of phase space is sampled. Typically, to observe any significant conformational change of biological molecules, the simulations time needs to be on the order of hundreds of nanoseconds. A useful measure of the accuracy of a simulation is the total energy of the system, which means that typically throughout the simulations, the total energy should not increase. The size of deviations of the total energy gives the measure of precision of the integration. The accuracy of the integration procedure depends on the value of the time step. For the simulations to be as long as possible, the time step needs to be chosen as large as possible. However, the maximum time step is limited by the highest frequency motions in the system, which is typically on the order of a femtosecond.
1457
M
investigation. Moreover, the MD simulation method is capable of providing a complete microscopic description of the dynamical processes involved. Most importantly, the reliability of MD simulations depends on accurate force fields for all system components. The classical MD simulation method also prevents study of processes like proton transfer, tunneling effects, zero-point motions, behavior of electrons and atomic nuclei, and making/breaking chemical bonds. Problems like these require modeling that applies the laws of quantum mechanics, the fundamental physics equations that describe electrons. However, because of the complex nature of the equations, these models can handle only a few atoms at a time. Some efforts are underway to create force fields that overcome these barriers. Considerable improvements in force fields have also been achieved, making simulations more reliable and accurate.
Advantages and Limitations
Application of MD Simulations to Study Nano-Biosystems In the subsequent sections, an overview of utilizing MD simulations for investigations of a variety of properties of nano-biosystems is provided. Figure 1 summarizes the types of systems that are included in this entry and specific simulation conditions for five simulations highlighted here are summarized in Table 1. The overview begins with carbon nanoparticles, which are the most actively and successfully used systems for simulations of nanomaterials. Then, the application of MD technique to other nanobiosystems, such as functionalized gold nanoparticles, will be addressed. The focus is on the interactions of these nanoparticles with lipids, proteins, and nucleic acids that have been investigated with MD simulations are then summarized.
The challenges in using an MD description of complex nano-biosystems are the severe limitations of time and length scales, limited to several hundreds of nanoseconds, and moderately sized biomolecules and NPs. However, as computer speed and memory increase, so do the simulation time and size limits. For example, for all-atom simulations the reasonable simulation time 10 years ago was on the order of 1 ns, but now about100 ns time length can be obtained for the same size of a system. The main advantage of an MD simulation is that only details of the microscopic interactions need to be specified and no assumptions are made about the character of the processes under
Interactions of Carbon NPs with Biological Molecules The most common type of carbon NPs are fullerenes or buckyballs (Fig. 1b) and carbon nanotubes (Fig. 1a). Both of these carbon NPs are important in many technologically relevant areas due to their unique properties. Buckyballs have potential as drug delivery mechanisms and medical diagnostic devices because of their size, spherical shape, stability, and the ability to be functionalized with other atoms; however, their hydrophobic character is a drawback to some medical applications.
M
M
1458
Molecular Dynamics Simulations of Nano-Bio Materials
Molecular Dynamics Simulations of Nano-Bio Materials, Table 1 Summary of the simulations highlighted in this work CNT-protein nanobiosensor [11] All-atom
C60-potassium channel [12] All-atom for C60, proteins; united atoms for lipids NAMD CHARMM27 Particle mesh Ewald
NP-membrane [13] Coarsegrained
Software Force field Electrostatics
Gromacs Amber99 Particle mesh Ewald
Ensemble
NPT (1 atm., 300 K)
NPT (1 atm., 300 K)
NPT (1 bar, 305 K)
Solvent
TIP3P water + 0.1 M NaCl 1.5 10.14–11.25 8,200–40,338
TIP3P water
water
NP-protein [14] Coarse-grained NP; all-atom protein CHARMM Param22 Generalized Born and explicit solvent Constant T (300 K, 450 K) TIP3P water
2 10–213 90,300–173,000
20 60 65,000
2 8 7,000
Level of MD
Time step (fs) Simulation time (ns) Number of atoms
Carbon nanotubes (CNTs) have unique mechanical and electrical properties, as well as strong chemical reactivity due to their high surface area. Their large diameter-to-length ratio provides unique applications, such as microcatheter reinforcements for tissue engineering scaffolds. CNTs can have an impact on biomolecules by enhancing their electrochemical reactivity and promoting protein electron transfer. They are very attractive materials for the design of electrochemical biosensors ranging from an amperometric enzyme electrode to DNA hybridization biosensors. For example, a CNT-based sensor has been developed for detection of hepatitis B virus (HBV), which enables a diagnostic application for nontreated human blood. A significant amount of MD simulations at various levels of detail have been employed to explore the interactions of CNTs and buckyballs with lipids, proteins, and nucleic acids [16–18]. These simulations have revealed the physicochemical details for a variety of applications. Such focus areas include potential mechanisms for toxicity of carbon nanoparticles and the design of engineering devices such as sensors, drug delivery vehicles, and actuators. Factors for the carbon NPs that were studied include diameter size, structural alignments, aromaticity, nanoparticle curvature, degree of hydrophobicity, and the local environment. Two recent exemplary articles are summarized next.
Gromacs Martini Particle mesh Ewald
NP-DNA [15] All-atom
NAMD CHARMM Particle mesh Ewald
NVT (300 K)
TIP3P + 0.5 M NaCl 2 8 32,000
MD Simulations of a Carbon NP Nanobiosensor
A nanobiosensor has been shown experimentally to be a feasible device for detecting adenovirus proteins, particularly the Knob proteins. This nanodevice is comprised of a single-walled CNT that is functionalized with the coxsackie-adenovirus receptor (CAR) protein. However, many questions remain about the system, including how structural rearrangements and dynamics may impact its efficiency, as well as the specificity of CAR when bound to the CNT outside its native environment. Johnson and coworkers [11] recently performed MD simulations on this nanobiosensor system to provide some molecular insights into these questions. They investigated three different types of systems in its physiological aqueous conditions CAR with a bound Knob protein, and then both systems attached to a CNT. The physiological conditions were created by adding a 100 mM concentration of Na+ and Cl- ions to the aqueous system comprised of TIP3P water molecules; an additional three Na+ ions were added to also neutralize the slightly negative charge on CAR. The CNT was covalently attached to residue Asn98 on CAR through a carboxylic acid defect on the CNT surface, which replicates the experimental conditions that are done via diimide-activated amidation. The systems were built from a crystal structure of the CAR-Knob complex from the Protein Data Bank. Since CAR is known to only undergo minor structural
Molecular Dynamics Simulations of Nano-Bio Materials
rearrangements upon binding Knob, this crystal structure was claimed by the authors to be sufficient for both the bound and unbound configurations, after the structural properties were equilibrated from an MD run; structural equilibrium was defined as when the rootmean-square-deviation of the Ca atoms no longer varied with time. The CNT was built the same length as the MD box, connecting terminal atoms across that box boundary, thus creating an infinite tube with the use of periodic boundary conditions. In addition, the positions of the CNT atoms were constrained with a harmonic potential, which reduces the computational cost of the simulations with minimal effect on the results. The atomistic MD simulations were run with a 1.5 fs time step, for around 11 ns, and with the NPT ensemble at 1 atm and constant temperature of 300 K. The use of the Amber99 force field was first validated by comparing the simulations on the systems without the CNT in the physiological aqueous environment to known experimental information, such as NMR observations of the structural and dynamical properties of CAR. For the MD simulations of the systems with the CNT present, some interesting observations were made. First, for the systems with the CNT, physisorption of the proteins onto the CNT surface was observed, which still preserved most of the structural features of the unbound CAR and unbound CAR-Knob; the most significant changes occurred in the loop regions and C-terminus since the CNT can restrict the configurational space of the bound proteins. In addition, for the CAR-Knob-CNT complex, a salt bridge and some hydrogen bonds were lost between the proteins, but not enough to impact its overall stability. Therefore, it was concluded that the CNT does not significantly perturb CAR, except for reducing the available configurational space due to confinement. These results indicate that comprised of protein receptors that rely on large conformational changes as part of their mechanisms could be significant. MD Simulations of C60 Embedded in Potassium Channels
Potassium (K+) channels are ion channels that span the cell membrane and are selective for K+. They control a wide variety of cell functions that are involved in biological processes such as nerve impulses, or smooth muscle contraction. The selectivity is due to the
1459
M
so-called selectivity filter, which is a highly conserved sequence. K+ channels are also targets for a variety of pore-blocking toxins that (1) bind to the extracellular pore entrance and prevent ion conduction; or (2) permeate the pore channel and disrupt binding sites within the pore. Since studies have indicated that fullerenes can cross the lipid membranes of cells, it is expected that they could also impact K+ channels. Kraszewski and coworkers [12] recently performed extensive MD simulations to investigate molecular aspects of how K+ channels are impacted by the presence of the C60 buckyball. They investigated three different types of channels, specifically those comprised of the proteins KcsA, MthK, and Kv1.2 embedded into a fully hydrated DOPC lipid bilayer, which included water molecules and counter ions to neutralize the overall system charge. The total system sizes ranged from 90,300 (KcsA) to about 173,000 atoms (Kv1.2). Up to six C60 molecules were put near the pore region. As a comparison, systems were built of the C60 molecules in a water box and with a fully hydrated DOPC layer without the proteins. The atomistic MD simulations were run with the NAMD software program, using the CHARMM27 force field, where united atoms were used to represent the lipid molecules and chemical bonds between hydrogen and heavy atoms were constrained to their equilibrium values elsewhere. The C60 molecules were first put near the pore on the extracellular region. The MD simulations revealed that no C60 molecules anchor or block the selectivity filter of the K+ channels. In addition, for the voltagegated K+ channel, Kv1.2, the C60 molecules bound to residues within the voltage sensor domain, which triggered a large conformational movement. When the C60 molecules were put near the pore on the intracellular region, a different behavior was observed. Quick binding of the C60 molecules was observed, which resulted in conformational changes that plugged the ionic conduction pathway. Finally, when the C60 molecules were placed within the membrane core, the C60 molecules were observed to migrate laterally toward the membrane proteins, and eventually bind to interhelical domains, which can alter their function. Therefore, these simulations revealed a variety of potential molecular mechanisms for binding of C60 nanoparticles, depending on the structure and properties of the channel as well as the initial locations of the nanoparticles.
M
M
1460
Interactions of Functionalized Nanoparticles with Biological Molecules Ligand-protected nanoparticles (Fig. 1c) consists of metallic nanocrystals coated by a monolayer of thiolated molecules. The ligands control the specific properties of the nanoparticle, such as solubility, electro-optical and catalytic behavior. Such nanoparticles can be dispersed in various solvents without irreversible aggregation. Functionalized NPs are a much more challenging case for simulations and modeling than carbon NPs due to their complexity, overall size, and features at multiple length and timescales. In addition, not only are functionalized NPs composed of various organic and inorganic components, but the nanoparticle corona of functionalized ligands is also a dynamic entity by itself. However, MD simulations have been performed to elucidate details about the structure and dynamics of these complex NPs. Recent atomistic MD simulations of NPs functionalized with alkanethiol ligands provided atomistic data on the ligand shell organization and indicated that the ligand’s tilt angle is affected by solvent, temperature, nanoparticle size, and the ligand tail length [19, 20]. The complexity of ligand behavior, recognition, dynamic interactions, and assembly between ligand functionalized NPs and biomolecules is not trivial to simulate and challenges the current limits of the MD simulation techniques. To date, most of the simulations that address the interactions between functionalized NP and biological molecules (proteins, ligands, and nucleic acids) have a simplified representation of the nanoparticle and a few simulations have explicit ligands represented via a united-atom approximation, which have provided insights into their biorecognition properties. A few such studies are summarized next. Functionalized NPs with Lipids
Properties of functionalized NPs, such as size and type of ligands, can directly control the ability of a NP to enter the cell. The properties that control the NP uptake in cells can cause cellular response and undesirable toxic effects and can also control the performance of NPs in imaging, biosensing, and drug delivery. An understanding of the processes controlling the uptake and accumulations of the NPs by the cell is important for NP design, assessment of nanoparticle toxicity, and improvement of the NP properties for use in imaging
Molecular Dynamics Simulations of Nano-Bio Materials
and drug delivery applications. However, the rules of uptake kinetics have not yet been determined. In order to gain a deep understanding of NP translocation through a membrane, a complete picture of the processes at the molecular level is needed, which can be delivered through MD simulations. To the best of our knowledge, only coarse-grained MD simulations of NP interactions with the membrane have been used. The advantage of the coarse-grained representation of lipids and NPs is the ability to use larger time and size scales at the expense of observing the intricate behavior of H-bonding network formation and the atomic interactions. Lin and coworkers [13] recently performed coarsegrained MD simulations of penetration of lipid membrane by gold nanoparticles. They investigated two kinds of lipids, DPPC and DPPG, and NPs with alkyl thiol chains that can contain two different functional group, ammonium and carboxylate, using GROMACS 3.3.2 package [9]. The coarse-grained representation of NPs and lipids was compatible with the MARTINI force field [10]. The gold NP was 2.2 nm in diameter functionalized with 104 alkyl thiol ligands. The NPs were charged with 70% cationic or anionic coverage represented by the charge on the end functional ligand group and both electronegative and electroneutral bilayers were examined. The coarse-grained force field for the NP was parameterized to represent experimental values of the radius of gyration, the diffusion coefficient, and the average carbon distance. The membrane bilayer was composed of 1,152 coarse-grained lipids in a periodic box filled with explicit water and ions. The initial placement of NPs was chosen to be 7 nm above the bilayer center. The system was carefully equilibrated for a cumulative time of 20 ns before the production run of 40 ns was performed with a 20 fs time step. The simulations revealed that NP charge and surface change density (electrostatic interactions) control the ability of the NP to penetrate, adhere, or repel from the lipid bilayer. Interestingly, NP penetration into the lipid bilayer causes lipid disarrangement and formation of a hydrophilic pore that can transport water molecules. The extent of membrane disruption was determined to depend on the NP charge density. Thus, the simulations indicate that control over the NP cellular uptake can be achieved through charge density of the functional groups.
Molecular Dynamics Simulations of Nano-Bio Materials
Functionalized NPs with Proteins
The effect of protein interactions with various NPs and nanomaterials can shed light on their possible toxicity and aggregation behavior linked to some neurological disorders. The interface, which is composed of covalent or secondary bonding interactions between NPs and proteins, is highly complex. The properties of this interface may largely affect the protein structure and function. Only a few MD studies were devoted to study protein–NP interactions. Simulations by Aubin-Tam and coworkers [14] were performed on nanoparticle–protein conjugates comprised of Cyt c from Saccharomyces cerevisiae to understand the effect of the NP ligand on protein structure. The NP was modeled as a dummy atom that is 1.5 nm in diameter with Lennard Jones parameters. A total of 21 bathophenanthroline disulfonate (BPS) ligands were uniformly distributed on its surface. The coordinates for the Cyt c protein structure was obtained from the PDB databank (PDB id 2BCN), mutated and covalently linked to the NP surface. MD simulations were performed for 8 ns at 300 and 450 K using CHARMM and param22 all-atom force-fields and the generalized Born implicit solvent. Simulations were also performed in explicit solvent to verify the results in implicit solvent. MD simulations of Au nanoparticle/Cyt C conjugates indicated that NP labeling induces sitespecific, local-to-global structural perturbations to the protein [14]. Protein conformations obtained at 300 and 450 K showed that side chains of positively charged residues interact with BPS ligands, and that perturbation of the secondary structures depends on the NP attachment site, which agrees well with the experimental observations. MD simulations of the semirigid, coarse-grained protein/nanoparticle model revealed the importance of lysine and arginine amino acids on the interactions of Cyt C and nanostructured surfaces, which suggest the design rules for peptides with enhanced surface adsorption [21, 22]. Thus, MD simulations can be successfully employed for identifying mechanisms responsible for fine tuning of the biorecognition properties of the nanocomplex, and more detailed investigations are needed in this area. Functionalized NPs with Nucleic Acids
MD studies have been successful at corroborating and elucidating experimental findings, or exploring fundamental aspects of DNA dynamics and structure [23]. For
1461
M
example, MD simulations were applied to a 12-basepair DNA duplex tethered on a neutral silica surface to assess its structural and orientation changes [24]. MD simulations were used to study the conformation of seven DNA strands attached to the (111) gold surface via a six-carbon alkylthiolate linker [25]. These simulations indicated that while the overall structure is disordered, both base stacking and non-Watson-Crick base pairing can be formed between short single-stranded regions of adjacent DNAs. Moreover, the separation distance between DNAs is similar to one calculated by theory and estimated from experiments. MD simulations of a 2-nm gold nanoparticle functionalized with four single-stranded DNA molecules were used to determine the structure and dynamics of DNA thiolates on a gold surface [15]. The 10-mer of ssDNA polythymines and poly adenines were symmetrically distributed on four (111) gold surfaces of an NP. The interactions between gold atoms were represented by Lennard Jones parameters, and the CHARMM force field was used for the ssDNAs. The resultant DNA-functionalized NP structure was placed in a box with water and 0.5 M NaCl. After equilibration, the NVT Langevin production simulations were carried out using NAMD at 300 K for 8 ns. The results indicated that base–base stacking is better for poly adenines than for poly thymines and the effective radius of DNA-NPs agrees well with experiment. Thus, the MD simulations provided an insight onto the dynamics of DNA strands on the gold NP surface. In addition, atomistic MD was used to study the behavior of the DNA strands during the assembly of layer-by-layer DNA capsules, which produced a prediction of the optimal length for the NP assembly [26]. Thus, MD simulations were able to provide important information on the behavior of DNA strands on the surface of the NPs; however, the DNA-driven assembly is outside of the reach for atomistic MD simulations. A recent coarse-grained model [27] of DNA-driven NP assembly indicated the possibility to explain the DNA-induced crystallization of NPs and established a general phase diagram. Summary and Future Directions The future development of many biomedical devices and the optimization of current systems are dependent on the fundamental knowledge of properties of the interface between NPs and biomolecules. MD simulations are capable of producing such fundamental
M
M
1462
knowledge, but on limited time and length scales. While the interactions of carbon-nanostructures are significantly explored by MD simulation techniques, the use of MD simulations for other systems such as colloidal gold and magnetic and inorganic nanoparticles is not covered as much. With the increase in computational power and improvements of algorithms, as well as the development of new hardware systems such as GPU systems, it is expected that more MD simulations will be used for detailed studies of interactions between biomolecules and NPs. Further work is needed on the understanding of interactions between nanomaterials and biomolecules, such as on the surface of inorganic and organic nanoparticles and nanofilms. The investigations at the atomistic and molecular levels will provide information on collective biomolecular behavior and change in properties and will ultimately enable access to new and valuable future applications.
Cross-References ▶ Ab Initio DFT Simulations of Nanostructures ▶ Computational Study of Nanomaterials: from Large-Scale Atomistic Simulations to Mesoscopic Modeling
References 1. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N., Bourne, P.E.: The Protein Data Bank. Nucleic Acids Res. 28, 235–242 (2000) 2. Guvench, O., MacKerell Jr., A.D.: Comparison of protein force fields for molecular dynamics simulations. Methods Mol. Biol. 443, 63–88 (2008) 3. Case, D.A., Darden, T.A., Cheatham I, T.E., Simmerling, C.L., Wang, J., Duke, R.E., Luo, R., Crowley, M., Walker, R.C., Zhang, W., Merz, K.M., Wang, B., Hayik, S., Roitberg, A., Seabra, G., Kolossva´ry, I., Wong, K.F., Paesani, F., Vanicek, J., Wu, X., Brozell, S.R., Steinbrecher, T., Gohlke, H., Yang, L., Tan, C., Mongan, J., Hornak, V., Cui, G., Mathews, D.H., Seetin, M.G., Sagui, C., Babin, V., Kollman, P.A.: AMBER 10. University of California, San Francisco (2008) 4. MacKerell Jr., A.D., Banavali, N., Foloppe, N.: Development and current status of the CHARMM force field for nucleic acids. Biopolymers 56, 257–265 (2000) 5. MacKerell, A.D., Bashford, D., Bellott, M., Dunbrack, R.L., Evanseck, J.D., Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph-McCarthy, D., Kuchnir, L., Kuczera, K., Lau, F.T.K., Mattos, C., Michnick, S., Ngo, T., Nguyen, D.T., Prodhom, B., Reiher, W.E., Roux, B., Schlenkrich, M., Smith, J.C., Stote, R.,
Molecular Dynamics Simulations of Nano-Bio Materials
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Straub, J., Watanabe, M., Wiorkiewicz-Kuczera, J., Yin, D., Karplus, M.: All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616 (1998) Duan, Y., Wu, C., Chowdhury, S., Lee, M.C., Xiong, G.M., Zhang, W., Yang, R., Cieplak, P., Luo, R., Lee, T., Caldwell, J., Wang, J.M., Kollman, P.: A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J. Comput. Chem. 24, 1999–2012 (2003) Wang, J., Wolf, R.M., Caldwell, J.W., Kollman, P.A., Case, D.A.: Development and testing of a general amber force field. J. Comput. Chem. 25, 1157–1174 (2004) Damm, W., Halgren, T.A., Murphy, R.B., Smondyrev, A.M., Friesner, R.A., Jorgensen, W.L.: OPLS_2002: a new version of the OPLS-AA force field. Abstr. Pap. Am. Chem. Soc. 224, U471–U471 (2002) Christen, M., Hunenberger, P.H., Bakowies, D., Baron, R., Burgi, R., Geerke, D.P., Heinz, T.N., Kastenholz, M.A., Krautler, V., Oostenbrink, C., Peter, C., Trzesniak, D., van Gunsteren, W.F.: The GROMOS software for biomolecular simulation: GROMOS05. J. Comput. Chem. 26, 1719–1751 (2005) Marrink, S.J., Risselada, H.J., Yefimov, S., Tieleman, D.P., de Vries, A.H.: The MARTINI force field: coarse grained model for biomolecular simulations. J. Phys. Chem. B 111, 7812–7824 (2007) Johnson, R.R., Rego, B.J., Johnson, A.T.C., Klein, M.L.: Computational study of a nanobiosensor: a single-walled carbon nanotube functionalized with the coxsackieadenovirus receptor. J. Phys. Chem. B 113, 11589–11593 (2009) Kraszewski, S., Tarek, M., Treptow, W., Ramseyer, C.: Affinity of C-60 neat fullerenes with membrane proteins: a computational study on potassium channels. ACS Nano 4, 4158–4164 (2010) Lin, J.Q., Zhang, H.W., Chen, Z., Zheng, Y.G.: Penetration of lipid membranes by gold nanoparticles: insights into cellular uptake, cytotoxicity, and their relationship. ACS Nano 4, 5421–5429 (2010) Aubin-Tam, M.E., Hwang, W., Hamad-Schifferli, K.: Sitedirected nanoparticle labeling of cytochrome c. Proc. Natl. Acad. Sci. USA 106, 4095–4100 (2009) Lee, O.S., Schatz, G.C.: Molecular dynamics simulation of DNA-functionalized gold nanoparticles. J. Phys. Chem. C 113, 2316–2321 (2009) Monticelli, L., Salonen, E., Ke, P.C., Vattulainen, I.: Effects of carbon nanoparticles on lipid membranes: a molecular simulation perspective. Soft Matter 5, 4433–4445 (2009) Makarucha, A.J., Todorova, N., Yarovsky, I.: Nanomaterials in biological environment: a review of computer modelling studies. Eur. Biophys. J. Biophys. Lett. 40, 103–115 (2011) Ke, P.C., Lamm, M.H.: A biophysical perspective of understanding nanoparticles at large. Phys. Chem. Chem. Phys. 13, 7273–7283 (2011) Lane, J.M.D., Grest, G.S.: Spontaneous asymmetry of coated spherical nanoparticles in solution and at liquidvapor interfaces. Phys. Rev. Lett. 10(4), 235501 (2010) Ghorai, P.K., Glotzer, S.C.: Molecular dynamics simulation study of self-assembled monolayers of alkanethiol
Molecular Modeling on Artificial Molecular Motors
21.
22.
23.
24.
25.
26.
27.
surfactants on spherical gold nanoparticles. J. Phys. Chem. C 111, 15857–15862 (2007) Hung, A., Mwenifumbo, S., Mager, M., Kuna, J.J., Stellacci, F., Yarovsky, I., Stevens, M.M.: Ordering surfaces on the nanoscale: implications for protein adsorption. J. Am. Chem. Soc. 133, 1438–1450 (2011) Auer, S., Trovato, A., Vendruscolo, M.: A condensationordering mechanism in nanoparticle-catalyzed peptide aggregation. PLoS Comput. Biol. 5, e1000458 (2009) Mitchell, J.S., Laughton, C.A., Harris, S.A.: Atomistic simulations reveal bubbles, kinks and wrinkles in supercoiled DNA. Nucleic Acids Res. 39, 3928–3938 (2011) Wong, K.Y., Pettitt, B.M.: A study of DNA tethered to surface by an all-atom molecular dynamics simulation. Theor. Chem. Acc. 106, 233–235 (2001) Yao, L., Sullivan, J., Hower, J., He, Y., Jiang, S.: Packing structures of single-stranded DNA and double-stranded DNA thiolates on Au(111): a molecular simulation study. J. Chem. Phys. 12(7), 195101 (2007) Singh, A., Snyder, S., Lee, L., Johnston, A.P.R., Caruso, F., Yingling, Y.G.: Effect of oligonucleotide length on the assembly of dna materials: molecular dynamics simulations of layer-by-layer DNA films. Langmuir 26, 17339–17347 (2010) Lara, F.V., Starr, F.W.: Stability of DNA-linked nanoparticle crystals I: effect of linker sequence and length. Soft Matter 7, 2085–2093 (2011)
Molecular Encapsulation ▶ Nanoencapsulation
Molecular Layer Deposition (MLD) ▶ Atomic Layer Deposition
Molecular Modeling on Artificial Molecular Motors Yunfeng Shi Department of Materials Science and Engineering, MRC RM114, Rensselaer Polytechnic Institute, Troy, NY, USA
Synonyms Catalytic bimetallic nanorods; Catalytic Janus particle; Nanomotors
1463
M
Definition Molecular motors are nanoscale objects capable of converting a certain energy source (chemical energy, photons, etc.) to mechanical work.
Background One universal driving force for nanotechnology is miniaturization. The realization of nanometer scaled machineries needs nanometer scaled motors to provide mechanical power for nanofluidic and nanorobotic devices. Another underlying theme for nanotechnology is biomimetic. The fascinating biological world, as optimized by billions of years of evolution, provides answers of science and engineering challenges. In this regard, biomolecular motors play important roles in almost all important physiological processes in a cell, including transportation, actuation, nanostructure assembly, and disassembly. All the aforementioned functionalities are performed at the nanometer scale, thus molecular motors are important enablers in nanotechnologies inspired by biology. Molecular motors can be classified as active nanostructures (capable of energy conversion or structural transformation upon a local stimuli), as opposed to passive nanostructures (usually operate within a bulk and are not designed to function upon local stimuli). According to Roco’s classification, molecular motors belong to the second generation of nanotechnology applications and will play an important role in the third generation (nanosystems) [1]. Survey of Experiments Artificial molecular motors have been fabricated with supramolecules [2], metal nanocrystal, [3] or magnetic colloidal chains [4]. The operation of these motors is mostly powered by an external agitation such as light, an electric/magnetic field, or oscillating chemical concentration. Therefore, unlike biomolecular motors, the motions of the above artificial motors are not strictly autonomous. Recently, it has been shown that catalytic bimetallic nanorods (illustrated in Fig. 1) exhibit interesting self-propelled motion in hydrogen peroxide solutions [5, 6]. Interestingly, these nanorods have comparable size, speed, and energy source (catalytic chemical reactions) to bacteria.
M
M
1464
Molecular Modeling on Artificial Molecular Motors, Fig. 1 Illustration of a bimetallic nanorod which relies on decomposition of hydrogen peroxide
Molecular Modeling on Artificial Molecular Motors
Au 2H2O2
Pt 2H2O + O2
Various propulsion mechanisms have been proposed, including bubble recoil [5], interfacial tension [6], self-electrophoresis [7], and self-diffusiophoresis [8]. The driving force for different molecular motors is thought to be system-dependent, and multiple propulsion mechanisms might operate at the same time. For instance, self-electrophoresis seems to be responsible for propulsion in bimetallic nanorod systems, while self-diffusiophoresis is likely to play an important role for systems without electrochemical reactions. An apparent universal feature is to achieve asymmetric reactions by placing the catalysts on one side of the motor or partially blocking the release routes of the products. The smallest dimension of the catalytic molecular motors investigated in the above studies is about one third of a micron. This is largely due to the technical difficulty in discerning directional motion for objects at submicron sizes [9]. Therefore, it is unclear whether this type of catalytic system can operate when being scaled down to the tens of nanometers (or less) scale. In addition, the thermodynamic efficiency of catalytic molecular motors has not been directly measured. By considering the viscous drag force instead of coupling external loads, Paxton et. al. [10] estimated the efficiency of bimetallic nanorods to be on the order of 109. The fact that multiple (sometimes competing) propulsion mechanisms operate simultaneously makes it difficult to rationally optimize these nanomotors. Roles of Simulations Molecular-level simulations are ideal tool to gain mechanistic details in complex systems, such as the energy conversion process in a molecular motor. In the spirit of taking simulations as “controlled numerical experiments,” the goal of molecular simulation is to simplify, instead of reproduce the exact
Molecular Modeling on Artificial Molecular Motors, Fig. 2 Illustration of asymmetric product distribution caused by the catalytic patch
experimental condition, so as to distil the precise mechanism at work. Specific objectives include enhancing the overall energy conversion efficiency and exploring the feasibility of molecular motors that are at least one to two orders of magnitude smaller than current experimental motors. On what follows, three main simulation approaches to model artificial molecular motors based on asymmetric chemistry are described. Constitutive Simulation Approach The first modeling strategy stems from existing treatment of interfacial forces at the continuum level. Although not strictly at the molecular level, this method considers molecular-level events such as reaction and diffusion. Generally, the initial step is to solve the diffusion–reaction equation written below, subjected to the proper boundary conditions such as the motor dimension, catalytic site locations. @t rð~ r ; tÞ DH2 rð~ r ; tÞ ¼
dNp ðtÞ 3 d ð~ r ~ rs Þ dt
(1)
r is the density of the products, D is the diffusivity, dNp ðtÞ is the catalytic reaction rate at site ~ rs . As and dt illustrated in Fig. 2, a concentration gradient is established due to the chemical reactions localized near the catalytic site. Thus, products diffuse from high concentration region to low concentration region. Moreover, solvent flow can also occur driven by
Molecular Modeling on Artificial Molecular Motors
1465
M
osmotic forces. Both processes can deform the interfacial layer between the motor–fluid interface and lead to motion of the motor itself. From the interfacial theory [11], the drift velocity of the motor is proportional to first momentum of the product concentration distribution. In other words, if the product distribution is symmetric, the drift velocity of the motor is zero, and vice versa. As the catalytic site is decorated on the motor, a concentration gradient will be quickly established even though the motor is moving around. Thus, the molecular motor is self-propelled without any external intervention. Monte Carlo Simulation Approach The second approach to treat catalytic chemical reaction is through Monte Carlo (MC) operations, which can be coupled to molecular dynamics (MD) simulations [12]. As illustrated in Fig. 3, a nanodimer consists of a catalytic sphere and a noncatalytic sphere. A collision between the reactant particle and the catalytic sphere will trigger a Monte Carlo operation that will convert it to a product particle at a probability determined by reaction kinetics and temperature. However, neither a collision between the product particle with the catalytic sphere nor collisions between any particles with the noncatalytic spheres will lead to chemical reaction. Thus, the irreversible catalytic chemical reaction can be accomplished. This method has been used in conjunction with multiparticle collision (MPC) to study the self-motile behavior of nanodimers. MPC is a mesoscopic particle-based simulation method, which is capable of capturing the essential features of MD simulations. As shown in Fig. 3, the product concentration is again distributed asymmetrically around the nanodimer which is a result of the catalytic chemical reaction. By controlling the interparticle interaction, dimer separation, and temperature, the directed motion of the nanodimer can be controlled. The orientational relaxation time for this simulated nanodimer is very long so that the directed motion of the nanodimer is not overwhelmed by the random Brownian motion. The main disadvantages of this method are the crude representation of chemistry. Particularly, the exothermicity of the chemical reaction is not described in this approach. Moreover, chemical reaction is only possible at the catalytic site. Thus noncatalytic chemical reactions, which are nonetheless present, are not described here.
Molecular Modeling on Artificial Molecular Motors, Fig. 3 Illustration of a nanodimer consisting of an inert particle and a catalytic particle
Reactive Molecular Dynamics Simulation Approach The third approach is to utilize reactive molecular dynamics simulations to describe the full dynamics of molecular motors. The most important component of this approach is a reactive force field to describe catalytic chemical reaction. Many reactive force fields (REBO [13], ReaxFF [14], etc.) based on the chemistry of bonding have been used to successfully describe hydrocarbon systems, among other systems. However, little work has been done to model catalytic reactions, for instance, the decomposition of hydrogen peroxide with platinum as catalysts. An alternative approach is the recently proposed reactive state summation (RSS) scheme, which does not aim for accuracy but for efficiency and tunability [15]. RSS scheme has been used to describe phase transformations in silica and boron oxide, detonation of nitrogen cubane, growth of nanoporous carbon, and catalytic chemistry in a model fuel-catalyst system [16]. A chemical reaction is accomplished by a transition from one reactive state to another. The reactive state can be characterized conveniently by partial coordination. For instance, oxygen next to just another oxygen is in diatomic oxygen, while oxygen next to just two hydrogen is in a water molecule. The essence of the RSS scheme is to use nonreactive two-body force field to describe individual reactive states, which will be switched on or off by
M
M
1466
Molecular Modeling on Artificial Molecular Motors
system can be written as a combination of all nonreactive pair interactions weighted by the above switching function: PE ¼
XX p and the stored energy <Estored> are related [16] by
Damping C
2p < Edissip > ¼ < P > ð2p=oÞ ¼ C< x_ > o 2
¼ pCox20
x=0
< Estored > ¼ M o2 þ o20 x20 =4 :
Mass
(3a) (3b)
M
If the average energy stored in the resonator is very large compared to the energy dissipated (¼ energy supplied, for steady-state) per cycle, one may anticiF cos(wt) pate lower noise operation when driven, and a slower decay of the oscillation amplitude envelope when the Nanomechanical Resonant Sensors and Fluid Interactions, driving force is switched off, allowing for a timeFig. 3 Single degree of freedom model for resonator domain measurement of damping. These are measures of the quality of a resonator, expressed by the may not be much larger than 1. A summary is included magnitude of the quality factor, defined [16] as later in this section. <Estored > The displacement with time, {x(t)}, of a resonator Q ¼ 2p (4) with mass M, damping C, and stiffness K, driven har<Edissip > monically with a forcing amplitude F0 at a frequency o, as seen in Fig. 3, is Using Eqs. 3a and 3b, this yields M€ x þ Cx_ þ Kx ¼ F0 cosðotÞ
(1)
A surrogate form of the Q-factor, defined using material properties alone, is the dimensionless ratio pffiffiffiffiffiffiffiffi Qp , given by Qp ¼ MK =C ¼ Mo0 =C; where pffiffiffiffiffiffiffiffiffiffi o0 ¼ K=M is the undamped resonance frequency. Qp is related to the commonly used dampingp ratio ffiffiffiffiffiffiffiffi z; through z ¼ C=Cc ¼ 1=ð2 Qp Þ, where Cc ¼ 2 MK is the critical value of damping. Damping decreases the resonance frequency od, from the undamped resonance value o0, as given by the relation, h i12 1 od ¼ o0 1 1=ð2Q2p Þ ¼ o0 1 2z2 2 :
(2)
For steady-state resonance, the mean power supplied by the driving force balances the energy dissipation rate, and the total stored kinetic energy and spring potential energy, averaged over many cycles,
M o20 þ o2 K þ Mo2 Q¼ ¼ : 2Co 2Co
(5)
For forcing at the damped resonance frequency od, using Eq. 5 with the definition for Qp, o ¼ od, with od given by Eq. 2, the expression for the quality factor reduces to Q2p ð1=4Þ Q ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : Q2p ð1=2Þ
(6)
Most applications aim for quality factors Q >> 1, for which od reduces to o0 and Q to Qp, as seen from Eqs. 2 and 6. It is also easy to show [16] that for high Q-factors, the measured resonance spectrum, expressed as a plot of amplitude-squared against forcing frequency, has a full width at half maximum (FWHM) height, as seen in Fig. 1, equal to damping
N
N
1636
Nanomechanical Resonant Sensors and Fluid Interactions
Nanomechanical Resonant Sensors and Fluid Interactions, Table 1 Significant deviations at low Qp Qp 1,000 100 10 5 2 1
z 0.0005 0.0050 0.0500 0.1000 0.2500 0.5000
(Q – Qp)/Qp 3.12E-14 3.13E-10 3.14E-06 5.10E-05 2.22E-03 6.07E-02
(od – o0)/o0 2.50E-07 2.50E-05 2.50E-03 1.01E-02 6.46E-02 2.93E-01
coefficient divided by mass, C/M, which allows for a simple measurement of damping. The ringdown method is an alternative to the spectral technique above. It is a time-domain technique for measuring damping. In this method, the forcing for a driven nanoresonator is turned off, and the amplitude of the oscillation begins to decay with time t as exp(t/t), where the time constant t ¼ C/M ¼ o0/Qp [16]. Qp can be obtained from t, the time taken for the measured amplitude envelope to reduce to 1/e of the previous value. Nanoresonator operation in a fluid environment is typically associated with high damping (C). Some consequences of low Qp due to high damping may now be examined. Low Q-factor spectra, Q 1, have been encountered in measurement attempts in liquids. For low Q-spectra, the spectral width (FWHM) may be hard to obtain and the spectra are noisier – the signalto-noise ratio, SNR Q1/2. Using the best-fit Lorentzian curve solution of Eq. 1 to low-Q experimental spectra is preferable for the evaluation of the damping. As seen in Table 1, obtained using Eqs. 2 and 6, the deviations of Q from Qp and od from o0, become significant and increase rapidly as Qp approaches a value of 1. As an example, the fractional shift of the resonance peak frequency due to damping, changes from 0.25 ppm (parts per million) at Qp of 1,000 to 65,000 ppm at Qp of 2. Sensitive mass detection aims for resolution of frequency shifts of the order of 10 ppm or less, between mass-loaded and unloaded measurement states. For Qp 1, small variations in fluid damping between these states could themselves cause frequency shifts that will exceed the frequency shift from small added mass. The experimental evaluation of the Q-factor (sketch in Fig. 1) shows that the accuracy of any sensing scheme dependent on location of the peak frequency will also degrade significantly for broad peaks (low Q). In liquids and dense gases,
there are cases where fluid damping can be high enough to prevent the detection of the resonance peak altogether. Ringdown measurements also become less accurate or impossible, if Q is low enough to reduce the displacement amplitude to zero in a time comparable to one oscillation time period. Finally, when an electrostatic or piezoelectric drive is used for exciting and driving the nanostructure through resonance, the large range of swept frequency required to capture low Q-factor spectra, has the potential to create nonconstant forcing F0. This arises due to changes in the amplitude of the driving signal, due to significant frequency-dependent changes in the impedance of the driving circuit (parasitics). Techniques for enhancing the Q-factor Qp of nanoscale resonators in fluids could include the use of resonance at higher frequency harmonics (higher o0), and an increase of effective string-like stiffness (higher K, o0), by using tensioned beams [7]. These have demonstrated viable resonance with high quality factor in some fluid measurements, and should prove useful for many applications. However, their reduced sensitivity to fluid damping makes them less suitable as measurement devices for well-resolved and accurate measurement of the value of fluid damping, when that is the object of the study. To complete this section, the experimental steps used for the evaluation of gas damping force on nanomechanical resonators are reviewed. A few results from mechanics of materials and vibration of continuous beams will be useful to see the interplay of geometry, material characteristics, and measurement artifacts related to nanoresonator fabrication and transduction schemes that are characteristic of nanoresonator measurements. Consider fluid interactions with the nanoresonator that are linear – that is, they are not functions of products of the displacement x or its time derivatives. The nanoresonator motion may then be represented by the equation governing the displacement x, ðMs þ Mf Þ€ x þ ðCs þ Cf Þx_ þ ðKs þ Kf Þx ¼ F0 cosðotÞ:
(7)
In Eq. 7, subscripts s, f denote structural intrinsic (solid material) and fluid quantities respectively. The fluid force on the beam in phase with the velocity is represented using fluid damping Cf ; the force on the
Nanomechanical Resonant Sensors and Fluid Interactions
1637
beam due to the fluid in phase with the acceleration can be significant in liquids, and is represented by the fluid added mass term Mf ; a restoring spring force component is represented by a fluid stiffness term Kf. A measurement at high vacuum, where fluid molecules have no appreciable effect, will involve only the intrinsic quantities. In nanoresonators, Eq. 2, with typical Qp,s >> 1, based on intrinsic damping Cs, shows that Cs is not large enough to cause an appreciable shift of the resonance peak frequency from the undamped value. The undamped resonance frequency for transverse flexural vibrations of a solid beam having l, Ms, Ks, rs, E, and I as the length, mass, stiffness, density, Young’s modulus, and cross-sectional moment of inertia, is [17] os o0 ¼ ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ks =Ms ¼ b2n EI=Ms l3 b2n
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1=12Þ ðE=rs Þ ðt=l2 Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ b2n ð1=16Þ ðE=rs Þ ðd=l2 Þ
(8)
(8b)
(9a) (9b)
with Eq. 9a for the rectangular beam and (Eq. 9b) for the circular section nanowire. If squeeze-film damping is negligible, the effective stiffness in the fluid remains unchanged from its intrinsic value in vacuum, so that Ksþf = Ks = Ms o02. Substituting for effective mass and damping, Msþf = Ms þ Mf and Csþf = Cs þ Cf, yields the damped resonance frequency os+f and quality factor Qp,s+f for the beam in fluid, osþf ¼ o0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1 1=ð2Q2p;sþf Þ
(10)
(11)
For rarefied gas damping experiments with nanoresonators, the situation is usually simplified with fluid added mass Mf being negligible. A first measurement may be made at high vacuum, with no gas damping (Cf ¼ 0), for which the Q-factor value, Qs, depends only on the structural properties of the solid nanowire, Ms, Cs, and Ks. Following this, the ambient gas pressure in the chamber is increased in steps, and the frequency osþf and quality factor Qs+f are measured at each step. In experiments, we assume the surrogate property-related Q-factor (Qp) relations hold true for the measured Q-factor values. This yields Qs ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ms Ks =Cs ¼ Ms o0 =Cs ¼ ð1=Cs Þ b2n Ms EI=l3 (9)
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ð1=Cs Þ b2n ðErs =12Þ ðwt2 =lÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ð1=Cs Þ b2n ðErs =16Þ ðpd 3 =4lÞ
Qp;sþf ¼
(8a)
where Eq. 8a is for a rectangular beam of width w, and thickness t parallel to the direction of motion and Eq. 8b is for a circular section beam (nanowire) of diameter d. For rigid clamping, b0 is 1.88 for cantilever beams and 4.73 for doubly-clamped beams vibrating in the fundamental mode (n ¼ 0). The surrogate Q-factor for the beam in vacuum is given by, Qp;s ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðMsþf Ks Þ=Csþf "
0:5 # 1 þ Mf =Ms ¼ Qp;s 1 þ ðCf =Cs Þ
N
Qsþf ¼
Ms o0 Cs
(12a)
Ms o0 : Cs þ Cf
(12b)
Using Eq. 12a and mass Ms from known density and geometry of nanoresonator, the value of intrinsic damping Cs may be obtained from measurement at high vacuum. Subsequent measurements of the Q-factor Qs+f, at each value of ambient gas pressure, are related using Eqs. 12a and 12b to obtain
Cf 1 Qsþf ¼ Qs 1 þ ; Cs
(13)
from which the gas damping Cf can be calculated. For liquids or dense gases, the added fluid mass Mf is not negligible, and the full Eq. 11 must be considered (see [9]), in place of its subset Eq. 13. These relations provide a summary for the conduct of experimental work and are useful in a number of ways. For measurements in vacuum, they establish the expected scaling of frequency and Q-factor with parameters that depend on geometry, material, mode shape, and type of clamping. They allow evaluation of the rigidity and repeatability of the clamps (through the value of mode constant bn in Eq. 9) if the geometric parameters and material properties are accurately known. Most importantly, they allow experimental
N
N
1638
testing of models for fluid damping Cf and fluid added mass Mf. These relations also show how nanoresonator materials may be chosen to increase or decrease the Qfactor range for a given range of fluid damping. This is discussed in the following section detailing the coupling of fabrication, materials, and transduction choices.
Nanomechanical Resonant Sensors and Fluid Interactions
a
Fabrication and Materials Nanomechanical resonators are suspended structures – most often long and prismatic, clamped at one end (cantilevered) or both ends (doubly-clamped). How does the choice of geometry, material, fabrication, and transduction schemes influence the operation of nanomechanical resonant sensors in fluids? Earlier discussion of Fig. 2 showed the strong coupling between these choices and the necessity of using the framework of mechanics of fluids, mechanics of materials, and nonlinear/stochastic dynamics as guides to make sensor design choices. Fabrication choice plays a dominant role and it is worth briefly reviewing two fabrication paradigms – termed top-down and bottom-up fabrication and an example of the synergy that may be obtained by hybrid methods combining features of each. Top-down Methods Top-down methods are named for the processes of patterning and etching from larger planar sheets of material (top-level) to obtain smaller desired structures (bottom-level), typically as large arrays. No assembly is required as the structures are built in the desired locations. Typically, a sacrificial layer is sandwiched between the substrate and a film of the intended nanoresonator material (device layer). An etch mask, usually made from a photosensitive chemical that resists etching (photoresist) is used while etching through the device layer, leaving it in the shape of the desired resonator array. A second etch targeted at the sacrificial layer material is then used to remove material beneath the devices and create suspended resonators. Doubly-clamped resonators created by this method are shown in Fig. 4. Optical lithography allows transferring micrometer scale features on to large (upto centimeter-scale) areas of photoresist for patterning thousands of devices using a single exposure through a patterned mask. This well-established technique, long used in
b
Nanomechanical Resonant Sensors and Fluid Interactions, Fig. 4 (a) Doubly-clamped nanoresonator beams of different lengths with the same undercut. Figure reprinted from Ref. [18], with permission. (b) Distortion of long thin metal cantilevers attempted using a top-down process (Figure reprinted from Ref. [19], with permission)
microelectronics to make integrated circuits (ICs), is easy to implement, and is almost always preferred at microscale. Minimum feature sizes for ICs, limited by the wavelength of light used, have been extended to submicron scale with deep-UV light sources and phase-shift masks. Nanoresonators, involving smaller features, from tens to hundreds of nanometers, typically have been made by electron-beam (e-beam) lithography. Here, no mask is used, but an electron beam is rastered across a film of e-beam resist to create the pattern for nanoscale structures. Time-consuming coverage of large areas required to make commercial sensor arrays, the cost of procuring and operating e-beam equipment, and the fabrication sensitivity to small perturbations do not allow inexpensive production of large-area sensor arrays by e-beam lithography. An alternative pattern transfer technique of imprinting,
Nanomechanical Resonant Sensors and Fluid Interactions Nanomechanical Resonant Sensors and Fluid Interactions, Fig. 5 Film stress and stress gradients create residual axial stresses in doubly-clamped beams and distort cantilevered beams, during their release from the substrate, by etching (Figure reprinted from Ref. [20], with permission)
N
1639 TF
Thin film Substrate
σtotal Thin film TFb
σ0
σ1
Cantilever
ΔL
TFc
σ0
molding, and stamping known as microcontact printing, or soft lithography, is similar to the use of an inked rubber stamp. This technique, somewhat surprisingly, has been used to demonstrate the ability to transfer pattern features below 100 nm in size. The master pattern in this case may be made by e-beam lithography and transferred to one or more soft “stamp pads,” thus reducing the requirements for e-beam lithography, with the promise of lowered costs. Mature techniques, relatively well-controlled shape, size and position, the ability to make devices with well-ordered internal structure through the use of single-crystal layers, are important strengths of top-down fabrication [18]. There are limitations of top-down fabrication for nanoresonator sensor arrays. Besides the cost of large-area coverage, the limited choice of local material selection (e.g., difficult to have resonator material different from clamp material) can restrict functionality. Geometries such as smooth rods or hollow tubes, with a high cross-sectional moment of inertia allowing for a higher stiffness per unit mass, are prohibitively difficult to produce in large numbers by top-down techniques. Undercuts, or significant underetching of supports during the removal of the sacrificial layer [18] are seen as a lighter-gray region around the periphery of the harp-like structure of Fig. 4a. Undercuts depend
on the sacrificial layer etch time and are of the same depth for all structures released on a chip. They can cause significant increase in the compliance, particularly for short, stiff, nanoresonators seen in Fig. 4a. Energy losses and cross talk between devices through undercuts can cause significant changes in frequencies and Q-factors. During release by etching of the sacrificial layer, shape distortion (see Fig. 4b) in the case of cantilevered structures [19] or residual axial stress in doubly-clamped nanoresonators (see Fig. 5) typically arises due to stress gradients in the device layer [20]. For long thin metal film structures, these effects are particularly difficult to control. Higher residual axial stress adds an increasing string-like component to the restoring force (normally flexural), changing the simple frequency scaling presented in Eq. 9 and makes the nanoresonator more prone to geometric stiffness nonlinearities. High axial stress and string stiffness can, however, be used to increase the effective stiffness K and the peak frequency, to enable a higher Q-factor value for given damping, provided a sufficient linear range exists. Distorted cantilevers also cannot be described by Eq. 9 – furthermore, it may be difficult or impossible to drive and detect resonance in such structures. Film stress gradient effects can be controlled in device materials such as single-crystal
N
N
1640
silicon, silicon nitride or carbide, using silicon oxide as a sacrificial layer. These materials are frequently used for laboratory experiments. However, stress gradient effects (Fig. 5) are very pronounced for metal films. It is therefore difficult for top-down methods to yield long, thin metal cantilevers which are free of distortion – a “convex-up” example [19] of such distortion for titanium cantilevers is seen in Fig. 4b. Though long thin cantilevered resonators made of metals may have desirable surface functionalization attributes, they are difficult to achieve using top-down methods. Finally, for multiplexing, or the sensing of multiple analytes on the same chip, functionalization of different devices to capture different targets is tedious, and one may anticipate the value of nanostructures that can be made and functionalized off-chip, then directed to predetermined areas of a chip and integrated to form nanoresonators. While top-down methods remain the primary choice for most microcantilever fabrication, in spite of their obvious strengths, many of the shortcomings of established top-down techniques discussed above have led to an interest in alternatives that bottom-up fabrication could provide. This is particularly true for nanoscale resonators, when long thin cantilevers made of metals are required for their attachment chemistries, or when undercut reduction or multiplexed sensor arrays are important. Bottom-Up Methods Bottom-up methods are identified by their use of assembly of smaller units into larger structures, for example, single nanoresonator elements assembled and integrated into arrays for sensing. While the possibilities for improved control of individual element geometry and clamping are attractive, making arrays of a few tens of thousands of elements by an individual pick and place technique is unfeasible. Bottom-up methods are thus defined by the use of physical or chemical forces for self-assembly or directed assembly of nanostructures. A recent review of the directed assembly of nanowires [21] discusses assembly by molecular forces, electrostatic interactions, shear forces, magnetic fields, and dielectrophoresis. A significant advantage over top-down methods is that special vapor or solution phase synthetic chemistry can be used to tailor the shape, composition, and crystallinity of nanoscale elements, before their assembly and integration on a sensor chip. This greatly widens the scope of material properties (e.g., nanowires
Nanomechanical Resonant Sensors and Fluid Interactions
made of magnetic, piezoelectric, semiconductor, or multiple-metal segments) and available geometries (e.g., cylindrical or hexagonal section rods, hollow tubes, single and multi-walled carbon nanotubes) that can be put to use. The elements may also be functionalized off-chip to capture different sensing targets on the same chip (multiplexing), with each target-type assembled in a predetermined area on the chip. This requires the development of assembly and integration techniques that do not destroy the capability of molecules used for functionalization. Results Using a Hybrid Fabrication Strategy for Nanoresonator Arrays Hybrid strategies combine the ease of patterning simple planar structures, borrowed from top-down methods, with the flexibility of tailoring individual nanostructures provided by bottom-up methods, while partially overcoming the difficulties of each method. As an example of this synergy, consider the hybrid bottom-up method reported for nanoresonator array assembly and integration, and designed to accommodate a multiplexed biosensor chip [22]. In this study, both metallic rhodium and semiconducting silicon were synthesized off-chip. Polycrystalline rhodium nanowires were grown by electrodeposition in nanoporous aluminum oxide membranes, while predominately single-crystal silicon nanowires were grown by the Au-catalyzed vapor–liquid–solid (VLS) technique. Bottom-up directed assembly was used to position single nanowires at lithographically defined locations on a silicon chip for fabricating multiplexed arrays. To do so, patterned arrays of wells in a sacrificial insulating photoresist layer covering metal guiding electrodes were defined on the chip surface. Dielectrophoretic forces were used to align the nanowires within the wells. Thick metal clamps were then built over the guiding electrodes by electrodeposition. The photoresist was dissolved to leave arrays of freestanding single-NW cantilevers as shown in scanning electron micrograph images of Fig. 6. The figures show the electrodeposited metal in intimate contact with the nanowire at each nanowire clamp. The clamps were shown to be of repeatable rigidity, with no undercuts. Besides securing the nanowire to the chip, the choice of gold for the clamp material, irrespective of the nanowire material, enables a good electrical connection between the chip and conductive nanowire, which can be used for electrical
Nanomechanical Resonant Sensors and Fluid Interactions SiNW
SiNW detail
RhNWs
RhNW Array
Nanomechanical Resonant Sensors and Fluid Interactions, Fig. 6 Silicon and rhodium nanowire resonators made by a hybrid scheme using top-down patterning of electrodes and a bottom-up directed assembly scheme. Scale bar is 10mm, for picture at lower right. (Figure reprinted from Ref. [22], with permission)
drive and detection of single nanowires in the array. Figure 6 shows that the distortion found in metal cantilevered nanowires from thin film stress evolution in the top-down process is absent. The photoresist used as a sacrificial layer was removed without the need for etchants that could damage the surface functionalization for biosensing. The example hybrid method above eliminates expensive e-beam lithography (reduces cost), shows the viability of long cantilevered metal nanoresonators. It uses lithographic patterning to enable directed assembly and nanoresonator clamping, allows choice of different materials for clamps and nanoresonators, and shows the potential to use different nanoresonator cross-section shapes, materials, and sensor functionalization chemistries for multiplexing on a single chip. As seen above, the choice of fabrication technique plays a strong role in determining the nanoresonator properties available for sensing. In particular, flexible methods such as the hybrid technique described above, allow a wider choice of nanoresonator materials and cross sections. The density, Young’s modulus and intrinsic damping within the nanowire material (Cs) can serve to provide some measure of tuning of the Q-factor for a given fluid damping, depending on the application desired. Figure 7 provides an example of this. It shows the Q-factor change from moderate vacuum to atmospheric pressure for a rhodium and a silicon nanowire of roughly comparable dimensions and fluid damping Cf. Although fine-grained, polycrystalline Rh should be expected to have a much higher
1641
N
intrinsic damping Cs compared to the single-crystal silicon nanowire, its higher elastic modulus and density compared to silicon compensate (Eq. 9), giving it a vacuum Q-value (QRh,s) as high as a fourth of that of the silicon nanoresonator (QSi,s). The experimental data [22] of Fig. 7 further agrees with a prediction that may be made from Eq. 13 – the Q-factor for the silicon nanowire, with smaller intrinsic damping Cs, should begin to deviate from its vacuum value at a lower pressure value (point QSi,i vs. QRh,i) and show a sharper fall-off with pressure, than the rhodium nanowire. The Q-factor of the rhodium nanowire at atmospheric pressure (QRh,a) is seen to be higher than that of the silicon nanowire (QSi,a). Gas damping of the nanoresonator motion increases as ambient gas pressure is raised from high vacuum to atmospheric pressure. Eq. 13 shows that the high intrinsic damping in the rhodium nanowire serves as a buffer to delay the onset of Q-factor reduction and to limit its rate of change. This makes it more suitable for applications such as small mass sensing where reduction in Q-factor would adversely affect performance. However, the value of the Q-factor for the silicon nanowire begins to decrease at a lower pressure and then shows a larger change in measured Q-factor over the same pressure range, compared to the rhodium nanowire. As an instrument for measuring gas damping force, the silicon wire begins to show response at lower pressure and has a greater dynamic range than the rhodium nanowire. Silicon nanowires would thus serve as better instruments for studying gas damping as a function of pressure than rhodium nanowires. In this section, the link between fabrication technique and nanoresonator material choice has been reviewed. Well-established top-down methods remain a strong choice for most batch-fabricated resonant sensors at microscale. For nanoresonators with dimensions far smaller than the wavelength of UV light sources, the costs of patterning nanostructures by rastering an electron beam, increase with chip area. Bottom-up methods, which rely on assembly of nanostructures made off-chip, begin to look more favorable for such structures. A larger variety of materials and shapes may be synthesized and assembled by these techniques, but precise location and individual addressing of assembled structures can be a challenge. An example of a hybrid fabrication scheme that borrows simple top-down photolithography steps for patterning relatively large microscale structures to enable
N
1642
Nanomechanical Resonant Sensors and Fluid Interactions, Fig. 7 Measured quality factor variation with ambient gas pressure for silicon and rhodium nanowires of comparable size. (Figure reprinted from Ref. [22], with permission)
Nanomechanical Resonant Sensors and Fluid Interactions QSi,i
QRh,i
104
103
Rh
Quality factor
N
QSi,s QRh,a 102
QRh,s Si
QSi,a 101
10−6
10−5
spatially directed assembly and clamping to create nanoresonator arrays was reviewed. Nanoresonators of two different materials –polycrystalline rhodium and single-crystal silicon, illustrated how this method may be used to match material with different performance requirements for two different nanoresonator applications. The data show rhodium to be a more favorable nanoresonator material for a sensitive mass detector operating in air at near atmospheric pressure. The material properties of silicon make it a better choice for a nanoresonant sensor of gas damping force over a large range of pressure, spanning the transition and free molecular flow regimes of rarefied gas dynamics.
Summary and Related Reviews This section provides a summary of the choice of topics presented in this entry and points to kindred reviews of complementary material, which could not be accommodated here. Nanoscale resonant sensors were chosen due to their potential for creating new instrument systems of unprecedented sensitivity – the promise of their application to biosensing has been included in a recent comprehensive review of mechanical biosensors [23]. Table 1 of Ref. [23] indicates detection limits of 300 fM (femto-molar) and 1.5 fM
10−4 10−2 10−1 10−3 Ambient Air Pressure (atmospheres)
100
concentration have been achieved with mechanical microcantilever resonators with and without massamplifying labels, respectively, compared to 1,000 fM for the nonmechanical standard test ELISA (enzyme-linked immunosorbent assay). Biomolecular target capture kinetics, target depletion upon capture from small analyte volumes, target sequestration of low-abundance molecules by abundant nontarget molecules are a few other areas where a research effort is focused on extending the current boundaries of knowledge for molecular diagnosis with resonant cantilevers. Related work may be found in the entries on “Cantilever Biosensors” and “Nanowire Biosensors” in this Encyclopedia. Work in nanoresonators constitutes a large field and the focus in this entry has been restricted to nanoresonators interacting with fluids – either gases or liquids. The goal has been to establish notions and the simplest useful quantitative framework to interrogate fluid interaction with nanoresonators. Damping in fluid environments can be strong and lead to low Q-factor operation, associated with a divergence in values calculated from different definitions of the Q-factor. Low Q-factor operation demands large frequency sweeps, which can bring about nonconstant forcing due to frequency-dependent circuit parasitics. Figure 2 provides an example summary of the coupled interdisciplinary work required to deal with the
Nanomechanics
interdependencies between material properties desired for an application, fabrication schemes available, and transduction artifacts. Such work both requires and can extend knowledge in mechanics of materials, and fluid mechanics of continuum and rarefied gas flows at small scales [9, 14]. Details of transduction artifacts in nanoresonator measurements are worthy of a separate review and have not been included here. Ekinci [24] provides a useful review of nanoresonator transduction schemes. While the field of micro/nanofluidics, dealing with small-scale internal flows has acquired some maturity, small-scale external flows at high frequencies realized with nanoresonators hold promise of both technological application and an improved understanding of fluid mechanics.
Cross-References ▶ AFM in Liquids ▶ Electric-Field-Assisted Deterministic Nanowire Assembly ▶ Nanomechanical Properties of Nanostructures
References 1. Jensen, K., Kim, K., Zettl, A.: An atomic-resolution nanomechanical mass sensor. Nat. Nanotechnol. 3, 533– 537 (2008) 2. Hickman, S.A., et al.: Batch-fabrication of cantilevered magnets on attonewton-sensitivity mechanical oscillators for scanned-probe nanoscale magnetic resonance imaging. ACS Nano 4, 7141–7150 (2010) 3. Radmacher, M.: Studying the mechanics of cellular processes by atomic force microscopy. In: Wang, Y.L., Discher, D.E. (eds.) Cell Mechanics, pp. 347–372. Elsevier, Amsterdam (2007) 4. Engler, A.J., Rehfeldt, F., Sen, S., Discher, D.E.: Microtissue elasticity: measurements by atomic force microscopy and its influence on cell differentiation. In: Wang, Y.L., Discher, D.E. (eds.) Cell Mechanics, pp. 521–545. Elsevier, Amsterdam (2007) 5. Lavrik, N.V., Sepaniak, M.J., Datskos, P.G.: Cantilever transducers as a platform for chemical and biological sensors. Rev. Sci. Instrum. 75, 2229–2253 (2004) 6. Naik, T., Longmire, E.K., Mantell, S.C.: Dynamic response of a cantilever in liquid near a solid wall. Sensor. Actuator. A 102, 240–254 (2003) 7. Verbridge, S.S., Craighead, H.G., Parpia, J.M.: A megahertz nanomechanical resonator with room temperature quality factor over a million. Appl. Phys. Lett. 92, 013112 (2008)
1643
N
8. Arlett, J.L., et al.: BioNEMS: nanomechanical systems for single-molecule biophysics. In: Nobel Symposium 131, B€ackaskog Slott, Sweden, 31 Aug 2005 9. Bhiladvala, R.B., Wang, Z.J.: Effect of fluids on the Q factor and resonance frequency of oscillating micrometer and nanometer scale beams. Phys. Rev. E 69, 036307 (2004) 10. Ramanathan, S., Koch, D.L., Bhiladvala, R.B.: Noncontinuum drag force on a nanowire vibrating normal to a wall: simulations and theory. Phys. Fluid 22, 103101 (2010) 11. Zheng, Y.S., Reese, J.M., Struchtrup, H.: Comparing macroscopic continuum models for rarefied gas dynamics: a new test method. J. Comput. Phys. 218, 748–769 (2006) 12. Karabacak, D.M., Yakhot, V., Ekinci, K.L.: High-frequency nanofluidics: an experimental study using nanomechanical resonators. Phys. Rev. Lett. 98, 254505 (2007) 13. Svitelskiy, O., et al.: Pressurized fluid damping of nanoelectromechanical systems. Phys. Rev. Lett. 103, 244501 (2009) 14. Ekinci, K.L., et al.: High frequency nanofluidics: a universal formulation of the fluid dynamics of MEMS and NEMS. Lab Chip 10, 3013–3025 (2010) 15. Sahai, T., Bhiladvala, R.B., Zehnder, A.T.: Thermomechanical transitions in doubly-clamped micro-oscillators. Int. J. Non Linear Mech. 42, 596–607 (2007) 16. Feynman, R.P., Leighton, R.B., Sands, M.: The Feynman Lectures on Physics. Addison Wesley Longman, Reading (1970) 17. Weaver, W., Timoshenko, S.P., Young, D.H.: Vibration Problems in Engineering. Wiley, New York (1990) 18. Carr, D.W., et al.: Measurement of mechanical resonance and losses in nanometer scale silicon wires. Appl. Phys. Lett. 75, 920–922 (1999) 19. O’Mahony, C., et al.: Titanium as a micromechanical material. J. Micromech. Microeng. 12, 438–443 (2002) 20. Fang, W., Wickert, J.A.: Determining mean and gradient residual stresses in thin films using micromachined cantilevers. J. Micromech. Microeng. 6, 301–309 (1996) 21. Wang, M.C.P., Gates, B.D.: Directed assembly of nanowires. Mater. Today 12, 34–43 (2009) 22. Li, M.W., et al.: Bottom-up assembly of large-area nanowire resonator arrays. Nat. Nanotechnol. 3, 88–92 (2008) 23. Arlett, J.L., Myers, E.B., Roukes, M.L.: Comparative advantages of mechanical biosensors. Nat. Nanotechnol. 6, 213–215 (2011) 24. Ekinci, K.L.: Electromechanical transducers at the nanoscale: actuation and sensing of motion in nanoelectromechanical systems (NEMS). Small 1, 786–797 (2005) 25. Tian, M., Wang, J., Kurtz, J., Mallouk, T.E., Chan, M.H.W.: Electrochemical growth of single-crystal metal nanowires via a two-dimensional nucleation and growth mechanism. Nano Lett. 7, 919–923 (2003)
Nanomechanics ▶ Reliability of Nanostructures
N
N
1644
Nanomedicine
Nanomedicine
Definition
▶ Molecular Modeling on Artificial Molecular Motors
Nanoparticle cytotoxicity is defined as the extent to which the interaction of nanoparticles with cells disrupts cellular structures and/or processes essential for cell survival and proliferation. Cytotoxicity assays are a quick and simple way to perform initial acute toxicity assessments. Coupling nanoparticle cytotoxicity data with other safety testing data can help predict the biocompatibility of nanoparticles.
Nano-optical Biosensors
Overview
▶ Nanophotonic Structures for Biosensing
When evaluating the biocompatibility of a chemical, often the first step is to screen for adverse effects in cultured cells before using animal models. Compared to animal studies, in vitro testing is less ethically ambiguous, easier to control and reproduce, and less expensive. Conducting preliminary evaluations in vitro also provides a basis for prioritizing chemical testing and reducing the number of animals used for in vivo toxicity testing. Generally, the first endpoint tested in chemical biocompatibility assessments is cell lethality. Cytotoxicity assays are used to measure the extent of cell death and inhibition of cell proliferation after exposure to a chemical. Since toxicity is often dose-dependent, a range of concentrations is tested to determine the LC50 or the concentration at which 50% cell death occurs. The LC50 serves as a point of comparison of the toxicity of chemicals and is used to group chemicals into toxicity classes ranging from extremely toxic (e.g., botulinum toxin) to relatively harmless (e.g., water). Knowledge of dose is essential as many chemicals that are beneficial at low doses can be toxic at high doses. When evaluating the toxicity of a new chemical, characterizing the composition, purity, stability, and solubility is necessary before testing. In the case of nanoparticles, more extensive characterization, including particle size and size distribution, shape, crystal structure, surface charge, and surface chemistry, is warranted since these factors may also contribute to the observed effects. Generally size, shape, and agglomeration state are thought to be governing properties; however, surface composition and structure also influence the way nanoparticles interact with biological environments. Careful attention should be paid to how the various physiochemical properties of
▶ Acoustic Nanoparticle Synthesis for Applications in Nanomedicine
Nanomotors
Nano-optics Biosensing ▶ Nanophotonic Structures for Biosensing
Nano-optics, Nanoplasmonics ▶ Optical Properties of Metal Nanoparticles
Nano-optomechanical Systems (NOMS) ▶ Optomechanical Resonators
Nanoparticle ▶ Toxicology: Plants and Nanoparticles
Nanoparticle Cytotoxicity Nastassja A. Lewinski Department of Bioengineering, Rice University, Houston, TX, USA
Synonyms Cellular toxicity; In vitro toxicity
Nanoparticle Cytotoxicity
1645
N
Nanoparticle Cytotoxicity, Table 1 Summary of common cytotoxicity assays Assay Trypan blue Neutral red
Mechanism of action Dye exclusion Lysosomal retention
Calcein AM
Dye retention
Fluorecein diacetate
Dye retention
Ethidium homodimer
DNA marker
Propidium iodide
DNA marker
LDH
LDH release
MTT
Mitochondrial reduction
Alamar blue
Mitochondrial reduction
Luciferin/luciferase
ATP-dependent oxidation
Annexin V-FITC
Phosphatidylserine binding
Indication Dead cells stained blue Live cells stained red, abs ¼ 540 nm Live cells fluoresce green, em ¼ 520 nm Live cells fluoresce green, em ¼ 520 nm Dead cells fluoresce red, em ¼ 620 nm Dead cells fluoresce red, em ¼ 620 nm LDH presence converts dye, abs ¼ 490 nm Live cells stained purple, abs ¼ 570 nm Live cells fluoresce red, em ¼ 590 nm Live cells luminesce, em ¼ 560 nm Apoptotic cells fluoresce green, em ¼ 520 nm
Limitations in testing NPs Dye adsorption to NPs in the absence of cells results in absorbance readings
Dye adsorption to NPs quenches fluorescence
Substrate adsorption to NPs reduces absorbance Substrate adsorption to NPs quenches fluorescence
NPs nanoparticles, abs absorbance, em emission
nanoparticles modify their toxicity. Although a detailed assessment of the various particle characterization techniques is beyond the scope of this definition, the reader is referred to review articles on this topic [1, 2]. Given the combinatorial nature of nanoparticle testing, in vitro assays facilitate quick screening of nanoparticle formulations to identify which nanoparticles elicit adverse effects. Depending on their intended use, the results from initial in vitro tests can be used to prioritize them for further development and/or in vivo testing. The following sections will describe some of the most common cytotoxicity assays, the potential mechanisms of nanoparticle toxicity that have been determined to date, and some challenges for future nanoparticle cytotoxicity research.
Common Cytotoxicity Assays Cytotoxicity assays can be categorized by the endpoints they evaluate. Markers of cell injury and death include changes in membrane integrity, mitochondrial function, and cell proliferation. While some of these endpoints can be identified by microscopy, visual inspection of cellular changes provides only
qualitative information. Instead, a majority of published nanoparticle cytotoxicity studies present measurements of cell death using colorimetric methods. A compilation of the assays discussed is presented in Table 1 toward the end of section. Membrane Integrity The cell membrane serves as a physical barrier separating the cell from its surrounding environment, with traffic into and out of the cell regulated by transporters, receptors, and secretion pathways. Exposure to nanoparticles could compromise the cell membrane, making it “leaky” which is the basis for the following assays. Membrane integrity is typically assessed by measuring either (1) the uptake of a cell impermeable dye, (2) the retention of a cell permeable dye, or (3) the release of a cell biomolecule into the extracellular media. Cell impermeable dyes can only enter cells with compromised membranes; therefore, injured or dead cells are stained while intact or live cells remain unstained. Examples of dyes normally excluded from intact cells that are utilized for cytotoxicity assays include trypan blue, ethidium bromide (EtBr), and propidium iodide (PI). Trypan blue is an absorbing dye, staining dead cells blue which can be manually
N
N
1646
counted using a hemocytometer. Although trypan blue assay results have been published by a few groups, it is generally considered a qualitative assay as it is less sensitive and less reliable than other assays. EtBr and PI are red fluorescent dyes, whose peak emission intensities at 620 nm are greatly enhanced upon binding to DNA and can be measured using a flow cytometer or spectrofluorometer. Since quantitative measurements can be derived from these fluorescent dye based assays, they are more widely used. In addition, EtBr and PI can be used to identify necrotic cells and when co-stained with annexin V-FITC differentiate them from apoptotic cells. Refer to the Apoptosis Versus Necrosis section for further discussion. A complementary technique to dye exclusion uses cell permeable dyes, which tend to accumulate within cells, to assess membrane integrity. If the cell membrane is damaged, the dye can leak out of cells or its uptake is decreased, with fluorescence intensity differentiating live and dead cells. Examples of commonly used dyes include neutral red, calcein acetoxymethyl (calcein AM), and fluorescein diacetate (FDA). Neutral red absorbs at 540 nm and can be measured with a spectrophotometer. Calcein AM and FDA are nonfluorescent dyes, which are converted by intracellular esterases to their fluorescent counterpart once within the cell. Cytotoxicity can be quantified by measuring the fluorescence intensity at 520 nm using a flow cytometer or spectrofluorometer. While the fluorescent dyes can be used individually, they are more commonly used in a double “live/dead” assay. The cell sample is exposed to fluorescent dye pairs (e.g., calcein AM/EtBr, FDA/PI) where live cells fluoresce green and dead cells fluoresce red. Another membrane integrity assay detects lactate dehydrogenase release. Lactate dehyrogenase (LDH) is an enzyme normally present in the cytosol and can only be detected outside the cell after cell damage has occurred. In this assay, LDH released from damaged cells oxidizes lactate to pyruvate which promotes the conversion of tetrazolium salt INT to formazan. The amount of LDH released, determined by measuring the absorption of formazan at 490 nm, is proportional to the number of cells damaged or lysed. Significant LDH release from the cytosol is associated with necrosis. Mitochondrial Function Changes in cell function due to toxicity can occur before serious cell damage, such as compromised membrane integrity, are detected. Indicators of
Nanoparticle Cytotoxicity
metabolic activity can be used to detect early cell injury. Considered the “gold standard” for cytotoxicity, the MTT assay is a colorimetric assay that measures the activity of mitochondrial enzymes. MTT, or 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide, is yellow in solution but is converted into an insoluble purple formazan product in living cells. Absorbance measurements at 570 nm can be made after solublizing the formazan with DMSO or isopropanol, and the absorbance intensities can be correlated to cell viability. In addition, the absorbance is also representative of cell number and can be used to measure cell proliferation. While MTT is less expensive and more common, other tetrazolium dyes (MTS, XTT, or WST) similarly produce a formazan that absorbs around 570 nm in live cells but have greater sensitivity with less reaction time. Similar to the MTT assay, the alamar blue assay is another colorimetric assay that measures metabolic enzyme activity. The assay is based on the reduction resazurin, a nonfluorescent dye, to resorufin, a pink fluorescent dye, mainly by acting as an electron acceptor for enzymes such as NADP and FADH during oxygen consumption. The converted dye fluoresces at 590 nm and can be detected using a fluorometer. In addition, alamar blue shows little to no cytotoxic effects in tested cells, and thus has been used to perform several tests or kinetic measurements on the same set of cells. Measurement of ATP levels and oxygen consumption rate are less commonly used methods of assessing nanoparticle cytotoxicity. This could due to the assays being more expensive or requiring specialized equipment. Briefly, ATP bioluminescence assay measures the concentration of ATP, which is proportional to the number of metabolically functional cells, using a luciferin-luciferase system. Healthy cells maintain ATP concentrations through a balance of production and consumption pathways. Since the oxygen consumption rate is disrupted upon cell death, cells lose the ability to produce ATP, and since ATPases can continue to consume ATP after cell death, the concentration of ATP decreases. The decrease in ATP concentration corresponds to a loss of luciferin–luciferase luminescence. The oxygen consumption rate of cells can be measured using electron paramagnetic resonance oximetry, Clark-type polarographic oxygen electrode, and fluorescent oxygen sensors. However, to date, little or no nanoparticle cytotoxicity studies use these approaches.
Nanoparticle Cytotoxicity
Proliferation If nanoparticle exposure induces cell senescence but not cell death, this could be considered a cytotoxic effect as cell proliferation has been disturbed. Comparing the number of cells present at different exposure time points can indicate effects on cell proliferation. Cell number can be measured by direct cell counting, nuclei counting, or by the measurement of total cell protein or DNA content, which are proportional to the number of cells. Fluorescence from dyes that label DNA, such as Hoechst or PI, is directly proportional to the sample DNA content. Compared to the control, higher fluorescence indicates proliferation and lower fluorescence signals cytotoxicity. This method is relatively cheap and fast. In comparison, markers such as 3 H-thymidine and 5-bromo-2’-deoxyruidine (BrdU) incorporation into DNA can also be used to measure cell proliferation; however, both methods require significant sample processing. As mentioned previously, MTT can also be used to measure cell proliferation since the dye absorbance is proportional to cell number.
Apoptosis Versus Necrosis Differentiating the type of cell death can elucidate information about the mechanism of toxicity. For example, the plasma membrane of apoptotic cells remains largely intact, and therefore cell death via may not be accounted for when using assays based on compromised membrane integrity, such as PI staining. Apoptosis involves a series of signaling cascades that are initiated by the cell after disruption of intracellular homeostasis. During apoptosis, phosphatidylserine is transported from the inner surface of the cell membrane to the outer surface. Annexin V binds specifically to phosphatidylserine and therefore can be used to label apoptotic cells. In contrast, necrosis is passive and results from irreversible cell damage, such as disruption of the plasma membrane. As mentioned previously, PI is a nuclear stain that only fluoresces in cells with compromised plasma membranes. Combined cell staining with annexin V–FITC (fluorescein isothiocyanate) and PI allows for the discrimination of intact cells (no staining), early apoptotic (FITC staining only), and late apoptotic or necrotic cells (both PI and FITC staining). Staining can be observed using fluorescence microscopy or quantified using flow cytometry.
1647
N
Additional Study Design Considerations Choice of Cell Type Most nanoparticle cytotoxicity data is generated from monocultures of cells representative of specific organs of the body. Recently co-culture in vitro systems have been developed to better simulate the cellular environment in vivo [3]. Cell lines recommended in the ASTM E2526-08 and ISO 10993-5 cytotoxicity testing standard protocols include LLC-PK3 porcine kidney cells, HepG2 liver cancer cells, and 3T3 mouse embryo fibroblasts. Other cell lines often used in nanoparticle cytotoxicity studies include A549 lung cancer cells, HEK293 human embryonic kidney cells, human epidermal keratinocytes, human dermal fibroblasts, and HeLa human epithelial cancer cells. Epithelial cells are often used because they form the outer barrier layer of organs and are usually the first cells to come into contact with nanoparticles during exposure. Regardless of which cell lines chosen, it is important to recognize that there may be differences in sensitivity between cell types [4]. In addition, if primary cells are chosen, these cells may have greater sensitivity than established cell lines.
Suitability for High-Throughput Screening
A considerable amount of time and money can be lost on nanoparticle development when they fail late in the process due to toxicity issues. The growing number of nanoparticles needing to be screened places new demands on cytotoxicity tests, with more attention being drawn to the use of high throughput methods. High-throughput screening (HTS) utilizes automated systems to rapidly conduct in vitro tests on microtiter plates. If multiple measurements are obtained from each well, then the process is referred to as high content screening. Several of the established cytotoxicity assays mentioned have been successfully adapted to 96-well and 384-well plates for use in high-throughput plate reader and flow cytometer setups [5]. Since the advantage of in vitro cytotoxicity screening is the ability to evaluate the toxicity of a large number of compounds, testing nanoparticle libraries with HTS is one strategy toward generating the volume of data needed to identify toxic nanoparticle formulations, elucidate nanoparticle structure-activity relationships, and better prioritize nanoparticles for further in vivo testing.
N
N
1648
Nanoparticle Cytotoxicity
Nanoparticle Cytotoxicity, Table 2 Toxicological effects of some widely used NPs NP C60
Toxicity observations Unmodified C60 more toxic than functionalized (hydroxylated and carboxylated) C60, residual THF solvent can contribute to toxicity observed from unmodified C60 cluster (nC60) toxicity, photocatalytic behavior contributes to toxicity CNTs Unpurified as-prepared CNTs exhibit toxicity possibly due to presence of catalytic iron, single-walled CNTs elicit more toxic response than multiwalled CNTs Au Relatively “nontoxic,” cationic AuNPs more toxic than anionic AuNPs, CTAB and citrate coatings contribute to toxicity but can be mitigated with PEG Ag Silver ion release contributes to toxicity, ROS generation and oxidative stress are two major mechanisms, different surface chemistries induce different DNA damage responses QDs Heavy-metal (e.g., Cd) leaching contributes to toxicity, shell and surface coating can reduce toxicity by containing core and modulating surface charge, toxicity also influenced by oxidative & photolytic stability TiO2 Anatase form ~10 times more cytotoxic than rutile form, photoactivity enhances toxicity, oxidant driven inflammation and protein nitration also associated with TiO2 NP exposure
Reference [6]
[7] [8] [9] [10] [11]
CNTs carbon nanotubes, NPs nanoparticles, THF tetrahydrofuran, CTAB cetyl trimethylammonium bromide, PEG polyethylene glycol
Potential Mechanisms of Nanoparticle Toxicity Critical determinants of nanoparticle-induced cytotoxicity include exposed cell type, composition, and size of nanoparticles. In addition, through one-by-one evaluation of nanoparticles, structure-activity relationships specific to certain nanoparticle formulations have been observed. Several reviews provide detailed discussions on general and nanoparticle-type specific cytotoxicity, and recent examples are included in the References section. A summary of the current understanding on the mechanisms behind toxicity observed from six widely used nanoparticles with reference to recent reviews is presented in Table 2. The following section elaborates on some of the effects that are common across nanoparticle types. Aggregation and Agglomeration Nanoparticle stability can vary in different dispersion solutions, such as deionized water and cell culture media. Aggregation, or the strong binding of nanoparticles together, and agglomeration, or the clustering of nanoparticles held together by relatively weak forces, can change the concentration and physicochemical properties of nanoparticles in solution. The properties affected are their size, size distribution, surface area, and therefore their surface reactivity. Since these parameters play a role in the toxicity of nanoparticles, it is important to keep in mind the potential contribution of aggregation and agglomeration during study design and analysis. Methods of reducing nanoparticle
aggregation and agglomeration as well as homogeneously dispersing them in liquids include surface modifications and sonication before exposure. While surface modifications may improve nanoparticle stability, they may also influence their bioactivity. The durability of the surface coatings once inside biological environments is another issue, and release of the surface coatings may also contribute to the toxicity of nanoparticles. Cellular Uptake Uptake of nanoparticles into cells through endocytosis or phagocytosis processes is size-dependent and can vary greatly among cell types. The amount of nanoparticle internalization by cells can contribute to the observed toxicity as internalized nanoparticles may interact with cell organelles and disrupt cellular equilibrium or homeostasis. Transmission electron microscopy has been the preferred method of studying the cellular uptake of nanoparticles due to its high resolution, resolving the intracellular location of nanoparticles. Most studies have found internalized nanoparticles localize in endosomes or lysosomes. The low pH in endosomes and lysosomes has been suggested to dissolve or degrade nanoparticles releasing their core atoms. This can result in cytotoxicity, especially with metal nanoparticles, since metal ions can disrupt calcium homeostasis and/or generate reactive oxygen species. Oxidative Stress and Inflammation In vitro studies have shown that nanoparticles of various compositions generate free radicals and reactive
Nanoparticle Cytotoxicity
oxygen species (ROS) more than larger particles, due to their higher surface area. ROS can damage cells by inducing oxidative stress, which involves a redox imbalance within cells often due to increased intracellular ROS and decreased antioxidants. The oxidative stress associated with nanoparticle exposure can result from: • The presence of transition metals (ex. iron, copper, nickel) catalyzing Fenton reactions. • Generation of ROS at the particle surface. • Phagocytosis of nanoparticles by inflammatory cells, such as alveolar macrophages and neutrophils. • Physical damage due to nanoparticles interacting and/or entering mitochondria, altering mitochondrial function. The cellular damage resulting from ROS generation and activation of inflammatory cells can also lead to inflammation. Markers of inflammation include the proinflammatory cytokines IL-1b, IL-6, and TNF-a plus the chemokine IL-8. Cytokines can be quantified using the enzyme-linked immunosorbant assay (ELISA), and ELISA results have been reported for metal oxides, carbon-based nanoparticle, silica, and quantum dots. However, carbon nanoparticles and metal oxides have also been reported to adsorb cytokines IL-8 and IL-6, interfering with the ELISA assay. These observations should be considered when interpreting results from nanoparticle inflammation testing. DNA Damage Of the cellular events that can occur in the absence of cell death, DNA damage is of critical concern as it can initiate disease development, such as cancer, and genetic abnormalities in reproductive cells. Both the nanoparticles themselves as well as nanoparticleelicited events, such as overload and persistent inflammation, have been suggested to play a role in nanoparticle-induced genotoxicity. However, to date, limited genotoxicity testing of nanoparticles has been conducted. Genotoxicity may result from nanoparticle exposure due to their (a) small size which may facilitate their transport across cell and nuclear membranes where they may interact directly with DNA, (b) high surface area and/or chemical components (e.g., heavy metals) could generate ROS, which damage DNA. The comet assay has been used to detect DNA damage in individual cells using gel electrophoresis. Cells with damaged DNA appear as “comets” with intact DNA residing in
1649
N
the head portion and broken DNA pieces migrating away, forming the tail. A DNA-specific dye such as PI is used to read the gel, and the amount of DNA found in the tail is proportional to the amount of DNA damage. The comet assay is the most widely used genotoxicity test in published nanoparticle studies; however, it only nonspecifically indicates the presence of DNA strand breaks. The most investigated oxidative lesion caused by hydroxyl radical attack, 8-OHdG, has been shown to be induced in vitro by nanoparticles, including carbon nanotubes, CeO2, gold, and TiO2 nanoparticles [12]. Isotope-dilution gas chromatography–mass spectrometry (GC-MS) can be used to detect and quantify DNA damage mediated by ROS produced from nanoparticle interactions in cells. This method is more robust compared to the commonly used comet assay since it has been shown to differentiate among more than 20 DNA base lesions. Since oxidatively modified DNA lesions have different mutagenic potential, it is important to identify the types of DNA modifications that occur due to nanoparticle exposure.
Challenges for Future Nanoparticle Cytotoxicity Testing It is important to consider that in vitro dose–response data serves as a starting point for evaluating the cytotoxicity of nanoparticles. Often in vitro data cannot be extrapolated to nor is representative of what would occur in vivo. For example, cells exposed to nanoparticles suspended in cell culture media poorly reflects inhalation exposure conditions, where lung cells at the air–liquid interface are exposed to aerosols of nanoparticles. In addition, in vitro cell culture systems are controlled environments, often containing a single cell line at low cell density, which may not accurately model normal cell–cell interactions or tissue injury signaling cascades. However, these shortcomings arise from the inherent differences between in vitro and in vivo systems. Cytotoxicity assays also presents unique challenges for nanoparticle testing since particulates differ from soluble chemicals. The following section describes areas for improvement in nanoparticle cytotoxicity testing. Dose Metric Compared to soluble chemicals, nanoparticles can agglomerate, aggregate and, as a consequence, settle
N
N
1650
out of solution depending on their size, concentration, and surface chemistry. These processes can affect the cellular dose, making quantification of the exposure concentration more complicated and less comparable across nanoparticle types. As a result, questions have been raised on the applicability of standard cytotoxicity assay protocols, developed to test soluble chemicals, for nanoparticles. Nanoparticle–cell interactions may be enhanced due to nanoparticles settling to the cell surface, leading to potential nanoparticle overloading of the cells. Resulting cytotoxicity may therefore be due to the higher exposure level from poor stability. It is important to account for this phenomenon when conducting nanoparticle in vitro dose– response assessment. In addition, the exposure dose is often inconsistently reported. There are several dose metrics that can be used, such as mass per volume, particles per volume, mass per surface area, or mass per cell number. To aid study interpretation, enough data should be reported so that the reported values can be converted to different dose metric units. Most in vitro studies assess dosimetry by observing the dose–response relationship after external introduction of different concentrations of nanoparticles. Yet the internalized concentration of nanoparticles may better correlate to cytotoxicity [13]. A more informative representation of exposure dose would be to quantify the amount of nanoparticles externally attached or internalized by the cells (delivered dose) in addition to the concentration of nanoparticles added to the media (administered dose). Teeguarden et al. and more recently Wittmaack provide thoughtful discussions on how the particulate nature of nanoparticles introduces issues such as gravitational settling that impact dosimetry [14, 15]. Despite these examples, few studies make a distinction between administered and delivered dose. Knowing the external and internal fractions of the delivered dose could also give insights into the mechanisms of nanoparticle toxicity; however, experimental methods to resolve these concentrations need to be developed. Future research should work to correlate the deposited and internalized nanoparticle concentrations to observed cytotoxicity. Nanoparticle Interference Due to their high surface area, and in some cases optical properties, the presence of nanoparticles in the test solutions can interfere with cytotoxicity assay results. The increased adsorption capacity and
Nanoparticle Cytotoxicity
increased catalytic activity resulting from their high surface area can either bind or interact with assay dyes leading to false readings. For example, carbon nanomaterials, especially single-walled carbon nanotubes (SWCNTs), have been reported to interact with cytotoxicity indicator dyes, including MTT, WST-1, Coomassie blue, alamar blue, and neutral red. Adsorption of the dyes to the carbon nanotubes results in either false negative data for the absorbing dyes or false positive data for the fluorescent dyes. Fluorescence quenching by carbon nanomaterials has also been reported with adsorption of the dye suggested as a reason for the quenching response. Other ways nanoparticles can contribute optical interference include physically blocking the emitted light, scattering the excitation light, and adding their own fluorescence signal (e.g., quantum dots or polystyrene beads). The reader is also referred to a review that describes in more detail these and other assay interferences derived from nanoparticles [16, 17]. Future developments could include designing label-free cytotoxicity assays, which would eliminate the issue of nanoparticle–dye interactions. As a safeguard, conducting multiple tests is advantageous to ensure valid conclusions are drawn. Inter-Laboratory Consistency While many studies using two or more different cytotoxicity assays have reported reasonable agreement in the results, large discrepancies have also been reported. Lanone et al. showed that for highly toxic or not toxic nanoparticles, inter-laboratory reproducibility was good, with LC50 values very similar for toxic nanoparticles [4]. However, their data also indicated that the reproducibility for nanoparticles with intermediate toxicity was lower in comparison. Since the cell lines and methodology were the same between labs, the variation between labs found suggests that there may be other factors unaccounted for in protocol development to ensure reliability in performance. These could include differences in nanoparticle storage methods, age of nanoparticle solution at testing, instrument calibration, and user performance may have contributed to the variability observed. Until data generated in different labs come within an accepted margin of error, inconsistencies in study findings and questions on methodology appropriateness during nanoparticle cytotoxicity assessment will remain.
Nanoparticle Self-Assembly
Conclusion Cytotoxicity testing constitutes just one of many steps required for nanoparticle biocompatibility and human health risk assessment. Traditionally, in vitro toxicity testing focuses on whether or not exposure to a potentially toxic agent results in cell death. However, changes in cellular function may serve as early signs of cytotoxicity when no cell damage or death may be apparent. As not all disruptive effects result in cell death, more extensive toxicity studies have attempted to determine the sublethal effects of nanoparticles. Given the variety of assays developed for cytotoxicity testing, it is important to verify that the endpoints of the assay used are appropriate for the hypothesis being tested. In addition, many of these assays were developed for testing soluble chemicals and may not be suitable for nanoparticle testing. Evidence of nanoparticle interferences with the indicator dyes has raised questions on the accuracy of currently available nanoparticle cytotoxicity data. Researchers should take care when interpreting the available literature as well as when conducting their own nanoparticle cytotoxicity testing. Ultimately, the goal of nanoparticle cytotoxicity assessments is to provide initial toxicity data to guide further risk assessment, which will inform decision making to protect human health.
Cross-References ▶ Genotoxicity of Nanoparticles ▶ Quantum-Dot Toxicity ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles
References 1. Card, J.W., Magnuson, B.A.: A method to assess the quality of studies that examine the toxicity of engineered nanomaterials. Int. J. Toxicol. 29, 402–410 (2010) 2. Powers, K.W., Palazuelos, M., Moudgil, B.M., Roberts, S.M.: Characterization of the size, shape, and state of dispersion of nanoparticles for toxicological studies. Nanotoxicology 1, 42–51 (2007) 3. Stone, V., Johnston, H., Schins, R.P.F.: Development of in vitro systems for nanotoxicology: methodological considerations. Crit. Rev. Toxicol. 39, 613–626 (2009)
1651
N
4. Lanone, S., Rogerieux, F., Geys, J., Dupont, A., MaillotMarechal, E., Boczkowski, J., Lacroix, G., Hoet, P.: Comparative toxicity of 24 manufactured nanoparticles in human alveolar epithelial and macrophage cell lines. Part. Fibre Toxicol. 6, 14 (2009) 5. Riss, T., Moravec, R., Niles, A.: Assay development for cell viability and apoptosis for high-throughput screening. In: Chen, T. (ed.) A Practical Guide to Assay Development and High-Throughput Screening in Drug Discovery. CRC Press, Hoboken (2009) 6. Johnston, H.J., Hutchison, G.R., Christensen, F.M., Aschberger, K., Stone, V.: The biological mechanisms and physicochemical characteristics responsible for driving fullerene toxicity. Toxicol. Sci. 114, 162–182 (2010) 7. Cui, H.-F., Vashist, S.K., Al-Rubeaan, K., Luong, J.H.T., Sheu, F.-S.: Interfacing carbon nanotubes with living mammalian cells and cytotoxicity issues. Chem. Res. Toxicol. 23, 1131–1147 (2010) 8. Alkilany, A.M., Murphy, C.J.: Toxicity and cellular uptake of gold nanoparticles: what we have learned so far? J. Nanopart. Res. 12, 2313–2333 (2010) 9. Ahamed, M., AlSalhi, M.S., Siddiqui, M.K.J.: Silver nanoparticle applications and human health. Clin. Chim. Acta 411, 1841–1848 (2010) 10. Pelley, J.L., Daar, A.S., Saner, M.A.: State of academic knowledge on toxicity and biological fate of quantum dots. Toxicol. Sci. 112, 276–296 (2009) 11. Johnston, H.J., Hutchison, G.R., Christensen, F.M., Peters, S., Hankin, S., Stone, V.: Identification of the mechanisms that drive the toxicity of TiO(2)particulates: the contribution of physicochemical characteristics. Part. Fibre Toxicol. 6, 33 (2009) 12. Petersen, E., Nelson, B.: Mechanisms and measurements of nanomaterial-induced oxidative damage to DNA. Anal. Bioanal. Chem. 398, 613–650 (2010) 13. Lewinski, N., Colvin, V., Drezek, R.: Cytotoxicity of nanoparticles. Small 4, 26–49 (2008) 14. Teeguarden, J.G., Hinderliter, P.M., Orr, G., Thrall, B.D., Pounds, J.G.: Particokinetics in vitro: dosimetry considerations for in vitro nanoparticle toxicity assessments. Toxicol. Sci. 95, 300–312 (2007) 15. Wittmaack, K.: Novel dose metric for apparent cytotoxicity effects generated by in vitro cell exposure to silica nanoparticles. Chem. Res. Toxicol. 24(2), 150–158 (2011) 16. Doak, S.H., Griffiths, S.M., Manshian, B., Singh, N., Williams, P.M., Brown, A.P., Jenkins, G.J.: Confounding experimental considerations in nanogenotoxicology. Mutagenesis 24, 285–293 (2009) 17. Kroll, A., Pillukat, M.H., Hahn, D., Schnekenburger, J.: Current in vitro methods in nanoparticle risk assessment: limitations and challenges. Eur. J. Pharm. Biopharm. 72, 370–377 (2009)
Nanoparticle Self-Assembly ▶ Electric Field–Directed Assembly of Bioderivatized Nanoparticles
N
N
1652
Nanoparticle Tracking Analysis Matthew Wright NanoSight Limited, Amesbury, Wiltshire, UK
Synonyms
Nanoparticle Tracking Analysis
liquid suspension. As populations of particles can be simultaneously analyzed on an individual basis, it is ideally suited for analysis of polydisperse systems ranging from 10–20 nm to 1–1.5 mm (depending on particle type). Concurrent measurements also allow users to acquire information on concentration, zeta potential, and relative intensity of light scattered and also to analyze fluorescently labeled particles as discussed by Carr et al. [1].
LM10; LM20; NS500; NTA
Definition Nanoparticle Tracking Analysis (NTA) is an analytical methodology which facilitates the direct and real-time visualization and characterization of nanomaterials sized between 10 and 1,500 nm in liquid suspension. The aim of this entry is to address the need to characterize different properties of particles in the nanoscale, explain the principles of operation of Nanoparticle Tracking Analysis, explore the concentrations and size ranges measurable, look at other parameters characterized by Nanoparticle Tracking Analysis (NTA) as well as functions and applications of the technique. The use of nanoscale materials is widespread throughout many different industries. Whether this is in pigments for the paint industry, or viruses used in the development and manufacture of vaccines, materials in this size range play a large part in the effectiveness and constancy of many products and processes. As differences in not only the size and concentration, but also shape and both magnetic and electrical charge of these materials can make a substantial difference to the end product, it is clear there is a need for processes and techniques that have the ability to characterize the different properties of these materials. There are numerous techniques that allow different aspects of this combination of properties to be determined, each with their own sets of benefits and limitations. These techniques include: Dynamic Light Scattering, also known as Photon Correlation Spectroscopy, Analytical Disk Centrifuge, Flow Cytometry, Transmission Electron Microscopy, and Scanning Electron Microscopy. NTA is a multiparameter methodology that allows the direct and real-time visualization, sizing, and counting of materials between 10 nm and 1.5 mm in
Principles of Operation and Analysis NTA obtains particle size distributions of samples by utilizing properties of both light scattering and Brownian motion. A finely focused laser beam (usually from laser diodes operating at 635, 532, 488, or 404 nm) is passed through an optical flat with a beveled edge within a sample chamber. The angle that the laser is introduced to the flat is such that when the beam reaches the interface between the flat and the liquid layer of solvent containing the sample above it, it is refracted almost flat. This compression of the laser to such a low profile gives a very high power density. As such, a region of intense illumination is created within which particles can be visualized by utilizing the light scattered from them by Rayleigh Scattering. Figure 1 shows a simplified version of the configuration of laser (running from left to right), glass flat and objective lens used by systems using NTA. This allows particles to be visualized using light scattered. A conventional microscope arrangement with a 20 magnification objective fitted is used in order to visualize the particles in suspension. Coupled with this microscope arrangement, is a video camera (typically a CCD, emCCD, or CMOS). This camera, operating at 30 frames a second, is used to capture a video field of approximately 106 mm 80 mm 6 mm, within which particles in the sample can be seen moving under Brownian motion. Figure 2 shows the movement of a system of polydisperse nanoparticles under Brownian motion. On average, smaller particles move at a greater rate than larger particles, as can be seen by the relative step lengths of the trails following the particles (each change in direction or “step” indicating one video frame). Once a field of view containing particles in suspension moving under Brownian motion has been
Nanoparticle Tracking Analysis Nanoparticle Tracking Analysis, Fig. 1 Configuration of components used by NTA in order to visualize nanoparticles (Figure courtesy: NanoSight Ltd.)
1653
N
Microscope
Particles scattering from laser beam Particles suspended in liquid Metalized surface
Laser beam (Approx. 50um wide)
Glass
using a proprietary version of a Centroid tracking algorithm to give x and y coordinates for each particle. Eq. 1: Centroid tracking algorithm Cx ¼
n X m X i¼1 j¼1
ðx : Iij Þ Pn iPm i¼1
j¼1 Iij
Where xi ¼ Coordinate of pixel on the x-axis Iij ¼ Pixel intensity [2] The center of the same scattering point is then found in the next frame, and the distance that the particle has moved in that time is calculated . Eq. 2: Finding displacement of particles between frames (r) r2 ¼ Dx2 þ Dy2 Nanoparticle Tracking Analysis, Fig. 2 Showing the movement of polydisperse nanoparticles under Brownian motion (Figure courtesy: NanoSight Ltd.)
established, the NTA software is used to record a video file of the movement, typically 30–60 s in duration. This video file is then analyzed by the software which automatically finds the center of each scattering point
By tracking this movement over several consecutive frames, the software determines the mean displacement and hence the diffusion coefficient of each particle within the field of view. Provided that the temperature (T) and the viscosity (Z) of the system are known, the Stokes-Einstein equation (Eq. 3) is then used to relate the diffusion coefficient to the
N
N
1654
Nanoparticle Tracking Analysis
hydrodynamic sphere equivalent particle diameter of the material. The Stokes-Einstein equation is as follows: Eq. 3: The Stokes-Einstein equation Dt ¼
KB T 3pZdh
where Dt ¼ Diffusion coefficient KB ¼ Boltzmann constant T ¼ Temperature Z ¼ Viscosity dh ¼ Sphere equivalent hydrodynamic diameter particle size As Brownian motion occurs in three dimensions, a modified version of the Stokes-Einstein equation is needed for NTA as this technique only observes the motion in two dimensions. However, it is possible to determine Dt from the mean squared displacement of a particle in one, two, or three dimensions (Eq. 4a–c respectively). Eq. 4: (a, b and c): Tracking Brownian motion in different numbers of dimensions ðxÞ2 ¼
2TKB T 3pZd
ðx; yÞ2 ¼
4TKB T 3pZd
ðx; y; zÞ2 ¼
2TKB T pZd
(a)
(b)
(c)
So in the case of Nanoparticle Tracking Analysis where measurements are made in two dimensions: Eq. 5: Tracking Brownian motion in two dimensions to determine particle size ðx; yÞ2 KB T ¼ Dt ¼ 3pZdh 4 Equation 5 is applied to the trajectories of each of the particles within the field of view allowing the user to obtain a reading of particle size (hydrodynamic diameter, in nm) against concentration (absolute in particles per ml).
Other Measurable Parameters Concentration The field of view interrogated by NTA systems is of known dimensions and therefore known volume which means that determining the concentration of particles in terms of particles per ml can easily be achieved using the NTA software. As the software not only tracks and sizes every particle within the field of view but also counts the number of particles identified, a concentration can be achieved by extrapolation of this number. The field of view measures 106 mm 80 mm 6 mm. This leads to a volume of 50,880 mm3. Equation 6 shows how this is used in order to determine particle concentration. Eq. 6: Extrapolation of particle count to concentration in particles per ml
1012 mm3 Concentration ¼ Pcount 50; 880mm3 Concentration ¼ Pcount 1:97 107 where Pcount ¼ Number of particles within the field of view In a 2010 study Filipe et al. investigated the ability for NTA to measure concentration and concluded that the capability of NTA to determine approximate concentration ranges for submicron particles offers an advantage over other techniques for particle sizing [3]. A study conducted by Du et al. found that concentrations obtained using NTA correlated very well with those they calculated using other methodologies [4].
Scatter Intensity While the size of particles is determined by the NTA software using Brownian motion, the system also measures the relative amount of light scattered by each particle in the scattering volume. These two properties can be plotted as a function of one another allowing differentiation of particles not only by size, but also by composition. For a mixture of similar sized particles with significantly different refractive indices (for example, gold particles of 100 nm and polystyrene particles of 100 nm), both would appear in the same size distribution, but can be separated when size is plotted against relative refractive index. Figure 3
Nanoparticle Tracking Analysis
Nanoparticle Tracking Analysis, Fig. 3 Illustrating differentiation of particle material by comparing particle size (x-axis), relative intensity (y-axis), and number of particles (z-axis) (Figure courtesy: NanoSight Ltd.)
shows a distribution of 30-nm and 60-nm gold nanoparticles mixed with a population of 100-nm polystyrene particles, separated by both particle size and the relative amount of light scattered. Particle size is given on the x-axis, with relative intensity on the y-axis and number of particle on the z-axis. These populations may have merged with one another on a simple particle size versus concentration graph. However, by employing the third dimension of relative intensity of light scattered, it is easy to define the three discrete populations of materials. Fluorescence Being a direct visualization and tracking method, NTA does not rely on interference from coherent light sources. This means that fluorescently labeled or inherently fluorescent materials can be analyzed without affecting the results that are obtained. In order to do this, typically a green laser (532 nm) or a blue/violet laser (404 nm) is used in place of the usual red laser used within the system allowing the fluorescent excitation of particles with absorption spectra that span these wavelengths. By comparing profiles generated with and without a filter removing the excitation wavelength present in the optical path of the system, the user is able to determine the effectiveness of their labeling process, or the proportion of labeled to unlabelled material within their sample, and whether or not the presence or absence of fluorescence is in any way related to the hydrodynamic sphere equivalent particle size.
1655
N
Suspension Viscosity NTA can also be used to determine unknown viscosities of fluids. The system uses the Stokes-Einstein equation in order to relate the diffusion coefficient of the material to the sphere equivalent particle size. However, if the size of the particles in question is already well characterized, then it is possible to rearrange the Stokes-Einstein equation as shown in Eq. 7 and use the known parameters of diffusion coefficient, particle size, and temperature in order to use the collected data to determine the viscosity of the system. This will be briefly discussed in the section on drug delivery studies. Eq. 7: Stokes-Einstein equation to find viscosity Z¼
KB T 3pDt dh
where Dt ¼ Diffusion coefficient KB ¼ Boltzmann constant T ¼ Temperature Z ¼ Viscosity dh ¼ Sphere equivalent hydrodynamic diameter particle size Zeta Potential The zeta potential of a system is a measure of charge stability and has an effect on all particle-particle interactions within a system. Understanding zeta potential is of critical importance in controlling dispersion and determining the stability of a nanoparticle suspension. The zeta potential is the electric potential at the slip plane between the bound layer of solvent molecules surrounding the particle, and the overall solution. This can be closely linked to the particle’s surface charge in simple systems but is also heavily dependent on the properties of the solvent in which the particles are suspended. A higher level of zeta potential results in greater electrostatic repulsion between the particles, which results in less aggregation and flocculation. In order to carry out investigations into the zeta potential of a system, the hardware used for NTA is adapted so that the sample chamber is fitted with platinum electrodes through which an electric field can be applied to the cell. This field causes motion of both the sample particles (electrophoresis) and the diluent (electroosmosis). The NTA software records the apparent drift velocity
N
N
1656
Nanoparticle Tracking Analysis
Nanoparticle Tracking Analysis, Fig. 4 Diagram showing profile map positions where six standardized depth locations intersect the laser beam (Figure courtesy: NanoSight Ltd.)
Top surface
Thumbprint at bottom surface
Negative flow velocity
Nanoparticle Tracking Analysis, Fig. 5 Total velocity profile measured using NTA (red) and electroosmotic velocity profile (blue) inferred by subtraction of a contact offset to obtain a velocity profile that sums to zero over the channel depth (Figure courtesy: NanoSight Ltd.)
for each tracked particle, which will be a superposition of both electrophoresis and electroosmosis. This total velocity is measured at different depths within the sample chamber (as shown in Fig. 4, each dot corresponding to a depth of measurement), making it possible to separate these components and obtain a measurement of the electrophoretic velocity (due to the force impinged directly on the particles), and hence the zeta potential of the particles. By employing the assumption that the net electroosmotic flow summed over the chamber depth will be zero, as should be the case for a closed system, the electroosmotic profile can be inferred, as shown in Fig. 5. Using the calculated flow profile, the velocity component caused by electroosmosis is removed from the tracked velocities for each particle. The corrected drift velocities then give a measure of the electrophoretic mobility of individual particles.
O
Positive flow velocity
Offset
The electric field strength (E) within the flow cell is determined using the Voltage (V) applied through the sample by the electrodes, and the distance between the electrode surfaces (d) (Eq. 8). Eq. 8: Determining electric field strength V d The particle velocities can then be converted into electrophoretic mobilities (velocity divided by electric field strength). By application of the Henry equation using the Smoluchowski approximation (appropriate for aqueous diluent media with moderate electrolyte concentration, seen in Eq. 9) the zeta potential (ZP) for each particle can be calculated: Eq. 9: Calculation of zeta potential mZ ZP ¼ e0 er E¼
Nanoparticle Tracking Analysis
N
Accuracy of Concentration Measured by NTA Measured NTA concentration (E+08) particles per ml
Nanoparticle Tracking Analysis, Fig. 6 Relationship between absolute concentration (x) versus that measured by NTA(y) (Figure courtesy: NanoSight Ltd.)
1657
40 35 30 25 20 15 10 5 0 0
10
20 30 40 Theoretical Concentration (E+08) particles per ml
where m is the electrophoretic mobility of the particle, e0 is the permittivity of free space, er is the relative sample solution permittivity, and Z is the sample solution viscosity. This particle-by-particle approach allows different charge populations within a sample to be determined rather than being lost in the “noise” of an ensemble technique.
Concentration and Size Ranges Measurable The ability of NTA to analyze polydisperse samples comes from the ability to track and analyze each particle in the system individually, irrespective of the others around it. It is therefore important that a sufficient number of particles are analyzed during analysis to ensure that the data produced is statistically viable. Repeatable and accurate results can be obtained when analyzing particles in concentrations between 107 and 109 particles per ml. Below this concentration, the number of particles within the field of view is so low that the typical analysis time of 30–60 s will only allow a very small number of particles to be analyzed, leading to inaccuracies of a statistical nature. Above 109 particles per ml, the field of view becomes very crowded. As Brownian motion is a random process, when two particles cross over, the tracking software is unable to determine which particle is which and must
50
60
therefore stop tracking both to avoid erroneous tracking. When the field of view is crowded, the rate at which crossover of particles occurs is so great that there is insufficient time to determine diffusion coefficient and therefore size. A high concentration of particles also leads to erroneous results when trying to identify solution concentration. Figure 6 shows the relationship between the concentration of particles in a solution and the concentration as measured using NTA. It is clear that the correlation between the actual and measured concentration remains stable until a concentration of around 3.5 109 particles per ml whereupon the accuracy of the NTA concentration reading begins to deteriorate. The typical size range between which NTA operates is from 10–20 nm up to 1.5 mm. The lower limit of this figure is mainly dependent upon the amount of light scattered by the particles. This value in turn varies in relation to particle size, shape, refractive index (real and imaginary), illumination power, wavelength, power, and polarization as well as solvent refractive index. The amount of light falling on the detector also fluctuates according to many variables including efficiency of optics, sensitivity, and spectral response of the camera. In their 1983 work, Bohren et al. comprehensively describe the theory of light scatter and the formula for Rayleigh scattering of small particles [5]:
N
N
1658
Nanoparticle Tracking Analysis
Eq. 10: Rayleigh scattering of light by small particles
I 16p4 a6 n2 1 ¼ sin2 c Iin n2 þ 2 r2 l4 where a ¼ Particle radius n1 ¼ Refractive index of particle n2 ¼ Refractive index of solvent l ¼ Wavelength of light Iin ¼ Incident power per unit area I ¼ The scattered power per unit area r ¼ Distance from the scattering region and c ¼ Angle between input polarization and scattering direction The total scattering (Pscat) into an aperture of collection angle y (numerical aperture NA ¼ sin y) is then: Eq. 11: Total light scattered into an aperture Pscat
64p4 a6 n2 1 ¼ Z Iin n2 þ 2 0 l4
yÞ yÞ where Z0 ¼ ð1cos þ ð1cos 4 12 As the system has fixed laser wavelength and power as well as a fixed aperture and detection angle, the variables that affect the lower limit of detection consist primarily of the particle size and the relative refractive index difference between the particles and the solvent in which they are suspended. As light scattering at these scales varies as a function of radius [6], as particles reduce in size, they scatter considerably less light. The functionality of NTA relies on the ability to visualize the particles, and so if the amount of light scattered is not sufficient that the particles can be seen, then it is not possible to size them. For strongly scattering materials such as colloidal gold, the lower limit of detection is around 10 nm. For those materials with a weaker scattering potential, such as biological materials, the lower limit of detection is more likely to be 20–40 nm. The upper size limit of the system is determined by the rate of Brownian motion. When particles reach a certain size the rate of Brownian motion becomes so low that inherent centering errors in the software start to negate the tracking of the movement leading to inaccuracies in sizing. For dense, high Ri particles such as gold, their rapid sedimentation rate and excessive scattering 3
restricts the upper size at which they can be analyzed to approximately 1,173 K). In common CVD reactors, source molecules react and produce both molecular and cluster
Nanoparticles
N
1665
Nanoparticles, Fig. 6 Schematic drawing of a spray pyrolysis system
Electrical precipitator or Bag filter
Gas carrier
furnace
exhust
starting solution atomizer
precipitator
Electrical furnace or flame
Two fluid nozzle or ultrasonic nebulizer
intermediates in the gas phase. The intermediates become supersaturated and cause the nucleation in the gas phase due to the low vapor pressure. Nucleated NPs collide vigorously with each other and form agglomerated particles [10]. Growth and agglomeration of the particles are mitigated via rapid expansion of the two-phase gas stream at the outlet of the reaction chamber. Subsequent heat treatment of the synthesized nanopowders in various high-purity gas streams allows compositional and structural modifications, including particle purification and crystallization, as well as transformation to a desirable size, composition, and morphology. CVD is widely used to produce carbon nanotubes [11]. Bottom-Up: Liquid Phase Methods
Liquid phase methods include the set of processes that involve a wet chemistry route such as Hydrothermal (solvothermal) synthesis, sol-gel synthesis, and microemulsion synthesis. The particle formation mechanism is the same as in the vapor phase process. While in gas condensation it is not easy to control grain size and crystal shape, and furthermore its product yield is low, the wet chemistry route provides reasonable control over growth by utilizing organic ligands that act as growth confining agents.
Bottom-Up: Hydrothermal Synthesis
Hydrothermal synthesis of NPs provides precise control over particle size, shape distribution, and crystalline fraction. This method takes advantage of the solubility of almost all inorganic substances in water at elevated temperatures and pressures and subsequent crystallization of the dissolved material from the fluid. Water at high temperatures plays an essential role in the precursor material transformation because the vapor pressure is much higher and the structure of water at elevated temperatures is different from that at room temperature. The properties of reactants, including their solubility and reactivity, also change at high temperatures. The solvent is not limited to water but includes also other polar or nonpolar solvents, such as benzene, and the process is more appropriately called solvothermal synthesis in different solvents [12]. Bottom-Up: Sol-gel Synthesis
The sol-gel process is a recent technique widely used in the field of material science and ceramic engineering for the generation of colloidal NPs from liquid phase. Sol-gel method is well adapted for oxide NPs and
N
N
1666
Nanoparticles
composite nanopowder synthesis starting from a chemical solution (sol) which is followed by the formation of a gel. It is based on inorganic polymerization reaction including hydrolysis, polycondensation, drying and thermal decomposition. Typical starting materials are inorganic precursors: metal or nonmetal alkoxides (MOR) dispersed in an appropriate solvent. Precursors hydrolyze with water or alcohols, according to the reaction scheme:
Microemulsion 1
Microemulsion 2
Aqueous phase
Oil phase
Aqueous phase
Microemulsion 1+2
MOR þ H2 O ! MOH þ ROH ðHydrolysisÞ
Oil phase
Collision and coalescence of droplets
MOH þ MOR ! MOM þ ROH or MOH þ MOH ! MOM þ H2 O ðCondensationÞ The rates of hydrolysis and condensation are important parameters that affect the properties of final products. Slower and more controlled hydrolysis typically leads to smaller particle size and more unique properties. After solution condensation to gel, the solvent should be removed. This requires a drying process, which is typically accompanied by a significant amount of shrinkage and densification. Generally, higher calcination temperatures are needed to decompose the organic precursors [11]. Bottom-Up: Microemulsion Method A microemulsion is a thermodynamically stable dispersion of at least three components: two immiscible fluids and a surfactant with amphiphilic properties (stabilizer). A surfactant is a molecule that possesses both polar and nonpolar moieties. In very diluted water (or oil) solutions, it dissolves and exists as monomer, but when its concentration exceeds a certain minimum critical micelle concentration (CMC), the stabilizer molecules associate spontaneously to form aggregates-micelles [13]. The formation of microdroplets can be in the form of oil-swollen micelles dispersed in the aqueous phase as the oil/water (O/W) microemulsion or waterswollen micelles dispersed in oil as the water/oil (W/O) microemulsion (reverse microemulsion) [12]. The formation of O/W or W/O micelles is driven by strong hydrophobic interactions of the hydrophobic tail of the surfactant molecule (O/W micelle) or by hydrophilic interactions of the polar head of the surfactant molecule (W/O micelle).
Chemical reaction Precipitate (metal or metal oxide)
Nanoparticles, Fig. 7 Proposed mechanism for the formation of metal particles within micelle “nanochamber”
Micelles in these systems can be described as “nanoreactors” providing a suitable environment for controlled nucleation and growth. Figure 7 illustrates the mixing of two microemulsions containing the appropriate reactants (metal salt and reducing agent) during the collision of the water droplets. This intermicellar interchange process is very fast (from 10 ms to 1 ms) and it is the rate-limiting step for particle growth, but it is slow when compared to diffusion of reagents inside the nanoconfinement (“nanoreactor”). The attractive interaction between the droplets, which is responsible for the percolation process, is of great importance for the kinetics [13]. Characterization of Nanoparticles NPs are generally characterized in particle size distribution, morphology, charge determination, and surface hydrophobicity. More specific characterizations depend on the properties of NPs synthesized such as magnetic characterizations for NPs featuring magnetic properties (e.g., Fe, Fe3O4) and electric characterizations for conductive or dielectric NPs (e.g., Ag, Cu, Au, SiO2, BaTiO3, etc.). Different parameters versus principal characterization methods used for NPs are given in Table 1.
Nanoparticles Nanoparticles, Table 1 Parameters versus principal characterization methods for Nps Parameters Characterization methods Particle size, size Scanning Electron Microscopy distribution, morphology (SEM), Transmission Electron Microscopy (TEM), Atomic Force Microscopy (AFM), X-ray Diffraction (XRD), Laser diffractometry Charge determination Zeta potentiometer Surface hydrophobicity Water contact angle measurements, hydrophobic interaction chromatography Magnetic features Vibrating Sample Magnetometry (VSM), Alternating gradient force magnetometry (AGFM), Magnetic Force Microscopy (MFM) Electronic properties Percolation threshold measure (both current-voltage DC and AC), dielectric characterization
Magnetic Nanoparticles Magnetic NPs containing pure 3D transition metals – or some mixture of their oxides – are one of the most studied nanomaterials in view of the prospective, ubiquitous applications in quite different areas, the most notable being biomedicine, sensor technology, and magnetic recording [14]. According to their usage, magnetic NPs are often either embedded in a (typically diamagnetic) solid [15], or dispersed in a fluid; in some cases they are surrounded by an outer shell of a diamagnetic material. This entry deals with particles having sizes not exceeding a few tens of nanometers. The lower limit is taken to be a few (>2) nanometers. As a result, the number of magnetic atoms per NP typically ranges between hundreds and a few millions. Smaller aggregates should be more properly termed magnetic clusters; their magnetic properties are closer to those of magnetic surfaces and are described in a quite different way. On an even smaller scale, magnetic molecules (mainly organic molecules containing a small number, about 10, of transition metal ions) are the subject of much study even if their applicability is still controversial. Nowadays, magnetic NPs are routinely prepared by a variety of techniques which can be termed as physical (e.g., laser-assisted deposition, sputtering, etc.) or chemical (e.g., wet chemical routes) [14, 16].
1667
N
The main feature of bona fide NPs made of a particular element or compound is that their magnetic properties are quite different from those of the corresponding bulk material. In a simplified picture, an ideal magnetic NP can be figured out as provided of a highly symmetrical body (e.g., spherical, ellipsoid, acicular, etc.) and characterized by a homogeneous magnetization provided by the ferro- (or ferri-) magnetic alignment of elementary magnetic moments, giving rise to a mesoscopic permanent magnetic moment associated to the particle. Of course this picture can be improved to account for the role played by the NP surface, whose chemical and magnetic properties can be quite different from those at the particle core. Strictly speaking, the locution “core-shell nanoparticle” [17] means a particle whose chemical or stoichiometric properties are substantially different when moving from the center to the surface (e.g., a metallic core surrounded by an oxide shell). In a sense, however, all magnetic NPs fall in this category at least when their magnetic properties are considered, even when surface and core do not substantially differ in their chemistry. The simplified approach is instrumental in understanding the reason why magnetic properties of NPs change with respect to bulk counterparts. Any macroscopic magnetic material is usually divided into magnetic domains in order to minimize the magnetostatic energy; the number of domains, as well as their shape, is dictated by the interplay and balance between magnetic energies, such as exchange, anisotropy, and domain-wall energy. When the size of a piece of magnetic material is ideally shrunk down to the sub-micrometer range without changing its aspect ratio, all the associated magnetic energies decrease accordingly; however, they decrease following different laws. Both magnetostatic energy and anisotropy energy are volume energies; however, domain wall energy being a surface energy, it has a slower reduction rate with reducing particle size. This makes the nucleation of magnetic domains unfavorable when the size of a particle becomes smaller. Below a critical size which strongly depends on the material’s magnetic properties and morphology, the total energy at remanence is no longer decreased by magnetic domains, and the NP remains in the single-domain state. The magnetic moment points along one of the magnetically “easy” directions corresponding to the minima of the anisotropy energy. In order to have it switched away
N
N
1668
from the easy direction, one must apply a magnetic field; on removing it, the magnetic moment will be again aligned along any one of the easy directions. This is the so-called single-domain (magnetically) blocked state [18]. However, on further decreasing the particle size, the anisotropy energy keeps decreasing as the particle volume, and eventually becomes so small that it falls below kBT, where kB is the Boltzmann’s constant and T the absolute temperature, that is, the typical thermal energy steadily exchanged at equilibrium by the NP with the thermal bath. As a consequence, the magnetization is no longer “blocked” along an easy direction, being continuously affected by thermal (random) torques of thermal origin, and the magnetic moment vector behaves in a sense as a free rotator which can be aligned by an external magnetic field, very much as paramagnetic substances do. However, the NP’s magnetic moment is usually much larger than that of a single atom or ion, so that it can be termed as a “supermoment,” and the particular magnetic behavior of such a NP is known as “superparamagnetism” [19]. The size below which a NP ceases to be in the magnetically blocked state at a given temperature to enter the superparamagnetic (SP) region, and the temperature above which a NP of given size begins to follow the SP behavior are termed critical size (temperature) for the onset of superparamagnetism, respectively. Both depend on NP size, shape, and the leading anisotropy term. The transition from blocked to SP state is not a thermodynamic phase transition, because the SP behavior emerges only when the magnetic moment’s direction randomly fluctuates at a rate high enough that during the time needed to perform a single measurement it virtually takes all directions in space, so that its time average is zero at zero applied field. The observation time plays a crucial role in determining whether a NP at a given temperature is in the SP state or not. From this viewpoint, the blocked-SP transition resembles, in a sense, that between an undercooled liquid and a glass. SP particles exhibit a reversible magnetization versus field behavior, as paramagnetic substances do. Therefore, no hysteresis loop appears when an alternating magnetic field is applied to a system of genuine SP particles. The material’s coercivity and remanence disappear. On the other hand, blocked single-domain NPs usually exhibit coercivity and
Nanoparticles
remanence values which can be much higher than the ones measured in a macroscopic specimen of the same material. This means that a system containing blocked NPs made of a magnetically soft material acts as a magnetically hard material [18]. Therefore, two quite opposite magnetic behaviors are expected at the same temperature, strictly depending on NP size (and shape) alone. Magnetic NPs in the SP phase are attractive for biomedical applications, because they can be easily driven by a magnetic field (an oscillating field makes them their “free” moments to oscillate; an inhomogeneous field exerts a force on them, displacing them toward a target point). Oscillating magnetic moments dissipate energy as heat, and may effectively contribute in the local heating of organic tissue, for example, malignant tissue. Magnetic NPs may be properly functionalized with diamagnetic, organic shells in order to act as the portable inner core of a larger, multi-layered particle aimed to a specific intra-body target where it can release specific drugs carried by the organic shell (drug delivery). Other applications of functionalized particles include cell separation and DNA reconnaissance. Blocked NPs are interesting because they can be exploited as low-cost substitutes of more expensive high-coercivity materials in permanent magnet industry and in magnetic data storage [16]. Hard-disk magnetic memories are often based upon nanoparticulate media where the information can be safely stored owing to the high coercivity of the medium. In this case, superparamagnetism is detrimental because it causes a loss of information. The SP limit introduces a lower boundary to the size of the NPs which a magnetic memory can be composed of, and poses a major, still partially unsolved technological problem in the development of higher-density magnetic memories. A major drawback of almost all techniques proposed so far to produce magnetic NPs (at least, all of the low-cost ones) is the lack of full control on their size, which results in a (more or less) broad size distribution. Considering how size matters in determining the magnetic properties of these magnetic nanomaterials, it is apparent why many efforts of both fundamental and applied research are now aimed to develop preparation techniques resulting in monodisperse nanoparticle systems.
Nanoparticles
Application of Nanoparticles Here it is proposed a selection of the application fields of NPs, in addition to those of magnetic NPs which were summarized in the previous paragraph. As references, the same literature previously cited was kept. Drinking water purification is a new and promising application, performed making use of Fe, Ni, Au, Ag, or Cu NPs, for example for pesticide removal and organic contaminant detection. The NPs feature an enhanced reactivity and selectivity toward some dangerous elements (such as heavy metals) and organic substances (pesticides); when combined with a MEMS structure, this property is used to detect the chemicals by monitoring certain physical properties of the system (resonance frequency, impedance, etc.). The enhanced chemical activity of metal NPs is also well known in catalysis, where transition metal NPs are used in the synthesis of organic compounds. Catalytic NPs may be surrounded by a dendritic polymeric structure (dendrimer) in order to allow catalyst recovery at the end of the chemical reaction. A peculiar form of catalysis, called water splitting, is performed using semiconductor NPs (Cu2O, Fe3O4), where sunlight is directly converted in a current (either electrons or holes, depending on the semiconductor properties) and used to split the water molecule into oxygen and hydrogen, to produce an environmentally friendly fuel. Cellular imaging, intended as optical microscopy performed on cells using fluorescence, Raman spectroscopy and SERS, is normally enhanced by uptake of NPs into the cell body. For this purpose are often used Au and, TiO2 NPs, and so-called quantum-dots, where this definition includes semiconductor NPs, as CdSe and ZnS. Photoluminescence of semiconductor NPs (quantum dots) is well known and studied in CdTe, Au/CdTe NPs assemblies, PbS, CdS, etc. Biomedical applications of noble metal NPs include targeted drug delivery and targeted heat release; similar applications are more commonly approached by using ferromagnetic NPs (e.g., hyperthermia treatment on malignant cells). Also polymeric NPs, which are not a matter of discussion here, are currently used as vectors for drug delivery. Inhibition of microorganisms and antimicrobial effects are obtained using Ag and/or TiO2 NPs. For solar cells optimization noble metal (Au, Ag) NPs have been proposed as electromagnetic field
1669
N
enhancement media due to their SPR, even though TiO2 and SiO2 NPs possess more interesting antireflection properties. Flexible electronics: Metallic NP tracks (Au, Ag, Cu, Sn), deposited by means of inkjet printing technology, are sintered toward electrical percolation at very low temperatures, allowing the realization of conductive paths on flexible polymeric substrates [20]. Currently, research is going on, considering possible solutions to assemble NPs and realize functional devices. Besides the direct application of NPs, they are currently added to either metallic or ceramic or polymeric matrices to realize nanocomposite materials featuring enhanced mechanical, electronic, thermal, magnetic, and optical properties, which are not a matter of discussion in this entry.
Cross-References ▶ Chemical Vapor Deposition (CVD) ▶ Chitosan Nanoparticles ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles ▶ Nanostructures for Energy ▶ Physical Vapor Deposition
References 1. ESRF newsletter n 38. December 2003 and references therein 2. Patel, J.K., Patel, D.J., Pandya, V.M.: An overview: nanoparticles. Intl. J. Pharmaceut. Sci. Nanotechnol. 1(3), 215–220 (2008), and references therein 3. Hosokawa, M., Nogi, K., Naito, M., Yokoyama, T.: Nanoparticle Technology Handbook, pp. 5–48. Elsevier, Oxford (2007), and references therein 4. Ung, T., Liz-Marza´nand, L.M., Mulvaney, P.: Redox catalysis using Ag@SiO2 colloids. J. Phys. Chem. B. 103(32), 6770–6773 (1999), and references therein 5. Scientific Committee on emerging and newly identified health risks (SCENIHR): The appropriateness of existing methodologies to assess the potential risks associated with engineered and adventitious products of nanotechnologies (2007) and references therein 6. Muno˜z, J.E., Cervantes, J., Esparza, R., Rosas, G.: Iron nanoparticles produced by high-energy ball milling. J. Nanopart. Res. 9, 945–950 (2007), and references therein 7. Chen, Y., Pui Li, C., Chen, H., Chen, Y.: One-dimensional nanomaterials synthesized using high-energy ball milling and annealing process. Sci. Technol. Adv. Mater. 7, 839–846 (2006), and references therein
N
N
1670
8. Gutsch, A., Kr€amer, M., Michael, G., M€ uhlenweg, H., Prido¨hl, M., Zimmermann, G.: Gas-phase production of nanoparticles. KONA 20, 24–37 (2002), and references therein 9. Iskandar, F.: Nanoparticle processing for optical applications – a review. Adv. Powder Technol. 20, 283–292 (2009), and references therein 10. Adachi, M., Tsukui, S., Okuyama, K.: Nanoparticle formation mechanism in CVD reactor with ionization of source vapor. J Nanopart Res 5, 31–37 (2003), and references therein 11. Kumar, M., Ando, Y.: Chemical vapor deposition of carbon nanotubes: a review on growth mechanism and mass production. J. Nanosci. Nanotechnol. 10(6), 3739–3758 (2010), and references therein 12. Tavakoli, A., Sohrabi, M., Kargari, A.: A review of methods for synthesis of nanostructured metals with emphasis on iron compounds. Chem. Pap. 61(3), 151–170 (2007), and references therein 13. Capek, I.: Preparation of metal nanoparticles in water-in-oil (w/o) microemulsions. Adv. Coll. Interf. Sci. 110, 49 (2004), and references therein 14. Reiss, G., H€ utten, A.: Magnetic nanoparticles: applications beyond data storage. Nat. Mater. 4, 725–726 (2005) 15. Allia, P., Tiberto, P., Coisson, M., Chiolerio, A., Celegato, F., Vinai, F., Sangermano, M., Suber, L. and Marchegiani, G.: Evidence for magnetic interactions among magnetite nanoparticles dispersed in photoreticulated PEGDA-600 matrix. J. Nanopart. Res. (2011), in press, doi:10.1007/ s11051-011-0249-7 16. Gubin, S.P., Koksharov, Y.A., Khomutov, G.B., Yurkov, G.Y.: Magnetic nanoparticles: preparation, structure and properties. Russian Chem. Rev. 74(6), 489–520 (2005) ` ., Labarta, A., Batlle, X.: Exchange bias phenom17. Iglesias, O enology and models of core/shell nanoparticles. J. Nanosci. Nanotechnol. 8, 2761–2780 (2008) 18. Tannous, C., Gieraltowski, J.: The Stoner–Wohlfarth model of ferromagnetism. Eur. J. Phys. 29, 475–487 (2008) 19. Knobel, M., Nunes, W.C., Socolovsky, L.M., De Biasi, E., Vargas, J.M., Denardin, J.C.: Superparamagnetism and other magnetic features in granular materials: a review on ideal and real systems. J. Nanosci. Nanotechnol. 8, 2836–2857 (2008) 20. Chiolerio, A., Maccioni, G., Martino, P., Cotto, M., Pandolfi, P., Rivolo, P., Ferrero S., Scaltrito, L.: Inkjet printing and low power laser annealing of silver nanoparticle traces for the realization of low resistivity lines for flexible electronics. Microelectron. Microeng 88, 2481–2483 (2011) and references therein
Nanoparticles Toxicity Examined In Vivo in Animals ▶ In Vivo Toxicity of Titanium Dioxide and Gold Nanoparticles
Nanoparticles Toxicity Examined In Vivo in Animals
Nanopatterned Substrata ▶ Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior Jason P. Gleghorn1 and Celeste M. Nelson1,2 1 Department of Chemical and Biological Engineering, Princeton University, Princeton, NJ, USA 2 Department of Molecular Biology, Princeton University, Princeton, NJ, USA
Abbreviations 1D 2D 3D AFM ECM ES MSC RGD SAM
One-dimensional Two-dimensional Three-dimensional Atomic force microscopy Extracellular matrix Embryonic stem cell Mesenchymal stem cell Argine-glycine-aspartic acid Self-assembled monolayer
Synonyms Nanopatterned substrata
Definition Surfaces (substrata) that are fabricated to have chemically or topographically nanometer-sized features, which are used to investigate cell morphology, phenotype, and/or behavior.
Overview Cells Sense and Respond to Their Environment Cellular environments consist of complex mechanical, chemical, electrical, and topological gradients ranging
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
MESOSCALE
MICROSCALE
NANOSCALE
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior, Fig. 1 Cell interaction and behavior are governed over multiple length scales. Cell-cell and cell-matrix interactions collectively form tissue architecture and function at the mesoscale. Microscale interactions guide cell shape and connectivity, whereas nanoscale interactions influence cell processes such as adhesion, motility, and gene expression
in size from the nanoscale (1–100 nm) to the macroscale (>1 mm) (Fig. 1). Cell behavior is governed by these multiscale environmental properties, which, when disrupted, lead to aberrant tissue function and
1671
N
disease. At the mesoscale (100–1,000 mm), aggregate cell behavior resulting from cell-cell interactions, coupled with properties of the extracellular matrix (ECM), create the tissue architecture and sculpt the mechanical and chemical microenvironment. Heterogeneity of the tissue architecture at the mesoscale produces unique microenvironments. Cell-cell and integrin-dependent cell-ECM interactions at the microscale (1–100 mm) regulate cell shape and adhesion, and consequently affect cell survival, growth, differentiation, and migration. At the nanoscale, molecular and subcellular processes such as integrin activation, focal adhesion formation, actin polymerization, and cytoskeletal organization enable the cell to be physically coupled to its ECM. Additionally, membrane-bound receptors signal through intracellular pathways to modulate gene expression and protein synthesis in response to chemical signals such as morphogens. Cells interact with and regulate their surroundings in a relationship referred to as dynamic reciprocity [1]. Through dynamic and reciprocal crosstalk, cell behavior is modulated by cell-cell and cell-ECM interactions, and that behavior sculpts and defines the microenvironment. One of the many ways in which the cell senses and reacts to its environment is through integrin-dependent signaling. Transmembrane integrins attach and mechanically couple to several ECM ligands, including the prototypic sequence arginine-glycine-aspartic acid (RGD). Integrin engagement leads to recruitment and clustering to form focal adhesions. Intracellularly, integrins within the focal adhesion associate with several proteins including vinculin, talin, tensin, a-actinin, paxillin, Src, and focal adhesion kinase. These proteins couple the focal adhesion to the actin cytoskeleton. This physical linkage enables the cell to adhere to, move along, and interact mechanically with the ECM. In addition, the focal adhesion activates several kinase cascades that biochemically transduce information about the microenvironment into the nucleus, which leads to changes in gene expression. These gene expression changes can result in the production and secretion of soluble signaling molecules, enzymes, and ECM components that shape and maintain the extracellular environment. The formation, maturation, and disassembly of focal adhesions are not only key for cell spreading and migration, but also appear to be central modulators of many cellular functions including proliferation and differentiation.
N
N
1672
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
Engineering Model Substrata The tissue microenvironment is complex and the responses of cells to microenvironmental interactions can occur through individual and synergistic mechanisms. As a result, model substrata are often employed to decouple and isolate the effects of various microenvironmental factors. These model surfaces typically fit into one of two categories: chemically or topologically patterned. Chemically patterned surfaces contain regions of active and inactive (inert) biomolecules and thus present spatial patterns of chemical ligands to the cells of interest. ECM proteins such as fibronectin, vitronectin, laminin, and collagen or their active binding motifs are often used. In contrast, topologically patterned surfaces alter the surface topology that is presented to the cells. In these systems, etched pits, posts, grooves, and gratings are used to characterize cell behavior. Regardless of the type of model substratum used, advances in micro- and nanotechnology have enabled the creation of such tailored surfaces to investigate the interactions between cells and their microenvironment at multiple length scales. Because of the existence of several simple, wellestablished techniques, micropatterned surfaces have become a popular tool to interrogate cell behavior. Micrometer-scale topology can be created using standard photolithography coupled with traditional etching processes; however, at the micrometer scale most studies have focused on chemically patterned substrata. For chemically patterned surfaces, photolithographic techniques can be used to spatially deposit or remove different materials and thus define regions for chemical functionalization. Microcontact printing is the most common method of microscale chemical functionalization. This technique employs the use of micropatterned elastomeric stamps that are “painted” with the ECM ligand of interest. The coated stamps are then pressed onto a treated surface, which deposits the cell-adhesive ECM ligands onto the surface in the same spatial pattern as that of the stamp. Micrometer-sized features can be used to control cell position and connectivity; and thus micropatterning is often employed to investigate both mesoscale and microscale cell behavior. Micropatterned surfaces have been used to coculture fibroblasts and hepatocytes to maintain normal hepatocyte phenotype [2]. Additional mesoscale behavior has been investigated with epithelial sheets grown on micropatterned islands of varying size and shape.
Patterns of cell proliferation emerge that result from cell-cell and cell-ECM interactions within the epithelial sheet [3]. Chemical micropatterning has also been used with single cells to identify the role of cell shape in determining cell fate [4]. Geometric control of cell growth and viability was demonstrated by varying the size of chemically patterned islands of ECM. Although micropatterning is a powerful tool to investigate cell behavior, particularly at the mesoscale, this approach has its limitations. Micropatterning techniques do not afford the ability to control the surface density of the patterned ligand. Additionally, the minimum feature sizes achieved with these techniques do not permit the exposure of multiple different ligands to a single cell. These drawbacks make it difficult to probe cell-ECM interactions at the scale of integrin engagement or focal adhesion formation. Additionally, cells in vivo are not exposed to flat two-dimensional (2D) surfaces but rather to a complex topological microenvironment. The ECM consists primarily of interwoven protein fibers ranging from 10 to 300 nm in diameter. Basement membranes consist of complex mixtures of nanoscale pits, pores, and fibers ranging in size from 5 to 200 nm with peaks and valleys of approximately 100 nm in height [5]. The investigation of cell behavior in response to these native nanotopologies cannot be explored using micropatterned surfaces. The advent of new tools and fabrication techniques has enabled the creation of spatial patterns in chemistry and topology that can vary greatly over the length scale of a cell. Nanoscale differences in surface roughness, fiber diameter, and ligand spacing can be produced using micropatterned substrata to more closely recapitulate the native tissue microenvironment. Thus, the goal of nanoscale patterning is to investigate molecular mechanisms and subcellular processes that determine fundamental cell behaviors including adhesion, proliferation, gene expression, and differentiation. Fabricating Nanopatterned Substrata Unlike micropatterned surfaces, fabrication of nanopatterned surfaces large enough for multicellular studies is technically challenging as well as time and labor intensive. As a result, studies to date have focused primarily on the behavior of single cells. The types of fabrication methodologies discussed herein are by no means comprehensive, but rather
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior Nanopatterned Surfaces for Exploring and Regulating Cell Behavior, Fig. 2 Cartoon representations of various fabrication modalities used to create nanopatterned surfaces
E-Beam Lithography
N
1673
Dip Pen Nanolithography
V
Molecular Self-assembly
a sampling of commonly used techniques to create chemically and topologically defined nanopatterned surfaces. Fabrication technologies adapted from the semiconductor industry as well as novel nanobiotechnology methods are rapidly changing, enabling faster and simpler creation of nanopatterned features. For a more comprehensive review of fabrication techniques, the reader is directed to specialized reviews [6]. Topologically nanopatterned substrata are created using several techniques. These surfaces can be categorized as unordered or ordered topologies. Unordered topologies typically arise spontaneously as a result of processing methods including chemical etching and polymer demixing. Surfaces with unordered topologies have randomly patterned features and lack orientation and geometrical control, but the fabrication processes are relatively simple, fast, and inexpensive. Ordered topologies, on the other hand, are created with precise control of feature pattern, orientation, and geometry. Electron-beam (e-beam) lithography (Fig. 2) is the most commonly used approach to create nanoscale ordered topologies. E-beam lithography uses an electron beam to develop a layer of resist
Electrospinning
coated on the substratum. Following e-beam lithography, further processing of the substratum using standard etching techniques results in nanoscale topologies including nanoscale grooves, gratings, or pits. Dip-pen nanolithography (Fig. 2) is a method to create chemically nanopatterned surfaces using an atomic force microscopy (AFM) tip [7]. The AFM tip is “dipped” into a reservoir, coating the tip with the ligand of interest, and is then used to repeatedly deposit small amounts of the ligand onto the substratum to form chemically patterned regions with an approximate spatial resolution of 5 nm. This technology enables the placement of different adhesive ligands in close proximity in defined shapes with nanoscale precision and dimensions. Additionally, this technology can be multiplexed to create a higher throughput chemical patterning system by combining multiple AFM tips in parallel. In addition to the “top-down” fabrication strategies used to deposit or remove material from a surface as described above, newer “bottom-up” technologies of molecular self- or templated-assembly (Fig. 2) have been used to create larger chemically nanopatterned surfaces. Molecular self-assembly is a broad category
N
N
1674
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
of techniques wherein (supra)molecular building blocks or colloids are spontaneously self-organized as a result of their interactions to form nanoscale topology and/or spatial patterns of cell-adhesive ligands [8]. Similarly, templated-assembly creates the same types of surfaces with the use of a reusable template to define an organization for the molecular building blocks. These newer fabrication techniques (e.g., selfassembled monolayers (SAMs) and phase-separated block copolymers) have enabled the reliable creation of large patterned areas. Whereas all of the fabrication techniques described above can be combined to create surfaces with heterogeneously patterned features that vary across length scales, other techniques such as electrospinning [9] do exist to create patterned surfaces that simultaneously expose cells to topologies with widely different dimensions. In electrospinning, an electric field is used to produce polymer fibers from a liquid solution. As the fiber orientation and diameter are easily controlled with simple experimental parameters, fibrous mats containing both nanoscale and microscale fibers can be created with aligned or randomly oriented topologies. Additionally, multiple fibers and complex fiber geometries can be spun simultaneously from different materials or conjugated with different cell-adhesive ligands. This produces substrata that more closely mimic the topology of tissues in vivo: complex fibrous meshes with nano- to micro-sized fibers and randomly sized micropores/pits. Thus, electrospun substrata are widely used to determine how cell behavior is regulated by fiber alignment and multiscale distributions of fiber size and surface ligands.
Key Research Findings Chemically nanopatterned substrata have been primarily used to investigate integrin engagement and focal adhesion formation in an effort to dissect out cell adhesion properties. The majority of these studies have been carried out on self-assembled surfaces containing RGD-functionalized gold nanoparticles [10] (Fig. 3). Integrin receptors are approximately 10 nm wide and these “nanodots” can be fabricated such that they bind to single integrins. As a result, spacing of bound integrins can be altered by changing the spacing of the nanodots on the substrata. These nanopatterned surfaces have revealed that cell
behavior is regulated by the spacing of adhesive ligands. Results from early studies showed that cell motility was affected on surfaces with RGD ligand spacings in the range of 6–300 nm and that spacings from 14 to 25 nm influenced cell-adhesion strength. Subsequent studies explored nanodot spacings between 28 and 85 nm [11] and found that RGD ligand spacings from 58 to 73 nm are optimal for integrin clustering and activation, whereas ligand separation greater than 73 nm causes significantly less cell adhesion, spreading, and integrin-mediated focal contact formation in osteoblasts. Additional studies have measured cell detachment forces from these substrata and demonstrated that spacings greater than 90 nm inhibited focal adhesion formation and decreased detachment forces compared to spacings less than 50 nm [12]. Overall, the data obtained using these chemically nanopatterned surfaces indicate that RGD ligand spacings should be less than approximately 70 nm to stimulate collective cell functions. Additionally, these stimulatory effects tend to increase as the ligand spacing decreases down to approximately 10 nm. Studies using topologically nanopatterned substrata can be grouped into two categories, ordered or unordered, based on the layout of the surface features. Ordered surfaces consist of repeated features that are arranged with consistent height, spacing, and orientation. Such surfaces include those with uniform posts or pits and gratings formed by regularly spaced groves. Conversely, surfaces with randomly oriented, varying sized features, such as electrospun meshes, are classified as unordered. The mechanisms by which nanotopographic cues regulate cell behavior are not well understood. The emerging picture is that changes in cytoskeletal organization and structure may drive some of these cell behaviors, potentially in response to the underlying geometry or feature size of the surface. Regardless, it is clear that there is a range of different responses, or differing levels of response that vary by cell type. Morphology and proliferation are altered when cells are cultured on surfaces with ordered nanotopography [14] (Fig. 4). Human mesenchymal stem cells (MSCs) grown on 350 nm-wide grooves have an aligned cytoskeleton along the direction of the features. Likewise, stem-cell-derived osteoblasts not only aligned and spread in response to nanogrooves of polystyrene, but also showed alignment in actin and
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
1675
N
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior, Fig. 3 Example substrata. (a, b) Ligand-coated gold nanodots can be spaced at varying distances to form chemically nanopatterned substrata (Reproduced with permission from SelhuberUnkel et al. [12]). (c, d) The basement membrane of the cornea has topographical variations with both micrometer- and nanometersized features (Reproduced with permission from Abrams et al. [5]). Electrospun meshes can be formed to create both unordered and ordered topographies (e) and fiber diameters can vary within individual samples (f) (Reproduced with permission from Carnell et al. [13])
N
mineralized matrix. Additional studies with human embryonic stem cells (ES) cultured on 600 nm polydimethylsiloxane (PDMS) ridges have demonstrated an altered organization of cytoskeletal components including F-actin, vimentin, g-tubulin, and a-tubulin. The alterations to morphology and proliferation were eliminated when the ES cells were exposed to actin-disrupting drugs [15]. Generally, cells seem to be more sensitive to the depth of the groove as opposed to the groove pitch or spacing. Additionally, cell orientation increases with increasing groove depth on grated surfaces but decreases as the groove width or pitch increases. Whereas the phenomenon of cell and cytoskeletal alignment is fairly conserved over a wide range of nanogratings of varying dimensions, cell behavior is more varied in response to ordered surfaces with pits and posts. The adhesion of osteoblasts was significantly increased when cultured on 11 nm-high posts
compared to either 85 nm-high posts or flat polystyrene [16]. Similarly, fibroblasts also demonstrated increased adhesion and spreading on 13 nm posts compared with 95 nm-high posts. However, endothelial cells displayed less spreading and lower cytokine production when cultured on 100 nm-high posts as compared to flat control surfaces. In general, cells cultured on surfaces that are uniformly or randomly textured with nanometer-sized pits and pillars do not display aligned or preferential orientations. Although comparing different unordered topologies is difficult due to differences in feature size, shape, uniformity, and chemistry, a trend does seem to emerge. In electrospun fiber meshes ranging from 283 to 1,425 nm in diameter, rat hippocampus-derived adult neural stem cells differentiate and proliferate in response to fiber diameter [17]. For fiber meshes smaller than 283 nm, cells did not demonstrate preferential patterns of spreading; however, as the fiber
N
1676
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior, Fig. 4 Examples of cell responses to nanotopography. Uniform cell spreading is observed on flat substrata (a), whereas cells spread and align along the direction of nanograted substrata (b) (Reproduced with permission from Yim et al. [14]). Cells spread, attaching to and around fibers on an unordered electrospun mesh (c–e). The cell is highlighted in yellow and scale bars are 10 mm (c) and 5 mm (d, e) (Reproduced with permission from Christopherson et al. [17])
diameter increased, cells extended along the length of the fiber. In contrast, proliferation increased as fiber diameter decreased. In another study investigating cell adhesion, fibroblasts demonstrated a higher level of adhesion to surfaces with unordered topology as compared to either flat surfaces or those with ordered topology [18]. In summary, the general trend is that cell behaviors such as proliferation, adhesion, and spreading are positively stimulated as the feature size decreases from 100 to 10 nm. While almost all of the studies to date have focused exclusively on either chemically or topologically patterned substrata, these parameters are difficult to decouple. Chemically nanopatterned surfaces inevitably have 5–10 nm variations in height across the substratum due to the islands of conjugated ligands. In topographically nanopatterned systems, differences in protein transport and adsorption due to structural features may result in differences in the presentation of adhesive areas and thereby alter cell behavior. Few studies have investigated the competitive or synergistic effects of both chemical and topographic cues [16].
When continuous chemically patterned strips of celladhesive ligand were orthogonally overlaid onto nanogrooves, fibroblasts preferentially oriented along the chemical patterns. However, when fibronectin was orthogonally overlaid discontinuously along the top of a nanogroove pattern, osteoblasts preferentially aligned along the topographic pattern. From these limited data it is clear that a variety of systems involving both chemical and topological patterning over varying nanometer-sized scales need to be employed to fully understand interaction effects.
Applications The basic science studies highlighted herein that seek to understand how nanoscale environmental cues influence cell behavior are fundamental to a range of applications. In particular, these data can be used to engineer biomaterial surface characteristics for improved performance. Materials that interact with the circulatory system, such as those used for stents or
Nanopatterned Surfaces for Exploring and Regulating Cell Behavior
cardiac catheters, could be designed with a nanopatterned surface to prevent cell adhesion and clotting. Conversely, orthopedic implants, which need to integrate strongly with the surrounding tissue, could be designed with a nanopatterned bioactive surface to encourage cell adhesion, proliferation, and motility into the surface. Thus, understanding the nanoenvironmental cues will enable engineers to recreate physiologically desirable microenvironments for various biomaterial-tissue applications. Nanopatterning also has applications for diagnostics and cell biology studies such as cell-based biosensors and cell sorting and culture systems. Integration of micro- and nanotechnologies with cell biology has yielded many new and exciting approaches. Use of microfluidics has created cellson-a-chip or lab-on-a-chip designs where differential cell adhesion based on cell type is critical. Introduction of nanopatterned surfaces can enable a passive physical modification to these systems that would save fabrication time and complexity. Nanopatterned surfaces can be used as passive cell sorting systems that leverage the fact that different cell types respond differently to the same nanotopology. Thus a heterogeneous mixture of cells could be exposed to a nanopatterned surface and only the desired cell population would adhere. Additionally, as understanding of stem cell behavior on nanopatterned substrata grows, one could envision designing substrata that control stem cell adhesion, proliferation, and differentiation – cell behaviors that are difficult to control and that are currently modulated using complex media formulations and cell handling procedures. Finally, studies with nanopatterned substrata can be integrated into the fields of tissue engineering and regenerative medicine. It is clear that the native in vivo environments and the chemical and physical cues to which cells are exposed are vastly different from those presented by traditional in vitro surfaces such as tissue culture plastic. Examination of signaling cascades within cells on nanopatterned surfaces can reveal how cells detect and respond to chemical and topographic signals in terms of short- and long-term cell functions, such as changes in gene expression and protein synthesis. These details of scaffold architecture and ligand presentation can improve tissue engineering technologies, and provide insight on disease processes where microenvironmental regulation is disrupted, such as in cancer. While some examples of
1677
N
nanopatterned tissue scaffolds exist, particularly with nanoporous materials for bone tissue engineering, the link between cell behavior in 2D and 3D is not straightforward. Cells in 3D environments adhere differently than those in 2D environments and contain different cell-ECM adhesions [19]. A recent study has even demonstrated that cells cultured on 1D patterns (lines) of adhesive ligands behave more similarly to cells in 3D matrices than to those on 2D chemically patterned islands [20]. Thus, while the link between cell behaviors in 2D and 3D is not certain, it is clear that further understanding gained using wellcontrolled nanopatterned surfaces can contribute to improved 3D tissue models and translational tissue engineering technologies.
Cross-References ▶ Dip-Pen Nanolithography ▶ Electron-Beam-Induced Deposition ▶ Electrospinning ▶ Nanoimprint Lithography ▶ Self-assembly
References 1. Bissell, M.J., Hall, H.G., Parry, G.: How does the extracellular matrix direct gene expression? J. Theor. Biol. 99(1), 31–68 (1982) 2. Bhatia, S.N., Balis, U.J., Yarmush, M.L., Toner, M.: Probing heterotypic cell interactions: hepatocyte function in microfabricated co-cultures. J. Biomater. Sci. Polym. Ed. 9(11), 1137–1160 (1998) 3. Nelson, C.M., Jean, R.P., Tan, J.L., Liu, W.F., Sniadecki, N.J., Spector, A.A., Chen, C.S.: Emergent patterns of growth controlled by multicellular form and mechanics. Proc. Natl. Acad. Sci. U.S.A. 102(33), 11594–11599 (2005) 4. Chen, C.S., Mrksich, M., Huang, S., Whitesides, G.M., Ingber, D.E.: Geometric control of cell life and death. Science 276(5317), 1425–1428 (1997) 5. Abrams, G.A., Goodman, S.L., Nealey, P.F., Franco, M., Murphy, C.J.: Nanoscale topography of the basement membrane underlying the corneal epithelium of the rhesus macaque. Cell Tissue Res. 299(1), 39–46 (2000) 6. Norman, J.J., Desai, T.A.: Methods for fabrication of nanoscale topography for tissue engineering scaffolds. Ann. Biomed. Eng. 34(1), 89–101 (2006) 7. Lee, K.B., Park, S.J., Mirkin, C.A., Smith, J.C., Mrksich, M.: Protein nanoarrays generated by dip-pen nanolithography. Science 295(5560), 1702–1705 (2002)
N
N
1678
8. Zhang, S.: Fabrication of novel biomaterials through molecular self-assembly. Nat. Biotechnol. 21(10), 1171–1178 (2003) 9. Pham, Q.P., Sharma, U., Mikos, A.G.: Electrospinning of polymeric nanofibers for tissue engineering applications: a review. Tissue Eng. 12(5), 1197–1211 (2006) 10. Lim, J.Y., Donahue, H.J.: Cell sensing and response to micro- and nanostructured surfaces produced by chemical and topographic patterning. Tissue Eng. 13(8), 1879–1891 (2007) 11. Arnold, M., Cavalcanti-Adam, E.A., Glass, R., Blummel, J., Eck, W., Kantlehner, M., Kessler, H., Spatz, J.P.: Activation of integrin function by nanopatterned adhesive interfaces. Chemphyschem 5(3), 383–388 (2004) 12. Selhuber-Unkel, C., Erdmann, T., Lopez-Garcia, M., Kessler, H., Schwarz, U.S., Spatz, J.P.: Cell adhesion strength is controlled by intermolecular spacing of adhesion receptors. Biophys. J. 98(4), 543–551 (2010) 13. Carnell, L.S., Siochi, E.J., Holloway, N.M., Stephens, R. M., Rhim, C., Niklason, L.E., Clark, R.L.: Aligned mats from electrospun single fibers. Macromolecules 41(14), 5345–5349 (2008) 14. Yim, E.K., Pang, S.W., Leong, K.W.: Synthetic nanostructures inducing differentiation of human mesenchymal stem cells into neuronal lineage. Exp. Cell Res. 313(9), 1820–1829 (2007) 15. Gerecht, S., Bettinger, C.J., Zhang, Z., Borenstein, J.T., Vunjak-Novakovic, G., Langer, R.: The effect of actin disrupting agents on contact guidance of human embryonic stem cells. Biomaterials 28(28), 4068–4077 (2007) 16. Lim, J.Y., Hansen, J.C., Siedlecki, C.A., Runt, J., Donahue, H.J.: Human foetal osteoblastic cell response to polymerdemixed nanotopographic interfaces. J. R. Soc. Interface 2(2), 97–108 (2005) 17. Christopherson, G.T., Song, H., Mao, H.Q.: The influence of fiber diameter of electrospun substrates on neural stem cell differentiation and proliferation. Biomaterials 30(4), 556–564 (2009) 18. Curtis, A.S., Casey, B., Gallagher, J.O., Pasqui, D., Wood, M.A., Wilkinson, C.D.: Substratum nanotopography and the adhesion of biological cells. Are symmetry or regularity of nanotopography important? Biophys. Chem. 94(3), 275–283 (2001) 19. Cukierman, E., Pankov, R., Stevens, D.R., Yamada, K.M.: Taking cell-matrix adhesions to the third dimension. Science 294(5547), 1708–1712 (2001) 20. Doyle, A.D., Wang, F.W., Matsumoto, K., Yamada, K.M.: One-dimensional topography underlies threedimensional fibrillar cell migration. J. Cell Biol. 184(4), 481–490 (2009)
Nano-patterning ▶ BioPatterning
Nano-patterning
Nanophotonic Structures for Biosensing Emiliano Descrovi1, Mirko Ballarini2 and Francesca Frascella1 1 DISMIC – Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, Torino, Italy 2 DIFIS – Dipartimento di Fisica, Politecnico di Torino, Torino, Italy
Synonyms Nano-optical biosensors; Nano-optics biosensing
Definition Nanophotonic structures for biosensing: onedimensional, two-dimensional, or three-dimensional structures made of suitable materials, having dimensions ranging from few nanometers to few hundreds of nanometers, said structures showing a well-defined interaction with light providing an optical transduction mechanism for detecting/revealing specific biomolecules in close proximity to the structure itself.
Overview Optical biosensors [1] constitute powerful detection and analysis tools with wide applications in the biomedical domain. They can provide parallel detection within a single device. Generally speaking, there are two detection methods that are implemented in optical biosensing: fluorescence-based detection and labelfree detection. Thanks to the recent advances of technological capabilities for the fabrication of nanometer-sized structures, most of the conventional techniques can be improved, and new original techniques are become feasible. In fluorescence-based detection, target molecules are labeled with fluorescent tags, typically organic dyes and/or chemically functionalized/decorated quantum dots; the overall intensity of fluorescence indicates the presence of the target molecules. Some quantification can also be inferred, but in general quantitative analysis is challenging due to the fluorescence
Nanophotonic Structures for Biosensing
signal bias, as the number of fluorophores on each molecule cannot be precisely controlled. While fluorescence-based detection is extremely sensitive, with the detection limit down to a single molecule, it suffers from laborious labeling processes that may also interfere with the biological/chemical function of the biomolecule and/or its interaction with the environment. In label-free techniques, target molecules are not labeled or altered, and are detected in their natural forms. This type of detection is relatively easy and cheap to perform, and allows for kinetic measurement of molecular interaction. Despite all these differences between fluorescencebased and label-free detection, both protocols can benefit of the use of proper nanostructures and provide complementary information regarding interactions among biomolecules. In this framework, the use of dielectric, metallic, and metallo-dielectric structures obtained by means of a number of methods, for example, optical, electronic or ionic lithographic processes, physical/chemical deposition/sputtering, chemical synthesis, or self-assembling can improve both fluorescence-based and label-free biosensing. In the following, a broad variety of structures are described with particular focus to their respective main applications, but additional original combinations are still under investigation. Label-free detection Label-free techniques include refractive index (RI) detection, optical absorption detection, and Raman spectroscopic detection. In a typical label-free biosensing application a mean should be provided in order to uniquely identify the desired target biomolecules. There are basically two strategies that are implemented: a chemical functionalization of the optically sensitive area of the photonic nanostructure and a spectroscopic measurement. In the first case, the specificity of the photonic biosensor is demanded to the chemical functionalization, therefore allowing the specific detection of desired target molecules on the unique molecular recognition mechanism embodied by the chemical functionalization. In the second case, the recognition of the target molecules is performed by detecting the corresponding spectroscopic fingerprint (either the IR absorption or the Raman emission spectra). Unfortunately, the low Raman activity of the majority of biomolecules makes them little suitable for Raman analysis. Furthermore, a detectable Raman
1679
N
spectrum cannot be attained by simply increasing the power of the laser excitation radiation, because irreversible damages to the biological matter can be induced. Photonic metallic nanostructures producing strong field enhancements by means of Localized Surface Plasmons (LSP) can overcome this limitation, thus allowing a Surface-Enhanced Raman Scattering (SERS) or a Tip Enhanced Raman Scattering spectroscopic detection. Nanostructures sustaining LSP can also provide for Surface-Enhanced Infrared Absorption (SEIRA) spectroscopic detection. Surface-enhanced Raman spectroscopy (SERS) was originally discovered in the 1970s where it was found that submonolayers of small organic molecules, when adsorbed onto the surface of silver nanoparticles (and a few other metals such as copper and gold) would exhibit greatly enhanced Raman intensities (enhancements of 106) [2]. These results have always held the promise for using this technique to observe very low concentrations of molecules on nanoparticles and nanostructured surfaces, but only recently has this promise started to be fulfilled in a predictable way. Thanks to exciting advances in techniques for making nanoparticles (such as nanosphere e-beam lithography and Focused Ion Beam), to characterize the surfaces using electron and scanning probe microscopies, to functionalizing the surfaces of the particles using self-assembled monolayers with attached chemical receptors, and to the laser and optics technology associated with measuring the Raman spectra, a number of important applications have been reported recently. SERS has almost exclusively been associated with three metals, silver (by far the most important), gold, and copper. The generalization to other metals and other materials has been explored since the discovery of SERS, but only in the last few years have the tools been available to generate well-characterized experiments where the reported enhancements are reliable. Light incident on the nanoparticles induces the conduction electrons in them to oscillate collectively with a resonant frequency that depends on the nanoparticles’ size, shape, and composition. As a result of these LSPR modes, the nanoparticles absorb and scatter light so intensely that single nanoparticles are easily observed by eye using dark-field (optical scattering) microscopy. The shape of the nanoparticle extinction and scattering spectra, and in particular the peak wavelength lmax, depends on nanoparticle
N
N
1680
composition, size, shape, orientation, and local dielectric environment [3]. The LSPR can be tuned during fabrication by controlling these parameters with a variety of chemical syntheses and lithographic techniques. Although silver and gold are the most commonly used materials, LSPR is theoretically possible in any metal, alloy, or semiconductor with a large negative real dielectric constant and small imaginary dielectric constant. When molecules adsorb onto a plasmonic nanoparticle or move to within a few nanometres of its surface, the local electromagnetic fields around the nanoparticle can enhance the Raman scattering by a factor of 106–108 for an ensemble of molecules and by as much as 1014–1015 for single molecules. Two principal enhancement mechanisms are generally thought to explain the large SERS signals. First, an electromagnetic enhancement factor arises because LSPR modes in the metal nanoparticles focus the incident light energy at the nanoparticle surface and also increase the density of states at Stokes-shifted wavelengths. The electromagnetic enhancement factor is typically 105–108 for nanoprisms, but simulations indicate that it can be as large as 1013 for structures such as arrays of nanoprism dimers. Second, chemical enhancement factors arise from changes in the molecular electronic state or resonant enhancements from either existing molecular excitations or newly formed charge transfer states. Wet chemical synthesis methods have now made it possible to fabricate plasmonic nanoparticles having a variety of shapes (for example spheres, triangles, prisms, rods, and cubes) with controllable sizes and narrow size distributions. Furthermore, metallic– dielectric, and metallic–metallic core–shell nanoparticles and mixed metallic–alloy nanoparticles with different shapes (for example, nanoshells and nanorice) have been prepared. Fabrication of this large variety of different structures makes a variety of applications possible, as the spectral position of the surface plasmon resonance depends on both the shape and the size of the nanoparticle [4]. An alternative possibility of exploiting the localized near-field generated in close proximity to metallic nanostructures is to employ a metallic nanoprobe or tip where LSP are coupled. Such a structure allows a well localized source of highly confined electromagnetic field to be raster scanned on the surface of a given sample, in such a way that a high spatial resolution Raman mapping of the surface can be performed [5].
Nanophotonic Structures for Biosensing
Tip Enhanced Raman Scattering (TERS) has been observed for a number of molecules adsorbed at various substrates, including single-crystalline metal surfaces, showing thereby a more than million-fold enhancement of the Raman scattering [6]. In order to excite LSPs at the apex of a metallic nanoprobe, a further structuring can be performed. For example, gratings can be lithographed on the side surface of a gold cantilevered tip to ease the excitation of plasmon polaritons upon external lateral illumination of the tip. Alternatively, resonant nanoantennas can be fabricated directly on the top of the tip. An effect quite similar to SERS occurs in the midinfrared region: molecules on metal surfaces show infrared absorption 10–1,000 times more intense than would be expected from conventional measurements without the metal. This effect is referred to as SurfaceEnhanced Infrared Absorption (SEIRA) to emphasize the analogy to SERS [7]. Vacuum-evaporated thin metal films are used most frequently for SEIRA experiments. Thin metal films that show strong SEIRA are not continuous but consist of metal islands smaller than the wavelength of the incident light. The average dimension of the islands is few tens of nanometers. The density, shape, and size of the islands depend on the mass thickness, the evaporation conditions, and the chemical nature of the substrate. As the mass thickness increases, the islands grow in size, contact each other, and eventually form a continuous film. Connection of the islands significantly reduces the enhancement, suggesting that the small metal islands play an important role in the enhancement. The metal islands are polarized by the incident photon field through the excitation of collective electron resonance, or localized plasmon, modes, and the dipole p induced in an island generates a local electromagnetic field stronger than the incident photon field around the island. Experimental results suggest that the collective electron resonance is excited even in the mid-infrared region and that the enhanced electromagnetic field enhances the infrared absorption of adsorbed molecules. As previously stated besides those label-free techniques based on spectroscopic measurements, there are quantitative methods based on the measurements of refractive index changes close to the nanostructured surface of the photonic biosensor. Among these, the Surface Plasmon Resonance (SPR) technique and the waveguide- (or optical fiber-) based technique play a major role [8].
Nanophotonic Structures for Biosensing
SPR is based on the coupling of propagating or stationary plasmons on silver/gold/copper thin films. The potential of surface plasmon resonance (SPR) for characterization of thin films and monitoring processes at metal interfaces was recognized in the late seventies. Generally, an SPR optical sensor comprises an optical system, a transducing medium which interrelates the optical and (bio)chemical domains, and an electronic system supporting the optoelectronic components of the sensor and allowing data processing. The transducing medium transforms changes in the quantity of interest into changes in the refractive index which may be determined by optically interrogating the SPR. The sensor sensitivity, stability, and resolution depend upon properties of both the optical system and the transducing medium [9]. SPR can be used in a prism-coupler configuration or with a grating-coupler configuration. If a metal–dielectric interface is periodically distorted, the incident optical wave is diffracted forming a series of beams directed away from the surface at a variety of angles. If the total component of momentum along the interface of a diffracted order is equal to that of the plasmon, the optical wave may couple to the plasmon. Gratingbased optical SPR sensors have been demonstrated which use the measurement of the light intensity variations at SPR and the wavelength interrogation. A high-sensitivity SPR grating-based gas sensor using silver as an SPR active metal has attained sensitivity 1,000 nm RIU1 in wavelength interrogation mode, in angular interrogation mode, the system’s sensitivity would be about 100 RIU1. Gold-based SPR grating sensors have been used for monitoring biomolecular interactions in aqueous environments, with estimated refractive index sensitivity of 30 RIU1 and 900% RIU1 in the angular interrogation and intensity measurement modes, respectively. The grating coupler has been also used for the excitation of an SPW in an optoelectronic SPR-sensing device based on a Schottky-barier semiconductor structure. Periodic and quasiperiodic structuring of smooth metal films sustaining plasmons can be used to create Bragg mirrors (Stationary plasmons) and Fabry-Perot cavities. Electromagnetic modes propagating at the interface between a homogeneous medium and a truncated periodic structure like a dielectric monodimensional photonic crystal (1DPC), also referred to as Bloch Surface Waves (BSW), have been proposed as an alternative to
1681
N
surface plasmon polaritons (SPP). The possibility to excite simultaneously several different BSW modes in the same 1DPC allows for self referencing sensing [10]. In analogy to surface plasmons on smooth thin films, a surface periodic patterning of 1DPC sustaining BSWs have been performed in order to fold BSW dispersion curves, open photonic band gaps in the dispersions, and further enhance the sensitivity of BSW upon surface perturbations. Dielectric micro and nanostructures can be used for biosensing based on either refractometric or absorption measurements. Such sub-wavelength structures include Bragg mirrors, gratings, microcavities, microrings and optical fibers and waveguides based on Photonic Crystals (PC). Provided that some of the photonic structures mentioned above produce strong field localization that can be exploited for field-enhancement effects, a large portion of their application consists in measuring shifts of the resonance conditions (light incident angle and/or wavelength) upon a perturbation of the refractive index of the external medium, containing the target biomolecules [11, 12]. Bragg mirrors consist of a periodic stack of dielectric materials acting like a lossless mirror for a given wavelength band. Periodic multilayers can be fabricated, for example, by sputtering, thermal evaporation, and Chemical Vapor Deposition (CVD) techniques. When one or more defect layers (spacers) are introduced therein, a defect (cavity) mode is obtained. In such a simple case, the cavity is a one-dimensional Fabry-Perot cavity that has been widely used for refractometric measurements. If the material is porous in nature, such as porous silicon eletrochemically etched, target molecules can penetrate inside the porous photonic structure (pores size for few nanometers to tens of nanometers), thus providing a much stronger perturbation and lowering the detection threshold. If fluorescent labels are used, onedimensional microcavities can also be used to improve the detection of fluorescence radiation, as described in the following. Photonic crystal surfaces can be designed to provide a wide range of functions that are used to perform biochemical and cell-based assays. Detection of the optical resonant reflections from photonic crystal surfaces enables high sensitivity label-free biosensing, while the enhanced electromagnetic fields that occur at resonant wavelengths can be used to enhance the detection sensitivity of any surface-based fluorescence
N
N
1682
assay. Fabrication of photonic crystals from inexpensive plastic materials over large surface areas enables them to be incorporated into standard formats that include microplates, microarrays, and microfluidic channels. The periodic modulation of refractive index within a PC, along with incorporation of intentional line or point “defects” in the PC can be used to concentrate and direct the electromagnetic fields associated with light to produce efficient wavelength selective reflectors, waveguides, optical circuits, beam steering devices, and optical multiplexors. PCs can be designed to interact strongly with particular optical wavelengths through selection of their materials and the period of their modulation. At the wavelengths of “optical resonance” (also known as “guided mode resonance”), light will couple strongly to the PC structure for a particular incident angle, resulting in electric fields inside the PC that can be many times higher than the electric field of incident radiation. These optical resonances can be observed by an external observer simply by illuminating the PC at normal incidence with a broad band of and observing a narrow band of wavelengths that are reflected back with nearly 100% efficiency. Two-dimensional photonic crystals used for sensing applications are typically microfabricated periodic nanostructures on a two-dimensional planar silicon on insulator (SOI) chip. Such structures can be designed to exhibit a two-dimensional photonic band gap (PBG) that can be tuned by varying the structural parameters. Photonic cavity states within this PBG are formed by systematically introducing defects into the perfect hexagonal lattice during the fabrication process. In the case of nanocavities, the modal volume is very small, of the order of V < 1mm3 and furthermore, the strongly localized electric field in the cavity is highly sensitive to local changes of the refractive index of the environment. The photonic crystal concept is fruitfully applied to optical fibers, in which the transverse cross section of the fiber is patterned with channels specifically arranged, wherein the refractive index contrast provided by the nanostructuration results in multiple effects such as optical field localization, multimodal guiding, etc. By combining the technology available for waveguide fabrication and the working principle of Fourier-Transform-based spectrometers, a suitable microdevice based on a high-density sample of the
Nanophotonic Structures for Biosensing
standing-wave interferogram produced inside a single waveguide has been obtained [13]. All the above mentioned techniques can be combined for obtaining devices and structures aimed at biosensing based on one or more of the described principles. For example, optical fibers presenting metal coatings or hosting metallic nanostructures can combine guiding and plasmonic effects, fiber grating can ease light coupling/decoupling, or resonance conditions, etc. Fluorescence-based detection As already mentioned, dielectric micro and nanostructures can provide a mean to improve the detection of labeled target molecules in close contact or impregnated within the photonic nanostructure. An example of such a structure is represented by onedimensional porous silicon microcavities [14], in which the fluorescence emitted by labeled molecules (e.g., proteins) is enhanced upon diffusion of said molecules inside the porous structures of the cavity [14]. Additional one-dimensional, two-dimensional, and three-dimensional (e.g., artificial opals) dielectric structures can be exploited for such enhancement effects, reaching very high Q factor of the order of 103 or more [15]. In recent years, there has been a growing interest in the interactions of fluorophores with metallic surfaces or particles. Near-field interactions are those occurring within a wavelength distance of an excited fluorophore. The spectral properties of fluorophores can be dramatically altered by near-field interactions with the electron clouds present in metals. These interactions modify the emission in ways not seen in classical fluorescence experiments and can be controlled by properly structuring the surface of plasmonic substrates [16]. While all metal–fluorophore interactions are based on the same physics [17], the effects can be different based on the geometry of the metal structure. Three possibilities are considered, in which a fluorophore is interacting with silver particles, a smooth metal surface and a metal surface with a regular pattern. Metal particles can be typically used to increase the fluorescence intensities. This increase occurs by a combination of enhanced fields around the metal and rapid and efficient plasmophore emission. These effects are usually called metal-enhanced fluorescence (MEF), and typically result in increased intensities and decreased lifetimes.
Nanophotonic Structures for Biosensing
In the case of a fluorophore interacting with a smooth metal film, it is typically about 40 nm thick silver or gold. In this case, the fluorophore creates plasmons which radiate at a defined angle into the substrate. Typically the intensities and lifetimes are not dramatically changed. This phenomenon is called surface plasmon-coupled emission (SPCE). Here, the emission spectrum is the same as the fluorophore but the polarization properties indicate that the plasmon is radiating. Finally, an exemplary fluorophore can be placed over regular patterns (nanorings, grooves, gratings, etc.). In this case, the emission at certain wavelengths is expected to show well-defined beaming into the substrate (or air), while other wavelengths are deflected from the normal. These effects are due to a combination of interactions with a smooth surface and with the sub-wavelength features (according to a diffraction phenomenon). This more general effect is called plasmon-controlled fluorescence (PCF). Plasmon-controlled fluorescence provides an opportunity to create ultra-bright probes based on complexes of metal particles with fluorophores. The basic idea is to create metal–fluorophore complexes in which the metal particle increases the brightness of the bound fluorophores. This can be accomplished by trapping the fluorophores in metal shells or by coating metal particles with fluorophores [18]. Theory has predicted that fluorophores in metal shells can be 100-fold brighter than the isolated fluorophore, even after consideration of the transfer efficiency of incident light into the shell and radiation out of the shell. Such probes may have the advantage over quantum dots because of the low toxicity of silver particles, even if the silver surface is completely exposed. Instead of organic fluorophores such as dyes, solid state semiconductor nanoparticles (quantum dots) can be used. The luminescent nanostructures are becoming more and more popular in labeling bio-assay [19].
Cross-References ▶ Active Plasmonic Devices ▶ Biosensors ▶ Chemical Vapor Deposition (CVD) ▶ Detection of Nanoparticulate Biomarkers in Blood ▶ Field Electron Emission from Nanomaterials ▶ Microfabricated Probe Technology ▶ Nanostructured Materials for Sensing
1683
N
▶ Nanostructures for Photonics ▶ Optical Properties of Metal Nanoparticles ▶ Optical Techniques for Nanostructure Characterization ▶ Polymer Coatings ▶ Scanning Near-Field Optical Microscopy ▶ Self-Assembled Monolayers
References 1. Fan, X., White, I.M., Shopova, S.I., Zhu, H., Suter, J.D., Sun, Y.: Sensitive optical biosensors for unlabeled targets: a review. Anal. Chim. Acta 620, 8–26 (2008) 2. Kneipp, K., Moskovits, M., Kneipp, H.: Surface Enhanced Raman Scattering. Springer, Heidelberg (2006) 3. Anker, J.N., Hall, W.P., Lyandres, O., Shah, N.C., Zhao Van Duyne, J.R.P.: Biosensing with plasmonic nanosensors. Nat. Mater. 7, 442–453 (2008) 4. Lal, S., Link, S., Halas, N.J.: Nano-optics from sensing to waveguiding. Nano Lett. 1, 641–648 (2007) 5. Steidtner, J., Pettinger, B.: Tip-enhanced Raman spectroscopy and microscopy on single dye molecules with 15 nm resolution. Phys. Rev. Lett. 100, 236101 (2008) 6. Alkire, R.C., Kolb, D.M., Lipkowski, J., Ross, P.N.: Tipenhanced Raman spectroscopy – recent developments and future prospects. In: Pettinger, B. (ed.) Advances in Electrochemical Science and Engineerigns: Diffraction and Spectroscopic Methods in Electrochemistry, vol. 9. Wiley, Weinheim (2008) 7. Osawa, M.: Surface-enhanced infrared absorption. In: Kawata, S. (ed.) Near-Field Optics and Surface Plasmon Polaritons. Springer, Heidelberg (2001) 8. Le Ru, E.C., Etchegoin, P.G.: Surface-Enhanced Raman Spectroscopy and Related Plasmonic Effects. Elsevier, Amsterdam (2009) 9. Homola, J.: Surface Plasmon Resonance Based Sensors. Springer, Berlin (2006) 10. Konopsky, V.N., Alieva, E.V.: A biosensor based on photonic crystal surface waves with an independent registration of the liquid refractive index. Biosens. Bioelectron. 25, 1212–1216 (2010) 11. Zourob, M., Lakhatakia, A.: Optical Guided-Wave Chemical and Biosensor II. Springer, Heidelberg (2010) 12. Arregui, F.J., Matias, I.R., Claus, R.O.: Fiber optic biosensors. In: Encyclopedia of Optical Engineering. Taylor & Francis, London (2004) 13. le Coarer, E., et al.: Wavelength-scale stationary-wave integrated Fourier-transform spectrometry. Nat. Photon. 1, 473– 478 (2007) 14. Sciacca, B., Frascella, F., Venturello, A., Rivolo, P., Descrovi, E., Giorgis, F., Geobaldo, F.: Doubly resonant porous silicon microcavities for enhanced detection of fluorescent organic molecules. Sens. Act. B 137, 467–470 (2009) 15. Dorfner, D., Zabel, T., H€ urlimann, T., Hauke, N., Frandsen, L., Rant, U., Abstreiter, G., Finley, J.: Photonic crystal nanostructures for optical biosensing applications. Biosens. Bioelectron. 24, 3688–3692 (2009)
N
N
1684
16. Lakowicz, J.R.: Plasmon-controlled fluorescence: a new paradigm in fluorescence spectroscopy. Analyst 133, 1308–1346 (2008) 17. Lakowicz, J.R.: Principles of Fluorescence Spectroscopy, 3rd edn. Springer, New York (2006) 18. Burns, A., Ow, H., Wiesner, U.: Fluorescent core–shell silica nanoparticles: towards “Lab on a Particle” architectures for nanobiotechnology. Chem. Soc. Rev. 35, 1028–1042 (2006) 19. Viste, P., Plain, J., Jaffiol, R., Vial, A., Adam, P.M., Royer, P.: Enhancement and quenching regimes in metal semiconductor hybrid optical nanosources. ACS Nano 4, 759–764 (2010)
Nano-piezoelectricity ▶ Piezoelectric Effect at Nanoscale
Nano-pillars ▶ Nano-sized Nanocrystalline and Nano-twinned Metals
Nanopolaritonics Daniel Neuhauser1, Christopher Arntsen1 and Kenneth A. Lopata2 1 Department of Chemistry and Biochemistry, UCLA, Los Angeles, CA, USA 2 William R. Wiley Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA, USA
Synonyms FDTD; Maxwell–Bloch simulations or equations or studies; Maxwell–Schro¨dinger simulations (or equations); Molecular plasmonics; Near-field molecular optics
Definition Molecular nanopolaritonics is the study of nanoscale control of near-field energy transport via combined plasmon–molecule excitations.
Nano-piezoelectricity
Introduction Absorption and emission of light (from near-IR to UV) on the nanoscale is qualitatively associated with one of two types of entities: excitons and plasmons. Excitons are associated with transition of electrons between occupied and unoccupied orbitals, i.e., a particle-hole excitation. Plasmons, on the other hand, are classicallike collective excitation of electrons in metals that are analogous to tidal waves. Specifically, when an electric field excites a plasmon, all the valence electrons in the metal shift, in tandem, creating a surface dipole; the remaining background exerts a restoring force and the resulting oscillation is in the IR to UV range, depending on geometry. The coupling of an exciton and a plasmon is known as a polariton. Polaritons are usually associated with surface excitons. They have been applied in various fields, and at low temperatures even exhibit Bose– Einstein condensates [1]. Here, the combination of excitons and plasmons on the nanoscale is discussed, where the plasmon is part of a relatively small metal structure and even more importantly, the exciton is associated with a molecule or a cluster. Such a combination is labeled as nanopolariton [2], and its properties and approaches for simulation are discussed here. The greatest emphasis here is on the manipulation of transport of nanopolaritons by varying the molecular properties, as detailed below. Broadly speaking plasmon–molecule interactions come in two forms: the influence of plasmonic fields on molecules, and the influence of molecules on plasmonic fields. The first, which is the typical focus of plasmon–molecule studies, is where plasmons focus light into molecules driving them into highly excited states, such as in surface-enhanced Raman spectroscopy (see SERS – Raman Spectroscopy). SERS relies on the fact that classically, an electric field near a metal surface is very strong, especially near the plasmon resonance. In SERS, the massive electric fields due to plasmons on a metal lead to huge enhancements in Raman scattering, allowing for the detection of single molecules [3]. Other examples of plasmons acting on molecules include plasmon-enhanced fluorescence, near-field optical microscopy, and enhancements in FRET [4]. The second direction, where single molecules have a significant effect on plasmons, is more subtle and less understood. Molecules have a much smaller transition
Nanopolaritonics
dipole moment than metal nanostructures (as they have few electrons, versus hundreds to million and more electrons in metal structures) and therefore molecules will generally have a negligible effect on overall light absorption of a polaritonic system, except possibly at very low temperatures. However there is one place where plasmonic interactions are weak: in the transport of plasmons from one structure to another. As this transport involves comparatively weaker coupling between plasmon modes, under certain circumstances it can be influenced by nearby molecules. While this article focuses primarily on molecule-mediated plasmon transport, linked DNA molecules have been shown to alter the dielectric properties of arrays of metal nanoparticles [5]. The length scales relevant for nanopolaritonics are significantly smaller than even UV wavelengths. At such scales, the electromagnetic fields are essentially “near fields,” i.e., the field due to a dipole is closer to its electrostatic form, so it is strong and decays rapidly with distance; the far-field-radiated part, proportional to the acceleration of the dipole, is much weaker over these scales (the ratio between the far and near fields for a distance scale R is approximately R2 =l2 , where l is the wavelength). The simulation of the near-field combined plasmon–molecule excitations (polaritons) involve a complicated interplay of distinct mechanisms, including quantum excitations, classical electrodynamics, and charge transfer, which involve a range of length and energy scales. This makes concise treatment of these systems very difficult: molecules on the order of a nanometer must be treated quantum mechanically, whereas plasmonic materials with sizes of up to hundreds of nanometers are usually too large to treat quantum mechanically in a computationally efficient way, and thus must be treated classically in most cases. Therefore, a primary thrust of theoretical molecular nanopolaritonics is and will be the development of methods which can effectively treat and provide an interface between these classical (metal) and quantum (molecular) regions.
1685
molecule. Some examples are described in greater detail below. Briefly, the simplest method evolves the spatial charge density of the metal in time and couples plasmons to a molecular density matrix via dipole interactions. A more sophisticated treatment models classical regions with the finite-difference timedomain (FDTD) method (see the entry in this volume on ▶ Finite-Difference Time-Domain Technique), and couples the classical electric field to the molecular density matrix again via dipole interactions. This particular technique has the advantage that the classical description can easily be interfaced with essentially any type of quantum mechanical treatment of a molecule, including time-dependent density functional theory (TDDFT), time-dependent Hartree– Fock (TDHF), or semiempirical methods, as described later. Plasmon Surface Modes Coupled to Electronic Density Matrices In the simplest treatment, appropriate for molecules near metallic structures, a set of modes is first used to represent the plasmons via surface modes on the metals [6]. The charge density on the metals is written as sðr; tÞ ¼
Overview There are several techniques used to model behavior in nanopolaritonic systems, but in general they involve coupling classical fields to a quantum mechanical
X
CJ ðtÞsJ ðrÞ;
(1)
J
where sJ ðrÞ is a surface-localized charge mode on a metal particle (e.g., dipole, quadrupole, etc., for a spherical metal particle, or a more general function in other cases), and the time-dependent coefficients which determine the evolution of the surface modes were introduced. The dynamics of these surface modes is associated with plasmons; the collective oscillation of the charges causes a time-dependent surface charge. The surface charges are then coupled via dipole interaction to specific molecules; the resulting equation for the motion of the modes, including interaction with other modes and with specific molecules is then [2]: X
Nanopolaritonics: Methodology
N
MJK C€K ðtÞ ¼
K
X
X
VJK CK ðtÞ
K
rij ðtÞqij;J vJ ðtÞ:
X K
MJK
C_ K ðtÞ tK
(2)
ij
The first three terms are purely plasmon related, including mass terms ðMJK Þ associated with the kinetic
N
N
1686
Nanopolaritonics
energy of the surface modes, potential terms ðVJK Þ associated with coulomb interactions between the surface modes, and Drude damping terms (1=tK ) responsible for the dephasing and damping of the surface modes. The next term contains a dipole integral qij;J coupling the plasmon modes to the molecule; the molecular density is represented as a density matrix with a constant (pre-interaction) part, r0ij , and a timedependent part induced by interaction with the plasmons, rij ðtÞ. Finally, the last term is the interaction with the external field, vJ ðtÞ. The other relevant equation is the time-dependence of the density matrix for the molecules. For each molecule, schematically, apply the Liouville equation:
i
drij ðtÞ ¼ ½FðrÞ; r ij igij rij ðtÞ r0ij dt " # X þ CJ ðtÞqJ ; r ; J
(3)
In the generalized Liouville coupling matrix, L is straightforwardly shown to be 0
0 B V B L¼@ 0 q
M1 t1 0 0
0 q g D Z2B
1 0 0 C C: D A g
The specific terms in the Liouville matrix result from a linear-response analysis of Eqs. 2 and 3. Maxwell–Schro¨dinger Simulations The plasmonic-mode model is appropriate for qualitative studies, especially because of its very modest computational requirements. However, a quantitative investigation requires a more realistic treatment of the electric field distribution. The best-known approach for simulating the electromagnetic fields is FDTD, where one propagates the Maxwell equations [7]. Assuming for simplicity no magnetic susceptibility effects, these read:
ij
where the following variables are introduced: the Fock matrix (FðrÞ), which depends itself on the density matrix; the damping coefficients (gij ) which determine the return of r to the equilibrium density matrix, r0 ; and the coupling to the classical fields of the plasmons via the dipole coefficients qij;J . Details of the different possibilities for a Fock operator are presented later. The simplest way to view Eqs. 2–3 is when the excitation is not very large, as then they can be linearized in the density matrix. Then, the linearized equations have a simple form: df i ¼ Lf; dt
(4)
@Eðr; tÞ 1 ðH Hðr; tÞ Jðr; tÞÞ ¼ @t eeff ðrÞ (6) @Hðr; tÞ 1 ¼ H Eðr; tÞ; @t m0 where the electric and magnetic fields and the frequency-independent part of the susceptibility are introduced. The current, J, is made from two parts, J ¼ Jp þ Jm ; the first term is the usual plasmonic contribution, which in FDTD is simulated as a sum of terms, Jp ðr; tÞ ¼
X
Jp;n ðr; tÞ;
(7)
n
where the individual terms fulfill where f is the nanopolariton, i.e., the combined vector describing both the plasmon modes and the density matrix. Schematically, f is made from four parts: the surface charge coefficients and their time derivatives, as well as the real and imaginary part of the excitation of the density matrix away from the ground state:
dC 0 0 f ¼ CJ ; ; Re r r ; Im r r : dt T
(5)
dJp;n ðr; tÞ ¼ an ðrÞJp;n ðr; tÞ þ bn ðrÞEðr; tÞ: dt
(8)
The sum typically includes 1–5 terms, and the individual terms (an ðrÞ and bn ðrÞ) in the sum are usually fitted to give the correct dielectric function in the frequency range of interest. Other, somewhat more sophisticated time evolution techniques for Eq. 8 have also been suggested.
Nanopolaritonics
1687
The second part of the current is the molecular contribution, which for a single molecule will be d TrðrxÞ dt ¼ idðr rm Þ Trð½F; r xÞ
JM ðr; tÞ ¼ dðr rM Þ
(9)
where x is the electronic 3-D coordinate, and the delta function in the position of the molecule is represented numerically on the FDTD grid. The molecular current is obtained from a time-dependent study of the density matrix (or alternatively from propagating the molecular orbitals). In the simplest version, where only a dipole electric field is assumed, these take a form similar to Eq. 3 but now the coupling is directly due to the electric field: i
drij ðtÞ ¼ ½Ftot ðrÞ; r ij igij rij ðtÞ r0ij ; dt
(10)
where Ftot ðrÞ ¼ FðrÞ m E:
(11)
Molecular Simulation Methodologies Throughout, the molecules are simulated in real time. For the present purposes, the molecules will be treated by a Liouville equation or a time-dependent orbital equation, where the initially occupied orbitals are propagated as: i
@cn ¼ FðrÞcn : @t
(12)
There are different possible Fock operators; the most established are TDDFT and TDHF. Both methods, however, could be numerically expensive, so a simple alternate approach is to propagate in time a semiempirical approach; an example is TD-PM3 (time-dependent Parameter-Method 3) where the Fock operator has the generic TDHF form: Fij ¼ H0;ij þ
X
Vijkl rkl ;
(13)
kl
but with a very sparse set of matrix elements, Vijkl . TDPM3 generally gives excitation energies with a 15% accuracy [8], so is very suitable for the
N
qualitative questions which arise in the study of nanopolaritonics. As the method uses a minimal basis set and scales well with system size, it is ideal for simulating large dye molecules. Splitting Fields It turns out that when there is one molecule, the delta function associated with having a localized charge is problematic and leads to numerical singularities. The reason is that this strong delta function source due to the molecule acts back on the molecule, creating an artificial self-energy problem. More precisely, for a single electron this is a self-energy problem as the electron interacts with itself; for a general molecule, the interaction with the molecule-induced current induces a double-counting problem, since the Fock operator already includes the electron–electron interaction and the interaction with the molecular current has essentially the same role. This difficulty was alleviated through the development of a new formalism, in which the electric field is divided into two parts, a molecular and a plasmonic part [9]. The molecule is only affected by the external part, while the total field acts on the plasmon. Specifically, for the electric field the resulting equations are:
@Ep 1 1 1 ¼ HHþ H Hm eeff eeff e0 @t
Jp ; eeff
@Em 1 ¼ ðH Hm Jm Þ; e0 @t
(14) (15)
while the magnetic fields equations are simpler @Hp 1 ¼ H Ep ; m0 @t
(16)
@Hm 1 ¼ H Em : m0 @t
(17)
As mentioned, only the electric field that acts on the molecule enters into the evolution equation for the density matrix. Only the plasmonic field affects now the molecule, so that the Fock operator for the evolution of the density matrix is now: Ftot ðrÞ ¼ FðrÞ m Ep :
(18)
N
N
1688
Nanopolaritonics
The advantage in this approach is that the decoupling prevents numerical singularities. The disadvantage is that a separate electric field is required for each molecule. Therefore, future approaches will mix the FDTD treatment of the plasmonic field with a dipole and quadrupole treatment of the molecules, i.e., write the total electric field as Eðr; tÞ ¼ Ep ðr; tÞ þ
X j
þ Quad:;
mðr; t0 ÞH
1 jr rj j
Array of spherical gold nanoparticles with nearby molecule
1
5 nm
(19)
where “Quad.” refers to quadrupolar interaction with nearest neighbors, and the molecular dipole is introduced. The magnetic field is written similarly in terms of the molecular current and the plasmonic field, and the equation of evolution of the total field can be derived directly from Eq. 14. This mixed field-dipole approach will allow the treatment of a host of molecules interacting with an external field. In the extreme limit that a continuum of molecules are required, the FDTD equations will be solved directly; then, the effect of each molecule will be relatively small so that the separation between plasmonic and molecular fields will not be needed, and the resulting, single electric field equations are the familiar Maxwell–Bloch equations, i.e., there will be no need to split the electric fields.
Key Findings The fundamental conclusion from plasmon mode and Maxwell–Schro¨dinger (Sect. Nanopolaritonics: Methodology) simulations is that under certain circumstances single molecules have a marked effect on the near-field transport of plasmons, as detailed below. Molecule-Mediated Plasmon Transfer in Waveguides An illustrative example of the type of system studied is shown in Fig. 1, where a single two-level dipolar molecule is located near a linear array of gold nanoparticles. In the absence of a molecule, an initial x-polarized plasmon excitation on the leftmost (input) nanoparticle will propagate down the array, which after a few femtoseconds results in an x-polarized plasmon on the rightmost (output) nanoparticle. Since
2
3
Molecule
Nanopolaritonics, Fig. 1 Schematic of the computed setup. Three small metal spheres are placed close to each other. The leftmost sphere is excited; the middle sphere mediates this transfer. The molecule is placed between the middle and the final, rightmost sphere, and mediates this transport. The direction of the molecule influences the final polarization of the molecule
the x, y, and z plasmon modes are decoupled in this linear configuration, no y or z excitation is induced on the output nanoparticle. Noble metal plasmonic waveguides of this type have been extensively studied both theoretically and experimentally [10]. Perhaps surprisingly, the addition of a single molecule to the waveguide can have a drastic effect. Take, as a simple example, an xy-oriented two-level molecule with a strong resonance at om ¼ 2.55 eV, which is close to the plasmon frequency of the array (2.60 eV). Figures 2 and 3 show results of a FDTD/linearresponse TDDFT simulation. Figure 2 shows the clear transmission effect of a molecule in a timedomain simulation; the presence of the molecule causes significant enhancement in the amount of transmitted light in the nonnative initial polarization. The frequency-domain picture is even more revealing. Figure 3 shows that the Fourier transform of the total current on the output nanoparticle displays a Fano-like resonance. As a reminder, Fano resonance are generally the result of coupling of single (or a few, in the generalized case) “dark” quantum level(s) to a continuum of states. Their most obvious mark is a sine-like modification of the observable (spectrum or transmission profile) which in the original Fano resonance case will actually lead to a dip of zero absorption due to destructive interference caused by the presence of the quantum level. In the present simulation case, a clear Fano-like transmission profile results due to the presence of the molecule; the excitation of the x-polarization is drastically modified. Specifically, the molecule enhances the x-polarized plasmon transmission over a range of
Nanopolaritonics
1689 9.6 fs
9.9 fs
10.2 fs 8 ⫻ 10−4 4 ⫻ 10−4 0 −4 ⫻ 10−4 −8 ⫻ 10−4
No molecule
With molecule
5 nm
Nanopolaritonics, Fig. 2 Three snapshots in time of the ycomponent of the electric field with (bottom) and without (top) a molecule, for an excitation starting on the left sphere. The
dipolar molecular field serves to scatter x-oriented energy (not shown) into y-oriented energy (From Ref. [9])
Current enhancement and scattering
Frequency resolved current on third nanoparticle
0.006
400 350 300
0.004
|J{x, y} (ω)| (a.u.)
tot |J{x, y} (ω)| (a.u.)
0.005
tot Jx
0.003
ω˜ m
0.002 0.001 0
N
2.2
2.4
2.6
200 150 100
J ytot 2
250
2.8
3
Frequency (eV / h ¯)
50 0 2
Nanopolaritonics, Fig. 3 Fourier transform of the currents on the output nanoparticle of a linear waveguide, showing that a single two-level xy-oriented molecule rotates incident plasmon polarization from x ! y (From Ref. [9])
frequencies below om and decreases the transport for a range above. Further, the molecule rotates the excitation from the x-polarization to the y-polarization, resulting in a significant y-current (approximately 20% of the x-current). Physically, the molecule, which has a transition dipole moment aligned along the xy axis, scatters x-polarized near-field light and reemits y-polarized light. Despite the fact that the molecule does not affect the overall absorption of the array, the intense molecule– plasmon resonance in the gap between nanoparticles strongly influences plasmon transport, to the point of significantly scattering plasmons between otherwise orthogonal modes of excitation. Realistic Simulations To study this problem in a more rigorous manner, i.e., by using a more realistic quantum-chemistry
3.2
3.4 3.6 Frequency (eV/h¯ )
3.8
4
Nanopolaritonics, Fig. 4 Frequency-resolved current on the third silver nanoparticle of the system shown in Fig. 1 with 2-nm nanoparticles with (bold lines) and without (thin line) a molecule in the x-direction (solid) and y-direction (dashed) (From Ref. [11])
treatment, the FDTD approach has also been combined with a time-dependent semiempirical method, TDPM3, mentioned above, to show rotation of current for large molecules with a significant number of possible excitations [11]. The system examined used silver nanoparticles, and thus tartrazine, a yellow dye molecule having a strong transition dipole moment near the plasmon resonance of silver, was chosen. The realistic simulations verify that molecules with an excitation energy near the plasmon resonance of the metal scatter the x-oriented current into the y-direction, as observed with the two-level molecule. The results are shown in Fig. 4. These results are encouraging as they confirm that even a molecule with a broad range of excitations can influence the transport of plasmons.
N
1690
These studies offer insight into near-field light transport in heterogeneous nanosystems. Single molecules, although too weak to play a role in overall light absorption, can have drastic effects on transport if strategically placed in “hot spots,” where they can modulate the comparatively lower energy coupling between nearby metal structures. Since this is a resonance effect, it is highly dependent on the materials involved. The molecule, in order to scatter current, must have a strong optical transition at or near the coupling (i.e., plasmon resonance) of the waveguide. On one hand, this limits the diversity of the systems that can be used; however, this specificity is actually advantageous for sensing applications and tunable frequency-dependent wave guiding. Controlling Plasmon Transport with Single Molecules The strong effect of a single molecule on nanopolariton transport can be exploited to selectively control plasmon transport on the nanoscale. This has implications in the field of plasmonics, where molecule–plasmon interactions have promise as device components such as switches or higher harmonic generators. A molecule placed near a fork-like arrangement of gold nanoparticles, e.g., can selectively divert incoming plasmon waves into either the top or bottom paths, as demonstrated in Fig. 5, which shows transmission through the fork calculated using the plasmon mode approach. Beyond being sensitive to the molecule’s location, this effect is also highly dependent on its orientation: When the molecule points upward, the plasmons are directed into the upper branch of the fork (blue curve in Fig. 5) much more than to the lower branch (red curve). In the absence of a molecule (black curve), the transition probabilities of the two branches are equal. The controllability can be extended to use in molecular switches: Molecules can potentially be used to gate the transfer of near-fields in a specific and deliberate manner. Likewise, a similar configuration can act as a type of highly specific molecular sensor, sensitive to a molecule’s identity, orientation, and position. Limitations There are several important caveats to these findings. First, compared to metals, molecules have a much smaller capacity to absorb and radiate light, as they are unable to further absorb light (of a given frequency) once already in their excited state. FDTD/
Nanopolaritonics
a 1.1
b x -Transmission
N
c
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 2.2
tM = ∞
tM = 24 fs
T
I 5 nm
B
tM = 12 fs
2.3
2.4 2.5 w (eV)
2.6
2.7
Nanopolaritonics, Fig. 5 Simplified-model study of control of radiation transfer through a fork by placing a molecule aligned parallel to the top branch of the fork. Blue top branch, red bottom, black either branch without a molecule. The frames differ by the amount of dephasing on the excited state of the molecule. The gold dots have the experimentally fitted dephasing time, 6 fs (From Ref. [2])
TDDFT simulations of nanopolaritonic systems which treat the molecule beyond linear response have shown that under increasing field intensities the polarization rotation effects saturate rapidly; this transition region is especially interesting, since it can be used as a nonlinear gating mechanism.
Future Directions Nanopolaritonics is still in its infancy. Although a variety of simulations have demonstrated proof of
Nanopolaritonics
concept for controlling near-field light with molecules, there is much untapped potential for novel plasmon– molecule applications, as well as many outstanding theoretical and experimental questions. This last section briefly overviews future directions for both experiments and theory. Transport-Based Sensing Among the most promising potential applications is transport-based sensing, where extremely dilute concentrations of molecules are detected and uniquely identified based on how they alter the plasmon transport properties of a nanostructure system. In contrast to a SERS-like device, which is used to enhance and probe the response of a molecule, here the plasmon transport is measured, which is predicted to be extremely sensitive to the presence of select molecules. A pair of nanoscale metal tips, e.g., can act as a sensor for molecules which drift between them. Alternatively, one can build a parallel detector using a chip containing a series of different plasmonic junctions, each with a slightly different plasmon resonance, which are tuned to different target analyte molecules and excited by different frequencies of light. Plasmonic Metamaterials Another potential application of nanopolaritonics is in the development of more sophisticated metamaterials. Beyond simply mediating wave guiding, molecules could also mediate coupling between elements of a plasmonic metamaterial. This could lead to, e.g., tunable metamaterials where the presence of different molecules shifts the resonant frequency of the device. Need for Experimental Studies There is a pressing need for high-quality experimental measurements of plasmon transport in the presence of strongly interacting molecules. Whereas there have been a wealth of experiments concerning the effects of intense fields on single molecules, studies on the effects of molecules on plasmons are lagging. Improved Multiscale Simulation Methodologies From a theoretical standpoint, progress needs to be made on efficiently and accurately capturing plasmon–molecule interactions, especially in the “chemical” regime where molecules are bound to the surface of a metal and share electrons. In reality both charge and energy are shared on a surface, and
1691
N
developing real-time embedding approaches capable of describing this is necessary to drive the field forward. This is a daunting task as it necessitates treating the quantum mechanical and classical regions as open systems. Moreover, seamless integration of the two regions requires careful choice of either boundary conditions or embedding potentials.
Cross-References ▶ Finite-Difference Time-Domain Technique ▶ Nanoparticles
References 1. Kasprzak, J., Richard, M., Kundermann, S., Baas, A., Jeambrun, P., Keeling, J.M.J., Marchetti, F.M., Szymanska, M.H., Andre, R., Staehli, J.L., Savona, V., Littlewood, P.B., Deveaud, B., Dang, L.S.: Bose-Einstein condensation of exciton polaritons. Nature 443, 409–414 (2006) 2. Neuhauser, D., Lopata, K.: Molecular nanopolaritonics: Cross manipulation of near-field plasmons and molecules. I. Theory and application to junction control. J. Chem. Phys. 127, 154715 (2007) 3. Nie, S., Emory, S.R.: Probing single molecules and single nanoparticles by surface-enhanced Raman scattering. Science 275, 1102–1106 (1997) 4. Govorov, A.O., Lee, J., Kotov, N.A.: Theory of plasmonenhanced Fo¨rster energy transfer in optically excited semiconductor and metal nanoparticles. Phys. Rev. B. 76, 125308 (2007) 5. Lazarides, A.A., Schatz, G.C.: DNA-linked metal nanosphere materials: Fourier-transform solutions for the optical response. J. Chem. Phys. 112, 2987–2993 (2000) 6. Prodan, E., Nordlander, P.: Plasmon hybridization in spherical nanoparticles. J. Chem. Phys. 120, 5444–5454 (2004) 7. Gray, S.K., Kupka, T.: Propagation of light in metallic nanowire arrays: Finite-difference time-domain studies of silver cylinders. Phys. Rev. B. 68, 045415 (2003) 8. Bartell, L.A., Wall, M.R., Neuhauser, D.: A time-dependent semiempirical approach to determining excited states. J. Chem. Phys. 132, 234106 (2010) 9. Lopata, K., Neuhauser, D.: Multiscale Maxwell– Schro¨dinger modeling: A split field finite-difference timedomain approach to molecular nanopolaritonics. J. Chem. Phys. 130, 104707 (2009) 10. Maier, S.A., Atwater, H.A.: Plasmonics: Localization and guiding of electromagnetic energy in metal/dielectric structures. J. Appl. Phys. 98, 011101 (2005) 11. Arntsen, C., Lopata, K., Bartell, L.A., Wall, M.R., Neuhauser, D.: Modeling molecular effects on plasmonic transport: Silver nanoparticles with tartrazine. J. Chem. Phys. 134, 084101 (2011)
N
N
1692
Nanopores
Definition
Nanopores ▶ Nanochannels for Nanofluidics: Fabrication Aspects
Nanopowders ▶ Nanoparticles
Nanoribbons ▶ Graphene
Nanorobotic Assembly ▶ Robot-Based Automation on the Nanoscale
Nanorobotic Manipulation of Biological Cells ▶ Robot-Based Automation on the Nanoscale
Nanorobotic Spot Welding Lixin Dong1, Xinyong Tao2, Zheng Fan1, Li Zhang3, Xiaobin Zhang4 and Bradley J. Nelson5 1 Electrical and Computer Engineering, Michigan State University, East Lansing, MI, USA 2 College of Chemical Engineering and Materials Science, Zhejiang University of Technology, Hangzhou, China 3 ETH Zurich, Institute of Robotics and Intelligent Systems, Zurich, Switzerland 4 Department of Materials Science and Engineering, Zhejiang University, Hangzhou, China 5 Institute of Robotics and Intelligent Systems, ETH Zurich, Zurich, Switzerland
Synonyms Nanobonding; Nanointerconnection; Nanosoldering
Nanojoining;
Spot welding, a widely used resistance welding process, was originally developed by Elihu Thompson in the early 1900s [1]. The most common application of spot welding is in the automobile industry, where it is widely used and most often performed automatically by preprogrammed industrial robots on assembly lines. With the continuing development of bottom-up nanotechnology fabrication processes, the nanoscale version of robotic spot welding, or nanorobotic spot welding [2], may likewise play an important role in interconnecting carbon nanotubes (CNTs), nanowires, nanobelts, nanohelixes, and other nanomaterials and structures for the assembly of nanocircuits and nanoelectromechanical systems (NEMS).
Overview Van der Waals forces [3], electron-beam-induced deposition (EBID) [4], focused-ion-beam chemical vapor deposition (FIB-CVD), high-intensity electronbeam welding [5], Joule-heating-induced joining [6], and nanomechanochemical bonding [7] are experimentally demonstrated interconnection strategies, although all have limitations. Van der Waals forces are generally very weak, Joule-heating-induced joining and nanomechanochemical bonding are promising but not yet mature, and the other methods involve high-energy electron or ion beams, which significantly limits their applications. Another interconnection approach is to use CNTs, with their hollow cores and large aspect ratios, as possible conduits for nanoscale amounts of various materials that can be used to fuse CNTs together. The concept of nanorobotic spot welding is schematically shown in Fig. 1. A metal-filled carbon nanotube is positioned with a nanorobotic manipulator to the interconnection site and the metal is then deposited to solder the nano-building blocks, such as CNTs and nanowires, together, or to weld them onto electrodes. To facilitate the deposition, the encapsulated metallic materials may need to be melted or evaporated, which means only materials with a lower melting point than that of CNTs can be applied.
Nanorobotic Spot Welding
1693
N
Nanorobotic Spot Welding, Fig. 1 Concept of nanorobotic spot welding
Nanorobotic Spot Welding, Table 1 Melting point of pure metals and alloy solders Materials W CNTs in vacuum Fe Y Cu Au Ag CNTs in air Pb Bi Sn
Melting point ( C)a 3,410 2,800 [8] 1,535 1,523 1,083 1,064.43 961.93 750 [8] 327.5 271.3 231.9
Sn50Pb50 Ga
183–212 29.78
Possibility to be filled into CNTs
N Produced by pyrolysis of ferrocene powder [9] Yttrium carbide synthesized from yttrium oxide powder mixed with isopropanol [10] Synthesized using an alkali doped Cu catalyst by a thermal CVD method [11] Capillary filling [12] Capillary filling [13] Capillary filling [14] Capillary filling Synthesized by catalytic deposition of acetylene using nanocrystalline SnO2 as a catalyst [11] Synthesized in a vertical radiofrequency furnace from a homogeneous mixture of Ga2O3 and pure, amorphous, active carbon under a flow of pure N2 gas [15]
a
Data source for melting points not specified: http://www.chemicalelements.com/show/meltingpoint.html
Fortunately, most pure metals have a lower melting point than that of CNTs in vacuum and a large portion of them have a lower melting point than that of CNTs in air (Table 1). Alloy solders routinely used in the larger scale for electronics and silversmithing may also serve as excellent candidates of nanosolders and they
typically have lower melting points than the pure metals. These include soft solders (melting range: 90–450 C, typically tin-lead (Sn-Pb) alloys), leadfree solders (melting point 5–20 C higher than soft solders, alloys contain tin, copper (Cu), silver (Ag), bismuth (Bi), indium (In), zinc (Zn), antimony (Sb),
N
1694
Nanorobotic Spot Welding
Nanorobotic Spot Welding, Fig. 2 (a–d) Copper-filled CNTs. (a) FESEM images of Cu-filled CNTs. Observation shows that all the CNTs have sharp tips filled with metal nanoneedles. These CNTs are up to 5 mm long with outer diameters in a range of 40–80 nm. (b) A typical copper-filled CNT synthesized for 30 min. (c) HRTEM image reveals that the Cu nanoneedles are encapsulated in graphite walls approximately 4–6 nm thick. The inset is the corresponding SAED pattern of the Cu nanoneedle along the zone axis, showing that
the Cu nanoneedle is single crystalline. The appearance of a pair of arcs in the SAED pattern indicates some orientation of the (002) planes in the carbon tubes. (d) The magnified image of the rectangular region in (c). The large arrows indicate the growth direction of the CNTs. The interlayer spacing of carbon nanotube is about 0.34 nm, consistent with the (002) plane lattice parameter of graphite. It can also be seen that the graphite layers are not parallel to the tube axis
and traces of other metals), and hard solders (melting point >450 C, alloys of copper with either zinc or silver). Figure 2 shows CNT samples synthesized using an alkali-doped Cu catalyst by a thermal CVD method [16]. Figure 2a–d show their structures imaged by field emission scanning electron microscopy (FESEM, Sirion, FEI), transmission electron microscope (TEM, JEM-2010, 200 kV), selected area electron diffraction (SAED), and high-resolution TEM (HRTEM). It can be seen from these observations that the yield of the Cu-filled CNTs was high. The CNTs were up to 5 mm long with outer diameters in a range of 40–80 nm. The single-crystalline Cu nanoneedles were encapsulated in graphite walls approximately 4–6 nm thick in the first sections of the bamboo-structured CNTs. The graphite layers were not parallel to the tube axis.
Key Research Findings An experimental investigation of controlled melting and flowing of single crystalline copper from CNTs has been performed using nanorobotic manipulation [3], and its application in spot welding of nanotubes has been demonstrated using this copper. Because copper is a good conductor of heat and electricity and has a very low binding energy (0.1–0.144 eV/atom) when bound to carbon, copper-filled CNTs were used in the investigation. The experiments were performed in a CM30 TEM equipped with a scanning tunneling microscope (STM) built in a TEM holder (Nanofactory Instruments AB, ST-1000) serving as a manipulator, as shown in Fig. 3a or schematically in Fig. 3b. The material consisting of a CNT bundle was attached to a 0.35 mm thick Au wire
Nanorobotic Spot Welding
a
STM Unit Tip Probe
Piezotube
Ball
1695
Wire Specimen Holder
Spring Leg
Tip Holder Hat
b Cu-tip Sample holder
Probe First Bamboo Section V
I
Nanorobotic Spot Welding, Fig. 3 (a, b) Nanorobotic manipulation system in a TEM
using silver paint, and the wire was held in the specimen holder. The probe was an etched 10 mm thick tungsten wire with a tip radius of approximately 100 nm (Picoprobe, T-4-10-1 mm). The probe can be positioned in a millimeter-scale workspace with subnanometer resolution with the STM unit actuated by a three-degree-of-freedom piezo-tube, making it possible to select a specific CNT. Physical contact can be made between the probe and the tip of a nanotube. Applying a voltage between the probe and the sample holder establishes an electrical circuit through a CNT and injects thermal energy into the system via Joule heating. By increasing the applied voltage, the local temperature can be increased past the melting point of the copper encapsulated in a tube. The process was recorded by TEM images and real-time video. In the experiment, the tip of a Cu-filled CNT was first brought into contact with the tungsten probe. Then a bias voltage was applied on the two ends of the CNT with the tungsten probe serving as the anode. The voltage was slowly increased from 0 mV with 100 mV steps. When the voltage reached 1,500 mV, a vacant section inside the carbon shells appeared, indicating that the copper has begun melting. Transportation of the copper to the probe-tip contact moved the vacant section to the root of the first bamboo section of the Cu-filled CNT. After the vacant section reached the root, the copper core started to flow to the tip of the CNT when bias was increased to 2,500 mV.
N
Figure 4a is a series of time-resolved TEM images taken from video frames showing the flowing process. The copper core started to flow inside the carbon shell from the bottom to the tip of the first bamboo section as the bias voltage reached 2.5 V. The entire process continued for about 70 s. The flow rate was found to be 11.6 nm/s according to the change of apparent length of the copper core (Fig. 4b). Accordingly, the mass change was calculated as shown in Fig. 4c, and the mass flow rate can then be determined by fitting the data to the curve 3 109 t2 1:2 103 t þ 0:12, yielding approximately 120 ag/s, which is strikingly slow and well controllable, allowing precise delivery of mass at the attogram scale for time-based control that can readily reach sub-second precision. Figure 4d shows time-resolved current versus voltage characteristics under a constant positive bias of 2.5 V. The current density under 2.5 V when flowing occurred was then calculated according to the cross-sectional area as 2.60 3.07 106 A/cm2. This is comparable to the observed value for electromigration of iron in CNTs (ca. 7 106 A/cm2) [17]. The difference can be a result of the lower binding energy of copper to the carbon shells (0.1–0.144 eV/atom) than that of irons to carbon shells (0.3 eV/atom). The high current densities employed here will lead to resistive heating. Temperatures as high as 2,000–3,000 C have been estimated according to the lattice spacing change in electric breakdown experiments on non-filled multiwalled nanotubes (MWNTs) [18] at a slightly higher bias (3 V) than those used here. The current density J and the mass flow rate m_ were then correlated as shown in e. The relation m_ ¼ 0:3135 J 2 1:6206 J þ 2:1373 sug_ gests that a real positive value of m(42.9 ag/s) can only be given when the current density J surpasses 2.5847 106 A/cm2. The existence of this threshold also implies that the mechanism of the observed flowing is most possibly electromigration [17]. Under a negative bias, that is, when the tungsten probe served as a cathode, the flow was observed in the opposite direction. Other possible mechanisms for flow can be excluded. Capillary force can induce filling/flowing, but the direction should be opposite to the observed flow, that is, from the tip to the bottom of the carbon shells. Thermal expansion can enable flow, but the flow should be isotropic, heading toward both the tip and the bottom. A recent investigation showed that the
N
N
1696
a
Nanorobotic Spot Welding
0s
10s
20s
30s
40s
50s
70s
60s
100nm
c
1000 800
Cu Mass Flow [fg]
Apparent Length of Copper Core [nm]
b
L = −11.6t + 822.9
600 400 200 0
0
20
40
60
6 5 4 3 2
0
80
m = 1E-09t3 − 0.0006t2 + 0.1249t − 0.0613
1 0
20
Time [s]
Mass Flow Rate dm/dt [ag/s]
Current [μA]
e 42 36 30 2.4
2.45
2.5
0s 20s 31s 54s 59s 64s −2
−1
0
1
60
80
3
3.2
Time [s]
d 50 40 30 20 10 0 −10 −20 −30 −40 −50 −3
40
2
3
Voltage [V]
150
100
50 dm/dt = 0.3J2 − 1.6J2 − 1.6
0 2.4
2.6
2.8
Current Density J [x106 A/cm2]
Nanorobotic Spot Welding, Fig. 4 (a) Time-resolved TEM images from video frames showing the flowing process. The copper core started to flow inside the carbon shells from the root to the tip as the bias voltage reached up to 2.5 V. The whole process was continued for about 70 s. (b) The flow rate has been found to be 11.6 nm/s according to the change of apparent length of the copper core. The tungsten probe has been positively
biased. (c) The mass changes along the time. The mass flow rate can be then drawn out from the fitting curve as approximately 0.12 fg/s. (d) Time-resolved current versus voltage characteristics under a constant bias of 2.5 V. The current density under 2.5 V as flowing occurred is about 2.60 3.07 106 A/ cm2. (e) Correlation of the current density J and the mass flow rate dm/dt
irradiation of MWNTs can cause a large pressure buildup within the nanotube core that can plastically deform, extrude, and break encapsulated solid material [19]. In the experiments, however, no contraction of the carbon shells was observed. Although there is no way of measuring the actual temperature of the nanotubes or the copper core, this process provides the possibility for estimating the temperature according to the resistivity change if
Matthissen’s rule is applied on temperature dependence of resistivity, provided that the resistivity was known at a certain temperature. Unfortunately, the latter is unknown. However, it has been noted that the temperature for CNT growth from 100-nm scale Cu particles in the synthesis (700 C) is far lower than the melting point of bulk Cu (1,083 C) [16]. It is well known that the surface-to-volume ratio with respect to a nanosized particle can affect the melting point.
Nanorobotic Spot Welding
1697
N
Nanorobotic Spot Welding, Fig. 5 Self-soldering of CNTs with copper encapsulated in a CNT. (a) A copper-filled tube, CNT1, was attached to a tungsten probe. (b) A section of CNT1 was soldered onto CNT2 by the melted copper. (c–h) Video frames showing the soldering process. A copperfilled tube, CNT1, was first attached to a probe, and brought into contact with another tube, CNT2. The probe (cathode) has a 10 V bias. (c–e) Three different positions as the probe is approaching CNT2. (f) Showing how contact has been made. The shape change of the copper suggested a melting process has occurred. With a higher bias (15 V), CNT1 was broken (g) and its end section remains soldered to the tip of CNT2 (h)
N
Although the exact reason for melting at such a low temperature is not yet known, a rough estimation of the melting point of the copper core can be made, that is, around 700 C. This is comparable to the in situ electron microscope observations of the melting point of the encapsulated 20–60 nm diameter Cu nanocrystals in multilayer graphitic carbon spheres, which was reported at 802 C [20]. It is also known according to the experiment that the melting point of the carbon shells is much higher than that of the copper core. This provides further evidence of molten copper inside carbon shells.
Examples of Application The application of such controlled transportation of the copper core was then investigated. Self-soldering of CNTs with copper encapsulated in a CNT is shown in Fig. 5. A copper-filled tube, CNT1, was first attached to a tungsten probe (Fig. 5a). A section of CNT1 was then attached to and soldered onto CNT2 by the melted copper (Fig. 5b). Figure 5c–h are video frames showing the soldering process. A copper-filled tube, CNT1, was attached to a probe, and brought into contact with another tube, CNT2. The probe has a 10 V bias. Parts
N
1698
Nanorobotic Spot Welding
Nanorobotic Spot Welding, Fig. 6 Related investigations and applications
ce of Fig. 5 show three different positions as the probe was approaching CNT2, and Fig. 5f shows how the contact has been made. The shape change of the copper suggests a melting process has occurred. It has been found from the video frames (25 fps) that the melting of the copper happened in a very short interval ( directions, that is, the smallest Young modulus direction on a < 001 > substrate, in which the bilayer will start to roll up when it has been freed from the substrate. Several techniques have been proposed to improve the control over the fabrication process in terms of the resulting length, shape, and orientation of structures. It has been demonstrated that the misaligned angle between the stripe and etching direction and the edge effect of the helical-shaped ribbon are crucial to determine the chirality and pitch of a helical nanobelt [5, 9]. Moreover, various materials, such as metal, dielectrics, and polymers, can be integrated into the helical nanobelt by thin film deposition and lithographic patterning [9]. It is worthy of note that nanohelices with a minimum diameter of 7 nm were obtained using 6-monolayer-thick InGaAs/GaAs bilayers [4].
N
N
1716
Nanorobotics for NEMS Using Helical Nanostructures
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 1 Various helical nanostructures. Bottom left is reproduced with permission from ref. [8], (Copyright 2003 APEX/JJAP. The others are in courtesy of L. Zhang and D. Gr€utzmacher)
Key Research Findings Nanorobotic manipulation [18] enables property characterization to be performed after intermediate processes, and in situ characterization can be performed using manipulation rather than conventional static observations. Nanorobotic approach extends the lower limit of robotic manipulation to the dimension of nanometers, and it paves the way for the design and fabrication of nanocomponents having applications in NEMS. Mechanical Properties of Helical Structures By in situ manipulation using a nanoprobe, mechanical properties of various helical structures have been investigated, such as superlattice-structured ZnO nanohelices, carbon nanocoils, and rolled-up nanosprings [11, 12, 14]. It is found that the superelasticity (shape memory) of superlattice-structured ZnO nanohelices enables the nanohelix to recover its shape after an extremely large axial stretching and compressing, which could be of great interest for fabricating nanoscale elastic energy storage in microelectromechanical systems (MEMS) and NEMS. The manipulation experiments of rolled up helical nanostructures can be performed by a nanomanipulator and an atomic force microscope (AFM) cantilever inside a scanning electron microscope (SEM). The nanohelix was picked up by welding its one end using platinum (Pt) deposition onto the tip of a tungsten nanoprobe, and the
mechanical deformation was recorded via extending or compressing the nanohelix using the nanoprobe. As a result, its spring constant can be increased continuously for up to 300–800%. In Fig. 4a, b, it is evident that the welded nanohelix can be pulled to an almost straightened shape, with its twist cycles preserved. Next, the welded nanohelix is released and started to restore its original shape, as shown in Fig. 4c. With comparison to the initial nanohelix dimensions in Fig. 4a, it indicates that the nanohelix has an almost identical dimensionality including pitch and radius, suggesting a complete elastic recovery from stretching. Similar mechanical behavior, that is, the superelasticity (shape memory) behavior, has also been observed when applying a compression load on the welded nanohelix (Fig. 4d, e). The spring constant of ZnO nanohelices is in a range of 0.07–4.2 N/m, and the transverse fracture force is in the micro-Newton range [12, 17]. Nanomanipulators can also be used to carry out the mechanical characterization of rolled-up helical structure. A probe picked up an InGaAs/GaAs nanospring and attached it to the AFM tip (Fig. 5a, b). Then, a tensile force was applied to the nanospring, while continuous SEM images were taken to detect the AFM tip displacement and the nanospring deformation. When the tensile force was increased dramatically, the attachment between the nanospring and the AFM cantilever
Nanorobotics for NEMS Using Helical Nanostructures
1717
N
3D view
a
10) (01
right hand helix
right handed
b
ring
left hand helix
left handed
Top view
) (0001 Cross-section view O2+ Zn2+
O2+ Zn2+
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 3 Illustration of the effect of the angle between the mesa stripe and preferred etch direction (gray arrows) on the formation of rings or helices with right- and left-handed chirality. The thickness of the SiGe/Si/Cr ribbons and the diameter of the helices are 40 nm and 3 mm, respectively (Reproduced with permission from ref. [9], Copyright 2006 ACS. Reproduced with permission from ref. [5], Copyright 2005 IOP)
O2+ Zn2+
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 2 Initiation and formation process of a nanohelix. (a) Low magnification SEM image of a ZnO nanohelix showing its starting point and finishing end. (b) A three-dimensional (3-D) schematic model of a nanohelix, showing its initiating point and finishing end. The periodicity of the superlattices may result in the formation of periodic piezoelectric domains (Reproduced with permission from ref. [17], Copyright 2005 AAAS)
broke, and the nanosprings returned to their initial shape at the zero-displacement position. From the displacement data and the known stiffness of the AFM cantilever, the tensile force acting on the nanosprings versus the nanospring displacement was plotted (Fig. 5c). For the nanospring, an overall elastic strain of 48% was measured when the attachment to the AFM tip broke. Ultra-flexible rolled-up Si-based nanospring with a spring constant of 0.003 N/m was also reported which exhibits long-range linear elasticity (with ca. 189% relative elongation) and excellent mechanical stability [19]. The rolled-up structure is promising for use in electromechanical sensors, such as force or pressure sensors, because of intrinsic piezoresistive and piezoelectric properties of these materials.
Linear-to-Rotary Motion Converter The conversion between various forms of motions will play an important role in future NEMS applications. Among various nano building blocks, 3-D helical nanostructures, for example, dual-chirality helical nanobelts (DCHNBs), could realize conversion between linear and rotary motion [20]. As schematically shown in Fig. 6a, the motion converter consists of a DCHNB with a left-handed and a right-handed part. By linearly stretching both ends of the DCHNB, the central part was mobilized in rotary motion; allowing for linear-to-rotary motion conversion, which has not been previously demonstrated in other nanostructures. With an extended arm, the output can be linear (small displacement) or rotary (large displacement) motion, as seen in Fig. 6b. When a tensile force F is applied to the DCHNB, it elongates and rotates about the unwinding direction. To implement and characterize motion converters, experimental investigations have been performed in the SEM using a nanomanipulator equipped with a tungsten probe and an AFM cantilever. Manipulation of an as-fabricated DCHNB was performed by cutting the lower end of the DCHNB shown in Fig. 7a.
N
N
1718
Nanorobotics for NEMS Using Helical Nanostructures
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 4 Manipulation process of a nanohelix during axial stretching and compressing. (a) One end of a nanohelix was welded with Pt onto a tungsten nanoprobe. (b) A SEM image shows the extremely stretched nanohelix. (c) A complete restoring of the nanohelix shape after releasing. (d, e) Compressed deformation process of a nanohelix induced by the nanoprobe. Both stretching and compressing suggest the superelasticity behavior (Reproduced with permission from ref. [12], Copyright 2006 ACS)
A “sticky” probe was then prepared by dipping a tungsten probe (Picoprobe, T-4-10-1 mm) into a silver tape and attached on the lower end of the DCHNB. Rotation was then generated in the central part by moving the probe downward (Fig. 7b–e).
Examples of Application A fundamental application of the rolled up nanohelix is the 3-D microscopy based on the linear-to-rotary
motion conversion [20], in which samples are rotated so that different views are exposed to light, electron beams, or focused-ion beams (FIBs). Due to their extremely compact size, the dual-chirality helical nanobelts are well suited to serve as rotary scanners for creating 3-D scanning probe microscope (SPMs), rotary stages for a microgoniometer for observing an object from different crystal surfaces, and components of nanomachines. Figure 8 shows 3-D imaging of a pollen grain. The pollen grain is picked up by a Picoprobe and attached
Nanorobotics for NEMS Using Helical Nanostructures Nanorobotics for NEMS Using Helical Nanostructures, Fig. 5 Mechanical characterization. (a, b) Break and pick up nanospring and attach it to AFM tip. (c) Simulation and experimental results of axial force versus nanospring displacement (Reproduced with permission from ref. [14], Copyright 2006 ACS)
a
b
5
c
Nanospring 1 Nanospring 2 Nanospring 3 Simulation N1 Simulation N2 Simulation N3
0.6
4.5
0.5 Force [μN]
4 3.5 Force [μN]
N
1719
3 2.5
0.4 0.3 0.2 0.1
2
0
1.5
0
6 4 Displacement [μm]
2
1
8
10
0.5 0 0
5
10
15
20
Displacement [μm]
a Nanorobotics for NEMS Using Helical Nanostructures, Fig. 6 Motion converter based on the dual-chirality helical nanobelts (DCHNB). (a) Linear-to-rotary motion converter using a DCHNB. (b) Transmission converting linear motion to extended linear (small displacement)/ rotary (large displacement) motion (Reproduced with permission from ref. [20], Copyright 2009 IEEE)
F,d
t, f
N F,d
b F, d
to the DCHNB. The motion converter has been characterized before loading the sample. Figure 8a–d shows that the sample rotates counterclockwise (top view) for approximately 180 when releasing the extended DCHNB. Figure 8e–h shows further stretching of the DCHNB, and the sample rotates clockwise (top view) for another 180 . It can be seen that the different aspects of the sample can be exposed
t, f
F, d
T, F
P, D
to the electron beam simply by extending or releasing the DCHNB with a small displacement. Further possibilities for this converter include goniometry for a transmission electron microscope (TEM), tomography for a SEM or a TEM, and 3-D SPM. By integrating in the fabrication processes of DCHNBs, rolled-up spirals can serve as claws for holding relatively large samples [6].
N
1720
Nanorobotics for NEMS Using Helical Nanostructures
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 7 In situ characterization of motion conversion using nanorobotic manipulation. (a) One end of an as-fabricated DCHNB is cut and attached to a “sticky” probe. (b, e) Rotation of the central part while moving the probe downward (Reproduced with permission from ref. [20], Copyright 2009 IEEE)
Nanorobotics for NEMS Using Helical Nanostructures, Fig. 8 Three-dimensional microscopy of a pollen grain. (a–d) When releasing the extended DCHNB, the sample rotates
counterclockwise (top view). (e–h) When stretching further the DCHNB, the sample rotates clockwise (top view) (Reproduced with permission from ref. [20], Copyright 2009 IEEE)
References 1. Amelinckx, S., Zhang, X.B., Bernaerts, D., Zhang, X.F., Ivanov, V., Nagy, J.B.: A formation mechanism for catalytically grown helix-shaped graphite nanotubes. Science 265(5172), 635–639 (1994) 2. Kong, X.Y., Wang, Z.L.: Spontaneous polarization-induced nanohelixes, nanosprings, and nanorings of piezoelectric nanobelts. Nano Lett. 3(12), 1625–1631 (2003) 3. Schmidt, O.G., Eberl, K.: Nanotechnology – thin solid films roll up into nanotubes. Nature 410(6825), 168–168 (2001) 4. Prinz, V.Y., Seleznev, V.A., Gutakovsky, A.K., Chehovskiy, A.V., Preobrazhenskii, V.V., Putyato, M.A., Gavrilova, T.A.: Free-standing and overgrown InGaAs/ GaAs nanotubes, nanohelices and their arrays. Physica E 6(1–4), 828–831 (2000) 5. Zhang, L., Deckhardt, E., Weber, A., Schonenberger, C., Grutzmacher, D.: Controllable fabrication of SiGe/Si and
6.
7.
8.
9.
SiGe/Si/Cr helical nanobelts. Nanotechnology 16(6), 655–663 (2005) Zhang, L., Dong, L.X., Bell, D.J., Nelson, B.J., Scho¨nenberger, C., Gr€ utzmacher, D.: Fabrication and characterization of freestanding Si/Cr micro- and nanospirals. Microelectron. Eng. 83(4–9), 1237–1240 (2006) Schmidt, O.G., Schmarje, N., Deneke, C., Muller, C., Jin-Phillipp, N.Y.: Three-dimensional nano-objects evolving from a two-dimensional layer technology. Adv. Mater. 13(10), 756–759 (2001) Kubota, K., Fleischmann, T., Saravanan, S., Vaccaro, P.O., Aida, T.: Self-assembly of microstage using micro-origami technique on GaAs. Jpn. J. Appl. Phys 42(6B), 4079–4083 (2003). Part 1 – Regular Papers Short Notes & Review Papers Zhang, L., Ruh, E., Grutzmacher, D., Dong, L.X., Bell, D.J., Nelson, B.J., Schonenberger, C.: Anomalous coiling of SiGe/Si and SiGe/Si/Cr helical nanobelts. Nano Lett. 6(7), 1311–1317 (2006)
Nanoscale Printing 10. Cho, A.: News focus: nanotechnology: pretty as you please, curling films turn themselves into nanodevices. Science 313, 164–165 (2006) 11. Chen, X.Q., Zhang, S.L., Dikin, D.A., Ding, W.Q., Ruoff, R.S., Pan, L.J., Nakayama, Y.: Mechanics of a carbon nanocoil. Nano Lett. 3(9), 1299–1304 (2003) 12. Gao, P.X., Mai, W.J., Wang, Z.L.: Superelasticity and nanofracture mechanics of ZnO nanohelices. Nano Lett. 6(11), 2536–2543 (2006) 13. Bell, D.J., Sun, Y., Zhang, L., Dong, L.X., Nelson, B.J., Gr€utzmacher, D.: Three-dimensional nanosprings for electromechanical sensors. Sens. Actuator A-Phys. 130–131, 54–61 (2006) 14. Bell, D.J., Dong, L.X., Nelson, B.J., Golling, M., Zhang, L., Grutzmacher, D.: Fabrication and characterization of threedimensional InGaAs/GaAs nanosprings. Nano Lett. 6(4), 725–729 (2006) 15. Zhang, H.F., Wang, C.M., Buck, E.C., Wang, L.S.: Synthesis, characterization, and manipulation of helical SiO2 nanosprings. Nano Lett. 3(5), 577–580 (2003) 16. Zhang, D.Q., Alkhateeb, A., Han, H.M., Mahmood, H., McIlroy, D.N., Norton, M.G.: Silicon carbide nanosprings. Nano Lett. 3(7), 983–987 (2003) 17. Gao, P.X., Ding, Y., Mai, W.J., Hughes, W.L., Lao, C.S., Wang, Z.L.: Conversion of zinc oxide nanobelts into superlattice-structured nanohelices. Science 309(5741), 1700–1704 (2005) 18. Dong, L.X., Nelson, B.J.: Robotics in the small, part II: nanorobotics. IEEE Robot. Autom. Mag. 14(3), 111–121 (2007) 19. Dai, L., Zhang, L., Dong, L.X., Shen, W.Z., Zhang, X.B., Ye, Z.Z., Nelson, B.J.: Long-range linear elasticity and mechanical instability of self-scrolling binormal nanohelices under a uniaxial load. Nanoscale 3, 4301–4306 (2011) 20. Dong, L.X., Zhang, L., Kratochvil, B.E., Shou, K.Y., Nelson, B.J.: Dual-chirality helical nanobelts: linear-torotary motion converters for three-dimensional microscopy. J. Microelectromech. Syst. 18(5), 1047–1053 (2009)
Nanorods ▶ Nanostructures for Energy ▶ Physical Vapor Deposition3
Nanoscaffold ▶ Ligand-Directed Gold-Phage Nanosystems
Nanoscale Drug Vector ▶ Fullerenes for Drug Delivery
1721
N
Nanoscale Fluid Mechanics ▶ Computational Micro/Nanofluidics: Unifier of Physical and Natural Sciences and Engineering
Nanoscale Heat Transport ▶ Thermal Conductivity and Phonon Transport
Nanoscale Particle ▶ Gas Phase Nanoparticle Formation
Nanoscale Printing Timothy J. Merkel1 and Joseph M. DeSimone1,2,3,4,5,6,7,8 1 Department of Chemistry, University of North Carolina, Chapel Hill, NC, USA 2 Department of Pharmacology, Eshelman School of Pharmacy, University of North Carolina, Chapel Hill, NC, USA 3 Carolina Center of Cancer Nanotechnology Excellence, University of North Carolina, Chapel Hill, NC, USA 4 Institute for Advanced Materials, University of North Carolina, Chapel Hill, NC, USA 5 Institute for Nanomedicine, University of North Carolina, Chapel Hill, NC, USA 6 Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC, USA 7 Department of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh, NC, USA 8 Sloan–Kettering Institute for Cancer Research, Memorial Sloan–Kettering Cancer Center, New York, NY, USA
Synonyms Dip-pen nanolithography; Microcontact printing;
Imprint lithography; Molecular printing;
N
N
1722
Nanoscale Printing
Nanoimprint lithography; Nanotransfer printing; PRINT ®; Replica molding; Soft lithography
new industries, especially in the life sciences and energy sciences fields.
Definition
Patterning Surfaces
Nanoscale printing refers to the patterning of surfaces in two or three dimensions with at least one feature on the submicron length scale. The printing techniques may resemble macroscopic printing, in which a material, or ink, is transferred to a substrate by a stamp or a stylus. However, nanoscale printing may also refer to embossing or molding processes, in which topographical features are transferred from a threedimensional mold to a surface by either elevating or depressing areas or by the addition of material. For the case of patterning features in three dimensions, two distinct outcomes are possible: the features generated can remain connected by an underlying layer of material, yielding an embossed film or, alternately, the molded features can be free of the connecting layer, yielding discrete particles. Arrays of features have been used to direct the assembly of materials for applications in molecular electronics, genetic analysis, and disease detection. Particles generated by these methods have found a myriad of uses, including those in drug delivery and biomedical imaging.
Two complementary nanoscale printing techniques have emerged for the transfer of reagents, both chemical and biological to a surface: techniques based on scanning probe microscope-based lithography (SPL) and soft lithography-based techniques. Based on the development of scanning probe microscopy in the 1980s, SPL nanofabrication techniques have emerged that exploit the use of the tip of the scanning probe to manipulate matter on a surface. These methods offer the advantage of extremely high resolution as these probes are capable of manipulating individual atoms or molecules. However, this typically requires high vacuum and temperatures which approach absolute zero, and is not readily scalable for patterning of large areas [2]. While these methods represent powerful tools for high-resolution patterning over a small area, probe tips were not used in a printing method until the advent of dip-pen nanolithography (DPN) in 1999 [3].
Overview Methods used to generate nanoscale structures are generally divided into “top-down” and “bottom-up” approaches. In bottom-up approaches, features are generated from the atomic or molecular level to the supermolecular or macroscopic level and utilize interactions between molecules or particles to achieve twoor three-dimensional structures [1]. For top-down methods, materials are processed at the desired size scale, with external forces directing the material to form the desired architecture. The development of nanoscale printing methods for drug delivery and imaging applications have been enabled by the remarkable developments in the semiconductor industry over the last 50 years to reduce feature sizes in order to increase the number of transistors in a microchip [2]. Now these top-down technologies are poised to be exploited for benefit in a variety of
Patterning with Dip-Pen Nanolithography Using an “ink”-coated atomic force microscope (AFM) tip as a stylus, DPN is a direct-write method in which soft or hard materials can be deposited to a surface with sub 50 nm resolution [3]. Developed in the Mirkin laboratory at Northwestern University in 1999, DPN relies on humidity to cause the spontaneous formation of a meniscus between the AFM tip and the surface (Fig. 1). The ink is deposited along this meniscus, with the deposition rate dependent upon factors such as the nature of the ink, the contact time between the tip and the substrate, and the coverage of the ink on the AFM tip. The first example of DPN deposited an alkanethiol onto a gold substrate, with the strong interactions between the gold substrate and the thiol overcoming the weaker interaction between the tip and the thiol to drive the formation of patterns on the surface. A variety of metallic, semiconducting and insulating substrates have been used with DPN [3]. The technique has proven to be quite versatile with respect to the composition of the ink; structures have been formed from organometallics, polymers, colloidal nanoparticles, metal ions, and a range of biologicals
Nanoscale Printing
1723
N
Heating of the AFM tip (thermal DPN) can allow for deposition of insoluble ink molecules that are solids at room temperatures.
Nanoscale Printing, Fig. 1 A schematic showing the dip-pen nanolithography (DPN) technique. In DPN, ink molecules are transferred from the tip of an atomic force microscope stylus to a surface through a water meniscus
including DNA, proteins, peptides, and affinity templates which allowed for the subsequent immobilization of viruses [2]. A primary application of DPN techniques is the generation of biological nanoarrays for the study of cellular genetics and rapid screening of diseases. DPNgenerated nanoarrays of an antibody have been used to screen for the human immunodeficiency HIV-1 virus antigen in serum samples. This study demonstrated an increase in detection limits of orders of magnitude over conventional enzyme-linked immunosorbent assays [3]. Nanoarrays generated in this way have advantages of holding 104–105 more features than conventional microarrays, offering the opportunity for either decreased sample volume or for the screening of a larger number of targets per experiment. While printing with a single AFM tip over large areas would be impractically slow, massively parallel arrays of cantilevers have been fabricated to increase the throughput of the technique. Arrays with as many as 1.3 million cantilevers have been microfabricated toward this end, increasing the throughput by four orders of magnitude. A number of variations to DPN have emerged to broaden the applicability of the technique, with modification of the tips a common theme. Coating the AFM tip with a silicone polymer increases the amount of ink that can be loaded to the tip, allowing for printing of larger areas. The inclusion of microfluidic channels to deliver ink to the tips, termed nanofountain pen (NFP) probes, ensures a constant supply of ink to the surface. Features with greater than 50 nm resolution have been printed with NFP probes. The technique also serves to keep the ink molecules solvated, an aspect which can be critical for biological applications.
Patterning with Soft Lithography Techniques Pioneered in the 1990s in the Whitesides laboratory at Harvard University, soft lithography begins with the generation of a rigid template which is used to generate an elastomeric mold or stamp which replicates the three-dimensional shape of the original template. Master templates are typically prepared by photolithographic techniques, which were pioneered in the electronics industry. The molds of these master templates are typically generated from poly (dimethyl siloxane) (PDMS), although polyurethanes, polyamides, perfluoropolyethers, and natural polymers such as agarose have been used for this purpose. To fabricate the mold, the desired polymer is poured onto the master template and cured, generating a relief structure with the raised portions of the stamp defining the pattern to be transferred. Microcontact printing (mCP) is the process by which the mold is used as a stamp after it is coated with an ink, and subsequently brought into contact with a substrate to transfer the ink (Fig. 2). The ink transfers from the stamp by chemisorption or physisorption to the surface. While alkanethiol inks transferred to a gold substrate are, perhaps, the most common usage of mCP, the technique has also been used to pattern proteins, DNA, cells, colloids, and polymers [4] onto a variety of surfaces. The flexibility of the elastomeric stamp allows for conformal atomic level contact to be achieved on flat and curved surfaces, and over large stamp areas. The mCP technique has advantages as it is low-cost, is high-throughput and is easy to use which has earned it popularity as a tool among researchers. While the low cost of the elastomeric molds has rendered mCP accessible to researchers in many fields, there are some limitations to the technique. Patterns are determined by the original master template, so that each stamp can only produce one pattern. The mechanical properties of the stamp can limit the flexibility of the design; features spaced too widely (> 20 times the feature height for typical PDMS) or with aspect ratios outside of the 0.2–2 range can cause defects from deformation of the stamp [2]. Stamping of features smaller than 150 nm is a challenge using conventional methods, though features as small as 80 nm have
N
N
1724
Nanoscale Printing
Nanoscale Printing, Fig. 2 A schematic for microcontact printing (mCP). (a) The fabrication of a mold (top to bottom). The mold material (gray) is poured onto a master template with topographical features (black). The liquid mold materials solidifies and is removed from the master template, producing a mold
with the inverse of the features found on the master template. (b) The mCP process (top to bottom). The elastomeric mold is coated with a chemical ink (red). Ink is transferred to the surface by contact to the raised features of the mold
been prepared by using DPN to modify the surface of a flat stamp to passivate surrounding areas with a perfluorinated monolayer. Nanotransfer printing (nTP) is the process of transferring a thin solid film from a nanoscale-featured stamp to a substrate. In this case, the stamp can be either rigid or flexible and the material transferred is typically a metal. In a typical process, the stamp is coated with a thin film (e.g., 20 nm of gold) and brought into contact with the desired surface. The film can transfer from the stamp by a number of methods, including binding to reactive self-assembled monolayers (SAMs) like alkanethiols, cold welding between two metal layers, or condensation between silanol or titanol groups [5]. Applications for nTPderived structures are typically electronic in nature, including organic thin film transistors, capacitors, and electrostatic lenses. Features as small as 70 nm have been fabricated with nTP. The recent innovation of polymer pen lithography (PPL) in 2008 combines many of the advantages of mCP and DPN by using an elastomeric array of tips to print a pattern. With PPL, a PDMS stamp consisting of an array of tips is affixed to a piezo scanner and delivers an ink with spot sizes from 65 nm to over 10 mm determined by varying the force and time over which the ink is delivered [6]. The elastomeric arrays of PPM pens, which can contain as many as 10 [8] pyramid-shaped tips, are made from a master prepared by traditional photolithographic techniques. The cost
of preparing new arrays of tips is greatly reduced for PPL compared to DPN, with the cost of a new PPL array estimated at less than $1 after the fabrication of the initial master template [2]. The resulting arrays are transparent, allowing for registration of the cantilevers with previously printed features. The master template used to generate the arrays can be filled by injet printing, allowing it to be used as an array of inkwells with which the array of tips is perfectly aligned. The entire elastomeric array absorbs ink, acting as a reservoir and allowing for printing over large areas without reinking. Molding and Embossing Methods
The molding process involves pouring a liquid precursor onto a nanopatterned template and curing or solidifying it into a solid which replicates the original features. Molding features on the nanoscale can be performed with both hard and soft molds. This is the basis for techniques such as step-and-flash imprint lithography (SFIL), replica molding (RM), and pattern replication in non-wetting templates (PRINT). Embossing or imprinting methods use a mold to transfer features to an initially flat polymer film. Embossing methods include nanoimprint lithography (NIL), solvent-assisted micromolding (SAMIM), and PRINT. Hard molds have been fabricated out of quartz, silicon, and metals, though quartz and silicon are the most commonly used. Features as small as 10 nm have been transferred into silicon, while quartz molds have a minimum size of 20 nm [4]. Though soft molds are
Nanoscale Printing
1725
N
less costly and easier to fabricate, rigid molds offer several advantages over softer molds. Rigid molds will show little deformation over large areas with the pressures required for embossing. Hard molds are stable at the temperatures required for cross-linking polymer precursors and are chemically inert to them. Release of the mold from the patterned substrate can be difficult, depending on the surface properties of the mold and the molded polymer, and release coatings of fluoroalkyl silanes are typically linked to the mold surface to facilitate release. Quartz substrates are transparent to UV light and visible light, while silicon is not, allowing for photochemical cross-linking of prepolymer with quartz molds. Step-and-Flash Imprint Lithography
Step-and-flash imprint lithography replicates the topography of a quartz mold using a photocurable prepolymer solution as the molded material at room temperature using low applied pressure. The prepolymer solution consists of a monomer and a photoinitiator with hardening of the material proceeding by free-radical polymerization upon exposure to UV light. An important consideration for complete filling of the mold is the viscosity of the prepolymer, or monomer, solution; low molecular weight monomers are used due to their low viscosity ( Psat) condensation prevails over the evaporation. The ratio of the two values is the relative humidity, RH ¼ PV/Psat. For a concave interface, the equilibrium occurs at PV/Psat < 1, whereas for a convex interface is occurs at PV/Psat > 1. The exact value of PV/Psat for a meniscus of a given curvature is given by the Kelvin equation. The pressure drop of water with density r risen for the height h in a capillary tube is DP ¼ rgh. Using the Laplace equation and the hydrostatic formula for vapor pressure P ¼ P0exp(gh/RT) – where R is the gas constant, P0 is the pressure at the surface (h ¼ 0), and T is the temperature – yields the Kelvin equation: gLV
1 1 þ R 1 R2
¼ rRT ln
PV Psat
(1)
where R1 and R2 are the principle radii of curvature of the meniscus (Rk ¼ (1/ R1 + 1/ R2)1 is referred to as the Kelvin radius). Equation 1 relates the interface curvature at the thermodynamic equilibrium with the ratio of actual and saturated vapor pressure, PV/Psat (relative humidity). According to the Kelvin theory, a concave meniscus with a negative curvature given by (3) may form at any relative humidity [1]. Water at the Nanoscale Normally, the phase state of water is uniquely characterized by its pressure and temperature. The phase diagram of water shows whether water is in its solid, liquid, or vapor state at a given temperature and
Nanoscale Water Phase Diagram
0
Laplace pressure drop
Standard conditions Solid
N Critical point
Liquid
Pressure
pressure. However, at the nanoscale, the situation may be different. Both molecular dynamics (MD) simulations and experimental studies with the atomic force microscope (AFM) and the surface force apparatus (SFA) show that the state of nanoscale volumes of water at a given pressure and temperature is not always the same as prescribed by the macroscale water phase diagram. In particular, confined small water volumes demonstrate quite unusual properties, such as the melting point depression. Several effects that can be detected by AFM are related to this unusual phase behavior of the nanoscale capillary bridges. First is the metastability of small volumes of water. When pressure is below the liquid–vapor transition line of the water phase diagram, vapor is the most stable state. However, the liquid–vapor transition involves energy barriers, and thus a metastable liquid state is possible [1]. At the macroscale, random fluctuations are normally large enough to overcome the barriers. The metastable states are very fragile, and thus they are not normally observed, except for special circumstances (e.g., superheated water). However, at the nanoscale, the barriers are large compared with the scale of the system, and metastable states can exist for long intervals of time [1]. Second, it was argued that ice could form at room temperature inside the capillary bridges. The water in such a bridge can be neither liquid, nor solid, but form a liquid–ice condensate. This explains the slow rearrangement of the bridges between tungsten AFM tips and graphite surfaces reported in recent experiments [2]. The presence of ice-like structured water adsorbed at the silicon oxide surface was suggested to be responsible for the large RH dependence of the adhesion force in the single-asperity contact between silicon oxide surfaces in AFM experiments. Third, pressure inside the capillary bridge is lower than the pressure outside. At the reduced pressure, the water boiling point is lower than 100 C. There is experimental evidence that boiling in the capillary bridges indeed happens in concave water menisci. Investigating nanoscale water phase transitions using AFM data can lead to modifications of the conventional water phase diagram by introducing a scale dependence. Such a diagram should depend on the characteristic size of the system, in addition to the pressure and temperature dependencies [3].
1735
Boiling
Temperature Metastable liquid Vapor Tensile strength of water (spinodal)
Nanoscale Water Phase Diagram, Fig. 1 Schematic of a phase diagram of water showing solid, liquid, vapor, and metastable liquid phases. While the energy barriers separating the metastable states are small, at the nanoscale these barriers are very significant
Stretched Water When a phase transition line in the phase diagram is crossed, it is normally expected that the substance would change its phase state. However, such a change would require additional energy input for nucleation of seeds of the new phase. For example, the liquid–vapor transition requires nucleation of vapor bubbles (this process is called cavitation), while the liquid–solid transition requires nucleation of ice crystals. If special measures are taken to prevent nucleation of the seeds of the new phase, it is possible to postpone the transition to the equilibrium state phase leading to supercooled, superheated, or stretched water (Fig. 1). Two factors affect a vapor bubble inside liquid, needed as a nucleation center for vapor in superheated or stretched water: the pressure and the interfacial tension. While the negative pressure (tensile stress) is acting to expand the size of the bubble, the interfacial tension is acting to collapse it. For a small bubble, the interfacial stress dominates; however, for a large bubble, the pressure dominates. Therefore, there is a critical radius of the bubble, and a bubble with a radius larger than the critical radius would grow, whereas one with a smaller radius would collapse. The value of the critical radius is given by: RC ¼
2gLV Psat P
(2)
N
N
1736
Nanoscale Water Phase Diagram
where gLV is the liquid–vapor surface tension, Psat is the saturated vapor pressure, and P is the actual liquid pressure [1]. With the further decrease of pressure in the negative region, the so-called spinodal limit can be reached. At that pressure, the critical cavitation radius becomes of the same order as the thickness of the liquid–vapor interface. In this case, there is a lower energy path of nucleation connecting the liquid to the gas phase by expansion of a smoothly varying density profile. In the phase diagram, the line that corresponds to the spinodal limit is expected to go all the way to the critical point. This critical spinodal pressure at a given temperature effectively constitutes the tensile strength of liquid water. Various theoretical considerations, experimental observations, and MD simulation results have been used to determine the value of the spinodal limit. At room temperature, the spinodal pressure is between 150 and 250 MPa [1].
a
Disjoining Pressure Another important effect is the so-called disjoining pressure. The adhesion force between the solid and water has a certain range x0 and decreases with the distance x from the interface. As a result, water in the layer next to the interface is subjected to higher pressure P ¼ P0 + Pd(x) than the bulk pressure, where Pd(x) is the so-called disjoining pressure. For a thin liquid layer of thickness H < x0, the evaporation/condensation equilibrium will be shifted toward condensation, since an additional attractive force acts upon water molecules from the solid surface. As a result, a thin water layer can be in equilibrium with undersaturated vapor, PV/Psat < 1, so the effect of the disjoining pressure on the thermodynamic equilibrium is similar to the effect of concave meniscus [4]. The influence of the disjoining pressure and the related wetting films in narrow confinement have to be considered as they may significantly change the meniscus curvature (Fig. 2). A typical effect of nanoscale confinement is the appearance of capillarity-induced negative pressure [3].
Condensation
b Evapo-
P(x)
ration Evaporation
Disjoining pressure
Condensation
Condensation
Evaporation Vapor
Vapor x0 G(x)
Vapor
x
Liquid
Liquid
Energy x0
Liquid
x
Evaporation
Liquid
Condensation
Vapor
Condensation Evaporation
Liquid
Vapor
x Liquid Solid
Solid
Nanoscale Water Phase Diagram, Fig. 2 (a) Disjoining pressure near a solid surface (b) The effect of meniscus curvature ( flat, concave, and convex) and liquid layer thickness on the evaporation–condensation equilibrium
Nanoscale Water Phase Diagram
1737
Key Findings Negative Pressure in Water Capillary Bridges It is very difficult to obtain deeply negative pressure at the macroscale because of bubble nucleation. The pressure values that have been reported constitute 19 MPa with the isochoric cooling method, 17.6 MPa with the modified centrifugal method, and 25 MPa with the acoustic method. The situation is different at the micro- and nanoscale. The pressure of 140 MPa in the microscopic aqueous inclusions in quartz crystals was reported. Water plugs in hydrophilic nanochannels can be at a significant absolute negative pressure due to tensile capillary forces [1]. Negative pressure of 160 MPa was reported in nanoscale liquid capillary bridges. The capillary component of the adhesion force was determined by the comparison of the adhesion force in air (at different RH levels) and in ultra-high vacuum (UHV), where capillary bridges are absent. Very small values of the Atomic force microscope (AFM) tip radius (on the order of 10 nm) correspond to a small size of meniscus (meniscus foundation area on the order of 1016 m2) and thus even a small force of 10 nN corresponds to huge pressure differences inside and outside the meniscus on the order of 100 MPa or 1,000 atm (Fig. 3). The reason for the existence of such large tension limits lies in the strength of the hydrogen bonds which must be broken in order to form a cavity that is large enough to act as a stable nucleus for cavitation. The metastable state of the capillary bridges is observed because the size of the bridges is smaller than the critical cavitation radius given by (2) and, therefore, no cavitation occurs [1]. Room Temperature Ice in Capillary Bridges Most AFM results show that the capillary force has a strong dependence on RH: first it increases with increasing relative humidity up to approximately 30% and then decreases with a further increase of the relative humidity. For low humidity, the pressure difference inside and outside the bridge is large; however, the size of the bridge is small due to high curvature of the meniscus. For very low humidity, the size of the bridge may be insufficient to generate a capillary force. On the other hand, for high humidity, the pressure difference is small; however, the meniscus is large (but, of course, its size is limited by the radius of the
N
F
Water meniscus
Solid asperity h
Solid surface
2R0
F
Water
Nanoscale Water Phase Diagram, Fig. 3 A water capillary bridge between an AFM tip and a flat sample. The Laplace pressure inside the bridge is reduced because the bridge is concave. The bridge is under tension, similarly to water in a sealed tube with a piston with the applied tensile force F
tip). Thus, it is expected that in the limiting cases of RH ¼ 0% and RH ¼ 100% the capillary force would vanish, since at RH ¼ 0% there is no capillary bridge and at RH ¼ 100% the Laplace pressure drop is zero. The capillary force achieves its maximum at a certain value of RH [3]. It has been argued that the capillary force alone cannot explain the observed magnitude of the humidity dependence. The origin of the large RH dependence is due to the presence of ice-like structured water adsorbed at the silicon oxide surface at room temperature. Thus, the force is due to the rupture of an ice–ice bridge at the center of the contact region with the structure, thickness, and viscoelastic behavior of the adsorbed water layer influencing the adhesion force [2]. Water often forms adsorbed layers at the surface of many materials, and these layers may have an ice-like structure and be connected to the capillary bridge, as indicated by the study of the molecular configuration of adsorbed water as a function of relative humidity using attenuated total reflection (ATR) infrared spectroscopy [4]. A completely hydrogen-bonded ice-like network of water grows up to three layers as the
N
N
1738
Nanoscale Water Phase Diagram
relative humidity increases from 0% to 30%. In the relative humidity range of 30–60%, the liquid water structure starts appearing, while the ice-like structure continues growing to saturation. Above 60% relative humidity, the liquid water configuration grows on top of the ice-like layer. During the contact between a tungsten tip and a graphite surface it was found that water in the nanoscale capillary bridges may include ice or be a mixture of liquid water with ice [2]. Such a mixture would have high viscosity and may act like glue, demonstrating elastic response. Low-Temperature Water Boiling Another important consequence of the water pressure reduction in the capillary bridges is the possibility of boiling at temperatures lower than 100 C. If a capillary bridge is large enough, so that cavitation occurs, the pressure inside the bridge is reduced compared with the ambient, and the boiling point is depressed by the amount DT. The Clausius–Clapeyron equation provides the change of pressure DP ¼ P2 P1 as related to the change of the boiling temperature DT ¼ T2 T1 as lnðP2 =P1 Þ ¼ Dh=Rð1=T1 1=T2 Þ
(3)
where Dh is the change of specific enthalpy [54]. Note that DP is given by (1). For example, for a meniscus between an AFM cantilever and a flat surface with a gap between them of 3 mm and the principal radii of curvature of the meniscus R1 ¼ R2¼2 mm, the Kelvin radius is Rk ¼ (1/R1 + 1/R2)1 ¼ 1 mm and the pressure drop, based on (6) is DP ¼ 72.9 kPa (or 0.72 atm). Thus, the pressure inside the meniscus is P ¼ 28.1 kPa (or 0.28 atm), which corresponds to the boiling point of about 333 K [3]. Nanobubbles There is no adequate theoretical explanation why ice can be present in the capillary bridges at room temperature; however, it is clear that this effect is related to the adsorption of water to solid surfaces. The ability of water to form thin layers on hydrophobic surfaces is associated with the disjoining pressure caused by the spatial dependence of the van der Waals adhesion force. Due to the dependence of the adhesion force of the distance (on the scale of nanometers), the energy of the water molecules in the vicinity of the solid surface
is lower than the energy of those far away. Thus, it is often not only energetically profitable for water to cover the solid, but also to create a thin layer. It is noted that not only the solid phase can be adsorbed at a surface. Very thin (5–80 nm) gas films could exist for a long time at the interface between a hydrophobic solid and water. Despite certain thermodynamic objections for nanobubble existence (such as expected high pressures in bubbles leading to their dissolution), some experimental studies with the AFM and other equipment have demonstrated the existence of nanobubbles at the solid/water interface [5,6]. Nanobubbles can play a significant role in a number of processes, such as the long-range hydrophobic attractive force. Water in Confinement A number of studies are devoted to water phase behavior in confinement. MD simulations show that water confined to carbon nanotubes of a critical size under ambient conditions (1 bar, 300 K) can undergo a transition into a state having ice-like mobility with an amount of hydrogen bonding similar to that in liquid water. Maeda [7] investigated experimentally phase transitions of capillary-held liquids in a slit-like pore. When the surfaces were placed at a separation very close to the Kelvin length, it was possible to detect a stage in which the system was in an apparent kinetic equilibrium between two physical states – with and without the liquid connecting the two surfaces. Giovambattista et al. [8] used MD simulation to investigate the phase behavior of water confined between two nanoscale hydrophobic or hydrophilic plates at room temperature, varying the pressure (0.15 GPa P 0.2 GPa) and plate separation (0.4 nm d 1.6 nm). With hydrophobic plates, capillary evaporation occurred between the plates at low P. The threshold value of d at which this transition occurs decreases with P (e.g., 1.6 nm at P ¼ 0.05 GPa, 0.5 nm at P ¼ 0.1 GPa), until, at high P, no capillary evaporation occurred. For d ¼ 0.6 nm and P > 0.1 GPa, the system crystallizes into a bilayer ice. They developed a P–d phase diagram showing the vapor, liquid, and bilayer ice. With hydrophilic plates, water always remained liquid even at the P at which bulk water cavitates. This indicates that size is important for a phase diagram due to the finite range of the adhesion force in addition to the critical cavitation radius and curvature of the meniscus, which were discussed in the preceding sections.
Nano-sized Nanocrystalline and Nano-twinned Metals
1739
N
Cross-References
Definition
▶ AFM in Liquids ▶ Disjoining Pressure and Capillary Adhesion ▶ Wetting Transitions
• Nano-pillars are pillar-shaped specimens at the nanoscale, whose aspect ratio (height/diameter) is generally within 3 7. • Nanocrystalline material refers to a class of polycrystalline materials whose grain size is smaller than 100 nm. • Nano-twin refers to a class of crystalline materials that contain a number of parallel coherent twin boundaries with the inter-boundary distance at the nanoscale.
References 1. Yang, S.-H., Nosonovsky, M., Zhang, H., Chung, K.-H.: Negative pressure in water capillary bridges at nanocontacts. Chem. Phys. Lett. 451, 88–92 (2008) 2. Jinesh, K.B., Frenken, J.W.M.: Capillary condensation in atomic scale friction: how water acts like a glue. Phys. Rev. Lett. 96, 166103 (2006) 3. Nosonovsky, M., Bhushan, B.: Phase behavior of capillary bridges: towards nanoscale water phase diagram. Phys. Chem. Chem. Phys. 10, 2137–2144 (2008) 4. Asay, D.B., Kim, S.H.: Effects of adsorbed water layer structure on adhesion force of silicon oxide nanoasperity contact with humid ambient. J. Chem. Phys. 124, 174712 (2006) 5. Steitz, R., Gutberlet Hauss, T., Klosgen, B., Krastev, R., Schemmel, S., Simonsen, A.C., Findenegg, G.H.: Nanobubbles and their precursor layer at the interface of water against a hydrophobic substrate. Langmuir 19, 2409– 2418 (2003) 6. Mashl, R.J., Joseph, S., Aluru, N.R., Jakobsson, E.: Anomalously immobilized water: a new water phase induced by confinement in nanotubes. Nano Lett. 3, 589–592 (2003) 7. Maeda, N.: Phase transitions of capillary-held liquids in a slitlike pore. J. Phys. Chem. B 110, 25982–25993 (2006) 8. Giovambattista, N., Rossky, P.J., Debenedetti, P.G.: Effect of pressure on the phase behavior and structure of water confined between nanoscale hydrophobic and hydrophilic plates. Phys. Rev. E 73, 1–14 (2006)
Nanoscratch Tester ▶ Nanoindentation
Nano-sized Nanocrystalline and Nano-twinned Metals Dongchan Jang Department of Applied Physics and Materials Science, California Institute of Technology, Pasadena, CA, USA
Synonyms Electroplating; Nanocrystalline materials; Nano-pillars; Nano-twinned Materials
Overview Traditionally, the most fundamental principle in materials science is that the properties of materials are strongly influenced by their internal microstructure [1]. During the plastic deformation of many metallic systems, dislocations carry most of the plasticity and hence their interactions with the internal microstructure, such as grain boundaries, solute atoms, precipitation or even other dislocations, are the most influential in determining mechanical responses of those materials [2]. However, recent studies found that the microstructure is not the only factor that affects the mechanical properties, but the external dimension of the specimen also significantly changes the mechanical behavior when it is reduced to the submicron and nanometer scale [3]. Therefore, when the effects from those two factors, microstructure and specimen size at the nanoscale, are combined, the materials often exhibit very unique properties distinguished from both of the bulk samples containing same kind of microstructure and nanosamples not containing the corresponding microstructure [3]. Especially, the homogeneous internal interfaces, such as ordinary grain boundary (GB) and coherent twin boundary (TB), are one of the most interesting microstructures as their interaction to dislocations is a key parameter in understanding plasticity of metallic materials. In this entry, the review of the recent studies investigating the mechanical properties of nanosized metallic samples containing ordinary GBs and coherent TBs is presented. The main focus is placed on the results from uniaxial tension/compression experiments on the nanometer-sized pillar-shaped specimen (nano-pillars). Fabrication techniques to produce the nano-pillars with various internal interfaces are also introduced.
N
N
1740
Nano-sized Nanocrystalline and Nano-twinned Metals
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 1 Strength of polycrystalline materials as a function of grain size: HallPetch relation and transition to inverse Hall-Petch (Reprint with permission from [3])
Mechanical Properties Nanocrystalline Nano-pillars GB is the interface between differently oriented grains in the polycrystalline materials. Generally, there is atomic mismatch across the GB as the crystallographic orientation of one grain abruptly changes to that of the adjacent one [1]. In conventional bulk polycrystalline materials, GBs usually act as a barrier to the dislocation motion across the boundary because of the discontinuity of atomic planes at the GB [1]. As a result, the strength of the bulk polycrystalline materials is inversely correlated with the grain size because total GB area increases with decreasing grain size, so dislocations are impeded more often with smaller grain size. For many conventional polycrystalline metals, the strength is inversely proportional to the square root of grain size, i.e., sy ¼ so þ ky d1=2 , where sy and d are the yield strength and grain size, respectively, and so and ky are constants for a particular material. This expression is called Hall-Petch relation [1, 4, 5]. However, this Hall-Petch relation cannot be extrapolated down to infinitesimally small grain sizes. Instead, it is reported to break down when grain size is reduced below a critical value (usually 10 20 nm). Below this critical size, the strength is often saturated or decreases with further reduction of grain size. This reversed relation between the strength and grain size is called inverse Hall-Petch equation. The deviation
from the Hall-Petch relation is likely due to the transition in plastic deformation mechanism from dislocation-driven in Hall-Petch to GB-mediated in inverse Hall-Petch. The dependence of strength on the grain size and the corresponding deformation mechanism change are schematically summarized in Fig. 1. Further details of bulk nanocrystalline materials are available in the entry “▶ Mechanical Properties of Nanocrystalline Metals.” As briefly stated in the Overview section, the specimen with external dimensions at the nanoscale shows different mechanical responses from those of their bulk counterpart. For example, the strengths of single-crystalline nano-pillars drastically increase when their diameters decrease below submicron scale, showing the well-known “Smaller is stronger” trend (See the entry “▶ Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures” for further details). In nanocrystalline nano-pillars, where both of the external specimen and internal grain sizes are in the nanometer range, the mechanical properties are clearly distinguishable from those of bulk nanocrystalline materials and single-crystalline nano-pillars. Figure 2 shows the 3D plot, where the strengths for nearly the same materials, Ni, are plotted as functions of nano-pillar diameter (D) and grain size (d). Quantities t and m in this plot are the shear strength and shear modulus, respectively. When D d, the sample dimension approaches to the bulk
Nano-sized Nanocrystalline and Nano-twinned Metals
1741
N
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 2 3D representation of strength of Ni as functions of nano-pillar diameter (D) and grain size (d). The quantities t and m are the shear strength and shear modulus, respectively (Reprint with permission from [3])
poly-/nanocrystalline materials, and the strength becomes independent of the sample size, D. Therefore, the strength is the function of grain size, d, only and follows the ordinary Hall-Petch and inverse Hall-Petch relations of the bulk nanocrystalline Ni. In Fig. 2, the data points (solid squares) on the vertical plane perpendicular to the D axis at D 1 show these relations. When D ¼ d, there is only one grain across the diameter, and the resultant microstructure is single crystal (another possibility is bamboo structure, where the GBs are orthogonal to the pillar axis, but this structure is not discussed in this entry). The single-crystalline nano-pillars display “Smaller is stronger” trend, where the strength increases as decreasing the pillar diameter, D. In Fig. 2, the green plane satisfies the condition of D ¼ d, and the data points on that plane (open diamonds) represent the strength of single-crystalline Ni nano-pillars as a function of pillar diameter, D (or equivalently as a function of d). The linear regression line in this log-log scale plot shows the power-law slope of 0.6. Further details on the single-crystalline nano-pillars are available in the entry “Plastic Properties of Single Crystalline at the Nanoscale” or in Refs. [3, 6]. While both of bulk nanocrystalline Ni and single-crystalline Ni nano-pillars exhibit “Smaller is stronger” trend, where the strength increases as decreasing either d or D (in bulk nanocrystalline
samples, this statement is valid only down to the critical grain size of 20 nm), the strength of nanocrystalline nano-pillars shows different aspects from both of them. In Fig. 2, the vertical plane perpendicular to the d axis at d ¼ 60 nm displays the strength of nanocrystalline Ni nano-pillars with fixed grain size of 60 nm as a function of D ranged from 1,600 to 100 nm. Unlike bulk nanocrystalline Ni or singlecrystalline Ni nano-pillars, this nanocrystalline Ni nano-pillars show “Smaller is weaker” trend, where the strength decreases with decreasing D. Possible explanation to this trend can be due to the contribution of the free surface, whose relative fraction increases with decreasing the pillar diameter. In general, the strength of an individual grain embedded within the surrounding ones (i.e., a grain in bulk polycrystalline materials) is higher than that of the same but completely freed grain. It is because for plastic deformation in bulk polycrystalline materials, all grains must deform cooperatively by matching the displacement across the GB, and it applies additional constraints to the plastic deformation. However, when the adjacent grains are removed, those constraints are released, and the grain can deform more easily. In nanocrystalline nano-pillars, the grains located at the free surface (surface grains) have less constraint than those in the pillar (inner grains) since the surface grains
N
N
1742
face the empty space at least to one direction. Therefore, the strength of the surfaces grains is lower than that of the inner grains, dividing the whole nano-pillars into hard inner core and soft outer shell. The thickness of the outer shell is approximately same with the average grain size of the nano-pillar. Therefore, when they have a fixed average grain size, the relative fraction of the soft outer shell increases as the diameter of the nanocrystalline nano-pillars decreases. As a result, the strength of the nanocrystalline nano-pillar lowers [3]. Another important observation on the nanocrystalline nano-pillars is the onset of GB-mediated plasticity at larger grain size than that of bulk nanocrystalline samples. In bulk nanocrystalline metals, activation of GB-mediated plasticity is usually correlated with transition from the Hall-Petch to inverse Hall-Petch relation, and this transition grain size for Ni is around 20 nm, as can be seen in Fig. 2. However, recent study reveals that reduction of the nano-pillar diameter to 100 nm facilitates the activation of GB-mediated process, such as grain boundary sliding and grain rotation, at 60-nm grain size, which is significantly larger than the transition size of bulk samples, 20 nm [3]. Nano-twinned Nano-pillars In general, the misorientation angles between the crystallographic planes across the ordinary GBs described in the previous section are randomly distributed, and atomic structure of those GBs is incoherent in that the lattice planes lose the crystallographic registry across the interface. However, there is a different type of internal coherent and homogeneous interface, called twin boundary (TB). Twinned microstructure usually refers to a special type of relative adjacent grain orientations, where one of the grains has the mirror-symmetric crystallographic structure of the other, and the mirror plane is called a twin boundary [1]. In this structure, one of the grains is called matrix and the other twin. Particularly, when the lattice planes do not lose the crystallographic registry across the boundary, the TB is called coherent. Because the TBs are also a special type of the GB, their strengthening mechanism shares many common factors with the ordinary GBs, i.e., TBs also interfere with the glide of dislocations. However, their influence on strengthening is known to be less pronounced because they are less effective at resisting the penetration of the moving dislocations, especially when the distance between the neighboring TBs is in micron range [7].
Nano-sized Nanocrystalline and Nano-twinned Metals
Over the last decade, the so-called nano-twinned materials (usually metals) have attracted a lot of scientific and industrial interest as promising candidates for the next generation structural and/or electrical material because of their simultaneous attainment of high strength, ductility, and electrical conductivity. In these nano-twinned metals, the nano-twinned area consists of alternative pileup of nanometer-thick planar matrix and twin grains (nano-twin lamellas). The transmission electron microscope (TEM) image of a typical nano-twinned Cu and the corresponding stress-strain curve is presented in Fig. 3 [8]. An important difference between ordinary GB and coherent TB in their roles during plastic deformation is that while the dislocation can hardly move along or across the ordinary GBs because of their disordered or discontinued atomic structure, the coherent TBs relatively easily allow dislocations to glide along and across the boundary [7]. The consequence of this difference is that dislocation mobility is highly limited in the presence of GBs, resulting in increased strength with sacrificing the ductility in regular nanocrystalline metals. On the other hand, because of its coherent nature, the TBs are capable of accommodating more various interactions with moving dislocations other than simply blocking their glide. For example, a moving dislocation can be absorbed to the TB, leaving a sessile dislocation on the interface plane, or can be dissociated into two gliding partial dislocations on the TB plane, leaving the stacking fault, or can transmit across the TB, leaving a step at the TB [7]. All of these interactions gradually degrade coherency of the TBs, hinder subsequent dislocations in interacting in the similar manners, and hence change the characteristic of coherent TB close to that of ordinary incoherent GBs. As a result, it gradually requires higher applied stress to keep the dislocations moving, and the material is workhardened [7]. Usually, the increase of work-hardening coefficient correlates with higher resistance to instability, such as necking [4, 9], so it can contribute to the effective increase of ductility. Most of nano-twinned samples fabricated to date are polycrystalline in the sense that the nano-twin lamellas are embedded within polycrystalline grains as sub-grain structures as can be seen in Fig. 3. Properties measured in the polycrystalline nano-twinned samples are convoluted by the effects from many different microstructures like GBs, and differently orientated TBs, etc. Therefore, the specific role of TBs and
Nano-sized Nanocrystalline and Nano-twinned Metals
a
1743
N
b 1200
True stress (MPa)
1000
C
800 B 16 mm
600
Original
A
A C
IGC-Cu
400
200 CG-Cu 0
2
4
6
8
10
12
14
16
18
True strain (%)
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 3 (a) TEM observations of typical microstructure in nano-twinned Cu sample. (b) Typical tensile stress-strain curves for the as-deposited Cu samples with average nano-twin thickness
of 93, 30, and 15 nm for samples A, B, and C, respectively. For comparison, the stress-strain curves for coarse-grained polycrystalline Cu sample (CG-Cu) and nanocrystalline Cu (IGC-Cu) are also displayed (Reprint with permission from [8])
nano-twinned lamellae in the enhanced mechanical properties is far from being understood. However, recent studies on the nano-twinned nano-pillars, where the specimen contains well-aligned TBs only, offered an insight into understanding the specific role of TB during plastic deformation. The TEM images in Fig. 4a–c show the Cu nano-pillars with 500-nm diameters, whose microstructures are nanocrystalline (NC), purely nano-twinned (NT), and nanocrystalline nanotwinned (NCNT), respectively [10]. NC nano-pillars only contain the randomly distributed ordinary GB with average grain size of 160 nm, while NT nanopillars have only coherent TBs orthogonal to the pillar axis. Average twin thickness of NT nano-pillars is 8.8 nm. NCNT nano-pillars contain both of GBs and TBs, where nano-twin lamellae are embedded within the regular grains. Average grain size and twin thickness of NCNT nano-pillars are 250 and 6.7 nm, respectively. The strengths of those Cu nano-pillars measured by uniaxial nano-compression experiments are shown in Fig. 4d, where SC stands for singlecrystalline Cu nano-pillars. These results suggest that in Cu nano-pillars with 500-nm diameter, the two different interfaces, ordinary GB and coherent TB, appear to affect the strengths differently. Given that
strength of NC is lower than that of SC by 20%, the ordinary GBs seem to lower the strength of the nanopillars. On the other hand, strength of NT is higher than that of SC by a factor of 3, implying that orthogonally aligned TBs drastically increase the strength of nano-pillars. Comparing the strengths of NC and NCNT, TBs also increase the strength even in the polycrystalline case. The influence of GBs and TBs on the strength of 500-nm Cu nano-pillars is summarized in the diagram in Fig. 4.
Fabrication of Nano-pillars via Electroplating Traditionally, electroplating is a widely used metallic surface coating technique on a conductive surface. During electroplating, the metallic ions in the electrolyte migrate to the surface of the cathode under externally applied electrical voltage between the cathode and anode, and deposit a thin metallic layer on the cathode. Basically, the fabrication of nano-pillars shares the same physics and chemistry with this traditional electroplating technique. However, rather than forming a metallic layer of uniform thickness, it requires the ions to deposit only onto prescribed
N
N
1744
Nano-sized Nanocrystalline and Nano-twinned Metals
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 4 (a) TEM bright-field image and diffraction pattern (inset) of a typical nanocrystalline (NC) Cu nano-pillar. TEM dark-field images and diffraction patterns (upper-right insets) of typical (b) nano-twinned (NT) and (c) nanocrystalline nano-twinned (NCNT) Cu nano-pillars. The lower-right inset
in (c) is the zoomed-up image of the region within the white box. (d) The average yield strength of each microstructure. (e) Illustration showing the effect of grain and nano-twin boundaries on the strength of 500-nm Cu nano-pillars (Reprint with permission from [10])
locations. This can be achieved by using the prepatterned template as the cathode. The template consists of the chemically etchable (e.g., polymethylmethacrylate (PMMA)) patterned nonconductive sacrificial layer on top of a conductive metallic thin film deposited on an insulating substrate (Si, SiO2 or sapphire). A pattern of desired shapes can be written
via e-beam lithography, and the resist is subsequently developed to generate arrays of through-holes to the underlying conductive metallic film. The underlying thin metallic layer provides an electric-conducting path for the electroplating, and the ions deposit onto the metallic layer through the open spaces in the patterned template. After electroplating, the PMMA
Nano-sized Nanocrystalline and Nano-twinned Metals
Silicon (1) Evaporate Au/Ti seed layer
Silicon (2) Spin and bake PMMA resist
1745
Nano-sized Nanocrystalline and Nano-twinned Metals, Table 1 Polymethylmethacrylate (PMMA) resist and electron beam exposure parameters for template fabrication [11]. The quantity d is the diameter of the circular patterns, which is effectively the same as the diameter of the electroplated nanopillars, and L is the thickness of PMMA (Reprint with permission from [11]) Pillar dimensions
Silicon (3) Electron beam lithography and development Compression samples
Tension samples
Silicon (4) Electroplate metal nanopillars
Silicon (5) Strip PMMA in acetone
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 5 Schematic representation of template fabrication and electroplating (Reprint with permission from [11])
sacrificial layer can be stripped off in acetone. Therefore, the resulting electroplated structure becomes a negative of the pattern. The nonconductivity of the substrate prevents the ions from depositing to the backside of the template. The schematic illustration of this process is presented in Fig. 5. The specific details of the template fabrication are available in [11]. For example, a 100-nm thick Au seed layer on top of the 20-nm thick Ti adhesion layer is deposited on the Si wafer by electron beam evaporation. The Au layer provides the electric-conducting path for the electroplating and Ti layer enhances the adhesion of the Au layer to the Si substrate. The PMMA resist diluted in anisole is spin-coated on top of the Au seed layer. The thickness of the PMMA resist determines the maximum height of the nano-pillars, as illustrated in Fig. 5. Table 1 shows the list of some examples of the resist parameters and corresponding layer thicknesses [11]. The fabricated template can be used as the part of the cathode during the electroplating. The cathode is
N
d (nm) 25 50 100 150 200 250 500 750
L (nm) 100 200 400 600 800 1,000 2,000 3,000
Resist parameter Dilution Speed (%) (rpm) 4 2,500 4 2,500 7 3,000 7 1,500 9 4,500 9 2,000 11 2,500 11 1,500
Exposure parameters Dose (mC/cm2) 2,600 1,600 1,450 1,600 1,550 1,500 1,600 1,750
split into two parts: the template and a Cu-coated dummy chip. The purpose of the dummy chip is to precisely control the current density at the cathode (J), which is defined as the total current divided by the cross-sectional area of the cathode. With the template alone, the cross-sectional area, which is the sum of the area of the patterns on the template, is extremely small (on the order of nm2 or mm2) so that the current density can sharply change by a small error in the area measurement. By adding the dummy chip whose crosssectional area is a few orders of magnitude larger than the patterns, the total cross-sectional area of the anode becomes insensitive to the small error in the area measurement, so that the current density at the cathode can be reliably controlled. As the anode material, either insoluble platinized titanium mesh or soluble metallic plate can be used, depending on the material to be plated. Microstructure of the electroplated nanopillars is highly dependent on the plating condition, including composition of electrolyte, waveform of the applied current, and current density at the cathode. For example, in order to fabricate the Cu nano-pillars, the platinized titanium mesh was used as the anode material, and rectangular pulsed current was applied between the cathode and anode, as shown in Fig. 6a, where the current density is maintained at Jpeak during on time, ton and becomes zero during off time, toff [10]. Electrolyte composition is also an important factor to control the microstructure of the Cu nano-pillars. In Table 2, the electroplating parameters to create the NT,
N
1746
Nano-sized Nanocrystalline and Nano-twinned Metals
Nano-sized Nanocrystalline and Nano-twinned Metals, Fig. 6 (a) The waveform of the pulsed electroplating current. (b) SEM image of a typical as-fabricated Cu nano-pillar (Reprint with permission from [10])
a
Current density
N
b t on
J peak
t off 0 Time 500 nm
Nano-sized Nanocrystalline and Nano-twinned Metals, Table 2 The electroplating condition for the fabrication of 500-nm Cu nano-pillars for each microstructure (Reprint with permission from [10])
Microstructure Nano-twinned (NT) Nanocrystalline (NC) Nanocrystalline Nano-twinned (NCNT)
Electrolyte composition 125 g/L of CuSO4∙5H2O + 50 g/L of H2SO4 100 g/L of C6H8O7∙H2O + 28 g/L of CuSO4∙5H2O + 50 g/L of (NH4)2SO4 50 g/L of C6H8O7∙H2O + 28 g/L of CuSO4∙5H2O + 50 g/L of (NH4)2SO4
NC, and NCNT Cu nano-pillars are summarized, including the electrolyte composition, Jpeak , ton, and toff [10]. The Scanning Electron Microscope (SEM) image of a typical as-fabricated pillar is shown in Fig. 6b. Fundamentally, all conventional electroplatable metals can be also fabricated into the nano-pillar forms utilizing the method in the previous paragraph. However, determining electroplating parameters for specific microstructure, especially the waveform of the applied current, can be challenging. While specific details have to be developed for each material, the general guideline is that (1) the grain size inversely scales with the plating overpotential, and hence reducing forward current density often yields the larger grain size. (2) Sometimes, using pulsed current is efficient in producing larger grain size as it periodically reduces the depletion region near the cathode, and consequently lowers the overpotential. The recipes to create single crystalline Cu [11], polycrystalline Au [11], and polycrystalline In [12] nano-pillars are available elsewhere.
Jon (A/cm2) 0.4 1.2
ton (ms) 1 1
toff (ms) 100 100
Feature sizes Twin Grain thickness size (nm) (nm) N/A 8.8 160 N/A
0.72
1
100
250
6.7
Cross-References ▶ Mechanical Properties of Nanocrystalline Metals ▶ Nanomechanical Properties of Nanostructures ▶ Plasticity Theory at Small Scales ▶ Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
References 1. Callister, W.D.: Materials Science and Engineering: An Introduction. Wiley, New York (1997) 2. Hirth, J.P., Lothe, J.: Theory of Dislocations. Wiley, New York (1982) 3. Greer, J.R., De Hosson, J.T.M.: Plasticity in small-sized metallic systems: intrinsic versus extrinsic size effect. Prog. Mater. Sci. 56, 654–724 (2011) 4. Courtney, T.H.: Mechanical Behavior of Materials. McGraw-Hill, Boston (2000) 5. Meyers, M.A., Mishin, Y., Benson, D.J.: Mechanical properties of nanocrystalline materials. Prog. Mater. Sci. 51, 427–556 (2006)
Nanostructure Field Effect Transistor Biosensors
1747
6. Zhu, T., Li, J.: Ultra-strength materials. Prog. Mater. Sci. 55, 710–757 (2010) 7. Lu, K., Lu, L., Suresh, S.: Strengthening materials by engineering coherent internal boundaries at the nanoscale. Science 324, 349–352 (2009) 8. Shen, Y.F., Lu, L., Lu, Q.H., Jin, Z.H., Lu, K.: Tensile properties of copper with nano-scale twins. Scripta Mater. 52, 989–994 (2005) 9. Dieter, G.E.: Mechanical Metallurgy. McGraw-Hill, New York (1988) 10. Jang, D., Cai, C., Greer, J.R.: Influence of homogeneous interfaces on the strength of 500 nm diameter Cu nanopillars. Nano Lett. 11, 1743–1746 (2011) 11. Burek, M.J., Greer, J.R.: Fabrication and microstructure control of nanoscale mechanical testing specimens via electron beam lithography and electroplating. Nano Lett. 10, 69–76 (2010) 12. Lee, G., Kim, J., Budiman, A.S., Tamura, N., Kunz, M., Chen, K., Burek, M.J., Greer, J.R., Tsui, T.Y.: Fabrication, structure and mechanical properties of indium nanopillars. Acta Mater. 58, 1361–1368 (2010)
N
Nanostructure Field Effect Transistor Biosensors Jason Li1, Steve To2, Lidan You1 and Yu Sun3 1 Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada 2 Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada 3 Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering and Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada
Synonyms Bio-FET; Chem-FET; CNT biosensor; CNT-FET; DNA FET; Nano-FET; Nanowire FET biosensor
Nano-sized Particle ▶ Gas Phase Nanoparticle Formation
Definition
Nanoslits ▶ Nanochannels for Nanofluidics: Fabrication Aspects
Nanosoldering ▶ Nanorobotic Spot Welding
Overview
Nanospring ▶ Nanorobotics Nanostructures
for
NEMS
Nanostructure field effect transistor (nano-FET) biosensors are nanoscale electronic devices used to detect and measure the presence and concentration of specific biological molecules within a given sample solution. These sensors utilize a quasi-one-dimensional semiconducting nanostructure as the active sensing element to detect the presence of local electric fields produced by charged biological molecules.
Using
Nanostructure ▶ Structure and Stability of Protein Materials
Helical
Biomolecular Detection Technologies Methods for detecting and measuring the presence, abundance, and activity of biological molecules, such as proteins, DNA, and RNA, enable the investigation of basic biological function, molecular diagnosis of diseases, and development of therapeutic treatments. The combination of polyacrylamide gel–based electrophoresis separation (and related blotting techniques) and absorption-based chromogenic dye staining (e.g., Coomassie brilliant blue or silver staining) has traditionally formed the core technologies for protein and
N
N
1748
Nanostructure Field Effect Transistor Biosensors
a
b Gate Electrode
PDMS channel
Fluid outlet
Source Electrode Receptors Drain Electrode
Fluid inlet
Chip Semiconducting Nanostructure Oxide Layer
Substrate
Chip carrier (Platform)
Contact pad
Nanostructure Field Effect Transistor Biosensors, Fig. 1 (a) Schematic of a single nano-FET biosensor. (b) Schematic of a nano-FET biosensor chip with integrated microfluidic
sample delivery system. The nano-FET biosensor array is housed within the microfluidic system (Adapted with permission from Macmillan Publishers Ltd: [5], copyright 2006)
nucleic acid detection. Subsequent development of fluorescence- and luminescence-based labeling stains offered the opportunity for multicolor labeling, making multiplexed analysis possible. However, these methods were limited to the detection of biological molecules on gels and blots. Immuno-labeling of fluorescent and luminescent probes allowed specific protein detection and facilitated the emergence of immunoassays able to detect and quantify the abundance of a target protein molecule in solution. These optical detection methods suffer from large sample volume consumption, high reagent costs, and long assay times which, when combined with the drawback of using labeling molecules, preclude the application of these technologies to many laboratory and clinical applications (e.g., real-time protein quantification and point-of-care clinical diagnostics). A number of novel label-free protein-detection strategies have since been developed in an effort to overcome some of the aforementioned limitations. Many of these strategies leverage nonoptical transduction modalities to circumvent the limitations imposed by optical probes. Examples include micro-/nanosurface plasmon resonance sensors [1], micro- and nano-cantilevers that translate biomolecular interactions into mechanical deformations [2], nanostructured field effect transistors (FETs) that measure intrinsic biomolecular charge [3], and electrochemical sensors that translate biomolecular adsorption to changes in redox current [4]. Of these technologies, nanoFET biosensors offer unique device characteristics including ultrasensitive, real-time, and label-free measurement capability, compact physical form,
CMOS-compatible and simple on-chip integration with circuitry and microfluidic systems, direct electrical readout, and multiplexed detection capability. Structure Nano-FET biosensors are composed of a semiconducting quasi-one-dimensional nanostructure (e.g., nanowire or nanotube) bridging the source and drain electrodes (Fig. 1). The nanostructure serves as the transistor’s channel and functions as the sensing element of the device. The biosensor structure deviates slightly from that of the transistor in the placement and structure of the gating electrode. While the gate electrode in a nano-FET is typically located directly above the channel, separated by a thin dielectric layer, the nano-FET biosensor uses a solution gate electrode that is typically composed of platinum or Ag/AgCl and is immersed within the aqueous solution being measured. Furthermore, the nanostructure surface is functionalized with a layer of analyte-specific receptors to impart a degree of specificity to the sensor. This may be accomplished via chemical bonding using bifunctional linker molecules, or through physical interaction due to adsorption, van der Waals forces, and weak electrostatic forces. While the specific nanostructure employed may vary in shape, size, and composition, it is important that these structures be quasi-one-dimensional and exhibit semiconductive material properties. Typical nanostructures used in the construction of nano-FETs include single and multi-walled carbon nanotubes, graphene, and silicon nanowires. Devices constructed from other semiconductive materials such as
Nanostructure Field Effect Transistor Biosensors
N
1749
a
b 2.0 1.60 Conductance (μS)
Conductance (μS)
1.8 1.55
1.50
1.45
NW1 1.6 1
2
3
4
1.4
5
NW2
1.2 1.0
1.40 0
30
60
90
120
0
1,000
2,000
Time (min)
3,000
4,000
5,000
Time (s)
Nanostructure Field Effect Transistor Biosensors, Fig. 2 Real-time nanowire FET sensing results. (a) Conductance versus time data recorded following alternate delivery of prostate-specific antigen (PSA) and pure buffer solution (1 mM phosphate (potassium salt) containing 2 mM KCl, ph 7.4). Subsequent PSA concentrations were 5 ng ml1, 0.9 ng ml1, 9 pg ml1, 0.9 pg ml1, and 90 fg ml1, respectively.
(b) Complementary sensing of PSA using p-type (NW1) and n-type (NW2) nanowire FET devices. Vertical lines correspond to addition of PSA solutions of (1, 2) 0.9 ng ml1, (3) 9 pg ml1, (4) 0.9 pg ml1, and (5) 5 ng ml1. Arrows on the bottom represent the injections of sensing buffer solution (Adapted with permission from Macmillan Publishers Ltd: [5], copyright 2006)
metal-oxides and nanostructures of various shapes including nanotubes, nanobelts, nanorods, nanoribbons, and nanobars have also been reported.
the current passing through the nanowire (ID) for a given source-drain voltage (VDS) is therefore:
Physics Nano-FET biosensors utilize semiconductive nanostructures to detect the presence of charged molecules in the immediate vicinity around the nanostructure surface. The sensing mechanism is based on the observation that an externally applied electric potential is capable of modulating the number of free charge carriers within a semiconductive material, thus leading to a measurable change in electrical conductance. For a cylindrical semiconducting nanostructure with radius R and length L, the charge carrier concentration may be described as a function of the applied potential: n0 ¼
CðVG Vt Þ qpR2 L
(1)
where C is the gate capacitance, q is the elementary charge, VG is the gate voltage, and Vt is the threshold voltage [6]. As the conductance of the semiconducting nanowire is described by: 2 pR (2) G ¼ nqm L
ID ¼
pR2 n0 qmVDS L
(3)
where m is the carrier mobility. Charged functional groups present on the surface of bound biological molecules such as proteins and nucleic acids provide the necessary electric potential required to induce a conductance change in the semiconductive nanostructure, thus enabling the direct electrical detection of such molecules using nano-FET devices (Fig. 2). The strength of a molecule’s surface charge is a function of electrolyte pH and is characterized by the molecule’s isoelectric point. The simplified analysis qualitatively illustrates the physical sensing mechanism behind nano-FET biosensors. However, this treatment does not accurately describe the complexities of the nano-FET biosensor system. Equations (1–3) describe the electrical behavior of a cylindrical semiconducting crystal subjected to a uniform charge density over the entire nanostructure surface. This assumption fails to take into consideration the nonuniform nanostructure surface charge distribution resulting from discrete biomolecular binding events. The discrete gating mechanism by individual
N
N
1750
Nanostructure Field Effect Transistor Biosensors
molecules in solution introduces several additional phenomena that significantly influence the sensor’s response including depletion of analyte from solution, analyte transport mechanisms to the nanostructure surface, and analyte-receptor binding and unbinding kinetics. These effects become particularly important in the detection of rare analytes. Equation (3) also neglects the influence of electrostatic screening – the spatial arrangement of mobile ions around a charged molecule to effectively neutralize that charge. Importantly, target analytes must be sufficiently close to the sensor surface in order to elicit a conductance change due to electrostatic screening by mobile charges around the analyte molecule (e.g., ions in solution and charge carriers in the semiconducting crystal). The Debye screening length, the distance away from a charged molecule at which the electrostatic potential is diminished to zero due to screening, defines the required proximity. Debye length (aqueous electrolyte): 1 k1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8plB NA I
(4)
where I is the ionic strength of the electrolyte (mole/ m3), NA is the Avogadro number, and lB is the Bjerrum length of the medium (0.7 nm for water). Debye length (silicon): sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e0 er k B T LD ¼ (5) q2 N D where er is the dielectric constant of the semiconducting material, kB is the Boltzmann’s constant, T is the absolute temperature in Kelvin, q is the elementary charge, and Nd is the density of donors in a substrate. The implications of charge screening and a finite Debye length make several important predictions on nano-FET biosensing performance. Namely, these relations suggest a conductance dependence on charge carrier concentrations (ionic strength of an electrolyte solution and doping density of a semiconductor), size of the semiconducting crystal, strength of the molecule’s net surface charge, and the nanostructure-analyte separation distance. Inclusion of these phenomena into a unified analytical model that accurately describes experimental nano-FET biosensor response is an active area of research.
Performance Parameters Nano-FET Sensitivity Nano-FET sensitivity is the induced change in device conductance (DGsd), upon exposure to a certain biomolecular stimulus for a constant source-drain voltage (Vds). Sensitivity is defined as: Sensitivity ¼
DGsd G0
(6)
where DGsd is the direct conductance change observed in sensing experiments and G0 is the initial device conductance [7]. Because DGsd depends on the specific nanowire parameters (e.g., diameter, mobility, etc.) this value does not reflect the intrinsic sensitivity. Instead, it is more meaningful to characterize the sensitivity using a dimensionless parameter (6). Normalization enables comparison across devices with different physical dimensions. Sensitivity is closely related to three additional parameters: detection limit (the smallest analyte concentration that can be measured, which is dictated by the system signal-to-noise ratio); measurement resolution (the smallest detectable change in analyte concentration, which is dictated by noise levels); and measurement range (the measurable analyte concentrations, with the upper and lower bounds defined by nanostructure saturation and the detection limit, respectively). A higher sensitivity directly translates into lower detection limit, finer measurement resolution, and a broader measurement range. Typical protein and DNA detection limits reported in literature are in the pM (picomolar) to fM (femtomolar) range [8, 9]. Single virus detection has also been achieved [10]. Specificity
Specificity is a measure of a biosensors relative responsiveness to the target analyte as compared to all other molecules present within the sample solution. Very specific sensors should possess high sensitivity toward the target analyte and low sensitivity toward all other biomolecules. This attribute is imparted onto a nano-FET biosensor by functionalizing the nanowire surface (either covalently or noncovalently) with a confluent monolayer of analyte-specific receptors/ ligands that bind specifically to the molecule of interest.
Nanostructure Field Effect Transistor Biosensors
a
1751
b
Antibody
Debye length
Source
N
Aptamer
Drain Linker
SiO2
CNT
Si
Si
Nanostructure Field Effect Transistor Biosensors, Fig. 3 Schematic representation of nano-FET biosensors functionalized with (a) antibodies and (b) aptamers. Aptamers (and antibody fragments) are smaller than whole antibodies. The use of these capture probes thus results in increased nano-FET
sensitivity. For certain experimental conditions, aptamers may serve as the ideal probe candidate as they are shorter than the system Debye length while antibodies are not (Reprinted with permission from [14]. Copyright 2007 American Chemical Society)
Response Time
may be obtained in a situation where the effective screening length is much larger than the radius of the nanowire. Improved device sensitivity can thus be obtained through careful design and optimization of several device parameters that influence the effective screening length:
Response time is the time it takes for the biosensor system to obtain a stable biomolecular concentration measurement upon introduction of the sample solution to the sensor surface. As with any dynamic system, this characteristic is captured in the step response of the system by the settling time. For a nano-FET biosensor, this parameter is influenced by the analyte-receptor binding and unbinding kinetics, transport of analyte molecules to the sensor surface, concentration and degree of mixing of the analyte solution, and the electrical response behavior of the measurement system. The typical response time of nano-FET biosensors is on the order of minutes.
Key Research Findings Active areas of research in the field of nano-FET biosensing include: (1) optimization of biosensor performance based on the systematic examination of how fundamental device parameters affect sensitivity, specificity, and response time; (2) development of robust large-scale fabrication strategies for the manufacturing of biosensor arrays; and (3) development of an analytical model capable of accurately describing and predicting nano-FET biosensor performance. Performance Optimization Minimizing the Effects of Charge Screening As the effect of charge screening acts to minimize the total nanostructure volume gated by surface charges, thus reducing device sensitivity, strategies that increase screening length are expected to improve device sensitivity. The maximum device sensitivity
Diameter Reduction of the nanostructure diameter will dramatically increase the surface-to-volume ratio, thus improving device sensitivity. This is the primary advantage of decreasing the channel size and explains why nanostructure FET sensors exhibit significantly higher performance as compared to traditional planar FET devices. Dielectric Thickness Silicon nanowires form a native oxide layer (1–2 nm) on their surface that serves as a dielectric layer reducing the electric field strength acting on the nanowire itself. This layer also increases the separation distance between surface-bound molecules and the semiconductive core of the nanowire. Thus, a thinner oxide layer is expected to improve sensitivity. Complete removal of the native oxide layer surrounding silicon nanowires has been shown to increase device sensitivity [11, 12]. Functionalization Scheme and Receptor Size Decreasing the separation distance between the receptor-bound analyte molecule and the nanostructure surface minimizes the effect of electrolyte screening. While antibodies are commonly used to selectively bind analyte molecules to the biosensor surface, the use of smaller capture probes such as antibody fragments or aptamers in place of antibodies can increase device sensitivity without the loss of selectivity (Fig. 3) [13, 14]. Similarly, utilizing shorter
N
N
1752
bifunctional linker molecules to attach the capture probes to the nanostructure surface may also increase device sensitivity. The choice of functionalization strategy may also dictate the capture probe density and coverage on the nanostructure surface. A poor capture probe density can result in the nonspecific adsorption of molecules onto unmodified areas of the nanowire surface and may also limit the upper detection limit of the sensor due to capture probe saturation. Electrolyte Ion Concentration and pH The ionic strength of the electrolyte solution influences the efficiency of charge screening as the density of free ions is directly related to size of the electric double layer formed around a charged molecule. Higher ionic strength electrolytes are thus able to screen a charged molecule over a shorter distance. It is therefore advantageous to perform biosensing experiments in low-ion conditions to maximize sensitivity [15]. However, this is not always feasible as many sensing applications are performed with physiological samples (e.g., serum, whole blood, etc.). The pH of a solution also influences device sensitivity by changing the effective electrical charge on a biological molecule. Biological molecules such as proteins and nucleic acids contain both acidic and basic functional groups that may be positively or negatively charged depending on the availability of protons (H+) present in solution. The pH for which the negative and positive charges on the molecule are balanced, resulting in zero net charge, is defined as the isoelectric point (pI). The net charge on a molecule is therefore a function of the electrolyte pH and can become increasingly more positively or negatively charged as the term jpI-pHj increases. While tuning this parameter may be useful in specific applications, the pH of a sample is usually fixed at physiological levels (pH ¼ 7.4). Carrier Concentration The screening length within a semiconductive nanowire dictates the depth of penetration into the nanostructure for which surfacebound charged molecules are able to gate. For example, the screening length in a silicon nanowire operating with a typical carrier concentration of 1018–1019 cm3 is 1–2 nm, suggesting that only a small portion of the total nanowire volume (a 1–2-nm thick layer at the nanowire surface) is
Nanostructure Field Effect Transistor Biosensors
affected by surface charges. Decreasing the charge carrier concentration will therefore increase device sensitivity by increasing the screening length and the total gated volume. This can be achieved by varying either of two device parameters: the nanostructure doping density or the bias voltage applied through a gate electrode on the nano-FET biosensor. In terms of semiconductors, lower doping densities result in longer screening lengths and therefore produce higher sensitivity devices [16]. It is instructive to note that the optimum biosensor sensitivity is obtained when the device is operating in the subthreshold regime, as under these conditions, the transconductance is at a maximum [7, 17, 18]. Optimization of Analyte Delivery Efficiency
Biomolecular detection using nano-FET sensors requires the delivery of analyte to the sensor surface. The efficiency of this process can therefore limit the performance of nano-FET biosensors. Several phenomena that influence this process include diffusive and convective molecular transport mechanisms of analyte to the biosensor surface, depletion of free analyte from solution, and nonspecific analyte adsorption and binding. Molecular Transport As with many surface-based biosensor systems, the transport of analyte in solution to the sensor surface plays a crucial role in governing binding kinetics and ultimately sensor performance. This is especially true when analyzing samples with dilute concentrations of target molecules and/or employing microfluidic systems for efficient and automated handling of small sample solution volumes. A number of competing physical processes influence target transport to the sensor surface. Analyte molecules suspended in solution may diffuse randomly within the solution, be convected along with flowing fluid, bind to adjacent surface-bound receptors, or subsequently unbind to reenter solution. The binding (or collection) of biomolecules onto the sensor surface simultaneously depletes the surrounding solution to form a so-called depletion region around the sensor (Fig. 4). As the depletion region grows in size, the analyte diffusive flux decreases and thus the collection rate becomes slower. Through finite element analysis, Sheehan and Whitman determined that the size and shape of the sensor profoundly affects the total analyte flux to the sensor surface (Fig. 5a) and further argues
Nanostructure Field Effect Transistor Biosensors
U
0.9 0.8
Bound analyte (b/beq)
1.0 2 μm 10 nm
0.7
Time (min) 0 7 13 20 26 33 39 46 52 59 65
Bound analyte (molecules)
Nanostructure Field Effect Transistor Biosensors, Fig. 4 The steady-state depletion zone around a nanowire sensor from a pure mass-transport problem. The depletion zone is thick compared to the nanowire, but is substantially thinner than the channel itself (ds LPes½ ¼ 100 nm). Reaction-limited binding onto the nanowire follows the Langmuir binding curve (inset top right) (Adapted with permission from Macmillan Publishers Ltd. Nature Biotechnology [20], copyright 2006)
N
1753
0.9 0.0036 0.0032 0.8 0.7 0.0028 0.6 0.0024 0.5 0.0020 0.4 0.0016 0.3 0.0012 0.2 0.0008 0.1 0.0004 0 0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3 Dimensionless time (t/τCRD)
0.6 200 nm
0.5 1 μm 0.4
100 nm
0.3 −200 nm −100 nm
0.2
100 nm 200 nm
0.1 0
a
b 100
1 week
100th
1 day
10th
1 hour
1 min 10n
1st m o
lecule
Total Flux (#min)
Accumulation Time
1 month
us
m 0μ
i rad
1
1 μm
100 nm
10
10 nm 1 nm
1 100n
1μ 10μ sensor radius (m)
100μ
0.1
1
10 100 Volumetric Flow (μL/min)
1000
Nanostructure Field Effect Transistor Biosensors, Fig. 5 (a) Theoretical calculation of the time required for a 10-mm-long hemicylindrical sensor to accumulate 1, 10, and 100 molecules due to diffusive molecular transport. The sensor lies at the bottom of a channel whose width is equal to the sensor’s length and which is filled with a 1 fM analyte solution.
(b) Finite element analysis (points) and theoretical calculation (lines) of the total flux of molecules to a hemicylindrical sensor in a microchannel (800 mm wide, 100 mm high). The total flux plotted is the steady-state value at the given volumetric flow rate for a 1 fM analyte concentration (Adapted with permission from [19]. Copyright 2005 American Chemical Society)
that without directed transport of biomolecules, individual nanoscale sensors would be limited to picomolar order sensitivity for practical time scales (hours to days) [19]. While the diffusive depletion zone grows indefinitely, albeit at an ever-decreasing rate, introduction of convective flow (assumed to be laminar) halts this
growth resulting in a steady-state depletion zone with a length scale defined by the balance between the convective analyte flux delivered to the depletion zone boundary and the diffusive flux out of the depletion zone. By adjusting the convective flow rate through the channel, it is possible to tune the size of the depletion layer and thus control the rate at which
N
N
1754
molecules are delivered to the sensor surface (Fig. 5b). However, Squires et al. argue that the total mass transport varies only weakly with the flow rate as flow rates must be increased 1,000-fold to enhance flux by a factor of 10 [20]. The treatment thus far has assumed instantaneous analyte-receptor coupling upon transport of target molecules to the sensor surface. This represents one of two sensing regimes that can be described as being “transport-limited” as the availability of analyte adjacent to the sensor surface limits sensor response. A second biosensing regime arises when considering the kinetics of analyte-receptor association and dissociation (assumed to be first-order Langmuir kinetics) [20]. If delivery of target molecules to the sensor surface occurs at a faster rate than the net analytereceptor association rate, then the transport is said to be “reaction limited.” In general, the reaction kinetics is determined by the fidelity of the immobilized reagents. While each of the two regimes imposes an independent maximum target collection rate, the “reaction-limited” condition is most desirable for nano-FET operation as, for a given device, biomolecular sensing is purely a function of analyte concentration. The sensor size and flow rate should thus be adjusted to ensure nano-FET operation within this regime. Alternatively, introduction of turbulent flow or fluid mixing through innovative microfluidic design may also be employed to overcome the limitations imposed by convective and diffusive molecular transport [8]. Depletion of Free Analyte As silicon and metaloxide nanostructures share similar surface chemistries with several popular microfabrication substrates (e.g., silicon wafers, glass, quartz), surface modification techniques used to attach receptor molecules to the nanostructure surface can also functionalize the device substrate. In this situation, the total device surface area exposed to the sample solution during biomolecular detection experiments is proportional to the number of receptors available for reaction. Analyte binding events with substrate-bound receptors do not induce a change in nanostructure conductance. Rather, these reactions serve only to deplete free analyte molecules from solution, thus reducing the number of analyte molecules available for reaction with nanostructurebound receptors. Minimizing the total surface area exposed to sample solutions is, therefore, important
Nanostructure Field Effect Transistor Biosensors
Streptavidin Source +
Drain −
Biotin
Polymer
+
Gate
Nanostructure Field Effect Transistor Biosensors, Fig. 6 Schematic of a nanotube field effect transistor coated with a polymeric functional layer. Co-functionalization with a molecular receptor (biotin) and a nonreactive polymer (e.g., polyethylene glycol or polyethylene oxide) prevents nonspecific adsorption of molecules onto the nanotube surface (Adapted with permission from [22]. Copyright 2003 American Chemical Society)
for maximizing device sensitivity. This effect may be circumvented using nanostructure-specific surface modification chemistries. Nonspecific Adsorption Random adsorption of biological molecules onto the nano-FET sensor produces a false-positive measurement signal. This effect may be minimized by grafting a dense layer of long nonreactive polymer chains (e.g., polyethylene glycol) onto the nanostructure surface alongside receptor molecules (Fig. 6). The polymer chains prevent molecular adsorption directly onto the nanostructure surface by acting as a spacer while the co-functionalized receptors are able to bring analyte molecules into intimate contact with the nanostructure through molecular binding. The length of the polymer chains used for this application should exceed the screening length of the system. Nonspecific adsorption of random molecules is especially problematic when working with samples containing numerous types of biological molecules such as whole blood. One strategy that enables detection of a specific analyte within a whole blood sample involves the use of a sample pre-purification procedure before nano-FET measurement [21]. In this method, the whole blood is introduced into a chamber in which analyte-specific receptors have been immobilized on to the chamber walls. Upon binding of free analytes to the
Nanostructure Field Effect Transistor Biosensors
surface-bound receptors, the whole blood solution is flushed out of the chamber and discarded. The receptors are then detached from the chamber surfaces and transported to the nano-FET biosensor for analysis. Large-Scale Methods of Fabrication Reliable large-scale fabrication of nano-FET biosensors remains to be a significant technological challenge, impeding the adoption of this technology in medical diagnostic and biomolecular detection applications. Unlike traditional microfabrication methods for producing micrometer-level structures, the difficulties encountered in nano-FET biosensor fabrication arise from the process of establishing electrical contact with nano-sized structures in a high-throughput and reproducible manner. Early methods for manufacturing nano-FETs involve (1) the growth of individual nanostructures using a chemical vapor deposition process, (2) transfer of as-grown nanostructures from the growth substrate onto a device substrate (a process which results in the random placement and orientation of the deposited nanostructures), and (3) identification of a suitable nanostructure on the device substrate followed by the patterning of metallic electrodes onto the ends of that nanostructure using a mask-less nano-lithographic tool (e.g., electron beam lithography or focused ion beam lithography), which has a low throughput and is only suitable for proof-of-principle use. Recently devised batch-fabrication strategies for the wafer-scale production of nano-FET arrays can be broadly categorized into “bottom-up” or “top-down” fabrication methods. Bottom-Up Fabrication Methods
One strategy for bottom-up fabrication is to make use of pre-synthesized freestanding nanostructures using epitaxial growth methods such as chemical vapor deposition. This strategy affords several advantages over top-down and other bottom-up approaches including (1) strict control over nanostructure dimensions and electrical properties, (2) the ability to assemble nano-FET devices onto flexible and transparent substrates, and (3) the ability to incorporate nanostructures with unique chemical compositions and architecture into nano-FET devices. Successful incorporation of these nanostructures into nano-FET devices relies primarily on the ability to transfer, align, and position the as-grown nanostructures onto a device
1755
N
substrate. Contacting electrodes may then be patterned using conventional photolithography to form individual sensors or sensor arrays. Flow Alignment Flow alignment is a method that makes use of flowing fluid to align suspended nanostructures along a single orientation in the direction of flow [23]. By passing a nanostructure suspension through a microfluidic structure formed between a poly(dimethylsiloxane) mold and the device substrate, free-flowing nanostructures assemble onto the device substrate due to surface forces and remain relatively aligned parallel to the direction of flow. The resulting degree of alignment is a function of the flow rate as increased flow velocities result in a narrower angular distribution among deposited nanostructures. Furthermore, varying the total flow duration dictates the nanostructure deposition density, with longer flow durations favoring higher nanostructure densities. This method is useful in controlling the degree of nanostructure alignment and the deposition density, thus allowing subsequent electrode formation. Nano-FET fabrication can therefore be achieved by patterning a pair of electrodes parallel to the flow direction using conventional microfabrication processes. For this physical arrangement, minimizing the electrode pair separation distance increases the likelihood of a bridging nanostructure between the electrode pair; however, the exact number of bridging nanostructures resulting from this process is probabilistic. Nanostructure Contact Printing Nanostructure contact printing is a method that provides similar results to the flow alignment strategy. This method, however, relies on mechanical shear forces to detach freestanding nanostructures from the growth substrate, and friction to align the deposited the nanowires onto the device substrate [24]. In this method, the growth substrate is inverted and placed on top of the device substrate such that the nanostructures are sandwiched in between (Fig. 7). Weights are added on top of the growth substrate to control the applied pressure between the two substrates in order to optimize the deposited nanostructure density. The growth substrate is then pulled over the device substrate at a continuous velocity in order to deposit and align the nanostructures onto the device substrate. The shear velocity and the amount of friction between the two
N
N
1756
Nanostructure Field Effect Transistor Biosensors, Fig. 7 Contact printing of nanowires. (a) Schematic of the contact printing process for producing well-aligned nanowire arrays. (b) Darkfield optical and (c) SEM images of 30-nm nanowires printed on a Si/SiO2 substrate showing highly dense and aligned monolayer of nanowires. The self-limiting process limits the transfer of nanowires to a single layer, without significant nanowire bundling (Adapted with permission from [24]. Copyright 2008 American Chemical Society)
Nanostructure Field Effect Transistor Biosensors
a
donor substrate
Contact Printing
patterned receiver substrate
weight
Resist Lift-off
b
lubricant Removal of the donor substrate
c
Nanostructure Field Effect Transistor Biosensors, Fig. 8 Blown bubble film approach. Nanostructures are dispersed in a polymer solution. A volume of solution is expanded as a bubble using a die to produce well-aligned nanostructures
suspended within a thin film. The film is then contacted with a substrate to deposit well-aligned nanostructures onto the substrate ([25] Reproduced with permission of The Royal Society of Chemistry)
substrates influence the resulting degree of nanostructure alignment. The latter may be adjusted with the use of a lubricant.
when that film is elongated. Therefore, this method involves the formation of a large balloon from a nanostructure suspension (Fig. 8) [25]. The aligned nanostructures within the thin film can then be transferred to a substrate by placing the substrate in contact with the balloon. Nano-FET fabrication would subsequently proceed as in the case of flow alignment.
Thin-Film Nanowire Suspension Thin-film nanowire suspension is a method that makes use of the observation that nanostructures suspended within a liquid thin film will align along a single direction
Nanostructure Field Effect Transistor Biosensors Nanostructure Field Effect Transistor Biosensors, Fig. 9 Linker-free directed self-assembly. (a) The patterning process of octadecyltrichlorosilane selfassembled monolayer on a solid substrate via photolithography. (b) Selfassembly suspended nanostructures onto the bare surface regions on the substrate. Subsequent photolithography and metallization can be used to pattern electrodes onto the ends of the deposited nanostructures (Adapted with permission from Macmillan Publishers Ltd. Nature Nanotechnology [26]: copyright 2006)
1757
a
b Bare surface
N
Inert molecular patterns
Substrate
Substrate
Inert molecular patterns
Substrate
Nanowire solution
Photoresist patterning with short soft baking
Anhydrous hexane
Nanowire assembly vai van der Waals interactions Substrate
Removal of residual water
Nanowires
N
OTS solution Aligned nanowire patterns
Electrodes Substrate
Molecular deposition Electrode fabrication
Directed Self-Assembly Directed self-assembly relies on the minimization of surface energy to assemble suspended nanostructures into predefined locations on a substrate [26]. In this method, surface modification strategies are used to render certain portions of the device substrate either hydrophobic or hydrophilic (Fig. 9). Upon immersion of the substrate into a nanostructure suspension, individual nanostructures will
self-assemble onto the hydrophobic portions of the substrate. This strategy can be used to deposit individual nanostructures at specific locations over a large area on a substrate. A similar strategy using analyteligand interactions has also been employed in which suspended nanostructures are functionalized with biotin molecules. Upon immersion of a substrate with bound streptavidin molecules at predefined
N
1758
Nanostructure Field Effect Transistor Biosensors
a
b
w S
t
200 nm
D G
Nanostructure Field Effect Transistor Biosensors, Fig. 10 (a) Schematic of nano-FET device after anisotropic etching. The silicon-on-insulator active channel (yellow, width w and thickness t) is undercut etched, whereas degenerate leads (red) are
etch-resistant. The source (S), drain (D), and underlying backgate (G) are labeled. (b) Scanning electron micrograph of a complete device (Adapted with permission from Macmillan Publishers Ltd: Nature [26], copyright 2007)
locations, biotin-streptavidin conjugation forces the nanostructures to assemble onto the device substrate at predefined locations.
resulting nanostructure orientation. This method usually produces numerous nanostructures and is, therefore, useful in fabricating nano-FET sensors that utilize a network of nanostructures.
Dielectrophoresis Dielectrophoresis (DEP) utilizes electrical fields to manipulate nanostructures. In this method, metallic contact electrodes are patterned on the device substrate prior to the assembly of nanostructures onto the device. By applying a biased alternating voltage across the electrodes, a local nonuniform electric field is produced, exerting a force on suspended semiconducting nanostructures. This force causes the nanostructures to assemble across the two electrodes. This process is easily amenable to the large-scale production of nano-FET arrays and can be used to fabricate single nanostructure FET devices; however, typical DEP trapping processes yield networked multi-nanostructure FET devices [27]. This process is highly susceptible to the experimental conditions including the size, shape, and properties of the nanostructure to be manipulated, and parameters of the electrical signals, as well as the electrical properties of the surrounding medium. In Situ Growth of Nanostructures In situ growth of nanostructures is a method that utilizes patterned catalysts to selectively synthesize nanostructures at certain regions on the device substrate. These catalysts (e.g., iron or gold particles) are required to initiate and/or propagate nanostructure growth. In some cases, the direction of nanostructure growth can be influenced with the application of an electric field, therefore enabling control of the
Top-Down Fabrication Method
Anisotropic Lateral Wet Etching Top-down methods are typically based on the anisotropic lateral wet etching of nanometer-thin SOI (silicon-on-insulator) wafers to produce well-defined nanostructures from the device layer (Fig. 10) [8]. Micro- or nanosized etch masks are patterned via conventional or ebeam photolithography and anisotropically time-etched to produce nanometer-wide nanostructures. Source and drain electrodes are degenerately doped, rendering them conductive and unaffected by the etchant. This approach enables wafer-scale formation of semiconducting nanostructures at precise locations on a wafer-scale, making subsequent electrode fabrication relatively straightforward using conventional photolithography. As this method relies on time-controlled anisotropic etching for the removal of material, it is highly susceptible to small changes in processing conditions such as etch time, processing temperature and mixing, and device-layer doping density and crystal orientation.
Future Directions At this point in time, the development of nano-FET biosensors is still in the proof-of-concept phase. The promise of highly sensitive, label-free electrical
Nanostructure Field Effect Transistor Biosensors
sensors with real-time measurement capability remains attractive for numerous medical and basic science research applications. While great progress has been made in nano-FET fabrication and detection, continued research into large-scale fabrication methods for batch-manufacturing nano-FET sensors will be essential for the ultimate commercial success of this technology and its application to clinical and research applications. In addition to low cost and high yield, these fabrication methods must produce devices with consistent sensing performance across the nanoFET sensors such that measurements made with these devices can be standardized. Another critical hurdle that must be overcome is the issue of poor analyte specificity and nonspecific binding. While some strategies have been put forth to remedy these issues, these solutions must be further improved upon before nanoFET biosensors are able to reliably analyze whole blood, serum, and other specimens. Lastly, efforts in the area of nano-FET biosensor integration with microfluidic and lab-on-chip devices will facilitate point-of-care diagnostic applications and real-time closed-loop drug delivery systems. Successful lab-onchip integration will also enable the creation of multiplexed nano-FET biosensor arrays, which could hold great promise in the areas of basic chemical and biological research, high-throughput screening systems for drug development, and novel in vitro biology experimentation. The future will likely see many point-of-care biosensors using electrical-based detection, and nano-FET biosensing technologies can possibly play an important role.
Cross-References ▶ Biosensors ▶ Carbon Nanotubes ▶ Chemical Vapor Deposition (CVD) ▶ Dielectrophoresis ▶ Dielectrophoretic Nanoassembly of Nanotubes onto Nanoelectrodes ▶ Electric-Field-Assisted Deterministic Nanowire Assembly ▶ Functionalization of Carbon Nanotubes ▶ Nanophotonic Structures for Biosensing ▶ Nanostructured Materials for Sensing ▶ Nanotechnology ▶ Self-assembly of Nanostructures
1759
N
References 1. Lee, S.W., et al.: Highly sensitive biosensing using arrays of plasmonic au nanodisks realized by nanoimprint lithography. ACS Nano 5, 897–904 (2011) 2. Fritz, J.: Cantilever biosensors. Analyst 133, 855–863 (2008) 3. Cui, Y., Wei, Q., Park, H., Lieber, C.M.: Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species. Science 293, 1289–1292 (2001) 4. Soleymani, L., Fang, Z., Sargent, E.H., Kelley, S.O.: Programming the detection limits of biosensors through controlled nanostructuring. Nat. Nanotechnol. 4, 844–848 (2009) 5. Patolsky, F., Zheng, G., Lieber, C.M.: Fabrication of silicon nanowire devices for ultrasensitive, label-free, real-time detection of biological and chemical species. Nat. Protoc. 1, 1711–1724 (2006) 6. Fan, Z., Lu, J.G.: Gate-refreshable nanowire chemical sensors. Appl. Phys. Lett. 86, 123510–123513 (2005) 7. Gao, X.P., Zheng, G., Lieber, C.M.: Subthreshold regime has the optimal sensitivity for nanowire FET biosensors. Nano Lett. 10, 547–552 (2010) 8. Stern, E., et al.: Label-free immunodetection with CMOScompatible semiconducting nanowires. Nature 445, 519–522 (2007) 9. Zheng, G., Patolsky, F., Cui, Y., Wang, W.U., Lieber, C.M.: Multiplexed electrical detection of cancer markers with nanowire sensor arrays. Nat. Biotechnol. 23, 1294–1301 (2005) 10. Patolsky, F., et al.: Electrical detection of single viruses. Proc. Natl. Acad. Sci. U.S.A. 101, 14017–14022 (2004) 11. Bunimovich, Y.L., et al.: Quantitative real-time measurements of DNA hybridization with alkylated nonoxidized silicon nanowires in electrolyte solution. J. Am. Chem. Soc. 128, 16323–16331 (2006) 12. Zhang, G.J., et al.: Highly sensitive measurements of PNA-DNA hybridization using oxide-etched silicon nanowire biosensors. Biosens. Bioelectron. 23, 1701–1707 (2008) 13. Zhang, G.J., et al.: DNA sensing by silicon nanowire: charge layer distance dependence. Nano Lett. 8, 1066–1070 (2008) 14. Maehashi, K., et al.: Label-free protein biosensor based on aptamer-modified carbon nanotube field-effect transistors. Anal. Chem. 79, 782–787 (2007) 15. Stern, E., et al.: Importance of the Debye screening length on nanowire field effect transistor sensors. Nano Lett. 7, 3405–3409 (2007) 16. Nair, P.R., Alam, M.A.: Design considerations of silicon nanowire biosensors. Electron Devices, IEEE Trans. 54, 3400–3408 (2007) 17. Heller, I., Mannik, J., Lemay, S.G., Dekker, C.: Optimizing the signal-to-noise ratio for biosensing with carbon nanotube transistors. Nano Lett. 9, 377–382 (2009) 18. Lu, M.P., Hsiao, C.Y., Lai, W.T., Yang, Y.S.: Probing the sensitivity of nanowire-based biosensors using liquid-gating. Nanotechnology 21, 425505 (2010) 19. Sheehan, P.E., Whitman, L.J.: Detection limits for nanoscale biosensors. Nano Lett. 5, 803–807 (2005)
N
N
1760
20. Squires, T.M., Messinger, R.J., Manalis, S.R.: Making it stick: convection, reaction and diffusion in surface-based biosensors. Nat. Biotechnol. 26, 417–426 (2008) 21. Stern, E., et al.: Label-free biomarker detection from whole blood. Nat. Nanotechnol. 5, 138–142 (2010) 22. Star, A., Garbiel, J.P., Bradley, K., Gruner, G.: Electronic detection of specific protein binding using nanotube FET devices. Nano Lett. 3, 459–463 (2003) 23. Huang, Y., Duan, X., Wei, Q., Lieber, C.M.: Directed assembly of one-dimensional nanostructures into functional networks. Science 291, 630–633 (2001) 24. Fan, Z., et al.: Wafer-scale assembly of highly ordered semiconductor nanowire arrays by contact printing. Nano Lett. 8, 20–25 (2008) 25. Yu, G., Li, X., Lieber, C.M., Cao, A.: Nanomaterialincorporated blown bubble film for large-area aligned nanostructures. J. Mater. Chem. 18, 728–734 (2008) 26. Lee, M., et al.: Linker-free directed assembly of highperformance integrated devices based on nanotubes and nanowires. Nat. Nanotechnol. 1, 66–71 (2006) 27. Vijayaraghavan, A., et al.: Ultra-large-scale directed assembly of single-walled carbon nanotube devices. Nano Lett. 7, 1556–1560 (2007)
Nanostructured Antireflective Surfaces Arrays ▶ Moth-Eye Antireflective Structures
Nanostructured Functionalized Surfaces Lorenzo Lunelli1, Cristina Potrich1, Laura Pasquardini2 and Cecilia Pederzolli2 1 Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation and CNR-IBF, Povo, TN, Italy 2 Biofunctional Surfaces and Interfaces, FBK-CMM Bruno Kessler Foundation, Povo, TN, Italy
Synonyms Functionalized nanomaterials
Definition Nanostructured functionalized surfaces are materials that contain microscale and nanoscale features produced in a well-controlled manner with nanocoating,
Nanostructured Antireflective Surfaces Arrays
film deposition, or in situ surface modifications at nanoscale level, with the aim to improve the material performances. The potential applications of nanostructured functionalized surfaces are wide; some examples are in vivo imaging and diagnostics (nanosized contrast agents for medical imaging, miniaturized devices); in vitro miniaturized diagnostics (lab-on-a-chip, pointof-care measurements); drug development, delivery, and therapy (nanoparticles, nanobody, affibodies); tissue engineering; and implants of drug dispenser/factory.
Overview Nanostructures or nanomaterials have peculiar physical and/or chemical properties, different from their bulk material. Main differences rely on topography or chemistry or on both aspects of the outer layer of the material which becomes “nanostructured” when these topographical and/or chemical modifications are made at a nanometric level. Nanostructured surfaces can be produced by surface modification of materials conventionally interacting with biological systems such as human body (biomaterials). Nanofunctionalization techniques can be divided in two main categories: (a) nanocoating and film deposition and (b) in situ surface nanofunctionalization. These two types of techniques are often combined to produce surfaces with hybrid nanostructures such as a coating/film and a nanostructured zone. Typical techniques belonging to the first category are, for example, plasma spraying, plasma immersion ion implantation and deposition, chemical or physical vapor deposition, cold spraying, and self-assembly, whereas the second category is represented by methods like laser etching, shot blasting, acid and alkali treatments, anodic oxidation, micro-arc oxidation, ion implantation, and similar [1]. The nanofunctionalization of the surface can improve the performances of biomaterials, helping their use in clinical applications. In fact, cells life in complex tissues is controlled through modulation of cell binding to the extracellular matrix, which is composed by nanostructures. Cells respond to micro- and nanofeatures with different chemistries and topographies, resulting in changes in cell alignment, polarization, elongation, migration, proliferation, and gene expression. Through micro- and nanopatterning techniques it is possible to precisely position selected
Nanostructured Functionalized Surfaces
biomolecules on a substrate for driving cell growth and, possibly, regulating cell functions [2]. While flat nanostructured surfaces are mainly suitable for cells growing in different conditions depending on chemical and topographical patterning of substrates, other nanostructures are also deeply studied for various applications. Nanostructures such as metal or metal oxide nanoparticles, quantum dots, silica, carbon nanotubes, C60, hydrogel, block copolymer, and dendrimer are reported [3]. Main applications of these different types of nanostructures are imaging (both biolabeling and biosensing, comprising diagnostics purposes) and drug or gene delivery (for instance nanobodies or affibodies against cancer cells). Nanostructured surfaces, such as possibly functionalized nanoparticles or even nanotubes, are also included in micro- and nanodevices aimed at diagnostics and/or medical treatments. Various methods for the fabrication of nanostructured functionalized surfaces are available, either adapted from macro-scale technologies or ad hoc developed, while fewer are the techniques suitable for the production of nanoarrays containing multiple features such as a range of different proteins or small ligands. Topographical and chemical nanopatterning of surfaces by methodologies substantially developed for fabrication in the microelectronic field are not reported in this chapter, whereas scanning probe–based techniques utilized to obtain functional structures at nanoscale (such as nanoarray) with different features are described in the following paragraphs. They are based on, for example, dip pen nanolithography, SNOM lithography, and nanografting.
Flat Nanostructured Surfaces Flat nanostructured functionalized surfaces are surfaces with defined regions of different chemical functionality to achieve site-specific attachment of one species in some areas while minimizing unwanted surface interactions in other areas. Two surface patterning types can be distinguished, a structural topography and a chemical patterning, but at nanometric scale these features can be mixed. Moreover, a topographical pattern can be performed maintaining the surface chemistry, but the majority of chemical techniques induce a change in topography. Different strategies to achieve such patterned surfaces have been
1761
N
proposed, mainly using techniques developed in the microelectronics field to chemically modify some parts of the surface, for example, via electron irradiation or extreme UV interference lithography or via microcontact printing. Alternatively, the combination of different substrate materials for the creation of patterned surfaces, commonly via photolithography, is also reported [4]. The production of patterns of low cost and high complexity (three or more addressable chemistries) over large areas was recently described [5]. The method proposes the creation of a long-rangeordered array of microscale “cups” with multiple material regions by a combination of colloidal lithography and the bottom-up deposition techniques of plasma polymerization and evaporation as well as sputtering via physical vapor deposition. Plasma polymerization possesses indeed the unique practical advantages that offer single-step coating with nanometer-scale control over thickness and a wide range of functional groups can be presented with control over density. Patterned surfaces gained great attention since cells respond to micro- and nanofeatures with different chemistries and topographies. The influence of various topographic features such as grooves, ridges, stops, pores, wells, and nodes in microscale or nanoscale has been analyzed by culturing a wide variety of cells in the presence of such patterned nanosurfaces. Fibroblasts are greatly studied with nanotopography, showing as a general behavior that the adhesion increases with decreasing feature size or height. Morphology and cytoskeleton arrangement also change. The details of the nanotopography, and hence randomness and pattern, play an important role in eliciting different cell responses [6]. Muscle cells, endothelial cells, blood cells, hepatocytes, and many other cell lines were also studied, as revised by Yim et al. [6], concluding that cells vary their adhesion, proliferation, migration, and gene expression properties if grown on nanostructured substrates. This behavior could be ascribed to the fact that cells in their natural environment interact with extracellular matrix components which are in the nanometer scale. For example, the basement membrane of many tissues displays features of pores, fibers, and ridges in the nanometer range. Moreover, surface nanofunctionalization was demonstrated to significantly affect cellular and subcellular functions, as reviewed by Liu et al. [1], stating that clinical applications of biomaterials can be improved
N
N
1762
by producing a nanostructured surface. Apart from physical or chemical treatments aimed at the creation of nanostructures on a material surface, the modification of a substrate with specific ligand which improves cell attachment and differentiation is also well documented. Bell et al. [7], for example, reported the use of oligopeptides to coat titanium surfaces for osteointegration of orthopedic implants by favoring attachment of osteoprogenitor cells and promoting osteoblastic differentiation. Stem cell differentiation to desired phenotypes was also deeply studied in response to nanostructured growing substrates [2]. In particular, it was demonstrated that it is possible to drive cell adhesion, migration, proliferation, and even differentiation of stem cells, by the design and fabrication of growing surfaces or scaffolds with controlled topography and chemistry at the micro- and nanoscale. Topography acts on cell shape and alignment through mechanical interaction of the plasma membrane and the specific ligands present on it with the nanostructures on the substrate surface. However, the most important parameter at micro- and nanometer level seems the order or randomness characteristic of the surface. Chemical patterning regulates cell functions by mechanical interaction (e.g., adhesive microislets) or by direct receptor-ligand interaction with cell-membrane proteins. Topography and chemistry can also be combined to have synergistic effects and mimic the natural cell micro-environment. Apart from eukaryotic cells, bacteria can also colonize surfaces possibly responding to nanotopography. Main difference is however the fact that bacteria live on surfaces as a community within a species- and strain-specific extracellular polymer, forming the socalled biofilm. Little is known about the response of bacteria toward nanostructures and only indirect experimental results support the idea that bacteria are able to use molecular features of their cell membrane as sensors and implement intracellular signaling pathways to sense the surface and to react to the stimuli created by the surface [8].
Functionalized Nanostructures Various nanostructures or nanomaterials sensitive to environmental or biological parameters have been reported. Usually they are made of metals, metal oxides, semiconductors, silica, carbon nanotubes, C60, hydrogel,
Nanostructured Functionalized Surfaces
block copolymer, and dendrimer [3]. Nanometer-sized metallic structures, such as platinum, gold, and silver nanoparticles, exhibit unique optical properties. In fact, the characteristics of the plasmons excited by the light on their surface are depending on the dielectric constant of the media that constitute the interface. The adsorption of molecules to the nanoparticle changes this parameter, leading to detectable optical changes of the plasmon resonance. Moreover, these nanostructures can be functionalized to specifically interact, for example, with cells for site-specific drug delivery or imaging purposes. It was clearly demonstrated that differently decorated nanoparticles interact in different extent with cell membranes and cell uptake is indeed greatly influenced [9]. Iron oxide magnetic nanoparticles have been used widely in biological detection because of their good biocompatibility and ability to be manipulated under an external magnetic field. Quantum dots are also extensively used for biolabeling and biosensing, since they exhibit advantages in signal brightness, photostability, multicolor-light emission, and size-tunable properties. Carbon-based nanomaterials are recently gaining interest for biomedical applications. Carbon nanotubes have been used as drug or gene carriers and for biosensing, whereas certain fullerene C60 derivatives and fluorescent carbon dots have been developed as nontoxic alternatives to semiconductor quantum dots for biomedical imaging [10]. Self-assembly has been also extensively explored to functionalize the material surface and build nanostructures or nanomaterials, including ordered mono- and multilayers, liposomes or vesicles, nanoparticles, nanotubes, and nanofibers. Moreover, self-assembled nanostructures from block copolymers have also frequently been reported. They consist of two or more covalently bounded blocks with different physical and chemical properties, resulting in a supramolecular assembly that can respond to changes in temperature, pH, and redox potential. In fact, they can generate various microdomain morphologies by means of intramolecular forces. Since active selectivity is often desirable, targeted drug delivery carriers are functionalized with antibodies or antibody fragments to provide such active localization. The conjugation of multiple antibodies to each nanocarrier enhances their specificity and ability to detect their targets; nanocarriers can thus be surfacefunctionalized with multiple distinct antibodies to overcome tumor heterogeneity. Functional, single-
Nanostructured Functionalized Surfaces
domain heavy-chain antibodies, referred to as nanobodies, have been raised against cancer targets with the purpose to either antagonize receptor function or deliver an enzyme for prodrug activation. Affibodies against a variety of cancer-related targets have been developed and are now commercially available, including: EGFR, HER2, and transferrin [11]. Nanoparticles have also been capped with various biocompatible polymers such as PVA, dextran, silica, and inorganic capping agents like biotin, citric acid, avidin, carbodiimide, SiO2, and chitosan-PEG. These molecules then act as linkers for the coupling of cytotoxic drugs or target antibodies. A step forward is, for example, the design of luminomagnetic nanocarriers functionalized with the ligand folic acid [12]. This approach combines the luminescent and magnetic properties of the luminomagnets with the specificity of folic acid making them feasible for folate receptormediated molecular imaging and targeted drug delivery. Besides these, the small size of these luminomagnetic particles (about 8 nm) can be exploited for their easy elimination from the body after delivery of the drugs. Furthermore, micro- and nanostructures are included in the development of smaller medical devices or systems aimed to drug and gene delivery, tissue engineering, biosensors, and diagnostic systems. The final goal relies in developing implantable sensing and/or treatment devices to collect information for diagnosing disease and administrating treatment in vivo [3]. Finally, another class of nanostructured materials is often mentioned as potentially suitable for bio-applications. Hydrogels and molecular imprinted polymers can be considered either as flat nanofunctionalized surfaces or nanosystems similar to those described in this paragraph. Hydrogels are 3D high-molecular-weight networks composed of a polymer backbone, water, and a cross-linking agent. They can be stimuli sensitive, thus being of great interest in drug delivery, cell encapsulation, and tissue engineering [3]. Molecular imprinted polymers are also described as powerful nanotools for specific applications. They are tailor-made biomimetic receptors often used for functionalizing the surface of nanomaterials with sensing or separation purposes. Molecular imprinted polymers are produced by polymerization in the presence of molecular templates that are removed afterward, thus creating complementary nanostructured surfaces. These new surfaces can bind selectively the target (or template) used to prepare them [3].
1763
N
Nanoscale Patterns Fabrication Nanoscale patterns are generally quite difficult to work with, both in terms of preparation and characterization. Besides downscaling the fabrication methods originally developed for microconditions to nanoscale (such as photolithography, nanoimprint lithography), dedicated fabrication methods have been proposed for these applications. The following paragraphs describe the scanning probe–based techniques most commonly utilized to obtain functional structures at nanoscale such as nanoarrays containing different features. Dip Pen Nanolithography A scanning probe technique that has been developed for the production of nanoarrays is the ▶ Dip-Pen Nanolithography (DPN), first reported by Mirkin’s group in 1999 [13]. DPN makes use of one or several AFM tips that are dipped into an ink and subsequently used to write a nanoscale pattern on a surface. The method was presented as a powerful technique to deposit molecules on a substrate from AFM tips at resolutions comparable to those achieved with much more expensive lithographic methods. The principle takes advantage of the water meniscus that naturally forms between the ink-coated probe tip and the substrate when the scanning was done in contact mode under ambient laboratory conditions. The ink moves on the substrate by capillary transport through the meniscus, as shown in Fig.1, Panel A. A key point is the correct choice of the ink and the substrate, which have to possess reciprocal chemical affinities, thus favoring the chemical adsorption of the ink molecules onto the substrate. An useful application of this technique is related to the precise functionalization of nanoscale devices previously prepared by more conventional lithographic methods. The resolution of DPN depends on several parameters such as the grain size of the substrate, the molecule diffusion rate, the tipsubstrate contact time and the scan speed, and finally on the environmental conditions such as humidity. The minimum feature sizes producible by DPN are in the range of 15 nm [8]. Nanografting An effective way to introduce chemical and functional features with nanoscale lateral resolution down to few nanometers has been developed by Liu and coworkers starting from 1997 [14, 15] using an AFM (▶ AFM) tip.
N
N a
1764
Nanostructured Functionalized Surfaces
AFM tip
DIP PEN LITOGRAPHY
writing direction
Lateral Resolution
molecules deposition The patterning is obtained directly depositing the molecules.
Down to 15 nm
water meniscus
b
Force
NANOGRAFTING The patterning is obtained replacing some molecules with others.
c
SNOM LITOGRAPHY The patterning is obtained partially removing some molecules or photoactivating specific areas.
AFM tip
Aperture probe tip
3–5 nm
Apertureless probe tip Aperture tips: down to 50 nm
Far-field light Far-field light
Near-field
Apertureless tips: down to 10 nm
Near-field
Nanostructured Functionalized Surfaces, Fig.1 Summary of scanning probe–based techniques for functional surface nanopatterning. A brief description of the principle, an illustrative cartoon, and the lateral resolution of each technique are reported
This method is based on the local modification of a selfassembled monolayer (▶ Self-assembled monolayers) surface, previously deposited starting from an alkanethiol solution, on a flat gold substrate. The surface is then mounted in a liquid environment AFM cell, allowing its observation and nanomanipulation with the AFM tip. While scanning the surface in a low load regime (< 1 nN) results in the three-dimensional imaging of the surface, increasing the load over a threshold value one obtains the removal of the previously deposited thiolated molecules under the tip, locally disrupting the monolayer and exposing the gold surface. If thiolated
molecules (usually different from that used in the initial SAM deposition) are present in solution, a newly deposited SAM is obtained in areas where the tip is removing the old one, locally changing morphology and chemistry of the surface. Patches with different height with respect to the surrounding surface can be deposited and detected in topography AFM images. Nanografted structures can also introduce characteristic groups in specific areas that show a chemical contrast in AFM friction images. The principle is illustrated in Fig.1, Panel B. Nanografting can be applied even with a solution of the same molecule previously used for the initial SAM
Nanostructured Functionalized Surfaces
formation (homogeneous nanografting), obtaining in this case the removal of defects that were possibly present in the SAM [15]. Having obtained a nanopatterned surface endowed with specific chemical groups in well-defined patches, one can then exploit this method to create nanoarrays of biomolecules. Liu and coworkers [16] showed how to take advantage of both non-covalent and covalent attachments to form such protein nanoarrays, down to a lateral scale of tens of nanometers. This method generally leads to non-oriented protein patches, as demonstrated by the AFM analysis, that shows the presence of proteins with different heights with respect to the surface, corresponding to the different protein orientation. The nanografting technology can be also successfully applied to the realization of DNA nanoarrays, when thiolated single-stranded DNAs are used in the solution during the nanografting process. DNA lines as narrow as 10 nm can be indeed fabricated. Moreover, nanografting allows a controlled density deposition of DNA strands, depending on the number of tip passages over the same area [17]. Successful DNA hybridization has been demonstrated on such nanoarrays. It is also possible to extend such a technique to realize protein nanoarrays: Patches of single-stranded DNAs of different sequences are deposited using nanografting in specific places of a surface previously covered with a SAM able to discourage aspecific protein adhesion. Then, proteins linked with DNA sequences that are complementary to that of the patch where the protein deposition is desired are incubated over the surface. Thanks to the DNA complementarity, a specific protein localization is then obtained. The effectiveness of this methodology for the realization of nanoscale protein sensors has been demonstrated using human serum enriched with antibodies against specific proteins that were previously immobilized on nanografted patches, showing that antibodies bind to the patches where their targets are present [18]. A method that in similar fashion takes advantage of the DNA hybridization properties to direct the localization of nanosized objects with the help of an AFM probe (although not based on nanografting) has been demonstrated by Kufer and coworkers in 2008 [19]. Their method is based on the fabrication of specific areas on a glass surface, namely, depot areas – where DNA-coupled nanoobjects are picked up by a single-stranded functionalized AFM tip, and target areas – where the nanoobjects are delivered with a precision of some nm to form the desired pattern.
1765
N
More than 5,000 object transfers with the same tip have been demonstrated. SNOM Lithography In 1983, it was proved the Scanning Near-field Optical Microscopy (SNOM), shortly after the invention of the first proximal probe technology, the Scanning Tunneling Microscope (STM). These two techniques are based on a similar principle, but in the case of STM the detection is performed on the electron tunneling currents, while in the case of SNOM the detection is performed on the tunneling of photons. This characteristic makes the SNOM suitable for the analysis of nonconductive samples [20]. By illuminating the sample with near-field light, the probe tip is held very close to the sample without touching. An evanescent light field is created in very close proximity of surface. The evanescent light through the aperture is incident on the sample, exciting the atoms in the surface to reradiate propagating waves. The same probe can then collect the propagating light and transmit it to an appropriate detector. The probe tip can be designed in two different ways, an aperture-based tip and an apertureless-based tip (as illustrated in Fig.1, Panel C). In an aperture-based SNOM, however, the nanometer-scale aperture limits the light throughput and the resolution. To overcome these limitations an apertureless SNOM was developed that uses an apertureless probe, similar to an STM or AFM tip, to scan within the near field of the sample. In this case the sample is first irradiated from above by far-field light in order to produce the evanescent field on the surface of the sample. The evanescent photons then excite the atoms in the probe tip, which reradiate the propagating photons. An interesting application of this technique is related to the lithography-like process. The SNOM probe is used as an energy source for modifying a variety of materials at the nanoscale level, including conventional photoresists and non-conventional electro-magnetooptical materials as well as ferroelectric and SAM. The photoresists layer deposited on a substrate may be changed for instance by laser pulses emitted from the SNOM probe, offering resolutions unachievable by conventional optical and laser systems (well beyond the diffraction limit). Patterns are generated by scanning the probe over these target surfaces and both aperture and apertureless probe tips are used, obtaining different resolution limits (50 nm for aperture probe and down to 10 nm for the apertureless probe).
N
N
1766
Nanostructured Hydrogels
Cross-References ▶ AFM ▶ Chitosan Nanoparticles ▶ Dip-Pen Nanolithography ▶ Functionalization of Carbon Nanotubes ▶ Microfabricated Probe Technology ▶ Nanomedicine ▶ Nanoparticles ▶ Nanostructures for Surface Functionalization and Surface Properties ▶ Self-assembled Monolayers ▶ Self-assembly of Nanostructures
References 1. Liu, X., Chu, P.K., Ding, C.: Surface nano-functionalization of biomaterials. Mat. Sci. Eng. R 70, 275–302 (2010) 2. Martı´nez, E., Lagunas, A., Mills, C.A., Rodrı´guez-Seguı´, S., Este´vez, M., Oberhansl, S., Comelles, J., Samitier, J.: Stem cell differentiation by functionalized micro- and nanostructured surfaces. Nanomedicine (Lond.) 4(1), 65–82 (2009) 3. He, J., Qi, X., Miao, Y., Wu, H.L., He, N., Zhu, J.J.: Application of smart nanostructures in medicine. Nanomedicine (Lond.) 5(7), 1129–1138 (2010) 4. Briand, E., Humblot, V., Landoulsi, J., Petronis, S., Pradier, C.M., Kasemo, B., Svedhem, S.: Chemical Modifications of Au/SiO2 Template Substrates for Patterned Biofunctional Surfaces. Langmuir 27(2), 678–685 (2011) 5. Ogaki, R., Cole, M.A., Sutherland, D.S., Kingshott, P.: Microcup arrays featuring multiple chemical regions patterned with nanoscale precision. Adv. Mater. 23(16), 1876– 1881 (2011) 6. Yim, E.K.F., Leong, K.W.: Significance of synthetic nanostructures in dictating cellular response. Nanomed. Nanotechnol. Biol. Med. 1, 10–21 (2005) 7. Bell, B.F., Schuler, M., Tosatti, S., Textor, M., Schwartz, Z., Boyan, B.D.: Osteoblast response to titanium surfaces functionalized with extracellular matrix peptide biomimetics. Clin. Oral Impl. Res. 22(8), 865–872 (2011) 8. Anselme, K., Davidson, P., Popa, A.M., Giazzon, M., Liley, M., Ploux, L.: The interaction of cells and bacteria with surfaces structured at the nanometre scale. Acta Biomater. 6, 3824–3846 (2010) 9. Verma, A., Uzun, O., Hu, Y., Hu, Y., Han, H.S., Watson, N., Chen, S., Irvine, D.J., Stellacci, F.: Surface-structureregulated cell-membrane penetration by monolayerprotected nanoparticles. Nat. Mater. 7, 588–595 (2008) 10. Xing, Y., Dai, L.: Nanodiamonds for nanomedicine. Nanomedicine 4(2), 207–218 (2009) 11. Murday, J.S., Siegel, R.W., Stein, J., Wright, J.F.: Translational nanomedicine: status assessment and opportunities. Nanomedicine 5(3), 251–273 (2009) 12. Dutta, R.K., Sharma, P.K., Pandey, A.C.: Design and surface modification of potential luminomagnetic
13. 14.
15.
16.
17.
18.
19.
20.
nanocarriers for biomedical applications. J. Nanopart. Res. 12, 1211–1219 (2010) Piner, R.D., Zhu, J., Xu, F., Hong, S., Mirkin, C.A.: “Dip-Pen” Nanolithography. Science 283, 661–663 (1999) Xu, S., Liu, G.: Nanometer-scale fabrication by simultaneous nanoshaving and molecular self-assembly. Langmuir 13(2), 127–129 (1997) Xu, S., Miller, S., Laibinis, P.E., Liu, G.: Fabrication of nanometer scale patterns within self-assembled monolayers by nanografting. Langmuir 15(21), 7244–7251 (1999) Wadu-Mesthrige, K., Xu, S., Amro, N.A., Liu, G.: Fabrication and imaging of nanometer-sized protein patterns. Langmuir 15(25), 8580–8583 (1999) Mirmomtaz, E., Castronovo, M., Grunwald, C., Bano, F., Scaini, D., Ensafi, A.A., Scoles, G., Casalis, L.: Quantitative study of the effect of coverage on the hybridization efficiency of surface-bound DNA nanostructures. Nanoletters 8(12), 4134–4139 (2008) Bano, F., Fruk, L., Sanavio, B., Glettenberg, M., Casalis, L., Niemeyer, C.M., Scoles, G.: Toward multiprotein nanoarrays using nanografting and DNA directed immobilization of proteins. Nanoletters 9(7), 2614–2618 (2009) Kufer, S.K., Puchner, E.M., Gumpp, H., Liedl, T., Gaub, H.E.: Single-molecule cut-and-paste surface assembly. Science 319, 594–596 (2008) Tseng, A.A.: Recent developments in nanofabrication using scanning near-field optical microscope lithography. Opt. Laser Technol. 39, 514–526 (2007)
Nanostructured Hydrogels ▶ Nanoengineered Hydrogels for Cell Engineering
Nanostructured Materials ▶ Nanomaterials for Electrical Energy Storage Devices
Nanostructured Materials for Sensing Mariangela Lombardi IIT - Italian Institute of Technology @ POLITO Centre for Space Human Robotics, Torino, Italy
Synonyms Nanomaterials for sensors
Nanostructured Materials for Sensing
Definition Materials having nanometric dimensions are able to improve their functional properties, which allow their exploitation in highly performing sensors.
Overview Materials with at least one dimension smaller than 100 nm are classified as nanostructured. Compared to micro-sized ones, nanomaterials are characterized by a higher surface to bulk ratio, usually leading to an enhancement of optical, mechanical, electrical, structural, and magnetic properties [1]. In the last few years, research focused its efforts on the development of multifunctional composite materials, able to perform multiple non-structural functions, sometimes combined with a suitable mechanical response. If nanostructured, they can also imply additional functional properties with respect to conventional composites. Multifunctional materials can be used for developing highly performing sensors, particularly in the case of single-phased and composite materials containing carbon nanotubes and piezoelectric phases [2].
Piezoelectric Materials Piezoelectric materials are able to directly convert a mechanical signal into an electrical one and vice versa. For this reason piezoelectric materials produce a charge when subjected to a stress (direct effect) or a strain when an electric field is applied (converse effect). These materials can then be used as actuators or sensors, by exploiting their converse or direct piezoelectric effect, respectively. Generally piezoelectric sensors are used for measuring several parameters, like acceleration, angular rate, pressure, force, temperature, mass, magnetic field, and amount of several chemical substances [3]. Piezoelectricity can be expressed by several materials, such as single crystals, ceramics, polymers, and composites [4, 5]. Even if they naturally present this behavior, piezoelectric single crystals, as quartz and Rochelle salt, cannot be easily used being isotropic, e.g., they have different properties depending on the
1767
N
cut and the orientation directions. Most piezoelectric ceramics applied in commercial sensing devices are made of polycrystalline ferroelectric ceramics [6], which present a higher strength and an easier fabrication with respect to single crystals [4]. Ferroelectrics are characterized by a spontaneous polarization which can be orientated under a realizable electric field. In fact, they initially contain randomly oriented, polarized regions, named domains. These polarized regions form upon cooling at a temperature called the Curie temperature to minimize the total elastic energy of the ceramic structure. In this state, they do not present piezoelectric properties. When an electric field is applied, the domains align; this process is known as poling. When the electric field is removed, most domains remain aligned and piezoelectricity is induced. Generally the piezoelectric ceramics present a composition that implies a structural instability such as polymorphic phase transition (PPT) or morphotropic phase boundary (MBP) [7]. PPT-based systems are characterized by a large temperature dependence of piezoelectric properties and a rapid depoling, due to temperature changes close to PPT temperature. In the case of MBP-based materials, piezoelectric features are slightly influenced by temperature, but their working temperatures approach a half of Curie temperature. Among polycrystalline ferroelectric ceramics, the first system studied for its piezoelectric properties was barium titanate (BaTiO3) in 1943 and it remained the primary electroceramic material until the discovery of the lead zirconate titanate or PZT (PbZrxTi1-xO3) in 1954 [4]. The former is a PPT-based, highperformance piezoelectric ceramic, the latter presents a MBP from tetragonal to rhombohedral phase at about 48 mol% PbTiO3 content. For a long time PZT was the most used piezoceramic for its very attractive behavior in sensor applications [5, 7]. Notwithstanding this, in 2003 the European Union claimed PZT as a hazardous substance and research efforts were then focused on lead-free piezoceramics having high performances. For this reason, in the last decade, alternative systems were deeply studied modifying the composition of the traditional ferroelectric ceramics. In particular, this compositional engineering approach involved perovskite-based materials, exploiting their solid solutions with other ferroelectric systems or their doping, e.g.,
N
N
1768
the introduction of different types of cations within their structure. The perovskite structure ABO3 consists in a face-centered cubic array, in which the large cations A are placed at the corners of the unit cell, the smaller cations B in the body center, and the oxygen atoms in the center of each face, as illustrated in Fig. 1 [5–7]. Even if the high symmetry of the perovskite structure could lead to improved poling and consequently higher piezoelectric performances, currently no innovative lead-free ceramics are able to compare the piezoelectric and thermal behavior of PZT-based materials. On the other hand, a structural engineering approach, based on template grain growth, grain size optimization, and domain engineering, is able to enhance the electromechanical performances of existing piezoelectric materials. For instance, in the case of BaTiO3, a microstructural refinement can double the piezoelectric coefficient. In this case, the finer grain size implies a decrease of the domain size and an increase in density of domain walls, which are able to influence the piezoelectric properties [5, 8]. Moreover, a fourfold increase of the piezoelectric coefficient can be reached synergically combining the effect of submicrometric domain sizes and grain orientation obtained by template grain growth technique. The positive effect of the grain orientation was also demonstrated in the binary system (Bi1/2K1/2)TiO3–BaTiO3 (BKT–BT) [5]. Generally piezoelectric ceramics can be applied in sensors as bulk components or thin films, by using micromachining techniques [2, 3]. Thin films from one tenth to several tens of microns in thickness can present different properties with respect to the corresponding bulk material. This is due to the dependence of the piezoelectric features to the morphology and density of the film, as well as to the compositional aspects related to the stoichiometry and presence of impurities. Moreover, the piezoelectric behavior can also be influenced by the orientation and nature of the substrate material. All the deposition methods are based on three fundamental steps; firstly, the atomic or molecular species are yielded by chemical or physical processes. In the former case chemical vapor deposition or sol-gel techniques can be used; in the latter one, sputtering or pulsed laser deposition can be employed. After this step, deposition on the substrate is achieved through transport and condensation
Nanostructured Materials for Sensing
A
O B
Nanostructured Materials for Sensing, Fig. 1 The perovskite structure common in most ceramic piezoelectric materials (With permission from [5])
mechanisms and finally, the film is obtained by crystallization and annealing [3]. In order to achieve a long-term stability of piezoelectricity at room temperature and to develop flexible devices, piezoelectric polymers can be employed [9]. Generally ferroelectric materials based on polyvinylidene fluoride (PVDF), odd nylon, vinylidene cyanide (VDCN), and aromatic and aliphatic polyurea families are exploited [9, 10]. In the case of PVDF-based polymers, the residual polarization after poling is due to the alignment of CF2 dipoles present in their structure (CH2–CF2)n and trapped charges. Dipoles are oriented in the crystalline phases inside lamellae and in the interfaces between crystalline and noncrystalline phases, while the charge traps are present at the surface of crystallites [10]. The electromechanical performances of these polymeric materials can be improved by the polymerization of PVDF with other fluorinated polymers, which implies an increase of the crystallinity [9]. In the case of odd-nylon and polyurea families the fieldinduced orientation of N–H and C–O dipoles is stabilized by the hydrogen bonding present between adjacent polymer chains. On the other hand, in the case of amorphous polymers such as VDCN-based materials, an easier orientation of the C–CN dipoles is obtained by poling at temperatures near to the glass transition
Nanostructured Materials for Sensing
1769
particles in a polymer (0–3)
spheres in a polymer (1–3)
diced composite (1–3)
PZT rods in a polymer (1–3)
sandwich composite (1–3)
glass-ceramic (1–3)
transverse reinforcement (1–2–3–0)
PZT-polymer-air (1–3–1)
Moonie (3–0)
PZT-air (3–0)
perforated (3–2)
Replamine (3–3)
P3
P3
N
P3
honeycomb (3–1P)
honeycomb (3–1S)
perforated (3–1)
BURPS (3–3)y
Sandwich (3–3)y
Ladder Structure (3–3)y
Nanostructured Materials for Sensing, Fig. 2 Possible connectivity in bicomponent piezoelectric composites (With permission of [5])
temperature and freezing the residual polarization, through the cooling down of the system under the electric field [9, 10]. Among piezoelectric polymers, PVDF presents the best electromechanical behavior at room temperature, but its performances decrease at higher temperature [9]. This drawback can be partially limited through the copolymerization of PVDF [11]. Other ferroelectric polymers can be used, but they are characterized by less piezoelectric properties [9, 11]. Compared to the pure polymer performances, piezoelectric composites offer significantly higher functional properties. It is in fact possible to synergically couple the best features of the ceramic and polymeric phases while minimizing the poorest properties [4, 11]. Composites can be classified on the basis of the connectivity between the constituent phases according to Newnham notation [4]. This notation describes the number of dimensions each phase is physically in contact with itself. When just two components are considered, ten configurations are possible: 0-0, 1-0, 2-0, 3-0, 1-1, 2-1, 3-1, 2-2, 3-2, or 3-3, illustrated in Fig. 2 [5].
Among them, 0-3, 1-3, and 2-2 connectivity have been recently developed, in order to tailor the piezoelectric response through the exploitation of anisotropic composites having different ceramic phase contents [11]. In the last few years, miniaturization of devices based on piezoelectric behavior involves the exploitation of carbon nanotubes as functional materials [12].
Carbon Nanomaterials Multidimensional carbon nanomaterials include graphite, graphene, and carbon nanotubes. They are made of carbon atoms in an exagonal configuration, bonded each other with sp2 bonds [13]. Graphene is a two-dimensional sheet of carbon atoms and represents the repetitive unit of the graphite. Carbon nanotubes (CNT) consist in graphene sheets rolled to form cilynders, as shown in Fig. 3. In particular, carbon nanotubes can be classified as single-walled (SWCNT) or multi-walled (MWCNT); the former, consisting in
N
N
1770
Nanostructured Materials for Sensing
Nanostructured Materials for Sensing, Fig. 3 Structure of the multidimensional carbon nanomaterials: (a) CNT, (b) graphene, and (c) graphite (With permission of [13])
a single graphene cylinder, the latter, being made of concentric cylinders with an interlayer spacing of 0.34 nm. CNT nanostructure, atomic arrangement, diameter, and length strongly affect their mechanical, electric, and thermal properties [12]. Carbon nanotubes show an electromechanical behavior that, unlike to piezoelectric ceramics, is due to the phonon frequency shifts as a consequence of a charge injection [12]. This feature is mainly exploited in carbon nanotubes composites, which are characterized by a change in resistance when subjected to a chemical, thermal, or mechanical loading. Among them, the most promising sensing devices based on carbon nanotubes are the chemiresistive-type ones. In fact, their high sensitivity, small size, high surface area, and high aspect ratio imply the ability to detect smaller concentrations of analyte than traditional sensors. In comparison to carbon nanotubes, the homogenous distribution of electrochemically active sites at a nanometer scale and the higher purity could make graphene an ideal material for sensing and detection. Notwithstanding this, no consolidated studies have been carried out about its application in sensing devices to date [14].
Cross-References ▶ Active Carbon Nanotube-Polymer Composites ▶ Carbon Nanotubes ▶ Graphene ▶ NEMS Piezoelectric Switches ▶ Piezoelectric Effect at Nanoscale ▶ Piezoelectric MEMS Switch ▶ Sol-Gel Method ▶ Synthesis of Carbon Nanotubes ▶ Synthesis of Graphene
References 1. Fauchais, P., Montavon, G., Lima, R.S., Marple, B.R.: Engineering a new class of thermal spray nano-based microstructures from agglomerated nanostructured particles, suspensions and solutions: an invited review. J. Phys. D: Appl. Phys. 44, 53 (2011). 093001 2. Gibson, R.F.: A review of recent research on mechanics of multifunctional composite materials and structures. Compos. Struct. 92, 2793–2810 (2010) 3. Tadigadapa, M., Mateti, K.: Piezoelectric MEMS sensors: state-of-the-art and perspectives. Meas. Sci. Technol. 20, 30 (2009). 092001 4. Tressler, J.F., Alkoy, S., Newnham, R.E.: Piezoelectric sensors and sensor materials. J. Electroceram. 2, 257–272 (1998) 5. Cook-Chennault, K.A., Thambi, N., Sastry, A.M.: Powering MEMS portable devices-a review of non-regenerative and regenerative power supply systems with special emphasis on piezoelectric energy harvesting systems. Smart Mater. Struct. 17, 33 (2008). 043001 6. Rodel, J., Jo, W., Seifert, K.T.P., Anton, E.M., Granzow, T., Damjanovic, D.: Perspective on the development of leadfree piezoceramics. J. Am. Ceram. Soc. 92, 1153–1177 (2009) 7. Leontsev, S.O., Eitel, R.E.: Progress in engineering high strain lead-free piezoelectric ceramics. Sci. Technol. Adv. Mater. 11, 13 (2010). 044302 8. Setter, N., Damjanovic, D., Eng, L., Fox, G., Gevorgian, S., Hong, S., Kingon, A., Kohlstedt, H., Park, N.Y., Stephenson, G.B., Stolitchnov, I., Taganstev, A.K., Taylor, D.V., Yamada, T., Streiffer, S.: Ferroelectric thin films: review of materials, properties, and applications. J. Appl. Phys. 100, 46 (2006). 051606 9. Eberle, G., Schmidt, H., Eisenmenger, W.: Piezoelectric polvmer electrets. IEEE T. Dielect. El. In. 3, 624–646 (1996) 10. Fukada, E.: History and recent progress in piezoelectric polymers. IEEE T. Ultrason. Ferr. 47, 1277–1290 (2000) 11. Lushcheikin, G.A.: New polymer-containing piezoelectric materials. Phys. Solid State 48, 1023–1025 (2006) 12. Li, C., Thostenson, E.T., Chou, T.W.: Sensors and actuators based on carbon nanotubes and their composites: a review. Compos. Sci. Technol. 68, 1227–1249 (2008)
Nanostructured Thermoelectric Materials
1771
13. Pumera, M., Ambrosi, A., Bonanni, A., Chng, E.L.K., Poh, H.L.: Graphene for electrochemical sensing and biosensing. Trac-Trend. Anal. Chem. 29, 954–965 (2010) 14. L€u, P., Feng, Y.Y., Zhang, X.Q., Li, Y., Feng, W.: Recent progresses in application of functionalized graphene sheets. Sci. China Tech. Sci. 53, 2311–2319 (2010)
Nanostructured Solar Cells ▶ Nanomaterials for Excitonic Solar Cells
Nanostructured Thermoelectric Materials Menghan Zhou and Jian He Department of Physics and Astronomy, Clemson University, Clemson, SC, USA
Synonyms Nanostructured nanomaterials
thermoelectrics;
Thermoelectric
Definition Nanostructured thermoelectric materials are materials that contain nanoscale constituents and exhibit enhanced thermoelectric performance due to nanoscale phenomena.
Introduction To a great extent, the development of civilization has evolved with the technological advent of energy production, storage, delivery, and use. As electricity will remain the most convenient form of energy in the foreseeable future, eco-friendly technology to generate electricity is of the utmost importance. Heat, however, has always been an abundant but low-quality form of energy as more than half of all the energy generated by mankind is lost as heat. As such, harvesting and converting the waste heat emanating from many different sources (e.g., solar, geothermal, and exhaust
N
from automobiles or other industrial processes) into electricity is highly desirable. One emerging market is to harvest the large amount of waste heat (2/3 of the generated power) from an automobile’s engine and convert it into “onboard” electrical energy using thermoelectric (TE) devices. Meanwhile, there has always been a strong need for solid-state cooling (heat management) in microelectronics, which often requires responsive spot size cooling instead of traditional environmental cooling. This is best satisfied by TE refrigeration, the gain from temperature stabilization and device performance can be significant. The simplest technology applicable for direct heatelectricity energy conversion is thermoelectricity [1]. Thermoelectricity is based on two basic effects, the Seebeck effect: generating electricity from a temperature gradient (“power generation mode”), and the Peltier effect: generating a temperature gradient when electrical current is applied (“refrigeration mode”). As shown in Fig. 1 [2], a basic thermoelectric couple consists of one leg made of n-type material and one leg made of p-type material in order to maximize the amount of power generated or the amount of heat absorbed, many of these couples are connected electrically in series and thermally in parallel to form a thermoelectric (TE) module or device. The working media of TE materials are the electrons, rather than the atomic or molecular gases or liquids used in mechanical power generation and refrigeration systems. Accordingly, the TE device is a solid-state assembly without moving parts or green house emissions, lightweight and compact, responsive, and feasible for miniaturization, the TE device can readily work in tandem with other alternative energy technologies, such as photovoltaics. This is an important feature because no single technology can meet the world’s energy needs in the twentyfirst century; a combination of many technologies is necessary. Given the ubiquitous heat sources and the modular aspects of the TE device, thermoelectricity is guaranteed a position as an alternative and complementary energy technology of the twenty-first century. At present, however, the conversion efficiency of thermoelectricity is inferior to that of equivalent mechanical systems by a factor of 2–4. For this reason, thermoelectricity has been long confined to niche applications, where the issue of efficiency is less of a concern than the issues of energy availability,
N
1772
a
b + Electrical power input Current
Heat input
Heat ejected
Hot junction
Hot junction
n-type
p-type
p-type
Current
Nanostructured Thermoelectric Materials, Fig. 1 The basic thermoelectric module, made of n-type and p-type materials, can work either in the (a) power generation mode or (b) refrigeration mode; and (c) a thermoelectric device is made of many basic modules [2]
Nanostructured Thermoelectric Materials
Cold junction
n-type
N
Cold junction
c
Ceramic plate
Heat absorbed
Heat ejected Contact pads n-type Electrical power output
p-type
reliability, compactness of the device, and quiet operation. For example, radioisotope TE power generators are the primary power source used in the Voyager and Cassini spaceships for NASA’s deep-space missions beyond Mars, where fuel cells, solar cells, and nuclear power are not feasible. The efficiency of a TE device can be calculated by regarding it as a heat engine working between a heat reservoir and a heat sink. The maximum efficiency of power generation, , and that of TE refrigeration, f, of a TE device [1] is determined by the Carnot efficiency, Carnot, and the dimensionless figure of merit, ZT, of the TE materials of which it is composed. ¼ Carnot TE
2 3 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ ZTm 1 5 Thot Tcold 4 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ Tcold Thot 1 þ ZTm Thot þ
f¼
2 ffi Thot 3
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 þ ZT m Tcold 4 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tcold 5 Thot Tcold 1 þ ZTm þ 1
(1)
(2)
The Carnot efficiency, Carnot, is the ratio of the temperature difference between the hot-end
temperature, Thot, and the cold-end temperature, Tcold, to Thot, and Tm is the mean temperature [1]. ZT is the material parameter of prime importance in thermoelectrics study: ZT ¼
sa2 T PF sa2 T
¼ ¼ k k kph þ ke
(3)
where s is the electrical conductivity, a the thermopower (i.e., Seebeck coefficient), k the total thermal conductivity (often expressed as the sum of the lattice thermal conductivity, kph, and the carrier thermal conductivity, ke), PF the power factor, and T the temperature in Kelvin. The Wiedemann–Franz relationship, ke ¼ LsT, is often used to estimate ke, where L is the Lorenz number. Although there is no theoretical upper limit for ZT values, the current stateof-the-art bulk TE materials have a maximum ZT 1 in their respective temperature ranges of operation because the physical quantities (s, a, and k) that govern ZT are inherently interdependent: A helpful modification of any of these quantities often adversely affects the others. These materials ideally have conversion efficiencies of 7–15% depending on the specific materials and the temperature differences involved.
Nanostructured Thermoelectric Materials
Nanostructured Thermoelectrics: Big Gains from Small Sizes Equations (1)–(3) shift the fundamental issue of TE energy conversion to the materials’ physical transport properties. As such, modern thermoelectrics studies are materials oriented; tremendous efforts have been contributed continuously to developing high ZT TE materials. A high ZT TE material is ideally a “phonon-glass electron-crystal” (PGEC) system that also possesses a large thermopower. A PGEC system is a system that simultaneously possesses a glass-like poor lattice thermal conductivity and a crystal-like good power factor. Over the past decade, the efforts of pursuing higher ZT materials have culminated into a two-pronged strategy, depending on the type of material studied. One approach is to develop novel bulk materials in line with the concepts of “PGEC,” while the other is to search for higher ZT in nanostructured thermoelectric materials (NTMs) that possess reduced dimensionality or nanometer characteristic lengths [3, 4]. The main themes of this entry are the strategies, implementations, and challenges of NTMs. Interested readers are referred to other reviews [5–7] and the references therein for more details about the topics discussed herein. Classical and Quantum Size Effects The paradigm of NTMs dates back to the early proposition by Hicks and Dresselhaus [3] that a low-dimensional material may have enhanced TE properties compared to its bulk counterpart due to quantum size effects. Low-dimensional material is the material whose system size in one or more directions is comparable with the wavelength or the mean free path of a quantum particle or an excitation. Accordingly, the charge-carrying electrons and heatcarrying phonons may be subject to different size effects, providing an extra layer of control of the charge and heat flows via altering the characteristic length scales of a low-dimensional (“confined”) system. The experimental proof-of-principle studies in “custom nano-engineered” quantum dot and superlattice systems [4] show that (i) enhancement of s and a can be achieved simultaneously with the reduction of k in a nanostructured material, (ii) the degrading of the electrical transport property is
1773
N
minimized by an optimal choice of band offsets in these heterostructures, and (iii) the primary gain in ZT is from the reduction of kph, rather than the predicted enhancement of the PF, which has been attributed to the less-dispersed phonon modes or phonon scattering due to interfacial lattice mismatch. To date the record high ZT is still held by superlattice and quantum dot systems, but the present cost of the nanofabrication processes restricts the large-scale commercial use of these “custom nano-engineered” systems. Theoretical studies of the mechanism of heat conduction in superlattice systems suggest that the thermal boundary resistance and a high density of interfaces, rather than the atomically perfect interfaces or the periodicity of nanostructures, accounts for such a significant reduction of kph. It naturally leads to the idea of using bulk nanostructured materials (namely, nanocomposites) as a potential costeffective and scalable alternative to these “custom nano-engineered” systems for high ZT materials [8]. Important advantages of nanocomposites include being made using a bulk synthesis procedure and in a form compatible with existing commercial TE devices, possessing better mechanical properties, and being more isotropic. Despite the variety in micro-morphologies and TE properties of superlattice, quantum dot systems, and nanocomposites, the mechanisms underlying the enhancement of ZT are largely the same: While classical size effects limit the mean free paths of both electrons and phonons, quantum size effects may alter the electron band structures or phonon dispersion relations individually. As shown Fig. 2 [6], many advances in the enhancement of ZT are attained in these nanostructured thermoelectric materials (NTMs). It is not always easy to distinguish classical and quantum effects. For example, the wave-particle dual nature of electrons may manifest itself in nanostructures, where the electron transport is between the coherent transport mode as a wave and the incoherent transport mode as a particle. Nonetheless, the combined and quantum size effects can make the physical properties of a NTM so different from its bulk counterpart(s) that the NTM is eligible to be treated as a new material, even they may have the same chemical composition. Si is a good example [5, 6].
N
N
1774
Nanostructured Thermoelectric Materials Bi2Te3/Sb2Te3 SL - Venkatasubramanian et al. PbSeTe NDSL - Harman et al. Si NW - Boukai et al. Si NW - Hochbaum et al. BiSbTe - Poudel et al. AgPbmSbTem+2 - Hsu et al. Na1−xPbmSbyTem+2 - Poudeu et al.
2.5 n-type
PbSnTe-PbS - Androulakis et al.
2.0
In4Se3 Rhyee et al.
p-type
Tl-PbTe - Heremans et al.
1.5
CsBi4Te6 Chung et al.
Bi2Te3 Goldsmid et al.
Ba8Ga16Ge30 - Saramat et al. SiGe - Wang et al.
ZT
Bi2Te3 Goldsmid et al. PbTe QW - Harman et al. 1.0
0.5
0 1950 1960 1970 1980 year
1200 900
1990 600
2000 2010
300
ture
(K)
a per
Tem
Nanostructured Thermoelectric Materials, Fig. 2 ZT of the state-of-the-art TE materials, both n-type and p-type, as a function of temperature and year [6]. Many of recent advances
in the enhancement of ZT are achieved in NTMs, while it should be noted that some results have not yet been reproduced or verified
Micromorphology and Preparations of NTMs The micromorphology of a NTM determines the material’s TE properties. Hence creating a controlled micromorphology using a bulk synthesis procedure is crucial for the development of NTMs. As shown in Fig. 3 [9], the micromorphologies of NTM are comprised of three major aspects: (i) the composition, structure and characteristic dimension of nanostructures; (ii) the connectivity of nanostructures (i.e., the topology); and (iii) interfaces [10]. When the characteristic length scale becomes smaller, the aspects (ii) and (iii) often become more important than the aspect (i) in determining the TE properties of the NTMs. In certain cases, interfaces should be treated as an individual phase. Note that the effective medium theory suggests that the ZT of a composite cannot exceed the best performing constituent if there is no contribution from the interfaces.
The recent advent of nanomaterials and nanotechnology provides a variety of ways to prepare NTMs, to name a few: molecular epitaxial growth, laser ablation, chemical vapor transport, vapor–liquid–solid growth, etching, wet chemistry and hydrothermal synthesis, ball milling, mechanical alloying, melt spinning and phase separation. High-energy ball milling is an effective scalable method to produce large amount of nanoparticles of a variety of compositions. Another notable nanostructuring method is melt spinning, which can create rich nanostructures via highly dynamic quenching process. The current focus is mainly on nanostructuring bulk materials with well-known TE properties, such as Bi2Te3, PbTe, CoSb3, and SiGe. The impacts of nanostructuring are thus amenable to a certain degree of analysis, prediction, and optimization. In particular, nanocomposites can be prepared by means of ex situ or in situ methods. The ex situ approach
Nanostructured Thermoelectric Materials
1775
N
Nanostructured Thermoelectric Materials, Fig. 3 Typical micromorphologies of NTMs [9]
N is typically performed in a two-step fashion: Nanomaterials are first prepared using the aforementioned methods, and then densified into bulk objects [11]. The two most widely used densification methods are hot pressing (HP) and spark plasma sintering (SPS). The SPS process can densify nanoparticles in a relatively short time, compared to HP process, to avoid the interdiffusion and growth of grains. The main drawback of the ex situ method is the lack of coherence in interfaces, which may severely degrade the carrier mobility. An alternative way is the in situ or self-assembled preparation [12], including precipitation nucleation and growth, spinodal decomposition, and matrix encapsulation (“arrested precipitation”), with the resulting interfaces being mostly coherent or semi-coherent.
Strategies and Implementation of Nanostructuring This section is devoted to several important strategies and implementation of nanostructuring, each of which
is directed toward one of those physical transport quantities that determine ZT (Eq. 3). The rationale behind these strategies and implementation is in line with the PGEC concept. Reducing the Lattice Thermal Conductivity Although the paradigm of NTMs started with the theoretical speculation that quantum confinement may induce enhanced electrical transport properties, to date almost all high-performance NTMs are a result of a very low kph. An interesting question thus concerns the lower limit of kph in NTMs. In a bulk material, the lower limit of kph was set by the so-called alloy limit obtained by amorphous solids, in which the phonon mean free path is close to half the phonon wavelength. Historically, the alloying approach was employed to reduce kph as the atomic scale lattice imperfections can strongly scatter short-wavelength phonons, but alloying can often lead to deterioration in electrical conductivity. A somewhat oversimplified relationship, kph (1/3)(vSCVLph), is often employed to estimate the phonon mean free path, Lph, where vS is
N
1776
Nanostructured Thermoelectric Materials
b 100
100
90
90
80
80
Percent Accumulation
Percent Accumulation
a
70 60 50 40 30 20
0
100
101
60 50 40 30 20
Callaway MD Simulation
10
70
Callaway MD Simulation
10 102
Wavelength (nm)
0
10−2
100
102
104
Mean Free path (μm)
Nanostructured Thermoelectric Materials, Fig. 4 Cumulative distribution of kph of pure bulk Si with respect to (a) wavelength and (b) mean free path predicted by the Callaway model and a more exact molecular dynamics simulations [13]
the sound velocity and CV is the heat capacity at a constant volume. This relationship is derived from the classical kinetic theory of heat conduction, regardless of the details of the scattering mechanisms. At temperatures comparable and above the Debye temperature, yD, both vS and CV are weakly dependent on temperature, kph is thus governed by Lph. At T 3 is certainly desired but the return in the conversion efficiency becomes marginal [1]. At present, unless people can employ coherent or correlated phonon scattering effects to actively reduce the group velocity and the number of heat-carrying phonon modes [6], any further reduction of kph below the “alloying limit” is not only difficult but also insufficient to reach a ZT 3. Due to the interdependence of s and ke, the enhancement in the PF must come mainly from the enhancement in the thermopower. In general, the magnitude of thermopower is a measure of the entropy per unit charge carrier. For the purpose of illustration, one can estimate the minimum thermopower required for a viable TE material by assuming a hypothetical material in which kph is zero and the Wiedemann–Franz relationship holds. As such, Eq. 3 is rewritten as ZT ¼ a2/L, requiring that a 157 mV/K for a ZT ¼ 1, 225 mV/K for a ZT ¼ 2, and 275 mV/K for a ZT ¼ 3. Multifold Preferential Scattering at Interfacial Boundaries
There are numerous results underscoring the important role of interfacial boundary scattering [10]. A well-known mechanism is the energy-dependent scattering of electrons, namely, the electron-filtering mechanism [4]. The introduction of appropriate energetic barriers at interfaces restricts the energy of carriers entering a material. At an interface, those carriers with a mean energy substantially above the Fermi level, EF, will pass through the interface preferentially, thereby enhancing the thermopower that directly depends on the excess energy (E–EF) of charge carriers [7].
1777
N
This leads to an important strategy that implements multifold preferential scattering at interfacial boundaries: scattering phonons more effectively than electrons as for a “PGEC”; and scattering those carriers with a mean energy substantially lower than the Fermi level more effectively than those with higher mean energy to ensure a large thermopower. Certainly, the interfacial boundary may somewhat degrade the electrical conductivity, but the effect should not be significant because most TE materials are heavily doped and the electron mean free path is very short. Such multifold preferential scattering mechanisms can be implemented via controlling the physical parameters of the barrier at interfacial boundaries [10]. Electron Band Structure Engineering Based on semiclassical theory, Pichanusakorn and Bandaru recently proposed that [7] (i) the maximum PF of a bulk material is a function of the reduced Fermi level, an energy-dependent carrier scattering constant, and the dimensionality; (ii) for any bulk materials and at any temperatures, the PF maximizes when the magnitude of thermopower is in the range of 130–187 mV/K, while the magnitude of PF can be increased by tuning the detailed material parameters; (iii) in both bulk and nanostructured materials, the PF is eventually limited by the onset of the bipolar conduction; and (iv) the PF of a NTM can be enhanced beyond the bulk value, say, via quantum-confinement-induced enhancement of magnitude of the DOS, rather than the change in shape of the DOS. Several Transformative Concepts There are several interesting concepts that are viable to implement in conjunction with the nanostructuring approach. In a recent work, Heremans et al. [15] showed that a 2% doping of Tl in bulk PbTe significantly increased ZT from 0.7 to 1.5, which has been attributed to the increase in DOS(EF) induced by resonant levels. Also in the PbTe system, Pb and Sb co-nanostructuring was found to significantly enhance the electron mobility and lead to a ZT 1.4, while this phenomenon was not observed with Pb or Sb precipitates alone [8], the underlying mechanism are not clear. Another promising class of materials are the strongly correlated semiconductors, in which the virtual excitations from d- or f-levels into the conduction band give rise to a sharp feature in the DOS(EF), and a giant thermopower at low temperatures. For example, despite a very high kph, FeSb2 exhibits a colossal thermopower and
N
N
1778
a record high PF at low temperatures [16], the PF is about 65 times higher than that of Bi2Te3. The high PF at low temperatures makes these materials prime candidates for transient TE cooling (supercooling in a Peltier cooler). There are currently very few materials, such as BiSb and CsBi2Te3, available for cooling in the temperature range below room temperature [1].
Challenges in Nanostructured Thermoelectrics The effectiveness of “nanostructuring” as a means to improve ZT has been corroborated by numerous studies. While a systematic search and subsequent thorough investigation of promising materials may eventually yield the desired NTMs, the present trial-and-error approach is not sufficient, and the concerted efforts of theoreticians and experimentalists are needed to address the key question as to why certain nanostructures are thermoelectrically favorable while others are not. Some of the challenges will be briefly addressed in this section. Understanding the Phonon and Electron Transport Current understanding of phonon and electron transport in NTMs is fairly limited. This lack of understanding affects the ability to improve and design higher performance NTMs. For example, although it is experimentally shown that NTMs are able to exhibit kph lower than the “alloy limit,” and the multiscale structural complexity can suppress kph by strongly scattering a wide spectrum of phonons, the existing models largely fail to provide satisfactory predictions for the spectral dependence of kph or the phonon transmissivity at interfacial boundaries. A pertinent question is what is the lower limit of kph? [6, 13]. There is also much work to be done on understanding electron transport in NTMs. The primary difficulties are (i) the standard analysis based on the Boltzmann transport equation loses its validity when the characteristic length scale of NTMs is comparable or less than the electron de Broglie wavelength; and (ii) modeling the electron transport across the interfacial boundaries. Models based on molecular dynamics and nonequilibrium Green’s functions are promising options. Creation and Evolution of Nanostructures At present, the factors that control the nanostructure formation and evolution in solids are not well
Nanostructured Thermoelectric Materials
understood. More importantly, nanostructures are a metastable state of the material. As NTMs will work under high temperatures and temperature gradients, it is important to study and optimize their long-term stability, especially the impacts of interdiffusion and coarsening on the TE performance of the NTMs. It is noted that the nanoinclusions in the AgPbmSbTem+2 (LAST-m) NTMs are thermodynamically and kinetically stable to at least 800 K over a long period of time [8]. Cost of Raw Materials Even with the desired improvement in ZT, a TE device can only compete with the equivalent mechanical systems in some decentralized applications with a power scale from mW to kW. After performance, the cost of raw materials and fabrication is a major consideration for any large-scale industrial application of NTMs. For example, the price of tellurium (Te) has significantly increased in the past few years due largely to the use of Te in solar cells. This somewhat vitiates the enhancement of ZT enhancement in NTMs and especially limits the applications of Te-containing NTMs such as those based on Bi2Te3 and PbTe. High-performance NTMs based on abundantly available and nontoxic elements are crucial for scaled-up applications of NTMs in power generation and refrigeration. Figure 5 presents a histogram of the relative abundance of the elements [17].
New Concepts The advance in thermoelectrics studies hinges primarily upon the discovery of novel materials and novel concepts that implement decoupling the interrelated TE quantities (s, a, and k) in order to optimize the charge flow and entropy flow (the magnitude of a is a measure of the entropy per charge carrier) in the material. Spin Caloritronics The charge and lattice are the two degrees of freedom most relevant to the conventional thermoelectricity while the role of the spin degree of freedom has received much less attention with the exception of the layered cobaltates. A notable new area is spin caloritronics, basically a combination of spintronics and thermoelectrics. While the central theme of
Nanostructured Thermoelectric Materials
N
1779
Nanostructured Thermoelectric Materials, Fig. 5 Relative abundance of the elements in a logarithmic scale [17]
Na
Mg
K Ca Ti Sc
V
Rb Sr
Al
Si N
Y Zr Cs
Co Ni
Nb
Ba La Hf
Mo Ta
Cu
P Zn
Ru Re
Os
Ir
Rh Pt
Pd Ag Cd In Au Hg Tl
F S
Cl
Ga
Ne Ge
W
O
B
Fe Cr Mn
Ar
As
Sn
Br
Se Sb
Pb
Te
Kr
I Xe
Bi Rn
N
Nanostructured Thermoelectric Materials, Fig. 6 (a) Screw-type lattice dislocations (shown in the inset) form topologically protected 1-D conduction paths in a 3-D topological
insulator. (b) The contour plot of room temperature ZT as a function of chemical potential, m, and the density of lattice dislocations, n, for a sample’s length of 1 mm [20]
thermoelectrics is controlling charge flows by means of heat flows and vice versa, that of spintronics is the active manipulation of the population and the phase of spins, and that of spin caloritronics is the controlling of spin currents by means of heat flows and vice versa. The recently discovered spin Seebeck effect is conceptually different from the traditional charge Seebeck effect in that the spin Seebeck effect is not related to the charge-particle diffusion [18]. Uchida et al. [19] proposed an exciting possibility of using the spin Seebeck effect in magnetic insulators, in combination with the inverse spin Hall effect, to build TE devices “operating on an entirely new principle.”
Topological Insulators Beyond the local order parameter and long-range correlation in the Landau symmetry-breaking description, topological order results from long-range quantum entanglement of quantum states and is described by a new set of quantum numbers (variables), such as ground state degeneracy, quasiparticle fractional statistics, edge states, and topological entropy. Topological insulators (TIs) have recently attracted great attention in condensed matter physics. Among the state-of-the-art bulk TE materials, Bi2Te3, (Bi,Sb) and certain half-Heusler compounds have been recently identified as TIs. With the bulk insulating and topologically protected surface conducting nature, TIs have been suggested to be promising candidates for
N
1780
high-performance TE materials. As shown in Fig. 6, a possible approach is to fabricate TI nanocomposites to combine enhanced electric conductivity and thermopower from the topologically protected 1-D channels (engineered line dislocations) and suppressed thermal conductivity (due to the vast phonon scattering at the interfaces and the insulating bulk states) [20].
Perspective Remarks As a fundamental phenomenon discovered in the nineteenth century, thermoelectricity has been widely used in thermocouples for temperature measurements, but it has only found some niche applications in power generation and refrigeration in the twentieth century. The current state-of-the-art bulk TE materials such as Bi2Te3, PbTe, CoSb3, and SiGe have been in the place for 20–30 years; expansion of thermoelectricity into broader applications is limited by the low ZT of materials. This could soon change due to the progress in the emerging field of NTMs in the past decade. Quantum and classical size effects in NTMs offer a means to tune the thermoelectric conversion efficiency beyond that of bulk materials. It is plausible to expect the recent advances in NTMs, especially the creative preparation methods, new materials, novel concepts, and better understanding, will lead to eco-friendly, cost-effective, higher performance thermoelectricity and a broader application of TE power generation and refrigeration. Acknowledgments M.H. Z. and J. H. acknowledge Drs. Don Liebenberg and Malcolm Skove for useful discussions, the support of the DOE/EPSCoR Implementation Grant (#DE-FG0204ER-46139), and the SC EPSCoR cost-sharing program.
Cross-References ▶ Self-assembly of Nanostructures ▶ Thermal Conductivity and Phonon Transport ▶ Thermoelectric Heat Convertors
References 1. Nolas, G.S., Sharp, J., Goldsmid, H.J.: Thermoelectrics basic principles and new materials developments. Springer, Berlin/Heidelberg (2001)
Nanostructured Thermoelectric Materials 2. Vining, C.B.: Semiconductors are cool. Nature 413, 577–578 (2001) 3. Hicks, L.D., Dresselhaus, M.S.: Effect of quantum-well structures on the thermoelectric figure of merit. Phys. Rev. B. 47, 12727–12731 (1993) 4. Dresselhaus, M.S., Chen, G., Tang, M.Y., Yang, R., Lee, H., Wang, D., Ren, Z.F., Fleurial, J.-P., Gogna, P.: New directions for low-dimensional thermoelectrics. Adv. Mater. 19, 1043–1053 (2007) 5. He, J., Tritt, T.M.: Thermal-to-electrical energy conversion from the nanotechnology perspective. In: Garcia-Martinez, J. (ed.) Nanotechnology for the energy challenge, pp. 47–78. Wiley-VCH, Weinhaim, Germany (2010) 6. Vineis, C.J., Shakouri, A., Majumdar, A., Kanatzidis, M.G.: Nanostructured thermoelectrics: big efficiency gains from small features. Adv. Mater. 22, 3970–3980 (2010) 7. Pichanusakorn, P., Bandaru, P.: Nanostructured thermoelectrics. Mater. Sci. Eng. Rep. 67, 19–93 (2010) 8. Kanatzidis, M.: Nanostructured thermoelectric: the new paradigm? Chem. Mater. 22, 648–659 (2010) 9. Li, J., Liu, W., Zhao, L., Zhou, M.: High-performance nanostructured thermoelectric materials. NPG Asia Mater. 2(4), 152–158 (2010) 10. Medlin, D.L., Snyder, G.J.: Interfaces in bulk thermoelectric materials: a review for current opinion in colloid and interface science. Curr. Opin. Colloid Int. Sci. 14, 226–235 (2009) 11. Lan, Y., Minnich, A.J., Chen, G., Ren, Z.F.: Enhancement of thermoelectric figure-of-merit by a bulk nanostructuring approach. Adv. Funct. Mater. 19, 1–20 (2009) 12. Sootsman, J.R., Chung, D.-Y., Kanatzidis, M.G.: New and old concepts in thermoelectric materials. Angew. Chem. Int. Ed. 48, 8616–8639 (2009) 13. Minnich, A.J., Dresselhaus, M.S., Ren, Z.F., Chen, G.: Bulk nanostructured thermoelectric materials: current research and future prospects. Energy Environ. Sci. 2, 466–479 (2009) 14. Humphery, T.E., Linke, K.: Reversible thermoelectric nanomaterials. Phys. Rev. Lett. 94, 096601(1)–096601(4) (2005) 15. Heremans, J.P., Jovovic, V., Torberer, E.S., Saramat, A., Kurosaki, K., Chroenphaskdee, A., Yamanaka, S., Snyder, G.J.: Enhancement of thermoelectric efficiency in PbTe by distortion of the electronic density of states. Science 321, 554–557 (2008) 16. Bentien, A., Johnsen, S., Madsen, G.K.H., Iversen, B.B., Steglich, F.: Colossal Seebeck coefficient in strongly correlated semiconductor FeSb2. Euro. Phys. Lett 80, 17008(1)– 17008(5) (2007) 17. Vaqueiro, P., Powell, A.V.: Recent developments in nanostructured materials for high-performance thermoelectrics. J. Mater. Chem. 20, 9577–9584 (2010) 18. Jaworski, C.M., Yang, J., Mack, S., Awschalom, D.D., Heremans, J.P., Meyers, R.C.: Measurement of the spinSeebeck effect in a ferromagnetic semiconductor. Nat. Mater. 9, 898–903 (2010) 19. Uchida, K., Xiao, J., Adachi, H., Ohe, J., Takahashi, S., Leda, J., Ota, T., Kajiwara, Y., Umezawa, U., Kawai, H.,
Nanostructures Based on Porous Silicon Bauer, G.E.W., Maekawa, S., Saitoh, E.: Spin Seebeck insulator. Nat. Mater. 9, 894–897 (2010) 20. Tretiakov, O.A., Abanov, Ar, Murakami, S., Sinova, J.: Large thermoelectric figure of merit for three-dimensional topological Anderson insulators via line dislocation engineering. Appl. Phys. Lett. 97, 073108(1)–073108(3) (2010)
1781
N
be produced in many different ways, not only (electro) chemical ones, and ordered (electro) chemical etching is in principle possible.
Overview
Nanostructured Thermoelectrics ▶ Nanostructured Thermoelectric Materials
Nanostructures Based on Porous Silicon Luca Boarino and Giampiero Amato Quantum Research Laboratory, Electromagnetism Division, Istituto Nazionale di Ricerca Metrologica, Torino, Italy
Definition The term Porous Silicon (or, the more general, Porous Semiconductor) has been and is largely used to define a class of semiconducting material in which pores are opened, basically by means of electrochemical or chemical processes. According to the present definition (and to IUPAC), three different material types are identified, say, nanoporous, mesoporous, or macroporous. The important point is that these types are defined according to the dimension of the pores, instead of the residual structures. In practice, a correlation between pore diameters and structure sizes is found, e.g., ensembles of macropores and nanostructures are found very seldom, and are the product of more complicated technological steps, not the simple (electro)chemical etching. The term Porous Silicon has been extended in the past to define some kinds of disordered arrays of Silicon nanowires. This extended definition does not keep into account for some regular macroporous structures obtained by means of electrochemical etching coupled with photolithography. Moreover, disorder is a typical consequence for the (electro) chemical processes employed, and cannot be considered as distinctive for the material. In other words, nanostructured arrays can
Historical Notes Porous silicon (PS) was accidentally discovered by Arthur and Ingenborg Uhlir at the Bell Labs, in 1956, in the framework of a research work on silicon and germanium electropolishing. If the electropolishing current was not reached for the electrochemical etching of the silicon wafer, instead of a clean and polished surface, a brown or black layer appeared, due to a nonuniform dissolution of silicon atom (see Fig. 1). The material is porous, full of fine holes, and the electropolishing regime is reached only at the pore tips in these conditions. This former evidence of PS was published in the Bell Lab’s technical notes [1] and the study of this system abandoned until the 1970s–1980s, when the huge specific surface of the material, of the order of hundreds of square meters per square centimeter gave origin to spectroscopic studies, as a precursor for thick oxide growth on silicon and to the first applications to gas sensing. In 1977 the first patent of a humidity sensor based on porous silicon was deposited [2]. PS was also applied to Silicon On Insulator (SOI) technology [3]. At the end of the 1980s, Leigh T. Canham working at the British Defense Research Agency obtained the experimental confirmation of his intuition that the nanofilaments of porous could exhibit good luminescence properties, due to quantum confinement effects. Also other authors were investigating on the phenomenon, V. Lehmann and U. Go¨sele, but Canham was the first to publish, in 1990, the experimental evidence of PS room temperature photoluminescence [4, 5]. This experiment stimulated the work of many research groups all over the world in the nonlinear optical and electrical properties, and the number of published papers arose exponentially from 1991 to 1995. The premises of a brilliant future for silicon-based lasers were not maintained, due to the low efficiency of electroluminescence of this material, slow decay times and to the chemical and mechanical stability. In the mid 1990s, the main applications of porous silicon were micromachining, gas sensing and with
N
N
1782
Nanostructures Based on Porous Silicon
PVC (Fig. 2). To obtain homogeneous and thick layers, a pulsed bias is preferable, so giving time to the solution at the pore tips to recover its title by Fluorine ions diffusion from the bath. The ethanol is added to the hydrofluoric acid to favor the hydrogen bubbles removal from the porous silicon layer during formation. Formation mechanisms The selective removal of the silicon atoms (local electropolishing at the pore tips) or complete electropolishing is performed by the Fluorine ions in presence of two or four holes in the bulk of the semiconductor, allowing the breaking of the Si-Si backbonding (Fig. 3). After the initial work of Turner, Memming, and Schwandt proposed the overall dissolution reaction: Nanostructures Based on Porous Silicon, Fig. 1 Porous Silicon formation voltammogram. Over the IPS, electropolishing regime occurs. At lower currents, porous silicon is formed
1995 the medical and biological applications received a big boost, due to the demonstration of hydroxyapatite growth on porous silicon end to the proven biocompatibility and hemocompatibility of nanostructure silicon, again by L. T. Canham [6]. Other studies showed that B50 rat hippocampal cells have a better adhesion onto porous silicon respect to untreated surfaces, so indicating suitability of porous silicon for cell culturing purposes. In 2001 a group of researchers at the Technical University of Munich [7] announced that hydrogenated porous silicon, in contact with liquid oxygen, reacts explosively, releasing more energy than an equivalent amount of TNT with higher speed, and with no necessity of a detonating material also for small quantities. The application of this discovery, substituting the liquid oxygen requiring very low temperatures with salts rich in oxygen, is in the air bag igniters in the automotive field, in the satellite propulsion, and in atomic emission spectroscopy [8]. Fabrication of Porous Silicon Anodization
Porous silicon is realized by electrochemical anodization of a silicon substrate, in a solution of HF and Ethanol in different ratio, depending on the silicon wafer resistivity on the desired etch rate and thickness and characteristics of the porous layer. The cathode is a platinum wire or a grid; the beaker can be Teflon or
Si þ 2HF þ lhþ ! SiF2 þ 2Hþ þ ð2 lÞe
(1)
SiF2 þ 2HF ! SiF4 þ H2
(2)
SiF4 þ 2HF ! H2 SiF6
(3)
Where h+ is the hole, e- the electron, and l is the number of charges exchanged. Similar mechanisms have been proposed in literature, including silicon oxide formation in presence of H2O and dissolution in presence of HF, like in the case of the electropolishing regime: Si þ 2H2 O ! SiO2 þ 4Hþ þ 4e E ¼ 0:84V=NHEðNormal Hydrogen ElectrodeÞ (4) SiO2 þ 2HF2 þ 2HF ! SiF6 2 þ 2H2 O
(5)
Stain Etching
Porous silicon formation occurs also in absence of an external biasing, by stain etching with hydrofluoric acid, nitric acid, and water. The vacancies necessary to silicon dissolution in this case are provided by HNO3, the etchant concentrations are generally higher, and the etch rates are lower than in the electrochemical process. The first studies were carried on during the 1960s to probe p-n junction, then the discovery of room temperature photoluminescence of porous silicon obtained electrochemically renewed the interest for stain etching, and
Nanostructures Based on Porous Silicon
1783
N
Nanostructures Based on Porous Silicon, Fig. 2 Electrochemical cell scheme for Porous Silicon fabrication, (exploded, left, and assembled, right)
Nanostructures Based on Porous Silicon, Fig. 3 Dissolution sequence of silicon in presence of HF and holes
a systematic study regarding the structural and optical properties of porous films have been exploited [9]. The reactions are the following: Anode: Si þ 2H2 O þ mhþ ! SiO2 þ 4Hþ þ ð4 mÞe (6) SiO2 þ 6HF ! H2 SiF6 þ 2H2 O
(7)
HNO3 þ 3Hþ ! NO þ 2H2 O þ 3hþ
(8)
Cathode:
Overall: 3Si þ 4HNO3 þ 18HF ! 3H2 SiF6 þ 4NO þ 8H2 O þ 3ð4 mÞhþ þ 3ð4 mÞe (9) Recently, K. Kolasinsky [10] developed a new type of PS stain etching oxidants like, Fe(III), Mn(VII), and Cr(VI) drastically reducing hydrogen bubbling so obtaining very homogeneous and highly luminescent materials.
Metal-Assisted Etching
Metal-assisted etching (MAE) has been recently reported, by Li, Bohn and coworkers [11], it is a sort of self-catalytic etching of silicon, with no need of polarization, since the formation mechanism is ruled by local redox potential of the metal respect to silicon. This drives the holes necessary to dissolution at the metal-semiconductor interface. Noble metals like gold, silver, or platinum can be used and obtained by precipitation starting from liquid metallic salts like AgNO3 and others, but also deposited in form of thin solid films by sputtering or evaporation. The metal clusters at the surface of the silicon wafer, rapidly sink following the crystallographic directions, and define silicon walls or wires (Fig. 4), depending on the metal particle distribution at surface, on the temperature and concentration of the etchants. When the metal is patterned on the surface, MAE leads to regular morphologies. The dissolution of silicon in HF and an oxidizing agent is a combination of chemical and electroless process, having strong similarities with porous silicon formation or electropolishing by stain etching in HF–HNO3 [12].
N
N
1784
Nanostructures Based on Porous Silicon
Nanostructures Based on Porous Silicon, Fig. 4 Dissolution sequence of silicon metal assisted etching. In light blue the silver clusters are depicted, for different etching times (from a.1 to a.3)
In electroless metal-assisted etching, holes are injected in silicon by H2O2, independently from the original substrate doping. In this case, the oxidizing agent concentration can be considered equivalent to current density in electrochemical process. HF concentration rules dissolution and surface chemistry at the same way for both the processes. The ratio between oxidizing agent and HF deeply affects the etching regimes and the morphologies. The reaction mechanisms involved in the electroless dissolution of silicon in presence of metal particles are rather complex and will be given in summary following the work of Chartier and coworkers [13] for the case of Ag on Si. The Ag metallic nuclei at the silicon surface are charged of electrons, and for charge compensation, localize holes at silicon-silver interface, in majority coming from H2O2 dissociation at cathode in presence of catalyzing Ag, as shown in Eq. 10 and forming SiO2 or removing Si at anode, as in Eq. 11. Cathodic reaction The Ag particles at cathode, catalyze the reduction of H2O2 as in Eq. 3 taking advantage from a favorable redox potential, and enhancing the etch rates by supplying free holes in proximity of Si-Ag interface. H2 O2 þ 2Hþ ! 2H2 O þ 2hþ E0 ¼ þ1:76V=NHEðNormal Hydrogen ElectrodeÞ (10) Anodic reaction In analogy with the well-known chemical dissolution of Si in HF–HNO3 [14], summarizing the different mechanisms proposed in literature and following the work of Chartier, the following anodic reaction is proposed: Si þ 6HF þ nhþ ! H2 SiF6 þ nHþ þ ½ð4 nÞ=2 H2
(11)
The general form is valid for the two already cited electrochemical regimes, porous silicon formation and electropolishing, where n, ranging from 2 to 4, is the number of holes per dissolved Si atom. In the case of n ¼ 2, porous silicon formation occurs, with the following overall dissolution anodic reaction: Si þ 4HF2 ! SiF6 2 þ 2HF þ H2 þ 2e E0 ¼ 1:2V=NHE
(12)
also valid for HF/HNO3 stain etching well below the critical current density of electropolishing (JPS). In the case of n ¼ 3, ½ H2 per dissolved silicon atom, as in the case of 6 M HF – 6 M HNO3 solutions or in stable electrochemical dissolution at JPS [15]. When n ¼ 4, the electropolishing regime occurs, as in the case of low HF/HNO3 ratio or for current densities higher than JPS, with a different mechanism involving the formation and dissolution of intermediate species like SiO2 and without H2 formation (Eqs. 13a and 13b). Si þ 2H2 O ! SiO2 þ 4Hþ þ 4e E0 ¼ 0:84V=NHE SiO2 þ 2HF2 þ 2HF ! SiF6 2 þ 2H2 O
(13a) (13b)
The local mechanism of silicon dissolution in proximity of Ag particle is still unclear, but the pore morphology indicates hole consumption close to the silicon/metal clusters interface, also thanks to the catalytic reduction of H2O2. This phenomenon can be ascribed to a faster surface reaction with respect to holes diffusion due to coupling between silicon valence bonds and energy levels of silver, or a depletion layer around Ag nanoparticles allowing an easier access of holes in proximity of Ag/Si interface with respect to HF/Si interface to be produced.
Nanostructures Based on Porous Silicon
1785
N
Classification of Porous Silicon Porosity Porosity is defined as the volume fraction between silicon structures and voids in the porous layer, and is determined by gravimetry, weighting the silicon wafer before the etching (m1), after the etching and drying (m2), and after PS removal by alkaline solution (m3) as in Fig. 5. Porosity is a number between 0 and 1, and is obtained by the formula: Pð%Þ ¼ ðm1 m2Þ=ðm1 m3Þ The most critical parameters affecting PS structure and porosity are current density and HF concentration. Porosity can range from 4% of macroporous silicon to 95% for mesoporous layers. Porosity is only and indicative parameter and it does not yield information regarding the microstructure of the PS layer, structure size, and relative size distributions are the much more significative for the prediction and interpretation of physical phenomena like optical, thermal, and electrical response of the system. From gravimetry measurements, also the thickness of the porous layer can be determined using the following formula:
Nanostructures Based on Porous Silicon, Fig. 5 The three weights used in gravimetry for porosity evaluation. The first, m1, is the weight of the silicon sample before anodization, the second, m2, is the weight of the sample after PS formation and at least 20 min of drying, the third, m3, is the weight of the sample after alkaline removal of the PS layer
W ¼ ðm1 m3Þ=S d Where d is the silicon density and S the silicon surface exposed during etching. Other destructive methods to evaluate thickness are SEM sections, profiling after the porous layer removal, and among the nondestructive methods is ellipsometry, even if the final result is dependent on the adopted model. Pore Size, Morphology, and Orientation
According to an IUPAC convention [16], Porous silicon has been divided into three categories based on the size of its pores; macroporous, mesoporous, and microporous. Type Pore width (nm)
Microporous Mesoporous Less than 2 Between 2 and 50
Macroporous Larger than 50
The pores are strongly dependent from the substrate characteristics, like crystalline orientation and impurity concentration. In general, pores initiate at the wafer surface, and tend to follow the crystalline
directions, an example for different orientations is shown in Fig. 6. Since the electrochemical etching is dependent on the number of holes available in the substrate, PS morphology is deeply affected by the original impurity content of donors or acceptors in the substrate. In low resistivity N-type silicon, no etching occurs in absence of light or high bias. In presence of light, if the energy hn of the radiation is greater or equal the silicon band gap, 1.12 eV at room temperature, the energy excess is transferred to the crystal lattice by rapid thermal processes on the timescale of picoseconds, and the amount equal to the silicon band gap creates an electron-hole couple in Conduction Band and Valence Band, respectively. The holes can so contribute to silicon dissolution, in presence of Fluoride ions of the solution, removing the silicon-silicon bonding. If the illumination is from the top of the silicon surface, form the same side of the etching solution, the pores will initiate from the top and will follow the 100 direction, forming regular or branched macropores, depending on parameters like electric field and free carriers concentration available, but also
N
N
1786
Nanostructures Based on Porous Silicon
Nanostructures Based on Porous Silicon, Fig. 6 Some PS morphologies: from left to right, first row: N-type PS silicon from (111) substrate orientation, N-type PS silicon from (100) substrate orientation and backside illumination, N-type
PS silicon from (100) substrate orientation and front illumination. Second row: N + type PS silicon from (100), P + type PS silicon from (100), P-type PS silicon from (100)
a microporous phase, due to the quantity of holes generated by light. The depth of this layer depends form the spectral component of the lamp. If the light reaches the substrate from the bottom, the holes will drift through the silicon substrates, reaching the pore pits at the upper interface with HF and ethanol. There is a high probability that the holes will reach the pore tips, so contributing to maintain a stable and regular pore growth. In N-type macroporous silicon pore etching, there are several critical parameters to be monitored for a regular and sufficiently fast etching, temperature, current density, electrode potential, and illumination intensity. For the current density, galvanostatic or potentiostatic conditions can be used. Typically constant current conditions (i.e., galvanostatic) are adopted, so allowing the formation of pores under constant dissolution regime. So regular macropores on n-type silicon with backside illumination can be controlled in two ways: (1) By using galvanostatic/potentiostatic control maintaining the illumination intensity constant, (2) Fixing current and potential and adjusting the illumination intensity via feedback loop. The highly doped N-type and P-type silicon give origin to mesoporous structures. In the case of P-type the great availability of free holes due to doping and the low resistivity of the material give origin to high etch rates, of the order of hundreds of nanometers per second, depending on the current density. A typical
value can be 300 nm/s with 300 mA/cm2 in HF:Eth 1:1 starting from HF 49%. Mesoporous silicon has an intrinsic anisotropy. The material shows an interconnected columnar structure, with structure size distribution extending from 1.5 to 15 nm [17]. This material is scarcely luminescent, it is mechanically robust and its morphology is determined by the presence of impurities, since the etching does not remove selectively them (e.g., Boron) but the anodization deactivates the trivalent atoms leaving them in subsurface position [18]. For the highly doped N-type silicon the mechanism is rather different. The right branch of the curve is determined by the applied current density, and the regime is electrochemical, the left branch of the curve, is more a chemical dissolution, with slow etch rates, and so reaching high porosities. Low resistivity P-type silicon gives origin to a tiny coral structure, with typical sizes around 1–2 nm. These PS layers are fragile and show room temperature photoluminescence. Optical Properties and Photoluminescence Generally, nano-PS emits bright visible light at room temperature, in spite of the fact that, the pristine c-Si is a well-known indirect-gap semiconductor, with a fair emission in the near-IR, at cryogenic T.
Nanostructures Based on Porous Silicon
The application of Photoluminescence (PL) technique in PS involves several different physical processes, which make the final product (the PL spectrum) of difficult interpretation. Absorption of the incident photon creates an electron-hole pair. If nanocrystallites are somehow interconnected, or imbedded in a medium, these particles can escape through different mechanisms. If not, they can decay to lower states in the same nanocrystal, or could recombine radiatively, directly from the excited state. If localized states are present, trapping into these shallow states can occur. On the contrary, deep states, e.g., unpassivated dangling bonds, can enable the competing non-radiative recombination of electron-hole pairs. After several years of investigation, there is a general consensus about the origin of such strong light emission. Resonant PL technique proves that emission arises in the crystalline core of the material (instead of other Si-O, Si-H, or siloxene compounds) [19] where the energy gap is broadened thanks to Quantum Confinement (QC) effects. On the other hand, the efficient passivation of the non-radiative pathways (due to H-, or O- termination of the Si nanostructures) plays an important role. PL decay times are in the order from ms to ms, which makes practical application of PS in integrated optoelectronics unlikely. In particular, the red PL shows non-exponential decay, with typical times ranging in the ms and ms range at room T and at low T, respectively. The competition between radiative, and non-radiative recombination processes depends on T. Generally, the PL intensity reaches a maximum around 50–200 K (depending upon the sample), the simple meaning being that one of the competing processes, say, the radiative one, prevails at those T values. The T dependence of the PL lifetime is the consequence of exchange splitting of excitons. Indeed, the electron and hole, with spin 1/2, can be combined into a singlet (S ¼ 0) and a triplet (S ¼ 1) state, the exchange interaction stabilizing the triplet. While at high T both of them are populated, the PL arising from the “fast” decay of the singlet state, at low T the PL becomes much slower because the triplet state is virtually stable. In the T range at which the PL emission is maximum, the radiative process is still related to population of singlet states, whereas the non-radiative one in quenched. When T is further decreased, radiative
1787
N
processes are now related to the slow decay of triplet states, and non-radiative ones become competing again. PS Structures PS has been fruitfully employed as a sacrificial layer since it is a strongly reactive material, and easy to etch [20]. Generally, a wide class of materials, from metals to semiconductors and insulators can be deposited onto PS. Due to the intrinsic reactivity of PS, it can be easily etched, giving rise to large undercutting of the layers deposited on it. This method is proved to be very efficient for realizing suspended membranes, bridges, or cantilevers PS has been also employed for the epitaxial deposition of c-Si, because it conserves the lattice of the pristine material and can be used as a seed for further growth. Thanks to its reactivity and fragility, it allows transfer of the epitaxial film onto foreign substrates. Macroporous Silicon The former studies of n-type silicon anodization in the dark started in the early 1970s. Etch pits and silicon dissolution in dark is due to and breakdown between surface states tunneling in conduction band. From these studies, some evidences where clear, the channels formation is not due to crystal defects, the pores direction follow the orientation of the substrate, the growth rate is in general constant with anodization time, so the diffusion within the pores is not a rate-limiting process, and the pores’ walls are passivated. Beale proposed a depletion of porous layer, explaining the high resistivity of the material after anodization, and localizing carriers at the pore tips. Several concepts introduced by this physical model, were then adopted in successive works, but no chemical mechanisms were included, so providing poor comprehension of pore size variations and morphologies. Smith and coworkers, at the end of 1908s, postulated that the rate of pore growth is limited by the diffusion of the holes randomly walking in the bulk of silicon. This model was mainly ruled by the diffusion length L, a quantity that is function of dopant concentration and of potential. A sticking factor varying from point to point of the partially oxidized pore tip was
N
N
1788
necessary to explain different morphologies and pore diameter variations. This model was too general to explain conduction mechanisms and morphologies in different type of substrate and like the previous was not taking in account the electrochemical reactions at the interface. Zhang et al., in the same period, thanks to a systematic study of the i-V characteristics, clarified that the PS formation or electropolishing is not dependent by electronic properties of the substrate, but as the result of the two competing reactions occurring at the pore tips, direct dissolution by fluoride species and indirect removal of silicon through silicon dioxide formation and dissolution in HF. In the following years, Zhang also proposed that surface curvature of the pore tips could reduce the effective width of the space charge layer greatly increasing the interface tunneling current and the dissolution rates of silicon dissolution and oxide formation and removal. This model is also applicable to other substrates and PS morphologies. Lehmann and Fo¨ll in 1990 described the formation of macropores from low-doped n-type silicon in aqueous HF-electrolyte, backside illumination and pore nucleation predefined lithographically. They experimentally verified the hypothesis that the space charge region (SCR) around the pores can focus the holes diffusing in the silicon bulk on the pore tips, so enabling the growth of regular macropores. In further studies, Lehmann demonstrated that the equilibrium of the carrier transport in the semiconductor and the mass transfer in the electrolyte determines the etch rates and the morphologies in n-type silicon. At low current densities, there is no accumulation of holes and dissolution takes place only at the pore tips, leaving the walls unaffected thanks to carrier depletion. For high current densities, the reaction is mass transport limited and holes accumulate at the pore tips, dissolving walls and enlarging the pores. Lehmann tried to unify the models describing the formation mechanisms of all types of PS, using the assumption that all the porous structure except the pore tips are passivated by carrier depletion. Depletion is due to quantum confinement for microporous silicon and to space charge layer for porous structures larger than few nanometers. Fo¨ll et al. developed the current burst theory, one of the most recent and comprehensive models of PS formation, it is so far the only model considering the
Nanostructures Based on Porous Silicon
electrochemical reactions under formation conditions in the different PS morphologies. The main assumptions are that the electrochemical reaction of silicon dissolution operates in microscopic units with a definite spatial and temporal distribution. Other important points are the following: Current flow is inhomogeneous in space and time, from here the local current burst. Current flow induces direct silicon dissolution and oxidation and removal following current burst. As a result of hydrogen termination and desorption, the position of Fermi level and so of the space charge layer oscillates with the time The rate of hydrogen termination is fastest on (111), determining the probability of current burst on surfaces of different orientations. This model is rather powerful in explaining many aspects involved in the PS formation but it is extremely difficult to be expressed in mathematical form [21]. Porous Silicon Based Photonics In nano- and mesoporous cases, application in 1D photonics structures deserves a mention. From an optical point of view, PS is seen from visible light as a homogeneous material. In fact, the sizes of the pores and the Si nanocrystals are well below the light wavelength. When porosity is varied, the macroscopic dielectric constant (and the refractive index) is varied, too. In principle, it is possible to “choose” the refractive index of the layer between the limits of n ¼ 3.4 (porosity ¼ 0) to n ¼ 1 (porosity ¼ 1). The range of variation for n is smaller, but n can be varied in a continuous way in this wide range. Being the dependence of n from porosity relatively simple and considering that porosity monotonically increases with the anodization current, by changing the current during anodization in the right way, one is able to obtain optical devices like Bragg Reflectors or Fabry-Perot microcavities done by PS. The advantage of this technology is evident: it allows obtaining optical devices directly integrated on Si [22]. A Bragg reflector is a periodical structure of layers with high and low refractive index. The optical thickness of both is chosen to be l/4, l being the light wavelength at which maximum rejection is wanted. The Bragg reflector is then a sort of 1D photonic crystal, there the rejection band stands for photons like energy gap stands for electrons in semiconductors.
Nanostructures Based on Porous Silicon
Correspondingly, one can choose to insert defects states within such gap: this can be accomplished by means of a Fabry-Perot microcavity, where a layer with optical thickness equal to l/2 is sandwiched between two Bragg reflectors. The PS anodization technology allows for realizing Bragg reflectors with >98% reflectivity at the rejection band and Fabry-Perot microcavities with peak FWHM nair.
1791
N
nair θ1
nfilm
θ2
d
nair
Nanostructures for Coloration (Organisms other than Animals), Fig. 1 Thin film interference in the case of, for example, soap bubbles
The condition for constructive interference for a soap bubble is given in (Eq. 1), the condition for destructive interference is given in (Eq. 2): 2n film d cosðy2 Þ ¼ ðm 1=2Þl
(1)
2n film d cosðy2 Þ ¼ ml
(2)
where d equals the film thickness, nfilm equals the refractive index of the film, y2 equals the angle of incidence of the wave on the lower film–air boundary, m is an integer, and l is the wavelength of the light. Summing up, there are two conditions that should be satisfied for constructive interference: firstly, the thin film should be thin enough to crest the reflected waves and secondly, the two reflected waves should be in one phase. Multilayer Interference As described above, a light wave can be reflected from both boundaries of a thin film layer and dependent on the phase of the two reflected light waves either destructive or constructive interference occurs. The same phenomenon occurs in a series of thin films, a multilayer. A schematic of a multilayer is given in Fig. 2. With nB > nA, between each A-B interface a 180 change in phase takes place, while at the B-A
N
N
1792
Nanostructures for Coloration (Organisms other than Animals)
m=−2
A
m=−1 m=0 m=1
m=2 m=3
m=−3 B dA
nA
dB
nB
θA
A θB
a
m=−4 b
B
m=4 b d
A B
Nanostructures for Coloration (Organisms other than Animals), Fig. 2 Multilayer interference. nB > nA. A change in viewing angle corresponds to a change in the perceived color, due to changes in the phases of the interfering waves
interface there is no phase change. Reflection or refraction at the surface of a medium with a lower refractive index causes no phase shift. Reflection at the surface of a medium with a higher refractive index causes a phase shift of half a wavelength. Constructive interference: 2ðnA dA cos yA þ nB dB cos yB Þ ¼ ml
Nanostructures for Coloration (Organisms other than Animals), Fig. 3 Schematic of the color generating mechanism by the diffraction of light by a diffraction grating (black bar on bottom) 1
2
1
1′
2
1′
a
a
b
2′
x 2′ y
Nanostructures for Coloration (Organisms other than Animals), Fig. 4 Coherent (left) and incoherent scattering (right)
(3) dðsin a sin bÞ ¼ ml
(5)
Maximum reflection: 2nA dA cos yA ¼ ðm 1=2Þl
(4)
In ideal multilayers all waves interfere constructively, resulting in colorful reflections, whereas in non-ideal multilayers some waves interfere destructively and the reflection is less colorful [17]. Diffraction Gratings The grating equation (Eq. 5) correlates the spacing of the grating (d) with the angles of the diffracted and incident beams (a, b) and the wavelength of illumining light (l) (Fig. 3). The coloration in a particular direction (viewing angle) is generated by the interfering components from each slit of the grating. The diffracted light has maxima at angles where there is no phase difference between the waves. The 0th order reflection corresponds to direct transmission and is denoted m = 0. Other maxima occur at angles with m = 1, 2, 3, . . . .
CDs and DVDs are examples for this phenomenon. Very recently, diffraction gratings that produce colors have been described in flowers [18]. Scattering The term scattering is a more general denomination for the interference of light waves with different wavelengths reflected from scattering objects either in a constructive or destructive way (Fig. 4). In terms of coloration, scattering can yield either strong or weak colors. Examples of scattering, and more precisely, coherent scattering are the thin film interference and the multilayer film interference described above. In coherent scattering there is a definite phase relationship between incoming and scattered waves, whereas in incoherent scattering this is not the case. The two main types of scattering are termed Rayleigh scattering and Tyndall scattering. In both types of scattering, the intensity of the scattered light depends on the fourth power of the frequency.
Nanostructures for Coloration (Organisms other than Animals)
In Rayleigh scattering, the responsible particles are much smaller than the wavelength of the incident light, whereas in Tyndall scattering, the particles are macroscopic (e.g., dust particles in air or small fat particles in water, as in milk). Since blue light (i.e., shorter wavelengths) is scattered more than red light (i.e., long wavelengths), milk, tobacco smoke, and the sky are blue or have a blue hue. In plants, for example, the blue of the blue spruce can be explained by scattering [14]. Multiple scattering within regular structures produces highly reflective bands in the reflection spectrum, and leads to the brilliant colors observed in, for example, iridescent fruits ([12] and references therein). Photonic Crystals Photonic crystals exhibit a periodicity a in the refractive index (Fig. 5). There are one-, two-, and three-dimensional photonic crystals. Depending on their wavelength, photons either can be transmitted through the crystal or not (allowed and forbidden energy bands). For effects in the visible range, the periodicity of the photonic crystal has to be between about 200 nm (blue) and 350 nm (red). Vegetable opal is an example for a photonic crystal produced by plants. Opal is the only gemstone that is known to be formed also as a plant-produced product. There are three different types of opal produced by photosynthetic organisms: minuscule iridescent diatoms, phytoliths, and tabasheer. Tabasheer (bamboo opal, plant opal) is a hard, translucent to opaque whitish substance extracted from the joints of bamboo [5]. It can have beautiful blue hue, and was – cut as cabochons – used in China for fashion jewelry. Cholesteric Liquid Crystals Cholesteric liquid crystals are layered helicoidal structures, also known as chiral nematic liquid crystals, within which molecules take a preferred direction that gradually changes from layer to layer (Fig. 6). The variation of the director axis tends to be periodic in nature. The period of this variation (the distance over which a full rotation of 360 is completed) is known as the pitch, p. Certain iridescent tropical understory ferns contain a helicoidal layering of cellulose microfibrils that produces multiple interference layers ([12] and references therein), similar to the helicoidal microstructure
N
1793 1-D
a
2-D
3-D
Nanostructures for Coloration (Organisms other than Animals), Fig. 5 Schematics of one-, two-, and three-dimensional photonic crystals (1-D, 2-D, 3-D). The colors represent materials with different refractive indices. The spatial period of the material is called the lattice constant, a
p/2
Nanostructures for Coloration (Organisms other than Animals), Fig. 6 The chiral nematic phase in a cholesteric liquid crystal. p refers to the chiral pitch. Image reproduced with permission under the terms of the GNU Free Documentation License
in a cholesteric liquid crystal and the characteristic liquid crystals in beetles [1]. The peak wavelength of the reflected light, l, is determined by the helicoidal pitch, p, where l ¼ pn
(6)
and n is the average refractive index of the anisotropic material [3].
Plants (and Other Nonanimals) with Structural Colors Structural colors in nature are found in various animals, plants, microorganisms, and even viruses [1, 10, 12, 19]. Studies of structural colors date back to the seventeenth century when the earliest scientific
N
N
1794
description of structural colors appeared in “Micrographia,” written by Robert Hooke in 1665. In his book “Opticks” Sir Isaac Newton already in 1704 related iridescence to optical interference: The finely colour’d feathers of some birds, and particularly those of the peacocks’ tail, do in the very same part of the feather appear of several colours in several positions of the eye, after the very same manner that thin plates were found to do.
Since then, various organisms with structural colors (almost all in animals) have been reported. This chapter focuses on organisms outside of the animal kingdom, primarily plants, but also macroalgae, slime molds, diatoms, and viral particles. For plants, the discussion is organized along the physical phenomena as well as the various organs where such color is found, that is, leaves, flowers, fruits, and seeds. Other organisms with simpler organization are classified by the major groups, for example, the red algae (Rhodophyta). The first report on structural colors in plants was by Lee and Lowry [13], who described the physical basis and ecological significance of iridescence in blue plants. The most common mechanisms in plants leading to structural coloration are multilayer interference and diffraction gratings. Multilayer interference is found predominantly in shade-plant leaves, suggesting a role either in photoprotection or in optimizing capture of photosynthetically active light ([12] and references therein). Surprisingly, diffraction gratings may be a common feature of petals, and recent work has shown that bees use them as cues to identify rewarding flowers. Structural colors may be surprisingly frequent in the plant kingdom, and still much remains to be discovered about their distribution, development, and function [7]. Nonanimals with Coloration Caused by Thin Film Interference True Slime Molds Slime molds are organisms that normally live as single cells and that can aggregate to a multicelled organism and form fruiting bodies that produce spores. They have the characteristics of single-celled microorganisms as well as of fungi and occur worldwide on decaying plant material. Inchaussandague and coworkers [9] described structural colors produced by a single thin layer in the slime mold Diachea
Nanostructures for Coloration (Organisms other than Animals)
leucopoda. Interference within the peridium, a 200 nm transparent layer invariant along the different cells of the “organism,” is the basis for this color. Plants with Coloration Caused by Multilayer Interference Tropical Understory Ferns The “peacock fern,” Selaginella willdenowii (actually a lycophyte, or fern ally), is a common plant in the Malaysian rainforests. It exhibits blue iridescence; investigation with TEM reveals various layers (with less than 100 nm thickness each) on the surface. The photo shown in Fig. 7 was taken in the Bukit Wang Recreational Forest in Malaysia, with very long exposure time as to better reveal the blue coloration of the leaves. Seen with the naked eye, the fern is bluish green and seems to glow in semidarkness. The blue coloration is iridescent and changes with the viewing angle. Holding and tilting the leaves results in color change from green to nearly total blue, although not as strong as seen in the long exposure time photograph shown in Fig. 7. Also another species of Selaginella shows iridescence: Selaginella uncinata. As in S. willdenowii, a multilayer system where each layer is thinner than 100 nm is responsible for the structural coloration. When the leaves of these two Selaginella species are dried, the coloration disappears, but comes back again when the plants are hydrated again. When a droplet of water is put on a Selaginella leaf, the blue coloration disappears. Other tropical understory ferns where coloration is caused by multilayer interference are Diplazium crenatoserratum, Lindsaea lucida, Danaea nodosa, and Trichomanes elegans (the neotropical bristle fern, also called the “dime store plant” because it looks like it is made from plastic and is shade tolerant, is native to the American tropics) ([12] and references therein). Tropical Understory Begonias
Gould and Lee (1996) reported structurally modified chloroplasts (termed “iridoplasts”) in the Malaysian tropical understory plant Begonia pavonina (the peacock begonia) ([12] and references therein). The structural coloration is generated by many layers, each less than 100 nm thick. In addition to B. pavonina, the authors have observed several other iridescent blue species on the Malaysian peninsula. Modified plastids produce blue colors in Begonia, Phyllagathis (see
Nanostructures for Coloration (Organisms other than Animals)
Nanostructures for Coloration (Organisms other than Animals), Fig. 7 The “peacock” fern Selaginella willdenowii, a common plant in the Malaysian rainforest. # Mr. Foozi Saad, IPGM, Malaysia. Image reproduced with permission
section “Other Tropical Understory Plants”), and Trichomanes (see section “Tropical Understory Ferns”). Species with modified chloroplasts are not helicoidal. Other Tropical Understory Plants
A traditional Malaysian medicinal plant, the melastome Phyllagathis rotundifolia (in Malay known as tapak sulaiman) and also the Malaysian understory species P. griffithii exhibit iridoplasts, similar to the ones in Begonia pavonina described above. Iridescent Macroalgae
Among the red algae (Rhodophyta) several species of Mazzaella (= Iridaea) are strikingly iridescent [6]. Green and brown algae have been recorded as iridescent, but the phenomenon is most commonly found in red algae, where blue and green colors appear on the surface at certain stages of the lifecycle. Iridescent red algae (e.g., Mazzaella flaccida, Mazzaella cordata) exhibit shiny blue, emerald green, and deep red colors. The blades are relatively thin and vary in shape and size (up to 3 ft long and 10 in. wide) depending on the habitat and the amount of wave exposure. Mazaella flaccida is iridescent yellow-green with purple or brown near the base of the blade, and the iridescent blade of Mazzaella splendens, is dark purple with a blue iridescent sheen. It is also known as the rainbow-leaf seaweed. The coloration disappears when the algae are dried, and is restored when the algae become wet again.
1795
N
Nanostructures for Coloration (Organisms other than Animals), Fig. 8 Mazzaella sp., a highly iridescent red algae. # 2003 by J. Harvey, http://www.JohnHarveyPhoto.com. Image reproduced with permission
Wet Iridaea look like they have been coated with oil (Fig. 8), because of their rainbow sheen; iridescent spots, shaped like a teardrop, with a multilayer system of 17 electron opaque and translucent layers, some tens to a few hundred of nanometers thick, are responsible for the coloration. The iridescence might be a byproduct of a wear-protection mechanism of the algae. The edible seaweed Chondrus crispus (carragheen, Irish moss) is a red algal species that shows blue iridescence due to multilayer interference effects on the tips [15]. The function of algal iridescence is still unclear. Suggestions include a role in camouflage or a role in optimizing photosynthesis by enhancing the absorption of useful wavelengths of light at the expense of increased reflection of other wavelengths. Plants with Coloration Caused by Diffraction Gratings Glover and Whitney [7] reviewed the original research on iridescence in plants and presented very recent results on floral iridescence produced by diffractive optics. They identified iridescence in flowers of Hibiscus trionum (the inner part of its petals has an oily iridescence overlying red pigmentation) and tulip species (e.g., Tulipa kolpakowskiana) and demonstrated that iridescence was generated through diffraction gratings. The iridescence in the hibiscus is obvious to human eyes (appearing blue, green, and yellow depending on the angle from which it is viewed), whereas the iridescence in the tulip is in the ultraviolet spectrum, which bees can see, but
N
N
1796
people cannot. The surface striations in the diffraction gratings observed in the tulips are about one micrometer apart. Whitney and her coworkers investigated 22 tulip species from around the world, and found ordered striations in 18 of them. They report iridescence generated by diffraction gratings in 10 families of angiosperms, tulips and hibiscus being just two of them: From the mallow family the species example they give is Hibiscus trionum, from the lily family, Tulipa sp.; from the aster, daisy or sunflower family, Gazania klebsiana; from the pea family, Ulex europaeus; from the Loasaceae family (of bristly hairy and often climbing plants), Mentzelia lindleyi; from the buttercup or crowfoot family, Adonis aestivalis; from the willowherb or evening primrose family, Oenothera biennis; from the nightshades, Nolana paradoxa; from the iris family, Ixia viridiflora; and from the peony family, Paeonia lactiflora. According to this extensive list, it might well be that many more flowers are iridescent than previously thought – since they are iridescent in parts of the spectrum that people cannot see. The principal reason why not more species with diffraction-based coloration are known might simply be that until now nobody has looked. Plants with Coloration Caused by Scattering Some green leaves look white or silvery because microscopic air spaces in surface hairs reflect the light. Other leaves (such as the wax palm) and some fruits (such as plums and grapes) have a white “bloom” which is a surface deposit of wax. Whitish, silvery, and other metallic finishes of leaves are generally produced by reflection of the light off microstructures, nonliving plant hairs called trichomes. In deserts this increased reflection, especially of infrared radiation, off the microstructures is important since it reduces heating of the leaf. However, a reflecting leaf has a much lower photosynthetic capacity than would an equal leaf without the reflectant trichomes. The desert brittlebush Encelia farinosa produces nonreflectant green leaves during the cooler springtime, when it is beneficial to have warmer leaves. Silvery, grey, white, or blue coloration in plants might arise due to light scattering on threedimensional epicuticular wax structures (structural wax). Leaves with a bluish gray waxy surface are called glaucous.
Nanostructures for Coloration (Organisms other than Animals)
1 μm
Nanostructures for Coloration (Organisms other than Animals), Fig. 9 Monohelical tubular epicuticular wax structures in Wollemia nobilis. Scale bar 1 mm. From [4], reproduced with permission
Generally, radiation is scattered across the spectrum, with increased reflectance in the ultraviolet, visible, and infrared windows. The size, distribution, and orientation of wax crystals and other surface features determine the extent to which light is scattered at the plant surface. In some plant species such as the blue spruce (Picea pungens) preferential scattering of shorter wavelengths and enhanced reflectance of UV radiation occurs. Epicuticular wax is a complex mixture of longchain aliphatic and cyclic compounds. The bluish leaf hues of the evergreen desert mahonia Berberis trifoliolata, for example, arises from wax structures on nipple-like projections from the epidermal cells, called epidermal papillae. In some cases, for example in the Blue Finger, Kleinia mandraliscae, and in cabbage, Brassica oleracea, powdery wax can be rubbed from the surface, causing the grey or bluish hue to disappear, and revealing the green leaf color beneath. The Wollemia pine Wollemia nobilis, an evergreen tree with bluish green mature foliage, has monohelical tubular epicuticular wax structures with a diameter of about 100 nm (Fig. 9) [4]. Depending on the wax compound present, epicuticular wax crystals can have various shapes [11]: The glaucous eucalyptus tree (Eucalyptus gunnii) has nanoscale b-diketone epicuticular wax tubules; the blue-grey foliage of the dusty meadow rue (Thalictrum flavum glaucum) has nanoscale nonacosanol tubules. The Black Locust tree (Robinia pseudoacacia) has dark, dull, blue-green
Nanostructures for Coloration (Organisms other than Animals)
summer foliage with nanoscale epicuticular wax platelets that are arranged in rosettes. The leaves of wild cabbage (Brassica oleracea) show simple nanoscale rodlets, whereas the nanoscale rodlets on the Silky Sassafras (Sassafras albidum) leaves are transversely ridged. The epicuticular wax in the skin of plums (Prunus domestica L.) consists of an underlying amorphous layer adjacent to the cuticle proper together with crystalline granules of wax protruding from the surface. The epicuticular waxes in olive leaves serve as radiation inceptors. Coloration Caused by Photonic Crystals: Plants, Diatoms, and Viruses Iridescent Blue Fruits The fruits of the blue quandong (Elaeocarpus angustifolius, syn. E. grandis) and the blue nun (Delarbrea michiana) have brilliant blue coloration ([12] and references therein). The iridescent blue color of the quandong fruit is actually enhanced by wetting, not like in Selaginella, mentioned above, where it disappears when wet. Iridosomes, multilayered systems arranged in 3-D structures, similar to the structure yielding coloration in Morpho butterflies, are responsible for the coloration of these fruits. Iridosomes are different from the iridoplasts described above for leaves: They are secreted by the epidermal cells of the fruit, and are located outside the cell membrane but inside the cell wall. The structures of the iridosomes are at least partly cellulosic. The stone of the silver quandong fruit is termed rudraksha. It is hard and highly ornamented, and is used throughout India and Southeast Asia in religious jewelry. Edelweiss
Edelweiss (Leontopodium nivale subsp. alpinum) is a European alpine flower that grows at high elevations up to 3,400 m. Its body is covered with white hairs. Vigneron and coworkers showed in 2005 [16] that the internal structure of these hairs acts as a twodimensional phonic crystal. The hairs are hollow tubes, about 10 mm in diameter, with an array of parallel striations with a lateral diameter of 176 nm around the external surface. Through diffraction effects determined by the optical fiber with photonic crystal cladding, the hairs absorb the majority of the UV light, thereby acting as an efficient sun-block for the structures beneath.
1797
N
Spores
Some ferns produce two types of spores: microspores and megaspores (size: some hundreds of micrometers). Hemsley et al. [8] reported iridescent megaspores in Selaginella (fossil and modern). They report that spore iridescence is not generally visible until the outer layers of the spore wall (with different organization to the iridescent layer) are removed. The three-dimensional pattern yielding the iridescence, for example, in megaspores of Selaginella exaltata, is still unresolved. According to SEM images, an ordered structure of 240 nm diameter particles, perhaps an opal-like colloidal crystal that is generated via selfassembly, is the reason for the coloration. Tabasheer
Opal is the only gemstone that can be produced in biotic and abiotic ways. Tabasheer (vegetable opal, pearl opal) consists of pure silica and is produced when bamboo is hurt; the plant sap comes to the surface and dries in small nodules. These little nodules are opaque or translucent white, sometimes with a faint hint of blue. They can be polished as cabochons and are often used in the orient as jewelry. Tabasheer is highly porous and hygroscopic, and has been used in ancient times to remove snake poison from the body – therefore it is also known as snake stone [5]. Diatoms
Diatoms (Baccilariophyta) are found in both freshwater and marine environments, as well as in damp soils and on moist surfaces. They are unicellular microalgae with a cell wall consisting of a siliceous skeleton enveloped by a thin organic case. The cell walls of each diatom form a pillbox-like shell consisting of two parts that fit within each other. Diatoms exhibit an amazing diversity of nanostructured frameworks (Fig. 10), including two-dimensional inverse photonic crystals. The cell wall exhibits periodic arrangement of pores in the micrometer to nanometer range – each diatom species has its own specific morphology. Individual diatoms range in size from 2 mm up to several millimeters, although only a few species are larger than 200 mm. A patch of diatoms can look like iridescent scum to the naked eye. Under the microscope, beautiful iridescence can be seen on the single cell level, and a microscope slide covered with a monolayer of diatoms also exhibits iridescence to the naked eye
N
N
1798
Nanostructures for Coloration (Organisms other than Animals)
500 μm
500 μm
500 μm
200 μm
Nanostructures for Coloration (Organisms other than Animals), Fig. 10 Sample with various diatoms from New Zealand, laid by hand by Elger in 1925. The slide was photographed by Friedel Hinz under the optical microscope with slightly varying angles of view, revealing color change in
some of the diatoms (top images and bottom left). Bottom right: Zoom. Marine fossil diatoms from Gave Valley, New Zealand, Sample T1/21 prepared by Elger, 1925. Image (c) 2011, F. Hinz, AWI Bremerhaven, Germany. Image reproduced with permission
(with good illumination or in sunlight). Iridescent effects are used widely in color cosmetic products and personal care packaging, and there is great potential for using diatoms in this industry.
extremely brilliant islands. Electron microscope examination reveals numerous viruses in paracrystalline arrays (see Fig. 11) [20].
Viruses
Iridescent viruses (Iridoviridae) are between 120 and 140 nm in diameter and infect insects, fishes, and frogs. Iridescent viruses produce systemic infections, with the highest concentrations in the outer layer of the skin (the epidermis) and the fat bodies. Insect larvae infected with this virus develop iridescent lavender-blue, blue, and blue-green coloration. In older infections, the iridescent color is generally more vivid, and frequently there are also small,
Plants with Coloration Caused by Cholesteric Liquid Crystals In three iridescent tropical understory fern species, structurally modified chloroplasts with helicoidal structures, similar to the characteristic liquid crystals in beetles [1], might be the reason for the blue coloration. These ferns are Danaea nodosa (with many multilayers, each less than 100 nm thin), the necklace fern Lindsea lucida (17 layers, 192 nm spacing) and Diplazium tormentosum (20–23 layers, 141 nm spacing, see Fig. 12) [12].
Nanostructures for Coloration (Organisms other than Animals)
Nanostructures for Coloration (Organisms other than Animals), Fig. 11 Paracrystalline array of virus particles within an infected cell. This array gives rise to the iridescent phenomenon. Image Source: http://www.microbiologybytes.com/virology/ kalmakoff/Iridoviruses.html. Permission pending
1799
N
Additional to the iridescence in red algae with multilayers responsible for the coloration (see section “Iridescent Macroalgae” above), the following red algae exhibit structural coloration: The blue branching seaweed Fauchea laciniata shows deep red with iridescent blue; thin layer interference (whether from one layer or a multilayer system still needs to be clarified) is the reason for its coloration. Fryeella gardneri exhibits iridescent purple; Cryptopleura ruprechtiana is mildly iridescent. In Chondracanthus corymbiferus the younger blades are smooth and iridescent. Other iridescent red algae are Chondria coerulescens and Drachiella spectabilis. Also Maripelta rotata and Botryoglossum farlowianum are iridescent. Slime Molds
Nanostructures for Coloration (Organisms other than Animals), Fig. 12 Ultrastructure of a Diplazium tomentosum leaf. Scale bar 0.5 mm. Image # with one of the authors (DWL)
Structural Coloration Caused by Not Yet Described or Not Yet Identified Mechanisms Macroalgae Various red algae (Rhodophyta) and brown algae (Phaeophyta) exhibit iridescence, and are completely unstudied. For example, of yet unknown origin is the bright blue, purple, or red iridescence in the red alga Ochtodes secundiramea. Dictyota are widespread brown algae (Phaeophyta) along the Atlantic coast and very well known. Dictyota mertensii has blue/green iridescence; Dictyota dichotoma is also known as the purple peacock algae; Dictyota barayresii shows a translucent, iridescent blue; Dictyota sandvicensis exhibits a yellow green iridescence; and Dictyota humifusa is light brown, often with brilliant blue iridescence. Iridescent bodies in the brown algae Cystoseira stricta are specialized vacuoles (vesicles) that contain numerous dense globules inside. The physics of their iridescence has not yet been clarified. The Hawaiian brown alga Stypopodium hawaiiensis has a beautiful blue or green color, and Dictyopteris zonarioides sometimes is iridescent blue [6].
As described above in section “True Slime Molds,” the physical reason for the iridescent color in at least one type of slime mold is a single transparent thin layer. Slime mold iridescence is also described for some other species; however, its physical basis still needs to be determined. Spores of Diachea subsessilis exhibit dull greenish grey varying to blue iridescence; the peridium in Diachea deviata is iridescent and persistent. Beautiful bronze iridescent colors, sometimes tinged with blue, can be observed in the peridium of D. subsessilis (with the membranous peridium being colorless in water mounts, suggesting that pigments are not involved in the color production). Plants
Bryophytes Bryophytes include the mosses, liverworts, and hornworts. The moss Schistostega pennata exhibits golden-green iridescence when it is grown in caves where the light always comes from only one direction. This moss (also known as dragon’s gold) shines like emerald jewels from the darkness of a rock crevice or cave. This unusual property is the result of lens-shaped cells with curved upper surface that focus the light on one point in the interior of the cell, where the chloroplasts aggregate. It remains to be determined if nanostructures are responsible for the coloration of this moss (Lee [12] and references therein). Ferns Elaphoglossum herminieri, a strap fern that is native to tropical rainforests of the neotropics shows iridescent blue color (Fig. 13) that is not removed by wetting. The reason of the coloration is unknown, as it is
N
N
1800
Nanostructures for Coloration (Organisms other than Animals)
iridescent inflorescences are the small grass Pennisetum alopecuroides, little bunny, and the even smaller variety little honey. In late summer, these grasses are covered with blooms that look like hairy caterpillars.
Nanostructures for Coloration (Organisms other than Animals), Fig. 13 The iridescent blue strap fern Elaphoglossum herminieri, at Heredia, Sarapiqui near Puerto Viejo, La Selva Biological Station, Costa Rica. # 2008 R.C. Moran, The New York Botanical Garden (
[email protected]) [ref. DOL23374]. Image reproduced with permission
in Elaphoglossum wurdackii and E. metallicum. The oil fern Microsorum thailandicum (Microsorum steerei) from Southeast Asia has blue-green iridescent leaves. Also ferns of the family Vittariaceae can have iridescent scales. In this family, especially ferns of the genus Antrophyum (e.g., Antrophyum formosanum) and Haplopteris can be densely covered with iridescent scales. Antrophyum mannianum, a mountain fern liking full shade, has more or less iridescent leaves. There are also reports on an Antrophyum species found in Borneo (Mt. Mulu, or Gunong Mulu) with an iridescence that makes the plant look like it is covered with metallic eye shadow. Haplopteris ferns live in tropical and subtropical Asia. Obscure iridescence is exhibited by Haplopteris amboinensis. H. doniana, H. taeniophylla, H. himalayensis, H. mediosora, H. fudzinoi, H. linearifolia, H. hainanensis, H. sikkimensis, H. elongata and H. anguste-elongata show bright iridescence. Also H. flexuosa is iridescent, but not as bright. Malaysian individuals of the fern Didymochlaena truncatula are reported to be iridescent. The Common maidenhair Adiantum capillus-veneris has iridescent stems, and the Western maidenhair Adiantum aleuticum has iridescent blackish foliage. Grasses (Poaceae) Iridescence can be observed in the rosy inflorescences of the ornamental grass Miscanthus sinensis. Further grasses with startling
Sedges (Cyperaceae) A sedge is a grass-like or rush-like plant growing in wet places having solid stems, narrow grass-like leaves, and spikelets of inconspicuous flowers. Leaves and seedlings of sedges can exhibit iridescence. The understory sedge Mapania caudata from Malaysia and also M. graminea exhibit iridescence in their leaves. The smooth black sedge grass Carex nigra shows whitish iridescence and the green sedge grass Carex dipsacea has iridescent olive green leaves. Carex flagellifera “toffee twist” has iridescent, slender leaves. The seedlings of the pond flat sedge Cyperus ochraceus are dark iridescent gray because of an outer one-cell thick covering of translucent cells, whereas the seedlings of the marsh flat sedge Cyperus pseudovegetus are brown with a very thin translucent-iridescent layer of cells. Also Carex squarrosa has blackish seedlings with iridescent superficial cells (when fully mature). In Fimbristylis annua and F. vahlii the seedlings are often iridescent. The Harper’s fimbry F. perpusilla has pale brown seedlings with iridescent tints, and the southern fimbry F. decipiens has seeds of whitened-iridescent to brown. Hypoxidaceae: Hypoxis The stargrass Hypoxis sessilis has seedlings that are black but with iridescent membranous coats. Orchids (Orchidaceae) Iridescent foliage in orchids occurs, for example, in Macodes petola, Masdevallia (Byrsella) caesia, and Aulosepalum pyramidale, where it is bright green with an iridescent shine. Flowers of many species of orchid (such as Ophrys apifera and the mirror bee orchid Ophrys speculum) have been reported to have an iridescent patch, the speculum. The shape and iridescence of this iridescent patch is thought to mimic the closed wings of female wasps or bees, and sexually deceive the respective males, thereby pollinating the orchid (see [7] and references therein). Others There are various further plant species that exhibit iridescence. Some of them are mentioned
Nanostructures for Coloration (Organisms other than Animals)
below. The main point is that there is a lot to study and potentially several novel mechanisms of color production in organisms are yet to be identified. Purslanes (Portulacaceae): Portulaca. Seeds mostly glossy black or iridescent gray have been reported for various species of Portulaca. The reason of their iridescence is still unknown. In the moss-rose purslane Portulaca grandiflora Hooker the mature seeds are steely gray and often iridescent; in the pink purslane Portulaca pilosa the seeds are black, with very slight purplish iridescence when mature; in Portulaca psammotropha, the seeds are black, turning iridescent gray when fully mature, and in the shrubby purslane (copper purslane) Portulaca suffrutescens Engelmann and P. halimoides, the seeds are leaden and slightly iridescent. Grapes (Vitaceae): Ampelopsis. The porcelain berry vine, Ampelopsis brevipedunculata, has slightly iridescent fruits showing white, green, turquoise, blue, brown, and violet coloration. Lardizabalaceae (no common name): Decaisnia. The blue beans (Decaisnea fargesii) from Bhutan have iridescent, 4-in. large blue fruits. Pinks (Caryophyllaceae): Spergularia. Other plants with iridescent seeds are the blackseed sandspurry, Spergularia atrosperma, whose black seeds are often iridescent. Aizoaceae (ice plants): Sesuvium. Another plant with iridescent seeds is the slender sea purslane, Sesuvium maritimum, with brownish-black, smooth, and somewhat iridescent seeds. Plantaginaceae (no common name): Bacopa. The blue hyssop, bacopa caroliniana, has iridescent seeds. Goosefoot (Chenopodiaceae): Chenopodium. The purple goosefoot (tree spinach), Chenopodium giganteum, is an annual plant in which the young shoots are covered with a fine iridescent magenta powder. Rapataceae: Stegolepis. The leaves of Stegolepis ligulata have blue-green iridescence; even more impressive is the larger S. hitchcockii. Mallows (Malvaceae): Hibiscus. The red leaf hibiscus (H. acetosella) produces iridescent lobed maroon leaves on woody stalks. Palms (Arecaceae): Chamaedorea. Very impressive are the leaves of the metallic palm Chamaedorea metallica. Hydrangea (Hydrangeaceae): The United States patent PP18294 refers to the invention of
1801
N
a Hydrangea macrophylla plant named “HYMMAD I” whose inflorescences mature to an iridescent lime green. Star Flowers (Hypoxidaceae): The flowering plant Spiloxene capensis can have white, cream yellow, or pink flowers with a dark center. In the pink form the center is iridescent blue-green, whereas in the white form the center is iridescent bronze-green. Ice Plants (Aizoaceae): The inflorescences of the hardy ice plant Delosperma cooperi “Mesa VerdeTM” have iridescent hues of salmon-pink. Gesneriads (Gesneriaceae): Rhynchoglossum obliquum is a highly shade-resistant flowering plant with deep blue iridescent inflorescences. Sea hollies (Apiaceae): Eryngium maritimum, sea holly and E. x tripartitum, big blue, have surprisingly iridescent blue flowers. Spikemoss (Selaginellaceae): Selaginella kraussiana “gold tips,” also known as golden jade, with the common names Krauss’s spikemoss and African clubmoss, is a spikemoss found naturally in the Canary Islands, the Azores, and parts of mainland Africa. It has dark green foliage with golden tips. Begoniaceae: Begonia. Many begonia species have iridescent leaves; however, the coloration of the leaves is strongly dependent on the environment the plants are grown in. In Begonia Rex the leaves can be of shiny metallic red. A begonia plant named “lady rose” with iridescent foliage, flower bud, petals, and tepals was patented by Mr. Terry McCullough in December 1998 (patent number plant 10,736). There are no reproductive organs formed in this invention. Other iridescent begonias are B. burkillii, B. limprichtii, B. congesta, B. hahiepiana, B. sizemoreae and the hybrids Begonia bandit, Bethlehem star begonia, B. comedian, B. Palomar prince, B. wild pony, B. his majesty, B. merry Christmas, B. little darling, B. max gold, B. stained glass, B. regal minuet, B. shirtsleeves, B. Venetian red, and B. hocking tutu terror. The authors have observed several iridescent members of this genus in Malaysia, besides the B. pavonina described previously.
Conclusion and Outlook In this entry, the authors give a review of the physical mechanisms leading to structural colors, introduce
N
N
1802
plant species as well as microorganisms and viruses with the respective nanostructures thought to be responsible for or contribute to the coloration and present quite a number or organisms where the physical reason for the structural coloration has not yet been determined. Many animals have senses that either work in different bandwidths compared to the ones in humans, or they sense completely different properties (such as with echolocation or the magnetic sense). Since plants and animals interact on various levels and for various reasons, plants have evolved various properties tuned to animals. Structural colors are increasingly being used in current and emerging technology as well as in the arts. Some examples: The Austrian company Attophotonics develops structural colors as sensors for lab-on-a-chip applications. The Japanese company Teijin Fibers Limited produced one blue dress made from Morphotex ® structural colored fiber, a flattened polyester fiber mimicking Morpho butterfly structures. Qualcomm’s mirasol™ display technology was inspired by butterfly wings. ChromaFlair ® and SpectraFlair ® color-shifting paints from the company JDSU create color-shifting effects in cars. The emerging contemporary design practice Biornametics explores a new methodology to interconnect scientific evidence with creative design in the field of architecture, with role models from nature (e.g., structural colors in organisms) being investigated and the findings applied to design strategies. More often than not, natural structural colors are the inspiration for such technical products and applications. The field of biomimetics is dealing with the identification of deep principles in living nature, and their transfer to humankind [2]. General biological principles identified by the German biologist Werner Nachtigall that can be applied by engineers who are not at all involved in biology are, for example, integration instead of additive construction, optimization of the whole instead of maximization of a single component feature, multifunctionality instead of mono-functionality, energy efficiency and development via trial-anderror processes. We are hopeful that biomimetic colors will not only show the bright and shiny properties as the natural examples do, but will also
Nanostructures for Coloration (Organisms other than Animals)
exhibit one of the most important properties ensuring continuity of the biosphere: sustainability. Currently, with nanotechnology as booming new promising field, the sciences have started to converge, with nanobioconvergence as one of the most promising examples. New materials, structures, and processes can arise from this novel approach, perhaps exhibiting some of the multifunctional properties of the inspiring organisms, where various levels of hierarchy (with functionality on each and every level) are so refined. The convergence of the ways of thinking and novel approaches promises an interesting future. Acknowledgments The National University of Malaysia funded part of this work with its leading-edge research project scheme “Arus Perdana,” and the Austrian Society for the Advancement of Plant Sciences funded part of this work via the Biomimetics Pilot Project “BioScreen.” Living in the tropics and exposure to high species diversity at frequent excursions to the tropical rainforests is a highly inspirational way to do biomimetics. Profs. F. Aumayr, H. Sto¨ri, and G. Badurek from the Vienna University of Technology are acknowledged for enabling ICG extensive research in the inspiring environment of Malaysia; Bettina G. Neunteufl is acknowledged for carefully reading the manuscript. Mrs. Friedel Hinz and Mr. Foozi Saad are acknowledged for providing Figs. 10 and 7.
Cross-References ▶ Bioinspired Synthesis of Nanomaterials ▶ Biomimetics ▶ Biornametics – Architecture Defined by Natural Patterns ▶ Lotus Effect ▶ Nanoimprinting ▶ Nanomechanical Properties of Nanostructures ▶ Nanostructured Functionalized Surfaces ▶ Nanostructures for Photonics ▶ Nanostructures for Surface Functionalization and Surface Properties ▶ Nanotechnology ▶ Optical and Electronic Properties ▶ Rose Petal Effect ▶ Scanning Electron Microscopy ▶ Self-assembly of Nanostructures ▶ Self-repairing Materials ▶ Transmission Electron Microscopy
Nanostructures for Energy
References 1. Berthier, S.: Iridescences: the Physical Colors of Insects. Springer, New York (2007) 2. Bruckner, D., Gruber, P., Hellmich, C., Schmiedmayer, H.-B., Stachelberger, H., Gebeshuber, I.C. (eds.): Biomimetics – Materials, Structures and Processes. Examples, Ideas and Case Studies. Springer, Berlin (2011) 3. de Gennes, P.G., Prost, J.: The Physics of Liquid Crystals. The International Series of Monographs on Physics. Oxford University Press, Oxford (1995) 4. Dragota, S., Riederer, M.: Epicuticular wax crystals of Wollemia nobilis: morphology and chemical composition. Ann. Bot. 100, 225–231 (2007) 5. Eckert, A.W.: The World of Opals, p. 16 and 183. Wiley, New York (1997) 6. Gerwick, W.H., Lang, N.J.: Structural, chemical and ecological studies on iridescence in Iridaea (Rhodophyta). J. Phycol. 13(2), 121–127 (1977) 7. Glover, B.J., Whitney, H.M.: Structural colour and iridescence in plants: the poorly studied relations of pigment colour. Ann. Bot. 105(4), 505–511 (2010) 8. Hemsley, A.R., Collinson, M.E., Kovach, W.L., Vincent, B., Williams, T.: The role of self-assembly in biological systems: evidence from iridescent colloidal sporopollenin in Selaginella megaspore walls. Phil. Trans. R. Soc. Lond. B 345, 163–173 (1994) 9. Inchaussandague, M., Skigin, D., Carmaran, C., Rosenfeldt, S.: Structural color in myxomycetes. Opt. Express 18(15), 16055–16063 (2010) 10. Kinoshita, S.: Structural Colors in the Realm of Nature. World Scientific Publishing Company, Singapore (2008) 11. Koch, K., Barthlott, W.: Superhydrophobic and superhydrophilic plant surfaces: an inspiration for biomimetic materials. Phil. Trans. R. Soc. A 367(1893), 1487– 1509 (2009) 12. Lee, D.: Nature’s Palette: the Science of Plant Color. University of Chicago Press, Chicago (2007) 13. Lee, D.W., Lowry, J.B.: Physical basis and ecological significance of iridescence in blue plants. Nature 254, 50–51 (1975) 14. Martin, J.T., Juniper, B.E.: The Cuticles of Plants, p. 109. Edward Arnold, Edinburgh (1970) 15. Pederse´n, M., Roomans, G.M., Hosten, A.V.: Blue iridescence and bromine in the cuticle of the red alga Chondrus crispus Stackh. Bot. Mar. 23, 193–196 (1980) 16. Vigneron, J.P., Rassart, M., Ve´rtesy, Z., Kerte´sz, K., Sarrazin, M., Biro´, L.P., Ertz, D., Lousse, V.: Optical structure and function of the white filamentary hair covering the edelweiss bracts. Phys. Rev. E 71(8p), 011906 (2005) 17. Vukusic, P., Sambles, J.R.: Photonic structures in biology. Nature 424, 852–855 (2003) 18. Whitney, H.M., Kolle, M., Andrew, P., Chittka, L., Steiner, U., Glover, B.J.: Floral iridescence, produced by diffractive
1803
N
optics, acts as a cue for animal pollinators. Science 323(1), 130–133 (2009) 19. Williams, T.: The Iridoviruses. Adv. Vir. Res. 46, 345–412 (1996) 20. Willis, D.B. (ed.).: Iridoviridae. Current Topics in Microbiology and Immunology, vol. 116. Springer, Berlin/Heidelberg/New York/Tokyo (1985)
Nanostructures for Energy Stefano Bianco, Angelica Chiodoni, Claudio Gerbaldi and Marzia Quaglio Center for Space Human Robotics, Fondazione Istituto Italiano di Tecnologia, Torino, Italy
Synonyms Nanodevices; Nanomaterials; Nanorods; Nanotubes; Nanowires
Nanoparticles;
Definition Objects whose dimensions range between 100 and a few nanometers can be classified as nanostructures. Since the physical-chemical properties of matter change dramatically at the nanometer-scale, the introduction of this classification results particularly important to open a new field of investigation for next-generation materials and devices.
Introduction The worldwide need for power is mainly satisfied by the use of well-known energy resources as oil, coal, hydropower, natural gas and nuclear power. The environmental impact of most of these energy sources is tremendous, mainly because of the emission of greenhouse gases that causes the increase in the globally averaged temperature. In the next decades a rise of the global energy demand is expected, mainly because of the growth of world population especially in the developing countries. It is interesting to notice how the preindustrial world’s
N
N
1804
energy consumption grew about 0.3 terawatts (TW) year per year, and this value dramatically increased to 14 TW in the years 2000s and it is expected to double to 28 TW by year 2050. In the same period the world population increased only five times, clearly showing how the energy consumption is not uniformly distributed. An improvement in life quality in developing countries will cause a further tremendous increase in energy demand in the next future [1, 2]. A fast growing country needs far more energy than a developed country, essentially because a mature industrialized economy has learned how to produce and use energy more efficiently [2]. In this scenario, exploring renewable, efficient and green energy sources is the greatest challenge to be won for sustainable human progress. Innovation is one of the key strategies to address this challenge, and some of the most promising solutions are related to nanotechnologies. Materials at the nanometer-scale are able to show completely different properties than those people are used to in their everyday life. Understanding the huge potentialities of materials at the nanoscale and learning how to drive them in real devices opens new extraordinary opportunities. The field of alternative energy is certainly the most challenging platform for some of nanotechnology’s most exciting contributions [3]. New nanomaterials and nanostructures showed recently their potentiality in the development of devices for both energy production and storage, like third-generation photovoltaic cells, fuel cells and lithium-based batteries. Solar energy conversion systems are limited mainly by the poor properties of current materials, showing limited charge–carrier mobility and significantly high charge recombination rates for materials with sustainable costs. Moreover, theoretical limitations (Shockley/Queisser limit) are well established and fix a superior insurmountable efficiency threshold. Nanotechnology could help in the design of innovative architectures for a newgeneration solar cells, actually opening to a possible real alternative to traditional silicon solar cells. Valuable examples are dye-sensitized solar cells (DSCs). In the last few years, significant efforts were spent in designing new dye molecules with improved light harvesting properties as well as optimized mesoporous semiconducting layers with
Nanostructures for Energy
lower charge recombination rates. Devices based on photosensitized nanostructured TiO2 films and on liquid electrolytes have demonstrated light-tocurrent conversion efficiency greater than 10% [4]. Other approaches, like hybrid organic solar cells based on conjugated polymers or quantum dots solar cells with a light harvesting performed by semiconductor nanocrystals, are showing promising laboratory results. However, the achievement of satisfactory efficiency values still remains a challenge, mainly because of consistent losses during photoinduced charge separation and charge transport across the nanoassemblies. Fuel cells represent another technological field in which new concepts from nanotechnology can be beneficial. The development of this kind of devices for large-scale use is still limited by the lack of adequate systems for hydrogen storage. At the same time nanotechnologies offer new interesting opportunities, directly storing hydrogen into the fuel cells by nanostructures of carbon, zeolites, stacked clays, or light metals hydrides [5]. The introduction of nanoengineering also opens to the design of new electrodes. Nanotechnologies can lead to a more reduced use of precious metals (e.g., platinum nanoparticles for high surface area and low volume) along with improved membrane function and durability [6]. Used batteries and accumulators represent a tremendous environmental problem in current days. Nanomaterials could help not only in developing batteries showing higher energy densities but also devices that will reduce the use of disposal energystorage systems [7]. For example, it was demonstrated that silicon-based nanowires could significantly improve the specific capacity of Li-ion accumulators, currently powering-up electronic devices such as laptops, video cameras, cell phones, and many others [8]. In fact, such a new technology can grant a ten times increase of the capacity of traditional Li-ion batteries. Also, nanotechnology can reduce the possibility of batteries catching fire by providing less flammable electrode materials. New applications of nanotechnology are underway in the energy field. Their use will undoubtedly impact on modern society, hopefully changing and improving the way people are currently thinking to energy production/storage, leading to a more sustainable human growth.
Nanostructures for Energy
Energy Supply Systems The traditional approach to energy supply is based on the production of huge amounts (from MW to GW) of energy in localized production sites and its following delivery for minute use, causing significant losses across the distribution net. Petroleum and coal industry, nuclear power stations, hydroelectric dams, wind farms, and geothermal power installations are common examples of this approach (concerning both nonrenewable and renewable energy sources). Nanotechnologies are expected to dramatically increase the efficiency of alternative energy devices in all the different fields [3], creating the starting point for a new approach to a more sustainable energy management. In fact, it is reasonable to believe that the current scenario will be slightly modified in the next future. The centralized production of high quantities of energy will be more intensively joined by a distributed system of energetic hot spots, able to produce very low amounts of energy, with high spatial density. By this way, the production of energy occurs in direct connection or in close proximity with its use in devices (often miniaturized or microdevices), thus avoiding losses. In this framework, the efficiency of the single energy harvester could be no more a crucial point, while other aspects (like flexibility, integration in complex systems, and even aesthetics) gain importance. These kinds of approaches cover a broad interest in different fields, starting from new-generation buildings to new smart portable devices. The firsts are designed and fabricated according to a sustainable construction approach and are able not only to reduce the total energy consumption, but also to produce a large part of the energy required inside the building [9]. The next future devices (as household electrical appliances, cellular phones, computers, and biomedical devices) will be able to manage energy more efficiently, thanks to the introduction of new power supply systems. These systems will present an energy source, hopefully a renewable one as solar, mechanical harvesting, or thermoelectric, connected to a storage system as lithium-ion batteries, both managed by proper electronic power management circuits [10]. In this new futuristic scenario the role of nanomaterials and nanostructures in devices for energy production and storage will become more and more crucial.
1805
N
Two complementary strategies can be considered to manage energy in energy supply systems. A first class of devices, as batteries, fuel cells, and supercapacitors, need a finite quantity or a continuous supply of an active material or fuel to properly work. Devices of this kind can be classified as non-regenerative power supplies. A second class of tools, as photovoltaic cells, mechanical harvesters, and thermoelectric devices, do not need any material or fuel to work and can be classified as regenerative devices. In the next sections some valuable examples of both categories will be briefly introduced.
Nanostructures for Energy Production Dye-Sensitized Solar Cells (DSCs) Dye-sensitized solar cells (DSCs) were invented in the early 1990s by Michael Gr€atzel and Brian O’Regan at the E´cole Polytechnique Fe´de´rale de Lausanne (CH) [11], following the extensive research in semiconductor photo-electrochemistry started in the 1970s. These cells have immediately drawn a huge interest in the scientific community, because of their potentiality for the fabrication of devices with relatively high efficiency and potential low cost. In fact, these kinds of solar harvesters can be produced with abundantly available organic materials and widely diffused inexpensive semiconductors, using simple and scalable technologies and avoiding costly and energy-intensive high-vacuum treatments and materials purification steps. The working principle of the cell [4] is based on the presence of a molecular light absorber, which is able to harvest photons from sunlight and convert them to electrons, pumped up from a lower to a higher energetic level, thus creating an electric potential that can be used to produce electric work. The lightabsorbing dye is attached as a monolayer to the surface of a nanocrystalline film of a wide band gap semiconductor (TiO2 in a pure anatase form in the most common DSC architecture). The semiconductor works as electron collector, since the photogenerated charges are fast injected in the conduction band of the nanostructured solid. A careful design of the molecular energy levels of the dye makes the charge transfer toward the oxide strongly favored from an energetic point of view, thus drastically reducing the electronhole recombination process. The dye is electrically
N
N
1806
regenerated by charge donations from an electrolyte, which penetrates the pores of the nanofilm to have a close proximity with the molecules of the sensitizer, creating a 3D interpenetrating network junction. The electrolyte is an ionic conductor able to accept hole from (or donate electrons to) the oxidized dye, and the best performance are noticed for liquid electrolytes containing a I/I3 redox system. The electrons percolate through the nanoparticles, which are interconnected because of a short sintering process, and are collected by an external circuit, thus generating electric power without causing any permanent chemical transformation inside the cell. The role of the nanostructures in designing suitable photoelectrodes for high-performance cells is crucial. In fact, as reported from the first experiments of Gr€atzel, it emerges that it is possible to move from a sunlight-to-current conversion efficiency of 1.2% using a polycrystalline anatase film [12] to the 7.9% using a nanostructured mesoporous TiO2 layer [11], just varying the active exposed surface area of the electrode. In fact, on a flat surface a monolayer of dye is able to absorb only a little percentage of the impinging photons, since its optical cross section is sensibly smaller with respect to the geometrical area it occupies. Thus, for good energy harvesting efficiency it is mandatory that a big area is exposed to the incident flux of photons, and this is possible only with a nanostructuring of the photoanode. By carefully controlling the porosity of the film, it is possible to find the optimal equilibrium between the thickness of the film and the path that the photogenerated charges are able to cover (e.g., the diffusion length) before recombination. Harvesting Energy harvesting is the direct extraction of background energy from the environment and its conversion in electrical power [13]. Harvestable ambient energies can be classified in three main classes, namely radiations (see Sect. 3.1), kinetic/vibration energy, and thermal energy. Energy recovery is possible through properly designed devices. The heart of energy harvesting devices is an active material that is actually able to change/modify at least one of its properties as a response to the stimulus of ambient energy. The change of the material behavior, induced by environmental energy, is usually turned into electrical power. The diffusion of harvesting devices
Nanostructures for Energy
has been limited till now by their low efficiency in electrical energy production if compared to traditional energy sources, mainly due to the lack of well-performing materials. The development of new nanostructures and nanomaterials is promising to fasten up the development of a new generation of highly efficient devices for energy harvesting [3]. Energy from Vibrations
Several sources of environmental vibrations can be used to harvest energy, such as vibrations from household appliances, moving equipments, trains, helicopters flying overhead, or human movement. To extract that energy from the ambient a piezoelectric material is needed as the active material. Piezoelectric materials are able to directly convert the mechanical strain induced by vibrations into an electrical energy. The conversion is possible trough the structural deformation of the material that induces a local spatial charge separation through an electric dipole. Piezoelectricity can occur in polycrystalline ferroelectric ceramic materials such as barium titanate (BaTiO3), lead zirconate titanate (PZT) and polymeric materials as poly(vinylidene fluoride), PVDF. Traditional piezoelectric ceramics are very brittle, needing to be handled carefully to avoid breakage. The usability is also affected by their hard and fragile behavior since they have limited ability in following complex curved surfaces. This aspect is particularly limiting when sensors and harvesters are considered, since the capability of devices as harvesters to capture and extract energy is also related to the ability of the material (and, consequently, of the whole device) to well couple to the surface of the object generating the vibrations. Traditional polymeric piezoelectric materials can overcome this limitation being flexible and compliant, but unfortunately their electromechanical coupling coefficient is not comparable to traditional ceramics. These limitations have increased the attention to other approaches, and hybrid nanocomposite materials are one of the most promising alternatives. The final properties of nanocomposite materials are related to the connectivity between phases [10]. Nanocomposite materials are so promising for energy harvesting devices because of high coupling factors, to which they add low acoustic impedance, mechanical flexibility, and broad bandwidth coupled with low mechanical quality factor that make them interesting candidates also for transducer and acoustic applications.
Nanostructures for Energy
1807
Energy from Temperature Gradients
Heat adsorption
Heat and Charges flow
A thermoelectric material is able to generate a voltage when a temperature gradient is applied to it (Seebeck effect) and vice versa (Peltier effect). The thermoelectric effect is related to the ability of charge carriers to move freely in metals and semiconductors (like molecules in a gas); while moving they can conduct charge and heat too. If a temperature gradient is applied to the material, charge carriers tend to move from the hot part to the cold region of the material through a diffusion process. While moving to the cold side charges are accumulated, creating a voltage that can be externally exploited [14]. The conversion of thermal energy in electrical energy is possible through p-n junctions made of thermoelectric materials as described in Fig. 1. Thermoelectric generators are characterized by no moving parts that make them silent. They are reliable and scalable, and so they are considered quite interesting as power generators. Efficiency for thermoelectric materials and devices can be described by a nondimensional figure of merit that can be evaluated as ZT ¼ sS2T/k, where S is the Seebeck coefficient, s is the electrical conductivity, k is the thermal conductivity, and T is the absolute temperature. For a long time, thermoelectric materials have been considered too inefficient to be cost effective for applications. The current state-of-the-art material in this field is a Bi2T3-based alloy showing a ZT value close to one. New interest in thermoelectrics began in the middle of the 1990s, since theory suggested that thermoelectric efficiency could be significantly enhanced by the use of nanostructures. A well-performing thermoelectric material needs a variety of conflicting properties to be well optimized in order to work properly. It needs a large thermopower, which means a large value of the Seebeck coefficient; it must be a good electrical conductor, showing at the same time a very low thermal conductivity. In traditional materials, these transport characteristics are dependent the on one another, being deeply linked to interrelated material properties. Based on both theoretical and experimental predictions, the current approach to increase ZT is to use nanostructured materials, both in form of films than quasi-1D structures, like nanowires, nanobelts, and nanotubes. Nanostructured films are quite interesting because they can use the same fabrication routes of traditional materials, while showing improved thermal behavior thanks to phonon scattering at the interfaces
N
p
n
h+
e−
Heat rejection External Load
Nanostructures for Energy, Fig. 1 Schematic of a thermoelectric device based on a p-n junction between semiconducting thermoelectric materials
among nanograins. The main issue in nanostructured films is to control electron scattering at grain boundaries between randomly oriented grains that could decrease also electrical conductivity [14]. In 1D materials as nanowires, the thermoelectric parameters (S, s and k) can be optimized independently. A high value of ZT is predicted for 1D nanostructures, as a result of an increase of charge mobility due to the quantum confinement of the density of the states, together with a reduction in thermal diffusivity because of scattering at the grain boundaries [3]. Fuel Cells A fuel cell (FC) is an electrochemical cell that converts directly chemical energy into electricity. This characteristic makes it an appealing system to provide a relatively cheap power source, by using hydrogen or ethanol as fuels, and by producing water as waste. They are constituted by an anode (hydrogen or ethanol fuel source) and a cathode (usually oxygen from air) separated by an electrolyte. The basic operation is depicted in Fig. 2. Over the past few years, fuel cells have demonstrated increased reliability and lower costs because of the use of nanomaterials. The use of nanotechnology in the manufacturing process results in increased surface area, which originates large power and energy densities, long shelf life, and ease of miniaturization. These features, combined with MEMS technology
N
N
1808
Nanostructures for Energy e−
O2
Fuel (H2)
Pt
Anode
Pt
Membrane
H2O Cathode
Nanostructures for Energy, Fig. 2 Basic operation of a fuel cell. Hydrogen enters at the anode, where it is catalytically dissociated and ionized. Protons migrate by ionic conduction to the cathode through the polymer electrolyte membrane. Electrons travel through the external circuit where they do electrical work before arriving at the cathode. At the cathode, oxygen molecules react with protons and electrons to form water molecules
[10], are important for the development of a more powerful fuel cell for portable electronic devices. Nanomaterials are also increasingly used in the production, purification and storage of hydrogen. In fact, at present, the ability to store hydrogen at high volumetric and gravimetric densities, as well as the ability to extract/insert it at sufficiently rapid rates, is still a hot topic that could condition further improvements of the fuel cell technology. In particular, a big effort has been done by the research community in the past decades to identify candidates that could fit such strict characteristics. Examples of investigated nanostructures are CNTs, considered a promising solution, even if recently partially abandoned, light metal hydrides [5], and graphene. There are different types of fuel cells, depending on the type of electrolyte: the most diffused are proton exchange membrane fuel cells (PEMFCs) and high temperature fuel cells such as solid oxide fuel cells (SOFCs) and molten carbonate fuel cells (MCFCs). Among these PEMFC technology has become the dominant choice for the portable electronics market. Nowadays, nanotechnology is deeply embedded in the design of fuel cell components and, therefore, the development of nanostructured catalysts at the electrodes or optimized proton exchange membranes (PEM) has become an active area of research for fuel
cell applications. At present, many efforts are devoted to an optimization of the PEM characteristics, but despite great advancements in the development of nanostructured membranes, some restrictions and limitations remain. One of the challenges in current PEM fuel cell development is the lack of appropriate materials for the fabrication of protonconductive membranes with improved performances. The high cost of the commercially available Nafion® membrane, the most diffused polymeric electrolyte in FC, has induced the development of new protonconducting materials, including polybenzimidazoles, polyamides, polyether imides, polysulfones, polyphenylene sulfides, polyetheretherketones, and polyphenyquinoxalines. However, the Nafion® membranes show superior performance because of their high ionic conductivity compared to other nonfluorinated membranes. Proton conductivity is the most important property of membrane efficiency in fuel cells. Membranes with acceptable conductivity at high temperatures are desirable for fuel cell applications, but there is always the issue of decreased performance at high temperatures due to dehydration of the membrane and a resulting reduction in ionic conductivity. To overcome the problems faced by proton exchange membranes, numerous studies were conducted to develop alternative materials for membrane preparation. One of the approaches to improve water retention in the membrane is to incorporate hydrophilic, inorganic materials [6]. However, with this method it is difficult to avoid nanoparticle self-agglomeration in the polymeric matrix, and the method therefore induces non-homogeneous dispersions and poor electrochemical performance. In order to achieve a homogeneous distribution, addition of nanosized inorganic particles is carried out by various methods, such as the solgel method, the solution casting method, and PECVD to obtain high-performance nanocomposite membranes. The use of surfactants can assist in creating a uniform distribution of nanoparticles as well. A second challenge for the ultimate commercialization of fuel cells is the preparation of active, robust, and low-cost catalysts to replace platinum. Emphasis is placed on nanoengineering-based fabrication, processing, and characterization of multimetallic nanoparticles (transition metals) with controllable size (1–10 nm), shape, composition, and morphology.
Nanostructures for Energy
Nanostructures for Energy Storage Systems Many of the sustainable energy alternatives described produce or require electricity. Therefore, novel more efficient ways to store electricity are very much needed in the way to a more sustainable production, transformation, and use of energy. In this respect, electrochemical systems, such as batteries and supercapacitors, which can efficiently store and deliver energy on demand in stand-alone power plants, as well as provide power quality and load-leveling of the electrical grid in integrated systems, are playing a crucial role in the present energy economy. In addition, they are also seen as the power sources of choice for the progressive diffusion of sustainable hybrid/electric vehicles at high levels. Lithium-Ion Secondary Batteries Lithium-ion batteries (LIB) are one of the greatest successes of modern electrochemistry of materials. Their science and technology have been extensively reported in reviews [7] and dedicated books [8] to which the reader is referred for more details. A commercial LIB does not contain lithium metal; it comprises a negative electrode (generally, graphitic carbon), and a positive electrode (generally, layered lithium metal oxides such as LiCoO2), both capable of reversibly intercalate Li+ ions, these being separated by a nonaqueous lithium-ion conducting electrolyte (generally, a separator diaphragm soaked in a mixed ethylene – diethyl carbonate solution of LiPF6). During discharge, Li+ ions carry the current from the negative to the positive electrode, through the nonaqueous electrolyte [8]. During charge, an external electrical power source (the charging circuit) applies a higher voltage (but of the same polarity) than that produced by the battery, forcing the current to pass in the reverse direction. Lithium ions then migrate from the positive to the negative electrode, where they become embedded in the porous electrode material in a process known as “intercalation.” The current-generation LIBs are increasingly being used by virtue of their high (volumetric and gravimetric) energy density; but, although commercially successful, some disadvantages arise, related to relatively low power density, large volume change on reaction, safety, and costs. In fact, the limits in performance using the current electrodes/electrolytes are being reached and further breakthroughs in materials
1809
N
chemistry are essential, especially modifying and improving already known materials. The aforesaid shortcomings can be reduced by the application of nanotechnology. It is important to design and fabricate nanostructured electrode materials that provide high surface area and short diffusion paths for ionic transport and electronic conduction. This gives rise to a larger electrode/electrolyte surface and enables Li-ion diffusion through the electrode material to be limited as much as possible. Moreover, the volume changes and structural evolution that occur as the electrodes accept and release ions can be quite different for nanoscale materials, thus greatly affecting the cycle/service life. The potential advantages and disadvantages associated with the development of nanomaterials applied to LIBs can be summarized as follows [7]. Advantages include (a) the strain associated with lithium ions insertion/removal is better accommodated (along with a wider solid-solution region), thus improving service life as shown in Fig. 3; (b) new electrode reactions are enabled to occur that are not possible for materials composed of micrometer-sized particles; (c) a high surface area permits a more extensive electrode/electrolyte contact area, leading to higher charge/discharge rates; and (d) the reduced dimensions significantly reduce the path lengths for electronic and Li+ transport, thus permitting operation with low electronic conductivity and/or at higher power. As for disadvantages, the following have been identified: (a) an increase in undesirable (side) reactions with the electrolyte due to the high electrode/electrolyte contact area, possibly leading to self-discharge, poor cycling, and calendar life; (b) the volume of the electrode increases because of poor particle packing, leading to lower volumetric energy densities unless special compaction methods are developed; and (c) potentially, nanoparticles may be more complex to synthesize, especially their dimensions may be difficult to control. There are numerous examples in which nanostructured materials have been used in Li-ion batteries as both electrodes and electrolytes. In meso- and/or nanoporous architectures, for instance, the nanoscale domains of the material are accessible to molecular reactants because of rapid mass transport via diffusion through the continuous volume of mesopores. Very recently, B. Scrosati, Y.-K. Sun, and colleagues developed a high-capacity, nanostructured tin-carbon anode, and a high-voltage, high-rate
N
N
1810
Nanostructures for Energy, Fig. 3 Example of how nanostructuring can help in improving the performances of Li-ion battery electrodes by reducing the strain associated with lithium ions insertion/ removal (Readapted from Chan et al. [15])
Nanostructures for Energy
Initial substrate
Bulk material
Nanoparticles
Nanopillars
After cycling
Bulk material
Nanoparticles
Nanopillars
Pulverization Loss of contact with current collector Poor electron transport
Li[Ni0.45Co0.1Mn1.45]O4 spinel cathode. When the two parts are put together, the result is a high-performance battery with a high energy density and rate capacity, highly suitable for powering low or zero emission HEV or EV vehicles [15]. The ultimate expression of the nanoscale in rechargeable lithium batteries is the formation of 3D nanoarchitectured cells, in which pillared anodes and cathodes are interdigitated [16]. The prospect of having active and passive multifunctional components interconnected within a 3D architecture is envisioned to lead to superior energy-storage capacity, high-volume synthesis, and improved safety. A nanowire battery has been invented by a team led by Dr. Y. Cui at Stanford University in 2007 [17]; it produces ten times the amount of electricity of existing LIBs. The team’s invention consisted of a stainless steel anode covered with silicon nanowires, to replace the traditional graphite anode. Silicon, which stores ten times more lithium than graphite, allows a far greater energy density on the
Efficient electron transport
Facile strain relaxation
Good contact with current collector
anode, thus reducing the mass of the battery. The large surface area further allows for fast charging and discharging and efficient electron transport down the length of each nanowire. Electrochemical Capacitors Electrochemical capacitors (ECs), also named “supercapacitors,” are capacitors with relatively high energy density (typically on the order of thousands of times greater with respect to conventional electrolytic capacitors) [18]. The term “supercapacitor” is usually used to describe an energy-storage device based on the charge storage in the electrical double layer (EDL) of a high-surface-area carbon in aqueous electrolytes. In fact, they store electrical energy, like batteries, but using a different mechanism: while batteries do it chemically, supercapacitors store electricity physically, by separating the positive and negative charges. Their lifetime is in principle infinite, as they operate solely on electrostatic surface charge
Nanostructures for Energy
accumulation. In comparison with conventional batteries or fuel cells, ECs also have a much higher power density. They can be charged or discharged at a rate that is typically limited by current heating of the electrodes. So, while existing ECs have energy densities that are perhaps one tenth that of a conventional battery, their power density is generally 10–100 times greater. Supercapacitors can be used as uninterruptable power sources (UPSs), can be coupled with batteries to provide peak power, and can replace batteries for memory backup. The supercapacitor is ideal for energy storage that undergoes frequent charge and discharge cycles at high current and short duration. They have applications as energy-storage devices used in vehicles and for smaller applications like home solar energy systems where extremely fast charging is a valuable feature. Since their discovery, these devices have attracted considerably less attention with respect to batteries as energy-storage devices. Nevertheless, thanks to both the contribution of nanotechnology and the better understanding of the charge storage mechanisms (ion behavior in small pores) the interest on ECs has noticeably increased very recently [18]. Nanotechnology is important because it enables large surface areas to be obtained to optimize performance in terms of capacitance, ionic and electronic resistance. There are numerous examples in the literature showing that nanosized crystallites and/or mesoporous materials exhibit significantly higher specific capacitance as compared to nonporous materials or micron-sized materials. The assembly of nanoscale materials is also important. Recent trends in supercapacitors involve the development of high-surface-area activated carbon electrodes to optimize the performance in terms of capacitance and overall conductivity. Attention has been focused on nanostructured carbons, such as aerogels, nanotubes, and nanotemplates [19]. One structure envisioned to be of interest is an array of vertically aligned carbon nanotubes where the spacing between the tubes is matched to the diameters of the solvated electrolyte ions. The main focus has been the optimization of interparticle contact resistance and electrolyte wettability of the pores. The capacitance increase is almost 50% compared with the best performing commercially available activated carbons. Nanostructured metal oxide materials have also been investigated for supercapacitor applications, especially RuO2 and RuO2·xH2O nanotubes composites [20].
1811
N
The Impact of Nanostructures on Traditional Energy Sources: An Overview As the worldwide demand for energy rises at an everincreasing rate, there is a new urgency to improve the efficiency and sustainability of actual power generation technologies. One of the keys to address this challenge is innovation, and some of the most promising solutions are occurring at the nanoscale. Perhaps the most mainstream acceptance within renewable technology has come in the form of wind energy, which is approaching cost competitiveness with traditional energy sources. Countries such as Germany, Spain, and Denmark are already beginning to utilize substantial amounts of wind energy to meet their growing electricity needs. Nanotechnology impacts the wind industry by improving turbine performance and reliability to allow for longer lifetime, less fatigue failure, and lower costs of generation. New lubricants containing nanoparticles act like mini ball bearings, thus reducing the friction generated from turbine rotation, consequently improving its life cycle. Advancements in nanocoatings, such as deicing and self-cleaning technologies, also help improving efficiencies, thus rendering ice and dirt buildup on the turbines virtually nonexistent. In spite of the optimistic predictions of its proponents, alternative energy sources such as wind, hydro, and biomass are unlikely to seriously challenge coal, oil, natural gas, or nuclear power for a meaningful share of the worldwide energy needs in the near future. However, how these traditional energy sources are produced will rapidly change. In fact, nanotechnology is already at work modifying how coal, oil, and gas are being produced. One of the more prominent examples can be found in the application of new nanoscale catalysts to the production and refining processes of fossilfuel resources. Moreover, modern fossil-fuel power plants, including coal-fired systems, require components that are able to withstand higher temperatures and pressures to improve efficiency, lower maintenance costs and decrease emissions levels. In this field, nanotechnology-based innovation is being credited with improvements in efficiencies and life spans of existing technologies, as well as the introduction of novel and disruptive power-generating components. Nanotechnology has the potential to revolutionize the electric utility industry as well. Many of the
N
N
1812
Nanostructures for Energy
A
B
C
armchair
zigzag
chiral
Nanostructures for Energy, Fig. 4 Structure of single-walled carbon nanotubes (SWCNTs); carbon nanotubes can be metallic or semiconducting, depending on how their chickenwire-like lattice rolls into a tube
improved insulation materials could have a noticeable impact on overall energy demand. All of these products are made possible by advancements in nanotechnology. Nevertheless, the greatest long-term cost savings will probably come from the creation of “intelligent energy networks,” that is, integrated systems of more energy-efficient computers and sensors that gauge and regulate customer energy use. Such networks already exist today, but as nanotechnology enables computers and sensors to become smaller, but correspondingly more powerful, the cost effectiveness of deploying such devices will increase, along with the number of possible applications.
Conclusions components used in the electric utility industry can be modified to yield incremental improvements in the overall product performance. The end-benefit is that electric utility providers can improve their operating margins by making existing equipment both last longer and more efficient. Nanotechnology can provide immediate improvements to the electrical utility system of moving power by efficiently improving the transmission of electrons, as well as create newer and even better conducting materials. The real payoff may be carbon nanotubes (CNTs), which are unbelievably thin, possess 100 times the strength of steel, have only one sixth the density of aluminum, and efficiently conduct electricity. Most promising are the singlewalled carbon nanotubes (SWCNTs). The broad range of uses for SWNTs is connected to the different nanotube properties according to how the hexagonal lattice is oriented (see Fig. 4), which together with the nanotube diameter determines the chirality. This combination of strength and conductivity suggests that in the future, energy providers should not have to dig up city streets to lay new wires. The existing lines could be simply replaced with super strong, highly conductive carbon nanotube wires. Reduction of energy consumption is another active area in nanotechnology, targets being the development of more efficient lighting, better combustion systems, and/or use of lighter and stronger materials in the transportation sector. Light-emitting diodes (LEDs) are interesting examples of how energy consumption can be reduced; or “smart” windows, which can change reflective properties to attract or deflect sunlight depending on the customer’s needs. Similarly,
New materials hold the key to fundamental advances in energy production and storage, both of which are vital in order to meet the challenge of global warming and the finite nature of fossil fuels. Nanomaterials in particular offer unique properties or combinations of properties as electrodes and electrolytes in a range of energy devices. Through innovation, the efficiencies of present technologies can be improved and discover new ways by which mankind can prosper. Nanotechnology provides researchers with the opportunity to attain sustainable development, thus overcoming one of the greatest challenges of modern times, by using some of the simplest and smallest available means. In conclusion, nanotechnology, known as paying attention to structures technology of the twentieth century, owing to benefits such as being potentially inexpensive, effective, and promising for to come. Hence, many companies and researchers are getting curious day by day in order to explore especially its energy application. Indeed, to minimize energy waste, it should become common and applicable as soon as possible in spite of the fact that it is quite a new technology.
Cross-References ▶ Carbon-Nanotubes ▶ Hybrid Solar Cells ▶ Nanomaterials for Electrical Energy Storage Devices ▶ Nanomaterials for Excitonic Solar Cells
Nanostructures for Photonics
▶ Nanostructured Materials for Sensing ▶ Piezoresistivity ▶ Thermoelectric Heat Convertors
References 1. US DOE Office of Basic Energy Sciences, Alivisatos, P., Cummings, P., De Yoreo, J., Fichthorn, K., Gates, B., Hwang, R., Lowndes, D., Majumdar, A., Makowski, L., Michalske, T., Misewich, J., Murray, C., Sibener, S., Teague, S., Williams, E.: Nanoscience research for energy needs. Report of the Department of Energy NanoSummit: Nanoscale Science and Our Energy Future, Wardman Park Marriott, NW, Washington, DC, 23–24 June 2004. This document is a work of the U.S. Government and is in the public domain. Printed in the United States of America, 2nd edn (2005) 2. Arunachalam, V.S., Fleisher, E.L.: Harnessing materials for energy. MRS Bull. 33, 261 (2008) 3. Liu, J., Cao, G., Yang, Z., Wang, D., Dubois, D., Zhou, X., Graff, G.L., Pederson, L.R., Zhang, J.G.: Oriented nanostructures for energy conversion and storage. ChemSusChem 1, 676–697 (2008) 4. Gr€atzel, M.: Recent advances in sensitized mesoscopic solar cells. Accounts Chem. Res. 42, 1788–1798 (2009) 5. Crabtree, G.W., Dresselhaus, M.S.: Harnessing of materials for energy – the hydrogen fuel alternative. MRS Bull. 33, 421–428 (2008) 6. Thiam, H.S., Daud, W.R.W., Kamarudin, S.K., Mohammad, A.B., Kadhum, A.A.H., Loh, K.S., Majlan, E.H.: Overview on nanostructured membrane in fuel cell applications. Int. J. Hydrogen Energ. 36, 3187–3205 (2011) 7. Arico`, A.S., Bruce, P., Scrosati, B., Tarascon, J.M., van Schalkwijk, W.: Nanostructured materials for advanced energy conversion and storage devices. Nat. Mater. 4, 366–377 (2005) 8. van Schalkwijk, W., Scrosati, B. (eds.): Advances in Lithium-Ion Batteries. Kluwer/Plenum, New York (2002) 9. Sartori, I., Hestnes, A.G.: Energy use in the life cycle of conventional and low-energy buildings: a review article. Energ. Buildings 39, 249–257 (2007) 10. Cook-Chennault, K.A., Thambi, N., Sastry, A.M.: Powering MEMS portable devices – a review of non-regenerative and regenerative power supply systems with special emphasis on piezoelectric energy harvesting systems. Smart Mater. Struct. 17, 1–33 (2008) 11. O’Regan, B., Gr€atzel, M.: A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO2 films. Nature 353, 737–740 (1991) 12. Vlachopoulos, N., Liska, P., Augustynski, J., Gr€atzel, M.: Very efficient visible light energy harvesting and conversion by spectral sensitization of high surface area polycrystalline titanium dioxide films. J. Am. Chem. Soc. 110, 1216–1220 (1988) 13. Angrist, S.: Direct Energy Conversion. Allyn & Bacon, Boston (1982) 14. Snyder, G.J., Toberer, E.S.: Complex thermoelectric materials. Nat. Mater. 7, 105–114 (2008)
1813
N
15. Hassoun, J., Lee, K.S., Sun, Y.K., Scrosati, B.: An advanced lithium ion battery based on high performance electrode materials. J. Am. Chem. Soc. 133, 3139–3143 (2011) 16. Long, J., Dunn, B., Rolison, D., White, H.: Three-dimensional battery architectures. Chem. Rev. 104, 4463–4492 (2004) 17. Chan, C.K., Peng, H., Liu, G., McIlwrath, K., Zhang, X.F., Huggins, R.A., Cui, Y.: High-performance lithium battery anodes using silicon nanowires. Nat. Nanotechnol. 3, 31–35 (2008) 18. Conway, B.E.: Electrochemical Supercapacitors. Kluwer/ Plenum, New York (1999) 19. Frackowiak, E., Be´guin, F.: Carbon materials for the electrochemical storage of energy in capacitors. Carbon 39, 937–950 (2001) 20. Kim, I.H., Kim, J.H., Kim, K.B.: Electrochemical characterization of electrochemically prepared ruthenium oxide/carbon nanotube electrode for supercapacitor application. Electrochem. Solid St. 8, A369–A372 (2005)
Nanostructures for Photonics Andrea Toma1, Remo Proietti Zaccaria1, Roman Krahne1, Alessandro Alabastri1, Maria Laura Coluccio1,2, Gobind Das1, Carlo Liberale1, Francesco De Angelis1, Marco Francardi2,3, Federico Mecarini1, Francesco Gentile1,2, Angelo Accardo2,4, Liberato Manna1 and Enzo Di Fabrizio1,2 1 Nanobiotech Facility, Istituto Italiano di Tecnologia, Genoa, Italy 2 Lab. BIONEM, Dipartimento di Medicina Sperimentale e Clinica, Universita` “Magna Grecia” di Catanzaro, Catanzaro, Italy 3 International School for Advanced Studies (SISSA), Edificio Q1 Trieste, Italy 4 Soft Matter Structures Group ID13 – MICROFOCUS Beamline, European Synchrotron Radiation Facility, Grenoble Cedex, France
Definition “Nanostructures for Photonics” represents a collection of devices whose working principles are based on photons at the nanoscale level.
Overview In the last decade, new disciplines emerged where the terms nano and photon happen to play a fundamental
N
N
1814
descriptive role. Their origin dates back to the ancient Greece, with nano (¼nanoz) indicating something of small dimensions, and photon (¼jϖΨψρz) indicating the light. In Physics, but also in Chemistry, Biology, Material Science, and many more fields these terms have been used for indicating the studies of extremely small objects by means of light. This new technological frontier has already produced some amazing results: LCD based mobile phones, nanoparticles for optical tagging of biological molecules and hyperthermia, or devices for single-molecule detection are just few of many examples where nanophotonics plays a crucial role. In terms of optical devices, their functionality can be divided in three general key parts: light generation, light guiding and light harvesting. Here will be introduced some approaches describing these aspects, starting from colloidal nanocrystals for photonics to end, through the introduction of photonic crystals and metamaterials, with plasmonic devices for Raman spectroscopy. The aim is to prove that photonics at nanoscale level is not just an exotic research field but a mature discipline capable of offering devices for a multitude of uses.
Light Generation: Semiconductor Nanocrystals for Lasing Applications Colloidal semiconductor nanocrystals are a very attractive alternative as light emitting materials to more conventional nanostructures, because they can be fabricated in a cost-effective bottom-up approach by wet chemical synthesis. In the last decade, great progress has been achieved regarding the control on nanorcrystal shape and composition, leading to size dispersions of less than 5% in spherical particle samples, and to rod- and multibranched shapes. One particular property of colloidal nanocrystals is their exposed surface that needs to be stabilized via surfactant molecules. The related obstacle that has to be overcome toward highly efficient light emission is the nonradiative recombination of the electrons and holes at surface states. The most promising solution to this problem are core-shell architectures, where a low band gap material nanocrystal is capped by a second material with higher band gap. In such structures, the photoexcited charges are confined by the nearly defect-free internal interface between the two materials (see band alignment illustration in Fig. 1).
Nanostructures for Photonics
Nanostructures for Photonics, Fig. 1 Schemes illustrating the core-shell architectures and the related band structure of spherical (a) and dot/rod nanocrystal structures (b)
Here, the emission wavelength is determined on one hand by the band gap of the core material, and on the other hand by the quantum confinement effects, since typical diameters of nanocrystals are in the range of 1–8 nm. The most popular systems for light emission in the visible consist of CdSe, CdS, ZnSe, and ZnS materials. For example, the emission of CdSe based systems can be tuned by the nanocrystal size in the range from 1.8 to 2.5 eV, as demonstrated in Fig. 2a. Two very successful core-shell architectures are sketched in Fig. 1, the spherically symmetric coreshell structure, and the dot-in-a-rod (dot/rod) structure that can be fabricated by the seeded growth technique [1]. The dot/rod architecture in Fig. 1b combines the advantageous optical properties of rod-shaped nanocrystals, such as strongly linearly polarized emission and oriented assembly, with a high quantum yield (ratio photons in/photons out) comparable to that of spherical nanocrystals (Fig. 2). For lasing devices two factors are important: the optical gain of the light emitting material and the quality of the resonant cavity that provides the optical feedback. The optical gain can be improved by the composition and architecture of the nanocrystals, and by the homogeneity and density of their assembly. The optical feedback can be obtained from semitransparent mirrors or distributed feedback resonators that consist of Bragg reflectors. The optical gain of the material is typically characterized by transient (time-resolved) optical absorption spectroscopy, which reveals the population dynamics of the energetic levels of the nanocrystals [3]. If the emitting ground state population maintains an occupation larger than unity, optical gain is achieved. One peculiar property of the dot/rod core-shell nanocrystals is that optical gain can be obtained from the optical transitions related to the core and from the transitions in the shell material.
Nanostructures for Photonics
(a)
N
10K 56Å 33Å
26Å
Luminescence
22Å
19Å
Absorbance
Nanostructures for Photonics, Fig. 2 (a) Emission and absorbtion spectra of spherical CdSe nanocrystals (With permission from Ref. [2]). (b) Transmission electron microscopy images of dot-ina-rod core-shell nanorods with similar core size and different rod length. The scale bar corresponds to 50 nm (With permission from Ref. [1])
1815
16Å 15Å
13Å 12Å
2.0
2.5 3.0 Energy (eV)
This effect extends the spectral range significantly, for example in the CdSe/CdS system from 570 to 650 nm (core emission) to 490 nm emission wavelength of the shell [4]. So far, the fabrication of efficient optical cavities for nanocrystal assemblies remain challenging since, for example, conventional sequential evaporation of Bragg reflectors does not work due to the incorporation of the inhomogeneous nanocrystal layer. Hence, novel approaches to obtain cavities have been sought, like the loading of cylindrical microcavities with nanocrystal solutions via the capillary effect [5], or the hot embossing of two pre-fabricated Bragg reflectors which embed the nanocrystal layer in a sandwich structure [6]. Another innovative approach consists in the laser emission from core-shell nanorod assemblies where the layer of nanorods itself formed the resonant cavity [3]. Here, circular rings with micron scale width and sharply defined lateral edges were formed via the coffee stain effect by simple deposition of nanocrystals dissolved in toluene (see Fig. 3a). The lateral edges of the rings themselves functioned as optical interfaces and were suitable for achieving feedback in a Fabry– Perot resonator scheme (Fig. 3b). Such microlasers
3.5
could be very interesting for a new generation of photonic devices integrated onto substrates for lab-ona-chip or for other compact, integrated applications.
Light Guiding: From Photonic Crystals to Metamaterials Transportation of light to a detector is one of the fundamental aspects of any optical device. The light, during its path, might be affected by the surrounding world thus undergoing a modification of its properties. This change can be temporary, namely existing only during the path toward the detector or even permanent. An example of the first kind is given by the effective wavelength of surface plasmon polaritons on a metallic interface whereas permanent modification can be well described by white light entering a fiber. In fact, in this case the light will loose most of its spectral components once the detector will be reached. A further example is given by metamaterials, usually defined as exotic devices capable of strongly modifying the standard properties of the light. These different situations indicate a strong relation between the intended
N
Nanostructures for Photonics
b
Intensity (arb. Units)
a
1816
Output energy (arb. un.)
N
580
0
2.8 3.2 3.6 4.0 Pump fluence(mJ/cm2)
590
600 610 wavelength (nm)
620
Nanostructures for Photonics, Fig. 3 (a) Optical microscope images of a self-assembled circular ring formed via coffee stain evaporation dynamics. The ring consists of densely packed core-shell nanorods and acts as a microcavity for laser feedback.
(b) Emission spectra from a Fabry–Perot self-assembled microcavity made of nanorods with 25 nm length recorded at different pump fluences. Inset: emission intensity as a function of the pump fluence (With permission from Ref [3])
application and the method of transportation of light which must be properly chosen.
characteristics of samples embedded in the crystal. For the present purposes, these devices will be seen either as filters or waveguides. In fact, according to their geometrical and material properties, PCs can realize optical gaps in specific spectral regions. Consequences are the impossibility of the light to propagate inside the crystal. Figure 5a is an example of a square-like photonic crystal waveguide where light at 2.85 mm is obliged in following the free path of the crystal. The structure is formed by Silicon cylinders of 200 nm radius in air. The periodicity (lattice constant) of the crystal is 1 mm. When the optical gap concept is associated to the fiber world, Photonic Crystal Fibers are obtained. In this case, light propagates along the axis of the cylinders as in Fig. 5b. Among the PCFs special attention deserves the Photonic Quasicrystal Fibers (PQFs). Their geometry is similar to photonic crystals with the fundamental characteristic of loosing the translational periodicity. Only rotational and mirror symmetry are proper of a quasicrystal [14, 15]. Figure 6 gives some examples of quasicrystal. One of the advantages of considering PQFs versus PCFs is in a wider single-mode operation regime than the PCFs, maintaining similar crosssectional dimensions. The typical approach for determining the optical characteristic of a fiber is the calculation of its
Periodic and Quasiperiodic Fibers A standard and well-established method for the transportation of light is by means of optical fibers. These devices are nowadays commonly used for applications such as Internet, allowing a high-speed data transfer (Fig. 4). A different kind of fiber known as Photonic Crystal Fiber (PCF) was introduced in [8] with the idea of improving the classical fibers in many aspects, such as single-mode operation regime in a wide range of wavelengths, air-core guidance or reduced crosssectional dimensions. In fact, while standard fibers have diameters of some hundreds of micrometers, PCFs can be reduced down to few microns [9–11]. The physics behind the working principle of PCFs fibers lies in their geometrical characteristics. Photonic crystals (PCs) are structures showing a translational periodicity of the medium permittivity [12, 13]. A three dimensional example of a face centered cubic PC is given in Fig. 4. The main characteristic of a PC is the capability of controlling the interaction between radiation and matter in order to obtain new physics. For example, they can induce a local modification of the field, which in turn will affect the optical
Nanostructures for Photonics
1817
N
Nanostructures for Photonics, Fig. 4 The scanning electron microscope image of the scaffold type a fcc optical lattice. The lattice constant is about 1.5 mm. (a) A bird’s eye view. (b, c) Magnified images of (b) top surface of the structure, and (c)
cross section of a crack in the rim of the structure. (d, e) Simulated cross section of fcc lattice in (a) < 1,1,1 > and (b) < 1,1,1 > (With permission from Ref. [7])
a
a multimode fiber, which makes single-mode fibers the best choice for applications related to long distance communications. This explains why fibers working as single mode in a wide range of frequencies are very desirable for a number of applications. However, in case of PQFs the V parameter method has been demonstrated not to be working in many cases [15]. For this reason, a different approach consisting in the direct calculation of the effective area of the second-order mode is preferable. Figure 7 shows a comparison between these two approaches for three kinds of fibers. Besides a disagreement between the V parameter and the effective area approach as in Fig. 7b, c, it is possible to notice that the single-mode operation regime for both the eightfold and tenfold fibers is much more extended than the sixfold one. This makes these kinds of fiber more robust for transporting signal at long distances.
b
Z (μm)
10
0
−10 −10
0 X (μm)
10
Nanostructures for Photonics, Fig. 5 (a) Two-dimensional photonic crystal waveguide. Transverse magnetic polarized light at 2.85 mm coming from the bottom of the figure travels along the free path. The period (distance between two Silicon cylinders) is 1 mm, the radius is 200 nm. (b) Photonic crystal fiber. The blue arrow indicates the direction of propagation of the light
V parameter. This value is usually considered as the quantity capable of identifying the single-mode operation regime of a fiber. In fact, in the case of singlemode fiber, for a given wavelength and polarization, only a particular mode can propagate. This avoids the problem of the destructive interference that occurs in
Fishnet Structures: A First Step Toward Metamaterials Metamaterials are artificial man-made materials exhibiting particular properties such as negative refraction [16–19], which can lead to several
N
N
1818
Nanostructures for Photonics
b d
c
a A
Nanostructures for Photonics, Fig. 6 Cross section of three kinds of PQFs: (a) sixfold, (b) eightfold, and (c) tenfold. Gray filled circles denote air holes drilled in silica fiber. The central
hole (empty circle) is absent in the fibers. The quantities A and d represent the lattice constant and the hole diameter, respectively (With permission from Ref. [15])
Veff & Eff. area/A2
a Nanostructures for Photonics, Fig. 7 Effective V parameter (black squares and curves) and the secondorder mode normalized area (red circles and curves) as a function of A/l for (a) sixfold, (b) eightfold, and (c) tenfold quasicrystal fibers. The dashed lines are drawn at p. A is the lattice constant (With permission from Ref. [15])
b
7
c
7
7
5
5
3
3
1
1
Veff 5 3
2nd-order mode area
1 0
10 A/λ
20
applications like the realization of superlenses or invisibility cloak. In order to reach a negative refractive index, simultaneously negative electric permittivity and magnetic permeability need to be reached as pointed out by Veselago in 1968 [16]. A negative permittivity is a relatively easy task to obtain, since all metals below their plasma frequency exhibit it. On the other hand, a negative permeability is more difficult to achieve; the lack of magnetic charges and the weak interaction between matter and the magnetic part of light, make scientists capable to attain it only in certain frequency regions. Combining thin metallic wires and rods in a three layers Metal/Dielectric/ Metal structure (as shown in Fig. 8), constitutes the so called fishnet structure which has been proved to exhibit peculiar characteristics both in metamaterials and plasmonics fields.
0
10 A/λ
20
0
10 A/λ
20
Fishnet structures have been widely investigated in metamaterials research field both from an experimental and a theoretical point of view [20]. The reason is that their geometry is simple and its fabrication is relatively easy. However, the mechanism behind the negative refractive index behavior is nowadays not completely understood. The main idea is that properly polarized light passing through the structure can experience at the same time both negative permittivity and permeability. Considering, for instance, an impinging electromagnetic field polarized as shown in Fig. 9b: the electric field oscillates along thin wires allowing a negative electric permittivity and the magnetic field can induce a resonance through the aligned sets of rods, just like in a LC circuit. In fact, whereas the capacitor-like response is directly given by the vertical structure of the fishnet (two metallic plates separated
Nanostructures for Photonics
b 100 nm
a
N
600 nm
Nanostructures for Photonics, Fig. 8 (a, b) SEM images of a fabricated fishnet structure. (c) Sketch of the vertical structure. The inset in (a) shows the cell unit of the fishnet structure
1819
300 nm 300 nm
c Metal
25 nm
Dielectric 25 nm
a
35 nm
Metal
b
Nanostructures for Photonics, Fig. 9 (a) Transmissivity spectrum for the presented fishnet structure. (b) SEM micrograph of measured fishnet structure and fields direction of the incident light
Transmission
0.6 E 0.4
H
0.2
0.0
400 nm 0.6
0.8
1.0 1.2 1.4 Wavelength (μm)
by a dielectric, see Fig. 8c), the inductive response is expressed by the electric currents, generated by the oscillating magnetic field, that circularly flow in the plates (as regular currents) and across the dielectric (as displacement current). The resonant behavior can be observed, for example, in the transmission spectrum (Fig. 9a): if the magnetic resonance occurs, a dip in the transmission coefficient can be noticed. Before performing the experiments, also according to previous studies where similar structure were investigated, a magnetic resonance between 1,300 nm and 1,500 nm was expected [20]. In Fig. 9b, this kind of resonance is not clearly visible, probably because of high losses in that spectral range which damp transmission or reflection features. Nevertheless, it is possible to notice a resonant behavior at shorter wavelengths. Indeed current loops, which are supposed to induce the magnetic resonances, are generated by internal gap surface plasmons polaritons (SPPs). This
1.6
means that, more than one resonance exist, actually several of them may occur as far as the matching condition subsists: kSPP ¼ kwave þ mGx þ nGy with : Gx ¼
2p 2p and Gy ¼ ax ay
In this relation, kSPP and kwave are, respectively, the wave vector of the excited SPP and the one of the incident light. ax , ay are the lattice periodicities along the x and y fishnets axis. Therefore, higher frequency resonances can be reached as far as the lattice provides the missing momentum through Gx and/or Gy . However, a negative refractive index is not guaranteed in the entire spectrum since the equivalent plasma frequency of the medium is dominated by the cutoff wavelength of the holes of the fishnet. To conclude, fishnet structures represent a good example on the concept of metamaterials and how
N
N
1820
their fascinating properties are related to plasmonic effects. A deep study of these structures can be considered a first step to understand the underlying physics of negative refraction enabling, eventually, the realization of more complex devices exploiting the effects of negative index materials.
Light Harvesting: Supported Metal Nanoarrays for Plasmonic Devices Light harvesting plays an important role for the operation of many kinds of devices, such as biosensors or photovoltaic cells. In fact, besides the generation and transportation of light, an efficient coupling mechanism between the guiding device and the investigated bio-sample in order to produce a readable signal is a fundamental step of a generic photonic device. For example, metallic nanostructures have the property of realizing a high localized electric field which can be efficiently coupled to a bio-sample. This, in turn, means high signal-to-noise ratio from the sample, namely a good signal for the detector can be produced. Metallic nanoparticles exhibit specific optical properties in the visible/near-infrared range due to the collective excitation of the conduction electrons (Localized Surface Plasmon Resonances – LSPRs). The field enhancement provided by metallic nanostructures has been used to manipulate lightmatter interactions and enhance nonlinear phenomena. The strong effects associated to the LSPR excitation on metal nanoparticles are the core of many interesting perspective applications such as single- molecule detection and photovoltaic conversion. Here two specific examples are introduced, that is, nanoaggragate and nanoantenna arrays where light harvesting and field enhancement will be exploited with respect to high sensitivity Raman spectroscopy. The use of advanced lithographic techniques allows precise control of particle shape and size, as well as interparticle separation [21, 22]. Electron beam lithography (EBL) has become the method of choice for fabricating single or periodic arrays of metal nanostructures endowed with the desired morphological features [23]. It is versatile and provides high resolution and precise control over the geometry and separation of nanostructures, guaranteeing a fabrication reproducibility, and precision down to the nanometer scale.
Nanostructures for Photonics
Nanoaggregate Array The fabrication of an active SERS device was achieved by means of site selective deposition of Ag nanoparticles, exploiting electroless technique on Si pre-patterned substrates [23]. A large array of nanostructures with high degree of reproducibility at the nanometer scale was realized using electron beam lithography onto a Si substrate spin-coated with 50 nm thick ZEP layer. The Ag deposition was realized by means of a redox reaction between the metal and the Si surface remained uncovered after the lithographic process. The reaction occurs in a dilute (0.1–0.5 M) HF solution in which is dissolved a metal salt, that is, AgNO3, at a concentration of about 1 mM. The global chemical reaction is: 4Agþ þ SiðsÞ þ 6HF ! 4Ag0 þ H2 SiF6 þ 4Hþ (1) which in its turn can be separated into two half-cell reactions: Si þ 2H2 O ! SiO2 þ 4Hþ þ 4e ðAnode Si oxidationÞ
(2)
Agþ þ e ! Ag0 ðCatode Ag reductionÞ
(3)
The driving force of the whole process is coming from the potential difference in the two half-reactions, thus fostering the growth of Ag grains and the formation of well-organized metallic nanostructures (see Fig. 10a, b). In order to probe the enhancement behavior in Raman spectroscopy related to the plasmon excitation, the sample was immersed in Rhodamine 6G (R6G) solutions with different concentration (105 M and 1020 M in H2O) for 30 min. The sample was gently rinsed in DI-water to remove molecules in excess not chemisorbed on the metal surface and then dried in N2 flow. In Fig. 10c, Raman spectra of R6G molecules, adsorbed on a Si substrate with a 60 s Ag deposition, are shown. There are various bands for R6G compound at around 1,650, 1,509, and 1,361 cm1, attributed to the xanthenes ring stretching of C-C vibrations, 1,575 cm1 attributed to C-O stretching; while the band at 1,183 cm1 is related to the C-H bending and N-H bending vibration of xanthenes ring. As can be clearly observed in Fig. 10c, the Raman intensity
Nanostructures for Photonics
N
8000 6000
R6G (10−5 M)
4000 Intensity (a.u.)
Nanostructures for Photonics, Fig. 10 SEM micrographs of (a) array of empty holes fabricated by EBL technique and (b) Ag nanoaggregates selectively grown inside the nanohole pattern. (c) SERS spectra acquired from 105 to 1020 M R6G, absorbed on the silver structures
1821
2000 0 800 600
R6G (10−20 M)
400 200 0 400
600
800 1000 1200 1400 1600 1800 2000 Raman shift (cm−1)
N
Nanostructures for Photonics, Fig. 11 (a) The nanoantenna concept: fabrication of a highly sensitive and specific nanobiosensor based on extraordinary optical signal enhancement. (b, d) Representative SEM images of two different
nanoantenna arrays fabricated by EBL technique. The dimensions of the NAs are of length 410 nm (b), 200 nm (d) while width and height both are 60 nm for all NA. (c) SEM of coupled nanoantenna array
decreases with the decreasing concentration of R6G, as expected. Though, the concentration is very low, various bands are clearly visible even for R6G molecule with concentration of 1020 M, keeping accumulation time 60 s and laser power 0.0018 mW.
overview of different nanoantenna patterns are reported in Fig. 11. The sample topography has been characterized recurring to scanning electron microscopy (SEM) and atomic force microscopy (AFM) operated in tapping mode. The optical properties of the nanoantenna array, shown in Fig. 12c, were investigated by means of spectroscopic transmission of polarized light in the range between 400 and 900 nm. The polarization of the incident light was varied from transverse magnetic TM (electric field perpendicular to the short NA axis) to transverse electric TE (electric field parallel to the long NA axis). The optical transmittance spectra present evidence of a clear anisotropic behavior; for TM
Nanoantenna Array The ability to tailor the optical properties of structured plasmonic materials has proven to be crucial for a wide variety of applications, using them for surfaceenhanced spectroscopy and biosensing. While a symmetrical geometry allows polarization independent plasmon excitation, a dichroic absorption is well expected for nanoantenna (NA) fabrication. Large scan
1822
a
b
30000
CV NA
15000
60 nm
0 1000
1 μm
CV Au 500 0 200 NA 100 0 200
400
600
800 1000 1200
Raman shift (cm−1)
polarization a localized minimum around 620 nm is found. This is the typical behavior exhibited by subwavelength metal nanoparticles sustaining a localized surface plasmon resonance. Evidence of plasmon-induced SERS effects is given in Fig. 12a in which the different Raman emission lines from cresyl violet (CV) molecules (with a concentration of 3.46 mM) deposited on the gold nanoantenna arrays are shown. The sample was excited by 633 nm laser wavelength (laser power ¼ 0.14 mW and accumulation time ¼ 50 s) through a 150X objective. The measurements were performed on bare nanoantenna arrays (sample NA), on cresyl violet molecules chemisorbed on flat Au film (sample CV Au) and on CV deposited on the nanoantenna arrays (sample CV NA). To ensure the quality of the sample surface, the background measurements were performed at different positions of the nanoantenna SERS device. As shown in Fig. 12a (NA line) no characteristic Raman band, except at around 320 cm–1 related to the Ca-F vibration from the substrate, is observed. The characteristic vibrational bands of CV are observed in the SERS spectrum (CV NA trace). Intense Raman bands centered at around 591, 882, 927, and 1,189 cm1 can be attributed to the N-H2 rocking vibration, two benzene group bending, out-of phase N–H2 rocking vibration, and combination of N–H2 rocking and C–HX rocking, respectively. CV molecules chemisorbed over flat Au film are also shown in Fig. 12a. A giant enhancement in Raman signal for CV NA sample with respect to the CV Au
c Transmittance (%)
Nanostructures for Photonics, Fig. 12 (a) SERS spectra of cresyl violet molecules deposited on NA sample (blue line) and flat Au film (red line). Background measurement performed on bare NA structure is also reported (black line). (b) AFM 3D image. (c) Transmission optical spectra of the NA array of Fig. 1.11b. Red/black lines are for TM/TE polarization respectively
Nanostructures for Photonics
Intensity (a. u.)
N
0 nm 100 95 90 TM TE
85 450
600
750
900
Wavelength (nm)
one can be clearly observed. The maximum enhancement of the Raman emission is obtained when the excitation is polarized along to the nanoantenna short axis, in correspondence to the resonant excitation of the LSP resonance. The evaluated SERS enhancement for the fabricated nanoantenna device is 107.
Novel Nanophotonic Devices The three standard aspects of a photonic device, namely light generation, guiding, and harvesting can be combined in a number of ways to obtain many kinds of functionalities. For example, metallic-nanorods can be used for shrinking the wavelength of the light to improve the coupling with detectors such as nanoantenna arrays, or optical fiber can be properly modified for three-dimensional optical trapping [24]. Here will be introduced two specific examples: photonic crystals combined to plasmonic devices for the realization of background-free Raman probes, and superhydrophobic devices for single-molecule detection. Integration of Photonic Crystals and Plasmonic Devices for Label-free Chemical and Structural Detection A combination of different nanostructures for photonics give rise to a novel photonic-plasmonic device that allows label-free detection, through Raman spectroscopy, of single inorganic nanoparticles and of monolayers of organic compounds [25] and that allows,
Nanostructures for Photonics Nanostructures for Photonics, Fig. 13 (a) SEM image of the device fabricated on a silicon nitride AFM cantilever. (c) Zoom on the photonic crystal cavity and the tapered plasmonic waveguide with radius of curvature at the apex of 5 nm. (b) Sketch of the experimental setup, showing the integration of the photonic–plasmonic device into an AFM–Raman microscope
1823
a
b
Infrared laser for AFM detection
5 μm
N
532-nm incoming laser
QPD
Condenser
Modified AFM cantilever Sample on scanning stage
c
¥63 1.4 NA
1 μm
integrated on an atomic force microscope (AFM) cantilever, to obtain topographic, chemical, and structural information at an unprecedently high spatial resolution (below 10 nm) [26]. The device consists of a two-dimensional dielectric photonic crystal cavity, which can be patterned on a Si3N4 AFM cantilever, together with a tapered silver plasmonic waveguide placed at the center of the cavity (Fig. 13a). The photonic crystal consisted of a triangular lattice of air holes (lattice constant ¼ 250 nm, hole diameter ¼ 160 nm) patterned on a 100-nm-thick Si3N4 membrane. Three missing holes in the center generated a photonic crystal cavity, termed L3, tuned at l ¼ 532 nm. The plasmonic waveguide is a 2.5-mmtall metallic cone, with a base diameter of 300 nm, and the minimum radius of curvature at the apex is controllable in the range 2.5–5 nm (Fig. 1c). The entire device is fabricated by means of Focused Ion Beam milling and electron-beam induced deposition [27]. Ion beam milling was used to define the photonic crystal whereas the silver tapered waveguide was grown in the center of the cavity using electron-beam induced deposition from a gas precursor containing a platinum–carbon polymer (CH3)3Pt(CpCH3). The photonic crystal cavity enables an efficient coupling between the external laser source (used for Raman excitation) and the tapered waveguide, together with optimal sample illumination. In fact,
Collection objective (oil immersion)
Raman signal
the focal plane from which the SPPs are launched is 2.5 mm away from the cone apex where the Raman excitation is generated. This drastically lowers the background intensity at the sample plane. Thus, the spatial separation between SPP generation and Raman excitation efficiently increases the signal-to-noise ratio in the Raman spectrum, giving rise to a new optical configuration, namely Surface Plasmon Polariton Enhanced Spectroscopy (SPPERS). The focusing of SPPs follows an adiabatic compression mechanism, which causes a strong localization of the electrical field in a region with a size comparable to the apex radius of curvature [28]. This device can be fully integrated with an inverted Raman microscope combined with an AFM stage, as shown in Fig. 13b. The SPP spatial confinement in the near field is comparable to the radius of curvature of the tapered waveguide apex, which lies between 2.5 and 5 nm. The expected electric field enhancement E due to adiabatic compression is defined as E ¼ Etip/Ebase, where Etip and Ebase are the maximum electrical fields at the apex and bottom of the waveguide, respectively, and reaches values of practical interest in the range of 100 when the radius of curvature of the tip is below 5 nm. For successful spectroscopy applications, this necessitates state-of-the-art fabrication quality. The architecture of the device and the mechanical stiffness of the tapered waveguide allow using it both
N
N
1824
Nanostructures for Photonics
b
SiOx
Raman shift (cm−1)
Height (nm)
a 120 60 0 120
Si 90
x
30
an
sc
60
m
(n )
0
125
y
530 520 510 0 kcps
500
500
105
90
75
60 45 x scan (nm)
30
15
0
100
16
75
14
50 12
25 0
Intensity (kcps)
Height (nm)
0
375 250 ) m n ( scan
16 kcps
540
10 105
90
75
60 45 30 x scan (nm)
15
0
Nanostructures for Photonics, Fig. 14 (a) SEM image of the device fabricated on a silicon nitride AFM cantilever. (c) Zoom on the photonic crystal cavity and the tapered plasmonic waveguide with radius of curvature at the apex of 5 nm. (b) Sketch of
the experimental setup, showing the integration of the photonic– plasmonic device into an AFM–Raman microscope (With permission from Ref. [26])
as an AFM tip and as a nanometer light source for nearfield Raman excitation. In other words, it is possible to obtain topographic and chemical mapping to yield structural and stress information with the same spatial resolution as is imposed by the geometry of the tip, that is, limited only by its radius of curvature. Device performances have been assessed by measuring a test sample for which the chemical composition had periodic spatial variations in the range of few nanometers. The test sample consisted of patches and gratings of silicon nanocrystals having different pitches (from several micrometers to a submicrometer scale) obtained by laser melting of a silica substrate previously fabricated by plasma-enhanced chemical vapor deposition. The AFM–Raman setup was operated in tapping mode and wet conditions (in distilled water) to measure simultaneously the topography and Raman intensity map by scanning the sample and acquiring Raman data across the lithographic structures, point by point. A severe test on the described sample was performed by scanning the sample in a patterned region in which the crystalline and amorphous topography forms a grating with submicrometer pitch. From the AFM measurements, it could be seen that silicon,
upon laser-induced crystallization, has its height reduced by 100 nm, which results in the formation of a trench on the sample surface, as shown in the threedimensional AFM map in Fig. 14a. The AFM scan step size was 7 nm. The red line in Fig. 14a is a representative AFM line scan which provides simultaneously the Raman intensity map shown in Fig. 2b (upper panel), in which the Raman spectrum extends from 504 to 540 cm1. The acquisition time of each Raman map was 5 s. Fig. 14b (lower panel) shows the topography and Raman intensity for the same line scan. The Raman intensity is taken at the maximum of the silicon phonon peak at 520 cm1. The Raman intensity and relative peak amplitudes change from point to point, indicating an optical resolution of better than 7 nm. In conclusion, the development of a device based on the combination of photonic–plasmonic nanostructures that is fully compatible with AFM and Raman spectroscopy opens up a number of significant opportunities in the nanoscale chemical mapping of materials. The approach could be of particular benefit in the nanoscale analysis of biological matter, such as in the investigation of cell membrane protein spatial distributions, in which chemical/physical bond strengths and amino acid composition in vivo
Nanostructures for Photonics
1825
N
4
1
the solution gets concentrated over time
D.I. Solution containing biological molecules
2 the substrate is a superhydrophobic surface
5 upon evaporation the solute is confined to an infinitesimal area
3
nano plasmonic devices allow Raman measurements with unprecedented accuracy
intensity
the structure is hierarchical: at smaller scales nano plasmonic devices induce SERS effects
Raman shift
Nanostructures for Photonics, Fig. 15 Biophotonic nanostructures integrated with superhydrophobic surfaces can lead to the fabrication of novel devices with advanced sensing capabilities
conditions could be revealed. Device architectures based on the adiabatic generation of SPPs have considerable potential for applications in which a reliable nanometer-sized light source is of utmost importance. Integration of Biophotonic and Superhydrophobic Devices for the Detection of Low-concentrated Biomolecules Here, will be investigated how biophotonic nanostructures can be integrated to superhydrophobic surfaces (SHSs) to obtain devices with advanced sensing capabilities, thus offering new possibilities in the field of clinical medicine, especially toward the early diagnosis. Superhydrophobicity is a phenomenon whereby a drop post upon a surface would preserve its original spherical shape rather than spreading or wetting indefinitely the plane of contact (Fig. 15). In the celebrated model of Cassie [29], a surface would be superhydrophobic on account of the pockets of air that remain trapped between the liquid and the substrate, and the smaller the fraction of solid in contact with the drop the larger the apparent contact angle. SHSs retain unique properties in terms of wettability that can be reviewed as follows: (a) SHSs have superior adhesive properties, in the sense that they exhibit
vanishing friction coefficients; (b) a droplet, deposited upon these surfaces, would accordingly preserve a quasi-spherical shape while evaporates, and the contact area at the interface would thus progressively reduce; (c) SHSs can be artificially reproduced using micro and nanofabrication techniques. Using the properties above, micro textured surfaces may be successfully exploited to concentrate tiny amounts of moieties over micrometric areas, and consequently measure these moieties with unprecedented accuracy [30, 31]. SHSs typically comprise a periodic hexagonal lattice of cylindrical Si micro pillars with a certain diameter and pitch (Fig. 16a). The micro pillars are realized combining optical lithography, electroless growth, and Bosch Reactive Ion Etching (RIE) techniques. Nonconventional biophotonic nanostructures such as arrays of silver nanograin aggregates, conveniently positioned upon the pillars, would complete a hierarchical structure thus permitting the identification of substances in the single-molecule regime. These “two-stages” micro and nanostructures act as superhydrophobic surfaces with an increased contact angle ranging from 155 to about 175 . A particular kind of nanogeometry based plasmonic device is here reported as example, which is an electroless grown random assembly of silver nanograins
N
N
1826
Nanostructures for Photonics, Fig. 16 (a) SEM micrograph of periodic hexagonal lattice of cylindrical Si micro pillars used as SHS. (b) The spherical shape of a drop positioned upon the SHS is preserved. (c) Arrays of Ag nanograin aggregates positioned upon the pillars. (d) SEM image of the residual solute of R6G at the end of process of evaporation. The intensity of fluorescence is shown in (e), while Micro-Raman mapping measurements are reported in (f)
Nanostructures for Photonics
a
b
c
d
−80
e
750.0
f
675.0 600.0
Y axis (μm)
−40
525.0 450.0
0
375.0 300.0
40
225.0 150.0 75.00
80
0
120 −105 −65
[23]. Nevertheless, the method can be extended to a number of different plasmonic nanostructures, as those described in the sections above. Imagine to deposit a drop of an extremely diluted solution upon a textured, superhydrophobic substrate. The drop would evaporate over time and thus the solution would get more and more concentrated (Fig. 15). At the end of the process, the residual solute would be confined within an exceptionally small region of the plane. With a suitable design, few molecules may be conveniently enforced to confine into the smallest area conceivable, at the limit upon a sole
−25 15 55 X axis (μm)
95
135
pillar. Nanogeometry based biophotonic devices, conveniently tiling these surfaces, would probe/detect the moieties with extremely small resolution limits. Here, the beneficial effects of superhydrophobicity and nanogeometry based spectroscopy are combined and conveyed into a unique platform, and from the combination of the two novel properties arise permitting the identification of proteins or analytes with unprecedented accuracy. In order to clarify the concept, an experiment performed using R6G molecules, is reported. Evaporation processes of drops of D.I. water were followed
Nanostructures for Photonics
over time. Few molecules were conveniently enforced to confine into a small area. A solution was investigated with initial concentration as small as 1018 M. Figure 16d shows a SEM image of the residual solute of R6G at the end of process of evaporation. The intensity of fluorescence as in Fig. 16e is correctly proportional to the quantity of substance deposited upon the pillars, and thus the intensity signal of fluorescence follows the solute distribution upon the substrate. Micro-Raman mapping measurements were also performed (Fig. 16f). The mapping analysis was performed by referring the band centered at 1,650 cm1. The results demonstrate that the method is effective in measuring extremely small diluted solutions.
Cross-References ▶ Biosensors ▶ Electron Beam Lithography (EBL) ▶ Light Localization for Nano-Optical Devices ▶ Nanoparticles ▶ Nanophotonic Structures for Biosensing ▶ Optical Properties of Metal Nanoparticles ▶ Surface Plasmon-Polariton-Based Detectors
1827
7.
8.
9.
10.
11.
12.
13.
14.
15.
References 1. Carbone, L., Nobile, C., De Giorgi, M., Sala, F.D., Morello, G., Pompa, P., Hytch, M., Snoeck, E., Fiore, A., Franchini, I.R., Nadasan, M., Silvestre, A.F., Chiodo, L., Kudera, S., Cingolani, R., Krahne, R., Manna, L.: Synthesis and micrometer-scale assembly of colloidal Cdse/Cds nanorods prepared by a seeded growth approach. Nano Lett. 7, 2942 (2007) 2. Efros, A.L., Rosen, M., Kuno, M., Nirmal, M., Norris, D.J., Bawendi, M.: Band-edge exciton in quantum dots of semiconductors with a degenerate valence band: dark and bright exciton states. Phys. Rev. B-Condens. Mat. 54, 4843 (1996) 3. Zavelani-Rossi, M., Lupo, M.G., Krahne, R., Manna, L., Lanzani, G.: Lasing in self-assembled microcavities of Cdse/Cds Core/Shell colloidal quantum rods. Nanoscale 2, 931 (2010) 4. Krahne, R., Zavelani-Rossi, M., Lupo, M.G., Manna, L., Lanzani, G.: Amplified spontaneous emission from core and shell transitions in Cdse/Cds nanorods fabricated by seeded growth. Appl. Phys. Lett. 98, 063105 (2011) 5. Kazes, M., Lewis, D.Y., Ebenstein, Y., Mokari, T., Banin, U.: Lasing from semiconductor quantum rods in a cylindrical microcavity. Adv. Mater. 14, 317 (2002) 6. Martiradonna, L., Carbone, L., De Giorgi, M., Manna, L., Gigli, G., Cingolani, R., De Vittorio, M.: High Q-factor
16.
17.
18.
19.
20.
21.
22.
N
colloidal nanocrystal-based vertical microcavity by hot embossing technology. Appl. Phys. Lett. 88, 181108 (2006) Shoji, S., Proietti Zaccaria, R., Sun, H., Kawata, S.: Multistep multi-beam laser interference patterning of threedimensional photonic lattices. Opt. Express 14, 2039 (2006) Knight, J.C., Birks, T.A., Russell, P.St.J., Atkin, D.M.: Allsilica single-mode optical fiber with photonic crystal cladding. Opt. Lett. 21, 1547 (1996) Galli, M., Agio, M., Andreani, L.C., Atzeni, L., Bajoni, D., Guizzetti, G., Businaro, L., Di Fabrizio, E., Romanato, F., Passaseo, A.: Optical properties and photonic bands of GaAs photonic crystal waveguides with tilted square lattice. Eur. Phys. J. B 27, 79 (2002) Malvezzi, A.M., Vecchi, G., Patrini, M., Guizzetti, G., Andreani, L.C., Romanato, F., Businaro, L., Di Fabrizio, E., Passaseo, A., De Vittorio, M.: Resonant second-harmonic generation in a GaAs photonic crystal waveguide. Phys. Rev. B 68, 1613061 (2003) De Vittorio, M., Todaro, M.T., Stomeo, T., Cingolani, R., Cojoc, D., Di Fabrizio, E.: Two-dimensional photonic crystal waveguide obtained by e-beam direct writing of SU8-2000 photoresist. Microelectron. Eng. 73–74, 388 (2004) Proietti Zaccaria, R., Verma, P., Kawaguchi, S., Shoji, S., Kawata, S.: Manipulating full photonic band gaps in two dimensional birefringent photonic crystals. Opt. Express 16, 14812 (2008) Zhao, H., Proietti Zaccaria, R., Verma, P., Song, J., Sun, H.: Single-mode operation regime for 12-fold index-guiding quasicrystal optical fibers. Appl. Phys. B 100, 499 (2010) Zhao, H., Proietti Zaccaria, R., Song, J., Kawata, S., Sun, H.: Photonic quasicrystals exhibit zero-transmission regions due to translational arrangement of constituent parts. Phys. Rev. B 79, 115118 (2009) Zhao, H., Proietti Zaccaria, R., Verma, P., Song, J., Sun, H.: Validity of the V parameter for photonic quasi-crystal fibers. Opt. Lett. 35, 071064 (2010) Veselago, V.: The electrodynamics of substances with simultaneously negative values of e and m. Sov. Phys. Uspekhi 10, 509 (1968) Papasimakis, N., Luo, Z., Shen, Z.X., De Angelis, F., Di Fabrizio, E., Nikolaenko, A.E., Zheludev, N.I.: Graphene in a photonic metamaterial. Opt. Express 18, 8353 (2010) Nikolaenko, A.E., De Angelis, F., Boden, S.A., Papasimakis, N., Ashburn, P., Di Fabrizio, E., Zheludev, N.I.: Carbon nanotubes in a photonic metamaterial. Phys. Rev. Lett. 104, 153902 (2010) Sa´mson, Z.L., MacDonald, K.F., De Angelis, F., Gholipour, B., Knight, K., Huang, C.C., Di Fabrizio, E., Hewak, D.W., Zheludev, N.I.: Metamaterial electro-optic switch of nanoscale thickness. Appl. Phys. Lett. 96, 143105 (2010) Mary, A., Rodrigo, S.G., Garcia-Vidal, F.J., Martin-Moreno, L.: Theory of negative-refractive-index response of doublefishnet structures. Phys. Rev. Lett. 101, 103902 (2008) Pe´renne`s, F., Marmiroli, B., Matteucci, M., Tormen, M., Vaccari, L., Di Fabrizio, E.: Sharp beveled tip hollow microneedle arrays fabricated by LIGA and 3D soft lithography with polyvinyl alcohol. J. Micromech. Microeng. 16, 473 (2006) Tormen, M., Businaro, L., Altissimo, M., Romanato, F., Cabrini, S., Perennes, F., Proietti, R., Sun, H.-B., Kawata, S., Di Fabrizio, E.: 3D patterning by means of nanoimprinting.
N
N 23.
24.
25.
26.
27.
28. 29. 30.
31.
1828
Nanostructures for Surface Functionalization and Surface Properties
X-ray and two-photon lithography. Microelectron. Eng. 73–74, 535 (2004) Das, G., Mecarini, F., Gentile, F., De Angelis, F., Kumar, M.H.G., Candeloro, P., Liberale, C., Cuda, G., Di Fabrizio, E.: Nano-patterned SERS substrate: application for protein analysis vs. temperature. Biosen. Bioelect. 24, 1693 (2009) Liberale, C., Minzioni, P., Bragheri, F., De Angelis, F., Di Fabrizio, E., Cristiani, I.: Miniaturized all-fibre probe for three dimensional optical trapping and manipulation. Nat. Photonics 1, 723 (2007) De Angelis, F., Patrini, M., Das, G., Maksymov, I., Galli, M., Businaro, L., Andreani, L.C., Di Fabrizio, E.: A hybrid plasmonic photonic nanodevice for label-free detection of a few molecules. Nano Lett. 8, 2321 (2008) De Angelis, F., Das, G., Candeloro, P., Patrini, M., Galli, M., Bek, A., Lazzarino, M., Maksymov, I., Liberale, C., Andreani, L.C., Di Fabrizio, E.: Nanoscale chemical mapping using three-dimensional adiabatic compression of surface plasmon polaritons. Nat. Nanotech. 5, 67 (2010) Cabrini, S., Carpentiero, A., Kumar, R., Businaro, L., Candeloro, P., Prasciolu, M., Gosparini, A., Andreani, C., De Vittorio, M., Stomeo, T., Di Fabrizio, E.: Focused ion beam lithography for two dimensional array structures for photonic applications. Microelectron. Eng. 2005(78–79), 11 (2005) Stockman, M.I.: Nanofocusing of optical energy in tapered plasmonic waveguides. Phys. Rev. Lett. 93, 137404 (2004) Lafuma, A., Que´re´, D.: Superhydrophobic states. Nat. Mater. 2, 457 (2003) Accardo, A., Gentile, F., Mecarini, F., De Angelis, F., Burghammer, M., Di Fabrizio, E., Riekel, C.: In situ x-ray scattering studies of protein solution droplets drying on micro- and nanopatterned superhydrophobic PMMA surfaces. Langmuir 26, 15057 (2010) De Angelis, F., Gentile, F., Mecarini, F., Das, G., Moretti, M., Candeloro, P., Coluccio, M.L., Cojoc, G., Accardo, A., Liberale, C., Zaccaria, R.P., Perozziello, G., Tirinato, L., Toma, A., Cuda, G., Cingolani, R., Di Fabrizio, E.: Breaking the diffusion limit with super hydrophobic delivery of few molecules to plasmonic nanofocusing structures. Nat. Photon. 5, 682 (2011)
Nanostructures for Surface Functionalization and Surface Properties Paola Rivolo Dipartimento di Scienza dei Materiali e Ingegneria Chimica – Politecnico di Torino, Torino, Italy
Synonyms Chemical modification; Functionalization; Organosilanes; Patterning; Physical modification; Plasma grafting; Plasma polymerization; Selfassembled monolayers; Surface properties; Thiols
Definition Modification and functionalization of substrates can lead to the presence at materials surfaces of molecular nanostructures. Their 2D packing and chemical functionalities may impart to surfaces particular properties that can be very different from the pristine ones. Modification is a very general term concerning every kind of treatment leading to change the physical morphology or surface chemistry of a given material in order to obtain tailored properties with dependence on the material application field. Functionalization is mainly related to the introduction at a surface of chemical groups with a variable surface density which are requested to subsequently react with other species. When a process or a group of processes are applied to the surface of an inorganic or organic material, affecting both topography/morphology and surface chemistry at the micro- and/or nanoscale level, the material surface results physically and chemically patterned. In the following the main modification, functionalization, and patterning surface techniques will be described according to process methods, substrate chemistry, and nanotechnological application fields.
Surface Modification and Functionalization In general, some materials which have excellent bulk physical and chemical properties often do not possess suitable surface properties required for specific applications. For this reason, surface-modification techniques, that can transform these materials into valuable finished products, become an important part of surface science and technology. Processes, developed in the last decade, are able to simply modify surface morphology of materials or give rise to nanostructured coatings aimed to simply affect the adhesion or anti-adhesion properties, mainly hydrophilicity and hydrophobicity, or to promote specific bonding with other species. These features can be a crucial advantage in several applications in the fields of protective, anti-adhesion, adhesion coatings, friction and wear, composites, microelectronics and thinfilm technology, molding and lithography, technical fabrics, biomaterials, sensing and in particular biosensing.
Nanostructures for Surface Functionalization and Surface Properties
Generally, these surface-modification methods can be divided into two classes: physical modifications and chemical modifications [1]. In this entry will be reported the main topics regarding surface chemical modifications, both the surface derivatization or grafting method which depend on the chemistry of the substrate surface and the functional organic coating deposition which can be performed on the surface of very different materials without the need of a specific chemical binding with the surface (no-dependence on chemistry). Finally, some surface patterning procedures devoted to get functional structures will be described as following process of the previously reported methods or based on other modification techniques including physical processes.
= Si-Alkoxy, Si-Alogenide, SH = body chain = CH3, OH, COOH, NH2, SO3H
tail
tail
Inter-chain Van der Waals Interactions
head
Chemical Derivatization or Grafting The key advantage of these techniques is that the surface of the materials can be modified or tailored to acquire very distinctive properties through the choice of different grafting precursors, while maintaining the substrate properties. The crucial point is to single out the suitable molecule to be grafted according to the chemistry of the material substrate, that is, the kind of atoms/groups with which the substrate is truncated with. Before the grafting step, commonly a surface activation is needed in order to create reactive sites on the substrate surface or enhance the amount of the ones naturally already present. Practically, one can generate reactive groups through UV, high-energy electrons, g irradiation, plasma treatment, ozone exposure [1] and chemical reactions performed in liquid and vapor phase. The kind of activation treatment heavily depends on materials. For sake of simplicity, the different types of derivatizing treatments and the related materials activation procedures will be considered according to substrate chemistry. Generally, among the surface-modification techniques, the formation of a Self-Assembled Monolayer (SAM) has been intensively studied. SAM is a kind of “bottom-up” technique which provides advantages regarding the surface modification. It leads to a structure which is at, or close to, thermodynamic equilibrium, and thus tends to self-healing/defect rejection and leads to a closely packed, well-ordered, and stable configuration on the surface [1]. SAMs offer
N
1829
tail
Inter-chain Van der Waals Interactions
head
head
SUBSTRATE
Nanostructures for Surface Functionalization and Surface Properties, Fig. 1 Scheme of a general SAM formed on the surface of a substrate
unique opportunities to increase fundamental understanding of self-organization, structure-property relationships, and interfacial phenomena. The ability to tailor both head and tail groups of the constituent molecules makes SAMs excellent systems for a more fundamental understanding of phenomena (e.g., ordering and growth, wetting, adhesion, lubrication, corrosion, etc.) affected by competing intermolecular, molecule-substrate and molecule-solvent interactions. SAMs provide the needed design flexibility, both at the individual molecular and at the material levels, and offer a vehicle for investigation of specific interactions at interfaces and of the effect of increasing molecular complexity on the structure and stability of twodimensional assemblies. The key factors characterizing the SAM precursor molecules consist in (a) the “head,” (b) the “tail,” and (c) the body chain length of the molecule (Fig. 1). The “head” is one of the terminal group of these normally linear species and has the role to anchor the surface by a favored and fast reaction producing the formation of a usually stable (covalent) bond linking the surface. The choice of this termination is driven by the chemistry of the material substrate. The “tail” of the molecule, that is, the termination opposite to the “head,” is deputed to perform the
N
N
1830
Nanostructures for Surface Functionalization and Surface Properties
desired effect at the SAM surface. So as a rule of thumb, it can be a methyl termination (CH3), a perfluoroalkyl termination that generate at the most a hydrophobic or super-hydrophobic and repulsive behavior or a functional group such as –SH, –OH, –NH2, –COOH able to carry out a subsequent reaction with other molecule (for example, biomolecules, polymers, suitable linker molecules) or simply express hydrophilic and adhesion properties. The molecule length (the hydrocarbons chains can range, for instance, from C3 to C24) will strongly depend on the application for which SAM structures are needed. Some reviews on the self-assembled monolayers have covered the SAM formation and structure their characterization by electrochemical and scanning tunneling microscopy [1], by PM-IRRAS vibrational spectroscopy techniques and XPS analysis [2], their application to biosensors [1], anti-stiction in MEMS devices and in nano-imprinting lithography techniques [3], and their advantages in providing controlled surface properties and unique reactions [1]. In particular, such monolayers have been generated mostly on metal surfaces, such as Au, Ag, and Cu, using the thiol chemistry [1]. Nuzzo and Allara, reported in 1983, [4] the procedure of bifunctional organic sulfides species self-assembly on the surface of Au (111) freshly evaporated substrates in the liquid phase by simply putting in contact the samples with a diluted (milliMolar range) solution of the reagent, for some hours. Successively many authors observed with other n-alkyl thiolates that in the first few seconds the species are able to bind the gold surface and in more than 1 h, they self-assemble into a compact, well-ordered flat intercrossed molecular monolayer. The –SH head is able to establish a strong Au–S bond and the chain-to-chain interaction performed by the adjacent grafted molecules is dominated by weak dispersion forces if nonfunctional interacting groups are present in the middle of the aliphatic chain. In this case, H-bond or other stronger van der Waals forces can be exerted. The interchain forces are able to produce, as calculated by Godin [5] an equilibrium separation between two alkyl chains of 0.44 nm. The alkanethiol SAM is formed as a densely packed structure showing (√3 √3)R30 overlayer on gold surface where the molecule is tilted about 30 from the surface normal with the sulfur atom chemically bonded to gold atoms. This implies that a compressive surface stress has to be
Terphenyldithiol (P3)
AcS
Quarterphenyldithiol (P4)
AcS
Fluorenedithiol (F1)
AcS
Terthiophenedithiol (T3)
AcS
Quarterthiophenedithiol (T4) AcS
SAc SAc SAc
S
S
S
S S
SAc S
S
SAc
Nanostructures for Surface Functionalization and Surface Properties, Fig. 2 Schematic structures of conjugated molecules studied in [7]
generated during the self-assembly, and this stress will increase with increasing alkyl chain length. Patel et al. [6] studied the effect of chain length on the reactivity of SAMs toward protein immobilization generating SAMs from short- and long-chain carboxylic acid terminating alkyl-thiols, 11-MercaptoUndecanoic Acid (11-MUA), 3-MercaptoPropanoic Acid (3-MPA) and from the mixture, 1:10, of the both. The accessibility of the carboxylic groups by the amino group of a lysine termination exposed by the protein, catalase, to be anchored, decreases in the order Mixed > 11-MUA > 3-MPA. The experiment was performed by previously activating the carboxylic groups by a water soluble 1-ethyl-3-[3-(dimethylamino) propyl]carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS). Successful formation of the NHS ester intermediate is reliant on the accessibility of the terminal carboxylate groups. It was assessed that a steric packing of these acid groups can limit the rate of intermediate formation with full conversion of accessible acid groups only occurring after several repeated reaction cycles. The co-adsorption of different alkyl-thiols provides the system a degree of disorder which is strategic for an efficient protein immobilization. SAMs of conjugate alkyl-dithiols are one of the more interesting solutions focused by research in molecular electronics which is open to the possibility of substituting inorganic materials with organic molecules in the core regions of devices. Jiang [7] reported STM characterization of the electronic transport properties of SAM of containing double bonds (insaturation) in the body chain such as the aromatic dithiols (Fig. 2). Co-assembly of conjugated dithiols and alkanemonothiols is suggested as a route to obtain dithiols that are denser packed and more vertical (with respect to what is obtained from pure dithiol routes). This surface
Nanostructures for Surface Functionalization and Surface Properties
N
1831
Nanostructures for Surface Functionalization and Surface Properties, Fig. 3 (a) STM image (95 95 nm) of octanethiol SAMs on Au(111); (b) STM image (70 70 nm) of decanedithiol SAMs on Au(111); (c) STM image (75 75 nm) of the mixture SAMs on Au(111)
condition is suitable to the evaporation on the top formed SAM of an Au layer and to the formation of other Au–S bonds with the free thiol end of the dithiol molecules. Kobayashi [8] reported the study by Ultra High Vacuum Scanning Tunneling Microscopy (UHVSTM) of alkyl-thiol and alkyl dithiol SAMs on Au surfaces, characterizing the morphology of resulting nanostructures and the strength of molecules interaction with the surface. Figure 3 shows the different appearance of the surface according to the nature mono- or dithiolic of the molecule used to form the SAMs both by a liquid process in ethanolic solution (24 h) and by a vapor functionalization process (from a 10 mM ethanolic solution, for 2 h in a closed vessel). In (a) the 1,8-octanethiol SAM results in nanometric islands where molecules assume a surface-normal arrangements. In (b) the nanometric stripe structures are evident due to the molecules of dithiols (1,6hexanedithiol, 1,10-decanedithiol) lying with their axes parallel to the surface (the bright spots are assigned as sulfur atoms). The different arrangement between (a) and (b) is probably ascribed to the stronger interaction between both thiol groups at the molecular ends with the gold atoms with respect to the case of alkanethiol SAMs. The parallel arrangement is more preferred than the normal arrangement in the case of alkanedithiol SAMs. In (c) the SAM is obtained by a mixture of the vapor of the ethanol solution which contained octanethiol and octanedithiol molecules at 1 mM each. The UHV-STM image shows stripe phase regions corresponding to the areas where alkanedithiol SAMs are formed. On the other hand, the island regions may be attributed to alkanethiol SAMs. Ag (Silver) and other materials such as GaAs (Gallium Arsenide) can be functionalized by alkyl-thiols. Dubowski et al. [9] recently studied the formation mechanism of T-Alkanethiols (where T¼NH2,
SH
S
alkanethiol
S
S
dialkyl disulfide
dialkyl sulfide
O S
N
N S−
alkyl xanthate
S
S−
dialkylthiocarbamate
Nanostructures for Surface Functionalization and Surface Properties, Fig. 4 Surface-active organosulfur compounds that form monolayers on metals and semiconductors [10]
COOH, CH3) SAMs on III-V semiconductors surfaces, in particular GaAs (001). They proposed the formation at the interface of a S–As or S–Ga bond according to which of the two elements is the richer in the exposed surface. They demonstrated the dependence of PhotoLuminescence signal on the terminal T group and the possibility to use these functionalized semiconducting materials in biosensing by PL technique. In 1996 Ulman [10] reported other organosulfur molecules (Fig. 4) in addition to alkane thiols able to strongly graft and self-assemble on semiconductors surface, such as GaAs and InP, and other metals such as copper, platinum, mercury, iron, nanosize g-Fe2O3 particles.
N
1832
Nanostructures for Surface Functionalization and Surface Properties
Nanostructures for Surface Functionalization and Surface Properties, Fig. 5 (a) Different diphosphonic acids used in self-assembled multilayer preparation. (b) A schematic description of fatty acid monolayers on AgO and on Al2O3
PO32−
a PO32−
PO22−
S
PO32−
S
N
S
I
OH
S
PO32−
PO32−
HO
N
N
N
N
PO32−
PO32−
PO32−
PO32−
II
III
IV
V
N
PO32− VI
b
O
O− O
O− O AgO
In Fig. 5a are displayed the diphosphates molecules used for Zirconium (Zr4+) ions exploiting their ability to form highly insoluble salts with tetravalent transition metal ions. At last, fatty acids, which are longchain n-alkanoic acids (CnH2n+1COOH), are known, by an acid-base reaction, to form a surface ordered salt between the carboxylate anion and a surface metal cation, for example, with AgO and Al2O3 (Fig. 5b). For inorganic oxides, the typical molecules used to form SAMs monolayers are organosilanes. They are characterized by a variability of possible structures and by the easiness of reaction performed both in the liquid and in the vapor phase. In particular, Silicon/Silica (Si/SiO2), glass and quartz surfaces, Aluminum/Alumina (Al/Al2O3), phyllosilicates such as mica, magnetic materials such as maghemite (g-Fe2O3) nanoparticles, germanium oxide [1], and Zinc oxide (ZnO) [11] exposing
O− O
O−
O
O O−
O O−
O O−
O O−
O−
Al2O3
naturally hydroxyl-terminated (OH) surfaces or zinc selenide [1] and Gallium Nitride (GaN) Aluminum Nitride (AlN) and Silicon Nitride (SiN) [12–14], where the –OH species can be easily created by a pretreatment, react with alkylchlorosilanes and alkylalkoxysilanes. Stable siloxy linkages [2, 3] are formed between hydroxylated Silica and 3-aminopropyltriethoxysilane (3-APTES). The reaction proceeds through the hydrolysis of silane alkoxy terminations and the condensation with surface hydroxyls and elimination of an alcohol molecule. Further condensation between adjacent molecules already grafted to the surfaces take part stabilizing the SAM through the formation of siloxane bridges (Si–O–Si). Glass and Si/SiO2 are the most common inorganic materials used in many technological applications as optical and electrical materials and as MEMS (MicroElectroMechanical Systems). One major factor
Nanostructures for Surface Functionalization and Surface Properties
that limits the widespread use and reliability of MEMS is adhesion. Adhesion is a result of the dominance of surface forces, such as capillary, hydrogen bonding, electrostatic, and van der Waals forces, over body forces at the microscale. When internal restoring forces of microstructures cannot overcome surface adhesive forces, the devices are said to suffer from stiction. Many procedures based on organosilanes SAMs formation are reported to impart to MEMS anti-stiction properties. Comparing the SAM formation from dimethyldichlorosilane ((CH3)2SiCl2, DDMS) and tridecafluoro-1,1,2,2-tetrahydrooctyltrichlorosilane (CF3(CF2)5(CH2)2SiCl3, FOTS) [3], by a liquid (thermal annealing about 65 C in anhydrous toluene solution) and a vapor procedure (dosage of organosilane vapor pressure on an Oxygen plasma pre-activated surface in a low pressure CVD style reactor), it results that the vapor phase method avoid vertical polymerization of precursor molecule producing a flatter SAM with the same efficiency in anti-stiction properties. 3-APTES has been shown to be an efficient SAM precursor [2, 15] for biosensing of protein tumoral markers based on micro-cantilever devices. Both the functionalization routes described above are efficient in reducing the mass effect of the functional coating which is detrimental for the device sensitivity and in providing a maximum coverage in the order of 1014 –NH2 groups/cm2 [2]. In the field of magneto-chemotherapeutics, where spatial control of conjugated active biomolecules under magnetic field and Magnetic Resonance Imaging (MRI) contrast can be achieved simultaneously are becoming always more relevant functionalized magnetic nanoparticles (NP) able to conjugate specific biomolecules, for instance, Matrix MetalloProteinase (MMP), zinc-containing endopeptidases, which are responsible for the regulated degradation and processing of extracellular matrices (ECM), with physiological functions such as wound healing, tissue remodeling, tumor growth, and metastasis [16]. The silanization of the NP was conducted under anoxic (N2 purging) suspension consisting of glycerol, aqueous methanol, and acetic acid. Mercaptopropyltrimethoxysilane (MPTMS) was added slowly to the suspension to ensure homogeneous hydrolysis of the silane agent, followed by reflux heating for 16 h. It is demonstrated that by this procedure it is possible not only to get at the NP surface a –SH terminated SAM able to react
1833
N
with MMP by the cross-linker Sulfosuccinimidyl-4-(Nmaleimidomethyl)cyclohexane-1-carboxylate (SulfoSMCC), but also to reduce the iron oxide phase from g-Fe2O3 to the higher magnetic moment phase of Fe3O4. Other methods to functionalized Iron Oxide NP are reported by Wu et al. in [17]. For instance, alkyl phosphonates and phosphates are used which bind efficiently to iron oxide particle surfaces, and favor, due to their good biocompatibility, the utilization of encapsulated magnetic NPs in nanomedicines. The no-oxide surfaces, such as nitrides, can be easily hydroxylated in order to exploit organosilanes chemistry in order to achieve surface functionalization. The well-assessed procedure consists in the substrate immersion in H2SO4:H2O2 (3:1) solution, also called Piranha solution, for 20 min, rinsed with deionized water and dried under Nitrogen flux prior to silanization. The highly oxidizing reagents produce, for GaN (deposited by MetalOrganic Chemical Vapor Deposition – MOCVD), an oxide layer exposing hydroxyl species. For AlN (also obtained by MOCVD) which is already covered by a native oxide layer, the only effect is surface hydroxylation [12]. The Piranha solution is used also on crystalline Silicon and on SiO2 thermally grown from crystalline Silicon. On SiN (deposited by Plasma Enhanced Chemical Vapor Deposition, PECVD), a solution with 10% nitric acid (HNO3) at 80 C for 20 min [13] or a mixture of CH3COOH and H2O2 for 3 h [14] is used to activate the silanols (Si–OH); for the both water is used for rinsing and N2 for drying. Also polymers can be activated in order to react with organosilane “head” (trialkoxy- or trichloroterminations) as suggested by K€uhn et al. in [18]. The double procedure consists in an Oxygen plasma step in order to produce mixed oxygen containing species OH, C¼O, C–O–C, CHO, COOH, C–O–OH, CO3, C¼C, etc., on a polyolefin polymer such as PolyEthylene (PE) or PolyPropylene (PP) and then a reduction step in diborane (gaseous procedure) or LiAlH4 solution in order to obtain a monotype homofunctional surface exposing hydroxyls. Since long time, Silicon surface reconstruction has been studied in order to understand the most suitable Silicon face (100) and (111) to get an organic functionalization avoiding detrimental side oxidation effects. Bent [19] reported some example of Silicon hydrosilylation by alkenes or alkines, molecules
N
N
1834
Nanostructures for Surface Functionalization and Surface Properties, Fig. 6 Proposed mechanism for surface hydrosilylation. Initial loss of silicon hydride generates a silicon dangling bond. Reaction between the silicon and an alkene molecule leads to an attached alkyl radical, which may abstract a hydrogen atom from a neighboring silicon
Nanostructures for Surface Functionalization and Surface Properties R H
C H
C H
H Si
Si
R H H
C C
H
Si
H Si
R H H
C C Si
H Si
bearing C¼C terminal groups that react with the Si–H of the Si surface under different activation methods. A radical initiator, a species used to create free radicals in reactions, starts the reaction by abstracting H from a surface Si–H bond, producing a silicon dangling bond (which is also referred to as a silicon radical). The mechanism is illustrated in Fig. 6 on a (111) silicon surface. The dangling bond is represented as an orbital with a dot, where the dot represents a single, unpaired electron. The silicon dangling bond then reacts with the alkene molecule, forming a new Si–C bond at the surface and leading to the attachment of the organic molecule, which itself is left with a carbon radical. This surface bound organic radical is thought to abstract a hydrogen atom either from an unreacted alkene molecule or from a neighboring Si–H group on the surface. The net reaction sequence produces a stable, closed-shell organic group bound directly to the Si (111) surface. Hydrosilylation can be achieved also under UV irradiation in the range 200–400 nm, thermal activation at T > 150 C (but a low quality of resulting organic monolayers has been noticed), introducing in the reactant mixture acidic molecules (called Lewis acids) to catalyze the reaction. The main applications of organic layers on Silicon are related to protection of the underlying substrate from attack by etchants, attribution of chiral properties to the surface, addition of new optical properties to
silicon and recently the substitution of organosilanes in sensing technology when the presence of oxide has to be avoided, also if the organic monolayers lacks in terms of order and cohesion, typical of SAMs. Silicon/Silica and Diamond surface can be also directly derivatized through polymers grafting. On their own, polymers expose functionalities which are able to give further reactions. Surface graft polymerization is a chemical modification method. In surface graft polymerization, the modification is achieved by grafting suitable macromolecular chains on the surface of materials through covalent bonding. The key advantage of these techniques is that the surface of the materials can be modified or tailored to acquire very distinctive properties through the choice of different grafting monomers, while maintaining the substrate properties. It also ensures an easy and controllable introduction of graft chains with a high density and exact localization onto the surface. Compared with the physically coated polymer chains, the covalent attachment of the grafted chains onto a material surface avoids their desorption and maintains a long-term chemical stability of the introduced chains. Chiari et al. [20] used a ter-polymer of N,Ndimethylacrylamide, N-acryloyloxysuccinimide, and 3-(trimethoxysilyl)propyl methacrylate. The copolymer (DMA-NAS-MAPS) graft the SiO2 surface of oxidized silicon or glass by a simple and robust procedure consisting in the dip coating in an aqueous solution, of substrates of different shapes in a short time span (less than 1 h) and in mild conditions without harmful treatments with harsh washing solutions. The film formed on the surface is thin, on the order of a few nanometers. For fluorescent microarray chips, this is a suitable coating as it does not affect the fluorescence intensification provided by the silicon/silicon oxide substrate. The trimethoxysilyl terminations react with surface silanols, as well as an organosilane, the acryloyloxysuccinimidyl ends give rise to covalent bonds with amino groups of biomolecules and the DMA units form the structural backbone of the polymer. Recently, Oxidized Ultrananocrystalline diamond (UNCD) has been successfully coated with poly (styrene) (PS) using the self-initiated photografting and photopolymerization (SIPGP) approach with styrene as the monomer, within 16 h, under constant irradiation with UV light (lmax ¼ 350 nm) [21]. The
Nanostructures for Surface Functionalization and Surface Properties
obtained coating stable grafted to surface is then easily derivatized by sulphonation, nitration, and amidoalkylation carried out according to well-known liquid chemistry routes and so allowing the incorporation of additional functional molecules, such as fluorescent labels. This will open the way for designing more advanced biosensing schemes, incorporating multifunctional elements and with a higher loading capacity for biomolecules. Ruckenstein et al. reported the graft polymerization reaction on already SAMs coated surfaces of glass and silicon wafers in order to introduce additional reactive sites. The method involves the aniline substitution at Br-terminated SAM through the aniline amino termination and further the surface oxidative graft polymerization of aniline on the modified glass surface via the covalently immobilized aniline sites. A PolyAniline (PANI) coating is so obtained, performing the reaction in the polymerization solution containing aniline, and due to its well-known conductivity (conductive polymer) this can be a suitable substrate to be used as biosensor and biomaterial. Another route to further derivatize an organosilane surface for biosensing purposes consists in the grafting, from a 8.5 pH solution, of homofunctional linkers such as Glutaraldehyde (OHC–CH2–CH2– CH2–CHO) which is able through an aldehyde end to react with NH2 exposed at amino-SAM modified surface and, then, through the second aldehyde termination to immobilize biomolecules exposing amino terminations. This method has been successfully applied by Ricciardi et al. [15] for functionalization of cantilever based biosensors aimed to the detection of tumoral markers (Angiopoietin). Surface Modification: No-Dependence on Materials Chemistry Among the surface-modification techniques, deposition of coating obtained by plasma polymerization presents, respect to surface derivatization discussed in the previous paragraph, the advantage to allow the tailoring of properties of any kind of surface without involving the substrate surface chemistry in the functionalization mechanism. Macromolecular plasma chemistry [22] has developed in the last two decades in the following directions: plasma-enhanced synthesis (deposition and/or grafting) of thin layer macromolecular structures;
1835
N
surface functionalization of polymeric materials; and etching of inorganic or polymeric substrate surfaces. Plasma-enhanced synthesis involves the dissociation of starting materials and reorganization of the resulting neutral and charged molecular fragments into macromolecular structures on the surfaces located inside or outside of the plasma zone. Outside the plasma zone, recombination mechanisms usually result in the incorporation of molecular fragments with lower degrees of dissociation into the nascent macromolecular structures. When the starting materials are common monomers the recombination processes are more complex due to the development of simultaneous conventional polymerization reactions along with the fragmentrecombination mechanisms initiated by the plasmacreated and surface-attached active species (e.g., ions, free radicals). In these cases, complex structures result and structural identification is difficult. However, an advantage of conventional monomerbased processes is that specific functionalities can be retained in the polymer structures and consequently some characteristics can be more easily predicted and designed into the final product. Macromolecular thin layers can also be synthesized by the plasma generation of active sites (e.g., ions, free radicals) or reactive functionalities (primary amine groups, etc.) on substrate surfaces, followed by the development of graft-polymerization reactions in the absence of the plasma in the presence of “true monomers.” Remote (the recombination processes are developed outside the plasma zone) and pulsed (plasma sustained according to preselected duty cycles) plasma processes, allow recombination mechanisms in the absence of the plasma state in time or in space, and are characterized by significantly minimized dissociation processes. These approaches do not change dramatically the nature of molecular fragmentation processes (the nature and relative ratios of charged and neutral active species); however, they limit significantly the intensities of plasma-induced fragmentation mechanisms, and allow the development of recombination processes in the partial or total absence of plasma state. As a result, molecular precursors can be physically or through covalent bonding incorporated through surface-located charged and neutral (e.g., free radicals) active species, into the nascent macromolecular layers. Owing to the high reactivity of plasma species, surface functionalization reactions of even the most
N
N
1836
Nanostructures for Surface Functionalization and Surface Properties
inert polymeric substrates can be conveniently achieved. These mechanisms usually involve nonpolymerizing gas plasmas, which generate active molecular fragments that covalently attach during the plasma reactions to the activated substrate surfaces. Recently, it has been demonstrated that even neutral molecules can be connected to active sites with covalent bonds on plasma-exposed substrates under in situ conditions [22]. At high particle energies, etching processes dominate both the deposition and functionalization mechanisms. Depending on the nature of the plasma gases, inert (sputtering) or reactive etching mechanisms can be developed. It should, however, be noted that these mechanisms can have practical importance; they allow the creation of specific surface morphologies by selective ablation, for instance, of amorphous and crystalline surface regions. The nature of plasmas, the modalities of transferring electric or electromagnetic field intensities to the reaction systems, the geometry of reactor (shape and volumes of the reaction vessel, geometrical location of electrodes and substrates, etc.) and the selected experimental conditions (pressure, power, flow rates of gases, temperature of the substrates, etc.) crucially influence the gas-phase and surface-related plasma chemistry. Consequently, comparing experimental results from a large variety of plasma systems, even for identical starting components, is extremely difficult. Cold-plasma-mediated processes involve both gasphase and surface reaction mechanisms. The gas-phase reaction mechanisms involve the interaction of ab initio existing and nascent neutral and charged plasmacreated species, including atoms, molecules, free radicals, ions of either polarity, excited species, electrons, and photons. Besides the recombination mechanisms developed on the surfaces which confine the plasma, the active species of the discharge interact and continuously tailor the artificially exposed (reactor walls, various substrates, etc.) and self-generated (e.g., plasmasynthesized macromolecular structures) surface layers. The competition between the recombination-deposition (deposition, grafting, functionalization) processes and “destructive interaction” of plasma species (etching) with the nascent macromolecular structures will control the intensities and the predominance of ablation, surface functionalization and macromolecular-filmformation reactions.
The mechanisms of surface functionalization of polymeric substrates are different from the gas-phase processes. While electrons play the most important role in the plasma state, positive ions also play a significant role in the surface chemistry during the interactions of plasma species with polymers. The resulting valence-ionized polymer chains undergo neutralization reactions leading to sufficiently intense localized internal energy concentrations (electronically excited states), which can induce homolytic bond cleavages. Free-radical and unsaturated-bond development can result in cross-linking of polymeric layers. Free radicals can further induce chemical reactions involving plasma species in situ and affect reactions controlled by specific chemical environments (gas-phase or condensed-phase compounds, including, oxygen, monomer molecules, etc.) in the absence of plasma. Qualitative and quantitative evaluation of charged and neutral species of the gas phase and identification of charged and neutral surface functionalities on substrate surfaces is essential for understanding the plasma-induced reaction mechanisms and physical processes. Adaptation of in situ diagnostic techniques (including real time measurements) for relatively high and low pressure (100–1,000 mTorr) plasma environments is a research area where many questions still remain unanswered. Continued research will promote the development of novel and/or improved processes for tailoring materials surface characteristics and the development of advanced laboratory plasma installations, required for scaling up plasma processes to industrial levels. Recently, surface treatments and modification based on vapor phase plasma-assisted techniques have been widely applied to several biomedical fields. These processes are able to provide specific monotype chemical functionalities, exposed at materials surfaces. Respect to traditional liquid phase procedures, advantages such as the high process control and the use of small quantities of reagents, are crucial for the enhancement of efficiency of the interactions that materials must perform according to the application. In particular, procedures for in situ plasma polymerization, starting from vapor released by liquid monomers, are presently used to synthesize innovative polymeric thin films applicable as substrates for cell culture [23], as adhesion promoter in prosthetic implants and for molecular recognition in sensors and biosensors.
Nanostructures for Surface Functionalization and Surface Properties
The low pressure plasma polymerization, starting from the same reagent and only varying the process parameters (modulation of the plasma discharge, monomer or mixture of monomers/process gas, reactant vapor partial pressure, etc. ) leads to the deposition of polymeric materials with different features in terms of density of functional groups, wettability (hydrophilicity/hydrophobicity), and chemical stability [24]. The most studied polymers in biomedical and sensing fields, display as functional groups mainly carboxyls (COOH) and amines (NH2), hydroxyls (OH), and aldehydes (CHO). Fewer reports exist on hydroxy and aldehyde surfaces prepared by plasma methods. Hydroxy surfaces can be prepared by water plasma treatment or the plasma polymerization of alkyl alcohol vapors. Water plasma treatment on many polymer substrates suffers from aging, with surface adaptation leading to the movement of surfacemodification effects into the polymer. Both hydroxy and aldehyde surfaces have been used for the covalent immobilization of biologically active molecules. Aging effects are less well documented than for amine surfaces. Aminated surfaces have been fabricated using various plasma vapors or mixtures (allylamines, ethylenediamine, diaminopropane, n-heptylamine [24]) and have found wide use for biointerface applications. However, in many cases the amine surfaces have a rather limited shelf life, with post-plasma oxidation reactions and surface adaptation leading to the disappearance of amine groups from the surface. Aging is a widespread phenomenon that often has not been recognized, particularly in some of the earlier studies on the use of plasma-fabricated surfaces for biointerfacial applications, and can markedly alter the surface chemistry [23]. Plasma-fabricated surfaces that contain carboxyl groups have also been well documented and are obtained, for example, by acrylic acid [23] and vinyl acetic acid or applying CO2 plasma on olefin polymers [24]. In order to better tune the distribution and density of functional groups, are also widely known processes of plasma copolymerization, where the monomer containing functional groups of interest is mixed with a precursor which does not contain specific functionalities (e.g., styrene or octadiene [25]) and which has the function to dilute the reactive groups at the
1837
N
surface. For example, carboxylate copolymer surfaces have shown excellent ability to support the colonization of some human cell lines of clinical interest. Immobilization of proteins onto plasmacarboxylated surfaces is also well established. Generally, these surface chemistries show good ability to support cell colonization, though the effectiveness seems to depend on the process vapor and the plasma conditions. An example in which process parameters affect the final properties of the plasma-polymer film is reported by Detomaso et al. [23]. Starting from Acrylic Acid vapor, processes obtained by Continuous Wave (CW) plasma discharge and by Modulated Wave (MW), that is pulsed, plasma discharge are compared respect to the efficiency of the final polymer coating toward cells growth. The –COOH terminated polymer obtained by MW, with a Duty Cycle (D.C. ¼ (Ton / Ton + Toff ) of 5, where Ton ¼ 5 ms and Toff ¼ 95 ms, is less useful to murine fibroblast growth than the CW plasma film, because more soluble in water due to the higher density of –COOH groups and the higher retention of the monomer (acrylic acid) structure (low cross-linking and backbone chain propagation). Materials Surface Modification: Morphology Effects Fabrication of micro- and nano-patterned surfaces is a fundamental step for the development of engineered materials. For biometerials, it is well known that a series of interactions occur between their surface and the biological environment after they have been implanted into the human body. Therefore, the biomaterials surface plays an extremely important role in the response of artificial medical devices to the biological environment. The efficacy of artificial implants is determined mainly by their surface characteristics such as surface morphology, microstructure, composition, and properties. Liu et al. [26] reported many surface-modification methods mainly based on inorganic nano-coatings deposition and functionalization at the nanoscale of obtained structures. Typical coating and film deposition techniques are plasma spraying, plasma immersion ion implantation and deposition (PIII&D), solgel, chemical vapor deposition (CVD), physical vapor deposition (PVD), cold spraying, self-assembly, and so on, whereas in situ surface-modification techniques
N
N
1838
Nanostructures for Surface Functionalization and Surface Properties
Nanostructures for Surface Functionalization and Surface Properties, Fig. 7 SEM images of a titania nanorod formed on titanium substrate: (a) low magnification, (b) high magnification
include laser etching, shot blasting, acid and alkali treatments, anodic oxidation, micro-arc oxidation, ion implantation, etc. In bioceramics field, for example, a critical grain size for osteoblast adhesion between 49 and 67 nm for alumina and 32 and 56 nm for titania, provides evidence about the ability of nanophase alumina and titania to simulate the desirable materials characteristics of bones in a physiological environment, particularly those associated with the enhancement of protein interactions such as adsorption, configuration, and bioactivity as well as subsequent osteoblast adhesion. Metal surfaces with low micrometer to nanophase topography enhance adhesion of osteoblasts. A surface with various nanostructures can be obtained by chemically treating titanium and its alloys. Chemical treatments are suitable for the creation of micrometer-scale and nanometer-scale textures on large-area surfaces and multifaceted devices with complex 3D shape due to the non-line-of-sight nature. Titania nanorods can be produced at low temperature on a large scale via direct oxidation of titanium with hydrogen peroxide. The ordered nanorod layer with a thickness of approximately 1 mm is deposited on a 2 mm thick condensed anatase layer as illustrated by Fig. 7. Concerning chemical patterning, microstructured surfaces containing opposite bio-properties such as cell and/or protein adhesive features in a non-bioadhesive matrix provide valuable tools in the field of biotechnology and numerous methods have been successfully developed for the production of such surfaces. Besides, the development of chemical nanopatterning technology is great of interest due to its ability to provide powerful tools for studying fundamental aspects of cell adhesion at the level of single protein/receptor molecules or for the implementation
a
200 mm 150X
b
0 2.0μm 4.0μm
10.0μm 8.0μm
6.0μm 6.0μm 4.0μm
8.0μm 2.0μm 10.0μm
Nanostructures for Surface Functionalization and Surface Properties, Fig. 8 (a) SEM image of the patterned PANI film on a silicon substrate; and (b) selected edge area of (a) under AFM
of new generations of miniaturized biochips requiring smaller reagent quantities with lower detection limits [27]. In general [28], the chemical patterning techniques commonly used in bioengineering can be divided into
Nanostructures for Surface Functionalization and Surface Properties
N
a
CH3O+
b
C3H7O+
250 200 Intensity
Nanostructures for Surface Functionalization and Surface Properties, Fig. 9 Positive single ion ToF-SSIMS images of ppMA patterned photolithographically on Si wafer precoated with ppTTg (500 500 mm2). Unique secondary molecular fragments are presented in inset (a). The pixel intensity plot for secondary ions C3H7O+ (ppTTg) and CH3O+ (ppMA) shown in inset (b). The image resolution was calculated to be 10 mm, based on 80:20 definition
1839
C3H7O+
150 100
CH3O+
50 0 0
four categories: (a) lithographic patterning (photoassisted and electron-beam assisted), (b) scanning probe microscopies (SPM); (c) transfer via polymeric stamps (soft lithography), and (d) physical patterning. The challenge with plasma polymerization lies in developing patterning techniques by which, reproducible spatial resolution and distribution of chemistries is achieved and the specific functional properties of the thin films are retained. The simplest patterning approach reported by literature is the one based on the use of physical mask (solid materials deposited on defined areas of the substrate), e.g., polymeric masks and TEM grids. Sometimes, the substrates are previously coated with another plasma polymer obtained starting from a precursor which does not contain specific functionalities (e.g., styrene or octadiene [25]) with the role of non-bio-adhesive matrix. Ruckenstein et al. [1] reported a method for graft polymerization patterning of polyaniline on Si surface previously coated by SAM monolayers from phenyltrichlorosilanes and further subjected to a laser photolithographic process. The subsequent treatment with trifluoromethanesulfonic acid (triflic acid; HOTf) results in the generation of reactive silyl cation sites
100
200 300 Distance (μm)
400
500
on the backbones, which can react with suitable monomers to form functionalized polysilanes. The aniline monomer undergoes to nucleophilic substitution reaction followed by the surface oxidative graft polymerization of aniline on the covalently immobilized aniline sites. The grafted PANI thus formed provided a negative of the lithographic mask. Figure 8a presents the scanning electron microscopy (SEM) image of the patterned PANI after the graft polymerization. The white strips represent the grafted PANI areas, whereas the black strips, the nongrafted ones. A relatively clear boundary between the patterned and nonpatterned areas could be observed in the SEM image. A closer look at the boundaries between the two can be achieved under a higher magnification by AFM (Fig. 8b), which demonstrated that this method provides a high line edge acuity of the patterned PANI. Mishra et al. [28] studied, by using a TEM grid placed onto an already plasma polymer coated silicon wafer, a wide range of standard size patterns in order to vary the resultant pattern features from approximately 5 to 200 mm. After the plasma polymerization process starting from, for example, glycidyl methacrylate (ppGMA), maleic anhydride (ppMA), acrylic acid
N
N
1840
Nanostructures for Surface Functionalization and Surface Properties
Nanostructures for Surface Functionalization and Surface Properties, Fig. 10 SEM images of MG 63 osteoblasts on hydrophobic structured samples showing in detail how filopodia could interact with nanostructures on the surface
(ppAAc), allylamine (ppAAm), tetraglyme (ppTG), and grid remotion, AFM characterization highlighted the presence, at the physical boundaries, of a diffused interface rather than a sharp step (Fig. 9) and Time-ofFlight Secondary Ion Mass Spectrometry (ToF-SIMS) analysis showed that there was diffusion of the plasma deposit below the physical mask during the second polymer layer deposition. On the other side, ion beam etching and plasma etching through masks as well as electron beam lithography (EBL) [27], have been shown to be able to produce chemically resolved microscale and in some cases nanoscale plasma polymer patterns. Bretagnol et al. [27] showed that chemically active features of polyacrylic acid (pAA),with lateral size down to 300 nm, exposing –COOH groups, obtained by EBL and the use of a photoresist mask, were able to graft a model protein (bovine serum albumin), even if a wet chemical step in acetone was performed. Photolithographic patterning for plasma polymers is a particularly attractive approach for the fabrication of complex chemical arrays and the opportunity presented by plasma polymerization to produce thin films from a very wide range of different monomers with several properties such as surface charge, hydrophilicity, biospecificity, and adhesion properties alternated to the opposite such as hydrophobicity and, generally, non-fouling characteristics. Recently, it has been demonstrated that specific properties including protein resistance are maintained after performing the photolithographic patterning technique, with both spatial and chemical resolution down to 1 mm. [27]. Finally, an exotic method for surface nanopatterning, based on plasma polymerization has been reported by Gristina et al. [29]. Teflon ribbon-like nanostructures (Fig. 10) have been obtained by C2F4
MW (low D.C.) plasma polymerization and it has been demonstrated that these structures promote fibroblast cell adhesion through proteins interaction of cell filopodia with the nanostructures surface notwithstanding the well-known teflon super-hydrophobicity.
Cross-References ▶ Applications of Nanofluidics ▶ Bioadhesion ▶ Biosensors ▶ BioPatterning ▶ Charge Transport in Self-assembled Monolayers ▶ Microfabricated Probe Technology ▶ Nanophotonic Structures for Biosensing ▶ Nanostructured Functionalized Surfaces ▶ Self-assembled Monolayers
References 1. Ruckenstein, E., Li, Z.F.: Surface modification and functionalization through the self-assembled monolayer and graft polymerization. Adv. Colloid Interface Sci. 113, 43–63 (2005) 2. Fiorilli, S., Rivolo, P., Descrovi, E., Ricciardi, C., Pasquardini, L., Lunelli, L., et al.: Vapor-phase self-assembled monolayers of aminosilane on plasma-activated silicon substrates. J. Colloid Interface Sci. 321, 235–241 (2008) 3. Ashurst, W.A., Carraro, C., Maboudian, R., Frey, W.: Wafer level anti-stiction coatings for MEMS. Sens Actuator A-Phys. 104, 213–21 (2003) 4. Nuzzo, R.G., Allara, D.L.: Adsorption of bifunctional organic disulfides on gold surfaces. J. Am Chem Soc. 105, 4481–3 (1983) 5. Godin, M.: Surface stress, kinetics and structure of alkanethiol self-assembled monolayers. PhD Dissertation, McGill University, Canada (2004)
Nanotechnology 6. Patel, N., Davies, M.C., Heaton, R.J., Roberts, C.J., Tendler S.J.B., Williams, P.M.: A scanning probemicroscopy study of the physisorption and chemisorption of protein molecules onto carboxylate terminated self-assembledmonolayers. Appl Phys A. 66, S569–74 (1998) 7. Jiang, W., Zhitenev, N., Bao, Z., Meng, H., Abusch-Magder, D., Tennant, D., et al.: Structure and bonding issues at the interface between gold and self-assembled conjugated dithiol monolayers. Langmuir. 21, 8751–7 (2005) 8. Kobayashi, K., Horiuchi, T., Yamada, H., Matsushige, K.: STM studies on nanoscopic structures and electric characteristics of alkanethiol and alkanedithiol self-assembled monolayers. Thin Solid Films. 331, 210–5 (1998) 9. Dubowski, J.J., Voznyy, O., Marshall, G.M.: Molecular self-assembly and passivation of GaAs (0 0 1) with alkanethiol monolayers: a view towards bio-functionalization. Appl Surf Sci. 256, 5714–21 (2010) 10. Ulman, A.: Formation and structure of self-assembled monolayers. Chem. Rev. 96, 1533–54 (1996) ˚ rd, L., Khranovskyy, V., So¨derlind, F., Vahlberg, C., 11. SelegA Ahre´n, M., K€all, P.-O., Yakimova, R., Uvdal, K.: Biotinylation of ZnO nanoparticles and thin films: a twostep surface functionalization study. ACS Appl. Mater. Interfaces 2, 2128–2135 (2010) 12. Baur, B., Steinhoff, G., Hernando, J., Purrucker, O., Tanaka, M., Nickel, B., Stutzmann, M., Eickhoff, M.: Chemical functionalization of GaN and AlN surfaces. Appl. Phys. Lett. 87, 263901-1–263901-3 (2005) 13. Tlili, A., Jarboui, M.A., Abdelghani, A., Fathallah, D.M., Maaref, M.A.: A novel silicon nitride biosensor for specific antibody–antigen interaction. Mater Sci Eng C-Biomimetic Supramol Syst. 25, 490–5 (2005) 14. Lee, S-H., Lee, C-S., Shin, D-S., Kim, B-G., Lee, Y-S., Kim, Y-K.: Micro protein patterning using a lift-off process with fluorocarbon thin film. Sens Actuator B-Chem. 99, 623–32 (2004) 15. Ricciardi, C., Fiorilli, S., Bianco, S., Canavese, G., Castagna, R., Ferrante, I., et al.: Development of microcantilever-based biosensor array to detect Angiopoietin-1, a marker of tumor angiogenesis. Biosens Bioelectron. 25, 1193–8 (2010) 16. Li, D., Teoh, W.Y., Djunaedi, D., Gooding, J.J., Selomulya, C., Amal, R.: Facile functionalization and phase reduction route of magnetic iron oxide nanoparticles for conjugation of matrix metalloproteinase. Adv. Eng. Mater. 12, B210–4 (2010) 17. Wu, W., He, Q., Jiang, C.: Magnetic iron oxide nanoparticles: synthesis and surface functionalization strategies. Nanoscale Res. Lett. 3, 397–415 (2008) 18. K€uhn, G., Weidner, S., Decker, R., Ghode, A., Friedrich, J.: Selective surface functionalization of polyolefins by plasma treatment followed by chemical reduction. Surf Coat Technol. 116–119:796–801 (1999) 19. Bent, S.F.: Organic functionalization of group IV semiconductor surfaces: principles, examples, applications, and prospects. Surf. Sci. 500, 879–903 (2002) 20. Cretich, M., Di Carlo, G., Longhi, R., Gotti, C., Spinella, N., Coffa, S., et al.: High sensitivity protein assays on microarray silicon slides. Anal. Chem. 81, 5197–203 (2009) 21. Steenackers, M., Lud, S.Q., Niedermeier, M., Bruno, P., Gruen, D.M., Feulner, P., et al.: Structured polymer grafts on diamond. J. Am Chem Soc. 129, 15655–61 (2007)
1841
N
22. Denes, F.S., Manolache, S.: Macromolecular plasma-chemistry: an emerging field of polymer science. Prog Polym Sci. 29, 815–85 (2004) 23. Detomaso, L., Gristina, R., D’Agostino, R., Senesi, G.S., Favia, P.: Plasma deposited acrylic acid coatings: surface characterization and attachment of 3 T3 murine fibroblast cell lines. Surf Coat Technol. 200, 1022–5 (2005) 24. Siow, K.S., Britcher, L., Kumar, S., Griesser, H.J.: Plasma methods for the generation of chemically reactive surfaces for biomolecule immobilization and cell colonization – a review. Plasma Process Polym. 3, 392–418 (2006) 25. Swaraj, S., Oran, U., Friedrich, J.F., Lippitz, A., Unger, W.E.S.: Surface chemical analysis of plasma-deposited copolymer films prepared from feed gas mixtures of ethylene or styrene with allyl alcohol. Plasma Process Polym. 4, 376–89 (2007) 26. Liu, X., Chu, P.K., Ding, C.: Surface nano-functionalization of biomaterials. Mater Sci Eng R-Rep. 70, 275–302 (2010) 27. Bre´tagnol, F., Ceriotti, L., Valsesia, A., Sasaki, T., Ceccone, G., Gilliland, D., et al.: Fabrication of functional nanopatterned surfaces by a combination of plasma processes and electron-beam lithography. Nanotechnology. 18, 135303–7 (2007) 28. Mishra, G., Easton, C.D., McArthur, S.L.: Physical vs photolithographic patterning of plasma polymers: an investigation by ToF-SSIMS and multivariate analysis. Langmuir. 26, 3720–30 (2010) 29. Gristina, R., D’Aloia, E., Senesi, G.S., Milella, A., Nardulli, M., Sardella, E., et al.: Increasing cell adhesion on plasma deposited fluorocarbon coatings by changing the surface topography. J. Biomed Mater Res Part B: Appl Biomater. 88B, 139–49 (2009)
Nanotechnology Bharat Bhushan Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA
Synonyms BioMEMS/NEMS; MEMS; Micro/Nanodevices; Micro/Nanomachines; Micro/Nanosystems; MOEMS/ NOEMS; NEMS; RF-MEMS/NEMS
Definition Nanotechnology literally means any technology done on a nanoscale that has applications in the real world. Nanotechnology encompasses the production and
N
N
1842
application of physical, chemical, and biological systems at scales ranging from individual atoms or molecules to submicron dimensions, as well as the integration of the resulting nanostructures into larger systems.
Overview Nanotechnology is likely to have a profound impact on world economy and society in the early twenty-first century, comparable to that of semiconductor technology, information technology, or cellular and molecular biology. Science and technology research in nanotechnology promises breakthroughs in areas such as materials and manufacturing, nanoelectronics, medicine and healthcare, energy, biotechnology, information technology, and national security. It is widely felt that nanotechnology will be the next industrial revolution. Nanometer-scale features are mainly built up from their elemental constituents. Examples include chemical synthesis, the spontaneous self-assembly of molecular clusters (molecular self-assembly) from simple reagents in solution, biological molecules (e.g., DNA) used as building blocks for the production of three-dimensional nanostructures, or quantum dots (nanocrystals) of an arbitrary diameter (about 10–105 atoms). The definition of a nanoparticle is an aggregate of atoms bonded together with a radius between 1 and 100 nm. It typically consists of 10–105 atoms. A variety of vacuum deposition and nonequilibrium plasma chemistry techniques are used to produce layered nanocomposites and nanotubes. Atomically controlled structures are produced using molecular beam epitaxy and organo-metallic vapor phase epitaxy. Micro- and nanosystem components are fabricated using top-down lithographic and nonlithographic fabrication techniques and range in size from micro- to nanometers. Continued improvements in lithography for use in the production of nanocomponents have resulted in line widths as small as 10 nm in experimental prototypes. The nanotechnology field, in addition to the fabrication of nanosystems, provides impetus to the development of experimental and computational tools. The discovery of novel materials, processes, and phenomena at the nanoscale and the development of new experimental and theoretical techniques for research provide fresh opportunities for the
Nanotechnology
development of innovative nanosystems and nanostructured materials. The properties of materials at the nanoscale can be very different from those at a larger scale. When the dimension of a material is reduced from a large size, the properties remain the same at first; then small changes occur, until finally when the size drops below 100 nm, and dramatic changes in properties can occur. If only one length of a three-dimensional nanostructure is of the nanodimension, the structure is referred to as a quantum well; if two sides are of nanometer length, the structure is referred to as a quantum wire. A quantum dot has all three dimensions in the nanorange. The word quantum is associated with these three types of nanostructures because the changes in properties arise from the quantummechanical nature of physics in the domain of the ultra-small. Materials can be nanostructured for new properties and novel performance. This field is opening new venues in science and technology. Micro- and nanosystems include Micro/NanoElectroMechanical Systems. MEMS refers to microscopic devices that have a characteristic length of less than 1 mm but more than 100 nm and combine electrical and mechanical components. NEMS refers to nanoscopic devices that have a characteristic length of less than 100 nm and combine electrical and mechanical components. In mesoscale devices, if the functional components are on the micro- or nanoscale, they may be referred to as MEMS or NEMS, respectively. These are referred to as an intelligent miniaturized system comprising of sensing, processing, and/or actuating functions and combine electrical and mechanical components. The acronym MEMS originated in the USA. The term commonly used in Europe is microsystem technology (MST), and in Japan it is micromachines. Another term generally used is micro/ nanodevices. MEMS/NEMS terms are also now used in a broad sense and include electrical, mechanical, fluidic, optical, and/or biological function. MEMS/ NEMS for optical applications are referred to as micro/nanooptoelectromechanical systems (MOEMS/ NOEMS). MEMS/NEMS for electronic applications are referred to as radio-frequency-MEMS/NEMS or RF-MEMS/RF-NEMS. MEMS/NEMS for biological applications are referred to as BioMEMS/BioNEMS. To put the dimensions of MEMS/NEMS and BioNEMS in perspective, see Fig. 1 and Table 1 [4]. Individual atoms are typically a fraction of
Nanotechnology
1843
N
MEMS - characteristic length less than 1 mm, larger than 100 nm NEMS - less than 100 nm
Mirror
Human hair 50-100 μm
Hinge
Yoke
Landing tip
DMD 12 μm 500 nm Source
Drain
Quantum-dots transistor Red blood cell 300 nm 8 μm
Molecular gear 10 nm-100 nm
A SWNT
1 μm
SWNT chemical sensor 2 nm
N C atom 0.16 nm 0.1
DNA 2.5 nm 1
10
100
1000
10000
100000
Size (nm)
Nanotechnology, Fig. 1 Dimensions of MEMS/NEMS and BioNEMS in perspective. Examples shown are of a singlewalled carbon nanotube (SWNT) chemical sensor [1], of molecular dynamic simulations of carbon-nanotube–based gears [2],
quantum-dot transistor obtained from van der Wiel et al. [3], and DMD obtained from www.dlp.com. For comparison, dimensions and weights of various biological objects found in nature are also presented
a nanometer in diameter, DNA molecules are about 2.5 nm wide, biological cells are in the range of thousands of nm in diameter, and human hair is about 75 mm in diameter. Smaller length of BioNEMS shown in the figure is about 2 nm, NEMS ranges in size from 10 to 300 nm, and the size of MEMS is 12,000 nm. The mass of a micromachined silicon structure can be as low as 1 nN, and NEMS can be built with mass as low as 10–20 N with cross-sections of about 10 nm. In comparison, the mass of a drop of water is about 10 mN, and the mass of an eyelash is about 100 nN.
MEMS/NEMS and BioMEMS/BioNEMS are expected to have a major impact on people’s lives, comparable to that of semiconductor technology, information technology, or cellular and molecular biology [5, 6]. MEMS/NEMS and BioMEMS/ BioNEMS are used in electromechanical, electronics, information/communication, chemical, and biological applications. The MEMS industry in 2004 was worth about $4.5 billion with a projected annual growth rate of 17%, Fig. 2 [7]. The NEMS industry was worth about $10 billion in 2004, mostly in nanomaterials,
N
1844
Nanotechnology, Table 1 Characteristic dimensions and weights in perspective
Nanotechnology (a) Characteristic dimensions in perspective NEMS characteristic length MEMS characteristic length SWCNT Chemical sensor Molecular gear Quantum-dots transistor Digital micromirror Individual atoms DNA molecules Biological cells Human hair (b) Weight in perspective NEMS built with cross-sections of about 10 nm Micromachines silicon structure Eyelash Water droplet
Fig. 2 [8]. Growth of Si-based MEMS/NEMS may slow down, and nonsilicon MEMS may pick up during the next decade. It is expected to expand in this decade, in nanomaterials and biomedical applications as well as in nanoelectronics or molecular electronics. For example, miniaturized diagnostics could be implanted for the early diagnosis of illness. Targeted drug delivery devices are under development. Due to the enabling nature of these systems and because of the significant impact they can have on both commercial and defense applications, industry as well as federal governments have taken special interest in seeing growth nurtured in this field. MEMS/NEMS and BioMEMS/BioNEMS are the next logical step in the “silicon revolution.”
Background and Research Expenditures On December 29, 1959 at the California Institute of Technology, Nobel Laureate Richard P. Feynman gave a talk at the Annual Meeting of the American Physical Society that has become one of the twentieth century classic science lectures, titled “There’s Plenty of Room at the Bottom” [9]. He presented a technological vision of extreme miniaturization in 1959, several years before the word “chip” became part of the lexicon. He talked about the problem of manipulating and controlling things on a small scale. Extrapolating from known physical laws, Feynman envisioned a technology using the ultimate toolbox of nature,
< 100 nm < 1 mm and > 100 nm 2 nm 10 nm 300 nm 12,000 nm typically fraction of a nm in diameter 2.5 nm wide In the range of thousands of nm in diameter 75,000 nm in diameter As low as 10–20 N As low as 1 nN 100 nN 10 mN
building nanoobjects atom by atom or molecule by molecule. Since the 1980s, many inventions and discoveries in the fabrication of nanoobjects have been testaments to his vision. In recognition of this reality, the National Science and Technology Council (NSTC) of the White House created the Interagency Working Group on Nanoscience, Engineering and Technology (IWGN) in 1998. In a January 2000 speech at the same institute, former President W. J. Clinton talked about the exciting promise of “nanotechnology” and the importance of expanding research in nanoscale science and technology more broadly. Later that month, he announced in his State of the Union Address an ambitious $497 million federal, multi-agency national nanotechnology initiative (NNI) in the fiscal year 2001 budget, and made the NNI a top science and technology priority [10, 11]. The objective of this initiative was to form a broad-based coalition in which the academe, the private sector, and local, state, and federal governments work together to push the envelope of nanoscience and nanoengineering to reap nanotechnology’s potential social and economic benefits. The funding in the USA has continued to increase. In January 2003, the US senate introduced a bill to establish a National Nanotechnology Program. On December 3, 2003, former President George W. Bush signed into law the twenty-first century Nanotechnology Research and Development Act. The legislation put into law programs and activities supported by the National Nanotechnology Initiative. The bill gave
Nanotechnology
1845
Nanotechnology, Fig. 2 Global MEMS and nanotechnology market segments
N
Global MEMS market segment 8 7
US $ billions
6
Microbolometers RF MEMS
5
Bio & Microfluidics comp. DLP (micromirrors)
4
MOEMS
3
Gyroscopes Accelerometers Pressure sensors Inkjet Head
2 1 0 2004
2003
2005
2006
2007
Global nanotechnology market segment 30 Nanodevices
US $ billions
Nanotools 20 Nanomaterials 10
0 2002
N 2003
2004
nanotechnology a permanent home in the federal government and authorized $3.7 billion to be spent in the 4year period beginning in October 2005 for nanotechnology initiatives at five federal agencies. The funds would provide grants to researchers, coordinate R&D across five federal agencies (National Science Foundation (NSF), Department of Energy (DOE), NASA, National Institute of Standards and Technology (NIST), and Environmental Protection Agency (EPA)), establish interdisciplinary research centers, and accelerate technology transfer into the private sector. In addition, the Departments of 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 for more than one-third of the federal budget for nanotechnology. The European Union (EU) made nanosciences and nanotechnologies a priority in the Sixth Framework
2005
2006
2007
2008
Program (FP6) in 2002 for a period of 2003–2006. They had dedicated small funds in FP4 and FP5 before. FP6 was tailored to help better structure European research and to cope with the strategic objectives set out in Lisbon in 2000. Japan identified nanotechnology as one of its main research priorities in 2001. The funding levels increased sharply from $400 million in 2001 to around $950 million in 2004. In 2003, South Korea embarked upon a 10-year program with around $2 billion of public funding, and Taiwan has committed around $600 million of public funding over 6 years. Singapore and China are also investing on a large scale. Russia is well funded as well. Figure 3a shows the public expenditure breakdown in nanotechnology R&D around the world, with about $5 billion in 2004, being about equal by USA, Japan, and Europe. Next public expenditure on a per capita basis is compared. The average expenditures per
N
1846
R&D expenditure (US $ billions)
a
Applications in Different Fields
Public expenditure in nanotechnology R&D 4 FP6 (EU) 3 Europe NNI (USA)
2
Japan USA
1
0 1997
b
R&D Expenditure (US $ billions)
Nanotechnology
Others 1998
1999
2000
2001
2002
2003
Public and private expenditure in nanotechnology R&D in 2004
3 Private 1.3 Private 1.7
2
1
0
Member States+ Associated 1.4 EC 0.48 Europe
States 0.40
Private 1.4 Public 1.9
Federal 0.99
Public 0.90
USA
Japan
Others
Nanotechnology, Fig. 3 Breakdown of public expenditure in nanotechnology R&D (a) around the world (European Commission, 2003), and (b) by public and private resources in 2004 (European Commission 2005; figures from private sources based upon Lux Research)
capita for the USA, EU-25, and Japan are about $3.7 billion, $2.4 billion, and $6.2 billion, respectively [12]. Figure 3b shows the breakdown of expenditures in 2004 by public and private sources with more than $10 billion spent in nanotechnology research. Two thirds of this came from corporate and private funding. The private expenditure in USA and Japan was slightly larger than that of the public, whereas in Europe it was about one-third. Figure 4 shows the public and private expenditure breakdown in 2004 in various countries. Japan and USA had the largest expenditure, followed by Germany, Taiwan, South Korea, UK, Australia, China, France, and Italy. Figure 5 shows a breakdown of worldwide publications and patents. USA and Canada led, followed by Europe and Asia. Figure 6 shows the breakdown in start-up companies around the world (1997–2002). Entrepreneurship in USA is clearly evident followed by Europe.
Science and technology continue to move forward in masking the fabrication of micro/nanodevices and systems possible for a variety of industrial, consumer, and biomedical applications (e.g., [4, 13, 14]). A variety of MEMS devices have been produced, and some are commercially used [4, 15–23, 45]. A variety of sensors are used in industrial, consumer, defense, and biomedical applications. Various micro/nanostructures or micro/nanocomponents are used in micro-instruments and other industrial applications such as micromirror arrays. The largest “killer” MEMS applications include accelerometers (some 90 million units installed in vehicles in 2004), silicon-based piezoresistive pressure sensors for manifold absolute pressure sensing for engines and for disposable blood pressure sensors (about 30 million units and about 25 million units, respectively), capacitive pressure sensors for tire pressure measurements (about 37 million units in 2005), thermal inkjet printheads (about 500 million units in 2004), and digital micromirror arrays for digital projection display (about $700 million revenue in 2004), and optical cross-connections in telecommunications. Other applications of MEMS devices include chemical/biosensors and gas sensors, microresonators, infrared detectors and focal plane arrays for earth observations, space science and missile defense applications, pico-satellites for space applications, fuel cells, and many hydraulic, pneumatic, and other consumer products. MEMS devices are also being pursued in magnetic storage systems [24], where they are being developed for super compact and ultrahigh-recordingdensity magnetic disk drives. NEMS are produced by nanomachining in a typical top-down approach and bottom-up approach, largely relying on nanochemistry [4, 25–30, 46]. Examples of NEMS include microcantilevers with integrated sharp nanotips for scanning tunneling microscopy (STM) and atomic force microscopy (AFM), quantum corral formed using STM by placing atoms one by one, AFM cantilever array for data storage, AFM tips for nanolithography, dip-pen lithography for printing molecules, nanowires, carbon nanotubes, quantum wires (QWRs), quantum boxes (QBs), quantum-dot transistors, nanotube-based sensors, biological (DNA) motors, molecular gears by attaching benzene molecules to the outer walls of carbon nanotubes, devices incorporating nm-thick films (e.g., in giant
Nanotechnology
1847
Nanotechnology, Fig. 4 Breakdown of public and private expenditures in nanotechnology R&D in 2004 in various countries [8]
N
Public and private expenditure in nanotechnology R&D in 2004 Government/institutional
Corporate/private
Japan USA Germany UK Taiwan South Korea Australia China
2004 global total Corporate/private - $6.6 billion Government/institutional - $3.7 billion
France Italy 0
1
2
3
4
R&D expenditure (US $ billions)
magnetoresistive or GMR read/write magnetic heads and magnetic media) for magnetic rigid disk drives and magnetic tape drives, nanopatterned magnetic rigid disks, and nanoparticles (e.g., nanoparticles in magnetic tape substrates and magnetic particles in magnetic tape coatings). Nanoelectronics can be used to build computer memory using individual molecules or nanotubes to store bits of information, molecular switches, molecular or nanotube transistors, nanotube flat-panel displays, nanotube integrated circuits, fast logic gates, switches, nanoscopic lasers, and nanotubes as electrodes in fuel cells. BioMEMS/BioNEMS are increasingly used in commercial and defense applications (see, e.g., [4, 31–37]. They are used for chemical and biochemical analyses (biosensors) in medical diagnostics (e.g., DNA, RNA, proteins, cells, blood pressure and assays, and toxin identification) [37], tissue engineering [38], and implantable pharmaceutical drug delivery [39, 40]. Biosensors, also referred to as biochips, deal with liquids and gases. There are two types of biosensors. A large variety of biosensors are based on micro/ nanofluidics. Micro/nanofluidic devices offer the ability to work with smaller reagent volumes and shorter reaction times, and perform analyses multiple
times at once. The second type of biosensors includes micro/nanoarrays which perform one type of analysis thousands of times. Micro/nanoarrays are a tool used in biotechnology research to analyze DNA or proteins to diagnose diseases or discover new drugs. Also called DNA arrays, they can identify thousands of genes simultaneously [33]. They include a microarray of silicon nanowires, roughly a few nm in size, to selectively bind and detect even a single biological molecule, such as DNA or protein, by using nanoelectronics to detect the slight electrical charge caused by such binding, or a microarray of carbon nanotubes to electrically detect glucose. After the tragedy of September 11, 2001, concern of biological and chemical warfare has led to the development of handheld units with bio- and chemical sensors for detection of biological germs, chemical or nerve agents, and mustard agents, and chemical precursors to protect subways, airports, water supply, and population at large. BioMEMS/BioNEMS are also being developed for minimal invasive surgery including endoscopic surgery, laser angioplasty, and microscopic surgery. Other applications include implantable drug-delivery devices – micro/nanoparticles with drug molecules encapsulated in functionalized shells for site-specific
N
N a
1848
Nanotechnology
targeting applications, and a silicon capsule with a nanoporous membrane filled with drugs for long-term delivery.
Worldwide publications in nanotechnology 1997-99 Europe 34% Others 5%
Candidate Countries and Russia 8%
b
Various Issues USA and Canada 28%
Asia 25%
Worldwide patents in nanotechnology
Europe 39%
Others 3% Asia 13%
USA and Canada 45%
Nanotechnology, Fig. 5 Breakdown of (a) worldwide publications and (b) worldwide patents (European Commission 2003)
There is an increasing need for a multidisciplinary, system-oriented approach to the manufacturing of micro/nanodevices which function reliably. This can only be achieved through the cross-fertilization of ideas from different disciplines and the systematic flow of information and people among research groups. Common potential failure mechanisms for MEMS/NEMS requiring relative motion that need to be addressed in order to increase their reliability are: adhesion, friction, wear, fracture, fatigue, and contamination [4, 41–44]. Surface micro/nanomachined structures often include smooth and chemically active surfaces. Due to large surface area to volume ratio in MEMS/NEMS, they are particularly prone to stiction (high static friction) as part of normal operation. Fracture occurs when the load on a microdevice is greater than the strength of the material. Fracture is a serious reliability concern, particularly for brittle materials used in the construction of these components, since it can immediately or would eventually lead to catastrophic failures. Additionally, debris can be formed from the fracturing of microstructures, leading to other failure processes. For less brittle materials, repeated
Start-up companies in nanotechnology (1997-2002) Asia 4%
Rest of world 11%
France 4% Switzerland 4%
Europe 29%
Nanotechnology, Fig. 6 Breakdown of start-up companies around the world (1997–2002) (CEA, Bureau d’Etude Marketing)
USA 55%
UK 6% Germany 11%
Others 5%
Nanotechnology
loading over a long period of time causes fatigue that would also lead to the breaking and fracturing of the device. In principle, this failure mode is relatively easy to observe and simple to predict. However, the material properties of thin films are often not known, making fatigue predictions error-prone. Many MEMS/NEMS devices operate near their thermal dissipation limit. They may encounter hot spots that may cause failures, particularly in weak structures such as diaphragms or cantilevers. Thermal stressing and relaxation caused by thermal variations can create material delamination and fatigue in cantilevers. In large temperature changes, as experienced in the space environment, bimetallic beams will also experience warping due to mismatched coefficients of thermal expansion. Packaging has been a big problem. The contamination that probably happens in packaging and during storage also can strongly influence the reliability of MEMS/NEMS. For example, a particulate dust landed on one of the electrodes of a comb drive can cause catastrophic failure. There are no MEMS/NEMS fabrications standards, which make it difficult to transfer fabrication steps in MEMS/ NEMS between boundaries. Obviously, studies of determination and suppression of active failure mechanisms affecting this new and promising technology are critical to high reliability of MEMS/NEMS and are determining factors in successful practical application. Adhesion between a biological molecular layer and the substrate, referred to as “bioadhesion,” and reduction of friction and wear of biological layers, biocompatibility, and biofouling for BioMEMS/BioNEMS are important. Mechanical properties are known to exhibit a dependence on specimen size. Mechanical property evaluation of nanometer-scaled structures is carried out to help design reliable systems since good mechanical properties are of critical importance in such applications. Some of the properties of interest are: Young’s modulus of elasticity, hardness, bending strength, fracture toughness, and fatigue life. Finite element modeling is carried out to study the effects of surface roughness and scratches on stresses in nanostructures. When nanostructures are smaller than a fundamental physical length scale, conventional theory may no longer apply, and new phenomena
1849
N
emerges. Molecular mechanics is used to simulate the behavior of a nano-object. The societal, ethical, political, and health/safety implications are getting major attention [12]. One of the prime reasons is to avoid some of the public skepticism that surrounded the debate over biotech advances such as genetically modified foods, while at the same time dispelling some of the misconceptions the public may already have about nanotechnology. Health/safety issues need to be addressed as well. For example, one key question is what happens to nanoparticles (such as buckyballs or nanotubes) in the environment and are they toxic in the human body, if digested.
Cross-References ▶ Biomimetics ▶ MEMS ▶ Micromachines ▶ Nanocalorimetry ▶ Nanofabrication ▶ NEMS
References 1. Chen, R.J., Choi, H.C., Bangsaruntip, S., Yenilmex, E., Tang, X., Wang, Q., Chang, Y.L., Dai, H.: An investigation of the mechansims of electrode sensing of protein adsorption on carbon nanotube devices. J. Am. Chem. Soc. 126, 1563–1568 (2004) 2. Srivastava, D.: Computational nanotechnology of carbon nanotubes. In: Meyyappan, M. (ed.) Carbon Nanotubes: Science and Applications, pp. 25–36. CRC Press, Boca Raton (2004) 3. van der Wiel, W.G., De Franceschi, S., Elzerman, J.M., Fujisawa, T., Tarucha, S., Kauwenhoven, L.P.: Electron transport through double quantum dots. Rev. Mod. Phys. 75, 1–22 (2003) 4. Bhushan, B.: Springer Handbook of Nanotechnology, 3rd edn. Springer, Heidelberg (2010) 5. Anonymous.: Microelectromechanical systems: advanced materials and fabrication methods, NMAB-483, National Academy, Washington, DC (1997) 6. Roukes, M.: Nanoelectromechanical systems face the future. Phys. World 14, 25–31 (2001) 7. Eloy, J.C.: Status of the MEMS industry 2005. Report Yole Developpement, France. Presented at SPIE Photonics West, San Jose, California (2005)
N
N
1850
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
8. Lawrence, S.: Nanotech grows up. Technol. Rev. 108(6), 31 (2005) 9. Feynman, R.P.: There’s plenty of room at the bottom. Eng. Sci. 23, 22–36 (1960). www.zyvex.com/nanotech/feynman. html 10. Amato, I.: Nanotechnology, www.ostp.gov/nstc/html/iwgn/ iwgn.public.brochure/welcome.htm or www.nsf.gov/home/ crssprgm/nano/nsfnnireports.htm (2000) 11. Anonymous.: National nanotechnology initiative, www. ostp.gov/nstc/html/iwgn.fy01budsuppl/nni.pdf or www.nsf. gov/home/crssprgm/nano/nsfnnireports.htm (2000) 12. Anonymous.: Towards a European Strategy for Nanotechnology, European Commission, Research Directorate, Brussels (2004) 13. Anonymous.: Small Tech 101 – An Introduction to micro and nanotechnology, Small Times (2003) 14. Schulenburg, M.: Nanotechnology – Innovation for Tomorrow’s World. European Commission Research DG, Brussels (2004) 15. Bhushan, B.: Tribology Issues and Opportunities in MEMS. Kluwer, Dordrecht (1998) 16. Elwenspoek, M., Wiegerink, R.: Mechanical Microsensors. Springer, Berlin (2001) 17. Fujimasa, I.: Micromachines: A New Era in Mechanical Engineering. Oxford University Press, Oxford (1996) 18. Hierlemann, A.: Integrated Chemical Microsensor Systems in CMOS Technology. Springer, Berlin (2005) 19. Hsu, T.R.: MEMS and Microsystems: Design and Manufacture. McGraw-Hill, Boston (2002) 20. Kovacs, G.T.A.: Micromachined Transducers Sourcebook. WCB McGraw-Hill, Boston (1998) 21. Madou, M.: Fundamentals of Microfabrication: The Science of Miniaturization, 2nd edn. CRC Press, Boca Raton (2002) 22. Senturia, S.D.: Microsystem Design. Kluwer, Boston (2000) 23. Trimmer, W.S. (ed.): Micromachines and MEMS, Classical and Seminal Papers to 1990. IEEE Press, New York (1997) 24. Bhushan, B.: Tribology and Mechanics of Magnetic Storage Devices, 2nd edn. Springer, New York (1996) 25. Dresselhaus, M.S., Dresselhaus, G., Avouris, Ph.: Carbon Nanotubes – Synthesis, Structure, Properties, and Applications. Springer, Berlin (2001) 26. Drexler, K.E.: Nanosystems: Molecular Machinery, Manufacturing and Computation. Wiley, New York (1992) 27. Timp, G. (ed.): Nanotechnology. Springer, New York (1999) 28. Goddard, W.A., Brenner, D.W., Lyshevski, S.E., Iafrate, G.J. (eds.): Handbook of Nanoscience, Engineering, and Technology. CRC Press, Boca Raton (2002) 29. Nalwa, H.S. (ed.): Nanostructured Materials and Nanotechnology. Academic, San Diego (2002) 30. Poole, C.P., Owens, F.J.: Introduction to Nanotechnology. Wiley, New York (2003) 31. Cheng, J., Kricka, L.J. (eds.): Biochip Technology. Harwood Academic, Philadephia (2001) 32. Heller, M.J., Guttman, A. (eds.): Integrated Microfabricated Biodevices. Marcel Dekker, New York (2001) 33. Lai Poh San, C., Yap, E.P.H. (eds.): Frontiers in Human Genetics. World Scientific, Singapore (2001) 34. Manz, A., Becker, H. (eds.): Microsystem Technology in Chemistry and Life Sciences, Topics in Current Chemistry 194. Springer, Heidelberg (1998)
35. Mastrangelo, C.H., Becker, H. (ed.): Microfluidics and BioMEMS. In: Proceedings of SPIE, vol. 4560, SPIE, Bellingham, Washington (2001) 36. Becker, H., Locascio, L.E.: Polymer microfluidic devices. Talanta 56, 267–287 (2002) 37. van der Berg, A. (ed.): Lab-on-a-Chip: Chemistry in Miniaturized Synthesis and Analysis Systems. Elsevier, Amsterdam (2003) 38. Lanza, R.P., Langer, R., Vacanti, J. (eds.): Principles of Tissue Engineering, 2nd edn. Academic, San Diego (2000) 39. Oeberg, P.A., Togawa, T., Spelman, F.A.: Sensors in Medicine and Health Care. Wiley, New York (2004) 40. Park, K. (ed.): Controlled Drug Delivery: Challenges and Strategies. American Chemical Society, Washington, DC (1997) 41. Bhushan, B.: Principles and Applications of Tribology. Wiley, New York (1999) 42. Bhushan, B.: Handbook of Micro/Nanotribology. CRC Press, Boca Raton (1999) 43. Bhushan, B.: Introduction to Tribology. Wiley, New York (2002) 44. Bhushan, B.: Nanotribology and Nanomechanics I – Measurement Techniques and Nanomechanics, II – Nanotribology, Biomimetics, and Industrial Applications, 3rd edn. Springer, Heidelberg (2011) 45. Muller, R.S., Howe, R.T., Senturia, S.D., Smith, R.L., White, R.M.: Microsensors. IEEE Press, New York (1991) 46. Rietman, E.A.: Molecular Engineering of Nanosystems. Springer, New York (2001)
Nanotechnology Applications in Polymerase Chain Reaction (PCR) Kuo-Sheng Ma IsaCal Technology, Inc., Riverside, CA, USA
Synonyms Nucleic acid amplification
amplification;
Oligonucleotide
Definition Polymerase chain reaction (PCR) is the most prevalent technology used in modern molecular biology research. Generally, it is a technique to amplify the amount of a piece of DNA with a specific sequence. Although the PCR technique is mature, improvement of its efficiency is still an emerging area of research. Recently, nanotechnology is getting more attention in
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
this application. Several nanometer-sized materials such as carbon nanotubes, gold nanoparticles, quantum dots, and metal oxide nanoparticles have been employed.
Introduction The polymerase chain reaction (PCR) process is a technique to amplify the number of copies of a specific DNA sequence, producing hundreds of thousands of copies of DNA. This revolutionary technique was first developed by Kary Mullis [1] in 1983. Since then, it has been broadly used in medical and biological research laboratories for a variety of applications [2, 3]. These include DNA sequencing, DNA cloning, identification of functional gene, the disease diagnosis, and the identification of genetic fingerprints. A basic PCR reaction needs several components in the mixture. These include: (a) A small amount of original DNA (mRNA in some applications) as a template which contains the oligonucleotides segment to be amplified. (b) Primers. These are short sequences of oligonucleotides in which their sequences are designed complementary to the specific site of the template DNA. Two primers are used for one amplification process and each primer binds to the different end of the segment on one of the paired DNA strands. (c) DNA polymerases. These are the enzymes that perform DNA synthesis in the amplification process. The Taq polymerase is most prevalently used in the conventional PCR process. (d) Deoxynucleotide triphosphates (dNTPs). These are the building blocks of DNA which are polymerized by the enzyme through the sequence of DNA templates. (e) Buffer solution. This is the salt solution to provide optimum condition for DNA synthesis. Divalent cations such as Mg2+ or Mn2+ are normally used in buffer solution to mediate DNA mutagenesis or decrease the error rate of DNA polymerase. The technique is based on the cycles of repeated temperature changes of the reactant in reaction tubes. The temperature changes are normally carried out in a thermal cycler which performs heating and cooling of the tubes. A typical PCR process consists of 20–40 cycles, with each cycle consisting of three basic heating and cooling steps on discrete temperatures. The temperatures employed and the duration of specific temperature in a cycle depends on the reactant
1851
N
used in the mixture for each specific process. In a conventional PCR process, as the first step (or called as denature step), the mixture is heated up to 94–98 C for 20–30 s to break the hydrogen bonds between complementary double-stranded DNA (templates) and yield single-stranded DNA molecules. In the second step (or called as annealing step), the temperature of the solution is lowered to 50–65 C for 20–40 s allowing the primers bind to the separated strands of DNA template. Sequentially, the enzyme (DNA polymerase) binds to the hybridization sites of primer and template to start DNA amplification. In the next step (also called the extension/elongation step), the temperature of solution is risen to a temperature at which the enzyme has its optimum activity. A temperature of 72 C is used for the commonly used Taq polymerase. At this step, the amount of DNA is doubled due to the enzyme synthesizing a new complementary strand to each DNA template by polymerizing dNTPs in 50 –30 direction. These three basic steps form a cycle and repetitions of this cycle lead to exponential amplification of the amount of DNA fragments. Figure 1 schematically shows a conventional PCR process.
Variations on the Basic PCR Technique PCR is such a powerful technique to provide a relatively easy, efficient method to amplify the amount of desired DNA sequence. Since it was invented, many modifications to tailor the basic PCR thermal cycle approach to fit specific applications have been made. The features of these modified techniques are summarized in Table 1.
Nanotechnology in PCR Although the PCR technique is mature and has been broadly employed in various applications, there is still an effort to improve the specificity and efficiency of it by using different advanced technologies. In the nanotechnology, the development and discovery of more applications of nanoparticles is an emerging area of research. Nanoparticle-assisted PCR has attracted more attentions and many nanomaterials have been studied for their potential in developing better PCR technique. Various types of nanoparticles such as carbon nanotubes, metal nanoparticles (especially
N
N
1852
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
Nanotechnology Applications in Polymerase Chain Reaction (PCR), Fig. 1 Schematic illustration of PCR processes
5 3
3 5 1. Denature Step 3
5 3
5 2. Annealing Step
5 3
3 3
5
3
5
5
3. Elongation Step
3
5
3
3
5
5
5
3
1. & 2. & 3.
3
5 3 3
5 3
5 5
3 5
3 5
3 5 5 3
More cycles and exponential growth
5′
3′ Primer DNA polymerase
gold nanoparticles), semiconductor quantum dots, and metal oxide nanoparticles have been introduced into PCR and their effect on reaction efficiency studied. Some studies showed these nanoparticles dramatically improve PCR in efficiency and specificity but some studies showed contrary results. In this entry, the development of the technique which employed the nanoparticles in PCR would be introduced. The different types of nanoparticles utilized in the studies and their effect in assisting PCR as well as mechanisms will be discussed. Carbon Nanotubes Gao et al. [14] investigated the effects of single-walled carbon nanotubes (SWCNTs) on PCR. Their results showed the amount of PCR product increased by adding single-walled carbon nanotubes into the PCR
reactant mixture. They further discovered that adding higher concentrations of SWCNTs (> 3 ug/ul) reversed the effect (amount of PCR product decreased). Using SEM and HRTEM, they showed that DNA templates and Taq enzymes were attached to bundles of SWCNTs. XPS results further suggested that there might be a chemical reaction between SWCNTs and components in PCR solution. The authors concluded that the SWCNTs have the potential to act as catalysts in the solution, which increased the yield of PCR products. Zhang et al. [15] had done the similar study and reported the beneficial effect of using single-walled carbon nanotubes (SWCNTs) and multi-walled carbon nanotubes (MWCNTs) to enhance the amplification efficiency of long PCR up to 14 kb. Rather than the pristine carbon nanotubes, the authors employed
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
1853
N
Nanotechnology Applications in Polymerase Chain Reaction (PCR), Table 1 Variation techniques based on PCR process Name 1. Quantitative PCR (qPCR)
2. Reverse transcription PCR (RT-PCR)
3. Polymerase cycling assembly (PCA)
4. Helicase-dependent amplification
5. Hot start PCR
6. Methylation-specific PCR (MSP)
7. Allele-specific PCR
8 Solid phase PCR
Functions A technique used to quantify the PCR products during the amplification cycles. Because the measurement is in “real time” of amplification, it is also called quantitative real-time PCR (qRT-PCR). The technique employed the fluorescent dyes in the PCR solution to measure the amount of amplified products in the log phase of the reaction before the plateau. The commonly used dyes are Syber Green or TaqMan which is a DNA probe containing a fluorophore A preferred technique used to detect and quantify mRNA. In mRNA detection applications, the target RNA needs to be amplified by PCR. Thus, RNA samples need to be first reverse transcribed into cDNA by reverse transcriptase and cDNA sequences will be amplified by PCR. This technique is widely used in profiling of gene expressions A technique to synthesize long DNA sequences by a PCR. It employed a pool of oligonucleotides with designed short overlapping sequences in each piece of oligonucleotides. Rather than using two primers, this technique uses multiple short oligonucleotides which are part of the single strand of the target sequences. During the elongation cycle, the rest of sequences will be filled in by polymerase. The technique is popular for the production of synthetic genes and even synthetic genomes This technique is similar to traditional PCR but uses a constant temperature instead of temperature cycling in denaturation and annealing steps. Rather than thermal denaturation, a DNA helicase is used to separate the double strand of DNA which provides single-stranded DNA template for primers and polymerase binding A modified process to reduce nonspecific amplification of DNA by inactivating the Taq polymerase at a lower temperature in the initial stages of the conventional PCR. It can be performed manually by adding a primer to start amplification after heating the reaction components to the melting temperature. Common methods for this technique include chemical modifications, wax-barrier methods, inhibition by a Taq-directed antibody, and by the presence of covalently bound inhibitors. Due to lack of nonspecific hybridization of primers to template, the amplified DNA bands tend to be cleaner A technique is used to detect methylation of CpG islands in the genomic DNA. DNA is first treated with sodium bisulfite, which converts unmethylated cytosine bases to uracil. A special primer is designed to distinguish methylated from unmethylated DNA. The process takes advantage of the sequence differences resulting from bisulfite modification. MSP using qPCR can also be performed to obtain quantitative rather than qualitative information about methylation It is a variation of the PCR which is used as a diagnostic or cloning technique to identify single-nucleotide polymorphism (SNPs) (single-base differences in DNA). It requires prior knowledge of a DNA sequence and differences between alleles. The allele-specific PCR uses primers whose 30 ends encompass the SNP In a solid phase PCR (SP-PCR), the amplification takes place directly on a solid substrate. The technique encompasses multiple modifications, including polony amplification where PCR colonies are derived in a gel matrix; bridge PCR in which primers are covalently linked to a solid-support surface; conventional solid phase PCR in which an asymmetric PCR is employed. A primer is immobilized on a solid support which has the sequence matching one of the aqueous primers
the CNTs with surface functionalization. Both hydroxylic and carboxylic functionalized CNTs had been studied and had similar enhancing effects of long PCR. The opposite effect on CNTs enhancing PCR had also been reported. Yang et al. [16] studied the effects of multi-walled carbon nanotubes and single-walled carbon nanotubes functionalized with and without carboxylic groups on PCR. Their
References [4]
[5]
[6]
[7]
[8, 9]
[10]
N
[11]
[12, 13]
results showed that CNTs reduced the efficiency of PCR. From their report, the inhibition effect on PCR was increased in the order of CNT-COOH > pristine CNT and SWCNT > MWCNT. The authors attributed the inhibition effect of nanomaterials on PCR to the decrease of the DNA polymerase activity, which they believed to be a significant effect of how nanomaterials impact biological systems.
N
1854
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
Gold Nanoparticles Small gold nanoparticles (Au-NPs) or nanogold had also been studied of their influences on PCR. Hu et al. [17] reported nanogold significantly improved the specificity and yield of PCR. The author attributed the enhancement to the greater affinity of Au-NPs to single-stranded DNA (ssDNA) than to doublestranded DNA (dsDNA) which reduced the mispairing of nucleotides to the template sequence in the elongation step. In their another work [18], the authors did the multiple rounds of PCR with 10 nm Au-NPs. The result showed the Au-NPs effectively enhanced the specificity and yield of error-prone amplification. As the result showed, after six rounds of amplification, the target band was still manifested by a single bright band in gel electrophoresis. However, the target band disappeared after four rounds of amplification of the sample without adding Au-NPs. This study has shown with Au-NPs enhancement, a large amount of target DNA can be produced even with very low copies of original DNA samples. Liu et al. [19] had done the similar study using 13 nm Au-NPs in the PCR reagent and discovered the efficiency was increased. In this study, the authors also compared different PCR systems by using different DNA polymerases, various DNA sizes for amplification, and complex samples. Their results demonstrated that Au-NPs increased the efficiency by fivefold to tenfold in a conventional PCR and by 104-fold in a real-time PCR. The authors hypothesized that the enhancement is due to the excellent heat transfer property of the Au-NPs in PCR solution. Although Au-NPs enhanced PCR had been investigated, different results had been reported. Emslie et al. [20] reported the contrary results from their study. The authors investigated the efficiency of real-time quantitative polymerase chain reaction (qPCR) by adding AuNPs into the PCR solution. It was found that Au-NPs affect the qPCR amplification due to fluorescence quenching by Au-NPs. Rather than the quenching effect, it was discovered that Au-NPs destabilized the double-stranded PCR products which further reduced the efficiency of qPCR. In contrast to those previous reports, Willson et al. [21] discovered that Au-NPs suppressed the amplification of longer products and rather enhanced the amplification of shorter products. In their discovery, when Au-NPs were added in PCR solution, some of the DNA polymerases were nonspecifically absorbed on nanoparticles, and thus reduced the amount of polymerase and the efficiency
of amplification. In their recent report, Hu et al. [22] also discovered that the Au-NPs can enhance the PCR efficiency only by employing native Taq polymerase. No improvement was found in a PCR based on the recombinant Taq polymerase. However, the underlying mechanism was still unclear. Yeow et al. [23] had done the study of the proposed mechanism for Au-NPs affecting the PCR. The authors studied the PCR efficiency with different Au-NPs size (5 nm, 10 nm, and 20 nm) and different concentrations. Evidence from their study showed that the possible mechanism for Au-NPs affecting the PCR is through DNA polymerase-AuNPs binding which modulated the enzyme activity for the amplification. Besides, the binding strongly related to the total surface area of the nanoparticles in the reaction solution. Semiconductor Quantum Dots Another type of nanoparticles such as semiconductor quantum dots (QDs) has been studied in affecting the PCR efficiency. Li et al. [24] reported that the QDs dramatically improved the yield and specificity of PCR. In their study, by employing mercaptoacetic acid (MAA)-coated QDs, yield and specificity of PCR enhanced even at lower annealing temperature. Rather than MAA surface coated QDs, streptavidin surface–modified QDs also effectively improved the specificity of PCR which demonstrated that the effect was not due to the molecules of surface modification but instead due to the QDs itself. Further study is needed to discover the underlying mechanism. Another study using QDs in PCR was done by Xu et al. [25]. In this study, the authors also discovered that the QDs could increase the specificity of the PCR at different annealing temperatures. Thus, as the authors concluded, this method could make the PCR widely applicable, especially in the multi-round PCR with different annealing temperatures. Other Nanoparticles Several different types of nanoparticles have also been studied of their affects in PCR. Mahapatra et al. [26] recently reported that the TiO2 nanoparticles with the size of 25 nm diameter significantly enhanced the PCR efficiency even with various types of templates (i.e., plasmid DNA, genomic DNA, and complementary DNA). Besides, the yield was also enhanced by utilizing TiO2 nanoparticles in the PCR. Investigations of the mechanism by simulations showed
Nanotechnology Applications in Polymerase Chain Reaction (PCR)
evidences of fast and efficient heat transfer in the PCR solution in the presence of TiO2 nanoparticles to enhance the PCR efficiency. Consistent with the simulation results, experimentally, the authors observed the increase of denaturation of genomic DNA in the presence of TiO2 nanoparticles in PCR solution which indicated higher thermal conductivity through the reaction buffer. This technique may be useful for reduction of the overall PCR reaction time and further enhanced the efficiency and specificity of PCR in a variety of samples. The effect of C-60 on PCR was studied by Jiang et al. [27] using a qPCR system. The authors reported a dramatical enhancement of the qPCR at the beginning of the exponential phase but a significant inhibition at the plateau region. In addition, the resulting melting curve demonstrated that C-60 decreased the melting temperatures (Tm) of the template DNA fragments. The authors hypothesized this is due to the enhancement of qPCR efficiency by C-60 in the initial phase of PCR. However, when there is more DNA amplicon in solution, C-60 can disrupt DNA replication by binding to DNA to decrease the access of templates and changing the conformation of DNA templates which reduced the amplification efficiency at the later cycles of the reaction.
Conclusion Although the utilization of nanoparticles in conventional PCR is just emerging, the potentials to enhance the process have been demonstrated by some of the works. Due to its excellent thermal conductivity, high surface-to-volume ratio, and high affinity to bind molecules on surface, nanoparticle-assisted PCR can greatly reduce the total process time and specificity of the process. This is critical for the technique to be used in medical diagnosis and in-field detection applications. In addition, due to the high efficiency of process with NPs enhancement, less amount of expensive reagents is needed which significantly reduce the experimental costs. However, until now, the NP-assisted technology has not yet been revolutionized in practice. To enable users to embrace the technique, future development in identifying type of NPs, size of NPs, concentration of NPs in PCR reagent, and specific PCR components are all essential. In addition, optimization of process design to maximize the
1855
N
enhancements and further reduce the time and cost of the process is also important to evolve this technology out of research and to be used routinely.
Cross-References ▶ Carbon-Nanotubes ▶ Functionalization of Carbon Nanotubes ▶ Gold Nanorods ▶ Nanoparticles
References 1. Mullis, K., et al.: Specific enzymatic amplification of DNA invitro – the polymerase chain-reaction. Cold Spring Harb. Symp. Quant. Biol. 51, 263–273 (1986) 2. Saiki, R.K., et al.: Enzymatic amplification of beta-globin genomic sequences and restriction site analysis for diagnosis of sickle-cell anemia. Science 230(4732), 1350–1354 (1985) 3. Saiki, R.K., et al.: Primer-directed enzymatic amplification of DNA with a thermostable DNA-polymerase. Science 239(4839), 487–491 (1988) 4. Kellogg, D.E., Sninsky, J.J., Kwok, S.: Quantitation of HIV-1 proviral DNA relative to cellular DNA by the polymerase chain-reaction. Anal. Biochem. 189(2), 202–208 (1990) 5. Mackay, I.M., Arden, K.E., Nitsche, A.: Real-time PCR in virology. Nucleic Acids Res. 30(6), 1292–1305 (2002) 6. Stemmer, W.P.C., et al.: Single-step assembly of a gene and entire plasmid from large numbers of oligodeoxyribonucleotides. Gene 164(1), 49–53 (1995) 7. Vincent, M., Xu, Y., Kong, H.M.: Helicase-dependent isothermal DNA amplification. EMBO Rep. 5(8), 795–800 (2004) 8. Chou, Q., et al.: Prevention of pre-PCR mis-priming and primer dimerization improves Low-copy-number amplifications. Nucleic Acids Res. 20(7), 1717–1723 (1992) 9. Kellogg, D.E., et al.: Taqstart antibody(Tm) – Hot start PCR facilitated by a neutralizing neutralizing monoclonalantibody directed against Taq DNA-polymerase. Biotechniques 16(6), 1134–1137 (1994) 10. Herman, J.G., et al.: Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc. Natl. Acad. Sci. U.S.A. 93(18), 9821–9826 (1996) 11. Newton, C.R., et al.: Analysis of any point mutation in DNA – the amplification refractory mutation system (arms). Nucleic Acids Res. 17(7), 2503–2516 (1989) 12. Bing, D.H., Sawosik, T.M., Word, C.J.: Assay performance results with the AmpliType(R) PM PCR amplification and typing kit. Crime Lab. Dig. 23(2), 27–45 (1996) 13. Khan, Z., Poetter, K., Park, D.J.: Enhanced solid phase PCR: mechanisms to increase priming by solid support primers. Anal. Biochem. 375(2), 391–393 (2008) 14. Cui, D.X., et al.: Effects of single-walled carbon nanotubes on the polymerase chain reaction. Nanotechnology 15(1), 154–157 (2004)
N
N
1856
15. Zhang, Z.Z., et al.: Aqueous suspension of carbon nanotubes enhances the specificity of long PCR. Biotechniques 44(4), 537–545 (2008) 16. Yi, C.Q., et al.: Interactions between carbon nanotubes and DNA polymerase and restriction endonucleases. Nanotechnology 18(2), 6 (2007) 17. Li, H.K., et al.: Nanoparticle PCR: nanogold-assisted PCR with enhanced specificity. Angew. Chem.Int. Ed. 44(32), 5100–5103 (2005) 18. Pan, J.K., et al.: Nanogold-assisted multi-round polymerase to chain reaction (PCR). J. Nanosci. Nanotechnol. 7(12), 4428–4433 (2007) 19. Li, M., et al.: Enhancing the efficiency of a PCR using gold nanoparticles. Nucleic Acids Res. 33(21), 10 (2005) 20. Haber, A.L., et al.: Addition of gold nanoparticles to realtime PCR: effect on PCR profile and SYBR Green I fluorescence. Anal. Bioanal. Chem. 392(5), 887–896 (2008) 21. Vu, B.V., Litvinov, D., Willson, R.C.: Gold nanoparticle effects in polymerase chain reaction: favoring of smaller products by polymerase adsorption. Anal. Chem. 80(14), 5462–5467 (2008) 22. Yang, W.C., et al.: Evaluation of gold nanoparticles as the additive in real-time polymerase chain reaction with SYBR Green I dye. Nanotechnology 19(25), 9 (2008) 23. Wan, W.J., Yeow, J.T.W.: The effects of gold nanoparticles with different sizes on polymerase chain reaction efficiency. Nanotechnology 20(32), 5 (2009) 24. Ma, L., et al.: Maximizing specificity and yield of PCR by the quantum dot itself rather than property of the quantum dot surface. Biochimie 91(8), 969–973 (2009) 25. Wang, L.B., et al.: Effects of quantum dots in polymerase chain reaction. J. Phys. Chem. B 113(21), 7637–7641 (2009) 26. Khaliq, A., et al.: Enhancement in the efficiency of polymerase chain reaction by TiO2 nanoparticles: crucial role of enhanced thermal conductivity. Nanotechnology 21(25), 11 (2010) 27. Liang, Y., et al.: C-60 affects DNA replication in vitro by decreasing the melting temperature of DNA templates. Carbon 47(6), 1457–1465 (2009)
Nanotechnology in Cardiovascular Diseases Lu Dai1, Yongfen Qi1, Lixin Jia1, Wei Yu1 and Jie Du2 1 The Key Laboratory of Remodeling-related Cardiovascular Diseases, Capital Medical University, Ministry of Education, Beijing, China 2 Beijing Institute of Heart Lung and Blood Vessel Diseases, Beijing Anzhen Hospital, Beijing, China
Definition Nanomolecular imaging is a new type of imaging with a combination of nanomaterial and molecular biology,
Nanotechnology in Cardiovascular Diseases
using nanoparticles to visualize, characterize, and measure biological processes at the molecular and cellular levels in living systems, including radiotracer imaging/nuclear medicine, magnetic resonance (MR) imaging, MR spectroscopy, optical imaging, ultrasound, and other nanomolecular imaging.
Overview Recently, the emergence of nanomolecular imaging has set the stage for an evolutionary leap in the diagnosis and therapeutics of diseases such as cancer, and neurological and cardiovascular diseases (CVD) [1–3]. CVD is the leading cause of morbidity and mortality in the world. The pathology underpinning CVD is chronic inflammation involving the injury of arterial wall, infiltration leukocytes, and remodeling of vasculature. The main benefits of nanomolecular imaging are derived from combination of cell-type-specific molecular-based, molecular-molecular recognition with nanoparticle-based probes, leading to accurate and high resolution of diseased sites, and to the early diagnosis and treatment for CVDs, including atherosclerosis, thrombosis, and myocardial infarction [4–6]. The progress in nanomolecular imaging will play an important role not only in molecular imaging, but also in the development of biosensors, biomarkers, and new drug delivery system. Nanoparticles are defined as ranging from 5 to 700 nm in diameter and can enter the cells and function inside the cells (such as generation of intracellular imaging and targeting-specific molecular pathways) [4] (see ▶ Nanoparticles). For example, new types of nanoparticles have been recently introduced to generate contrast enhancement and improved sensitivity, such as infrared dyes and quantum dot (QD) nanocrystals for optical imaging [7], gold colloids or radiotracer-based nanoassemblies for X-ray contrast [4], gas-filled microbubbles for contrast using ultrasound [8], and nanoparticles of gadolinium or iron oxide for magnetic resonance contrast [4]. These abilities open up a large number of exciting possibilities for early detection, the potential treatment of disease, and a novel type of drug delivery system. Nanomolecular imaging has also been applied to imaging techniques including luminescence imaging and spectroscopy. Indeed, current imaging device can be applied for nanomolecular imaging, such as positron emission tomography (quantitative-PET) and
Nanotechnology in Cardiovascular Diseases
magnetic resonance spectroscopy (MRS), magnetic resonance spectroscopic imaging (MRSI), diffusion spectroscopy (d-MRI) and functional magnetic resonance (f-MRI).
Key Research Findings Nanomolecular Imaging and Potential Therapy in Cardiovascular Disease CVD is the largest cause of mortality in the world, accounting for about 40% of deaths. CVD occurs at a higher rate in diabetes with mortality exceeding 50%. Risk factors, such as visceral obesity, hypertension, and dyslipidemia, are all associated with increased risk of CVD. The major underlying disease of most CVD is atherosclerosis, characterized by a complex multifactorial pathophysiology such as the slow accumulation of lipid and inflammatory products in the vessel wall that further leads to calcification and fibrosis in atherosclerotic plaques. Inflammation in the vessel wall is now considered to play an essential role in the initiation, progression, and the final steps of atherosclerosis, namely, plaque destabilization and eventually plaque rupture (Fig. 1) [9]. Atherosclerotic plaques can be divided into two types: stable plaque and unstable plaque (a vulnerable plaque, causing myocardial infarction and stroke). Unstable plaques are at higher risk for plaque disruption, characterized by fibrous cap rupture, ulceration following spontaneous denudation of the endothelium, or intraplaque hemorrhage. This plaque disruption leads to local thrombus formation and vessel occlusion, recognized as the proximate cause of the majority of clinical CVD events such as a heart attack or stroke. Although there have been significant advances in vascular imaging (such as intravascular ultrasound and MRI), current imaging methods are still insufficiently capable for elucidating plaque composition. Therefore, it is critical to develop novel techniques to identify vulnerable plaques. In recent years, nanomolecular imaging provides exciting opportunities to identify vulnerable plaque to reduce the risk of acute coronary syndromes. Endothelial Cells
The initiating event for the development of atherosclerosis is the endothelial layer inflammation or dysfunction. Increased adhesiveness of the endothelium leads to the infiltration of leukocytes which could engulf oxidized lipid particles and develop into inflammatory foam cells
1857
N
[10]. These processes can be determined by measuring inflammatory markers, such as vascular cell adhesion molecule-1 (VCAM-1), a molecule found in blood vessel walls at the onset of inflammation – a first stage in the development of atherosclerotic plaques. Kelly et al. identified VCAM-1 internalizing peptide-28 nanoparticles (VINP-28), a novel VCAM-1 peptide affinity ligand from phage display that undergoes internalization and trapping in cells that express VCAM-1, resulting in higher signal amplification ion [11]. That offers an opportunity for detecting early atheromata (fatty streaklike lesions), and targeting VCAM-1-expressing cells of human carotid atherosclerotic plaques. Macrophages
Macrophages have emerged as a key component of atherosclerosis, which accumulate cholesterol esters to form lipid-laden foam cells in the arterial intima and generate proteinases that lead to plaque destabilization and rupture. The accumulation of macrophage in the subendothelial space could be observed through all stages of vascular lesion development, and the increased macrophage density is associated with lesion progression and rupture [9]. Additionally, the macrophage may serve as an ideal target for the detection and potential therapy of vulnerable inflamed lesions. Recently, macrophagespecific lipid-based nanoparticles were created which improved cardiac magnetic resonance detection and characterization of human atherosclerosis [12]. These nanoparticles were made from phospholipids, a surfactant (Tween 80), and an aliphatic gadolinium complex, which targets CD36, a macrophage class B scavenger receptor. The macrophage-specific (CD36) nanoparticles bind human macrophages and improve cardiac magnetic resonance (CMR) detection. A lightactivated theranostic nanoagent for targeted macrophage ablation was invented [13]. This nanoparticle was based upon cross-linked dextran-coated iron oxide (CLIO). It was modified with near-infrared fluorophores and lightactivated therapeutic moieties. The therapeutic component was activated at 650 nm and could result in eradication of inflammatory macrophages. Moreover, other potential targets, such as scavenger receptor class B type 1 (SRB1), Lectin-like oxidized LDL receptor-1 (LOX-1), Receptor for Advanced Glycation Endproducts (RAGE), and urokinase-type plasminogen activator (uPA) receptor, could be more efficacious and provide an integrated imaging and therapeutic nanoplatform for atherosclerosis [14–17].
N
N
1858
Nanotechnology in Cardiovascular Diseases Fibrin
Macrophage
a v b 3 integrin
VCAM-1 Endothelial cell CD36
Lumen
Plaque
Intima
Media Smooth muscle cell
Nanotechnology in Cardiovascular Diseases, Fig. 1 The structure of plaque in atherosclerosis. VCAM-1 vascular cell adhesion molecule-1
Atherosclerotic Plaque
A major component of atherosclerosis plaques is fibrin, which forms the fibrils that entangle the blood platelets and seal up the rent in the artery wall. Fibrin deposition is one of the earliest signs of plaque rupture or erosion and intraplaque hemorrhage. Densely packed fibrintargeted nanoparticles can be detected by scanning electron micrographs (SEM) [18]. These nanoparticles were formulated with gadolinium DTPA-bis-oleate (Gd-DTPA-BOA) and presented a highly detectable, homogeneous T1-weighted contrast enhancement that improved with increasing gadolinium level. Higherresolution scans and scanning electron microscopy revealed that the nanoparticles were present as a thin layer over the clot surface. Then, the development of other matrix-targeted nanoparticles could provide accurate detection and localization of matrix deposits and may allow early, direct identification of vulnerable plaques. Angiogenesis
In atherosclerosis, angiogenesis is not only related to plaque growth, but also related to plaque instability and rupture. Integrins, cell surface receptors involved in cell–cell and cell–matrix interaction, such as avb3, are associated with angiogenesis and have been extensively used as a target for nanomolecular imaging [19]. Patrick Winter et al. developed a paramagnetic nanoparticle contrast agent targeted specifically to avb3 integrins to permit noninvasive molecular imaging of plaque-associated angiogenesis. Moreover, avb3 integrins-targeted
paramagnetic nanoparticles can deliver fumagillin and elicit a marked antiangiogenic response with a minimal drug dosage [19].
Future Directions for Research Nanomolecular imaging is a new evolving field, which represents the beginning of a revolutionary new age for the clinicians to enhance the diagnosis and treatment of the cardiovascular diseases relying on the powerful imaging probes and hardware. While it is still in the developing phase, the nanomolecular imaging has resulted in lots of nanoparticles that vary in size, shape, change, chemistry, coating, and solubility. More importantly, nanomolecular imaging must utilize the progress of molecular and cellular biology in pathological conditions. At the same time, the potential toxicity of these nanoparticles should also be paid more attention to [20] (see ▶ Cellular Mechanisms of Nanoparticle’s Toxicity and ▶ Nanoparticle Cytotoxicity). The nano-imaging agent synthesis requires the efforts on their subsequent translation to clinical practice. The rapid growth of nanomolecular imaging will change the field of imaging and imaging-guided interventions for cardiovascular diseases.
Cross-References ▶ Cellular Mechanisms of Nanoparticle’s Toxicity ▶ Nanoparticle Cytotoxicity ▶ Nanoparticles
Nanotribology
References 1. Ferrari, M.: Cancer nanotechnology: opportunities and challenges. Nat. Rev. Cancer 5, 161–171 (2005) 2. Bhaskar, S., et al.: Multifunctional nanocarriers for diagnostics, drug delivery and targeted treatment across blood– brain barrier: perspectives on tracking and neuroimaging. Part. Fibre Toxicol. 7, 3 (2010) 3. Yang, X.: Nano- and microparticle-based imaging of cardiovascular interventions: overview. Radiology 243, 340–347 (2007) 4. Wickline, S.A., Neubauer, A.M., Winter, P., Caruthers, S., Lanza, G.: Applications of nanotechnology to atherosclerosis, thrombosis, and vascular biology. Arterioscler. Thromb. Vasc. Biol. 26, 435–441 (2006) 5. Godin, B., et al.: Emerging applications of nanomedicine for the diagnosis and treatment of cardiovascular diseases. Trends Pharmacol. Sci. 31, 199–205 (2010) 6. Douma, K., et al.: Nanoparticles for optical molecular imaging of atherosclerosis. Small 5, 544–557 (2009) 7. Stinaff, E.A., et al.: Optical signatures of coupled quantum dots. Science 311, 636–639 (2006) 8. Anderson, D.R., Tsutsui, J.M., Xie, F., Radio, S.J., Porter, T.R.: The role of complement in the adherence of microbubbles to dysfunctional arterial endothelium and atherosclerotic plaque. Cardiovasc. Res. 73, 597–606 (2007) 9. Tabas, I.: Macrophage death and defective inflammation resolution in atherosclerosis. Nat. Rev. Immunol. 10, 36–46 (2010) 10. Sitia, S., et al.: From endothelial dysfunction to atherosclerosis. Autoimmun. Rev. 9, 830–834 (2010) 11. Nahrendorf, M., et al.: Noninvasive vascular cell adhesion molecule-1 imaging identifies inflammatory activation of cells in atherosclerosis. Circulation 114, 1504–1511 (2006) 12. Lipinski, M.J., et al.: Macrophage-specific lipid-based nanoparticles improve cardiac magnetic resonance detection and characterization of human atherosclerosis. JACC Cardiovasc. Imaging 2, 637–647 (2009) 13. McCarthy, J.R., Korngold, E., Weissleder, R., Jaffer, F.A.: A light-activated theranostic nanoagent for targeted macrophage ablation in inflammatory atherosclerosis. Small 6, 2041–2049 (2010) 14. Stephen, S.L., et al.: Scavenger receptors and their potential as therapeutic targets in the treatment of cardiovascular disease. Int. J. Hypertens. 2010, 646929 (2010) 15. Farmer, D.G., Kennedy, S.: RAGE, vascular tone and vascular disease. Pharmacol. Ther. 124, 185–194 (2009) 16. Oka, H., et al.: Lysophosphatidylcholine induces urokinasetype plasminogen activator and its receptor in human macrophages partly through redox-sensitive pathway. Arterioscler. Thromb. Vasc. Biol. 20, 244–250 (2000) 17. White, S.J., Sala-Newby, G.B., Newby, A.C.: Overexpression of scavenger receptor LOX-1 in endothelial cells promotes atherogenesis in the ApoE(/) mouse model. Cardiovasc. Pathol. (2010) 18. Flacke, S., et al.: Novel MRI contrast agent for molecular imaging of fibrin: implications for detecting vulnerable plaques. Circulation 104, 1280–1285 (2001) 19. Winter, P.M., et al.: Endothelial alpha(v)beta3 integrintargeted fumagillin nanoparticles inhibit angiogenesis in
1859
N
atherosclerosis. Arterioscler. Thromb. Vasc. Biol. 26, 2103–2109 (2006) 20. Stone, V., Donaldson, K.: Nanotoxicology: signs of stress. Nat. Nanotechnol. 1, 23–24 (2006)
Nanotechnology in Dental Medicine ▶ Nanodentistry
Nanothermodynamics ▶ Surface Energy Nanoscale
and
Chemical
Potential
at
Nanotoxicology ▶ Physicochemical Properties of Nanoparticles in Relation with Toxicity
Nanotransfer Printing ▶ Nanoscale Printing
Nanotribology Bharat Bhushan Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA
Synonyms Adhesion; Friction; Lubrication; Wear
Definition The word “Tribology” was coined in 1966. It is derived from the Greek word tribos, meaning rubbing so the
N
N
1860
literal translation would be the science of rubbing. However, it encompasses studies of surface characterization, adhesion, friction, wear, and lubrication. According to Webster dictionary, Tribology is defined as the science and technology of interacting surfaces in relative motion and of the related subjects and practices. Tribology is an interdisciplinary field. Rather complex surface interactions in a tribological interface require knowledge of various disciplines including physics, chemistry, mechanical engineering, solid mechanics, materials science, rheology, applied mathematics, and reliability. At most interfaces of technological relevance, contact occurs at numerous asperities. It is of importance to investigate a single asperity contact in the fundamental tribological studies and thus emphasizing the importance of nanotribology.
Introduction The mechanisms and dynamics of the interactions of two contacting solids during relative motion, ranging from atomic- to microscale, need to be understood in order to develop fundamental understanding of adhesion, friction, wear, indentation, and lubrication processes. For most solid–solid interfaces of technological relevance, contact occurs at multiple asperities. Consequently, the importance of investigating single asperity contacts in studies of the fundamental micro/nanomechanical and micro/nanotribological properties of surfaces and interfaces has long been recognized. The recent emergence and proliferation of proximal probes, in particular scanning probe microscopes (the scanning tunneling microscope and the atomic force microscope), the surface force apparatus, and of computational techniques for simulating tip-surface interactions and interfacial properties, has allowed systematic investigations of interfacial problems with high resolution as well as ways and means for modifying and manipulating nanoscale structures. These advances have led to the appearance of the new field of nanotribology, which pertains to experimental and theoretical investigations of interfacial processes on scales ranging from the atomic- and molecular-scale to the microscale, occurring during adhesion, friction, scratching, wear, indentation, and thin-film lubrication at sliding surfaces [2, 4–8, 10–12, 15, 18, 22, 25]. Proximal probes have also been used for mechanical and electrical
Nanotribology
characterization, in situ characterization of local deformation, and other nanomechanics studies. Nanotribological and nanomechanics studies are needed to develop fundamental understanding of interfacial phenomena on a small scale and to study interfacial phenomena in micro/nanostructures used in magnetic storage devices, nanotechnology, and other applications [1–12]. Friction and wear of lightly loaded micro/nanocomponents are highly dependent on the surface interactions (few atomic layers). These structures are generally coated with molecularly thin films. Nanotribological and nanomechanics studies are also valuable in the fundamental understanding of interfacial phenomena in macrostructures, and provide a bridge between science and engineering. The surface force apparatus (SFA), the scanning tunneling microscopes (STM), atomic force and friction force microscopes (AFM and FFM) are widely used in nanotribological and nanomechanics studies. Typical operating parameters are compared in Table 1. The SFA was developed in 1968 and is commonly employed to study both static and dynamic properties of molecularly thin films sandwiched between two molecularly smooth surfaces. The STM, developed in 1981, allows imaging of electrically conducting surfaces with atomic resolution, and has been used for imaging of clean surfaces as well as of lubricant molecules. The introduction of the AFM in 1985 provided a method for measuring ultra-small forces between a probe tip and an engineering (electrically conducting or insulating) surface, and has been used for morphological and surface roughness measurements of surfaces on the nanoscale, as well as for adhesion measurements. Subsequent modifications of the AFM led to the development of the FFM, designed for atomic-scale and microscale studies of friction. This instrument measures forces in the scanning direction. The AFM is also being used for various investigations including scratching, wear, indentation, detection of transfer of material, boundary lubrication, and fabrication and machining. Meanwhile, significant progress in understanding the fundamental nature of bonding and interactions in materials, combined with advances in computer-based modeling and simulation methods, has allowed theoretical studies of complex interfacial phenomena with high resolution in space and time. Such simulations provide insights into atomic-scale energetics, structure, dynamics, thermodynamics, transport, and rheological aspects of tribological processes.
Nanotribology
1861
N
Nanotribology, Table 1 Comparison of typical operating parameters in SFA, STM and AFM/FFM used for micro/ nanotribological studies Operating parameter Radius of mating surface/tip Radius of contact area Normal load Sliding velocity
Sample limitations
STMa 5–100 nm
SFA 10 mm 10–40 mm 10–100 mN 0.001–100 mm/s
AFM/FFM 5–100 nm
N/A N/A 0.02–200 mm/s (scan size 1 nm 1 nm to 125 mm 125 mm; scan rate VGS VT IDS ¼
mFE Co W ðVGS VT Þ2 2L
(2)
where W and L are the channel width and length, Co is the gate dielectric capacitance, VT is the threshold voltage, VDS is the drain-source voltage, VGS is the gate-source voltage, and mEF is the field effect mobility that can be extracted from:
mEF
pffiffiffiffiffiffiffiffiffi 2 @IDS 2L pffiffiffiffiffiffiffiffiffiffiffi ¼ WCo @VGS
(3)
Figure 3 shows the molecular structure of pentacene, a well-known p-type organic material, which is currently under extensive investigation for use in TFTs as the active layer because of its highest field effect mobility among all organic semiconductors reported to date. Since the carrier mobility strongly depends on the p-orbital overlap of neighboring molecules, the ordering of the organic materials controlled
by relatively weak Van der Waals interactions is the key to improvement in the mobility. It has been well known that the formation of highly ordered pentacene films resulting in large grains is significantly affected by both growth conditions and substrate surfaces such as deposition rate, substrate temperature, substrate morphology, and chemical properties of the substrate. Since the nucleation of the first monolayer is extremely sensitive to defects and/or impurities on the substrate, careful treatment of the substrate surface is strictly required to obtain large as well as highly ordered grains. Grain boundaries associated with trap sites act as carrier barriers that dominate charge transport by trapping at the boundaries. This intergrain transport is mostly responsible for the low mobility of the devices. Thus, the investigation of close correlation between the morphology and the structural properties of organic semiconductors should be first required to obtain the most advantageous structures and morphological characteristics for improved performance of organic devices.
Organic Sensors Over the last decade, micro-fabrication technologies have been extensively applied to yield a miniaturized biochemical and physical sensor in order to monitor various parameters such as ion diffusion, temperature, strain, and electrolyte conductivity leading to the
Organic Sensors
2035
O
Organic Sensors, Fig. 3 A molecular structure of pentacene
C
H
successful development of multi-parameter measuring systems including electronic nose/tongue, micro total analysis system (mTAS), and lab-on-chip. It is obvious that microsensors based on the principle of the field effect and solid state transport in semiconductor structures offer a number of advantages over traditional physiobiological sensors with regard to real-time electronic readout, small size, and rapid response. In the last several years, much research effort has been devoted to the use of OTFTs as multi-parameter sensors, pursuing good levels of selectivity, reliability, and reproducibility at low cost. This is due to the fact that OTFT sensors can provide a variety of information from changes in multiple properties on exposure to biochemical as well as physical quantities, such as the bulk conductivity of an organic thin film, the saturation current, the threshold voltage, and the field effect mobility. It is obvious that the multifunctionality in such a multi-parameter sensor operating in different modes enables both envisioning and realization of a simple “higher order” sensor network, which leads to low-cost manufacturing and easy commercialization. In addition, the inherent characteristics of field effect devices such as signal amplification are expected to provide transistor-based sensors with enhanced detection limits and sensitivity compared to usual chemiresistor, amperometric, or potentiometric sensors. Recently, the possibility of using OTFTs for sensing devices has been demonstrated by many research groups. In this case, the conductivity of the organic semiconductor is modulated due to the direct interaction with gas, pressure, humidity, and strain. As one of the promising applications of organic electronics, large-area strain sensors integrated with transistor matrix capable of monitoring shape, strain, force, and/or pressure in a subject material have been
envisioned for a broad spectrum ranging from microsurgery, point-of-care (POC) diagnostics, and monitoring of respiration rate to arrays such as flexible artificial skins of robots and structural health monitoring systems for aging aircrafts. Especially, the key to POC diagnostics is the development of small, light, portable, and implantable sensor devices that are capable of monitoring and controlling diseases in real time so that health-care professionals can make rapid and accurate diagnosis of diseases. This will ensure the effectiveness of therapy along with early detection. Thus, there has been high demand at the POC for development of biochemical and physical microsensors that possess good portability, high sensitivity, lowpower dissipation, and compact device integration. It is obvious that such flexible strain sensors will help construct a POC system for continuous health monitoring when integrated into clothing to provide patients with a safe and comfortable environment for home health care, illness prevention, and citizen medicine. The system monitors various physiological vital parameters measured from the skin including thermal, electrical, geometrical, and mechanical parameters to enable early detection, which help suppress the occurrence of acute events and complications leading to hospitalization and extended hospital treatment. From geometrical and mechanical parameters, it is expected that strain sensors constituting a wearable monitoring system are capable of collecting health data such as respiration rate and amplitude, heart rate, blood pressures, and detection of falls. Strains sensors, usually bonded to the structures to be tested, measure the change in the resistance of conductors as directly proportional to mechanical motion (strain) experienced by the sensor. Inorganic semiconductors such as amorphous and
O
O
2036
microcrystalline silicon have recently been employed as the strain-sensing element and fabricated on a flexible film to overcome the limitation of conventional strain sensors based on metallic foils and rigid piezoresistive semiconductors [12, 13]. However, the high Young’s modulus (200 GPa) of the inorganic materials limits the use for applications with large flexible polymeric substrates due to the large stiffness mismatch that is generated in the interface of the inorganic semiconductor elements and the flexible films. This may lead to irreversible plastic substrate deformations, degrading the sensor performance in terms of reliability and repeatability. Thus, the use of organic materials with low Young’s modulus (5 GPa) as the sensing element is expected to minimize the induced stress concentration. Ji et al. first developed organic flexible strain sensors that employed a thin-film transistor-like structure in conjunction with a Wheatstone bridge configuration, as shown in Fig. 4 [14]. The conductivity of the sensor is field controllable by the bias applied to the gate electrode, which tailors the sensor’s sensitivity. This strain sensor consisted of five electrodes, one of which was the gate (G) electrode capable of controlling the field-induced conducting layer of pentacene to turn the sensor on and off, thereby tailoring the sensor sensitivity. This pseudo-TFT sensor operated in a depletion mode (normally on) in which it exhibited typical field effect transistor characteristics in a gate voltage range of 40 to 60 V, revealing a good saturation behavior at high drain current. Pressure sensors, consisting of OTFTs and pressure-sensitive rubber developed by Someya et al. [15], provide the possibility of realizing a practical artificial skin thanks to the inherent flexibility and low-cost processing of organic circuits. A variation of drain current is measured under the application of different pressures from 0 to 30 kPa. Subramanian et al. recently developed printed RFID electronics including passive components (inductors and capacitors) and organic transistors for item-level tracking applications [16]. Zhu et al. [17] proposed a humidity sensor based on OTFTs that can be envisioned to monitor relative humidity in moisture-sensitive environments such as use in semiconductor manufacturing and packaging, medical, pharmaceutical, and food manufacturing. Humidity can affect the charge transport properties of organic active layers in many ways. When H2O
Organic Sensors
T1 Perpendicular σ T4
T2 Diagonal Parallel I G
T3
Organic Sensors, Fig. 4 Field controllable flexible strain sensors employing a transistor-like structure in conjunction with a Wheatstone bridge configuration. T1and T3: input bias, T2 and T4: bridge output, G: gate, and J: input current flow direction. Strains (s) were applied either parallel (0 ), diagonal (45 ), or perpendicular (90 ) to the direction of the input current flow (J)
molecules diffuse into the crevices of polycrystalline organic materials and react with the trapped carriers in there while changing their morphology, they will alternate the electric field at the grain boundaries, which decrease the rate of charge transport by increasing the amount of energetic disorder through charge–dipole interactions. All these factors will affect the carrier mobility, saturation current, on/off ratio, threshold voltage, and so on. They measured the sensor output with respect to time dependence of the saturation current (IDS) when the OTFT was alternatively exposed to vacuum and N2 gas with different levels of relative humidity (RH). It turned out that the saturation current of the OTFT drops dramatically by nearly one order of magnitude when exposed to 70% RH. Ion-sensitive field effect transistors (ISFET) using OTFTs, as shown in Fig. 5, have been used to detect ions where the gate insulator is exposed to an electrolyte containing the analyte. The source-drain current is modulated by the electric field across the insulating gate dielectric which is controlled by ions at the electrolyte–insulator interface. Bartic et al. presented OTFTs fabricated with poly (3 hexylthiphene) regioregular (P3HT) as the semiconductor layer to be used as a pH sensor that can detect both charged and uncharged chemical species in aqueous media, as shown in Fig. 6 [18]. This pH sensor detects pH
Organic Sensors
2037
Organic Sensors, Fig. 5 Schematic cross section of an ion-sensitive organic field effect transistor where the insulator is exposed to an electrolyte containing the analyte, and the gate electrode is separated from the insulator by the electrolyte. The electric field across the insulating gate dielectric is controlled by ions at the electrolyte–insulator surface
Source
Drain Organic semiconductor
Insulator
Electrolyte
Drain Support
D
IDS S
P3HT
P3HT
IG
Ta2O5 Source
RE (gate)
Ia2O5
b Sensing area “O”-ring
Si rubber
Gate
a Silicon rubber
O
VGS
Organic Sensors, Fig. 6 (a) Schematic of the device structure used for pH sensor, and (b) schematic of the measurement setup [18]. This pH sensor detects Ph variations occurring in the bulk
of the solution through field-enhanced conductivity of the organic semiconductor using Ta2O5 material that possesses a number of proton-sensitive surface sites
variations occurring in the bulk of the solution through field-enhanced conductivity of the organic semiconductor using Ta2O5 material that possesses a number of proton-sensitive surface sites. They also realized a glucose biosensor with the same structure by anchoring an enzymatic layer of glucose oxidase (GOx) on the dielectric of the organic transistor where the biocatalytic reaction between glucose and GOx modifies the local concentration of proton on the dielectric surface leading to the variation in drain current. Crone et al. proved that a variety of organic semiconducting materials being used in OTFTs can offer advantages to develop sensor arrays for selective gas detections [19]. Eleven different organic sensor materials based TFTs were exposed to 16 various gaseous analytes including alcohol, ketones, thions, and ring compounds while measuring the source-drain current of the TFTs which gave rise to selective patterns associated with each analyte due to different semiconductor-analyte pair response mechanisms. Crone et al. have also pointed out the fact that the OTFT sensor can be readily integrated into a circuit system with
superior performance to discrete chemiresistor sensors. For instance, the ring oscillator sensor they developed exhibited a higher sensitivity to octal vapor with a faster recovery time than discrete transistor sensor [20].
Future Directions for Research The tremendous progress in the performance of organic electronics, including OTFTs, opens wide horizons for the use of organic devices as compact sensing systems. However, developments in sensors using organic semiconductors still remain at the laboratory stage. It can be predicted that the impact of organic electronics on sensing applications will be dependent on the stability of the sensors in a real environment, the ability to design a new structure of sensors that is suitable for the integration of sensor arrays, and the cost of the sensors themselves, which determines whether this technology will move from the research state to potentially commercial phases.
O
O
2038
Cross-References ▶ Microfludic Whole-cell Biosensor ▶ CMOS MEMS Biosensors ▶ Flexible Electronics ▶ Nanostructure Field Effect Transistor Biosensors ▶ Nanostructured Materials for Sensing ▶ NEMS Resonant Chemical Sensors ▶ Organic Actuators ▶ Organic Bioelectronics ▶ Organic Photovoltaics: Basic Concepts and Device Physics
References 1. Pope, M., Swenberg, C.E.: Electronic Processes in Organic Crystals and Polymers, 2nd edn. Oxford University Press, New York (1999) 2. Chamberlain, G.A.: Organic solar cells: a review. Sol. Cell 8, 47–83 (1983) 3. Horowitz, G., Hajlaoui, R., Delannoy, P.: Temperature dependence of the field effect mobility of sexithiophene. Determination of the density of traps. J. Phys. III 5, 355–371 (1995) 4. Vissenberg, M.C.J.M., Matters, M.: Theory of the field effect mobility in amorphous organic transistors. Phys. Rev. B 57, 12964–12967 (1998) 5. Mattheus, C.C., Dros, A.B., Baas, J., Oostergetel, G.T., Meetsma, A., de Boer, J.L., Palstra, T.T.M.: Identification of polymorphs of pentacene. Synth. Met. 138, 475–481 (2003) 6. Lilienfeld, J.E.: Device for controlling electric current. US Patent 1,900,018, 7 March 1933 7. Weimer, P.K.: The TFT-new thin-film transistor. Proc. IRE 50, 1462–1465 (1962) 8. Spear, W.E., LeComber, P.: Substitutional doping of amorphous silicon. Solid State Commun. 17, 1193–1196 (1975) 9. LeComber, P.G., Spear, W.E., Gaith, A.: Amorphoussilicon field effect device and possible application. Electron. Lett. 15, 179–181 (1979) 10. Ebisawa, F., Kurokawa, T., Nara, S.: Electrical properties of polyacetylene/polysiloxane interface. J. Appl. Phys. 54, 3255–3259 (1983) 11. Mitzi, D.B., Chondroudis, K., Kagan, C.R.: Organicinorganic electronics. IBM J. Res. Dev. 45, 29–46 (2001) 12. Caputo, D., Cesare, G., Gavesi, M., Palma, F.: a-Si:H alloy for stress sensor application. J. NonCryst. Solid 338–340, 725–728 (2004) 13. Zhou, L., Jung, S., Brandon, E., Jackson, T.N.: Flexible substrate micro-crystalline silicon and gated amorphous silicon strain sensors. IEEE Trans. Electron. Dev. 53, 380–385 (2006) 14. Ji, T., Jung, S., Varadan, V.K.: Field-controllable flexible strain sensors using pentacene semiconductors. IEEE Electron Dev. Lett. 28, 1105–1107 (2007)
Organic Solar Cells 15. Takao, S., Tsuyoshi, S., Shingo, I., Yusaku, K., Hiroshi, K., Takayasu, S.: A large-area, flexible pressure sensor matrix with organic field-effect transistors for artificial skin applications. Proc. Natl Acad. Sci. U.S.A. 101, 9966–9970 (2004) 16. Subramanian, V., Chang, P.C., Lee, J.B., Molesa, S.E., Volkman, S.K.: Printed organic transistors for ultra-lowcost RFID application. IEEE Trans. Compon. Packag. Technol. 28, 742–747 (2005) 17. Zhu, Z.T., Mason, J.T., Diekmann, R., Malliaras, G.G.: Humidity sensors based on pentacene thin-film transistors. Appl. Phys. Lett. 24, 4643–4645 (2002) 18. Bartic, C., Campitelli, A., Borghs, S.: Field-effect detection of chemical species with hybrid organic/inorganic transistors. Appl. Phys. Lett. 82, 475–477 (2003) 19. Crone, B., Dodabalapur, A., Gelperin, A., Torsi, L., Katz, H. E., Lovinger, A.J., Bao, Z.: Electronic sensing of vapors with organic transistors. Appl. Phys. Lett. 78, 2229–2231 (2001) 20. Crone, B., Dodabalapur, A., Sarpeshkar, R., Gelperin, A., Katz, H.E., Bao, Z.: Organic oscillator and adaptive amplifier circuits for chemical vapor sensing. J. Appl. Phys. 91, 10140–10146 (2002)
Organic Solar Cells ▶ Nanomaterials for Excitonic Solar Cells ▶ Organic Photovoltaics: Basic Concepts and Device Physics
Organic–Inorganic Heterojunctions ▶ Hybrid Solar Cells
Organic–Inorganic Photovoltaics ▶ Hybrid Solar Cells
Organosilanes ▶ Nanostructures for Surface Functionalization and Surface Properties
Orientation Sensors ▶ Gyroscopes
P
Passive Pumping
Peltier Couple
▶ Capillary Flow
▶ Thermoelectric Heat Convertors
Patterned Hydrogels
Peltier Effect ▶ Thermoelectric Heat Convertors
▶ Nanoengineered Hydrogels for Cell Engineering
Peltier Element Patterning ▶ Nanostructures for Surface Functionalization and Surface Properties
▶ Thermoelectric Heat Convertors
Peltier Module ▶ Thermoelectric Heat Convertors
Peltier Coefficient ▶ Thermoelectric Heat Convertors
Peltier-Seebeck Effect ▶ Thermoelectric Heat Convertors
Peltier Cooler
Percutaneous Absorption
▶ Thermoelectric Heat Convertors
▶ Dermal and Transdermal Delivery
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
P
2040
Perfluorocarbon Nanoparticles Patrick M. Winter1, Gregory M. Lanza2 and Samuel A. Wickline2 1 Department of Radiology, Imaging Research Center, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA 2 C-TRAIN Labs, Washington University, St. Louis, MO, USA
Synonyms PFC nanoparticles; PFOB nanoparticles
Definition Perfluorocarbon (PFC) nanoparticles represent a multifunctional platform consisting of a PFC core encapsulated within a phospholipid monolayer. The particles have been developed for magnetic resonance (MR) molecular imaging to noninvasively map the expression of important biomarkers of disease.
Introduction Perfluorocarbon (PFC) nanoparticles consist of a liquid PFC core encapsulated by a phospholipid monolayer. PFCs represent a family of modified hydrocarbon molecules in which the hydrogens are replaced with fluorine atoms [1–3]. The carbon–fluorine bond is very stable and biologically inert due to the dense electron cloud, which prevents chemical reactions with other compounds [2]. PFCs with molecular weights between 460 and 520 are biocompatible, showing no toxicity, carcinogenicity, mutagenicity, or teratogenic effects. PFCs have tissue halflife residencies ranging from 4 to 65 days. They are not metabolized. Instead, they are reintroduced into the blood circulation from clearance organs, such as the liver, by lipid carriers and are exhaled via the lungs [2]. PFC nanoparticles have been formulated for a wide range of molecular imaging and targeted drug delivery applications. The particles are typically 250 nm in diameter, which is large enough to constrain them to the intravascular space. As a result, PFC
Perfluorocarbon Nanoparticles
particles are often targeted to intraluminal biomarkers expressed by endothelial cells, such as integrins, selectins, or adhesion molecules [2, 3]. Magnetic resonance (MR) molecular imaging aims to identify and map the expression of important biomarkers on a cellular scale. Typical MR imaging techniques lack sufficient resolution and sensitivity to directly detect these molecular signatures due to their very low concentrations in vivo. To overcome this limitation, nanoparticle agents are often employed to specifically bind to the biomarker of interest, accumulate at the site of pathology, and generate sufficient image contrast. Nanoparticles offer a large surface area for the incorporation of multiple binding ligands (to improve targeting efficacy) and multiple paramagnetic chelates (to amplify the signal enhancement). In some cases, the nanoparticle may also serve as a drug delivery agent, providing diagnostic and therapeutic information via noninvasive MR imaging. A number of other medical imaging modalities are capable of molecular imaging, including positron emission tomography (PET), single photon emission computed tomography (SPECT), and optical imaging [4]. Like MR imaging, these modalities typically rely on specifically designed contrast agents that bind to the molecular biomarker of interest. For PET and SPECT imaging, the contrast agents consist of a radioactive element for detection. In optical imaging, the contrast agent will utilize fluorescence or bioluminescence for mapping expression of cellular receptors. MR imaging, however, has certain important advantages over these other modalities for molecular imaging applications. MR imaging offers higher spatial and temporal resolution than PET or SPECT and provides greater tissue penetration than optical imaging. This allows MR imaging to combine anatomical and/or functional information with molecular imaging, much like the multimodality benefits of PET/CT. In addition, MR imaging may be better suited for serially tracking disease progression or therapeutic response compared to nuclear imaging due to concerns of cumulative radiation dose. Molecular imaging with PFC nanoparticles has been applied to a wide array of disease biomarkers, therapeutic drugs, and imaging modalities [2, 3]. PFC nanoparticle formulations have been developed for ultrasound imaging [5], computed tomography [6], nuclear imaging [7], optical imaging [8–10], MR [3], as well as multi-modality applications. For example, combined 99mTc SPECT and MR
Perfluorocarbon Nanoparticles
imaging of anb3-integrin-targeted PFC nanoparticles has been used to map tumor angiogenesis with high sensitivity and high resolution [7]. The anb3-integrin is a heterodimeric transmembrane glycoprotein expressed by numerous cell types, including endothelial cells, macrophages, platelets, lymphocytes, smooth muscle cells, and tumor cells [2]. Endothelial expression of anb3-integrin is required for the migration and invasion of vascular endothelial cells, and therefore, serves as a biomarker of angiogenesis [11]. Since PFC nanoparticles are constrained to the vasculature, they are not exposed to nonendothelial integrin-expressing cells, which improves binding specificity to the neovasculature [2]. Nuclear imaging offers very high sensitivity, which is preferred for the detection of occult tumors or metastases within large tissue regions. However, the limited spatial resolution of nuclear imaging precludes detailed characterization of the pathology. Incorporating a 99mTc nuclear imaging agent onto anb3-integrin-targeted gadolinium nanoparticles yielded a high-sensitivity nuclear signal that was used to guide further high-resolution MR imaging and characterization of tumor angiogenesis [7]. MR molecular imaging demonstrated a coherent, asymmetric, patchy pattern of angiogenesis along the outer aspects of the tumor mass, consistent with an angiogenic phenotype. In the case of ultrasound imaging, PFC nanoparticles represent a significant departure from conventional contrast agent design. Most ultrasound contrast agents consist of gas bubbles to maximize the backscatter of ultrasound waves and thereby increase the image contrast. However, these bubbles are quite large (more than 1 mm in diameter) and are quickly entrapped within the lungs. This severely limits the practical application of bubbles in molecular imaging because they do not circulate long enough to bind to biomarkers of disease and accumulate at the sites of pathology. PFC nanoparticles, however, have a long circulatory half-life and are inherently echogenic due the differences in the speed of sound between water and the PFC core. This ultrasound backscatter is typically not sufficient to visualize particles in the blood stream. Instead, the particles must accumulate at the target site, forming an “acoustic mirror” to generate significant image contrast. Targeted PFC nanoparticles have been used for molecular imaging of fibrin in acute arterial thrombus, high-frequency intravascular imaging, and imaging with novel ultrasound receivers, called information theoretical detectors [5].
2041
P
Along with MR molecular imaging, PFC nanoparticles have been employed to label cells and noninvasively track implantation and migration of cell therapies [1, 3]. While MR molecular imaging with PFC nanoparticles typically utilizes paramagnetic chelates for detection with 1H MR, tracking of cells labeled with PFC nanoparticles often relies on 19F MR. 19F has a relatively high sensitivity, 83% compared to 1H, and there is virtually no native background signal. Thus, the PFC core can provide a definitive and quantitative MR signature [1]. PFC nanoparticles can be internalized by stem cells without the need of a transfection agent and 19F MR imaging can quantitatively track the cells in vivo [1]. The high sensitivity of 19F MR allows the in vivo detection of as few as 2,000 labeled cells. For some applications, modification of the surfactant surface of the particles improves cellular uptake. For example, PFC nanoparticles formulated with cationic phospholipids on the particle surface can be effectively internalized by neural stem cells [1].
Molecular Imaging of Therapeutic Response One potential clinical application of MR molecular imaging with PFC nanoparticles is to serially map biomarkers of disease progression and/or therapeutic response. Noninvasive imaging could be used to monitor the biochemical signatures of disease, providing earlier and more specific detection of pathology compared to tracking symptomology or the occurrence of clinical events. For example, molecular imaging of angiogenesis can serve as a biomarker of atherosclerosis and can be used to monitor the physiological effects of a range of therapeutic treatments [3]. Paramagnetic anb3-integrin-targeted PFC nanoparticles have been utilized to characterize angiogenesis in the aortic wall of an atherosclerotic rat model and to monitor the effects of benfluorex (an appetite suppressant) on aortic angiogenesis [12]. The JCR:LA-cp rat is a model of metabolic syndrome characterized by obesity, insulin resistance, hyperlipidemia, and vasculopathy. The obese rats displayed higher food consumption and body weight, as well as increased insulin, leptin, cholesterol, and triglyceride levels compared to the lean controls. In conjunction with this typical metabolic syndrome phenotype, MR
P
P
2042
Perfluorocarbon Nanoparticles Full FOV
Bsl
2 Hr
Enh. Map
Lean
150 %
Obese
100 %
50 % Obese Treated 0%
Perfluorocarbon Nanoparticles, Fig. 1 Full field-of-view (“Full FOV”) images of JCR rats showing large abdominal fat deposits in the obese (middle) and obese-treated (bottom) animals that were absent in the lean (top) rat (arrows denote aortas). Magnified views of the abdominal aorta in lean, obese, and obese-treated rats at baseline (“Bsl”) and 2-h (“2 Hr”) post
injection of targeted nanoparticles displaying signal enhancement in the aorta wall. Percent signal enhancement maps (“Enh. Map”, false-colored from blue to red) demonstrate higher MR signal enhancement in the untreated obese animal, indicating active angiogenesis supporting development of atherosclerotic lesions (Reprinted with permission from Cai et al. [12])
molecular imaging revealed increased aortic angiogenesis in the obese animals, indicative of actively developing atherosclerotic plaques in the arterial wall (Fig. 1). Benfluorex reduced food consumption, body weight, insulin, and leptin in obese-treated rats, but had no effect on serum cholesterol or triglyceride levels. In a similar fashion, benfluorex reduced aortic angiogenesis in the obese-treated animals to the same level as the control lean rats. These molecular imaging results were corroborated by histological measurements of microvascular density. Therefore, MR molecular imaging with anb3-integrin-targeted PFC nanoparticles can noninvasively monitor atherosclerosis in a rat model of obesity and can monitor the neovascular effects of an appetite suppression therapy. However, atherosclerosis does not always lead to an increase in angiogenesis. In a rabbit model of peripheral vascular disease, hypercholesterolemia causes a reduction in the angiogenic response to chronic muscle ischemia, which can be normalized with a pro-angiogenic therapy, such as L-arginine supplementation [13]. MR molecular imaging was performed 10 days after femoral artery ligation with paramagnetic
anb3-integrin-targeted PFC nanoparticles. Molecular imaging revealed diffuse angiogenesis throughout the ischemic limb for both L-arginine-treated and untreated groups (Fig. 2). However, the angiogenic response was significantly increased in L-argininetreated animals compared to the untreated controls. Furthermore, the integrin-targeted formulation produced two times higher MR enhancement compared to nontargeted particles, demonstrating improved identification of angiogenic vasculature with biomarker targeting. X-ray angiography was performed 40 days postligation to measure the development of collateral vessels feeding the ischemic muscle tissue. More collateral vessels were observed in L-arginine-treated rabbits compared to the untreated controls. Histologic staining of muscle capillaries revealed a denser pattern of microvasculature in L-arginine-treated animals, confirming the MR and X-ray imaging results. Noninvasive molecular imaging with PFC nanoparticles could be used for early detection of response to angiogenic therapies directed at ischemic diseases, such as peripheral vascular disease or myocardial infarction, to enable personalized optimization of treatment strategies.
Perfluorocarbon Nanoparticles Perfluorocarbon Nanoparticles, Fig. 2 Molecular imaging of angiogenesis shows highly diffuse enhancement throughout the ligated (right) leg of cholesterol-fed rabbits with only slight enhancement of the control (left) leg. The L-arginine-treated rabbit (right panel) shows a more dense distribution of angiogenesis in the ligated limb compared with the control animal (left panel) (Reprinted with permission from Winter et al. [13])
2043
P
Molecular Imaging of Angiogenesis Right Leg (Ligated)
Left Leg Right Leg (Control) (Ligated)
Tap Water
Targeted Antiangiogenic Therapy In the realm of targeted drug delivery, PFC nanoparticles have been utilized to carry a number of antiangiogenic drugs. In particular, the potent antiangiogenic drug, fumagillin suppresses angiogenesis by inhibition of methionine aminopeptidase 2 (MetAP2) [9]. Targeting the particles to a biomarker of angiogenesis, such as the anb3-integrin, provides a means to specifically direct potent drugs to the sites of pathology, improving the therapeutic efficacy and decreasing the occurrence of unwanted side effects. Combining a specifically targeted molecular imaging agent with a therapeutic drug could benefit patients in a wide range of clinical scenarios. The imaging agent would allow confirmation and quantification of local drug delivery and enable personalization of treatment protocols based on the pharmacokinetics in the target tissue. In addition, specific targeting of the agent could improve uptake and retention of the drug at the sites of disease while lowering the exposure to other susceptible organs and reducing the occurrence of side effects. Paramagnetic PFC nanoparticles targeted to the anb3-integrin have been used for sensitive detection of angiogenesis in Vx2 tumors implanted in New Zealand white rabbits [2]. In vivo competition studies demonstrated that the targeted particles specifically bind to the anb3-integrin. Furthermore, MR molecular imaging with the targeted nanoparticles was able to differentiate between growing tumors and
Left Leg (Control)
L-Arginine
inflammatory lesions resulting from rejected cancer cells. These tumor cell remnants appeared identical to the viable growing tumors on standard T2-weighted MR images, but molecular imaging of angiogenesis showed negligible neovasculature in these masses. Antiangiogenic therapy has also been demonstrated in this animal model using anb3-integrin-targeted fumagillin nanoparticles [14]. Treatment with anb3integrin-targeted fumagillin nanoparticles reduced the tumor volume significantly compared with animals given nontargeted fumagillin nanoparticles, anb3integrin-targeted nanoparticles without drug or saline (Fig. 3). The MR imaging data was used to generate a 3D map of tumor angiogenesis, which showed a coherent asymmetric peripheral distribution characterized by dense neovessel regions interspersed with a finer, reticular pattern. Another possible biomarker of angiogenesis is Robo4, an endothelial specific guidance receptor that is expressed on tumor neovasculature but not mature vessels [15]. Paramagnetic PFC nanoparticles were coupled to a Robo4 antibody and evaluated in a mouse melanoma tumor model [15]. Both Robo4targeted and anb3-integrin-targeted nanoparticle formulations generated greater MR image contrast than particles lacking a targeting ligand or coupled to scrambled targeting ligands. Microscopic analysis of particle binding was performed with Robo4-targeted rhodamine nanoparticles and anb3-integrin-targeted AlexaFluor 488 nanoparticles. Binding of anb3integrin-targeted particles was generally more
P
P
2044
Perfluorocarbon Nanoparticles, Fig. 3 Left: Molecular imaging of targeted antiangiogenic therapy (fumagillin) in rabbit Vx-2 tumors showing diminished MR contrast enhancement in the treated animal (top) versus the control animal (bottom). Middle: MR enhancement, color coded in yellow (arrows), demonstrating reduced tumor angiogenesis in the fumagillin-treated animal. Right: 3D maps of tumor angiogenesis showing an asymmetric distribution of neovasculature (blue) that is dramatically reduced in the fumagillin-treated rabbit (Reprinted with permission from Winter et al. [14])
Perfluorocarbon Nanoparticles
Fumagillin Treated
1 cm 1 cm
Control
prevalent than the binding of Robo4-targeted nanoparticles, particularly in the tumor periphery. Thus, PFC nanoparticles can be used for both macroscopic MR molecular imaging as well as light microscopy to define cell binding of the particles and the coexpression of different cellular receptors. MR molecular imaging of angiogenesis has also been demonstrated in human melanoma tumors (40 mm3) implanted in immunocompromised mice using anb3-integrin-targeted PFC nanoparticles [2]. Again, in vivo competition studies demonstrated the high specificity of ligand-directed targeting, which was further corroborated microscopically using anb3integrin-targeted fluorescent PFC nanoparticles. As an alternative to targeting the anb3-integrin for mapping angiogenesis, the a5b1-integrin has been explored as a biomarker of tumor neovasculature [8]. The a5b1integrin is an important adhesion molecule, which regulates endothelial cell migration and survival during angiogenesis. The a5b1-integrin is not expressed on normal quiescent blood vessels, but its expression is induced on tumor blood vessels and in response to angiogenic factors, including basic fibroblast growth factor, interleukin-8, tumor necrosis factor-alpha, and the angiomatrix protein Del-1. The utility of a dual a5b1(anb3)-integrin targeting ligand for molecular imaging and targeted delivery of fumagillin was
compared with the use of a targeting ligand specific for only the anb3-integrin. MR signal enhancement maps revealed a sparse network of neovessels predominantly located in the tumor periphery. Treatment with a5b1(anb3)-integrin-targeted nanoparticles containing fumagillin further decreased the already small amount of neovessel enhancement to an almost negligible level; however, the tumor volume was not decreased. Since this tumor model displayed a very sparse neovasculature and was relatively slow growing, it is not surprising that an antiangiogenic therapy did not significantly decrease the tumor size. Clinically, MR molecular imaging of angiogenesis could be used to characterize small tumors, such as breast, lung, or colon, and stratify patients into antiangiogenic therapies based on the abundance of neovascular biomarkers. Such a procedure for screening potential patients may yield improved therapeutic benefits with decreased risk of side effects. Similar to the growth of tumors, the development of atherosclerotic plaques requires the recruitment of new blood vessels to meet the high metabolic demands of the abnormal tissue. In atherosclerosis, angiogenesis occurs in the vasa vasorum of large caliber arteries, such as the aorta, carotids, and coronaries. Since antiangiogenic therapies can slow the progression of cancer, it is reasonable to investigate if these
Perfluorocarbon Nanoparticles
Week 0 Segmentation
2045
Week 0 Enhancement
P
Week 1 Enhancement
50
Week 2 Enhancement
Week 3 Enhancement
Week 4 Enhancement
0
Perfluorocarbon Nanoparticles, Fig. 4 Serial imaging of aortic wall angiogenesis for monitoring targeted antiangiogenic therapy with fumagillin nanoparticles. Top: Black blood image of the thoracic aorta (arrow) prior to antiangiogenic treatment (Week 0) demonstrating segmentation of the vessel wall (outlined in yellow) and patchy areas of high angiogenesis within the aortic wall (color-coded as percent enhancement of MR
signal). At week 1, the MR enhancement is drastically reduced due to the antiangiogenic effect of fumagillin. Bottom: The level of signal enhancement gradually increases at Weeks 2 and 3 after fumagillin treatment. At Week 4, the level of enhancement is practically identical to the Week 0 image (Reprinted with permission from Winter et al. [9])
treatments could be employed to normalize atherosclerotic vessels. The acute antiangiogenic response to anb3-integrin-targeted paramagnetic nanoparticles containing fumagillin has been studied in atherosclerotic rabbits [16]. MR imaging was used to estimate drug deposition in the aortic wall. One week later, the level of neovascular anb3-integrin expression was assessed using anb3-targeted paramagnetic nanoparticles without fumagillin. Aortic areas displaying high MR enhancement at the time of treatment had the largest subsequent reduction in anb3integrin-targeted MR signal 1 week later, suggesting that combining imaging with therapy may not only confirm and quantify the local delivery of chemotherapeutics but may also provide early predictions of the subsequent treatment effects. In addition to the acute response to therapy, MR molecular imaging could provide serial monitoring of the end organ effects with the ultimate goal of optimizing drug treatment regiments and customizing patient protocols. The antiangiogenic effect of
atorvastatin with and without targeted delivery of fumagillin was serially monitored for 8 weeks by MR imaging in atherosclerotic rabbits [9]. Atorvastatin was dosed continuously via incorporation into the feed, while fumagillin treatment was provided once every 4 weeks. During 8 weeks of study, atorvastatin treatment did not reduce anb3-integrin expression in the aortic wall. Fumagillin treatment, on the other hand, resulted in a transient (2–3 weeks) reduction in image enhancement, indicating successful antiangiogenic treatment (Fig. 4). Combining the fumagillin and atorvastatin treatments yielded a persistent decrease in the MR signal, suggesting that chronic statin treatment could be used to prolong the effects of discrete doses of an antiangiogenic agent. Histology revealed that the MR image enhancement was strongly correlated to the microvascular density in the aortic wall in a logarithmic fashion. An effective and sustained antiangiogenic treatment may halt plaque progression and stabilize atheromas by decreasing intramural hemorrhage and inflammation.
P
P
2046
One of the benefits of using a molecular imaging contrast agent for targeted drug delivery is that imaging can provide a noninvasive measurement of drug deposition at the site of pathology and elimination from the body. Paramagnetic anb3-integrin-targeted PFC nanoparticles have been utilized to quantify the pharmacokinetics of the particles in both the blood pool as well as the target tissue in an atherosclerotic rabbit model [3]. Noninvasive monitoring of the particle biodistribution and elimination could be used as a surrogate marker for the tissue levels of lipophilic drugs incorporated in the nanoparticle formulation. Typically, pharmacodynamic measurements rely on blood sampling without consideration of the drug accumulation in the target tissue. Noninvasive mapping of drug concentrations in the end organ could provide significant clinical advantages in monitoring targeted therapeutic agents and determining the local drug concentrations beyond the blood pool. Atherosclerotic rabbits were injected with either anb3targeted or nontargeted paramagnetic nanoparticles at a dose of 1 mL/kg (0.0046 mmol Gd3+/kg), which is 20-fold lower than the standard dose of conventional gadolinium agents (0.1 mmol Gd3+/kg). Blood sampling was performed to determine the bulk pharmacokinetic behavior of the nanoparticles, and T1-weighted MR images of the descending aorta were collected over 24 h to ascertain binding to the angiogenic microvasculature in the vessel wall. A three-compartment pharmacokinetic model was developed to describe the in vivo behavior of the nanoparticles in both the blood pool as well as the aortic wall. Compartments 1 and 2 represent the bulk distribution of nanoparticles throughout the blood stream. The third compartment consists of the vasa vasorum of the aortic wall where the nanoparticles can specifically bind to the anb3-integrin. The constants that describe the transfer into and out of this compartment, k13 and k31, are lumped parameters that are used to describe both passive transfer and active binding. Blood sampling after IV injection of nanoparticles demonstrated the half-life for the distribution phase was 20.2 min, while the elimination half-life was 11.9 h. MR imaging of the aortic wall revealed that the targeted nanoparticle agent generated at least twofold higher signal enhancement at 0.5 and 1 h post injection than the nontargeted nanoparticles. Fitting both the blood clearance and tissue uptake data with
Perfluorocarbon Nanoparticles
a three-compartment model showed no significant differences in k12, k21, v1, or k31 between the targeted and nontargeted groups. However, k13 (representing transfer from the blood into the aortic wall) and ke (elimination from the blood pool) were significantly higher for the targeted formulation. While k13 represents the passive transfer from the blood into the third compartment in a standard three-compartment model, the nanoparticles are not just passively transferred but also actively binding within the third compartment. These two effects cannot be differentiated with this model and combine together, resulting in an apparent increase in k13 for the targeted nanoparticles compared to the nontargeted formulation. The pharmacokinetic parameters were used to calculate the tissue exposure to the nanoparticles. The local tissue concentration of targeted nanoparticles was doubled compared to nontargeted particles (1.64 vs 0.84 nmol Gd3+/g tissue/h, respectively). The amount of gadolinium that reached the aortic vasa vasorum was 0.18% of the total dose for nontargeted particles and 0.38% for targeted nanoparticles. These results demonstrate that MR can be used to noninvasively track binding of a molecularly targeted contrast agent, which has previously only been accomplished with nuclear and optical imaging. The gadolinium chelate on the particle surface could provide a surrogate marker for tracking the biodistribution of a lipophilic drug payload [9, 16, 17]. Rather than relying on blood pool measurements of drug concentrations, noninvasive imaging could serially monitor the actual concentration of the drug in the end organ and lead to more accurate modeling of therapeutic response.
Targeted Antiproliferative Therapy In addition to antiangiogenic drugs, PFC nanoparticles have been formulated to deliver drugs that inhibit the pathological proliferation of various cell types. The progression of both cancer and atherosclerosis typically involve uncontrolled cell division and a wide range of antiproliferative drugs have been developed to fight these diseases. For example, smooth muscle cells line the luminal aspect of arteries and are involved in the progression of atherosclerosis. Tissue factor, a transmembrane glycoprotein with a prominent role in angiogenesis, thrombosis, cell signaling, hemostasis, and mitogenesis, is expressed on
Perfluorocarbon Nanoparticles
smooth muscle cells in response to vascular injury. MR molecular imaging of tissue factor targeted PFC nanoparticles showed that the targeted particles bind to cultured smooth muscle cells at a concentration of 468 pM, while nontargeted particles only reach a concentration of 88 pM [3]. The MR measurements were validated by gas chromatography of the samples, revealing a particle concentration of 530 pM for the targeted nanoparticles and 111 pM for the nontargeted nanoparticles. Smooth muscle cell proliferation portends restenosis following coronary angioplasty. Therefore, the utility of PFC nanoparticles for delivery of antiproliferative drugs, including doxorubicin and paclitaxel, to smooth cells has been studied [2]. Exposure of the cells to targeted doxorubicin or paclitaxel nanoparticles significantly inhibited proliferation. Including the targeting ligand on the particle surface significantly increased the antiproliferative effect, particularly for paclitaxel. In vitro dissolution studies revealed that nanoparticle drug release persisted over 1 week. Targeted antiproliferative results were dependent on the hydrophobic nature of the drug and noncovalent interactions with other surfactant components. MR molecular imaging of the bound nanoparticles was demonstrated with both 1H and 19F MR imaging. A targeted paramagnetic nanoparticle carrying an antiproliferative payload to specific cell types could provide a novel, MR-visualizable, and quantifiable drug delivery system for the prevention of restenosis after angioplasty. Clinically, drug-eluting stents are often used to locally release antiproliferative drugs for inhibiting restenosis following angioplasty. However, in-stent thrombosis can occur with these devices due to endothelial damage from the antiproliferative drugs. An alternative approach has been pursued with PFC nanoparticles loaded with antiproliferative agents and targeted to cellular biomarkers or the extracellular matrix [18]. Furthermore, these nanoparticle agents can be used to visualize ruptures in the artery wall resulting from the angioplasty balloon. Paramagnetic anb3-integrin-targeted PFC nanoparticles containing rapamycin were used to locally deliver a drug payload through microfractures created by balloon overstretch injury. The local delivery of rapamycin inhibited arterial stenosis without impeding endothelial healing [18]. MR imaging demonstrated that the integrintargeted nanoparticles remained bound to the injured arterial wall 30–40 min after balloon overstretch and
2047
P
reestablishment of blood flow. Two weeks after injury, MR imaging showed significant luminal stenosis in vessels treated with control formulations: anb3integrin-targeted nanoparticles without drug, nontargeted nanoparticles with rapamycin or saline. In vessels treated with anb3-integrin-targeted rapamycin nanoparticles, no lumen irregularities were observed on MR angiograms. Histology showed anb3-integrin-targeted rapamycin nanoparticles decreased neointimal proliferation by 40% compared to controls. Endothelial healing was evaluated 1, 7, 14, and 28 days after balloon injury and showed no differences between anb3-integrin-targeted rapamycin nanoparticles and the controls. Another therapeutic approach to inhibit cellular proliferation uses host defense peptides that embed in the phospholipid cell membrane to create pores and disrupt normal cellular physiology. Melittin is a cytolytic peptide derived from honeybee venom that ruptures lipid membranes. PFC nanoparticles conjugated with melittin have been used as a therapeutic delivery vehicle in cancer models [19]. Melittin PFC nanoparticles were injected intravenously in mice bearing MDA-435 tumors [19]. Ultrasound imaging showed that treated tumors grew 25% slower than controls. In a mouse melanoma model, four intravenous injections of melittin nanoparticles reduced the tumor size by 80% compared to control injections of saline or nanoparticles lacking melittin. In a genetically engineered model of squamous carcinoma, treatment with integrin-targeted melittin nanoparticles resulted in an 80% reduction in dermal papillae, the site of dysplasia in this model. In cultured cells, free melittin causes necrosis resulting in the release of lactate dehydrogenase (LDH) due to the disruption of cell membranes and cytochrome c indicating mitochondria damage. Cultured cells exposed to targeted melittin nanoparticles showed cytochrome c release, but only minimal LDH, suggesting a shift to apoptotic cell death rather than necrotic cell death. Thus, drug delivery with PFC nanoparticles could be utilized to modulate the therapeutic mechanism of action. 19
F MR Molecular Imaging
MR detection of paramagnetic PFC nanoparticles typically relies on the effect of gadolinium on the 1H MR
P
P
2048
Perfluorocarbon Nanoparticles Concentration of Nanoparticles (nM)
a
b
c
>1.5
1.0
0.5
Optical Image
19F
MRI
Nanoparticle Concentration Map
0
Perfluorocarbon Nanoparticles, Fig. 5 Direct detection of fibrin-targeted PFC nanoparticles utilizing 19F MR imaging. (a) Optical image of a cross section of a human carotid endarterectomy sample showing moderate luminal narrowing as well as several atherosclerotic lesions. (b) A 19F projection image of the
carotid artery sample showing high signal along the lumen due to nanoparticles bound to fibrin. (c) Concentration map of bound nanoparticles on the carotid artery (Reprinted with permission from Morawski et al. [20])
signal. The paramagnetic nature of gadolinium increases the water relaxation rate and generates signal enhancement on T1-weighted MR images. An alternative approach achieves direct detection of the PFC core of the nanoparticles via 19F MR imaging [1, 3]. 19F has a relatively high sensitivity, 83% compared to 1H, and there is virtually no native background signal. Thus, the PFC core can provide a definitive and quantitative MR signature [1]. For example, 19F MR of fibrintargeted PFC nanoparticles could be used to map the formation of thrombi on ruptured atherosclerotic plaques and quantify the extent of ruptures in the fibrous cap. Human carotid endarterectomy samples were treated with fibrin-targeted paramagnetic nanoparticles and imaged at 4.7 T [20]. 1H MR imaging showed high levels of signal enhancement along the luminal surface due to nanoparticle binding to fibrin deposits (Fig. 5). A 19F projection image of the artery showed an asymmetric distribution of nanoparticles around the vessel wall corroborating the 1H signal enhancement. Co-registration of the quantitative nanoparticle map with the 1H image permitted visualization of both anatomical and pathological information in a single image. Combining information from 1H and 19F MR imaging could allow prediction of subsequent occlusion or distal embolization from unstable or disrupted plaques, and aid clinical decision-making for acute invasive intervention versus pharmaceutical therapies. PFC nanoparticles can be formulated with different perfluorochemicals to yield particles that can be
differentiated by 19F MR. Fibrin-targeted nanoparticles consisting of perfluorooctylbromide (PFOB) or perfluoro-15-crown-5-ether (CE) have been distinguished based on the unique chemical shifts of each compound [1]. Both 19F imaging and spectroscopy could identify PFOB and CE nanoparticles. The signal to noise for PFOB was lower than CE (10 vs 25, respectively), presumably due to the single CE peak (20 equivalent fluorine atoms) compared to the multiple PFOB peaks (17 fluorine atoms distributed over 5 peaks). A clear linear relationship between the 19F signal intensity and nanoparticle concentration was demonstrated for both PFOB and CE using both imaging and spectroscopy. Multiple PFC nanoparticle agents could be used to target different epitopes and achieve a noninvasive analogy to immunohistochemistry. For example, simultaneous quantification of angiogenesis in the vessel wall and fibrin deposition on the plaque cap could be used to evaluate the pathophysiological stage of an obstructive atherosclerotic lesion. Tissue inflammation is another biological marker of unstable atherosclerotic plaques, inducing the release of cytokines and upregulating the production of proteins like vascular cell adhesion molecule-1 (VCAM-1). PFC nanoparticles targeted to VCAM-1 were injected into genetically engineered ApoE/ mice, which mimic the clinical progression of atherosclerosis, to map inflammation [3]. These mice display focal inflammation and macrophage infiltration in the kidneys. To definitively identify nanoparticle binding in the kidneys, 19F MR imaging was performed on a 11.7 T
Perfluorocarbon Nanoparticles
research scanner 2 h after injection of VCAM-1 targeted PFC nanoparticles. The 19F signal arising from the PFC core provided an unambiguous marker of particle accumulation, without the innate signal variations that can confound typical 1H signal enhancement with traditional paramagnetic or superparamagnetic MR contrast agents. VCAM-1-targeted nanoparticles accumulated in ApoE/ kidneys to a greater extent than nontargeted nanoparticles (3.7 billion particles per gram of tissue vs 0.9 billion particles/gram). The uptake of targeted nanoparticles was also higher in the kidneys of ApoE/ mice compared to non-ApoE/ controls (3.7 vs 1.6 billion particles/gram). Control animals displayed no significant difference in the uptake of targeted versus nontargeted nanoparticles (1.6 vs 1.5 billion particles/gram). Typically, targeted MR contrast agents generate image enhancement as a result of both nonspecific blood pool signal as well as the specific binding of the agent to the biomarker of interest. Separating out these two contributions can be very difficult to achieve in vivo. One method to suppress the nonspecific signal utilizes diffusion-weighted 19F spectroscopy to null the signal arising from moving particles, representing the unbound fraction in the blood pool, while retaining the signal from stationary particles that are specifically bound to the target epitope [1]. A genetically engineered mouse model of squamous cell cancer, derived by incorporating human papilloma virus into the mouse genome, was studied with a 11.7 T research scanner. Transgenic and age-matched control mice were scanned 90 min after injection of anb3-integrintargeted PFC nanoparticles to assess angiogenesis in these precancerous lesions. Both the transgenic and control mice displayed decreased 19F signal with increasing b-values. However, 60–100% of the 19F signal remained at b-values near 60,000 s/mm2 in the transgenic animals, while no detectable 19F signal was observed in control mice at b-values of 1,500 s/mm2. The calculated ADC of PFC nanoparticles was 33.1 m2/s in the transgenic mice, significantly lower than the 19,563 m2/s ADC in the controls. Angiogenesis is a prominent feature in the progression of aortic valve stenosis. Cholesterol-fed rabbits develop aortic valve sclerosis, characterized by gross thickening, macrophage infiltration, calcification, and eventual bone formation that mimics the clinical presentation of the disease. Cholesterol-fed rabbits underwent 19F MR imaging after injection of
2049
P
anb3-integrin-targeted nanoparticles to quantify angiogenesis in the aortic valve leaflets [1]. The cholesterol feeding caused gross thickening of the aortic valves accompanied by extensive foam cell infiltration, noncalcified bone deposition, activation of myofibroblasts, abnormal microvascular proliferation, and upregulation of anb3-integrin expression. None of these abnormalities were observed in the normal valve tissue from control animals. The crown ether (CE) signal arising from the nanoparticle contrast agent was readily detected and distinguished from a perfluorooctylbromide (PFOB) quantification reference based on the chemical shifts of these perfluorocarbon species, allowing quantification of the total volume of bound nanoparticles. The volume of targeted nanoparticles bound to the valves was 19.5 nL, which was more than three times higher than the amount of nontargeted nanoparticles (5.6 nL). Competitive inhibition of anb3-integrin binding reduced the amount of nanoparticles in the valves by about half (10.3 nL). Valves from healthy rabbits treated with targeted nanoparticles contained almost nine times fewer nanoparticles (2.3 nL) than the valves from cholesterol-fed rabbits. These techniques may be useful for assessing atherosclerotic components of preclinical aortic valve disease in patients and could assist in defining efficacy of medical therapies. The sensitivity of this approach for molecular detection of sparse quantities of inflammatory epitopes in very thin structures at high field strengths establishes a basis for future efforts to develop localized spectroscopic methods at clinical field strengths that could be useful for detecting disease and monitoring therapies.
Conclusions PFC nanoparticles have been utilized for a wide range of molecular imaging, cell tracking, and targeted drug delivery applications. The relatively large size of the particles constrains them to the intravascular space, allowing specific targeting to endothelial expression of integrins, selectins, and adhesion molecules. PFC nanoparticles have a large surface area, which provides a means to incorporate multiple binding ligands and multiple paramagnetic chelates to the particle surface. The particle surface can also serve as a drug reservoir that selectively delivers a therapeutic payload to the targeted cell population. The dual role of PFC
P
P
2050
nanoparticles as both a molecular imaging contrast agent and a drug delivery vehicle can be exploited to provide diagnostic and therapeutic information via noninvasive MR imaging, or most of the other common clinical imaging modalities, including ultrasound, nuclear imaging, and computed tomography. Although PFC nanoparticles not currently approved for clinical use, the future applications of these agents in the management of patients with chronic diseases could aid in diagnostics, therapeutics, and/or followup of treatment efficacy or disease recurrence, by noninvasively mapping biomarkers associated with cancer or cardiovascular disease and specifically directing highly potent drugs directly to the sites of pathology.
Cross-References ▶ Angiogenesis ▶ Imaging Human Body Down to Molecular Level ▶ Intravital Microscopy Analysis ▶ Liposomes ▶ Microfabricated Probe Technology ▶ Nanomedicine ▶ Nanotechnology in Cardiovascular Diseases
References 1. Chen, J., Lanza, G.M., Wickline, S.A.: Quantitative magnetic resonance fluorine imaging: today and tomorrow. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2, 431–440 (2010) 2. Pan, D., Caruthers, S.D., Chen, J., Winter, P.M., Senpan, A., Schmieder, A.H., Wickline, S.A., Lanza, G.M.: Nanomedicine strategies for molecular targets with MRI and optical imaging. Future Med. Chem. 2, 471–490 (2010) 3. Winter, P.M., Caruthers, S.D., Lanza, G.M., Wickline, S.A.: Quantitative cardiovascular magnetic resonance for molecular imaging. J. Cardiovasc. Magn. Reson. 12, 62 (2010) 4. Culver, J., Akers, W., Achilefu, S.: Multimodality molecular imaging with combined optical and SPECT/PET modalities. J. Nucl. Med. 49, 169–172 (2008) 5. Lanza, G.M., Winter, P.M., Caruthers, S.D., Hughes, M.S., Hu, G., Schmieder, A.H., Wickline, S.A.: Theragnostics for tumor and plaque angiogenesis with perfluorocarbon nanoemulsions. Angiogenesis 13, 189–202 (2010) 6. Winter, P.M., Shukla, H.P., Caruthers, S.D., Scott, M.J., Fuhrhop, R.W., Robertson, J.D., Gaffney, P.J., Wickline, S.A., Lanza, G.M.: Molecular imaging of human thrombus with computed tomography. Acad. Radiol. 12(Suppl 1), S9–13 (2005)
Perfluorocarbon Nanoparticles 7. Lijowski, M., Caruthers, S., Hu, G., Zhang, H., Scott, M.J., Williams, T., Erpelding, T., Schmieder, A.H., Kiefer, G., Gulyas, G., et al.: High sensitivity: high-resolution SPECTCT/MR molecular imaging of angiogenesis in the Vx2 model. Invest Radiol. 44, 15–22 (2009) 8. Schmieder, A.H., Caruthers, S.D., Zhang, H., Williams, T.A., Robertson, J.D., Wickline, S.A., Lanza, G.M.: Threedimensional MR mapping of angiogenesis with alpha5beta1(alpha nu beta3)-targeted theranostic nanoparticles in the MDA-MB-435 xenograft mouse model. FASEB J. 22, 4179–4189 (2008) 9. Winter, P.M., Caruthers, S.D., Zhang, H., Williams, T.A., Wickline, S.A., Lanza, G.M.: Antiangiogenic synergism of integrin-targeted fumagillin nanoparticles and atorvastatin in atherosclerosis. JACC Cardiovasc. Imaging 1, 624–634 (2008) 10. Zhou, H.F., Chan, H.W., Wickline, S.A., Lanza, G.M., Pham, C.T.: Alphavbeta3-targeted nanotherapy suppresses inflammatory arthritis in mice. FASEB J. 23, 2978–2985 (2009) 11. Horton, M.A.: The alpha v beta 3 integrin “vitronectin receptor”. Int. J. Biochem. Cell Biol. 29, 721–725 (1997) 12. Cai, K., Caruthers, S.D., Huang, W., Williams, T.A., Zhang, H., Wickline, S.A., Lanza, G.M., Winter, P.M.: MR molecular imaging of aortic angiogenesis. JACC Cardiovasc. Imaging 3, 824–832 (2010) 13. Winter, P.M., Caruthers, S.D., Allen, J.S., Cai, K., Williams, T.A., Lanza, G.M., Wickline, S.A.: Molecular imaging of angiogenic therapy in peripheral vascular disease with alphanubeta3-integrin-targeted nanoparticles. Magn. Reson. Med. 64, 369–376 (2010) 14. Winter, P.M., Schmieder, A.H., Caruthers, S.D., Keene, J.L., Zhang, H., Wickline, S.A., Lanza, G.M.: Minute dosages of alpha(nu)beta3-targeted fumagillin nanoparticles impair Vx2 tumor angiogenesis and development in rabbits. FASEB J. 22, 2758–2767 (2008) 15. Boles, K.S., Schmieder, A.H., Koch, A.W., Carano, R.A., Wu, Y., Caruthers, S.D., Tong, R.K., Stawicki, S., Hu, G., Scott, M.J., et al.: MR angiogenesis imaging with Robo4vs. alphaVbeta3-targeted nanoparticles in a B16/F10 mouse melanoma model. FASEB J. 24, 4262–4270 (2010) 16. Winter, P.M., Neubauer, A.M., Caruthers, S.D., Harris, T.D., Robertson, J.D., Williams, T.A., Schmieder, A.H., Hu, G., Allen, J.S., Lacy, E.K., et al.: Endothelial alpha(v)beta3 integrin-targeted fumagillin nanoparticles inhibit angiogenesis in atherosclerosis. Arterioscler. Thromb. Vasc. Biol. 26, 2103–2109 (2006) 17. Lanza, G.M., Yu, X., Winter, P.M., Abendschein, D.R., Karukstis, K.K., Scott, M.J., Chinen, L.K., Fuhrhop, R.W., Scherrer, D.E., Wickline, S.A.: Targeted antiproliferative drug delivery to vascular smooth muscle cells with a magnetic resonance imaging nanoparticle contrast agent: implications for rational therapy of restenosis. Circulation 106, 2842–2847 (2002) 18. Cyrus, T., Zhang, H., Allen, J.S., Williams, T.A., Hu, G., Caruthers, S.D., Wickline, S.A., Lanza, G.M.: Intramural delivery of rapamycin with alphavbeta3-targeted paramagnetic nanoparticles inhibits stenosis after balloon injury. Arterioscler. Thromb. Vasc. Biol. 28, 820–826 (2008) 19. Soman, N.R., Baldwin, S.L., Hu, G., Marsh, J.N., Lanza, G. M., Heuser, J.E., Arbeit, J.M., Wickline, S.A., Schlesinger,
Photolithography P.H.: Molecularly targeted nanocarriers deliver the cytolytic peptide melittin specifically to tumor cells in mice, reducing tumor growth. J. Clin. Invest. 119, 2830–2842 (2009) 20. Morawski, A.M., Winter, P.M., Yu, X., Fuhrhop, R.W., Scott, M.J., Hockett, F., Robertson, J.D., Gaffney, P.J., Lanza, G.M., Wickline, S.A.: Quantitative “magnetic resonance immunohistochemistry” with ligand-targeted (19)F nanoparticles. Magn. Reson. Med. 52, 1255–1262 (2004)
Perfluoropolyethers ▶ Boundary Lubrication
2051
P
Photolithography Marc Madou1 and Chunlei Wang2 1 Department of Mechanical and Aerospace Engineering & Biomedical Engineering, University of California at Irvine, Irvine, CA, USA 2 Department of Mechanical and Materials Engineering, Florida International University, Miami, FL, USA
Synonyms Optical lithography
Petal Effect ▶ Rose Petal Effect
Petroleum Lubricants ▶ Boundary Lubrication
PFC Nanoparticles ▶ Perfluorocarbon Nanoparticles
PFOB Nanoparticles ▶ Perfluorocarbon Nanoparticles
Photoetching ▶ Chemical Milling and Photochemical Milling
Photofabrication ▶ Chemical Milling and Photochemical Milling
Definition Photolithography is a process used in microfabrication to selectively pattern a thin film substrate using light to transfer a geometric pattern from a photomask to a light-sensitive photoresist.
Introduction Lithography is a technique used to transfer copies of a master pattern onto the surface of a solid material such as a silicon wafer. The word lithography (Greek for the words stone [lithos] and to write [gra´phein]) refers to the process invented in 1796 by Aloys Senefelder, who inked Bavarian limestone and transferred a carved image from stone onto paper. The most widely used form of lithography is photolithography. In the IC industry, pattern transfer from masks onto thin films is accomplished almost exclusively via photolithography. Photomasking, followed by chemical processing, led to the photolithography now used in fabricating ICs (integrated circuits) and in miniaturization science. The first applications of the printed circuit board (PCB) (also printed wiring board or PWB) were invented in 1943 by the German Paul Eisler. By 1961, Fairchild Semiconductor had introduced the first commercial integrated circuit in which photo-etching processes produce large numbers of transistors on a thin slice of silicon (Si). In 2008, integrated circuits were ubiquitous and they could be found in microprocessors, audio and video equipment,
P
P
2052
dishwashers, garage openers, security systems, automobiles, etc. With the first ICs, patterns had a resolution not better than 5 mm. Today, photolithography, x-ray lithography, and charged particle lithography all achieve submicron-printing accuracy. Intel introduced the Prescott Pentium IV chip in the fourth quarter of 2003, an IC chip incorporating 90 nm sized features, using UV photolithography. Chips based on a 65 nm technology became available in 2006 and the 32 nm node was achieved in 2009. The combination of accurate registration and exposing a series of successive patterns leads to complex multilayered ICs. This essentially twodimensional (2D) process has a limited tolerance for nonplanar topography, creating a major constraint for building non-IC miniaturized systems, which often exhibit extreme topographies. Photolithography has matured rapidly and its ability to resolve ever-smaller features is constantly improving. For the IC industry, this continued improvement in photolithography resolution has impeded the adaptation of alternative, higherresolution lithography techniques, such as x-ray lithography. Research over the last 10 years, in high-aspectratio resists features to satisfy the needs of both IC and non-IC miniaturization, is also finally improving dramatically photolithography’s capacity to cover wide ranges of topography. Performance of a photolithographic process is determined by its resolution, the minimum feature size that can be transferred with high fidelity, the registration, how accurately patterns on successive masks can be aligned and throughput, the number of wafers that can be transferred per hour. Photolithography and pattern transfer involve a set of process including photoresist coating, UV exposure, baking, developing, etc.
Wafer Cleaning and Clean Room An important step, even before lithography proper, is wafer cleaning. Contaminants include solvent stains (methyl alcohol, acetone, trichloroethylene, isopropyl alcohol, xylene, etc.), dust from operators and equipment, smoke particles, etc. Stains or films may lead to adverse effects during oxidation and evaporation processes. Particulates, chunks of granular matter, may cause undesirable masking effects and scratches on the photomask during contact printing. All lithography processes take place inside a semiconductor clean room, which is a specially constructed enclosed area
Photolithography
environmentally controlled with respect to airborne particulates, temperature ( 0.1 F), air pressure, humidity (from 0.5% to 5% RH), vibration, and lighting. In a Class 1 clean room, the particle count does not exceed 1 particle per cubic foot with particles of a size of 0.5 mm and larger, and in a Class 100 clean room, the particle count does not exceed 100 particles per cubic foot with particles of a size of 0.5 mm and larger. The allowable contamination particle size in IC manufacture has been decreasing hand in hand with the ever-decreasing minimum feature size. Many different dry and wet methods for wafer cleaning currently in use, such as RCA1 and RCA2, developed by W. Kern, use mixtures of hydrogen peroxide and various acids or base followed by deionized (DI) water rinses. Others include vapor cleaning; thermal treatment, for example, baking at 1,000 C in vacuum or in oxygen; and plasma or glow discharge techniques, for example, in freons with or without oxygen. Mechanical methods include ultrasonic agitation, polishing with abrasive compounds, and supercritical cleaning. Ultrasonic cleaning, which is excellent for removing particulate matter from the substrate, is unfortunately prone to contamination and mechanical failure of deposited films.
Masks The stencil used to repeatedly generate a desired pattern on resist-coated wafers is called a mask. In typical use, a photomask – a nearly optically flat glass (transparent to near ultraviolet [UV]) or quartz plate (transparent to ˚ deep UV) with an absorber pattern metal (e.g., an 800 A thick chromium layer) – is placed above the photoresistcoated surface, and the mask/wafer system is exposed to UV radiation. The absorber pattern on the mask can be generated by e-beam lithography which yields higher resolution than photolithography. A positive or dark field mask is a mask on which the pattern is clear with the background dark. A negative or clear field mask is a mask on which the pattern is dark with the background clear. A light field or dark field image, known as mask polarity, is then transferred to the semiconductor surface. This procedure results in a 1:1 image of the entire mask onto the silicon wafer. Masks, making direct physical contact with the substrate, are called contact masks. Unfortunately, these masks degrade faster due to wear than
Photolithography
noncontact, proximity masks (also referred to as soft contact masks), which are slightly raised, say 10–20 mm, above the wafer. Contact mask and proximity mask printing are collectively known as shadow printing. A more reliable method of masking is projection printing where, rather than placing a mask in direct contact with (or in proximity of) a wafer, the photomask is imaged by a high-resolution lens system onto the resist-coated wafer. The imaging lens can reduce the mask pattern by 1:5 or 1:10, making mask fabrication less challenging. Lithography is still mostly carried out using masks, but due to problems caused by masks, such as expense and time in fabricating them, contamination introduced by them, their disposal, and the difficulties in their alignment, research into maskless optical projection lithography (MOPL) is growing rapidly and broadly. One MOPL, approach already on the market (http:// www.intelligentmp.com), is based on the Digital Micromirror Device (DMD) chip from Texas Instruments Inc. (TI), and relies on the same spatial light modulation technology used in DLP (digital light processing) projectors and HDTVs (high definition television) arrays in a DMD chip to project images on the photoresist. The resolution of DMD-based maskless photolithography (about 5 mm) is significantly less than with e-beam lithography (0.25 mm) or laser writers ( 300 C, a low dielectric constant e, good chemical stability, and ease of processing. Most resists in IC and MEMS fabrication are deposited as liquids, whereas resists used in printed wiring board (PWB) manufacture are usually dry film resists that come in rolls (ranging from 2 to 60 in. wide and 125 to 1,000 ft. long) and are laminated onto the substrate instead of being spin coated on it. Dry film
resist formulations are sandwiched between a polyolefin release sheet and a polyester base, rolled up on a support core. Dry film resists offer advantages such as excellent adhesion on most substrates, no liquid handling since there is no solvent, high process speed, excellent thickness uniformity over a whole substrate, simple handling, no formation of edge beads, low exposure energy, low cost, short processing time, and near vertical sidewalls. There are a variety of dry film photoresists widely used and commercially available, examples include Riston ®, Ordyl BF 410, Etertec ® 5600, DF 4615, and DFR-15. All are used in the manufacture of circuit boards and can be made quite thick and all are candidates for broad use in MEMS as well. Photoresist stripping, in slightly oversimplified terms, is organic polymer etching. The primary consideration is complete removal of the photoresist without damaging the device under construction. There are many commercial strippers available; some are specific for positive resists (e.g., ACT-690 C from Ashland), others for negative resists, and still others are universal strippers (e.g., ACT-140 from Ashland). Other popular commercial strippers are Piranha and RCA clean. Wet stripping solutions lose potency in use, causing stripping rates to change with time. Accumulated contamination in solutions can be a source of particles, and liquid phase surface tension and mass transport tend to make photoresist removal difficult and uneven. Dry stripping or oxygen plasma stripping, also known as ashing, has become more popular as it poses fewer disposal problems with toxic, flammable, and dangerous chemicals. Dry stripping is more controllable than liquid stripping, less corrosive with respect to metal features on the wafer, and, more importantly, it leaves a cleaner surface under the right conditions. Finally, it does not cause the undercutting and broadening of photoresist features that can be caused by wet strippers.
Critical Dimension and Overall Resolution The absolute size of a minimum feature in an IC or a miniature device, whether it involves a line-width, spacing, or contact dimension, is called the critical dimension (CD). The overall resolution of a process describes the consistent ability to print a minimum size image, a critical dimension, under conditions of
Photolithography
2057
P
Photolithography, Table 1 Comparison of traditional negative and positive photoresists Characteristic Adhesion to Si Available compositions Baking Contrast g Cost Developer Developer process window Influence of oxygen Lift-off Mask type Opaque dirt on clear portion of mask Photospeed Pinhole count Pinholes in mask Plasma etch resistance Proximity effect Residue after development Resolution Sensitizer quantum yield F Step coverage Strippers of resist over Oxide steps Metal steps Swelling in developer Thermal stability Wet chemical resistance
Resist type Positive resist Fair (priming required) Many In air (+) Higher, e.g., 2.2 More expensive Temperature sensitive () and aqueous based (Ecologically sound) Small No (+) Yes Dark-field: lower-defects Not very sensitive to it
Negative resist Excellent (priming not required) Vast In Nitrogen () Lower, e.g., 1.5 Less expensive Temperature insensitive (+) and organic solvent () Very wide, insensitive to overdeveloping Yes () Yes Clear-field: higher-defects Causes printing of pinholes
Slower Higher Prints mask pinholes Not very good Prints isolated holes or trenches better Mostly at 1 mm) 0.5–1 Lower
Acid Simple solvents No Good Fair
Acid Chlorinated solvent compounds Yes Fair Excellent
reasonable manufacturing variation. Many aspects of the process, including hardware, materials, and processing considerations can limit the resolution of lithography. Hardware limitations include diffraction of light or scattering of charged particles (in the case of chargedparticle lithography or hard x-rays), lens aberrations, mechanical stability of the system, etc. The resist material properties that impact resolution are contrast, swelling behavior, thermal flow, and chemical etch resistance, etc. The most important process-related resist variables include swelling (during development) and stability (during etching and baking steps). Resolution frequently is measured by line-width measurements using either transmitted or reflected light or other metrology techniques. Optical techniques perform satisfactorily for features of 1 mm and larger, providing a precision of
0.1 mm. The successful performance of devices depends upon the control of the size of critical structures across the entire wafer and from one wafer to another, referred to as line-width control. A rule of thumb is that the dimensions must be controlled to tolerances of at least 1/5 of the minimum feature size. Correct feature size must be maintained within a wafer and from wafer to wafer, as device performance depends on the absolute size of the patterned structures. In the shadow printing mode, including contact and proximity arrangements of mask and wafer, optical lithography has a resolution with limits set by a variety of factors. These include diffraction of light at the edge of an opaque feature in the mask as the light passes through an adjacent clear area, alignment of wafer to mask, nonuniformities in wafer
P
P
2058
Photolithography
flatness, and debris between mask and wafer. Diffraction causes the image of a perfectly delineated edge to become blurred or diffused. The theoretical resolution, R, that is, the minimum resolved dimension in a grating mask (bmin for a line or a space) is given by: R ¼ bmin
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi z ¼ k lðsþ Þ 2
(1)
where bmin ¼ half the grating period and the minimum feature size transferable s ¼ the gap between the mask and the photoresist surface l ¼ wavelength of the exposing radiation z ¼ the photoresist thickness k ¼ a constant which theoretically is 1.5 Depending on the relative magnitude of s and z, one distinguishes between contact and proximity printing. The square root relation in contact and proximity printing is a consequence of the near field or Fresnel diffraction theory valid in the near field region just below the mask openings, this contrasts with the far field behavior (Fraunhofer diffraction) of projection lithography (see below). Self-aligned printing can be regarded as an extreme form of contact printing where there is no gap between the mask and the substrate. In contact printing, a photomask is pressed against the resist-covered wafer with pressures in the range of 0.05–0.3 atm., and the s, in Eq. 8.13 is zero. The theoretical maximum resolution is seldom achieved, however, as only diffraction effects were taken into account to derive Eq. 1. The other factors mentioned above (wafer flatness, mask alignment, etc.) and shadowing effect (penumbral blur) usually conspire to make the resolution worse. In proximity printing, spacing of the mask removed from the substrate (by at least 10 mm) minimizes defects that result from contact. On the other hand, diffraction of the transmitted light reduces the resolution. The degree of reduction in resolution and image distortion depends on the wafer-to-substrate distance, which may vary across the wafer. In projection printing, wafer contact is completely avoided; a high-resolution lens projects an image of the photomask onto the photoresist-covered wafer. Imaging in this case is far-field or Fraunhofer diffraction dominated and the Rayleigh criterion for far-field
resolution can be considered. A dimensionless constant k1 can be used to define the practical limiting resolution R in projection printing as: R ¼ k1
l NA
(2)
where k1 ¼ experimentally determined dimensionless parameter that depends on resist parameters, process conditions, mask aligner optics, etc. To achieve higher resolutions, instrumentation designers want to work at shorter wavelengths l, higher NA, and smaller k1. Since the mid-1980s, the demise of optical lithography has been predicted as being only a few years away, but each time a new resolution limit approaches, some new method extends the useful life of the technology. The three main technologies involved in printing ICs and other miniaturized devices are resist technology, mask technology, and the exposure tool, all of which need to be addressed to optimize lithography resolution. Strategies for improved resolution can be achieved through using chemically amplified resists, antireflection coatings, improved mask technology (such as Optical Proximity Correction), improved Exposure Equipment (such as off-axis illumination), etc. DUV photolithography is expected to continue until the 45 nm node through these resolution enhancing techniques. The 32 nm node is viewed as beyond the scope of DUV lithography.
Next-Generation Lithographies (NGLs) and Lithography Research In the IC industry, continuous improvements to DUV photolithography have postponed the industrial adoption of alternative next-generation lithographies (NGLs), such as extreme ultraviolet lithography (EUVL), x-ray lithography, charged particle beam lithography based on electrons and ions (such as electron and ion projection techniques), and imprint lithography (Nanoimprint lithography [NIL] and Step-and-Flash Imprint Lithography [SFIL]), because of the huge financial investment in existing photolithography equipment. These NGLs are regarded today as important technologies for the beyondthe-DUV-lithography-era, either for mask making or for actual IC production. In addition, many lithography
Photolithography
approaches in the R&D stage, including lithography based on very thin resist layers and block copolymers, zone plate array lithography (ZPAL), quantum lithography (two-photon lithography) and proximal probe–based techniques such as atomic force microscopy (AFM), scanning tunneling microscopy (STM), dip-pen lithography (DPL) and near-field scanning optical microscopy (NSOM), and apertureless near-field scanning optical microscopy (ANSOM), are attracting more and more research interest. It is quite possible that some of the alternative lithography tools in the R&D category will emerge as serious next-generation lithographies (NGL) in the coming years. Extreme Ultraviolet Lithography (EUVL) Extreme ultraviolet lithography (EUVL), using wavelengths in the 10–14 nm range to carry out projection imaging, is perhaps the most natural extension of optical projection lithography as, in principle, it only differs in terms of the wavelength. This type of radiation is also referred to as soft-x-ray radiation and vacuum UV. Sources for this type of radiation are laserproduced plasmas and synchrotrons. EUV is strongly absorbed in virtually all materials and, consequently, imaging must be carried out in vacuum; also, all camera optics as well as masks used must be reflective rather than refractive. New resists and processing techniques must be developed as well. X-ray Lithography X-ray lithography employs a shadow printing method similar to optical proximity printing. The x-ray wave˚ ) is much shorter than that of UV light length (4–50 A (Table 2). Hence, diffraction effects are reduced and higher resolution can be attained. For instance, for an ˚ and a gap of 40 mm, resolution x-ray wavelength of 5 A can be around 0.2 mm. In contrast, with electron lithography and ion-beam lithography, no charged particles are directly involved in x-ray exposures, which makes the need for vacuum less stringent. X-ray lithography is superior to optical lithography due to its use of shorter wavelengths and its very large DOF, and because exposure time and development conditions are not as stringent. Reproducibility is high since results are independent of substrate type, surface reflections, and wafer topography. Another important benefit is that x-ray lithography is immune to low-atomic-number (Z) particle contamination
2059
P
Photolithography, Table 2 Light sources for various types of lithography Wavelength [nm] 436 405 365 248 193 157 10 1
Source Hg arc lamp Hg arc lamp Hg arc lamp Hg/Xe arc lamp, KrF excimer laser ArF excimer laser F2 laser
Range G-line H-line I-line Deep UV (DUV)
DUV Vacuum UV (VUV) Laser-produced plasma Extreme UV sources (EUV) X-ray tube, synchrotron X-ray
(dust). With an x-ray wavelength on the order of ˚ or less, diffraction effects generally are negligi10 A ble, and proximity masking can be used, increasing the lifetime of the mask. In x-ray lithography, there are essentially no optics involved and, although this sounds like an advantage, it also presents one major disadvantage since one can only work with 1:1 shadow printing. No image reduction is possible, so the mask fabrication process is very complicated. In the United States, IBM remains the only major champion of x-ray lithography for next-generation lithography. At the end of 1999, IBM fabricated several PowerPC 604e microprocessor batches to demonstrate the viability of the method. In Japan, there are still many players involved such as NTT, Toshiba, Mitsubishi, and NEC. LIGA The LIGA technique (a German acronym for Lithographie, Galvanoformung, Abformung) was invented about 20 years ago. LIGA exploits all of the advantages of x-ray lithography listed above. The LIGA process involves a thick layer of resist (from micrometers to centimeters), high-energy x-ray radiation, and resist development to make a resist mold. By applying galvanizing techniques, the mold is filled with a metal. The resist structure is removed, and metal products result. Alternatively, the metal part can serve as a mold itself for precision plastic injection molding. Several types of plastic molding processes have been tested, including reaction injection molding, thermoplastic injection molding, and hot embossing. LIGA enables new building materials and a wider dynamic range of dimensions and possible shapes.
P
P
2060
Photomilling
Cross-References
Introduction
▶ BioPatterning ▶ Biosensors ▶ CMOS MEMS Biosensors ▶ CMOS-CNT Integration ▶ Dip-Pen Nanolithography ▶ Dry Etching ▶ DUV Photolithography and Materials ▶ Electron Beam Lithography (EBL) ▶ Focused-Ion-Beam Chemical-Vapor-Deposition (FIB-CVD) ▶ MEMS Packaging ▶ Microcontact Printing ▶ Nanoimprint Lithography ▶ NEMS Piezoelectric Switches ▶ NEMS Resonant Chemical Sensors ▶ SU-8 Photoresist ▶ Wet Etching
On-chip manipulation of light is of great importance for applications ranging from optoelectronics and nonlinear optics to biochemical sensing and cavity quantum electrodynamics. Ultra-high quality factor (Q) optical resonators are indispensable for these applications since they enhance the probability of light-matter interaction due to their ability to store photons for many optical cycles [1]. Traveling-wavebased resonators, and micro-toroid resonators in particular, have been used to achieve some of the largest Q factors to date [1]. For many applications, however, the figure of merit is proportional to the ratio between resonator’s Q and mode volume (Vmode), which can be quite large in traveling-wave geometries. Therefore, standing-wave optical resonators, capable of having very small mode volume, have been explored as an attractive alternative [1]. These structures, based on a Fabry-Perot geometry, typically consist of the region in which light can propagate (e.g., an optical waveguide), terminated at its both ends with optical mirrors. The mirrors are often in the form of a 1-D Bragg mirror, as in the case of vertical cavity surface emitting laser (VCSEL), for example. In 1987, the Bragg mirror idea was extended to 3-D by Yablonovitch [2] and John [3] who proposed 3-D photonic crystals as a platform suitable for realization of ultimate cavities with wavelength-scale mode volumes and ultra-high Qs. Unfortunately, fabrication of these 3-D periodic optical structures has proven to be challenging and in spite of great effort no ultra-high Q small mode volume cavity based on 3-D photonic crystals has been realized to date. This prompted researchers to consider alternative approaches, based on lower-dimensionality photonic crystals. In particular, planar photonics crystals, based on 2-D periodic lattices realized in thin semiconductor membranes [4] received considerable interest due to the ease of fabrication and compatibility with standard planar micro-fabrication technology. State-of-the-art fabrication techniques, combined with creative cavity designs, have resulted in wavelengthscale optical cavities with Q > 106 (experimental) and Q/Vmode values larger than traveling wave resonators. Recently, however, it has been demonstrated that photonic crystals with even lower dimensionality, resembling some of the early distributed feedback (DFB) cavity designs, can be used to realize cavities with performance rivaling and even outperforming those
References 1. Madou, M.: Fundamentals of Microfabrication, 2nd edn. CRC Press, Boca Raton (2002)
Photomilling ▶ Chemical Milling and Photochemical Milling
Photonic Crystal Nanobeam Cavities Parag B. Deotare and Marko Loncar Electrical Engineering, Harvard School of Engineering and Applied Sciences, Cambridge, MA, USA
Definition Photonic crystal nanobeam cavities are high quality factor, small mode volume optical resonators formed by using photonic crystal Bragg mirrors in the longitudinal direction and total internal reflection in the others.
Photonic Crystal Nanobeam Cavities
based on 2-D planar photonic crystals. These photonic crystal nanobeam cavities (PCNCs), based on chirped 1D optical lattices, also have a physical footprint which is exactly the same as that of an optical waveguide and therefore, represent the smallest cavity geometry that can be made using dielectric materials only. Furthermore, nanobeams enable realization of high-Q smallVmode cavities even in low-index materials (e.g., SiO2, polymers) [5, 6], thus vastly increasing the library of photonic materials suitable for realization of ultra-high Q/Vmode cavities. The near-field of PCNCs is highly accessible, which allows for easy light-material interaction and which is useful for sensing [6], optical trapping [7], and optomechanics [8, 9]. PCNCs can be easily and naturally integrated with optical waveguides for efficient delivery of light [10, 11]. Cavities that support ultra-high Q modes with both TE and TM polarization [12] can be easily realized. All of these features, many of which are hard to achieve using 2-D photonic crystals, make photonic crystal nanobeam cavities an ideal platform for practical applications.
Design Evolution The idea of waveguide-based micro-cavities has been around for more than 15 years. It was first proposed at MIT [13] in 1995 and then experimentally realized 2 years later [14]. Figure 1a shows the schematic of the device which consisted of a waveguide segment surrounded with two lattices of holes that play the role of Bragg mirrors. Light trapping is achieved in this system via a combination of Bragg scattering (in longitudinal direction) and total internal reflection (in two transverse directions), resulting in an optical resonance as shown in Fig. 1b. The theoretical value of Q was calculated to be Q ¼ 280 whereas the experimentally measured value was 265 (Fig. 1c). Such a low value of Q factor was due to the suboptimal design. In particular, light scattering at the interface between the Bragg mirror and the central cavity region was responsible for the low Q factor (Fig. 2a) due to the mismatch in the effective mode indices of the two regions. In order to increase the Q of the cavity it is necessary to taper propagating mode of the cavity region into the exponentially decaying mode of the Bragg mirror [15, 16]. This mode matching, analogous to impedance matching problem in electronics, can be achieved using a multi-hole taper toward the cavity
2061
P
[15]. By doing so, the waveguide mode index can slowly be reduced to match the Bragg index which significantly reduces the scattering and provides exponential increase in cavity Q, at an expense of slight, linear increase in the cavity mode volume. By increasing the number of holes used in the taper, near adiabatic conversion between the two modes could be achieved. By putting two of such engineered tapered (chirped) mirrors back to back, ultra high-Q cavity could be realized. The electric field of a cavity mode formed by using a 7-hole taper is shown in Fig. 2a. The effect of tapered holes on the Q and mode volume was first studied in silicon nitride material system for operation in visible [16] and the results are shown in Fig. 2b. The figure shows Q as a function of cavity length, defined as distance between two central holes in the structure, for two different numbers of tapered holes used in the mirror. It can be seen that the drastic increase in Q can be achieved for an optimal value of the cavity length. Also, it is apparent that the Q values depend strongly on the precise cavity length which puts stringent requirement on the fabrication quality needed for the realization of ultra-high Q cavities.
Ultra-High Q Nanobeam Cavities In order to validate the designs, devices were fabricated in silicon-on-insulator (SOI) material platform (Fig. 3) using state-of-the art nanofabrication technology. The highest Q measured in such cavities that operated in telecom wavelength range was 750,000 [17]. The cavities were probed using a free-space cross-polarization technique popularly known as resonance scattering. This technique is relatively straightforward to implement but suffers from poor in- and out-coupling of light, in particular in the case of high-Q cavity modes. Other characterization techniques that have been used to characterize nanobeam cavities include evanescent coupling using a tapered fiber [8] and butt coupling [10] (Fig. 4a). The latter technique, particularly suitable for realization of integrated onchip networks, takes advantage of spot-size converters to couple light from tapered optical fiber into the onchip optical waveguide. In this case, the converter is in the form of polymer (SU-8 photoresist) waveguides (ref. Fig. 4b) whose cross-section and resulting mode profile is comparable to the spot emerging from the tapered optical fiber. Light is then adiabatically
P
P
2062
Photonic Crystal Nanobeam Cavities
a
c ad
Si
a
SiO2 a = 0.42 μm ad = 0.63 μm r = 0.10 μm w = 0.47 μm tSi = 0.20 μm tetch = 0.55 μm
b 1.0
Transmission
0.8 0.6 0.4 0.2 1.0 μm 0.0
1,300
1,400
1,500
1,600
1,700
Wavelength (nm)
Photonic Crystal Nanobeam Cavities, Fig. 1 (a) Schematic of photonic crystal nanobeam cavity (PCNC) (b) simulated transmission of the device showing the bandgap and the cavity
b
scattering
Q7-hole
106
r Quality factor
mirror less scattering
r1
r2
taper
r
1 V7-hole
Q4-hole
0.8 V4-hole
105 0.6
Vmode/(λ/n)3
a
resonance [13] (c) SEM image of the fabricated device. The measure Q of the device was 265 [14]
mirror 104 70
0.4 80
90 100 Cavity length (nm)
110
120
Photonic Crystal Nanobeam Cavities, Fig. 2 (a) Schematic showing the effect of taper on scattering loss due to mismatch of the waveguide mode and the Bloch mode. Cavity mode electric field (Ey) for a 7-hole taper (b) simulated results from [16]
showing the variation in Q and mode volume for different taper and cavity lengths. Cavity length was defined as the distance between two innermost holes (indicated in Part a)
coupled from polymer waveguide into the silicon waveguide using the inverse taper geometry, and then from the silicon waveguide into the nanobeam cavity using adiabatically tapered photonic crystal [10, 11].
Efficient coupling of light into the cavity is achieved by using hole taper similar to the one described earlier [11]. Light transmitted through the system is collected by a second spot-size converter and
Photonic Crystal Nanobeam Cavities
2063
tapered fiber. Figure 4b shows an optical micrograph of such a fabricated structure with the input coupling pads, silicon waveguides, cavities, and output coupling pads. Inset shows the zoomed-in image of the polymer coupling pad. Coupling efficiencies as high as 80% over more than 150 nm could be achieved using this technique. The transmission spectrum shown in Fig. 4c features the excellent quality of signal to noise achieved using the above technique. The inset shows the zoomed region around the cavity resonance. There have been quite a few variations in the design of the periodic structure of the ultra-high Q nanobeam cavity, including ladder cavities [18] with rectangular
Photonic Crystal Nanobeam Cavities, Fig. 3 SEM image of the fabricated structure with a measured Q of 7.5 105 [17]. The inset shows the zoomed cavity region
a
P
holes, bookshelf/domino structures [18], and parabolic width-taper cavity [19]. However, the working principle of all of these approaches is the same and is based on efficient effective index matching between cavity and mirror regions. Another common feature of all the different design strategies is that they are all intuitionbased and involve extensive parameter search and trial-and-error-based optimization in order to find the optimal taper profile (the radius and spacing of the holes in the taper section). To overcome these limitations and streamline the design process, thus significantly reducing the time needed for the design of ultrahigh Q structures, a deterministic method to design ultra-high Q cavities has recently been proposed and experimentally demonstrated [10]. This approach not only eliminates trial-and-error-based shifting and resizing of holes, but also results in cavities that feature high transmission at the cavity resonance: Q > 107 with transmission of 97% (Fig. 5). As mentioned in the introduction, nanobeam cavities enable realization of ultra-high Q cavities, with small mode volumes, even in materials with low refractive index. This is typically very hard to achieve using 2-D planar photonic crystals since the bandgap is typically closed due to low-index contrast. Nanobeam cavities do not suffer from this problem since 1-D bandgap is always open even for the smallest index
b
cou p
lers
Optical micrograph couplers
cavity waveguide
rs fibe
c 0.25 Intensity (au)
0.08
0.20 0.15 0.10
0.04 0.00
polymer waveguide
1508.25
cavities Waveguides
0.05 0.00
1500
1550 1600 Wavelength (nm)
1650
Photonic Crystal Nanobeam Cavities, Fig. 4 (a) Schematic showing butt coupling technique using spot-size converter couplers based on polymer waveguides and inverse-tapered silicon waveguides. (b) Optical micrograph showing the fabricated
silicon waveguide 2mm
couplers
device with the in- and out-coupling pads, silicon waveguide, and PCNC. Inset shows an SEM image of the polymer coupling pads. (c) Transmission from one of the devices shown in (b)
P
P
2064
Photonic Crystal Nanobeam Cavities
a
b
Modulated Bragg mirror
1 mm
Photonic Crystal Nanobeam Cavities, Fig. 5 Ultra-high Q cavity designed using deterministic approach [10] (a) Electric field profile (b) SEM image of structure fabricated in SOI
modulation. Recently, high-Q all-polymer cavities with index contrast between core (cavity material) and cladding as low as 1.15 have been experimentally demonstrated [6]. Finally, it has been demonstrated that photonic crystal nanobeam cavities also play a role of phononic crystals and are capable of localizing phonons [9]. Furthermore, simultaneous confinement of photons and phonons allows for strong interaction between mechanical and optical excitations, thus enabling both exploration of exciting physics at nanoscale and realization of novel devices. For example, electromagnetically induced transparency (EIT) in photonic/ phononic crystal nanobeam cavity was demonstrated and successfully used to alter the group velocity of light thus realizing optical delay lines of interest for optical buffering, filtering, and signal processing [20].
Applications One promising application of nanobeam cavities is realization of compact optical filters and spectrometers. However, due to fabrication limitations, some degree of post-fabrication wavelength trimming is needed in order to bring the resonance of the filter (cavity) to the desired wavelength. Moreover, optical devices that could be dynamically and reversibly configured to operate at vastly different wavelengths (optical analog to FPGAs) would be of great interest for practical applications. Conventional techniques, based on thermal effects and free carrier injection, suffer from high steady-state power consumption and small tuning range, respectively. These limitations can be overcome by taking advantage of mechanical degrees of freedom of suspended photonic crystal nanobeam resonators.
Using coupled photonic crystal nanobeam cavities [21], a reconfigurable photonic crystal filter [22] that can be dynamically, continuously, and reversibly tuned over a wide wavelength range was proposed and demonstrated. These devices, which combine the fields of NEMS and nanophotonics, consist of two doubly clamped photonic crystal nanobeam cavities separated by a gap smaller than 100 nm. Coupling between cavity modes results in two super modes with even and odd symmetry [8, 21] – the situation analogous to coupled quantum wells. The symmetric mode of such coupled system is extremely dispersive to the separation between the nanobeams which allows for the tuning of this mode by controlling the separation. On the other hand, the asymmetric mode of the system is virtually insensitive to the changes in the separation, which can be attributed to higher-order effects such as a coupling-induced frequency shift [21]. The separation between nanobeams can be controlled using capacitive force, for example: An external bias voltage can be applied to two gold electrodes (Fig. 6a) that are in contact with Si with which the cavities were made of. In this way, the two cavities form the plates of a parallel plate capacitor. The capacitive force is strong enough to bend the beams toward each other as shown in Fig. 6b, which in turn has a strong effect on the resonant wavelength of the structure. Using this approach, tunable filters with a tuning range of 10 nm for less than 6 V bias and negligible steady-state power consumption were demonstrated. The latter is a very important feature, because once reconfigured, they do not dissipate any power. In addition, due to their low mass and flexibility, the energy required to deflect nanobeams is extremely low, and is estimated to be well below the femtojoule level. Therefore, nanoscale optomechanical devices can be used to realize programmable optical systems that
Photonic Crystal Nanobeam Cavities
P
2065
c 1558
50
Wavelength (nm)
1556 V+
40
1554 30 1552 20
Wavelength δλ/δV
1550
V-
δλ/δV (nm/V)
a
10
1548 Experiment 1546 0
1
2
4 3 Voltage (V)
5
6
0
b 0V
6V
3V
6.8V
Photonic Crystal Nanobeam Cavities, Fig. 6 (a) SEM image of the fabricated structure with gold contact pads. The insets show the field profiles for even and odd super modes. (b) SEM images showing the reduction of gaps for different applied
potentials between the two nanobeams; (c) red graph shows the tuning of the even mode for various applied voltages while the blue graph plots the sensitivity at the operating points [22]
can be pre-configured (and reconfigured) using electrical signals, and will likely find applications in lowpower optical circuit switching and programmable photonics. It should also be noted that when properly biased, only a few mV of additional electrical signal is enough to move the nanobeams and induce the shift in cavity resonance larger than one linewidth, thus completely detuning the cavity from the incoming optical signal (originally in resonance with the structure) [22]. Significant improvements in the tuning range can be obtained by using more sophisticated actuation methods. For example, Fig. 7 shows a tuning technique that uses a mechanical comb drive to move one of the cavities thus changing the gap between the cavities and resonance of the system [23]. This technique can achieve extremely small gaps as it does not suffer from the pull-in effects which can be detrimental for capacitive actuation approach discussed above. In addition to electrical actuation, nanoscale optical devices can be actuated using light [8, 24]. Figure 8a shows a structure consisting of freestanding coupled
nanobeam cavities interfaced with optical waveguides and fiber couplers (for efficient in- and out-coupling of light). Optical transmission through the system (experimental) is shown in Fig. 8b, whereas the tuning of cavity resonance for 6 mW of pump power at 1,583 nm is shown in Fig. 8c. The cavity tuning was due to contributions from thermo-optical, nonlinear optical, and optomechanical effects [24]. Alternatively, optomechanical systems can be used to read out mechanical motion optically. Figure 9 shows the mechanical spectrum of the structure in air (mechanical modes excited by Brownian motion), detected by monitoring optical transmission through the device. However, it is important to note that this measurement also affects the mechanical properties of the devices, as seen in Fig. 9: Optical signal can stiffen or soften mechanical resonances (induce an additional effective spring constant) via optical gradient force [8]. Owing to their small mode volume and footprint, nanobeam cavities are also excellent candidates for the realization of low-threshold nanolasers with a
P
P
2066
Photonic Crystal Nanobeam Cavities
Photonic Crystal Nanobeam Cavities, Fig. 7 SEM image showing the PCNCs and comb drive used to achieve tuning of the resonant wavelength [23]
Output Elliptical suspension Coupled 1D PhC nanocavities Nanomechanical comb-drive actuators Cavity A
ad
g
Tilted-folded beam suspension
Cavity B
Suspended nanowire waveguides
1.0 0.8 0.6 0.4 Probe 0.2
Pump
0.0 1500
400nm
1520
1540
1560
1580
1600
Wavelength (nm)
c 1.0 Normalized transmission
Photonic Crystal Nanobeam Cavities, Fig. 8 Integrated optomechanical system based on coupled nanobeam cavities (a) SEM image showing the complete device with the fibercoupling pads, interferometer, waveguides, and the suspended nanobeam cavity region (inset). (b) A transmission spectrum of the fabricated cavity. The even cavity mode – probe – is at 1,507 nm (Q 15,000) and the shaded region can be used to pump outside the bandgap. (c) Cavity resonance was tuned from 1,507.72 to 1,508.19 nm when 6 mW of pump power at 1,583 nm was used. Cavity tuning was due to contribution from thermooptical, nonlinear, optical, and optomechanical effects
b Normalized transmission
a
5μm
Input
0mW 6mW
0.8 0.6 0.4 0.2 0.0
large spontaneous emission (b) factor [19, 25]. Furthermore, it is possible to realize tunable/programmable nanolasers by taking advantage of mechanical degrees of freedom of coupled nanobeam cavities, as
1507.6 1508.0 Wavelength (nm)
1508.4
discussed above [26]. PCNCs can also be designed to support TE and TM modes simultaneously, which is of interest for nonlinear frequency conversions [12]. Scaling properties of photonic crystals can be used in
Photonic Crystal Nanobeam Cavities h1d
h1c
a
P
2067
V1
h3
−85 −90 −95
h2
−100 −105
−73 −75 −77 −79 −81
V3
V2
h4
−110
V6
V1
6.0
h5
7.0
V7
V5
8.0
10−15
9.0
h7
−115
10−14
h1c,h1d
h9
h8
h6
10−16
−120 0
50
100
h1 displacement sensitivity (m Hz−1/2)
Radio-frequency PSD (dBm Hz−1)
X
h1
−75 −80
150
b
Radio-frequency power (dBm)
Frequency (MHz)
−5 −10 −15 −20 −25 −30 −4
−3
−2
−1
0
1
2
3
4
8.6 Frequency (MHz)
8.4 8.2 8 7.8 7.6 7.4 7.2
−4
−3
−2
−1
0
1
2
3
4
Normalized detuning, 2D¢º /G
Photonic Crystal Nanobeam Cavities, Fig. 9 Zipper cavities (a) FEM simulation showing the lowest-order common and differential mechanical modes. Detected radio frequency spectrum showing various in-plane and out-of-plane mechanical
modes of zipper structure. (b) Power and resonance frequency of fundamental mechanical mode (common) for various optical probe detunings
P
P
2068
a
Photonic Crystal Nanobeam Cavities
Trapped Nanoparticle
b
Bus-Waveguide
1-D Photonic Crystal Resonator Buried Oxide Layer Particle trapped and released back on waveguide
Photonic Crystal Nanobeam Cavities, Fig. 10 (a) Schematic of side-coupled PCNC used for trapping polystyrene particles; (b) snapshots showing particle being trapped by the cavity field and propelled by the guided modes of the waveguide [26]
PCNCs to design cavities in microwave regime allowing for integrating of RF with photonics. Finally, PCNCs are excellent candidates for applications in sensing [6, 7] and for manipulation of particles suspended in fluid (optical trapping). Figure 10a shows the schematic of the device while Fig. 10b reports snapshots from a movie showing particle manipulation. Polystyrene particles were propelled along the waveguides by the guided modes and trapped by the intense field in the cavity region when the laser was tuned to the cavity resonance.
Outlook Nanobeam photonic crystal cavities have developed significantly since they were first proposed more than a decade ago. Greater understanding and improvements in the design methods along with the advancement in nanofabrication processes has significantly helped research outgrow the challenges in making ultra-high Q wavelength-scale optical resonators. The last few years of research has seen nanobeam photonic crystal cavities mature as a photonic device and allowed researchers to not only solve current technological problems but also explore new research areas involving enhanced light-matter interaction. Their use in tunable filters, switches, delay lines, sensing, particle manipulation, etc., has already been demonstrated. Their wavelength-scale dimensions, flexible structure, and small footprint area along with the scalable nature make them ideal to be used for applications in the entire electromagnetic spectrum. Emerging areas of research include RF, Terahertz, and Mid-IR photonics.
Cross-References ▶ Capacitive MEMS Switches ▶ Light Localization for Nano-optical Devices ▶ Nanophotonic Structures for Biosensing ▶ Nanostructures for Photonics ▶ Whispering Gallery Mode Resonator Biosensors
References 1. Vahala, K.J.: Optical microcavities. Nature 424(6950), 839–846 (2003) 2. Yablonovitch, E.: Inhibited spontaneous emission in 3 dimensionally modulated periodic dielectric structures. Journal De Physique 48(C-5), 615–616 (1987) 3. John, S.: Strong localization of photons in certain disordered dielectric superlattices. Phys. Rev. Lett. 58(23), 2486–2489 (1987) 4. Joannopoulos, J.D., et al.: Photonic Crystals: Molding the Flow of Light. Princeton University Press, Princeton (2008) 5. Gong, Y., Vuckovic, J.: Photonic crystal cavities in silicon dioxide. Appl. Phys. Lett. 96(3), 031107 (2010) 6. Quan, Q.M., et al.: High-Q, low index-contrast polymeric photonic crystal nanobeam cavities. Opt. Express 19(22), 22191–22197 (2011) 7. Mandal, S., Serey, X., Erickson, D.: Nanomanipulation using silicon photonic crystal resonators. Nano Lett. 10(1), 99–104 (2010) 8. Eichenfield, M., et al.: A picogram- and nanometre-scale photonic-crystal optomechanical cavity. Nature 459(7246), 550-U79 (2009) 9. Eichenfield, M., et al.: Optomechanical crystals. Nature 462(7269), 78–82 (2009) 10. Quan, Q.M., Deotare, P.B., Loncar, M.: Photonic crystal nanobeam cavity strongly coupled to the feeding waveguide. Appl. Phys. Lett. 96(20), 203102 (2010)
Physical Vapor Deposition 11. Zain, A.R.M., et al.: Ultra high quality factor one dimensional photonic crystal/photonic wire micro-cavities in silicon-on-insulator (SOI). Opt. Express 16(16), 12084–12089 (2008) 12. McCutcheon, M.W., et al.: High-Q transverse-electric/ transverse-magnetic photonic crystal nanobeam cavities. Appl. Phys. Lett. 98, 111117 (2011) 13. Villeneuve, P.R., et al.: Air-Bridge Microcavities. Appl. Phys. Lett. 67(2), 167–169 (1995) 14. Foresi, J.S., et al.: Photonic-bandgap microcavities in optical waveguides. Nature 390(6656), 143–145 (1997) 15. Lalanne, P., Hugonin, J.P.: Bloch-wave engineering for high-Q, small-V microcavities. IEEE J. Quantum. Electron. 39(11), 1430–1438 (2003) 16. McCutcheon, M.W., Loncar, M.: Design of a silicon nitride photonic crystal nanocavity with a quality factor of one million for coupling to a diamond nanocrystal. Opt. Express 16(23), 19136–19145 (2008) 17. Deotare, P.B., et al.: High quality factor photonic crystal nanobeam cavities. Appl. Phys. Lett. 94(12), 121106 (2009) 18. Notomi, M., Kuramochi, E., Taniyama, H.: Ultrahigh-Q nanocavity with 1D photonic gap. Opt. Express 16(15), 11095–11102 (2008) 19. Ahn, B.H., et al.: One-dimensional parabolic-beam photonic crystal laser. Opt. Express 18(6), 5654–5660 (2010) 20. Safavi-Naeini, A.H., et al.: Electromagnetically induced transparency and slow light with optomechanics. Nature 472(7341), 69–73 (2011) 21. Deotare, P.B., et al.: Coupled photonic crystal nanobeam cavities. Appl. Phys. Lett. 95(3), 31102 (2009) 22. Frank, I.W., et al.: Programmable photonic crystal nanobeam cavities. Opt. Express 18(8), 8705–8712 (2010) 23. Chew, X.Y., et al.: Dynamic tuning of an optical resonator through MEMS-driven coupled photonic crystal nanocavities. Opt. Lett. 35(15), 2517–2519 (2010) 24. Deotare, P.B., et al.: All optical reconfiguration of optomechanical filters, accepted Nature Communications. DOI: 10.1038/ncomms1830 (2012) 25. Zhang, Y., et al.: Photonic crystal nanobeam lasers. Appl. Phys. Lett. 97(5), 051104 (2010) 26. Perahia, R., et al.: Electrostatically tunable optomechanical “zipper” cavity laser. Appl. Phys. Lett. 97(19), 191112 (2010)
Photonics in Nature ▶ Biomimetics of Optical Nanostructures
Photovoltaic Devices ▶ Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
2069
P
Physical Colors ▶ Nanostructures for Coloration (Organisms other than Animals)
Physical Dry Etching ▶ Dry Etching
Physical Modification ▶ Nanostructures for Surface Functionalization and Surface Properties
Physical Vapor Deposition Yoke Khin Yap Department of Physics, Michigan Technological University, Houghton, MI, USA
Synonyms Anodic arc deposition; Arc discharge; Boron nitride nanotubes (BNNTs); Carbon nanotubes (CNTs); Catalyst; Cathodic arc deposition; Chemical beam epitaxial (CBE); Electron beam evaporation; Electron beam physical vapor deposition (EBPVD); Gallium arsenide (GaAs); Gas-phase molecular beam epitaxy (gas-phase MBE); Graphene; Ion beam; Magnetron sputtering; Metal-organic molecular beam epitaxy (MOMBE); Molecular beam epitaxy (MBE); Nanorods; Nanotubes; Nanowires; Plasma; Pulsed-laser deposition (PLD); Si nanotubes; Sputtering; Thermal evaporation
Definition Physical vapor deposition (PVD) is referred to deposition processes of thin films and nanostructures through the evaporation of solid precursors into their vapor phase by physical approaches followed by the
P
P
2070
Physical Vapor Deposition
condensation of the vapor phase on substrates. The whole process consists of three stages: (1) evaporation of the solid source, (2) vapor phase transport from the source to the substrates, and (3) vapor condensation on the substrates. Since PVD technique can convert solid materials into vapor phase without chemical processes, they are very convenient and possible to deposit many types of materials into thin films and nanostructures.
Classification PVD techniques can be classified based on the techniques used to evaporate the solid source materials into their vapor phase [1, 2]. These techniques include (1) thermal evaporation, (2) ion sputtering, and (3) arc discharge. In the following subsections, examples of PVD based on these evaporation techniques will be described. PVD by Thermal Evaporation Molecular Beam Epitaxy
Molecular Beam Epitaxy (MBE) is a PVD technique conducted under ultrahigh vacuum (UHV, better than 109 mbar) where epitaxial deposition of thin film crystal on crystalline substrates can be obtained by using one or more vapor sources (Effusion cells or sometimes called Knudsen cells). An effusion cell consists of a crucible (made of graphite, pyrolytic boron nitride, quartz, or tungsten) where the solid source can be heated and evaporated by filaments (Fig. 1). For example, powders of gallium (Ga) and arsenic (As) can be evaporated in separated cells until they are sublimate into vapors. When these separated beam of element vapors are condensed on a heated Si substrate, crystallized thin films of gallium arsenide (GaAs) will be formed. A typical MBE system has a reflection highenergy electron diffraction (RHEED) system to provide in situ monitoring of the crystal growth processes. The major advantage of MBE comes from its controllable deposition rate at the atomic or molecular scale. This can be achieved due to the use of effusion cells that enable the generation of “molecular beam,” where the generated vapors have low interparticle collisions before they reach to the substrate surface. This means the evaporated atoms escape from the orifice of the cells through effusion and have long mean free path. They neither interact with each other nor with the gases in the vacuum chamber. Due to this capability, MBE was recently used for the growth of graphene
Physical Vapor Deposition, Fig. 1 Schematic drawing of a typical molecular beam epitaxy system
but the outcome is debatable [3, 4]. Finally, gases were also used as the precursors instead of solid sources and such configuration is called gas-phase MBE or more appropriately chemical beam epitaxial (CBE) [5]. In this case, the pyrolysis of either trimethylindium (TMIn) or triethylindium (TEIn), and trimethylgallium (TMGa) or triethylgallium (TEGa) at the heated substrates were used as the In and Ga source, while As2 and P2 were obtained by thermal decomposition of the trimethylarsine (TMAs) and triethylphosphine (TEP). Such approach is more like a combined MBE and metal-organic chemical vapor deposition (MOCVD) and is sometimes called metal-organic molecular beam epitaxy (MOMBE). In fact, MBE is an advanced form of thermal evaporation technique with highly controllable deposition thickness (and thus with the drawback of slow deposition rate) due to the use of effusion cells and high vacuum. For the deposition of amorphous this films, much simple setup can be employed with less demanding vacuum level (102 mbar in a glass bell jar chamber). This approach is simply called direct resistive evaporation and can be achieved by passing a large current through a resistive wire or foil containing the solid material to be deposited. The heating element is often referred to as an “evaporation source.” The evaporation source can appear in many forms including tungsten wires in the form of filaments or baskets, and thin tungsten, tantalum, or molybdenum foils in the form of boats. PVD by Electron Beam (E-Beam) Evaporation
In contrast to MBE, electron beam physical vapor deposition (EBPVD) provides very high deposition
Physical Vapor Deposition
rate on large deposition areas. EBPVD is referring to thin film deposition initiated by electron beam evaporation. The deposition rate is typically ranging from tens of nanometer/minute to hundreds of micrometer/minute, although a rate as high as 50 mm/s was reported [6]. EBPVD is based on electron beam evaporation of anode materials (usually in ingot form) in high vacuum (104 mbar). A typical EBPVD system can have multiple electron guns and sources. The electron beams can be generated by either thermionic emission or field emission of filaments. These beams are then accelerated to a high kinetic energy and focused on the ingot. The energy of these electrons is converted into thermal energy as the beam bombards the surface of the ingot. The high deposition rate of EBPVD without corrosive products meets the needs of industrial applications. A potential drawback of EBPVD is that filament degradation in the electron gun results in a nonuniform evaporation rate. In addition, EBPVD may cost contamination of the deposition chamber as the vapors generated from the source materials are coated everywhere in the vacuum chamber, the so-called line-of-sight process. Pulsed-Laser Deposition
Pulsed-laser deposition (PLD) is a PVD technique where a high-power pulsed-laser is used to generate vapors (i.e., laser ablation) from a solid target followed by the condensation of the vapors on substrates. The experimental setup for PLD is relatively simpler than many PVD techniques but the laser ablation mechanism is quite complex [7]. As shown in Fig. 2a, a PLD system consists of a pulsed-laser, a vacuum chamber with a rotating target holder and a substrate holder with a heater. PLD can be performed in vacuum at various substrate temperatures. During film deposition, laser pulses are focused on the surface of a solid target through a viewport of the chamber by using a thin lens. This will generate a laser spot with very high irradiance (I, power per unit area) and local electric field, where I ¼ ½eo c Eo2 [8]. Here, Eo is the amplitude of electric field of the laser, c is the speed of light in vacuum, and ϵ0 is the vacuum permittivity. The local electric field is sufficient to cause atomic bond breaking and dielectric breakdown of the target materials to initiate vaporization of the target materials. This vapor will absorb the remaining portion of the laser pulse, and the electric field of the pulse will accelerate electrons and ions
2071
P
inside the vapor to create rapid heating and to form laser plasma. Continued absorption of the laser pulse will cause a highly directional expansion of the laser plasma perpendicular to the target surface. This expanding vapor is called laser plume that propagates toward the substrate placed several centimeters opposite to the target. PLD can be conducted in vacuum or in a lowpressured gas ambient. When PLD is conducted in reactive gas ambient (say oxygen or nitrogen) the technique is called reactive-PLD. In this case, the expanding laser plume will form shock waves and cause chemical reaction between the target vapor and the gas molecules. PLD can also be modified into various configurations. For example, plasma can be applied to enhance the interaction of the vapors with the gas ambient and the technique is called plasmaenhanced or plasma-assisted PLD. For example, such an approach was shown to control the phases of nitride thin films [9, 10]. Sometimes, an auxiliary ion beam is irradiated on the surface of the substrates during thin film deposition as shown in Fig. 2b, and the technique is called ion beam-assisted PLD. In fact, the mechanism of laser ablation and the quality of samples deposited by PLD depend strongly on the wavelength and energy of the pulsed-laser. This can be understood by the absorption of the laser light by the laser plasma when the plasma index of refraction, n(o), becomes complex. According to the relation n2 (o) ¼ 1 [op/o]2, a complex n(o) will occur when op > o [8]. This means, shorter laser wavelength (larger o) will reduce the chances of such plasma heating effect. However, one should also consider that the laser plasma frequency, op ¼ [N e2 me e0]1/2, where me is the mass of electron and N is the electron density. In this case, to avoid plasma heating, N must be minimized, which can be obtained by keeping the laser energy or irradiance low. Plasma heating will lead to the so-called photothermal ablation. For example, photothermal ablation of h-BN target will transfer h-BN target to the substrates and prevent the formation of cubic phase (c-BN) [8]. Photothermal ablation of graphite target at longer laser wavelength will also prevent the formation of diamond-like carbon film [11] as the sp2 species from the graphite are transferred to the substrates. PVD by Sputtering Sputtering is a process where atoms are released from a solid target material due to bombardment of the
P
P
2072
Physical Vapor Deposition
Physical Vapor Deposition, Fig. 2 Schematic drawings of a typical pulsed-laser deposition (PLD) system operated (a) in vacuum or with auxiliary plasma, or (b) with auxiliary ion beam
target by energetic particles such as ions and atoms. The atoms released from the target can then be deposited as thin film on substrates and is commonly referred as PVD by sputtering. The physics behind the process is momentum transfer from the incident ions/atoms to the target materials through collisions. The average number of atoms ejected from the target per incident particle is called the sputter yield. The yield is depend on the ion incident angle, the energy of the ion, the masses of the ion and target atoms, the surface binding energy of atoms in the target, and the crystalline orientation of the target when crystals are used as the target. There are many approaches to generate sputter including the use of DC, AC, and RF plasmas or external ion beams. Sputtering offers several advantages over other PVD techniques including: (1) large area deposition as compared to PLD; (2) convenient for thin film deposition of alloy/composite and materials with high melting points; and (3) avoid device damaging from X-rays generated by electron beam evaporation. However, sputtering also has the following disadvantages: (1) more complicated setup than PLD; (2) contamination is more likely to occur due to the use of plasma as well as relatively lesser vacuum level; and (3) the film morphology may be rougher and even damaged due to bombardment of energetic growth species and clustering of the growth species in the relatively high deposition pressures.
yield and uniform glow discharge. These Ar ions will bombard the target surface to generate vapors that will be deposited to the substrates. Despite the simplicity, the DC configuration will have the following issues: (1) positive charges will build up on the target surface and cause sparking within the plasma and (2) not suitable for insulating target. To avoid these issues, AC and RF configurations are often used. In these cases, the potential between the target and substrate is alternating so that the target will be negative in potential more often (to generate sputtering) than positive potential. These configurations will also allow lower operation pressure (10–30 mTorr). The use of magnetron can further enhance the deposition rate at even lower pressures by the use of DC, AC, or RF electric fields. RF potential is of advantage since it can be used for insulating targets and is applicable in an antenna configuration, that is, without positive bias voltages on the substrates (less electron bombardments and less substrate heating). In this case, the target is placed on the magnetron surrounded by strong electric and magnetic fields. Electrons in the RF plasma will follow helical paths around the magnetic field lines and generate more ionizing collisions near the target surface with the surrounding gas. Thus, more ions will be created and will lead to a higher sputtering and deposition rates. Because of higher ionization rates, plasma can be sustained at a lower pressure.
PVD by DC, AC, RF, and Magnetron Sputtering
PVD by Ion-Beam Sputtering
PVD can be obtained by applying plasma in various configurations. The simplest one is to apply a DC potential across the substrate (anode) and target (cathode). In this case, plasma can be generated when gases are introduced (10s–100s of mTorr) at sufficient electric field strength ( a few kV/cm). For example, argon gas is often used as argon ion can lead to high sputter
Ion-beam sputtering (IBS) is a PVD technique in which the target is evaporated by an external ion source (ion beam). In a typical ion source, ions are generated by collisions of neutral gas atoms/molecules with electrons. They are then accelerated by the electric field by a grid electrode toward the target. Before these ions leave the source, they are neutralized by electrons from
Physical Vapor Deposition
a second external filament. The major advantage of PVD by IBS is that the energy and flux of ions can be controlled independently. In addition, the IBS approach is applicable to insulating and conducting targets since the flux of “ions” (which are neutralized as they are emitted from the ion beam) that strikes the target is composed of neutral atoms/molecules. Reactive Sputtering
Like the PLD process, reactive sputtering can be achieved by mixing reactive gases (N2, O2, etc.) with the inert gas (Ar, Kr, etc.). While the inert gas is responsible for the sputtering of the target materials, the reactive gases are used to initiate chemical reactions with the target vapors to form oxide or nitride films. Like PLD, reactive sputtering can also be achieved by various configurations. For example, plasma can be applied to enhance the interaction between the target vapors and the reactive gas ambient and the technique is called plasma-enhanced or plasma-assisted deposition. Sometimes, an auxiliary ion beam is irradiated on the surface of the substrates during thin film deposition, and the technique is called ion beam-assisted deposition (IBAD). PVD by Arc Discharge In addition to PVD by thermal evaporation and sputtering, the desired vapors for thin film deposition can be obtained by arc discharge of the target. This is sometimes called arc vapor deposition [1], where a high-current (104–106 A/cm2), low-voltage (tens of voltages) discharge was applied on the target. For example, cathodic arc deposition is referring arc vapor deposition using the target as the cathode. In this case, a high-voltage “trigger arc” will be ignited between the cathode and a sharp auxiliary anode. This will form “seed” electrons and ions to initiate a lowvoltage, high-current discharge between the target cathode and the adjacent anode. On the other hand, target materials can be evaporated by using the target as the anode of the arc discharge. This PVD process is referred as anodic arc deposition. In fact, this process is similar to electron beam physical vapor deposition (EBPVD) discussed earlier, where arc discharge is generated by irradiating an unfocused electron beam on the target anode. The electrons can be made to spiral in a magnetic field to further ionize the evaporated materials from the target prior to deposition. Like most other PVD processes,
2073
P
reactive arc vapor deposition can be performed by both anodic arc deposition and cathodic arc deposition using reactive gases like nitrogen and oxygen. This will lead to the formation of nitride and oxide films. The deposition rates of arc vapor deposition are usually higher than that of PVD by sputtering. This is one of the advantages of anodic arc deposition and cathodic arc deposition in addition to the low operation voltage. On the other hand, contamination is the potential drawback of most plasma-based PVD processes. In addition, particulates or the so-called macros are undesired for thin films deposited by anodic arc deposition and cathodic arc deposition. Macros are formed by ablation of molten or solid particles by thermal shock during arc discharges.
Examples of PVD Approaches for Nanotechnology Physical evaporation techniques like arc discharge [12] and laser ablation [13] are important on the discovery of carbon nanotubes (CNTs). However, PVD techniques discussed here are not popular for the growth of nanomaterials in comparison to chemical vapor deposition (CVD). The major reason is the tendency of forming thin films and amorphous byproducts on the surfaces of catalysts which cause “poisoning” effect and prohibit the formation of nanotubes or nanowires [14]. However, there have been some pioneering works reported on the growth of nanotubes and nanowires by PVD techniques such as sputtering, PLD, and MBE. For example, vertically aligned carbon nanotubes (CNTs) were first successfully deposited by RF plasma-assisted pulsed-laser deposition [15]. In this approach, two RF plasmas were employed. The first one was applied on the substrates to form a negative bias voltage during deposition. The second one was applied on a ring electrode located between the graphite target and the substrates. Hydrogen gas was used for the plasma generation and Fe powders (20 nm in diameter) were employed as the catalysts. Although the density of these CNTs is low, these nanotubes have high structural order with internal nanotube diameter as small as 0.4 nm. The formation of these smallest possible CNTs was confirmed by transmission electron microscopy (TEM). Later, CNTs were deposited by magnetron sputtering with a dc bias voltage on substrates [16]. Prior to the growth of CNTs,
P
P
2074
Ni films (10 nm thick) were deposited by e-beam evaporation to form nanoparticles that later function as the catalyst for the vapor-liquid-solid (VLS) process. However, these CNTs are defective with bamboo-like structures. Later, RF plasma-assisted pulsed-laser deposition was used to achieve low-temperature growth of boron nitride nanotubes (BNNTs) [14, 17]. In this case, pyrolytic boron nitride pallets were used as the targets and a RF plasma was applied on the substrates to create bias voltages, which are important to eliminate the formation of boron nitride thin films. Again, these BNNTs have internal nanotubes of 1 nm in diameter with high structural order. Various PVD approaches were also employed for the synthesis of nanowires/nanorods. For example, ZnO and Zn1xMgxO nanorods were grown by MBE on oxidized Si substrate coated with Ag catalyst [18]. In both cases, an ozone/oxygen mixture was used as the oxidizing source. The Zn and Mg cation fluxes were provided by a Knudsen effusion cell using high purity Zn and Mg metals as the sources. The growth of vertically aligned ZnO [19] and MgO [20] nanowires can also be obtained by PLD using ZnO or MgO targets, respectively. Gold particles were used as the catalysts for both cases.
Cross-References ▶ Atomic Layer Deposition ▶ Carbon Nanotube-Metal Contact ▶ Carbon Nanotubes for Chip Interconnections ▶ Carbon-Nanotubes ▶ Chemical Vapor Deposition (CVD) ▶ Electron-Beam-Induced Deposition ▶ Focused-Ion-Beam Chemical-Vapor-Deposition (FIB-CVD) ▶ Graphene ▶ Growth of Silica Nanowires ▶ Nanomaterials for Electrical Energy Storage Devices ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanotechnology ▶ Synthesis of Carbon Nanotubes
Physical Vapor Deposition
2. 3.
4.
5. 6.
7. 8.
9.
10.
11.
12. 13. 14. 15.
16.
17.
18.
References 19. 1. Mattox, D.M.: Handbook of Physical Vapor Deposition (PVD) Processing: Film Formation, Adhesion, Surface
Preparation and Contamination Control (Hardcover). Noyes, New Jersey (1998) Mahan, J.E.: Physical Vapor Deposition of Thin Films. Wiley-Interscience, New York (2000) Park, J., et al.: Epitaxial graphene growth by carbon molecular beam epitaxial (CMBE). Adv. Mater. 22, 4140–4145 (2010) Hackley, J., et al.: Graphitic carbon growth on Si (111) using solid source molecular beam epitaxy. Appl. Phys. Lett. 95, 133114 (2009) Tsang, W.T., et al.: Chemical beam epitaxy of InP and GaAs. Appl. Phys. Lett. 45, 1234–1236 (1984) Schiller, S., J€asch, G.: Deposition by electron beam evaporation with rates of up to 50 mm s1. Thin Solid Films 54, 9–21 (1978) Chrisey, D.B., Hubler, G.H. (eds.): Pulsed Laser Deposition of Thin Films. Wiley-Interscience, New York (1994) Wang, J., Yap, Y.K.: Growth of adhesive cubic phase boron nitride films without argon ion bombardment. Diam. Relat. Mater. 15, 444–447 (2006) Yap, Y.K., Kida, S., Aoyama, T., Mori, Y., Sasaki, T.: Influence of negative dc bias voltage on structural transformation of carbon nitride at 600 C. Appl. Phys. Lett. 73, 915 (1998) Yap, Y.K., Aoyama, T., Kida, S., Mori, Y., Sasaki, T.: Synthesis of adhesive c-BN films in pure nitrogen radio-frequency plasma. Diam. Relat. Mater. 8, 382–385 (1999) Yamamoto, K., Koga, Y., Fujiwara, S., Kokai, F., Heimann, B.: Dependence of the sp3 bond fraction on the laser wavelength in thin carbon films prepared by pulsed laser deposition. Appl. Phys A 66, 115–117 (1998) Iijima, S.: Helical microtubules of graphitic carbon. Nature 354, 56–58 (1991) Iijima, S., Ichihashi, T.: Single-shell carbon nanotubes of 1-nm diameter. Nature 363, 603–605 (1993) Wang, J., et al.: Low temperature growth of boron nitride nanotubes on substrates. Nano Lett. 5, 2528–2532 (2005) Yap, Y.K., Yoshimura, M., Mori, Y., Sasaki, T., Hanada, T.: Formation of aligned-carbon nanotubes by RF-plasmaassisted pulsed-laser deposition. Special issue for Tsukuba symposium on carbon nanotubes in commemoration of the 10th anniversary of its discovery; S. Ijima et al. eds., Physica B 323, 341–343 (2002) Lee, K.-F., et al.: Synthesis of aligned bamboo-like carbon nanotubes using radio frequency magnetron sputtering. J. Vac. Sci. Technol. B 21, 1437–1441 (2003) Xie, M., Wang, J., Yap, Y.K.: Mechanism for low temperature growth of boron nitride nanotubes. J. Phys. Chem. C 114, 16236–16241 (2010) Heo, Y.W., Kaufman, M., Pruessner, K., Norton, D.P., Ren, F., Chisholm, M.F., Fleming, P.H.: Optical properties of Zn1xMgxO nanorods using catalysis-driven molecular beam epitaxy. Solid-State Electron. 47, 2269–2273 (2003); Heo, Y.W., Varadarajan, V., Kaufman, M., Kim, K., Norton, D.P., Ren, F., Fleming, P.H.: Site-specific growth of Zno nanorods using catalysis-driven molecularbeam epitaxy. Appl. Phys. Lett. 81, 3046–3048 (2002) Rahm, A., et al.: Pulsed-laser deposition and characterization of ZnO nanowires with regular lateral arrangement. Appl. Phys. A 88, 31–34 (2007)
Physicochemical Properties of Nanoparticles in Relation with Toxicity 20. Nagashima, K., Yanagida, T., Tanaka, H., Kawai, T.: Epitaxial growth of MgO nanowires by pulsed laser deposition. J. Appl. Phys. 101, 124304 (2007)
Physical-Chemical Etching ▶ Dry Etching
Physicochemical Properties of Nanoparticles in Relation with Toxicity Je´roˆme Rose1,2, Me´lanie Auffan3,4, Olivier Proux2,5, Vincent Niviere2,6 and Jean-Yves Bottero3,4 1 CEREGE UMR 6635– CNRS–Universite´ Paul Ce´zanne Aix–Marseille III, Aix–Marseille Universite´, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France 2 GDRI ICEINT: International Center for the Environmental Implications of Nanotechnology, CNRS–CEA, Europoˆle de l’Arbois BP 80, Aix–en–Provence Cedex 4, France 3 CEREGE, UMR 6635 CNRS/Aix–Marseille Universite´, Aix–en–Provence, France 4 iCEINT, International Consortium for the Environmental Implications of Nanotechnology, Center for the Environmental Implications of NanoTechnology, Aix–En–Provence, Cedex 4, France 5 OSUG, Universite´ Joseph Fourier BP 53, Grenoble, France 6 Laboratoire de Chimie et Biologie des Me´taux, UMR 5249, iRTSV–CEA Bat. K’, 17 avenue des Martyrs, Grenoble Cedex 9, France
Synonyms Engineered nanoparticles; Nano-effects; Nanomaterials and nanoproducts; Nanotoxicology
Definition Enhancement or modifications of properties in the nanometric range is the core of nanotechnologies. Nanoparticles exhibit size-related properties that differ significantly from those observed in fine
2075
P
particles or bulk materials. Properties can be intrinsic (e.g., electrical, magnetic, optical, etc.) or extrinsic (e.g., aggregation state, surface area, etc.). Then what makes nanoparticles so attractive also raises questions of “new” toxic “nano-effects.” It is then worth trying to identify which specific properties can generate biological harmful effects. More than these “new” nano-properties aspects, the second issue is to manufacture materials that incorporate those nanoparticles. Then exposure to nanoparticles will depend not only on intrinsic properties of nanoparticles, but also on extrinsic properties related to the stability and durability of their incorporation.
Overview Introduction The success story of nanoparticles in many scientific and technological fields is based on their unique physicochemical properties and reactivity. Indeed engineered nanoparticles are designed to exhibit specific optical, electronic, magnetic, or chemical properties that are strongly related to the exponential increase in the number of atoms localized at the surface as the size decreases [1]. However, the uniqueness of nanoparticles that makes them so attractive may also raise questions about the environmental and health risk they could generate. Indeed is there any relation between their new physicochemical properties and potential new biological hazards? Moreover the tremendous development of nanotechnologies leads to a huge variety of manufactured nanoparticles with varying chemical composition, size, shape, crystal structure, surface area, surface chemistry and charge, solubility. . . etc., e.g., [1]. It is therefore challenging to determine which of those unique intrinsic or extrinsic properties can be at the origin of their biological danger. In the following, we will focus on the different properties, how to determine their evolution in biological media and how they can affect living organisms.
Physicochemical Properties of Nanoparticles Even if there is no consensus concerning the definition of engineered nanoparticles, most of the definitions indicate that besides the size, “Novel properties
P
P
2076
Physicochemical Properties of Nanoparticles in Relation with Toxicity
differentiate nanoparticles from the bulk material typically develop at a critical length scale of under 100 nm” or “phenomena not observed in larger structures start to become apparent . . .”. But what are these novel properties or phenomena not observed for larger structures? What is the size at which the latter properties occur? One of the most emblematic example concerns gold. Indeed gold has been a topic of fascination since its discovery in almost all ancient and modern civilizations. Gold preserve its brightness for years which made his high value for art. Everybody has in mind that the magnificent death mask of Egyptian pharaoh Tutankhamun is made of gold. Moreover, gold chemical inertness is related to its high monetary value [2]. But its chemical inertness is a matter of size since at the nanoscale it can be exceptionally active as a catalyst for instance but only for particles below 2–3 nm [3]. Among the vast diversity of engineered nanoparticles, many other properties can be tailored in different scientific and technological fields. Indeed optical properties can be modified to use nanoparticles as biological dyes for instance. QDots are of good examples to trace, label, or image biological organisms or cells. In the case of QDots, the band gap energy (related to the fluorescence wavelength that is to say the color) is constant for large particles whatever their size but only if larger than 6–8 nm. The bandgap energy strongly increases when size of particles decreases below 6–8 nm. Photocatalytic properties may also be tailored not only as function of size, but also as function as the crystal structure. TiO2 anatase phase exhibits optimum photocatalytic degradation of organic contaminants at various sizes that are much below 100 nm (e.g., 7 nm for trichloroethylene or around 11 nm for chloroform). But in any case, anatase is much more efficient than rutile, indicating that besides size, crystal structure also strongly affects physicochemical properties. Thermal properties may also vary in relation with size decrease. For instance, tin and indium melting point can be strongly reduced by 80 C and 120 C, respectively (e.g., in [1]) when their diameters decrease from 100 to 10 nm. More surprisingly, their normalized heat of fusion that is supposed to remain constant in classical thermodynamics exponentially decreases when size decreases below 20 nm in diameter. Solubility is also affected by size decrease. It is intuitive to predict that the dissolution kinetics of
particles is function of the specific area and aggregation state, with of course faster dissolution kinetic for the smallest particles, but what about the thermodynamic crystal solubility Kb? Kb is supposed to be constant and generally similar to the solubility product Ksp. ln Kb ¼ ln Ksp þ f ðg=rÞ (with g is the surface tension and r is the characteristic length of the crystal) For particles larger than 25 nm, f ðg=rÞ can be approximated as a constant function. But this simplification fails for smaller nanoparticles as the size dependence of the morphology and g cannot be ignored. Other examples of physicochemical properties modification with size could be given (magnetic, electrical, etc.). Besides these intrinsic properties, it is worth detailing nanoparticles when introduced in aquatic media. Indeed some confusion may exist between dissolution and dispersion. Due to their small size, when nanoparticles are homogeneously distributed in aquatic media, particles cannot be detected by eyes and appear as if particles may be totally dissolved to ionic species. However, (colloidal) dispersion is the term referred to suspended material in the 1–250 nm range. Colloidal dispersions are thermodynamically unstable even if the aggregation rate can be very slow. Latter examples are good illustrations of the high variety of properties that can change or be enhanced as size of particles decreases in the nanometric size range. It is therefore challenging to prioritize which properties are of main concern in relation with toxicity.
Determination of Nanoparticle’s Properties in Biological Media Generalities Even if challenging, determining the evolution of physicochemical properties of nanoparticles in relation to human and environmental exposure requires to answer to, at least, two questions, in order to define the appropriate useful techniques. First of all, which structural and characterization information are considered to be important? As detailed before their aggregation in the media, their (mean) size and shape, dissolution or solubility, their surface area, redox stability, surface charge, and surface chemistry are of prime interest.
Physicochemical Properties of Nanoparticles in Relation with Toxicity
Secondly, in which conditions the nanoparticles have to be characterized? A three-step procedure have been formalized in which the characterizations need ideally to be performed (1) on as-synthesized nanoparticles in their native state (dry or hydrated state, depending on the synthesis), (2) in situ on nanoparticles dispersed in the media used for the environment and toxicity tests (ultrapure water, phosphate-buffered saline solution cell culture media. . .), and (3) in vivo or in vitro on particles interacting with the living systems tested (bacteria, cells, animals, plants, etc.). Numerous characterization and detection techniques exist to determine these physicochemical properties of nanoparticles in a toxicological or biological context. There is no best technique, each of them has their specificities, advantages, and disadvantages. However, in order to limit at the maximum the discrepancies between the results, the same technique has to be employed for each step of the characterization. Several recent review papers summarize different and complementary analytical techniques (e.g., [4]. . .). We will here focus on recent techniques using X-ray. Their large penetration depth, especially in the hard X-rays regime, is interesting in a large range of applications because of its compatibility with in situ or in vivo conditions, in most of the cases without any special sample preparation, directly on the nanoparticles in suspension, for example. Moreover, specific interactions of X-rays with matter provide elemental and chemical sensitivity that have made X-ray techniques a very attractive tool. X-Ray Imaging Advances in X-ray spectromicroscopy techniques based on new third-generation synchrotron sources have opened unprecedented opportunities in terms of spatial resolution to analyze nano-sized compounds in biological systems [5]. Different possibilities exist to improve the contrast using X-ray microscopes with respect to optical or electron microscopes. They can be schematically described looking at the different interaction processes occurring between the incident photons (or the associated wave) and the sample. (1) The photon absorption which strongly depends on the chemical nature of the element (the higher the atomic number of an element, the stronger will be the absorption). (2) The fluorescence emission resulting from the decay process. (3) The phase shift
2077
P
of the incoming wave induced by the matter/sample interaction. (4) The photon diffraction by the sample. These processes give rise to different kinds of X-ray microscopy, among them Transmission X-ray Microscopy (TXM), Scanning Transmission X-ray Microscopy (STXM), X-ray fluorescence with microbeam (mXRF), and X-ray Diffraction Microscopy (XDM). Examples of imaging performed with these techniques are shown on Fig. 1, on fibroblast cells exposed to nanoparticles using STXM in absorption (ABS) and differential phase contrast (DPC) modes combined with mXRF (Fig. 1, left) [5], and on yeast using XDM compared with Scanning Electron Microscopy (SEM) and STXM (Fig. 1, right) [6]. First example deals with the toxicity of CoFe2O4 nanoparticles, nano-sized magnets that can be potentially used in combined cancer therapy if their biocompatibility is demonstrated. mXRF maps indicate that, for the probed concentrations, these nanoparticles do not penetrate the nucleus. Precise localization of the elements with respect to the different constituting parts of the cell is possible by acquiring at the same time an image, the best contrast being obtained in DPC mode (Fig. 1, left, upper part). Spatial resolution in this softX-ray microscope is sub-50 nm. An improvement of the resolution can be achieved by shifting to DCM microscopy as shown by Nelson et al. [6] who obtained a “resolution between 11–13 nm, the highest resolution in X-ray imaging of a biological specimen.” Saccharomyces cerevisiae yeasts were imaged after labeling using colloidal gold particles. These labels were positioned to correlate the images obtained by the three different ways, to have a probe to estimate the resolution and the probing depth. The 3D XDM image obtained after reconstruction allows to image the internal part of the cell (which can be compared to the image obtained in transmission using STXM) or the two opposite walls of the cell (which can be compared in this case to the two images obtained in SEM). This technique is really promising even if it still remains quite challenging, to image thick cells, too thick to be studied by transmission electron microscopy. From Long-Range to Short-Range Orders in Nanomaterials X-ray scattering and X-ray Absorption Spectroscopy techniques (XAS) techniques provide valuable information on long- and short-range orders existing inside
P
P
2078
Physicochemical Properties of Nanoparticles in Relation with Toxicity
Physicochemical Properties of Nanoparticles in Relation with Toxicity, Fig. 1 Left. Uptake and interaction of CoFe2O4 nanoparticles with fibroblast cells imaged using absorption (ABS) and differential phase contrast (DPC) combined with
mXRF analysis of the C, O, Fe, and Co (From [5]). Right. Saccharomyces cerevisiae yeast imaged using XDM (three images corresponding to three different analyzing depth), STXM, and SEM ([6])
(X-Ray Diffraction, XRD) and between (Small-Angle X-ray Scattering, SAXS) nanoparticles, with a great sensitivity to their shape. SAXS experiments have been performed to follow the kinetics of aggregation or dissolution of ZnS nanoparticles as a function of the mean size of the particles and of the pH of the solution [7]. Levard et al. [8] have combined SAXS and XAS in situ experiments to study the growth mechanisms of imogolitelike aluminogermanate nanotubes at various stages of their formation, from the precursor (rooftile-shaped particles) to the nanotubular structure. 20% of Ge vacancies have been also determined in the Ge layer from XAS analysis. Finally, structure of iron oxide nanoparticles in aqueous suspension was probed by high-energy powder XRD on a large angular range coupled to atomic pair distribution function (PDF) analysis, well suited for determining the internal atomic ordering and the geometry of nanoparticles [9]. This PDF technique provided detailed structural information for g-Fe2O3 spherical and tetrapod-shaped nanoparticles. These techniques are performed using synchrotron radiation in order to perform kinetics experiments and XAS spectroscopy, but SAXS and XRD experiments can also be performed using laboratory sources, preferably at high energy (Mo or Ag Ka anode).
Electronic Structure and Short-Range Order in Nanomaterials XAS is now a well-established structural technique that allows to determine the electronic structure (valence) and the local order (geometry of the site, chemical nature of the neighbors, etc.) of a selected element in a sample. Being only sensitive to the local order, XAS is well suited to study either amorphous, ill-ordered, or well-crystallized samples, in the solid or liquid states, and has been then quite naturally used for the nanoparticles studies [10]. Numerous examples of the application of this technique for nanotoxicology studies can be found (e.g., [11, 12]). In this particular field, two conditions are required. Acquisitions need to be performed in the fluorescence mode, to be sensitive to a small amount of nanoparticles, using, for example, of conventional solid state detector, with a medium energy resolution (ca. 200–250 eV). Moreover, a high resolution is necessary to detect accurately the subtle variations upon interaction with a solution or a biological media. This second requirement is especially true in the pre-edge and edge part of the signal (X-ray Absorption Near Edge Structure, XANES). High-Energy Resolution Fluorescence Detection (HERFD) allows to further enhance XANES sensitivity. Even if not extensively used at present, this technique has gained a lot of interests and under development on many synchrotron beamlines. Using a detector with an
Physicochemical Properties of Nanoparticles in Relation with Toxicity Physicochemical Properties of Nanoparticles in Relation with Toxicity, Fig. 2 TFY XANES pre-edge and edge region compared to HERFD spectra for three CoPc samples. The pre-edge is magnified by 5 for clarity (From [13])
P
2079
Co
N C
H
CoPc HERDF
Intensity (arbitrary units)
CoPc inside MW carbon NT
TFY x5
Importance of the detection mode to enhance the XANES pre-edge features
CoPc
x5
CoPc/MWNT mixture CoPc@MWNT
7706
energy resolution better than the core hole lifetime broadening gives XANES spectra with sharper spectral features compared to conventional total fluorescence (TFY) spectra. Example of such sharpening is illustrated in Fig. 2. The structure of Cobalt phthalocyanine (CoPc), free or encapsulated inside multiwalled carbon nanotubes has been probed using either TFY or HERFD [13]. The differences in the pre-edge using conventional TFY spectra are really small, until those obtained by HERFD clearly put in evidence a modification of the Co conformation when CoP is free or encapsulated. Complete analysis shows that Co atoms move out of the molecular plane by a fraction of Angstrom.
7708
7710
7712 7714 Energy (eV)
7716
7718
7720
Physicochemical Properties: Biological Effects Relationships
[16]. But despite the growing body of literature, many conclusions remain contradictory. For instance, dose–response relationship are not always observed even for equivalent systems. Lovern et al. [17] clearly observed such relationship in the case of daphnia mortality exposed to 25 nm TiO2, while Hund-Rinke et al. [18] were unable to determine a LC50 value for 30 nm TiO2 in contact with Daphnia due to chaotic results. In many cases, the degree of physicochemical characterization of nanoparticles is not equivalent and one cannot identify why results are inconsistent. As previously discussed, nanoparticles can differ by many aspects due to variation in size, shape, redox properties, surface charge, crystal structure, chemical composition, adhesion to surfaces, etc., and beyond a list of results, it is worth addressing the effects of various physicochemical parameters of NPs on their toxic effects.
Some recent reviews inventoried NPs that are toxic to living organisms, e.g., [14–16]. There are now obvious examples of the toxic effects of C60 fullerenes to bacteria, to daphnia, etc., Nano-TiO2, nano-CeO2, nano-Ag, etc., can also be toxic to various organisms ([14, 15] etc.), or even colloidal silica to rat and mice
Effects of Aggregation State and Size Most of pioneer published results dealing with NPs toxicity mainly focused on size effects. Can toxic effects only be due to the small size of particles? Size was mainly measured by TEM that implied dry state characterization. However, the size was not characterized in
P
P
2080
Physicochemical Properties of Nanoparticles in Relation with Toxicity
nutritive or aquatic test media. Indeed the dispersion state of NPs (aggregation vs. dispersion) in contact with living organisms can change from one study to another. At this stage, it is worth detailing that here aggregation is referred to an assembly of particles which are loosely attached to each other. It means that aggregation is a reversible phenomenon and that NPs can aggregate in aquatic media but can disaggregate in presence of stabilizing agents like proteins. Then one has to keep in mind that aggregation can be reversible. There is a generally accepted idea assuming that aggregation of NPs impedes the determination of specific properties of NPs. It is obvious that the transport and mobility are strongly affected. Nevertheless, intrinsic or specific “nano” physicochemical properties of NPs are not necessarily (or even significantly) affected by aggregation processes. In the particular case of surface chemical reactivity, Auffan et al. [1, 19] have shown that surface normalized quantity of probe molecules sorbed to iron oxide particles is strongly enhanced for particles smaller than 12 nm. Using arsenite as surface reactivity molecular probe, the quantity of arsenite adsorbed to the maghemite (gFe2O3) surface per surface unit reached 3.6 As/nm2 for 300 nm and 20 nm particle size while it increased up to 11 As As/nm2 for 12 nm particles. This surface reactivity was not affected by aggregation [19]. The example of Nano-Fe2O3 in contact with E. coli can illustrate that the aggregation state as well as the size may not be the most important parameters in terms of toxicity [20]. Indeed, [20] have shown that 8 nm maghemite (Fe2O3) had no toxic effect toward E. coli regardless of the aggregation state of the NPs. Effects of the Redox State Evolution and Instability As function of their chemical composition redox states of NPs can change in biological redox/pH conditions. Indeed most elements exhibit various possible redox states that are or are not stable as function of Eh and pH values. It is obvious for all metals that they can exist under metallic or oxidized forms. But even when oxidized, some elements can exist under various redox states. Metals like Fe, Mn, Cr, or rare earth like Ce illustrate such variability (Fe2+, Fe3+; Mn2+, Mn3+, Mn4+, Mn7+; Cr3+, Cr6+; Ce3+, Ce4+, etc.). Living organisms and especially eukaryotic cells metabolism cannot support large pH variation. For most eukaryotic cells, pH is buffered in the 7–8 pH range. In the case of redox potential (noted Eh), oxidative phosphorylation in mitochondria is based on a series of redox reactions at near
circumneutral pH for which potentials are from 0.32 (NAD+/NADH) to 0.29 V (cytochromes). Besides oxidative phosphorylation (respiration), extracellular Eh is generally controlled by thiol/disulfide redox systems (mainly GSH/GSSH and Cys/CySS) for which Eh vary in the 0.140/0.08 V range. In such conditions, many elements can be redox unstable leading to electron exchange between NP surface and surrounding media. This could then be the starting point of disequilibrium of the redox balance and then to oxidative stress (See Box 1). In the specific case of iron, Auffan et al. [21] have shown that nano-zero valent iron as well as nano-magnetite (Fe3O4) can be toxic to E. coli. The study reported that the oxidation of iron in contact to bacteria can be at the origin of the toxicity. The potential of the redox couple characteristic of Fe0 nanoparticles at pH ¼ 7 is in the EhFe2+/Fe(0) ¼ 0.40/0.7 V range as function of the Fe concentration [22] which is lower than the typical Eh in biological media. This allowed the oxidation of Fe0(s) in contact with E. coli. This oxidation is directly responsible of toxic effects through the generation of an oxidative stress, as demonstrated by an enhanced toxicity of these Fe NPs toward a mutant strain of E. Coli deprived of defense mechanism against oxidative stress. Other examples in the literature confirm the importance of the redox “instability” of nanoparticles. Using XANES at Ce L3-edge, Thill et al. [12] and Auffan et al. [11] have shown that the cytotoxicity/genotoxicity of CeO2 was due to the reduction of surface Ce(IV) ions to Ce(III). Effects of NPs Shape and Aspect Ratio The toxicity of particles as function of their shape does not concern only nanoparticles. Indeed, many studies concerning asbestos toxicity have demonstrated that the toxicity can be partly related to the aspect ratio. Then comparison between asbestos and carbon nanotubes legitimated the worry concerning probable toxic effects of NPs. Poland and coworkers [23] have shown that “long multiwalled carbon nanotubes result in asbestos-like, length dependent, pathogenic behavior.” However, the mechanisms at the origin of the toxicity of long (>15–20 mm) nanotubes still remain unclear. In the case of nano-silver, it is possible to produce spherical, truncated triangular, rods nanoparticles. Pal et al. [24] have shown that that the biocidal effects depend on silver NPs shape with the higher effect for truncated triangular NPs. It has been emphasized that the {111} lattice plane exhibit the
Physicochemical Properties of Nanoparticles in Relation with Toxicity
2081
P
Box 1
Oxidative stress is a major problem that concerns all the cells living in the presence of dioxygen. It originates from the production of highly toxic reactive oxygen species (ROS), including superoxide radical O2•, hydrogen peroxide H2O2 and hydroxyl radical HO•. Although cells have developed efficient antioxidant systems to detoxify ROS, any deficiency of these systems or overload by overproduction of ROS can result in severe cellular damages such as proteins, lipids, and DNA oxidations, which can ultimately lead to cell death. It is now well documented that oxidative stress is involved in the pathogenesis of a variety of diseases such as cancers, asthma, pulmonary diseases, Alzheimer, Parkinson, and in ageing [1]. e−
O2 → Oxygen
O2•− Superoxide anion
e−
→ H2O2 2H + Hydrogen peroxide
e−
→
↓
OH •
Hydroxyl OH − radical
e−
→ H2O ↑
: formation and elimination of ROS
H+
Several studies have related the toxicity of some NPs to an oxidative stress toward biological models [2]. In order to better understand the cellular oxidative stress pathways induced by NPs, identification and quantification of the specific ROS produced directly or indirectly by NPs represent an important issue. In vivo detection and quantification of ROS are far from being trivial and require the use of specific techniques [3, 4]. This is mainly due to the fact that ROS are unstable species and are present at very low steady-state concentration in the cells. Various methods using synthetic probes have been developed to detect ROS in vivo. Some of these probes lead to a fluorescence signal upon oxidation by ROS, which is very well adapted for in cellulo detection by fluorescence microscopy. However, some of them lack of specificity and may lead to false interpretations for the presence of ROS [3, 4]. For example, the use of dichlorofluorescein (DCF) as a fluorescent peroxide probe is complex and crucially lacks of selectivity in vivo [5]. Hydroethidine (HE, MitoSOX™), a cell permeable compound which is specifically oxidized by O2• into the fluorescent 2-hydroxyethidium (2-OH-E+) compound, has been described as an interesting O2• probe [6]. However, in the cell, HE also generates other fluorescent compounds that are not related to the reaction with O2•. Consequently, direct quantification of 2-OH-E+ production in whole cells by fluorescence microscopy is not possible, and chromatography analysis of extracts obtained after cell lysis must be carried out in order to quantify 2-OH-E+ production [7]. In vitro, ROS detection is much less challenging since control experiments can be usually carried out in order to make corrections for nonspecific reactions. Probes like cytochrome c and tetrazolium salts XTT have been widely used to detect O2• in vitro [3, 4]. Spin trap nitrones, like DMPO or DEPMPO, associated with electron paramagnetic resonance spectroscopy (EPR) are well adapted to detect the unstable radical species O2• and HO• [3]. However, in vivo, their use is more difficult since the nitrone adducts resulting from the reaction of the spin trap with O2• or HO• are unstable and can be easily reduced by cellular reductants [3]. 1. Winterbourn, C.: Reconciling the chemistry and biology of reactive oxygen species. Nat. Chem. Biol. 5, 278–286 (2008) 2. Nel, A., Xia, T., Madler, L., Li, N.: Toxic Potential of Materials at the Nanolevel. Science 311, 622–627 (2006) 3. Bartosz, G.: Use of spectroscopic probes for detection of reactive oxygen species. Clin. Chim. Acta 368, 53–76 (2006) 4. Wardman, P.: Fluorescent and luminescent probes for measurement of oxidative and nitrosative species in cells and tissues: Progress, pitfalls, and prospects. Free Rad. Biol. Med. 43, 995–1022 (2007) 5. Zielonka, J., Kalyanaraman, B.: ROS-generating mitochondrial DNA mutations can regulate tumor cell metastasis”— a critical commentary. Free Rad. Biol. Med. 45, 1217–1219 (2008)
P
P
2082
Physicochemical Properties of Nanoparticles in Relation with Toxicity
6. Robinson, K.M., Janes, M.S., Pehar, M., Monette, J.S., Ross, M.F., Hagen, T.M., Murphy, M.P., Beckman, J.S.: Selective fluorescent imaging of superoxide in vivo using ethidium-based probes. Proc. Natl. Acad. Sci. 103, 15038–15043 (2006) 7. Zielonka, J., Vasquez-Vivar, J., Kalyanaraman, B.: Detection of 2-hydroxyethidium in cellular systems: a unique marker product of superoxide and hydroethidine. Nat. Protocol. 3, 8–21 (2008)
highest reactivity to sulfur- or phosphorus-containing soft bases, such as R-S-R, R-SH, RS, or PR3. Then sulfur-containing proteins in the membrane interacting with Ag NPs (with high affinity {111} lattice plane) can lead to membrane disruption or strong damage. Effects of NPs Solubility As discussed in the “Physicochemical properties of Nanoparticles” part, NPs solubility may substantially vary from thermodynamic prediction deduced from larger particles. The small size and consequently large surface area is of high concern to address possible NPs dissolution. It has been mentioned that in the case of toxic metals the dissolution will increase the concentration of free metal ions in contact with living organism. This is particularly true for silver nanoparticles for which the dissolution is associated with a change of the redox state (oxidative dissolution). It has been reported that upon contact with a bacterial cell, biocide Ag+ is released from the surface of NPs, which then interacts with bacterial cell membranes and inhibits respiratory enzymes (e.g., [25]). In the case of ZnO NPs, dissolution and release of Zn2+ play an important role in the biological toxicity (e.g., [26]). Dissolution is also discussed in the case such as Quantum Dots, even if recent developments prevent or strongly decrease QDots solubility via specific surface coating.
Effects of NP Adhesion to Cells NPs can be chemically unstable (dissolution, oxidation, reduction, etc.), inducing biological effects. However, the toxicity level may be enhanced or alleviated by additional parameters. For instance, it has been shown that nano-CeO2 can be reduced when both E. coli and Synechocystis are exposed [27]. But the reduction of Ce4+ to Ce3+ induced a much more intense effect on E. coli. Indeed, at the highest nano-CeO2 concentration, the percentage of survival cells was still 35% for Synechocystis while no cell was alive in
the case of E. coli. This difference may be explained by the production of extracellular polymeric substances (EPS) preventing direct contact between cell membranes and nano-CeO2 due to strong adsorption and adhesion of NPs. Thus, in the event of redox phenomena, the added distance between NPs and the membranes and the EPS shield itself may provide at least partial protection against the toxicity. Effects of Chemical Composition, Crystal Structure. . . Besides NPs properties discussed above, other parameters can induce biological effects. The chemical composition of NPs is of utmost importance. Many metals and metalloids are known to be toxic like for instance Cd or Se. Their use by nanotechnology to elaborate nanoproducts or nanomaterials must be done with care. Any dissolution of NPs incorporating Cd (for instance) will lead to toxicity, and it is worth developing any specific design to enhance their chemical stability and prevent at best their direct contact with external media. Most importantly, one has to define whether or not other less toxic element can be used to reach the same level of performance. From Nanoparticles to Nano-Residues of the Nanomaterials Life Cycle Nanoparticles and nanomaterials may be released from point sources like factories or landfills that may be more related to accidental events or technical defects during the processes. But most of the release may be due to indirect exposure through the life cycle of the nanomaterials. Indeed, for most applications, nanoparticles are surface modified and generally are embedded in the final product. Therefore, they do not come into direct contact with consumers or the environment. Their surface modification can strongly affect their behavior and especially their toxicity. NPs in commercial sunscreens can illustrate the difference between bare nanoparticles and surface modified nanoproducts. Indeed, the nanoparticles of TiO2 are
Physicochemical Properties of Nanoparticles in Relation with Toxicity
2083
P
Box 2: Exposure During Life Cycle of Nanoparticles and Nanomaterials
Most of the time, nanoparticles are surface modified prior to their incorporation in final products. Therefore, it is very likely that nanoparticles will not be directly released in the environment.
The degradation of products due to external conditions (sun, rain, etc.) may generate the release of by-products containing nanoparticles. For instance:
P
coated with AlOOH and polydimethylsiloxane for their use as sunscreens. The release of nano-residues from commercial NPs-based sunscreens is no longer a hypothesis, but has become a fact [28–30]. The major effect of the remaining Al-based coating is a protection
against the generation of superoxide ions from the photoactive/phototoxic TiO2 core. This surface modification of the nano-TiO2 results in a substantially lower toxic potential of the residue as long as the protective layer remains stable (See Box 2).
P
2084
Physicochemical Properties of Nanoparticles in Relation with Toxicity
Conclusion
nanoparticles from an environmental, health and safety perspective. Nat. Nanotechnol. 4, 634–641 (2009) Hutchings, G.J., Brust, M., Schmidbaur, H.: Gold - an introductory perspective. Chem. Soc. Rev. 37, 1759–1765 (2008) Turner, M., Golovko, V.B., Vaughan, O.P.H., Abdulkin, P., Berenguer-Murcia, A., Tikhov, M.S., Johnson, B.F.G., Lambert, R.M.: Selective oxidation with dioxygen by gold nanoparticle catalysts derived from 55-atom clusters. Nature 454, 981–U931 (2008) Hassello¨v, M., Readman, J.W., Ranville, J.F., Tiede, K.: Nanoparticle analysis and characterization methodologies in environmental risk assessment of engineered nanoparticles. Ecotoxicology 17, 344–361 (2008) Kaulich, B., Thibault, P., Gianoncelli, A., Kiskinova, M.: Transmission and emission x-ray microscopy: operation modes, contrast mechanisms and applications (Topical Review). J. Phys. Condens. Matter 23(8), 083002 (2011) Nelson, J., Huang, X., Steinbrener, J., Shapiro, D., Kirz, J., Marchesini, S., Neiman, A.M., Turner, J.J., Jacobsen, C.: High-resolution x-ray diffraction microscopy of specifically labeled yeast cells. Proc Natl Acad Sci 107, 7235–7239 (2010) Zhang, H., Chen, B., Banfield, J.F.: Particle size and pH effects on nanoparticle dissolution. J. Phys. Chem. C. 114, 14876–14884 (2010) Levard, C., Rose, J., Thill, A., Masion, A., Doelsch, E., Maillet, P., Spalla, O., Olivi, L., Cognigni, A., Ziarelli, F., Bottero, J.Y.: Formation and growth mechanisms of imogolite-like aluminogermanate nanotubes. Chem. Mater. 22, 2466–2473 (2010) Petkov, V., Cozzoli, D., Buonsanti, R., Cingolani, R., Ren, Y.: Size, shape, and internal atomic ordering of nanocrystals by atomic pair distribution functions: a comparative study of g-Fe2O3 nanosized spheres and tetrapods. J. Am. Chem. Soc. 131, 14264–14266 (2009) Modrow, H.: Tuning nanoparticle properties - the X-ray absorption spectroscopic point of view. Appl. Spectrosc. Rev. 39, 183–290 (2004) Auffan, M., Rose, J., Orsiere, T., De Meo, M., Thill, A., Zeyons, O., Proux, O., Masion, A., Chaurand, P., Spalla, O., Botta, A., Wiesner, M.R., Bottero, J.-Y.: CeO2 nanoparticles induce DNA damage towards human dermal fibroblasts in vitro. Nanotoxicology 3, 161–171 (2009) Thill, A., Zeyons, O., Spalla, O., Chauvat, F., Rose, J., Auffan, M., Flank, A.M.: Cytotoxicity of CeO2 nanoparticles for Escherichia coli. Physico-chemical insight of the cytotoxicity mechanism. Environ. Sci. Technol. 40, 6151–6156 (2006) Swarbrick, J.C., Weng, T.-C., Schulte, K., Khlobystov, A.N., Glatzel, P.: Electronic structure changes in cobalt phthalocyanine due to nanotube encapsulation probed using resonant inelastic X-ray scattering. Phys. Chem. Chem. Phys. 12, 9693–9699 (2010) Klaine, S.J., Alvarez, P.J.J., Batley, G.E., Fernandes, T.F., Handy, R.H., Lyon, D.Y., Mahendra, S., Mclaughlin, M.J., Lead, J.R.: Nanomaterials in the environment: behavior fate, bioavailability, and effects. Environ. Toxicol. Chem. 27, 1825–1851 (2008) Handy, R.H., von der Kammer, F., Lead, J.R., Hassello¨v, M., Owen, R., Crane, M.: The ecotoxicology and chemistry of manufactured nanoparticles. Ecotoxicology 17, 287–314 (2008)
This section discussed the relationships between the physicochemical properties of nanoparticles and their biological effects. As described above, nanoparticles differ from each other by many aspects and not only size and shape. Aggregation state, chemical stability, solubility, etc. can change from one to another NP leading to difficult comparison when toxicity occurs. Then, even if many issues remain unsolved, it clearly appears that the toxicity strongly depends on an accurate physicochemical characterization of nanoparticles and nanomaterials. However, while characterizing several properties like chemical composition, size, shape, crystal structure, surface area, surface chemistry, charge, and solubility of the initial NPs is now well accepted (but not always performed!), information is still lacking as, more than initial product characterization, it is worth determining the evolution of important parameters during the interaction between NPs and biological organisms. However, such accurate determination of nanoparticles in biological media remains highly challenging as discussed in this section. Acknowledgements The authors would like to thank the Centre National de la Recherche Scientifique (CNRS) and the Commissariat a` l’Energie atomique (CEA) for funding the International Consortium for the Environmental Implications of NanoTechnology (GDRi iCEINT).
2. 3.
4.
5.
6.
7.
8.
9.
10.
Cross-References ▶ Cellular Mechanisms of Nanoparticle’s Toxicity ▶ Effect of Surface Modification on Toxicity of Nanoparticles ▶ Electron Microscopy of Interactions Between Engineered Nanomaterials and Cells ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles ▶ In Vivo Toxicity of Titanium Dioxide and Gold Nanoparticles ▶ Nanoparticle Cytotoxicity ▶ Selected Synchrotron Radiation Techniques
References 1. Auffan, M., Rose, J., Bottero, J.Y., Lowry, G.V., Jolivet, J.P., Wiesner, M.R.: Towards a definition of inorganic
11.
12.
13.
14.
15.
Piezoelectric Effect at Nanoscale 16. Napierska, D., Thomassen, L.C.J., Lison, D., Martens, J.A., Hoet, P.H.: The nanosilica hazard: another variable entity. Part. Fibre Toxicol. 7(1), 39 (2010) 17. Lovern, S.B., Klaper, R.: Daphnia magna mortality when exposed to titanium dioxide and fullerene [C60] nanoparticles. Environ. Toxicol. Chem. 25, 1132–1137 (2006) 18. Hund-Rinke, K., Simon, M.: Ecotoxic effect of photocatalytic active nanoparticles (TiO2) on algae and daphnids. Environ. Sci. Pollut. Res. 13, 225–232 (2006) 19. Auffan, M., Rose, J., Proux, O., Borschneck, D., Masion, A., Chaurand, P., Hazemann, J.L., Chaneac, C., Jolivet, J.P., Wiesner, M.R., Van Geen, A., Bottero, J.Y.: Enhanced adsorption of arsenic onto maghemites nanoparticles: As (III) as a probe of the surface structure and heterogeneity. Langmuir 24, 3215–3222 (2008) 20. Auffan, M., Decome, L., Rose, J., Orsiere, T., De Meo, M., Briois, V., Chaneac, C., Olivi, L., Berge-Lefranc, J.L., Botta, A., Wiesner, M.R., Bottero, J.Y.: In vitro interactions between DMSA-coated maghemite nanoparticles and human fibroblasts: A physicochemical and cyto-genotoxical study. Environ. Sci. Technol. 40, 4367–4373 (2006) 21. Auffan, M., Achouak, W., Rose, J., Roncato, M.A., Chaneac, C., Waite, D.T., Masion, A., Woicik, J.C., Wiesner, M.R., Bottero, J.Y.: Relation between the redox state of iron-based nanoparticles and their cytotoxicity toward Escherichia coli. Environ. Sci. Technol. 42, 6730–6735 (2008) 22. Pourbaix, M.: Atlas of electrochemical equilibria in aqueous solutions, 2nd edn. National Association of Corrosion Engineers, Houston (1974) 23. Poland, C.A., Duffin, R., Kinloch, I., Maynard, A., Wallace, W.A.H., Seaton, A., Stone, V., Brown, S., MacNee, W., Donaldson, K.: Carbon nanotubes introduced into the abdominal cavity of mice show asbestos-like pathogenicity in a pilot study. Nat. Nanotechnol. 3, 423–428 (2008) 24. Pal, S., Tak, Y.K., Song, J.M.: Does the antibacterial activity of silver nanoparticles depend on the shape of the nanoparticle? A study of the gram-negative bacterium Escherichia coli. Appl. Environ. Microbiol. 73, 1712–1720 (2007) 25. Choi, O., Hu, Z.: Size dependent and reactive oxygen species related nanosilver toxicity to nitrifying bacteria. Environ. Sci. Technol. 42, 4583–4588 (2008) 26. Xia, T., Kovochich, M., Liong, M., Madler, L., Gilbert, B., Shi, H.B., Yeh, J.I., Zink, J.I., Nel, A.E.: Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties. ACS Nano 2, 2121–2134 (2008) 27. Zeyons, O., Thill, A., Chauvat, F., Menguy, N., CassierChauvat, C., Orear, C., Daraspe, J., Auffan, M., Rose, J., Spalla, O.: Direct and indirect CeO2 nanoparticles toxicity for E. coli and Synechocystis. Nanotoxicology 3, 284–295 (2009) 28. Auffan, M., Pedeutour, M., Rose, J., Masion, A., Ziarelli, F., Borschneck, D., Chaneac, C., Botta, C., Chaurand, P., Labille, J., Bottero, J.Y.: Structural degradation at the surface of a TiO2-based nanomaterial used in cosmetics. Environ. Sci. Technol. 44, 2689–2694 (2010) 29. Labille, J., Feng, J.H., Botta, C., Borschneck, D., Sammut, M., Cabie, M., Auffan, M., Rose, J., Bottero, J.Y.: Aging of TiO2 nanocomposites used in sunscreen. Dispersion and fate
2085
P
of the degradation products in aqueous environment. Environ. Pollut. 158, 3482–3489 (2010) 30. Botta, C., Labille, J., Auffan, M., Borschneck, D., Miche, H., Cabie´, M., Masion, A., Rose, J., Bottero, J.-Y.: TiO2-based nanoparticles released in water from commercialized sunscreens in a life-cycle perspective: structures and quantities. Environ. Pollut. 159(6), 1543–1550 (2011)
Phytotoxicity ▶ Toxicology: Plants and Nanoparticles
Piezoelectric Effect at Nanoscale Zhong Lin Wang and Ying Liu School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA, USA
Synonyms Nano-piezoelectricity
Definition Piezoelectricity is a phenomenon of strain induced electric polarization in certain crystals, which can be used to create a mechanical action by applying an external voltage for sensors and actuators, or an mechanical straining can produce a voltage for energy conversion.
Piezoelectricity Piezoelectricity was discovered by Jacques and Pierre Curie during their systematic studies of the effect of pressure on the generation of electrical charge by crystals in 1880. It is a reversible physical phenomenon, as there is the direct piezoelectric effect and the reverse piezoelectric effect. Direct piezoelectric effect is defined as: electric polarization produced by mechanical strain in crystals belonging to certain classes, the polarization being proportional to the strain and changing sign with it [1].
P
P
2086
Reverse (or inverse) piezoelectric effect is defined as: a piezoelectric crystal becomes strained, when electrically polarized by an amount proportional to the polarizing field.
Structures of Piezoelectric Crystals The nature of piezoelectricity comes from the lack of centrosymmetry/inversion symmetry in the crystal (Fig. 1). 21 out of all 32 crystal classes lack inversion symmetry, and 20 of them exhibit piezoelectric effect [2]. The two most important classes include perovskite structure, which occurs in ceramics such as Lead Zirconate Titanate (PZT) and Barium Titanate (BaTiO3); and wurtzite structure, which occurs in semiconducting materials such as Zinc Oxide (ZnO), Gallium Nitride (GaN), Indium Nitride (InN), and Zinc Sulfide (ZnS). Perovskite Structure The perovskite structure gets its name from the mineral CaTiO3 and is adopted by many oxides that have the chemical formula of ABO3 [3]. The most widely used piezoelectric ceramics, including Barium Titanate (BaTiO3) and Lead Zirconate Titanate (PZT) are formed in this structure. As shown in Fig. 2a, in the unit cell of perovskite compound, type “A” atom is in the cube corner positions (0, 0, 0), type “B“ atom is in the body center position (1/2, 1/2, 1/2) and oxygen atoms is in the face centered positions (1/2, 1/2, 0). However, in real cases, since ion sizes hardly meet the stringent requirement for stability, most perovskite structures are distorted and lose their cubic symmetry, and commonly shift into orthorhombic or tetragonal phases [4]. When “B” atoms are displaced in such distortion, as shown in Fig. 2b, the center for negative charge and positive charge in the crystal no longer coincide with each other, and results in an electric dipole. This is where the piezoelectricity comes from. Since the piezoelectricity originates from rather randomly oriented distortion, it is much more complicated and is correlated with the concept of ferroelectricity. In most cases, perovskite ceramics forms polycrystalline structures with multiple electric “domains.” Each domain has a size in the order of 10–100 nm, and even though each individual domain is piezoelectric itself, the whole ceramic does not show macroscopic dipole
Piezoelectric Effect at Nanoscale
without poling, as they have differently oriented electric dipoles. Electrical alignment of the dipoles, or electric poling as stated in most cases, are required to activate the piezoelectric effect in the ceramics. Similar to ferromagnetic materials, a characteristic temperature, the Curie temperature (TC) exists in perovskite ceramics, above which they lose their electric polarization induced in the poling process. There are multiple methods for fabricating perovskite ceramic nanostructures [5]. The top-down methods include electron-beam assisted fabrication, nanoprinting, and self-assembling, and they provide high-precision positioning and size control. However, top-down methods are limited in resolution, and prone to processing damage. In comparison, the bottom-up approaches allow much smaller feature sizes but suffer from poor registration and thus require substrate patterning that in turn involves top-down processing. For one dimensional nanostructure, solution-phase decomposition and hydrothermal chemical synthesis [6] can result in single crystalline nanorods and nanowires, sol-gel method associated with nanochannel templates or electrospinning can produce polycrystalline nanowires, and wetting of the porous templates can produce nanotubes. In all of the conventional piezoelectric materials, PZT has the best performance and the widest application. Its piezoelectric coefficient d33 can be as high as 400 pC/N, and can be higher than 600 pC/N when doped with lanthanum [2, 7]. However, since it is a lead containing chemical, it is not environmentally and biologically friendly without proper packaging, and its applications in the nanoscale region are thus restricted, as such applications are often related to biomedical or environmental issues. Wurtzite Structure (ZnO, GaN) The wurtzite crystal structure is named after the mineral wurtzite, and is the crystal structure for various binary compounds. Wurtzite structure occurs in most piezoelectric semiconductors, including Zinc Oxide (ZnO), Gallium Nitride (GaN), Cadmium Sulfide (CdS), Silicon Carbide (SiC) and so on, and among them, ZnO is most widely used in one dimensional nanostructure, and the key material for nanogenerators, piezotronics and piezophototronics, as will be discussed later. As shown in Fig. 3, in a Wurtzite structure, each of the two individual atom types forms a sublattice which
Piezoelectric Effect at Nanoscale
32 Crystal classes
21 Noncentrosymmetric
2087
10 Spontaneously polarized
Perovskite ceramics
10 Other structures
Wurtzite semiconductors
P
20 Piezoelectric
Piezoelectric Effect at Nanoscale, Fig. 1 Scheme of piezoelectric crystal classes
is hexagonal close pack (HCP) type, and they form tetrahedral coordination with each other [8]. Distinct from Perovskite structure, Wurtzite structure crystals do not need distortion or external poling to break the centrosymmetry. In single crystal Wurtzite materials, there are strong and uniform dipoles along the c axis, and the dipole direction is determined by the stacking order of the anions and cations, and more specifically, +c for from cation to anion and –c for from anion to cation. The orientation and polarity determines the direction of the piezoelectricity in the crystal. In ZnO, the interaction of the polar charges at the surface results in the growth of a wide range of unique nanostructures, such as nanobelts, nanosprings, nanorings, and nanohelices. The synthesis of ZnO nanostructures is easy and convenient. To fabricate large scale nanowire arrays, chemical hydrothermal at 80 C can be carried out on any substrate, while physical vapor liquid solid (VLD) or physical or chemical vapor solid (PVD or CVD) growth can give results with higher aspect ratio and higher density on substrate with proper lattice match. As will be discussed in latter sections, one dimensional wurtzite nanostructures, especially ZnO nanowires and nanobelts, promotes the new concept of “piezotronics,” which is the coupling between piezoelectricity and semiconductor transport property; and “piezo-phototronics,” which is the result of threeway coupling of piezoelectricity, photonic excitation, and semiconductor transport property. Piezoelectric Polymer Piezoelectricity also occurs in semicrystalline polymer materials. The most famous and widely used of them are polyvinylidene fluoride (PVDF) and its copolymers [9].
A atom O atom
b
B atom
a
+
–
Piezoelectric Effect at Nanoscale, Fig. 2 (a) Ideal perovskite structure; (b) perovskite structure with distortion and polarization along c-axis
In PVDF, hydrogen atoms are positively charged and fluoride atoms are negatively charged with respect to the carbon atoms, so that each monomer has an inherent dipole moment. As shown in Fig. 4, PVDF has several conformations: a (tg+tg), b (all trans), and g (tttg+tttg). In a conformation, the dipoles along the chain cancel each other, so that the overall piezoelectric effect is zero, and this conformation is the nonpiezoelectric conformation. In the b conformation, the alignment of all its dipoles is in the same direction normal to the chain axis. In this way, once strain is applied on the b phased polymer, and the chain is either stretched or compressed, the dipole strength will change and extra net charge will be induced, so that this conformation is piezoelectric. Since the nature of PVDF is ferroelectric, electric poling is also needed for the dipoles on the polymer chains to align so that the polymer will have piezoelectric response. The crystallinity of pure PVDF is usually 5060%. In its copolymer with TrFE and TFE, the crystallinity can be increased to nearly 100% after annealing, and
P
P
2088
Piezoelectric Effect at Nanoscale, Fig. 3 ZnO with wurtzite structure and illustration for the formation of dipoles ([8], Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission)
Piezoelectric Effect at Nanoscale
a b
O
Zn
c
F
+ −
P
F
a
b t
t
g+
t
C atom F atom t H atom
t g−
t t t
t t +
g
t t t
Piezoelectric Effect at Nanoscale, Fig. 4 Illustration of the two main conformation of PVDF. (a) a conformation; (b) b conformation
t
g− t t
Piezoelectric Effect at Nanoscale
the portion of b conformation is also larger, and the piezoelectric response is thus greater. The piezoelectric coefficient in PVDF can be as high as a negative value of 33 pm/V, and that of PVDF-TrFE can be as high as 38 pm/V. The synthesis of piezoelectric polymer nanofibers can be easily carried out through the process of electrospinning.
Theory of Piezoelectricity in One Dimensional Nanostructures In theoretical concern, piezoelectricity refers to the linear coupling of electricity and mechanics [10]. Considering electrical behavior, we have Maxwell’s equation: D ¼ eE, where D is electric displacement, e is permittivity, and E is electric field. Considering mechanical behavior, we have Hooke’s law: S ¼ sT, where S is strain, s is compliance, and T is stress. To describe piezoelectricity, the coupled equations are given as, f Sg ¼ s E f T g þ ½ d t f Eg fDg ¼ ½dfT g þ eT fEg Where [d] is the direct piezoelectric effect matrix and its transpose [dt] is the inverse piezoelectric effect matrix, and their specific form is related to the point group to which the crystal belongs. The most frequently used parameter is the piezoelectric coefficient d33, with expression d33 ¼ ð@D3 =@T3 ÞE ¼ ð@S3 =@E3 ÞT , so the unit of d33 is either C/N or m/V, and generally in materials d33 is often in the order of 1010 to 1012 C/N or m/V. When normal stress is applied along the polarization axis and on the surface charge is collected, d33 can be simplified as the volume change when subject to an electric field, or the electric polarization when subject to stress. A similar parameter d31 is applied when stress is applied at right angle with the polarization axis. Theoretical approaches have been applied on calculating piezoelectric potential distribution in piezoelectric nanostructures. A continuum model was built for calculating the electrostatic potential in a laterally bent nanowire (NW) [11]. To derive the relationship between the potential distribution in a NW and the dimensionality of the
2089
P
NW and magnitude of the force applied, four sets of equations for static piezoelectric material are first listed: ðbÞ 1. Mechanical equilibrium equation, = s ¼ ~ fe ¼ 0, where s is the stress tensor. ( sp ¼ cpq eq ekp Ek 2. Constitutive equations, , Di ¼ eiq eq þ kik Ek where cpq is the linear elastic constant, ekp is the linear piezoelectric coefficient, and kik is the dielectric constant. 3. Geometrical compatibility equation, 2 eilm ejpq @ emp @xl @xq ¼ 0, where eilm and ejpq are Levi–Civita antisymmetric tensors. Small deformation of the NW is also assumed for simplicity of the derivation. 4. Gaussian equation of electric field, ðbÞ ~ ¼ re ¼ 0, under the assumption that the =D piezoelectric NW has no conductivity. For analytical analysis, a perturbation theory has been developed based on these equations. Under different orders of approximation, the four equations correspond to different extends of decoupling and coupling between the electric field and mechanical deformation: the zeroth order solution stands for purely mechanical deformation without piezoelectricity; the first order is the result of direct piezoelectric effect, where stress/strain generates an electric field in the NW; and the second order shows the first coupling of the piezoelectric field to the strain in the material. For the calculation of piezoelectric potential in an insulating NW, the first order approximation is sufficient, and it gives analytical results less than 5% away from finite element method (FEM) numerical calculation for potential distribution in a typical NW with diameter d ¼ 50 nm and length l ¼ 600 nm at a lateral bending force of f ¼ 80 nN. For further investigation on the interaction of piezoelectric potential with semiconductor transportation properties, charge carrier concentration can be substituted into the Gaussian equation of electric field, and the equation is modified as: þ ~ ¼ ð@=@xi Þ eiq eq þ kik Ek ¼ rðbÞ =D e ¼ ep en þ eND eNA , where p is the hole concentration in the valance band, n is the electron concentration in the conduction band, NDþ is the ionized donor concentration, and NA is the ionized acceptor concentration. For n-type nanowires, it can be assumed that p ¼ NA ¼ 0, and vice versa for p-type. Considering band structure and charge carrier
P
P
2090
Piezoelectric Effect at Nanoscale
a
j (V ) 0.4
b
0.3 0.2 0.1 0 −0.1 −0.2 y
z
x
y
distribution equations, there can be numerical solution of the modified equations. As shown in Fig. 5, FEM solutions based on these equations is given for a typical ZnO NW with moderate doping [Ref. 36 of [12]. Further investigation is also carried out considering the contact of semiconducting NW with metals [13]. When a ZnO NW is under nonuniform deformation, the local piezoelectric potential distribution at different positions shows significant effect on and distinct trend of variation in the charge carrier transport characteristic. This can help understanding the characteristics of piezotronic devices based on NW of wurtzite materials by controlling local contact position and contact size.
Effect of Size on Piezoelectricity Size Effect in Wurtzite Nanocrystals At nanoscale, many interesting effects occur and this can affect physical properties of materials, including piezoelectric properties. In nanowires (NW) and nanobelts (NB), crystal defects such as dislocation and vacancies will concentrate and eliminate on the surface, and perfect single crystallinity is more easily achieved at this scale of dimension. As a result, during piezoelectric measurement of individual Zinc Oxide NB with dimension of tens of nanometers in thickness, hundreds of nanometers in width and tens of micrometers in length, the effective piezoelectric coefficient d33
−0.4
permission from Ref. 36 of [12] “Gao, Y.F., Wang, Z.L.: Equilibrium potential of free charge carriers in a bent piezoelectric semiconductive nanowire. Nano Lett. 9, 1103–1110 (2009).” Copyright 2009 American Chemical Society)
30
Piezoelectric coefficient (pm/V)
Piezoelectric Effect at Nanoscale, Fig. 5 Plot of calculated piezoelectric potential ’. Regarding a ZnO NW with diameter d ¼ 50 nm and length l ¼ 600 nm at a lateral bending force of f ¼ 80 nN for the NW are the same as in Fig. 4 except that it is ntype semiconducting with ND ¼ 1 1017 cm3 (Adapted with
−0.3
ZnO nanobelt
25
ZnO (0001) bulk x-cut quartz
20
15
10
5
0 10
100 Frequency f (KHz)
1000
Piezoelectric Effect at Nanoscale, Fig. 6 Frequency dependence of piezoelectric coefficient of ZnO nanobelts, ZnO bulk material, and x-cut quartz (Adapted with permission from Ref. 41 of [12] “Zhao, M.H., Wang, Z.L., Mao, S. X.: Piezoelectric characterization of individual zinc oxide nanobelt probed by piezoresponse force microscope. Nano Lett. 4, 587–590 (2004).” Copyright 2004 American Chemical Society)
is revealed to be frequency dependent and varies from 14.3 to 26.7 pm/V, which is about one time larger than that in bulk ZnO, around 10 pm/V, as shown in Fig. 6. Size Effect in Perovskite Nanostructures Since in perovskite materials, piezoelectric performance is tightly related to the concept of domain, the
Piezoelectric Effect at Nanoscale
a
P
b 100 nm
Remanence
200 nm
Unpatterned
−20
−10 0 10 DC bias to tip (V)
case of size effect is much more complicated, there are multiple factors to take into consideration. Intrinsically, at nanoscale, competition exists between thermal vibrations and the correlation energy which aligns the dipoles, but greater surface area also helps eliminate defects. Extrinsically, grain boundaries and other microstructural factor can also have various influences. Between 10 and 1,000 nm, which is bigger than or equal to the scale of a typical single domain and similar to the state of art piezoelectric nanodevices, controversial size effects can occur in differently processed materials regarding the sequence of patterning and crystallization of the nanostructure. This controversy is caused by extrinsic factors related with post annealing. In structures that were patterned with high-energy ion etching after crystallization, which leads to higher concentration of oxygen and lead vacancies, several reports reveal the inherent reduction in the piezoelectric coefficient in ultrathin PZT films. With post annealing, nonswitching of pinned domains help stabilize and enhance piezoelectric effect. In thin film studies, the observed increase of piezoelectric response amplitude can be as high as 300% when decreasing the feature size of patterns on epitaxial PZT thin film from 200 to 100 nm, as shown in Fig. 7. The crystalline nature of patterns might also have contributed to this difference: single crystalline tends to have better piezo performance
20
Saturation
Amplitude of piezoresponse (a.u.)
140 nm
Amplitude of piezoresponse (a.u.)
Piezoelectric Effect at Nanoscale, Fig. 7 Ferroelectric loop showing increased piezoelectric response with shrinking size. (a) Loops on patterns with different lateral sizes and on the unpatterned film; and (b) remnant polarization and saturation as a function of lateral size [14]
2091
0
0.2
0.4 0.6 0.8 Lateral size (μm)
1.0
1.2
with smaller scale than polycrystalline. In template method, synthesized PZT NW enhanced piezoelectric response than the value in thin film was also discovered [5]. Under 10 nm, intrinsic factors have more influence on the piezoresponse. People have used phenomenological models and first principal/ab initial calculations to predict the behavior of such nanostructures, and their prediction is based on the competition and cancelation of thermal drift. In general, the results are shown as: (1) depolarizing fields exist near the surfaces to reduce polarization and (2) piezoelectric state becomes unstable below certain sizes, which is estimated to be 7–10 nm along each of the three directions of well-known materials such as BaTiO3. Microscopy study in this scale has been done and verified these predictions to some extent. This gives us important information that fabricating piezoelectric devices under the scale of 10 nm using perovskite structure materials will hardly result in better performance.
Applications of Nanoscale Piezoelectric Properties As mentioned in previous sections, for materials with perovskite structure, the enhancement of performance at the nanoscale is not always significant, and related
P
P
2092
application is very limited. For materials with wurtzite structures however, especially for ZnO NW and NB, the interaction of piezoelectricity with semiconducting transportation process brings about new phenomena and draws great research interests, into the new concept of “piezotronics” and even “piezo-phototronics.” These applications includes piezoelectric field-effect transistor, piezoelectric diode, piezoelectricity based nanosensors, piezoelectricity enhanced photodetector, and most importantly, the piezoelectric nanogenerators. Piezoelectric Nanogenerator The future of nanotechnology research is likely to focus on the areas of integrating individual nanodevices into a nanosystem that acts like living specie with sensing, communicating, controlling, and responding. A nanosystem requires a nanopower source to make the entire package extremely small yet with high performance. The goal is to make selfpowered nanosystem that can operate wirelessly, independently, and sustainably. Harvesting energy from the environment is a choice for powering nanosystems especially for biomedical applications. Since mechanical energy is a conventional form of energy that exists in our living environment, it is essential to explore innovative nanotechnologies for converting mechanical energy (such as body movement and muscle stretching), vibration energy (such as acoustic/ultrasonic wave), and hydraulic energy (such as body fluid and blood flow) into electric energy that will be used to power nanodevices without using battery. This is a key step toward self-powered nanosystems [15]. Dr. Z.L. Wang’s group in Georgia Institute of Technology is the first to invent an innovative approach for converting mechanical energy into electric energy by piezoelectric zinc oxide nanowire arrays [16]. The operation mechanism of the nanogenerator (NG) relies on the piezoelectric potential created by an external strain; a dynamic straining of the nanowire results in a transient flow of the electrons in the external load due to the driving force of the piezopotential. The nanogenerator technology has been developed from fundamental science, to engineering integration, and to technological scale-up. As today, a gentle straining can output 1–3 V at an instant output power of 2 mW from an integrated nanogenerator, using which a selfpowered nanosensor has been demonstrated [6]. This
Piezoelectric Effect at Nanoscale
technology has the potential applications for powering MEMS/NEMS that requires a power of the microwatt to milliwatt range. The fundamental of nanogenerator relies on the presence of the piezoelectric potential (piezopotential) generated in a nanowire/nanobelt via dynamic straining. As demonstrated in previous sections, ZnO has the wurtzite structure, in which the tetrahedral coordinated O2 and Zn2+ are stacked layer by layer along the c-axis. The lack of central symmetry results in the piezoelectric property of ZnO, which is vital for the mechanic-electric energy conversion with the ZnO-based nanogenerator. The cations and anions are tetrahedral-coordinated in the wurtzite-structured ZnO. At the strain free status, the charge-center of the cations and that of anions coincide with each other. When an external strain is applied, the structure is deformed so that the chargecenters for cations and anions separate and result in an electric dipole. Because the ionic charges are not free to move and the intrinsic free charge carriers can only partially screen them if the doping level is low, the piezoelectric field is preserved as long as the NW is strained. The potential created by the polar ions is called piezoelectric potential, or piezopotential. The presence of piezopotential is the fundamental physical basis of the nanogenerators and piezotronics (Fig. 8). When a strained crystal is connected to an external load, the electrons in the circuit are driven to flow in to partially screen the piezopotential, which is the energy conversion process. Therefore, the principle of the nanogenerator is the transient flow of electrons in external load as driven by the piezopotential created by dynamic straining. Piezopotential-driven transient flow of electrons in an external load is the principle of the nanogenerator. Figure 9 presents a single wire generator (SWG) using a laterally packaged ZnO wire. In brief, a single ZnO wire was placed laterally on the flexible polyimide film. Silver paste can be used to fix two ends of the ZnO wire to the substrate and link the wire to the external measuring instrument through metallic wires. Electric measurement shows that a functioning SWG usually has rectifying I–V characteristic, indicating the existence of the Schottky contact at least at one end. Before the SWG is deformed, there is no measurable potential drop from the ZnO wire. When the substrate is bent, both the substrate and the ZnO wire are under strain. Because the thickness of the substrate is much larger than the dimension of the ZnO wire, the ZnO wire is solely
Piezoelectric Effect at Nanoscale
a
2093 c
P
0.2 0.15
f
c
f
Piezcelectric Potential (V)
b
0.1 0.05 0 −0.05 −0.1 −0.15 −0.2
Piezoelectric Effect at Nanoscale, Fig. 8 Numerical simulation of the piezopotential distribution in an insulating ZnO nanowire (a); The piezopotential distribution and the deformed shape of a ZnO nanowire grown along c-axis under a stretching force of 85 nN (b); or a compressing force of 85 nN (c) (Adapted
with permission from Ref. 13 of [12] “Gao, Z.Y., Zhou, J., Gu, Y.D., Fei, P., Hao, Y., Bao, G., Wang, Z.L.: Effects of piezoelectric potential on the transport characteristics of metal-ZnO nanowire-metal field effect transistor. J. Appl. Phys. 105, 113707 (2009).” Copyright 2009, American Institute of Physics)
Piezoelectric Effect at Nanoscale, Fig. 9 Design and power output from a single wire generator. (a) and (b) schematic diagram of single wire generator before and after stretching. Open-circuit voltage (c) and short-circuit current (d) of the single wire generator, which was cyclically fast stretched and fast released (Adapted from Ref. 16 of [12] “Yang, R. S., Qin, Y., Dai, L.M., Wang, Z.L.: Power generation with laterally packaged piezoelectric fine wires. Nat. Nanotechnol. 4, 34–39 (2009)”)
experiencing tensile strain when the bend is concave downward, as shown in Fig. 9b. As discussed earlier, tensile strain will result in piezopotential, such that the +c-axis side gains positive potential (light yellow) and c-axis side gains negative potential (dark red). The potential difference between the two ends can then be measured either as open-circuit voltage (Fig. 9c) or short-circuit current (Fig. 9d). The working mechanism of the SWG can be understood with the energy band diagram in Fig. 10. The
P
Schottky contact at least at one end is necessary for the energy generation. It is assumed the Schottky contact is at the +c-axis side (left side in Fig. 10a), and the energy generation follows in a similar way when the Schottky contact is on the other side or both sides. The entire wire is in equilibrium state without any strain and power output (Fig. 10a, e). When the wire is stretched, the tensile strain induces polarization of atoms, as well as piezoelectric field, in the crystal. Consequently, the +c-axis side holds positive potential
P
2094
Piezoelectric Effect at Nanoscale
Piezoelectric Effect at Nanoscale, Fig. 10 Current generation mechanism explained with energy band diagram of the cyclically stretched NW. Energy band diagram of the ZnO NW in equilibrium and free condition (a); non-equilibrium and tensile strained condition (b); re-reached equilibrium and tensile strained condition (c); and nonequilibrium and free condition (d). After (d), the NW will reach equilibrium as in (a) after charges flow. EF, CB, VB, FSB, and DEp indicate Fermi level
of the electrode, conduction band, and valence band of ZnO, Schottky barrier height, and piezopotential difference respectively. (e) Short-circuit current measurement in which the labels of a, b, c, and d indicate the corresponding process shown in subparts a–d (Adapted from Ref. 16 of [12] “Yang, R.S., Qin, Y., Dai, L.M., Wang, Z.L.: Power generation with laterally packaged piezoelectric fine wires. Nat. Nanotechnol. 4, 34–39 (2009)”)
and the c-axis side holds negative potential. The potential difference is DEp . The conduction band edge, as well as the Fermi level of the metal of the right side electrode, rises up for the same amount of DEp . The electrons in the external circuit should flow from the right-hand side to the left-hand side to compensate the energy difference, which will generate the first output signal if the flow rate is sufficiently large, as indicated in Fig.10b and e. The electrons cannot flow across the interface due to the presence of a Schottky barrier fSB . The accumulation of the electrons at the interface will partially screen the piezopotential built in the wire. When the electrons and the piezoelectric field reach equilibrium, the Fermi energy on both sides is on the same level and there is no more current flow (Fig. 10c, e). When the SWG is fast released, there is no strain inside the NW. As a result, the polarization and piezoelectric field vanish as well and the equilibrium with the accumulated electrons is broken. The accumulated charge carriers flow back from the lefthand side to the right-hand side through the external circuit, producing the second output signal in the opposite direction, as shown in Fig. 10d, e. Integration of SWGs is a major step toward the practical applications. Taking advantage of the small size of the NW, multiple SWGs can be built on a single substrate such that the hamster can drive all SWGs simultaneously. Integration of up to four
SWGs in serial has been demonstrated with the open-circuit voltage of 0.1–0.15 V. The output current can also be improved with multiple SWGs in parallel. The work on the human finger and living hamster clearly demonstrated the capability of the SWG to scavenge mechanical energy from the biosystem. It also confirms the feasibility of using the SWG for harvesting the energy created by regular and irregular motions. Piezotronics A most simple FET is a two ends bonded semiconductor wire, in which the two electric contacts at the ends are the source and drain, and the gate voltage can be applied either at the top of the wire through a gate electrode or at its bottom on the substrate [8, 13]. When a ZnO NW is strained axially along its length, the piezoelectric potential continuously drops from one side of the NW to the other, which means that the electron energy continuously increases from the one side to the other. Meanwhile, the Fermi level will be flat all over the NW when equilibrium is achieved, since there is no external electrical field. As a result, the effective barrier height and/or width of the electron energy barrier between ZnO and metal electrode will be raised at one side and lowered at the other side; thus, it has a nonsymmetric effect on the source and drain. This is the piezotronic effect [11].
Piezoelectric Effect at Nanoscale
a VDS
IO = O
VGS
ID CMOS FET
Piezoelectric Effect at Nanoscale, Fig. 11 A comparison of a CMOS FET and a piezotronic FET. The key difference is that the externally applied gate voltage VGS is replaced by the stain generated piezopotential, which tunes the charge transport insets are the enlarged view of the boxed area for one cycle of deformation. This is the core of piezotronics
P
2095
W GATE SOURCE
DRAEN
L
b VDS
Piezopotential (tensile strain)
Metal
ZnO
VDS
Piezopotential (compressive strain)
Once a strain is created in the semiconductor that also has piezoelectric property, a negative piezopotential at the semiconductor side effectively increases the local SB height, while a positive piezopotential reduces the barrier height. The polarity of the piezopotential is dictated by the direction of the c-axis for ZnO. The role played by the piezopotential is to effectively change the local contact characteristics through an internal field, thus, the charge carrier transport process is tuned/gated at the metal-semiconductor (M-S) contact. Figure 11 compares the piezopotential gated FET and the traditional CMOS FET, which are different only by replacing the gate voltage by the inner crystal piezopotential. With considering the change in piezopotential polarity by switching the strain from tensile to compressive, the local contact characteristics can be tuned and controlled by the magnitude of the strain and the sign of strain. This is the core of piezotronics. The fundamental working principle of the p-n junction and the Schottky contact is that there is an effective
barrier that prevents the charge carriers of both sides from crossing. The height and width of the barrier are the characteristic of the device. In piezotronics, the role played by the piezopotential is to effectively change the width of p-n junction or height of Schottky barrier (SB) by piezoelectricity, which has applications in maximizing the performance of the solar cell, photon detector and LED by tuning the height of the local SB for achieving the optimum charge separation. A simple piezotronic device is a polarity switchable diode that is made of a ZnO NW contacted with metal contacts at the two ends on an insulating polymer substrate. From the initial I-V curve measured from the device before applying a strain as shown in Fig. 12, the symmetric shape of the curve indicates that the SBs present at the two contacts are about equal heights. The equivalent circuit model of the device is a pair of backto-back Schottky diodes, as illustrated in the inset in Fig. 12. Under tensile strain, the piezoelectric potential at the right-hand side of this NW was lower (denoted
P
P
2096
Piezoelectric Effect at Nanoscale
Original Stretched Compressed
2
Current (μA)
1
0 Original −1
Stretched
−2 −3
Compressed
−2
−1
0
1
2
Voltage (V)
Piezoelectric Effect at Nanoscale, Fig. 12 Piezotronic strain sensor/switch. Changes of transport characteristics of an Ag/ ZnO-nanowire/Ag device from symmetric I–V characteristic (black) to asymmetric rectifying behavior when stretching (red) and compressing (green) the wire. The inset is the equivalent circuit models of the device in corresponding to the observed I–V curves, different sizes of diode symbol are used to illustrate the asymmetric Schottky contacts at the two ends of the nanowire. The blue side is the negative potential side, and the other side is positive side (Ref. 22 of [12], Adapted with permission from “Zhou, J., Gu, Y.D., Fei, P., Mai, W.J., Gao, Y.F., Yang, R.S., Bao, G., Wang, Z.L.: Flexible piezotronic strain sensor. Nano Lett. 8, 3035–3040 (2008).” Copyright 2008 American Chemical Society)
by blue color in the inset in Fig. 12), which raised the local barrier height (denoted by a large diode symbol in the inset). Since the positive piezoelectric potential was partially screened by free electrons, the SB height at the left-hand side remained almost unchanged. As a result, under positive bias voltage with the left-hand side positive, the current transport was determined by the reverse biased SB at the right-hand side. While under the reverse biased voltage with the right-hand side positive, the current transport depended on the reverse biased SB at the left-hand side, which had a much lower barrier height than the right-hand one. Experimentally, the device thus exhibited a rectifying behavior in the positive voltage region, and the I–V curve in the negative voltage region overlapped with that of the original curve without straining. By the same token, under compressive strain the device exhibited a rectifying behavior in the negative voltage region, and the I–V curve in the positive voltage region overlapped with that of the original curve without straining, as shown by the green line in Fig. 12.
Piezotronic Logic Operations The existing semiconductor NW logic devices are based on electrically gated field effect transistors (FETs), which function as both the drivers and the active loads of the logic units by adjusting the conducting channel width [17]. Moreover, the currently existing logic units are “static” and are almost completely triggered or agitated by electric signals, while the “dynamic” movable mechanical actuation is carried out by another unit possibly made of different materials. It is highly desirable to integrate the static logic operations with the dynamic mechanical action into one unit utilizing a single material. The first piezoelectric trigged mechanical-electronic logic operation has been demonstrated using the piezotronic effect, through which the integrated mechanical actuation and electronic logic computation are achieved using only ZnO NWs. A strain gated transistor (SGT) is made of a single ZnO NW with its two ends, which are the source and drain electrodes, being fixed by silver paste on a polymer substrate (Fig. 13a). Once the substrate is bent, a tensile/compressive strain is created in the NW since the mechanical behavior of the entire structure is determined by the substrate. Utilizing the piezopotential created inside the NW, the gate input for a NW SGT is an externally applied strain rather than an electrical signal. IDS–VDS characteristic for each single ZnO-NW SGT is obtained as a function of the strain created in the SGT (Fig. 13a) before further assembly into logic devices. A NW SGT is defined as forward biased, if the applied bias is connected to the drain electrode (Fig. 13a). For a SGT, the external mechanical perturbation induced strain (eg) acts as the gate input for controlling the “on”/“off” state of the NW SGT. The positive/negative strain is created when the NW is stretched/compressed. The SGT behaves in a similar way to an n-channel enhancement-mode MOSFET, apparently indicating the working principle of the SGT. The working principle of a SGT is illustrated by the band structure of the device. A strain free ZnO NW may have Schottky contacts at the two ends with the source and drain electrodes but with different barrier heights of FS and FD, respectively (Fig. 13b-a). When the drain is forward biased, the quasi-Fermi levels at the source (EF,S) and drain (EF,D) are different by the value of eVbias, where Vbias is the applied bias (Fig. 13b-b). An externally applied mechanical strain (eg) results in both band structure change and piezoelectric potential field in a ZnO NW. The former leads
Piezoelectric Effect at Nanoscale
2097
P
eT
a 6
SGT under tensile strain
5
IDS (μA)
4 SGT under compressive strain
3 2 1 0 −1.5
−1.0
−0.5
0.0 ea (%)
0.5
1.0
b b
a
c Φd
EC, ZnO
EF, D
EF, ZnO
Φs
Vbias
eVbias
Φd
Φs EF, S
EF, S
EF, ZnO
EF, D
EV, ZnO c
Tensile strained
d
Compressive strained
Φ″s≅ Φs+ΔE′p
Φ′s≅Φs−ΔEp EF, S
EF, S Φ′d
eVbias EF, ZnO
EF, D
−
Piezopotential
+
Φ″d EF, D
eVbias EF, ZnO Piezopotential +
−
Piezoelectric Effect at Nanoscale, Fig. 13 Strain-gated transistor (SGT). (a) Current – strain (IDS – eg) transfer characteristic for a ZnO SGT device with strain sweeping from eg ¼ 0.53– 1.31% at a step of 0.2%, where the VDS bias values were 1, 0.75, and 0.5 V, respectively; (b) The band structures of the ZnO NW SGT under different conditions for illustrating the mechanism of SGT. The crystallographic c-axis of the nanowire directs from drain to source. (a) The band structure of a strain-free ZnO NW SGT at equilibrium with different barrier heights of FS and FD at the source and drain electrodes, respectively;
(b) The quasi-Fermi levels at the source (EF,S) and drain (EF,D) of the ZnO SGT are split by the applied bias voltage Vbias; (c) With tensile strain applied, the SBH at the source side is reduced from FS to F0S ffi FS DEP . (d) With compressive strain applied, the SBH at the source side is raised from FS to F00S ffi FS þ DE0P (Adapted with permission from Ref. 29 of [12] “Wu, W.Z., Wei, Y.G., Wang, Z.L.: Strain-gated piezotronic logic nanodevices. Adv. Mater. 22, 4711–4715 (2010).” Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission)
to the piezoresistance effect, which is a non-polar and symmetric effect at the both source and drain contacts. Since ZnO is a polar structure along c-axis, straining in axial direction (c-axis) creates a polarization of cations and anions in the NW growth direction, resulting in a piezopotential drop from V+ to V along the NW,
which produces an asymmetric effect on the changes in the SB heights (SBHs) at the drain and source electrodes. Under tensile strain, the SBH at the source side reduces from FS to F0S ffi FS DEP (Fig. 13b-c), where DEP denotes the effect from the locally created piezopotential and it is a function of the strain,
P
P
2098
Piezoelectric Effect at Nanoscale
a
resulting in increased IDS. For the compressively strained SGT, the sign of the piezopotential is reversed, thus the SBH at the source side is raised from FS to F00S ffi FS þ DE0P (Fig. 13B-d), where DE0P denotes the piezopotential effect on the SBH at source side, resulting in a large decrease in IDS. Therefore, as the strain eg is swept from compressive to tensile regions, the IDS current can be effectively turned from “on” to “off” while VDS remains constant. This is the fundamental principle of the SGT. Logic operations of NW strain-gated NOR gate have been realized by integrating SGTs, which are gated individually by the applied strains, according to corresponding layout rules (Figs. 14a–b). The output voltages of NOR gate versus the input gate strains are shown in Fig. 14c. It can also be seen that NW straingated NOR gate with active loads exhibit better overall performance, such as larger logic swing, compared to passive-load NOR gate. Using the SGTs as building blocks, other universal logic components such as inverters and NAND and XOR gates have been demonstrated for performing piezotronic logic calculations, which have the potential to be integrated with the current NEMS technology for achieving advanced and complex functional actions in nanorobotics, microfluidics, and micro/nanosystems.
Vout
SGI C
Vbias SGI D
Vbias
b
εg, D
εg, C 6
7
5
8
Vout
c Logic “1” output 1.0
Vout (V)
0.8
0.6
Strain on SGI C
“00” (0.965 V)
“0” (−1%)
“1” (1%) “0” (−1%) “1” (1%)
0.4
“1” (1%)
eg “1” (1%)
“0” (−1%)
Strain on SGI D “0” (−1%)
0.2 “01” (1E-4 V)
“11” (1E-4 V)
“10” (1E-4 V)
0.0 Time (a.u.)
Logic “0” output
NOR gate
Piezoelectric Effect at Nanoscale, Fig. 14 Strain gated piezotronic logic NOR gate. (a) Schematic of the ZnO NW strain-gated NOR logic gate, which is composed of two SGIs, SGI C and SGI D. The strain input C for SGI C is defined in reference to the strain applied to SGT 5 and the strain input D for SGI D is defined in reference to the strain applied to SGT 8; (b) Layout for ZnO NW strain-gated NOR logic gate connecting two ZnO NW SGIs; (c) Logic operations and experimental truth table of the ZnO NW strain-gated NOR logic gate. Red line is the electrical output of the NOR gate. Blue and green lines represent the strain inputs applied on SGI C and SGI D, respectively (Adapted with permission from Ref. 29 of [12] “Wu, W.Z., Wei, Y.G., Wang, Z.L.: Strain-gated piezotronic logic nanodevices. Adv. Mater. 22, 4711–4715 (2010).” Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission)
Summary With reduction in materials size, its mechanical robustness, elastic modulus, and toughness improve dramatically. It can easily go way beyond the nonlinear but still elastic mechanical deformation without causing plastic damage. The piezoelectric effect at nanoscale is expected to be largely enhanced, so that they can be effectively used for ultrasensitive sensors/actuators and energy converters. The two most recent important advances in using the nano-scale piezoelectric effect have been introduced here. One is the nanogenerator that converts mechanical energy into electricity utilizing tiny physical motion. Besides its potential for large-scale applications such as harvesting ocean wave and wind energy, it finds application in driving wireless nanodevices and nanosystems for medical science, environmental/infrastructure monitoring, defense technology, and even personal electronics. It is a key step for building self-powered nanotechnology.
Piezoelectric MEMS Switches
The second important concept in science is the piezotronics. Electronics fabricated by using innercrystal piezopotential as a “gate” voltage to tune/control the charge transport behavior is named piezotronics. Piezo-phototronic effect is a result of three-way coupling among piezoelectricity, photonic excitation and semiconductor transport, which allows tuning and controlling of electro-optical processes by strain induced piezopotential. These designs are different from the CMOS field effect transistor by replacing the externally applied gate voltage using a crystal generated inner potential (piezopotential) for controlling the charge transport. It is a transistor controlled by a force/strain. It will find applications in human-computer interfacing, mechanical triggered electronics, nanomachine, electronic signature, medical diagnostics, and sensor.
Cross-References ▶ Ab Initio DFT Simulations of Nanostructures ▶ AFM Probes ▶ Biosensors ▶ Carbon Nanotube-Metal Contact ▶ CMOS MEMS biosensors ▶ Electrospinning ▶ Finite Element Methods for Computational Nano-Optics ▶ Flexible Electronics ▶ Hybrid Solar Cells ▶ Nanorobotics ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanostructures for Energy ▶ Nanotechnology ▶ NEMS Piezoelectric Switches ▶ Piezoelectric MEMS Switch ▶ Sol-gel Method
References 1. Cady, W.G.: Piezoelectricity: An Introduction to the Theory and Applications of Electromechanical Phenomena in Crystals. Dover, New York (1964) 2. Haertling, G.H.: Ferroelectric ceramics: history and technology. J. Am. Ceram. Soc. 82(4), 797–818 (1999) 3. Bhalla, A.S., Guo, R., Roy, R.: The perovskite structure – a review of its role in ceramic science and technology. Mat. Res. Innovat. 4, 3–26 (2000) 4. Johnsson, M., Lemmens, P.: Crystallography and chemistry of perovskites. In: Kronm€ uller, H. (ed.) Handbook of Magnetism and Advanced Magnetic Media. Wiley, New York (2006)
2099
P
5. Gruverman, A., Kholkin, A., Ramesh, R.: Nanoscale ferroelectrics: processing, characterization and future trends. Rep. Prog. Phys. 69, 2443–2474 (2006) 6. Xu, S., Hansen, B.J., Wang, Z.L.: Piezoelectric-nanowireenabled power source for driving wireless microelectronics. Nat. Commun. 1, 93 (2010) 7. Turner, R.C., Fuierer, P.A., Newnham, R.E., Shrout, T.R.: Materials for high temperature acoustic and vibration sensors: a review. Appl. Acoust. 41, 299–324 (1994) 8. Wang, Z.L.: Nanopiezotronics. Adv. Mater. 19, 889–892 (2007) 9. Murayama, N., Nakamura, K., Obara, H., Segawa, M.: The strong piezoelectricity in polyvinylidene fluoride (PVDF). Ultrasonics 14, 15–24 (1976) 10. Yang, J.S.: Introduction to the Theory of Piezoelectricity. Springer, New York (2005) 11. Wang, Z.L.: Towards Self-Powered Nanosystems: From Nanogenerators to Nanopiezotronics. Adv. Funct. Mater. 18, 3553–3567 (2008) 12. Wang, Z.L.: Piezopotential gated nanowire devices: piezotronics and piezo-phototronics. Nano Today 5(6), 540–552 (2010) 13. Yan, Z., Hu, Y.F., Xiang, S., Wang, Z.L.: Effects of piezopotential spatial distribution on local contact dictated transport property of ZnO micro/nanowires. Appl. Phys. Lett. 97, 033509 (2010) 14. B€ uhlmann, S., Dwir, B., Baborowski, J., Muralt, P.: Size effect in mesoscopic epitaxial ferroelectric structures: increase of piezoelectric response with decreasing feature size. Appl. Phys. Lett. 80, 3195–3197 (2002) 15. Wang, Z.L.: Self-powering nanotech. Sci. Am. 298, 82–87 (2008) 16. Wang, Z.L., Song, J.H.: Piezoelectric nanogenerators based on zinc oxide nanowire arrays. Science 312, 242–246 (2006) 17. Thorsen, T., Maerkl, S.J., Quake, S.R.: Microfluidic largescale integration. Science 298, 580–584 (2002)
P Piezoelectric MEMS Switches R. G. Polcawich1, J. S. Pulskamp1 and R. M. Proie2 1 US Army Research Laboratory RDRL-SER-L, Adelphi, MD, USA 2 US Army Research Laboratory RDRL-SER-E, Adelphi, MD, USA
Synonyms AlN; Aluminum nitride; Lead zirconate titanate; PZT; Relay; Relay logic
Definition A piezoelectric switch is an electromechanical relay or switch relying on piezoelectric actuation to provide
P
2100
the force necessary to close or produce a gap between electrical contacts. In practical use, these switches can be designed for systems operating with carrier frequencies from DC to tens or hundreds of gigahertz.
Piezoelectric MEMS Switches
a Ohmic Contact
Transmission Line
b
Overview The routing of electrical signals is controlled by switches of various types. For mechanical switches, a number of transduction techniques have been utilized including electrostatic, electromagnetic, thermomechanical, and piezoelectric. Fundamental to most radio frequency (RF) circuits, a switch is used not only to control the path of electrical circuits but also the phase and timing of the circuits. The continuous miniaturization of military radar and communication systems requires development of smaller, more energy-efficient switches for continuous control of a wide variety of electronic signals [1]. Additionally, switches are required for ever increasing frequencies, making existing solid state technologies less attractive to the system designer because of larger power losses and decreases in linearity as the frequency increases [2]. In short, microelectromechanical systems (MEMSs) switches provide a very low loss, low power, highly linear, with respect to input power, alternative to existing solid state switch technologies. Moreover, they are significantly smaller than macroswitch technologies. Unlike electronic switches, PIN diodes, and field effect transistors (FETs), a welldesigned MEMS switch is extremely linear with respect to input power and does not create distortion, such as harmonics or intermodulation products, in the electrical signal [2]. Next generation missile seeker radar, smart cellular phone technologies, and communication-on-the-move systems require highly linear, low-power, broadband RF switches as a fundamental building block [1, 3].
History and Brief Background of MEMS Switch Development The development of MEMS switches began with the effort of Petersen at IMB in 1979 [4] and then continued with Hughes Research Laboratories on micromachined microwave actuators (MIMACs) that
Metal
Contact
Capacitive Contact
Dielectric
Contact
Transmission Line
Piezoelectric MEMS Switches, Fig. 1 Schematic illustration of (a) ohmic and (b) capacitive switch contacts
produced the first switch capable of operating up to 50 GHz [5]. Following this research, a concentrated effort began in developing switches in both series and shunt configurations for a wide variety of applications, including lower frequency commercial cell phones and high-frequency military radar and communication systems [1–3]. All MEMS switches have a few basic components for their operation. This includes the signal line, switch contact(s), and actuator(s). The signal line can be designed for use with signals ranging from DC to sub-THz. For low-frequency operation, a simple electrical trace commonly comprised of gold or other conductive metal can be utilized. For higher frequency operation, the signal line requires greater design consideration. For MEMS switches, the signal line is commonly configured using coplanar waveguide or microstrip transmission line geometries. MEMS switches are defined by two different types of contacts: ohmic and capacitive (see Fig. 1). Capacitive contact switches utilize a dielectric thin film between two metals with adjustable gap. At elevated frequencies, the RF energy is capacitively coupled through the dielectric to the switch contact. As a result, capacitive switches are generally limited to higher frequency operation (>6 GHz) where capacitive coupling is feasible with reasonably sized capacitive contacts. In contrast, ohmic contacts use two metals to create the switch contact. Ohmic switches provide the opportunity for very broadband operation using carrier frequencies from DC on upward. It should be noted that the type of contact material is crucial for high-performance switches with long lifetime. Although beyond the scope of this entry,
Piezoelectric MEMS Switches
dielectric materials with low charge trapping densities are preferred for capacitive contacts (e.g., silicon nitride and alumina) and metals with high hardness and low probability of creating frictional polymers are preferred for ohmic contacts (e.g., ruthenium and ruthenium oxide). The final critical component is the actuator. Historically, most reported MEMS switches have utilized electrostatic actuators in which a metal beam is suspended over both a bias electrode and one of the switch contacts [2]. The remainder of this entry, however, will focus on an alternative actuation mechanism, piezoelectricity, and the piezoelectric actuators’ role in MEMS switch technologies. When compared with switches actuated by electrostatic force, one critical advantage of piezoelectric switches is the full control of their displacement when the switch is making contact. In contrast to piezoelectric switches, electrostatic switches experience a “pulldown voltage” wherein there is an instability in the gap between the actuator and its bias electrode. Once the pull-down voltage is reached and the gap of the electrostatic capacitor is reduced to two-third of the gap at 0 V, a positive feedback is created in the device. A smaller gap creates larger closing force, which in turn moves the switch in the direction of even smaller gaps. This process of constant acceleration ends when the contact is closed against a typically immovable transmission line. In contrast, piezoelectric MEMS switches move only in response to the magnitude of the DC actuation voltage. They do not exhibit positive feedback and their dynamic behavior is well controlled. As a result, switches with piezoelectric actuation can be operated where the contacts close with only a modest acceleration resulting in minimal contact bounce thus improving switch cycle reliability [6].
Piezoelectricity Piezoelectricity is comprised of two electromechanical responses: a direct effect and a converse effect. The direct effect refers to the generation of charge under the application of an applied stress and is governed by Eq. 1 where the induced electric displacement (Di) is proportional to the piezoelectric coefficient (dijk) and the applied stress (Tjk). The converse effect refers to the generation of strain under an applied electric field
2101
Piezoelectric
P
Electrodes
Elastic Layer Applied Stress
Generated Surface Charge
Piezoelectric MEMS Switches, Fig. 2 Illustration of the direct piezoelectric effect applied to a piezoelectric unimorph membrane
Piezoelectric
Electrodes
Elastic Layer
V
E P
Piezoelectric MEMS Switches, Fig. 3 Illustration of the converse piezoelectric effect applied to a unimorph cantilever with the vertical scale exaggerated
P and is governed by Eq. 2, where the induced strain (xjk) is proportional to the piezoelectric coefficient and electric field (Ei). Di ¼ dijk Tjk
(1)
xjk ¼ dijk Ei
(2)
In application, the direct effect can be used for sensing and energy harvesting in which applied stresses are used to generate surface charges on the piezoelectric layer (see Fig. 2). For actuators, an applied electric field generates a strain in the piezoelectric layer producing flexure in unimorph actuators (see Fig. 3) and bimorph actuators (see Fig. 4). In switch applications, the piezoelectric unimorph can be combined with overhanging contact cantilevers to create a functioning switch (see Fig. 5).
P
2102
Piezoelectric MEMS Switches
Piezoelectric Layer 2
Electrodes
Piezoelectric Layer 1
coefficients, d31 and e31 (see Eq. 4) by compensating for the clamping effect experienced by piezoelectric thin films in composite structures [8, 9]. e31; f ¼ e31
V
(4)
Piezoelectric Materials: Ferroelectrics and Non-ferroelectrics
Piezoelectric MEMS Switches, Fig. 4 Illustration of the converse piezoelectric effect applied to a bimorph cantilever with the vertical scale exaggerated
Contact Location
Au
actuated profile (E ) Pt PZT
cE31 e33 d31 ¼ E cE33 s11 þ sE12
Au
P
Ti/Pt Elastic Layer
Elastic Layer
Silicon
Silicon
Piezoelectric MEMS Switches, Fig. 5 Cross section schematic of a PZT unimorph and electrical contacts configured for use as switch
The tensor notation for the aforementioned relationships can be simplified using matrix notation where the subscripts i, j, and k having values 1–3 are replaced with the subscripts p and q with values of 1–6 [7]. As a result, dijk is transformed to dip. The dip (strain) and eiq (stress) piezoelectric coefficients are the most useful for thin film sensor and actuator applications and are related by the elastic constant, cEpq , in Eq. 3: eiq ¼ dip cpq
(3)
For thin film composites where the piezoelectric layer is rigidly attached to a supporting elastic layer (i.e., thin film or substrate), the out of plane stress, X3, is generally set to zero while the longitudinal strain, x3, varies from a combination of the piezoelectric effect and elastic coupling from the Poisson effect. As a result, an effective thin film piezoelectric coefficient, e31,f, can be related to the common piezoelectric
The important piezoelectric and dielectric material properties for the three most common thin film piezoelectrics, ZnO, AlN, and lead zirconate titanate (PZT) are listed in Table 1. A few distinctive features to recognize are the differences in crystal structure and the effective thin film piezoelectric coefficient, e31,f, of these three compounds. The difference in crystal structures can be summarized as a distinction between purely polar materials and ferroelectric materials. Pure polar materials such as AlN and ZnO do not permit the polarization vector to change orientations relative to the crystal lattice. The polar direction is crystallographically defined along the c-axis of the Wurtzite crystal structure. In contrast, ferroelectric materials, such as PZT, have a spontaneous electric dipole moment that can be reoriented between crystallographically defined stable states by a realizable electric field (i.e., before dielectric breakdown occurs) [10]. As a result, extremely large polarization values and subsequently large dielectric constants and piezoelectric coefficients can be achieved in ferroelectric materials. Another important characteristic distinction between purely polar materials and ferroelectric materials is the use of the constitutive equations of piezoelectricity, Eqs. 1 and 2, which describe the linear response of piezoelectric materials. These, generally, can be applied successfully to polar piezoelectric thin films like AlN and ZnO in most circumstances. However ferroelectrics, like PZT, can display pronounced stress and electric field–induced nonlinear material responses. This is particularly true for thin films and actuators. Thin films permit the application of extremely large electric fields with modest operating voltages due to the film thicknesses involved. Piezoelectric MEMS actuator applications typically drive the material well beyond the coercive field. For a half
Piezoelectric MEMS Switches
Crystal structure e31,f (C/m2) d33,f (pC/N) e33 tan d Density (g/cm3) Young’s modulus (GPa) Acoustic velocity (m/s) References
micron thin film of PZT (52/48) for example, the coercive field corresponds to an applied voltage of approximately 2.5 V. Ferroelectrics even show significant nonlinearity below the coercive field [17]. Despite this fact it is surprising that, for both bulk material and thin film applications, the linear piezoelectric equations are often used to describe the device response in ferroelectric materials. To accurately model ferroelectric MEMS device response, under high-operating-field (actuators) and/or bias field (sensors) conditions, these significant nonlinearities must be taken into account. At the high fields encountered in MEMS actuators, the total strain response is due to the combination of the linear response, nonlinear piezoelectricity including saturation effects, domain wall motion, and electrostriction. All dielectrics display the property of electrostriction, whereby externally applied electric fields induce a strain response that is proportional to the square of the field strength. Equation 5 describes the total strain response in ferroelectric materials where Q is the appropriate electrostrictive coefficient, Ps is the spontaneous polarization, E is the applied electric field, and k is the appropriate dielectric constant. The first term defines the remnant strain. The second term is equal to the linear piezoelectric coefficient (dij) multiplied by the electric field. The last term is the strain due to electrostriction. As the applied field strength increases, the electrostrictive term contributes more to the overall strain response in the material. x ¼ QP2s þ 2e0 kQPs E þ Qðe0 kEÞ2
(5)
At large applied stresses and electric fields, significant contributions to the total strain response (Eq. 5) are due to nonlinear piezoelectricity and the extrinsic response. These nonlinear material properties used for
ZnO Wurtzite 0.4 ! 0.8 10 ! 17 8 ! 12
AlN Wurtzite 0.9 ! 1.1 3.4 ! 6.5 10.1 ! 10.7
5.68 110 ! 150 6.07 103 [11–14]
3.26 260 ! 380 11.4 103 [11, 12, 15, 16]
Pb(ZrxTi1-x)O3 Perovskite 8 ! 18 90 ! 150 800 ! 1200 0.02 ! 0.05 7.5 ! 7.6 60 ! 80 2.7 103 [11, 12]
14 12 Displacement (μm)
Piezoelectric MEMS Switches, Table 1 Comparison of thin film piezoelectric materials
P
2103
10 8 6 4 2 0 −2 −20
Wafer 5124 PZT = 0.5054 mm
−15
−10
5 −5 0 Voltage (V)
10
15
20
Piezoelectric MEMS Switches, Fig. 6 Displacement versus voltage plot for a PZT thin film cantilever comprised of a 0.5 mm elastic layer (SiO2/Si3N4/SiO2)/0.1 mm Ti/Pt bilayer, 0.505 mm PZT (52/48), and a 0.1 mm Pt top electrode. Key aspects of this diagram include the hysteretic displacement characteristics upon bipolar voltage swings. Additionally, all but a small fraction of the displacement occurs in one direction
design purposes are best measured for specific materials and processing conditions. Due to these complex nonlinear effects, the ratio of total strain to applied electric field is referred to as the effective electroactive coefficients. These coefficients can be used with the models described above using field-dependent functions. Perhaps the most important consequence of these nonlinear effects is the absence of a large bipolar strain response when operated above the coercive field. Figure 6 shows voltage-displacement data from a simple unimorph PZT MEMS–actuated cantilever, illustrating the so-called butterfly-loop. The plot was obtained by sweeping the actuation voltage up to
P
P
2104
Piezoelectric MEMS Switches
40
IDT electrodes
Effective e31 (N/V-m)
30 20 unimorpha actuator
10
contact
0 −10
substrate
transmission lines
−20 Wafer 5124 PZT = 0.5054 mm
−30 −40 −30
−20
−10
10 0 Voltage (V)
20
30
Piezoelectric MEMS Switches, Fig. 7 Effective piezoelectric stress constant, e31,eff, versus voltage extracted from the displacement-voltage measurements for the PZT thin film unimorph described in Fig. 6
a positive voltage, then reducing to zero, then sweeping down to a negative voltage, and finally returning to zero field. The extracted effective electro-active piezoelectric stress constant, e31eff, corresponds to the displacement data in Fig. 6 and is illustrated in Fig. 7. This effective piezoelectric constant includes the small signal piezoelectric effect, electrostriction, and strain induced by large polarization charges associated with overcoming a pinched ferroelectric hysteresis loop. The response of the device is seen to be bipolar only in the regime between the positive and negative coercive voltages where the negative displacement reverts to a positive displacement (the “V” regions of the lower curves). Thus, for large electric fields (i.e., in excess of the coercive field) applied to the piezoelectric material, the sense of the piezoelectric strain is independent of the applied field polarity; therefore, only a single sense of the piezoelectric strain is possible. In terms of the device, the in-plane contraction of the piezoelectric material at large fields gives a negative sense to the piezoelectric actuation force. The standard configuration of the d31 mode MEMS unimorph, with the neutral axis below the midplane of the piezoelectric layer gives a positive sense of the moment arm. Piezoelectric MEMS devices based on ferroelectric materials are most often operated under unipolar conditions and hence avoid much of the hysteresis observed in Fig. 6.
Piezoelectric MEMS Switches, Fig. 8 Illustration of a thin film PZT unimorph switch developed at Penn State University [20] (Reprinted with permission. Copyright 2003 American Institute of Physics)
Piezoelectric Switch Actuator History One of the key limitations of electrostatically actuated MEMS switches is the inability to combine a low actuation voltage with excellent DC and RF performance (i.e., contact force and insertion loss, respectively) and reliability. Attempts to lower the actuation voltage with electrostatic switches generally rely on reducing either the mechanical stiffness of the released structures or the gap between the mechanical bridge/ cantilever and the corresponding biasing pad [18]. Reductions in the stiffness can limit the restoring force of the switch and can lead to stiction failures [19]. Decreasing the electrode gap limits the highfrequency RF performance (in particular, the isolation in the open state for series switches) [2]. The first functioning piezoelectric MEMS DC relay was demonstrated by Gross et al. using in-plane poled PZT film actuators [20]. These devices used a cantilever beam with a supporting elastic layer, a PZT thin film as the actuator, and patterned gold structures to act as the electrodes (see Fig. 8). This device operated as low as 20 V with a switching on time as low as 2 ms. The first reported working RF MEMS switch using PZT thin films used bulk micromachining to release a unimorph actuator and a suspended transmission line [21]. This design utilized a cantilever that is perpendicular to the wave propagation direction along the RF conductor of the coplanar waveguide (CPW). The switch operated as low as 2.5 V with an insertion loss better than 0.7 dB and isolation better than -21 dB up to 20 GHz (see Fig. 9). The first surface micromachined approach for creating a PZT-actuated RF MEMS switch was developed by
Piezoelectric MEMS Switches
G
G
S
S
G
G
Isolation [dB]
CPW ground lines
−10
−0.5
−20
−1.0
−30
−1.5
−40
Isolation 2.5 V 3.5 V 4.0 V
−50 PZT capacitor
Contact electrode
−60 0
5
Bias pad
15 10 Frequency [GHz]
−2.0
Insertion Loss [dB]
0.0
0 RF signal lines
P
2105
−2.5 −3.0 20
Piezoelectric MEMS Switches, Fig. 9 SEM image of a bulk micromachined RF MEMS series switch using a thin film PZT actuator to complete the signal path [21] (Reprinted with permission. Copyright 2005 IEEE)
Polcawich et al. [22]. These devices operated at less than 10 V and demonstrated temperature stable performance from 25 C to 100 C for normally open series switches with isolation less than 20 dB and insertion loss of better than 1.0 dB up to 40 GHz [22, 23] (Fig. 10). The interest in RF-compatible and cell-phonefrequency-compatible AlN-based filters has motivated the development of integrated piezoelectric switches and contour mode filters, both operated with AlN thin films. Mahameed et al. created the first AlN-based MEMS switches demonstrating good RF performance up to the low GHz regime [24]. These devices feature an innovative approach to address residual stress deformation, temperature sensitivity, and the modest piezoelectric coefficients of AlN. A symmetric “dual-beam” design features pairs of actuators that experience similar residual stress- and temperature-induced deformations that minimize variations in the open state contact gap despite the mechanical fluctuations (see Fig. 11). The device shows immunity to residual stresses, fast switching speeds (2 ms), moderate actuation voltage (20 V), and good low-frequency isolation (>26 dB) and insertion loss ( RHNB1 > RHNB2 > RHNB3. Hence, the average and maximum velocities and accelerations depend on the resistance of HNB as was expected from the Eq. 3. Highly doped HNB2 showed better swimming performance than lowdoped HNB1. This can be explained by the fact that the InGaAs layer thickness (estimated as 8–10 nm) of HNB2 is slightly thinner than the one of HNB1 (11 nm). It results in better performance in HNB2 than HNB1 from the reduced gravity effect and hydrodynamic friction, regardless of the higher conductivity. It reveals that the doping concentration does not play much role in the swimming performance compared to the geometry or volume effect. It should also be noted that the HNB3 and HNB4 are slightly thicker than HNB1 and HNB2 because of additional metallic or dielectric layers (Cr/Ni 10/10 nm, AZ5214 photoresist around 50 nm Table 1). However, the HNB3 had lower performance, while HNB4 was better than the others. In the case of HNB3, this can be explained by the increase in hydrodynamic friction during swimming which slow down HNB. Furthermore, in case of metallic HNB3, the
gravitational force can also increase the surface interaction force by dragging it down to the substrate. The detailed calculation and comparison of buoyancy force and gravitational force are not described here; however, the resulting nonfluidic forces (as the difference between the gravitational force and the buoyancy force) on HNB3 is 0.94 pN, which is two times higher than on HNB1 (0.43 pN). These negative effects on HNB3 reduce its swimming performance compared to the others (Fig. 8). The result of HNB4 is contrary to the HNB3 while additional layer was deposited onto the surface. It can be explained by the fact that the nonfluidic force of HNB4 was decreased by the increased buoyancy force from the photoresist, which has much lower density than both gold and nickel. The result clearly shows the high dependency on the surface charge and thus zeta potential in dielectric surface coating to HNB compared to conductive or semiconductor materials (Fig. 8). Swimming Performance Comparison with Other Swimmers The force generated by the HNB’s swimming is estimated from the drag it overcomes, considering its shape approximately as an ellipsoid [8]. 4pav F ¼ 2a 1 ln b 2
(4)
where a and b are the dimensions along the longitudinal and lateral axes respectively, Z is the viscosity of the liquid medium; and v is speed of the HNB. For
R
R
2236
Remotely Powered Propulsion of Helical Nanobelts
Remotely Powered Propulsion of Helical Nanobelts, Fig. 8 Dependence of normalized velocity and acceleration of propulsion on four types of HNBs. High- and low-doped semiconductor HNB showed higher swimming performance than metal (Cr/Ni)-coated HNB, and best performances are achieved with type 4 (dielectric)
Conclusion
Remotely Powered Propulsion of Helical Nanobelts, Fig. 9 Comparison of swimming performances of swimmers: Magnetic1 [6], Magnetic 2 [8], E. coli [1, 19] and Electroosmotic HNB
a 74-mm-long HNB with a diameter of 2.1 mm moving at 1.8 mm/s in isopropyl alcohol medium, the generated force of 1.3 nN (Eq. 4) is an order of magnitude higher than the conventional optical trap force which is less than 200 pN. Furthermore, HNBs can apply a large pressure up to 375.5 Pa, which can be advantageous for local manipulation, such as on biological membranes. Compared to two rotating magnetic propulsions of helical nanostructures, the demonstrated electroosmotic propulsion of HNBs showed much higher swimming performance in terms of maximum swimming velocity and manipulation force. The demonstrated velocity (1,785 mm/s) can reach 90 times faster and the manipulation force and pressure reaching at around 1.3 nN and 375.5 Pa which are much higher than others respectively (Fig. 9).
This entry introduced biologically inspired microswimmers based on helical nanobelts. Helical nanobelts microswimmers have excellent electromechanical and mechanical properties which can mimic their original model in nature. The demonstrated direct pulling of HNBs by electro-osmotic pumping was proven to be an efficient energy conversion mechanism for microscopic artificial swimming objects and overcomes easily the viscous drag and gravitational forces in an aqueous environment at low Reynolds numbers. Swimming velocities as fast as 1.8 mm/s (24 times the body length per second) and pressure as high as 375.5 Pa are achieved by HNBs propelled by an electro-osmotic force under an external electric field. These performances are higher than other state-of-the-art microscale artificial swimmers with physical energy conversion. Note that the velocity of 24 body lengths per second is even faster than natural bacterial propulsion which is their original model. Although helical morphology with a thin film is beneficial to such a fast swimming at low Reynolds numbers, the swimming performance appears to depend highly on the surface electrokinetic effects. Among the tested four different types, dielectric polymer coated HNB showed the best swimming performance compared to others with metallic or semiconductor surfaces. These HNBs can be used as wireless liquid manipulators, assemblers and nanorobots for biomedical or MEMS/
Reversible Adhesion
NEMS applications or remote physical or chemical detection by functionalizing with proper readouts.
Cross-References ▶ Biomimetics ▶ Dielectrophoresis ▶ Microfabricated Probe Technology
References 1. Berg, H.C.: E. coli in Motion. Springer, New York (2004) 2. Purcell, E.M.: Lift at low Reynolds number. Am. J. Phys. 45, 3–11 (1977) 3. Sundararajan, S., Lammert, P.E., Zudans, A.W., Crespi, V.H., Sen, A.: Catalytic motors for transport of colloidal cargo. Nano Lett. 8, 1271–1276 (2008) 4. Honda, T., Arai, K.I., Ishiyama, K.: Micro swimming mechanisms propelled by external magnetic fields. IEEE Trans. Mag. 32, 5085–5087 (1996) 5. Behkam, B., Sitti, M.: Bacterial flagella-based propulsion and on/off motion control of microscale objects. Appl. Phys. Lett. 90, 023902(1–3) (2006) 6. Zhang, L., Abbott, J.J., Dong, L.X., Peyer, K.E., Kratochvil, B.E., Zhang, H., Bergeles, C., Nelson, B.J.: Characterizing the swimming properties of artificial bacterial flagella. Nano Lett. 9, 3663–3667 (2009) 7. Dreyfus, R., Baudry, J., Roper, M.L., Fermigier, M., Stone, H.A., Bibette, J.: Microscopic artificial swimmers. Nature 437, 862–865 (2005) 8. Ghosh, A., Fischer, P.: Controlled propulsion of artificial magnetic. Nanostructured propellers. Nano Lett. 9, 2243– 2245 (2009) 9. Abbott, J.J., Peyer, K.E., Lagomarsino, M.C., Zhang, L., Dong, L.X., Kaliakatsos, I.K., Nelson, B.J.: How should microrobots swim? Intl. J. Rob. Res. 28, 1434–1447 (2009) 10. Martel, S., Tremblay, C., Ngakeng, S., Langlois, G.: Flagellated magnetotactic bacteria as controlled MRI-trackable propulsion and steering systems for medical nanorobots operating in the human microvasculature. Intl. J. Rob. Res. 28, 571–582 (2009) 11. Kosa, G., Jakab, P., Hata, N., Jolesz, F., Neubach, Z., Shoham, M., Zaaroor, M. and Szekely, G.: Flagella swimming for medical micro robots: theory, experiments and application. Proceeding of the 2nd IEEE RAS-EMBS International Conference on Biomedical Robotics and Biomechatronics, Scottsdale (2008), pp. 258–263 12. Chang, S.T., Paunov, V.N., Petsev, D.N., Velev, O.D.: Remotely powered self-propelling particles and micropumps based on miniature diodes. Nat. Mater. 6, 235–240 (2007) 13. Gao, P.X., Mai, W., Wang, Z.L.: Superelasticity and nanofracture mechanics of ZnO nanohelices. Nano Lett. 6, 2536–2543 (2006) 14. Hwang, G., Hashimoto, H., Bell, D.J., Dong, L.X., Nelson, B.J., Schon, S.: Piezoresistive InGaAs/GaAs nanosprings with metal connectors. Nano Lett. 2, 554–561 (2009)
2237
R
15. Wang, Z.L., Song, J.H.: Piezoelectric nanogenerators based on zinc oxide nanowire arrays. Science 312, 242–246 (2006) 16. Xu, S., Qin, Y., Xu, C., Wei, Y., Yang, R., Wang, Z.L.: Self-powered nanowire devices. Nat. Nanotechnol. 5, 366–373 (2010) 17. Karrai, K., Grober, R.D.: Piezoelectric tip-sample distance control for near field optical microscopes. Appl. Phys. Lett. 66, 1842 (1995) 18. Hunter, R.J.: Foundations of Colloid Science. Oxford University Press, New York (2010) 19. Berg, H.C., Brown, D.A.: Chemotaxis in Escherichia coli analysed bythree-dimensional tracking.Nature239,500–504 (1972)
Replica Molding ▶ Microcontact Printing ▶ Nanoscale Printing
Resonant Evanescent Wave Biosensor ▶ Whispering Gallery Mode Resonator Biosensors
Retina Implant ▶ Artificial Retina: Focus on Clinical and Fabrication Considerations
Retina Silicon Chip ▶ Artificial Retina: Focus on Clinical and Fabrication Considerations
Retinal Prosthesis ▶ Epiretinal Prosthesis
Reversible Adhesion ▶ Gecko Effect
R
R
2238
RF MEMS Switches
▶ Shark Skin Drag Reduction
translational level by introducing synthetic doublestranded or short interference RNA (dsRNA, siRNA) into the cell. Gene expression is then reduced by employing an existing cellular mechanism for processing the dsRNA into the active, single-stranded form incorporated in a complex called RISC, which targets and leads to cleavage of the specific messenger RNA (mRNA) substrate. Biomedicine is the field of medicine that develops concepts or mechanisms discovered or used in basic research into clinical applications. This process is called From Bench to Bedside. In the context of RNAi or drugs that are delivered to the organism or target tissue by using nanoparticles, biomedicine is sometimes referred to as nanomedicine. The size of the particles ranges between 1 and 300 nm.
Rigorous Maxwell Solver
Overview
▶ Finite Element Methods for Computational Nanooptics
The recent advances in the understanding of the molecular mechanisms of RNA interference (RNAi) have led to the concept that dsRNA can be delivered into cells to modulate pathogenic functions of genes. This mechanism opens new possibilities to cure diseases since the underlying cause of nearly every pathological process is the dysregulation of one or several genes at different steps in a physiological process. This therapeutic application of RNAi as a new type of drug has been quickly embraced by many researchers as a promising strategy for diseases which were hitherto considered to be difficult or nigh impossible to treat. However, the delivery of dsRNA into cells is not a trivial task due to their large size of approximately 14 kDa and negative surface charge. Additionally, RNA has a rather short lifespan under physiological conditions due to the efficient removal by nucleases. In order to successfully employ RNAi to target pathological processes, the stability of siRNA has to be addressed. This is generally achieved by introducing chemical modifications to the nucleotides without inducing toxicity or through the selected drug vesicle. Since RNAi is highly sequence-specific, an appropriate sequence in the gene of interest has to be found and preferentially tested using different siRNAs against the same target. Further important factors which should be controlled for are off-target effects (OTEs), for example, cleavage of other mRNA than the sequencedefined target, which can arise despite the precision
RF MEMS Switches ▶ Capacitive MEMS Switches
RF-MEMS/NEMS ▶ Nanotechnology
Riblets
RNA Interference (RNAi) ▶ RNAi in Biomedicine and Drug Delivery
RNAi in Biomedicine and Drug Delivery Miche`le Riesen and Nuria Vergara-Irigaray Department of Genetics Evolution and Environment, Institute of Healthy Ageing, University College London, London, UK
Synonyms RNA interference (RNAi); Short-interfering RNA (siRNA); siRNA delivery
Definitions RNA interference (RNAi) is the targeted down regulation of gene expression, most commonly on the
RNAi in Biomedicine and Drug Delivery
of the mechanism, and activation of the immune response which can be induced both sequence dependently and independently. The major challenge is still the delivery of the siRNA to the target organ for the treatment. This then expands the list of considerations to the choice of delivery vehicles and routes of administrations. For an excellent and comprehensive introduction into the field please see [11, 16]. Some of these issues, in particular the delivery of the siRNA, are solved by combining the expertise of material, chemical, and physical science to devise efficient vectors for delivery of the dsRNA into the target tissue. The development of carrier molecules in the nano-range, such as polymers, liposomes, peptides, and aptamers, goes hand in hand with the design of new therapeutics for various pathologies, such as cancer, genetic, and viral diseases. Notably, in less than 10 years since the discovery of RNAi, several drugs have been successfully developed and are currently tested in Phase I–III clinical trials in the United States (www.clinicaltrials.gov). It is worth mentioning that RNAi can originate from either synthetic siRNA, which has to reach the cytosol for final processing and activation, or as plasmid-borne short hairpin RNA (shRNA). These plasmids must be delivered to the nucleus where the RNA is transcribed and undergoes a first step of processing before it is transported to the cytosol. This type of RNAi is generally carried by viral vectors and has been subject of extensive investigation. However, the focus of this review is nonviral delivery of siRNA; hence, the reader is referred to elsewhere for a further discussion of this area. Finally, bacteria have recently been employed as delivery vehicles for shRNA or siRNA in a process termed transkingdom RNAi (tkRNAi) or bacteriamediated RNAi (bmRNAi). In this entry, the basic cellular mechanism of RNAi will be introduced. Important criteria for the design of siRNA such as sequence design and chemical modifications will be discussed together with observed offtarget effects (OTEs). This will be followed by an overview over physiological barriers and approaches chosen to deal with these challenges. Further, different types of delivery strategies will be presented which are currently employed in clinical applications according. Finally, how nanotechnology is integrated with RNAi will be discussed with an overview and discussion of a selection of particles presently utilized for in vitro, in vivo, and clinical trials.
2239
R
Basic Mechanisms RNA Interference: A Shared Strategy to Defend the Genome In 2006, Andrew Fire and Craig Mello were awarded the Nobel Prize in Medicine or Physiology for their discovery of gene silencing through doublestranded RNA (dsRNA) in Caenorhabditis elegans (C. elegans). Notably, their work describing this phenomenon had only been published for 8 years before they received the honors [5]. The observation that gene silencing could be achieved by the introduction of plasmid DNA into plants was first described some 20 years ago, but the importance of this curious finding was not recognized until it was reported by Andrew Fire and colleagues in one of the most widely used model organisms, the nematode worm C. elegans. Although the exact mechanisms of RNA interference (RNAi) were unknown when it was first reported, not much time passed before it became clear that the strategies and functions of RNAi are conserved in plants, invertebrates and vertebrates, as well as many of the components of the pathway. Once long dsRNA has entered the cell, it is trimmed into small duplexes of approximately 20–25 nucleotides in the cytosol by an RNase III enzyme called DICER1. This step is followed by integration into a complex called preRNA induced silencing complex (pre-RISC), also thought to be mediated by DICER1, where further processing into single-stranded RNA (ssRNA) takes place. RISC contains DICER1 and several proteins from the Argonaute family, AGO1–4. The passenger (sense) strand is then degraded and the mature RISC is formed with the guide (antisense) strand integrated into RISC. The guide strand is complementary to mRNA and determines the specificity of RISC. When RISC encounters its target mRNA and perfect Watson–Crick base-pairing occurs, the mRNA is cleaved by the RISC-member AGO2. If the basepairing is imperfect, the protein translation is disabled, but the mRNA is not cleaved by AGO2. This complex can be assembled stably over an extended period of time and catalytically regulate gene expression on the translational level (Fig. 1). Since the discovery of RNAi, the knowledge of the field has progressed tremendously. Species possess a vast arsenal of their own small RNAs, for example, microRNAs (miRNA) or PIWI-interacting RNAs (piRNAs). The different classes of RNAs fulfill
R
R
2240
RNAi in Biomedicine and Drug Delivery
on the understanding of gene regulatory processes over the past decade must not be diminished. For comprehensive and recent reviews over the different pathways, processes, and the components of RNAi please refer to [12, 19].
Risc AG02
target mRNA
pre-Risc DICER1
discarded passenger strand
single stranded siRNA
guide strand passenger strand duplex siRNA
cytosol
RNAi in Biomedicine and Drug Delivery, Fig. 1 siRNAinduced silencing. Double-stranded siRNA is delivered to the cytosol, where it is recognized by DICER1, and incorporated into a pre-RISC complex upon which the duplex is separated and the passenger strand discarded. This processing step is also mediated by DICER1. The mature RISC complex is then formed and guided to target mRNA. Perfect match interactions between the guide (antisense) strand and mRNA lead to cleavage of the substrate mRNA by AGO2. A fully assembled RISC complex is stable and can catalytically inactivate a large number of mRNA. The model is a simplified depiction of the siRNA pathway only and omits transport of siRNA into the cell as well as other small RNA processing pathways. The nomenclature refers to human proteins
different functions such as regulating gene expression at the translational level, or they are expressed in specific tissues, for example, the germ line or neurons. Furthermore, RNA was found in the nucleus where it was found to regulate gene silencing at the transcriptional level by modulating histone modifications or DNA methylation levels. Further evidence proposes RNAi as some kind of nuclear pest control to defend the genome against insertion of mobile elements such as L1 transposons and viral DNA. Despite the rapid succession of discoveries of RNAi-governed processes, the detailed mechanisms by which these regulations are executed remain yet to be elucidated. Nevertheless, the impact the study of RNAi has had
Design of Synthetic Short Interference RNA for Clinical Applications Sequence Design Intensive research into sequence design resulted in a set of criteria that greatly facilitates the development of suitable RNA duplexes for any given gene. There are a number of web tools available from either commercial or academic sources in order to identify appropriate RNAi target sequences [15]. The basic structure of the siRNA generally contains a central domain of 19 basepairs with an overhang of two bases at the 30 -end and a 50 -phosphate group. The reason for this type of synthetic siRNA to be common is because it is identical with the end product of the cellular RNAi pathway and DICER1. However, other siRNA varying in length between 19 and 27 basepairs and with either blunt or asymmetric ends have been shown to be efficient in vitro and to be processed by DICER1. Secondary structures of the siRNA as well as the target sequence should be considered together with the overall thermodynamic properties (Tm), which influence loading into RISC. As a rule, siRNA is more efficient when the 50 end of the guide strand has a lower melting temperature than the 50 -end of the passenger strand by adjusting the A/U and G/C content, respectively. The positioning of individual bases has also been found to have an effect on efficiency of siRNA [1, 15]. Chemical Modifications to the Sugar or Phosphodiester Backbone
Chemical modifications of the dsRNA in order to increase the stability (Tm) under physiological conditions or reduce the susceptibility to RNase attacks include the introduction or replacement of various reactive groups on the base, sugar or phosphodiester backbone of the RNA. In general, the sugar backbone is modified at the 20 O- position, for example, to 20 -Fluoro. This modification does not interfere with loading into RISC or cleavage by DICER1. Another commonly used modification is 20 -O-Methyl. This nucleotide is occurring naturally in tRNA or ribosomal components and insertion of these modified nucleotides has been shown to abolish the induction of the
RNAi in Biomedicine and Drug Delivery
innate immune system in mice. Locked nucleic acids (LNA), where the sugar ring is locked in the 30 -endo conformation due to the presence of a methylene bridge between the 20 -O and the 40 -C of the ribose, have been reported to increase stability successfully. Modifications to the phosphodiester backbone include, for example, phosphorothioate, which has been shown to increase Tm and resistance to nuclease-mediated degradation. This modification can be easily integrated at defined positions, but it has been related to toxicity, loss of function, and nonspecific protein binding if large numbers of nucleotides in the siRNA are replaced with this modification. The replacement of the phosphodiester bond with a boranophosphate greatly increases stability of the dsRNA, but synthesis of this type of backbone requires in vitro transcription and the position of the insertion of boranophosphate can therefore not be defined. For excellent and detailed reviews of the design of siRNA and modifications, please refer to [1, 2, 8]. Off-Target Effects
A serious concern when developing siRNA as therapeutics are OTEs. They arise most commonly for the following reasons: activation of the innate immune system, sequence homology with other mRNAs additionally to the specific target, and when siRNA enters the cellular miRNA pathway. Initially, it was assumed that the short RNA duplexes could not induce an immune response. However, with the increasing number of work employing siRNA in vitro and in vivo, siRNAs were reported to stimulate the innate immune system and induce a strong interferon response. This activation has been shown to be both sequence dependent and independent, as well as cell-type specific. Intracellular receptors from the Toll-like family, in particular TLR3, 7, and 8 recognize RNA, either dsRNA (TLR3) or ssRNA (TLR7 and 8), and mediate the induction of IFN-a or IFN-b. These receptors are found either on intracellular organelles such as the endosome or lysosomes, as well as cells that constitute the innate immune system such as B cells or monocytes. Under given circumstances, siRNA can therefore induce the innate immune system [22]. In order to avoid OTEs based on sequence homology, it is recommended to carefully screen the transcriptome for similarities with the required siRNA. It is important to note that although single mismatches have been reported to abolish the
2241
R
efficiency of siRNA, short similarities of five to eight bases can direct RISC to unintentional targets. If siRNA engage in the miRNA pathway, the specific effect of siRNA on one gene is replaced by effects on many genes due to the nature of miRNA-mediated gene regulation. Generally, OTEs are dose-dependent and can therefore be circumvented by establishing dose–response curves and choosing the lowest efficient dose. Since it has been shown that downregulations of the targeted gene transcripts were in fact due to OTEs and not to the direct mediation of the administered siRNA, it is strongly recommended to test at least two or three different siRNAs against the same gene. This should result in reproducible effects and confirm that the expected and observed downregulation of the transcripts is due to the direct effect of siRNA [8, 15]. Delivery of Synthetic Short Interference RNA for Clinical Applications Physiological Barriers After the successful design, synthesis, and in vitro tests of a suitable siRNA compound, the next and likely more challenging obstacle to be overcome is the delivery of siRNA to its site of action. There is a consensus in the field that the vast potential of RNAi cannot be fully exploited yet because of the lack of efficient delivery vehicles to the target tissue. The body has various highly efficient strategies for clearance of foreign particles from the system, which are briefly described below in order to understand the difficulties in delivery and how they are addressed by the choice of appropriate vesicles or siRNA modification or both combined. Tremendous efforts are directed toward the development of a range of particles with different physicochemical properties which are equipped to reach the target organ and pass the physiological barriers [7]. If siRNAs are not immediately degraded by free nucleases then the next barrier is renal clearance, which is thought to determine the half-life of modified or naked siRNA in vivo. The small size of around 14 kDa, which is far below the cutoff of 40 kDa for glomerular filtration, and the hydrophilic nature of siRNA target these molecules for removal by the kidneys. The next trial is the clearance via the reticuloendothelial system (RES). The RES consists of mobile, circulating phagocytic cells, such as monocytes, or resident tissue macrophages that remove pathogens and cellular debris which remains from apoptosis,
R
R
2242
wound healing, or tissue remodeling. Large numbers of RES-associated cells reside in the liver, the Kupffer cells, and in the spleen or generally in other tissues with high blood flow and fenestrated vasculature. Should carriers have successfully dodged clearance by any of the above-described pathways, they will then encounter the endothelial barrier lining the vascular lumen. Endothelial cells are tightly connected with the underlying extracellular matrix through integrins and they are joined together via a number of strictly regulated junctions. Endothelial cells are also perfo˚ rated with a large number of small pores of about 45 A ˚ . Most of the transport of and rare large pores of 250 A vesicles or siRNA is thought to be mediated via leaks in junctions rather than the pores themselves; however, there is some controversy regarding this matter. In many tissues, particles larger than 5 nm will not be able to extravasate through the endothelial lining, whereas organs such as the liver or spleen are easily accessible because of the extensive vasculature of these organs. One specific type of endothelial barrier is the blood brain barrier (BBB) at the boundary to the astrocytes and pericytes where the endothelium forms an exceptionally tight lining to protect the brain. As progress in research into neurodegenerative diseases such as Alzheimer’s reveals an ever-increasing number of genes which are potentially “drugable,” the brain becomes an attractive target for siRNA. Under normal physiological conditions, this barrier is intact. However, the integrity of the BBB can be compromised during disease or through administration of certain drugs. Also, specific endothelial reporters have been described which could be employed to aid passage of particles across the BBB to the brain. Therefore, development of carriers that can deliver drugs across the BBB will become increasingly important to treat this kind of diseases. The final step for successful delivery is the cellular uptake and, importantly, endosomal release into the cytosol. In general, particles enter a cell via receptormediated endocytosis. Recognition and binding to the receptor induces the formation of a vesicle which leaves the cell membrane on the cytosolic site. From there, the vesicles undergo a series of intracellular trafficking steps and maturation until, finally, late endosomes or lysosomes have been formed. The last challenge the particles face is the release from the endosome into the cell. The proton sponge effect has
RNAi in Biomedicine and Drug Delivery
been described as a mechanism by which the trapped drug can exit the endosome. Briefly, acidification of the endosome leads to protonation of amine groups on the vesicle. This leads to an influx of protons and chloride ions, followed by water to equilibrate the changes in osmolarity, eventually causing rupture of the vesicle and release of its contents into the cytosol. For further reading, please see [7, 20]. Types of Delivery
In vivo application of siRNA can be classified, for example, into active or passive systemic and local/ cellular delivery. Passive systemic delivery does not require any specific targeting sequences or molecules. It is based on the property of nanoparticles to accumulate in certain tissues under certain conditions. For example, nanoparticles or naked siRNA have been shown to accumulate in liver or spleen because of the specific structure of the endothelium, with large fenestrations of the microvasculature of these organs. Delivery of vesicles to inflamed tissue is thought to be mediated by the RES. Accumulation of particles in tumors has been attributed to the enhanced permeability and retention effect (EPR) in tumors. This is thought to be caused by reduced lymphatic drainage, leaky microvasculature, and enhanced angiogenesis of cancerous tissue. Active or targeted systemic delivery is more complex because it requires the carrier to be equipped with components such as antibodies, ligands against cellsurface receptors, or peptides from viral proteins targeting the nanoparticle to the desired cell type or tissue. Antibodies are used as whole entities, or fragments thereof, and they are generally directed against receptors which are increasingly expressed on tumor cells. Receptors that have been targeted include, for example, the epidermal growth factor receptors (EGFR) or human epidermal growth factor receptor-2 (HER-2), the iron-binding protein transferrin (Tf), which targets the transferrin receptor (TfR) responsible for iron uptake, as well as Toll-like receptors (e.g., TLR3). Successfully employed peptides are the rabies virus glycoprotein (RVG) or arginine-glycineasparagine (RGD). Sugars, i.e., glucose or mannose, are recognized by a number of cell-type specific receptors and have also been successfully employed in delivery to specific cell types or across the BBB. Naked siRNA has been injected using the hydrodynamic delivery method into mouse tail veins and was
RNAi in Biomedicine and Drug Delivery
subsequently found to accumulate in the liver and to a lesser extent in other organs such as the lungs. This method requires large volumes to be rapidly injected, and although successfully employed in mice and pigs, it is doubtful whether this way of administering systemic siRNA would be of therapeutic use in humans. The volume for injection needed is around 10% of the body weight, and this would most likely lead to heart failure and death. Direct injection of siRNA is feasible if the target is easily accessible, such as is used, for example, for treatment of wet age-related macular disease or other diseases of the eyes where siRNA is injected directly into the eye. Several Phase I–III clinical trials have been approved by the US FDA. Other means of delivery that have been tested include topical administration which is useful for exposed sites, such as the skin. Inhalation is the means of choice for treatment of respiratory tract diseases. Electroporation has been used for delivery into muscle tissue. For more detailed information, please see [14, 18, 21, 22]. Delivery Vesicles and Nanoparticles
Considerations for the Design of an siRNA Carrier Investigation into suitable carriers over the past decade has resulted in a vast amount of materials and combinations of several components which all have advantages and disadvantages for delivery of specific cargo, the route of administration, and the target tissues. Criteria to be considered are the size and size distribution of particles, which is, for example, inversely related to cellular uptake and delivery to tumor tissue. Since the smallest microvessels have a diameter of 5–6 mm, particles must be considerably below this diameter in order to avoid accumulation and clogging which could lead to embolism. The surface properties of nanoparticles dictate whether they will be recognized and removed by the host immune system; non-modified hydrophobic particles are generally swiftly removed, whereas vesicles decorated with biodegradable materials which are hydrophilic such as polyethylene glycol (PEG) or polysorbate 80 (Tween 80) are better tolerated. Drug loading capacities and the release characteristics are further important points. Loading can be achieved by two methods. The drug can be incorporated at the time of nanoparticle assembly or it can be absorbed after the formation of the carrier by incubation in a concentrated drug-solution. Highly efficient loading has been shown for siRNA due to the negative
R
2243
a
RNA-aptamer
siRNA
b
c
SNALP
siRNA
targeted uncharged lipid particle
lipid bilayers
PEG chains
HA chains
targeting moieties
RNAi in Biomedicine and Drug Delivery, Fig. 2 Delivery vehicles for siRNA administration. Three types of vesicles are shown for delivery of siRNA. (a) RNA-aptamer with siRNA ligated. (b) Stable nucleic acid particle decorated with PEG (SNALP). (c) Uncharged lipid particle decorated with targeting moieties, i.e., single chain antibody fragment directed against Tf and HA chains
charge of the oligonucleotide and positively charged vesicles, such as cationic lipids or polymers. Drug release depends, for example, on the degradation of the carrier or how fast the drug can diffuse across the polymer membrane, as well as the solubility of the drug. A concise selection of the most frequently and successfully employed carriers is briefly described below. For a detailed description of these design considerations, please see [4, 18]. The reader is also referred to cross-referenced entries which elaborate in more detail the properties and design of nanoparticles specifically for drug delivery. Aptamers Among biological macromolecules RNA has a unique place because of the ease of manipulation and synthesis similar to DNA, which is combined with a structural versatility found in proteins. These useful properties predestine RNA to be used itself as
R
R
2244
a building block for the assembly of nanoparticles called RNA-aptamers (Fig. 2a). Generally, aptamers are highly structured oligonucleotides that are obtained by SELEX (systematic evolution of ligands by exponential enrichment) for the required purpose. This is a protocol whereby a library of RNA-aptamers is subjected to iterative rounds of counter-selection and selection to yield aptamers which recognize, for example, specific receptors on target cells, proteins, as well as organic compounds. The prime features which make aptamers attractive carriers are their small size of a few nanometers compared to cationic polymers or liposomal carriers (upto 300 nm), that modified nucleotides can easily be integrated to increase stability in physiological conditions, they have low toxicity and are biodegradable. Instances where aptamers have been employed as carriers for siRNA include murine xenografts of prostate cancer where tumor growth and regression was mediated by aptamer-mediated siRNA delivery, or treatment against viral infection by either targeting HIV gp120, a glycoprotein which recognizes and utilizes the CD4 receptors on T lymphocytes for entry, or the CD4 receptor itself [6, 10, 23]. Cationic Polymers and Dendrimers The cationic polymer cyclodextrin (CDP) has been widely and successfully used as a carrier directed in particular against solid tumors. Advantages of this material are the low toxicity; it is not metabolized in humans and does not induce an immune response. CDPs are commonly of a tri-composite structure, where CDP is employed as the carrier for siRNA due to strong ionic interactions between the cationic polymer and the anionic nucleic acids. The particle is stabilized with the hydrophilic cycloalkane adamantane conjugated to PEG (AD-PEG). Targeting of the vesicle to tumors is achieved by adding an AD-PEG-Transferrin (AD-PEG-Tf) moiety because Tf is expressed abundantly in a number of tumors. This formulation carrying siRNA against the ribonucleotide reductase subunit 2 (RRM2), called CALAA-01 by Calanda Pharmaceuticals, was approved for Phase I clinical trials by the FDA in 2008. The reader is referred to [3, 9] for a comprehensive review on the development of CALAA-01. Another cationic species that is frequently used for similar properties is polyethylene-imine (PEI), which can be synthesized to be of linear or branched structure
RNAi in Biomedicine and Drug Delivery
and is then classified as a dendrimer. PEI efficiently forms stable nanoparticles with oligonucleotides and promotes endosomal release via the proton sponge effect. Because of its branched structure, it can carry a large payload. Despite the supreme quality of PEI as a siRNA delivery vesicle, its identified toxicity toward most cells is not a desirable side effect. Three approaches have been taken to address the toxicity issue of PEI. First, the dendrimer is equipped with targeting moieties such as RGD to reduce the toxicity to nontarget tissue. Second, maintaining the particle size below 10 kDa substantially reduced toxicity compared to larger particles of 25 kDa. Finally, ketalized carriers have been synthesized with the added advantage that these linkages are sensitive to acidic pH, and hence readily dissolve in endosome to release the cargo [13]. Liposomal Carriers Lipids as gene delivery vehicles have been used since the early 1980s. Cationic lipids are often used as vesicle due to the facile handling and preparation since they readily associate with the anionic nucleic acids. Moreover, routinely used reagents for in vitro transfection of cell culture are based on cationic liposomes, such as Lipofectamine 2000 by Invitrogen or other equivalent products. The most commonly used cationic lipid is 1,2-dioleoyl-3trimethylammonium-propane (DOTAP) because it efficiently associates with nucleic acids. However, DOTAP was again shown to be cytotoxic; hence, this carrier was equipped with modifications which reduce this side effect, such as lactosylation. Another strategy to inhibit the growth of pancreatic xenografts on a murine model involved DOTAP decorated with histidine-lysine peptides. This facilitated endosomal escape and single-chain antibody fragment directed against Tf, and this vesicle did not induce an IFNresponse and could be administered at low doses due to the target specificity. Another promising type of vesicle based on cationic lipids is the stable nucleic acid-lipid particle (SNALP) (Figure 2b). The siRNA is embedded within the lipid envelope and the composition of SNALPs contains reduced cationic lipid content and is PEG-ylated. SNALPs have been subjected to extensive study as siRNA carrier against in vivo viral models with considerable success of full protection against infection of guinea pigs with the deadly Zaire strain of the Ebola
RNAi in Biomedicine and Drug Delivery
virus. Another trial employed SNALPs with siRNA against apolipoprotein B levels in nonhuman primates, which was well tolerated and showed reduction of Apolipoprotein B (ApoB) and serum cholesterol among other physiological improvements. Following on from these promising results, several therapeutic siRNA-SNALP formulations against tumors or metabolic diseases are now in preclinical or have been admitted to Phase I clinical trials, and targets include VEGF and ApoB among others [8]. Further information on these trials can be found on www.clinicaltrials. gov or the web sites listed in section Further Reading. Since cationic liposomes have been associated with significant toxicities, development of vesicles composed of uncharged lipids has been resumed with considerable success. One lipid in particular has emerged as a promising carrier: 1,2-dioleoyl-sn-glycero-3phosphatidylcholine (DOPC). These vesicles successfully reduced tumor xenografts of different origins. Targeted vesicles against the thrombin receptor (TR) expressed on some tumors greatly reduced not only TR but also angiogenic and invasive tumor markers, such as vascular enodothelial growth factor (VEGF) and matrix metalloproteinase 2 (MMP2). Techniques developed in order to maximize the payload of these nanoparticles utilize lyophilization followed by rehydration with siRNA. The half-life in circulation under physiological conditions of these particles was increased by surface coating with hyaluronic acid (HA), a naturally occurring glucosaminoglycan with properties equivalent to PEG (Figure 2c). Further information can be found in [17]. Other Types of Carriers: Atellocollagen, Cationic Cell Penetrating Peptides (CPPs), Polylysine There are a number of other options for building siRNAnanocarriers. Atellocollagen is a preparation of digested cationic calf dermal collagen successfully employed and well tolerated in animal models of inflammatory disease or tumor xenografts. Despite the fact that the studied atellogen-siRNA nanoparticles were not decorated with a specific targeting moiety, they were found to accumulate in tumors and inflamed areas presumably due to the EPR effect in these tissues. CPPs are short peptides with a high content of positively charged amino acids, in particular arginine, which can be synthesized or are derived from naturally occurring peptides, e.g., TAT from HIV-1 to facilitate
2245
R
cellular uptake. Initially thought to be fully soluble, it appears that their efficiency is due to aggregation into nanoparticles. Successful trials in vivo used polyarginine fused to cholesterol and siRNA against VEGF in a tumor xenograft. Another polyargininebased CPP coupled with RVG was found to cross the BBB and reduce green fluorescent protein expression in a mouse model. Another type of carrier is based on polylysine (K) stretches. Initial problems with toxicity such as complement activation were overcome by PEGmodifications and the addition of histidine (H), which facilitates endosomal escape. Further, it was found that the efficiency as siRNA carriers was improved by increasing the histidine ratio compared to lysine [9].
Further Reading Currently, 20 studies are listed on www.clinicaltrials. gov when searched for the keyword siRNA. This clearly shows that the challenges of delivery posed by the physiological barriers or physicochemical properties of the particles can be overcome. The marriage between advances in material sciences and the developing understanding of the physiological and pathophysiological processes culminates in the development of novel RNAi-based therapeutics for the treatment of diseases ranging from viral to inherited and age-related pathologies. Due to space restraints, a number of additional excellent reviews could not be referenced in this entry; however, they can be found when searching Pubmed or Web of Science with keywords such as “siRNA AND nanotechnology.” Further information on research and development of siRNA-based therapeutics can be found on the web sites of the leading companies. http://www.alnylam.com/ http://www.calandopharma.com/ http://www.opko.com/ http://www.quarkpharma.com/ http://www.silence-therapeutics.com/ http://www.sirna.com/ http://www.tekmirapharm.com/Home.asp An insightful and curated web site on the recent development of RNAi therapeutics, career opportunities, consulting, and overview over the market and enterprise portfolios can be found here: http://rnaitherapeutics.blogspot.com/
R
R
2246
Cross-References ▶ Genotoxicity of Nanoparticles ▶ Lab-on-a-Chip for Studies in C. elegans ▶ Liposomes ▶ Microfabricated Probe Technology ▶ Nanomedicine ▶ Nanoparticle Cytotoxicity ▶ Nanoparticles ▶ Nanotechnology ▶ Nanotechnology in Cardiovascular Diseases
References 1. Behlke, M. A.: Chemical modification of siRNAs for in vivo use. Oligonucleotides 18, 305–319 (2008) 2. Behlke, M. A.: Progress towards in vivo use of siRNAs. Mol. Ther. 13, 644–670 (2006) 3. Davis, M. E.: The first targeted delivery of siRNA in humans via a self-assembling, cyclodextrin polym. based nanopart: from concept clinic. Mol. Pharm. 6, 659–668 (2009) 4. Fattal, E., Barratt, G.: Nanotechnologies and controlled release systems for the delivery of antisense oligonucleotides and small interfering RNA. Br. J. Pharmacol. 157, 179–194 (2009) 5. Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., Driver, S. E., Mello, C. C.: Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391, 806–811 (1998) 6. Guo, P.: RNA nanotechnology: engineering, assembly and applications in detection, gene delivery and therapy. J. Nanosci. Nanotechnol. 5, 1964–1982 (2005) 7. Juliano, R., Bauman, J., Kang, H., Ming, X.: Biological barriers to therapy with antisense and siRNA oligonucleotides. Mol. Pharm. 6, 686–695 (2009) 8. Lares, M. R., Rossi, J. J., Ouellet, D. L.: RNAi and small interfering RNAs in human disease therapeutic applications. Trends Biotechnol. 28, 570–579 (2010) 9. Leng, Q., Woodle, M. C., Lu, P. Y., Mixson, A. J.: Advances in systemic siRNA delivery. Drugs Future 34, 721 (2009) 10. Levy-Nissenbaum, E., Radovic-Moreno, A. F., Wang, A. Z., Langer, R., Farokhzad, O. C.: Nanotechnology and aptamers: applications in drug delivery. Trends Biotechnol. 26, 442–449 (2008) 11. Martin, S. E., Caplen, N. J.: Applications of RNA interference in mammalian systems. Annu. Rev. Genomics Hum. Genet. 8, 81–108 (2007) 12. Moazed, D.: Small RNAs in transcriptional gene silencing and genome defence. Nature 457, 413–420 (2009) 13. Mok, H., Park, T. G.: Functional polymers for targeted delivery of nucleic acid drugs. Macromol. Biosci. 9, 731–743 (2009) 14. Oliveira, S., Storm, G., Schiffelers, R. M.: Targeted delivery of siRNA. J. Biomed. Biotechnol. 2006, 63675 (2006) 15. Peek, A. S., Behlke, M. A.: Design of active small interfering RNAs. Curr. Opin. Mol. Ther. 9, 110–118 (2007)
Robot-Based Automation on the Nanoscale 16. Raemdonck, K., Vandenbroucke, R. E., Demeester, J., Sanders, N. N., De Smedt, S. C.: Maintaining the silence: reflections on long-term RNAi. Drug Discov. Today 13, 917–931 (2008) 17. Schroeder, A., Levins, C. G., Cortez, C., Langer, R., Anderson, D. G.: Lipid-based nanotherapeutics for siRNA delivery. J. Intern. Med. 267, 9–21 (2010) 18. Singh, R., Lillard Jr., J. W.: Nanoparticle-based targeted drug delivery. Exp. Mol. Pathol. 86, 215–223 (2009) 19. Siomi, H., Siomi, M. C.: On the road to reading the RNAinterference code. Nature 457, 396–404 (2009) 20. Tiemann, K., Rossi, J. J.: RNAi-based therapeutics-current status, challenges and prospects. EMBO Mol. Med. 1, 142–151 (2009) 21. Tokatlian, T., Segura, T.: siRNA applications in nanomedicine. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2, 305–315 (2010) 22. Weinstein, S., Peer, D.: RNAi nanomedicines: challenges and opportunities within the immune system. Nanotechnology 21, 232001 (2010) 23. Zhou, J., Rossi, J. J.: Aptamer-targeted cell-specific RNA interference. Silence 1, 4 (2010)
Robot-Based Automation on the Nanoscale Sergej Fatikow, Daniel Jasper, Christian Dahmen, Florian Krohs, Volkmar Eichhorn and Michael Weigel-Jech Division Microrobotics and Control Engineering (AMiR), Department of Computing Science, University of Oldenburg, Oldenburg, Germany
Synonyms AFM; Atomic force microscopy; Carbon-nanotubes; DNA manipulation; Manipulating; Nanorobotics; Nanorobotic assembly; Nanorobotic manipulation of biological cells; Nanorobotics for bioengineering; Nanorobotics for MEMS and NEMS; Scanning electron microscopy; Visual servoing for SEM
Definition The handling of micro- and nanoscale objects is an important application field of robotic technology. Nanohandling includes the finding, grasping, tracking, cutting, etc., of objects as well as different characterization methods such as indenting or scratching on the
Robot-Based Automation on the Nanoscale
nanoscale. Measurement of different features of the object, probe positioning with nanometer accuracy, structuring or shaping of nanostructures, and generally all kinds of changes to matter at the nanolevel could also be defined as nanohandling in the broadest sense. This entry addresses several approaches that can be automated with the help of nanohandling robots. The development of nanohandling robot systems working in constricted work spaces (in a vacuum chamber of an SEM or under an optical microscope) and with a nanometer resolution and accuracy is a big technological challenge for the robotics research community. Advanced actuator and sensor technologies suitable for nanohandling have to be investigated and implemented as well as vision-based feedback methods for tracking and data acquisition. Another crucial issue is the development of real-time robot control methods that meet the demands of automated nanomanipulation. This entry presents several promising solutions and applications of robot-based nanohandling.
Robot-Based Automation on the Nanoscale Introduction to the Topic The handling of micro- and nanoscale objects is an important application field of robotic technology. It is often referred to as nanohandling, having in mind the range of aspired positioning accuracy for the manipulation of micro- and nanoscale objects of different nature. The nanohandling of objects may include their finding, grasping, moving, tracking, releasing, positioning, pushing, pulling, cutting, bending, twisting, etc. Additionally, different characterization methods like indenting or scratching on the nanoscale, measurement of different features of the object, probe positioning with nanometer accuracy, structuring or shaping of nanostructures, and generally all kinds of changes to matter at the nanolevel could also be defined as nanohandling in the broadest sense. As in the field of industrial robotics, where humans leave hard, unacceptable work to robots, robots with nanohandling capabilities can help humans to handle extremely small objects with very high accuracy. The size of these robots also plays an important role in many applications [1, 2]. Highly miniaturized robots, often referred to as microrobots, are able to operate in constricted work spaces, for example, under a light
2247
R
microscope or in the vacuum chamber of a scanning electron microscope (SEM). Microsystem technology (MST) and nanotechnology require this kind of robot, since humans lack sensing capabilities, precision, and direct manipulation at those scales. Automated nanohandling by microrobots will have a great impact on both these technologies [3]. This entry will address several aspects of nanohandling automation. The development of nanohandling robot systems is a big technological challenge for the robotics research community. Advanced actuator and sensor technologies suitable for nanohandling have to be investigated and implemented. Another crucial task is the development of real-time robot control methods that meet the demands of automated nanomanipulation. The latter is challenging, especially due to the difficulty of getting real-time visual feedback and the lack of advanced control strategies able to deal with changing and uncertain physical parameters and disturbances. Trends in Nanohandling
The following three approaches are being pursued in the nanohandling research community: • Top-down approach utilizing serial nanohandling by microrobotic systems. The main goal is the miniaturization of robots, manipulators, and their tools as well as the adaptation of robotic technology (sensing, actuation, control, automation) to the demands of MST and nanotechnology. This approach is the major topic of this entry. • Bottom-up approach or self-assembly, utilizing parallel nanohandling by autonomous organization of micro- and nanoobjects into patterns or structures. • The use of a scanning probe microscope (SPM) as a nanohandling robot. In this approach, the (functionalized) tip of an atomic force microscope (AFM) probe or of a scanning tunneling microscope (STM) probe acts as a robot end-effector that is applied to a nanoscale part. This approach has been actively investigated in recent years [1, 2] and is discussed in section “AFM as a Nanohandling Robot.” Several other approaches like the use of optical tweezers or electrophoresis might also be adapted for automated nanohandling. Automated Microrobot-Based Nanohandling
Microrobotics for handling micro- and nanoscale parts has been established as a self-contained research field
R
R
2248
Robot-Based Automation on the Nanoscale
for nearly 15 years and, in recent years, a trend toward the microrobot-based automation of nanohandling processes has emerged [4, 5]. Process feedback is the most crucial aspect of nanohandling automation. It is difficult to obtain reliable information when handling microscale and especially nanoscale parts. Vision feedback often is the only way to control a nanohandling process. For this reason, the vacuum chamber of a scanning electron microscope is, for many applications, the best place for a nanohandling robot. It provides an ample work space, very high resolution up to 1 nm, and a large depth of field. However, real-time visual feedback from changing work scenes in the SEM containing moving nanorobots is a challenging problem.
The positioning accuracy of the microrobots during automated nanohandling is affected by several factors, so that a powerful robot control system is required. On the low level of the control system, driving voltages for the robot actuators are calculated in real time, which keep the robot and its end-effector on the desired path. The high-level control system is responsible among others for path planning, error handling, and the timesaving parallel execution of tasks. GUI and haptic interface are supported by the high-level control system as well. The AMNS concept has been implemented in different application fields for (semi-) automated nanomanipulation [2, 3]. Relevant aspects of this work will be introduced in the following sections.
Automated Microrobot-Based Nanohandling Station (AMNS)
Structure of the Essay
Figure 1 presents a generic concept of an automated microrobot-based nanohandling station (AMNS). The microrobots are usually driven by piezoactuators that enable positioning resolutions down to subnanometer ranges. The microrobots of the station have a nanomanipulator integrated in their mobile platform, which makes them capable both of moving over longer distances and of manipulating with nanometer accuracy. This leads to more flexibility as the robots can be deployed anywhere inside the SEM vacuum chamber. Stationary robots can also be used in an AMNS. They are easier to control compared to mobile robots, which makes them more suitable for high-throughput automation. Depending on the application, different combinations of both robot types can be implemented. The sensor system of the AMNS includes SEM, video cameras, force sensors, as well as position sensors integrated into the robots axes. The sensor data is sent to the station’s control system for real-time signal processing. The control systems task is to calculate the positions of the robots and their tools as well as the positions of the parts to be handled or other objects of interest. The calculated positions serve as input data for the closed-loop robot control. Teleoperation often is the first step on the way to automation. For this reason, a user interface is an important component of the AMNS. The user can influence the handling process by a haptic interface and/or by a graphical user interface (GUI). A good overview of state of the art teleoperation techniques and applications is given in [6].
The following is a brief summary of the topics covered in the following sections. Section “Real-Time Vision Feedback” covers the vision feedback in nanohandling robot stations. The focus is on the use of SEM in combination with realtime image processing algorithms, which is proposed as the near-field sensor for the automation of nanohandling tasks. Section “Position Control Inside SEM” introduces fundamental considerations for automated positioning inside the SEM. The section encompasses actuation types, motion principles, closed-loop control mechanisms for nanorobots, SEM-based visual servoing, as well as the difference between macro- and microscale control engineering. Section “AFM as a Nanohandling Robot” introduces the atomic force microscope as a robotic system for manipulation at the nanoscale. The AFM tip acts as a nanohandling tool to change the position or orientation of a nanoobject or to modify the surface of a substrate with nanometric precision. Different application fields are introduced and current challenges in this field of research are discussed. Section “Application Example: Handling of CNT” describes nanorobotic strategies for the microgripperbased pick-and-place handling of individual carbon nanotubes. For a reliable alignment in z-direction, a shadow-based depth detection technique is presented. Furthermore, special handling strategies are introduced for placing and releasing the CNT on the target structure. As an example, the automated
Robot-Based Automation on the Nanoscale
2249
R
Robot-Based Automation on the Nanoscale, Fig. 1 Generic concept of the AMNS [2]
assembly of so-called CNT-enhanced AFM probes is demonstrated. Section “Application Example: Handling of Biological Materials” deals with the characterization and manipulation of biological objects by an atomic force microscope. An overview of relevant parameters for AFM measurements of samples in liquids and of soft samples is given. After a short introduction of the biological background, the current state of the art in automated AFM-based handling of biomaterials is given, ranging from the design of DNA-based nanoelectric circuits to the structuring of biosensor chips for medical and forensic use. Real-Time Vision Feedback Manipulation and handling of objects on the nanoscale is subject to strong uncertainty on position and status of objects. This is due to the low mass of the objects of interest and the relatively high parasitic and interacting forces. Additionally, the choice of imaging modalities available at this scale is limited. Optical microscopes cannot deliver resolutions sufficient for imaging nanosized objects. Scanning probe microscopes like the AFM, though having the advantageous possibility
to be used as a manipulator, do not exhibit the necessary speed for closed-loop feedback. The most flexible solution for imaging in this context is the SEM. It allows for observing processes at the nanometer range and still is fast enough to enable closed-loop control feedback. Drawbacks of the SEM are the sensitivity to noise and movement artifacts during fast scans, the possibility of distortions due to magnetic or electric influences generated by the AMNS, and the lack of three-dimensional information. Various approaches have been developed to deal with these conditions. The important properties of the SEM imaging and the tasks and approaches to solve the problems associated with it will be described in the following sections. Imaging Source Properties
Certain properties of the SEM and its imaging process are important for the applied computer vision approaches. The most important negative property is the high level of noise. Noise can be reduced by frame averaging or pixel averaging which, however, also reduces the frame rate. Figure 2 shows the SEM images with low speed and high speed and the corresponding noise level.
R
R
2250
Robot-Based Automation on the Nanoscale
Robot-Based Automation on the Nanoscale, Fig. 2 SEM images at different scan speeds, acquired with a LEO 1,450 tungsten emitter SEM, (a) slow scanning, (b) fast scanning
The noise in the SEM can be modeled as additive noise: gSEM ¼ g þ gnoise With g the true gray level value, gnoise the noise component, and gSEM the acquired gray level value. Because the generation of secondary electrons is a statistic process and the number of electrons generated and detected can be assumed as Poisson distributed, the gray level fluctuation also has a Poisson characteristic. Due to the scanning principle of the SEM, fast moving objects appear distorted in the images. This is due to the fact that the object has moved during the time needed to finish each line of the scan. This can significantly reduce the performance of computer vision algorithms. Therefore, the speed of robots has to be restricted if computer vision algorithms are required. Typical Visual Feedback Tasks
For a successful nanohandling automation, visual feedback is needed for specific tasks. These can be described as object recognition, classification, and position determination. The latter can be executed in two dimensions and in three dimensions, respectively. Important especially for the position determination is speed in order to allow for closed-loop control. Still,
even with fast imaging, control is often restricted to a look-then-move strategy, because the actuation is orders of magnitude faster than the imaging and subsequent visual feedback. When using SEM image acquisition with computer vision approaches as an image-based sensor, this is a limitation which cannot be overcome easily. Object Recognition
Object recognition has the task of determining if an object of interest is in the view field or not. Many different approaches are known in traditional macroor microscale imaging and computer vision. The approaches used at the nanoscale include template matching, statistical pattern recognition, and recognition based on local features. Object recognition algorithms used for nanohandling automation in many cases do not have special and elevated needs for noise robustness, because the recognition task is not time critical and low frame rates with high averaging can be used. Approaches used in nanohandling applications use support vector machines and various feature extraction algorithms. Applications include CNT-detection and detection of TEM lamellae [7]. Two-Dimensional Position Determination and Tracking
The positions of objects and tools are not known precisely through any sensor except for the SEM. Object position determination and tracking allows
Robot-Based Automation on the Nanoscale
image-based visual servoing approaches to work. Example methods which have been developed to work on SEM images rely on template matching [8], active contours [3], and rigid body models [9]. What is common in all these methods is that some measures have been included to diminish the effect of noise on the algorithm performance, either by use of inherent noise robustness like in the case of correlation-based template matching, the use of regions to cancel noise influence, or the usage of multiple or even manifold features and the selection and weighting of these features according to their expected significance and robustness. Different algorithms may be suitable for different needs, depending on the degrees of freedom and the boundary conditions imposed by the system setup and the object properties. If the objects of interest are rigid, their appearance constant and known, their expected movement restricted to translations, and no magnification changes are required, template matching using cross correlation has turned out to be a robust and powerful yet simple approach. The idea is to calculate the cross correlation of a template image T and an input image I: Cðx; yÞ ¼ Tðx; yÞ Iðx; yÞ In the resulting matrix, each value corresponds to the similarity of the image part to the template. Assuming that the object of interest is in the image, the position is with very high probability at the position of the maximal value in this matrix: Cðx0 ; y0 Þ ¼ maxðCðx; yÞÞ The algorithm delivers a single position. It may well determine the position with sub-pixel accuracy, using weighting or interpolating techniques. The speed can be improved by using the frequency domain for calculation. In Fig. 3, the result of the algorithm can be seen. The template is marked at the place the best match is found. If the objects may rotate or change in size, active contours or rigid body models are more suitable, depending on whether an a priori model of the object is available or not. Also, the occurrence of occlusions is less problematic with these algorithms. The principle behind active contours is that a contour represented by a polygon or spline is combined with an energy
2251
R
10μm
Robot-Based Automation on the Nanoscale, Fig. 3 Tip of a microgripper tracked with correlation tracking
function which will determine its shape and position. The energy function consists of two parts: J ¼ Jint þ Jext Jint is the internal energy which purely depends on contour properties like length, compactness, or smoothness. For compactness, the internal energy Jint can be formulated as:
Jint ¼
L2 A
The second part is the external energy, which depends on the image content or image features. Possible implementations of this evaluate the distance to edges or the region statistics. The energy function is minimized for each frame by applying transformations to the contour. These transformations can be either single point translations or Euclidean transformations of the whole contour. Figure 4 shows an active contour tracking an object in SEM images. Using single point translations, deformable and nonrigid objects can be tracked, but occlusion is problematic. Using Euclidean transformations of the entire contour, the algorithm is robust against occlusion at the expense of trackability of deformable objects. Using these kinds of algorithms, full automation of nanohandling applications has been performed.
R
R
2252
Robot-Based Automation on the Nanoscale, Fig. 4 Object tracked with active contours in SEM images
Three-Dimensional Position Determination
A difficult issue with image-based position determination using SEM images is the extraction of the third dimension. Any setup used for nanohandling and nanomanipulation is essentially a three-dimensional one, which means that in many cases, a twodimensional position extracted from an image is insufficient for reliable and flexible automation. Several approaches exist to solve this problem: – Restrict the setup or degrees of freedom to a plane, which is an unpractical solution. – Use different image sources to enable depth estimation from multiple views. This includes the usage of additional microscope cameras to roughly estimate the information missing in the SEM image, as well as the use of beam deflection mechanisms to acquire SEM images from different angles [3]. – Use models of the SEM image generation and the lens distortions to extract the third dimension. This has been successfully done with smaller magnifications by [9].
Robot-Based Automation on the Nanoscale
– Use additional information from the images, such as defocus [20] or shadow to estimate the missing information or at least determine if the object or tool is at the desired location in space. All these approaches have different advantages and disadvantages. Restricting the setup or degrees of freedom as mentioned has the advantage that no additional measures are needed to deal with the threedimensional problem. This restricts the possibilities for handling and manipulation to an unacceptable extent. The use of different sources may deliver everything from rough estimates in case of microscope cameras to very precise (nm resolution) measurements in case of well-designed stereo imaging SEMs or SEM/FIB combinations. Drawbacks here are the reduction in speed when using stereo SEM and the complex correspondence problem in very dissimilar images when using a camera or the SEM/ FIB combination. The use of models for the SEM imaging has produced quite good results, but depends strongly on the accuracy of the imaging model and on the accuracy of the rigid body model of the object which is tracked. This restricts the application to some extent to objects for which a model can be assumed, and to a certain parameter range of the SEM imaging. The use of additional information from the image mainly concentrates on defocus and shadow. Using defocus of objects can deliver an accuracy up to 5 mm using time-consuming focus sweeps, and it can also be used to estimate the outof-focus displacement and therefore the position of any object for a given frame, but with reduced accuracy and reliability. This is partly due to the complex nature of the electron beam optics and the high number of factors influencing the image, especially image and object sharpness. Position Control Inside SEM A common challenge for the automation of handling and assembly tasks in the SEM is the precise alignment of two or more objects carried by different robots. Such objects can be wafers with micro- or nanostructured components as well as different tools including grippers, probes, and cantilevers. Dependent on the size of the involved objects, the accuracy of this alignment has to be in the micrometer or even nanometer range. Accuracy in this sense stands for the precision of the relative alignment, that is, a nanoobject needs to be brought precisely into the opening of a gripper.
Robot-Based Automation on the Nanoscale
For automated positioning, a closed-loop controller is necessary. Actuation Types and Motion Principles
To facilitate closed-loop motion control inside SEMs, actuators capable of generating the required resolution need to be designed. Although there are efforts to scale conventional actuators such as electromotors to enable such a resolution, for example, employing backlashfree gears, novel actuation principles are better suited. In general, an actuator capitalizes on a physical effect in order to create motion. Most of the currently employed actuators for nanohandling are based on the inverse piezoelectric effect, that is, the controlled deformation of a piezoelectric element using an electric field. Piezoelectric actuators are popular owing to their large forces, high precision, and fast response time. Such actuators can create forces in the kNrange and respond to an input signal within microseconds. Major downsides are the limited deformation of about 0.1% as well as a strongly nonlinear behavior affected by hysteresis and creep. Other physical effects that are used include electrostatic, thermal, and magnetostrictive actuators as well as shape memory alloys [4]. Piezoelectric actuators can be further categorized into scanning and step-wise actuators. Scanning actuators use a backlash-free cinematic structure based on flexure hinges to amplify the relatively small displacements of the piezoelectric elements. Although commercial actuators can reach working ranges of up to 500 mm, the working range is a severe limitation and usually a second actuator is required for coarse positioning. Furthermore, the electric field used for actuation cannot be generated with arbitrary precision and without noise. Thus, there is a trade-off between working range and resolution. Step-wise actuators are a solution to overcome these downsides. Instead of amplifying the piezoelectric element’s motion using a mechanical structure, the stroke is used to perform small steps. Step-wise actuators have a virtually unlimited working range. Most step-wise actuators are hybrid actuators, so that the piezoelectric element can also be used in a scanning mode to create movements smaller than single steps. Scanning actuators have the advantage of generating a smooth motion over the entire working range. Furthermore, due to the high force that can be generated by a piezoelectric actuator, very high
2253
R
accelerations and speeds are possible. Another advantage is that a rough knowledge of the actuator’s position can be directly derived from the applied voltage. Homing is not required for scanning actuators and even open-loop movements are reliable to a certain extent. A downside is that a scanning actuator always returns to its initial position on power loss. Thus, microrobotic systems have to be designed in a way that this movement cannot result in a severe damage to the employed robots, tools, or objects. Furthermore, the nonlinear characteristics including hysteresis and creep are scaled with respect to the full working range. Thus, these effects have a significant influence and make sensor feedback necessary. In contrast, step-wise actuators have opposed properties. Their working range is only restricted by the mechanical guides, usually in the centimeter range. They require a sensor because the executed steps are not sufficiently repeatable. Furthermore, dependent on the sensor, homing is required to find the initial location of the actuators. In addition to the large working ranges, major advantages of step-wise actuators are that they do not perform a significant movement if the power is lost and that hysteresis and creep are scaled with the step length. The most commonly applied step-wise motion principle based on piezoelectric actuators is the stick–slip actuation principle [4]. This principle makes use of the high movement speed and acceleration of a piezoelectric actuator. A mobile part is connected to a piezoelectric actuator mechanically using a contact with a specific friction. If the piezoelectric actuator is deformed slowly, the friction forces the mobile part to perform the same motion. If the actuator is deformed rapidly, however, the inertia of the mobile part keeps it in place and the friction force is overcome generating a net displacement between actuator and mobile part. Alternating slow and rapid deformations into opposite directions generate a step-wise motion over long travel ranges. Vision-Based Closed-Loop Control
On the macroscale, the location of the manipulated objects in a world coordinate system is either known from the process design or can be measured by a variety of available sensors. To perform closedloop positioning, a robot can move to this location with sufficient accuracy using its internal sensors. On the nanoscale, the position of objects has to be derived
R
R
2254
by processing images of the employed SEM (see section Real-Time Vision Feedback). SEMs, however, are designed to create highly magnified images and not to create highly accurate measurements. Thus, the resulting position is often time variant and nonlinear, so that the object’s location in a world coordinate system cannot be accurately derived. In addition, the positioning itself has limited accuracy due to the size ratio between the robotic system and the handled objects. Commonly, a robot with a size of a few centimeters is used to manipulate an object with a precision of a few nanometers, that is, the robot is about six orders of magnitude larger than the required accuracy. Effects such as thermally- induced drift and limited mechanical stiffness lead to a nonlinear and time-variant relation between the position measured by a robot’s internal sensors and the real position of the robot’s tool. Thus, internal sensors are insufficient for reliable nanopositioning. As known from the macroworld, the limitations of positioning based on internal sensors can be overcome by using the information generated by a visual sensor. Using vision feedback as a sensor for robotic manipulation is promising as it is a noncontact measurement and mimics human behavior. The initial technique, which is still used for most SEM-based positioning, extracts a position deviation from an image and then moves the robot either open loop or based on internal sensors to decrease this deviation. When the robot stops, another image is taken, again extracting and compensating for the deviation. This is called the look-then-move approach [10]. Two advantages of the approach are that low throughput rates of image processing are acceptable and that the entire setup can be handled in a quasi-static fashion neglecting dynamic aspects. The effectiveness, however, entirely depends of the accuracy of the visual sensor and the robot’s open-loop control. An alternative is to use the visual feedback as a lowlevel element in the motion control loop of a robot. This is called visual servoing [10]. Two challenges need to be solved to achieve effective visual servoing. First, the image processing needs to be fast. As the available processing power has significantly increased over the recent years, several image processing algorithms can be used for visual servoing. More complex algorithms, however, still require computation times on the order of seconds and do not allow for effective servoing. Second, the dynamic aspects of visual
Robot-Based Automation on the Nanoscale
servoing need to be considered. The robot’s movement as well as resulting vibrations leads to movement during image acquisition. This movement creates image distortions rendering most of the current image processing algorithms useless. Enabling an effective visual servoing, however, is a necessary step on the way of high-speed micro- and nanohandling viable for industrial application. Closed-Loop Control
Based either on an internal sensor or on vision feedback, a closed-loop controller can be implemented. Closedloop control on the micro- and nanoscale is significantly different from macroscale control [1]. Macroscale control mostly solves challenges arising from the dynamic behavior of systems, for example, inertia. On the microand nanoscale, these effects become negligible. As an example, a microrobot might carry a mass m of 30 g and can create a force F of approximately 0.02 N. Thus, its maximum acceleration a is: a ¼ F=m ¼ 2=3 m=s2 If it needs to accelerate to a velocity v of 500 mm/s, which is high for a micro- or nano-operation, it requires: t ¼ v=a ¼ 0:375½ms This is significantly faster than the update rate of any visual sensor. Thus, acceleration effects can be neglected. Therefore, for many applications and actuator types, a simple proportional controller is sufficient to perform an effective alignment. AFM as a Nanohandling Robot Since its development by Binnig, Quate, and Gerber [11], AFM has been increasingly recognized not only as a tool for imaging purposes, but also for manipulating at the micro-, nano-, and atomic scale. The main advantages of the AFM are its high resolution in the sub-nanometer range as well as its flexibility for operating under physiological conditions and in vacuum or liquid environments, depending on the specific application. In addition to its visualization capabilities, the AFM can be used for force spectroscopy and, with functionalized tips, for measuring electric, magnetic, or chemical properties. Moreover, the AFM tip can be
Robot-Based Automation on the Nanoscale
2255
R
Robot-Based Automation on the Nanoscale, Fig. 5 Schematic view of manipulation tasks that can be performed by AFM-based robotic systems [1, 2]
used to manipulate objects or materials with nanometric precision and thus the AFM can also be seen as a nanohandling robot. AFM-based manipulation benefits from the fact that the AFM directly measures forces that are exerted during the manipulation and thus gives important feedback on the progress of the manipulation. Potential applications of AFM-based manipulations are manifold, ranging from the fabrication of nanoscale devices, reparation or modification of nanostructures to the characterization and handling of biological samples. Different types of manipulation can be conducted with the AFM including pushing, pulling, cutting, indenting, and structuring of a wide range of materials (Fig. 5). Starting with unintended manipulations of AuClusters, reported 1987 by Baro´ et al. in [12], the AFM as a tool for manipulating single nanoentities became subject of increasing interest in the research community. A controlled manipulation of nanoscale objects by an AFM has been actively investigated over the last decade. Many publications exist describing the utilization of the AFM as an instrument for manipulating spherical nanoobjects such as gold nanoparticles as well as longish nanoobjects such as carbon nanotubes (CNTs), nanorods, or also biological objects such as DNA [2]. Figure 6 illustrates a successful manipulation of multiwalled CNTs that was conducted by the AFM tip. A commercially available dynamic-mode cantilever was used for both – the image acquisition as well as the manipulation. For imaging, conventional intermittent mode was used, whereas for manipulation the tip was brought into proximity with the surface and then moved laterally to mobilize the CNTs.
To perform a controlled movement of a nanoscopic object, it has to be localized and identified on the substrate first. A major drawback of AFM-based manipulation is the lack of an additional, visual sensor. Therefore, before any manipulation is conducted, the AFM usually has to be utilized to obtain a detailed image of the scenery in order to plan the execution of the manipulation. This imaging scan is most often performed in dynamic mode (intermittent or noncontact) in order to avoid any damage or unintended manipulation of the sample. For the manipulation, the AFM tip is brought into proximity with the object and moved – either in dynamic or contact mode, but without any feedback to control the z-direction – toward a predefined position. As a result, the object is then pushed forward by the repulsive forces. After a manipulation step, reimaging of the area of interest often becomes necessary to reveal the result of the manipulation. The disadvantage of this “look and manipulate” scheme is that the image acquisition process usually needs several minutes depending on the required quality. Moreover, the manipulation itself cannot be monitored by means other than the measured forces exerted by the cantilever and thus have to be performed in a “blind” way. Unfortunately, due to a variety of uncertainties, the results of the manipulation are often not satisfying and require frequent trial and error experiments. Real-time imaging would greatly help to make the manipulation process faster, more robust, and accurate. However, optical microscopy cannot be used as an additional real-time capable imaging method because its resolution does not fit the requirements for the manipulation of objects with
R
R
2256
Robot-Based Automation on the Nanoscale, Fig. 6 AFM topography star with diameter of 10 mm, fabricated on a gold layer (left) using mechanical AFMlithography and the replica of this structure fabricated on polypropylene (PP) using the microinjection molding technique (right)
nanometric dimensions. An alternative is given by the integration of the AFM into the vacuum chamber of an SEM to allow for real-time observation of the manipulated object and the AFM probe. However, this entails the disadvantage of restriction to vacuum and to some extent conductive sample materials, which is not feasible for many applications. Besides its ability to manipulate nanoscale objects, the AFM tip can also be used to modify surfaces with nanometric precision or to change an object’s shape, for example, by scratching, indenting, cutting, and dissecting. For nanomachining purposes, the AFM tip can be exploited as a nanohandling tool in order to act as, for example, a milling cutter, nanoscalpel, or nanoindenter. However, the type of interaction between AFM tip and substrate is not only mechanical, but may also involve chemical, thermal, or electrical impact. Over the last two decades, a large number of promising nanolithographic technologies have evolved and are still emerging that are based on such AFM-based nanomanipulation of surfaces. These techniques can be used among others for mask-free lithography on the nanoscale and provide a promising alternative to conventional technologies. Using AFMbased surface modification, material can either be removed (e.g., through direct mechanical scratching), chemically changed (e.g., resist exposure techniques or oxidation), or deposited (e.g., dip-pen lithography). A detailed review of such SPM-based lithography can be found [13]. An example for a novel process chain for small series production of nanostructures that is based on AFM-based mechanical lithography, which is the most basic type of AFM lithography, was introduced [14]. Figure 7 depicts how the mechanical AFM-lithography is utilized to fabricate a nanostructure that was subsequently replicated several times by means of the microinjection molding technique.
Robot-Based Automation on the Nanoscale 60,0 nm
60,0 nm
40,0
40,0
30,0
30,0
20,0
20,0
10,0
10,0
0,0
0,0
Up to now, the AFM as a robot for the nanomanipulation can only be found in laboratories for prototypical issues. For industry, the serial AFM-based approach is still not reliable and fast enough to be successfully applied for productive use. To increase throughput and reliability of this sequential process, automation can be considered as a desirable goal. To reach this goal, however, a large number of challenging prerequisites have to be fulfilled. Besides the lack of real-time imaging of the manipulation process discussed above, spatial uncertainties constitute an import error source. Although positioning inaccuracies caused by creep and hysteresis of piezoactuated scanning stages can nowadays most often be compensated by closed-loop operation, thermal drift of the AFM components lead to spatial positioning errors that are challenging to counteract. Solutions to overcome this problem have been addressed [1, 2, 15]. Moreover, sticking effects also play a crucial role in AFM-based nanohandling and therefore, substrate, cantilever, and tip material have to be carefully chosen in order to minimize these effects. A promising strategy that is pursued by several research groups aims at the precise modeling of the nanohandling task. To compensate for the uncertainty introduced by the inevitable “blind” manipulation, a model of the manipulation process is utilized enabling a manipulation in open-loop mode. Having a valid model of the nanomanipulation, including all relevant interactions between nanoobject, tip, and substrate, it is possible to calculate the expected position of the nanoobject during manipulation in real time. Using such a model for an augmented reality system, where a virtual reality is augmented by the calculated nanoworld situation, enables an additional visual feedback. However, this approach requires exact knowledge of the nanomanipulation phenomena, which is not yet available in the current state of nanosciences.
Robot-Based Automation on the Nanoscale
2257
R
25,0 nm 500 nm 20,0 15,0 10,0 5,0 0,0
Robot-Based Automation on the Nanoscale, Fig. 7 Multiwalled carbon nanotubes on HOPG substrate manipulated by the AFM. When moving the AFM tip laterally, the height control loop was turned off in order to avoid slipping
of the AFM tip over the CNT. It could be observed that, depending on the parameters chosen for manipulation (tip velocity, damping amplitude), cutting as well as pushing and pulling of the CNTs can be accomplished
Application Example: Handling of CNT Nanomaterials such as different kinds of nanowires and especially carbon nanotubes (CNTs) [16] have unique physical properties and can thus improve and enhance existing microsystems in a multitude of applications. Two main application areas have been identified where in particular the integration of CNTs into known microsystems is of high impact: CNTenhanced AFM probes [17] for the analysis of highaspect ratio structures and CNT-interconnects in microchip technology [18]. Even though parallel fabrication techniques will be required for a future bulk production of such devices, enabling technologies for the reliable manipulation of individual nanostructures and their prototypic integration into microsystems have to be developed. Automated robot-based assembly techniques are very promising to implement a fast and systematic prototyping of CNT-based devices. In the following section, a microgripper-based strategy for the automated assembly of CNT-enhanced AFM probes is presented.
developed. Most important are methods for the z-approach of gripper and CNT and for the reliable gripping and placing of CNTs that lead to a welldefined system state.
Nanorobotic Strategies for Microgripper-Based Handling of CNTs
In prior experiments, electrothermally actuated microgrippers have already been used for teleoperated pick-and-place handling of individual CNTs [19]. The CNTs used for the experiments have a typical diameter of 200 nm and are grown vertically aligned on a substrate with a spacing of 50 mm between nanotubes. In order to automate the microgripper-based handling of CNTs, special nanorobotic strategies have been
Methods for z-Approach of Microgripper and CNT An AMNS has been developed that can be integrated into the SEM [2], so that visual feedback from the microscope is available during the nanohandling tasks in addition to the integrated sensors of the positioning systems. Image processing algorithms for object recognition and tracking have been developed (compare section Real-Time Vision Feedback) providing the x,y-position of gripper and nanotube. This two-dimensional position information can be used to automate the approach in x,y-direction. However, to receive information on the gripper’s and CNT’s relative z-position, additional contact sensors or threedimensional systems are required. Such systems cause additional costs and reduce the available space within the SEM’s vacuum chamber [20]. A novel so-called shadow-based depth detection method has been developed that, together with the focus information of the SEM, facilitates the automated z-approach of microgripper and CNT in any available SEM. First of all, the “depth-from-focus” technique [20] is used to coarsely align both objects in z-direction down to 50 mm. Then, for the fine approach of gripper and CNT, a shadow can be detected as soon as the CNT is located exactly between the gripper jaws. The gripper jaws block the secondary electrons from the detector. Figure 8 shows SEM
R
R
2258
Robot-Based Automation on the Nanoscale, Fig. 8 Shadow-based depth detection for the alignment of microgripper and CNT in z-direction. (a) CNT is located below the gripper jaws, no shadow is detectable. (b) CNT is between the gripper jaws, a shadow is clearly visible
Robot-Based Automation on the Nanoscale
a
b
10 μm
images illustrating the shadow-based depth detection method. The clearly detectable shadow on the CNT in Fig. 8b is used as a stop criterion for the z-approach. Gripping and Removing CNTs from the Growth Substrate Different gripping techniques for the reliable picking and removing of individual CNTs that are grown on a silicon substrate by chemical vapor deposition (CVD) have been discussed in [21]. The socalled shear gripping technique has been identified as the most reliable approach, requiring smaller gripping forces compared to tensile gripping. The microgripper is closed around the nanotube and moved sideways to break the CNT off the substrate. After that, the gripped CNT can be transferred to the target structure. Placing and Releasing CNTs on the Target Structure A novel handling strategy for placing and releasing CNTs onto the target structure has been developed. This strategy relies on the adhesion forces between CNT and microgripper jaws and thus avoids the need of additional joining techniques. After reopening the microgripper, the nanotube can be oriented in four different configurations. Either the CNT is connected by a line contact to one gripper jaw (see Fig. 9a, b) or the CNT is connected by two-point contacts to the gripper jaws (see Fig. 9c, d). In case of the two-point contacts, the adhesion forces are smaller compared to the line contacts. For this reason, the two-point contact is preferred in order to easily place and release the CNT on the target structure. In addition, obtaining a well-defined system state after reopening the gripper is required to allow full automation of the placing and releasing process. For the automated microgripper-based mounting of an individual CNT onto the tip of an AFM probe, the strategy shown in Fig. 10 has been developed.
a
b Gripper
c
CNT
d Robot-Based Automation on the Nanoscale, Fig. 9 The four possible configurations after reopening the microgripper. (a), (b) Line contact between CNT and gripper jaw. (c), (d) Twopoint contact between CNT and gripper jaws
d
a
b
AFM probe
e
c f
Robot-Based Automation on the Nanoscale, Fig. 10 Nanorobotic handling strategy for automated placing a CNT onto the tip of an AFM probe and releasing it from the microgripper
Robot-Based Automation on the Nanoscale
The closed microgripper holding the CNT (Fig. 10a) is approached to the tip of an AFM probe until a bending of the nanotube can be detected (Fig. 10b). This bending is detected automatically and used as the stop criterion of the z-approach. At the same time, the exerted force assures that CNT and gripper jaws are forming a two-point contact with defined orientation after reopening the microgripper (Fig. 10c). After this, the CNT is moved toward the AFM tip until a line contact is achieved (Fig. 10d). Because the adhesion forces of the line contact between CNT and AFM tip are much higher than the forces of the two-point contacts between CNT and gripper jaws, the nanotube can be easily released from the gripper jaws by retracting the microgripper (Fig. 10e–f). Automated Assembly of CNT-Enhanced AFM Probes
The position control techniques described in section “Position Control Inside SEM” have been used to fully automate the nanohandling task of mounting a CNT onto the tip of an AFM probe. For this purpose, the whole automation sequence has been divided into three subsequences: The initialization process, the gripping and removing process, and the process of placing and releasing the CNT. In the following, these three subsequences are described in detail with respect to automation purposes. Initialization Sequence The first step of an automated robotic micro-nano-integration is the initialization. For this purpose, all actuators are moved to a well-known starting position and the coordinate system of the internal position sensors are mapped to the coordinates of the SEM image. Then, the electrothermal microgripper mounted to the mobile microrobots is calibrated. A pattern of the microgripper is recorded that allows for tracking the position of the gripper by applying cross-correlation algorithms to the SEM images (compare section “Real-Time Vision Feedback”). After this, the CNT substrate is scanned automatically to index all CNTs. For each CNT, length and orientation are checked as a quality criterion and only the positions of suitable CNTs are stored to the database. Finally, the AFM probes are indexed by saving the exact position of the probes’ tips to the database. Picking Sequence As described in section “Gripping and Removing CNTs from the Growth Substrate”, the
2259
R
shear gripping technique is used to grip and remove an individual CNT from the sample substrate. Figure 11 shows SEM images of the gripping sequence illustrating the process of breaking the CNT off the substrate. First, so-called coarse and fine positioning steps are performed placing the CNT between the gripper jaws (see Fig. 11a). Second, the microgripper is actuated and the CNT is detached by moving the gripper sideways (see Fig. 11b). Third, the gripped CNT can be transferred to the target structure. Placing Sequence For the automated placing of the CNT onto the tip of an AFM probe, the nanorobotic strategy described in section “Placing and Releasing CNTs on the Target Structure” is used. Figure 12 shows SEM images of the automated placing sequence. After the coarse and fine approach between CNT and AFM probe has been realized in x,y-direction (see Fig. 12a), the final z-approach is performed until a bending of the CNT is detected (see Fig. 12b ). Then, the gripper is reopened and a two-point contact between CNT and gripper jaws as well as a line contact between CNT and AFM probe tip is established (see Fig. 12c). Finally, the CNT can be easily released from the gripper jaws and placed onto the tip of the AFM probe by retracting the microgripper (see Fig. 12d). Application Example: Handling of Biological Materials Current research areas in molecular and cell biology, medicine, and process sensor technology (sensor systems monitoring various processes in industry and science) often require advances in the nanoengineering technologies to look for molecular phenomena with the highest possible resolution in order to enlighten the “black-box” which still shades most metabolic reactions. The aim of present work, starting from current medical problems, is to propose novel studies especially in molecular and cellular microbiology as well as virology with direct applications in medicine and industry, closing the bottom-up cycle. Thanks to advances in micro- and nanofabrication and microand nanorobotics, robotic systems which offer the possibility of characterizing and manipulating biological objects can be developed today. The use of such systems enable new studies of single cell phenomena with nanometer resolution (e.g., studying the local mechanical and electrical properties of single bacteria for characterization and evaluation of the resistance to
R
R
2260
Robot-Based Automation on the Nanoscale
a
b
c
5 μm Robot-Based Automation on the Nanoscale, Fig. 11 SEM images of the automated gripping sequence. (a) Coarse and fine alignment of gripper of CNT. (b) Removing the CNT from the
Robot-Based Automation on the Nanoscale, Fig. 12 SEM images of the automated assembly. (a) Coarse and fine alignment of gripped CNT and AFM tip. (b) Final z-approach until a bending of the CNT is detected. (c) Creating a twopoint contact between CNT and gripper jaws and a line contact between CNT and AFM probe. (d) Releasing the CNT by retracting the gripper
substrate by shear gripping. (c) Closed gripper holding the detached CNT
a
b
c
d
antibiotics or to the spread of infections) up to complete cell compounds (e.g., studying the mechanical properties of bacterial biofilms) and deepen the appreciation of the processes at the nanoscale. The following section will give a brief overview of the handling of biological objects using automated AFMbased approaches. Manipulation and Handling of Biomaterials Using Atomic Force Microscopy
Section “AFM as a Nanohandling Robot” has already shown the basics of the AFM for the automated handling of nanoscale materials. In this subsection, a brief overview of the differences for biomaterials is given. For characterization or manipulation, samples have to be immobilized on an adequate substrate. A common method is the immobilization on freshly cleaved mica. Mica has a hydrophilic, charged surface that
10 μm
binds proteins and other biomolecules easily. Occasionally, mica is treated with buffer systems, which change the charging of the surface. Gold substrates can also be used if coated with protein reactive monolayers. The characteristics of the used cantilevers are determined by material, tip shape, and mechanical parameters such as spring constant, resonant frequency, and Q factor. For biological applications, cantilevers are usually made of silicon or silicon nitride. These materials are chemically inert and can be doped to induce electrical charges. Soft cantilevers (C < 1 N/m) are used for the contact mode and hard cantilevers (C 42 N/m) are used for the tapping mode. If operated in liquids, the pH value and the presence of electrolytes influence the forces between tip and sample. Although a smaller tip radius results in more accurate images, it may be advantageous for soft biological
Robot-Based Automation on the Nanoscale
samples, to use a duller tip reducing the risk of accidentally damaging the sample. Compared to other methods (e.g., optical tweezers, dielectrophoresis) the AFM has some major advantages for the handling of biomaterials. The most important points are [1, 2, 22]: • The AFM can be used in liquids, in vacuum, and at ambient conditions meaning that biological samples can be characterized and manipulated at their necessary physiological conditions. Furthermore, living samples and their reaction to mechanical stimulation can be studied. • The AFM can be used for imaging as well as for manipulation nearly at the same time. • The possible resolution lies in the sub-nanometer range. • Three-dimensional topological information can be obtained. • No special preparation methods are necessary. However, compared to the other methods AFMbased manipulation entails major drawbacks and limitations. Knowing these limitations is important to choose the right method or tool to handle or characterize a specific sample. The most important limitations are: • A three-dimensional force measurement is not yet possible. • The scanning speed is generally too low for realtime mapping. • The scanning range for high-speed AFMs is still very low (about 3 mm 3 mm). Examples of Automated Handling of Biomaterials with an AFM
In the following sections, current work on the field of handling, characterization, and manipulation of biomaterials is described. Current research mainly deals with the development of fully automated solutions for handling and manipulation of DNA, for solving of packaging problems in the future of nanoelectronics, and for structuring of biosensors for medical and forensic use. Handling and Manipulation of DNA In accordance with the progressing miniaturization of electric circuits, new ways to realize smaller channel widths are needed. To overcome the limitations of silicon-based processing, the use of objects like CNTs or biomolecules like DNA is a promising solution. Nanowires
2261
R
produced via the metallization of DNA are promising candidates for nanoelectronic devices of future generations. DNA can be considered a basic building block for nanostructure fabrication because it provides unique self-recognition properties. DNA constitutes an ideal template for the organization of metallic and semiconductor particles into wire-like assemblies. Nanoscopic (thickness below 10 nm) and regular wire-like metallic structures can be efficiently produced because DNA molecules have a diameter of only two nanometers and a length in the micrometer range. Moreover, both the ends of DNA molecules and the surface of solid substrates can be functionalized through a variety of methods to create specific links between DNA and substrate, allowing for the integration of DNA molecules into specific, predefined sites of microstructured electronic circuits. By exploiting the specific recognition of complementary nucleotide sequences, complex structures made of DNA can be fabricated and metalized at a later stage. Packaging problems of next-generation nanoelectronics technologies, (e.g., single-electron transistors, quantum automata, molecular electronics, etc.) might also be addressed by using DNA strands. Recently, many researchers have proven the feasibility of DNA manipulation by an AFM tip, primarily in liquid conditions. The next step toward highthroughput fabrication is automation of DNA handling. Figure 13 show two alternative immobilization procedures of DNA, which offer the potential of automated DNA manipulation in dry ambient conditions. Figure 13a shows a topographic scan of a DNA strand. The height of the DNA strand is about 1.5 nm. Figure 13b shows the same DNA strand after several manipulations by the AFM cantilever tip. The DNA was manipulated and successfully moved about 25–100 nm. Figure 13c, d shows an experiment to demonstrate the transport of DNA. Two separate DNA strands are shown. The upper DNA strand (Fig. 13c) was moved by several AFM strokes up to 300 nm. As a result, this DNA strand was removed from its location (Fig. 13d), without affecting the position of the lower DNA. The presented methods for the handling of biomaterials show the potential for different usage scenarios in industry and scientific research. In case of the handling of DNA, the experiments demonstrate that a DNA manipulation and handling can be performed in dry ambient conditions.
R
R
2262
Robot-Based Automation on the Nanoscale
Robot-Based Automation on the Nanoscale, Fig. 13 (a, b) DNA immobilized on silicon without surface modifications (a) and the same DNA after the manipulation process (b). On the left side of the strand, two short distance movements from bottom to top were implemented. As a result, the corresponding DNA parts have been pushed about 25 nm. On the right side, two long
distance movements were implemented, resulting in DNA displacement by 100 nm. (c, d) DNAs immobilized on silicon without surface modifications (a) and the same substrate with upper DNA strand pushed away on the surface by the AFM tip, without affecting the lower strand (b)
Automated Structuring of Biological Materials for Biosensors Biosensors consist of a sensitive biological component, a transducer, or detector element, and associated electronics or signal processors. An accurate and well-defined preparation of the sensitive components made, for example, of microorganisms, cell receptors, proteins, enzymes, anti-bodies, or nucleic acids is crucial part of the sensor design. Commonly used methods are microstamping and microprinting, both of which are unable to decrease the structures of the biological component down to the nanometer range. However, higher density is necessary to increase the measurement accuracy and the number of different, simultaneously useable biocomponents. In this section, current work on automated design of high-density sensitive components using AFM-based lithography is introduced. The compensation of spatial uncertainties plays an important role in automation of AFM-based handling. Although the effects of hysteresis and creep of the PZT-based scanner can be reduced by closed-loop control, spatial uncertainties caused by thermal drift are crucial and less straightforward to deal with. Especially in life science applications, where the AFM is often operated in humid environments, the effect becomes strong and often prevents successful nanohandling, let alone automation of the handling
task. To achieve reliable results, the recently developed method for real-time drift compensation described in section “AFM as a Nanohandling Robot” and in [1, 15] was applied, both for the presented manipulation of DNA as well as for the AFM-based nanotooling described below. As a first step toward the automated fabrication of biosensor components, rectangular areas were successfully structured into an APTES monolayer on mica (Fig. 14). To further investigate the correlation between the forces applied during manipulation and the resulting machining depths, a series of experiments was performed to scratch multiple areas. By using the AFM, an automated structuring of biological materials for biosensors can be done. The automation of the necessary processes will also lead to a feasible industrial use, as well as for clinical and research laboratories. The size of the resulting structures correlates with the resolution of the commonly used read-out techniques such as fluorescence microscopy or surface plasmon resonance spectroscopy.
Future Trends In the future, closed-loop motion control in SEMs needs to be changed from the quasi-static
Robot-Based Automation on the Nanoscale
2263 6,3 nm
5,4 nm
2,2 nm
0,0 nm
R
Robot-Based Automation on the Nanoscale, Fig. 14 Height image of 1 1 mm area scratched into APTES monolayer by AFM machining. The two visible machining
depths arise from different force set points used during the processing (left) and AFM height image of a series of 1 1 mm sized scratched areas (right)
look-then-move-approach to an effective visual servoing in order to implement high-speed microand nanohandling viable for industrial application. Furthermore, to allow for a more effective avoidance of both collisions and vibrations, the commonly implemented position control needs to be extended to trajectory control so that the robot moves with a predetermined velocity, acceleration and jerk at all times. Considering visual feedback, an important future task is to enable more reliable and faster feedback especially in case of fast moving objects, because of the artifacts and distortions generated by the object movement. Also speed, accuracy, and resolution of the z-position estimation still is a topic where significant improvements can be made. The AFM has emerged as an important tool for the fabrication of nanoscale structures, as well as the handling and manipulation of biological objects and for the manipulation of nanoscale objects. To broaden the applicability of those techniques for productive use, it will become necessary to increase throughput significantly. For the lithographic techniques used to structure surfaces, the usage of multiple AFM tips in parallel will greatly enhance throughput. To speed up the assembly of individual nanoentities, achieving a high level of automation is an important research goal for the future. Further miniaturization of gripping tools can facilitate the nanorobotic pick-and-place handling of CNTs with smaller diameters. The optimization of the presented automated assembly sequence of CNT-enhanced AFM probes will lead to a highthroughput prototyping of CNT-based devices, which is an important pre-stage toward a direct integration of
CNTs into microstructures by CMOS-compatible CVD-based fabrication techniques.
Cross-References ▶ AFM in Liquids ▶ AFM Probes ▶ AFM ▶ Atomic Force Microscopy ▶ Basic MEMS Actuators ▶ BioPatterning ▶ Biosensors ▶ Carbon Nanotubes for Chip Interconnections ▶ Carbon-Nanotubes ▶ Chemical Vapor Deposition (CVD) ▶ Dielectrophoresis ▶ Dielectrophoretic Nanoassembly of Nanotubes onto Nanoelectrodes ▶ Dip-Pen Nanolithography ▶ DNA Manipulation Based on Nanotweezers ▶ Electric-Field-Assisted Deterministic Nanowire Assembly ▶ Electron-Beam-Induced Deposition ▶ Manipulating ▶ Microfabricated Probe Technology ▶ Nanogrippers ▶ Nanoparticles ▶ Nanorobotic Assembly ▶ Nanorobotic Manipulation of Biological Cells ▶ Nanorobotic Spot Welding ▶ Nanorobotics ▶ Nanorobotics for Bioengineering
R
R
2264
▶ Nanorobotics for NEMS using Helical Nanostructures ▶ Nanostructured Functionalized Surfaces ▶ Nanostructured Materials for Sensing ▶ Optical Tweezers ▶ Organic Bioelectronics ▶ Piezoelectric Effect at Nanoscale ▶ Piezoelectric MEMS Switch ▶ Scanning Electron Microscopy ▶ Scanning Tunneling Microscopy ▶ Self-assembled Monolayers ▶ Self-assembly ▶ Self-Assembly for Heterogeneous Integration of Microsystems ▶ Self-assembly of Nanostructures ▶ Thermal Actuators ▶ Transmission Electron Microscopy
Robotic Insects
13.
14.
15.
16. 17.
18.
19.
References 1. Sattler, K.D.: Handbook of Nanophysics – Nanomedicine and Nanorobotics. Taylor & Francis/CRC Press, Boca Raton (2010) 2. Fatikow, S., Wich, T., Dahmen, C., Jasper, D., Stolle, C., Eichhorn, V., Hagemann, S., Weigel-Jech, M.: Nanohandling robot cells. In: Sattler, K.D. (ed.) Handbook of Nanophysics – Nanomedicine and Nanorobotics, pp. 47.1–47.31. Taylor & Francis/CRC Press, Boca Raton (2010) 3. Fatikow, S. (ed.): Automated Nanohandling by Microrobots. Springer, London (2008) 4. Fatikow, S., Rembold, U.: Microsystem Technology and Microrobotics. Springer, Berlin/Heidelberg/New York (1997) 5. Fatikow, S.: Mikroroboter und Mikromontage (in german). B.G. Teubner, Stuttgart/Leipzig (2000) 6. Ferreira, A., Mavroidis, C.: Virtual reality and haptics for nanorobotics. IEEE Robot. Autom. Mag. 13, 78–92 (2006) 7. Fatikow, S., Dahmen, C., Wortmann, T., Tunnell, R.: Visual feedback methods for nanohandling automation. Int. J. Inf. Acquis. 6, 159–169 (2009) 8. Sievers, T., Fatikow, S.: Real-time object tracking for the robot-based nanohandling in a scanning electron microscope. J. Micromechatron 3(3–4), 267–284 (2006). Special Issue on Micro/Nanohandling 9. Kratochvil, B.E., Dong, L.X., Nelson, B.J.: Real-time rigidbody visual tracking in a scanning electron microscope. Int. J. Robot. Res. 28, 498–511 (2009) 10. Hutchinson, S., Hager, G.D., Corke, P.I.: A tutorial on visual servo control. IEEE Trans. Robot. Autom. 12, 651–670 (1996) 11. Binnig, G., Quate, C.F., Gerber, C.: Atomic force microscope. Phys. Rev. Lett. 56, 930–933 (1986) 12. Baro´, A.M., Bartolome, A., Vazquez, L., Garcı´a, N., Reifenberger, R., Choi, E., Andres, R.P.: Direct imaging of
20.
21.
22.
˚ -diam Au clusters using scanning tunneling micros13-A copy. Appl. Phys. Lett. 51, 1594–1596 (1987) Tseng, A.A., Notargiacomo, A., Chen, T.P.: Nanofabrication by scanning probe microscope lithography: a review. J. Vac. Sci. Technol. B 23, 877–894 (2005) Brousseau, E., Krohs, F., Caillaud, E., Dimov, S., Gibaru, O., Fatikow, S.: Development of a novel process chain based on atomic force microscopy scratching for small and medium series production of polymer nanostructured components. ASME J. Manuf. Sci. Eng. 132, 030901 (2010) Krohs, F., Onal, C., Sitti, M., Fatikow, S.: Towards automated nanoassembly with the atomic force microscope: a versatile drift compensation procedure. ASME J. Dyn. Syst., Meas, Control 131, 061106 (2009) Iijima, S.: Helical microtubules of graphitic carbon. Nature 354, 56–58 (1991) Dai, H., Hafner, J.H., Rinzler, A.G., Colbert, D.T., Smalley, R.E.: Nanotubes as nanoprobes in scanning probe microscopy. Nature 384, 147–150 (1996) Srivastava, N., Li, H., Kreupl, F., Banerjee, K.: On the applicability of single-walled carbon nanotubes as VLSI interconnects. IEEE Trans. Nanotechnol. 8, 542–559 (2009) Sardan, O., Eichhorn, V., Petersen, D.H., Fatikow, S., Sigmund, O., Bøggild, P.: Rapid prototyping of nanotubebased devices using topology-optimized microgrippers. Nanotechnology 19, 495503 (2008) Eichhorn, V., Fatikow, S., Wich, T., Dahmen, C., Sievers, T., Andersen, K.N., Carlson, K., Bøggild, P.: Depth-detection methods for microgripper based CNT manipulation in a scanning electron microscope. J. Micro-Nano Mechatron. 4, 27–36 (2008) Andersen, K.N., Petersen, D.H., Carlson, K., Mølhave, K., Sardan, O., Horsewell, A., Eichhorn, V., Fatikow, S., Bøggild, P.: Multimodal electrothermal silicon microgrippers for nanotube manipulation. IEEE Trans. Nanotechnol. 8, 76–85 (2009) Castillo, J., Dimaki, M., Svendsen, W.E.: Manipulation of biological samples using micro and nano techniques. Integr. Biol. 1, 30–42 (2009)
Robotic Insects ▶ Insect Flight and Micro Air Vehicles (MAVs)
Rolled-Up Nanostructure ▶ Nanorobotics Nanostructures
for
NEMS
Using
Helical
Rolled-Up Nanotechnology ▶ Electrical Impedance Tomography for Single Cell Imaging
Rose Petal Effect
2265
R
Theoretical
Rose Petal Effect Bharat Bhushan1 and Michael Nosonovsky2 1 Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA 2 Department of Mechanical Engineering, University of Wisconsin-Milwaukee, Milwaukee, WI, USA
Synonyms Petal effect
Definition Rose petal effect is the ability of certain rough surfaces to have a high contact angle with water simultaneously with high adhesion (large contact angle hysteresis) with water.
Recent experimental findings and theoretical analyses made it clear that the early Wenzel [1] and Cassie and Baxter [2] models do not explain the complexity of interactions during wetting of a rough surface which can follow several different scenarios. As a result, there are several modes of wetting of a rough surface, and therefore, wetting cannot be characterized by a single number such as the contact angle (CA). Contact angle hysteresis (the difference between the advancing and receding CA) is another important parameter which characterizes wetting (Fig. 1). Wetting of a solid is dependent upon the adhesion of water molecules to the solid. On the one hand, a high CA is a sign of low liquid–solid adhesion. On the other hand, low CA hysteresis is a sign of low liquid–solid adhesion as well. It is now widely believed that a surface can be superhydrophobic and at the same time strongly adhesive to water [3–9]. The so-called petal effect is exhibited by a surface that has a high CA,
a Air
b Liquid
Air
Liquid
θrec θadv Solid
R
Solid
c
d Liquid
Solid
Air
Air pockets
Rose Petal Effect, Fig. 1 (a) Schematics of a droplet on a tilted substrate showing advancing (yadv) and receding (yrec) contact angles. The difference between these angles constitutes the contact angle hysteresis. Configurations described by (b) the
Liquid
Air
Liquid
Solid
Wenzel equation for the homogeneous interface, (c) Cassie– Baxter equation for the composite interface with air pockets, and (d) the Cassie equation for the homogeneous interface
R
2266
Rose Petal Effect, Fig. 2 (a) Optical images, (b) scanning microscope micrographs, and (c) atomic force microscope roughness maps of petals of two roses (Rosa Hybrid Tea, cv. Bairage) (Rosa, cv. Bairage), and Rosa Hybrid Tea, cv. Showtime (Rosa, cv. Showtime) (Adapted from Bhushan and Her [3])
Rose Petal Effect Roses with superhydrophobic petals with high adhesion
with low adhesion
a
10 mm
10 mm
Rosa Hybrid Tea, cv. Bairage (Fresh)
Rosa Hybrid Tea, cv. showtime (Fresh)
b
20 μm
20 μm
Rosa, cv. Bairage (Dried)
Rosa, cv. Showtime (Dried)
c 7 μm
7 μm
0 μm
0 μm
100 μm
100 μm 100 μm
0 0
Bottom scan
but also a large CA hysteresis and strong adhesion to water.
Main Findings Plant leaves and petals provide an example of surfaces with high CA, and high and low CA hysteresis. Bhushan and Her [3] studied two kinds of superhydrophobic rose petals: (1) Rosa Hybrid Tea, cv. Bairage and (2) Rosa Hybrid Tea, cv. Showtime,
100 μm
0 0
Bottom scan
referred to as Rosa, cv. Bairage and Rosa, cv. Showtime, respectively. Figure 2 shows optical micrographs, scanning electron microscopy (SEM) images, and atomic force microscope (AFM) surface height maps of two rose petals. Figure 3 shows a sessile and a suspending water droplet on Rosa, cv. Bairage, demonstrating that it can simultaneously have high CA, high adhesion, and high CA hysteresis. The surface roughness of the two rose petals was measured with the AFM, and the results for the peakto-base height of bumps, the mid-width, peak radius,
Rose Petal Effect
2267
Rose Petal Effect, Fig. 3 Optical micrographs of water droplets on Rosa, cv. Bairage at 0 and 180 tilt angles. Droplet is still suspended when the petal is turned upside down ([3])
Droplet on Rosa, cv. Bairage
500 μm
500 μm
0° tilt angle
Rose Petal Effect, Table 1 Surface roughness statistics for the two rose petals [3]
Rose Petal Effect, Table 2 Wetting regimes of a surface with a single level of hierarchy of roughness
R
Rosa, cv. Bairage (high adhesion) Rosa, cv. Showtime (low adhesion)
State Cavities CA CA hysteresis
180° tilt angle
Peak-to-base height Mid-width (mm) (mm) 6.8 16.7
Peak radius (mm) 5.8
Bump density (1/10,000 mm2) 23
8.4
4.8
34
15.3
Cassie–Baxter Air High Low
Wenzel Water under droplet High Can be high
Impregnating Cassie Water everywhere High Low
R
Rose Petal Effect, Table 3 Different regimes of wetting of a surface with dual roughness
Air in nanostructure Water under droplet in nanostructure Water impregnating nanostructure
Air in microstructure Lotus, high CA, low CA hysteresis
Water under droplet in microstructure Rose, high CA, high CA hysteresis
Cassie (air-filled microstructure, water in Wenzel (water in micro- and nanostructure), high CA, low CA hysteresis nanostructure), high CA, high or low CA hysteresis Cassie-filled nanostructure Wenzel-filled nanostructure
Water impregnating microstructure Rose-filled microstructure Wenzel-filled microstructure Wenzel-filled micro and nanostructure
R
2268
Rose Petal Effect
Rose Petal Effect, Fig. 4 Schematics of nine wetting scenarios for a surface with hierarchical roughness Lotus
Rose
Rose filled microstructure
Cassie
Wenzel
Wenzel filled microstructure
Cassie filled nanostructure
Wenzel filled nanostructure
Wenzel filled micro/ nanostructure
Hierarchical structures with n-hexatriacontane Microstructures with 23 μm pitch
Rose Petal Effect, Fig. 5 SEM micrographs of the microstructures and nanostructures fabricated with two different masses of n-hexatriacontane for hierarchical structure. All images were taken at 45 tilt angle. All samples are positive replicas, obtained from negative replica with dental wax and Si micropatterned master template (14 mm diameter and 30 mm height) fabricated with epoxy resin coated with n-hexatriacontane [3]
105 μm pitch
100 μm
210 μm pitch
100 μm
100 μm
Low magnification images n-hexatriacontane (0.1 μg/mm2)
and bump density are summarized in Table 1. The data indicates that the low adhesion specimen (Rosa, cv. Showtime) has higher density and height of the bumps, indicating that the penetration of water between the microbumps is less likely. Wetting of a rough surface
n-hexatriacontane (0.2 μg/mm2)
2 μm
2 μm
High magnification images
with a single level of hierarchy of roughness details can follow several scenarios (Table 2). For a hierarchical structure with small bumps on top of larger bumps, a larger number of scenarios is available, and they are summarized in Table 3 and Fig. 4.
Rose Petal Effect
2269
R
Rose Petal Effect, Table 4 CA and CA hysteresis for surfaces with various micro- and nanoroughness (Based on Bhushan and Her [3])
0.1 0.12 0.16 0.2
Pitch 23 mm CA 164 165 166 167
CA hysteresis 3 3 3 3
Water can penetrate either into the micro- or nanostructure, or into both. In addition, the micro- or nanostructure can be impregnated by water or air. The regimes with water penetrating into the microstructure can have high solid–water adhesion and therefore, high CA hysteresis. Bhushan and Her [3] conducted a series of carefully designed experiments to decouple the effects of the micro- and nanostructures. They synthesized microstructured surfaces with pillars out of epoxy resin. The epoxy surfaces were reproduced from model Si templates and were created by a two-step molding process producing a dual replica (first a negative replica and then a positive replica of the original Si template). Surfaces with a pitch (the periodicity of the structure of the pillars) of 23, 105, and 210 mm and with the same diameter (14 mm) and height (30 mm) of the pillars were produced. After that, nanostructures were created on the microstructured sample by self-assembly of the alkane n-hexatriacontane (CH3(CH2)34CH3) deposited by a thermal evaporation method. Alkanes of varying chain lengths are common hydrophobic compounds of plant waxes. On smooth surfaces, alkanes can cause a large contact angle and a small contact angle hysteresis for water droplets. To fabricate the nanostructure, various masses of n-hexatriacontane were coated on a microstructure. The nanostructure is formed by three-dimensional platelets of n-hexatriacontane. Platelets are flat crystals, grown perpendicular to the surface. They are randomly distributed on the surface, and their shapes and sizes show some variation. Figure 5 shows selected images. When different masses of wax are applied, the density of the nanostructure is changed. For surfaces with a small pitch of 23 mm, while the mass of n-hexatriacontane is changed, there are only small changes in the static contact angle and contact angle hysteresis values, which means that they are
105 mm CA 152 153 160 168
Nanostructure mass (μg/mm2)
Mass of n-hexatriacontane (mg/mm2)
CA hysteresis 87 20 5 4
Lotus
0.2
210 mm CA 135 135 150 166
CA hysteresis 45 42 12 3
Rose
0.1 Cassie
0
23
Wenzel
105 Microstructure pitch distance (μm)
210
Rose Petal Effect, Fig. 6 Schematic of a wetting regime map as a function of microstructure pitch and the mass of nanostructure material. The mass of nanostructure material equal to zero corresponds to microstructure only (with the Wenzel and Cassie regimes). Higher mass of the nanostructure material corresponds to higher values of pitch, at which the transition occurs
always in the “Lotus” wetting regime. On the surface with a 210 mm pitch value, as the mass of n-hexatriacontane is increased, the static contact angle is increased, and the reverse trend was found for the contact angle hysteresis. This was interpreted as evidence that the nanostructure is responsible for the CA hysteresis and low adhesion between water and the solid surface. The results are summarized in Table 4. The wetting regimes are shown schematically in Fig. 6 as a function of the pitch of the microstructure and the mass of n-hexatriacontane. Small mass of the nanostructure material correspond to the Cassie and Wenzel regimes, whereas high mass of the nanostructure corresponds to the Lotus and Rose regimes. The Lotus regime is more likely for larger masses of the nanostructure material. Figure 7 shows a droplet on a horizontal surface of a hierarchical structure with 23 and
R
R
2270
Rose Petal Effect, Fig. 7 (a) Droplet on a horizontal surface of hierarchical structure with 23 mm pitch and n-hexatriacontane (0.1 mg/mm2) showing air pocket formation and (b) droplet on a hierarchical structure with 105 mm pitch and n-hexatriacontane (0.1 mg/mm2) and 0.2 mg/mm2 showing no air pocket and air pocket formation, respectively. Also shown is the image taken on the inclined surface with hierarchical structure with 0.1 mg/mm2 showing that droplet is still suspended [3].
Rose Petal Effect
a
Shape of droplets on hierarchical structure with 23 μm pitch Horizontal surface with n-hexatriacontane (0.1 μg/mm2) Regime B1
500 μm
b
Shape of droplets on hierarchical structure with 105 μm pitch Horizontal surface with different mass of n-hexatriacontane n-hexatriacontane (0.1 μg/mm2) Regime A
n-hexatriacontane (0.2 μg/mm2) Regime B2
500 μm
500 μm
Inclined surface with n-hexatriacontane (0.1 μg/mm2) Regime A vertical upside down
500 μm
105 mm pitches and n-hexatriacontane (0.1 mg/mm2). Air pockets are observed in the first case and not observed in the second case, indicating the difference between the two regimes [3]. To further verify the effect of wetting states on the surfaces, evaporation experiments with a droplet on a hierarchical structure coated with two different amounts of n-hexatriacontane were performed. Figure 8 shows the optical micrographs of a droplet
500 μm
evaporating on two different hierarchical structured surfaces. On the n-hexatriacontane (0.1 mg/mm2)coated surface, an air pocket was not visible at the bottom area of the droplet. However, the droplet on the surface has a high static contact angle (152 ) since the droplet still cannot completely impregnate the nanostructure. The footprint size of the droplet on the surface has only small changes from 1820 to 1791 mm. During evaporation, the initial contact area between
Rose Petal Effect
R
2271
Evaporation of a droplet on the surface with 105 μm pitch value and different mass of n-hexatriacontane Hierarchical structure with n-hexatriacontane (0.1 μg/mm2) -- Regime A Radius : 1147 μm
893 μm
729 μm
500 μm
no air pocket
Hierarchical structure with n-hexatriacontane (0.2 μg/mm2) -- Regime B2 Radius : 1106 μm
874 μm
653 μm
381 μm
Transiton 500 μm
air pocket
no air pocket
Rose Petal Effect, Fig. 8 Optical micrographs of droplet evaporation on the hierarchical structured surfaces with 105 mm pitch value. n-Hexatriacontane (0.1 mg/mm2)-coated sample has no air pocket formed between the pillars in the entire contact area until
evaporation was completed. Hierarchical structure with n-hexatriacontane (0.2 mg/mm2) has air pocket, and then the transition from the “Lotus” regime to the “Rose petal” regime occurred [3].
the droplet and hierarchical structured surface does not decrease until the droplet evaporates completely, which means complete wetting between droplet and microstructures. For the n-hexatriacontane (0.2 mg/mm2)coated surface, the light passes below the droplet, and air pockets can be seen, so to start with the droplet is in the Cassie-Baxter regime. When the radius of the droplet decreased to 381 mm, the air pockets are not visible anymore. The footprint size of the droplet on the surface is changed from 1177 to 641 mm, since droplet remained on only a few pillars until the end of the evaporation process [9]. The experimental observations of the two types of rose petals show that hierarchically structured plant surfaces can have both adhesive and non-adhesive properties at the same time with high CA. This is due to the existence of various modes of wetting of a hierarchical surface, so that water can penetrate either into macro- or nanoroughness, or into both.
Water penetration into the microroughness tends to result in high adhesion with the solid surface, whereas the presence of the nanoroughness still provides high CA. As a result, two distinct modes of wetting are observed. One can be called the “Lotus” mode (with low adhesion) and another is the “rose” mode with high adhesion.
Cross-References ▶ Biomimetics ▶ Lotus Effect
References 1. Wenzel, R.N.: Resistance of solid surfaces to wetting by water. Indust. Eng. Chem. 28, 988–994 (1936)
R
R
2272
2. Cassie, A., Baxter, S.: Wettability of porous surfaces. Trans. Faraday Soc. 40, 546–551 (1944) 3. Bhushan, B., Her, E.K.: Fabrication of superhydrophobic surfaces with high and low adhesion inspired from rose petal. Langmuir 26, 8207–8217 (2010) 4. Bormashenko, E., Stein, T., Pogreb, R., Aurbach, D.: “Petal Effect” on surfaces based on lycopodium: high-stick surfaces demonstrating high apparent contact angles. J. Phys. Chem. C 113, 5568–5572 (2009) 5. Chang, F.M., Hong, S.J., Sheng, Y.J., Tsao, H.K.: High contact angle hysteresis of superhydrophobic surfaces: hydrophobic defects. Appl. Phys. Lett. 95, 064102 (2009) 6. Feng, L., Zhang, Y., Xi, J., Zhu, Y., Wang, N., Xia, F., Jiang, L.: Petal effect: a superhydrophobic state with high adhesive force. Langmuir 24, 4114–4119 (2008) 7. Gao, L., McCarthy, T.J.: Teflon is hydrophilic. Comments on definitions of hydrophobic, shear versus tensile
Rotation Sensors hydrophobicity, and wettability characterization. Langmuir 24, 9184–9188 (2008) 8. Jin, M.H., Feng, X.L., Feng, L., Sun, T.L., Zhai, J., Li, T.J., Jiang, L.: Superhydrophobic aligned polystyrene nanotube films with high adhesive force. Adv. Mater. 17, 1977–1981 (2005) 9. Bhushan, B., Nosonovsky, M.: The rose petal effect and the modes of superhydrophobicity. Philos. Trans. Royal. Soc. A. 368, 4713–4728 (2010)
Rotation Sensors ▶ Gyroscopes
S
Sand Skink
Definition
▶ Friction-Reducing Sandfish Skin
A scanning electron microscope (SEM) is an instrument that uses high-energy electrons in a raster-scan pattern to form images, or collect other signals, from the three-dimensional surface of a sample.
SANS ▶ Small-Angle Scattering of Nanostructures and Nanomaterials
SAXS ▶ Small-Angle Scattering of Nanostructures and Nanomaterials
Scanning Electron Microscopy Yimei Zhu1 and Hiromi Inada2 1 Center for Functional Nanomaterials, Brookhaven National Lab, Upton, NY, USA 2 Hitachi High-Technologies America, Pleasanton, CA, USA
Synonyms Scanning electron microscopy and secondary-electron imaging microscopy
Introduction The scanning electron microscope (SEM) is one of the most popular and user-friendly imaging tools that reveal the surface topography of a sample. It is also widely used for structural characterization of materials and devices, especially in the field of nanotechnology. Today there are in excess of 50,000 SEMs worldwide and it is often seen as a “must-have” apparatus for research institutes and industry laboratories. In the SEM, incident electrons interact with the atoms that make up the sample-producing signals that contain information about the sample’s surface morphology, composition, and other physical and chemical properties. The most common imaging mode in SEM lies in using secondary electrons. Since secondary electrons have very low energies, they are generated in, and escape from regions near the sample surface. Combined with various detection systems, a SEM also can be used to determine the sample’s chemical composition through energy-dispersive x-ray spectroscopy and Auger electron spectroscopy, and identify its phases through analyzing electrondiffraction patterns, mostly via high-energy backscattered electrons. Besides backscattered electrons, Auger electrons, characteristic x-rays, other signals are generated from the interactions of the incident
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
S
2274
Scanning Electron Microscopy, Fig. 1 Schematics of configurations of the scanning electron microscopes proposed and developed by some notable pioneers in the SEM history. (a) Knoll in 1935, (b) Zworykin in 1945, and, (c) Cambridge Instruments in 1965
Scanning Electron Microscopy
a
b
Filament
Filament
Facsimile recorder
Condenser lens
Scan coil
Fluorescence screen
PMT
Specimen
Specimen/Scan stage
CRT
c
Filament
Condenser lens
Photograph CRT
Scan coil Objective lens
Display CRT Specimen
beam and the sample under a SEM include plasmons, bremsstrahlung radiation (noncharacteristic x-rays), cathodoluminescence, and electron-beam-induced current. This entry is focused on secondary-electron imaging related instrumentation, signal-generation processes, and state-of-the-art imaging capabilities in SEM.
A Brief History SEM was invented by Max Knoll in 1935 in Germany [1] to study the targets of television tubes. The instrument consisted of electron-beam deflector coils that scan the beam on a plate as the sample in a cathode ray tube (CRT), and an amplifier that boosts the plate current to display the signal on another CRT (Fig. 1a). Two years later Manfred von Ardenne built an electron microscope with a highly demagnified probe using two condenser lenses for scanning transmission electron microscopy and also tried it as an SEM. Zworykin and his coworkers of the RCA Laboratories in the USA designed and built a dedicated SEM in 1942. Its electron optics includes three electrostatic lenses with scan coils placed between the second and third ones. A photomultiplier was first used to detect secondary
SE detector
electrons (Fig. 1b). The essential components of this apparatus are similar to those used in modern SEMs. The probe size of the incident, or primary, electron beam had a diameter of about 10 nm. However, compared with transmission electron microscopes (TEMs) at that period, it could not image secondary electrons satisfactorily due to the poor signal-to-noise ratios of the images [3]. Sir Charles Oatley and Dennis McMullan built their first SEM at Cambridge University in 1948. The SEM technology was further pioneered by many postgraduate students at Cambridge including Gary Stewart. The first commercial instrument, named as “Stereoscan,” was launched in 1965 by the Cambridge Scientific Instrument Company for DuPont [4]. The instrument consists of electron multiplier detector with beryllium–copper dynodes to detect scattered electrons from the specimen surface. Images were displayed on a CRT, while another synchronized CRT recorded them on camera film. An Everhart-Thornley type secondary-electron detector [5] significantly improved the detection efficiency of the low-energy secondaryelectron signals (Fig. 1c). JEOL produced first commercial Japanese SEM, JSM-1, in 1966, while Hitachi commercialized its SEM, HSM-2, in 1969.
Scanning Electron Microscopy
2275
a
Pt / graphite core / shell
Virtual lens
S
d
SE image
Pb on Carbon substrate
C Cs corrector Pt
Coil Yoke
SE detector PMT
scan coil metal cover BSE
Specimen ADF image
b
preamplifier C
SE
Hitachi in-lens design pole piece
Pt 10 kVdc
sample
5 nm SE suppressor Vb 0 V-50 Vdc
c
C
ADF detector
Pt
BF image
BF detector
TE
Scanning Electron Microscopy, Fig. 2 (a–c) Three different objective lens designs. (a) Out-lens, (b) in-lens, and (c) semi-inlens. (d) Schematic of the lens-detector configuration of the aberration-corrected scanning electron microscope, Hitachi HD2700C, that routinely achieves an atomic resolution in imaging using secondary electrons and transmitted electrons. Two examples are shown on the left; one is Pb particles on a carbon
support, the other is Pt particles with carbon-graphite shells. BF bright-field, ADF annular dark-field, SE secondary electrons, TE transmitted electrons, and BES backscattered electrons. The SE images clearly give a topographic view of the area, and higher brightness of the light element C, compared with the corresponding ADF images
Instrumentation
The typical energy range of the electron beam used in SEM is from 0.5 to 40 keV. The electron-condenser lens system usually demagnifies the electron source more than hundreds of times to form a small probe on the sample. The beam passes through pairs of deflection coils, or scanning coils, in the electron column, typically in the final lens, which deflect the beam in the x and y axes by applying an incremental current into the scan coils, so that it scans in a raster fashion over a rectangular area of the sample surface. Historically, SEM incorporates an objective lens. Unlike the objective lens in optical microscopes or transmission electron microscopes (TEMs), its purpose in SEM is not to image the sample, but to focus the small probe on the sample. There are three types of objective lenses: out-lens (Fig. 2a); in-lens (Fig. 2b); and, semi in-lens (Fig. 2c). Most early SEMs had the simplest out-lens design, in which the sample sits beneath the lens leaving a large area available in the sample chamber. However, the yoke gap across the optical axis acts as a lens with leakage field, thus yielding significant imaging aberration. The in-lens pole piece, originally designed for TEMs, was
A typical SEM consists of an electron gun, an electron lens system, various electron-beam deflection coils, electron detectors, and display and recording devices [6]. In a SEM, an electron beam is emitted from an electron gun sitting on a thermionic filament cathode, or from a field-emission needle tip. Tungsten is often used for thermionic electron guns due to its low cost, high melting point, and low vapor pressure so that it tolerates heating to about 2,800 K for electron emission. Other types of electron emitters include lanthanum hexaboride (LaB6) cathodes, which offer higher brightness but require a better vacuum to avoid oxidizing the gun. Field emission guns (FEGs) often used in modern SEMs can be the thermally assisted Schottky type, using emitters of zirconium oxide (ZrO), or the cold-cathode type using tungsten single crystal emitters and operated at room temperature. A cold field-emission gun has a much smaller source size (5 10 nm) than a tungsten filament (1 10 mm) with three to four orders of magnitude larger current density and brightness [7].
S
S
2276
adopted for SEM to reduce spherical aberration. Hitachi developed the first commercial SEM with such a design in 1985 [8]. Their microscope operating at 30 kV reaches a probe size of 0.5 nm with spherical aberration coefficient Cs of 1.6 mm. The drawback of the design is that a conventional thin sample, similar to that used in TEM, is required because the sample sits inside the pole-piece gap. A design resulting from a compromise between the out-lens and in-lens is the semi in-lens pole piece (Fig. 2c), offering reasonable spatial resolution but with larger open space for the sample chamber so that a thick sample can be used. Such a design allows the sample to be placed a few mm from the pole piece, and a large magnetic field to be applied to produce a smaller focal length and less spherical aberration. SEMs with the semi in-lens design played a significant role in characterizing devices for the semiconductor industry in early 1990s. It is noteworthy that the in-lens design shown in Fig. 2b is very similar to that used in conventional TEM and/or STEM (scanning transmission electron microscope). Thus, with an efficient secondaryelectron detector, a TEM, or an STEM also can image a sample surface [9, 10]. Figure 2d shows a schematic of the objective lens (in-lens design), the sample, and the arrangement of detectors in the Hitachi HD2700C aberration-corrected SEM/STEM [11]. The instrument operates at 80 ~ 200 kV, allowing simultaneous acquisition of bright-field (BF), annular dark-field (ADF), and secondary-electron (SE) images. In the BF and ADF modes, the transmitted electrons are used to form the images, thus providing structural information from the sample’s interior. In contrast, in the SE mode electrons emerging from the surface with low energies, or short escape length, are used, and thus, the signals are surface sensitive. Different imaging modes have their own advantages and limitations. Since the BF signals are close to the phase contrast seen in TEM, they offer high spatial resolution, but are difficult to interpret. ADF images, on the other hand, are based on Rutherford scattering, and thus their image intensity is directly related to the sample’s atomic number Z (the so-called Z-contrast imaging). Since BF- and ADF-images are projected images they give little structural information in the direction of the beam’s trajectory. In contrast, SE imaging offers depth information on the surface topography, and is more sensitive to the light elements than is ADF imaging (see Fig. 2d). Combining these different imaging
Scanning Electron Microscopy
modes in an electron microscope with an in-lens design, it has been demonstrated that simultaneously imaging both surface (SE) and bulk (ADF) at atomic resolution is possible in thin samples for a wide range of elements, from uranium and gold to silicon and carbon [11, 12]. The electron detector is another important part of the SEM instrumentation. Although detector itself does not determine the image resolution in SEMs, it is essential to improving the SEM resolving power in terms of the signal-to-noise ratio. The commonest detector used in SEMs today still is that developed by Everhart and Thornley in 1957 [5], the so-called E-T detector. The detector consists of a photomultiplier, a light guide, and a positively biased scintillator. To attract low energy electrons effectively, the scintillator is applied a 10 kV dc bias to accelerate the electrons. The energized electrons cause the scintillator to emit flashes of light (cathodoluminescence) that then are transmitted to the photomultiplier. The amplified output of the electrical signals by the photomultiplier is displayed as a two-dimensional intensity distribution that can be viewed and photographed on an analogue video display. The Everhart-Thornley detector, which normally is positioned to one side above the specimen, exhibits low efficiency in detecting backscattered electrons because few such electrons are emitted in the solid angle subtended by the detector. Furthermore, its positive bias cannot readily attract the high-energy backscattered electrons (close to the energy of the incident electrons). Backscattered electrons are usually collected above the sample in a “doughnut type” arrangement, concentric with the electron beam, to maximize the solid angle of collection.
Secondary-Electron Signal Generation Secondary-electron (SE) imaging is the most frequently used mode of imaging in SEM. Secondary electrons, defined as the electrons with energy below 50 eV, are generated along the primary electrons’ trajectories within the sample, but are subject to elastic and inelastic scattering during their passage through the sample. These electrons can be valence electrons or are ejected from the orbits of the inner shells (most likely the k-shells) of the sample. The consequence of their low kinetic energy is their shallow escape depth, which is about 1 nm for metals, and up to 10 nm for
Scanning Electron Microscopy
2277
S
Scanning Electron Microscopy, Fig. 3 SEM micrographs of a polymer membrane that consists of electrospun fibrous scaffold for water filtration. The fibrous scaffold has an average diameter of 200 nm. The high-resolution SEM image on the right shows cellulose nanofibers (5–10 nm in diameter) infused into the scaffold form a cellulose network to enhance the
membrane’s ability to remove bacteria and viruses. The images were taken with JEOL7600 SEM at operation voltage of 0.5 eV. A thin layer of carbon was coated on the sample to avoid charging. Note the bright contrast at the edge of the fibers generates a topological view of the fibrous structure
insulators. The probability of escape decreases continuously with the increase of the depth below the surface. However, there is a nonzero probability of secondaryelectron emission arising from inelastic scattering below the escape depth, as, for example, when a primary electron creates a fast secondary electron (energy > 50 eV) that travels toward the surface and generates a lower-energy secondary electron within the escape depth. The production of the secondary electron signals involves the generation, propagation, escape from the surface, and arrival at the detector. These four processes are detailed in the atomic-imaging section. The secondary electrons discussed here are often referred to as SEI, i.e., the secondary electrons generated by the incident beam upon entering the sample. Secondary electrons generated by backscattered electrons when leaving the sample are termed SEII. Secondary electrons generated when the backscattered electrons strike a lens pole piece or the sample chamber’s wall, and by primary electrons hitting the aperture are, respectively called SEIII and SEIV. Although SEIII contain the information on the sample, SEIV do not. Furthermore, there are fast secondary electrons, with energy higher than 50 eV. Bias experiments, wherein a positive dc voltage is applied to the sample to suppress the emission of the secondary electrons, were mainly designed to separate the SEI from backscattered electrons; they cannot distinguish SEI from SEII, which for a very thin specimen should be negligible. Measuring other types of secondary electrons, including those with high energy, would require a different bias experiment. Heavy elements (high
atomic number) that backscatter electrons more strongly than do light elements (low atomic number), and thus appear brighter in the image, can be used to yield contrast that contains information of a sample’s chemical composition. Figure 3 shows SEM images of a nanofibercontaining polymer membrane developed for water filtration with enhanced ability to remove bacteria and viruses. The image on the right shows the structural network formed by the 5–10 nm diameter cellulose nanofibers. It reveals astonishing topographic details on how the cellulose nanofibers are interwoven with scaffold. In SEM images flat surfaces give even contrast while curved surfaces and sharp edges often appear brighter (high image intensity). This is due to the increased escape of secondary electrons from the top and side surfaces when the interaction volume intercepts them. Since the secondary electron detector is biased with a high acceleration voltage, a surface facing away from the detector still can be imaged. Surfaces tilted away from the normal to the beam allow more secondary electrons to escape [13].
Atomic Imaging Using Secondary Electrons In the last decade or so, high-resolution SEM has proven an indispensable critical dimension metrology tool for the semiconductor industry. The roadmap for semiconductor nanotechnology identifies the need for ultra-high-resolution SEM in the quest for ever-decreasing device sizes. In a SEM, the size of
S
S
2278
Scanning Electron Microscopy
Scanning Electron Microscopy, Fig. 4 (a–b) Simultaneous atomic imaging using secondary electrons (secondary-electron mode, a) and transmitted electrons (annual dark-field mode, b) of uranium individual atoms on a carbon support (raw data). The circles mark the single uranium atoms. The atoms shown in (b) but not in (a) are presumably those on the back side of the support. (c–d) Simultaneous atomic imaging using secondary electrons (c) and transmitted electrons (d) of YBa2Cu3O7 superconductor viewing along the [010] direction (raw data). The scale bar in (c) is 0.7 nm
the imaging probe often determines the instrument’s resolution power. The probe size d (measured in fullwidth and half maximum) is a function of the beam convergence half-angle a is an incoherent sum of contributions from source size, diffraction limit, spherical aberration, and chromatic aberration, and is given by 4ip 2 0:6l 2 d ¼ þ þ ð0:5Cs a3 Þ2 a bp2 a2 DE 2 þ Cc a E
2
where ip is the probe current, a is the convergence half-angle, b is the source brightness, l is the electron wavelength at beam energy E, DE is the energy spread, and Cs and Cc are, respectively, the spherical- and chromatic-aberration coefficients of the probe-forming lens. The first term on the right side of the equation is the probe size defined by the source size, the second term is due to the diffraction limit, and the third and the fourth terms are due, respectively, to the spherical and chromatic aberrations. Recent advancement on correcting spherical aberration in SEM and STEM diminishes Cs to zero, thus eliminating the third term and produces a small probe. It is important to note that the probe size also depends on the energy spread in the fourth term that includes the energy distribution of the primary electron beam and the fluctuation of the instruments’ acceleration voltage.
In SEM, the image resolution depends not only on the instrument, but also on the sample, or, more accurately, on the sampling volume of the sample from which the signal is generated When the incident electrons impinge on a point of the sample’s surface, they interact with atoms in the sample via elastic- (change trajectory) and inelastic- (lose energy) interactions. The size of the interaction volume depends on the electron’s landing energy, the atomic number of the sample, and its density. The simultaneous energy loss and change in trajectory spreads beam into the bulk of the sample and produces an interaction volume therein, that, for a thick sample, can extend from less than 100 nm to around 5 mm into the surface, i.e., more than an order of magnitude larger than the original probe size. For a very thin sample (a few nm thick), the interaction volume, as the first approximation, might be defined as the probe size. Figure 4 illustrates the atomic resolution images using secondary electrons on single uranium atoms (a) and the (010) surface of a YBa2Cu3O7 crystal (c) recorded on the Hitachi HD2700C SEM/STEM (Fig. 2d). For comparison, the corresponding annular dark-field images using transmitted electrons, (b) and (d), respectively, are included. The equally sharp images in SE and ADF suggest negligible imaging delocalization. Such an attainable resolution was attributed to the combination of several factors: Better design of the electro-optics of the instrument (including ultrahigh electric and mechanical stabilities) and
Scanning Electron Microscopy
the detector; aberration correction that reduces the probe size and increases the probe current; and, the higher operation voltage that beneficially assures a very small volume of beam interaction for a thin sample. The physical mechanism of producing the low-energy secondary electrons traditionally is ascribed to inelastic scattering and decay of collective electron excitation with the incident electron giving up, say 20 eV, to produce a secondary electron with energy 20 eV minus the work function of the surface. Since the momentum transfer (scattering angle) of the scattering with 20 eV energy loss is small, the transfer should be delocalized to an area >1 nm; thus atomic imaging using secondary electrons was not considered possible. Recent studies suggest that this is not the primary mechanism for secondary-electron imaging at least on a thin sample [11]. The secondary electrons responsible for atomic-scale resolution are generated by inelastic scattering events with large momentum transfer, including those from inner shell orbitals, which give rise to a sharp central peak in the pointspread function for signal generation. In general, four steps are involved in producing the signal that is used to form a secondary electron (SEI) image [12]. (1) The generation of secondary electrons through the inelastic scattering of primary electrons in the sample, at a generation rate, G; (2) random motion of these secondary electrons, which are scattered by atoms of the specimen both elastically and inelastically (potentially creating other secondary electrons of lower energies), such that, on average, T electrons reach the sample surface for each secondary electron generated; (3) the escape of secondary electrons over the potential barrier at the sample surface of the specimen, with an average probability P; and, (4) the acceleration of the emitted electrons in vacuum, such that a fraction D reaches the electron detector. The secondary-electron signal S is a product of these four factors: S ¼ G T P D. To generate contrast in a scanned-probe image, one or more of the above steps must depend on the x-coordinate of the electron probe in the scan direction, i.e., dS dG dT dP dD ¼ I0 TPD þ I0 GPD þ I0 GTD þ I0 GTP dx dx dx dx dx For most non-atomically resolved SE images obtained in an SEM, dT dx provides the main contrast
2279
S
mechanism: Secondary electrons created at an inclined surface or close to a surface step have an increased probability of escape, resulting in surface-topography contrast [14]. Less commonly, variations in surface work function contribute additional contrast by providing a nonzero dP dx . In voltage-contrast applications, changes in surface voltage provide a nonzero dD dx . Atomic-number contrast is possible if the specimen is chemically inhomogeneous and G varies with atomic number, yielding a nonzero dG dx . However, for atomic imaging in a thin sample (Fig. 4), the dominant mechanism can be quite different. dG dx is likely to play an important role in the atomic-scale contrast as a consequence of Z-dependence inelastic scattering cross section and channeling effect for crystals (for thickness in the order of extinction distance it should be minor). Because secondary electrons are generated through inelastic scattering of the incident electrons, dG dx is limited by the delocalization of the scattering process, which may be described by the point-spread function for inelastic scattering. The term dT dx should be small, for adatoms or surface atoms that lie on the detector side of the sample. dP dx would not become important unless the effective work function varies on an atomic scale, and dD dx must be also negligible at atomic scale. These assumptions are reasonable because the scattering process disperses secondary electrons over a range of x that is comparable to the escape depth, typically 1–2 nm. Consequently, T, P, and D are x-averages that vary little with x on an atomic scale. For uranium atoms on a carbon substrate, the argument is even simpler; these atoms lie outside the solid, so the terms T and P are not applicable.
Future Remarks Secondary-electron imaging is the most popular mode of operation of the scanning electron microscope (SEM) and traditionally is used to reveal surface topography. Nevertheless, this imaging method never was regarded as being on the cutting edge of performance, due to its perceived limited spatial resolution in comparison with its TEM or STEM counterparts using transmitted electrons. Recent work using aberrationcorrected electron microscopes demonstrated that secondary electron signals in the SEM can resolve both crystal lattices and individual atoms, showing SEM’s unprecedented and previously unrealized
S
S
2280
Scanning Electron Microscopy and Secondary-Electron Imaging Microscopy
imaging capabilities. Furthermore, the work demonstrates the incompleteness of present understanding of the formation of secondary-electron images. Secondary-electron imaging using high acceleration voltage with thin samples can now compete with TEM on spatial resolution, and provide new capabilities, such as depth-resolved profiles, at the atomic level. There seems to be no fundamental reason why atomic resolution in secondary-electron imaging could not be achieved at the accelerating voltages of 0.5–40 keV that are currently used in conventional SEMs. It remains to be seen whether the integrated spherical and chromatic aberration of a probe-forming objective lens can be corrected to a sufficient degree at low operation voltages. One clear outcome thus far is the importance of preparing samples with clean surfaces in order to obtain interpretable and reproducible results. Acknowledgments The work was supported by the US Department of Energy, Office of Basic Energy Science, Materials Science and Engineering Division, under Contract Number DEAC02–98CH10886.
9. 10. 11.
12.
13. 14.
11th International Congress On Electron Microscopy, Kyoto, pp. 2097–2100 (1986) Liu, J., Cowley, J.M.: High resolution SEM in a STEM instrument. Scanning Microsc. 2, 65–81 (1988) Howie, A.: Recent developments in secondary electron imaging. J. Microsc. 180, 192–203 (1995) Zhu, Y., Inada, H., Nakamura, K., Wall, J.: Imaging single atoms using secondary electrons with an aberrationcorrected electron microscope. Nat. Mater. 8, 808–812 (2009) Inada, H., Su, D., Egerton, R.F., Konno, M., Wu, L., Ciston, J., Wall, J., Zhu, Y.: Atomic imaging using secondary electrons in a scanning transmission electron microscope: experimental observations and possible mechanisms. Ultramicroscopy 111(7), 865–876 (2011). doi:10.1016/j.ultramic.2010.10.002. Invited articles for the special issue in honor of John Spence Joy, D.C.: Beam interactions, contrast, and resolution in the SEM. J. Microsc. 136, 241–258 (1984) Reimer, L.: Scanning Electron Microscopy, 2nd edn. Springer, New York (1998)
Scanning Electron Microscopy and Secondary-Electron Imaging Microscopy ▶ Scanning Electron Microscopy
Cross-References ▶ Robot-Based Automation on the Nanoscale ▶ Scanning Tunneling Microscopy ▶ Transmission Electron Microscopy
Scanning Force Microscopy in Liquids ▶ AFM in Liquids
References 1. Knoll, M.: Aufladepotentiel und sekund€aremission elektronenbestrahlter ko¨rper. Z. Tech. Phys. 16, 467–475 (1935) 2. von Ardenne, M.: Das Elektronen-Rastermikroskop. Praktische Ausf€ uhrung. Z. Tech. Phys. 19, 407–416 (1938) (in German) 3. Wells, O.C., Joy, D.C.: The early history and future of the SEM. Surf. Interface Anal. 38, 1738–1742 (2006) 4. Oatley, C.W.: The early history of the scanning electron microscope. J. Appl. Phys. 53, R1–R13 (1982) 5. Everhart, T.E., Thornley, R.F.M.: Wide-band detector for micro-microampere low-energy electron currents. J. Sci. Instrum. 37, 246–248 (1960) 6. Goldstein, G.I., Newbury, D.E., Echlin, P., Joy, D.C., Fiori, C., Lifshin, E.: Scanning Electron Microscopy and x-ray Microanalysis. Plenum, New York (1981) 7. Pawley, J.: The development of field-emission scanning electron microscopy for imaging biological surfaces. Scanning 19, 324–336 (1997) 8. Tanaka, K., Mitsushima, A., Kashima, Y., Osatake, H.: A new high resolution scanning electron microscope and its application to biological materials. In: Proceedings of the
Scanning Kelvin Probe Force Microscopy ▶ Kelvin Probe Force Microscopy
Scanning Near-Field Optical Microscopy Achim Hartschuh Department Chemie and CeNS, Ludwig-Maximilians-Universit€at M€unchen, Munich, Germany
Synonyms Near-field scanning optical microscopy (NSOM)
Scanning Near-Field Optical Microscopy
Definition Scanning near-field optical microscopy (SNOM) is a microscopic technique for nanostructure investigation that achieves sub-wavelength spatial resolution by exploiting short-ranged interactions between a sharply pointed probe and the sample mediated by evanescent waves. In general, the resolution of SNOM is determined by the lateral probe dimensions and the probesample distance. Images are obtained by raster-scanning the probe with respect to the sample surface corresponding to other scanning-probe techniques. As in conventional optical microscopy, the contrast mechanism can be combined with a broad range of spectroscopic techniques to study different sample properties, such as chemical structure and composition, local stress, electromagnetic field distributions, and the dynamics of excited states.
Introduction Optical microscopy forms the basis of most of the natural sciences. In particular, life sciences have benefited from the fascinating possibility to study smallest structures and processes in living cells and tissue. Optical techniques feature extremely high detection sensitivity reaching single molecule sensitivity in fluorescence, Raman scattering and absorption spectroscopy. Besides the direct visualization, chemically specific information is obtained through Raman spectroscopy. The resolution of conventional optical microscopes, however, is limited by diffraction, a consequence of the wave nature of light, to about half the wavelength. Concepts extending optical microscopy down to nanometer length scales below the diffraction-limit are distinguished into far-field and near-field techniques. Far-field techniques rely on the detection of propagating waves at distances from the source larger than the wavelength, while near-field techniques exploit short ranged evanescent waves. Scanning near-field optical microscopy, initiated by pioneering work of Pohl, Lewis and others in the late 1980s and the beginning of 1990s gave access to nanoscale resolution for the first time. The history of near-field optics is reviewed in [1]. In addition numerous review articles and books exist describing fundamentals and applications (see e.g., [2–4]).
2281
S
This entry introduces first the key physical principles beginning with the role of evanescent and propagating waves, and the loss of spatial information upon light propagation. Concepts of near-field detection are shown using different pointed probes. The next sections describe the experimental realization and present several applications of SNOM. The outlook addresses future prospects of SNOM and remaining challenges.
Key Principles and Concepts Near-field optics has its origin in the effort of overcoming the diffraction limit of optical imaging. The physical origin of this limit is sketched in the following starting with the distinction between propagating waves that form the optical far-field of a radiation source and their evanescent counterpart that dominate the optical near-field. A powerful tool to describe wave propagation is the so-called angular spectrum representation of fields expressing the electromagnetic field E in the detector plane at z as the superposition of harmonic plane waves of the form ~r iotÞ with amplitudes Eðk x ; ky ; z ¼ 0Þ emaexp ðik~ nating from the source plane at z ¼ 0 [2]. Z Zþ1 Eðx; y; zÞ ¼
x ; ky ; z ¼ 0Þeiðkx xþky yÞ eikz z dkx dky Eðk
1
(1) The wave vector ~ k describing the propagation direction of the wave is represented by its components k~ ¼ ðkx ; ky ; kz Þ while its length is fixed by the wavelength of light l and the refractive index of the medium qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n through k~ ¼ k2 þ k2 þ k2 ¼ 2pn=l. In Eq. 1, the x
y
z
time dependence of the fields has been omitted for clarity. For simplicity the following discussion is limited to the x-z-plane and n ¼ 1 such that qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi kz ¼ 4p2 =l2 kx2 . In Eq. 1 the term eikz z controls the propagation of the associated wave: For kx 2p/l the component kz is real and the corresponding wave x ; z ¼ 0Þ propagates along the with amplitude Eðk z-axis oscillating with eikz z . If kx > 2p/l the component kz becomes complex and ejkz zj describes an exponential decay of the associated wave that is therefore evanescent. As a result, only waves with kx 2p/l can
S
S
2282
Scanning Near-Field Optical Microscopy
Far-field
Spatial Propagation broadening
Propagating part of the kx spectrum kx,max = 2π/λ
z = 90 nm *exp(ik z) z
z = 30 nm
|E(kx, z)|
E(x,z) Near-field
Evanescent kz imaginary
10 nm z x Sample
x Real space
z = 0 nm
−0.5
Fourier-transformation
Evanescent kz imaginary
0
0.5 kx k-space
Scanning Near-Field Optical Microscopy, Fig. 1 Scheme illustrating the propagation of waves and the loss of spatial information. Initial field distribution E(x, z ¼ 0) at a 10-nm wide source in the x-z-plane (center) and corresponding angular x ; z ¼ 0Þj (right). Near the source the spectrum spectrum jEðk contains both evanescent and propagating waves. Upper panels
illustrate the evolution of the fields at z ¼ 30 nm and z ¼ 90 nm distance for a source wavelength of l ¼ 500 nm in vacuum. Only waves with kx 2p/l propagate. Evanescent waves decay exponentially following ejkz jz . The decay of high spatial frequencies leads to spatial broadening and loss of spatial information in the far-field
propagate and contribute to the field far from the source forming the far-field. Figure 1 schematically illustrates this behavior: In the center, the electric field E(x, z) emanating from a narrow sub-wavelength source at z ¼ 0 is shown together with its angular x ; z ¼ 0Þ calculated by the inverse of spectrum Eðk Eq. 1. The wave amplitudes E result from the Fourier-transformation of E with respect to the spatial coordinate x. As for the correlation between time and frequency domain, where a short optical pulse requires a broad frequency spectrum, a sharp field distribution requires a broad spectrum of spatial frequencies kx. Since only waves with limited spatial frequencies can propagate, the spectral width rapidly decreases with increasing distance from the source z leading to fast broadening of the electric field distribution in real space. In other words, propagation corresponds to low-pass filtering with frequency limit kx, max ¼ 2p/l. The far-field thus contains limited spatial frequencies equivalent to limited spatial information. To overcome this limitation different near-field concepts have been developed that are outlined in the following. The key concept of SNOM is the probing of the sample near-field that contains the evanescent waves using a sharply pointed probe. Since evanescent waves decay rapidly for increasing distance to the source, the probe needs to be in close proximity to the sample. Waves with large kx components that carry high spatial information decay most rapidly following ejkz jz as can
be seen from Fig. 1. Hence the spatial resolution obtained in an SNOM experiment drops fast with increasing tip-sample distance z. As for other scanning probe techniques that exploit short-ranged interactions, such as atomic force and scanning tunneling microscopy, AFM and STM, respectively, the lateral resolution is also determined by the lateral dimension of the probe. Two conceptually different types of probes can be distinguished: The first confines and samples electromagnetic fields using an aperture with subwavelength diameter (Fig. 2a). The second exploits the antenna concept that couples locally enhanced near-fields to propagating waves and vice versa (Fig. 2b–d). The two types, termed aperture and antenna probe, respectively, are illustrated in the following. Aperture Probes Aperture probes confine light by squeezing it through a sub-wavelength hole (Fig. 2a). This approach, termed aperture-SNOM, provides an enormous flexibility regarding signal formation. Different operation modes can be used that are capable of local sample excitation and/or local light collection. Depending on which step of the experiment exploits near-field interactions to obtain sub-wavelength resolution, apertureSNOM can be implemented in excitation ①, collection ② and excitation-collection ③ mode (Fig. 2a). The original scheme was proposed by Synge in 1928. He suggested to use a strong light source behind
Scanning Near-Field Optical Microscopy
a
2283
b 1
c
S
d
3
d
d
d
d
2
Scanning Near-Field Optical Microscopy, Fig. 2 Schematics of the most common SNOM probes and configurations: (a) Aperture probes are utilized in different modes: Near-field excitation – far-field collection ①, far-field excitation – near-field collection ② and near-field excitation – collection mode ③. (b) Tip-enhanced near-field optical microscopy (TENOM) using local field enhancement at a sharp metal antenna probe
upon far-field excitation. (c) Tip-on-aperture (TOA) probe using local field enhancement at an antenna probe that is excited in the near-field of an aperture probe. (d) Scattering-SNOM (s-SNOM) based on far-field illumination and detection of local scattering at a sharp antenna probe. The label d indicates the structural parameter that determines the achievable spatial resolution
a thin, opaque metal film with a 100-nm diameter hole in it as a very small light source. In 1984, two groups adopted this scheme and presented the first experimental realizations in the optical regime [1]. The aperture was formed at the apex of a sharply pointed transparent probe tip coated with metal. Raster-scanning the probe was made possible by the scanning technology developed in the context of scanning tunneling microscopy (STM). In Fig. 3a, a scanning electron microscopy (SEM) image of an aperture probe consisting of a metal-coated tapered glass fiber is shown. At the front surface a well-defined aperture with diameter of 70 nm is seen. Analytical expressions for the electric field distribution in a sub-wavelength aperture in a metallic screen were already presented in 1944 and 1950 by Bethe and Bouwkamp. Numerical simulations for metal-coated tapered fiber probes show that strong fields pointing in axial direction occur at the rim due to local fieldenhancement by the metal coating (see Fig. 4). The center of the aperture is dominated by a weaker horizontal component [2]. The optical field distribution can be determined experimentally by raster-scanning single fluorescent molecules that act as point-like dipoles across the aperture while recording the fluorescence intensity [5] (see section “Fluorescence Microscopy” and Fig. 3). Tapered aperture probes suffer from low light transmission due to the cut-off of propagating wave-guide modes. For probe diameters below the cut-off diameter only evanescent waves remain and the intensity decays
exponentially toward the aperture (Fig. 2c, d). Probe designs, therefore, aim at maximizing the cone angle that determines the distance between aperture and cut-off diameter. Hollow-cantilever probes feature relatively large cone-angles as compared to fiber-based probes (Fig. 2b). On the other hand, the input power needs to be limited because of the damage threshold of the metal coating in case of fiber-based probes. Due to the limited transmission and the skin-depth of the optical fields on the order of several tens of nanometers, most aperture-SNOM measurements are carried out with apertures of 50–100 nm. Antenna Probes Antenna probes act as transmitter and receiver coupling locally enhanced near-fields to propagating waves and vice versa (Fig. 2b–d) [8]. To distinguish this approach from the earlier implementations based on apertures, it is also termed apertureless-SNOM or aSNOM. Antenna probes can be used in two different techniques: (1) Scattering type microscopy [9, 10], also termed scattering-SNOM or s-SNOM, in which the tip locally perturbs the fields near a sample surface. The response to this perturbation is detected in the far-field at the frequency of the incident light corresponding to elastic scattering (Fig. 2d). (2) Tipenhanced near-field optical microscopy (TENOM) in which locally enhanced fields at laser-illuminated metal structures are used to increase the spectroscopic response of the system at frequencies different from
S
S
2284
Scanning Near-Field Optical Microscopy
a
c aluminum d = 250 nm
d = 160 nm
glass
a
evanescent decay
HE11
b
d
10−2
Pin/Pout
10−4 10−6 10−8 10−10
8μm
100 nm 20 nm aperture 10 nm
10−12 20
60
100
140
α[°]
Scanning Near-Field Optical Microscopy, Fig. 3 (a) Scanning-electron microscopy (SEM) image of an aperture probe formed by a metal-coated tapered fiber with an aperture diameter of 70 nm (scale bar 200 nm) (Reprinted with permission from Veerman et al. [5]. Copyright 1999, John Wiley and Sons) (b) SEM image of a metallic hollow aperture probe microfabricated on a Si cantilever. The aperture diameter is about 130 nm (Reprinted with permission from Mihalcea et al. [6]. Copyright 1996, American Institute of Physics) (c) Schematic of the mode
propagation in a tapered aperture probe. For probe diameters below the cut-off diameter, here d ¼ 160 nm, the intensity decays exponentially toward the aperture. (d) Transmission of tapered probes determined as the ratio of input versus output power Pin/Pout as a function of the cone angle a defined in (c). For smaller cone angles, the distance between cut-off and aperture increases leading to extremely low transmission (c and d) (Reprinted with permission from Hecht et al. [7]. Copyright 2000, American Institute of Physics)
that of the incident light [8, 11] (Fig. 2b, c). The flexibility of this technique allows the study of a variety of spectroscopic signals including Raman scattering (tip-enhanced Raman spectroscopy (TERS)), and fluorescence as well as time-resolved measurements. In the following, the signal formation in the case of optical antennas is sketched. Elastic scattering signal. The near-field interaction can be treated within a simplified model in which the tip is replaced by a polarizable sphere. Due to the antenna properties of the tip, laser excitation with
incident fiel Ei creates a dominating dipole oriented along the tip axis in z-direction normal to the sample surface. This dipole induces a mirror dipole in the sample depending on its dielectric properties. The mirror dipole’s field, decreasing with the third power of distance, interacts with the tip dipole. Solving the system of electrostatic equations that describes the multiple interaction between tip and mirror dipoles neglecting retardation yields an effective polarizability of the coupled tip-sample system which fully expresses the influence of the sample.
Scanning Near-Field Optical Microscopy x-,ypositioning
2285
S
at higher harmonics of the tapping mode frequency (see section “Instrumentation”). Besides the amplitude of the scattered field, its phase can be retrieved using interferometric heterodyne detection [9]. Raman scattering signal. In the case of Raman scattering, the total signal depends on the product of the excitation and emission rates kex(lex) krad(lrad). As a consequence, the total signal enhancement scales with the fourth power of the field enhancement for small differences between the excitation lex and emission wavelength lrad and assuming that the field enhancement at the tip does not depend sensitively on the wavelength.
Laser (a)
z-feedback Sample x-, y-scanner Objective
Beam splitter
Laser (b)
MRaman ¼
Flipmirror
APD
Spectrograph CCD
Scanning Near-Field Optical Microscopy, Fig. 4 Schematic of an experimental setup applicable to aperture-probe SNOM (laser (a), excitation mode) and TENOM (laser (b)) in case of transparent samples. In the case of TENOM the fiber probe would be replaced by an optical antenna, for example, an etched metal wire. The probe tip is positioned in the focus of the microscope objective by piezo-electric actuators. The sample is raster-scanned using a closed-loop x-y-scanner while both laser and probe position remain fixed. The tip-sample distance is controlled using a tuning-fork shear-force feedback scheme. The optical signal is collected by the objective and detected either by a highly sensitive avalanche photodiode (APD) or energetically resolved using a spectrograph and a CCD
aeff ¼
! ða ð1 þ bÞ = 1 ðabÞ ð16p ða þ zÞ3 Þ
(2)
The scattered field can then be calculated from Es / aeff Ei reflecting the short-ranged interaction required for sub-diffraction resolution. Since the laser illuminates a greater part of the tip and the sample, elimination of background-scattering contributions is crucial. Efficient background suppression can be achieved by demodulating the detected intensity signal
ex rad ktip ktip f4 k0ex k0rad
(3)
The factor f measures the ratio between tipenhanced Etip and non-enhanced electric field E0 in the absence of the tip. For the general case of surface-enhanced Raman scattering (SERS), Raman enhancement factors are reported reaching up to 12 orders of magnitude for particular multiple particle configurations involving interstitial sites between particles or outside sharp surface protrusions. Since the signal scales with the fourth power, already moderate field enhancement, predicted for a single spherical particle to be in the range of f ¼ 10–100 is sufficient for substantial signal enhancement. Fluorescence signal. The fluorescence intensity depends on the excitation rate kex and the quantum yield denoting the fraction of transitions from excited state to ground state that give rise to an emitted photon. The quantum yield is expressed in terms of the radiative rate krad and the non-radiative rate knonrad through ¼ krad/(krad + knonrad). Accordingly, the fluorescence enhancement due to the presence of the metal tip can be written as MFlu
2 tip tip Etip 2 ¼ ¼f : E0 0 0
(4)
Here, it is assumed that the system is excited far from saturation. From Eq. 4 it is clear that TENOM works most efficiently for samples with small fluorescence quantum yield 0 such as semiconducting singlewalled carbon nanotubes [11]. For highly fluorescent samples such as dye molecules, the quantum yield 0 is already close to unity and cannot be enhanced further.
S
S
2286
Because of the small separation between emitter and metal tip required for high spatial resolution, non-radiative transfer of energy from the electronically excited state to the metal followed by non-radiative dissipation in the metal has to be taken into account. This process represents an additional competing nonradiative relaxation channel and reduces the number of detected fluorescence photons. Metal-induced fluorescence quenching can also be exploited for image contrast formation (see e.g., [12]). In this case MFlu in Eq. 4 becomes smaller than unity. While the theory of energy transfer between molecules and flat metal interfaces is well understood in the framework of phenomenological classical theory, nanometer-sized objects are more difficult to quantify [2]. Tip- and particle-induced radiative rate enhancement and quenching has been studied in literature both experimentally and theoretically. Experiments on model systems formed by single dipole emitters such as molecules and semiconductor nanocrystals and spherical metal particles revealed a distancedependent interplay between competing enhancement and quenching processes. While semiconducting tips cause less efficient quenching, they also provided weaker enhancement because of their lower conductivity at optical frequencies. Polarization and angular-resolved detection of the fluorescence signal of single emitters demonstrated that the fluorescence rate enhancement provided by the optical antenna also results in a spatial redistribution of the emission [13]. The same redistribution can be expected to occur for tip-enhanced Raman scattering. The spatial distribution of the enhanced electric field follows approximately the outer dimensions of the tip apex. Since the signal enhancement scales with higher orders of field enhancement, the optical resolution can surpass the size of the tip [11]. Stronger fields and field confinement are observed for so-called gap modes formed by metal tips on top of metal substrates. The scheme depicted in Fig. 2b shows that in addition to the signal resulting from the near-field tip–sample interaction the confocal far-field signal contribution is detected, representing a background. This background originates from a diffraction-limited sample volume that is far larger than the volume probed in the near-field. The near-field signal has to compete with this background, and strong enhancement is required to obtain clear image contrast. This
Scanning Near-Field Optical Microscopy
requirement is relaxed in case of low-dimensional sample structures such as spatially isolated molecules or one-dimensional nanostructures [11]. The near-field signal to background ratio can be improved by exploiting the non-linear optical response of sample and tip. Examples include two-photon excitation of fluorescence using a metal tip antenna and the application of the four-wave mixing signal of a metal particle dimer as local excitation source.
Instrumentation Near-field optical microscopy exploits short-ranged near-field interactions between sample and probe. SNOM instruments thus require a mechanism for tipsample distance control working on the scale of nanometers. Typical implementations utilize other nonoptical short-ranged probe–sample interactions such as force or tunneling current used for topography measurements in AFM and STM, respectively. During image acquisition by raster-scanning the sample with respect to the tip, optical and topographic data are thus obtained simultaneously. Due to the strong tip-sample distance dependence of the near-field signal, cross-talk from topographic variations is possible that can lead to artifacts in the optical contrast. Tip-sample distance curves need to be measured to prove unequivocally the near-field origin of the observed image contrast. Since the optical signal results from a single sample spot only, SNOM instruments are often based on a confocal laser scanning optical microscope equipped with sensitive photodetectors. Scattering-SNOM is often implemented with intermittent-contact-mode AFM in which the tipsample distance is modulated sinusoidally at frequencies typically in the range of 10–500 kHz. The elastically scattered laser light intensity is demodulated by the tapping-mode frequency using lock-in detection for background suppression. Since Raman and fluorescence signals are typically weak requiring longer acquisition times, signal demodulation is more challenging. In the case of single-photon counting time-tagging can be used to retrieve the signal phase with respect to the tapping oscillation. While this can be applied to single-color experiments, spectrum acquisition using CCD cameras in combination with spectrometers is not feasible at typical tapping frequencies.
Scanning Near-Field Optical Microscopy
In most aperture-SNOM and TENOM experiments, the tip-sample distance is kept constant by either using STM, contact/non-contact AFM, or shear-force feedback. Sensitive piezoelectric tuning-fork detection schemes operating at small interaction forces have been developed and are used for fragile fiber and antenna probes [2]. Probe Fabrication The near-field probe forms the crucial part of an SNOM setup since both optical and topographic signals are determined by its short-ranged interactions with the sample. Instrument development, therefore, focuses mainly on the design of optimized tip concepts and tip geometries as well as on fabrication procedures for sharp and well-defined probes with high reproducibility. These continuing efforts benefit substantially from improving capabilities regarding nanostructuring and nanocharacterization, using, for example, focused ion beam milling (FIB) or electron beam lithography (EBL). Aperture Probes Optical fiber probes. Sharply pointed optical fiber probes were the first to provide sub-diffraction spatial resolution and are widely used as nanoscale light sources, light collectors or scatterers (Fig. 2a and 3a). Probe fabrication requires a number of steps starting with the formation of a tapered optical fiber. Typically two different methods are used. Chemical etching of a bare glass fiber dipped into hydro-fluoric (HF) acid yields sharp tips [4, 7]. The surface tension of the liquid forms a meniscus at the interface between air, glass, and acid. A taper is formed due to the variation of the contact angle at the meniscus, while the fiber is etched and its diameter decreased. Chemical etching allows reproducible production of larger quantities of probes in a single step. A specific advantage is that the taper angle can be tuned and optical probes with correspondingly large transmission coefficient can be produced. The second method combines local heating using a CO2 laser or a filament and subsequent pulling until the fiber is split apart. The resulting tip shapes depend heavily on the temperature and the timing of the heating and pulling, as well as on the dimensions of the heated area. The pulling method has the advantage of producing tapers with very smooth surfaces, which positively influences the quality of the evaporated
2287
S
metal layer. Etched probes, on the other hand, typically feature rough surfaces. Pulled fibers, however, have small cone angles and thus reduced optical transmission as well as flat end-faces limiting the minimum aperture size. The aperture is formed during the evaporation of aluminum. Since the evaporation takes place under an angle slightly from behind, the deposition rate of metal at the apex is much smaller than on the sides. This geometrical shadowing effect leads to the self-aligned formation of an aperture at the apex. The ideal aperture probe should have a perfectly flat end face to position a sample as close as possible into the near-field of the aperture. Conventional probes generally have a roughness determined by the grain size of the aluminum coating, which is around 20 nm at best. Due to the corrugated end face the distance between aperture and sample increases, which lowers the optical resolution and decreases the light intensity on the sample. Furthermore, these grains often obscure the aperture, which makes the probe ill-defined and not suited for quantitative measurements [4, 5, 7]. Subsequent focused ion beam (FIB) milling can be used to form high definition SNOM probes with well-defined end face (Fig. 3). Microfabricated cantilevered probes. Hollowpyramid cantilevered probes can be batch-fabricated with large taper angles. Lithographic patterning of an oxidized silicon wafer first defines the position of the aperture and the dimensions of the cantilever beam by structuring the oxide layer [6]. Anisotropic etching of the exposed silicon with buffered HF forms a pyramidal groove for the tip and trenches for the cantilever beam. After removing the oxide layer on the opposite side anisotropic etching is used to open a small aperture in the pyramidal groove. A 120-nm chromium layer is deposited on the back side forming the hollow pyramidal tip that is finally freed by isotropic reactive ion etching. Si3N4 tips can be fabricated by dry etching in a CF4 plasma and covering with thin aluminum films. Microfabricated probes based on quartz tips attached to silicon cantilevers have also been reported [2, 7]. First sharp quartz tips were produced in hydrofluoric acid followed by coating with thin films of aluminum and silicon nitride. Reactive ion etching was then used to selectively remove the silicon nitride from the tip apex, while the remaining film served as a mask for wet-etching of the protruding aluminum in a standard
S
S
2288
Al-etching solution, leaving a small aperture on the apex of the tip. For light coupling, windows are etched into the backside of the levers at the position of the tip. Microfabricated tips provide several advantages over fiber-based probes. The mechanical stability is typically increased and often sufficient to measure also in contact AFM mode without destroying the tip. A reproducible fabrication of the tips leads to a welldefined aperture shape. Large taper angles of around 70 shift the position of the mode cut-off closer to the aperture, resulting in higher transmission. Antenna Probes
Sharp metal tip are fabricated in a single step by simple electrochemical etching. In the case of gold, pulsed or continuous etching in hydro-chloric acid (HCL) or a mixture of HCL and ethanol routinely yields tips with diameters in the range of 30–50 nm. These tips can either be used in shear-force mode after gluing to the prong of a tuning-fork or in STM-mode. Silicon or silicon nitride cantilevered probes can be coated by a thin metal film through evaporation. Subsequent nanostructuring by FIB-milling can be used to optimize tip parameters. Tip-on-aperture probes need to be fabricated in a series of sequential steps [14] (Fig. 2c). First, a fiber-based aperture probe is produced on which a well-defined end face is formed by FIB-milling. In the second step, a nanoscale tip is grown by electron beam–induced deposition of carbon. Next, the carbon tip is coated by a thin layer of chromium to improve adhesion of an aluminum layer evaporated during the final preparation step. Since the length of the tip can be controlled during electron beam deposition, TOA probes can be tailored to provide optimum antenna enhancement for a chosen wavelength [8].
Applications in Nanoscience The energy of light quanta – photons – is in the range of electronic and vibrational excitations of materials. These excitations are directly determined by the chemical and structural composition of matter. Optical spectroscopy, the energy-resolved probing of the material response to light exposure, thus provides a wealth of information on the static and dynamical properties of materials. Combining spectroscopy with near-field microscopy is particularly interesting since spectral
Scanning Near-Field Optical Microscopy
information is obtained spatially resolved at the nanoscale. In the following several representative examples covering different material responses including fluorescence, Raman scattering and elastic scattering are briefly illustrated to highlight the capabilities of SNOM techniques. Fluorescence Microscopy Fluorescence measurements were among the first applications of SNOM. In these experiments, aperture-based probes were used to probe highly fluorescent dye-molecules on substrates. Aluminum-coated fiber tips were used for excitation while the fluorescence was collected in the far-field by a highnumerical aperture objective. A molecule is excited only if the optical electric field is polarized parallel to its transition dipole moment. The resulting fluorescence patterns rendered by a single molecule with known orientation can thus be used to visualize the local electric field distribution at the probe. Conversely, the molecular orientation can be determined for known field distributions. These experiments showed that the strongest electrical fields do not occur in the center of the aperture, but at the rims of the metal coating. This is the result of local field enhancement at the thin metal rim (see section “Aperture Probes”). Two lobes with strong fields oriented in axial direction occur located on opposite sides of the aperture in the direction of the polarization of the incident linearly polarized light. Molecules with a transition dipole moment oriented parallel to the tip axis are excited efficiently by these field components as can be seen in Fig. 5 [5]. Rotating the polarization of the incident linearly polarized light is seen to rotate the resulting double lobe pattern that indicates the area with strongest fields. Fluorescence measurements typically do not require high excitation densities and in many cases, the small light transmission of aperture probes represents no major drawback. In fact, fluorescence imaging of single dye molecules with 32 nm spatial resolution has been demonstrated using a microfabricated cantilevered glass tip covered with a 60-nm-thick aluminum film. Due to the thickness of the virtually opaque film, possible contributions from surface plasmons propagating on the outside of the film have been discussed [4]. Examples of fluorescence microscopy measurements based on aperture probes also include, for example, studies of single nuclear pore
Scanning Near-Field Optical Microscopy
a
2289
b
S
c
Scanning Near-Field Optical Microscopy, Fig. 5 Series of three successive aperture-SNOM fluorescence images of the same area (1.2 1.2 mm) of a sample of dye molecules embedded in a thin transparent polymer film. The excitation polarization, measured in the far-field, was rotated from linear vertical (a) to linear horizontal (b) and then changed to circular polarization (c). Circular features marked by arrows result from molecules with transition dipole moments oriented parallel to
the sample plane. The double lobe structure marked by the dashed circle results from a molecule with perpendicularly oriented transition dipole moment. This molecule senses the strong electrical fields that occur at the rim of the metal aperture at positions determined by the far-field polarization. Scale bar 300 nm (Reprinted with permission from Veerman et al. [5]. Copyright 1999, John Wiley and Sons)
complexes and the kinetics of protein transport under physiological conditions. Optical antennas have been used for a variety of samples and materials including photosynthetic proteins, polymers, semiconductor quantum dots, and carbon nano-tubes [11]. The image contrast was based on local field enhancement provided by the tip. The spatial resolution achieved in these experiments was essentially determined by the diameter of the tip and ranged between 10 and 20 nm. Imaging can be combined with local spectroscopy to visualize emission energies on the nanoscale. Single-molecule experiments revealed the field distribution at the tip-antenna in analogy to the discussion made above for aperture probes [14]. While most of the reported studies were exploiting local signal enhancement, distance-dependent metalinduced quenching of fluorescence can also be used for high-resolution imaging. This approach provides sub 10 nm spatial resolution and has been applied to single fluorescent organic molecules and inorganic semiconductor nanorods (see e.g., [12]). In these experiments, the spectrally integrated fluorescence signal was demodulated by the tapping-mode frequency of the AFM cantilever after recording photon-arrival times.
magnitude smaller than the cross-section of fluorescence in the case of organic molecules. Raman measurements thus require higher laser intensities and in many cases the low transmission of aperture probes prohibits their application. The signal enhancement provided by the antenna tip in TENOM is substantial for the detection of nanoscale sample volumina. In the following, a review of selected examples is given to illustrate the possibilities of tip-enhanced Raman scattering (TERS) (see e.g., [11, 15]). In Fig. 6, simultaneous near-field Raman and topographic imaging of individual single-walled carbon nanotubes is shown. The optical image in (a) reflects the intensity of the G’ band, a particular Raman-active vibrational mode of carbon nanotubes. The optical resolution obtained in this experiment was about 25 nm as can be seen from the width of the peaks in the cross-section in Fig. 6c. The strong fields required for sufficient enhancement of the Raman scattering signal can cause laserinduced decomposition and photochemical reactions in the presence of oxygen. TERS of single electronically resonant molecules has been demonstrated for ultra-high vacuum conditions. A review focusing on single-molecule surface- and tip-enhanced Raman scattering can be found in [15].
Raman Microscopy Raman scattering probes the unique vibrational spectrum of a sample and directly reflects its chemical composition and molecular structure. A main drawback of Raman scattering is the extremely low scattering cross-section which is typically 10–14 orders of
Elastic Scattering Microscopy Elastic scattering SNOM probes the dielectric properties of the sample and has been used from the visible to the microwave regime of the electromagnetic
S
S
2290
Scanning Near-Field Optical Microscopy
a
b
d
c
0
photons [kcts/s]
0
8
height [nm]
c
d
8k
2
6
6k 23 nm
4k
29 nm
1
2k 0
0 0
200
400 600 position [nm]
800
1000
0
100 200 position [nm]
Scanning Near-Field Optical Microscopy, Fig. 6 Tipenhanced Raman spectroscopy (TERS) of single-walled carbon nanotubes on glass. Simultaneous near-field Raman image (a) and topographic image (b). Scan area 1 1 mm2. The Raman image is acquired by detecting the intensity of the G’ band upon laser excitation at 633 nm. No Raman scattering signal is detected from humidity-related circular features present in the
topographic image. (c) Cross-section taken along the dashed line in the Raman image indicating a spatial resolution around 25 nm. (d) Cross-section taken along the indicated dashed line in the topographic image. Vertical units are photon counts per second for c and nanometer for d (Reprinted with permission from Hartschuh et al. [16]. Copyright 2003, American Physical Society)
spectrum. Reviews of the fundamentals of the technique and representative applications can be found in [9, 10]. The majority of s-SNOM experiments have been reported for the IR to THz spectral range. Applications include detection of the Mott-transition in nanodomains, mapping of the doping concentration in semiconductors, surface characterization with a sensitivity of a single monolayer, strain-field mapping, and infrared spectroscopy of a single virus. As an example nanoscale infrared spectroscopic near-field mapping of single nano-transistors is shown in Fig. 7. A cantilevered metallized Si-tip operating in tapping-mode with an oscillation frequency of 300 kHz and an amplitude of about 60 nm was used [17]. The data clearly demonstrates the potential of s-SNOM for infrared spectroscopic recognition of materials within individual semiconductor nanodevices.
Based on the antenna approach, s-SNOM typically provides 10–20 nm spatial resolution determined by the diameter of the tip-apex. In most of the s-SNOM experiments to date, monochromatic laser sources were used. Since only the optical response at this frequency is determined, the acquisition of scattering spectra or spectrally resolved images can only be obtained sequentially with a series of image scans at different laser frequencies. New developments exploiting broadband NIR laser sources aim at overcoming this limitation, and recently obtained a spectral bandwidth exceeding 400 cm1. Plasmonics and Photonic Nanostructures SNOM plays a vital role in the field of plasmonics which deals with the study of optical phenomena related to the electromagnetic response of metals
Scanning Near-Field Optical Microscopy
a
2291
c
d ω=895cm−1
10nm
S
e 1043cm−1
1083cm−1
max
NiSi
0nm
b
200nm a-SiO2
S2
min a-SiO2
Gate contact
80°
NiSi a-Si3N4 Source Si
Drain ϕ2
Si
−10°
Scanning Near-Field Optical Microscopy, Fig. 7 Materialspecific mapping of transistor components using s-SNOM: Cross-sectional images of a single transistor fabricated at the 65 nm technology node. (a) Topography. (b) Sketch of the transistor with materials indicated. (c–e) Near-field amplitude
and phase images recorded at three different laser frequencies. Amorphous SiO2 and Si3N4 render reversed optical contrast and are clearly distinguished. A spatial resolution better than 20 nm has been achieved (Reprinted with permission from Huber et al. [17]. Copyright 2010, IOP)
[18]. Near-field optical probes are particularly important for two reasons. First, they provide a means to locally excite propagating surface plasmon polaritons (SPPs) in metal films, a process that is not possible in case of propagating light waves because of momentum (k-vector) mismatch. The broad k-spectrum associated with the near-field of the probe contains sufficient bandwidth for efficient SPP-excitation (see Fig. 1). Second, near-field probes can be used simultaneously to convert SPPs back into propagating waves, thereby probing the local distribution of electromagnetic fields in the vicinity of metallic nanostructures. As an example, the near-field associated with SPPs has been visualized along gold nanowires using an aperture probe in collection mode (see Fig. 2a, [4]). This approach is also termed photon scanning tunneling microscopy (PSTM) to illustrate the analogy between evanescent electromagnetic waves and the corresponding exponentially decaying electron wavefunctions within the tunnel barrier of an STM. PSTM has been widely used to spatially resolve light wave propagation also in dielectric photonic nanostructures [4]. Besides the visualization of static field distributions, optical spectroscopy also allows for the study of their temporal evolution and the propagation of pulses. Figure 8 illustrates ultrafast and phase-sensitive
imaging of the plasmon propagation in a metallic waveguide by PSTM [19]. In this case the near-field microscope uses an aperture-probe in collection mode and incorporates a Mach-Zehnder–type interferometer enabling heterodyne time-resolved detection. Scattering-SNOM with antenna tips has been used extensively to study localized surface plasmon polaritons (LSPP) in different metal nanostructures. By varying the laser excitation frequency near-field optical imaging allowed for distinguishing higher order plasmonic resonances [20].
Perspectives During the last 25 years SNOM has demonstrated its capabilities for sub-wavelength optical imaging and spectroscopy of surfaces and sub-surface features. The strength of SNOM results from its enormous flexibility with respect to sample types as well as measurement configurations and in particular, from its combination with a broad range of spectroscopic techniques. Ongoing developments aim at increasing antenna efficiencies and new aperture-type schemes [18]. In addition, the combination of nano-optical approaches and ultrafast laser technique is explored
S
S
2292
Scanning Probe Microscopy
1 5μm
0
a
b
c
d
e
f
Scanning Near-Field Optical Microscopy, Fig. 8 Phasesensitive and ultrafast near-field microscopy of a surface plasmon polariton (SPP) waveguide. The local electric field is collected by an aperture-probe and detected interferometrically in a Mach-Zehnder–type configuration. (a–e) Normalized amplitude information of the SPP wavepacket E-field. Succeeding frames are new scans of the probe. In between the frames the delay line is lengthened to 14.4 mm. Therefore, the time between two frames is 48 fs. The scan frame is 15 110 mm2, scan lines run from top to bottom. (f) Topography of the SPP waveguide obtained by shear-force feedback (Reprinted with permission from Sandtke et al. [19]. Copyright 2008, American Institute of Physics)
to achieve enhanced light localization and the control of optical near-fields on the time scale of few optical cycles.
Cross-References ▶ Atomic Force Microscopy ▶ Confocal Laser Scanning Microscopy ▶ Light Localization for Nano-optical Devices ▶ Nanostructures for Photonics ▶ Scanning Tunneling Microscopy
References 1. Novotny, L.: The history of near-field optics. In: Wolf, E. (ed.) Progress in Optics, vol. 50, pp. 137–184. Elsevier, Amsterdam (2007) 2. Novotny, L., Hecht, B.: Principles of Nano-optics. Cambridge, Cambridge University Press (2006) 3. Kawata, S., Shalaev, V.M. (eds.): Advances in Nano-Optics and Nano-Photonics Tip Enhancement. Elsevier, Amsterdam (2007) 4. Kawata, S., Shalaev, V.M. (ed.): Handbook of Microscopy for Nanotechnology. Kluwer, Dordrecht (2006)
5. Veerman, J.A., Garcia-Parajo, M.F., Kuipers, L., van Hulst, N.F.: Single molecule mapping of the optical field distribution of probes for near-field microscopy. J. Microsc. 194, 477–482 (1999) 6. Mihalcea, C., Scholz, W., Werner, S., M€ unster, S., Oesterschulze, E., Kassing, R.: Multipurpose sensor tips for scanning near-field microscopy. Appl. Phys. Lett. 68, 3531–3533 (1996) 7. Hecht, B., Sick, B., Wild, U.P., Deckert, V., Zenobi, R., Martin, O.J.F., Pohl, D.E.: Scanning near-field optical microscopy with aperture probes: fundamentals and applications. J. Chem. Phys. 112, 7761–7774 (2000) 8. Bharadwaj, P., Deutsch, B., Novotny, L.: Optical antennas. Adv. Opt. Photon. 1, 438–483 (2009) 9. Keilmann, F., Hillenbrand, R.: Near-field microscopy by elastic light scattering from a tip. Phil. Trans. R. Soc. Lond. A 362, 787–805 (2004) 10. Br€ undermann, E., Havenith, M.: SNIM: scanning near-field infrared microscopy. Annu Rep Prog Chem Sect. C Phys. Chem. 104, 235–255 (2008) 11. Hartschuh, A.: Tip-enhanced near-field optical microscopy. Angew. Chem. Int. Ed. 47, 8178–8198 (2008) 12. Ma, Z., Gerton, J.M., Wade, L.A., Quake, S.R.: Fluorescence near-field microscopy of DNA at sub-10 nm resolution. Phys. Rev. Lett. 97, 260801–260804 (2006) 13. Taminniau, T.H., Stefani, F.D., Segerink, F.B., van Hulst, N.F.: Optical antennas direct single-molecule emission. Nat. Photon. 2, 234–237 (2008) 14. Frey, H.G., Witt, S., Felderer, K., Guckenberger, R.: High resolution imaging of single fluorescent molecules with the optical near field of a metal tip. Phys. Rev. Lett. 93, 200801–200804 (2004) 15. Pettinger, B.: Single-molecule surface-and tip-enhanced Raman spectroscopy. Mol. Phys. 108, 2039–2059 (2010) 16. Hartschuh, A., Sa´nchez, E.J., Xie, X.S., Novotny, L.: Highresolution nearfield Raman microscopy of single-walled carbon nanotubes. Phys. Rev. Lett. 90, 095503–4 (2003) 17. Huber, A.J., Wittenborn, J., Hillenbrand, R.: Infrared spectroscopic near-field mapping of single nanotransistors. Nanotechnology 21, 235702–6 (2010) 18. Schuller, J.A., Barnard, E.S., Cai, W., Jun, Y.C., White, S. W., Brongersma, M.L.: Plasmonics for extreme light concentration and manipulation. Nat. Mater. 9, 193–204 (2010) 19. Sandtke, M., Engelen, R.J.P., Schoenmaker, H., Attema, I., Dekker, H., Cerjak, I., Korterik, J.P., Segerink, F.B., Kuipers, L.: Novel instrument for surface plasmon polariton tracking in space and time. Rev. Sci. Instrum. 79, 013704–10 (2008) 20. Dorfm€ uller, J., Vogelgesang, R., Khunsin, W., Rockstuhl, C., Etrich, C., Kern, K.: Plasmonic nanowire antennas: experiment, simulation, and theory. Nano. Lett. 10(9), 3596–3603 (2010)
Scanning Probe Microscopy ▶ Atomic Force Microscopy
Scanning Thermal Microscopy
2293
Scanning Surface Potential Microscopy ▶ Kelvin Probe Force Microscopy
Scanning Thermal Microscopy Li Shi Department of Mechanical Engineering, The University of Texas at Austin, Austin, TX, USA
Synonyms Micro-thermal analysis; Scanning thermal profiler
Definition Scanning Thermal Microscopy (SThM) is a class of experimental methods for high spatial resolution mapping of the surface temperature distribution of an operating device or the thermal property variation of a structure with the use of a sensor fabricated on a scanning probe.
Thermal Probes In 1986, Williams and Wickramasinghe [1] pioneered a so-called scanning thermal profiler technique based on a thermocouple sensor fabricated at the end of a probe tip for Scanning Tunneling Microscopy (STM). The thermocouple sensor consisted of two dissimilar conductors that made a junction at the end of the STM tip. An insulator separated the two conductors in all areas remote from the tip. The size of the thermocouple junction could be made as small as 100 nm. When a temperature difference (DT) exists between the thermocouple junction and the ends of the two lead wires, thermal diffusion of electrons through the two dissimilar wires results in a thermoelectric voltage between the two lead wires. The magnitude of the thermovoltage measured with the use of a voltmeter is V ¼ ðS1 S2 ÞDT
(1)
S
where S1 and S2 are the Seebeck coefficient of the two thermocouple wires. The main purpose of this first scanning thermal profiler was not for mapping temperature distribution of a surface but to regulate the tip-sample distance using the relationship between the thermocouple tip temperature and the distance between the tip and the sample. The work of Williams and Wickramasinghe stimulated intense efforts to develop a scanning probe for high spatial resolution mapping of the temperature distribution or thermal properties on a surface. In 1993, Majumdar and co-workers [2] introduced a wire thermocouple probe that could be used in an atomic force microscope (AFM) for simultaneous mapping of topography and temperature. Since then, different designs of scanning thermal probes have been fabricated. A common feature of these thermal probes is a thermal sensor fabricated at the end of an AFM or STM tip. The thermal sensor can be a thermocouple, a resistance thermometer, a Schottky diode, or a fluorescence particle [3–5]. As discussed above, a thermocouple measures the temperature difference between the junction of its two constituent metals and the other ends of the two metal wires. The current-voltage (I-V) characteristics of a Schottky junction depends on the temperature, and can thus serve as a local temperature sensor. A resistance thermometer is usually made of a metal or degenerately doped semiconductor with a relatively large and constant temperature coefficient of resistance. Because the resistance of a nanoscale resistor can be too small for sensitive electronic detection, resistance thermometer cannot be miniaturized as readily as a thermocouple sensor. Another approach is to use the emission band of some fluorescence particles, which shifts with increasing temperature. This feature has been used for making a SThM probe with a fluorescence particle located at the end of a scanning probe tip [5]. Besides these temperature-sensing techniques, the thermally induced bending of a biomaterial AFM cantilever has been used for temperature measurements [4]. Moreover, the thermal expansion of a sample has been measured with a regular AFM tip, and used to infer the sample temperature [4]. Different methods have been reported for fabricating different SThM probes. Some of the methods use sequential (or one at a time) fabrication methods [4], whereas batch fabrication processes have been developed for wafer scale fabrication of the SThM probes
S
S
2294
Scanning Thermal Microscopy, Fig. 1 Scanning electron micrographs of a SiNx AFM cantilever probe (a) with a sub-micron Pt-Cr thermocouple junction (b) formed at the apex of the SiO2 tip
Scanning Thermal Microscopy
Pt Pt-Cr junction
Tip Laser reflector
Cr SiNx
1 μm
5 μm
[6]. While the probe batch fabrication processes are somewhat similar to those that have been developed for the manufacturing of ultra large scale integrated (USLI) devices, they often involve additional etching steps to make free standing cantilever probes, as well as novel processes to form a thermal sensor such as a thermocouple at the end of the cantilever probe tip. Figure 1 shows a batch-fabricated thermocouple SThM cantilever probe that consists of a 0.5 mm thick low stress silicon nitride (SiNx) cantilever, a 8 mm tall silicon dioxide (SiO2) pyramid tip with a 20 nm tip radius, and a sub-micron Pt-Cr thermocouple junction formed at the apex of the SiO2 tip [6]. Several other unique fabrication processes have been developed for wafer scale fabrication of a thermocouple junction, Schottky diode, Pt-C or doped silicon resistance thermometer [4]. Special probe holders with necessary electrical contacts or optical access can allow for the use of the SThM sensor probes in a commercial AFM or STM. STM requires the sample surface to be conducting because the tip-sample gap is controlled based on the tunneling current. This issue has limited the use of SThM probes in STM. In comparison, AFM with a cantilever SThM probe can allow simultaneous topography and thermal mapping of both conducting and non-conducting samples. In a typical contactmode AFM, the tip-sample spacing is regulated by the force acting on the probe tip. The tip-sample interaction force can be obtained from the AFM cantilever bending that is measured with the use of several methods. One of these methods is to measure the laser beam reflected by the cantilever with the use of a position-sensitive detector. Another method is to employ a built-in piezoresistive sensor in the cantilever to measure the cantilever bending. Although the optical detection method provides superior sensitivity,
the laser heating of the thermal sensor at the end of the cantilever needs to be minimized.
Theoretical Analysis of Heat Transfer Mechanisms For many of the various SThM sensor probes, the sensitivity and spatial resolution of the thermal imaging technique depend on the mechanisms of heat transfer between the thermal probe and the sample. For SThM measurements conducted in air, heat is transferred between the probe and the sample via conduction through the solid–solid contact, a liquid meniscus at the tip-sample contact, and the air gap between the probe, and via radiation, as illustrated in Fig. 2 heat transfer through the air gap and via radiation is not localized at the tip-sample contact, and can deteriorate the spatial resolution of the SThM probes. For a circular contact area with radius b, the spreading thermal resistance of the sample is obtained as [7] Rs ¼
1 4ks b
(2)
where ks is the thermal conductivity of the sample. For ks 5 W/m-K, b 30 nm, Rs 1.7 106 K/W. The thermal interface resistance through the solid– solid contact is given as [8] Ri;s
4 4K ¼ aCs vs pb2 3aks pb
(3)
where Cs and vs are the specific heat and phonon group velocity of the sample, a is the phonon transmission coefficient from the sample into the tip, K ls/b is the Knudsen number, and ls is the phonon mean free path
Scanning Thermal Microscopy
Tb
2295 Cantilever Mount, T0
Cantilever
An approximation expression for the tip-sample air thermal conductance per unit area is given here as
T0 RC
Solid-solid conduction
ga
Tb
Sample
Rt
SiO2
Pt
Liquid conduction
Sio2 tip
Air conduction
Tj
Radiation
Ts
Cr
Tj Ri
Liquid Ts
Scanning Thermal Microscopy, Fig. 2 Schematic diagram showing the heat transfer mechanisms between the thermal probe and the sample as well as a thermal resistance circuit where the heat transfer through the air gap is ignored (Reproduced from [6] with permission by ASME)
in the sample. The second equation is obtained with the use of the kinetic theory, ks ¼ Csvsl/3 [8]. For a ¼ 0.3, ks ¼ 5 W/m-K, b ¼ 30 nm, and ls ¼ 5 nm, Ri,s ¼ 1.5 106 K/W. A liquid meniscus often forms at the tip-sample junction when the tip is scanned on the sample in air. The interface thermal resistance through the liquid meniscus (Ri, l) at the tip-sample junction depends on the humidity and surface properties. The value of Ri, l has been estimated by Majumdar to be on the order of 105 K/W [4], which is lower than the solid-solid thermal interface resistance. The total interface tip-sample thermal resistance is 1 Ri ¼ ðRi;s þ R1 i;l Þ 1
(4)
The spreading thermal resistance of the conical tip can be estimated as Rt
1 pkt b tan y
S
(5)
where kt is the thermal conductivity of the tip and y is the half angle of the conical tip. For kt 5 W/m-K, b ¼ 30 nm, y ¼ p/8, Rt ¼ 5 106 K/W. Hence, if the thermocouple junction can be made as small as the contact size, the high spreading thermal resistance of the tip can be utilized to thermally isolate the junction from the cantilever.
ka z þ la
(6)
where ka and la are the thermal conductivity and mean free path of air molecules, and z is the air gap. In the ballistic limit of z la , this expression is reduced to ga ¼ ka/ la. Based on the kinetic theory, ka ¼ Cava la /3, where Ca and va are the specific heat and velocity of air molecules, so that ga ¼ Cava/3 for the ballistic limit. This result is close to the ballistic thermal conductance per unit area [8], CaVa/4, for heat transfer of gas molecules between two close parallel plates when the thermal accommodation coefficient is unity, corresponding to the case that the scattered molecules take the temperature of the surface. In the diffusive limit of z la , ga ¼ ka/z is the diffusive thermal conductance per unit area of the air gap. The total tip-sample thermal conductance through the air gap can be obtained as Z
bþH tan y
ka 2prdr l þ ðr bÞ= tan y a b b la tan y la þ H 2 ln ¼ 2pka H tan y 1 þ (7) H tan y H
Gg;ts ¼
where H is the tip height. For H ¼ 8 mm and y ¼ p/8, Gg, t-s 2.2 107 W/K, corresponding to a resistance Rg, t-s 1/ Gg, ts ¼ 4.4 106 K/W [6]. Far-field radiation conductance between the tip and the sample is approximated as that of a two-surface enclosure as Grad;ts
4At sT 3 ð1 et Þ=et þ 1=Fts þ At ð1 es Þ=As es (8)
where s is the Stefan–Boltzmann constant, At and As are the surface areas of the tip and sample, et and es are the surface emissivity of the tip and sample, and Fts and T are the view factor and the average temperature between the tip and sample. When the temperature is close to 300 K, Grad,ts is on the order of 1 1010 W/K. Both the far-field and near-field radiation heat transfer between the tip and the sample can be ignored compared to those via the air gap,
S
S
2296
Scanning Thermal Microscopy
solid–solid interface, and liquid meniscus [4], unless for a very high tip or sample temperature. Because of heating by the sample surface, the temperature of the air molecules surrounding the cantilever, Tg, can be different from the temperature at the cantilever mount, T0, which is usually close to the room temperature. In this case, heat transfer from the tip into the cantilever can be obtained as the fin heat transfer rate expressed as [7] qtc ¼ M
cosh mL y0 =yb sinh mL
(9)
where M ¼ (hPkcAc)1/2yb, m ¼ (hP/kcAc)1/2, h is the heat transfer coefficient between the cantilever and the surrounding gas, kc and Ac are the thermal conductivity and cross-section of the cantilever, P is the perimeter of the cantilever cross-section, y0 ¼ T0Tg, yb ¼ TbTg, and Tb is the temperature at the joint between the tip and the cantilever. When only a small sample area is at a temperature different from the room temperature so that Tg T0, the thermal resistance of the cantilever can be defined as Rc
Tb T0 tanh mL ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi qtc hpkc Ac
(10)
For kc 5 W/m-K, and 0.5 mm by 10 mm cantilever cross-section, L ¼ 150 mm, h ka(p/Ac)1/2, Rc ¼ 3 105 K/W. On the other hand, when a large area of the sample is at a temperature higher than T0, the air molecules surrounding the cantilever can be heated by the sample surface to a temperature Tg that is considerably higher than T0. In this case, the cantilever is heated by the surrounding air, resulting in an additional heat transfer path from the sample to the probe.
Experimental Characterization of Heat Transfer Mechanisms Experimentally, the heat transfer path through the air gap between the sample and the cantilever was found to be significant when the size of the heated zone on the sample surface was not small [6]. In the experiment [6], a 350-nm wide metal line was joule heated to 5.3 K above room temperature. The cantilever deflection and temperature rise of the thermocouple junction
were recorded simultaneously when the sample was approached and then retracted from a thermocouple SThM probe tip. When the sample approached the tip, the cantilever deflection signal remained approximately constant before the sample contacted the tip, as shown in the deflection curve in Fig. 3. In this region, the junction temperature rise was caused by air conduction between the probe and the sample. As the tipsample distance was reduced, the junction temperature rise due to air conduction increased slowly. Before the sample made solid–solid contact to the tip, the adsorbed liquid layers on the tip and the sample bridged each other. Initially, this liquid bridge pulled the tip down by a van der Waals force, as being seen in the dip labeled as “jump to contact” in the deflection curve. Coincidentally, there was a small jump in the junction temperature due to heat conduction through the liquid bridge. As the sample was raised further, both the solid–solid contact force and the junction temperature increased gradually, until the cantilever was deflected for more than 100 nm. After this point, the junction temperature remained almost constant as the contact force increased. As the sample was retracted from the tip, the junction temperature remained almost constant until at a cantilever deflection of 100 nm, after which the junction temperature rise decreased roughly linearly but at a smaller slope than that found in the approaching cycle. As the sample was lowered further, the tip was pulled down together with the sample by surface tension of the liquid bridge until after a certain point, the restoring spring force of the cantilever exceeded the surface tension, and the tip “snapped out of contact” with the sample. Associated with the breaking of the liquid bridge, there was a small drop in the junction temperature. This experiment shows several mechanisms. First, before the tip contacted the sample, air conduction contributed to a junction temperature rise up to 0.03 K per K sample temperature rise, which was about 60% of the maximum junction temperature rise at the maximum contact force of the experiment. Second, conduction through a liquid meniscus was responsible for the sudden jump and drop in junction temperature when the tip “jumped to contact” to and “snapped out of contact” from the sample, respectively. Third, solid–solid conduction resulted in the almost linear relationship between the junction temperature rise and the contact force. This is a well
Scanning Thermal Microscopy
S
2297
Scanning Thermal Microscopy, Fig. 3 Cantilever deflection and temperature response of a thermocouple probe as a function of the sample vertical position when a 350-nm wide heater line sample was raised toward and then retracted from the tip (Reproduced from [6] with permission by ASME)
Jump to/snapped out of contact
In contact
Out of contact
0.08
Deflection (nm)
Snapped out of contact
0.06
Approaching
0 Solid
0.04
Retracting
−100 Liquid 0.02 Air
−200 0.1
0.2
0.3
0.4
Temperature response (K/K)
Jump to contact
100
0.00
Sample vertical position (μm)
understood feature for macroscopic solid–solid contacts. When the contact force was increased further, the junction temperature approached a constant likely because the contact size approached the maximum possible value limited by the diameter of an asperity located at the tip-sample interface. The point contact experiment was repeated for a 5.8 mm wide and 2,000 mm long heater line on an oxidized silicon wafer. While the increase of the normalized junction temperature due to solid and liquid conduction was similar in magnitude to those observed in the narrower line, the normalized junction temperature rise due to air conduction was one order of magnitude higher than the corresponding one in the narrower line. In addition, the temperature rise of the thermocouple junction was measured when the tip was in contact with three heater lines of different line widths. It was found that the temperature rise at the thermocouple junction of the tip was about 53%, 46%, and 5% of that of a 50, 3, and 0.3 mm wide line, respectively, showing a trend of increasing temperature rise in the SThM probe with increasing heater line width. These results suggest that the cantilever was heated more by air conduction between the tip and the larger hot area of the wider heater lines. This
trend indicates that air conduction plays an important role in probe-sample heat transfer, especially when the hot area on the sample surface is large in comparison to the tip size. The unwanted heat transfer through the air gap can be eliminated by conducting SThM measurements in vacuum. In this case, the tip-sample air conductance Gg, ts ¼ 0, and the thermal resistance of the cantilever is the conduction thermal resistance given as Rc ¼
L kc Ac
(11)
For a cantilever length L ¼ 150 mm, thermal conductivity kc ¼ 5 W/m-K, and cross-section Ac ¼ 0.5 mm 10 mm, Rc ¼ 6 106 W/K based on the above equation. In addition, the heat transfer from the sample to the tip in vacuum can be described with the use of the thermal resistance circuit shown in Fig. 2. However, vacuum operation of the SThM probe is considerably much more challenging than operation in open air environment. Moreover, the liquid meniscus at the tip-sample junction is also eliminated by the vacuum environment. Although this can further enhance the spatial resolution of the thermal imaging
S
S
2298
Scanning Thermal Microscopy
technique, the tip-sample thermal interface conductance could be reduced considerably. Consequently, the thermocouple junction temperature at the tip end could become rather different from the sample temperature, resulting in low sensitivity and large uncertainty in the measured temperature. An alternative method to eliminate the unwanted air conductance while keeping the heat transfer via the liquid meniscus is based on a dual scan measurement approach [9]. In this approach, after a line scan of the SThM probe on the sample surface in the contact mode, the probe is scanned on the same line in the lift mode with a small fixed tip-sample distance between the tip and the sample. The difference between the two thermal sensor signals obtained in the contact mode and the lift mode is attributed to the local heat transfer through the solid–solid contact and the liquid meniscus at the tip-sample junction, and can be used to obtain the local sample surface temperature at the contact point, as discussed below.
Surface Temperature Mapping When the heat transfer through the probe-sample gap is eliminated by the dual scan technique, the difference in the measured thermocouple junction temperature values in the contact mode and the lift mode is still different from that for vacuum measurement given by the thermal resistance circuit of Fig. 2 as T j ¼ ðTs T0 Þ 1 þ
Ri Rc þ Rt
1 þ T0
(12)
where Ts is the local sample temperature at the contact point, T0 is the ambient temperature. It is important to increase Rt and Rc relative to Ri and Rs, so as to maximize DTj and minimize the sample temperature change due to the contact by the tip. For this reason, it is desirable to use low-thermal conductivity dielectrics such as SiO2 and SiNx to fabricate the pyramid tip and the cantilever, and low-thermal conductivity conductors such as Cr and Pt to fabricate the thermocouple sensor. The geometry of the sensor, tip, and cantilever also needs to be optimized to minimize heat loss. Even for state-of-the-art thermocouple SThM probes, Rt and Rc are still not much larger than Ri. Hence, the DTj obtained from the differential dual
scan measurement needs to be calibrated in a separate experiment, where the sample is heated with an external heater and the sample surface temperature is determined using a different method, such as a resistance thermometer fabricated on the sample surface [10]. SThM methods operated in the contact mode or the dual scan mode have been employed to measure the surface temperature distribution of silicon, carbon nanotube, and graphene nanoelectronic devices, with a spatial resolution better than 100 nm [10, 11]. The very high electric field in electrically biased nanoelectronic devices can result in non-equilibrium transport of electrons, optical phonons, and acoustic phonons [12]. Energetic hot electrons accelerated in the high field lose their energy to lattice vibration via emission of high frequency optical phonons. The optical phonons are further scattered with acoustic phonons that make a larger contribution to the lattice thermal conductivity because of a much higher group velocity than the optical phonons. The scattering time between optical phonons and acoustic phonons can be much longer than that between hot electrons and optical phonons in nanoelectronic devices made of silicon and carbon nanotubes. Consequently, energy transfer from optical phonons and acoustic phonons becomes a bottle neck in the heat dissipation process. This bottle neck can result in different temperatures for acoustic phonons and optical phonons. The local temperatures of different energy carriers in operating nanoelectronic devices can be measured with different methods. For example, infrared spectroscopy has been used to measure the electronic temperature of electrically biased graphene devices by fitting the infrared emission from the hot electrons in the device with the Planck distribution [13]. The intensity ratio between the anti-Stokes peak to the Stokes peak in the Raman spectrum can be used to probe the local temperature of the zone center or zone boundary optical phonons that are active in the Raman scattering processes [14]. Complementary to these optical thermometry methods, the thermocouple SThM probe can be used to map the low frequency phonon temperature distribution because the phonon transmission coefficient at the tip-sample interface increases with decreasing phonon frequency. This capability has been employed to observe biasdependant and asymmetric acoustic phonon temperature distribution in electrically biased graphene with a superior spatial resolution of about 100 nm [10], as illustrated in Fig. 4.
Scanning Thermal Microscopy
2299
S
150 μV
12 μm
12 μm
0 μV VDS = 4 V 0 VGS = −20 V
0 VGS = 0 V
0 VGS = 40 V
Scanning Thermal Microscopy, Fig. 4 Measured SThM thermovoltage maps of an electrically biased graphene device for a constant drain-source bias (VDS ¼ 4 V) and different
gate-source voltage of 20 V, 0 V, and 40 V, respectively. The scan size is 12 mm 12 mm
Besides the dual scan method, an active SThM method has been reported by Nakabeppu and Suzuki [15] to obtain quantitative temperature map of a sample surface in vacuum. Their probe for active SThM consists of two thin film thermocouples (TCs) and a micro-heater fabricated on a cantilever. When the cantilever was scanned on a heated sample, heat flow from the sample into the cantilever resulted in a temperature drop (DT) along the cantilever, which was measured using the differential thermocouple. The micro-heater power was adjusted by using a feedback loop until the differential thermocouple reads DT ¼ 0. Under this condition, the heat flow from the sample to the cantilever was zero so that the measured tip temperature is the same as the sample surface temperature. Alternatively, the active SThM method can also be implemented using the dual scan operation in air [9]. In this implementation, the SThM probe is heated to different temperatures and is scanned on a hot sample at each heating rate for the probe. When the measured sensor temperature of the heated SThM probe is the same during the contact-mode and lift-mode scanning of the probe, the sensor temperature is taken as the same as the local sample temperature, and the heat flow between the tip and the sample is nullified. It is worth emphasizing that the key issue in SThM measurements is that the thermal sensor temperature can be quite different from the sample temperature. This issue was addressed in the aforementioned measurements via either a detailed calibration or actively heating the sensor until the sensor temperature was the same as the sample temperature. While four-point measurement methods with minimum current leakage
into the voltage measurement devices are commonly used to address problems caused by contact electrical resistance in electrical measurements, a thermal analogue of this approach has not been effective for addressing the problem caused by thermal contact resistance in SThM measurements, because it is difficult to eliminate heat loss into a temperature measurement device, or the thermal probe in SThM.
Thermal Property Mapping Scanning thermal probes consisting of a resistance heater and thermometer at the tip end have been used to measure the local thermal property near the surface in a solid sample. For this measurement, the temperature of the heated probe tip decreases as the tip touches the sample. A larger drop in the tip temperature is expected when the tip touches a higher thermal conductivity region of the sample because of a smaller spreading thermal resistance in the sample. Alternatively, the probe heating rate can be feedbackcontrolled to maintain a constant tip temperature during tip scanning. In this case, a higher heating power is needed to achieve a constant tip temperature when the tip touches a higher thermal conductivity region. Although thermal conductivity contrast has been observed under dc operations, accurate thermal conductivity measurement remains a challenge especially for low-thermal conductivity samples [4]. This is in part because a significant fraction of the power may be dissipated through the cantilever instead of into the sample. The heat loss through the cantilever can be
S
S
2300
Scanning Thermal Microscopy
reduced and the spatial resolution can be improved by using microfabricated probes with a low-thermal conductivity cantilever and a sharp pyramid Si resistance heater and thermometer tip. However, with a decreased contact radius b, the tip-sample thermal interface resistance increases according to 1/b2, and can dominate the spreading thermal resistance of the sample that scales as 1/b. The large interface thermal resistance can make it difficult to quantify the sample thermal property when the tip size is small. The resistance thermometer probe can also be operated in the ac mode, as shown by Pollock, Hammiche, and coworkers [3]. The sensor probe used in their measurements was 5-mm-diameter Pt resistance wire bent to form a tip at the bent. An ac electrical heating current was used to modulate the probe temperature by 5 C at 10 kHz around a dc temperature of 40 C, the measured amplitude and the phase shift in the ac voltage drop across their Pt wire thermometer probe depended on the thermal properties of the sample. One advantage of measuring the phase lag is that it is independent of the temperature dependence of the electrical resistivity of Pt wire and the power input to the probe, because the phase lag is given as [4] tan f ¼
Gim Gr
(13)
where Gr and Gim are the real and imaginary parts of the complex AC thermal conductance, which can be expressed as rffiffiffiffiffiffiffi rffiffiffiffiffiffiffi o o Gr 2pks b 1 þ b þ Ac kc 2as 2ac rffiffiffiffiffiffiffi o Gim omC þ 2pks b2 2ac
(14)
where b is the contact radius, ks and as are the thermal conductivity and thermal diffusivity of the sample, kc and ac are the thermal conductivity and thermal diffusivity of the cantilever probe, and the mC product is the thermal mass of the temperature sensor. For simplicity, the thermal interface resistance is ignored in this analysis [4]. For the phase lag to be sensitive to the thermal properties of the sample, the thermal conductance of the cantilever probe cannot be much larger than that of the sample. In addition, it should be noted that both the amplitude and the phase of the signal are influenced by
pffiffiffiffiffiffiffiffiffiffiffiffiffi the penetration depth of 2as =o ¼ d. Hence, varying the frequency can allow for depth profiling and subsurface imaging. Pollock, Hammiche, and coworkers [3] have also used their Pt wire resistance heater and thermometer probe to perform localized ac calorimetry [16]. Here they ramped the temperature of both the sample and the probe at about 15 C/min while adding an ac temperature modulation of 1 C at 10 kHz by the probe. They found that any phase transition in the sample gave rise to a change in the phase signal in ac mode. This is observed more clearly in the first derivative of the phase as a function of temperature. By scanning the sample under this mode, calorimetric analysis can be performed locally.
Summary and Future Directions A number of SThM probes have been designed and fabricated. Different operating methods including the dual scan and the zero heat flux techniques have been employed for quantitative mapping of the surface temperature distribution of operating devices. Both dc and ac operations of a resistance heater and thermometer probe have allowed the mapping of local thermal property variation of nanostructured materials. Further enhancement of the spatial resolution and thermal measurement accuracy of the SThM methods can be made by continuous miniaturization of the sensor size at the probe tip as well as improved thermal isolation of the sensor. Detecting the phase lag in the ac SThM signal can potentially allow for locating sub-surface defects that generate localized heating in operating electronic devices, as suggested in [17]. The SThM probes based on a thermal-to-electrical signal transduction method can be complemented with non-contact optical methods for mapping the local temperature distribution of different phonon populations or different thermal properties. Although the spatial resolution of conventional far-field optical imaging methods is limited by diffraction to be on the order of the wavelength, the diffraction barrier has been broken by near-field scanning optical microscopy (NSOM) methods [18] as well as several far-field optical nanoscopy methods [19]. Near-field infrared evanescence wave emitted from a sample surface has been collected using a solid immersion lens [11] and/or scattered into far-field signal by an apertureless metal
Scanning Tunneling Microscopy
tip [20]. These techniques have the potential for profiling the surface temperature distribution of devices with a spatial resolution of a fraction of the IR wavelength. Similarly, the spatial resolution of Raman thermograph or thermal reflectance techniques can potentially be improved with either a near-field or far-field nanoscopy technique.
Cross-References ▶ Atomic Force Microscopy ▶ Carbon-Nanotubes ▶ Graphene ▶ Scanning Tunneling Microscopy ▶ Thermal Conductivity and Phonon Transport
References 1. Williams, C.C., Wickramasinghe, H.K.: Scanning thermal profiler. Appl. Phys. Lett. 49, 1587–1589 (1986) 2. Majumdar, A., Carrejo, J.P., Lai, J.: Thermal imaging using the atomic force microscope. Appl. Phys. Lett. 62, 2501– 2503 (1993) 3. Pollock, H.M., Hammiche, A.: Micro-thermal analysis: techniques and applications. J. Phys. D Appl. Phys. 34, R23–R53 (2001) 4. Majumdar, A.: Scanning thermal microscopy. Annu. Rev. Mater. Sci. 29, 505–585 (1999) 5. Aigouy, L., Tessier, G., Mortier, M., Charlot, B.: Scanning thermal imaging of microelectronic circuits with a fluorescent nanoprobe. Appl. Phys. Lett. 87, 3 (2005) 6. Shi, L., Majumdar, A.: Thermal transport mechanisms at nanoscale point contacts. J. Heat Trans-T. ASME. 124, 329– 337 (2002) 7. Incropera, F.P., Dewitt, D.P., Bergman, T.L., Lavine, A.S.: Fundamentals of Heat and Mass Transfer. John Wiley, Hoboken (2007) 8. Chen, G.: Nanoscale Energy Transport and Conversion: A Parallel Treatment of Electrons, Molecules, Phonons, and Photons. Oxford University Press, New York (2005) 9. Chung, J., Kim, K., Hwang, G., Kwon, O., Jung, S., Lee, J., Lee, J.W., Kim, G.T.: Quantitative temperature measurement of an electrically heated carbon nanotube using the null-point method. Rev. Sci. Instrum. 81, 5 (2010) 10. Jo, I., Hsu, I.-K., Lee, Y.J., Sadeghi, M.M., Kim, S., Cronin, S., Tutuc, E., Banerjee, S.K., Yao, Z., Shi, L.: Lowfrequency acoustic phonon temperature distribution in electrically biased graphene. Nano Lett. (2010). doi:10.1021/ nl102858c 11. Cahill, D.G., Goodson, K., Majumdar, A.: Thermometry and thermal transport in micro/nanoscale solid-state devices and structures. J. Heat Trans-T. ASME. 124, 223–241 (2002) 12. Tien, C.L., Majumdar, A., Gerner, F.M.: Microscale Energy Transport. Taylor & Francis, Washington, DC (1998)
2301
S
13. Berciaud, S., Han, M.Y., Mak, K.F., Brus, L.E., Kim, P., Heinz, T.F.: Electron and optical phonon temperatures in electrically biased graphene. Phys. Rev. Lett. 104, 227401 (2010) 14. Chae, D.H., Krauss, B., von Klitzing, K., Smet, J.H.: Hot phonons in an electrically biased graphene constriction. Nano Lett. 10, 466–471 (2010) 15. Nakabeppu, O., Suzuki, T.: Microscale temperature measurement by scanning thermal microscopy. J. Therm. Anal. Calorim. 69, 727–737 (2002) 16. Price, D.M., Reading, M., Hammiche, A., Pollock, H.M.: Micro-thermal analysis: scanning thermal microscopy and localised thermal analysis. Int. J. Pharm. 192, 85–96 (1999) 17. Kwon, O., Shi, L., Majumdar, A.: Scanning thermal wave microscopy (STWM). J. Heat Trans-T. ASME. 125, 156–163 (2003) 18. Dunn, R.C.: Near field scanning optical microscopy. Chem. Rev. 99, 2891–2928 (1999) 19. Hell, S.W.: Far-field optical nanoscopy. Science 316, 1153– 1158 (2007) 20. De Wilde, Y., Formanek, F., Carminati, R., Gralak, B., Lemoine, P.A., Joulain, K., Mulet, J.P., Chen, Y., Greffet, J.J.: Thermal radiation scanning tunneling microscopy. Nature 444, 740–743 (2006)
Scanning Thermal Profiler ▶ Scanning Thermal Microscopy
Scanning Tunneling Microscopy Ada Della Pia and Giovanni Costantini Department of Chemistry, The University of Warwick, Coventry, UK
Definition A scanning tunneling microscope (STM) is a device for imaging surfaces with atomic resolution. In STM, a sharp metallic tip is scanned over a conductive sam˚ while applying a voltage ple at distances of a few A between them. The resulting tunneling current depends exponentially on the tip-sample separation and can be used for generating two-dimensional maps of the surface topography. The tunneling current also depends on the sample electronic density of states, thereby allowing to analyze the electronic properties of surfaces with sub-nm lateral resolution.
S
S
2302
Scanning Tunneling Microscopy
Scanning Tunneling Microscopy, Fig. 1 Schematic representation of a STM. Tip and sample are held at ˚ and a distance s of a few A a bias voltage V is applied between them. The resulting tunneling current I is recorded while the tip is moved across the surface. The coordinate system is also shown
Overview and Definitions ˚ apart and a bias If two electrodes are held a few A voltage is applied between them, a current flows even though they are not in contact, due to the quantum mechanical process of electron tunneling. This current depends exponentially on the electrode separation, and even minute, subatomic variations produce measurable current changes. In 1981, Gerd Binnig and Heinrich Rohrer at IBM in Z€ urich realized that this phenomenon can be used to build a microscope with ultrahigh spatial resolution [1], if one of the electrodes is shaped as a sharp tip and is scanned across the surface of the other (Fig. 1). Moreover, since the tunneling current depends also on the electronic properties of the electrodes, this microscope has the ability to probe the electronic density of states of surfaces at the atomic scale. A few years later, Don Eigler at IBM in Almaden, showed that, due to the extremely localized interaction between tip and sample, it is also possible to use this instrument to manipulate individual atoms, to position them at arbitrary locations and therefore to build artificial structures atom-by-atom [2]. This remarkable achievement brought to reality the visionary predictions made by Richard Feynman in his famous 1959 lecture “There’s plenty of room at the bottom” [3]. The construction of this instrument, dubbed the scanning tunneling microscope (STM), was awarded the 1986 Nobel Prize in Physics and has since then revolutionized contemporary science and technology. The STM has enabled individual atoms and molecules to be imaged, probed, and handled with an unprecedented precision, thereby essentially contributing to our current understanding of the world at the nanoscale. Together with its offspring, the atomic force microscope (AFM) [▶ AFM], the STM is considered
as the main innovation behind the birth of nanotechnology. This entry will start with a discussion of the physical principles and processes at the heart of STM in section Theory of Tunneling. This will be followed by a description of the experimental setup and the technical requirements needed for actually operating such a microscope in section Experimental Setup. Section STM Imaging is dedicated to the most frequent use of STM, namely imaging of surfaces, while section Scanning Tunneling Spectroscopy gives a brief account of the spectroscopic capabilities of this instrument. Finally, section Applications discusses several applications and possible uses of STM.
Theory of Tunneling Figure 2 is a schematic representation of the energy landscape experienced by an electron when moving along the z axis of a metallic-substrate/insulator/metallic-tip tunneling junction. The following treatment can easily be extended to include also semiconducting tips or samples. Usually, the tip and the sample are not made of the same material and therefore have different work functions, fT and fS, respectively. At equilibrium, the two metals have a common Fermi level, resulting in an electric field being established across the gap region and in different local vacuum levels, depending on the difference fT – fS (Fig. 2a). Since the work functions in metals are of the order of several eV, the potential in the gap region is typically much higher than the thermal energy kT and thus acts as a barrier for sample and tip electrons. A classical particle cannot penetrate into any region where the potential energy is greater than its total energy because this requires a negative kinetic energy. However, this is
Scanning Tunneling Microscopy
2303 V=0
a
V>0
b Vacuum level
E
Vacuum level
E
φT
φs EFS
S
φT φs
EFT Tip
EFS
EFT
eV
Tip
Sample Sample
z 0
s
z 0
s
Scanning Tunneling Microscopy, Fig. 2 Energy potential perpendicular to the surface plane for an electron in a tipvacuum-sample junction. z is the surface normal direction, and s is the tip-sample distance. The gray boxes represent the Fermi– Dirac distribution at 0 K. fT,S and ET;S F are the work functions and the Fermi levels of tip and sample, respectively. (a) Tip and
sample in electrical equilibrium: a trapezoidal potential barrier is created. (b) Positive sample voltage V: the electrons tunnel from occupied states of the tip into unoccupied states of the sample. The thickness and the length of the arrows indicate the exponentially decreasing probability that an electron with the corresponding energy tunnels through the barrier
possible for electrons which, being quantum mechanical objects, are described by delocalized wave functions. This phenomenon goes under the name of quantum tunneling. In an unpolarized tip-sample junction the electrons can tunnel from the tip to the sample and vice versa, but there is no net tunneling current. On the contrary, if a voltage V is applied between sample and tip, the Fermi level of the former is shifted by –eV and a net tunneling current occurs, whose direction depends on the sign of V (Fig. 2b). Here the convention is adopted to take the tip as a reference since experimentally the voltage is often applied to the sample while the tip is grounded. If V is the bias voltage, the energy for an electron in the sample will change by –eV, that is, it will decrease for positive values of V. The tunneling current can be evaluated by following the time-dependent perturbation approach developed by Bardeen [4, 5]. The basic idea is to consider the isolated sample and tip as the unperturbed system described by the stationary Schro¨dinger equations:
ðT þ U T Þwn ¼ En wn ;
ðT þ U S Þcm ¼ Em cm and
(1)
(2)
where T is the electron kinetic energy. The electron potentials U S and U T and the unperturbed wavefunctions cm and wn are nonzero only in the sample and in the tip, respectively. Based on this, it can be shown [5] that the transition probability per unit time wmn of an electron from the sample state cm to the tip state wn is given by Fermi’s golden rule: wmn ¼
2p jMmn j2 dðEn Em Þ; h
(3)
S
where the matrix element is: Z Mmn ¼
w n ð~ xÞU T ð~ xÞcm ð~ xÞd 3~ x:
(4)
The d function in Eq. (3) implies that the electrons can tunnel only between levels with equal energy, that is, (3) accounts only for an elastic tunneling process. The case of an inelastic tunneling process will be considered in section Scanning Tunneling Spectroscopy. The total current is obtained by summing wmn over all the possible tip and sample states and by multiplying this by the electron charge e. The sum over
S
2304
Scanning Tunneling Microscopy
the states can be changed into an energy integral by considering the density of states (DOS) R rðEÞ : S ! 2 f ðeÞrðeÞde, where the factor 2 accounts for the spin degeneracy while f, the Fermi– Dirac distribution function, takes into consideration Pauli’s exclusion principle and the electronic state population at finite temperatures. As a consequence, the total current can be written as: Z 4pe 1 I¼ ½ fT ðETF eV þ eÞ fS ðESF þ eÞ h 1 ; (5) 2 T S rT ðEF eV þ eÞrS ðEF þ eÞjMj de where EF is the Fermi energy and the indexes T and S refer to the tip and the sample, respectively. Equation 5 already accounts for the movement of electrons from the sample to the tip and vice versa. Several approximations can be made to simplify Eq. 5 and to obtain a manageable analytical expression for I. If the thermal energy kBT 0. For V < 0 the integrand remains identical but the integration limits become e|V| and 0). In this case, only electrons with an energy differing from EF by less than eV can participate to the tunneling current. This can be directly seen in Fig. 2b for the case of positive sample bias: tip electrons whose energy is lower than ETF eV cannot move because of Pauli’s exclusion principle, while there are no electrons at energies higher than ETF . The main problem in determining expression (5) is, however, the calculation of the tunneling matrix elements M since this requires a knowledge of the sample and the tip wave functions, which can be very complicated. On the other hand, for relatively small bias voltages (in the 2 V range), Lang [6] showed that a satisfactory approximation of j Mj2 is given by a simple one-dimensional WKB tunneling probability. In the WKB approximation [7], the probability D(e) that an electron with energy e tunnels through a potential barrier U(z) of arbitrary shape is expressed as:
2 DðeÞ ¼ exp h
Z
s
1 2
½2mðUðzÞ eÞ dz :
(7)
0
This semiclassical approximation is applicable if (e 3 ac surface effects become negligible. Grazing incidence scattering geometries provide simultaneously enhanced surface sensitivity and limit the unwanted bulk scattering (diffuse or not) that may otherwise overcome the surface or nanostructure feeble signal. It opens additional experimental possibilities as compared to simple specular reflectivity for which incidence and exit angles are kept equal. Grazing incidence is thus a configuration of choice in order to nondestructively investigate nanostructures deposited on a substrate or buried under a substrate. Pattern Analysis The interpretation and simulation of the measured scattering patterns is challenging and requires accurate models and, if size/distance distributions are considered, is demanding tedious computing. However, a complete modeling is not always necessary and the data evaluation can go from quite simple for a crude first approximation to tedious when the full pattern reproduction is sought [21]. A practical illustration of all important terms is given in Fig. 1. In an extremely crude and semi-qualitative approach, and neglecting particle shape, correlation and anisotropy, one can deduce from the position and shape (width) of the diffusion lobes directly an approximation of the interisland distance (D), height (h), and size (R0) as illustrated in Fig. 1. The scattering signal contains a coherent contribution and an incoherent scattering term as soon as the size distribution of the nanostructures is not monodisperse. The coherent term is the product of the Fourier transform of the nanostructure shape with the interference function S(Q//). For the incoherent scattering the two limit cases are (1) the Decoupling Approximation (DA), assuming no nanostructure correlations, and (2) the Local Monodisperse Approximation (LMA), assuming full correlation between nanostructure sizes at a scale corresponding to the photon coherence length.
2327
S
Hard X-ray photons interact weakly with matter and, except for perfect crystals or at glancing angles, the kinematical Born Approximation (BA), that neglects multiple scattering effects, is valid. At angles below 3 ac dynamical effects of reflection and refraction at interfaces become important and have to be taken into account using the popular Distorted Wave Born Approximation (DWBA) [22]. Within surface physics and nano-science two approaches have to be distinguished depending on the level of interpretation that is sought: 1. The more intuitive Effective Layer Born Approximation (ELBA) which applies for isotropic buried islands with simple shapes. The index of refraction is that of the effective layer in which the islands are supposed to be buried. In ELBA the resulting scattering pattern is the weighted sum of the scattering arising from each sub-element since an assembly of objects can be considered as the sum of the individual scatterings as far as the relative intensities remain linked to the total number of electrons in each object within the X-ray coherence length. 2. The DWBA, which applies in the vicinity of critical scattering geometries for islands deposited on a substrate. In the complete treatment, an anisotropic signal originating from island facets can also be considered. The effective form factor corresponds to the coherent interference of four waves corresponding to the four possible scattering events (weighted by the corresponding reflection coefficients) experienced by the incoming and exiting beams on a given island. In both cases (ELBA and DWBA) specular reflectivity may be added, eventually evaluated using the Parratt recursive formalism when multilayer structures are involved. Within the ELBA, to first order and a rapid interpretation strategy assuming weak reflectivity, the scattered intensity I(Q) is proportional to the product of three terms: the form factor P(Q) which is the Fourier transform of the island volume, the interference function S(Q), and the transmission factor T(af) that is supposed to simulate the dependence of the scattered intensity as function of af. The interference function S(Q) is related to the correlation length, which characterizes a distribution of islands on the surface. It is also related to the real space correlation function g(r) via the formula:
S
S
2328
Selected Synchrotron Radiation Techniques
Z SðQÞ ¼ 1 þ 2p rs 0
r
ðgðrÞ 1Þ J0
(5)
ð Qc rÞ:r:dr where rs is the particle density per surface unit and J0 the Bessel function of zero order. Many analytical different pair-correlation functions can be used within models. One may consider, for example, a Gaussian pair-correlation function (illustrated in inset of Fig. 1 – top) given by: 8 0; for 0 r R0 > > > > 2 2 > ðrDÞ < o2 ðR0oDÞ 2 e e gðrÞ ¼ ; for R0 r D1 (6) ðR0 DÞ2 ð D1 DÞ2 > > > e o2 e o2 > > : 1; for D1 r 1 The most reliable input remains the direct evaluation of g(r) from real space measurements made, for example, by complementary microscopy techniques. Within the DWBA the effective form factor corresponds to the coherent interference of four waves corresponding to the four possible scattering events (weighted by the corresponding reflection coefficients) experienced by the incoming and exiting beams on a given island. The DWBA scattering intensity IDW from islands deposited onto a substrate may be summarized as: IDW ¼
1 n2
D ~wD q== ; qz þ Rf ~wD q== ; pz 2 Reð1 nD Þ
2 i i f þR ~wD q== ; pz þ R R ~wD q== ; qz (7)
where nD is the index of refraction of the islands,
~D q== ; q? is the form factor of the islands, w
q== ¼ qx ; qy is the component of the momentum transfer q parallel to the surface, qz corresponds to the net momentum transfer perpendicular to the substrate surface, pz ¼ kzf þ kzi where kzf and kzi are the normal to the substrate plane wave vectors of the incoming and exiting waves. The reflectivity coefficients for the incoming and outgoing waves are denoted Ri and Rf , respectively. The Born approximation is restored with Ri ¼ Rf ¼ 0. The specular reflectance of discontinued multilayer systems with interfacial roughness can be calculated by use of a recursive classical exact layer by layer
method, based on Fresnel coefficients within the Parratt optical formalism and the dispersion model. Interestingly, variations of electronic densities are the only requirement mandatory to enable the observation of GISAXS patterns. Objects (of lower electronic densities) in a larger electron density matrix will yield such signals. The extreme case will then be the presence of structured voids or pores. For reactive interfaces like NiO/Cu(111) [23], self-organized holes in the surface or the presence of Ni decorated corrals lead to GISAXS patterns similar to growing nanoparticles. Size distributions can be included in the pattern evaluation in any framework. The obvious effect of the size distributions is to smooth the scattering pattern. Height distribution and cross correlation between lateral size and height distributions are very difficult to extract from the GISAXS pattern. Examples of Application In 1989, GISAXS was introduced to study the dewetting of gold deposited on a glass surface and later, using synchrotron light metal agglomerates on surfaces and in buried interfaces were considered. Determining the morphology of islands on a substrate, embedded clusters in matrices as well as the fabrication control of nanometer-sized objects can be successfully addressed using GISAXS. It has become a popular technique because it can be applied to a broad range of samples in various sample environments: investigation of samples in standard conditions [21], porous layers, precipitates and quantum dots [24], growing particles in ultra-high vacuum conditions [13, 21], complex nanostructures like carbon nanotubes [25], and shape modifications upon various treatments. The method in itself does not require any particular sample preparation – it is the thin film growth method requiring ultra-high vacuum environment. A major step has been realized in the last decade because of an in-depth understanding of the underlying phenomena and the availability of algorithms enabling the simulation of the experimental patterns. In the most advanced studies coherent GISAXS experiments are considered enabling full particle shape reconstruction [24]. In general, GISAXS can be applied to characterize self-assembly and self-organization at the nanoscale in thin films. The technique becomes highly interesting in situations where charge build-up is a problem, e.g., for
Selected Synchrotron Radiation Techniques
investigating metal/insulating oxide systems. Electron-based techniques are generally hampered by the insulating character of the sample and the charge build-up. Many such systems are characterized by Volmer-Weber or Stranski-Krastanov 3D growth of particles following a nucleation, growth, and coalescence scheme. In GISAXS such a growth sequence leads to diffusion lobes moving toward the reflected beam in reciprocal space with respect to film thickness as illustrated in Fig. 2 for the Co/NiO(111) system. Even without pattern simulation the position and shape changes of the diffusion lobes indicates directly the growth mode. From sequences of images taken at various film thicknesses and sample conditions (like substrate temperatures or quality) one can extract the overall behavior of a metal/oxide interface and evaluate, for example, the effect of defects in the cluster and/or on the substrate. The study of Co deposits, performed in ultra-high vacuum conditions using substrates of different qualities, that is, with variable amounts of nucleation centers is reported in Fig. 2. The experiment was performed on BM32 beamline (ESRF, Grenoble, France) using a beam energy of 10 keV. GISAXS was also combined with grazing incidence diffraction measurements (not shown here) to access the crystallinity of the sample. When the growth is performed at low temperature (T ¼ 450 K) the Co clusters have numerous defects and poor crystalline quality. For Co thicknesses below 0.8 nm the island diameters are similar what so ever is the surface quality or Co island crystalline quality. Thus in the nucleation regime, for cluster sizes below the onset of defects in the clusters, the number of nucleation centers on the surface does not lead to noticeable effects. Above 0.8 nm the diameters fairly diverge indicating that when coalescence starts to play an important role, the island mobility increases with temperature. This effect is consistent with the inter-island distance behavior of the islands. From the island height behavior it appears that the deposition at high temperature (good Co crystalline quality) on a substrate with numerous nucleation centers is equivalent to low-temperature deposition (poor Co crystalline quality) on a high-quality substrate. Such investigations can hardly be made using other techniques. GISAXS is not only highly efficient to investigate in situ grown particles, it can also tackle very complex ex situ samples like carbon nanotubes (CNT) [25]
2329
S
organized in aligned long tubes perpendicular to the sample surface. Each tube includes on its top a Co nucleation particle used to promote the growth of the CNT. Figure 3 reports measurements made on beamline ID01 at the ESRF (Grenoble, France). The experimental pattern includes GISAXS scattering and reflectivity contribution. The calculated patterns required the use of a core-shell C-Co structure inspired from transmission electron microscopy additional measurements. The height/size distribution was fairly large and the DWBA adequate parameters were retrieved from a first ELBA approximation. Such an analysis enables the evaluation of the effects of preparation condition changes like the ammonia content in the reactive gas mixture [25].
High Resolution Micro-X-ray Diffraction (m-XRD) In a local probe diffraction approach (mXRD) is necessary to illuminate with the X-ray beam only a single object. The probe size yields the lateral resolution. As a consequence, the volume of nanostructures probed by the X-rays will be highly reduced (from several 104–106 objects in a “standard” XRD experiment to a single one), thus the detected scattered signal is reduced. Indeed, this one is proportional to the incident X-ray photon flux and the probed volume. In order to compensate for the very small probed volume, the X-ray beam has to be focused to small sizes. Several focusing alternatives are nowadays available. A Short Introduction to Hard X-Ray Focusing Optics The following discussion is centered mostly around hard X-ray (several kilo electron volts energy) optics used at synchrotron sources, although laboratory sources with X-ray beams of several 10 mm are available and can be used for certain XRD experiments with local (lateral) resolution. But a stable, highly brilliant and low divergent X-ray source is a “must have” for obtaining very small and intense X-ray spots. The synchrotron X-ray source is generally situated far from the experimental station (several 10 m) which allows for large demagnification factors for the focusing optics. Consequently, major progress was done at synchrotron facilities, where these focusing devices can be used up to their limits: beam sizes of several
S
S
2330
Selected Synchrotron Radiation Techniques
a
b
0.3 nm
c 50
Inter-lsland distance (Å)
d
Island heights (Å)
40 30 20 T=800K Hi quality NiO T=600K Hi quality NiO T=700K Low quality NiO T=450K Hi quality NiO
10 0
0
5
10
15
20
25
1.5 nm
240 200 160 120 80 dist T= 450 K Hi quality NiO dist T= 600 K Hi quality NiO dist T= 700 K Low quality NiO dist T= 800 K Hi quality NiO
40 0
30
0
5
10 15 20 Co thickness (Å)
Co thickness (Å)
25
30
e Island Diameter (Å)
120
80
40 diameter T=450 K Hi quality NiO diameter T=700 K Low quality NiO diameter T=600 K Hi quality NiO diameter T=800 K Hi quality NiO
0 0
5
10
15 20 Co thickness (Å)
Selected Synchrotron Radiation Techniques, Fig. 2 (a, b) Part of the GISAXS signal recorded on a CCD camera for (a) 0.3 nm and (b) 1.5 nm nominal thickness of Co deposited on NiO(111). The beam energy was 10 keV; the vertical of the CCD is parallel to the sample surface and the horizontal direction is perpendicular to it. The lead well at the bottom of the pictures hides the direct and reflected beams. The dashed
25
30
lines are guided for the eye to evidence the changes in shape and position of the GISAXS signal. (c, d, e) Morphology of Co islands grown on NiO(111): Island height (c), inter-island distance (d) and island diameter (e) with respect to the deposited Co thickness, as extracted from GISAXS measurements. Hi- and low-quality NiO substrates had mosaic spreads of 0.05 and 0.5 respectively
Selected Synchrotron Radiation Techniques
b
0,00
0,09
c
–0,09
Qy(Å–1)
Qy(Å–1)
–0,09
0,00
0,09 0,00
Qz(Å–1)
0,12
S
0,09
Qy(Å–1)
a
2331
0,00
–0,09 0,00
Qz(Å–1)
0,12
0,00
Qz(Å–1)
0,12
Selected Synchrotron Radiation Techniques, Fig. 3 (a) Experimental GISAXS pattern of CNTs prepared with 1% ammonia concentration in the reactive gas mixture, collected at ai ¼ 0.56 . The sample surface is vertical and located on the left side of the pattern. A horizontal beam-stop stops the intense direct and specular beams. (b) ELBA pattern including Co
cylinders, C tubes, Co–C intermixing (Co–C core-shell 14.8 nm average object and intermixed C–Co intermixed tubes); Co and C of radii 2.5 and 3.3 nm respectively, intertube distance 55.6 nm, (c) DWBA pattern using the parameters retrieved from the ELBA model. The color scale is logarithmic and identical for all patterns
10 nm have been demonstrated and highly intense beams in the sub-micrometer size range are available at a number of synchrotrons [1]. Various focusing/collimation schemes for X-rays were proposed (for reviews, see Refs. [26, 27] and references therein): – Refractive focusing optics: compound refractive lenses or planar nano-lenses – Reflective focusing optics: Kirkpatrick–Baez (KB) crossed bended mirrors; Capillaries (mono and poly-capillaries, single and multi-bounce versions) – Diffractive focusing optics: circular, linear, or crossed linear Fresnel Zone Plates, photon sieves – Waveguides – Asymmetrically cut crystals used as beam compressors or collimators Each of these solutions will have its own advantages and drawbacks [26], which will have to be carefully balanced when designing an experiment. Some key parameters to consider are: 1. Photon flux: the small investigated volumes make the scattered signal very small, so an intense and bright X-ray spot is required 2. Easiness in use: some of these devices are simple to use and align, some other require specific setups. The beam deflection from its initial direction (e.g., reflective optics) can also be a parameter which might complicate the experimental setup 3. Achromaticity: is important only for white beam microdiffraction or micro-spectroscopy experiments. The case and the examples detailed hereinafter require monochromatic beam.
4. Focal (working) distance and beam divergence: the use of bulky environments will need a longer working distance of the optical element, in order to allow for the necessary space. This in return might limit the minimum achievable spot size by reducing the geometrical demagnification factor. If a shortworking distance optics is used a smaller spot can be obtained, but there is an increased beam divergence, thus degrading the resolution for strain determination in high-resolution diffraction experiments [14, 15]. A compromise should be found depending on the sample and type of experiment. 5. Stability (mechanical), price, lifetime in the X-ray beam, etc. m-XRD Setup and Experimental Approach Using X-ray focusing optics, it is possible to build various X-ray microscopes working with different contrast mechanisms, or combinations of them: fluorescence, luminescence, transmission, and diffraction. The particular case in which a diffraction contrast is used will be briefly described here. The setup is a combination of an X-ray focusing element and an accurate diffractometer. It consists of a scanner stage used for laterally positioning the sample in the X-ray beam, coupled to precise rotation stages both for the sample angles (incidence, azimuth, tilts) and the X-ray detector (at or close to Bragg diffraction angle) for probing, in the reciprocal space, any desired lattice parameter. Optionally a high-resolution optical microscope pointing at the precise position of the diffractometer’s center (and X-ray beam) is used for sample
S
S
2332
Selected Synchrotron Radiation Techniques Detector
b
{111}
a α 1 = θΒ
[001] Sample
Transm itte y
d beam
x
Q⊥
c
d
Int. (cps)
2θB
Sample with nanostructures
e
3 × 104 1 × 104 –50
0 50 Sample translation (µm)
100
Substrate
Substrate
kf
]
10
[–1 [100]
kf
ki
Q⊥
[110 ]
X-ray scattered beam
Focusing device (Be CRL)
kf ki
ki
Q//
Q//
f
Selected Synchrotron Radiation Techniques, Fig. 4 Principle of the SXDM approach (see also Ref. [28]). (a) Optical image of the SiGe/Si pyramids sample. The inset shows a sketch of the shape of the SiGe square-based truncated pyramids with the (111) factes highlighted. The large rectangular zone shows the area imaged using the mXRD approach (panel f). The optical image, obtained in back-reflected light, makes that the side facets are deflecting the light out of the optical axis of the objective, thus appearing dark. Each small square represents a single SiGe island. (b) Illustration of the setup, ki and kf being the scattering vectors. (c, d) The X-ray spot (represented by the tip of the ki arrow) illuminates the substrate only or a single island
(respectively). Only in the latter case a broad X-ray diffraction signal (different lattice spacing distribution inside the island) is observed, at a different position with respect to the substrate signal. Tuning the diffraction angle to be sensitive in the reciprocal space to a position characteristic to the islands (dot pointed by the tip of the arrow kf) and scanning the sample laterally, higher intensity is observed only when the X-ray spot illuminates an island (panel e). If this is done in both x and y directions, a map of the surface of the sample (SXDM) is obtained (panel f), with enhanced intensity (various shades of grey) when an island is probed. Overlaid to this image, the position of the SiGe islands extracted from the optical image (shown in panel a) is superposed as small squares
pre-aligning and finding of the region of interest. The instrument can function in two ways: one in which the sample surface is imaged point by point like in a scanning probe technique, and a second one in which, for particular regions (microstructures) of the sample, detailed high resolution m-XRD data are recorded. Both approaches will be illustrated hereafter. Figure 4 shows the principle of the method. The sample considered here consists of SiGe islands (square-based truncated pyramids, inset Fig. 4, panel a) [16]. Panel (a) shows an optical image of the sample surface, image on which the individual SiGe islands can easily be identified (dark squares). The X-ray beam is focused onto a small spot on the sample. The elastically scattered X-ray photons are collected by the detector. By rotating the sample and detector angles,
the intensity distribution around different reciprocal lattice point positions can be probed [13, 15]. This is done by appropriately setting the Bragg angles (panel (b), beam impinging incidence angle given by the ki vector, and the position of the detector given by kf). If the sample is laterally translated in the X-ray beam without changing any of the angles, one expects to see differences in the scattered intensity depending on whether an island or the bare substrate is illuminated. In the case of the substrate, the intensity distribution shows a sharp Bragg peak (panel (c)), and a streak perpendicular to the sample surface, the so-called crystal truncation rod (CTR) [13]. If the focused X-ray beam hits an island, additional and broader features will appear in reciprocal space at different positions (panel (d)), since a different lattice parameter, the one
Selected Synchrotron Radiation Techniques
of the SiGe island, is probed. The origin of this different lattice parameter can be multiple, some of the possible causes can be: a various composition, strain, shape, relaxation, etc. Examples of Applications Scanning X-Ray Diffraction Microscopy (SXDM): Imaging with Diffraction Contrast It is now straightforward to perform a “microscopy”like experiment, the result being an image of the surface of the sample. The diffraction angles are fixed at the expected values to record diffraction from the crystalline planes inside the nanostructures and the position of the sample is scanned laterally while recording, at each point, the scattered intensity (Fig. 4 panel (e)). The resulting contrast is diffraction-based and is related to the presence of islands. If the mapping is performed in both directions (x and y), the resulting map is the result of the Scanning X-ray Diffraction Microscopy (SXDM) approach. Panel (f) shows an SXDM image (color scale) superimposed to the optical image of the same region of the sample (from panel (a)). The agreement is very good and validates the point-by-point microscopy approach, showing that it is possible to find precisely a particular object: in this case, the region in the vicinity of the “defect” exhibited as the vertical zone without any SiGe islands. Markers could also be envisaged. Indeed, it is of utmost importance not only to be able to measure an isolated object out of an ensemble, but to precisely know which particular object was measured: the properties of these SiGe islands can change dramatically from object to object and can be correlated with their shape obtained by ▶ SEM. The second example below is showing the power of the method and its sensitivity not only to (local) crystalline changes, but to thickness as well. The sample (metal oxide Magnetic Tunnel Junction, MTJ) consists of a stacking of several metallic and oxide layers on which structures of lateral size in the 10–100 mm range were made by optical lithography [29]. This particular sample suffered a mask misalign during one of the lithography steps. Well that this could be easily detected by simple optical microscopy examination of the surface, mXRD can also bring some information about how this happened and which layers suffered; this sample is a good example to show how the different structures can be imaged and differentiated. The corresponding XRD spectra recorded at these different
2333
S
locations are shown in Fig. 5 panel (b). Signal originating from the various crystalline structures of the layers (Pt, Fe3O4, and Co) can easily be differentiated and used as probe in imaging the surface of the sample, after tuning the diffraction angles to these positions (as highlighted by the colored arrows in panel b). The contrast on the surface images changes accordingly. Moreover, sensitivity to thickness is achieved: the arrow pointing to the maximum of the Pt signal also corresponds to a minimum contrast (interference) for the Fe3O4 layer. This explains the reversed contrast obtained in the left bottom panel (a) of Fig. 5. High-Resolution X-Ray Micro-diffraction (HR – MXRD) and Combination with SEM
With the approach depicted above it is possible to measure not only an individual object but a very particular one. The case of the SiGe islands will be used as example. On the SXDM image of the surface of the sample (Fig. 4), individual objects are chosen and measured in HR-mXRD. Once the lateral positions of the sample are fixed such that the X-ray spot illuminates the object of interest, the angles (sample and detector) are scanned in order to describe the reciprocal space in the vicinity of the chosen Bragg position. The recorded data show characteristics allowing to be grouped in three categories corresponding to three types of SiGe objects: the scattered intensity might vary slightly for the same type of object (depends on the particular imaged object), but, for one family the same characteristics are found. The differences are from one type to another. Figure 6a–h shows such measurements. The differences from the three mentioned types are visible both close to the (004) and (115) Bragg peaks. Since these mXRD measurements were performed at known positions of the sample (cf. image of the surface, as described in Fig. 4), it is possible, at the end of the diffraction experiment, to image in detail precisely the objects in question. The corresponding SEM images of the objects are reported as insets in Fig. 6a–h. Comparing the scattered intensities measured for SiGe pyramids and flat islands, some major differences can be highlighted: the intensity distribution close to the SiGe peak is changing. If the truncated pyramid exhibits scattered intensity concentrated around two positions (two maxima, labeled (1) and (2) in the figure), the flat islands show distributions mostly around the position (2). In the case of the (004) map, the position (1), closer to the Si Bragg
S
S
2334
Selected Synchrotron Radiation Techniques
b 106 a Fe3 O4 (111)
Intensity (arb.units)
105
α-Al2O3 (0006) (substrate)
Pt(111)
104
on junction between junctions / on track empty / substrate area
103 102
Co(0002)
101 100 20
21
22
23 24 theta(°)
25
26
Selected Synchrotron Radiation Techniques, Fig. 5 (a) Scanning mXRD image (SXDM) of the MTJ surface showing defects due to mask align in the optical lithography process. Areas where only the Pt layer (1) remained after the lithography process or the full MJT (2) is present are shown. (b) mXRD data (y-2y geometry) at different locations on the sample: ▼ on
a MTJ; ○ on contact track (Pt buffer); ● on the empty area (substrate). The arrows point to reciprocal space positions characteristic of the different crystalline structures (from left to right, Pt, Fe3O4, and Co), positions for which SXDM images of the sample surface (shown in panel a) were performed (top Fe3O4, left Pt and right Co respectively)
location, shows the presence of a lattice parameter (out of the surface plane) closer to the Si one, that is, indication of a lower Ge concentration. The signal at position (2) indicates a higher Ge concentration and a larger lattice parameter. The availability of two types of maps, around symmetric (004) and asymmetric (115) Bragg peaks, brings complementary information: the peak’s position on the (115) maps has also an in-the surface plane component of the SiGe lattice parameter [28]. This example shows the need and the power of combining several methods on the very same micro/ nano-object, for detailed characterization and modeling: the shape and the sizes are deduced from SEM images and then injected into a model used to simulate the diffraction data (finite element methods for strain distribution, then semi-kinematical scattering theory). Details can be found in references [14, 15, 26, 28].
in situ combination of a mXRD experiment with an Atomic Force Microscope (AFM) [30, 31]. This approach allows measuring in the same time and for the same individual small object, the mechanical behavior, the internal strain, and the response to external stress. Such measurements, especially when approaching the nanoscale, are of the highest importance for material characterization, both in the elastic and the plastic deformation regimes. The SiGe island sample described before was used as well as a model sample to prove the principle of this experiment. The experimental setup is schematically shown in Fig. 7a: an ▶ AFM is mounted on the diffractometer and the focused X-ray beam is aligned with the apex of the tip. During point by point mapping of the sample surface with the X-ray spot (SXDM), the electron current induced in the AFM tip by the X-ray beam is also recorded. The resulting two contrast images (acquired simultaneously) are then combined (green and red colors respectively) to yield the image shown in panel (b). Panel (c) shows precisely the same sample area imaged in SEM (ex situ) previous to the experiment – some of the missing islands in panel (b) were “destroyed” by plastic deformation attempts during the first part of the AFM experiment.
In Situ Combination of mXRD with Atomic Force Microscopy (AFM)
The previous example showed the combination of mXRD with (ex situ) SEM (▶ SEM) images, yielding to an understanding of the structure of microscopic semiconductor samples. The next example shows the
Selected Synchrotron Radiation Techniques
2335
S
Selected Synchrotron Radiation Techniques, Fig. 6 (a–h) Reciprocal space maps (logarithmic intensity scale) in the vicinity of the (004) and (115) Bragg peaks, recorded using a X-ray focused beam. The representative SEM image for each type of object is shown in the inset. The Si substrate Bragg peak is
indicated by (*). (i) Image of the sample surface, performed with diffraction contrast (SXDM), compared to a SEM image of the same area, with particular objects identified (circle). (j) SEM image (details) of the sample, around the region with the “vertical path” free of SiGe islands, in the center of the image
The elastic properties of individual SiGe islands while pushing on them with the AFM tip are well documented in Refs. [30, 31]: the applied force (and thus the pressure) can be extracted from the changes in the resonance frequency of the AFM tip; the lattice parameter (average value over the whole island) is extracted from the Bragg position in mXRD experiments, at each applied external pressure. It was also possible to access the plastic regime. Figure 7 (panels f and g) shows mXRD measurements (reciprocal space maps) close to the (004) Bragg position for an SiGe pyramid before and after pushing. The insets show the corresponding SEM images of the island and AFM tip before and after the experiment. Well that the AFM tip bends, the applied pressure can be large enough to completely destroy the SiGe island, which results in a major change of the mXRD signal.
This example validates the use of the coupled mXRD and AFM not only for studying the elastic properties of small structures, but also for in situ indentation of materials (▶ Nanoindentation) (a stiffer tip, and with an adapted shape for indentation, should then be used). In such an indentation experiment, by applying laterally resolved mXRD, the resulting strain field around the indented area could be mapped and modeled.
X-Ray Photo Emission Electron Microscopy XPEEM Instrumentation and Operating Principle XPEEM microscopy is a parallel imaging method that combines X-ray electron spectroscopy and electron microscopy. It is based on soft X-ray-in/Electrons-out
S
S
2336
Selected Synchrotron Radiation Techniques
a Focusing device (Be CRL)
b
c
Tip
d
C
C D
D
A
G
G
B
E
Tip Shadow
F
300 nm
A E
B
F
5mm
e
5mm
g
f 5mm
5mm
q001 (Å–1)
4.620
4.615
SIGe peak 4.610 –0.01
0.0
0.01
q110 (Å–1)
Selected Synchrotron Radiation Techniques, Fig. 7 (a) Cartoon of the combination of mXRD with AFM (simplified setup). Once the X-ray focused beam is aligned with the tip of the AFM, it is used to “image” (high-resolution XRD) the object placed beneath the tip. (b) superposed maps (images) of the sample, with the X-ray beam aligned on the AFM tip and a single SiGe island placed just beneath the tip: in green the detector signal (sensitive to SiGe) is recorded, while in red is the electron current induced in the tip when this one is hit by the X-ray beam. (c) SEM image of the sample showing precisely the region measured in panel (b). The different areas on the sample are labeled. Note the absence of the islands labeled “G” – during the experiment and previously taking this image, these islands
–0.01
0.0
0.01
q110 (Å–1)
were “smashed” (destroyed) or laterally displaced using the AFM tip. (e) SEM image of an SiGe single pyramid of 1 mm lateral size. (e) Image of the blunt tip used for pushing experiments. Note the relatively large apex radius of curvature (with respect to standard AFM tips) – this feature makes easier pushing on the small objects. (f) Reciprocal space map (close to the (004) Bragg peak) for a single SiGe island. The Si substrate Bragg position is not shown on the map (qSi(004) ¼ 0.4628 nm1). (g) Same map after pushing with the AFM and reaching the plastic regime (destroying the SiGe island). The intensity distribution changed completely. The insets show the SiGe island and the AFM tip (SEM images) before and after plastic deformation. The electron yield images of the tip are also shown for each case
Selected Synchrotron Radiation Techniques
2337
S
sample x-ray beam
objective illumination aperture
selected area aperture contrast aperture
electron gun
screen
slit
projector transfer condensor field lenses intermed. sector field lenses
analyzer
Selected Synchrotron Radiation Techniques, Fig. 8 Schematic sketch of the LEEM-PEEM microscope
principle and was originally pioneered by B. Tonner. The sample is illuminated by a monochromatic soft X-ray beam (90 97 100 >98 100 100 100 100 97 >95
Self-Assembly for Heterogeneous Integration of Microsystems, Table 1. Survey of current self-assembly process. Components are square in shape except where marked by C , and R, which are circular and rectangular respectively. Part sizes of components are taken to be the length for square and rectangular parts, and the diameter for circular components
Self-Assembly for Heterogeneous Integration of Microsystems 2357
S
S
S
2358
Self-Assembly for Heterogeneous Integration of Microsystems Au ring contact PR lift-off layer
d a
GaAs LED
10 μm GaAs LED structure GaAs substrate
Wax
1 μm AlAs Intermediate substrate
Au
b GaAs LED
Photoresist
e
1 μm AlAs
As Ga D LE
Ethanol
GaAs substrate
c
Si substrate GaAs substrate hν GaAs LED
f Wax
Intermediate substrate
GaAs LED Si substrate
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 3 Schematic diagram of the fluid selfassembly process: (a) Molecular beam epitaxy (MBE) grown structure with 1 mm AlAs etch-stop layer, (b) trapezoidal GaAs mesa definition, (c) bonding to intermediate substrate with wax,
(d) top-side ring contact metallization, (e) solution containing the GaAs blocks dispensed over patterned Si substrate and (f) Si substrate with GaAs, light-emitting diodes integrated by fluidic self-assembly. (Reprinted with permission from [5], # 1994 IEEE)
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 4 Scanning electron microscope (SEM) picture of Yeh’s devices: (a) a device assembled using fluidic self-assembly techniques – the metallic connections are
deposited over the device such that mechanical and electrical connection is established. (b) Devices placed on a US dime for scale. (From [6], reprinted with permission from Alien Technology)
When assembly was observed to be complete, the temperature of the self-assembly setup was lowered to room temperature, thereby solidifying the molten solder, establishing electrical and mechanical
connection between the template and the discrete components. While the fluidic flow transportation mechanism is similar to Yeh et al., the alignment of microcomponents is achieved through shape-matching
Self-Assembly for Heterogeneous Integration of Microsystems
and solder capillary action. Finally, binding of the discrete components was achieved through the solidification of the molten solder bumps. Final products are shown in Figs. 7 and 8. Parviz et al. added programmability to their fluidic shape-matching self-assembly process with a technique that involves the use of photosensitive materials to control the access of binding sites to Freestanding Microcomponents
Ethylene Glycol & HCI Plastic Template
Fluid Flow
Self-Assembled Components
Interconnect Empty Binding Sites 20° to 60° Tilt
S
microcomponents being flowed across the template [8]. Programmable binding locations were created by depositing photoresist within receptor sites that uses the principle of shape matching to capture microcomponents that are flowed over them. To activate a group of receptor sites, the region of interest is selectively exposed to ultraviolet radiation to remove the blocking photoresist material, thereby availing said sites to capture incoming components. By successive steps of site activation and component flowing, different components can be assembled onto a single type of receptor sites (Fig. 9). It is instructive, at this point, to definite “programmability” in the context of micro self-assembly as the ability to direct different reliably reproducible outcomes with the same set of components in a selfassembly process. Programmability, which will be revisited in other works, will add to the appeal of self-assembly in industry.
Fluidic Transport–Capillary Forces Alignment
Heat to 70 °C
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 5 The heterogeneous self-assembly process. Microcomponents are introduced over a template submerged in a liquid medium and moved with the fluid flow. Self-assembly occurs as microcomponents first fall into complementarily shaped wells and then become bound by the capillary forces resultant from a molten alloy. (Reprinted from [7], with permission from IOP)
a
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 6 Details of the selfassembly process for a single microcomponent. (a) The microcomponent approaches a binding site with complementary shape. (b) The microcomponent is held by capillary force resultant from molten-alloy-bridging the metal pads positioned on the microcomponent and on the template. (Reprinted from [7], with permission from IOP)
2359
Capillary forces dominate inertial forces and gravity as one approaches submillimeter scales, given that capillarity scales with length, while gravity scales with length cubed. There are many studies that concentrate on the alignment properties of capillary forces instead of an entire self-assembly process as defined in Fig. 2.
Silicon Microcomponents Au/Cr Interconnect
SU-8 Bin
din
Plastic Substrate
gS
ite
We ll
b SU-8
Low Melting Point Alloy
Embedded TiW/Au Interconnect
S
S
2360
a
Self-Assembly for Heterogeneous Integration of Microsystems 250 μm
b
250 μm
TiW/Au Interconnect Resistors Solder
FETs
Binding Sites
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 7 Self-assembly of field effect transistors (FETs). (a) Optical microscopic image of a plastic substrate with empty binding sites (two outlined with white lines for clarity);
(b) The template after completion of the self-assembly process showing the position of FETs and diffusion resistors. (Reprinted from [7], with permission from IOP)
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 8 (a) A glass template after the conclusion of the self-assembly process. The micro (m)-LEDs are assembled on the template. (b) 6 m-LEDs are turned ON on the same glass template via the application of a 5 V bias. (c), (d) Close-up images of 1 m-LED turned OFF and ON, on the template under
an optical microscope; (e) A flexible plastic template after the completion of the self-assembly process; (f, g) The same plastic template, bent over itself, with m-LEDs OFF and then ON upon the application of a 4 V bias. (Reprinted from [7], with permission from IOP)
Notably, work from Koyanagi et al. focuses on the aligning and bonding of silicon chips on microfabricated silicon pedestals with corresponding dimensions [9] (Fig. 10).
Droplets of an aqueous solution of hydrofluoric acid (HF) were placed on the pedestals before introducing silicon chips with a multi-chip vacuum pick system. The faces of the chips and the pedestals were treated to
Self-Assembly for Heterogeneous Integration of Microsystems
a
c
2361
e
S
g
Fabricated receptor site Assembled component 4-notch
b
d Blocking resist
Available receptor site
Assembled component 2-notch
f
h Assembled component 0-notch
Available receptor site Assembled component 1-notch
Available receptor site
Blocking resist
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 9 Optical microscope images of a sequence of events during the self-assembly process of four different microcomponent types. (a) Fabricated receptor site wells and C-shaped traps on a plastic template prior to selfassembly; (b) Positive photoresist was patterned on the template to block all the receptor sites, except the four receptor sites located on the bottom left. (c) The first type of components (circular 4-notch) was assembled within the 2 2 parallelogram pattern of available receptor sites. (d) A thin layer of photoresist
was coated on the whole template to fix the assembled microcomponents. The blocking resist was then removed from the four top right receptor sites designated for the second type of microcomponents. (e) The second type of components (circular 2-notch) was assembled in available receptor sites. (f) The third type of components (circular 1-notch) was assembled in the bottom left corner of the image in a similar fashion. (g) The resist was removed from the remaining four receptor sites on the top left. (h) Final type of components (circular 0-notch) was selfassembled. (Reprinted from [8], with permission from MRS)
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 10 Self-alignment process from Koyanagi et al. (a) Introduction of chips using a multi-chip vacuum pick system. (b) Chips self-aligned after placement on bonding areas. (Reprinted with permission from [9], # 2008 IEEE)
be hydrophilic so that the droplet underneath the chips will wet both surfaces and self-align the chips to the pedestals underneath to minimize the interfacial energies. Alignment will improve as the HF solution evaporates, and eventually dissipates entirely, while bonding chips to pedestals permanently with strong SiO2–SiO2 bonds.
S
There are many other works titled “self-assembly” that deal exclusively with alignment rather than the entire self-assembly process which were presented in Fig. 2. “Self-alignment” is probably a more appropriate title for such studies, drawing the distinction between investigations focused on alignment and works that also include a stochastic transportation
S
2362
Self-Assembly for Heterogeneous Integration of Microsystems
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 11 Optical micrographs of capillarydriven self-assembly of flat microparts: (a) Square parts on
quartz substrates in ring and grid configurations and (b) correct and wrong alignments for semicircle- and comma-shaped binding sites. (Reprinted with permission from [10], # 2001 IEEE)
mechanism that brings discrete components to the vicinity of the intended assemblies. Srinivasan et al. demonstrated a complete selfassembly technique with fluidic component transportation and capillary force capture and alignment [10]. They patterned a substrate with an array of hydrophobic self-assembled monolayer (SAM)–coated gold-binding sites. By inserting the substrate into water through a film of hydrophobic adhesive floating on the water surface, the hydrophobic binding sites were coated with the adhesive (Fig. 11). Microparts fabricated from silicon-on-insulator (SOI) wafers were then introduced through a pipette toward the substrate in water. When the hydrophobic pattern on the microparts came into contact with the adhesive, the parts were pulled to the hydrophobic binding sites and self-alignments occurred spontaneously. Finally, when all sites were occupied, the adhesive was polymerized by heat (thermoset) or UV radiation, depending on the type of adhesive used, bonding the chips to the substrate permanently. Binding sites of shapes with in-plane rotational symmetries such as squares gave alignment yields of up to 100%, with translation and rotational misalignments of less than 0.2 mm and 0.3 , respectively. Component–
binding site pairs without in-plane rotational symmetries (aimed at specific in-plane alignment) such as semicircles and commas gave alignments yields of approximately 30–40%. Jacobs et al. developed a similar self-assembly process that uses low-temperature melting solder to mount LED arrays on flexible cylindrical templates [11]. The assembly template was patterned with copper squares, and a simple dip-coating process applied molten solder on these copper squares, given that copper has good wetting characteristics for solders. The dip-coating process was performed in an acidic aqueous environment to prevent oxidation of the solder surface. Hundreds of LEDs and a flexible assembly template were placed inside a vial; the vial was filled with water and heated to a temperature above the melting point of the solder. The LEDs were tumbled inside the vial, allowing for the LEDs to come into contact with the molten solder, be pulled to a copper square by the solder, and self-aligned (Fig. 12c). The cathode and anode of the LEDs were distinguished by a small round contact and a larger square contact (the size of the copper squares on the template), respectively, on two faces of the discrete LEDs (Fig. 12a). The agitation intensity of the tumbling process was tuned to
Self-Assembly for Heterogeneous Integration of Microsystems
S
2363
ELEMENTS
a
Au contact binding site
d
‘receptor’-array
b solder copper
support
support (polymide, transparent)
e c
STEP 1: ASSEMBLY
STEP 2: ALIGNMENT
top electrode water
tumbling
H2O 90°C
solder bump
f
−
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 12 Procedure used to assemble a functional cylindrical display. (a) Top and bottom views of an LED segment that has two contacts: a small circular gold contact (cathode) on the front, and a large square gold contact (anode) on the back. (b) Array of solder-coated copper squares supported on a flexible substrate; these squares are of the same size as the anodes of the LEDs and act as receptors for the LEDs during the self-assembly. (c) The components are tumbled in
+ V
a vial at a temperature above the melting point of the solder. (d) Two-dimensional array of the assembled LEDs. (e) Alignment of a top electrode: The copper wires of the top electrode and the cathodic contacts on the front of the LEDs were first dip coated with solder. The array of wires is prealigned with the array of cathodic solder bumps. At a temperature above the melting point, electrical connections form and the structure self-aligns. (f) Test of the self-assembled display-prototype. (From [11], reprinted with permission from AAAS)
S dissociate assemblies between the solder and the smaller round contact while not severe enough to affect LEDs that were caught with their square contact face. This gives the assembly process side-selectively. A permanent bond between each LED and the template was established when the template was allowed to cool, solidifying the solder. To make the LED array functional, a transparent film patterned with electrical circuitry was aligned and bonded to the assembled array, establishing the electrical connections to the cathodes of the LEDs. By activating the LEDs (Fig. 13), the defect rate was visually determined to be approximately 2%.
Jacobs et al. also worked on self-assembly based on the delivery of components on a fluid-fluid interface [12]. Components were fabricated out of silicon and SU-8, with one face coated with gold. The gold surfaces were treated with a mercaptoundecanoic acid (MUA) SAM to render them hydrophilic. The silicon faces were treated to be hydrophobic using 3-glycidoxypropyltrimethoxysilane (GPTMS), creating an orientation preference for components when they are introduced at a silicone oil–water fluid-fluid interface (SU-8 surface are hydrophobic and thus required no further surface modification). By mildly agitating the system, components at the silicone
S
2364
Self-Assembly for Heterogeneous Integration of Microsystems
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 13 (a) Photograph of the display after self-alignment of the top electrode; (b–d) Photographs of the operating display after the alignment of the top electrode. The display contains 113 LEDs that are assembled in an interleaved
fully addressable array with eight columns of eight receptors interleaved with seven columns of seven on the bottom electrode that connect to 15 rows of crossing copper wires on the top electrode. (From [11], reprinted with permission from AAAS)
a
b Silicone Oil
Compaction -400mJ/m2 Glass Vial
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 14 (a) Procedure of surface tension–directed selfassembly at a liquid-liquid– solid interface. SEM of (b) SU-8 (20 mm side length) and (c) Si chiplets (20 mm and 60 mm side length) assembling in regular arrays and arbitrary text patterns (insets). (Reprinted from [12], # 2010 National Academy of Sciences, USA)
Si
Au
c -55mJ/m2
-90mJ/m2 -7*5=
-35 mJ/m2 Water
Liquid Solder
Temperature: 90 °C
oil–water interface will predominantly orient themselves such that the hydrophilic gold surface will face toward water, and the hydrophobic silicon/SU-8 will face the direction of the silicone oil. Two types of substrates have been used: silicon and flexible propylene terephthalate (PET). Copper contact pads, corresponding to the sizes of the components, were patterned on the surface. A dip-coating process
was used to deposit low-melting-temperature solder onto the copper pads. The transfer of the components is shown in Fig. 14. A container containing silicone oil, water, and components on the interface between the two fluids was first heated to a temperature beyond the melting point of the solder. Substrates were pulled upward through the liquid-liquid interface, allowing the components to
Self-Assembly for Heterogeneous Integration of Microsystems Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 15 Parts rearrangement by agitation. Darker squares are hydrophilic silicon surfaces
2365
a
S
b
Pull-up and pause
Substrate/water interface
c
d
After agitation
Example of 100% assembly
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 16 The experimental setup consists of a water container, linear electromagnetic actuator, dip coater, assembly substrate, magnets, and test parts floating at the air-water interface
come into contact with the molten solder. Upon contact, the molten solder will pull components in, transferring the components onto the substrates. While it is possible to deliver components to over 90% of the copper pads on a substrate in a single pass, multiple passes were required to achieve full coverage. Using
S
this method, chips 20 20 10 and 60 60 20 mm3 in size were assembled onto rigid and flexible substrates at reported speeds of up to 62,500 chips in 45 s. Bo¨hringer et al. studied the assembly of thin microcomponents in a fluidic environment, delivering components at the air-water interface [13].
S
2366
Components (1,000 1,000 100 mm3) with one face coated in gold are fabricated in silicon. Gold-coated silicon assembly templates were prepared with square openings, exposing silicon, that were designed to be slightly larger than the lateral dimensions of the components. All gold surfaces were then treated to be hydrophobic using dodecanethiol SAM. Components were introduced onto a water surface. By applying controlled agitation, the components will preferentially orient themselves with their gold faces upward due to the surface treatment. The assembly template was then pulled up, at an angle, through the air-water interface, in the immediate proximity of the floating components. When the level of the assembly template reaches the top edges of a row of assembly sites (Fig. 15a), controlled agitation was applied to cause the components to arrange themselves such that the edge of a single part was pinned to the edge of a single site (Fig. 15b, c). At this point, the assembly template was pulled up, allowing an entire row of parts to be transferred and self-aligned to the template. By repeating this pull-up, pause and agitate cycle, entire assembly templates can be perfectly filled. The most significant aspect of this work lies in the aspect ratio of the components being self-assembled; no other self-assembly technique has been demonstrated to work well with very thin components. The programmability of this process was demonstrated by the ability to assemble only on alternate rows of binding sites. After a single (first) row of parts have been transferred onto the assembly substrate, one can avoid assembling on the subsequent row by agitating the water surface while the substrate is extracted through the air-water interface, bypassing the second row and stopping at the assembly position of the third. Using similar methods, arbitrary row assemblies can be reproduced reliably.
Fluidic Transport–Capillary/Magnetic Forces Alignment By depositing nickel squares near the edges of chips, and positioning a magnet underneath each binding site underneath the assembly substrate, Bo¨hringer et al. extended the previous fluidic transport–capillary forces alignment self-assembly method to include orientation specificity (Fig. 16).
Self-Assembly for Heterogeneous Integration of Microsystems
a
before agitation
b
after agitation
c
assembly
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 17 Assembly process of 2,000 2,000 100 mm3 components on the assembly substrate: (a) A binding site does not yet have a corresponding component, and another has two test parts in its vicinity. (b) After applying specific Faraday waves using agitation, every binding site gets a single component. (c) Two rows of components assembled onto the assembly substrate
Similar to the previous method, controlled agitation of the water surface ensures one-to-one addressing of parts to binding sites. During this agitation, the interaction of the magnetic forces between the magnets underneath each binding site and the ferromagnetic nickel squares ensures also that chips approach the binding sites with the nickel-patterned edge, allowing for orientation specificity previously unachievable (Fig. 17).
Mechanical Agitation Transport–Capillary Forces Alignment For discrete components and/or substrates that are adverse to fluidic environments, there are also self-assembly processes that are conducted in dry or “semidry” [14–16] conditions. Dry environment
Self-Assembly for Heterogeneous Integration of Microsystems
a
assembly template
transfer substrate
b
2367
S
h
d
e
microparts
f final substrate
c
g
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 18 Full assembly process flow (a–g): (a) Mount assembly template on a gold-patterned transfer substrate. (b) Assemble parts. (c) Remove template. (d) Apply moisture. (e) Vibrate the setup gently to have the parts fall on
hydrophilic side and self-align. (f) Attach parts to destination substrate/device. (g) Complete the process by removing the assembled substrate/device from transfer template. (h) Top-down view of assembly template placed over transfer substrate. (Reprinted with permission from [15], # 2009 IEEE)
self-assembly processes predominantly require the use of mechanical vibration/agitation to distribute the microcomponents. Bo¨hringer et al. devised a suite of self-assembly methods to be performed in an air environment that uses a two-step assembly process consisting of a shapematching step and a capillary force–driven alignment step [14–16] for square silicon parts as well as surface mount technology (SMT) passive devices. The latest iteration in the technique for the assembly of silicon chips is shown in Fig. 18. A silicon fabricated assembly template was first aligned on top of a transfer substrate. The transfer substrate was fabricated by first depositing gold onto a silicon wafer. Square openings (silicon) corresponding to the size of the microparts were subsequently developed, and the gold surfaces were treated to be hydrophobic with a SAM coating. Each aperture in the assembly template addresses a single silicon opening underneath (Fig. 18h), and can only contain a single discrete component standing on its side. Square-shaped silicon microcomponents with one gold-coated (hydrophobic SAM-treated) side were delivered to every aperture on the assembly template. When the delivery was completed, the assembly template was removed, and steam was introduced to the system to form water droplets on the hydrophilic
silicon surfaces. When the droplets become large enough, they will exert enough force to pull the silicon face of the microparts flat onto the transfer substrate, and self-align the components to the squares. The aligned microcomponents can then be transferred onto a destination substrate to be permanently mounted. Similar techniques have been tested with parts ranging from 370 370 150 mm3 to 790 790 330 mm3 in size. Bo¨hringer et al. also developed a system of control for the precise manipulation of microcomponents on a vibrating surface [15]. Parts can be made to jump or walk (where parts seem to crawl across the assembly template) in predictable directions and speeds. Using these controllable part-motion modes, the transportation phase of the self-assembly process can be programmed to deliver components to arbitrary regions of apertures or even empty out filled apertures. In the case of a programmed “feedback-driven” delivery process [15] (Fig. 19), visual feedback is used to identify empty apertures and direct excess components toward them, guaranteeing the complete delivery of components. Finally, Bo¨hringer et al. also adapted the templatebased methodology (complete with the feedbackdriven assembly process) to the assembly of surface mount technology (SMT) thin-film resistors and
S
S
2368
Self-Assembly for Heterogeneous Integration of Microsystems
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 19 Stages of a feedbackdriven assembly: (a) shows the instance when jumping mode (with a top left corner bias) is switched to a walking mode with part-motion direction as indicated by the white arrow in the box. (b) and (c) show the progressive filling of the entire assembly area. Right after (c), when assembly-percentage, P(t), has plateaued, a walking mode in the upward direction as seen in (d–g) is activated, moving excess parts below the assembly area to the empty sites. After achieving 100% assembly in (h), a walking mode directed toward the lower-right corner is activated, moving excess parts away from the assembly area, finishing the process in (i); note: size-scale provided in image (i). (Reprinted with permission from [15], # 2009 IEEE)
monolithic ceramic capacitors of the 01005 standard (0.01600 0.00800 , 0.4 0.2 mm) [16]. In this instance, two templates were used, one stacked on top of another. Template 1 (Fig. 20) ensures that only one component will be allowed into template 2, and template 2 guides the in-plane orientation of the component. During the solder reflow process, the molten solder wets the metallized contact-ends of the SMT components and aligns them to the contact pads on the target substrate. At the end of the solder reflow, mechanical and electrical bonds between components and substrate were achieved.
Mechanical Agitation/Magnetic Forces Transport–Shape-Matching Alignment A dry environment self-assembly technique combining transportation via mechanical vibration and magnetic attractive force with shape-matching alignment
was devised by Ramadan et al. [17] (Fig. 21). A host substrate, featuring physical recesses, was superimposed to a master array comprising embedded NdFeB magnets, whose position exactly matched that of the recesses. Target chips, bearing a CoNiP soft magnetic film on their nonfunctional side, were distributed over the substrate, which was mechanically vibrated and tilted. This motion causes chips to be distributed across the entire substrate, until being collected in an enclosure at the lower end of the tilt. The amplitude of the substrate vibration and the thickness of chips were tailored so that the only stable configuration possible for the chips to be assembled is to have the functional side facing upward. The in-plane alignment of the chips was controlled by shape matching between the chips and the recesses of the host substrate. After assembly, the master array could be removed and reutilized while the assembled substrate could be further processed. An assembly of
Self-Assembly for Heterogeneous Integration of Microsystems
2369
a i
S
b
ii SMT component
Contact Pads
template 1 Assembly Stack template 2
Empty Binding Sites
target substrate iv
iii
solder bumps
c Assembled Sites
4mm
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 20 (a) Assembly process: (i) Walking mode [3] component delivery performed on assembly stack in Configuration 1; (ii) a single component is captured by template 1 near the binding location; (iii) Configuration 2 – component is allowed to drop into template 2; (iv) slightly agitating the system
aligns the component to the orientation of the aperture on top of the binding location on the target substrate. After step (iv), solder reflow is performed to bond the component to the target substrate mechanically and electrically. Assembly of SMT passive components onto a circular test device: (b) before assembly; (c) after assembly. (Reprinted with permission from [16], # 2009 IEEE)
a i
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 21 Assembly process (a): (i) Host substrate aligned to the master magnetic array, (ii) chip seeding, (iii) applying vibration, and (iv) chip final alignment and master magnetic array removal. Photo of 1 mm2 assembly (b) 2 mins and (c) 5 mins into the vibration process. (Reprinted with permission from [17], # 2007 American Institute of Physics)
b ii
c iii
2,500 chips in 5 min with 97% yield was reported for this technique. Fischer et al. proposed a scheme to implement through silicon vias (TSVs) by using magnetic force [18]. High-aspect-ratio holes are first etched on
iv
a silicon substrate using deep reactive ion etching (DRIE). Commercially available nickel wires (length: 350 mm, diameter: 35 mm) were deposited on the surface of the silicon substrate. A magnet was introduced at the bottom of the substrate, causing
S
S a
2370
Self-Assembly for Heterogeneous Integration of Microsystems
b
Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 22 Nickel wire assembly process: (a) nickel wires are introduced to a substrate etched with corresponding holes, (b) a magnet is placed under the substrate, causing the nickel wires to stand upright due to magnetic forces, (c) the magnet is moved laterally under the substrate, moving
c
d
wires to unoccupied holes and trapping them, (d) BCB is applied over the assembled wires. Processing steps after (d) include curing the BCB, polishing the substrate to expose the ends of the nickel wires, and depositing metal lines for electrical connection. (Reprinted with permission from [18], # 2011 IEEE)
TSVs are vertical electrical connections passing through die layers. They are important for heterogeneous integration solutions involving the stacking of dies. While seemingly out of place with previously mentioned assembly methodologies for chips/dies and passive devices, it is significant to show that there is a viable self-assembly solution for TSV placements, demonstrating that there are self-assembly solutions for every aspect of the device assembly process.
Conclusion Self-Assembly for Heterogeneous Integration of Microsystems, Fig. 23 SEM image of a 30 30 array with a pitch of 120 mm of nickel wires placed in via holes prior to the filling of BCB. (Reprinted with permission from [18], # 2011 IEEE)
the wires to stand upright. By moving the magnet laterally, wires were dragged along the surface (standing), and pulled into the holes (Fig. 22). After the wires have been assembled into the holes (Fig. 23), the excess space within a wire-occupied hole was filled by a manual application of the thermosetting polymer bisbenzocyclobutene (BCB). The BCB was then cured, permanently adhering the nickel wires within the silicon substrate. Subsequent steps involve steps to expose the ends of the wires on two sides of the silicon, and depositing metal contact lines per specific application. The filling rate of a 30 by 30 array of via holes with a pitch of 120 mm was reported at about 80% in 20 s, and near complete (>95%) in about 2 min.
Microscale self-assembly has been presented as a suite of techniques that can enable the advancement of the electronics industry through heterogeneous integration. A selection of self-assembly methodologies that are representative of techniques employing various modes of transportation and alignment, and the associated environments in which they are performed have been introduced. By identifying self-assembly techniques with suitable discrete component dimensions as well as carefully considering the transport and alignment mechanism and assembly environment, one can harness the parallel and scalable nature of self-assembly, thereby streamlining the existing assembly processes or enabling revolutionary pathways.
References 1. Morris, C.J., Stauth, S.A., Parviz, B.A.: Self-assembly for microscale and nanoscale packaging: steps toward selfpackaging. IEEE Trans. Adv. Packag. 28, 600–611 (2005)
Self-assembly of Nanostructures 2. Fearing, R.S.: Survey of sticking effects for micro parts handling. Presented at the proceedings of the international conference on intelligent robots and systems, vol. 2, 1995 3. Whitesides, G.M., Grzybowski, B.: Self-assembly at all scales. Science 295, 2418–2421 (2002) 4. Rothemund, P.W.K.: Folding DNA to create nanoscale shapes and patterns. Nature 440, 297–302 (2006) 5. Yeh, H.J.J., Smith, J.S.: Fluidic self-assembly for the integration of GaAs light-emitting diodes on Si substrates. IEEE Photonics Technol. Lett. 6, 706–708 (1994) 6. Alien Technology. Available: http://www.alientechnology. com (2011). Accessed 1 Mar 2011 7. Saeedi, E., Kim, S., Parviz, B.A.: Self-assembled crystalline semiconductor optoelectronics on glass and plastic. J. Micromech. Microeng. 18, 7 (2008) 8. Saeedi, E., Etzkorn, J.R., Parviz, B.A.: Sequential selfassembly of micron-scale components with light. J. Mater. Res. 26, 1 (2011) 9. Fukushima, T., Konno, T., Tanaka, T., Koyanagi, M.: Multichip self-assembly technique on flexible polymeric substrate. In: 58th Electronic Components and Technology Conference 2008 (ECTC 2008), pp. 1532–1537 (2008) 10. Srinivasan, U., Liepmann, D., Howe, R.T.: Microstructure to substrate self-assembly using capillary forces. J. Microelectromech. Syst. 10, 17–24 (2001) 11. Jacobs, H.O., Tao, A.R., Schwartz, A., Gracias, D.H., Whitesides, G.M.: Fabrication of a cylindrical display by patterned assembly. Science 296, 323–325 (2002) 12. Knuesel, R.J., Jacobs, H.O.: Self-assembly of microscopic chiplets at a liquid-liquid–solid interface forming a flexible segmented monocrystalline solar cell. Proc. Natl. Acad. Sci. U.S.A. 107, 993–998 (2010) 13. Park, K.S., Xiong, X., Baskaran, R., Bo¨hringer, K.F.: Fluidic self-assembly of millimeter scale thin parts on preprogrammed substrate at air-water interface. In: IEEE International Conference on Micro Electro Mechanical Systems (MEMS), pp. 504–507 (2010) 14. Fang, J., Bo¨hringer, K.F.: Parallel micro componentto-substrate assembly with controlled poses and high surface coverage. J. Micromech. Microeng. 16, 721–730 (2006) 15. Hoo, J., Baskaran, R., Bo¨hringer, KF.: Programmable batch assembly of microparts with 100% yield. In: Transducers, Denver, pp. 829–832 (2009) 16. Hoo, J., Lingley, A., Baskaran, R., Xiong, X., Bo¨hringer, K.F.: Parallel assembly of 01005 surface mount technology components with 100% yield. In: IEEE International Conference on Micro Electro Mechanical Systems (MEMS), pp. 532–535 (2010) 17. Ramadan, Q., Uk, Y.S., Vaidyanathan, K.: Large scale microcomponents assembly using an external magnetic array. Appl. Phys. Lett. 90, 172502/1–172502/3 (2007) 18. Fischer, A.C., Roxhed, N., Haraldsson, T., Heinig, N., Stemme, G., Niklaus, F.: Fabrication of high aspect ratio through silicon vias (TSVs) by magnetic assembly of nickel wires. In: IEEE 24st International Conference on Micro Electro Mechanical Systems 2011( MEMS 2011), Cancun (2011)
2371
S
Self-assembly of Nanostructures Wei Lu Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI, USA
Synonyms Self-assembled nanostructures
nanostructures;
Self-organized
Definition Self-assembly of nanostructures is a process where atoms, molecules or nanoscale building blocks spontaneously organize into ordered structures or patterns with nanometer features without any human intervention. It is the most promising practical low-cost and high-throughput approach for nanofabrication.
The Concept of Self-Assembly The term “self-assembly” can be understood from its two components. The first component “self” implies “spontaneous and on its own,” which suggests that it is a process that happens without human intervention or operation from outside of the system. The second component “assembly” indicates “forming or putting together,” which suggests that the result is a structure built up by lower level building blocks or parts. Selfassembly of nanostructures refers to structures or patterns with nanometer features that form spontaneously from the basic building blocks such as atoms, molecules, or nanoparticles. Cells and living organisms are perfect examples of fairly sophisticated structures selfassembled in nature. These functional assemblies have motivated the study and design of nonliving systems. From a scientific point of view, the spontaneous formation of ordered nanostructures from a random disordered state is inherently an intriguing process in any system. The fundamental mechanisms behind these behaviors raise significant research interests. In addition, studying self-assembly in nonliving systems may provide the critical understanding toward decoding the living systems, which is far more
S
S
2372
complicated. From an engineering point of view, selfassembly as the fundamental principle of a bottom-up approach is possibly the most promising technique for nanofabrication. Thus self-assembly plays a central role in the broad nanotechnology field. The selfassembly of nanostructures may enable a wide range of applications [1], such as nanoelectronic devices, ultrasensitive biosensors, carriers for drug delivery, high capacity lithium batteries, efficient photovoltaic devices, and advanced materials with unique mechanical, electrical, magnetic, or photonic properties.
Discrete System Versus Continuum System Depending on the type of building blocks, self-assembled systems can be classified into two categories: discrete and continuum. A discrete system uses prefabricated building blocks with fixed sizes and shapes. Figure 1 shows an example, where the building blocks are spherical nanoparticles with engineered bonding sites through surface treatment. When put in a liquid, these nanoparticles collide randomly due to the Brownian motion. When two particles are close to each other with matching orientations so that their bonding sites meet, a bond between them forms. A chain structure as shown in Fig. 1a will appear when the system has a single type of nanoparticles. Figure 1b shows a system composed of two types of nanoparticles. A yellow particle has two red bonding sites, indicating that it will bond to two red particles. A red particle has two yellow bonding sites, indicating that it will bond to two yellow particles. This selective bonding leads to a self-assembled chain of alternating particles. Figure 1c shows a case where each particle has four bonding sites. They self-assemble into a network of alternating particle chains. Similar approach can be used to construct three-dimensional structures as well. Various structures can self-assemble by engineering the bonding properties. In addition to local bonding properties, electric and magnetic fields as well as shear forces and spatial constraints have been used to direct the assembly of nanoparticles and nanorods into different configurations [2]. The electric and magnetic field method induce a dipole-type long-range interaction to act as the driving force to bring nanoparticles together, while the shear force method uses hydrodynamic interactions. The assembly of nanowire arrays is more
Self-assembly of Nanostructures
challenging than that of nanoparticles and nanorods due to their highly anisotropic shape. Nanowire selfassembly usually results in partially ordered, small superlattices. This issue has been addressed by several new methods to direct the process, including the use of microfluidic channels and electric fields [3, 4]. In contrast, a continuum system exploits the spontaneous formation of nanoscale domains. Examples include self-assembled domain patterns in binary monolayers [5], block copolymers [6], and organic molecular adsorbates on metal surfaces [7, 8]. A binary monolayer on an elastic substrate may separate into two phases, and self-assemble into ordered patterns, such as triangular lattice of dots, parallel stripes, or serpentine stripes. The feature size is on the order of 1 100 nm, and stable against annealing. Block copolymers are polymers consist of at least two chemically distinct, immiscible polymer fragments that are joined by a covalent bond. These systems have been shown to develop a variety of regular domain patterns via phase separation [9, 10]. The size and the period of the structures are typically on the order of 10–100 nm, depending on the conditions of preparation and the relative chain lengths of the participating polymers. Figure 2 shows an example of domain patterns formed by two phases. A notable feature is the dependence of the pattern on the average concentration. Take Cu and Pb monolayer on Cu (111) substrate as an example. The Pb and Cu mixture monolayer on a Cu substrate forms two phases: a Pb overlayer and a disordered Pb– Cu surface solution. The two phases in the monolayer self-assemble into ordered, nanoscale patterns. When the average concentration of Pb atoms increases, the system can experience a series of patterns from (a) to (e), where the bright phase is Pb while the dark phase is the Pb–Cu surface alloy. When the average concentration is very low, isolated dots are obtained, as shown in Fig. 2a. The bright phase is embedded in the continuous matrix of the dark phase. No long-range ordering is observed. When the average concentration increases, the area covered by the bright dots increases. The dots order and form patterns close to a triangular lattice, as shown in Fig. 2b. When the bright phase occupies roughly half of the surface, stripe structure as shown in Fig. 2c forms. “Inverted” dots are observed when the bright phase dominates, as shown in Fig. 2e. A continuum system offers several unique features. For instance, domains and their patterns self-assemble simultaneously, so that there is no need to pre-synthesize
Self-assembly of Nanostructures
2373
Self-assembly of Nanostructures, Fig. 1 An example shows the self-assembly of a discrete system
a
a
S
c
b
b
c
d
e
Self-assembly of Nanostructures, Fig. 2 An example shows the self-assembly of a continuum system. The domain patterns evolve from dots to parallel stripes, and to inverted dots with the increasing coverage of the bright phase
the building blocks; a significant degree of process flexibility and control can be achieved; and the approach may be applied to diverse systems.
Static Self-assembly Versus Dynamic Selfassembly Depending on the nature of interactions, self-assembly systems can be classified into two categories: static and dynamic. Static self-assembly is a process driven by energy minimization to form static structures which are at global or local equilibrium. Many examples fall into this category, such as colloidal crystals, microphase separation, or spontaneous domain patterns. Dynamic self-assembly involves interactions that only occur when the system is dissipating energy. The structure is determined by an energetic minimum relying on an influx of energy to the system. When the energy flux stops, the minimum configuration does not exist anymore, leading to disintegration of the structure. For instance, rotating nanodisks in a liquid can generate hydrodynamic interactions due to the flow of liquid. When the disks are close to each other, the hydrodynamic interaction causes them to form a nicely ordered hexagonal pattern with a regular inter-disk spacing. When the rotation stops, the interaction disappears and the pattern vanishes. Another example is the oscillating reaction-diffusion reaction, where an evolving pattern develops from a quiescent medium upon the influence of stimuli. A living organism is possibly the most typical example of dynamic self-assembly.
Compared to static self-assembly, dynamic selfassembly is even less understood [11]. This process involves interacting building blocks that can adapt or react to a chemical or physical stimulus from the environment. Unlike the forces that drive static selfassembly through a reduction of the free energy toward equilibrium, the interactions in dynamic self-assembly may drive the formation of structures and patterns away from the thermodynamic equilibrium sustained by a continuous energy input. Such patterns can thus be sensitive to external stimuli and adaptive in response to the surrounding conditions [12].
Forces That Drive Self-assembly Forces of different physical origins may contribute to the interactions between the building blocks. The short-range interaction is generally controlled by forces such as ionic, covalent, or metallic bonding; hydrogen bonding; and van der Waals forces. The long-range interaction has an effective range much longer than the atomic distance, and can originate from elasticity, electrostatic or magnetic field, colloidal and capillary forces, or hydrodynamic interaction. The short-range interaction alone typically leads to either a homogenous structure as shown in Fig. 1a, or a structure with only local modulation whose wavelength is on the order of the size of the building blocks, such as those in Fig. 1b and c. The long-range interaction is essential to form characteristic feature sizes larger than the dimension of the building blocks. Thus long-range interaction is especially important
S
S
2374
for the self-assembly in a continuum system to form nanoscale feature size, which is much larger than the building block of a single atom or molecule. In a discrete system of nanoscale building blocks, the feature size is determined by the competition between the attractive and repulsive forces. Thus force balance analysis is a useful approach. While such a concept may still be applied to a continuum system, the connection between the domain size and their interaction is much less direct. Therefore the energy method is usually used to analyze a continuum system. The feature size is determined by two competing actions: a coarsening action which tends to increase the domain size and a refining action which tends to reduce the domain size. The following discusses a few representative forces and their roles in self-assembly.
Surface Stress Surface stress plays crucial roles in a variety of surface phenomena in solids. To explain the idea, consider a uniform and infinitely large solid shown in Fig. 3a. The body is unstrained and is taken as the reference state. In Fig. 3b, the body is cut into two parts, and the atoms on the surfaces are allowed to find their equilibrium configuration. The energy in state (b) is higher than that in state (a). The energy difference between the two states, G0 , can be written as G0 ¼ 2g0 A0 , where g0 is the excess free energy per unit area owing to the existence of the surface, and A0 is the surface area. G0 is called surface energy. g0 is usually called surface energy density, which is the reversible work per unit area to create a surface. Figure 3c is the same solid in (a) but under strain e. In Fig. 3d, the strained body is cut into two parts. The surface energy can be written as G ¼ 2gA0 . The two energies, G and G0 , are different. The former depends on the strain, i.e., G ¼ GðeÞ. Accordingly, the surface energy density also depends on the strain state, g ¼ gðeÞ. Hence the surface energy depends on both the created surface area and the strain state, namely, dðgA0 Þ ¼ A0 fde, where f ¼ @g=@e is called surface stress. Surface stress becomes a tensor when generalized to biaxial strains. The physical origin of surface stress is related to the difference of bonding between atoms at the surface and the interior atoms. A typical approach to measure surface stress is the cantilever bending method. Upon
Self-assembly of Nanostructures
deposition of atoms on the top surface, surface stress causes the cantilever to bend. By measuring the curvature and some geometric parameters, such as the thickness of the cantilever, one can calculate surface stress. The surface stress of a domain depends on its composition. For a binary epilayer, a composition modulation will cause surface stress nonuniformity. Consider a thin binary layer that grows epitaxially on a solid substrate with fixed thickness, say a monolayer. The two species can relocate by diffusion within the layer. Several physical ingredients contribute to the spontaneous domain formation behavior. Phase separation is a commonly observed phenomenon in various material systems, which can be explained in terms of the free energy of mixing. Consider a mixture of two atomic species A and B, and define concentration by the fraction of one species. The free energy density is given by a double well function of the concentration. A tangent line contacts the function at two concentrations, corresponding to the equilibrium A-rich and B-rich phases. A mixture with an average concentration between these two concentrations will separate into the two phases to reduce the energy. The atoms at the phase boundaries have excess free energy. To reduce the free energy, the total area of the phase boundaries must reduce. Consequently, atoms leave small domains, diffuse in the matrix, and join large domains. Over time the small domains disappear, and the large ones become larger. Thus phase boundary energy causes phase coarsening, which cannot form stable patterns. The observed stable periodic patterns suggest that there must be a refining action in the regular nanoscale structures. This action is provided by the surface stress. Figure 4 shows an important concept: substrate deformation allows the nonuniform surface stress to reduce the total energy. The surface stress is f0 everywhere except for the portion of the surface between points P and Q, where the surface stress steps up by F. The nonuniformity causes P and Q to move toward each other, giving rise to a strain field in the substrate. Imagine that a pair of forces are applied to move the points P and Q away from each other. When the applied forces reach the magnitude F, the substrate becomes unstrained. Because the forces do work to the body, the unstrained state has a higher energy than the strained state. The energy change depends only on F, but not on f0 . Whether the overall surface stress is tensile, compressive, or vanishing does not
Self-assembly of Nanostructures Self-assembly of Nanostructures, Fig. 3 Schematic representation illustrates the concept of surface energy and surface stress
S
2375
a
b
c
d
A0
A0
ε
ε
G0 = 2g0A0 ε G = 2gA0
ε
ε
ε
make any difference to the energy reduction. When the domain size is refined, more pairs of F will contribute so that the energy is reduced further. Consequently, the elastic energy in the film-substrate composite tends to refine the domains. The refinement, however, adds more domain boundary, which increases the phase boundary energy. The phase boundary energy tends to coarsen the domains. The two competing actions – refining and coarsening – can select an equilibrium domain size. These physical ingredients have been incorporated into an energy framework to simulate various nanostructures [13]. Figure 5 clearly shows the refining effect of the surface stress. The average concentration is 0.5. Each phase takes half of the surface area. In sequence (a), shortly after phase separation, the two phases form serpentine stripes. The width of the stripes stabilizes very fast. From t ¼ 1,000 to t ¼ 106, the widths are almost invariant. In contrast, in sequence (b), the two phases try to increase their sizes as much as possible. The system finally evolves into a state that one phase takes half of the calculation cell and the other phase takes the rest half. This reproduces the classical spinodal decomposition. The pattern depends on the average concentration. Changing the average concentration away from 0.5 can generate dots instead of stripes. Anisotropic surface stress induces interesting patterns such as parallel stripes and herringbone structures [14].
f f0
in a phase in b phase
in b phase
F
x
Q
P
P f0
Q f0
f0 + F
Strained
F
substrate
Q
P
f0
F f0
f0 + F
Unstrained substrate
f
b
a
b
a
b
a
b
S Strained substrate
Self-assembly of Nanostructures, Fig. 4 Substrate deformation allows the nonuniform surface stress to reduce the energy, and surface stress provides a refining action
Guided or Templated Self-assembly Guided or templated self-assembly uses a coarse scale external field to influence the fine scale self-assembly behavior. An essence of the approach is that the wavelength of the guiding field or template can be much
larger than the nanoscale structures to be formed. Consequently, preparing the coarse scale field involves low cost and can be easily applied to a very large area. The guiding field can be electrostatic, geometric pre-patterns, surface chemistry, or elastic field [15].
S
2376
Self-assembly of Nanostructures, Fig. 5 Simulation starts with a random initial condition with an average concentration of 0.5. (a) The phases form nanoscale serpentine stripes with surface stress. (b) Spinodal decomposition without surface stress. The phases always coarsen. The times are (1) t ¼ 0, (2) t ¼ 10, (3) t ¼ 100, (4) t ¼ 1,000, (5) t ¼ 105, (6) t ¼ 106 (Adapted from [14])
Self-assembly of Nanostructures
a 1)
2)
3)
4)
5)
6)
1)
2)
3)
4)
5)
6)
b
Figure 6 shows an example to guide the surface patterns with strain field on the substrate surface. Surface strain is applied in the black region in Fig. 6a. Three cases are shown: (1) no guidance, (2) guiding strain in wide stripes, and (3) guiding strain in wavy stripes. In practice, there are many ways to induce an elastic field on the substrate surface. In addition to direct mechanical loading, pre-patterning a substrate with different materials by photolithography or applying an electric field to a substrate embedded with piezoelectric particles produce diverse well-defined
strain fields. Figure 6b shows patterns for an average concentration of 0.3. The monolayer separates into two phases and evolves into a multi-domained triangular lattice of dots when there is no guidance. The middle column shows that in case [2], the monolayer evolves into pattern colonies. The dots form an almost perfect triangular lattice within each colony, and line up with the close packed direction parallel to the edge of the stripes. The lattices in the white and black stripe areas are self-contained, maintaining their own lattice spacings. The right column shows that in case [3], the dots
Self-assembly of Nanostructures Self-assembly of Nanostructures, Fig. 6 Guided self-assembly with strain field on the substrate surface. (a) A constant strain is applied in the black region to guide the self-assembly on the substrate surface. (b) Patterns for an average concentration of 0.3. (c) Patterns for an average concentration of 0.5 (Adapted from [15])
2377
a
1)
2)
S
3)
b
c
orientate themselves along the wavy edges of the three curves. The narrow curves preclude the formation of any dots inside. Similar guiding effect is also shown for an average concentration of 0.5 in Fig. 6c. These results suggest that external loading can effectively change size, shape, and orientation of self-assembled surface structures. Similar guided self-assembly has also been achieved by surface chemistry [16]. Figure 7 shows another example where electric field is used to guide self-assembly. A three-dimensional model has been developed to allow the simulation of the entire self-assembly process and electric field design [17]. It is shown that a melted thin polymer film subjected to an electrostatic field may lose stability at the polymer-air interface, leading to uniform self-organized pillars emerging out of the film surface. The film thickness can affect the patterns formed. With patterned electrodes, parallel stripes replicating an electrode pattern emerge.
Electric Dipoles An adsorbate molecule usually carries an electric dipole. Even if the molecules are nonpolar, the act of binding onto a substrate breaks the symmetry and causes the formation of dipoles. A molecule can be
engineered to carry a large electric dipole moment by incorporating a polar group. Electric dipole interaction can give rise to domain patterns. Examples include Langmuir films at the air–water interface, ferrofluids in magnetic fields, and organic molecules on metal surface. Despite the difference of these systems, similar phenomenology and mechanism can be identified. The adsorbed molecules are mobile. Domains coarsen to reduce the domain boundary and refine to reduce the dipole interaction energy. The competition leads to equilibrium patterns. This mechanism may be used to make two-dimensional nanostructures. Specifically, molecule monolayers composed of electric dipoles can be manipulated with an electric field induced by an AFM tip, a ceiling above the layer, or an electrode array in the substrate. Multilayers of dipole molecules have the capability for making complex structures, especially the formation of nanointerfaces and three-dimensional nanocomposites. Studies have revealed self-alignment between layers, reduction of feature sizes, and guided self-assembly by layer–layer interaction and embedded electrodes [18]. Figure 8 gives an example. Each layer is composed of two kinds of molecules with different dipole moments, which form two domains. Figure 8a shows that the first layer self-assembles into a triangular lattice of dots at an average concentration
S
S
2378
Self-assembly of Nanostructures, Fig. 7 Self-assembled morphological patterns of a melted thin polymer film induced by two electrodes parallel to the initial flat film surface, one above the film and the other beneath the film. (a) Induced nanopillars in a thin film. (b) A relatively thick film. The film continues to evolve after touching the ceiling electrode and then spread. (c) Patterned top electrode causes the nanopillars to line up. (d) Replication of the top electrode pattern (Adapted from [17])
Self-assembly of Nanostructures
a
b
30
30
15 5 100
15
100 50
5 100 100 50
50
50
0 0
0 0
c
d
30
30
15
15 5 100
5 100
100
100 50
50
50
50
0 0
a
b
0 0
c
d
Self-assembly of Nanostructures, Fig. 8 Self-assembled patterns by electric dipole interaction. (a) First layer, average concentration 0.3. (b) Second layer on top of (a). (c) First layer
pattern replicating the electrode pattern. (d) Second layer on top of (c) (Adapted from [18])
of 0.3. The dots have uniform size and form multiple grains. Figure 8b shows the second layer pattern at an average concentration of 0.2 and grows on top of the pattern in Fig. 8a. The second layer evolves from a completely different random initial condition. The dots of the second layer stay at exactly the same positions as those in Fig. 8b, suggesting the anchoring effect of the first layer. The dot size of the second layer is smaller due to lower average concentration. The observations suggest that the first layer determines the ordering and lattice spacing, while the second layer determines the feature size. A scaling down of
size can be achieved via multilayers. The interesting behavior suggests a potential fabrication method. In additional to self-assembly, the first layer pattern can be defined by embedded electrodes, proximal probe technique, or nanoimprinting. Figure 8c shows the first layer pattern with the application of a high voltage, which sweeps off any self-assembled features so that the monolayer replicates the voltage pattern. The second layer evolves into a pattern shown in Fig. 8d. The dots align themselves along the edges of the first layer pattern, and form triangular lattice. It is interesting to note the formation of pairing
Self-assembly of Nanostructures
Self-assembly of Nanostructures, Fig. 9 Top view of structures self-assembled in a system of binary nanoparticles. (a) Gradient core-shell structure. (b) A continuous structure with isolated holes in between. (c) Network structure. (d) A single layer (Adapted from [19])
dark and white lines following the contour of the first layer pattern. Two nearby regions separated by the lines have different preference for the two dipole species. This causes local accumulation and depletion of the two dipoles, resulting in the observed phenomena. Another example of electric dipole interaction is the self-assembly of nanoparticles [19]. Under an applied electric field, each particle will acquire an effective induced dipole. The dipole interaction causes particles to move in a liquid medium and line up into chains following the electric field direction. The formation of an ordered three-dimensional structure is further affected by the interaction between the chains. The self-assembly of binary nanoparticles can lead to a wide class of nanocomposite materials with properties not attainable by a single particle component. Figure 9a shows a top view of the structure. Behind each visible particle shown in the figure is a particle chain lining up. The two kinds of particles are denoted by red and blue colors. They have the same diameter and are both more polarizable than the medium, while the red particle is more polarizable than the blue one.
2379
S
The volume fraction of the particles is 10.2%. Under these situations the particles assemble into isolated columns with a core-shell configuration. The core is composed of the more polarizable red particles, while the shell is composed of the blue ones. From an energetic point of view, the attraction between two red particle chains is stronger than the interaction between a red particle chain and a blue particle chain. The latter is strong than the attraction between two blue particle chains. As a result, the red chains aggregate to form columns. The blue chains tend to get close to the red columns as much as possible, leading to the formation of shells. It is exciting to note that this self-organized functionally gradient structure offers a gradual transition of the permittivity from that of the core to that of the medium. This approach could be potentially very useful to construct functional gradient nanocomposites from bottom up. Figure 9b shows a higher volume fraction of particles, 30.6%. The system evolves into a continuous structure with isolated holes in between. The blue particles enclose the red particles, and form the peripheral regions around the holes. The particle columns demonstrate local BCT structures. Another common feature is that same kind particles tend to aggregate. Figure 9c demonstrates the structure when one kind of particle is more polarizable than the medium while the other is less polarizable. The particles also form pure chains along the field direction with each chain composed of single kind particles. A distinct feature in Fig. 9c is that the red and blue chains are highly dispersed and form a network. This morphology is in contrast to that in Fig. 9a, where same color chains aggregate. Assembling a single layer of nanoparticles on a substrate has many potential applications. A relevant simulation is shown in Fig. 9d for a layer of two kinds of particles. The repulsion between the red particles causes them to form a nicely ordered triangular lattice. The attraction between the blue and red particles causes the formation of blue rings surrounding the red cores.
Sequential Activation of Self-assembly The lack of long-range order has been a major challenge for self-assembled nanostructures since high regularity is crucial to many applications. A general
S
S
2380
Self-assembly of Nanostructures
A
x3 x2 x1
Immobile V
Self-assembly of Nanostructures, Fig. 10 A schematic of sequential activation of self-assembly. Self-assembly is first activated in a finite mobile region, where atoms are allowed to diffuse and form domain patterns. This initial mobile region serves as a “seed.” The mobile region is then shifted like scanning. Self-assembly in the newly activated region will be under influence of patterns already formed in the seed (Adapted from [20])
kinetics-based template-free approach has been developed recently to grow nanostructured superlattices over a very large area [20]. The essence is sequential activation of self-assembly. The idea is illustrated in Fig. 10. A binary monolayer is used as an example. Typically multiple grains of dots will form due to simultaneous self-assembly at different locations, producing a pattern lack of long-range order. Here self-assembly is first activated in a finite mobile region, where atoms are allowed to diffuse and form domain patterns. This initial mobile region will serve as a “seed.” The seed does not need to have a perfect lattice. Then the mobile region is shifted like scanning. The self-assembly in the newly activated region will be influenced by the pattern already formed in the seed. In experiments this process can be achieved by laser or ion beam scanning to control the local temperature, so that diffusion is activated sequentially at each spot along the scanning path. This sequential activation will lead to a large longrange ordered domain pattern even if started with an imperfect seed. The pattern quickly improves and converges to a perfect superlattice along with the sequential activation. Figure 11 demonstrates the growth of long-range ordered superlattices from a seed. Rather than using
Self-assembly of Nanostructures, Fig. 11 Growth of superlattice from seeds. (a) A square seed. (b) A band of nicely ordered hexagonal superlattice formed after scanning over the width. The lattice improved to perfect along with the scanning, demonstrating tolerance of defects in the seeds (Adapted from [20])
a perfect lattice directly, the seed is grown on site. Take Fig. 11a as an example. The constraint of kinetics leads to a square seed composed of dots. The seed is larger than the size of a typical single grain. Therefore defects such as misalignment and multiple grains appeared in this seed. Next the mobile region is shifted to the right. This process is called scanning. The shift distance in each step is much smaller compared to the seed size, which creates a continuous scanning effect. The scanning velocity is chosen to ensure ample time for new domains to develop. Figure 11b shows the structure after scanning over the width of the calculation cell, which forms a band of nicely ordered hexagonal superlattice. Noticeably, the lattice improves to perfection along with the scanning, demonstrating the tolerance of defects in the seeds. High output is essential to nanostructure applications. Two schemes have been proposed to facilitate large-scale fabrication. (1) As shown in Fig. 12a, the scanning directions are alternated between left-right scanning and up-down scanning, using the superlattice created in each previous step as a large seed. This scheme allows the growth rate (area of lattice created per unit time) to increase exponentially with time, greatly
Self-assembly of Nanostructures
2381
S
Self-assembly of Nanostructures, Fig. 12 Two schemes for scaling-up growth. (a) Alternate the scanning directions, using the superlattice created in previous step as a large seed. (b) Increase the size of the mobile in two dimension (Adapted from [20])
accelerating large area fabrication. (2) As shown in Fig. 12b, the size of the mobile area is increased in two dimensions, rather than scanning along one direction. This scheme allows the growth rate to be quadratic of the scanning velocity.
Cross-References ▶ Charge Transport in Self-assembled Monolayers ▶ Nanomaterials for Electrical Energy Storage Devices ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanoparticles ▶ Nanostructured Functionalized Surfaces ▶ Nanotechnology ▶ Self-Assembled Monolayers ▶ Self-Assembly ▶ Self-Assembly for Heterogeneous Integration of Microsystems
References 1. Lu, W., Sastry, A.M.: Self-assembly for semiconductor industry. IEEE Trans. Semicond. Manuf. 20, 421–431 (2007) 2. Pileni, M.P.: Nanocrystal self-assemblies: fabrication and collective properties. J. Phys. Chem. B 105, 3358–3371 (2001) 3. Huang, Y., Duan, X., Wei, Q., Lieber, C.M.: Directed assembly of one-dimensional nanostructures into functional networks. Science 26, 630–633 (2001) 4. Smith, P.A., Nordquist, C.D., Jackson, T.N., Mayer, T.S., Martin, B.R., Mbindyo, J., Mallouk, T.E.: Electric-field assisted assembly and alignment of metallic nanowires. Appl. Phys. Lett. 77, 1399–1401 (2000) 5. Plass, R., Last, J.A., Bartelt, N.C., Kellogg, G.L.: Nanostructures – self-assembled domain patterns. Nature 412, 875 (2001) 6. Fasolka, M.J., Mayes, A.M.: Block copolymer thin films: physics and applications. Annu. Rev. Mater. Res. 31, 323–355 (2001) 7. Love, J.C., Estroff, L.A., Kriebel, J.K., Nuzzo, R.G., Whitesides, G.M.: Self-assembled monolayers of thiolates on metals as a form of nanotechnology. Chem. Rev. 105, 1103–1169 (2005)
S
S
2382
8. Schneider, K.S., Lu, W., Owens, T.M., Fosnacht, D.R., Banaszak Holl, M.M., Orr, B.G.: Monolayer pattern evolution via substrate strain-mediated spinodal decomposition. Phys. Rev. Lett. 93:166104 (2004) 9. Krausch, G., Magerle, R.: Nanostructured thin films via self-assembly of block copolymers. Adv. Mater. 14, 1579–1583 (2002) 10. Lopes, W.A., Jaeger, H.M.: Hierarchical self-assembly of metal nanostructures on diblock copolymer scaffolds. Nature 414, 735–738 (2001) 11. Whitesides, G.M., Boncheva, M.: Beyond molecules: selfassembly of mesoscopic and macroscopic components. Proc. Natl. Acad. Sci. USA 99, 4769–4774 (2002) 12. Ozin, G.A., Hou, K., Lotsch, B.V., Cademartiri, L., Puzzo, D.P., Scotognella, F., Ghadimi, A., Thomson, J.: Nanofabrication by self-assembly. Mater Today 12, 12–23 (2009) 13. Lu, W., Suo, Z.: Dynamics of nanoscale pattern formation of an epitaxial monolayer. J. Mech. Phys. Solids 49, 1937–1950 (2001) 14. Lu, W., Suo, Z.: Symmetry breaking in self-assembled monolayers on solid surfaces: anisotropic surface stress. Phys. Rev. B. 65, 85401 (2002) 15. Lu, W., Kim, D.: Engineering nanophase self-assembly with elastic field. Acta Mater. 53, 3689–3694 (2005) 16. Lu, W., Kim, D.: Patterning nanoscale structures by surface chemistry. Nano. Lett. 4, 313–316 (2004) 17. Kim, D., Lu, W.: Three-dimensional model of electrostatically induced pattern formation in thin polymer films. Phys. Rev. B 73, 035206 (2006) 18. Lu, W., Salac, D.: Patterning multilayers of molecules via self-organization. Phys. Rev. Lett. 94, 146103 (2005) 19. Park, J., Lu, W.: Self-assembly of functionally gradient nanoparticle structures. Appl. Phys. Lett. 93, 243109 (2008) 20. Zhao, Z., Lu, W.: Growing large nanostructured superlattices from a continuum medium by sequential activation of self-assembly. Phys. Rev. E. 83, 041610 (2011)
Self-Cleaning ▶ Lotus Effect
Self-Healing Materials
Self-Cleaning
Self-Organized Nanostructures ▶ Self-assembly of Nanostructures
Self-Regeneration ▶ Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
Self-Repairing Materials Michael Nosonovsky Department of Mechanical Engineering, University of Wisconsin-Milwaukee, Milwaukee, WI, USA
Synonyms Self-healing materials
Definition Self-repair or self-healing is the ability of a material (usually, a composite or nanocomposite) to partially or completely repair damage, such as voids and cracks, occurring during its service lifetime [1, 2]. Self-repair is achieved by embedding a nano-, micro-, or multiscale repair mechanism into the structure of the material (Fig. 1). Autonomous self-healing refers to the ability to repair damage without any external activation. Nonautonomous self-repair refers to healing mechanisms, which require external activation, such as heating the material.
▶ Self-Repairing Materials
Occurrence and Key Findings
Self-Organized Layers
Self-healing is related to the general ability of many systems for self-organization. To facilitate self-organization, usually a special nano-, micro-, or multi-scale structure of a material is needed. Self-healing in biological objects is a source of inspiration for this new
▶ Self-Assembled Monolayers for Nanotribology
Self-Repairing Materials
2383
Catalyst
Crack Fractured capsule
S
A different approach involves thermoplastic polymers with various ways of incorporating the healing agent into the material. In this approach, heating is often required to initiate healing [3]. A self-repair method involving two concentric cylinders of conducting material filled with a liquid solution containing electromagnetic particles of polystyrene or silica was suggested. When damage occurs, voltage is applied between the inner pipe and outer pipe, the current density naturally increases at the location of the damage, this increase in current density causes particle coagulation at the damage site, which in a way to heal the damage.
Capsule with a healing liquid
Self-Repairing Materials, Fig. 1 Self-repair by embedding microcapsules with healing agent
research area. Most living tissues or organisms can regenerate and heal themselves, provided the incurred damage is small or moderate. Most engineered materials, however, deteriorate with time irreversibly due to wear, brittle fracture, fatigue, creep, and other modes of degradation, which limits the life of various components and sometimes causes catastrophic damage. Biological mechanisms of healing are very complex and cannot be directly borrowed for artificial materials. Therefore, it is preferable to speak about bio-inspired, rather than biomimetic, self-healing materials. Polymers Self-healing has been applied most successfully in polymers since they have relatively high rates of diffusion and plasticity due the presence of the crossmolecular bonds. One way to create self-healing polymers is to use thermosetting polymers that have the ability to cure (toughening or hardening by crosslinking of polymer chains), such as the thermosetting epoxy. Epoxy is a polymer formed by a reaction of an epoxide resin with polyamine hardener. Epoxy can serve as a healing agent that is stored within thinwalled inert brittle macrocapsules embedded into the matrix along with a catalyst or hardener. The catalyst or hardener is also embedded in the matrix, but separately from the healing agent. When a crack propagates, the capsules fracture, the healing agent is released and propagates into the crack due to capillarity. Then the healing agent mixes with the catalyst embedded in the matrix, which triggers the cross-linking reaction and hardening of the epoxy that seals the crack [3].
Ceramics Besides the polymers, ceramic self-healing materials are being developed. For example, a concrete composite was produced with glass fibers containing an aircuring sealant embedded in the concrete matrix. This composite exhibited the self-healing behavior, but it suffered from a significant (10–40%) loss of stiffness compared with standard concrete due to fibers. This is a typical situation, when a compromise between selfhealing and mechanical properties should be sought. In another project involving self-healing ceramics, researchers have studied the crack-healing behavior and mechanical properties of a mullite composite toughened by the inclusion of 15% (by volume) SiC whiskers. Self-healing ceramic materials often use oxidative reactions because the volume of oxide exceeds the volume of the original material, and therefore, products of these reactions can be used to fill small cracks [3]. Self-healing nanocomposites constitute another area of research. Metallic Materials It is much more difficult to heal metallic materials, than polymers, because metallic atoms are strongly bonded and have small volumes and low diffusion rates. Currently, there are three main directions which have been taken in the development of self-healing metallic systems. First is the formation of precipitates at the defect sites that immobilize further growth until failure. This mechanism is called sometimes “damage prevention” because it prevents the formation of voids by the diffusion of the precipitate. The atomic transport of matter to voids and defects is provided by a supersaturated solid solution in alloys having decreasing solid solubility of solute elements with
S
S
2384
Self-Repairing Materials
Self-Repairing Materials, Table 1 Self-healing mechanisms in metals Mechanism Type Matrix material Reinforcement materials Microstructure parameter (c) Degradation measure (x) Healing measure (z) Characteristic length of degradation Characteristic length of the healing mechanism Phase transition involved Healing temperature Property improved
Precipitation Damage prevention (solid) Al–Cu, Fe–B–Ce, Fe–B–N, etc. – Solute fraction
SMA reinforcement Damage management (solid, possibly liquid-assisted) Sn–Bi, Mg–Zn
Healing agent encapsulation Damage management (liquid-assisted)
NiTi Concentration of microwires
Volume of voids Amount of precipitated solute Void size (microscale) Atomic scale (atomic diffusion) Solute precipitation Ambient Creep resistance
Volume of voids SMA strain
Sn–Pb Concentration of microcapsules or low-melting-point alloy Volume of voids Amount of released healing agent
Void/crack size (macroscale)
Void/crack size (macroscale)
Microwires diameter (macro or microscale) Martensite/austenite Martensite/austenite transition Restored strength
Microcapsule size (microscale)
decreasing temperature (e.g., Al–Cu). Such an “underaged” alloy, when quenched from high temperature, becomes supersaturated or metastable. However, in order to facilitate precipitation of the solute, heterogeneous nucleation sites are needed. Sites with high surface energy, such as voids, defects, grain boundaries, and free surfaces, become nucleation sites. The driving mechanism for the diffusion is the excess surface energy of microscopic voids and cracks that serve as nucleation centers of the precipitate that plays the role of the healing agent. As a result, the newly formed void is sealed before it grows and thus minimizing the creep and fatigue [1, 2]. Second is reinforcement of an alloy matrix with microfibers or wires made of a shape-memory alloy (SMA), such as Nitinol (NiTi). SMA wires have the ability to recover their original shape after some deformation has occurred if they are heated above certain critical temperature. If the composite undergoes crack formation, heating the material will activate the shape recovery feature of the SMA wires which then shrink and close the cracks [1, 2]. The third approach is to use a healing agent (such as an alloy with a low-melting temperature) embedded into a metallic solder matrix, similarly to the way it is done with the polymers. However, encapsulation of a healing agent into a metallic material is much more difficult task than in the case of polymers. The healing
Al
Solidification of the solder Melting of the low-melting-point alloy Restored strength and fracture toughness
agent should be encapsulated in microcapsules which serve as diffusion barriers and which fracture when crack propagates [1, 2]. Self-healing mechanisms in metals are summarized in Table 1, showing typical materials for the matrix and reinforcement, parameters that characterize microstructure, degradation, and healing, characteristic length scales for the degradation and healing mechanisms, the type of phase transition involved into the healing and what property is improved in the selfhealing alloy, the nature of the healing force and details of healing mechanisms will be discussed in the consequent section. Surface Healing Surface is often the most vulnerable area of a material. Several approaches have been suggested to for fixing surface damage. One approach is to mimic the healing of skin that has a network of vessels and veins so that cutting the skin triggers the blood flow, its coagulation, and sealing of the cut. The vascular network is a treelike hierarchical structure that provides a uniform and continuous distribution of fluids throughout the material volume. The vascularization has been successfully used for polymeric composites [4–6]. A biomimetic approach combining self-healing with the Lotus-effect has been suggested, mimicking healing of a leaf surface. In that approach,
Self-Repairing Materials
2385
micro-reservoirs with coating liquid (e.g., wax paraffin) were attached to the surface with the supply of coating to the damage area. Various types of self-lubrication (e.g., selfreplenishing solid and liquid lubrication, reservoirs for liquid lubricant, lubricant supply facilitated by oxidation or a tribochemical reaction, etc.) can also be viewed as surface healing mechanisms. Theoretical Considerations From the thermodynamic point of view, self-repair can be viewed as a nonequilibrium self-organization process that leads to increasing orderliness of the material and thus to decreasing entropy. In most schemes of self-repair, the self-healing material is driven away from the thermodynamic equilibrium either by the deterioration process itself or by an external intervention, such as heating. After that, the composite slowly restores thermodynamic equilibrium, and this process of equilibrium restoration drives the healing [5]. In most situations, damage repair at a certain length scale is achieved at the expense of the deterioration at the lower-length scale, making self-repairing materials multi-scale systems. For example, macroscale cracks are healed at the expense of fracturing macrocapsules of the healing agent embedded in the matrix of the material. In the precipitation-induced damage prevention mechanisms, micro-voids are healed by atomicscale material transport. Multi-scale organization is typical for biological materials and tissues, it is therefore not surprising that it plays a central role in biomimetic self-repairing materials. In order to characterize degradation, it is convenient to introduce a so-called “degradation parameter” x, to represent, for example, the wear volume, and a corresponding generalized thermodynamic force, Ydeg. When a self-healing mechanism is embedded in the system, another generalized coordinate, the healing parameter, z, can be introduced, for example, the volume of released healing agent together with the corresponding generalized force, Yheal. The generalized degradation and healing forces are external forces that are applied to the system, and flows are related to the forces by the governing linear Onsager’s equations J deg ¼ LY deg þ MY heal J heal ¼ NY deg þ HY heal
(1)
S
where L, M, N, H are corresponding Onsager coefficients [5]. The degradation force Ydeg is an externally applied thermodynamic force that results in the degradation. The healing force Yheal is an external thermodynamic force that is applied to the system. In most self-healing mechanisms, the system is placed out of equilibrium and the restoring force emerges, so this restoring force could be identified with Yheal. Since the restoring force is coupled with the degradation parameter x by the negative coefficients N ¼ M, it also causes degradation decrease or healing. Measure of Healing A quantitative measure is required in order to characterize the efficiency of a healing mechanism and to compare different mechanisms. Several parameters have been suggested, such as (1) restored strength divided by the original strength, (2) fracture toughness divided by the original toughness, and (3) creep life of the material (in the case of damage-prevention mechanisms). Most self-repair mechanisms are not capable to provide 100% restoration of the original properties of the material.
Cross-References ▶ Lotus effect
References 1. Ghosh, S.K. (ed.): Self-Healing Materials: Fundamentals, Design Strategies, and Applications. Wiley, Weinheim (2009) 2. van der Zwaag, S. (ed.): Self Healing Materials – An Alternative Approach to 20 Centuries of Materials Science. Springer, Dordrecht (2007) 3. van der Zwaag, S.: Self-healing behaviour in man-made engineering materials: bioinspired but taking into account their intrinsic character. Philos. Trans. R. Soc. A 367, 1689– 1704 (2009) 4. Nosonovsky, M., Amano, R., Lucci, J.M., Rohatgi, P.K.: Physical chemistry of self-organization and self-healing in metals. Phys. Chem. Chem. Phys. 11, 9530–9536 (2009) 5. Nosonovsky, M.: Self-organization at the frictional interface for green tribology. Philos. Trans R. Soc. A. 368, 4755–4774 (2010) 6. Nosonovsky, M., Bhushan, B.: Thermodynamics of surface degradation, self-organization, and self-healing for biomimetic surfaces. Philos. Trans. R. Soc. A367, 1607–1627 (2009)
S
S
2386
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components Moon-Ho Ham1, Ardemis A. Boghossian1, Jong Hyun Choi2 and Michael S. Strano1 1 Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA 2 School of Mechanical Engineering, Birck Nanotechnology Center, Bindley Bioscience Center, Purdue University, West Lafayette, IN, USA
Synonyms Biomimetic energy conversion devices; Photovoltaic devices; Self-regeneration; Solar cells
Definition Self-repairing photoelectrochemical cells are synthetic solar energy conversion devices that illustrate principles of self-regeneration and repair through a reversible assembly/disassembly cycle. Motivated by the repair mechanism in living chloroplasts, this cycle can potentially extend the lifetime of a photoelectrochemical cell indefinitely.
Introduction Advancements in nanotechnology toward the synthesis and manipulation of materials have enabled the direct interfacing of synthetic components with biological complexes to create new functions. Direct coupling of synthetic, nanoscale components with biological complexes has been used to study biological systems [1, 2] as well as develop new biomimetic electronics [3–5]. In most recent developments, protein complexes have been interfaced with nanoparticles and singlewalled carbon nanotubes (SWCNTs) in the synthesis of biomimetic solar cells [6–10]. This new generation of nano-bio solar cells often relies on a combination of biologically derived components and biologically inspired mechanisms for device fabrication to minimize costs and enhance device efficiencies.
As technology progresses in this field, the mechanisms demonstrated by the new devices increasingly mimic processes found in nature. In this work, the hydrophobicity of SWCNTs for the self-assembly of nanoscale components was employed to mimick the self-repair process used by plants [6]. In this sense, a thorough understanding of the natural processes is essential for developing such biomimetic devices.
Photosynthesis: Chloroplast Structure Natural light-harvesting mechanisms in plants are primarily carried out in the chloroplast, an organelle responsible for the conversion of sunlight into energy for the plant (Fig. 1) [11, 12]. Like most organelles, the chloroplast is surrounded by an outer as well as an inner membrane. A thick, aqueous fluid, the stroma, fills the organelle. The stroma contains genetic materials, including deoxyribonucleic acid (DNA), ribonucleic acid (RNA), and several ribosomes. The most pronounced containment within the chloroplast is the stacks of thylakoid membranes, or grana stacks. Each of the thylakoid membranes contains four key protein complexes used in photosynthesis: photosystem II (PSII), cytochrome b6f (cyt b6f), photosystem I (PSI), and adenosine-50 -triphosphate (ATP) synthase. These complexes are responsible for carrying out the first stage of photosynthesis, the light-dependent reactions [11]. The light-dependent reactions are initiated by the absorption of light in PSII. In PSII, light is primarily absorbed by the P680 site at approximately 680 nm. The light-harvesting antennas surrounding by the protein complexes contain an array of pigments and chlorophyll molecules that absorb light at various wavelengths throughout the solar spectrum, enhancing the overall absorbance of light by the plant. The photons absorbed by these antennas are then transferred to the P680 site via resonance energy transfer. Absorption of light at the P680 site excites an electron, transferring it to the pheophytin (Phe) site of the PSII reaction center (RC). The hole remaining in the P680 site is used in the oxidation of water for the production of oxygen and hydrogen. hv 2H2 O ! O2 þ 4e þ 4H þ
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
S
2387
Thylakoid membrane
Outer membrane Inner membrane
hv
D2 Protein
NADP+
hv
D1 Protein
3H
Fd QA
ADP+Pi
+
ATP
QB
Cyt b PQ Pool
Phe
Stroma P680
Granum
+ H
½ H2O
PSII
P700
FeS
Grana stacks ¼ O2 +
Fe + H
+ H
Cyt b6f
PC
PSI
ATP synthase
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components, Fig. 1 Chloroplast structure and function. The chloroplast (left) consists of an outer membrane surrounding stacks of thylakoid membranes, or grana. Each thylakoid membrane is embedded with several protein complexes (right) including
PSII, cyt b6f, PSI and ATP synthase. Electron-hole separation occurs in PSI upon light absorption. The hole is used in the oxidation of water into hydrogen and oxygen, while the electron is transmitted through cyt b6f and PSI. Synthesis of ATP for chemical storage of energy ultimately occurs in ATP synthase
Meanwhile, the electron is transferred to the QA and QB sites of PSII, where it is ultimately shuttled via a redox carrier to the next major transmembrane protein complex, cyt b6f. Cyt b6f is responsible for electron transfer between the two major RCs of the membrane, PSI and PSII. The electron transfer between the two centers is coupled to the establishment of a transmembrane proton gradient between the outer stroma and inner thylakoid space. The electron undergoes a series of redox reactions demoting electron energy in exchange for proton transfer from the outer stroma to the inner thylakoid lumen. At the conclusion of this series of redox reactions, the electron is emitted at a lower energy state. The electron emitted from cyt b6f is transferred via a redox carrier to the next major protein complex, PSI. Like PSII, PSI is surrounded by a variety of pigmentand chlorophyll-containing antennas. The P700 site of PSI is responsible for light absorption at approximately 700 nm. As before, the surrounding antennas broaden the absorbance of the chloroplast and transfer this absorbed energy to the P700 site via resonance energy transfer. Upon energy absorption, the photoelectric effect is used to excite the electron at the P700, where it is emitted toward the ferrodoxin site for pickup by the next major complex. Excited electrons emitted by PSI are then subject to a variety of chemical pathways, one of which is nicotinamide adenine dinucleotide phosphate (NADPH) production. In this
reduced state, the NADPH’s reducing power is used to power the Calvin cycle, or the second stage of photosynthesis in the light-independent series of reactions, which takes place in the stroma. The fourth, and final, major complex embedded within the thylakoid membrane, ATP synthase, is not directly involved in the preceding series of lightdependent reactions. In ATP synthase, a hydrogen ion is extracted from other sites in the membrane, including the water-oxidizing site of PSII, to produce ATP. ATP, which is also used in the Calvin cycle mentioned above, is the storage of light-absorbed energy in chemical form. Energy storage and release is carried out by phosphate bond formation and scission, respectively.
Natural Self-repair Cycles in Plants The aforementioned system of light-dependent reactions relies on a series of embedded protein complexes under nearly continuous illumination that become highly susceptible to photodamage. In particular, the D1 subunit, which contains the P680, Phe, and the QB sites of PSII, is vulnerable toward protein damage. To address this, plants have evolved sophisticated mechanisms of self-repair wherein the damaged D1 protein is replaced with a newly synthesized protein (Fig. 2a) [13–15]. The initiation of the self-repair cycle is the damage of the D1 protein, which undergoes a change
S
S
2388
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
a
CP43 subunit Stroma
D1 protein
Introduction of new D1
Photodamaging of old D1
Thylakoid Oxygen-evolving complexes
b Photosynthetic RC
MSP
Photosynthetic RC
SWCNT
SC removal
ND SWCNT
SC SC addition
SC-Lipid Micelle
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components, Fig. 2 Natural and synthetic mechanisms of self-assembly. (a) The self-repair in plants relies on the molecular recognition of components and trapping of meta-stable thermodynamic states. In the disassembled state (left), the damage D1 protein is released from the dysfunctional PSII and replaced with a new D1 protein. Introduction of the new protein triggers the self-
assembly of PSII into a functional state (right). (b) The selfassembly of the RC-ND-SWCNT relies on the removal of the surfactant from the system. In the presence of SC, the systems exists in the disassembled state with surfactant dispersion of individual components (left). Upon SC removal, the proteins, lipids and nanotubes self-assemble into the complex shown (right)
in conformation and phosphorylation levels when denatured. The dysfunctional protein signals the spontaneous, partial disassembly of the PSII complex. With complex disassembly, the damaged protein is released from the system and diffuses toward the unappressed, stromal region of the membrane. Meanwhile, new D1 protein biosynthesized elsewhere in the cell diffuses toward the appressed region of the membrane. The selective diffusion of the damaged protein toward unappressed regions and new protein toward appressed regions of the membrane is attributed to the varying phosphorylation levels of the two proteins, which vary the degree of protein hydrophilicity. The more hydrophilic protein is driven toward the appressed region of the membrane whereas the more hydrophobic, damaged protein is driven toward the unappressed region.
Introduction of the new protein to the disassembled complex triggers the spontaneous reassembly of a functional PSII incorporating the new D1 protein. The described self-repair cycle consists of transitions between a series of metastable states in an energy-minimized fashion. Aside from the synthesis of a new protein, this cycle requires the input of no additional energy; the reversible disassembly and reassembly of PSII is completely driven by the formation of thermodynamically and kinetically trapped states. The selfassembly in particular is driven by hydrophobic/ hydrophilic interactions and the assortment of diffusivities demonstrated by the proteins on the molecular scale. Synthetic replication of this reversible selfassembly process therefore hinges on the ability to control molecular interactions on the nanoscale.
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
Use of Biological Light-Harvesting Components in Solar Conversion Devices Biological components such as PSI and photosynthetic RCs have external quantum efficiencies of almost 100% and energy yields of approximately 58%. Recent works have demonstrated that the biological lightharvesting components can be incorporated into optoelectronic devices including photovoltaic devices [6–10]. Jennings and coworkers [10] reported that biomimetically inspired energy conversion devices were fabricated with spinach-derived PSIs which were bound to an Au electrode surface through covalent imine bonds. To increase the surface area of the electrode, the nanoporous leaf-like structure of an Au electrode was employed, which led to the threefold enhancement of the photocurrent due to the increased density of PSI complexes compared to planar electrodes. Carmeli and coworkers [8] directly bound PSI as a monolayer onto an Au surface. The PSIs were isolated from the cyanobacteria Synechocystis sp. PCC6803 and modified at cysteine sites near the P700 for covalent attachment. Kelvin probe force microscopy (KPFM) images show distinct photocurrent generation domains corresponding to PSIs covalently bound to the Au surface with unidirectional orientation. They extended their work to improve the photo-efficiency and apply it to other substrates. Orientated multilayers of PSI complexes were fabricated on the Au surface. In addition, they covalently attached PSI directly to carbon nanotubes and indirectly to GaAs surfaces via a small linker molecule. A self-assembled monolayer of RCs, isolated from the purple bacterium Rhodobacter (Rb.) sphaeroides, on an Au electrode with a His tag was incorporated into solid-state photovoltaic cells, as reported by Baldo and coworkers [7]. The photocurrent spectrum of the device matched the absorption spectrum of the RCs, producing internal quantum efficiencies of 12%. Lebedev and coworkers [9] used cytochrome c as a conductive wire between photosynthetic RCs and the electrode, where the RCs were the same as those in Baldo’s group, leading to a remarkable enhancement of the photocurrent. They also used arrayed carbon nanotube electrodes to improve the photo-conversion efficiency, where the RCs were encapsulated inside carbon nanotube arrays. The efficiency was
2389
S
considerably improved by increasing the number of RCs attached to the electrode surface by about fivefold compared to that obtained with the same proteins when immobilized on a planar graphite (HOPG) electrode. Until now, most of these efforts have been centered on binding photoactive components onto an electrode surface in a unidirectional orientation to enhance photo-efficiencies. However, even with such an enhancement, these arrangements, like those often found in nature, undergo photo-degradation and protein damage. To address issues of photo-degradation, plants have evolved self-repair mechanisms whereby damaged proteins are replaced with photoactive, newly synthesized proteins. In fact, without this selfassembling mechanism, plants would produce less than 5% of their typical photosynthetic yields with lifetimes on the order of minutes under intense illumination.
Synthetic Regeneration Cycles in Photoelectrochemical Cells One approach to mimicking this self-assembly process is to use nanoscaled materials with controllable properties to interface with biologically derived photoactive components to create a photoelectrochemical cell capable of plant-like regeneration. In a recent study [6], this approach was used to develop the first photoelectrochemical cell capable of mimicking key steps in the self-repair cycle. An aqueous solution consisting of SWCNTs, RCs isolated from R. Sphaeroides, the phospholipid dimyristoylphosphatidylcholine (DMPC), membrane scaffold proteins (MSPs), and the surfactant sodium cholate (SC) (Fig. 2b, left) is placed within a dialysis bag with a 10,000 molecular weight cutoff and dialyzed against buffer to selectively remove SC from the system. The removal of SC spontaneously triggers the selfassembly of the photoactive complex shown in Fig. 2b (right). This complex contains a series of agglomerates sequentially aligned along the length of a SWCNT. These agglomerates consist of a lipid bilayer disk, or nanodisk (ND) (Fig. 3). In the ND, the DMPC molecules arrange themselves such that their hydrophilic heads face outward toward the aqueous solution and their hydrophobic tails are sandwiched within the bilayer. Wrapped around the circumference of this disk are two strands of MSP,
S
S
2390
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components 10 nm
MSP
4 nm
DMPC bilayer
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components, Fig. 3 ND structure. NDs consist of DMPC lipids arranged into a bilayer surrounded by two strands of MSPs. Atomic force microscopy (AFM) measurements reveal that these lipid bilayer disks are approximately 10 nm wide and 4 nm high
the length of which determines the overall disk diameter. The particular strands of MSP used in this study result in NDs that are approximately 10 nm wide and 4 nm high. These NDs, which align along the length of the nanotube in the assembled state, house photoactive RCs. Although the self-assembly process occurs spontaneously, the RCs are specifically orientated such that their hydrophobic area (the region near the P680 site) faces the SWCNT and their hydrophilic area (the area near the QB site) faces outward toward its aqueous surroundings. This self-assembly process is completely reversible, such that the re-addition of surfactant to the solution decomposes the complex back into its initial micellar state, which is monitored using the photoluminescence shift of the nanotubes [6].
Understanding and Quantifying the Selfassembly Process This spontaneous assembly of nano- and bio-based materials into a precise configuration via chemical signaling alone can be attributed to the comparative scalabilities of the synthetic and biological components and the controlled, molecular interactions that occur on the nanoscale. The removal of surfactant from the system and, most importantly, from the hydrophobic SWCNT surface triggers the formation of (meta)stable hydrophobic/hydrophilic agglomerates that minimizes SWCNT exposure to its aqueous
surroundings. The precise locations and sizes of the hydrophobic/hydrophilic regions in the RC, the thickness of the ND bilayer, and the unidimensionality and length of the SWCNT are all factors that contribute to the precise molecular arrangement of the RCs along the nanotube length. Even the slightest alteration in sizes, diffusivities, and hydrophobicities of the nanocomponents are enough to perturb the dynamics of the system and even altogether inhibit the formation of such a photoactive complex. To develop a more thorough, quantitative understanding behind the formation of these complexes, the formation of various agglomerates upon surfactant removal was modeled as a series of reactions [16]. k1f ! ½SC A½SCfree micelle k1r
(1)
k2f B½DMPC þ C½SCfree ! ½SC DMPCmicelle k2r (2) k3f ! ½SC MSP ½MSP þ D½SCfree k3r k4f ! ½SC SWCNT ½SWCNT þ E½SCfree k4r
(3)
(4)
k5f
! F½DMPC þ ½SWCNT ½DMPC SWCNT (5) k5r
k6f ! ½MSP SWCNT (6) G½MSP þ ½SWCNT k6r k7f ! ½SWCNT 2½SWCNT 2 k7r
(7)
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
k8f ½SC ! ½SCremoved
(8)
k9f ! ½ND H½DMPC þ I½MSP k9r
(9)
k10f ! ½ND SWCNT J½ND þ ½SWCNT k10r
(10)
where [SC]free is free sodium cholate concentration in monomeric form, [SC]micelle is sodium cholate concentration in micellar form, [DMPC] is free lipid concentration in monomeric form, [SC DMPC]micelle is SC-lipid mixed micelle concentration, [MSP] is MSP concentration, [SC MSP] is SC-suspended MSP concentration, [SWCNT] is free SWCNT concentration, [SC SWCNT] is SC-suspended SWCNT concentration, [DMPC SWCNT] is lipid–SWCNT aggregate concentration, [MSP SWCNT] is protein– lipid aggregate concentration, [SWCNT2] designates SWCNT bundle concentration, [SC]removed is the concentration of SC that has been removed from the system, [ND] is a ND concentration, and [ND SWCNT] is ND–SWCNT concentration. In these reactions, kf is the forward rate constant, kr is the reverse rate constant, and A-J are stoichiometric coefficients. Fitting this model to experimental data on ND–SWCNT concentration over the course of dialysis, the best-fit rate constants for ND and ND-SWCNT formation were 79 mM1s1 and 5.4 102 mM1 s1, respectively. In a diffusion-controlled process, one would expect the ND–SWCNT rate constant to be smaller than that of the ND, since the bulkier NDs and SWCNTs are expected to have smaller diffusivities than the individual lipid and MSP molecules. However, the reverse is the case, and the calculated ND– SWCNT rate constant is larger than that of the ND, indicating that the system is not under diffusion-controlled conditions. Because of the strongly hydrophobic nature of the SWCNT, the adsorption of the ND is expected to be nearly instantaneous, and in this case, the interaction is expected to be closer to diffusion-controlled conditions than ND formation [16].
2391
S
The hydrophobic nature of SWCNTs not only induces the formation of ND–SWCNT complexes, but also the formation of other agglomerates, including SWCNT bundles. As discussed above, removal of surfactant from the SWCNTs forces the nanotubes to seek relatively hydrophobic surfaces to shield them from the aqueous surroundings. These surfaces may be lipid-formed agglomerates, such as NDs, or other nanotubes. In a diffusion-controlled dialysis, the structure with the largest diffusion coefficient is expected to diffuse toward the exposed nanotube surface more quickly. Large, bulky nanotubes with lengths on the order of microns demonstrate smaller diffusion coefficients than the more agile NDs; hence, they are expected to diffuse more slowly toward one another than the ND, favoring the faster ND-SWCNT formation. However, although diffusion occurs more slowly, the nanotube bundling is nearly irreversible: once nanotube bundles are formed, their dissociation would require the use of high-energy perturbations via sonication. Over extended periods of time, SWCNTs are expected to occupy their thermodynamically favored state in bundles, where strong, continuous, hydrophobic interactions along the entire length of the nanotube maintain irreversible bundle formation. Slower dialysis conditions would thus favor the formation of kinetically slower, but thermodynamically favored nanotube bundles. Faster dialysis conditions, on the other hand, favor the formation of kinetically faster, but thermodynamically metastable, ND–SWCNT complexes over smaller time scales [6].
Photoelectrochemical Measurements of Self-assembled Complexes To exemplify the advantages of using the kinetically and thermodynamically favored formation of nanotube-based complexes in dynamic solar cells, the self-assembly was used as a means of regeneration in photoelectrochemical measurements. Photoelectrochemical measurements were carried out using a three-electrode system: a casted SWCNT working electrode, a Pt counter-electrode, and a Ag/AgCl standard reference electrode. A ferricyanide/ubiquinone dual mediator was used to optimize device efficiency.
S
S
2392
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
a
Photoelectrochemical cell Potentiostat New RC/MSP/lipid
Dialyzer 1 for complex assembly
Pt counter electrode
Ag/AgCl reference electrode
Dialyzer 2 for complex disassembly
Used RC/MSP/lipid
Surfactant Surfactant
Tris buffer
Solution containing RC-based complex
Dialysis membrane (12–14 kDa)
Surfactant
Used RC/MSP/lipid
Surfactant Dialysis membrane (1000 kDa)
SWCNT working electrode
Glass substrate
Light illumination 785 nm laser
b
Complex regeneraiton
Relative photocurrent
1.0 0.8 0.6 0.4 0.2 0.0 0
20
40
60
80 100 Time (hour)
120
140
160
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components, Fig. 4 Regenerating photoelectrochemical cell with selfassembled RC-ND-SWCNT complexes. (a) Schematic of the setup of photoelectrochemical systems which consists of a photoelectrochemical cell incorporated to two re-circulating membrane dialyzers. Dialyzers 1 and 2 are used for assembly
and disassembly of the complex, respectively. The photodamaged RCs are removed during dialysis for disassembly of the complex and replaced with new ones while SWCNT are retained. (b) Temporal photoresponse of the RC-ND-SWCNT complexes with (red line) and without (black line) regeneration. (Reproduced from6 # 2010 Nature Publishing group)
Illumination of a 700 nM solution of RC-ND-SWCNT yielded a per nanotube external quantum efficiency (EQE) of approximately 40%. Upon introduction of SC to the illuminated solution, the photoactive complex disassembles, and no photocurrent is recovered [6]. Although the exact mechanism behind the demonstrated photoactivity of the assembled complex
remains unclear, it is hypothesized that electronhole generation occurs in the RC in a manner similar to that found in nature, where the hole ultimately occupies the P680 site and the electron is shuttled to the QB. In nature, the QB site contains ubiquione. Ubiquinone molecularly consists of an electronwithdrawing head and a conjugated tail. In vivo,
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
ubiquinone occupies a thermodynamically favored configuration within the RC wherein the head is docked in the QB site and the tail is extended outward toward to the surface of the protein complex. Since photoactivity is only observed in the presence of ubiquinone, it is hypothesized that the ubiquinone docks to the RC in an analogous manner, serving as a means of electron extraction from the QB site buried deep within the RC. On the other hand, the P680 site, which is much more easily accessible, faces the SWCNT, which may serves as a hole-conducting wire in this configuration. Although photoactivity has been demonstrated in the assembled state, it remains to be seen whether reversible self-assembly could be used as a means of solar cell regeneration, as is the case with the plants. To address this, photoelectrochemical measurements were taken over an extended period of continuous illumination, provoking photodamage and diminishing photocurrents (Fig. 4). When the photocurrent approaches approximately 30% of its initial value, SC is introduced to the system to disassemble the complex, the damaged proteins are replaced with new ones, and surfactant is once more removed from the system to reassemble the complex and recover photocurrent to its initial value. The synthetic regeneration cycle is hence conducted in analogy to the regeneration cycle used by plants, whereby chemical signaling is used to transition between the metastable assembled and disassembled states to replace photodamaged proteins with new ones. Using this regeneration cycle in the synthetic device, photoeletrochemical cell lifetime can be extended indefinitely while increasing efficiency by over 300% over 168 h.
Nanotechnology as a Means of Modulating Bio-inspired Devices The development of the regenerable, self-assembling solar device hinges on the ability to interface SWCNTs with biologically derived components. The precise arrangement of hydrophobic and hydrophilic regimes in biological complexes, such as RCs, is confined to regions that are at most a few nanometers apart. To utilize these variations in non-covalent interactions, one must subject these complexes to materials with properties that can be controlled on the nanoscale. For instance, nanotubes offer
2393
S
RC SWCNT ND
Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components, Fig. 5 High-density stacks for optimal efficiencies. Arrays of RC-ND-SWCNT can be aligned to maximize concentration of these complexes. Alignment of these complexes relies on recent advancements on SWCNT alignment
unidimensional confinement of hydrophobic surfaces that allows for the one-dimensional alignment of light-harvesting complexes. Confinement of these complexes into high-density, unidirectional arrays allows for device optimization and design. In fact, preliminary findings [6] indicate a linear increase in photo-efficiency with increasing RC–ND–SWCNT complexes, suggesting that maximum efficiency (e.g., 40%) could be achieved by maximizing concentration. Maximum concentration can be realized through high-density RC–ND–SWCNT stacks (Fig. 5). The key in synthesizing these stacks is controlling the alignment of not only RC-NDs along the nanotube, but also the nanotubes relative to one another. Recent advancements in nanotube alignment [17–20] have expanded researchers’ ability to precisely control device fabrication on the nanoscale, making the fabrication of such high-density stacks feasible.
Cross-References ▶ Carbon Nanotubes ▶ Hybrid Solar Cells ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanostructures for Energy ▶ Self-assembly of Nanostructures
S
S
2394
SEM
References
17. Xie, X.L., Mai, Y.W., Zhou, X.P.: Dispersion and alignment of carbon nanotubes in polymer matrix: A review. Mater. Sci. Eng. R Rep. 49, 89–112 (2005) 18. Vigolo, B., Penicaud, A., Coulon, C., Sauder, C., Pailler, R., Journet, C., Bernier, P., Poulin, P.: Macroscopic fibers and ribbons of oriented carbon nanotubes. Science 290, 1331– 1334 (2000) 19. Haggenmueller, R., Gommans, H.H., Rinzler, A.G., Fischer, J.E., Winey, K.I.: Aligned single-wall carbon nanotubes in composites by melt processing methods. Chem. Phys. Lett. 330, 219–225 (2000) 20. Murakami, Y., Chiashi, S., Miyauchi, Y., Hu, M.H., Ogura, M., Okubo, T., Maruyama, S.: Growth of vertically aligned single-walled carbon nanotube films on quartz substrates and their optical anisotropy. Chem. Phys. Lett. 385, 298–303 (2004)
1. Rosi, N.L., Mirkin, C.A.: Nanostructures in biodiagnostics. Chem. Rev. 105, 1547–1562 (2005) 2. Wong, S.S., Joselevich, E., Woolley, A.T., Cheung, C.L., Lieber, C.M.: Covalently functionalized nanotubes as nanometre-sized probes in chemistry and biology. Nature 394, 52–55 (1998) 3. Oregan, B., Gratzel, M.: Low-cost, high-efficiency solarcell based on dye-sensitized colloidal TiO2 films. Nature 353, 737–740 (1991) 4. Colfen, H., Mann, S.: Higher-order organization by mesoscale self-assembly and transformation of hybrid nanostructures. Angew. Chem. Int. Ed. 42, 2350–2365 (2003) 5. Lowe, C.R.: Nanobiotechnology: the fabrication and applications of chemical and biological nanostructures. Curr. Opin. Struct. Biol. 10, 428–434 (2000) 6. Ham, M.H., Choi, J.H., Boghossian, A.A., Jeng, E.S., Graff, R.A., Heller, D.A., Chang, A.C., Mattis, A., Bayburt, T.H., Grinkova, Y.V., Zeiger, A.S., Van Vliet, K.J., Hobbie, E.K., Sligar, S.G., Wraight, C.A., Strano, M.S.: Photoelectrochemical complexes for solar energy conversion that chemically and autonomously regenerate. Nat. Chem. 2, 929–936 (2010) 7. Das, R., Kiley, P.J., Segal, M., Norville, J., Yu, A.A., Wang, L.Y., Trammell, S.A., Reddick, L.E., Kumar, R., Stellacci, F., Lebedev, N., Schnur, J., Bruce, B.D., Zhang, S.G., Baldo, M.: Integration of photosynthetic protein molecular complexes in solid-state electronic devices. Nano Lett. 4, 1079–1083 (2004) 8. Frolov, L., Wilner, O., Carmeli, C., Carmeli, I.: Fabrication of oriented multilayers of photosystem I proteins on solid surfaces by auto-metallization. Adv. Mater. 20, 263 (2008) 9. Lebedev, N., Trammell, S.A., Spano, A., Lukashev, E., Griva, I., Schnur, J.: Conductive wiring of immobilized photosynthetic reaction center to electrode by cytochrome c. J. Am. Chem. Soc. 128, 12044–12045 (2006) 10. Ciesielski, P.N., Scott, A.M., Faulkner, C.J., Berron, B.J., Cliffel, D.E., Jennings, G.K.: Functionalized nanoporous gold leaf electrode films for the immobilization of photosystem I. ACS Nano 2, 2465–2472 (2008) 11. Asimov, I.: Photosynthesis. Basic Books, New York (1968) 12. Stern, K.R., Bidlack, J.E., Jansky, S.: Introductory Plant Biology, 11th edn. McGraw-Hill Higher Education, Boston (2008) 13. Aro, E.M., Virgin, I., Andersson, B.: Photoinhibition of photosystem-2 - inactivation, protein damage and turnover. Biochim. Biophys. Acta 1143, 113–134 (1993) 14. Melis, A.: Photosystem-II damage and repair cycle in chloroplasts: what modulates the rate of photodamage in vivo? Trends Plant Sci. 4, 130–135 (1999) 15. Melis, A.: Dynamics of photosynthetic membrane-composition and function. Biochim. Biophys. Acta 1058, 87–106 (1991) 16. Boghossian, A.A., Choi, J.H., Ham, M.H. Strano, M.S.: Dynamic and reversible self-assembly of photoelectrochemical complexes based on lipid bilayer disks, photosynthetic reaction centers, and single-walled carbon nanotubes. Langmuir 27, 1599–1609 (2011)
SEM ▶ Electron Microscopy of Interactions Between Engineered Nanomaterials and Cells
Semiconductor Nanocrystals ▶ Quantum-Dot Toxicity
Semiconductor Piezoresistance ▶ Piezoresistivity
Semicrystalline Morphology ▶ Spider Silk
Sense Organ ▶ Arthropod Strain Sensors
Sensors ▶ Arthropod Strain Sensors
Shark Skin Drag Reduction
2395
S
Shark Denticles ▶ Shark Skin Drag Reduction
Shark Skin Drag Reduction Amy Lang1, Maria Laura Habegger2 and Philip Motta2 1 Department of Aerospace Engineering & Mechanics, University of Alabama, Tuscaloosa, AL, USA 2 Department of Integrative Biology, University of South Florida, Tampa, FL, USA
Synonyms Riblets; Shark denticles; Shark skin separation control
Shark Skin Drag Reduction, Fig. 1 Lateral view of sectioned placoid scales of a shortfin mako shark Isurus oxyrinchus in the region midway between the leading and trailing edge of the pectoral fin. Note the relatively long scale base relative to the crown length. The pulp cavity (PC) of the scale on the left is visible, and the base of the scales is anchored by collagen fibers to the dermis (D) of the skin
Definition The scales, or denticles, on fast-swimming sharks have evolved two mechanisms for controlling the boundary layer flow over the skin surface leading to a reduction in drag. The first, and most widely known and studied, consists of the small streamwise keels covering the surface of the scales also known as riblets which reduce turbulent skin friction drag. The second mechanism is attributed to loosely embedded scales that are located on key regions of the body. When actuated to bristle by the flow, these scales potentially act as a means of controlling flow separation, thereby minimizing pressure drag during swimming maneuvers. Shark scales display a wide variation in geometry both across species while also varying with body location, but on faster swimming sharks, they typically range in size from 180 to 500 mm in crown length.
Overview The Shark Skin Sharks are covered with minute scales, also known as denticles or placoid scales because of their tooth-like nature. The scales have a pulp cavity and a hard enameloid covering and are anchored at the base of the scale to the collagenous layer of the skin known as
the stratum laxum (Fig. 1). The interlocking crowns of each scale make up the surface of the shark exposed to the water, and it is on the crown where many species have developed small riblets, or keels, orientated in the streamwise direction of the flow (Fig. 2). A reduction in the length of the base relative to the length of the crown and a change of shape of the base for some species over certain regions of the body appear to be the means by which certain scales have developed the capability to bristle or erect upon flow reversal (Fig. 3; flow would normally pass over the surface from left to right; flow reversal proceeding right to left can induce scale bristling as shown in Fig. 4 with a schematic shown in Fig. 5). The length of the scales is typically fixed for specific regions of the body within a species but differs among regions and species. Similarly, the number of keels per scale is also consistent per location for a species. For instance, on the fast-swimming shortfin mako (Isurus oxyrinchus), the flank scales have a crown length of approximately 0.18 mm; each crown typically has three keels, each having a height of 0.012 mm and a spacing of 0.041 mm. The slower swimming blacktip shark (Carcharhinus limbatus) has flank scales typically 0.32 mm in length; each crown typically has five keels with a height of 0.029 mm and a spacing of 0.065 mm (Fig. 2).
S
S
2396
Shark Skin Drag Reduction
Shark Skin Drag Reduction, Fig. 2 Scanning electron micrograph (200 ) of the placoid scales of a shortfin mako shark Isurus oxyrinchus (left) and blacktip shark Carcharhinus
limbatus (right) from the dorsal body wall anterior to the dorsal fin. The mako shark scale has three keels or riblets, whereas the blacktip shark has five. Anterior is to the left
Shark Skin Drag Reduction, Fig. 3 Lateral view of sectioned placoid scales in the flank region of a shortfin mako shark Isurus oxyrinchus in the region midway between the dorsal fin and the pectoral fin. Note the relatively long scale crown relative to the shorter base length
Shark Skin Drag Reduction, Fig. 4 Side view through the shortfin mako skin (from the flexible flank area) showing scales that have been manually erected. Because of the individual manual erection, not all scales are erected to the same degree. Flow would normally pass over the skin from left to right, and reversed flow, as occurs during separation, is hypothesized to cause bristling as shown
Specifications and Fluid Dynamics A shark swimming through water experiences two major sources of drag due to the viscous resistance of the fluid flow in the direction of motion. It is generally accepted that faster swimming sharks, such as the shortfin mako (Isurus oxyrinchus), have the capability to reduce both types of drag through evolutionary adaptations to their denticles. The first source of drag is skin friction drag; though not necessarily the largest contributor to overall drag, it is the one most associated with friction of the flow moving over the body of the shark. Skin friction drag is a result of the no-slip boundary condition between the water and the surface of the shark which results in the formation of a boundary layer. Because of the high speeds a shark
can achieve (often greater than U ¼ 10 m/s), this leads to a very high Reynolds number (Re ¼ Ux/n, where x is length and n is kinematic viscosity) in the boundary layer forming over most portions of the body and indicates the development of turbulent flow over most of the shark’s body. For instance, at this speed, the typical transition location when the local Re ¼ 5 105 occurs at a location just x ¼ 5 cm from the nose. While a turbulent boundary layer is less prone to separate from the body, it will have a skin friction magnitude 5–10 times larger than if the flow were to remain laminar. This riblet drag reduction mechanism has demonstrated the potential to reduce the turbulent skin friction drag by approximately 8–10% in manmade applications [1, 2]. The size and spacing of the
Shark Skin Drag Reduction
Shark Skin Drag Reduction, Fig. 5 Schematic showing forward flow in the boundary layer (blue arrows) subject to an adverse pressure gradient causing flow reversal (red arrows). The reversed flow close to the scales causes localized bristling to occur
riblets are indicative of the speed at which a shark may swim with faster species having closer spaced keels. The second type of drag, and the most important to be controlled, is due to a difference in pressure around the body and is often referred to as form drag. It is highly dependent on whether the flow remains attached or separates during swimming. Flow separation causes regions of low pressure on the downstream portions of the body leading to an imbalance of the pressure fore and aft, and thus a dramatic increase in drag. The first means of decreasing pressure drag is streamlining, which consists of smoothing sharp corners and elongating the body with a tapered downstream portion. As a result, sharks and other marine animals have evolved a very streamlined body shape. Body proportions for fast-swimming sharks such as the shortfin mako are in part determined by its thunniform swimming ability. However, when it undergoes turning maneuvers, the fusiform body will undergo relatively greater body curvature than when swimming in a straight line. Turning may thus induce flow separation, but recent experimental evidence suggests that loosely embedded scales on the flank and behind the gills, in the region of maximum girth and further downstream, may act as a means of flow control. Flow separation first involves a reversal of the fluid particles in a thin section adjacent to the surface; this is induced by a region of adverse pressure gradient occurring aft of the point of maximum girth in the streamwise direction. In other words, a suction pressure upstream induces the flow with the least momentum, that closest to the surface, to reverse. Scale bristling likely inhibits the process leading to reversed flow which thereby can control flow separation.
2397
S
Preventing separation will also favorably affect the flow further downstream over the caudal fin which can lead to higher thrust production. While for a swimming shark, the production of thrust and drag occur simultaneously over the body and are inexorably linked; evidence signifies that sharks have evolved mechanisms within the structure of their skin to reduce drag leading to increased thrust production and allow greater maneuverability at high speeds. Reif, a German biologist, working in the late 1970s is generally considered to be the first to report in literature the hypothesis as to the drag-reducing properties of shark skin [3]. Engineering research began at about the same time in both America [4] and Germany [5] as to the functional aspect of riblet surfaces for reducing skin friction drag. Seminal work into riblet design was completed by Bechert et al. in 1997, which demonstrated a maximum turbulent skin friction drag reduction for man-made, blade-like riblets of 9.9% [6]. The bristling of shark scales leading to a mechanism for separation control was always hypothesized to function in a manner similar to a man-made technique known as vortex generators utilized since the 1940s [7, 8]. In more recent years, Lang et al. [9] have posited a new mechanism, whereby flow reversal leads to a region of localized scale bristling, leading to a flowactuated separation control method which can be derived from the shark skin and for which research is ongoing.
Key Research Findings Since the 1970s when studies began, riblets have become a well-accepted means of reducing turbulent skin friction drag with an upper limit of reduction just below 10%. As previously stated, the most exhaustive testing of various riblet geometries, with comparisons made to other researchers working in the field, was completed in 1997 by Bechert et al. [6]. Previous work by Walsh [4] had focused on a sawtooth geometry, which resulted in a maximum drag reduction of about 5%. While flow field measurement and visualization were not carried out to fully understand the mechanism behind the drag reduction, an exhaustive series of drag measurement experiments was carried out in a specially designed oil channel facility [6]; these measurements allowed for a determination of the most effectual geometry for man-made riblet applications.
S
S
2398
Bernard and Wallace provide a summarization of the basic research that has been performed to map out the key characteristics found in turbulent flows [10]. To understand the mechanism whereby drag is reduced by surface modifications, some basic aspects of a turbulent boundary layer flow field need to be understood. A wall-bounded turbulent flow consists of a layer of vorticity within which fluid located further away from the surface is moving faster than that nearer to the surface due to the no-slip condition, and this results in an overall rotational characteristic of the flow in the clockwise direction for flow moving from left to right where the surface itself is stationary (this is the common reference frame to be used for studying a boundary layer flow). Within this layer, a complex assortment of horseshoe- or hairpin-shaped, vortices of various sizes and stages of growth/decay are formed and interact. Scaling of the flow within the boundary layer can be achieved by considering a viscous length scale defined as dv ¼ n/u∗, where u∗ is the friction velocity. The friction velocity is a function of the shear stress, or skin friction, at the surface (t) and the fluid density (r) such that u∗ ¼ (t/r)1/2. The viscous length scale determines the characteristic sizing of the fluid scales found in the boundary layer. For instance, in a given flow with fixed viscosity, n, and at a particular downstream location, x, within a boundary layer consider a variation in free-stream velocity, U. As U is increased, the thickness of the boundary layer at that location will decrease, and the local average shear stress at that same location will increase. This results in an increase in the friction velocity and thus a decrease in the viscous length scale. The resulting decrease in viscous length scale reduces the characteristic sizing of the vortices forming within the boundary layer. In the region close to the surface, longitudinal vortices with an axis of rotation in the streamwise direction are found to form and persist. These vortices have a characteristic diameter of approximately 30dv and a streamwise length ranging from a few hundred up to 1,000 viscous length scales. When these vortices pair up, a region of low-speed fluid is lifted up from the wall, resulting in the formation of a low-speed streak. It is the instability of these streaks located in a region of high shear that leads to the process known as a turbulent burst and subsequent turbulent sweep. A burst is a sudden ejection of low-speed fluid up into the boundary layer, and a sweep is a sudden
Shark Skin Drag Reduction
injection of high-speed fluid down toward the wall. It is the sweep of this high-momentum fluid onto the surface that results in localized, time-varying patches of high skin friction that are the main causation of increased drag in a turbulent boundary layer. It is the interaction of the shark skin with these components of the boundary layer flow that is essential to controlling the flow. Results from the work of Bechert et al. [6] plotted the relative change in skin friction (Dt/t) (compared to a flat surface) versus spacing (s+), where riblet spacing (s) is nondimensionalized such that s+ ¼ s/dv. They found that the reduction in skin friction increases and reaches a maximum in the vicinity of s+16 and thereupon begins to decrease such that larger spacing can actually result in an increase in drag. It is noteworthy that this value corresponds to half the characteristic diameter of the longitudinal vortices forming close to the wall. Thus the riblets sized correctly restrain these near-wall vortices, which for reduced skin friction and reduced viscous length scale will now form at a slightly larger size, and these in turn induce the formation of low-speed streaks [1]. It was also found that a height of the riblets corresponding to half the spacing (h ¼ 0.5 s) gave optimal results. The other key result to garner from these experiments is the variation in maximum decrease with geometry, where a bladelike shape provided the upper limit for skin friction reduction (9.9%). However, for durability of the surface in real applications, a trapezoidal configuration was tested and found to provide improved performance over the sawtooth geometry. Finally, it is remarkable that the geometry found on shark skin scales, with keels of a scalloped shape closely resembling the trapezoidal shape but with smoothed corners, was also tested by Bechert et al. [6] and found to perform comparably to that of the trapezoidal shape. Many shark species also appear to have the approximate h ¼ 0.5 s ratio found to be optimal in experiments. Both of these are indicators that sharks have indeed evolved a scale and riblet geometry for turbulent skin friction reduction; however, the keels may play a role in separation control as well. Experiments to discern the benefits of shark scale bristling began with Bechert et al. [8]; they used a shark skin replica whereby an overlapping array of individual shark scales was built and tested in their oil tunnel facility. In this man-made model of the shark skin, they meticulously built 800 small replicas of
Shark Skin Drag Reduction
a hammerhead (Sphyrna zygaena) shark denticle scaled up 100 times in size (resulting in a crown length of 19 mm and riblet spacing of 4 mm). Each scale was anchored with a compliant spring with variation in stiffness to discern if denticle bristling could result in any additional mechanism to further decrease turbulent skin friction drag. The model was used for cases where the scales laid flat or were wholly bristled at a collective angle of attack. Bristled cases only resulted in increased drag, with higher spring stiffness resulting in higher drag. Flat, aligned scales gave comparable results to riblet surfaces with the only difference being that the greatest drag reduction achieved around s+ ¼ 15 had a value of about 3%. This decrease in performance was attributed to the construction of the model with small gaps and other imperfections preventing a smoother surface. Thus if bristling of the scales is to be advantageous to the shark, it must be something that is only activated upon demand and thus concurs with the postulation that bristling is utilized as a means to control flow separation. Initial speculation began with the hypothesis that bristled shark scales act as vortex generators [8]. These devices produce streamwise vortices which energize the flow close to the surface. Vortex generators need to be placed at a specific downstream location within a boundary layer for maximum performance and typically upstream of the point of separation [9]. Another method of controlling flow separation, used to date at a more global scale, consists of movable flaps; for airfoil applications, these are placed close to the trailing edge (10% of chord length or larger) and have been shown to delay the onset of stall resulting in greater lift [8]. When the flap itself was given a 3D jagged trailing edge, they were also found to act more effectively. More recent work by the authors [11–13] has investigated the mechanism by which shark skin bristling may lead to the development of a passive, flowactuated mechanism for separation control. The current working hypothesis is that the passive, flexible scales of the shark work as microflaps to locally control flow separation as needed on crucial regions of the body where it most often occurs during swimming maneuvers. Recent observations on shark skin bristling angles on the shortfin mako (Isurus oxyrinchus) indicate that, for this particular species known for its swimming capability, only certain portions of the body have very flexible scales. Bristling capability
2399
S
was measured on dead specimens, and the effect of body pressurization was also considered as a potential bristling mechanism. However, results showed that subcutaneous skin pressurization did not cause scale bristling and had no effect on bristling angles; furthermore, such scales have no muscles attached to them as they lie in the deeper layer of the skin, the dermis. Because the flank scales are very loosely attached to the skin and highly mobile, these observations led us to infer that the scales are most likely bristled by reverse flow actuation. Sixteen body locations [12, 13] were considered, six on fins and ten on the body. Scales were manipulated by a fine acupuncture needle and remained bristled once manipulated as shown in Fig. 4. Scale angles vary with body location, but the most flexible scales are found along the flank of the body extending behind the gills to the tail; here scales are found to be easily flexible with slight manipulation on dead specimens to angles of 50 or greater. Highly flexible scales are also found at the trailing edge of the pectoral fins as compared to the leading edge where there was zero scale flexibility; this indicates the scales may be used to control dynamic stall (unsteady separation) to maintain lift forces on these surfaces during swimming and thereby maintain control. Contragility during swimming, or the ability to change direction quickly and easily, requires low pressure drag as well as high musculature control. These recent findings, herein reported as to variation in scale flexibility, corroborate the hypothesis that the flank region, extending from the location of maximum girth to the tail, is where a shark with a side-to-side swimming motion requires separation control to increase contragility. The scale flexibility appears to be a result of a reduction in size of the scale base anchored in the skin and a change in the shape of the base (as can be seen by the histological data shown in Figs. 1 and 3). The reduction appears to occur in the length of the base relative to its width, where the portion of the base that would pivot up and out of the skin at high bristling angles shows a decrease in length. Thus the scales with greater bristling angles are less firmly anchored in the anterior to posterior direction and can pivot more freely within the skin. Cassel et al. [14] describe the process leading to unsteady flow separation, as would occur in a turbulent boundary layer. In the presence of an adverse pressure gradient, the fluid closest to the wall, which also has the lowest momentum, is where flow reversal is first initiated (Fig. 5). This region of fluid moves back
S
S
2400
upstream and thickens and then subsequently erupts from the surface leading to a patch where the flow is separated from the surface. In a turbulent boundary layer, the fluid with the lowest momentum is that contained in the low-speed streaks, and separation of a streak results in the formation of a horseshoe-shaped structure as observed in computational studies of a separating turbulent boundary layer [15]. Thus for the shark skin, made up of a staggered array of scales, flow reversal in long, thin patches of fluid may locally bristle the scales, thereby interrupting the flow separation process. This feature of shark skin resulting in a surface with a favored flow direction is likely key to its capability to inhibit flow separation. Furthermore, previous experiments over a bristled shark skin model established the existence of embedded cavity vortices, axis of rotation in spanwise direction, forming between replicas of the scales [11]. Thus if flow is induced to form between the scales when bristled, there are two supplementary mechanisms that may aid to control the flow. The formation of embedded vortices, similar as occurs for golf ball dimples, would allow the flow to pass over the surface with an ensuing partial slip condition, thereby leading to higher momentum adjacent to the skin. Secondly, with a turbulent boundary layer flow forming above the cavities, there may be added momentum exchange whereby high-momentum fluid is induced at a greater rate to move toward the skin and into the cavities. This latter mechanism, resulting in turbulence augmentation [1], is another possible means to enhance the momentum overall in the flow closest to the wall. These three mechanisms may be working in combination to control flow separation over the skin of the shark.
Past and Future Applications The application of drag-reducing techniques to vehicles in both air and water has obvious advantages and at the same time limitations. Typical riblet spacing for both water and air applications, at speeds normally encountered, falls in the range of 0.035 mm. Thin plastic films with adhesive on one side and riblets on the other have been made commercially available; thus a wide array of testing and use on aircraft has already taken place where drag reduction was documented. However, practical limitations from cost, added
Shark Skin Drag Reduction
weight, and maintenance of the riblet film (particularly with respect to particulates clogging the surface) with only an associated total decrease in drag of 1–3% have prevented widespread use in most aircraft applications [1]. Separation control cannot only reduce drag but can also lead to increased maneuverability for vehicles. Decreasing drag overall can lead to increased fuel efficiency, payload and range in both military and commercial applications. Flow separation is also an important issue for maintaining use of control surfaces on vehicles (i.e., prevention of stall), which can also have relevance to helicopter rotors and turbine/ compressor blades. Passive control mechanisms, including those found on shark skin, have been and will continue to be applied in all these applications.
Cross-References ▶ Biomimetics ▶ BioPatterning ▶ Shark Skin Effect
References 1. Gad-el Hak, M.: Flow Control: Passive, Active and Reactive Flow Management. Cambridge University Press, Cambridge, UK (2000) 2. Bhushan, B.: Shark skin effect. In: Encyclopedia of Nanotechnology. Springer, Berlin (2012) 3. Reif, W.-E.: Protective and hydrodynamic function of the dermal skeleton of elasmobranchs. Neues Jahrb. Geol. Palaontol. Abh. 157, 133–141 (1978) 4. Walsh, M.: Drag characteristics of V-groove and transverse curvature riblets. In: Hough, G.R. (ed.) Viscous Flow Drag Reduction. Progress in Astronautics and Aeronautics, vol. 72, pp. 168–184. AIAA, New York (1980) 5. Bechert, D., Hoppe, G., Reif, W.: On the drag reduction of the shark skin. American Institute of Aeronautics and Astronautics (AIAA) Paper No. 85-0546 (1985) 6. Bechert, D., Bruse, M., Hage, W., Van der Hoeven, J., Hoppe, G.: Experiments on drag-reducing surfaces and their optimization with an adjustable geometry. J. Fluid Mech. 338, 59–87 (1997) 7. Bushnell, D., Moore, K.: Drag reduction in nature. Annu. Rev. Fluid Mech. 23, 65–79 (1991) 8. Bechert, D., Bruse, M., Hage, W., Meyer, R.: Fluid mechanics of biological surfaces and their technological application. Naturwissenschaften 80, 157–171 (2000) 9. Lin, J.: Review of research on low-profile vortex generators to control boundary-layer separation. Prog. Aerosp. Sci. 38, 389–420 (2002)
Shark Skin Effect 10. Bernard, P., Wallace, J.: Turbulent Flow: Analysis, Measurement, and Prediction. Wiley, Hoboken (2002) 11. Lang, A., Motta, P., Hidalgo, P., Westcott, M.: Bristled shark skin: a microgeometry for boundary layer control? Bioinspir. Biomim. 3, 046005 (2008) 12. Lang, A., Habegger, M., Motta, P.: Shark skin boundary layer control. In: Proceedings of the IMA Workshop “Natural Locomotion in Fluids and on Surfaces: Swimming, Flying, and Sliding,” 1–5 June 2010. IMA Volumes in Mathematics and its Applications (2012) 13. Lang, A., Motta, P., Habegger, M., Hueter, R., Afroz, F.: Shark skin separation control mechanisms. Mar. Technol. Soc. J. 45(4), 208–215 (2011) 14. Cassel, K., Smith, F., Walker, J.: The onset of instability in unsteady boundary-layer separation. J. Fluid Mech. 315, 223–256 (1996) 15. Na, Y., Moin, P.: Direct numerical simulation of a separated turbulent boundary layer. J. Fluid Mech. 374, 379–405 (1998)
Shark Skin Effect Bharat Bhushan Nanoprobe Laboratory for Bio- & Nanotechnology and Biomimetics, The Ohio State University, Columbus, OH, USA
Synonyms Low fluid drag surface
Definition The shark skin effect is the reduction of the fluid drag during swimming at fast speeds and protection of its surface against biofouling. The presence of surface microstructure on the skin surface is responsible for this effect.
Overview Many structures, materials, and surfaces found in nature can be exploited for commercial applications. As an example, nature has created ways of reducing drag in fluid flow, evident in the efficient movement of fish, dolphins, and sharks [10, 14]. The mucus secreted by fish causes a reduction in drag as they move through water, and also protects the fish from abrasion by
2401
S
making the fish slide across objects rather than scrape, and it makes the surface of the fish difficult for microscopic organisms to adhere to [28]. It has been known for many years that by adding as little as a few hundred parts per million guar, a naturally occurring polymer, friction in pipe flow can be reduced by up to two thirds. Other synthetic polymers provide an even larger benefit [18]. The compliant skin of the dolphin has also been studied for drag-reducing properties. By responding to the pressure fluctuations across the surface, a compliant material on the surface of an object in a fluid flow has been shown to be beneficial. Studies have reported 7% drag reduction [12]. Another set of aquatic animals which possess multipurpose skin is fast swimming sharks [14]. The skin of fast swimming sharks reduces the drag experienced by sharks as they swim through water and protects against biofouling. The tiny scales covering the skin of fast swimming sharks, known as dermal denticles (skin teeth), are shaped like small riblets and aligned in the direction of fluid flow (Fig. 1). Shark skin–inspired riblets have been shown to provide a drag reduction benefit up to 9.9% [5]. The spacing between these dermal denticles is such that the riblets may not be very effective against very small (micro-) organisms – they probably work best against larger organisms such as mussels, algae, and barnacles. Prevention of undesirable accumulation of microorganisms protects the surface against biofouling. Slower sharks are covered in dermal denticles as well, but these are not shaped like riblets and do not provide much drag reduction benefits. The effect of riblet structures on the behavior of fluid drag, as well as the optimization of their morphology, is the focus of this entry.
Mechanisms of Fluid Drag and Role of Riblets in Drag Reduction Fluid drag comes in several forms, the most basic of which are pressure drag and friction drag [14]. Pressure drag is the drag associated with the energy required to move fluid out from in front of an object in the flow, and then back in place behind the object. Much of the drag associated with walking through water is pressure drag, as the water directly in front of a body must be moved out and around the body before the body can move forward. The magnitude of pressure drag can be
S
S
2402
Shark Skin Effect, Fig. 1 Scale patterns on fast-swimming sharks (Squalus acanthias) [19]
reduced by creating streamlined shapes. Friction or viscous drag is caused by the interactions between the fluid and a surface parallel to the flow, as well as the attraction between molecules of the fluid. Friction drag is similar to the motion of a deck of cards sliding across a table. Fluids of higher viscosity – the attraction between molecules – have higher apparent friction between fluid layers, which increases the thickness of the fluid layer distorted by an object in a fluid flow. An increase in drag occurs as fluid velocity increases. The drag on an object is in fact a measure of the energy required to transfer momentum between the fluid and the object to create a velocity gradient in the fluid layer between the object and undisturbed fluid away from the object’s surface. The above discussion of friction drag assumes all neighboring fluid molecules move in the same relative direction and momentum transfer occurs between layers of fluid flowing at different velocities. Fully developed turbulent flow is commonly said to exhibit complete randomness in its velocity distribution, but distinct regions exist within fully developed turbulent flow that exhibit different patterns and flow characteristics [20]. As these vortices rotate and flow along the surface, they naturally translate across the surface in
Shark Skin Effect
the cross-flow direction. The interaction between the vortices and the surface, as well as between neighboring vortices that collide during translation, initiates bursting motions where vortices are rapidly ejected from the surface and into the outer boundary layers. As vortices are ejected, they tangle with other vortices and twist such that transient velocity vectors in the cross-stream direction can become as large as those in the average flow direction [20]. The translation, bursting of vortices out of the viscous sublayer, and chaotic flow in the outer layers of the turbulent boundary layer flow are all forms of momentum transfer and are large factors in fluid drag. Reducing the bursting behavior of the stream-wise vortices is a critical goal of drag reduction, as the drag reduction possibilities presented by this are sizable. The vortices have been visualized using flow visualization techniques to capture cross-sectional images, shown in Fig. 2, of the stream-wise vortex formations above both flat-plate and riblet surfaces [26]. The average cross-stream wavelength of these high- and low-speed streaks, the added widths of one high-speed streak and one low-speed streak, is equal to the added diameters of two neighboring vortices and has been measured at 70–100 wall units [6, 20, 34]. This corresponds to a vortex diameter of 35–50 wall units. The flow visualizations in Fig. 2 show vortex cross sections and relative length scales, demonstrating vortex diameters smaller than 40 wall units [26]. (As flow properties change, the dimensions of the turbulent flow structures change as well. As such, it is useful to use non-dimensional length values to better compare studies performed in different flow conditions. Dimensionless wall units, marked +, are used for all length scales, which are calculated by multiplying the dimensional length by Vt/n. For example, s+ ¼ sVt/n, where s+ is the non-dimensional riblet spacing, s is the dimensional riblet spacing, n is the kinematic viscosity, and Vt ¼ (t0/r)0.5 is the wall stress velocity, for which r is the fluid density and t0 is the wall shear stress. Wall shear stress can be estimated for round pipe flow using the equation t0 ¼ 0.03955n1/4rV7/4d-1/4, where V is the average flow velocity and d is the hydraulic diameter. For flow in rectangular pipes, the equation for hydraulic diameter d ¼ 4A/c can be applied, where A is the cross-sectional area and c is the wetted perimeter.) The small riblets that cover the skin of fast swimming sharks work by impeding the cross-stream translation of the stream-wise vortices in the viscous
Shark Skin Effect
Shark Skin Effect, Fig. 2 Turbulent flow visualization of stream-wise vortices in a vertical cross section over flat-plate and riblet surfaces (Adapted from [26])
sublayer. As vortices form above a riblet surface, they remain above the riblets, interacting with the tips only and rarely causing any high-velocity flow in the valleys of the riblets. Since the higher-velocity vortices interact only with a small surface area at the riblet tips, only this localized area experiences high-shear stresses. The low velocity fluid flow in the valleys of the riblets produces very low shear stresses across the majority of the surface of the riblet. By keeping the vortices above the riblet tips, the cross-stream velocity fluctuations inside the riblet valleys are much lower than the cross-stream velocity fluctuations above a flat plate
2403
S
[26]. This difference in cross-stream velocity fluctuations is the evidence of a reduction in shear stress and momentum transfer near the surface, which compensates the potentially drag-increasing effect of a larger surface area. Though the vortices remain above the riblet tips, some secondary vortex formations do occur that enter the riblet valleys transiently. The flow velocities of these transient secondary vortices are such that the increase in shear stress caused by their interaction with the surface of the riblet valleys is small. Protruding into the flow without greatly increasing fluid drag allows the riblets to interact with the vortices to reduce the cross-stream translation and related effects. As the riblets protrude into the flow field, they raise the effective flow origin by some distance. The amount by which the height of the riblets is greater than the apparent vertical shift of the flow origin is referred to as the effective protrusion height. By calculating the average stream-wise velocity in laminar flow at heights over riblet surfaces and comparing them to the average stream-wise velocities in laminar flow at heights over a flat plate, the effective streamwise protrusion height, hps, is found for laminar flow. The effective cross-stream protrusion height, hpc, is similarly found for laminar flow by comparing the cross-stream velocities over a riblet surface to those over a flat plate. A schematic of stream-wise and crossstream flow velocity profiles and effective protrusion heights is shown in Fig. 3. The difference between the vertical shifts in the stream-wise and cross-stream origin, Dh ¼ hps–hpc, for any riblet geometry has been proposed to be the degree to which that riblet geometry will reduce vortex translation for low Reynolds number (Re) flows [5]. As Re increases, the degree to which increased surface area affects the overall fluid drag increases, and the drag reduction correlation to the laminar flow theories deteriorates.
Optimization of Riblet Geometry The cross-sectional shape of riblets on fast swimming sharks varies greatly, even at different locations on the same shark. Many types of riblets have been studied, the shapes of which have been chosen for several reasons [14]. Riblet shapes have been chosen for their similarity to natural riblets, for their ease of fabrication, and for purposes of drag reduction optimization.
S
S
2404
Shark Skin Effect, Fig. 3 Schematic representation of the mean velocity profiles and effective protrusion heights for flow in both the stream-wise direction, hps, and in the cross-flow direction, hpc (Adapted from [5])
Two-dimensional (2D) riblets, which have a continuous extrusion of a simple cross section in the stream-wise direction, have been most extensively characterized. The most thorough characterization has been completed for symmetrical 2D riblets with sawtooth, scalloped, and blade cross sections as shown in Fig. 4a [1, 2, 4, 6, 7, 29–32, 35, 36]. Alternative riblet geometries have, in general, shown no increased benefit. These riblets, including asymmetrical riblets, hierarchical riblets, and riblets with rounded or notched peaks have been studied in detail and do not improve upon the benefit of standard riblet geometries [29, 30, 32]. Other 2D riblet shapes which have been studied include alternating brother-sister-type riblets [4] and hierarchical riblets with small riblets on top of larger riblets [35]. Three-dimensional (3D) riblets, which include segmented 2D riblets as well as shark skin moldings and replicas have also been studied. Riblet types characterized include aligned segmented-blade riblets [35], offset segmented-blade riblets [6], offset-3D blade riblets [6], and 3D shark skin replicas [7, 19, 24]. Most studies are done by changing the nondimensionalized spacing, s+, by varying only fluid velocity and collecting shear stress data from a shear stress balance in a wind tunnel or fluid flow channel. The use of non-dimensional characteristic dimensions for riblet studies, namely, nondimensional spacing, s+, is important for comparison between studies
Shark Skin Effect
performed under different flow conditions. Nondimensionalization accounts for the change in size of flow structures like vortex diameter, which is the critical value to which riblets must be matched. When comparing the optimal drag reduction geometries for sawtooth, scalloped, and blade riblets shown in Fig. 4b, it is clear that blade riblets provide the highest level of drag reduction, scalloped riblets provide the second most, and sawtooth riblets provide the least benefit [14]. A summary of comparison features for sawtooth, scalloped, and blade riblets is presented in Table 1. In general, it can be seen in Fig. 4b that each type of riblet is most beneficial near s+ 15, which is between 1/3 and 1/2, the width of the stream-wise vortices. Larger s+ will cause vortices to begin falling into the gap between the riblets, which increases the shear stress at the surface between riblets. As s+ decreases below optimum, the overall size of the riblets decreases to a point below which they cannot adequately impede vortex translation. An additional concern to the application of riblets is the sensitivity of drag reduction to yaw angle, the angle between the average flow direction and the riblet orientation. Figure 5 shows the effects of yaw angle on riblet performance for flow over sawtooth riblets. Yaw angle has a deleterious effect on the drag reduction benefits of riblet surfaces. Riblet surfaces become drag inducing above b ¼ 30 , but small drag reductions can still be seen up to b ¼ 15 [32]. Riblets on shark skin exist in short segments and groups, not as continuous structures. Riblets with 3D features have been created to better approximate the performance of actual shark skin and to determine if there are methods of drag reduction not yet understood from 2D riblet studies. Studies have explored the effects of compound riblet structures and 3D riblets comprised of aligned, segmented-blade riblets [36]. No improvement in net drag reduction was realized when compared to corresponding performance of continuous riblet geometries. More recently, experiments with similarly shaped segmented-blade riblets at spacing s with a matching set of segmented-blade riblets staggered between each row of blades at a spacing of s/2 from either side have been performed [5]. A schematic and image of staggered trapezoidal blade riblets is shown in Fig. 6a. Using these and other staggered riblets, Bechert et al. [5] hoped to achieve the same vortex elevation and anti-translation effects of continuous riblets with less effect on the flow
Shark Skin Effect
2405
a
seen in Fig. 6b. Again, no net benefit in drag reduction was achieved, and after comparison of data, the conclusion was made that it is unlikely that 3D riblets comprised of segmented 2D riblets will greatly outperform continuous 2D riblets. The scales have been molded in epoxy resin from the skin of the Spiny Dogfish (Squalus acanthias) by Jung and Bhushan [19], as shown in Fig. 7a. A decrease in pressure drop corresponding to a decrease in fluid drag was achieved as compared to a smooth surface in a rectangular flow cell experiment, as shown in Fig. 8a. Pressure drop from inlet to outlet of a pipe is a measure of drag with large pressure drop occurring as a result of high drag. Aligned riblets were fabricated on acrylic by using micromachining (Fig. 7b), and a minimal decrease in pressure drop was realized as compared to a smooth acrylic test section (Fig. 8b) [19].
Sawtooth riblets
α
h
s
Scalloped riblets t
h
Riblet Fabrication and Applications
s
Blade riblets
t h
s
b
α = 60 Sawtooth h / s = 0.86 Scalloped h / s = 0.7 Blade h / s = 0.5 t / s = 0.02
4 2
Data from Bechert et al., (1997b)
Δτ / τo (%)
0 –2 –4 –6 –8 –10
5
10
15
S
20 s+
25
30
35
Shark Skin Effect, Fig. 4 (a) Schematic representation of riblet dimensions and (b) drag reduction comparison for sawtooth, scalloped, and blade riblets (Adapted from [5])
origin. Experimental data comparing the largest drag reduction achieved with staggered segmented-blade riblets to optimum continuous blade riblets can be
Riblets have been fabricated for studies and large-scale applications [14]. Typical microscale manufacturing techniques are ill-fitted for large-scale application due to the associated costs. Even for studies, most researchers have opted for traditional milling or molding methods over the micro-fabrication techniques used in the microtechnology industry. Though nondimensional units allow for comparison between flow fields of different fluids and at different conditions, the accurate microscale manufacture of riblets for experimentation has been a field of study in its own right. The largest difficulty in optimizing riblet geometries has been the fabrication of riblet series with incremental changes in characteristic dimension. Riblets used in airflow require spacings at or below 1 mm due to the low viscosity of air and the high speed at which wind tunnels must operate to create accurately measurable shear stresses on a test surface. Conversely, studies in an oil channel have been carried out in flow that is both highly viscous and slower moving. This allows for riblets to be made with spacings in the 3–10 mm range [3]. Commercial and experimental application of riblets outside wind tunnels and test stands is also limited by the high costs for less-than-optimal riblet performance. Application of riblets on a large scale has been done for several studies as well as for competition and retail purposes. Sawtooth riblets on vinyl films produced by
S
S
2406
Shark Skin Effect
Shark Skin Effect, Table 1 Summary and comparison of optimum riblet geometry for various riblet shapes [14] Maximum drag reductionb 5% 6.5% 9.9%
Relative ranka 3 2 1
Riblet shape Sawtooth Scalloped Blade
Optimum geometryb h/s 1, a 60 h/s 0.7 h/s 0.5
Comments Most durable Drag reduction increases as riblet thickness, t decreases. Durability is an issue.
a
1 corresponds to greatest drag reduction Based on published data in [5]
b
Sawtooth riblets
α ~ 63
30
(Walsh and Lindemann, 1984)
Riblets β
25 β=0 β = 15 β = 30
20 Δτ / τo (%)
h / s ~ 0.63
Flow direction
15 10 5 0 −5
−10
5
10
15
20 s+
25
30
35
Shark Skin Effect, Fig. 5 Drag reduction dependence on yaw angle, b, of sawtooth riblets in free stream for h/s 0.62 (Adapted from [32])
3 M riblets have been applied on surfaces ranging from boat hulls to airplanes. Racing swimsuits produced by Speedo and others also employ a riblet pattern on the surface to reduce drag during the streamline portion of each lap of a race [23]. Additionally, a novel surfacescratching technique has been applied to the inside surface of pipelines to create a faux-riblet surface [33]. Beginning in the mid-1980s, vinyl film sawtooth riblets have been applied to boat hulls for racing. Both an Olympic rowing boat and an America’s Cup sailing yacht have been covered with riblets during competition. Because skin friction of an airplane accounts for as much as 48% of total drag, vinyl film riblets have also been applied to test planes of both Boeing and Airbus. These films have not seen use on standard commercial flights yet, but the benefits seen in testing should not go unmentioned. Application of riblets to an airplane requires that several concessions are made. Several locations that would be covered by riblets must be left uncovered due to environmental factors; windows are not covered for the sake of visibility; several locations where dust and debris contact the airplane during flight are left bare because the
riblets would be eroded during flight; and locations where deicing, fuel, or hydraulic fluid would come in contact with the riblets are left bare. After these concessions, the riblets covering the remaining 70% of the aircraft have provided 3% total drag reduction. This 3% drag reduction correlates to a similar 3% savings in fuel costs [4]. Another large commercial application for riblet technologies is drag reduction in pipe flow. Machining the surface or applying vinyl film riblets proves difficult in the confines of most pipes, and an alternate solution must be used. Experimental application of a scratching technique to the inside surface of pipes has created a riblet-like roughness that has provided more than 5% drag reduction benefit [33]. Stemming from an old sailors’ belief that ships sail faster when their hulls are sanded in the longitudinal direction, Weiss fabricated these riblets by using a steel brush moved through the pipeline to create a ridged surface. Studies have shown as much as a 10% reduction in fluid flow with the combined effect of cleaning the pipe and ridging the surface. Tests on a 10-mile gas pipeline section have confirmed this benefit during commercial operation [4]. The dominant and perhaps only commercial market where riblet technology for drag reduction is commercially sold is competitive swimwear [14]. The general population became aware of shark skin’s drag reduction benefits with the introduction of the FastSkin ® suits by Speedo in 2004. Speedo claimed a drag reduction of several percent in a static test compared to other race suits. However, given the compromises of riblet geometry made during manufacturing, it is hard to believe the full extent of the drag reduction. It is clear that creating surface structures by weaving threads is difficult. As a result, riblet geometries woven from thread have limited options of feasible riblet shapes. By the pattern woven into the FastSkin ® swimsuits, riblets are formed which resemble wide
Shark Skin Effect Shark Skin Effect, Fig. 6 Comparison of drag reduction over optimum continuous blade riblets with optimum segmented trapezoidal blade riblets. (a) Segmented riblets were staggered as shown. Spacing between offset rows is s/2, while spacing between corresponding rows is s. (b) Optimal hour to second ratio for staggered blade riblets is 0.4. Staggered blade riblets provide less drag reduction benefit than continuous blade riblets (Adapted from [5, 6])
2407
S
a
8.3
4.6
45° 45° 1.6 r
4
b
h / s = 0.5 h / s = 0.4
2.75
1.6 r
(Bechert et al., 2000a) (Bechert et al., 1997b)
2 t / s ~ 0.02
Δτ / τo (%)
0 Continuous blade –2 –4 Staggered –6 –8 –10
5
blade riblets with small grooves on top. The larger riblets are formed by the macro-weaving pattern, and the smaller riblets are created by the individual weaves of thread aligned with the macro-riblets. Both of these riblet-like shapes are distinguishable in Fig. 9. As shown in Fig. 9a, unstretched riblets are tightly packed. As the fabric stretches, the riblet width and spacing increase (Fig. 9b). The associated decrease in h/s ratio depends on the dimensions of each swimmer’s body, which is another compromising factor in the design. Riblet thickness is also a factor considered in the design. Aside from the limitations imposed by the weaving patterns available, flexibility in the riblet tips will hinder the fabric’s ability to impede the cross-stream translation of stream-wise vortices. Thicker riblets are probably needed, used for strength, and cause a decrease in the peak drag reduction capability compared to thinner riblets. 3 M vinyl film riblets [27] have been applied to many test surfaces, including the inside of various
10
15
20 s+
25
30
35
pipes for pipe flow studies [22], flat plates in flow channels and wind tunnels, boat hulls in towing tanks [11], airplane wings [17], and airplane fuselages. Similar riblet films have been fabricated using bulk micromachining of silicon to create a master for molding of Polydimethylsiloxane (PDMS) to create a thin, flexible riblet film. This film has been used in flow visualization tests [25]. Grinding and rolling methods of riblet fabrication have been studied for application in both research and large-scale application. A profiled grinding wheel has been used to fabricate several riblet geometries based on sawtooth riblets with h ¼ 20 mm and s ¼ 50 mm [15]. Dressing of the grinding wheel was done with diamond-profile roller used in a two-step process in which the profile roller dresses every second tooth on the first pass, shifts axially the distance of one riblet spacing, and dresses the remaining teeth on a second pass. One downside of the grinding process is the lack of hardening on the final riblet surface. Alternatively,
S
S
2408
Shark Skin Effect, Fig. 7 (a) Scanning electron microscope (SEM) micrographs of shark skin replica patterned in epoxy, and (b) segmented bade-style riblets fabricated from acrylic [19]
Shark Skin Effect
a
Shark skin (Squalus acanthias) replica Top view
45° tilt angle side view
100 μm
50 μm
45° tilt angle top view Swimming direction
100 μm
b
Rib patterned surface
200 μm
rolling methods can be used to strain harden the riblets during fabrication. Using a roller with the profile of two riblets on its outer face, a linearly patterned rolling process has been used to fabricate scalloped riblets in a titanium alloy with h ¼ 162 mm and s ¼ 340 mm [21]. The strain hardening, favorable grain patterns, and residual compressive stresses in the riblet surface after fabrication provide advantages in riblet strength for production applications.
Effect of Fish Mucus and Polymers on Fluid Drag Fish are known to secrete mucus during swimming [14]. Though it is not known whether the mucus is present at all times, it is known that certain
50 μm
100 μm
environmental factors cause, alter, or enhance the production of mucus. These environmental stressors may present a need for increased swimming speed to catch or avoid becoming prey, for protection against nonpredatory threats such as microorganisms, or to resist abrasion while swimming near rocky surfaces. Regardless of which events cause fish to secrete mucus, the drag reduction benefits during mucus-assisted swimming are known. Numerous experiments have demonstrated the drag reduction possible with fish-covered mucus compared to non-mucus-covered shapes [18]. In an experiment comparing the drag on wax models to a mucus-covered fish, a reduction in skin friction drag of 50% was seen [13]. Similar to these fish mucus experiments, polymer additives in pipe flows have been known for many years to reduce the drag in fluid flows by extreme
Shark Skin Effect
a 10 Predicted for hydrophilic surface Flat epoxy resin
Pressure drop (kPa)
8
Transitional flow (2300 < Re < 4000)
Epoxy shark skin replica
6 Laminar flow (Re < 2300) 4
Turbulent flow (4000 < Re)
2
0
0
2000
4000
6000
8000
b 10 Predicted for hydrophilic surface Flat acrylic surface
Pressure drop (kPa)
8
3D Riblet patterned surface
Transitional flow (2300 < Re < 4000)
6 Laminar flow (Re < 2300) 4 Turbulent flow (4000 < Re) 2
0
0
2000
4000
6000
8000
0.5 Predicted for hydrophilic surface Flat acrylic surface
0.4 Pressure drop (kPa)
Shark Skin Effect, Fig. 8 (a) Comparison of pressure drop in rectangular pipe flow over flat epoxy surface with shark skin replica surface. (b) Comparison of pressure drop in rectangular pipe flow over flat acrylic surface and segmented blade riblets. Data are compared with the predicted pressure drop function for a hydrophilic surface (Adapted from [19])
S
2409
3D Riblet patterned surface
S
0.3
0.2
0.1
0
0
100
200 300 Flow rate (μL / s)
400
500
S
2410
a
Shark Skin Effect Unstretched riblet pattern
1.0 mm
b
Stretched riblet pattern
1.0 mm
threads
Large riblets formed by weave
Shark Skin Effect, Fig. 9 Images of riblet geometries on (a) unstretched and (b) stretched Speedo FastSkin ® swimsuit. (c) Schematic showing apparent hierarchical riblet structure formed by threads [14]
amounts. Polymers are known to have low shear strength which results in low friction [8, 9]. In a pipe flow study comparing various injection techniques of polymer solutions into water, drag reductions of up to 80% were achieved [16]. Additionally, the drag reduction benefit increases with increased Reynolds number. While this works well for pipe flows, in which the polymer remains mixed and active throughout the length of the pipe, its application to external flows is
much more difficult. Mucus on fish does not mix well with water in static contact, but does mix and provide drag reduction during dynamic contact. By this feature, the mucus use of fish is minimized. Unfortunately for any long range application of polymer drag reduction on an external flow, the polymer solution must be continuously injected. This would cause large quantities of the solution to be used and likely render the strategy inefficient in terms of overall energy use. Though sharks do not secrete enough mucus to use this mechanism for drag reduction, small amounts of mucus are present on the skin of sharks [2]. It is possible that shark skin mucus secretion is similar to fish, where only environmental stressors or swimming causes an increase in output, but the total quantity of mucus on the surface at any given time is likely quite low. One possible mechanism by which this trace quantity of mucus could be useful is in changing the flow characteristics in the riblet valleys or at the riblet peaks, where shear stresses are highest. In the riblet valleys, a trace secretion of mucus could increase flow velocity and decrease the overall momentum transfer from the shark to the surrounding water by condensing the overall structure of the velocity gradient. Alternatively, injection at the riblet peaks may cause a reduction in shear stresses where they are at their maximum and again cause a reduction in drag. These small effects near the surface may propagate into a larger benefit as the lines of constant velocity in the flow shift and condense [14].
Closure Fluid drag in the turbulent boundary layer is in large part due to the effects of the stream-wise vortices formed in the fluid closest to the surface. Turbulence and associated momentum transfer in the outer boundary layers is in large part due to the translation, ejection, and twisting of these vortices. Additionally, the vortices also cause high velocities at the surface which create large shear stresses on the surface. Riblets impede the translation of the stream-wise vortices, which causes a reduction in vortex ejection and outer layer turbulence. In addition, riblets lift the vortices off the surface and reduce the amount of surface area exposed to the high-velocity flow. By modifying the velocity distribution, riblets facilitate a net reduction in shear stress at the surface.
Shark Skin Effect
Various riblet shapes have been studied for their drag-reducing capabilities, but sawtooth, scalloped, and blade riblets are most common. Drag reduction by riblet surfaces has been shown to be as high as nearly 10% given an optimal geometry of h/s 0.5 for blade riblets with a no-slip condition. The maximum reliable drag reduction provided by scalloped riblets and sawtooth riblets is about 6% at h/s 0.7 and 5% at ridge angle a 60 , respectively. Experimentation of other shapes has provided similar benefits of around 5% drag reduction. Though the optimum shape for drag reduction performance is blade riblets, the fragile nature of these blades makes their commercial application of little use. Scalloped and sawtooth riblets, which provide considerably less drag reduction benefit, are much stronger shapes mechanically speaking and should be used for application in environments where contact may occur with non-fluid materials. Commercial applications of riblets include competition swimsuits, which use a thread-based riblet geometry, as well as experimental applications to airplanes. Drag reductions in riblet application have been accomplished, and flight applications have seen fuel savings of as much as three percent. Manufacturing techniques for riblets must also be chosen specific to their application. Vinyl film riblets are the easiest method, as application of a film to a surface requires less for work small-scale application than other methods. Rolling or grinding methods of riblet application should be investigated for turbine blades or high volume commercially sold pieces.
Cross-References ▶ Biognosis ▶ Biomimetics ▶ Biomimicry
References 1. Becher, D.W., Hoppe, G.: On the drag reduction of the shark skin. Paper # AIAA-85-0564, presented at AIAA Shear Flow Control Conference, Boulder, CO, 12–14 Mar (1985) 2. Bechert, D.W., Bartenwerfe, R.M., Hoppe, G., and Reif, W.-E.: Drag reduction mechanisms derived from shark skin. Paper # ICAS-86-1.8.3, Proceedings of the 15th ICAS congress, vol. 2 (A86-48-97624-01), pp. 1044–1068. AIAA, New York (1986)
2411
S
3. Bechert, D.W., Hoppe, G., van der Hoven, J.G.T., Makris, R.: The Berlin oil channel for drag reduction research. Exper. Fluids 12, 251–260 (1992) 4. Bechert, D.W., Bruse, M., Hage, W., and Meyer R.: Biological surfaces and their technological application – laboratory and flight experiments on drag reduction and separation control. Paper # AIAA-1997-1960, presented at AIAA 28th Fluid dynamics conference, Snowmass Village, CO, June 29–Aug 2 (1997) 5. Bechert, D.W., Bruse, M., Hage, W., van der Hoeven, J.G.T., Hoppe, G.: Experiments on drag reducing surfaces and their optimization with an adjustable geometry. J. Fluid Mech. 338, 59–87 (1997) 6. Bechert, D.W., Bruse, M., Hage, W.: Experiments with three-dimensional riblets as an idealized model of shark skin. Exper. Fluids 28, 403–412 (2000) 7. Bechert, D.W., Bruse, M., Hage, W., Meyer, R.: Fluid mechanics of biological surfaces and their technological application. Naturwissenschaften 87, 157–171 (2000) 8. Bhushan, B.: Principles and Applications of Tribology. Wiley, New York (1999) 9. Bhushan, B.: Introduction to Tribology. Wiley, New York (2002) 10. Bhushan, B.: Biomimetics: lessons from nature – an overview. Phil. Trans. R. Soc. A 367, 1445–1486 (2009) 11. Choi, K.S., Gadd, G.E., Pearcey, H.H., Savill, A.M., Svensson, S.: Tests of drag-reducing polymer coated on a riblet surface. Appl. Sci. Res. 46, 209–216 (1989) 12. Choi, K.S., Yang, X., Clayton, B.R., Glover, E.J., Altar, M., Semenov, B.N., Kulik, V.M.: Turbulent drag reduction using compliant surfaces. Proc. R. Soc A 453, 2229–2240 (1997) 13. Daniel, T.L.: Fish mucus: in situ measurements of polymer drag reduction. Biol. Bull. 160, 376–382 (1981) 14. Dean, B., Bhushan, B.: Shark-skin surfaces for fluid-drag reduction in turbulent flow: a review. Phil. Trans. R. Soc. A 368, 4775–4806;5737 (2010) 15. Denkena, B., de Leon, L., Wang, B.: Grinding of microstructures functional surfaces: a novel strategy for dressing of microprofiles. Prod. Eng. 3, 41–48 (2009) 16. Frings, B.: Heterogeneous drag reduction in turbulent pipe flows using various injection techniques. Rheologica Acta 27, 92–110 (1988) 17. Han, M., Huh, J.K., Lee, S.S., and Lee, S.: Micro-riblet film for drag reduction. In: Proceedings of the Pacific Rim Workshop on transducers and micro/nano technologies, Xiamen, China (2002) 18. Hoyt, J.W.: Hydrodynamic drag reduction due to fish slimes. Swim. Fly. Nat. 2, 653–672 (1975) 19. Jung, Y.C., Bhushan, B.: Biomimetic structures for fluid drag reduction in laminar and turbulent flows. J. Phys.: Condens. Matt. 22, #035104 (2010) 20. Kline, S.J., Reynolds, W.C., Schraub, F.A., Runstadler, P.W.: The structure of turbulent boundary layers. J. Fluid Mech. 30, 741–773 (1967) 21. Klocke, F., Feldhaus, B., Mader, S.: Development of an incremental rolling process for the production of defined riblet surface structures. Prod. Eng. 1, 233–237 (2007) 22. Koury, E., Virk, P.S.: Drag reduction by polymer solutions in a riblet-lined pipe. Appl. Sci. Res. 54, 323–347 (1995) 23. Krieger, K.: Do pool sharks really swim faster? Science 305, 636–637 (2004)
S
S
2412
24. Lang, A.W., Motta, P., Hidalgo, P., Westcott, M.: Bristled shark skin: a microgeometry for boundary layer control? Bioinspir. Biomim. 3, 1–9 (2008) 25. Lee, S.J., Choi, Y.S.: Decrement of spanwise vortices by a drag-reducing riblet surface. J. Turbulence 9, 1–15 (2008) 26. Lee, S.-J., Lee, S.-H.: Flow field analysis of a turbulent boundary layer over a riblet surface. Exp. Fluids 30, 153– 166 (2001) 27. Marentic, F.J., Morris, T.L.: Drag reduction article. United States Patent No 5,133,516, 1992 28. Shephard, K.L.: Functions for fish mucus. Rev. Fish Biol. Fisheries 4, 401–429 (1994) 29. Walsh, M.J.: Drag characteristics of v-groove and transverse curvature riblets. Viscous Flow Drag. Red. 72, 169–184 (1980) 30. Walsh, M.J.: Turbulent boundary layer drag reduction using riblets. Paper # AIAA-82-0169, presented at AIAA 20th Aerospace sciences meeting, Orlando FL, 11–14 Jan 1982 31. Walsh, M.J., Anders, J.B.: Riblet/LEBU research at NASA Langley. Appl. Sci. Res. 46, 255–262 (1989) 32. Walsh, M.J., Lindemann, A.M.: Optimization and application of riblets for turbulent drag reduction. Paper # AIAA84-0347, presented at AIAA 22nd aerospace sciences meeting, Reno, NV, 9–12 Jan 1984 33. Weiss, M.H.: Implementation of drag reduction techniques in natural gas pipelines. Presented at 10th European drag reduction working meeting, Berlin, Germany, 19–21 March 1997 34. Wilkinson, S.P.: Influence of wall permeability on turbulent boundary-layer properties, Paper # AIAA 83–0294, presented at 21st aerospace sciences meeting of the american institute of aeronautics and astronautics, Reno, NV, 10–13 Jan 1983 35. Wilkinson, S.P., Lazos, B.S.: Direct drag and hot-wire measurements on thin-element riblet arrays. Presented at the IUTAM symposium on turbulence management and relaminarization, Bangalore, India, 19–23 Jan 1987 36. Wilkinson, S.P., Anders, J.B., Lazos, B.S., Bushnell, D.M.: Turbulent drag reduction research at NASA Langley: progress and plans. Inter. J. Heat Fluid Flow 9, 266–277 (1988)
Shark Skin Separation Control ▶ Shark Skin Drag Reduction
Short (Low Aspect Ratio) Gold Nanowires
Shark Skin Separation Control
Si Nanotubes ▶ Physical Vapor Deposition
Silent Flight of Owls ▶ Silent Owl Wings
Silent Owl Wings Thomas Bachmann1 and Hermann Wagner2 1 Institute for Fluid Mechanics and Aerodynamics, Technische Universit€at Darmstadt, Darmstadt, Germany 2 Institute for Biology II, RWTH Aachen University, Aachen, Germany
Synonyms Silent flight of owls
Definition Wings of owls are equipped with macro- and microstructured devices that reduce flight noises significantly. These birds of prey hunt in both twilight and darkness. Since visual information is limited at that time of day, owls use acoustic information to detect and track potential prey. The silent flight affords the detection of prey in flight and makes the owl inaudible for its prey. Different wing and feather specializations, such as serrations at the leading edge of the wing, a velvet-like upper surface and fringes along the inner vanes, influence the airflow and thus are important for the reduced flight noise in owls.
▶ Gold Nanorods
Overview
Short-Interfering RNA (siRNA) ▶ RNAi in Biomedicine and Drug Delivery
Fundamentals of Bird Flight In the course of evolution, birds have conquered the air for locomotion. A great variety of wing geometries
Silent Owl Wings
have evolved for different needs of birds. Wings can be long or short, narrow or broad, thick or thin. Furthermore, wings can be pointed or have a rounded tip, which may be smooth or fingered. Considering this variety, each wing appears to be adapted to the distinct flight conditions of the particular species and the ecological niche it uses. However, all wings have at least one aerodynamic feature in common: They are cambered to the dorsal side. Only curved wings are able to produce enough lift at moderate angles of attack and low flight speeds to allow a bird to become airborne. Bird wings are mostly cambered in proximal regions. Thickness as well as camber decrease towards the wing tip. The resulting reduction of mass at the wing tip causes a decrease of inertia to allow high wing-beat frequencies (2–12 Hz) [1] at low energy consumption [2]. A bird has to defy gravity to become airborne. Hence, flight consumes a lot of energy, especially during takeoff and landing [2]. High lift is achieved either by an intense beating of the wings or by forming wings with high-lift properties [3]. Such wings are characterized by a high camber and a large wing area. Apart from gravity, birds have to deal with an additional force, the drag. Drag is the resistance of the body and wings while moving through the air. It is comprised of three components [3]: first, the friction drag between the airflow and the surface, second, the form drag of the bird’s body and the leading edges of the wings, and third, the induced drag. The induced drag is the component of drag force on the bird’s wing that is caused by an induced downward velocity. The drag is also influenced by the microstructure of the plumage. Several anatomical specializations of feathers have been reported that influence the airflow around the bird wing, most of them in owls [4–8]. Function of Silent Flight in Owls Owls have an almost global distribution, which is reflected in the variety of species and subspecies. These birds spend a high proportion of their time searching for prey, either by sitting on a perch, or flying slowly over fields and vegetation [9, 10]. The prey of most owls is mainly active at night. Since visual information is limited in low-light conditions, owls use acoustical cues to detect their prey. The prey emits noise by rustling through the vegetation or through intra- and interspecific communication sounds which can be detected by the owl even in absolute darkness [11]. Several anatomical and behavioral
2413
S
adaptations of the hunting strategy have been discovered in owls. One conspicuous specialization in many owl species is the facial ruff with its specialized feathers that direct sound toward the ears like a dish antenna. The facial ruff is also responsible for the low hearing threshold [12]. Asymmetrically arranged ear flaps in front of the outer ear canals help to localize the elevation of sound sources. Furthermore, nuclei in the brain that process acoustic information are enlarged compared with other similar-sized birds [13]. With this set of adaptations, owls are able to detect and localize their prey precisely. Nevertheless, to be able to utilize this ability during flight, owls should fly silently. If that would not be the case, noise emitted by prey would be masked by the owl’s own flight noise and the prey might be warned by the approaching owl. Consequently, owls rely on a silent flight, which is achieved by adaptations of the wing, the plumage, and the flight behavior [4–10]. Specializations of the Owls’ Plumage Feathers are the main aerodynamic components of a bird’s wing. Interaction with the airflow during flight and protection against cold, heat, and wetness are essential functions. Furthermore, feathers are used for display or camouflage. A feather consists of a central shaft (rachis) and two laterally attached vanes. The feather vane is made up of parallel barbs that branch off from either side of the rachis. Barbs are interlinked via hook and bow radiates to form a closed surface. While each feather quill is inserted into the integument, all flight feathers (remiges) are additionally associated with the underlying skeleton. Secondary remiges are connected to the posterior edge of the ulna; primary remiges are supported by bones of the hand and fingers. Coverts arise from the anterior integument membrane, forming smooth and closed upper and lower wing surfaces. All wing feathers together form a wing that is light and flexible, but strong enough to meet the bird’s requirements to fly. In owls, several microstructured plumage specializations are known that influence the flow over the wing [4–8]: The dorsal surface of the feathers has an increased roughness caused by elongations of hook radiates, termed pennula. Those feathers that form the leading edge of the distal wing are equipped with serrations at their outer vanes. Every feather of the wing is surrounded by a fringed structure which is due to unconnected barb endings. Furthermore, owl
S
S
2414
feathers are very flexible and pliant due to a reduced number of radiates and thus a lower number of interlinks between the barbs compared to other birds [8]. Finally, cross-sectional profiles of the rachises are less structured compared to other birds of similar size. These plumage specifications have been implicated in noise suppression during flight by damping noise above 2 kHz [5]. Thus, flight noise is reduced within the typical hearing spectrum of the owls’ prey (>3 kHz) [5] and within the owl’s own best hearing range (5–9 kHz) [14]. The reduction of noise is mainly achieved by airflow control and friction reduction of single feathers.
Basic Methodology Morphometrics Bird wings, together with many other parameters, have been morphometrically characterized in order to classify bird species and to create a systematic order within birds (Aves). However, aerodynamic parameters are not included in this approach. Aerodynamic properties such as wing span, wing area, and aspect ratio can be measured in living birds or well-preserved museum specimens. Camber and thickness distribution are more prone to errors due to drying processes or unnatural fixation. Therefore, anesthetized or recently deceased birds yield better results. Best results are gained from experiments with living birds. Hereby, the wing geometry is obtained during free-flight experiments. Computer-aided analyses of three-dimensional reconstructions of the wings provide high-resolution profile data. Free-flight experiments are conducted to measure the noise emission of flying owls in different flight modes while the birds fly over an extreme sensible microphone array at a defined flight speed and height. The measured noise spectrum can then be correlated with the flight parameters. Finally, the results are compared to other non-silently flying birds [15]. Substructures of the feathers are investigated with different digital imaging techniques having high spatial resolutions. Light microscopy and scanning electron microscopy (SEM) enable to some extent two-dimensional measurements and morphometric characterizations. Confocal laser scanning microscopy (CLSM) and micro-computed-tomography (m-CT) allow the precise digitization of the three-dimensional
Silent Owl Wings
shape of microstructures. In all cases, the data are then processed and analyzed with adequate visualization and measuring software to determine relevant morphometric and aerodynamic parameters. After specifying the shape and geometry of the owl-specific structures, these can be artificially manufactured and applied to wing models for further analysis, e.g., in wind tunnel experiments. By this means, their aerodynamic function may be clarified. Behavioral Studies Birds use their wings to maneuver through the air. The form and geometry but also the movements of the wings are important for an efficient locomotion. Hence, studying the flight behavior of birds is one fundamental piece of the large puzzle of understanding how bird wings work. The study of wing motion in different flight modes, ideally in combination with flow visualization, helps to understand the functions of each wing component. Flight speed, wing-beat frequency, and wing-beat amplitude are some parameters that influence the flight of birds. Different flight modes like flapping flight, gliding, or soaring exist. Furthermore, some birds adapt their flight behavior during hunting, escaping, courtship, and other modes. Nowadays, high-speed cameras outperform traditional bird watching with binoculars and cameras in many aspects. Images with high temporal and spatial resolution of wing and feather movements during beating of the wings allow a much better investigation of each wing element and its aerodynamic function. Such techniques are used in field observations or wind tunnel experiments with living birds. Wind Tunnel Experiments Fixed wings of birds and artificial models of bird wings are investigated in wind tunnel experiments. Performing experiments in wind tunnels guarantee reliable and repeatable results. The use of wing models instead of living animals allows the manipulation of certain parameters of the wing. The impact of several wing geometries can then be compared and general conclusions can be drawn based on these results. Flow visualization techniques such as particle-image velocimetry (PIV), oil-flow pattern or pressure measurements are carried out to understand the influence of the wing profiling and each wing element on the airflow. In case of owl wings, the specific feather adaptations are added separately on a wing model or
Silent Owl Wings
are removed step by step from fixed owl wings. Their potential function is revealed by comparing wings either with or without the owl’s specializations in fluid-mechanical experiments.
Key Research Findings Owl Wings Owl wings differ from those of other birds. Figure 1a shows an example wing of a barn owl (Tyto alba). The large wing area, in combination with a relatively low body mass, results in a low wing loading allowing the owl to fly slowly and to carry large prey. Specific camber and thickness distributions (Fig. 1b, c) are also characteristic for owl wings [6]. While the proximal wing is highly cambered and anteriorly thickened, the distal wing is thin and less cambered. Anatomical studies revealed the construction plan of the wing (Fig. 1d). Skeletal elements, the muscle mass, and the coverts are responsible for the thick anterior part of the wing. By contrast, the distal wing and the posterior arm part of the wing are formed by remiges. These areas are extremely thin and lightweight. Hand and arm part of the wing have different functions (lift production and thrust) expressed by different geometries and profiling. At both wing parts, however, the airflow tends to separate in wind tunnel experiments when tested with a smooth surface without any surface or edge modifications [5, 16]. Owls, however, evolved microstructures on their feathers to influence the airflow around the wing in such a way that the airflow stays attached especially during critical flight maneuvers [5, 16] such as takeoff, landing, or striking. Hence, the wings produce a huge amount of lift even at low flight speeds. Owl Feathers Owl feathers share the anatomy of typical contour feathers [8]. However, they differ in some detail. As feathers are the main aerodynamic components of the bird wing, their geometry and dimensions affect the overall wing geometry. In owls, the feather vanes are large and at the same time extremely light and flexible. Fewer interlinks of the barbs cause these effects [8]. Hereby, owl feathers are very air-transmissive and flexible, whereby they can react rapidly to varying airflow conditions. Their bending behavior is passively driven. It depends on the material properties of feather
2415
S
keratin and the geometry of the rachis and the barbs, respectively [17]. Compared to other birds, owls stand out due to a large count of contour feathers [9]. Specifically, head, feet, and body are densely covered. This upholstery of the body serves the suppression of noise to some extent, but also thermal insulation since owls have only small fat reserves. The coloration of the plumage is mostly adapted to the environment of the relevant owl species with respect to an effective camouflage, but also to sexual display to some extent. Snowy owls (Bubo scandiacus), for instance, have a white plumage, while species that live in wooded environment have a brownish patterned plumage (see Fig. 2). Different color intensities between males and females are found, for instance, in barn owls with the females being slightly darker. The most conspicuous attribute of owl remiges are several anatomical microstructures that are responsible for the silent flight [4, 8]. Interestingly, fishing owls like the Malay fish owl (Bubo ketupu) lack such feather specialization [4]. Fish cannot hear the approaching owl. Hence, these owls do not need to fly silently. Microstructures of Owl Feathers Owls evolved several surface and edge modifications of their wings that are in turn specializations of the plumage. Three structures are mainly worth mentioning: the leading-edge comb-like serrations, the velvet-like upper surface, and the inner vane fringes [4, 7, 8]. Each feather that functions as a leading edge of the wing is equipped with comb-like serrations (Fig. 3). The number of remiges forming the leading edge varies among owl species, but is in the range of 1–3 remiges. Barb endings separate and bend upward, changing their form and function. Differences in the anatomical phenotype are small within one feather, though small variations occur (Fig. 3) [8]. While long serrations with small spacing are found in basal and central regions of the feathers, shorter serrations with a larger distance are located at the tip of the feather. Functionally, serrations affect the airflow and noise production only marginally during cruise flight [5]. The stagnation pressure at the leading edge of the wing prevents a functional airflow around the serrations. However, in critical flight maneuvers, such as landing or striking, the angle of attack is much higher and the serrations cause the airflow to stay attached to the wing. This is achieved by dividing the airflow into
S
S
2416
Silent Owl Wings
a Lesser secondary coverts Median secondary coverts Greater secondary coverts
Alular coverts Alula
Marginal coverts
10th greater primary covert Greater primary coverts
10th primary remex
10 9
15 8 14
7 13
12 11 10 1 9 8 7 6 4 3 2 Secondary remiges
b
d
1
3
2
6
5
4
Primary remiges
c
First digit Radiale & Carpometacarpus Ulnare Patagium Rachises of alular feathers Radius Postpatagium Second digit Ulna Rachis of 10th primary remex
Patagialis longus tendon
*
Humerus Humeral patagium
10 9 15
8 6
14 1
2
3
4
2 1 6 4 3 11 12 First primary remex 10 98 7 First secondary remex Secondary remiges
13
Silent Owl Wings, Fig. 1 Anatomy of a barn owl (Tyto alba) wing: (a) Topography of the dorsal wing (photograph). (b) Surface image of the digitized wing (surface scan). (c) Profiles of the owl wing in steps of 10% of the half wing
7
5 Primary remiges
span. (d) 3D reconstruction of skeletal elements, skin, and rachises of the remiges (processed CT scan). *The fifth secondary remex is missing in all owl species
Silent Owl Wings
2417
S
Silent Owl Wings, Fig. 2 Feathers of different owl species: (a) Tenth primary remex of a barn owl (Tyto alba) wing. The outer vane is equipped with tiny serrations, the inner vane is lined with fringes. (b) Contour feather from the body of a snowy owl (Bubo scandiacus). Note the symmetry of the vanes in comparison to the flight feathers (a) and (c) and the elongated plumulaceous barbs (fringes at the end of the vanes). (c) Primary remex of an eagle owl (Bubo bubo). As the eagle owl is the largest owl, so are its feathers. Scale bars represent 2.5 cm
S
Silent Owl Wings, Fig. 3 Serrations at the leading edge of a tenth primary of a barn owl (Tyto alba) in different magnifications: Barb endings separate and bow toward the upper side (a, b, c). Note also the fringes at the inner vane (a) and the dorsal surface texture
S
2418
Silent Owl Wings
Silent Owl Wings, Fig. 4 Scheme of the airflow at the leading edge of a distal bird wing: (a) Intact wing of a tawny owl (Strix aluco) in the region of the serrations. Note the laminar flow conditions around the wing. (b) After manual removing of
the serrations at the wing of the tawny owl, the flow separates early. (c) Flow conditions around a wing of a mallard duck (Anas platyrhynchos). The arrow indicates the flow direction; alpha is the angle of attack (Drawings after Neuhaus et al. [5])
many small micro-vortices that extent chordwise over the upper wing surface. As a consequence, the boundary layer is more energetic and flow separation is prevented (Fig. 4a). By this means, lift production is maintained even at high angles of attack and low flight speed. This effect gets lost after manual removal of the serrations at the leading edge (Fig. 4b). Here, the flow field around the wing resembles that of a mallard duck (Fig. 4c) which produces much more noise [5]. The second specialization is found on the upper wing surface. Each feather of the owl’s plumage is covered with a velvet-like dorsal texture (Fig. 5). This structure is formed by elongations of the hook radiates. Each hook radiate terminates in a filament structure, termed pennula. Due to their large number, the surface becomes very fluffy and porous. Functionally, two aspects are at least associated with this structure. On the one hand, the velvet-like dorsal surface of the inner vane, which is mostly covered by the adjacent feather, is very thick and well developed. The pennula are long and interconnect at large angles with the feather’s surface. This in turn causes a frictionreductive device on the upper surface and enables a smooth and silent gliding of the feathers. In all other birds investigated so far, the parallel-oriented barbs of adjacent feathers rub against each other producing a well-noticeable high-frequency sound, especially during flapping flight. In owls, however, such noises are prevented by the velvety surface. Those noises that still occur are dampened by the porous feather texture acting like acoustic absorbers. On the other hand, the velvet-like structure is also found at the outer vane. Here, the texture affects the aerodynamics of the wing. As mentioned above, the airflow tends to separate at wings with the specific camber and thickness distribution of owls and a smooth wing surface [16]. Figure 6 shows the PIV
results of the velocity distribution of the flow around an owl-based airfoil at Reynolds number 40,000, two different angles of attack (0 and 3 ), and two different surfaces. The velocity of the flow is color coded. At the wing with a clean and smooth surface (Fig. 6a, c), a separation bubble occurs on the suction side that increases with the higher angle of attack and leads to a complete flow separation beyond 3 angle of attack (not shown) [16]. Increasing the surface roughness as it is found in owls is one means in the direction of flow control. An artificial velvety surface texture applied to the owlbased wing geometry has a dramatic influence on the flow field (Fig. 6b, d). The velvety surface shifts the separation toward higher angles of attack indicated by a smaller separation bubble in comparison to the clean wing model. The boundary layer of the wing is controlled in such a way that occurring separation bubbles are reduced in size and shifted toward the leading edge of the wing. Consequently, the airflow is stabilized even at low flight speeds [16]. A further specialization can be found at the inner vanes of the remiges. Here, barb endings separate by a reduction of the hooklets of the hook radiates. Barbs are no longer interconnected and fringes are formed (Fig. 7). Since the tips of the inner vanes of remiges form the trailing edge of the wing, fringes are also found here. The appearance of fringes is very fluffy and unstable in shape. They are made up of three components: the central barb shaft and the laterally attached hook and bow radiates. The radiates support the formation of fringes by a parallel orientation to the barb shafts [8]. Since the barb shaft does not change its shape, as is the case for the serrations, the structure is pliant and the fringes can float freely resulting in a thin, almost two-dimensional structure. Functionally, fringes are assumed to decrease the sound intensity by a reduction of turbulence at the trailing edge of
Silent Owl Wings
2419
Silent Owl Wings, Fig. 5 Velvet-like dorsal surface of a barn owl’s fourth secondary in different magnification: Elongations of the hook radiates create this filigree texture which is supposed to influence the airflow and to reduce friction noise. (a, b) Photographs, (c) scanning electron microscope image
S
b
a
c
b
c
400 μm
z/c
0.10
*
0.05
0.20
z/c
0.15 0.10 0.05 0 0
0.20
3
0.15
*
0.15
2
#
1
α=0°
0 0
m/s 4
0.1
# 0.2
0.3
x/c
0.4
0.5
0.6
c
3
0.15
#
*
2 1
α=3°
# 0.2
0.3
x/c
0.4
0.5
0.6
0
0.10 0.05 0 0
#
0.1
2 #
0.2
0.3
x/c
0.4
0.5
0.6
*
#
3 2 1
α=3° 0.1
0
m/s 4
d *
3 1
α=0°
0 0 0.20
*
*
0.05
0
m/s 4
b
0.10
m/s 4
*
0.1
z/c
a
z/c
0.20
# 0.2
0.3
x/c
0.4
0.5
0.6
0
Silent Owl Wings, Fig. 6 Averaged velocity fields at Re ¼ 40,000 and an angle of attack of 0 and 3 for a clean wing (a, c) and the same wing geometry with a velvety surface (b, d). The velocity of the airflow is color coded (online version)
the wing [7, 18]. While sharp edges of wings and flat plates are known to produce a well-noticeable noise even at low Reynolds numbers [19], fringed trailing edges of owl remiges prevent such phenomena. Each wing feather has its own trailing edge that produces noise during flight independently from being part of
the wing or separated, for example, during flapping flight. In general, fringes act as a pressure release mechanism that interrupt scattering phenomena which can be found when turbulent eddies pass over the posterior wing region that gradually transitions from the loaded wing to the freestream conditions
S
S
2420
Silent Owl Wings
Silent Owl Wings, Fig. 7 Fringes at the inner vane of a barn owl’s fifth primary in different magnification:Hooklets of the hook radiates are absent causing the barbs to separate. Hook and bow radiates change their length and thereby support the formation of fringes. (a) and (b) dorsal, (c) ventral
[18, 19]. Without such fringes, turbulent boundary layers being convected past the trailing edge into the wake would generate an intense, broadband scattering noise. In owls, fringes prevent sharp edges at the trailing edge such that the turbulent boundary layer is merged smoothly into the airflow. Behavioral Adaptations Nocturnal owls are unable to benefit from thermal upwind like other birds of prey. Hence, owls combine active flapping flight and gliding phases [9, 10]. The large wing area and the specific profiling of the wing enable the owl to glide and maneuver with little movements of the wing. The flight speed is relatively low and depends on the illumination of the environment. Barn owls (Tyto alba), for instance, can reduce their flight speed down to 2.5 m/s [9, 10] which per se leads to a silent flight, since only little turbulences occur at that velocity. Apart from the slow gliding flight, some adaptations in the motion of the owl already prevent rising noises. During hunting, owls reduce their wing beat frequency and amplitude [9]. Consequently, less friction between feathers occurs during flight which would cause rubbing noises. Those noises that still rise are efficiently dampened by the porous upper wing surface [4, 7].
of noise pollution in the last decades. At ground level, airplanes produce an intense noise by their engines and high-lift configurations of the wings. Furthermore, wind farms are constantly upgraded. Their noise emission does not only annoy humans, but also disturbs migratory animals such as birds, or in case of offshore wind farms, whales, and fishes. How flight noise can efficiently be reduced is demonstrated by owls. These birds evolved noise-reduction and noise-suppression devices in the course of evolution. The implementation of the underlying mechanisms in modern aircraft and rotors of wind-power plants would help to reduce noise pollution. Until then more studies and experiments have to be carried out. In joint biological and fluidmechanical efforts, the basic aerodynamic characteristics and the fluid-mechanical details that define the drag and noise reduction mechanisms of owl wings have to be investigated in more detail. This will be done using preserved material from zoological collections, experiments with living owls, and wing models in wind tunnel experiments. The knowledge gained will be used to create devices that operate at higher Reynolds numbers, since copying the natural structure alone would not deliver satisfying results. In the long run, designing silent wings of modern aircraft will bring us closer to the goal of a quieter environment.
Future Directions for Research
Cross-References
Urbanization and the increasing demand for air freight and renewable energy have led to a dramatic increase
▶ Biomimetic Flow Sensors ▶ Confocal Laser Scanning Microscopy
Simulations of Bulk Nanostructured Solids
▶ Friction-Reducing Sandfish Skin ▶ Insect Flight and Micro Air Vehicles (MAVs) ▶ Scanning Electron Microscopy ▶ Shark Skin Drag Reduction
References 1. Pennycuik, C.J.: Wing beat frequency of birds in steady cruising flight: new data and improved predictions. J. Exp. Biol. 199, 1613–1618 (1996) 2. Rayner, J.M.V.: On aerodynamics and the energetics of vertebrate flapping flight. Contemp. Math. 141, 351–400 (1993) 3. R€uppell, G.: Vogelflug. Rowohlt Taschenbuchverlag GmbH, Hamburg, Germany (1980) 4. Graham, T.: The silent flight of owls. J. Roy. Aero. Soc. 38, 837–843 (1934) 5. Neuhaus, W., Bretting, H., Schweizer, B.: Morphologische und funktionelle Untersuchungen € uber den lautlosen Flug der Eule (Strix aluco) im Vergleich zum Flug der Ente (Anas platyrhynchos). Biol. Zentr. Bl. 92, 495–512 (1973) 6. Nachtigall, W., Klimbingat, A.: Messungen der Fl€ugelgeometrie mit der Profilkamm-Methode und geometrische Fl€ ugelkennzeichnung einheimischer Eulen. In: Nachtigall, W. (ed.) Biona Report 3, pp. 45–86. Gustav Fischer Verlag, Stuttgart und New York (1985) 7. Lilley, G.: A study of the silent flight of the owl. AIAA Paper, pp. 98–2340 (1998) 8. Bachmann, T., Kl€an, S., Baumgartner, W., Klaas, M., Schro¨der, W., Wagner, H.: Morphometric characterisation of wing feathers of the barn owl (Tyto alba) and the pigeon (Columba livia). Front. Zool. 4, 23 (2007) 9. Mebs, T., Scherzinger, W.: Die Eulen Europas. FranckhKosmos, Stuttgart, Germany (2000) 10. Taylor, I.: Barn Owls: Predator – Prey Relationship and Conservation. Cambridge University Press, Cambridge (1994) 11. Knudsen, E.: The hearing of the barn owl. Sci. Am. 245(69), 113–125 (1981) 12. Hausmann, L., Campenhausen, V.M., Endler, F., Singheiser, M., Wagner, H.: Improvements of sound localization abilities by the facial ruff of the barn owl (Tyto alba) as demonstrated by virtual ruff removal. PLoS One 4(11), e7721 (2009) 13. Konishi, M., Takahashi, T., Wagner, H., Sullivan, W., Carr, C.: Neurophysiological and anatomical substrates of sound localization in the owl. In: Auditory Function – Neurobiological Bases of Hearing, pp. 721–745. Wiley, New York (1988) 14. Konishi, M.: How the owl tracks its prey. Sci. Am. 61, 414–424 (1973) 15. Sarradj, E., Fritzsche, C., Geyer, T.: Silent owl flight: Bird flyover noise measurements. AIAA Paper 2010–3991 (2010) 16. Kl€an, S., Bachmann, T., Klaas, M., Wagner, H., Schro¨der, W.: Experimental analysis of the flow field over a novel owl based airfoil. Exp. Fluids 46, 975–989 (2008) 17. Bonser, R.H., Purslow, P.: The young’s modulus of feather keratin. J. Exp. Biol. 198, 1029–1033 (1995)
2421
S
18. Herr, M.: Experimental study on noise reduction through trailing edge brushes. Note. N. Fl. Mech. Mul. D. 92/2006, 365–372 (2006) 19. Moreau, D.J., Brooks, L.A., Doolan, C.J.: On the aeroacoustic tonal noise generation mechanism of a sharp-edged plate. JASA E. L (2011). doi:10.1121/1.3565472
Silica Gel Processing ▶ Sol-Gel Method
Silicon Microphone ▶ Electrostatic MEMS Microphones
Siloxanes ▶ BioPatterning
Silver (Ag) ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles
Simulations of Bulk Nanostructured Solids Donald W. Brenner Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA
Synonyms Plasticity of nanostructured solids
Definition A bulk nanostructured solid is a material available in macroscopic quantities that has nanometer-scale structural features in at least one dimension.
S
S
2422
Computer modeling has emerged to become a third pillar of research alongside experiment and theory. There are many reasons for this emergence, including an exponential increase in processing speed, relatively inexpensive platforms for parallel computing and data storage, new visualization capabilities, and the development of powerful algorithms that take full advantage of these advances in hardware. Advances in simulation methodologies have also made the results of atomiclevel computer modeling reliable to the extent that in many cases simulations can replace difficult and expensive experiments. Methods that enable atomistic dynamics using full electronic energies now allow processes involving up to several thousand atoms to be accurately modeled, with significantly larger systems on the horizon. Similarly, advances in materials theory have produced relatively simple analytic potential energy functions that can capture essentials of quantum mechanical bonding for simulations involving well over a billion atoms [1]. Computer modeling has played an especially important role in developing the current understanding of nanometer-scale structures and processes [2]. Indeed, among the many scientific and technological advances coming from nanotechnology is the ability for computer modeling and experiment to characterize phenomena on a common scale. Atomic-level computer modeling is now commonly used to explore and even predict new properties and processes that can be probed experimentally, to suggest new materials and structures with unique and desirable properties, to provide insight into the results of experiments, to generate data for larger-scale analysis, and to test scaling laws and analytic theories. In the case of nanostructured solids, computer modeling is allowing researchers to understand at the atomic level the structure, deformation mechanisms, and thermal-mechanical properties of these new materials with unprecedented detail [3, 4]. The two standard methods for modeling nanometerscale systems are molecular dynamics and Monte Carlo simulation. In the former, coupled classical equations of motion for the atoms are integrated stepwise in time. Time steps generally range from one-half to tens of femtoseconds depending on the highest frequency vibration of interest, and simulations are typically carried out for picoseconds to tens of nanoseconds depending on system size, the phenomena to be studied, and the method in which interatomic forces are calculated. In a typical equilibrium Monte Carlo
Simulations of Bulk Nanostructured Solids
simulation, atomic configurations are generated with a probability that is proportional to their Boltzmann factor. Thermodynamic quantities in a Canonical ensemble are then obtained by averaging over the properties of each configuration. Alternatively, in kinetic Monte Carlo simulation, time-ordered configurations are generated, typically using some rate expression. Equilibrium Monte Carlo modeling requires specifying a potential energy as a function of atomic positions to calculate Boltzmann factors, while molecular dynamics simulations require interatomic forces, which are typically obtained as partial derivatives of the potential energy. In general, two approaches are used to calculate interatomic energies and forces. In the least computationally intensive approach, the interactions between atoms due to the electrons are replaced with effective interactions that are described with an analytic potential energy function. At present, there is no definitive mathematical form for the potential energy function, and forms that range from relatively simple pairadditive expressions to complicated many-body forms are used depending on the system, type of bonding, and phenomena to be studied. In the second approach to calculating potential energies, explicit electronic degrees of freedom are retained, the energy for which is calculated either using first principles methods or through a simplified semiempirical Hamiltonian. The calculation of total energies from first principles is well defined in terms of basis sets, electron correlation for ab initio methods, or choice of density functional expression (and pseudopotential) for density functional theory calculations. This is in contrast to analytic potentials, where a set of parameters (and often entirely new functional forms) must be developed for each system. Nanostructured materials can be defined as materials that have at least one dimension in the “nanoscale” (typically 1–100 nm). Depending on which dimension this is, they can be classified into nanoparticles, layered or lamellar structures, filamentary structures, and bulk nanostructured materials. This entry is focused on the latter, which can be thought of as a traditional material with grain sizes at the nanometer scale. The nanometer-scale grains introduce unusual properties compared to more traditional micron-scale grain sizes, including unique combinations of strength and ductility as discussed in more detail below. The origin of these differences, and how they are being revealed by computer modeling, makes up the bulk of the material discussed below.
Simulations of Bulk Nanostructured Solids
Atomic positions for bulk nanostructured solids have been generated by a number of different methods. Some of the methods are based on geometrical constraints imposed by the simulation conditions, for example, ensuring that active slip systems are properly oriented with respect to periodic boundaries [5, 6]. Crystallization dynamics have also been used to generate nanostructures. These methods can be based on a Johnson-Mehl or a Potts model construction, both of which produces a log normal grain size distribution, or by using a molecular dynamics simulation to model crystallization from a melt [3, 7, 8]. Other researchers have generated nanostructures by simulating compaction and sintering of nanoclusters [9–11]. Other methods use random grain centers, with grain boundaries chosen based on a Voronoi construction. Variations on this method include picking grain orientations to produce a particular range of tilt angle (e.g., low angle grains), or picking grain centers to produce a log normal distribution of grain sizes [4, 12]. Many of the atomic simulations of sintering and grain growth dynamics – and many of the simulation studies in general on nanostructured materials – have focused on understanding how processes and rates at the nanometer scale differ from their counterparts in macroscopic-scale systems. At the macroscopic scale, six distinct mechanisms contribute to the sintering dynamics of crystalline particles: surface diffusion; lattice diffusion from the surface, from grain boundaries and through dislocations; vapor transport; and grain boundary diffusion. There is a much larger degree of surface curvature at the nanometer scale and a much higher ratio of interface to bulk atoms, both of which may lead to very different sintering mechanisms. The details regarding these differences, and some new and unexpected effects at the nanometer scale, have been revealed from computer simulations. Molecular dynamics simulations, for example, have been used to study surface energies, grain boundary mobility, and sintering of metal nanoparticle arrays at different temperatures [13, 14]. The results suggest that of the macroscopic-scale mechanisms associated with sintering, only surface and grain boundary diffusion contribute significantly to nanometer-scale sintering dynamics, and that these two processes are accelerated due to the large interfacial forces. Simulations have also suggested three unconventional mechanisms that contribute to the early stages of nanometer-scale sintering: mechanical rotation, plastic
2423
S
deformation via dislocation generation and transmission, and amorphization of subcritical grains. This grain rotation mechanism has also been observed in structures with columnar grains, where simulations suggest that necessary changes in the grain shape during grain rotation in the nanostructure can be accommodated by diffusion either through the grain boundaries or through the grain interior [15]. In conventional metals with micron-scale grains, plastic deformation occurs by motion of dislocations. This dislocation motion can be inhibited by grain boundaries, which leads to the well-established Hall-Petch relation that the yield strength is proportional to the inverse square root of the grain size. However, there appears to be a threshold grain size below which materials become softer with decreasing grain size. This so-called inverse Hall-Petch behavior has been attributed to a transition from dislocationmediated plastic deformation to grain boundary sliding for some critical grain size. This transition has been observed in several sets of molecular dynamics simulations that predict a transition grain size of about 10–15 nm in good agreement with experimental measurements. At the same time, the simulations have also revealed a rich and unanticipated set of dynamics near the threshold region that can be related to the fundamental properties of the bulk materials. These unique dynamics include an enhanced role of grain rotation (analogous to sintering dynamics), cooperative intergrain motion, and formation of stacking faults via motion of partial dislocations across grains. The deformation of strained nanocrystalline copper with grain sizes that average about 5 nm has been studied with molecular dynamics simulations. These simulations showed a material softening for small grain sizes, in agreement with experimental measurements. These simulations suggest that plastic deformation in the inverse Hall-Petch region occurs chiefly by grain boundary sliding with dislocation motion having a minimal influence on deformation mechanisms. In related studies, molecular dynamics simulations have been used to understand the deformation of nanostructured nickel and copper with grain sizes ranging from 3.5 to 12 nm [16]. For grain sizes less than about 10 nm, deformation occurred mainly by grain boundary sliding, while for the larger grain sizes, deformation occurred by a combination of dislocation motion and grain boundary sliding. Detailed mechanisms of strain accommodation have been characterized; these
S
S
2424
include both single atom motion and correlated motion of several atoms, as well as stress-assisted free volume migration [17]. Molecular simulations of the deformation of columnar structures of aluminum have shown emission of partial dislocations during deformation that form at grain boundaries and triple junctions [18]. Atomic simulations have also suggested that these structures can be reabsorbed upon removal of the applied stress, which may contribute to an inability to observe dislocations experimentally in systems of this type after external stresses are released. In addition, near the grain size where plastic deformation transitions from dislocation-mediated plasticity to grain boundary sliding, the motion of single partial dislocations across nanograins during tensile loading has been observed in simulations. Without emission of a trailing partial dislocation, an intrinsic stacking fault is formed along the width of the nanograin. From the simulations, it appears that the formation of a low-energy ordered grain boundary drives emission of a full dislocation and the resulting absence of a stacking fault. It has been argued that nucleation of the initial partial dislocation and the atomic rearrangement at the grain boundary associated with its emission sufficiently lowers the grain boundary energy such that emission of the trailing partial dislocation is not always needed to further relax the system [19]. Based on simulations of aluminum with a columnar nanostructure, it has been further suggested that the stacking fault width, and hence the intrinsic stacking fault energy, as determined by the distance between two partial dislocations, is the critical quantity that defines the transition from full to partial dislocation emission as grain sizes approach the critical size for the onset of inverse Hall-Petch behavior [20]. On the other hand, it has been argued that the relation between the emission of partial dislocations does not correlate well with calculated stacking fault energies for nickel, copper, and aluminum. Instead, it has been suggested that full dynamics associated with the nucleation of a partial dislocation from a grain boundary must be considered, and therefore that the full planar fault energy, which includes the stable and unstable stacking fault energy as well as twin fault energies, must be used to understand and ultimately predict the relation between mechanical deformation and grain size. Extensive simulations of crack propagation in nanostructured metals have also been carried out to better
Simulations of Bulk Nanostructured Solids
understand how fracture, fatigue, and toughness depend on grain scale [21–23]. These simulations have revealed crack propagation mechanisms that are similar to the plastic response of fully dense samples as well as key differences resulting from the presence of the crack tip. For example, in simulations of nanocrystalline nickel with grain sizes in the inverse Hall-Petch region, mode I crack propagation occurred by inter-grain decohesion via a mechanism involving coalescence of nanovoids that form in front of the crack tip. Plastic deformation leading to both full and partial dislocations was also observed in the neighboring grains. Compared to their mechanical properties under tensile loading, much less is understood about the influence of grain size on the shock loading properties of nanostructured solids. Atomic simulations that have been carried out, however, suggest a strong coupling between nanostructure and shock loading dynamics, as well as unique and very important properties of shocked-loaded nanostructured metals. It has been noted, for example, that the mechanisms associated with the mechanical deformation of nanostructured metals depend strongly on pressure, temperature, and strain rate, and therefore, these materials may show ultrahigh strength under shock loading depending on the shock loading conditions and system [24]. The fast temperature and pressure rises associated with shock fronts freeze out deformation mechanisms that require diffusion. Similarly, production of dislocations that requires nucleating events is inhibited. In the case of grain boundary accommodation, increasing the pressure results in an increase in the threshold stress for sliding plasticity. Similarly, the threshold for dislocation plasticity increases with increasing pressure because of an increase in the shear modulus with increasing pressure. Taking these effects into account, and assuming that the maximum in hardness as a function of grain size occurs where stress for sliding and dislocation plasticity are equal, it has been shown that ultrahigh hardness can be achieved by shock loading of nanostructured solids. These arguments have been validated by using molecular dynamics simulations to model the shock loading of nanostructured copper with different grain sizes. At relatively low shock velocities (i.e., low stresses), grain boundary sliding is observed, which results in a relatively low hardness value that increases with increasing grain size. At intermediate stresses, the hardness of the
Simulations of Bulk Nanostructured Solids
copper increases with increasing shock strength for all grain sizes, with a shift in the maximum hardness toward lower grain sizes compared to deformation at lower strain rates. This leads to a net increase in the maximum hardness of the material. At even higher stresses, simulations predict a drop in strength due to an increase in temperature and an associated increase in dislocation nucleation and motion. It is clear from the discussion in the preceding sections that atomic simulations have provided new and exciting insights into the unique properties of nanosystems in general and nanostructured solids in particular. This remains a very active area of research within which molecular simulation will continue to provide new insights into relations between structural mechanical and thermodynamic properties.
2425
7.
8.
9.
10.
11.
12.
Cross-References ▶ Computational Study of Nanomaterials: From Large-scale Atomistic Simulations to Mesoscopic Modeling ▶ Molecular Dynamics Simulations of Nano-Bio Materials ▶ Nanomechanical Properties of Nanostructures
13.
14.
15.
16.
References 17. 1. Brenner, D.W., Shenderova, O.A., Areshkin, D.A.: Quantum-based analytic interatomic forces and materials simulation. In: Lipkowitz, K.B., Boyd, D.B. (eds.) Reviews in Computational Chemistry, pp. 213–245. VCH, New York (1998) 2. Brenner, D.W., Shenderova, O.A., Schall, J.D., Areshkin, D. A., Adiga, S., Harrison, J.A., Stuart, S.J.: Contributions of molecular modeling to nanometer-scale science and technology, Chapter 24. In: Goddard, W., Brenner, D., Lyshevski, S., Iafrate, G. (eds.) Nanoscience, Engineering and Technology Handbook. CRC Press, Boca Raton (2002) 3. Wolf, D., Yamakov, V., Phillpot, S.R., Mukherjee, A., Gleiter, H.: Deformation of nanocrystalline materials by molecular dynamics simulations: relation to experiment? Acta Mater. 53, 1 (2005) 4. Kumar, K.S., Van Swygenhoven, H., Suresh, S.: Mechanical behavior of nanocrystalline metals and alloys. Acta Mater. 51, 5743 (2003) 5. Van Swygenhoven, H., Spaczer, M., Car, A., Farkas, D.: Competing plastic deformation mechanisms in nanophase metals. Phys. Rev. B 60, 22 (1999) 6. Yanakov, V., Wolf, D., Salazar, M., Phillpot, S.R., Gleiter, H.: Length-scale effects in the nucleation of extended
18.
19.
20.
21.
22.
23.
S
dislocations in nanocrystalline Al by molecular-dynamics simulation. Acta Mater. 49, 2713 (2001) Phillpot, S.R., Wolf, D., Gleiter, H.: Molecular-dynamics study of the synthesis and characterization of a fully dense, three-dimensional nanocrystalline material. J. Appl. Phys. 78, 847 (1995) Phillpot, S.R., Wolf, D., Gleiter, H.: A structural model for grain boundaries in nanocrystalline materials. Scripta Metall Mater. 33, 1245 (1995) Zhang, Y.W., Liu, P., Lu, C.: Molecular dynamics simulations of the preparation and deformation of nanocrystalline copper. Acta Mater. 52, 5105 (2004) Kodiyalam, S., Kalia, R., Nakano, A., Vashashta, P.: Multiple grains in nanocrystals: effect of initial shape and size on transformed structures under pressure. Phys. Rev. Lett. 93, 203401 (2004) Chatterjee, A., Kalia, R.K., Nakano, A., Omeltchenko, A., Tsuruta, K., Vashishta, P., Loong, C.-K., Winterer, M., Klein, S.: Sintering, structure, and mechanical properties of nanophase SiC: a molecular-dynamics and neutron scattering study. Appl. Phys. Lett. 77, 1132 (2000) Schiotz, J., Di Tolla, F.D., Jacobsen, K.W.: Softening of nanocrystalline metals at very small grain sizes. Nature 391, 1223 (1998) Zeng, P., Zajac, S., Clapp, P.C., Rifkin, J.A.: Nanoparticle sintering simulations. Mater. Sci. Eng. A 252, 301 (1998) Xiao, S., Hu, W.: Molecular dynamics simulations of grain growth in nanocrystalline Ag. J. Crystal Growth 286, 512 (2006) Haslam, A.F., Phillpot, S.R., Wolf, D., Moldovan, D., Gleiter, H.: Mechanisms of grain growth in nanocrystalline fcc metals by molecular-dynamics simulation. Mater. Sci. Eng. A 318, 293 (2001) Van Swygenhoven, H., Spazcer, M., Caro, A., Farkas, D.: Competing plastic deformation mechanisms in nanophase metals. Phys. Rev. B 60, 22 (1999) Van Swygenhoven, H., Derlet, P.M.: Grain boundary sliding in nanocrystalline fcc metals. Phys. Rev. B 64, 224105 (2001) Yamakov, V., Wolf, D., Salazar, M., Phillpot, S.R., Gleiter, H.: Length scale effects in the nucleation of extended dislocations in nanocrysalline Al by molecular dynamics simulation. Acta Mater. 49, 2713 (2001) Van Swygenhoven, H., Derlet, P.M., Froseth, A.G.: Stacking fault energies and slip in nanocrystalline metals. Nat. Mater. 3, 399 (2004) Yamakov, V., Wolf, D., Phillpot, S.R., Mukherjee, A.K., Gleiter, H.: Deformation mechanism map for nanocrystalline metals by molecular dynamics simulation. Nat. Mater. 3, 43 (2004) Latapie, A., Farkas, D.: Molecular dynamics investigation of the fracture behavior of nanocrystalline a-Fe. Phys. Rev. B 69, 134110 (2004) Farkas, D., Van Petegem, S., Derlet, P.M., Van Swygenhoven, H.: Dislocation activity in nano-void formation near crack tips in nanocrystalline Ni. Acta Mater. 53, 3115 (2005) Farkas, D., Sillemann, M., Hyde, B.: Atomistic mechanisms of fatigue in nanocrystalline metals. Phys. Rev. Lett. 94, 165502 (2005)
S
S
2426
24. Bringa, E.M., Caro, A., Wang, Y., Victoria, M., McNaney, J.M., Remington, B.A., Smith, R.F., Torralva, B.R., Van Swygenhoven, H.: Ultrahigh strength in nanocrystalline materials under shock loading. Science 309, 1838 (2005)
Single Cell Analysis ▶ Electrical Impedance Tomography for Single Cell Imaging
Single Cell Impedance Spectroscopy ▶ Electrical Impedance Cytometry
Single-Cell Electrical Impedance Spectroscopy ▶ Single-Cell Impedance Spectroscopy
Single-Cell Impedance Spectroscopy Jian Chen1, Nika Shakiba1, Qingyuan Tan1 and Yu Sun2 1 Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada 2 Department of Mechanical and Industrial Engineering and Institute of Biomaterials and Biomedical Engineering and Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada
Single Cell Analysis
signal on a single cell positioned in between two measurement microelectrodes. The current response is measured as a function of frequency. By interpreting the impedance profile, dielectric properties of a single cell such as cell membrane capacitance and cytoplasmic resistance are obtained.
Overview Historical Development In conventional cell electrical impedance spectroscopy, the impedance of multiple cells as a whole is measured using an AC excitation signal. The cell suspension is held in a container with two electrodes. Current is measured as a function of frequency to determine the electrical properties of the cells in the suspension. Recent advances in microfabrication and lab-on-achip technologies enable the development of electrical impedance spectroscopic devices to measure impedance profiles of cells at the single-cell level, providing useful biophysical characteristics of single cells and promising potentially noninvasive, label-free approaches for analyzing cells. Working Principle Single-cell impedance spectroscopy measures impedance profiles of a cell positioned in between two micro electrodes [1]. A low-magnitude AC voltage, U ðjoÞ, over a range of frequencies is used as the excitation signal. The current response, I ðjoÞ, is measured. The impedance Z ðjoÞ is Z ðjoÞ ¼
Synonyms Impedance measurement of single cells; Impedance spectroscopy for single-cell analysis; Single-cell electrical impedance spectroscopy
Definition Single-cell impedance spectroscopy is a technique that operates by applying a frequency-dependent excitation
U ðjoÞ
I ðjoÞ
¼ ZRE þj ZIM
(1)
where ZRE and ZIM are the real and imaginary parts of impedance. The real part is termed resistance while the imaginary part is termed reactance. The magnitude and phase angle are qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Z~ ¼ Z~2 þ Z~2 RE
IM
(2)
and
ff Z ¼ arctan
Z IM
Z RE
! (3)
Single-Cell Impedance Spectroscopy
Advantages and Applications Compared to impedance measurement on a cell suspension, single-cell impedance spectroscopy interrogates the property of a single cell (vs. a cell population). Theoretical analysis can also become simpler because it is not necessary to take into account electrical interactions among cells [2]. The technique of single-cell impedance spectroscopy has been used as a noninvasive method to quantify the physiological state of single cells. As a biophysical marker, single-cell impedance profiles have also been utilized to distinguish normal cells from abnormal cells (e.g., human cancer cells with different metastatic potential and malaria-infected red blood cells) [3, 4].
2427
a
S
Excitation Electrode Glass RDL CDL Cmem Cm
Rmem
Ri
Rm
Cmem
Rmem
CDL RDL Glass
Methodology Equivalent Circuit Model Figure 1a [1] shows an equivalent circuit for interpreting single-cell impedance measurement data [1]. In this circuit model, the cell is represented by a capacitor in series with a resistor, with elements connected in parallel with the capacitor and resistor (i.e., the cell) representing the suspending medium. At low frequencies, the thin cell membrane gives rise to a large capacitance. As the frequency increases, the reactive component of this element tends to zero out, and the cell internal properties are represented by the resistor. In this circuit model, the membrane is assumed to have a high resistance and the cytoplasmic capacitance is ignored. All the circuit elements are functions of the volume fraction or cell size. The electrical double layer has an influence on measurements at low frequencies (below 1 MHz for high-conductivity solutions). This is generally modeled as a constant phase angle, represented by a resistor (RDL) and capacitor (CDL) in series. As shown in Fig. 1a, the double layer is in series with the network model. In the simplified model shown in Fig. 1b, the double layer is assumed to be purely capacitive (CDL), with a value given by the product of the inverse Debye length and the permittivity of the medium. The simplified circuit in Fig. 1b [1] shows that at very low frequencies current flow is blocked by the double layer capacitor, and only the impedance of
Sensing Electrode Rm
b CDL
Cmem
Ri
CDL
Cm
Single-Cell Impedance Spectroscopy, Fig. 1 (a) The equivalent circuit model for single-cell impedance spectroscopy. RDL and CDL represent the resistance and capacitance of the electrical double layer, Rm and Cm the resistance and capacitance of the medium, Rmem and Cmem the resistance and capacitance of the cell membrane and Ri the resistance of the cytoplasm. (b) The simplified equivalent circuit model neglecting the electrical double layer resistance and the membrane resistance (Reproduced with permission from [1])
S the double layer is measured. As the measurement frequency increases, this capacitor is gradually shortcircuited and the excitation voltage charges the cell in suspension. It takes a finite amount of time to charge the membrane through the extracellular and intracellular fluid, resulting in two dielectric dispersions in the frequency range of 1–100 MHz. The lower frequency dispersion is governed by the polarization of the cell membrane. Measurement of this parameter provides information about the dielectric properties of the membrane. The higher-frequency dispersion is governed by the polarization of the cytoplasm and the suspending medium as the membrane is short-circuited at these
S
2428
Single-Cell Impedance Spectroscopy
frequencies. This second dispersion is generally small and difficult to measure using impedance spectroscopy techniques. With the effect of the double layer taken into account, the total impedance of the circuit is
Z ðjoÞ ¼
Analysis Cavity
Trapped Cells Fluid Interconnects
Via Impedance Electrodes
1 joCDL
Polyimide
Rm ð1 þ joRi Cmem Þ þ joRm Cmem þ ð1 þ joRi Cmem Þð1 þ joRm Cm Þ
Silicon Detection Electrodes
(4)
Single-Cell Impedance Spectroscopy on Stationary or Moving Cells Single-cell impedance spectroscopy in microfluidic devices can be divided into two main categories: measurements on either stationary or moving cells. For measuring a stationary cell, a cell can be positioned/ trapped in between microelectrodes using several approaches. Cells can also be controlled to pass the measurement electrodes at the speed of microfluidic flow. 1. Cell aspiration [5] (a) Working mechanism: In the aspiration approach, a negative pressure is applied to trap a single cell in a channel opening. Measurement electrodes can be built into such channel openings. A tapped cell is forced by the applied negative pressure to contact the electrodes (Fig. 2) [5]. The impedance profile of the cell is then obtained. (b) Advantages: The main advantage of the aspiration approach for capturing a cell is the seal formed between the measurement electrodes and the cell. The suction force is sufficient to hold the cell in place throughout the measurement process, especially when the cellcapturing channel has a more or less circular cross section. (c) Potential limitations: It has been suggested that mechanical deformations of a cell due to the suction force may lead to changes in the electrical properties of the cell, thus, affecting the impedance profile measured. 2. Hydrodynamic trapping [6] (a) Working mechanism: In hydrodynamic trapping, cells are trapped by microfabricated weirs, used as filters on top of electrodes. By
SiO2
Control Fluid Suction
Plastic
Single-Cell Impedance Spectroscopy, Fig. 2 Schematic view of an aspiration-based single-cell impedance spectroscopy system. The system is composed of an array of analysis cavities, each capable of analyzing a single cell at a time, each containing a mechanical fluid via with flow control to hold the cell in position during the impedance analysis operation, and multiple electrodes surrounding the circumference of the cell to measure the electrical characteristics of the captured cell (Reproduced with permission from [5]) The captured single cell
Au electrode
Flow Pillar microstructures
Single-Cell Impedance Spectroscopy, Fig. 3 Schematic view of a microfluidic device for single-cell impedance spectroscopy using the concept of hydrodynamic trapping for single-cell immobilization (Reproduced with permission from [6])
adjusting flow rates, cells can be captured on measurement electrodes for impedance data recording (Fig. 3) [6]. (b) Advantages: The main advantage of this approach is the feasibility to realize largearray single-cell trapping and potentially high-throughput impedance measurement. (c) Potential limitations: Contact between the trapped cell and the measurement microelectrodes underneath can be problematic since there is no force to hold the cell in close contact with the microelectrodes.
Single-Cell Impedance Spectroscopy Single-Cell Impedance Spectroscopy, Fig. 4 Schematic view of differential impedance micro flow cytometry. Each cell’s impedance signal is recorded by a differential pair of microelectrodes using media without a single cell passing as a reference. In this setup, the effect of the electrical double layer on impedance profiles is canceled (Reproduced with permission from [7])
2429
S
Function generator
Cell In
Out
− +G
Electrodes
A D
Objective
BC
Data acquisition
BE
488nm Lasers 633nm PMT 1 BE : beam expander BC : beam combiner : dichroic mirror
PMT 2
: pinhole : filter : lens PMT 3
3. Flow cytometry [7] (a) Working mechanism: Flow cytometry is a technique that allows for the analysis of cells in suspension with single-cell precision. A laminar flow carries suspended cells through the measurement location. Each cell’s impedance signal is recorded by a differential pair of microelectrodes, using the surrounding media without a cell as reference. Microfabricated devices and electronic circuits allow simultaneous impedance measurements at multiple frequencies (Fig. 4) [7]. (b) Advantages: Flow cytometry is advantageous in that it allows for rapid analysis of cells in suspension with a high throughput. Furthermore, flow cytometry can also allow for the sorting of cells into subpopulations based on their impedance measurement profiles. (c) Potential limitations: There is no contact between the cell and measurement electrodes
and thus, the impedimentary contribution of the current leakage in the medium around the cell is significant.
Key Research Findings Effect of Electrical Double Layer on Impedance Profile When a polarizable electrode (i.e., an electrode operated at a regime where no charge transfer reaction occurs at the surface) comes into contact with an electrolyte, charges from the electrolyte opposite in sign to the charges present on the surface of the electrode move the electrode/electrolyte interface and provide a localized condition of electroneutrality as well as a layer of charge, termed the electrical double layer. The electrical double layer has an influence on cell impedance measurements at low frequencies, which is generally represented as a capacitor. The most
S
S
2430
effective way to minimize the electrical double layer is to maximize the electrolyte-electrode interface area. This interface area enhancement can be achieved either by mechanically roughing the electrode surface to an electrode-electrolyte interface area that is effectively larger than the actual electrode surface, or to use chemical treatments that lead to a high electrode-electrolyte interface area. Comparison of Impedance Measurement Mechanisms The essential feature of measurements on a stationary cell trapped/positioned between electrodes is a tight seal between the cell and electrodes. In this configuration, the impedimentary contribution of the cell shunt (i.e., current leakage in the medium around the cell) is insignificant. The advantage of impedance cytometry is differential impedance measurement that minimizes the effect of the electrical double layer. In an alternating fashion, each measurement electrode-ground electrode pair without a cell serves as a reference to the other pair, over which a cell passes.
Single-Walled Carbon Nanotubes (SWCNTs) Tarnok, A.: Microfluidic impedance-based flow cytometry. Cytom. A 77A, 648–666 (2010) 5. Han, K.H., Han, A., Frazer, A.B.: Microsystems for isolation and electrophysiological analysis of breast cancer cells from blood. Biosens. Bioelectron. 21, 1907–1914 (2006) 6. Jang, L.S., Wang, M.H.: Microfluidic device for cell capture and impedance measurement. Biomed. Microdevices 9, 737–743 (2007) 7. Holmes, D., Pettigrew, D., Reccius, C.H., Gwyer, J.D., Berkel, C.V., Holloway, J., Daviesb, D.E., Morgan, H.: Leukocyte analysis and differentiation using high speed microfluidic single cell impedance cytometry. Lab Chip 9, 2881–2889 (2009)
Single-Walled Carbon Nanotubes (SWCNTs) ▶ Chemical Vapor Deposition (CVD)
siRNA Delivery ▶ RNAi in Biomedicine and Drug Delivery
Future Work A single-cell impedance spectroscopy measurement system ideally should have a high testing throughput and the capability of sorting single cells based on impedance measurement results. The potential combination of single-cell impedance spectroscopy with other detection/measurement methods, such as fluorescent detection and the use of biochemical markers, may prove powerful for a range of applications, such as disease diagnostics and rare cell isolation.
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
References
“Size-dependent plasticity” here refers to the strength of metallic samples being a strong function of their size when their dimensions are reduced to the micron and below scales. The notion of reduced sample size applies to all three dimensions, i.e., stand-alone, or one-dimensional (1D) nano and microstructures rather than thin films (2D), where only their thicknesses are reduced to nano- and micro-scales, or to the smallscale deformation volumes within a bulk matrix (3D) as would be the case during nanoindentation experiments, for example.
1. Morgan, H., Sun, T., Holmes, D., Gawad, S., Green, N.G.: Single cell dielectric spectroscopy. J. Phys. D: Appl. Phys. 40, 61–70 (2007) 2. Valero, A., Braschler, T., Renaud, P.: A unified approach to dielectric single cell analysis: impedance and dielectrophoretic force spectroscopy. Lab. Chip. 10, 2216–2225 (2010) 3. Sun, T., Morgan, H.: Single-cell microfluidic impedance cytometry: a review. Microfluid. Nanofluid. 8, 423–443 (2010) 4. Cheung, K.C., Di Berardino, M., Schade-Kampmann, G., Hebeisen, M., Pierzchalski, A., Bocsi, J., Mittag, A.,
Julia R. Greer Division of Engineering and Applied Sciences, California Institute of Technology, Pasadena, CA, USA
Definition
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
Introduction and Overview of Stress Vs. Strain for Bulk Metals Pure bulk single crystalline metals like Cu, Ni, Mo, Ti, etc. generally exhibit a convex, continuous stress– strain relationship upon uniaxial deformation – tension or compression – as can be found in any classic “Mechanical Behavior of Materials” book and as is also schematically shown in Fig. 1. For the lowsymmetry orientations, i.e., where only a single slip system is activated, the deformation immediately following yield is called the “easy glide” region, which is characterized by a low hardening rate. Such low hardening rate stems from the dislocations traveling large distances unimpeded in their glide planes as only one set of crystallographic slip planes is active, and the dislocations are not forced to overcome closely spaced obstacles such as impurities or other dislocations in the course of their motion. A simple way to think of it is the following: Each dislocation travels a distance L before encountering an obstacle, which pins it. The shear strain associated with this motion is: d g ¼ bLd r, where b is the Burgers vector, L is the dislocation mean free path, and r is the mobile dislocation density [1]. The distance dispffiffiffi traveled by peach ffiffiffi location then scales with 1 r, i.e., L ¼b r, where b may be on the order of 1,000 since the dislocations traveling in parallel crystallographic planes have very limited interactions. The hardening corresponding to this “easy glide” region is a result of dislocation storage through the well-known Taylor relation: pffiffiffi t ¼ amb r, where a 0.2, which results in a very low hardening rate dt dglowsymmetry ¼104 m. In highsymmetry orientations, multiple symmetric slip systems are activated, and dislocations come into close encounters with one another as they are traveling toward one another in the symmetric slip planes, resulting in a pronounced dislocation density increase, which in turn, drives the high rate of hardening: . dt dghighsymmetry ¼0:01m. This hardening in bulk single crystals stems from dislocation multiplication processes arising from the well-established dislocation interactions, which are followed by the dislocation networks formation processes. Therefore, in bulk single crystals – regardless of the sample size – the dislocations multiply in the course of deformation, thereby creating dense networks and dislocation substructures, which require the application of higher
2431
S
single crystal (symmetrical orientation)
polycrystal
s single crystal (non - symmetrical orientation) e
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures, Fig. 1 Schematic of typical stress–strain curves for bulk metals: comparing polycrystalline and single crystalline (in high- and low-symmetry orientations) microstructures of the same metal
stresses to move the additional gliding dislocations through these obstacles, thereby carrying out plastic deformation. This holds true for crystals of any size with greater-than-several-microns dimensions. As a result, bulk samples of the same single crystalline material also exhibit identical yield strengths, flow stresses, and hardening rates.
Emergence of Size Effects in Single Crystals at the Nanoscale In a striking deviation from this classical depiction, in the last 5–6 years, it was ubiquitously demonstrated that at the micron- and sub-micron scales, the sample size dramatically affects crystalline strength – even in the absence of any constraining effects, strain gradients, and grain boundaries – as revealed by roomtemperature uniaxial compression experiments on a wide range of single crystalline metallic nano-pillars (for reviews, see [2–4]). In these studies, cylindrical nano-pillars were fabricated mainly by the use of the focused ion beam (FIB), as well as with some FIB-less methodologies, and remarkably, the results of these reports for all metallic single crystals show power law dependence between the flow stress and sample size, regardless of fabrication technique, of the form s ¼ C Dk with the exponent k strongly dependent on the initial crystal structure (i.e., face-centered cubic (fcc), body-centered cubic (bcc), hexagonal closepacked (hcp), etc.), experimental aspects (i.e., sample aspect ratio, lateral stiffness of the instrument,etc.),and dislocation density. A particularly auspicious example is the unique scaling of strength with pillar diameter
S
2432
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures, Fig. 2 Resolved shear strength normalized by the shear modulus as a function of diameter (normalized by the Burgers vector) of experimentally determined compressive strengths of most face-centered cubic metallic nano-pillars published to date (Reprinted with permission from [2])
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures Au Greer et al. (Acta Mater. 2005) Au Greer et al. (PRB 2006) Au Volkert et al. (Phil. Mag. 2006) Au Kim et al. (Acta Mater. 2009) Au Kim et al. Tension (Acta Mater. 2009) Au Kim et al. Tens./Comp. (Acta Mat. 2009) Au Lee et al. (Acta Mater. 2009) Cu Jennings et al. (PRL 2010) Cu Jennings et al. Tens (Phil. Mag. 2010) Cu Jennings et al. Com (Phil. Mag. 2010) Cu Jennings et al. (unpublished) Cu Kiener et al.Tens (Acta Mater. 2008) Ni Frick et al. (Mat. Sci. Eng. A. 2008) Al Ng et al. (Acta Mater. 2008)
0.1
τ/μ
S
0.01
1E-3
1E-4 100
1000
10000 D/b
with the exponent of 0.6 exhibited by nearly all non-pristine (i.e., containing initial dislocations) fcc nano-pillars subjected to uniaxial compression or tension, as illustrated in Fig. 2. The notion of “size effects” applied specifically to the strength of single crystalline metals dates back to the works of Brenner [5, 6], who demonstrated that dislocation-free Cu and Ag whiskers attained very high strengths upon uniaxial tension. Although significant efforts have since been dedicated to studying single crystalline plasticity in confined dimensions, until 5 years ago, most of these research thrusts were focused on thin films, whose yield strengths increased with decreasing film thickness (for example, [7]). The renaissance of small-scale mechanical testing on free-standing vertical pillars is largely due to the original work of Uchic et al. who reported higher compressive strengths attained by focused ion beam (FIB)-machined cylindrical Ni micro-pillars [8]. Greer and Nix then extended this robust and elegant methodology into the nanoscale regime, where oriented Au nano-pillars with diameters below 1 mm were reported to be 50 times higher than bulk [9], and today numerous groups are pursuing this type of uniaxial nano-mechanical testing of materials ranging from single crystalline metals to lithiated battery anodic materials, ceramics, irradiated materials,
shape memory alloys, nano-twinned and nanocrystalline metals, metallic glasses, superalloys, and nanolaminates. The results of many of these studies are overviewed in detail in three existing reviews: [2–4]. Figure 3 shows representative images of beforeand after-testing single crystalline Nb nano-pillars, as well as the representative stress–strain curves with clear discrete characteristics and size effect [10]. Further adapting this methodology, the exploration of size effects in plasticity for a large variety of materials ensued, albeit mainly focusing on fcc structures. More recently, investigating plasticity in small volumes has been further advanced through uniaxial tensile experiments, usually conducted inside of in-situ mechanical deformation instruments custom built by some research groups [9–12].
Key Research Findings Experimental Findings To date, uniaxial compression and tension tests have been performed on Ni and Ni-based superalloys, Au, Cu, and Al (as-fabricated and intentionally passivated) [2–4]. Beyond single crystalline fcc metals, mechanical behavior of other single crystals has been published to date: bcc metals (W, Nb, Ta, Mo, and V), hcp metals
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
a
b
D 242 nm
1200 True stress (MPa)
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures, Fig. 3 SEM images of (a) single crystalline Nb nanopillar as fabricated by the focused ion beam, (c) same pillar after compression showing pronounced crystallographic slip lines. (b) Stress vs. strain curves clearly revealing the size effect (smaller is stronger) and the stochastic signature of smallscale deformation (Reprinted with permission from [10])
S
2433
900 D 409 nm
600
D 868 nm
300 0 0.00
0.03
0.06 0.09 True strain
0.12
c
(Mg, Ti-Al alloy, and Ti), tetragonal metals (In) [2–4]. The reports on nearly all single crystalline metals with non-zero initial dislocation densities (i.e., not whiskers or nano-wires) unanimously demonstrate that their strengths significantly rise with reduced size, with 100 nm-diameter samples sometimes attaining a flow stress 10 higher than bulk. Intriguingly, unlike in bulk, where the dislocation multiplication processes result in Taylor hardening (introduced in Section Introduction and Overview of Stress Vs. Strain for Bulk Metals), the flow stresses in small structures do not appear to scale with the dislocation density. Rather, the global dislocation density appears to decrease upon mechanical loading, while the applied stress required to deform the structure increases (see, for example [11]). Consistent with this “upside down” behavior in the nano- and micron-sized crystals, whereby samples containing fewer mobile dislocations are stronger than their bulk counterparts, it has been shown that introduction of additional dislocations into the structure (for example, via pre-straining) actually weakens these crystals [12, 13]. These findings are antipodal to classical plasticity, where dislocation interactions lead to multiplication, causing higher dislocation densities and requiring higher applied stresses
for deformation to continue, as described in the Section Introduction and Overview of Stress Vs. Strain for Bulk Metals. The initial dislocation density indeed plays a critical role in the onset of the size effect, as has now been shown by several research groups [2–4]. Computational Findings Several models attempting to explain the origins of size-dependent flow stress in the absence of strong strain gradients, as well as the stochastic nature of deformation, have been put forth. There are generally three classes of these models: (1) phenomenological theories, which attempt to describe the physical processes occurring in a nano-single-crystal upon deformation, (2) discrete dislocation dynamics (2- and 3-dimensional) simulations (DD), which rely on the entered initial dislocation density and distribution, as well as on the dislocation mobility laws, and (3) molecular dynamics (MD) simulations, which are generally limited to small (50 nm and below) scales and unrealistically high strain rates (108 s1). An example of (1) is the “hardening by dislocation starvation” theory which hypothesizes that the mobile dislocations inside a small nano-pillar have a greater probability of annihilating at a free surface than of interacting with one
S
2434 300
a
250
Case1 Case2
200 1 150
2
100 50
3
0 0
500
1000 1500 2000 Stress (MPa)
c
another, thereby shifting plasticity into nucleationcontrolled regime [14–16]. Other models include source exhaustion hardening, source truncation, and weakest link theory, all of whose general premise involves representing dislocation source operations in a discrete fashion, and then evaluating the effect of sample size on the source lengths, and therefore on the strengths of their operation [17]. These models capture the ubiquitously observed stochastic signature of the experiments results and show either marginal dislocation storage or no storage at all. An example of an evolved microstructure as revealed by dislocation dynamics simulations is shown in Fig. 4 and clearly shows that upon deformation, the dislocation density decreases, rendering multiplication processes unlikely and shifting plasticity into dislocation nucleationcontrolled regime.
Discussion: What Causes the Size effect? Despite the definition of “ultimate tensile strength” according to Wikipedia stating that UTS is an intensive property and “therefore its value does not depend on the size of the test specimen,” there is now a large body
d
2500
Plastic strain (%)
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures, Fig. 4 Variations in (a) dislocation density and (b) plastic strain with increasing stress. A dislocation avalanche occurs during the second stage. (c), (d), and (e) are the microstructures corresponding to points 1, 2, and 3 in (a), respectively (Reprinted with permission from [20])
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
Dislocation density (μm−2)
S
11 10 9 8 7 6 5 4 3 2 1 0
b Case1 Case2
0
500
1000 1500 2000 Stress (MPa)
2500
e
of work that convincingly demonstrates that this does not hold true in the micron and sub-micron size regime. The ubiquitously reported size-dependent strengths in small-scale single crystalline micro- and nano-pillars show the “smaller is stronger” phenomenon as revealed by the existence of a power law dependence between the attained flow stress (sflow) and pillar diameter (D): sflow ¼ C Dn , where n varies with the specific crystal structure, initial dislocation density, and deformation type. Intriguingly, it appears that the power law slope for all fcc metals is unified, on the order of 0.6, as shown in Fig. 2. This is not the case for body-centered cubic metals, whose power law slopes vary within this crystal structure family, possibly due to the unique potential energy landscape of each bcc metal that a gliding screw dislocation has to overcome. The power law slopes of bcc metals appear to correlate with the intrinsic lattice resistance of each metal, a.k.a. the Peierls barrier, and can be separated into groups according to the low, mid- and high Peierls barrier: low-barrier group containing Nb (0.93, 0.48, 1.07) and V (0.79); midbarrier group containing Mo (0.44, 0.38) and Ta (0.43, 0.41); and high-barrier group containing W (0.21, 0.44) [2–4]. Beyond these two cubic
Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures
2435
S
crystalline structures, other types of crystals were also found to exhibit size-dependent strengths, with smaller generally being stronger. For example, hexagonal close-packed (hcp) metals, Mg and Ti, oriented for slip (as opposed to twinning) were also characterized by power law size effects: [11–21] Ti oriented for prismatic slip has the critical resolved shear stress (CRSS) scale inversely with the diameter, resulting in the slope of 1, while the [3–964] orientation of Mg (basal slip) has the slope of 0.64 [2]. There is currently no physics-based theory that captures these size effects as a function of metal, size, and initial microstructure. Several semi-empirical theories have been proposed, and nearly all identify the initial dislocation density as a key factor in defining the natural microstructural length scale, which in turn, drives the size effect. In an attempt to explain the emergence of higher strengths in small-scale crystals, multiple deformation mechanisms have been proposed, with two distinct ones: (1) single-arm source (SAS) theory first developed by Parthasarathy and Rao, et al. [17] applicable to micron-sized pillars, and (2) hardening by dislocation starvation proposed by Greer and Nix [9, 14] for the much smaller, nanosized pillars. In the SAS theory, the creation of dislocations occurs by the operation of truncated Frank-Read spiral sources, a.k.a. single-arm sources (SAS), whose strength, ts, is inversely proportional to
the distance before it replicates itself, as defined by Gilman [18]. This is likely to be the case in the samples with diameters deeply in the submicrometer range since Ag, for example, has the breeding distance on the order of 0.7 mm. The nucleation stress for a surface
0 source may be represented as s ¼ QQ^ kQB^T ln kBE_TNn ^ , eO
their average length, l, through ts ¼ ks m lnðl=bÞ , where,
• Recent experimental and computational results convincingly demonstrate that the strength of nano- and micron-sized single crystalline metals is indeed size dependent and appears to be well represented by a power law of the form s ¼ C Dk with the exponent strongly dependent on the initial microstructure (i.e., dislocation density, lattice resistance, and the presence of impurities). • With the advent of highly capable nano-fabrication and analysis instruments, as well as of the sophisticated computational tools, the society is ever closer to developing a more complete understanding of size-dependent mechanical behavior of small-scale structures. To date, it is generally agreed that nonpristine (i.e., containing dislocations) micron-sized fcc structures attain higher strengths at smaller sizes through dislocation multiplication processes produced by the operation of single-arm dislocation sources, whose lengths scale with the sample size.
l=b
ks is a source-hardening constant [17]. The smaller micron-sized pillars are not capable of accommodating large single-arm sources, and therefore require the application of higher stresses to activate the stronger, “shorter” sources, giving rise to higher strengths in smaller pillars. In the dislocation starvation theory, applicable to the much smaller pillars, with diameters deeply in the sub-micron regime, new dislocations are created via surface nucleation rather than through a multiplicative process of SAS operation as is the case in the micronsized samples. The surface nucleation gets activated only after a significant fraction of the pre-existing gliding dislocations within the pillar has exited through the free pillar surface, i.e., “starving” the crystal of the necessary mobile dislocations to carry plastic strain. A necessary condition for dislocation starvation is that the largest distance that any gliding dislocation has to travel is smaller than the so-called breeding length, or
where the first term corresponds to the athermal contribution and the last term represents thermally activated nature of such events [19]. The activation of surface sources has a significant thermal activation component, as is evident from the T ln T and strain rate (_e) dependence of the nucleation stress. Recent experimental studies performed on very small (below 500 nm) Cu nano-pillars are consistent with this surface nucleation model and lay out the single-arm source vs. surface source operation regimes as a function of both the pillar size and strain rate [2]. While these theories may appear to be competing, it is likely that they both take place at different pillar sizes: with SAS strengthening occurring in the micron-sized pillars (perhaps, down to 500 nm) and with dislocation starvation followed by surface nucleation prevailing at the smaller sizes, deep in the sub-micron regime.
Summary
S
S
2436
The operation of these truncated sources, which become exhausted in the course of deformation, drives the corresponding strength increase in these small-scale structures. On the contrary, when the sample dimensions are further reduced to the nanosized regime, plasticity likely occurs via dislocation nucleation from surface sources (rather than through a multiplicative process), and the higher strengths arise due to the lower probability of finding weaker dislocation sources in smaller structures. Acknowledgments JRG gratefully acknowledges the financial support of the National Science Foundation (NSF) CAREER grant (DMR-0748267) and the Office of Naval Research (ONR) Grant No. N00014-09-1-0883. The author is particularly grateful to W.D. Nix, A.T. Jennings, D. Jang, J.-Y. Kim, Q. Sun, A. Ngan, C. Weinberger, J. Li, and D. Gianola for useful discussions.
References 1. Nix, W.D.: MSE 208: Mechanical behavior of materials class notes. (1980) 2. Greer, J.R., de Hosson, J.Th. M.: Critical review: plasticity in small-sized metallic systems: intrinsic versus extrinsic size effect. Progress in Materials Science 56, 654–724 (2011) 3. Kraft, O., Gruber, P.A., Monig, R., Weygand, D.: Plasticity in confined dimensions. Annu Rev Mater Res 40, 293 (2010) 4. Uchic, M.D., Shade, P.A., Dimiduk, D.M.: Plasticity of micrometer-scale single crystals in compression. Annu Rev Mater Res 39, 361–386 (2009) 5. Brenner, S.S.: Tensile strength of whiskers. J. Appl. Phys. 27, 1484–1490 (1956) 6. Brenner, S.S.: Growth and properties of “whiskers”. Science 128, 568 (1958) 7. Thompson, C.V.: The yield stress of polycrystalline thin films. J. Mater. Res. 8, 237 (1993) 8. Uchic, M.D., Dimiduk, D.M., Florando, J.N., Nix, W.D.: Sample dimensions influence strength and crystal plasticity. Science 305, 986 (2004) 9. Greer, J.R., Oliver, W.C., Nix, W.D.: Size dependence of mechanical properties of gold at the micron scale in the absence of strain gradients. Acta Materialia 53, 1821 (2005) 10. Kim, J.-Y., Jang, D., Greer, J.R.: Insights into deformation behavior and microstructure evolution in Nb single crystalline nano-pillars under uniaxial tension and compression. Scripta Materialia 61, 300 (2009) 11. Tang, H., Schwartz, K.W., Espinosa, H.D.: Dislocation escape-related size effects in single-crystal micropillars under uniaxial compression. Acta Materialia 55, 1607 (2007)
Skin Delivery 12. Bei, H., Shim, S., Pharr, G.M., George, E.P.: Effects of pre-strain on the compressive stress–strain response of Mo-alloy single-crystal micropillars. Acta Materialia 56, 4762 (2008) 13. Lee, S., Han, S., Nix, W.D.: Uniaxial compression of fcc Au nanopillars on an MgO substrate: the effects of prestraining and annealing. Acta Materialia 57, 4404 (2009) 14. Greer, J.R., Nix, W.D.: Nanoscale gold pillars strengthened through dislocation starvation. Phys. Rev. B 73, 245410 (2006) 15. Shan, Z.W., Mishra, R., Syed, S.A., Warren, O.L., Minor, A.M.: Mechanical annealing and source-limited deformation in submicron-diameter Ni crystals. Nat Mater, 7, 115– 119 (2008) 16. Weinberger, C., Cai, W.: Surface controlled dislocation multiplication in metal micro-pillars. Proc. Natl. Acad. Sci. USA 105, 14304 (2008) 17. Rao, S.I., Dimiduk, D.M., Parthasarathy, T.A., Uchic, M.D., Tang, M., Woodward, C.: Athermal mechanisms of sizedependent crystal flow gleaned from three-dimensional discrete dislocation simulations. Acta Materialia 56, 3245 (2008) 18. Gilman, J.J.: Micromechanics of flow in solids. McGraw-Hill, New York (1969) 19. Zhu, T., Li, J., Samanta, A., Leach, A., Gall, K.: Temperature and strain rate dependence of surface dislocation nucleation. Phys. Rev. Lett. 100, 025502 (2008) 20. Liu, Z.L., Liu, X.M., Zhuang, Z., You, X.C.: Atypical threestage-hardening mechanical behavior of Cu single-crystal micropillars. Scripta Materialia 60, 594 (2009)
Skin Delivery ▶ Dermal and Transdermal Delivery
Skin Penetration ▶ Dermal and Transdermal Delivery
Small angle X-Ray Scattering in Grazing Incidence Geometry ▶ Selected Synchrotron Radiation Techniques
Small Unilamellar Vesicle (SUV) ▶ Liposomes
Small-Angle Scattering of Nanostructures and Nanomaterials
Small-Angle Neutron Scattering ▶ Small-Angle Scattering of Nanostructures and Nanomaterials
Small-Angle Scattering ▶ Small-Angle Scattering of Nanostructures and Nanomaterials
Small-Angle Scattering of Nanostructures and Nanomaterials M. Laver Laboratory for Neutron Scattering, Paul Scherrer Institut, Villigen, Switzerland Materials Research Division, Risø DTU, Technical University of Denmark, Roskilde, Denmark Nano–Science Center, Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark Department of Materials Science and Engineering, University of Maryland, Maryland, USA
Synonyms SANS; SAXS; Small-angle neutron scattering; Small-angle scattering; Small-angle X-ray scattering
2437
S
the technique has made an astounding impact in many fields of research including polymer systems [1, 2], complex fluids [3], biology [4], condensed matter physics, and materials science [1, 5]. The methodology for performing small-angle scattering studies extends back to the 1930s [6], following the first small-angle X-ray scattering (SAXS) studies. Subsequently, synchrotron sources have enabled explorations with ever smaller sample amounts, in complex sample environments. Laboratory SAXS instruments are now routinely used for sample characterization in many research institutions. The creation of large-scale facilities producing neutrons for research has proved to be instrumental for smallangle neutron scattering (SANS) studies. As X-rays and neutrons interact differently with matter, SAXS and SANS are quintessentially complementary, and a particular technique may frequently become an indispensable probe for a particular application. This entry is organized as follows: in section Key Principles, the general concepts of small-angle scattering are explained. In section X-Rays or Neutrons, the advantages and practical aspects of X-ray and neutron techniques are compared. In section Magnetic Neutron Scattering, the discussion is focused on small-angle neutron scattering from magnetic structures, and neutron polarization analysis (Section Neutron Polarization Analysis) is provided as an example of a developing field for small-angle studies. The concluding section (Section Summary and Outlook) incorporates an outlook toward novel directions for the small-angle scattering technique. An overview of symbols used repeatedly throughout this entry is provided in Table 1.
Definition Small-angle scattering of nanostructures and nanomaterials encompasses the measurements of scattering from structures with length scales ranging between the near-atomic (nanometer) to the near-optical (micrometer), using beams of nanometer wavelengths or less.
Overview Small-angle scattering reveals structural features on length scales between the near-atomic (nanometer) to the near-optical (micrometer). With such versatility,
S
Key Principles Bragg Diffraction Small-angle scattering patterns are frequently continuous in nature, rather than consisting of crystalline diffraction peaks. Nevertheless, Bragg’s law encapsulates a fundamental relationship l ¼ 2d sin y ¼
4p sin y q
(1)
showing that the angle of diffraction y varies inversely with the separation d of the diffracting lattice
S
2438
Small-Angle Scattering of Nanostructures and Nanomaterials
Small-Angle Scattering of Nanostructures and Nanomaterials, Table 1 List of symbols used throughout this entry y k~i, ~ kf
Bragg angle of diffraction Wavevector of incoming, final (scattered) beams
l ~ q, q z d D ~ r, r
Wavelength of beam Scattering vector, its magnitude Angle of q in detector plane Spacing between Bragg planes Characteristic length scale within sample (Real space) position vector in sample or within nano-object, its magnitude Scattering length, that of X-rays, that of neutrons from nuclei, that of neutrons from atomic moments Scattering length density, that of solvent or matrix Form factor of single nano-object Scattered intensity Magnetic, nuclear parts of (neutron) form factor of single nano-object Form factor P ¼ |F|2 Structure factor Component of magnetic moment perpendicular to k~i Angle of magnetic moment in detector plane Component of magnetic moment parallel to ~ ki Halpern–Johnson vector
b, bx, bn, bm r, r0 F I M, N P S C f Z ~ A
planes. It can be seen that the scattering from objects on nano- to micrometer scales falls predominantly in ˚ the small-angle regime rc, and j1 ðxÞ ¼ sinx2 x cosx x is a spherical Bessel function. For brevity, rc and rs are measured relative to the scattering length density r0 of the surrounding solution or matrix; equivalently, r0 may be considered to be zero. Equation (5) may be simplified by introducing volume terms such as Vc ¼ 43 prc3 . Figure 4b exemplary P(q) ¼ |F(q)|2 curves for this core-shell model are plotted.
from a finite instrument resolution; nonetheless it is clear that sizeable benefits are gained if samples for small-angle scattering studies are as monodisperse as possible. Figure 4 furthermore illustrates how contrast variation may be exploited, where feasible: by matching the solvent r0 to the core rc (dotted green line in Fig. 4), the interference bumps in the profile are more discernible, notably in polydisperse situations. Regimes in the Scattering Profile
Polydispersity
In reality, most scattering systems show a distribution in their sizes and shapes; this is known as polydispersity. Form factors can be straightforwardly modified to average over a distribution in size or shape, and several functions defining distributions for the polydispersity are available [19]. The effects of polydispersity on the small-angle scattering profile are now examined briefly. In Fig. 4c the core-shell model is modified to include a distribution in core radii with standard deviation 20% of the mean core radius. By comparing this figure with Fig. 4b, the adverse effects of polydispersity – a smearing in q of the scattering profile – can readily be seen. A smearing of the scattering profile also results
The appearance of distinctive q regimes in the scattering profile may also be seen in Fig. 4. The solid sphere model (continuous blue lines in Fig. 4) features a characteristic length scale D, namely, the sphere ˚ . The scattering features a high-q regime radius 80 A for qD 1, where the scattering profile appears to decay algebraically with q, and a regime at low q where the profile falls off more slowly in q. These regimes emerge independently of the absolute intensity and of any model, and invariably provide useful clues as to the size and homogeneity of the nanoobject. In the high-q regime qD 1, known as the Porod regime, the exponent a of the power-law decay I(q) qa reveals the “fractal dimension” of the
Small-Angle Scattering of Nanostructures and Nanomaterials
a
2443
S
1000
b
100
Form factor, P(q)
Core-shell
Core only
10 1 0.1
Monodisperse model Core-shell
0.01
Core only 0.001
Solvent matched to core
0.0001
Solvent matched to core
2
3 4 56
0.001
2
3 4 56
0.01
2
0.1
Scattering vector, q (Å−1)
c
d
1000
Form factor, P(q)
10
q −4
1 0.1
Polydisperse model Core-shell
0.01
Core only Solvent matched to core
0.001 0.0001
3 4 56
2
3 4 56
0.01
2
0.1
Scattering vector, q (Å−1)
Slope
6
-Rg2/3
5 4 3 Monodisperse model 2 1
2
0.001
Logarithm of form factor, ln P(q)
7 100
Core only
0
1
2
3×10−3
Scattering vector squared, q 2 (Å−2)
Small-Angle Scattering of Nanostructures and Nanomaterials, Fig. 4 The spherical core-shell form factor P(q). (a) Schematic showing the various parameters for the model curves plotted in (b)–(d): in (b) the particles are monodisperse; in (c) the particles have polydisperse core sizes with the standard deviation of the polydispersity distribution set at 20% of the mean core radius rc. The dashed red and dotted
˚ and a shell 10 A ˚ thick; green lines indicate P(q) with rc ¼ 70 A the dashed red line shows P(q) when the core scattering length density rc is twice that of the shell rs; the dotted green line shows P(q) when the solvent r0 matches rc. The continuous blue ˚ ; in (d), this line shows P(q) for a core only model with rc ¼ 80 A line is redrawn on ln P(q) versus q2 axes. Black lines are guides to the eye
scattering objects [14]. The a ¼ 4 observed in Fig. 4c, for example, indicates smooth surfaces, as expected for the model of uniform spheres with boundaries sharp in r(r). In the Guinier regime as q ! 0, the scattering follows a Gaussian dependence expð 13 q2 R2g Þ where Rg is the “radius of gyration” of the nano-object [6], and is analogous to the radius of gyration in
mechanics. For uniform solid spheres of radius rc, R2g ¼ 35 rc2 . To reveal this behavior at low q, a “Guinier plot” is useful, where the logarithm of the scattering is plotted versus q2; the plot will be linear with slope 13 R2g at sufficiently low q. A Guinier plot for the sphere model is shown in Fig. 4d, together with the expected low-q slope [black line in Fig. 4d]. Comparing the two, it may be seen that the
S
S
2444
Guinier approximation holds reasonably well when q2 ˚ 2, that is, for qRg ≲ 1.3. The upper limit ≲ 0.0008 A for the Guinier regime is found to depend somewhat on the scattering model [13]. Almost every small-angle scattering article incorporates an analysis of the low- or high-q regimes. Extensive examples of Guinier plots and associated methods (e.g., the Zimm and Kratky plots) that quantify scattering regimes without resorting to detailed model fitting may also be found in the literature [3]. There have also been efforts to encapsulate both low-q and high-q regimes in unified models. A common example is the Beaucage model, applicable to situations where a nanostructure exhibits a hierarchy of scales, each discernible scale contributing a pair of Guinier and Porod regions to the scattering profile [15].
X-rays or Neutrons? Scattering Length The basic principles of small-angle scattering, as overviewed in previous sections, have the same form for both X-rays and neutrons. However, there are fundamental distinctions between the techniques arising from how the X-rays or neutrons interact with matter. One such distinction is patent in the atomic scattering length b [cf. (2)]. For neutrons, bn depends on the structure of the nucleus and is isotope dependent. For X-rays, bx depends on the electronic structure and scales with the total number of electrons Z in the atom. At the length scales probed by small-angle scattering, nuclei look like point sources of scattering and bn is q-independent, whereas for X-rays (atomic absorption edges neglected) bx(q) falls off slowly with q – this is essentially the Guinier regime of the form factor for the atomic electron clouds, with ˚ . Low-resolution modeling bx(0) ¼ Z 2.8 105 A of SAXS profiles can be achieved with bx considered constant. The scattering lengths for selected atoms are pictured in Fig. 5 for both neutrons [20] and X-rays. Generation of Contrast The scattering length density r of an elementary volume V (e.g., a molecular volume) within the sample is calculated by summing the scattering lengths of atoms within V, that is, r ¼ Sj bj/V. Here, a uniform electron density needs to be assumed for X-rays. The scattering length densities for neutrons of selected
Small-Angle Scattering of Nanostructures and Nanomaterials Neutrons
H
D
C
N
O
Al
Fe
X-rays (2x)
Small-Angle Scattering of Nanostructures and Nanomaterials, Fig. 5 Schematic indicating the propensity for scattering of X-rays or neutrons [20] by selected atoms. The radii of the circles are drawn proportional to the scattering length. For X-rays, the scale has been 2 magnified for clarity. The scattering length of hydrogen for neutrons is negative, in contrast to the other atoms depicted
Small-Angle Scattering of Nanostructures and Nanomaterials, Table 2 The scattering properties, for neutrons, of selected substances Substance H2O D2O C B 10 B 11 B Al Fe Cd
Neutron attenuation length (cm) 0.18 1.55 1.59 0.0036 0.00067 1.35 7.74 0.629 0.0031
Neutron scattering length ˚ 2) density ( 106 A 0.56 6.33 7.53 6.91 0.14 8.51 2.08 8.02 2.27
For the calculation of these values, room temperature densities are used, and the elements with no isotope specified are deemed to comprise of their isotopes in naturally abundant ratios
substances is listed in Table 2. Comparing with Fig. 5, it is clear that there is an enormous distinction between hydrogen and deuterium. The former has bn ¼ 3.74 ˚ , the latter 6.67 105 A ˚ [20], and this is 105 A reflected in r as shown for H2O and for D2O in Table 2. Great contrast may be achieved by using hydrogenated solutes in deuterated solvents. Polystyrene in toluene, ˚ 2 for example, presents a contrast of 0.48 106 A when both are hydrogenated, whereas the contrast of hydrogenated polystyrene in deuterated toluene ˚ 2. In contrast-variation studies, tuning is 4.2 106 A the solvent scattering length density r0 may be accomplished by using a binary mixture whose components have widely different scattering length densities; these
Small-Angle Scattering of Nanostructures and Nanomaterials
are mixed in a linear ratio to yield the desired r0. For X-rays, the two components would have different electron densities; for neutrons, they may be deuterated and hydrogenated versions of the same solvent. Isotope substitution may also be used to create SANS contrast inside a particle by specific isotope labeling of a particular part of interest, which can be used, for example, to determine how this part fits in with the rest of the particle. In SAXS, a novel way to induce contrast is to measure scattering profiles at different photon energies in the vicinity of an atomic absorption edge of one of the elements within the sample: for this element, the form factor bx varies across the absorption edge, while the rest of the scattering potential landscape is essentially invariant in the small range of photon energies measured. This contrast-variation technique is known as anomalous SAXS (ASAXS) and may be exploited for the large number of elements having an absorption edge in the range 5–25 keV [1, 9, 21]. Neutron Absorption
Another aspect of neutron scattering that is heavily dependent on the isotope is absorption. A few elements, for example boron and cadmium (c.f. Table 2), are outstanding neutron absorbers in their naturally abundant forms. Not surprisingly, boron-containing substances and cadmium are common shielding materials. For scattering explorations of these and other such materials, samples need to be prepared using elements devoid of the absorbing isotopes; boron-11, for example, is seen from Table 2 to have an acceptable attenuation length. Separating elements into their different isotopes, however, becomes increasingly difficult and expensive for heavier atoms. It is also apparent from Table 2 that some materials, such as aluminum, are rather poor neutron absorbers. This is providential for the construction of bulky or demanding sample environments such as cryostats or pressure cells. General Considerations The majority of elements have neutron attenuation lengths on the order of a centimeter, reflecting the weak interaction of neutrons with matter. In general, neutron scattering explores the bulk of samples and, for similar reasons, effectively no radiation damage to the sample occurs with the neutron fluxes currently available (up to 108 neutrons cm2 s1) on SANS
2445
S
instruments. For SAXS experiments at synchrotron sources, the high brilliance of the X-ray beam (1016 photons cm2 s1) can damage soft matter and biological samples within milliseconds [9]. This may affect the validity of the scattering profile measurement. For SAXS the sample thickness may also need to be chosen to obtain the optimum scattered intensity, since at the photon energies (10 keV) used in SAXS, the X-ray attenuation length is on the order of millimeters for the lighter elements [9], which make up most soft matter systems. The many orders of magnitude more intense fluxes available on SAXS beamlines are, of course, invaluable for high-resolution, low uncertainty measurements of scattering profiles. The associated fast data acquisition times also facilitate the characterization of time-dependent effects. Likewise, relatively small sample quantities are typically required ( LCST
External Signal
On
Off
Smart Hydrogels, Fig. 2 Schematic illustration of the “on-off release” from a squeezing hydrogel device for drug delivery
make them reliable candidates for drug delivery in environmental conditions in which the temperature varies. Under the critical point of negative temperaturesensitive hydrogels, hydrogen bonding between hydrophilic segments of the polymer chain and water molecules dominates, leading to enhanced dissolution in water. As the temperature increases, however, interactions among hydrophobic segments are strengthened, while hydrogen bonding becomes weaker. The net result is shrinking of the hydrogels due to inter-polymer chain association, through hydrophobic interactions [9]. The LCST can be changed by adjusting the ratio of hydrophilic and hydrophobic segments in the hydrogel system. It can also be done by developing copolymers of hydrophobic (e.g., N-isopropylacrylamide [NIPAAm]) and hydrophilic (e.g., acrylic acid [AA]) monomers. Hydrogels with negative temperature sensitivity have the potential to show an “on-off” drug release with “on” at low temperature and “off” at high temperature allowing pulsatile drug release. Figure 2 shows a schematic illustration of the on-off release. Moreover, polymers having LCST below human body temperature have the potential for injectable applications and they are often combined with polysaccharides, such as chitosan, alginate, cellulose, and dextran [11], that improve the material’s properties in terms of biocompatibility, swelling ratio, pH-sensitivity, swelling dynamics (swelling and de-swelling rate), modulation of LCST, and thermal stability. Another issue of the thermoresponsive polymers is represented by temperature-sensitive peptides often associated with the widely investigated elastin-like polypeptide (ELP) polymers. This polymer is composed of
S
S
2454
Smart Hydrogels, Fig. 3 pH-dependent ionization of hydrogels. (a) Poly (acrylic acid) and (b) poly (N, N/diethylamionethyl methacrylate) (adapted from [13])
Smart Hydrogels H
a
H
C
OH−
C
H
H
C
C
H
COO−
H+ COOH
H H
b
H OH−
C
C
H
H
C
C
H
C=O
H+ C=O
H
O
O +
CH2CH2N(CH2CH3)2 Low pH
pentapeptide monomers, derived from human tropoelastin [12], which tend to aggregate in a reversible way under heating, entrapping the payload. This biopolymer showed extraordinary biological properties: it is fully biocompatible, completely biodegradable without any immunogenic reaction, and moreover, it appeared to be ideal for the delivery of protein-based treatments. The degradation of the polymer is constant and slow when compared to the free form of the monomer. The temperature-sensitive peptides represented a breakthrough in polymer science in the biomedical field. The exploitation of standard recombinant DNA techniques via genetic engineering allowed them to be easily synthesized and modified in the primary sequence thereby changing chemical, physical, and biological properties of the polymer while permitting a controlled design of the peptide according to the nature of the cargo. Furthermore, they could easily be attached to a protein-based payload during the synthesis process. The two main features of ELP polymers are directionality and reversibility. Directionality describes the conformational state of the polymer under a specific stimulus. In particular, the ability of the peptide to associate or disassociate under heating is essential if the treatment requires fast or slow drug release. Reversibility regards the capacity of the polymer to return to its initial state after the thermal stimulus [9, 12]. A reversible polymer could be very useful for concentrating the payload in a region of the body and obtaining a targeted release of the drug after heating through an external stimulus.
H
CH2CH2N(CH2CH3) High pH
Conversely, an irreversible polymer could allow drug delivery in a controlled fashion through slow degradation. This category of thermosensitive polymers is emerging as the new opportunity in cancer cell therapy [9] and future research aims to increase and optimize the biocompatibility and the tenability of these delivery systems. pH-Sensitive Hydrogels Another class of smart hydrogels undergoes a volume transition with the variation of the environmental pH. These hydrogels are characterized by the presence of ionic moieties on the polymeric backbone. These ions either accept or release protons in response to pH changes. Weak acidic groups such as acrylic and cationic acids or weak basic groups such as amines provide pH sensitivity to the polymer chain. Swelling of hydrogels sharply changes in the vicinity of the pKA or pKB values of the acidic or basic functional groups. Carboxylic pendant groups accept hydrogen at low pH but exchange it for other cations above the pKA value, and become ionized at higher pH. The hydrodynamic volume and swelling capability of these polymer chains increases sharply when their carboxylic groups become ionized and reach plateau around pH 7. Amines, on the other hand, accept protons at low pH and become positively ionized at and below their pKB values. Hence, their swelling capability increases sharply in acidic solutions. Figure 3 shows structures of anionic and cationic polyelectrolytes and their pH-dependent ionization. Poly(acrylic acid) (PAA)
Smart Hydrogels
becomes ionized at high pH, while poly(N,N/diethylaminoethyl methacrylate) (PDEAEM) becomes ionized at low pH [13]. The swelling and pH-responsiveness of polyelectrolyte hydrogels can be adjusted by using neutral comonomers, such as 2-hydroxyethyl methacrylate, methyl methacrylate, and maleic anhydride [9]. Different comonomers provide different hydrophobicity to the polymer chain, leading to different pH-sensitive behavior. pH-responsive polymers were first applied as an oral drug delivery systems due to the variation of the pH that characterizes the different areas of the gastrointestinal tract [13]. Several anionic polymers such as methyl acrylic acid, methyl methacrylate, and hydroxypropyl methylcellulose phthalate have been commercially used as enteric coatings for oral delivery of protein-based drugs. In the acidic environment of the stomach, the carboxylic acid groups of the above-mentioned polymers are unionized and the particle retains its therapeutic cargo. They were shown to release the encapsulated drug molecules into the small intestine in response to the alkaline pH [14] because the carboxylic acid group becomes ionized and the particle swells. PNIPAAm can be classified as a thermal and pH-responsive polymer. It is used for oral drug delivery applications exploiting its degradation characteristics in the alkaline environment of the intestine. It was shown that the polymer synthesized in combination with butyl methacrylate (BMA) and acrylic acid (AAc) is very stable at low pH while resulting in full degradation and release of payload within the high pH environment [15]. Vesicles derived from endocytosis fuse their membranes with vesicles of the early endosomal compartment. Lysosome formation follows after several other membrane fusion steps (with late endosomal and lysosomal vesicles). During these membrane fusions inside the cytoplasm, the vesicles enrich themselves of protons, decreasing their internal pH through an active import of H+. This proton enrichment is permitted by an ATPase that works using ATP to pump in hydrogen ions and a negative counter ion (Cl) to satisfy electroneutrality [16]. Due to the presence of many ionizable groups in the backbone, a pH-responsive polymer is usually able to work as a “proton sponge” with a buffering effect on the surrounding environment. Therefore, the ATPase continues to work with the result of increasing the concentration of the counter ion and affecting the osmotic balance of the vesicles.
2455
S
The final result of this process is the breaking of the vesicle’s membrane caused by an increase in osmotic pressure. Moreover, many polymers increase their volume when subjected to an acid environment; this produces a mechanical stress on the lysosomal membranes that affects the integrity of these organelles. Lysosomal escape is very useful for nonviral gene delivery and could potentially be more efficient and less toxic than standard virus-based carriers. These molecules are usually cationic, thus likely efficient in forming macro-complexes with DNA (polyplexes) or new therapies based on RNA (e.g., siRNA, miRNA). Figure 4 represents a schematic of the lysosomal escape performed by pH-responsive cationic hydrogel. The polymer in this case plays several roles: (1) to physically carry the plasmid, (2) to shield the DNA payload from degradation, (3) to perform lysosomal escape and facilitate the nuclear delivery of the cargo. Poly(ethylene imine) (PEI) [17], poly(amidoamine) (PAMAM) [18], Poly (N,N-dimethylaminoethyl methacrylate) (PDEAEM) [19], poly (L-Lysine) [20] and modified chitosan represent a few examples of all the polymers studied for this application. Thermo- and pH-responsive hydrogels allow (with or without the use of copolymers) the loading and the triggered release of many drugs for the treatment of different pathological states. Table 1 reports some examples of these polymer-drug formulations and their potential clinical applications.
Other Responsive Compounds Electro-Responsive Hydrogels Recently, electro-responsive hydrogels stimulated the curiosity of the scientific community because they can un-swell or bend, depending on the shape and orientation of the gel along an electric field. The gel bends when it is parallel to the electrodes, whereas unswelling occurs when the hydrogel lies perpendicular to them. This kind of polymer is usually based on a polyelectrolyte matrix (polymers that contain a relatively high concentration of ionizable groups along the backbone chain), so these polymers are also pH-responsive [21]. The electric field generates an interaction between the mobile counter ions and the immobile charged groups of the gel’s polymeric network in order to induce the phase transition. For example, in partially hydrolyzed polyacrylamide gels, the
S
S
2456
Smart Hydrogels, Fig. 4 Schematic of the lysosomal escape. Upon cell internalization by endocytosis, hydrogel particles are trapped in the endolysosomal compartment. During their transition from early to late endosomes and finally to lysosomes, particles undergo increased swelling due to protonation. The proton sponge effect they exert eventually leads to the breakage of the lysosomal vescicle
Smart Hydrogels
Cell membrane Hydrogel Particles
Endocytic vesicle (pH: 7.1-7.4) 2) Vesicle trafficking: the endosomal compartment
1) Particle uptake
5) Lysosomal Escape: Lysosme swells as a consequence of the proton sponge effect of the pH responsive hydrogel
3) Acidification
4) Vesicle trafficking: fusion with lysosomal compartment
Lysosomes (pH: 4.0-4.8)
Early endosome (pH: 6.2-6.3)
Late endosome (pH: 5.0-5.5)
Smart Hydrogels, Table 1 Some applications of temperature and pH-responsive polymer Polymer PNIPAAm Glycidyl methacrylate b-glycerophosphate Poly(propylene)-g-AA Triblock polymer PNIPAAm PNIPAAm Poly(ethylene imine) poly(amidoamine) Poly (N,N-dimethylaminoethyl methacrylate) poly (L-Lysine)
Environmental stimulus Temperature Temperature Temperature Temperature Temperature Temperature pH pH pH pH
Copolymer Chitosan Chitosan Chitosan NIPAAm/Chitosan Poly(lactic-co-glycolic acid)/poly (lactic acid)/poly(ethylene glycol) Alginate Acrylic acid
pH
mobile H+ ions migrate toward the cathode while the negatively charged immobile acrylate groups in the polymer networks are attracted toward the anode (Fig. 5). Thus, the anionic polymeric gel network is pulled toward the anode. This pull creates a uniaxial stress along the gel axis, being maximal at the anode and minimal at the cathode. This stress gradient contributes to the gel deformation and when triggered by electric filed, these hydrogels undergo the swelling and contraction necessary to release the payload. Electro-responsive polymers represent one of the
Loaded drug Diclofenac Doxorubicin Chitosan Chitosan Adriamycin
Application Anti-inflammatory Chemotherapeutic Antibiotic Antibiotic Chemotherapeutic
Indomethacin Growth factors Nucleic acid Nucleic acid Nucleic acid
Anti-inflammatory Wound repair Gene delivery Gene delivery Gene delivery
Nucleic acid
Gene delivery
most complex challenges in biomedical polymer synthesis. Even if the electric stimulus can be finely tuned in terms of strength, duration, and impulse frequency, the use of these techniques is limited by objective safety restrictions. This kind of hydrogel was conceived to obtain a pulsatile delivery of hormones (e.g., insulin, hydrocortisone), proteins, and peptides, and glucose and a few clinical applications for electro-responsive polymers are currently being investigated for dermal and transdermal drug delivery [21, 22].
Smart Hydrogels Smart Hydrogels, Fig. 5 The effect of an electric field on a polyelectrolyte gel. Positively charged counterions migrate toward the cathode, while the immobile polymeric anionic groups are attracted toward the anodes (Adapted from Ref. [21])
2457
S
E
Cathode
- Counter Ions
Stimulus-Sensitive Connection This approach allows for the development of drug delivery tools responsive to the composition of the biological microenvironment. In this case, the polymer’s function is limited to being a carrier for the cargo and targeting the area with the correct environmental conditions for drug release. Polymers like N-(2-hydroxypropyl) methacrylate (HPMA), already exploited for steric stabilization of bioactive proteins, were used to coat biological or synthetic particles to provide them protection against antibodies and complement factors [23]. HPMA was also used to bind doxorubicin via a pH-sensitive linker. This polymer-bound drug showed significant improvement in the ability to kill MCF-7 breast cancer cells when compared with the drug alone [24]. Another example of drug delivery system based on using a peptide linker (Pro-Val-Gly-Leu-Ile-Gly) was composed of a chemotherapeutic agent and a dextran-based polymer. This linker is sensitive to the action of the overexpressed cancer metalloproteinases 2 and 9 while remaining stable in blood circulation. Therefore, the drug release is more concentrated at the site of the tumor where these enzymes are usually overexpressed [25].
Future Perspectives Smart hydrogels, with proper chemical and biological modifications, represent promising alternatives to current controlled release and functional biomaterials
Anode
- Matrix
- Negative Charge
- Stress
applications. Their capability of addressing specific biological and medical challenges will determine the success of the next generation of smart hydrogel materials. Hydrogels that are being used for load-bearing applications in tissue engineering will need to provide enhanced mechanical properties as well as the molecular signals that are present in the native or regenerating tissues. Similarly, the stimuli-guided controlled release of drugs from smart hydrogels is, in most practical cases, too slow due to the thickness of the matrix. Therefore, a fast-acting hydrogel comprised of thinner and smaller matrix is necessary for this purpose. However, this type of hydrogel would be characterized by low mechanical stability and would have limited applications. Synthesis of new polymers, crosslinkers, and reinforcing agents with higher mechanical properties, more biocompatibility, and better biodegradability is essential for success in all aspects of translational medicine. With the development of new materials and novel methods of engineering chemical, mechanical, and biological functionality into the hydrogel matrix, it can be anticipated that future smart hydrogels will provide even greater contributions to the field of drug delivery and regenerative medicine.
Cross-References ▶ Nanomedicine ▶ Nanoparticles
S
S
2458
Soft Actuators
References
21. Sudaxshina, M.: Electro-responsive drug delivery from hydrogels. J. Control. Release 92, 1–17 (2003) 22. Peppas, N.A., Bures, P., Leobandung, W., Ichikawa, H.: Hydrogels in pharmaceutical formulations. Eur. J. Pharm. Biopharm. 50, 27–46 (2000) 23. Urry, D.W., Luan, C.H., Parker, T.M., Gowda, D.C., Prasad, K.U., Reid, M.C., Safavy, A.: Temperature of polypeptide inverse temperature transition depends on mean residue hydrophobicity. J. Am. Chem. Soc. 113, 4346–4348 (1991) 24. Fisher, K.D., Seymour, L.W.: Hpma copolymers for masking and retargeting of therapeutic viruses. Adv. Drug Deliv. Rev. 62, 240–245 (2010) 25. Liu, W., MacKay, J.A., Dreher, M.R., Chen, M., McDaniel, J.R., Simnick, A.J., Callahan, D.J., Zalutsky, M.R., Chilkoti, A.: Injectable intratumoral depot of thermally responsive polypeptide–radionuclide conjugates delays tumor progression in a mouse model. J. Control. Release 144, 2–9 (2010)
1. Navarro, M., Michiardi, A., Castan˜o, O., Planell, J.A.: Biomaterials in orthopaedics. J. R. Soc. Interface 5, 1137–1158 (2008) 2. Ringsdorf, H.: Structure and properties of pharmacologically active polymers. J. Polym. Sci. Pol. Sym. 51, 135–153 (1975) 3. Liu, S., Maheshwari, R., Kiick, K.L.: Polymer-based therapeutics. Macromolecules 42, 3–13 (2009) 4. Keys, K.B., Andreopoulos, F.M., Peppas, N.A.: Poly(ethylene glycol) star polymer hydrogels. Macromolecules 31, 8149–8156 (1998) 5. Dirk, S.: Thermo- and Ph-Responsive polymers in drug delivery. Adv. Drug Deliv. Rev. 58, 1655–1670 (2006) 6. Anderson, J.M.: The future of biomedical materials. J. Mater. Sci. Mater. Med. 17, 1025–1028 (2006) 7. Alijotas-Reig, J., Garcia-Gimenez, V.: Delayed immunemediated adverse effects related to hyaluronic acid and acrylic hydrogel dermal fillers: clinical findings, long-term follow-up and review of the literature. J. Eur. Acad. Dermatol. Venereol. 22, 150–161 (2008) 8. Nikolaos, A.P.: Hydrogels and drug delivery. Curr. Opin. Colloid Interface Sci. 2, 531–537 (1997) 9. Chilkoti, A., Dreher, M.R., Meyer, D.E., Raucher, D.: Targeted drug delivery by thermally responsive polymers. Adv. Drug Deliv. Rev. 54, 613–630 (2002) 10. Prabaharan, M., Mano, J.F.: Stimuli-responsive hydrogels based on polysaccharides incorporated with thermoresponsive polymers as novel biomaterials. Macromol. Biosci. 6, 991–1008 (2006) 11. Webber, R.E., Shull, K.R.: Strain dependence of the viscoelastic properties of alginate hydrogels. Macromolecules 37, 6153–6160 (2004) 12. Urry, D.W.: Physical chemistry of biological free energy transduction as demonstrated by elastic protein-based polymers. J. Phys. Chem. B 101, 11007–11028 (1997) 13. Qiu, Y., and Park, K.: Environment-sensitive hydrogels for drug delivery. Adv. Drug Deliv. Rev. 53, 321–339 (2001) 14. Davis, B.K., Noske, I., Chang, M.C.: Reproductive performance of hamsters with polyacrylamide implants containing ethinyloestradiol. Acta Endocrinol. 70, 385–395 (1972) 15. Serres, A., Baudysˇ, M., Kim, S.W.: Temperature and Ph-sensitive polymers for human calcitonin delivery. Pharm. Res. 13, 196–201 (1996) 16. Varkouhi, A.K., Scholte, M., Storm, G., Haisma, H.J.: Endosomal escape pathways for delivery of biologicals. J. Control. Release 151, 220–228 (2011) 17. Godbey, W.T., Wu, K.K., Mikos, A.G.: Poly (Ethylenimine)-mediated gene delivery affects endothelial cell function and viability. Biomaterials 22, 471–480 (2001) 18. Dufe`s, C., Uchegbu, I.F., Sch€atzlein, A.G.: Dendrimers in gene delivery. Adv. Drug Deliv. Rev. 57, 2177–2202 (2005) 19. Vicent, M.J., Dieudonne´, L., Carbajo, R.J., Pineda-Lucena, A.: Polymer conjugates as therapeutics: future trends, challenges and opportunities. Expert Opin. Drug Deliv. 5, 593–614 (2008) 20. Brown, M.D., Gray, A.I., Tetley, L., Santovena, A., Rene, J., Sch€atzlein, A.G., Uchegbu, I.F.: In vitro and in vivo gene transfer with Poly(Amino Acid) vesicles. J. Control. Release 93, 193–211 (2003)
Soft Actuators ▶ Organic Actuators
Soft Lithography ▶ BioPatterning ▶ Microcontact Printing ▶ Nanoscale Printing
Soft X-Ray Lithography ▶ EUV Lithography
Soft X-Ray Microscopy ▶ Selected Synchrotron Radiation Techniques
Soil/Terrestrial Ecosystem/Terrestrial Compartment ▶ Exposure and Toxicity of Metal and Oxide Nanoparticles to Earthworms
Sol-Gel Method
Solar Cells ▶ Self-repairing Photoelectrochemical Complexes Based on Nanoscale Synthetic and Biological Components
Sol-Gel Method Bakul C. Dave and Sarah B. Lockwood Department of Chemistry and Biochemistry, Southern Illinois University Carbondale, Carbondale, IL, USA
Synonyms Chemical solution deposition; Gel chemical synthesis; Silica gel processing; Wet chemical processing
Definition The sol-gel method is a wet chemical process of making oxide-based materials starting from hydrolyzable precursors via hydrolysis and condensation. The precursors usually contain weaker ligands as compared to water such as halides, nitrates, sulfates, alkoxides, or carboxylates. The hydrolyzed precursors then condense together to form small colloidal nanoparticles suspended in a liquid called a sol. Further polycondensation of the sol particles leads to an extended oxobridged network of polymeric oxide-based materials. The as-formed gels obtained by the sol-gel method are biphasic materials that contain gel network along with a significant amount of liquid phase. Drying of these gels, either under ambient conditions or at elevated temperature, can lead to expulsion of solvent phase to form dense materials. Because the method utilizes molecular precursors, it offers appealing prospects for the design of nanomaterials with a degree of control over their structure, composition, dimension, morphology, organization, geometry, and bulk architecture. While the method can be used to make bulk monoliths, it is particularly suited for fabricating nanomaterials of varied compositions in different shapes, geometries, and dimensionalities including nanoparticles, fibers, thin films, and coatings.
2459
S
Overview The sol-gel method is a solution-based method of making ceramic particles, powders, coatings, films, and monolithic objects [1]. The process is commonly used for making oxides, however, silica-based materials constitute a primary archetype system. With highly electropositive metals, the precursor molecules can react rapidly with water to undergo facile hydrolysis under ambient conditions. However, precursors based on alkoxysilanes usually require the use of catalysts, refluxing, or use of ultrasound for the reaction to proceed at an appreciable rate. The liquid sol containing nanoscale particles can be used to form powders, fibers, coatings, or monolithic forms by allowing the reaction to proceed under specific processing conditions. In addition to the formation of pure oxides, the method can also be used for making organic-inorganic hybrids and organically modified silicas as well as bioceramic hybrids [2]. In the sol-gel method, a suitable molecular precursor is hydrolyzed to generate a solid-state polymeric oxide network (Fig. 1). Initial hydrolysis of the precursor generates a liquid sol, which ultimately turns to a solid, porous gel. The gels formed this way are porous and contain a substantial amount of solvent phase. Slow drying of the gels under ambient conditions (or at elevated temperature) leads to evaporation of the solvent phase. The dried gels – termed xerogels – are substantially less porous than wet gels. Since the method begins from a molecular species, it offers a degree of control over structure and properties that can be tailored at the molecular level. The method can be used to prepare a wide range of oxides. However, the most well-known and extensively studied examples are those based on silica chemistry where there is a good understanding of how the chemical parameters can be used to control the properties of the final product. In general, the applications of the sol-gel method can be loosely divided into three interdependent aspects of synthesis, processing, and properties, which are a direct function of the structural composition, morphology, and microstructure. The sol-gel method for synthesis of materials is particularly appealing because with a chemical modification of the precursor at molecular level, the functional properties of the final material can be changed. Therefore, by selectively integrating specific organic functional
S
S
2460
Sol-Gel Method, Fig. 1 Schematic depiction of the sol-gel method
Sol-Gel Method
Hydrolysis + Condensation
Precursor + Solvent + H2O + Catalyst
groups into the precursor at the molecular level, it is possible to introduce desired properties into the product sol-gel inorganic-organic hybrid material. The sol-gel chemistry is feasible with any metal precursor that is hydrolytically unstable. Typical precursors used are alkoxides, halides, or carboxylates. With silica chemistry, alkoxides [Si(OR4)] are predominantly used. In addition, it is possible to use organically modified alkoxides [(OR)3SiR0 ] to incorporate organic functionalities in glasses. The R0 group can be any organic group, including polymers or oligomers. When a precursor containing an organic group is hydrolyzed, it leads to formation of an ORganically MOdified SILicate (ORMOSIL) glass. The organic R0 groups can be simple hydrocarbons or can include other functional groups. In addition, it is possible to make glasses with precursors of the type [(OR)3Si-R-Si[(OR)3], where the organic functionality is present as a spacer group. The precursors can be used alone, or alternatively, a combination of precursors can be used to make hybrid or composite materials. Finally, prior to gelation, other species such as organic-, inorganic-, and biomolecules and/or polymers can be added to the liquid sol to make composite materials. When large biomolecules are added to the glass, they usually become physically entrapped in the glass but nonetheless retain their structure and properties and the resultant materials exhibit biological function and activities [3]. The sol-gel method is primarily a method at the interface of synthesis and processing. Synthesis refers to the making of new materials by means of specialized chemical precursors through arrangement of precise functionalities at the molecular level. On the other hand, processing refers to the specific organization of matter at the nano-, micro-, and macroscopic level. Processing of gels is greatly facilitated by the stability of the liquid sol, which enables construction of materials with good control over their nanostructure, shape,
Gelation
Sol
Gel
and overall geometry. The final gels can be fabricated in a variety of shapes, geometries, and configurations, including monoliths, coatings, thin films, fibers, and powders. The liquid sol can be used to make monolithic geometries whose final shape is determined by the container used. From the perspective of nanotechnology, the method provides a rather facile pathway to construction of different nanomaterials. Formation of oxide-based nanoparticles is a key area in use of the sol-gel method due to the unique opportunities provided by the solution-based method that proceeds via the sol state characterized by dispersed nanoscale particulate matter. The liquid sol can be used for dipcoating, spin-coating, and spray-coating to prepare films and coatings of varied thicknesses on diverse surfaces including glass, metals, composites, or plastics. The versatility of the method has resulted in a burgeoning utility of the sol-gel method in diverse fields. Applications are the key drivers in the utilization of the sol-gel method in technology. Because of its versatility, the method has found widespread applications in all walks of science and engineering including, optics, photonics, microelectronics, and biotechnology to name a few. The intrinsic properties of sol-gel materials such as optical transparency, porosity, surface area along with chemical and physical stability confer a unique vantage point for their applications in different areas. Different functions and properties can be imparted to the parent glass by incorporation of additional organic, inorganic, or biological entities. Particularly important in this direction are the organosilica sol-gels and hybrid materials, which integrate the properties of both organic and inorganic materials and constitute a novel platform for development of new materials and devices from a molecular perspective. Similarly, the access to formation of bio-hybrid via integration of biological species makes the method particularly suited for the fabrication of
Sol-Gel Method
biological-inorganic hybrids and composites. For the design of advanced materials, the advantage of using sol-gel derived glasses is that the parent silica material is structurally inert, functionally inactive, and operationally nonresponsive. Therefore, by selectively integrating specific (bio)chemical entities into the glass (either through precursor modification or through encapsulation), it is possible to introduce desired structural, functional, and operational properties in a modular fashion. The introduction of properties can be singular, sequential, or even multiple. The strategy offers a powerful approach for designing a diverse range of sol-gel-derived materials whose properties can be tailored with a degree of control over the compositional, morphological, and functional characteristics of the product materials. This flexibility enables exploring new opportunities, and materials made using the sol-gel method have found a variety of applications in optics, photonics, magnetics, biocatalysis, detection, sensing, controlled release, and separation in addition to providing access to a range of structures and morphologies in the form of particles, fibers, coatings, and other nanoscale, nanostructured, and nanoporous materials. The sol-gel method has brought about a fundamental change in synthesis and processing of inorganic materials. The fact that final materials in different geometries and morphologies can be prepared from a solution circumvents the need to go via conventional solid-state high temperature processing pathways. The solution-based method provides considerable flexibility in making pure, homogenous materials as well as composites and hybrids. The method can be used to make materials by using a given precursor, or alternatively, by mixing a plurality of different precursors, it is possible to form multicomponent mixed-oxides. Similarly, the method can be used to make hybrids and composites via integration of polymers or other organic/biological components to obtain materials with varied properties and applications (Fig. 2). The low temperature at which the materials can be prepared makes the method economical and environmentally friendly for practical applications.
Basic Methodology The sol-gel method of making glasses begins with the reaction of hydrolyzable precursors with water. The
2461
S
most commonly used precursors are alkoxides. The process is illustrated herein with the alkoxysilanes as an example; however, the reaction sequences are analogous for other metals. These precursors are dissolved in a compatible solvent such as an alcohol followed by addition of water to initiate hydrolysis. Often a catalyst is used to accelerate the reaction. The nonpolar nature of alkoxysilanes makes them immiscible with water, and therefore, the reactions proceed at very slow rates even in the presence of a catalyst. The mixing can be enhanced by use of ultrasound, or alternatively, the reaction mixture is refluxed at elevated temperature to facilitate the reaction. Under these conditions, first the alkoxides react with water and form hydroxylated species which then condense to form colloidal nanoparticles containing oxo-bridges. As the reactions proceed, the sol particles combine to form an extended network and the reaction mixture becomes viscous, ultimately leading to the formation of a solid gel. The as-prepared gels are biphasic porous materials with interstitial solvent phase. Drying of the gels under ambient conditions results in evaporation of a majority of the solvent phase (water and alcohol) and the gels shrink by up to 80% in volume. These dried glasses – termed xerogels – are mechanically and dimensionally stable materials that contain some solvent phase, which can be further dried at elevated temperatures to form dense glasses. The overall method of making oxides from alkoxides is represented as Precursor
Hydrolysis
Gelation
Drying
! Sol ! Gel ! Condensation Densification
Xerogel ! Glass
The sequence is characterized by a series of reactions along with physical and chemical changes that take place along the precursor to sol to gel to xerogel to glass transformation steps. The properties of materials depend upon the conditions employed and by controlling the parameters under which these steps take place one can modulate, regulate, and fine-tune the properties of the materials at each stage. The Precursor: The nature, composition, and structure of precursor molecules is central to the properties of final product. Controlling the number of hydrolyzable versus nonhydrolyzable groups on the precursor provides an easy pathway to influence the chemical makeup of the materials obtained by the
S
S
2462
Sol-Gel Method • Precursor Chemistries • Organically-Modified Precursors • Hydrogels • Organic-Inorganic Hybrids and Composites • Dopant-Matrix or Host-Guest Chemistry • Biological-Inorganic Hybrids and Composites
Synthesis
• Optical • Photonic • Electronic • Magnetic • Mechanical • Chemical • Biochemical
Properties
The Sol-Gel Method
Processing
• Particles • Spheres • Fibers • Coatings • Thin Films • Mesoporous Silica • Organosilica
Applications
• Sensors • Separation Media • Delivery and Release Vehicles • Immobilized Enzymes as Biocatalysts • Surface Treatments, Films, and Coatings
Sol-Gel Method, Fig. 2 Representative summary of significant aspects of the sol-gel method
sol-gel method. Therefore, in order to obtain pure oxide-based materials, one can use precursors of the type Si(OR)4 wherein all the alkoxides are replaced with oxygen atoms in the product due to hydrolysis and condensation. Use of nonhydrolyzable ligands on the precursors ensures that these groups will remain in the final material thereby altering the structural, morphological, and functional characteristics of the product. Precursors of the type (OR)3SiR0 where R0 is organic group provide an easy access to introduce organic functional groups that can impart specific properties. Additionally, use of multiple precursors with different functional groups provides further access to tailoring the final properties. Alkoxides are the preferred precursors of choice due to the fact that the by-product of their hydrolysis is the corresponding alcohol, which can easily be evaporated to obtain relatively pure materials [4].
The rates of hydrolysis and condensation depend upon the nature of the alkoxide groups, which can lead to dramatic variations in the rates of the reactions as well as product morphologies. The rates of hydrolysis of alkoxides depend upon the polarity of M-OR bond as well as the steric environment around the central metal atom which can influence the approach of water molecule for hydrolysis of the bond. In general, more polar alkoxides are more susceptible to hydrolysis in a polar solvent environment. The inductive and steric effects of the alkoxide ligands can also alter the hydrolysis rates by altering the polarity of the M-OR bonds, and by modulating the access of water molecules to effect the hydrolysis step. Usually, the hydrolysis step is much slower with longer chain and/ or branched alkoxides which inhibit the access of water molecules to the metal site. A catalyst is typically used, especially with silicon alkoxides, to accelerate the
Sol-Gel Method Sol-Gel Method, Fig. 3 Hydrolysis mechanisms in acid or base catalyzed reactions
2463
S
Acid Catalyzed
R RO
O
H O
+
Si
O OR
RO
OR
RO
H
Si
H
H O
H
R RO
−ROH −H+
Si OR
RO +
OR
HO
OR
RO HO
Si RO
OR
RO
H+
−H+
HO−
H
Si
OR
RO
−RO−
RO Base Catalyzed
hydrolysis step. On the other hand, alkoxides of transition metals and other highly electropositive elements usually exhibit much faster hydrolysis due to the intrinsic polarity of the M-OR bond along with the ability of these elements to accommodate the incoming water molecules via coordination to the metal site. Hydrolysis: The first step of the sol-gel reaction is the replacement of alkoxide groups in the precursor molecule with water. The alkoxide groups leave as the corresponding alcohols along with formation of hydroxylated silanol species. Since alcohol is a by-product of the hydrolysis reaction, carrying out the reaction in the same alcohol as the corresponding alkoxide can alter the reaction equilibrium. If a different alcohol is used as a solvent then, due to its predominance, there is usually a replacement of the alkoxide ligand via transesterification. The ease with which the reaction occurs depends upon the miscibility of the precursors with water, its hydration characteristics, and its ability to accommodate water molecules in the primary coordination sphere. The products of the hydrolysis reaction are the hydroxylated precursor and the corresponding alcohol. The general reactions is given as ðORÞ4 Si þ H2 O ! ðORÞ3 Si OH þ ROH The hydrolysis reaction proceeds via associative (or SN2) pathways such that the transition state is characterized by an increase in coordination number. The reaction is usually acid or base catalyzed to make it occur at an appreciable rate. The general reaction under both conditions proceeds via associative
pathways through the aid of the predominant catalytic species, namely, protons in acidic conditions and the hydroxide ions under basic conditions. Under acidic conditions, the reaction occurs in three steps (Fig. 3). First, the alkoxide group gets protonated, which makes the metal-alkoxide bond weaker and the silicon atom electron-deficient. Second, the water molecule binds to the silicon atom to form a positively charged pentacoordinate transition state. Finally, the release of alcohol molecule completes the reaction. Under basic conditions, the reaction also occurs in three steps (Fig. 3). First, the water molecule loses its proton to form the hydroxide ion. Second, the hydroxide ion binds to the silicon atom to form a negatively charged transition state. Finally, the alkoxide ion dissociates to form the hydroxylated species. The replacement of alkoxides to form hydroxylated species proceeds at different rates under acidic and basic conditions. Under acidic conditions, subsequent hydrolysis steps are retarded after the initial substitution of alkoxide, while in basic medium, these follow-on steps are accelerated. In other words, the hydrolysis of [(OR)4nSi(OH)n] (where n varies from 0 to 4) species in acidic medium is faster for smaller values of n while in a basic medium the hydrolysis is faster for larger values of n. As such, the formation of multihydroxylated species predominates in the basic medium. The rate and extent of hydrolysis can be altered by use of suitable solvents that can change the polarity of the medium, and consequently, the stability of polar versus nonpolar species to shift the equilibrium. Solvents that interact with water by hydrogen bonding interactions also alter the activity of water in
S
S
2464
Sol-Gel Method
Sol-Gel Method, Fig. 4 Condensation mechanisms in acid or base catalyzed reactions
Acid Catalyzed H H
O
HO
+
Si
+H+
OH
HO
OH
HO Si
OH
HO
−H3O+
Si HO
OH
HO
OH
HO
O
Si OH
O−
+HO HO −
HO +
Si HO + H O 2
OH
Si
HO Si HO
OH −HO−
OH
OH
OH
OH
Base Catalyzed
the solution and affect the rate and extent of hydrolysis. Solvents that can homogenize the reactants to make them more miscible augment the rates and shift the equilibrium toward products. Finally, use of ultrasound can also facilitate mixing of the precursor with water and can accelerate the reaction. Condensation: Once the reactive hydroxylated species are formed, they begin to undergo condensation reaction to form Si-O-Si linkages to generate oligomers and polymers. The reaction can be represented as ðORÞ3 Si OH þ OH SiðORÞ3 ! ðORÞ3 Si O SiðORÞ3 þ H2 O As with hydrolysis, the condensation reactions are dependent upon pH of the medium and can be acid- or base-catalyzed (Fig. 4). Under conditions of acid catalysis, the reaction proceeds in three steps. First, there is a rapid protonation of the silanol group to form a positively charged species. Second, the protonated positively charged species is attacked by a neutral hydroxylated species. Finally, an elimination of the hydronium ion completes the reaction. Base catalysis follows a slightly different sequence. First, there is rapid deprotonation of the silanol group to form an anionic silicate species. Second, the negatively charged species coordinates with the neutral hydroxylated species via formation of a pentacoordinate transition state. Finally, the expulsion of hydroxide ions completes the reaction sequence. Analogous to hydrolysis reactions, the rates of the condensation reaction are dependent upon pH of the medium. Condensation reaction for [(OR)4nSi(OH)n]
(where n varies from 1 to 4) species in acidic medium is faster for smaller values of n while in a basic medium it is faster for larger values of n. As such, basecatalyzed systems, with a predominance of multihydroxylated species, are characterized by faster gelation times as compared to acid-catalyzed reactions. Silanol (Si-OH) groups become more acidic as the silicon atom gets increasingly surrounded by the electron withdrawing oxo groups. This high acidity of silanols with the increasing extent of condensation results in maximum condensation rates in the intermediate pH range of 3–5. In very high pH region (pH 10), condensation reactions occur but gelation is thwarted by repulsion due to surface charges leading to formation of discrete particles. As water is a byproduct of condensation reactions, the amount of water in the reaction mixture influences the equilibrium. Reaction media high in water content favor hydrolysis but hinder condensation. Similarly, more polar and hydrogen bonding solvents can stabilize the hydroxylated species to retard the condensation process. Finally, the steric effects of the remaining alkoxo groups can also influence the rate with bulkier groups hindering the condensation reactions. Gelation: As the hydrolysis and condensation reactions continue, the reaction medium becomes more viscous due to increased degree of polymerization. The viscous sol continues to evolve until the viscosity reaches a critical point where the liquid can no longer maintain its fluidity. Gelation is characterized as the point along the hydrolysis-condensation reaction coordinate when the liquid sol turns to a solid gel. Gelation is typically defined by gelation time (tg) as the time taken from the start of the reaction till the formation of
Sol-Gel Method
a solid gel. Gelation times are dependent upon a variety of factors, as discussed above, that control the rate and extent hydrolysis and condensation reactions. At gelation time, the growing polymer structure essentially gets “frozen” into an immobile state. The initial gels formed have high viscosity but they lack the mechanical stability due to lack of an extensive interconnected network. Even though the physical state of the reaction mixture changes from liquid to solid, the hydrolysis and condensation reactions still continue to evolve leading to formation of additional Si-O-Si linkages. The increasing degree of condensation results in enhanced mechanical strength over time even after gelation has occurred. Aging: The as-formed gel is a porous structure with pendent silanols and even some remaining unhydrolyzed alkoxides on the polymeric network. Aging of the gels under closed conditions, usually extending over a period of hours to days, is employed to enable completion of these reactions. During aging, continuation of hydrolysis and condensation reactions leads to an increase in cross-linking and overall stiffening of the gel network. During aging, the gels shrink and expel the solvent phase due to a decrease in internal pore volume caused by extensive condensation of silanols to siloxane linkages. Aging is a necessary step in processing of bulk gels to ensure mechanical stability of these materials during subsequent stages. Drying: Drying of the gels either under ambient conditions or elevated temperature is used to expel solvent from the gels. The elimination of solvent from the pores of the gels compacts the network and facilitates further condensation reactions between silanols on the surface of pores. The ambiently dried gels can lose up to 70–80% of the solvent and shrink considerably due to compaction of the pore network. In order to completely dry the gels, use of elevated temperatures is necessary. Drying enhances the mechanical stability of the gels, and the dried gels are considerably stiffer and stronger materials as compared to aged gels. The dried gels have a greater density and are dimensionally invariant under ambient conditions. Sintering/Densification: Heat treatment of the gels further eliminates the porous structure to obtain ceramics or dense glasses. During heat treatment, adsorbed and hydrogen-bonded solvent is removed and the collapsing pore structure causes further condensation of silanols to siloxane linkages. Heat
2465
S
treatment makes the gels nonporous and harder with densities and structure approaching close to fused silica. Nonhydrolytic sol-gel method: The sol-gel method depends upon water to initiate the hydrolysis of precursors. This presents a challenge for system that are either unstable in water or too reactive. In order to circumvent these issues, the nonhydrolytic sol-gel method [5] utilizes nonaqueous solvent medium and does not require the hydrolysis step. While there exist many different variants, the general nonhydrolytic sol-gel method can be represented as MðORÞn þ Xn M ! ½ðORÞn1 M O MðORÞn1 þ RX where X ¼ halide, -OR0 , or -O2CR0 . The reaction bypasses the hydrolysis process altogether and leads to direct formation of oxo-bridged linkages. The reaction is essentially a condensation pathway that is usually carried out at elevated temperatures. The nonhydrolytic process is advantageous in a system wherein a greater degree of control over the relative rates of hydrolysis and condensation to fine tune product microstructure, porosity, and morphology may not be possible with the hydrolytic pathway. Formation of nanoparticles and powders: The molecule-nanomaterial interface along the precursor to sol reaction coordinate provides a convenient access to a variety of nanoscale systems. Powders such as precipitated or fumed silica have been long used in commercial applications. A key challenge has been to obtain uniform monodisperse nanoscale particles with a control over their nuclearity, shape, and dimensions. The issue of uniformity and homogeneity, to a certain extent, is addressed by utilizing specific reaction conditions to prevent interparticle interactions. Two methods commonly used to stabilize isolated discrete nanoparticles are based on manipulating electrostatic and steric factors. As discussed above, highly basic conditions favor rapid hydrolysis and condensation while producing negative surface charges which favor the formation of particulate structures. The Stober method of forming particles specifically takes advantage of this approach. Uniform size spherical particles can be obtained under basic conditions by using dilute solutions to prevent interparticle aggregation and to allow dissolution-growth phenomena to occur without precipitation. Usually the concentration
S
S
2466
Sol-Gel Method
Sol-Gel Method, Fig. 5 Schematic depiction of dip-coating (left) and spincoating (right) processes
of the precursor in the reaction mixture is optimized to prevent interparticle polycondensation and prevent formation of larger particles due to particle growth and coalescence. Sometimes, a synthesis strategy of combined acid and base catalysis is used for the initial formation of a stable sol under acidic conditions followed by addition of base to facilitate particle growth. Optimized reaction conditions that favor dynamic surface reorganization and reconstruction to provide uniform spherical particles are commonly utilized. Use of microemulsions has been another method that is employed to form spherical nanoparticles wherein the emulsion provides a sterically enclosing environment for isolated particles to form without aggregation. Formation of thin films and coatings: The sol-gel method has found wide applicability in the construction and formation of two-dimensional nanostructures such as thin films and coatings. The films made via the sol-gel method provide a better dimensional control with the ability to coat complex shapes, and can even be used to form multilayer coatings. In order to form a coating, a stable sol with controlled viscosity is desired. Usually stable sols can be obtained under acidic conditions and by adjusting the amount of the precursors, solvent, and pH it is possible to control the viscosity of the sol. The films can be formed by dipcoating, spin-coating, or spray-coating methods. The dip-coating method of thin film deposition is perhaps more important technologically since a uniform coating can be deposited onto substrates of large dimensions and complex geometries. In the dip-coating process, a suitable substrate immersed in a low viscosity sol is withdrawn slowly at a constant speed (Fig. 5). The adhesion of the sol layer on the substrate aided by gravitational draining leads to formation of a thin film. In order to deposit uniform homogeneous films, a steady vibration-free withdrawal of the substrate
from the sol is necessary. Film thickness is controlled by adjusting the viscosity of the sol and the rate of pulling the substrate. Another method of forming thin films is via spin-coating wherein an excess of sol is placed onto a substrate that is rotated at high speeds to disperse the liquid uniformly (Fig. 5). The film thickness in this case depends upon viscosity of the sol and the speed of rotation along with the rate of evaporation of the solvent. The conversion of deposited sol to a porous gel is caused mainly by solvent evaporation. Unlike the bulk gel system where gelation, aging, and drying occur sequentially over a period of several days, all of these processes typically occur within 30 s to 1 min in the thin-film such that the drying stage overlaps the gelation and aging stages. Loss of solvent molecules accelerates hydrolysis and condensation steps such that the two steps occur concurrently leading to the eventual formation of a porous xerogel structure. The important microstructural characteristics of coatings and films such as porosity, surface area, and pore size are determined by pH, concentration, and nature of precursors, viscosity of the initial sol, the solvent and its rate of evaporation, the rate of sol dispersion on the substrate, the ambient temperature, the rate of drying, and the relative rates of evaporation and condensation reactions during drying. Organically modified materials: Perhaps the most useful aspect of the sol-gel method is the ability to make materials with covalently integrated organic functionalities via hydrolysis of precursors containing hydrolytically stable organic groups. This enables a wide range of functional groups to be incorporated within the oxide-based network to from a diverse array of new materials with unique, novel, and interesting properties. More than one type of precursors with different functional groups can also be used to include a multiplicity of functional groups within a given material. An important consideration in use of multiple
Sol-Gel Method
precursors is their differential reactivities toward hydrolysis and condensation which can lead to phase separation and/or spatially isolated nanodomains. Strategies to overcome these are based on prehydrolysis of the more reactive precursor followed by addition of the other precursors as well as use of different alkoxides groups on the precursors to modulate their reactivities. The organically modified materials contain less than four siloxane linkages per silicon atom and are more elastic as compared to pure oxides. The ability to tune functionality in the organically modified materials provides appealing prospects for applications based on the nature of organic group incorporated. Hybrid Materials: Different types of hybrids can be made via the sol-gel method. These include organic-inorganic hybrids and nanocomposites. The organic-inorganic hybrids combine the properties of organic materials with the mechanical strength of the oxides. The main issue in formation of hybrids relates to their miscibility which can be enhanced by means of suitable solvents or by use of functional groups that can noncovalently interact with silica pore structure via hydrogen bonding or electrostatic interactions. The simplest way to make a hybrid material is by encapsulation of organic molecules in the pores of the gels. The organic molecules in the pores of gels can be used to alter the properties and structure of the inorganic network. Typical examples of these include glasses containing encapsulated catalysts, dyes, and other optically active organic molecules. Another approach is based on forming interpenetrating networks of organic polymer with the silica network. These nanocomposites combine the properties of both inorganic and organic polymers, and usually influence the structure, morphology, and properties of each other. These synergistic interactions make the hybrid materials structurally and functionally superior as compared to their component parts. An important development in formation of hybrid materials by the sol-gel method centers on formation of bio-hybrids. The solution based sol-gel pathway can be used for direct integration of biological species such as proteins, enzymes, DNA, RNA, and even whole cells. Optimized pH regions are necessary during preparation to stabilize the biological component. Typical procedure of enzyme immobilization involves initial acid hydrolysis followed by raising the pH to neutral region. The biological macromolecules are usually
2467
S
added to a buffered sol and get encapsulated in the growing gel network as the sol turns to a gel. The water-filled porous structure of the gel along with electrostatically charged and hydrogen bonding pore walls provide a noncovalently interacting environment for the biological molecules. The encapsulated biological entity retains its structural and functional characteristics and is able to interact with exogenous molecules that can diffuse in and out of the porous structure. The last two decades have seen tremendous advances in the design and development of bio-silica hybrids made by the sol-gel method [3].
Key Research Findings Structure and Synthesis: The sol-gel method has been used for several decades to make silica-based materials [6]. Oxides of different metals including mixed oxides have been prepared [7]. Initial research in this area focused on mainly pure and mixed oxides but in recent times the focus has shifted to organically modified, hybrid, and composite system. A classification of organic-inorganic materials synthesized using the solgel method has been developed to categorize them [8]. Nanotechnology has also spurred an active interest fabrication of sol-gel-derived particles [9]. These particles are usually synthesized in a microemulsion and can be solid, hollow, or contain core-shell structures. Synthesis of organically modified glasses has been studied in much detail and a wide range of precursor chemistries have been identified. The organically modified precursors provide access to a diverse range of materials with tunable properties including chemical reactivity, optical properties, electronic properties, enhanced mechanical stability, and elasticity. A major area of research using these precursors has been in the construction of mesoporous materials by means of using suitable templates wherein the assembly is aided by electrostatic and hydrogen bonding interactions between the template and the growing silica network [10]. Gels made from organosilicate precursors made with spacered linkers containing both hydrophobic and hydrophilic groups exhibit hydrogel-like properties and have been shown to undergo structural and volume change with respect to different environmental stimuli. These materials are characterized by enlarged pores due the
S
S
2468
presence of spacer units along with elasticity which confers the necessary flexibility to undergo shape and volume changes [11]. These materials also provide a suitable matrix for immobilization of biological molecules. An important aspect of protein and enzyme immobilization has been elucidation of local environment of the immobilized biomolecule and its interactions with the sol-gel matrix [12]. The immobilizing environment of the pore restricts translational diffusion, rotational mobility, and conformational flexibility. Proteins and enzymes have been shown to exhibit enhanced stability against denaturation when encapsulated in the sol-gel-derived glasses. The enzymes in the gel matrix retain their catalytic activities and the materials have been used as solidstate catalysts for various transformations. Studies have shown that when large proteins get encapsulated in the growing gel network, they act as self-specific templates for the formation of their own pore environment. Recent research also shows that cells can be used for directing assembly of sol-gel materials. Furthermore, proteolytic enzymes have been shown to catalyze the sol-gel reactions leading to formation of gels starting from alkoxide precursors. Function and Properties: The transparency of the sol-gel-derived glasses has been widely utilized for the development of functions related to modulation of photon flux through the materials [13]. Incorporation of dyes and luminescent fluorophores has been used as a means to develop materials for lasers, luminescent materials, and solar concentrators. Incorporation of photochromic molecules in the pores of the gels has been employed as a means to impart photochromic properties to the resulting glasses. Electronic properties have been developed in the sol-gel glasses via incorporation of conducting polymers, metal particles, or formation of mixed oxides. The nanopores of the gels have been used to form nanoscale superparamagnetic particles of transition metal oxides. The use of organically modified precursors also provides access to tailoring of mechanical properties such as strength, elasticity, and hardness depending upon the functional groups used. In addition to development of physical properties, use of precursors with specific functional groups has also been used to tune their chemical properties. Applications: The applications of the sol-gelderived systems have been continuously increasing
Sol-Gel Method
[14]. The transparency of silica-based systems makes them especially suitable for development of optical applications and sol-gel-derived materials have been used as optical sensors based on changes in color or luminescence upon reaction with an analyte [15]. The analytical applications of sol-gel materials as electrochemical sensors have also been developed [16]. Additionally, the presence of specific functional groups on organosilica materials confers selectivity in their interactions and these materials have been used as separation media in chromatography columns [17]. The use of porous particles for controlled release applications has opened up new opportunities in market and commercial products have been designed for acne treatment and sunscreen release based on particles made using the sol-gel method. A new area in this direction has been applications in fabric-care systems and environmentally sensitive organosilica hydrogels have been demonstrated as controlled release vehicles for water-triggered release of enzymes. Coatings made from sol-gel pathways have been utilized for an array of applications including corrosion protection and surface passivation, tailoring water affinity of surfaces, enhancing optical gloss, thermal insulation, scratch resistance, and fingerprint resistance among a host of other applications [18]. The ability to modify hydrophobicity and surface microstructure of these coatings has enabled construction of surface features that mimic the lotus leaf effects. An offshoot of this research has also resulted in textile treatments to make them stain- and waterrepellent. In recent times, the use of bio-sol-gel hybrids has been an area of much activity. Initial applications related to immobilization of protein and enzymes inside the pores of silica sol-gels to make solid-state biocatalysts for various chemical transformations. The enzyme sol-gel hybrids have also been used as optical and electrochemical sensors, and as media for bio-affinity chromatography. The use of sol-gelderived materials has also been established for immobilization of whole cells and microorganisms [19]. Organosilica gels as matrices for bioimmobilization have provided access to a wide range of biocatalysts whose activities can be exploited in practical applications. The applications domain of materials incorporating biological species continues to grow and the biological-inorganic composite area remains a topic of continued development.
Sol-Gel Method
Future Directions for Research The sol-gel method of making materials provides a versatile pathway to not only making new materials, but also making known materials more useful by obtaining them in different morphologies and length scales. The last decade has witnessed significant advances in construction of nanoscale particles, assemblies, and organized structures made using the sol-gel method. Fabrication of different oxides, nanohybrids, and nanocomposites characterized by distinct nanoscale structures, surfaces, and interfaces that can be tailored to obtain the desired functional responses would continue to remain an area of vital investigation in the near future. Use of nanoscale building blocks obtained by the sol-gel method and their organization into higher-order structures along with control, manipulation, and regulation of the emergent properties at different length scales to generate specifically tailored responses would provide an appealing paradigm for future studies. An additional key issue to address in the future development of nanoscale systems made via the sol-gel method would be resolving, manipulating, and controlling the functional dichotomy of properties originating from the nanoscale as well as collective properties due to the ensemble nature of the materials. Being able to tailor properties of materials made from molecular precursors with specific chemistries offers enticing prospects for controlling and manipulating their organization as they undergo their structural evolution from molecules to clusters to particles and other higher order structures. While there have been some advances in supramolecular assemblies of sol-gel-derived materials, the complexity and sophistication to achieve self-assembling structures with distinct interactions, interfaces, and organization over varied length scales remain as yet elusive. An area of ambitious exploration, given the formidable challenges, would be the development of new precursors and chemistries to fine-tune covalent and noncovalent interactions to generate supramolecular systems with programmed assembly and organization. Tuning this dynamic interplay of structure and organization would require new methods and techniques of sol-gel chemistry that would provide access to far from equilibrium structures. In the near term, the structuralorganizational paradigms from nature would continue to provide the necessary inspiration toward achieving these objectives.
2469
S
The ability to fabricate bio-inorganic hybrids and composites has opened another exciting avenue in the design of novel materials with useful properties and applications. The integration of biological entities into the nanopores of materials, to a certain extent, restores the spatial and supramolecular organization that exists in the native cellular environment which is lacking in freely mobile biomolecules dispersed in an isotropic solution phase. The next extension of this method would generate organized assemblies that would take advantage of the specialized interactions at the biological-inorganic interfaces. Consolidated nanostructures integrating biological materials would become increasingly significant in creating a new generation of bio-hybrids that would combine the structural integrity of the sol-gel-derived materials with the functional and operational diversity of biomaterials. During the last decade there have been significant advances in mimicking nature with sol-gel-derived materials including biocatalysis, immunodetection, surfaces that mimic lotus leaf, and biomineralized supramolecular architectures. The opportunities provided by the facile access to nanomaterials, internal porous architecture, surfaces, and interfaces would continue to evolve as chemists, materials scientists, biochemists, and physicists attempt to develop correlations between structure, morphology, form, and function. The next evolution of these materials lies in the design of nanomaterials with adaptive capabilities that can adapt to their environment and exhibit active functional responses to generate dynamically active systems capable of self-diagnosis and self-correction. The interplay of structure, form, function, and operation that exists at different length scale would provide the guidelines for the development of next generation of materials capable of intelligent responses. While recent developments have seen the emergence of materials with smart functions, many of the systems, for the most part, remain monofunctional and passive. In the long range, design and development of new materials with hierarchical functions that span a range of functional length scales and tunable active responses would generate the next wave in this line of research. The last decade has also seen utilization of sol-gel methods to fabricate devices for detection, diagnosis, separation, sensing, controlled release, and delivery to name a few. The real potential of these materials as nanoscale devices remains to be explored. The ability
S
S
2470
to tailor size, porosity, and the surfaces of sol-gel nanomaterials would find new avenues of research activity in the development of nanodevice technology based on a combination of inorganic, organic, and biological components. Multifunctional systems that can detect, sense, separate, sort, release, and deliver molecules to targeted sites and in a predetermined fashion would constitute the long range evolution of nanodevices made using sol-gel-derived systems. In the near term, effective utilization of strategies to adjust and modulate physical and chemical interactions of exogenous molecules with sol-gel systems would provide the necessary means of control and regulation to design nanoscale sensors, filters, separators, sorters, and delivery agents. Development of nanodevice systems for biomedical applications would also constitute an important area of research and development. Finally, another area that would become increasingly important is the development of new “green” technologies utilizing the sol-gel method. The opportunities offered by the sol-gel method would provide the ideal framework for development of novel (bio) catalytic systems, molecular filters, sorters, or separators to reduce, remove, and/or remediate toxic and harmful substances. The sol-gel method is a greener alternative to conventional methods, however, use of volatile solvents and formation of by-products remain as issues to be addressed. Development of new processing techniques that minimize waste and improve atom economy by eliminating by-products would be essential to the development of eco-friendly manufacturing processes based on sol-gel methods. In the long range, sustained development and refinement of the sol-gel method would be critical to displace the current manufacturing processes in large-scale production technologies.
Cross-References ▶ Bioinspired Synthesis of Nanomaterials ▶ Biosensors ▶ Self-Assembly of Nanostructures ▶ Smart Hydrogels
References 1. Wright, J.D., Sommerdijk, A.J.M.: Sol-Gel Materials— Chemistry and Applications. CRC Press, Boca Raton (2001)
Solid Lipid Nanocarriers 2. Sanchez, C., Belleville, P., Popall, M., Nicole, L.: Applications of advanced hybrid organic–inorganic nanomaterials: from laboratory to market. Chem. Soc. Rev. 40, 696–753 (2011) 3. Avnir, A., Coradin, T., Lev, O., Livage, J.: Recent bio-applications of sol–gel materials. J. Mater. Chem. 16, 1013–1030 (2006) 4. Hubert-Pfalzgraf, L.G.: Alkoxides as molecular precursors for oxide-based inorganic materials – opportunities for new materials. New J. Chem. 11, 663–675 (1987) 5. Mutin, P.H., Vioux, A.: Nonhydrolytic processing of oxide-based materials – simple routes to control homogeneity, morphology, and nanostructure. Chem. Mater. 21, 582–596 (2009) 6. Hench, L.L., West, J.K.: The sol-gel process. Chem. Rev. 90, 33–72 (1990) 7. Livage, J., Henry, M., Sanchez, C.: Sol-gel chemistry of transition metal oxides. Prog. Solid State Chem. 18, 259–341 (1988) 8. Sanchez, C., Ribot, F.: Design of hybrid organic-inorganic materials synthesized via sol-gel chemistry. New. J. Chem. 18, 1007–1047 (1994) 9. Ciriminna, R., Sciortino, M., Alonzo, G., de Schrijver, A., Pagliaro, M.: From molecules to systems – sol-gel microencapsulation in silica based materials. Chem. Rev. 111, 765–789 (2011) 10. RIvera-Munoz, E.M., Huirache-Acuna, R.: Sol-gel-dervied SBA-16 mesoporous material. Int. J. Mol. Sci. 11, 3069–3086 (2010) 11. Rao, M.S., Grey, J., Dave, B.C.: Smart glasses – molecular programming of dynamic responses in organosilica sol-gels. J. Sol-Gel. Sci. Technol. 26, 553–560 (2003) 12. Menaa, B., Menaa, F., Aiolfi-Guimaraes, C., Sharts, O.: Silica-based nanoporous glasses – from bioencapsulation to protein folding studies. Int. J. Nanotechnol. 7, 1–45 (2010) 13. Penard, A.-P., Thierry, G., Boilot, J.-P.: Functionalized solgel coatings for optical applications. Acc. Chem. Res. 40, 895–902 (2007) 14. Aegerter, M.A., Menning, M.: Sol-gel Technologies for Glass Producers and Users. Kluwer, Boston (2010) 15. Tran-Thi, T.-H., Dagnelie, R., Crunaire, S., Nicole, L.: Optical chemical sensors based on hybrid organic–inorganic sol–gel. Chem. Soc. Rev. 40, 62–639 (2011) 16. Walcarius, A., Collinson, M.M.: Analytical chemistry with silica sol-gels – traditional routes to new materials for chemical analysis. Annu. Rev. Anal. Chem. 2, 121–143 (2009) 17. Kloskowski, A., Pilarczyk, M., Chrzanowski, W., Namiesnik, J.: Sol-gel technique – a versatile tool for adsorbent preparation. Crit. Rev. Anal. Chem. 40, 172–186 (2010) 18. Aegerter, M.A., Almeida, R., Soutar, A., Tadanaga, K., Yang, H., Watanabe, T.: Coatings made by sol–gel and chemical nanotechnology. J. Sol-Gel. Sci. Technol. 47, 203–236 (2008) 19. Livage, J., Coradin, T.: Living cells in oxide glasses. Med. Mineral. Geochem. 64, 315–332 (2006)
Solid Lipid Nanocarriers ▶ Solid Lipid Nanoparticles - SLN
Solid Lipid Nanoparticles - SLN
2471
S
Surfactant/Co-surfactant Interface
Solid Lipid Nanoparticles - SLN Claudia Musicanti and Paolo Gasco Nanovector srl, Torino, Italy
Synonyms Lipospheres; Solid lipid nanocarriers; Solid lipid nanospheres; Solid lipid-based nanoparticles
Definition Solid lipid nanoparticles (SLN) are drug carriers in submicron size range (50–500 nm) made of biocompatible and biodegradable lipids solid at room and body temperature.
Introduction Solid lipid nanoparticles (SLN) were developed at the beginning of the 1990s as alternative colloidal carriers to emulsions, liposomes, and polymeric nanoparticles. SLN have attracted increasing attention as delivery system for hydrophobic drugs: prepared with lipids, they can be administered by different routes of administration such as oral, parenteral, dermal, transdermal, ocular and pulmonary. A typical model of the structure of SLN consists of a solid lipid core surrounded by an emulsifier interface which stabilizes the particle (Fig. 1). SLN can offer several advantages for drug delivery 1. Possibility of controlled release of drug from lipid matrix 2. Possibility of drug targeting (active and passive) 3. Protection of incorporated drug from degradation 4. Increasing drug bioavailability 5. Feasible incorporation of hydrophobic and also hydrophilic drugs (lipid drug conjugates or other techniques) 6. Very low when absent toxicity 7. Affinity for biological barriers 8. Modification of pharmacokinetic parameters and drug distribution in organs 9. Possibility of large-scale production
Lipid core
Solid Lipid Nanoparticles - SLN, Fig. 1 Model representation of general structure of SLN
On the other hand SLN show certain possible limitations: 1. Poor drug loading capacity 2. Possible drug release during storage 3. Possible stability problems during long-term storage (aggregation, component degradation, tendency to form gel) The principal components used for the preparation of SLN are: 1. Lipids, such as triglycerides (trilaurin, trimyristin, tripalmitin, tristearin), mono/di/triglycerides mixtures (glyceryl stearate, glyceryl palmitostearate, glyceryl behenate), fatty acids (stearic acid, palmitic acid, behenic acid), waxes (cetyl palmitate), cholesterol esters 2. Emulsifiers (surfactants), such as phospholipids (e.g., lecithins), polysorbates (Tween) and sorbitan esters (Span), polymers (e.g., poloxamers), generally used to stabilize interface between water and lipid 3. Co-emulsifiers (cosurfactants), such as bile salts, short-chain fatty acids and alcohols Emulsifiers are amphiphilic molecules which possess both hydrophobic and hydrophilic portions; they are able to stabilize emulsion system by placing at the oil/water (o/w) interface (orienting the hydrophobic portion to the oil phase and the hydrophilic portion to the aqueous phase) and lowering the interfacial tension between the two phases. Co-emulsifiers intercalate between the emulsifier molecules and
S
S
2472
contribute to reduce the interfacial tension between oil and water. Additional ingredients are used to specifically modify surface of SLN: 1. Stealth agents, such as polyethylene glycol (PEG) for improving circulation time 2. Charge modifiers to modify surface charge 3. Targeting molecules, such as peptide or antibody fragment, for linking to specific site receptor (active targeting) Antioxidants, antimicrobial agents, and thickening agents can be usually added to SLN (dried or in dispersion) as for other pharmaceutical products.
Preparation Methods of SLN General issue in the preparation of SLN is to obtain liquid phase where lipid and water can be emulsified before SLN formation: lipids, which are solid at room temperature, are melted or they are dissolved into solvents. Main methods to prepare SLN are reported below. High-Pressure Homogenization (HPH) High-pressure homogenization is a well-established technique on the large scale and it is already available in the pharmaceutical industry. High-pressure homogenizers have been used extensively in the production of nanoemulsions for parenteral nutrition. High-pressure homogenizers push a liquid with high pressure (100–2,000 bar) through a micron size gap: by applying high pressure, the liquid accelerates to high velocity (over 1,000 km/h), and the resulting shear stress and cavitation forces break down the accelerated particles to submicron size. HPH technique can be applied by two main different approaches: hot high-pressure homogenization (HHPH) technique and cold high-pressure homogenization (CHPH) [1]. For both techniques, the drug is dissolved or dispersed in the melted lipid at approximately 5–10 C above its melting point. For the HHPH, the drug-loaded lipid is dispersed under high-speed stirring in a hot aqueous surfactant solution maintained at same temperature, to form a pre-emulsion. The obtained pre-emulsion is then subjected to high-pressure homogenization, still at temperature above melting point of lipid, to produce
Solid Lipid Nanoparticles - SLN
a hot o/w nanoemulsion. The homogenization process can be repeated several times: in most cases three to five homogenization cycles at 500–1,500 bar are performed. Increasing the homogenization pressure or the number of cycles often results in an increase of the particle size due to particle coalescence occurring for the high kinetic energy of the particles. Obtained o/w nanoemulsion is then cooled down to room temperature where SLN solidify. In CHPH, the drug-loaded lipid is rapidly cooled using liquid nitrogen or dry ice to obtain a solid solution which is then ground (milled) to obtain microparticles, approximately in the range 50–100 mm, then the solid microparticles are dispersed in a cold surfactant solution to obtain a pre-suspension. The presuspension is subjected to high-pressure homogenization at room temperature or below to obtain solid lipid nanoparticles: usually larger particle size and broader size distributions are obtained compared to hot homogenization. Cold homogenization minimizes the thermal exposure of the drug, although it does not avoid it, due to dissolution of drug in melted lipid in the initial step. Main reported disadvantages of HPH are: high energy-intensive operating conditions and potential thermal degradation of drug and excipients during production process. Warm Microemulsion Technique Microemulsions are transparent (clear), optically isotropic and thermodynamically stable systems. They are dispersions of two immiscible liquids (oil and water) stabilized by surfactants and optionally by cosurfactants. Depending on the composition of the system different structures can be formed: 1. Oil dispersion in water medium – o/w microemulsion 2. Water dispersion in oil medium – w/o microemulsion 3. Bicontinuous structure In o/w and w/o microemulsions, the dispersed phases are in shape of very small drops (10–100 nm) stabilized by an interfacial film of surfactant and cosurfactant molecules. In bicontinuous structures, regions of water and oil are interdispersed with no spherical geometry: it occurs when the amount of water and oil are comparable. The thermodynamic stability of microemulsion is due to the very low interfacial tension which allows the
Solid Lipid Nanoparticles - SLN
spontaneous formation of microemulsions without high energy input. Warm microemulsions obtained with melted lipids have been used as precursor for the preparation of SLN: after their preparation at temperature ranging from 55 C to 85 C depending on melted lipid used, they are then dispersed in cold aqueous medium where melted lipid drops solidify into SLN [2]. The hydrophobic drug is dissolved in mixture composed of melted lipid and surfactant maintained at warm temperature, the cosurfactant is dissolved in water and the obtained solution is heated at the same temperature of the lipid mixture; cosurfactant aqueous solution is then added to the mixture composed of lipid, drug and surfactant and by mild mixing an o/w microemulsion is obtained and successively dispersed in cold water at 2–3 C under mechanical stirring. Also hydrophilic drugs have been loaded into SLN by water in oil in water (w/o/w) double microemulsion technique: water, where hydrophilic drug is dissolved, constitutes the internal phase of w/o microemulsion, and the melted lipid (55–85 C) the external phase; the w/o warm microemulsion is then added to an external aqueous solution of surfactant and cosurfactant maintained at the same temperature and after mixing w/o/w microemulsion is obtained and dispersed in cold water to obtain SLN. The SLN dispersion, obtained by warm microemulsion technique, is usually purified by tangential ultrafiltration in order to remove surfactant and cosurfactants not incorporated into SLN, and to concentrate dispersion when needed. Advantage of the production of SLN by warm microemulsion include: low mechanical energy input, flexibility of interphase composition that allow surface functionalization, regular spherical shape. Potential drawbacks are thermal exposure, which can cause degradation of drug and of excipients, use of higher amount of surfactant and cosurfactant, which sometimes need to be removed. High Shear Homogenization and Sonication Following this process, drug dispersed in melted lipid phase is homogenized by rotor-stator homogenizer with a hot surfactant water solution to obtain nanoemulsion that is then cooled to form SLN. Process parameters that can affect particle size are: emulsification time, stirring rate, and cooling conditions.
2473
S
Homogenization can be used in association with sonication: melted lipid phase loaded with drug is homogenized with a hot surfactant solution to obtain a coarse emulsion that is then ultrasonicated to obtain final nanoemulsion before SLN are formed upon its cooling to room temperature [3]. Main possible disadvantages of this method include high energy input, broad particle size distribution, and potential damage of sensitive biomolecules. When sonication is used, potential metal contamination and temperature increase have to be considered. Solvent Emulsification-Evaporation Method In this method, drug, lipid matrix, and emulsifier are dissolved in a water-immiscible organic solvent which is then emulsified in an aqueous phase containing a water-soluble cosurfactant and homogenized. Upon evaporation of the solvent under reduced pressure, SLN are formed by precipitation of the lipid in the aqueous medium [4]. Parameters that affect mean particle size of nanoparticles obtained by this method are: surfactant/ cosurfactant blend and lipid concentration in the organic phase. Advantages of this technique are: avoidance of heat application, production of small size nanoparticles that can be sterilized by filtration through 0.2 mm pores. The major possible disadvantage of this method is the presence of organic solvent residues which can cause toxicity. Solvent Emulsification-Diffusion Method The process consists in dissolving the lipid in a partially water-soluble solvent (previously saturated with water) at room or controlled temperature, depending on solubility of lipid in the solvent. This organic phase is emulsified with an aqueous solution (saturated with solvent) containing the stabilizing agent and maintained at the same temperature, by a rotor-stator homogenizer. This o/w emulsion is then diluted with water maintaining a constant stirring and controlled temperature for promoting the diffusion of solvent of the internal phase toward the external phase, causing lipid aggregation and precipitation in form of SLN. Depending on its boiling point, solvent can be eliminated from SLN dispersion under reduced pressure or by washing (ultrafiltration). The selection of the water-miscible solvent and the stabilizers are critical parameters to obtain lipid
S
S
2474
particles in the nanometric range: usually solvents with high water miscibility and stabilizers able to form stable emulsions are preferred. It is possible to reduce the particle size by increasing the process temperature, the stirring rate, the amount of stabilizer, and by lowering the amount of lipids [5]. This technique is efficient and easy to be implemented, so feasible for industrial upscaling: no high energy is required and low physical (thermal and mechanical) stress is applied. Possible drawbacks of this method are solvent residues which need to be cleaned up and large dilution produced to obtain diffusion of solvent which needs further concentration steps. Solvent Injection Method In this procedure, solid lipid is dissolved in a watermiscible solvent or solvents mixture, and then rapidly injected through an injection needle into a stirred aqueous phase with or without surfactants. The SLN production by this technique is based on the rapid diffusion of the solvent across the solventlipid and solvent-water interface, causing precipitation of nanoparticles. Two simultaneous effects contribute to the effective formation of SLN: 1. Gradual solvent diffusion out of lipid-solvent droplets into water causes reduction of droplet size and simultaneously increases lipid concentration 2. Diffusion of pure solvent from the lipid-solvent droplet causes local variations in the interfacial tension at droplet surface, inducing reduction of size of droplets In this process particle size of SLN can be influenced and controlled by variation of process parameters such as injected solvent, lipid concentration, injected volume of solvent, lipid concentration in the solvent phase and viscosity of the aqueous phase [6]. This technique is simple and fast, without the need of sophisticated equipment; possible disadvantage of this method is the use of organic solvent which has to be considered for pharmaceutical applications and for this reason a complete removal of the solvents by ultrafiltration, evaporation, or freeze-drying is required. Coacervation Method This process allows to obtain SLN made of fatty acids by acidification of micellar solution of their alkaline
Solid Lipid Nanoparticles - SLN
salts: fatty acids sodium salts are dispersed in a polymer water solution (use of steric stabilizer polymer is essential to avoid particle aggregation) and heated under stirring just above the Krafft point of the fatty acid sodium salt to obtain a clear solution (Krafft point is the temperature at which the solubility of sodium soap increases dramatically and the solution becomes isotropic and transparent): an acidifying solution is then added drop-wise until pH about 4 is reached and the obtained suspension is then cooled down to 15 C in a water bath under stirring. Parameters that can influence size of nanoparticles and polydispersity index are: type and molecular weight of polymer and lipid concentration in the dispersion [7]. Although average dimensions are reported in a bigger range of 260–500 nm, this process presents several advantages like absence of solvents and need for common apparatus: the method is feasible and suitable for laboratory production and it is easy to be scaled up; studies are ongoing for loading of different drugs. Membrane Contactor Method This method employs a cylindrical membrane module: aqueous phase containing a surfactant is circulated in internal channel of membrane and melted lipid is pressed through pores of membrane into internal water flow allowing the formation of small droplets which are swept away by the aqueous phase; water is maintained at lipid melting temperature. SLN are then formed by cooling the preparation to room temperature. The membrane contactor allows the preparation of SLN with a lipid phase flux between 0.15 and 0.35 m3/h/m2 [8]. The advantages of this new process are its simple use, the control of the SLN size by an appropriate choice of process and membrane parameters, and its scaling-up abilities. Drug loading and surface modification, for targeting needs, have to be fully assessed and developed.
Characterization of SLN Particle Size, Size Distribution and Shape Particle size, shape and stability over time may influence several important pharmaceutical features of SLN
Solid Lipid Nanoparticles - SLN
2475
S
Solid Lipid Nanoparticles - SLN, Fig. 2 TEM micrographs of SLN obtained by warm microemulsion method and containing 2.5% of tobramycin in different physiological media: (a)
aqueous dispersion before administration to rats, (b) lymph, and (c) plasma after duodenal administration to rats (bar ¼ 100 nm) [9]
such as drug release, suitable administration route (e.g., particle size is a critical issue for intravenous administration, where particle size must not aggregate or enlarge by protein binding to exceed diameter of blood vessels) and in vivo biodistribution. Particle size and morphology of SLN are mainly affected by method of production and composition (lipid matrix, type and amount of emulsifiers, viscosity of aqueous phase, etc.). Main techniques commonly used for characterization of SLN are summarized below. Photon correlation spectroscopy (PCS) also known as dynamic light scattering (DLS), is widely used method for the measurement of particle size in range from nanometer to few micrometers. This technique determines hydrodynamic diameter of particles by measuring fluctuation of the intensity of scattered light which is due to Brownian motion of particles. Instruments usually report hydrodynamic diameter (Zaverage) and polydispersity index (PI), which give estimation of width of particle size distribution. Electron microscopy methods are very useful techniques for determining shape, morphology and size of lipid nanoparticles. The most commonly reported method for morphological characterization is transmission electron microscopy (TEM): different techniques can be applied for sample preparation (negative staining, freeze-fracture, sample vitrification (cryo-TEM)) and can provide different information about colloidal lipid structures; sample preparation is a crucial point as it can lead to structural alterations of the sample that need to be taken into consideration. As an example,
TEM micrographs of SLN (obtained by warm microemulsion method) are reported in Fig. 2 [9]: images show SLN in water dispersion before administration and same SLN in biological fluids coming from in vivo sampling after administration. Other methods to determine SLN particle size and morphology are based on scanning electron microscopy (SEM). Nonconductive samples, such as lipid-based nanoparticles, may be visualized without metal coating using specialized SEM instrumentation such as environmental SEM (ESEM) in which the sample is placed in a internal chamber at higher pressure than vacuum: positively charged ions generated by beam interactions with the gas, present in the chamber, help to neutralize the negative charge on the surface of sample. In field emission SEM (FESEM), the electron beam is produced with a cold cathode field emitter instead of a thermoionic emitter (tungsten filament heated with an electric current) as in conventional SEM: this allows for much higher resolution showing particle shape and size. As an example, images obtained from two different formulations of SLN, produced by warm microemulsion technique, are reported in Fig. 3. Atomic force microscopy (AFM) is another technique used to determine particle size and shape. AFM can be operated in different modalities depending on the specific application requirements: sample does not require any special treatments (such as coating) and technique also allows to work in ambient conditions, even in liquid environment. As an example, AFM characterization of SLN produced by solventdiffusion method is reported in Fig. 4 [10]: SLN
S
S
2476
Solid Lipid Nanoparticles - SLN
Solid Lipid Nanoparticles SLN, Fig. 3 FESEM image of SLN obtained by warm microemulsion method: (a) SLN with lipid matrix of Stearic acid, (b) SLN with lipid matrix of cholesteryl butyrate
c
100
0
0
0
0
100
20 [nm]
0
0
10 [nm]
b
40 [nm]
a
500
200 200
300
300
500 500
[nm]
500 1000
1000 [nm]
[nm]
Solid Lipid Nanoparticles - SLN, Fig. 4 Atomic force microscopy images of SLN prepared by solvent diffusion method in a nanoreactor system. (a) Blank SLN, (b) SLN with 10% of
clobetasol propionate charged (c) SLN prepared by conventional method with 10% of clobetasol propionate charged [10]
produced in miniemulsion used as nanoreactor system is compared with SLN obtained by conventional method.
can express tendency of particles to repulse or to attract: higher values of zeta potential (absolute value > 30 mV) usually imply more stable dispersions due to electrical repulsion, while low values can indicate possible colloid instability which could lead to aggregation. Z potential affects not only dispersion stability but also interaction with the biological surroundings (interaction with proteins and cells) and the in vivo fate of nanoparticles (biodistribution); it depends on the composition of the particle and on its surrounding medium: pH, electrolyte concentration, and concentration of components of the formulation. Z potential is determined by measuring the velocity of the particles dispersed in a liquid medium under the influence of an applied electric field (electrophoresis measurements): charged particles move toward the electrode under applied electric field with velocity dependent on the strength of electric field or voltage
Surface Charge Nanoparticles dispersed in a liquid medium are surrounded by an electrical double layer composed by an inner layer (stern layer) where the ions are strongly bound to particle surface and an outer layer (diffuse layer) where ions are less firmly associated with the particle. Within the diffuse layer there is a boundary (shear plane) inside of the ions and the particle form a stable entity: when the particle moves, only ions within the boundary move with the particle, while ions beyond the boundary remain in the bulk dispersant. Zeta potential (Z potential) is the electric potential at this boundary and it gives measureof surface charge. It is used to assess stability of colloidal systems as it
Solid Lipid Nanoparticles - SLN
gradient, the dielectric constant and the viscosity of the medium, and the Z potential of the particle. Electrophoretic mobility, defined as the velocity of the particle in a unit electric field, is related to Z potential by Henry equation and is usually measured by light scattering technique. Crystallinity, Polymorphism, Structure, and Stability Crystallinity and polymorphic transitions of lipids in the dispersed state may differ from bulk material due to the small particle size of colloidal system, the presence of emulsifiers used for the stabilization of SLN and the preparation method. In thermal analysis SLN dispersions usually show lipid melting point slightly shifted to lower temperatures and broader melting peak than the bulk lipid, polymorphic transitions are generally faster in SLN than in the bulk material [1, 11]. Crystallization and polymorphic behavior of SLN are correlated with drug incorporation and drug release: highly ordered crystal lattices (monoacid triglycerides) in SLN matrix can lead to drug expulsion, while less ordered crystal lattice (mixture of mono-, di-, and triglycerides), because imperfections which provide space to accommodate the drug, can allow better drug incorporation. Polymorphic transitions may cause alterations of lipid packing and thus of the internal structure of the nanoparticles, which may have negative consequences for drug loading. The course of polymorphic transitions depends on the type of lipid matrix and can be modified by other components of the dispersions [11]. As an example, triglycerides crystallize in three main polymorphic forms: a, b0 , and b; the b form is highly ordered with high thermodynamic stability, the a form is a less ordered form with low thermodynamic stability, the b0 form has intermediate characteristics between a and b forms. SLN based on triglyceride matrix tend to crystallize in the metastable a form: generally thermodynamically unstable configurations allow lipid molecules to have higher mobility causing lower density and a higher capability to incorporate drug molecules, but during storage the a form transform via b0 form in the more thermodynamically stable configuration b, characterized by higher packing density which can cause expulsion of the drug form structure. SLN may change their shape during polymorphic transitions: as an example, tripalmitin SLN have a spherical shape in a form, but they have a platelet
2477
S
shape in the b form. During storage the polymorphic transition from the a to the b configuration is connected with an increase of the particle surface due to preferred formation of platelets; surfactant molecules could not provide any more complete coverage of the lipid surface and particle aggregation can occur, leading to gelation (irreversible transformation of SLN dispersion into a viscous gel) [1]. SLN can display a lower crystallization tendency than the bulk material and may not crystallize properly after preparation and storage, at a temperature below melting point of lipid, forming super-cooled melts, which are liquid lipid nanoemulsions and not solid lipid dispersion. This phenomenon is particularly pronounced in SLN made from short-chain monoacid triglycerides such as trimyristin and trilaurin, and it is mainly due to size dependency of crystallization process that requires a critical number of nuclei to start, which cannot be reached in small droplets: tendency to formation of super-cooled melts increases with decreasing of droplet size [1, 11]. Hence, the physical state is a very important parameter which affects performance of SLN as drug delivery system: the techniques usually used to investigate physical state of SLN are differential scanning calorimetry (DSC) and X-ray diffraction (XRD) techniques. DSC is a thermal analysis technique which measures the difference in heat flow between the sample and a reference when they are both subjected to the same controlled temperature program: DSC quantifies the enthalpy changes during endothermic and exothermic process and it is a useful technique to determine melting point, melting enthalpy, degree of crystallinity, glass transitions of amorphous material, and to identify different polymorphic forms. In XRD techniques, an X-ray beam is focused on the sample, the scattered X-rays interfere with each other giving particular diffraction patterns: crystalline materials display many diffractions bands which depend on the disposition of the atoms within the unit cell of the crystalline lattice, while amorphous compounds present more or less a regular baseline. XRD techniques allow to differentiate between crystalline and amorphous material and to identify different polymorphic forms in SLN dispersions: the most commonly used are powder X-ray diffraction (PXRD), small angle X-ray scattering (SAXS), and wide angle X-ray scattering (WAXS).
S
S
2478
Nuclear magnetic resonance (NMR) is another useful technique to investigate structure of SLN for surfactant distribution on particle surface, drug distribution in the particle and potential formation of super-cooled melts. Due to different chemical shifts it is possible to attribute the NMR signals to particular molecules or their segments (functional groups); NMR active nuclei of interest are 1H, 13C, 19F, and 31P. Width and amplitude of NMR signals can be related to liquid and semisolid/solid state of nuclei: nuclei in the liquid state give sharp signals with high amplitude, while nuclei in the semisolid/solid state give weak and broad signals [1]. Drug Incorporation A broad range of drugs, mainly with hydrophobic properties, has been already incorporated into SLN, although the crystalline nature of the lipid matrix can offer limited space for drug incorporation inside the particle core [11]. Drug loading capacity of SLN is defined as the percentage of drug incorporated into nanoparticles referred to the total weight of the lipid phase or to the content of dispersed material; value of this parameter is usually in range 1–20%. Drug entrapment efficiency is defined as the percentage of drug incorporated into nanoparticles referred to the total drug added for the SLN preparation; values of this parameter is usually relatively high (80–99%). The distribution (localization) of the drug within SLN structure (Fig. 5) can vary considerably and three drug incorporation models have been proposed [12]: 1. Solid solution model. The drug is molecularly dispersed in the solid lipid matrix (SLN matrix as a solid solution). 2. Drug-enriched shell model. The drug is concentrated in the outer shell of the SLN. This situation can be explained by a higher solubility of the drug in the aqueous-surfactant outer phase at increased temperature in the production process; during cooling, lipid in the core starts to solidify and becomes less/not accessible for the repartition of the drug into the lipid inner phase leading to enrichment in the particle shell. High temperatures employed in the preparation process can increase solubility of the drug in the aqueous-surfactant phase promoting drug localization at the surface region.
Solid Lipid Nanoparticles - SLN
3. Drug-enriched core model. The drug is concentrated in the core of the lipid particle and it is surrounded by a lipid shell. This situation can be explained by a precipitation of the drug before the lipid crystallizes during cooling: this takes place preferentially when drug concentration in the lipid at process temperature is at its saturation solubility, so during cooling a super saturation and subsequent drug precipitation are achieved Drug-loading capacity of SLN is affected by different factors: 1. Solubility of the drug in the melted lipid. High solubility is a prerequisite to obtain sufficient drug loading; the amount of drug that can be dissolved in the lipid formulation may exceed the pure value of its solubility in lipid alone and a greater solubility in the whole formulation suggests that the drug can localize both in inner lipid phase and in external part on SLN (interfacial region) where surfactants are present. Solubility of drugs in the lipid can be improved by preparing lipid prodrugs such as stearic or palmitic acid derivatives (lipid drug conjugates). 2. Physical structure of solid lipid matrix. Complex lipids which form less ordered crystals with many imperfections can favor drug incorporation. 3. Polymorphic state of lipid matrix. Lipid transformation in more stable form reduces the number of imperfections in crystal lattice and can determine drug expulsion. Determination of drug-loading capacity can be usually performed by separation of free drug from the dispersion medium by employing different techniques such as ultrafiltration (tangential or centrifugal), ultracentrifugation, gel filtration chromatography, and dialysis. Drug Release Drugs incorporated into SLN are released by degradation and surface erosion of the lipid matrix and by diffusion of drug molecules through the lipid matrix. SLN are composed of physiological lipids for which metabolic pathways exist in the body: most important enzymes involved in in vivo SLN degradation are lipases, which are present in various organs and tissues [1]. Drug release from SLN is mainly affected by localization of the drug [12]: 1. Localization of the drug within the core of solid lipid matrix offers the possibility to obtain a prolonged drug release.
Solid Lipid Nanoparticles - SLN
2479
S
Solid Lipid Nanoparticles SLN, Fig. 5 Schematic representation of drug incorporation models for SLN
Solid solution
2. Localization of drug molecules on particle surface often leads to burst effect (fast initial drug release). SLN can show a biphasic drug release profile: an initial burst release, due to the drug localized at the surface, is followed by a more gradual release due to the drug localized in the lipid matrix. Possible presence of alternative colloidal species always has to be taken into account for drug release characterization: stabilizing agent cannot be localized exclusively on the lipid surface as expected, but also in the aqueous phase forming micelles, mixed micelles or liposomes that can solubilize the drugs and constitute alternative drug incorporation sites [1]. Composition of SLN is other important parameter that can affect drug release: in vitro experiments indicate that SLN show different degradation rates by lipases as a function of their composition (kind of lipid matrix and emulsifier). Degradation of triglycerides SLN show dependence to the length of the fatty acid chain: longer fatty acid chain shows slower degradation, while degradation of SLN made of waxes (cetylpalmitate) is slower compared to glyceride matrices [1]. Emulsifiers containing polyethylene glycol (PEG) chains can reduce SLN degradation by a hindering effect which reduces the anchoring of the enzyme on the SLN surface [1]. Since SLN are degraded by surface erosion, in SLN with small size and higher surface area, drug release is expected to be more rapid; nonspherical particles, such as thin platelet shape, are characterized by larger surface area and require shorter time for degradation and drug diffusion to particle surface [11]. Drug release can be studied by different techniques such as dialysis membranes (flat or bag) and Franz diffusion cells. Release kinetics is affected by in vitro release conditions: sink or non-sink conditions, release medium, dilution of nanoparticles dispersion that has to
Drug enriched shell
Drug enriched core
be considered in particular if the formulation is intended for oral and intravenous administration; release experiments have to be performed mimicking in vivo conditions in order to avoid distorted release profile [11]. As an example of in vitro study [9], drug release was investigated, working in sink conditions, for three types of SLN incorporating different percentage of tobramycin (1.25%, 2.50%, and 5.00%) by using a multicells rotating tool, where donor and acceptor compartments were separated by double hydrophilic/ hydrophobic membranes: results showed that release kinetic was pseudo-zero order for all three types of SLN and the amount of drug released was higher for Tobra-SLN 5.00% than for Tobra-SLN 2.50% which was higher than for Tobra-SLN 1.25%, showing that in vitro release can be modified by changing amount of drug loaded into SLN and consequently their physical characteristics (Fig. 6) [9].
Routes of Administration/Applications SLN have been tested for drug delivery applications by different routes of administration including intravenous, oral, ocular, topical dermal, transdermal, and pulmonary. Parenteral Administration: Intravenous Route (IV) Overview The evaluation of SLN interaction with blood and other tissues is very important for IV administration: many studies have been performed to assess toxicity of SLN both in in vitro and in vivo tests. Cellular toxicity of SLN has been investigated using several cell lines: in vitro experiments demonstrate that toxicity is dependent on composition of SLN (nature of lipid matrix and type of surfactant used) and on concentration of SLN in the culture medium [13].
S
Tobramycin release (%)
S
2480 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
Solid Lipid Nanoparticles - SLN
SLN 5.00 (5%) SLN 2.50 (2,5%) SLN 1.25 (1,25%)
0
15
30
45
60
90
105
120
Time (minutes)
Solid Lipid Nanoparticles - SLN, Fig. 6 In vitro release – percentage of tobramycin diffused through a double membrane versus time for SLN loaded at three different percentages of tobramycin [9]
In vivo toxicity studies confirmed that SLN are well tolerated in living system after their iv administration [13]. In order to obtain SLN suitable for parenteral administration, pharmaceutically acceptable excipients must be employed, and sterility must be assessed. Studies showed some SLN formulations can be sterilized by sterilizing filtration (such filtration is applicable for particles with size 1 mm) can be deposited by casting too. However, in such layers a significant solvent concentration gradient was observed to remain after the subsequent softbake step. This tends to weaken fine structures. It was observed that depositing and softbaking several layers on top of each other does not alleviate this significantly, as solvent from an upper layer tends to diffuse into the dry baked lower layers. A better solution is to put a known mass of pre-dried SU-8 flakes on the wafer surface, and melt it into a smooth layer [2]. Softbake The purpose of the softbake is to lower the solvent concentration, preventing adhesion to a contact mask and improving the lithographic performance, that is, the ability to produce high aspect ratio structures with straight sidewalls, by reducing the diffusion of the photoinitiator. The softbake temperature must be below 120 C, which is the onset of thermally induced cross-linking. A too high solvent content causes the photoacid generated during exposure to diffuse easily to unexposed regions, creating bulges or protrusions in the structure. A too low solvent concentration can cause cracks and peeling off of structures. It is, therefore, advisable to tune softbake time and temperature to reach a certain solvent concentration. In thick layers, the solvent concentration can be easily monitored by weighing the wafer on a balance with milligram precision. An advised solvent concentration is 5–7%. An even lower solvent content decreases cross-linking [18] and increases stress deformations in high aspect ratio features [5]. After softbaking, a relaxation period of several hours is advised. Exposure Typically, SU-8 is exposed using contact exposure on a mask aligner equipped with a standard mercury arc UV lamp. During exposure, the photoinitiator’s
SU-8 Photoresist SU-8 Photoresist, Table 3 Absorbance (inverse of penetration depth) of unexposed SU-8 for typical wavelengths [24] Wavelength (nm) 313 334 365 405 436
Absorbance (mm1) 1.19 101 5.85 102 2.71 103 1.41 104 6.90 105
triarylsulfonium hexafluoroantimonate salts are split, releasing a Lewis acid [16]. The acid generated serves as a catalyst, allowing a two-step reaction to occur that links the epoxy groups of different SU-8 monomers together. In practice, the latter two-step reaction occurs only very slowly at room temperature. To speed up the reaction rate, the SU-8 is heated up after exposure, during the post-exposure bake (PEB) step. For any practical application, both the exposure dose and the wavelength of the light used are important. From the critical dose up, parts that are unsolvable in developer will be formed. The critical dose for i-line (365 nm) exposure was determined to be 30 0.5 mJ/cm2 [20], for a PEB of 60 at 65 C and 30 at 95 C. This dose was observed to be by far insufficient for a good lithographic result: The adhesion of the structures formed to the substrate becomes only sufficiently strong to survive development at a much higher exposure dose. A second problem of a low dose is the permanent deformation of vulnerable and weakly cross-linked structures caused by internal stress during the development step. On the other hand, it is not possible to increase the exposure dose indefinitely as two problems arise with overexposure: the broadening of the features compared to their designed sizes and the appearance of effects caused by stress in larger parts. Careful choice of process parameters is therefore required. A typical i-line exposure dose is 200 mJ/cm2. Furthermore, the choice of exposure wavelength is important. Some UV wavelengths produced by a standard pressurized mercury arc lamp are absorbed too quickly and are not suitable to exposure of relatively thick SU-8 evenly (Table 3: absorbance). The transmission of light down to a certain depth is illustrated in Fig. 2. This shows that for even exposure of a layer thicker than a few microns, the 313-nm line should always be filtered out. For layers thicker than a few hundred microns, filtering out all light below 400 nm is recommended. As the photoinitiator is less
SU-8 Photoresist
S
1 λ=313 nm λ=334 nm
0.8 Transmission coefficient
SU-8 Photoresist, Fig. 4 Transmission of typical UV lines to the bottom of layer versus thickness of that layer, assuming no reflection of light and using the absorbances quoted in Table 3. transmission coefficient = eabsorbance thickness
2535
λ=365 nm λ=405 nm
0.6
0.4
0.2
0 100
101
102
103
Thickness [μm]
sensitive for 405-nm light, the recommended exposure dose is higher. A dose of 10 J/cm2 suffices for most applications. When exposing layers of irregular thickness, the use of an index matching fluid such as glycerin to fill up the air gap between mask and wafer can be beneficial to reduce diffraction effects and increase resolution (Fig. 4). Post-exposure Bake (PEB) The baking step following the exposure of the resist causes a cross-linking reaction in the exposed parts, rendering those parts insoluble during the subsequent development step. Meanwhile, a few side effects occur. These effects are shrinking of the resist by the cross-linking reaction, the thermal expansion of the resist during baking, and the (dissimilar) thermal expansion of the substrate during baking. The situation is clearly different for the in-plane and the out-of-plane direction. In the out-of-plane direction, the SU-8 layer is unconstrained and shrinking can occur freely. A shrinkage of approximately 4.5% (std 0.25%) was measured over a PEB of 1 h at 65 C [5]. In-plane, although some relaxation occurs, an internal stress is built up (4 MPa for a PEB temperature of 95 C in a standard process [5]). The combination of exposure energy and PEB time and temperature must cater for enough cross-linking for the structures to resist delamination and deformation during the subsequent development step. Typically,
post-exposure baking raises the glass transition temperature, that is, the temperature above which plastic deformation is possible, to a value slightly larger than the PEB temperature. After a PEB of 20 min, the glass transition temperature lies exactly at the PEB temperature. After that, saturation sets in quickly [10]. Therefore, it makes sense to use a minimum PEB time of 20 min. Typical PEB temperatures are between 80 and 100 C. After PEB, there is some stress relaxation over time. Hammacher et al. [11] notice a 7.5% drop in stress during the first 8 h after the PEB. After 4 days, there is an additional drop of 2%. This indicates that it is beneficial to build a waiting period between PEB and development, diminishing adhesion loss and deformations caused by stress. Also, a ramp up and slow ramp down of the PEB temperature are beneficial. For example, a cool down step of four hours in an oven can be employed [21]. Development During development, the non-cross-linked parts of the SU-8 layer are removed in a solvent. Typically, propylene glycol methyl ether acetate (PGMEA) is employed, though ethyl lactate or diacetone alcohol can be used as healthier alternatives with lower vapor pressure [13]. Exposed structures will, even after postexposure baking, adsorb solvents and thus swell during development. Structures in the 10 mm range are typically saturated within a minute [18]. As the solvent evaporates again after development is over, this should not be a problem. However, swollen structures that are
S
S
2536
not cross-linked hard enough can become permanently deformed during development. Development is dependent on agitation and on the design of the structures fabricated, and development rates as well as development times are therefore to be taken with caution. A development time of 8 min for a 50-mm-thick layer without agitation is typically sufficient. Ultrasonic agitation was found to increase the development rate by approximately eightfold. However, it can damage fragile structures. Megasonic actuation does not cause this problem. Another option is developing the wafer in an upside down position, which was found to decrease the development time by a factor 3.5 without inducing extra damage [5]. Hardbake The hardbake is an optional extra step at the end of the SU-8 process. During the hardbake, the material is heated well above the threshold for thermally induced cross-linking. This causes, also in regions where no photoacid is present, additional cross-linking reactions to occur until saturation. Hardbaking is done for increasing the long-term stability of the material and to close small cracks that arise in some process sequences. The additional cross-linking causes a higher chemical resistance as well: it was found impossible to remove hardbaked SU-8 structures from their substrates with solvents such as acetone or NMP (N-methyl-2pyrrolidone). For this reason, a hardbake is often not advised when the SU-8 layer will have to be removed later in the fabrication process. For the hardbake, typically a temperature between 150 C and 250 C is used. Following the observations of Hammacher et al. on the PEB, it can be assumed that most of the extra cross-linking is complete after 30 min. Hardbaking at 150 C for 60 min increases internal stress to 12 MPa (tensile) in 45% humidity air [18]. In the latter work, the built-in stress was shown to be strongly dependent on the humidity in the environment.
Removal of SU-8 The downside of the high chemical resistance of SU-8 is the difficulty to remove the material if needed. Nonhardbaked SU-8 can still be cracked up in strong solvents such as warm NMP. When this step is followed by rinsing or ultrasonification to remove the remaining
SU-8 Photoresist
flakes, a surface having a not too high topography can be clean again. SU-8 can be cleanly removed in piranha acid, though this is aggressive toward many metals and polymers that may already be present on the wafer. Silicon and glass are not attacked at all. Also, platinum, tantalum and gold are inert to piranha. Chromium is etched only very slowly, allowing the cleaning of contact masks in piranha after contamination with SU-8. Another application of piranha is for cleaning SU-8 off wafers for reuse. Reactive ion etching (RIE) in oxygen plasma with a small fluorine content is used as well, though traces of a nonvolatile antimony containing residue may be formed. Typically, 4–5% of fluorine-containing gas such as SF6 or CF4 is added to the reactor atmosphere to boost the etch rate significantly. Other less widely used options are ashing in an air or oxygen atmosphere and molten salt removal. These need a rather high temperature, 450–500 C and 350 C, respectively. Finally, one can consider downstream chemical etching (DCE). DCE is an organic removal technique in which reactants produced in a plasma are blown on the wafer, which is not exposed to the plasma itself. With a wafer temperature of 225 C and a gas composition of 98% oxygen and 2% CF4,, an etch rate of 6.8 mm/min can be obtained [9]. Given the reasonable working temperature, the etch speed, and the low metal etch rate, DCE is likely the most promising method for industrial use.
Alternative Exposure Methods Next to classic UV lithography, a number of other methods to selectively cross-link SU-8 exist. Though much less commonly used, they open some extra possibilities and therefore, deserve consideration. A distinction can be made between direct write methods in which structures are written pixel by pixel or voxel by voxel, and wafer-scale methods using masks which are generally much faster as writing is done in parallel over a large surface. The photoinitiator in SU-8 can be activated by the simultaneous absorption of two 800-nm photons, instead of by a single 400-nm photon [17]. The probability of such activation rises with the square of the
SU-8 Photoresist
2537
S
SU-8 Photoresist, Fig. 5 Microsieve structure in microchannel, fabricated in SU-8 using inclined UV exposure ([14], used with permission) Left: overview. Right: detail of sieve
intensity of the exposure. This is the basis for so-called two-photon lithography. The nonlinear dependence makes it possible, by focusing a laser to create a high-intensity spot that is small with respect to the layer thickness, to write true 3D structures such as photonic crystals directly in a single SU-8 layer. Proton beam writing is another direct write methods found in literature. The penetration depth of the exposing proton beam is dependent on its energy. Thus, bridge-like structures can be written in a single SU-8 layer, but not the more complicated structures that are possible with two-proton lithography. X-ray exposure has been tested as well. Becnel et al. [22] used SU-8 as a more sensitive alternative for the PMMA normally used as photoresist in LIGA processing. The exposure time needed was in the order of 10 min, two orders of magnitude smaller than for PMMA. The very limited diffraction and low absorption of the X-rays allows for a higher resolution in thick layers than that can be offered by UV-based processing, and aspect ratios greater than 100. Finally, inclined light UV exposure is worth mentioning. Some researchers have experimented with exposing SU-8 with inclined UV light through a contact mask. Thus, cylinders and beams with inclined orientation can be realized. Unless countermeasures are taken, the UV light will reflect from the substrate and a mirror image of the structure will be created by the light reflecting back from the substrate surface. At intersection of two cylinders a ball-shaped structure forms that can be used as an in-plane microlens. Others use the technique to construct microsieves (Fig. 5). Due to the high refractive index of SU-8 it is hard to attain large inclinations as explained by Snell’s law: Light entering at an oblique angle from the air will be bent toward the perpendicular axis of the higher refractive index material it enters. A solution for this is to immerse wafer and mask in
a higher refractive index fluid such as glycerin. When X-rays are used instead of UV, this is not important as refractive indices do not vary much.
Multilayered, Freestanding Structures in SU-8 A few methods that can be used to fabricated freestanding structures such as cantilevers and microchannels out of SU-8 will be discussed here. Modulation of Exposure Light The most straightforward method is to modulate the exposure energy, that is, to use a high UV dose (i-line) to expose a single SU-8 layer from top to bottom where the anchors of the freestanding structures must come and a lower dose to define the freestanding structures (Fig. 6). However, as SU-8 is quite transparent the resulting layer thickness is very sensitive to small variations in the exposure energy and other process parameters. Also, the absorbance of SU-8 tends to increase during exposure. Tight control of all the different parameters is required which makes it hard to achieve good tolerances. In practice, the method is only useful for fabricating thick freestanding structures with reasonable accuracy in relatively thick layers (>250 mm). Using an antireflective layer or a low-reflectivity substrate reduces the sensitivity of the process for exposure energy variations [6]. The minimum layer thickness is then reduced to 100 mm. The above process can be improved by using different wavelengths in the two UV exposure steps. By using a wavelength that is quickly absorbed for the second exposure step, thinner freestanding structures can be made with increased repeatability. It is straightforward to use the 313-nm wavelength for this as it is
S
S
2538
SU-8 Photoresist
SU-8 Photoresist, Fig. 6 Principle of the creation of freestanding structures by exposure time control
Spin SU-8, softbake
Substrate
365 nm mask 2 high dose
365 nm mask 2 low dose
Postbake, develop
Top layer thickness [μm]
20
15
10
5
0
0
50 100 150 313 nm exposure energy [mJ/cm2]
SU-8 Photoresist, Fig. 7 UV exposure energy at 313-nm wavelength versus resulting freestanding layer thickness showing LSE fit. Data points from measurement with (tiny) error bars showing standard deviation are plotted as well [5]
already present in the spectrum of standard exposure tools. The practical thickness range that can be reached this way is between 6 and 25 mm (Fig. 7). Due to uneven exposure between the top and bottom parts of the freestanding structure, single-clamped beams were observed to bend down [5]. Buried Mask Process In this process two SU-8 layers are used, as illustrated in Fig. 8. Thus, a larger thickness range and a better
control of layer thicknesses are obtained at the cost of a higher process complexity. First, the lower layer is processed as usual up to the PEB step. Then, a UV blocking layer is deposited. The purpose of this layer is to shield unexposed parts of the lower layer from the UV light used to expose the upper layer, later in the process. Then, the upper layer is spun on and processed. The two layers are developed together at the end of the process sequence. Halfway in the development step, the UV blocking layer must be removed when it is not dissolving in the normal SU-8 developer, for example, when a metal UV blocking layer is used. It is important to deposit the UV blocking layer such that no excess UV or heat is produced, which would cross-link SU-8 unwantedly. Also, temperature must be kept at a minimum when processing the second SU-8 layer to avoid wrinkling [5]. Figure 9 shows some example structures fabricated with the process. Evaporation by resistive heating of a low melting temperature metal such as aluminum, zinc, or magnesium is a suitable method. Another author uses a metal layer transfer process based on a gold-covered silicone stamp [8]. Sacrificial Layer–Based Processes A third major category of processes used to fabricate freestanding SU-8 structures is based on the deposition of a sacrificial layer in another material. After the SU-8 layer is processed on top of the sacrificial layer the sacrificial layer is dissolved, creating freestanding
SU-8 Photoresist SU-8 Photoresist, Fig. 8 Buried mask process
2539 Spin SU-8, softbake
S
Substrate
365 nm Mask 1
Evaporate AI
Spin Second Layer, Vacuum treatment
365 nm Mask 2
Postbake, develop AI etch, develop
S SU-8 Photoresist, Fig. 9 Single- and doubleclamped freestanding beams, fabricated with the buried mask method, lying over a KOH-etched U-groove
SU-8 structures. The sacrificial layer can be patterned before SU-8 deposition to create anchors defining where the SU-8 structures that will be formed in the next step will be attached to the substrate. When no anchor sites are defined either a well-timed etch
stop can define anchors or loose SU-8 components are created. The main property a sacrificial layer must have is that there must be a way to remove it without damaging SU-8. Reversely, the solvent in SU-8 must not dissolve
S
2540
the sacrificial layer to prevent layer intermixing and unpredictable results. Another point of interest is the adhesion of SU-8 on the sacrificial layer, which must be sufficient to allow it to survive the processing. The other properties that discern the different sacrificial layers discussed here are their layer thickness range, processing cost and time, and achievable aspect ratio. A metal sacrificial layer is a first option. Many metals common to micromachining, such as aluminum and copper can be etched chemically with perfect selectivity with respect to SU-8. Also, the adhesion of SU-8 to metals is generally satisfying (Table 2). There are some disadvantages to the use of metals, though. The most straightforward way to apply a metal layer is by a thin film technique such as sputtering. Due to stress and relatively slow speed, these processes are limited to a thickness of a few microns. Furthermore, the attainable aspect ratio of the lower layer is determined by the etching process of the metal, and is typically around one unless reactive ion etching is used. Both disadvantages can be overcome at the cost of an extra process step by the use of a sacrificial layer that is electroplated in a photoresist mold. Still, when using a metal sacrificial layer, the etch process of the metal may pose constraints on other metal layers that might be present and attacked as well by the release etch. A final point is that metal deposition methods are never self-planarizing, limiting the use on uneven surfaces. A second candidate is polyimide sacrificial layers. Polyimides resist the solvent in SU-8 very well once properly cross-linked. However, their removal is a problem. Polydimethylglutarimide (PMGI), on the other hand, is etchable in alkaline solutions such as positive photoresist developer while still being resistant to solvents. Thus, it can be removed selectively with respect to cross-linked SU-8 and is easily patternable. Disadvantages are the limited lower layer thickness, and the fact that the weak alkaline solution will still etch aluminum and, after an incubation time in which native oxide is removed, even (slowly) silicon. In Ref. [5], comb actuator structures were fabricated based on this process (Fig. 10). Positive photoresists can also be considered as sacrificial layer. They are inexpensive, can be quickly applied, and are available in a thickness range from below 1 mm to about 100 mm. Therefore, they would be ideally suited as sacrificial layer for a wide range of applications. The main problem hindering their
SU-8 Photoresist
SU-8 Photoresist, Fig. 10 Comb actuator structure fabricated in SU-8 using a PMGI sacrificial layer
applicability as sacrificial layer for SU-8 is that they dissolve in the solvent (cyclopentanone or GBL) present in SU-8. There are two ways to prevent this. The first is to deposit a thin metal layer on top of the positive resist. The second one is to strongly hardbake the positive photoresist. However, in the latter case resist reflow typically occurs, which severely limits the attainable aspect ratio of the sacrificial layer. Other sacrificial polymers used by workers in the field are polystyrene (PS) and PMMA. Toluene can be used as a solvent for spincoating and release, as cured SU-8 is resistant to several hours of immersion in toluene. It is possible to use RIE to pattern the polymers or to include photosensitizers in the PMMA, such as is done for deep-UV photoresists. A final interesting type of sacrificial polymers is polymers that decompose into volatile components cleanly at a temperature SU-8 can withstand. Examples are polypropylene carbonate (PPC) and polyethylene carbonate (PEC) [12]. Dissolved in NMP (N-Methyl-2pyrrolidone), they can be deposited by spin coating. Structuring is possible using oxygen plasma. PEC starts to decompose thermally around 220 C in nitrogen at atmospheric pressure. For PPC this is around 240 C. The most remarkable property of this process is that the volatile decomposition products can diffuse throughout SU-8. Thus, the sacrificial layer can even be removed from closed cavities, and long channels can be cleared in
SU-8 Photoresist
2541
S
a time independent of channel length. For a 30-mm-thick SU-8 layer, the sacrificial layer removal process time is about 10 min. For very thick upper layers, this time might be considerably longer. Layer Transfer Processes A final major approach to fabricating freestanding structures in SU-8 is to fabricate a single SU-8 layer on a carrier substrate and then transfer that layer to a second substrate on which another SU-8 layer is already present. This has the advantage that closed cavities or long channels can be fabricated. As the sealed cavities may contain released micromechanical structures, these processes can be used for capping purposes as well. The different variations of the processes can be further categorized according to the process step after which the layer transfer and optionally the removal of the carrier substrate is done (Fig. 11). When the transfer is done after softbaking, typically by applying limited heat and pressure, the material is still thermoplastic. Therefore, in order to achieve good control over the shape of the fabricated structure care must be taken to select process parameters such that little flowing of the non-cross-linked layer occurs. A typical sign of too much reflow is the clogging of channels that were sealed by the process and rounded corners. Another way to limit channel clogging is to use a micron-thin softbaked SU-8 layer to bond two substrates together. Bonding is done at 2 bars in a vacuum bonder [4]. However, the advantage that the upper layer can be readily patterned, for example, integrating vertical microfluidic connectors, is thereby lost. As the transferred SU-8 layer still must be crosslinked, either the carrier wafer must be UV transparent or it must be removed before exposure. After exposure and PEB, the carrier wafer must be removed in any case, unless the SU-8 layer is used as a simple adhesive. The removal is not straightforward with common hard substrates covered with a sacrificial material as this involves under etching of that sacrificial material for several centimeters. A better way is to use a foil as carrier, which can be laminated on with relatively low pressure, avoiding air entrapment and channel clogging. Furthermore, a foil can be removed by simply peeling it off. A polyimide foil should be peeled off before exposure as it is not UV transparent. Teflon foil, on the other hand, is transparent to UV.
SU-8 Photoresist, Fig. 11 Layer transfer processes. Top: cross-linking after transfer. Sometimes the carrier wafer is not removed, and SU-8 serves as an adhesive between the two substrates. Bottom: transfer of partially cross-linked layers
Due to the rather rough nature of the carrier wafer removal process it is best used when the top layer contains relatively large features. Examples are device capping and the sealing of microchannels. Deformation of the transferred layer can be avoided altogether by cross-linking it first before transferring it. The transferred layer can then also be developed before transfer. To make sure sufficient reactivity remains available to enable thermocompression bonding, the exposure energy and PEB must be scaled down. Arroyo et al. [1] determined the optimal parameters as 140 mJ/cm2 for i-line exposure followed by a PEB of 40 at 85 C. Due to the decreased deformability of the SU-8, a much higher bonding force is needed during thermocompression bonding: good results (95% yield) were achieved with a pressure 3.25 bar applied at a temperature of 88 C for 12 min.
S
S
2542
SU-8 Photoresist
A straightforward method to achieve conductive SU-8 structures is to sputter deposit a thin metal layer. As during sputtering material gets deposited on top of as well as on sidewalls of structures, the mask layout must be designed such as not to cause short circuits, for example, by incorporating overhanging structures such as illustrated in Fig. 12 [5].
Conclusion
SU-8 Photoresist, Fig. 12 Metallization of SU-8 by sputter deposition of a thin metal layer (red). Top: isolated freestanding structure. Middle: connection to substrate. Bottom: isolation of bonding pad (left)
Furthermore, it is of course essential to have good thickness uniformity. The edge bead must certainly be removed, be it mechanically, chemically, or by designing the mask such that the rim of the wafer is not exposed. In this case too, it can be advantageous to use a Teflon or polyimide foil as carrier for the second layer as it can be easily pulled off.
Electrically Active Structures in SU-8 In this paragraph, a few strategies will be discussed to render SU-8 conductive, which is mandatory for many actuation and sensing applications such as comb drives and capacitive accelerometers. Conductive polymers such as polyaniline (PANI) can be added to the SU-8 resin. However, it was shown that high PANI quantities (up to 50%) are required in order to achieve useful conductance, a point at which many SU-8 properties such as adhesion degrade. A different route to create conductive SU-8 is to use a silver nanoparticle-based filler. A silver content of 6 vol% is enough to reach to the percolation threshold, and to create a conductivity of over 10 S cm1, which can go up a factor 100 with a higher silver content [23]. Such blends are commercially available. However, because of light reflected by the filler, the available aspect ratio and achievable thickness are limited to about 1 mm and 10 mm, respectively. Carbon nanotubes have been explored as filler material as well. However, conductivities are typically quite limited.
Concluding this oversight of SU-8 photoepoxy processing, there is little doubt that this material will find – and indeed already has found – its way into a wide range of applications that benefit from the unique combination of low cost wafer-scale fabrication and the high aspect ratios and very high layer thicknesses attainable. Examples of those are the fabrication of microchannels, labs-on-a-chip, integrated optics, ink jet heads, and electroplating molds for fabricating metal structures. For the latter, electroplating in SU-8 is a low cost alternative for the LIGA process, provided minimum features are several microns wide and more relaxed tolerances can be acceptable. As SU-8 is a low-temperature process, it even shows potential for the fabrication of sensors directly on CMOS wafers.
Cross-References ▶ Lab-on-a-Chip for Studies in C. elegans ▶ Microfludic Whole-cell Biosensor ▶ Plating ▶ Stereolithography
References 1. Arroyo, M.T., Ferna´ndez, L.J., Agirregabiria, M., Iban˜ez, N., Aurrekoetxea, J., Blanco, F.J.: Novel allpolymer microfluidic devices monolithically integrated within metallic electrodes for SDS-CGE of proteins. J. Micromech. Microeng. 17, 1289–1298 (2007) 2. Becnel, C., Desta, Y., Kelly, K.: Ultra-deep x-ray lithography of densely packed SU-8 features: I. An SU-8 casting procedure to obtain uniform solvent content with accompanying experimental results. J. Micromech. Microeng. 15, 1242–1248 (2005) 3. Bohl, B., Steger, R., Zengerle, R., Koltav, P.: Multilayer SU-8 lift-off technology for microfluidic devices. J. Micromech. Microeng. 15, 1125–1130 (2005)
Sub-wavelength Waveguiding 4. Carlier, J., Arscott, S., Thomy, V., Fourrier, J.C., Caron, F., Camart, J.C., Druon, C., Tabourier, P.: Integrated microfluidics based on multi-layered SU-8 for mass spectrometry analysis. J. Micromech. Microeng. 14, 619–624 (2004) 5. Ceyssens, F.: Micromachining in polymers and glass: process development and applications. PhD thesis, KULeuven, Belgium (2009) 6. Chuang, Y., Tseng, F., Cheng, J., Lin, W.: A novel fabrication method of embedded micro-channels by using SU-8 thick-film photoresists. Sensor Actuat A-Phys 103, 64–69 (2003) 7. Chung, C.K., Hong, Y.Z.: Surface modification of SU8 photoresist for shrinkage improvement in a monolithic MEMS microstructure. J. Micromech. Microeng. 17, 207–212 (2007) 8. del Campo, A., Greiner, C.: SU-8: a photoresist for highaspect-ratio and 3D submicron lithography. J. Micromech. Microeng. 17, R81–R95 (2007) 9. Dentinger, P.M., Miles, C., Goods, S.H.: Removal of SU-8 photoresist for thick film applications. Microelectron.Eng. 61–62, 993–1000 (2002) 10. Feng, R., Farris, R.: Influence of processing conditions on the thermal and mechanical properties of SU8 negative photoresist coatings. J. Micromech. Microeng. 13, 80–88 (2003) 11. Hammacher, J., Fuelle, A., Flaemig, J., Saupe, J., Loechel, B., Grimm, J.: Stress engineering and mechanical properties of SU-8-layers for mechanical applications. J. Microsyst. Technol. 14, 1515–1523 (2007) 12. Metz, S., Jiguet, S., Bertsch, A., Renaud, Ph: Polyimide and SU-8 microfluidic devices manufactured by heatdepolymerizable sacrificial material technique. Lab Chip 4, 114–120 (2004) 13. MicroChem: SU-8 datasheet and adhesion results-shear analysis. www.microchem.com. Accessed 18 Nov 2011. (2007) 14. Sato, H., Matsumura, H., Keino, S., Shoji, S.: An all SU-8 microfluidic chip with built-in 3D fine microstructures. J. Micromech. Microeng. 16, 2318–2322 (2006) 15. Schoeberle, B., Wendlandt, M., Hierold, C.: Long-term creep behavior of SU-8 membranes: application of the time-stress superposition principle to determine the master creep compliance curve. Sensor Actuat A-Phys 142, 242–249 (2008) 16. Teh, W.H., D€ urig, U., Drechsler, U., Smith, C.G., Gntherodt, H.-J.: Effect of low numerical-aperture femtosecond two-photon absorption on SU-8 resist for ultrahigh-aspect-ratio microstereolithography. J. Appl. Phys. 97, 4095 (2005) 17. Witzgall, G., Vrijen, R., Yablonovitch, E., Doan, V., Schwartz, B.J.: Single-shot two-photon exposure of commercial photoresist for the production of threedimensional structures. Opt. Lett. 23, 1745 (1998) 18. Wouters, K., Robert, P.R.: Diffusing and swelling in SU-8: insight in material properties and processing. J. Micromech. Microeng. 20, 095013 (2010) 19. Yang, R., Wang, W.: A numerical and experimental study on gap compensation and wavelength selection in UV-lithography of ultra-high aspect ratio SU-8 microstructures. Sensor Actuat B 110, 279–288 (2005)
2543
S
20. Zhang, J., Tan, K.L., Hong, G.D., Yang, L.J., Gong, H.Q.: Polymerization optimization of SU-8 photoresist and its applications in microfluidic systems and MEMS. J. Micromech. Microeng. 11, 20–26 (2002) 21. Williams, J.D., Wang, W.: Using megasonic development of SU-8 to yield ultrahigh aspect ratio microstructures with UV lithography. J. Microsyst. Technol. 10, 694–698 (2004) 22. Becnel, C., Desta, Y., Kelly, K.: Ultra-deep x-ray lithography of densely packed SU-8 features: II. Process performance as a function of dose, feature height and post exposure bake temperature. J. Micromech. Microeng. 15, 1249–1259 (2005) 23. Jiguet, S., Bertsch, A., Hofmann, H., Renaud, P.: Conductive SU8-Silver composite photopolymer. Proceedings of Micro Electro Mechanical Systems, 125–128 (2004) 24. Reznikova, E.F., Mohr, J., Hein, H.: Deep photo-lithography characterization of SU-8 resist layers. J. Microsyst. Technol. 11, 282–291 (2005)
Submicron Structures in Nature ▶ Biomimetics of Optical Nanostructures
Sub-retinal Implant ▶ Artificial Retina: Focus on Clinical and Fabrication Considerations
Subwavelength Antireflective Surfaces Arrays ▶ Moth-Eye Antireflective Structures
Subwavelength Structures ▶ Moth-Eye Antireflective Structures
Sub-wavelength Waveguiding ▶ Light Localization for Nano-optical Devices
S
S
2544
Superelasticity and the Shape Memory Effect Xiaodong Han1, Shengcheng Mao1 and Ze Zhang1,2 1 Institute of Microstructure and Property of Advanced Materials, Beijing University of Technology, Beijing, People’s Republic of China 2 State Key Laboratory of Silicon Materials and Department of Materials Science and Engineering, Zhejiang University, Hangzhou, China
Synonyms Hyperelasticity; Pseudoelasticity; Theoretical elasticity; Ultralarge strain elasticity
Definition Superelasticity, or pseudoelasticity, is a unique property of shape memory alloys (SMAs), wherein up to 13% deformation strain can be sustained and the material can recover its original shape after removing the stress. The shape memory effect occurs in SMA and is defined as when a material can remember its original shape upon heating or cooling.
Overview Superelasticity (SE), also called pseudoelasticity, normally refers to a phenomenon observed in shape memory alloys (SMAs), a kind of metal that can remember its original shape after heating or cooling [1–3]. SE is different from true elasticity in bulk metals, which is a reflection of the interatomic spacing variation within a material (Hooke’s law). Regular elasticity in bulk metals is normally less than 0.5%, whereas the elasticity of shape memory alloys can be up to 13% in Fe-based SMA [4]. This SE occurs through reversible martensitic transformations from austenite to martensite and vice versa. Upon external mechanical loading or temperature change (or a mixture of the two), the SMA deforms by
Superelasticity and the Shape Memory Effect
reversible martensitic phase transformations rather than by irreversible plastic dislocation glide. The martensitic transformation occurs through shear on an invariable plane (habit plane) along the shear direction. SE occurs when the metals are deformed in the austenite state, and the deformation shape is recoverable when the applied load is removed. Large strain elasticity (LSE) can be achieved in nanomaterials due to their defect-free structure, particularly in nanowires. The true elasticity of metallic nanowires can even approach the theoretical elastic limit (8%) [5], and this is typically associated to extremely high strength [6]. In theory and in computer simulations, some metallic nanowires, such as Cu nanowires, can reversibly deform up to 50% by a combination of large strain elasticity and pseudoelasticity [7]. Some of these exceptional mechanical properties have already been experimentally demonstrated in metallic nanowires. The shape memory effect (SME) also relates to SMAs, and it refers to the phenomenon where a metallic specimen changes shape in response to temperature changes [1–3]. Figure 1 shows a sketch comparing SME and SE. A deformed SMA can return to its original high-temperature shape upon heating or go back to the low-temperature shape upon cooling (see Fig. 1a). If a SMA is deformed by mechanical loading at a constant temperature higher than a critical temperature (the austenite finish temperature, defined below), the initial shape can be recovered by unloading, through a reverse martensitic transformation (Fig. 1b). The SME was first discovered by A. Olander in Au-Cd alloy in 1932. Investigations into SMA and their applications have been further promoted since the discovery of the SME in the near-equiatomic NiTi alloys, known as “Nitinol.” The SME has also been found in other metallic alloys such as Cu-Al-Ni, Cu-Zn, Fe-Mn-Si, Ni-Mn-Ga, and Ti-Ni-Hf. Because of these unique properties, SMAs have been used for a variety of applications, including orthodontic wire, stents, microactuators, pipe coupling, eyeglass frames, microelectromechanical systems, and a variety of biomedical devices [8, 9]. Superelasticity and the shape memory effect are correlated with thermoelastic and stress/strain-induced martensitic transformations. In the following, we introduce some basic concepts of martensitic transformations in NiTi SMAs.
Superelasticity and the Shape Memory Effect
a
b T>Af
Af Ms
F
Thermodynamic Consideration of the Thermoelastic Martensitic Transformation The martensitic transformation is a transformation between austenite (stable at high temperature) and martensite (stable at low temperature), and it is usually treated as a diffusionless first-order phase transformation. The Gibbs free energy, defined as DG ¼ DH TDS (where DH and TDS are the changes in chemical enthalpy and entropy), is the pertinent chemical potential of a system, which is at the minimum when reaches equilibrium at constant pressure and temperature. The martensite and austenite Gibbs free energies (DGM and DGA ) gradually decrease, following different slopes, with increasing temperature (T), as illustrated in Fig. 2a. At temperature To, the Gibbs free energies of the two phases reach a thermodynamic equilibrium (DGM ¼ DGA ). To drive the forward or reverse martensitic transformation, the energy gap between the two phases (DGA!M or DGM!A ) needs to be overcome by either temperature or mechanical loading, as expressed in Eqs. 1 and 2: [10].
DGM!A ¼ DH M!A TDSM!A þ
se r
di oa
g F
Martensitic Transformation
DGA!M ¼ DH A!M TDSA!M DEmech se ¼ DHA!M TDSA!M r
ng
F
in
oli
Mf
ad
g
tin
a He
Co
ng
Lo
Mf
nl
Af As
(1)
(2)
where r, s, and e are the density of materials, the external stress, and strain, respectively.
F
U
Superelasticity and the Shape Memory Effect, Fig. 1 A schematic illustration of (a) the shape memory effect and (b) superelasticity
S
2545
Figure 2b shows schematically the evolution of the martensite volume fraction with temperature during forward and reverse transformations. There are four characteristic temperatures defining a thermoelastic martensitic transformation: (1) The martensite start temperature, Ms, at which martensite starts to nucleate; (2) the martensite finish temperature, Mf, with Mf < Ms, at which the transformation has completed; (3) the austenite start temperature, As, at which the system starts moving from martensite to austenite upon heating; and (4) the austenite finish temperature, Af, with Af > As, at which the martensitic phase has fully transformed into the austenite phase. Aside from the thermally induced martensitic transformation, martensite can also be obtained by applying external stress. Figure 3a shows a typical stress–strain curve of NiTi SMA showing superelasticity. There are two stress plateaus, defined as the upper stress plateau and the lower stress plateau, corresponding to the forward and reverse martensitic transformation, as indicated in Fig. 3a. The martensite starts to nucleate after reaching a critical transformation stress, sc . The martensitic transformation then proceeds at nearly constant stress upon further straining (the physical mechanisms are described below). As shown in Fig. 3a, the upper plateau stress is lower than the critical transformation stress. As a result, the growth energy is smaller than that of the nucleation energy of martensite. The austenite nearly fully transforms to martensite at the end of the upper stress plateau. Upon unloading to a critical reverse stress, sr , martensite starts transforming back to austenite. The martensite phase is nearly fully reverted to the austenite phase at the end of the lower stress plateau. According to Eq. 1, higher stress is required to trigger
S
S
2546
Superelasticity and the Shape Memory Effect
Superelasticity and the Shape Memory Effect, Fig. 2 Schematic illustration of the evolutions of (a) the Gibbs free energies of the martensite and austenite phases and (b) the martensitic fraction with temperature
Superelasticity and the Shape Memory Effect, Fig. 3 (a) A typical stress– strain curve of NiTi SMA showing superelasticity. (b) Schematic illustration of the stress–temperature relationship of an SMA. The stress can be both tensile and compressive stress
martensite transformations as the ambient temperature increases. The stress–temperature relationship in stressinduced martensitic transformations is schematically illustrated in Fig. 3b. As shown in Figs. 2b and 3a, the forward and reverse deformation paths do not overlap, i.e., temperature and stress hysteresis exist during thermo- and stress-induced martensitic transformation. The austenite start temperature is higher than the martensite start temperature (As > Ms, As > To and Ms < To), and the unloading stress plateau is lower than the loading stress plateau. This is because the generation of irreversible energies (DEir ), including the frictional energy of the martensite–austenite phase interface propagation, the acoustic emission, and the production and motion of dislocations, is ineluctable during the martensitic transformation [11]. Therefore, additional chemical or mechanical energy is required to promote the forward and reverse martensitic transformations.
Crystallography of Stress-Induced Martensitic Transformation Critical Transformation Stress Two important characteristics, the critical transformation stress and transformation strain, define the stressinduced martensitic transformation. It has been suggested that the selection of a martensitic variant under external stress (uniaxial tension and compression) comes from satisfying Schmid’s law, as illustrated in Fig. 4 [12]. The stress component along the shear direction in the habit plane can be expressed as: ts ¼
Fs F cos l F ¼ ¼ cos l cos ’ As Ao = cos ’ Ao
¼ s cos l cos ’
(3)
where m ¼ cos l cos ’ is defined as the Schmid factor, l is the angle between the loading axis and the
Superelasticity and the Shape Memory Effect
2547
F
Ao Normal to habit plane
Shear direction λ
ϕ
τ
corresponds to LCMV #1 with lattice correspondences of [001]m–[001]a, [010]m–[011]a, and [001]m–[0–11]a. The lattice deformation matrix M0 in the coordinates of the martensite (i0 , j0 , k0 ) can be expressed using the lattice constants of the parent and the product phases. For the martensitic transformation in nearequiatomic NiTi, the lattice constant of the austenite is ao ¼ 0.3015 nm and those of the martensite are am ¼ 0.2889 nm, bm ¼ 0.412 nm, cm ¼ 0.4622 nm, and b ¼ 96:8o [14]; thus: 2
A
am ao
0 b pffiffim
cmpcos ffiffi b 2ao
3
6 7 0 7 M0 ¼ 6 40 5 2ao cp m sin ffiffi b 0 0 2ao 2 3 0:9582 0 0:1283 ¼4 0 0:9663 0 5 0 0 1:0763 F
Superelasticity and the Shape Memory Fig. 4 Schematic illustration of Schmid’s law
Effect,
shear direction, and ’ is the angle between the loading axis and the normal of the habit plane. According to Eq. 3, the variant with the highest Schmid factor and, consequently, the highest resolved shear stress will be triggered in one grain, which shows that the critical transformation stress of martensite is strongly related to the grain/crystal orientation. Maximum Recoverable Strain The two phases of NiTi alloys, austenite and martensite, are known to have cubic (B2) and monoclinic (B19’) structures. Figure 5 shows a schematic illustration of the lattice distortion of the B2–B19’ transformation in NiTi SMA. A martensitic unit cell is identified by the dashed line within the austenite lattice, as shown in Fig. 5a. The embryo may be transformed to the martensitic unit cell by a simple “bain distortion,” as indicated in Fig. 5b, which can be expressed as an expansion in c, contractions in the a and b directions, and a change of the beta angle from 90 to 96.8 . Such a transformation can happen in 12 equivalent lattice correspondence martensitic variants (LCMVs) in NiTi [13]. The lattice distortion shown in Fig. 5
S
(4)
The lattice deformation matrix M in the coordinates of austenite (i, j, k) can then be transformed from M0 with a coordinate transformation matrix R from the martensite to the austenite via: M ¼ RM0 RT
(5)
where RT is the transpose of R. Using the lattice correspondence of austenite and martensite, the coordinate transformation matrix R for the variant shown in Fig. 5 is expressed as: 2
3 1 0pffiffiffi 0pffiffiffi R ¼ 4 0 1=p2ffiffiffi 1=pffiffiffi2 5 0 1= 2 1= 2
(6)
The lattice deformation matrix M is then calculated to be: 2 3 a b b M ¼ RM0 RT ¼ 4 0 o g 5 (7) 0 g o where a ¼ 0:9582; b ¼ 0:0907; o ¼ 1:0213, and g ¼ 0:0550. With this, a vector x in the austenite is transformed to x0 upon the martensitic transformation by the following equation: x0 ¼ Mx
(8)
S
S
2548
Superelasticity and the Shape Memory Effect
a
i,i ′
i′
b
ao=0.3015nm
.8°
j
b =96
j′
j′ k′
c
m=
0.
46
22
nm
am=0.2889nm
12nm
.4 b m=0
k′
k
Superelasticity and the Shape Memory Effect, Fig. 5 Schematic illustration of the lattice distortion of the B2–B19’ martensitic transformation in NiTi: (a) atomic coordinate systems, with (i, j, k) representing the reference frame in the
austenite and (i0 , j0 , k0 ) representing the reference frame in the martensite, (b) lattice distortion from B2 (dashed lattice) to B19’ (solid lattice)
Consequently, the transformation strain can be calculated by:
A and B. The deformation process is referred to as detwinning or reorientation of the martensites. The twinned or detwinned martensitic variants can fully transform back to austenite when heating the specimen to a temperature above Af corresponding to the recovery of the macroscopic SMA to its original shape. There are two types of SME: one-way and two-way. The difference is whether the SMA remembers its low-temperature shape. The critical stress to drive the reorientation of martensite is known to be very small (about 100 MPa); therefore, it is very easy to set the shape of SMA at low temperatures (T < Mf). For the one-way SME, the shape of SMA will remain stable until heated above the reverse transformation start temperature (As) and reverse to its original when the temperature is higher than Af. Upon the second cooling to a temperature lower than Mf, no macroscopic shape change happens, indicating that only the hightemperature shape is “remembered.” The two-way SME is an effect where both the lowtemperature and high-temperature shapes can be “remembered.” Thus, when cooling the SMA to a temperature T < Mf, the high-temperature shape will change to the low-temperature shape; likewise, the low-temperature shape will transform back to the high-temperature shape on heating to T > Af. The shape variations between the two shapes are stress free. It is an intrinsic property of SMA to “remember” the high-temperature shape; however, additional training is required for SMA to “remember” the lowtemperature shape.
e¼
j x0 j j xj jxj
(9)
According to thermodynamic principles, the variant with the maximum strain in the direction of the applied load produces the maximum driving force for the stress-induced martensitic (SIM) transformation [15] and is triggered to form first.
Mechanisms of the Shape Memory Effect and Superelasticity The unique properties, superelasticity and the shape memory effect, of SMA can be expressed by a simplified 2D geometric depiction, as illustrated in Fig. 6. Upon cooling to a temperature T < Mf, two martensitic variants, defined as A and B, with the same crystal structure but different orientations were cooperatively triggered to accommodate and minimize the martensitic transformation strain. The interface between the thermally induced martensite, composed of the matrix and its twin, and the austenite is the habit plane. The martensite variant at the habit plane is then referred to as the habit plane variant. Under external stress (T < Mf), martensitic variant A grows at the expense of variant B by the movement of the interface of variants
Superelasticity and the Shape Memory Effect Superelasticity and the Shape Memory Effect, Fig. 6 Schematic illustration of the martensitic structure upon loading, unloading, heating, and cooling
2549
S
Austenite
Cooling
Loading/Loading+Cooling
Unloading/heating /Unloading+Heating A
Heating
B A B
Loading
A Twinned Martensite
In addition to the SME, SMAs have another unique property known as superelasticity, or pseudoelasticity, which occurs at temperatures above Af and distinguishes SMA from other metallic alloys. Under external mechanical loading, detwinned martensite nucleates in the austenite matrix after reaching the critical resolved shear stress. Martensite continuously propagates with increasing external strain in a L€uderstype manner under a nearly constant stress (see the upper stress plateau in Fig. 3). The nucleation of martensite is so concentrated that a region nearly fully transformed to martensite is the first to form in the SMA. There is a clear edge between the martensite and austenite, with a shear angle of about 55o to the loading axis [15]. The martensite band propagates by new nucleation of martensite ahead of the martensite– austenite edge [16]. The inhomogeneous martensitic transformation is similar to the stress-induced slip band in low-carbon steels. This transformation behavior was first reported by Guillaume Piobert and W. L€ uders in 1864, and the martensitic band is then referred to as a L€ uders-like deformation band (LBD). The stress-induced martensite is unstable at the testing temperature (higher than Af) and will transform back to austenite when the external stress is unloaded. The superelasticity of SMA is associated with the reverse martensitic transformation and recovery of the macroscopic deformation strain. Upon unloading to the lower stress plateau (see Fig. 3), the martensite
Detwinned Martensite
plates inside the LDB gradually transform back to austenite. At the macroscopic scale, the reverse transformation is characterized by shrinkage of the LDB.
Cross-References ▶ Nanomechanical Properties of Nanostructures ▶ Plasticity Theory at Small Scales ▶ Size-Dependent Plasticity of Single Crystalline Metallic Nanostructures ▶ Surface Tension Effects of Nanostructures
References 1. Otsuka, K., Wayman, C.M.: Mechanism of shape memory effect and superelasticity. In: Otsuka, K., Wayman, C.M. (eds) Shape memory materials, pp. 27–48. Cambridge, Cambridge University Press (1998) 2. Saburi, T., Nenno, S.: The shape memory effect and related phenomena. In: Proceedings of the International Conference on Solid-Solid Phase Transitions, pp. 1455–1479. Pittsburg (1981) 3. Otsuka, K., Ren, X.: Physical metallurgy of Ti–Ni-based shape memory alloys. Prog. Mater. Sci. 50, 511–678 (2005) 4. Tanaka, Y., Himuro, Y., Kainuma, R., Sutou, Y., Omori, T., Ishida, K.: Ferrous polycrystalline shape-memory alloy showing huge superelasticity. Science 327, 1488–1490 (2010) 5. Yue, Y.H., Liu, P., Zhang, Z., Han, X.D., Ma, E.: Approaching the theoretical elastic limit in Cu nanowires. Nano Lett. 11, 3151 (2011) 6. Wong, E.W., Sheehan, P.E., Lieber, C.M.: Nanobeam mechanics: elasticity, strength and toughness of nanorods and nanotubes. Science. 277, 1971 (1997)
S
S
2550
7. Park, H., Gall, S.K., Zimmerman, J.A.: Shape memory and pseudoelasticity in metal nanowires. Phys. Rev. Lett. 95, 255504 (2005) 8. Humbeeck, J.V.: Non-medical applications of shape memory alloys. Mater. Sci. Eng. A 273–275, 134–148 (1999) 9. Duerig, T., Pelton, A., Stӧckel, D.: An overview of nitinol medical applications. Mater. Sci. Eng. A 273–275, 149–160 (1999) 10. Wollants, P., Roos, J.R., Delaey, L.: Thermally- and stressinduced thermoelastic martensitic trans- formations in the reference frame of equilibrium thermodynamics. Prog. Mater. Sci. 37, 227–288 (1993) 11. Liu, Y., McCormick, P.G.: Thermodynamic analysis of the martensitic transformation in NiTi—I. Effect of heat treatment on transformation behaviour. Acta Metal. Mater. 42, 2401– 2406 (1994) 12. Gall, K., Sehitoglu, H.: The role of texture in tension– compression asymmetry in polycrystalline NiTi. Int. J. Plasticity 15, 69–92 (1999) 13. Matsumoto, O., Miyazaki, S., Otsuka, K., Tamura, H.: Crystallography of martensitic transformation in Ti-Ni single crystals. Acta Metall. 35, 2137–2144 (1987) 14. Wollants, P., Bonte, M.D., Roos, J.R.: A thermodynamic analysis of the stress-induced martensitic transformation in a single crystal. Z. Metallkd. 70, 113–117 (1979) 15. Shaw, J.A.: Thermomechanical simulations of localized thermo-mechanical behavior in a NiTi shape memory alloy. Int. J. Plast. 16, 541–562 (2000) 16. Mao, S.C., Luo, J.F., Zhang, Z., Wu, M.H., Liu, Y., Han, X. D.: EBSD studies of the stress-induced B2-B19’martensitic transformation in NiTi tubes under uniaxial tension and compression. Acta Mater. 58, 3357–3366 (2010)
Superhydrophobicity ▶ Lotus Effect
Superoleophobicity of Fish Scales Lei Jiang1 and Ling Lin2 1 Center of Molecular Sciences, Institute of Chemistry Chinese Academy of Sciences, Beijing, People’s Republic of China 2 Beijing National Laboratory for Molecular Sciences (BNLMS), Key Laboratory of Organic Solids, Institute of Chemistry Chinese Academy of Sciences, Beijing, People’s Republic of China
Synonyms Oil-repellency of fish scales
Superhydrophobicity
Definition Superoleophobicity of fish scales is a wetting phenomenon that fish scales show oil-repellency in water, with a static oil contact angle (CA) higher than 150 . Generally, underwater superoleophobicity is defined as a static oil CA higher than 150 on solid surface in an oil/water/solid three-phase system.
Chemical and Physical Principles Wettability is a fundamental property of solid surfaces, which not only affects the behavior of the creatures in nature, but also plays an important role in all aspects of our life. A direct expression of wetting behavior is a static CA of a liquid droplet sitting on a solid surface when the surface tensions at multiphase interface reach thermodynamic equilibrium. For an ideal flat surface in a liquid/gas/solid system, the CA (y) is given by the Young’s equation [1]: cos y ¼
gsg gsl glg
(1)
where gsg is the solid/gas interface tension, gsl is the solid/liquid interface tension, and glg is the liquid/gas interface tension. For a rough surface in the air atmosphere, the situation is more complex. The surface topographic structure has a great influence on the wettability. Two distinct models, Wenzel model and Cassie–Baxter model, are commonly used to explain the effect of roughness on the apparent CAs of liquid drops. As described by Wenzel’s model [2], liquid completely penetrates into the rough structures, leading the increase of contact area of solid/liquid interface. Therefore, the apparent CA (yw) can be described as: cos yw ¼ r cos y
(2)
where the surface roughness factor (r) is defined as the ratio of the actual contact area of solid/liquid interface to the projected contact area (r 1), and y is the intrinsic CA on the flat surface. As described by Cassie–Baxter’s model [3], liquid suspends on the rough surface rather than completely penetrates. The rough surface therefore can be considered as
Superoleophobicity of Fish Scales
a solid/gas composite surface, and the apparent CA (yc) can be described as: cos yc ¼ f cos y ð1 f Þ
S
2551
a
b
c
d
(3)
where f is the area fraction of solid/ liquid interface, ð1 f Þ is that of liquid/gas interface. Based on both theories, scientists can better understand different wetting phenomena in nature, which would help the design of artificial materials with functional surfaces [4–6].
500 μm
1cm
Key Research Findings Oil-Wetting Behaviors on Fish Scales with Micro/ Nanostructures Nature abounds with mysterious living organisms which have special surface properties. A famous example is the lotus leaf, a typical superhydrophobic surface in nature. When a water droplet falls on the lotus leaf, it keeps bead shape and rolls off the surface immediately taking away the adherent dirt particles. This so-called “lotus effect” has been understood as a result of the complementary roles of low surface free energy and micro/nanostructures on the leaf surface [8]. Inspired by this phenomenon, tremendous artificial superhydrophobic surfaces have been fabricated and have facilitated practical applications in broad fields [9–11]. In a similar manner to the lotus effect in the air atmosphere, fish can resist oil pollution in water, keeping itself antifouling. This fascinating phenomenon shows great potential for many applications in an oil/water/solid system, such as marine antifouling, prevention of oil spills, microfluidic technology, and bioadhesion [12, 13]. Liu et al. first reported the oil-wetting behavior on fish scales in the water environment [7]. It is common knowledge that fish scales are covered by a thin layer of mucus that leads to their hydrophilic nature. Liu et al. revealed that the hydrophilic surface of fish scales showed superoleophilicity in air (Fig. 1a) [14]. However, the surface of fish scales turned to be superoleophobic once it was immersed in water, with an oil CA larger than 150 (Fig. 1b). To figure out the reason of oil-wetting reversion, they firstly observed surface structures on fish scales in detail. Figure 1c–f display typical images of fish scales (Crucian Carp, Carassius
e
f
20 μm
2 μm
Superoleophobicity of Fish Scales, Fig. 1 (a) Fish scales show superoleophilicity in air (1,2-dichloroethane (DCE) as a detecting oil, density: 1.245 g cm3, surface tension at 25 C: 31.86 mN m1). (b) Fish scales become superoleophobic once it is immersed in water (droplet of DCE as in (a)). (c–f) SEM images of fish scales disclose the surface micro/nanostructures with the increase of magnification (Reproduced with permission. Copyright Wiley-VCH Verlag GmbH & Co. KGaA (2009))
carassius) using scanning electronic microscopy (SEM). The fan-shaped fish scales with diameters of 4–5 mm are densely arranged (Fig. 1c). The magnifying images disclose that there are oriented micropapillae on each fish scale. Each micropapillea is in length of 100–300 mm and in width of 30–40 mm (Fig. 1d). In high-magnification SEM images (Fig. 1e–f), nanoscale roughness is clearly observed on the surface of micropapillae. It was suggested that these hierarchical structures could trap water and form a composite interface on fish scales to resist oil, which might play an important role on the oil-wetting reversion. Mechanism of Wetting Behaviors on Fish Scales in Oil/Water/Solid System To better understand the reversion of oil-wetting behavior on fish scales, Liu et al. chose three kinds of designed surfaces of silicon wafers as the models: smooth surface, microstructured surface, and micro/nanostructured
S
S
2552
Superoleophobicity of Fish Scales
Superoleophobicity of Fish Scales, Table 1 Contact angles of oil (1,2-dichloroethane, DCE) or water on three solid surfaces with different surface structures in the air or water environment, respectively Surfaces systems C2H4Cl2 droplets (in air) Water droplets (in air) C2H4Cl2 droplets (in water)
Smooth silicon > eI. Metals are dispersive at optical wavelengths. Away from interband transitions, the Drude model for the permittivity captures the dispersive character of metals [1]:
Surface plasmon-polariton photodetectors er;m ¼ eR jeI
Definition A surface plasmon polariton is a transverse-magnetic optical surface wave propagating along the interface of a metal and a dielectric; it is a coupled excitation formed
¼1
o2p o2p =t j oðo2 þ 1=t2 Þ o2 þ 1=t2
(1)
In the above, op is the plasma frequency and t the relaxation time. The “Drude region” corresponds to
S
S
2592
Surface Plasmon-Polariton-Based Detectors y
6 SiO2
5 δd
εr, m
x, Re{Ey (y)}
δm
Surface Plasmon-Polariton-Based Detectors, Fig. 1 Singleinterface surface plasmon-polariton waveguide; the distribution of the main transverse electric field component (Ey) of the SPP is sketched on the structure as the thick gray curve
4 E (eV)
εr, d
3 2 1
that portion of the electromagnetic spectrum where Eq. 1 holds. For many metals, this region spans the range from visible wavelengths to the infrared. The metal approaches a perfect electric conductor as the wavelength increases through the infrared and beyond. The single-interface (Fig. 1) supports one purely bound (nonradiative) SPP mode. This mode is transverse magnetic (TM) and may propagate at any angle in the x-z plane (e.g., along + z). The SPP fields (Ey, Ez, and Hx) are confined along the y direction, peaking at the interface and decaying exponentially into both media. The distribution of the main transverse electric field component (Ey) of the SPP is sketched on the structure of Fig. 1 as the thick gray curve. The field penetration depth in the metal dm is much smaller than the field penetration depth in the dielectric dd. Confinement of the SPP arises because the metal and dielectric have Re{er} of opposite sign at the wavelength of operation (indeed over a large wavelength range). The wavenumber of the single-interface SPP is [1]: k
SPP
¼ k0
er;m er;d er;m þ er;d
12 (2)
For a lossless dielectric cladding (Im{er,d} ¼ 0), the above simplifies to the following approximate expressions for the real (phase) and imaginary (attenuation) parts of kSPP [1]: 1 eR er;d 2 and eR er;d 3 eI er;d eR 2 k00 ffi k0 2 2eR eR er;d k0 ffi k0
0
10
20
30
40
50
–1)
k⬘ (μm
Surface Plasmon-Polariton-Based Detectors, Fig. 2 Dispersion of the SPP along a Ag/SiO2 interface. The light line in SiO2 is plotted as the dash-dot curve
a Ag/SiO2 interface [2]; the light line in SiO2 is also plotted as the dash-dot curve. As the frequency decreases (increasing wavelength), the metal approaches a PEC, the confinement of the SPP decreases, and its dispersion curve merges with the light line (SPPs are not supported at the interface between a smooth PEC and a semi-infinite dielectric). As the frequency increases, the SPP approaches an “energy asymptote,” readily observed in Fig. 2 near E 3.4 eV (l0 360 nm) where k0 diverges. The group velocity decreases and the optical density-ofstates increases as the asymptote is approached. The SPP bend-back is also observed for wavelengths shorter than the asymptote (l0 < 360 nm, E > 3.4 eV) and links the nonradiative SPP on the right of the light line to the radiative one on the left. The nonradiative SPP is located to the right of the light line (Fig. 2) and so cannot be directly excited by an incident beam. An additional structure is required to increase the in-plane momentum of the beam to match that of the SPP. A prism or a corrugated grating is commonly used to accomplish this task; alternatively, the SPP can be excited by end-fire coupling [1].
(3)
Figure 2 plots the dispersion curve of the SPP (E ¼ hn in eV versus k 0 , h is Planck’s constant) on
Detection Mechanisms Figure 3a gives a schematic of a conventional Schottky photodiode on n-type Si (n-Si) [3], adopted here to
Surface Plasmon-Polariton-Based Detectors
a
Schottky contact
S
2593 n-Si
Metal
b
Vb
t
e
hv ΦB EHP
IPE
n-Si n+-Si
p
Metal
EC
EF
Ohmic contact hv EV
Surface Plasmon-Polariton-Based Detectors, Fig. 3 (a) Schottky diode on n-Si illuminated by light of photon energy hn. Reverse biasing (Vb < 0) is assumed. Electrons are depicted by filled circles and holes by unfilled ones. (b) Energy band diagram of a Schottky contact on n-Si and the three-step internal
photoemission process: p photoexcitation, t transport, e emission. EC and EV are the conduction and valence band edges, respectively, EF is the Fermi level, and FB is the Schottky barrier height
describe generic detection mechanisms that can also be used for the detection of SPPs. A rectifying Schottky contact is formed at the abrupt interface between the top metal and the semiconductor, and an Ohmic (nonrectifying) contact is formed at the interface between the bottom metal and the heavily doped body. The structure is reverse biased (for detection) by applying Vb < 0. The complementary structure, obtained by exchanging the dopants (n ↔ p, n+ ↔ p+), is also of interest. The Schottky diode is a convenient structure with which to detect SPPs because SPPs may propagate along its metal surfaces. Figure 3a shows two mechanisms used for photodetection. The first consists of the creation of electron–hole pairs (EHPs) in the semiconductor due to absorption therein of incident radiation of energy hn greater than the bandgap energy of the semiconductor Eg. This mechanism, sketched in Fig. 3a and labeled EHP, involves three steps: optical absorption and creation of EHPs, separation of EHPs and transport across the absorption region (with or without gain) under reverse bias, and collection of EHPs into the photocurrent at the device contacts. The second mechanism consists of the internal photoemission (IPE) of hot carriers created in the metal due to absorption therein of incident radiation of energy hn. This mechanism is sketched in Fig. 3a, labeled IPE, and is described in greater
detail via the energy band diagram of Fig. 3b. IPE is a 3-step process consisting of the photoexcitation of hot (energetic) carriers in the metal by optical absorption (p), the transport and scattering of hot carriers toward the Schottky contact (t), and the emission of hot carriers over the Schottky barrier into the semiconductor (e) where they are collected under a reverse bias as the photocurrent. This mechanism requires that hn be greater than the Schottky barrier energy FB, and is useful when FB < hn < Eg; i.e., for detection at energies below the bandgap of the semiconductor (because the creation of EHPs in the semiconductor is more efficient and dominates over IPE when hn > Eg). The internal quantum efficiency i (internal photoyield) is a useful measure to characterize and optimize the detection mechanism. It is defined as the number of carriers that contribute to the photocurrent Ip per absorbed photon per second: i ¼
Ip =q Sabs =hv
(4)
where Sabs is the absorbed optical power and q is the elemental charge. i 1 for detection via the creation of EHPs in high-quality (defect-free) direct bandgap semiconductors (assuming absorption in the semiconductor only). In the case of detection via IPE, assuming
S
2594
Surface Plasmon-Polariton-Based Detectors
absorption in the metal only near the Schottky contact, i is given approximately by [4]: rffiffiffiffiffiffi!2 1 FB 1 i ¼ 2 hv
kL
E Lx ϑϑ
(5) k SPP x
Typical Schottky barrier heights are FB ¼ 0.34, 0.8, 0.58, and 0.72 eV for Au/p-Si, Au/n-Si, Al/p-Si, and Al/n-Si, respectively [3]; for detection at, e.g., l0 ¼ 1,310 nm i ranges from about 0.3% to 9%. Metal silicides can also be used, providing lower Schottky barriers. The external quantum efficiency e (external photoyield) and the responsivity R describe how well the detector performs when inserted into a system. e is defined similarly to i except that it depends on the incident optical power Sinc: e ¼
Ip =q Sinc =hv
(6)
e and i are related by: e ¼ Ai
(7)
where A is the optical absorptance, defined as: A¼
Sabs Sinc
(8)
Ip q A q ¼ e ¼ i hv hv Sinc
k Lx
kG
SiO2
H
(9)
SPP detectors, or SPP-enhanced detectors, typically combine a metallic structure that supports SPPs with a detector structure such as that shown in Fig. 3.
Grating-Coupled Detectors An SPP detector structure of interest combines a Schottky detector with a grating designed to couple incoming optical waves to SPPs, as shown in Fig. 4 [5]. The structure consists of a top Al electrode on a thin
z
y
^ Ohmic aluminum contact
x
Surface Plasmon-Polariton-Based Detectors, Fig. 4 SPP Schottky detector integrated with a grating coupler to excite x-propagating SPPs along the top surface of the top Al contact (Adapted from [5]. # 1989 Optical Society of America.)
SiO2 tunneling barrier on p-Si, formed as a sinusoidal corrugated grating of height H, period L, and grating vector kG ¼ x2p/L (x is the x-directed unit vector). p-polarized light having a wavevector kL and an electric field EL is incident onto the grating at the angle #. The incident light couples to the SPP propagating along the air/Al interface in the x direction when the following momentum conservation equation is satisfied: kxL þ mkG ¼ kxSPP
The responsivity R is given by the ratio of the photocurrent to the incident optical power, and can be expressed in terms of e and i: R¼
EL
pAl Si um lic in on um
S
(10)
where kxL is the x-directed component of kL, m is an integer, and kxSPP is the SPP wavenumber (following the notation of [5]). Equation 10 holds for shallow gratings – H is a small perturbation to the surface. The Al Schottky contact is thin enough (35 nm) for the SPP to tunnel through and leak into the high refractive index p-Si layer for detection via the creation of EHPs at the wavelength of operation (l0 ¼ 646 nm). Grating-coupled SPP detectors are sensitive to the angle of incidence, polarization, and wavelength of the incident light through Eq. 10. The polarization sensitivity, for instance, can be exploited in an arrangement of two detectors to determine the polarization angle of linearly polarized normally incident (# ¼ 0) light [5]. Grating-coupled SPPs have also been used to increase the absorptance in the metal contact of Schottky detectors based on IPE (for sub-bandgap
Surface Plasmon-Polariton-Based Detectors
detection) leading to increased responsivity. For example, a 30 increase in responsivity was observed for a Au/p-InP Schottky detector by exciting SPPs along the air/Au interface at l0 ¼ 1150 nm via an integrated corrugated grating, relative to the same structure without the grating [6]. Prism coupling, in the Otto arrangement [1] for instance, has also been used to excite SPPs on detector structures as an alternative to grating coupling. For example, a 10–20 increase in IPE was observed for Au/n-GaAs Schottky detectors by exciting SPPs along the Au/GaAs interface at l0 ¼ 1150 nm in an Otto arrangement implemented monolithically through the substrate [7].
Hole-Coupled Detectors Optical transmission through a single sub-wavelength hole in a metal film can be significantly larger than predicted by Bethe’s theory for a hole in a perfect conductor [8]. This larger (extraordinary) transmission is due to the excitation of SPPs on the metal surface near the hole that then propagate through the hole and couple to radiation on the other side; i.e., the transfer of energy through the hole is mediated by SPPs. Transmission at visible wavelengths is largest for a hole that is 150–300 nm in diameter in 200–300 nm thick Au or Ag films [8]. Structuring the metal film into a corrugation, e.g., around a sub-wavelength hole or slit leads to increased collection of incident light and to greater transmission at resonant wavelengths [8]. Applying this concept to detectors leads to devices that may offer improved performance [9–11]. Ishi et al. [9] reported a Schottky detector on n-Si where the metal forming the Schottky contact was structured into a circular corrugated grating surrounding a 300 nm diameter hole (and covered by SiO2). The contact metal was a Ag/Cr stack 200/10 nm thick. Detection was reported at 840 nm based on absorption in a 300 nm diameter n-Si mesa located directly below the hole, and the creation of EHPs therein. A prospective advantage of the structure is highspeed operation due to the small size of the detection volume while maintaining good responsivity due to the light collection ability of the grating/hole combination; thus the detector exhibits a favorable trade-off between these two parameters (compared to conventional
2595
S
detectors). It has also been predicted that the grating/ hole (or slit) combination can lead to detectors having an improved signal-to-noise ratio [10, 11]. Arrays of holes in metal films also couple incident light to SPPs, in a manner analogous to a corrugated grating [8]. The momentum conservation equation for a two-dimensional (low perturbation) periodic structure on a metal surface is [8, 12]: kSPP ¼ kjj þ lGx þ mGy
(11)
where k|| is the in-plane wavevector of the incident light, Gx and Gy are reciprocal lattice vectors of the periodic structure, and l, m are integers. This equation models approximately the coupling of light to SPPs on a two-dimensional hole array, although the holes are actually neglected along with their associated effects such as scattering and transmission. For example, in the case of a hexagonal or triangular lattice of holes of lattice constant a, illuminated at normal incidence (k|| ¼ 0) and neglecting all losses, the above yields the wavelengths lc for which strong coupling to SPPs on the array occurs:
4 2 l þ lm þ m2 lc ¼ a 3
12
eR er;d eR er;d
12 (12)
In the above, er,d corresponds to the dielectric bounding the array on the illuminating side. If the structure is symmetric (same dielectric on both sides of the array), then these wavelengths also correspond to those at which transmission peaks occur. If the structure is asymmetric (different dielectrics bounding the array) then two sets of transmission peaks are observed. Hole arrays can be integrated with broadband detectors to introduce wavelength and polarization selectivity, and potentially, to enhance the responsivity [12, 13]. This is achieved by structuring the top or bottom contact of a detector into an array of subwavelength holes and then illuminating at normal incidence. The photocurrent is then largest at wavelengths for which extraordinary transmission occurs, which depend on the lattice constants of the array, e.g., via Eq. 12 for the case of a hexagonal array, and so are easily tunable. Polarization selectivity can be introduced by using elliptical or rectangular holes. The responsivity can be enhanced for a thin detection layer by enhancing the absorptance (A) via
S
S
2596
Surface Plasmon-Polariton-Based Detectors
a) of square holes (dimensions w) patterned onto the top contact (150-nm thick Ag) of an InAs mid-infrared quantum dot detector [13]. Figure 5b shows the measured spectral response of such detectors as a function of a, compared to an unpatterned control detector (upper trace). The peak wavelength response is observed to range from l0 ¼ 5.5–7.2 mm as the lattice constant ranges from a ¼ 1.83–2.38 mm.
a
w w a a 5 μm
Detectors Incorporating Nanoparticles
b Unpatterned
Spectral response (a.u.)
a = 1.83 μm a = 1.90 μm a = 1.97 μm a = 2.04 μm a = 2.10 μm a = 2.17 μm a = 2.24 μm a = 2.31 μm a = 2.38 μm 5
6 7 8 9 Wavelength (μm)
10
Surface Plasmon-Polariton-Based Detectors, Fig. 5 (a) Image of a square lattice of square holes; the lattice constant is a and the hole size is w. (b) Spectral response of quantum dot detectors each bearing a hole array of the indicated lattice constant; w ¼ 0.6a for all detectors (Adapted from [13]. # 2009 American Institute of Physics.)
mediation by SPPs. This is achieved by placing the detection layer close to the hole array such that SPPs propagating along the array overlap with the layer and are absorbed therein; thus the enhancement arises because SPPs propagate along the absorption region rather than through it perpendicularly. Wavelength selectivity and responsivity enhancement are particularly attractive for mid-infrared detector arrays, which are typically broadband and have a low responsivity due to the low absorption of the detector materials used in this wavelength range. Applications such as multi-color (multi-spectral) night vision or medical imaging would benefit from inexpensive spectral filtering integrated at the pixel level. Figure 5a shows a square array (lattice constant
Small metal particles exhibit resonant responses under optical excitation, characteristic of particle shape, size, and composition, the dielectric environment in which they find themselves, and the wavelength of illumination [14]. Resonances occur when the electron charge density oscillates coherently with the illuminating optical (electrical) field. The fundamental resonant mode of, e.g., a spherical metal nanoparticle (small compared to the illuminating wavelength) is dipolar with densities of opposite charge forming at opposite spherical caps of the particle along the polarization of the illuminating electric field. The wavelength of the dipolar resonance red-shifts as the radius of the particle increases or as the index of the background increases. If the particle is large enough then higher-order resonant modes can exist (e.g., quadrupolar). On resonance, the electric field near the particle (in the dielectric) is strongly enhanced compared to the illuminating field. Resonant features appear in the measured extinction and scattering spectra of metal nanoparticles, in correlation with the plasmon resonances that are excited thereon. Metal nanoparticles can be integrated with a detector most readily via deposition onto the surface through which light penetrates into the detector [15, 16]. As light propagates through, resonances can be excited, and strong scattering can occur, leading to improved detector performance. Figure 6a shows a scanning electron microscope image of Ag-island nanoparticles on a 30-nm thick LiF layer on a Si-on-insulator (SOI) detector [15]. These approximately hemispherical particles are 108 nm across, on average. They were formed by first depositing a thin Ag film, then annealing the film to induce islandization. This process has the advantage of being simple but it generates particles that are nonuniform in shape and size, albeit, with a controllable
Surface Plasmon-Polariton-Based Detectors
2597
a
Enhancement
b 20 15
Mean particle diameter (c) 108 nm
10
(b) 66 nm (a) 40 nm
5
0 400
500
600
700
800
900
1000
S
more effectively by the detector compared to perpendicularly incident light. Chemically synthesized (colloidal) nanoparticles can also be deposited and attached to detector surfaces. Spectral responses of detectors coated with 50, 80, and 100 nm diameter spherical Au nanoparticles showed strong photocurrent enhancement at wavelengths where peaks were measured in the extinction spectra of the particles [16]. In contrast to the devices of [15], the enhanced photocurrent was attributed to enhanced absorption in the Si regions in contact with resonating nanoparticles, around which the resonating electric fields are significantly enhanced compared to the illuminating fields. The effects induced by metal nanoparticles, i.e., increased scattering into detectors and increased detector absorption on resonance, are useful to increase the absorptance of detectors, particularly those based on a thin absorption layer, or at wavelengths where the absorption is low (e.g., near the bandgap of an indirect semiconductor). These effects can potentially improve the efficiency of, e.g., thin-film Si solar cells [17].
Wavelength (nm)
Surface Plasmon-Polariton-Based Detectors, Fig. 6 (a) Scanning electron microscope image of Ag-island nanoparticles on LiF-coated Si-on-insulator (SOI) detector. The particles are 108 nm across, on average. (b) Spectral response of pn junction SOI photodetectors coated with Ag islands similar to those of (a) as a function of average island size. The enhancement is defined as the ratio of the photocurrent from a detector with Ag islands to the photocurrent from a detector without the islands (Adapted from [15]. # 1998 American Institute of Physics.)
average size. Detectors were formed in the thin Si slab (165 nm thick) above the insulator but beneath the LiF (and the nanoparticles) as pn junction detectors. Figure 6b shows measured spectral responses of detectors coated with Ag islands similar to those of Fig. 6a as a function of average island size. The enhancement is defined as the ratio of the photocurrent from a detector with Ag islands to the photocurrent from a detector without islands. A strongly enhanced photocurrent (20 ) is noted at l0 ¼ 800 nm in the case of islands that are 108 nm in average size. The enhancement is attributed to the large scattering cross-section of the islands, particularly to mediation by the islands whereby dipole resonances excited thereon radiate into guided modes of the thin underlying Si slab. These modes propagate longitudinally and thus are absorbed
Waveguide Detectors The single-interface (Fig. 1) is one example of an SPP waveguide; other popular 1-D geometries include the thin metal slab bounded by dielectrics and the thin dielectric slab bounded by metals [2]. The latter are often termed the IMI and MIM, respectively (I – insulator, M – metal). The IMI and MIM support supermodes formed from the coupling of single-interface SPPs through the thin intervening layer. The structures can be designed such that they support super-modes that are low loss but weakly confined (symmetric IMI) or strongly confined but high loss (MIM), and thus that are at opposite ends of the confinement-attenuation trade-off [2]. SPP waveguides can be formed into SPP detectors and conveniently integrated with other plasmonic or photonic waveguide structures. SPP waveguide detectors based on absorption in, e.g., organics [18], semiconductors [19], or metals [20] have been reported. The detector described in [18] consists of a thick Ag film on which SPPs are excited by illuminating a slit in the film. The SPPs then propagate along the Ag film to the detector section, which consists of a pn junction fabricated from organic polymers directly on the Ag
S
S
2598
Surface Plasmon-Polariton-Based Detectors
Figure 7a gives a sketch of a SPP waveguide detector consisting of a thin narrow metal stripe on semiconductor (with air on top) [20]. The thin metal stripe is similar to the asymmetric IMI, but it supports SPPs with additional confinement along the lateral dimension (stripe width). Detectors have been fabricated as Au or Al stripes on n-Si forming Schottky contacts thereon [20]. The asb0 mode, localized to the bottom metal-Si interface, is excited via butt-coupling to a tapered polarization-maintaining single mode fiber (PM-SMF). This mode propagates along the stripe with strong absorption, creating hot carriers therein, some of which may cross the Schottky barrier and be collected as the photocurrent; thus detection occurs via IPE. Figure 7b gives the spectral response of a Au on n-Si detector, plotted in Fowler form. The intercept with the abscissa yields the cutoff photon energy (0.765 eV), corresponding in this case to a cutoff wavelength of l0 1620 nm. Responsivities up to 1 mA/W were reported with this detector scheme.
Excited asb0 Mode
a
Tapered PM-SMF
SPP waveguide
b 20 x 10−3
R ½hv (eV (A / W )½)
15
10
5
0
0.75
0.8
0.85
0.9
0.95
hv (eV)
Surface Plasmon-Polariton-Based Detectors, Fig. 7 (a) Sketch of an SPP waveguide detector consisting of a metal stripe forming a Schottky contact on Si, excited (in the asb0 mode) via butt-coupling to a tapered polarization-maintaining single mode fiber (PM-SMF). (b) Spectral response of a Au on n-Si detector (as in (a)), plotted in Fowler form (Adapted from [20]. # 2010 Optical Society of America.)
film. The pn junction is covered by another Ag film, thereby creating an MIM-type detector. SPPs incident from the Ag film onto this MIM detector excite SPPs therein, which are absorbed by the pn junction creating EHPs that are collected as the photocurrent. Current maps correlating the photocurrent with the excitation of SPPs at l0 ¼ 632.8 nm along the Ag film are reported. The structure described in [19] consists of an MIM (Au/HSQ/Au) waveguide with slits, fabricated on a GaAs wafer. A slit in the top Au layer is used to couple incident light to SPPs propagating in the MIM. A slit in the bottom Au layer, placed a short distance away from the input, couples the SPP into radiation directed into the GaAs wafer, which is absorbed therein creating EHPs that are collected as the photocurrent. Spectral responses and photocurrent maps demonstrating the operation of the detector are reported.
Concluding Remarks SPP detectors, or SPP-enhanced detectors, typically combine a metallic structure that supports SPPs, such as a planar or grating structure, metallic nanoparticles, or holes in metal films, with a semiconductor detector structure such as a Schottky or a pn junction. Properties inherent to SPPs are exploited to convey additional characteristics to a detector such as polarization, angular or spectral selectivity, or to enhance the absorptance (A) of the detector. The involvement of SPPs has led to detectors with improved performance and greater functionality.
References 1. Maier, M.A.: Plasmonics: Fundamentals and Applications. Springer, New York (2007) 2. Berini, P.: Figures of merit for surface plasmon waveguides. Opt. Express 14, 13030–13042 (2006) 3. Sze, S.M.: Physics of Semiconductor Devices. Wiley, New York (1981) 4. Scales, C., Berini, P.: Thin-film Schottky barrier photodetector models. IEEE J. Quant. Electr. 46, 633–643 (2010) 5. Jestl, M., Maran, I., Kock, A., Beinstingl, W., Gornik, E.: Polarization-sensitive surface plasmon Schottky detectors. Opt. Lett. 14, 719–721 (1989) 6. Brueck, S.R.J., Diadiuk, V., Jones, T., Lenth, W.: Enhanced quantum efficiency internal photoemission detectors by
Surface Tension Effects of Nanostructures
7.
8. 9.
10.
11.
12.
13.
14.
15.
16.
17. 18.
19.
20.
grating coupling to surface-plasma waves. Appl. Phys. Lett. 46, 915–917 (1985) Daboo, C., Baird, M.J., Hughes, H.P., Apsley, N., Emeny, M.T.: Improved surface plasmon enhanced photodetection at an Au-GaAs Schottky junction using a novel molecular beam epitaxy grown Otto coupling structure. Thin Solid Films 201, 9–27 (1991) Genet, G., Ebbesen, T.W.: Light in tiny holes. Nature 445, 39–46 (2007) Ishi, T., Fujikata, J., Makita, K., Baba, T., Ohashi, K.: Si Nano-photodiode with a surface plasmon antanna. Jpn. J. Appl. Phys. 44, L364–L366 (2005) Yu, Z., Veronis, G., Fan, S., Brongersma, M.L.: Design of midinfrared photodetectors enhanced by surface plasmons on grating structures. Appl. Phys. Lett. 89, 151116 (2006) Bhat, R.D.R., Panoiu, N.C., Brueck, S.J.R., Osgood, R.M.: Enhancing the signal-to-noise ratio of an infrared photodetector with a circular metal grating. Opt. Express 16, 4588– 4596 (2008) Chang, C.Y., Chang, H.Y., Chen, C.Y., Tsai, M.W., Chang, Y.T., Lee, S.C., Tang, S.F.: Wavelength selective quantum dot infrared photodetector with periodic metal hole arrays. Appl. Phys. Lett 91, 163107 (2007) Rosenburg, J., Shenoi, R.V., Vandervelde, T.E., Krishna, S., Painter, O.: A multispectral and polarization-selective surface-plasmon resonant midinfrared detector. Appl. Phys. Lett. 95, 161101 (2009) Kelly, K.L., Coronado, E., Zhao, L.L., Schatz, G.C.: The optical properties of metal nanoparticles: The influence of size, shape, and dielectric environment. J. Phys. Chem. B 107, 668–677 (2003) Stuart, H.R., Hall, D.G.: Island size effects in nanoparticleenhanced photodetectors. Appl. Phys. Lett. 73, 3815–3817 (1998) Schaadt, D.M., Feng, B., Yu, E.T.: Enhanced semiconductor optical absorption via surface plasmon excitation in metal nanoparticles. Appl. Phys. Lett. 86, 063106 (2005) Atwater, H.A., Polman, A.: Plasmonics for improved photovoltaic devices. Nat. Mater. 9, 205–213 (2010) Ditlbacher, H., Aussenegg, F.R., Krenn, J.R., Lamprecht, B., Jakopic, G., Leising, G.: Organic diodes as monolithocally integrated surface plasmon polariton detectors. Appl. Phys. Lett. 89, 161101 (2006) Neutens, P., Van Dorpe, P., De Vlaminck, I., Lagae, L., Borghs, G.: Electrical detection of confined gap plasmons in metal-insulator-metal waveguides. Nat. Photon. 3, 283– 286 (2009) Akbari, A., Tait, R.N., Berini, P.: Surface plasmon waveguide Schottky detector. Opt. Express 18, 8505–8514 (2010)
Surface Properties ▶ Nanostructures for Surface Functionalization and Surface Properties
2599
S
Surface Tension and Chemical Potential at Nanoscale ▶ Surface Energy and Chemical Potential at Nanoscale
Surface Tension Effects of Nanostructures Ya-Pu Zhao and Feng-Chao Wang State Key Laboratory of Nonlinear Mechanics (LNM), Institute of Mechanics, Chinese Academy of Sciences, Beijing, China
Synonyms Surface energy density; interface excess free energy
Definition The surface tension is the reversible work per unit area needed to elastically stretch/compress a preexisting surface. In other words, it is a property of the surface that characterizes the resistance to the external force. Surface tension has the dimension of force per unit length or of energy per unit area. For nanostructures with a large surface to volume ratio, the surface tension effects dominate the size-dependent mechanical properties.
Overview Nanostructures have a sizable surface to volume ratio as compared to bulk materials, which leads its mechanical properties to be quite different from those of bulk materials [1]. The size-dependent mechanical properties of nanostructures are generally attributed to the surface effects, in which surface tension is one of the most predominant factors. In the framework of the thermodynamics, the surface (interface) between two phases is first modeled as a bidimensional geometrical boundary of zero thickness, that is, the mathematical surface, as shown in Fig. 1a. The physical quantities between the two phases are discontinuous, and the surface (interfacial)
S
S
2600
Surface Tension Effects of Nanostructures
Surface Tension Effects of Nanostructures, Fig. 1 An illustration for the surface (interface) models. (a) The mathematical surface of zero thickness, the physical quantities are discontinuous. (b) Gibbs’s surface with an infinitesimal thickness, the physical quantities are continuous. Cahn–Hilliard model for surface with a finite thickness is also illustrated
y
a
b
y
Phase B WB Mathematical interface(surface) x
x
Thickness l
W(y) WA Phase A
tension is a jump in stress. This idealization was then extended by Gibbs [2], who proposed that the physical quantities should undergo a smooth transition at the surface (interface) while the surface is still modeled as an infinitesimal thin boundary layer, as shown in Fig. 1b. In order to preserve the total physical properties of the system, the excess physical properties have to be assigned to the geometrical surface. The surface tension, which is defined based on the interface excess free energy, can be written as Z
1
g¼
Z ½wðyÞ wA dy þ
0
0 1
½wðyÞ wB dy
(1)
where w is the free energy distribution of the actual surface, wA and wB are free energy in the two phases of A and B. In some other theoretical models, the surface is treated as an extended interfacial region with a nonzero thickness. According to Cahn–Hilliard theory [3], the surface tension can be derived as Z g ¼ NV
1
1
h
i w0 ðcÞ þ kðdc=dyÞ2 cmB ð1 cÞmA dy;
(2) in which NV is the number of atoms per unit volume, c is one of the intensive scalar properties, such as composition or density, w0(c) is the free energy per atom of a solution of uniform composition c, k reflects the
crystal symmetry, mA and mB are the chemical potentials per atom in the A or B phase. The surface (interfacial) thickness can be obtained by rffiffiffiffiffiffiffiffiffiffiffiffiffi k l ¼ 2Dce ; Dwmax
(3)
where 2Dce ¼ cBcA is the difference of the uniform composition in the two phases, Dwmax is the maximum of the free energy referred to a standard state of an equilibrium mixture. Cahn–Hilliard model gives a finite thickness through Eq. 3, which has been used in many applications [4]. From the standpoint of molecular theory, the surface tension effects arise due to the difference of the atomic interactions in the bulk and on the surface. Atoms are energetically favorable to be surrounded by others. At the surface, the atoms are only partially surrounded by others and the number of the adjacent atoms is smaller than in the bulk. Thus the atoms at the surface are energetically unfavorable. If an atom moves from the bulk to the surface, work has to be done. With this view, the surface tension can be interpreted as the energy required to bring atoms from the bulk to the surface [5]. Therefore the term “surface energy density” is often used to when the surface tension is referred to. For the surface of solid, Gibbs pointed out that surface tension and surface energy density are not identical. The surface tension is the reversible work per unit area needed to
Surface Tension Effects of Nanostructures
2601
elastically stretch/compress a preexisting surface. The surface energy density is the reversible work per unit area needed to create a new surface. The surface tension can be positive or negative, while the surface energy density is usually positive. The surface tension for liquid surface is a property that characterizes its resistance to the external force, which is identical to the surface energy density.
Basic Methodology Since the surface to volume ratio increases as the dimension scale decreases, surface tension effects of nanostructures can be overwhelming. In the absence of external loading, the surface tension effects would induce a residual stress field in bulk materials. The relations between surface stress and surface tension for small deformations can be described by the Shuttleworth-Herring equation [6, 7], sij ¼ gdij þ
@g ; @eij
(4)
where g is the surface tension, dij is the Kronecher delta, sij and eij are the surface stress tensor and the surface strain tensor, respectively. The ShuttleworthHerring equation interprets that the difference between the surface stress and surface tension is equal to the variation of surface tension with respect to the elastic strain of the surface. With the development of computational materials science, molecular dynamics (MD) simulations are wildly performed to investigate the surface tension effects on the mechanical properties of nanostructures. Especially for the surface elastic constant, atomistic simulation is almost the only way to get them up to now. According to the Gibbs’s definition, the surface tension of a solid is given by g ¼ ðES nEB Þ=A0 , where A0 is the total area of the surface considered, ES is the total energy of a n-layer slab and EB is the bulk energy per layer of an infinite solid. In cases where the surfaces of the slab are polar, the electrostatic energy of the slab contains an energy contribution, Epol, proportional to the substrate thickness and the surface energy needs corrections. Thus there is an alternative way to calculate the surface tension g ¼ ES nEB Epol =A0 , which does not rely on an exact knowledge of the lattice or polar energy. MD simulations have identified that the
S
surface tension induced surface relaxation is proved to be a dominate factor of the size-dependent mechanical properties. When the surface stress is negative, the surface relaxation is inward; otherwise, the relaxation is outward. Surface stress has been used as an effective molecular recognition mechanism. Surface stresses due to DNA hybridization and receptor-ligand binding induce the deflection of a cantilever sensor [8]. The curvature of bending beam under a surface stress is governed by Stoney’s formula [9]. Stoney’s formula serves as a cornerstone for curvature-based analysis and a technique for the measurement of surface stress, which is given as follows as a general form for a film/ substrate system, s¼
Et2s f 6ð 1 n Þ
(5)
in which E is the effective Young’s modulus, n is Poisson’s ratio of the sensor material, ts is the substrate thickness, f ¼ 3Dz/2 L2 is the sensor curvature, L is the length and Dz is the deflection. The applicability of the above Stoney’s formula relies on several assumptions, which are well summarized as the following six: (1) both the film and substrate thicknesses are small compared to the lateral dimensions; (2) the film thickness is much less than the substrate thickness; (3) the substrate material is homogeneous, isotropic, and linearly elastic, and the film material is isotropic; (4) edge effect near the periphery of the substrate are inconsequential and all physical quantities are invariant under change in position parallel to the interface; (5) all stress components in the thickness direction vanish throughout the material; and (6) the strains and rotations are infinitesimally small. However, the one or several of above six assumptions can be easily violated in reality, which is to say that the Stoney’s formula needs to be revised to fit in the real applications. To summarize for the solid cases, the behaviors of nanostructures can be affected significantly by either of the two distinct parameters, surface tension and surface stress. The relation between the two parameters can be obtained by the Shuttleworth-Herring equation. For the surface stress induced deflection of cantilever sensors, the curvature is by Stoney’s formula. MD simulation is helpful to understand the surface effects of nanostructures and partial results are comparable to the experiments.
S
S
2602
Surface Tension Effects of Nanostructures
Surface Tension Effects of Nanostructures, Table 1 Characteristic time related to the surface tension effects Name Capillary characteristic time
Expression pffiffiffiffiffiffiffiffiffiffiffiffiffi m=gLV
Meaning Characteristic time derived from the droplet mass and the liquid– vapor surface tension qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p The period of a free 3 rd0 =gLV tp ¼ droplet in free oscillation 4 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Characteristic time of tR rl3 =gLV droplet dynamics tc ¼
Lord Rayleigh’s period Lord Rayleigh’s characteristic time Viscous characteristic time
tvis l=gLV
Characteristic time related to the liquid viscosity
For liquid droplets in contact with the surface of a nanostructure, surface tension is responsible for the shape of liquid droplets in the equilibrium state, as well as the dynamics response in wetting and dewetting. There are several characteristic time which are related to the surface tension effects, listed in Table 1. Dimensionless number related to the surface tension effects are listed in Table 2. Surface tension is dependent on temperature T, concentration of surfactants c, and the electric field V. The gradient of surface tension caused by these factors can be described by [10] @g @g @g dT þ dc þ dV: @T @c @V
cos y ¼ cos y0
t e0 eD 2 A þ V þ ; (7) gLV R 2dgLV 12ph2 gLV
where
m mass of the droplet, d0 diameter, r density, viscosity, l characteristic length.
dg ¼
vapor interface meets the solid surface. The contact angle is affected by various factors, including the surface tension, the line tension, the applied voltage as well as the molecular interactions between the liquid and solid surfaces. It is proposed that the dependence of the contact angle on these factors can be represented by the generalized Young’s equation, which has a form of
(6)
To the first order, the dependency of the surface tension on temperature is given by Guggenheim– Katayama formula with power index n ¼ 1, g ¼ g0 ð1 T=Tc Þ. Here g0 is a constant for each liquid and Tc is the critical temperature. For the dependency of the surface tension on concentration c, the surface tension can be expressed as a linear function of the concentration, g ¼ g0 ½1 þ bðc c0 Þ. The solid–liquid surface tension can be changed by applying a voltage 0 eD V, gSL ¼ g0SL e2d V 2 , where g0SL is the solid–liquid surface tension in the absence of the applied voltage, e0 is the permittivity of vacuum, eD is the relative permittivity of the dielectric layer with a thickness d separating the bottom electrode from the liquid. The wetting properties of a solid surface can be described by the introduction of the contact angle, which is defined as the angle at which the liquid–
cos y0 ¼
gSV gSL gLV
(8)
is the classical Young’s equation, y0 is the equilibrium contact angle (also the Young contact angle); The subscripts S, L, and V denote solid, liquid, and vapor, respectively. The second term on the right-hand side of Eq. 7 is related to the line tension. t is the line tension and R is the radius of the contact area. For a more generalized case, R can be replaced by 1/k, in which k is the geodesic curvature of the triple contact line. e ¼ 2ge0 eDd V 2 is defined as the dimensionless LV electrowetting number. The last term in Eq. 7 measures the strength of the effective interaction energy compared to surface tension. The interaction energy is related to the disjoining R 1 pressure P(h), which can be obtained by WðhÞ ¼ h Pðh0 Þdh0 , where h is the film thickness and A is the Hamaker constant [11, 12]. As discussed in the following paragraphs, each term would be illustrated in detail. If only the classical Young’s equation is referred to, the wetting properties of the surface can be obtained directly if the three tensions are known. The spreading parameter S determines the type of spreading, which is defined as S ¼ gSV ðgSL þ gLV Þ. If S > 0, the liquid spreads on the solid surface and forms a macroscopic liquid layer covers the whole solid surface; it is the case for complete wetting. If S < 0, the contact angle 0 < y0 < 180 , which means the liquid forms a droplet with a finite contact angle; it is the case for speak of partial wetting. The wettability of the solid surface could be distinguished by the contact angle y0, as shown in Fig. 2. If 0 y0 < 90 , the solid substrate is hydrophilic. If 90 < y0 180 , the solid substrate is hydrophobic. Especially, if 150 < y0 180 , the solid substrate is superhydrophobic.
Surface Tension Effects of Nanostructures
2603
S
Surface Tension Effects of Nanostructures, Table 2 Dimensionless number related to the surface tension effects Name Adhesion number Bond number Capillary number Deborah number Elasto-capillary number Electrowetting number Laplace number
Expression Na ¼ cos ya cos yr Bo ¼ rgl2 =gLV Ca ¼ v=gLV .pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi De ¼t=tR ¼ t rl3 =gLV
Meaning ya and yr are the advancing and receding angle. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The capillary length lc ¼ gLV =ðrgÞ can be determined pffiffiffiffiffiffiffi Ca ¼ On We; the capillary velocity v ¼ gLV = can be obtained t is the relaxation time, Lord Rayleigh characteristic time can be derived
Ec ¼ tgLV =ðlÞ e ¼ e0 eD V 2 =ð2dgLV Þ
t is the relaxation time, viscous characteristic time can be obtained The ratio of the electrostatic energy to the surface tension. See Ohnersoge number
Ohnersorge number
La ¼ð1=OnÞ2 dg L DT Ma ¼ LV a dT pffiffiffiffiffiffiffiffiffiffiffi On ¼ rlgLV
Weber number
We ¼ rv2 l=gLV
Marangoni number
Thermal surface tension force divided by viscous force Viscousforce pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Inertiaforce Surfacetension The inertia force compared to its surface tension
g gravitational acceleration, a thermal diffusivity, DT temperature difference.
a
Complete wetting
Vapor
c
Partial wetting --Hydrophilic
gLV
Vapor q0 = 0°
Liquid
b
Vapor
q < 90° Liquid 0 gSL
gSV
Partial wetting --Hydrophobic gLV Liquid q0 > 90° gSL
gSV
Solid substrate
Surface Tension Effects of Nanostructures, Fig. 2 Wetting properties of the surface
The physics behind wettability is that, the solid surfaces have been divided into high energy and lowenergy types [11, 12]. The relative energy of a solid has to do with the bulk nature of the solid itself. (1) Highenergy surfaces such as metals, glasses, and ceramics are bound by the strong chemical bonds, for example, covalent, ionic, or metallic, for which the chemical binding energy Ebinding is of the order of 1 eV. The solid–liquid interface tension is given by gSV Ebinding a2 0:5 5N=m, in which a2 is the effective area per molecule. Most high-energy surfaces are hydrophilic, some can permit complete wetting. (2) For low-energy surfaces, such as weak molecular crystals (bound by van der Waals forces or in some special cases, by hydrogen bonds), the chemical binding energy is of the order of kBT. In this category, the surface tension is gSV kB T a2 0:01 0:05N=m. Depending on the type of liquid chosen, low-energy surfaces can be either hydrophobic or hydrophilic. When the liquid droplet comes to the nanoscale, the classical Young’s equation seems to be not applicable,
since it has been derived for a triple line without consideration of the interactions near the triple contact line. The molecules close to the triple line experience a different set of interactions than at the interface [10]. To take into account this effect, the “line tension” term has been introduced in the generalized Young’s equation. A sketch of line tension is shown in Fig. 3. Line tension was first introduced by Gibbs [2]. The line tension t was introduced as an analogue of the surface tension: Z F ¼ gLV
S0
0
dS þ ðgLV W Þ
Z
S
dS þ t
Z C
dC ; (9)
where F, S’, S∗, and C∗ are the free energy, the free surface, the adhering interface, and the contact line of the droplet. W ¼ gLV þ gSL gSV is the adhesion potential. The line tension, depending on the radius R of the contact line, should be connected to the classical Young’s equation to include the interactions near the triple contact line. In the three-phase equilibrium
S
S
2604
Surface Tension Effects of Nanostructures
Surface Tension Effects of Nanostructures, Fig. 3 Illustration of the line tension t
systems when R decreases, the contact angle y will increase at t > 0 and will decrease at t < 0. However, in accordance to the mechanical equilibrium stability conditions, while the surface tension can only be positive, the line tension can have either positive or negative values. The characteristic length of this problem l ¼ jtj=gLV (|t| < 109 N, gLV 0.1 N/m for water) ranges from 108 to 106 m, which means small droplet (with typical dimension of |l|) should appreciate the line tension effect. The line tension can be indirectly measured through the contact angle, t pffiffiffiffiffiffiffiffiffiffiffiffiffi 4d gSV gLV cot y, in which d denotes the average distance between liquid and solid molecules. Thus the line tension is negative for an obtuse contact angle, while it is positive for an acute contact angle. In the presence of a charged interface, which can be achieved by applying a direct or alternating-current electric field, the wetting properties of solid surface will be modified. The physics describing the electric forces on interfaces of conducting liquids and on triple contact lines is called “electrowetting” [10]. In the year of 1875, Gabriel Lippmann observed the capillary depression of mercury in contact with an electrolyte solution could be varied by applying a voltage between the mercury and electrolyte. This phenomenon is called electrocapillarity, which is the basis of modern electrowetting. Then, the idea was developed to isolate the liquid droplet from the substrate using a dielectric layer in order to avoid electrolysis. This concept has subsequently become known as electrowetting on dielectric (EWOD) and involves applying a voltage to modify the wetting behavior of a liquid in contact with a hydrophobic, insulated electrode. When an
a Electrode
Droplet
+ −
Bottom Electrode
b Dielectric Layer + −
Surface Tension Effects of Nanostructures, Fig. 4 Illustration of EWOD. The external voltage is applied between a thin electrode (the Pt wire) and the bottom electrode (typically the indium-tin-oxide glass). Partially wetting liquid droplet in the absence of an applied voltage (a) and after the voltage is applied (b)
electric field was applied to the system (as shown in Fig. 4), electric charges gather at the interface between the conductive electrodes and the dielectric material; the surface becomes increasingly hydrophilic (wettable), the contact angle is reduced and the contact line moves. The change in contact angle over the buried electrodes can be evaluated by the Lippmann–Young equation cos y ¼ cos y0 þ e0 eD V 2 =ð2dgLV Þ. The parabolic variation of cos y in the Lippmann–Young
Surface Tension Effects of Nanostructures
equation is only applicable when the applied voltage lies below a threshold value. If the voltage is increased above this threshold, then contact angle saturation starts to occur and cos y eventually becomes independent of the applied voltage. The physics of why a liquid film wets or dewets is found in the derivative of the effective interfacial potential W ¼ gLV þ gSL gSV with respect to film thickness h, called the disjoining pressure P(h) [11, 12]. PðhÞ ¼ dWðhÞ=dh, where the surface area remains constant in the derivative. The effective interface potential W(h), which arises from the interaction energies of molecules in a film being different from that in the bulk, is the excess free energy per unit area of the film. If the interactions between the molecules in the film and the solid substrate are more attractive than the interactions between molecules in the bulk liquid, W(h) > 0. Consequently, a liquid film with a thickness in a range where P(h) > 0 can lower its free energy by becoming thicker in some areas while thinning in others, that is, by dewetting. When P(h) < 0, wetting or spreading occurs. The van der Waals interaction wðrÞ / 1 r 6 includes all intermolecular dipole–dipole, dipole– induced dipole, and induced dipole–induced dipole interactions. Performing a volume integral over all molecules present in the two half spaces bounding the film one finds a corresponding decay P(h) A/6ph3 [11, 12], where the Hamaker constant A (1019 J) gives the amplitude of the interaction. In the “attractive” case, in which the layer tends to thin, A < 0. In the “repulsive” case, in which the layer tends to thicken, A > 0. Because of the disjoining pressure, there is a precursor film ahead of the nominal contact line of the liquid droplet [13]. The thickness of the precursor film can be defined by a molecular length pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [11, 12] hPF A=6pgLV , which is on the order of ˚. several A
Key Research Findings Elastic Models for the Nanostructures Gurtin and Murdoch established the theoretical framework of the surface elasticity under the classical theory of membrane [14]. Recently, studies [15] have shown that, even in the case of infinitesimal deformations, one should distinguish between the reference and the current configurations; otherwise the out-plane terms of
2605
S
surface displacement gradient, associated with the surface tension, may sometimes be overlooked in the Eulerian descriptions, particularly for curved and rotated surfaces. By combining elastic models for surface and bulk, the size-dependent elastic and properties of nanomaterials have been investigated. Usually, in the absence of external mechanical or thermal loadings, the surfaces of a nanostructure will be subjected to residual surface stresses, and an elastic field in the bulk materials will be induced by such residual surface stresses induce from the point of view of equilibrium conditions. This self-equilibrium state without external loadings is usually chosen as the reference configuration, from which nanostructures will deform (see Fig. 5). That is to say, the bulk will deform from the residual stress states. However, in the prediction of elastic and thermoelastic properties of nanostructures, the elastic response of the bulk is usually described by classical Hooke’s law, in which the aforementioned residual stress was neglected in the existing literatures. Considering a bulk material with the surface properties aforementioned, the surface tension would induce a stress field in the bulk. According to Young–Laplace equation, the surface tension will result in a nonclassical boundary condition. The boundary condition together with the equations of classical elasticity forms a coupled system of field equations to determine the stress distribution. To solve a problem considering the surface properties, the surface model and the bulk model are established separately and using the Young–Laplace equation to bridge the two models together. Here, an example named surface elasticity would be used to show how to establish a model combined the surface and bulk together. Assuming the surface to be isotropic and homogeneous, the constitutive relations of the surface in the Lagrangian description can be written as [15]
0s u0 Þ Ss ¼ g 0 I0 þ g 0 þ g 1 ðtrEs ÞI0 g 0 ðH þ g1 Es þ g 0 FðoÞ s ;
(10)
where Ss is the first kind Piola–Kirchhoff stress of the surface, I0 is the identity tensor on the tangent planes of the surface in the reference configuration; the constants g 0 , g 1 , and g1 are the surface tension and the 0s u0 , and FðoÞ denote, surface Lame moduli; Es , H s respectively, the surface strain tensor, the in-plane
S
S
2606
Surface Tension Effects of Nanostructures
Surface Tension Effects of Nanostructures, Fig. 5 The choice of the reference configuration: the stressed state
part of surface displacement gradient and the out-plane term of surface deformation gradient. In view of the importance of the linearization of the general constitutive equations, the linear elastic constitutive relations of the bulk with residual stresses can be written as follows: S ¼ TR þ uH TR þ ltrðEÞ1 þ 2mE;
(11)
where S is the first Piola–Kirchhoff stress, TR is the residual stress in the reference configuration, uH is the displacement gradient calculated from the reference configuration, E is the infinitesimal strain, and l and m are material elastic constants. In the absence of external loading, surface tension will induce a compressive residual stress field in the bulk of the nanoplate and there may be self-equilibrium states which correspond to plate self-buckling. The selfinstability of nanoplates is investigated and the critical self-instability size of simplify supported rectangular nanoplates is proposed. The critical size for selfq ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi 2Þ , buckling is b ¼ ph a2 þ 1 Eh 24g ð 1 n A 0 where b and h are the width and thickness of the nanoplate, respectively; E and ndenote the Young’s modulus and Poisson’s ratio, aA ¼ l=b is the aspect ratio [15].
Surface Tension Effects on the Mechanical Properties of Nanostructures Many models are developed to relax one or some of the six above assumptions to extend Stoney’s formula to a more generalized and realistic application. For example, the effects such as axial force, the damaged/ nonideal interface effect and gradient stress, which violates one or several of the above assumptions, are analyzed and the extended/revised Stoney’s formulas are given. Surface stress physically is a distributed one and this characteristic is not emphasized in many studies. The analysis by Zhang et al. shows that the Stoney’s formula is obtained by assuming the influence of surface stress as a concentrated moment applied at the free end of a cantilever beam [16]. However, if the influence of surface stress is modeled as a distributed axial load and bending moment, the following nonlinear governing equation is obtained: EI
d4 z d2 z dz sbðL xÞ 2 þ sb ¼ 0; 4 dx dx dx
(12)
in which EI is the cantilever effective bending stiffness; b, L, and z are the beam width, length, and deflection, respectively. To solve the above nonlinear equation with the (given) boundary conditions at the two ends of the beam is a two-point-boundary value
Surface Tension Effects of Nanostructures Surface Tension Effects of Nanostructures, Fig. 6 Sketch of deformation of PDMS membrane induced by a water droplet
2607
gLV
gSL
q
S
gLV sinq
gSV
gLV cosq
Z
Y
X
problem [16], which is rather difficult. The semianalytical series solutions can more or less ease the difficulty of solving Eq. 12. One implication of the Stoney’s formula is that because the beam curvature is constant, the beam deflection under a surface stress is an arc of a circle (or a parabola if the approximate curvature definition is used). There is no mechanism to guarantee such kind of deflection. In general the curvature of a beam under a surface stress is not a constant, which has been verified in the experiments. The Stoney’s formula has been used to explain recent experimental results, in which a hybrid device based on a microcantilever interfaced with bacteriorhodopsin (bR), undergoes controllable and reversible bending when the light-driven proton pump protein, bR, on the microcantilever surface is activated by visible light [17]. It should be pointed that the Young’s modulus of a nanostructure is size-dependent, that is, it would be enhanced or softened with decreasing the size of the nanostructure, which is generally attributed to the surface effects. Surface tension is one of the most important factors that cause the size effects of the Young’s modulus of a nanostructure. The surface tension can be introduced into mechanical model via energy method. Using the relation of energy equilibrium, the effective elastic modulus of nanobeams are dependent on the surface tension [18]. Surface Tension Effects Induced the Deformation of Nanostructures The classical Young’s equation describes the equilibrium of forces in the direction parallel to the solid surface, while the vertical component of liquid–vapor interfacial tension is ignored. That is, there is a net force gLV sin y acting normal to the smooth solid surface at the solid–liquid–vapor contact line. Due to the unbalance force, there will be a surface deformation, as
shown in Fig. 6. Indeed, several decades ago, surface deformation of semi-infinite solid was theoretically analyzed with the physical assumption that the liquid–vapor has a finite thickness (maybe at the order of tens nanometers) and the liquid–vapor interfacial tension acts uniformly in this region. Their research suggests there is a wetting ridge at the three-phase contact line. Later, Shanahan and Carre´ used dimensional analysis to characterize the maximum height at the order of gLV =G, where G is the shear modulus of solid [19]. For the material widely used at that time were very rigid (at the order of at least 100 GPa), such a deformation is too small to be considered. However, to meet with the rapid development of microelectromechanical systems (MEMS) and nanoelectromechanical systems (NEMS), polydimethylsiloxane (PDMS) is widely fabricated to channels or membrane, which has at least one dimension on the order of submillimeters or even nanometers. The surface deformation might no longer be neglected. Moreover, it should be noted that whether the theoretical solution for semiinfinite case can be extended to the case of thin flexible membrane. Recently, Yu and Zhao considered the deformation of thin elastic membrane induced by sessile droplet and gave a theoretical solution correspondingly [20]. There are two important conclusions. The first is that there exists a saturated membrane thickness at the order of millimeter, if the solid is thicker than this, it can be taken regard as semi-infinite; otherwise, it is better to consider the effect of membrane thickness. The second is that if the membrane has a very low Young’s modulus (for example, on the order of MPa or much less), the effect of membrane thickness will become significant. Apart from theoretical analysis, experimental investigations on surface deformation induced by droplet have also been reported. Because of the surface resolving ratio, it is difficult to get the detailed
S
S
2608
Surface Tension Effects of Nanostructures, Fig. 7 Huh and Scriven’s paradox: Not even Herakles could sink a solid
Surface Tension Effects of Nanostructures
hU 2 · In E~ – q trq =
2h r
L 0
( c cos q – d sin q)
Moving Contact Line
Water
information of surface deformation at the contact line. Moreover, there might be a highly stressed zone near the contact line so that such a question should be studied further when necessary. Surface Tension Effects of Wetting at the Three-Phase Contact Line When considering the surface tension effects at the three-phase contact line, there is a famous paradox named Huh–Scriven paradox. It was first pointed out by Huh and Scriven [21] that there is a conflict between the moving contact line and the conventional no-slip boundary condition between a liquid and a solid. The interface meets the solid boundary at some finite contact angle y. Owing to the no-slip condition, the fluid at the bottom moves with constant velocity U and viscosity , while the flux through the cross section is zero. The energy dissipation per unit time and unit length of the contact line is obtained, E_ U 2 lnðL=0Þ, where L is an outer length scale like the radius of the spreading droplet. Stresses are unbounded at the contact line, and the force exerted by the liquid on the solid becomes infinite. The energy dissipation is logarithmically diverging, “not even Herakles could sink a solid” (Fig. 7) In reality, dynamic wetting occurs at a finite rate with changes in the wetted area and liquid shape. These processes are thermodynamically irreversible and therefore dissipative. But, the energy dissipation is
finite. There are two typical theories in identifying the effective channel of energy dissipation for small Capillary and Reynolds number. One of the two approaches is the hydrodynamic theory emphasizes energy dissipation caused by viscous flow within the wedge of liquid near the moving contact line. The other is the molecular kinetic theory emphasizes energy dissipation caused by of attachment (or detachment) of fluid molecules to (or from) the solid surface [22]. The Huh–Scriven paradox is raised from four ideal assumptions summarized as: incompressible Newtonian fluid, smooth solid surface, impenetrable liquid/ solid interface, and no-slip boundary. Hence, the typical methods proposed to relieve the dynamical singularity near the contact line are the precursor film, surface roughness, diffuse interface, and nonlinear slip boundary, aiming the four assumptions respectively, as shown in Fig. 8.
Examples of Application Nanostructures of Silicon Used as the Anode Material of Lithium Ion Batteries Rechargeable lithium ion batteries become the most suitable energy carrier for portable electro-equipments, electromobiles, and high performance computing, not only for the high-energy density and low cost, but also for the environmental needs for energy storage. Silicon
Surface Tension Effects of Nanostructures
2609
S
Surface Tension Effects of Nanostructures, Fig. 8 Possible mechanisms to solve the Huh and Scriven’s paradox
Precursor film
Diffuse layer
Slippage
γ
Roughness ΔP
Molecular kinetic theory
Ridge induced by γ⊥
Trapped nanobubbles
a Target Molecule
Surface Tension Effects of Nanostructures, Fig. 9 Schematic illustration of the deflection of a cantilever sensor due to the surface stress induced by surface tension effects [8, 17]. (a) The cantilever is functionalized on one side with the probe molecules. (b) After the target binding, the cantilever bends and the deflection can be measured
Probe Molecule
Cantilever Beam
b
is selected as a promising anode material of the lithium ion batteries due to the high-energy density (about 4,200 mAhg1). Nevertheless, a loss of electrical contact due to the fracture and crack of the bulk silicon induced by huge volume change (400%) in charging and discharging cycles hinders its applications. In recent years, nanostructured silicon is investigated as the material for electrode effectively circumvented the fragmentation, since surface tension effects of the nanostructured silicon on diffusion induced stresses would have much effect on the stress distribution [23].
Target Binding
Deflection, Δz
S Surface Tension Effects Induced the Deflection of the Cantilever Sensor Surface stresses scale linearly with dimension and surface to volume ratio increases as micro/nanostructure scale decreases. Therefore, surface stress can be very important in microstructures in the size domain of MEMS/NEMS. Surface stress has been used as an effective molecular recognition mechanism. For a microcantilever used in a bioactuator, some biomaterials and biomolecules, which are immobilized on the microcantilever surface, are used to convert chemical energy into mechanical energy. When excitations
S
2610
such as DNA hybridization and receptor-ligand binding are applied, the microcantilever generates a nanomechanical deflection response due to the surface tension effects, as shown in Fig. 9. Surface Tension Effects Used in the Application of Electrowetting One particularly promising application area for electrowetting is the manipulation of individual droplets in digital microfluidic systems. Applications range from “lab-on-a-chip” devices to adjustable lenses and new kinds of electronic displays. Besides, electrowetting has been used for the application to displays showing video content since the switching speed is very high (only a few milliseconds) [10].
Summary The surface tension effects become particularly predominant in nanostructures since the surface to volume ratio is sizable. In this entry, the surface tension effects that lead to the size-dependent mechanical properties of nanostructures are considered. Some key research findings related to this issue were listed and several examples of application were given, which are expected to be helpful to readers. Understanding and controlling these surface tension effects is a basic goal in the design and application of nanodevices.
Cross-References ▶ Disjoining Pressure and Capillary Adhesion ▶ Electrowetting ▶ Nanoscale Properties of Solid–Liquid Interfaces ▶ Surface Energy and Chemical Potential at Nanoscale ▶ Wetting Transitions
References 1. Bhushan, B. (ed.): Handbook of Nanotechnology. Springer, New York (2010) 2. Gibbs, J.W.: On the equilibrium of heterogeneous substances. In: Gibbs, J.W. (ed.) The Scientific Papers of J. Willard Gibbs. Volume 1: Thermodynamics, pp. 55–353. Dover, New York (1961)
Surface Tension Effects of Nanostructures 3. Cahn, J.W., Hilliard, J.E.: Free energy of a nonuniform system. I. Interfacial free energy. J. Chem. Phys. 28, 258–267 (1957) 4. Lu, W., Suo, Z.: Dynamics of nanoscale pattern formation of an epitaxial monolayer. J. Mech. Phys. Solids. 49, 1937–1950 (2001) 5. Butt, H.J., Graf, K., Kappl, M.: Physics and Chemistry of Interface. Wiley-VCH, Weinheim (2003) 6. Shuttleworth, R.: The surface tension of solids. Proc. Phys. Soc. A. 63, 444–457 (1950) 7. Herring, C.: Surface tension as a motivation for sintering. In: Kingston W.E. (ed.) The Physics of Powder Metallurgy. pp. 143–179, McGraw Hill, New York (1951) 8. Fritz, J., Baller, M.K., Lang, H.P., Rothuizen, H., Vettiger, P., Meyer, E., Guntherodt, H.J., Gerber, C., Gimzewski, J.K.: Translating biomolecular recognition into nanomechanics. Science 288, 316–318 (2000) 9. Stoney, G.: The tension of metallic films deposited by electrolysis. Proc. R. Soc. Lond. A. 82, 172–175 (1909) 10. Berthier, J.: Microdrops and Digital Microfluidics. William Andrew, Norwich (2008) 11. De Gennes, P.G.: Wetting: statics and dynamics. Rev. Mod. Phys 57, 827–863 (1985) 12. De Gennes, P.G., Brochard-Wyart, F., Quere, D.: Capillarity and Wetting Phenomena. Springer, New York (2004) 13. Yuan, Q.Z., Zhao, Y.P.: Precursor film in dynamic wetting, electrowetting and electro-elasto-capillarity. Phys. Rev. Lett. 104, 246101 (2010) 14. Gurtin, M.E., Murdoch, A.I.: A continuum theory of elastic material surfaces. Arch. Ration. Mech. Anal. 57, 291–323 (1975) 15. Wang, Z.Q., Zhao, Y.P., Huang, Z.P.: The effects of surface tension on the elastic properties of nano structures. Int. J. Eng. Sci. 48, 140–150 (2010) 16. Zhang, Y., Ren, Q., Zhao, Y.P.: Modelling analysis of surface stress on a rectangular cantilever beam. J. Phys. D: Appl. Phys. 37, 2140–2145 (2004) 17. Ren, Q., Zhao, Y.P.: A nanomechanical device based on light-driven proton pumps. Nanotechnol. 17, 1778–1785 (2006) 18. Guo, J.G., Zhao, Y.P.: The size-dependent bending elastic properties of nanobeams with surface effects. Nanotechnol. 18, 295701 (2007) 19. Shanahan, M.E.R., Carre´, A.: Nanometric solid deformation of soft materials in capillary phenomena. In: Rosoff, M. (ed.) Nano-Surface Chemistry. CRC Press, New York (2001) 20. Yu, Y.S., Zhao, Y.P.: Elastic deformation of soft membrane with finite thickness induced by a sessile liquid droplet. J. Colloid Interf. Sci. 339, 489–494 (2009) 21. Huh, C., Scriven, L.: Hydrodynamic model of steady movement of a solid/liquid/fluid contact line. J. Colloid Interface Sci. 35, 85–101 (1971) 22. Wang, F.C., Zhao, Y.P.: Slip boundary conditions based on molecular kinetic theory: The critical shear stress and the energy dissipation at the liquid-solid interface. Soft Matter 7, 8628–8634 (2011) 23. Cheng, Y.T., Verbrugge, M.W.: Evolution of stress within a spherical insertion electrode particle under potentiostatic and galvanostatic operation. J. Power Sources 190, 453–460 (2009)
Surface-Modified Microfluidics and Nanofluidics
Surface Tension–Driven Flow ▶ Capillary Flow
Surface Tension–Powered Self-Assembly ▶ Capillary Origami
Surface-Modified Microfluidics and Nanofluidics Shaurya Prakash Department of Mechanical and Aerospace Engineering, The Ohio State University, Columbus, OH, USA
Synonyms Heterogeneous walls; Lab-on-a-chip (LOC); Micrototal analytical systems (m-TAS)
Definition Fluid phenomena in micrometer- or nanometer-sized channels are governed by coupled principles from fluid mechanics, surface chemistry, electrochemistry, and electrostatics. The smaller length scales compared to traditional fluid mechanics provide several new and interesting phenomena due to significant enhancement of the surface-area-to-volume ratio, which makes surface-mediated flows important to the overall field of microfluidics and nanofluidics.
Introduction to Surfaces in Microfluidics and Nanofluidics Microfluidic and nanofluidic devices and systems are characterized by high surface-area-to-volume (SA/V) ratio. For example, a rectangular cross-section channel with 100 mm width and 100 nm depth and 1 cm length
2611
S
will have a SA/V ratio on the order of 106 m1. In fact, operational devices incorporating components with SA/V ratio on the order of 109 m1 have already been reported [1]. Consequently, the influence of device and system walls in affecting phenomena within confined spaces can no longer be ignored since the walls (i.e., surfaces) of the devices and systems interact extensively and directly with the species contained within these devices. Therefore, the surface of interest here is defined as an interface or thin region in space (often only a few nm in extent) that influences transport and reaction phenomena in its vicinity. In this surface region, properties such as chemical composition, refractive index, mechanical strength, and conductivity can significantly differ from the bulk material of the underlying substrate.
Scaling of Surface Forces in Microchannels and Nanochannels One non-dimensional parameter often used to characterize flow regimes, is the ratio of the inertial forces of the fluid to the viscous (or frictional) forces, i.e. the Reynolds number, Re, and is given by
Re ¼
rVLc ; m
(1)
where r is the fluid density, m is the fluid viscosity, and V is the fluid velocity. As seen from Eq. 1, Re is directly proportional to the characteristic length, Lc, of the flow. With Lc decreasing (and SA/V increasing), the inertial forces decrease, and as a result the Re becomes smaller. The direct consequence is that for given flow parameters of fixed velocity and fluid type, a decreasing Lc implies increasing influence of viscous forces in contrast to the inertial forces. In most microfluidics and nanofluidics applications, a particle (ion, colloid, biomolecule, AFM probe tip, etc.) will interact with the channel walls. Considering three important forces, electrostatic or Coulombic, Fel, van der Waals, Fvdw, and hydrodynamic forces, Fh, will provide a better insight into surface-particle interactions and subsequent phenomena. For an infinite flat surface separated by a distance D from a particle of radius R, Fel is given by [2]
S
S Fel ¼
2612
Surface-Modified Microfluidics and Nanofluidics
2pRLD D 2D þ s2S þ s2P exp 2sS sP exp ee0 lD lD
(2) where lD is the Debye length and ss and sp are the surface charge densities of sample and particle, respectively. Fvdw is given by [2] Fvdw ¼
AH R ; 6D2
(3)
where AH is the Hamaker constant, which serves as an indicator of the interaction between the particle and the sample surface, and Fh is given by [2] Fh ¼ bs V ¼ f
6pmR2 V ; D
(4)
where V is the particle velocity, bs is the hydrodynamic damping coefficient, and m is the viscosity of the fluid. The coefficient f * is related to the boundary slip properties. If the no-slip assumption holds true at the solid/ water interface, then f * ¼ 1 and if slip exists then f * < 1. As an aside, it should be noted that Eq. 4 is often used in literature for determining slip lengths using a colloidal AFM probe tip. As the operational length scales (here, characteristic length scale is separation distance, D) decrease, it can be seen from Eqs. 2–4 that contributions for Fel displays an exponential dependence, Fvdw an inverse quadratic dependence, and Fh an inverse linear dependence on the separation distance. Therefore, consideration of each force term at the microscale, and even more so at the nanoscale is essential toward completely describing importance of the wall-species interactions. Most microfluidic and nanofluidic devices with liquid flows are driven by electric fields and fall under the category of electrokinetic flows. The main reason for using applied voltages to generate electric fields for driving flows is the scaling of pressure forces. For example, for a circular nanochannel with laminar, incompressible, Poiseuille flow with water as the working fluid, the necessary pressure drop across a 100 mm long channel which is 1 nm in diameter for only an attoliter (10–18 m3) per second would be greater than 3 GPa, which is impractical for any device. Electrokinetic flows can sustain higher flow rates through nanometer channels without excessive pressures [1].
Methods for Surface Modification Surface modification methods can be divided in two broad categories: physical and chemical methods [3]. Physical methods, in most cases, do not change the chemical composition of the surface but may change the surface roughness, grain sizes and grain boundaries, and faceting. Physical methods often use lasers, plasmas, temperature, ion beams, ball-milling, and polishing or grinding, to alter the surface state of a material of interest. While, the main intent with physical modification methods is to not alter the chemical composition of the material, in some cases, physical surface modification methods can lead to changes in the chemical composition of the surface due to removal or addition of material or chemical reactions on surfaces, as in the case of selective or ion-beam sputtering, or by selective cross-linking in presence of plasmas. Temperature gradients and thermal treatments have been used to change surface roughness, grain sizes, and grain boundaries, and create nanoscale features, facets, textures, and nanoparticles on ceramics, metals, polymers, and semiconductors. Thermal treatments in the presence of gases such as oxygen or water vapor can cause creation of steps or induce other forms of nanostructures. For example, as was discussed in a recent review [3] that for crystalline a-Al2O3 surfaces heat-treated at 1,500 C in Ar/O2 and H2/He/O2 led to step formation and roughening as quantified through AFM topology images. Chemical methods introduce a change in the chemical composition at the surface by introducing chemical properties (e.g., surface charge density or surface energy) different from the bulk material. Among the methods for chemical modification, formation of surface layers, either covalently bonded or physisorbed, has been most common. Other chemical methods include treatment with UV light and reactive plasmas. Modification schemes are governed by a wide range of parameters including sample type (polymers, metals, ceramics, etc.), stability to treatment conditions (e.g., thermal or structural), and eventual applications. For example, polymeric surfaces are often modified by photochemical methods of which UV irradiation in air, other reactive atmospheres such as ozone combined with lasers, and grafting surface layers is fairly common. Use of reactive plasmas has been gaining popularity due the wide compatibility of
Surface-Modified Microfluidics and Nanofluidics Surface-Modified Microfluidics and Nanofluidics, Fig. 1 Schematic representation of the electric double layer. The surface charge and potentials are depicted based on the theories developed over the past century (Figure from [1])
2613 Chi Potential, FIHP
Surface or Volta Potential, Fo
S
Zeta Potential, z
FOHP
Hydrated Anions
Hydrated Cations
Substrate
Stern Layer
Diffuse Layer Shear Plane
Inner Helmholtz Plane (IHP)
materials and integration to microfabrication processes for device development. Many gas plasmas have been used such as air, oxygen, water vapor, ammonia, and argon for modification of polymer surfaces. Plasma modification processes generate new chemical species on polymer surfaces. The new chemical species can arise due to surface reactions with reactive gases or due to physical sputtering (such as with Ar plasma) caused by active gas-phase species. These new surface chemical species can provide an anchor for attaching a series of different molecules that display different properties from the underlying bulk polymer. For example, preferential hydroxylation of poly(methyl methacrylate) or PMMA surfaces by use of a water vapor plasma as opposed to an oxygen plasma has been used to activate surfaces toward trichlorosilane modification and subsequent “click” chemistries to form surface scaffolds of desired functionalities. The “click” chemistry method has also been used to modify glass channels for systematic control over electroosmotic flow velocity [4].
Outer Helmholtz Plane (OHP)
Applications In this section, applications utilizing modified surfaces are presented along with short discussions of the underlying physics. Chemically modified surfaces are influenced by changes to surface charge density, as discussed above. Any charged surface in contact with a liquid forms an electric double layer (EDL). A schematic diagram illustrating the classic EDL structure is shown in Fig. 1, depicting the positions to which different characteristic surface charge–related potentials are referenced within the EDL. One direct consequence of surface modification is a change in the interaction forces between the surface and the surroundings. Specifically for microfluidics and nanofluidics, changes to surface charge density, surface roughness, and surface energy alter the electrostatic and van der Waals forces that determine flow (ionic and fluid) phenomena. Following the scaling of pressure forces discussed above, electric fields and capillary forces are the two most common tools used
S
S
2614
Surface-Modified Microfluidics and Nanofluidics
to fill microchannels and nanochannels with liquids. For capillary forces, consider the Washburn equation, which provides one approach to quantifying the rate of channel filling, dlf ag cos yc ¼ ; dt 4mlf
(5)
u
u
y
b
where lf is the fill length, a is the radius (or characteristic length) of the channel, yc is the contact angle between the fluid wall and the fluid, g is the interfacial (or surface) tension, and m is the fluid viscosity. Equation 5 shows the dependence of the capillary (or micro/nanochannel) filling as a function of surface properties governed by yc. Chemical surface modification can easily affect a change to the surface wettability and therefore change yc. Therefore, for a given liquid such as water in a micro- or nanochannel the rate of filling decreases as the surface becomes progressively more hydrophobic. Given the importance of walls for microfluidics and nanofluidics, a contrast to the boundary conditions with traditional fluid mechanics should be considered. For example, macroscale fluid mechanics assumes a no-slip condition at the wall, which implies that the velocity of the fluid layer in contact with the walls is the same as that of the walls. However, in the early 1800s, Navier postulated that fluid slip may be possible and suggested the slip boundary condition with uw, the velocity at the wall (tangential component), can be expressed by the relationship in Eq. 6: uw ¼ b
@ub @y
(6)
where b is the slip length, ub is the bulk velocity, and y the axis perpendicular to the wall. Figure 2 shows a conceptual schematic for defining the physical effect of slip flow. Conceptually, one can imagine the slip length as the additional distance the wall must be extended to where the fluid velocity would be zero. Literature has noted that there are three types of slip [5]. First is molecular or intrinsic slip which occurs when molecular forces (e.g., van der Waals) are balanced by viscous forces. Intrinsic slip is usually observed at very high shear rates (e.g., shear rate of 1012 s1 for water). Second is apparent slip in which there is a finite distance between the slip and no-slip planes. Electrokinetically driven microfluidics and
Surface-Modified Microfluidics and Nanofluidics, Fig. 2 Conceptual schematic for the slip boundary condition. The left panel shows the no-slip boundary condition for a stationary wall, with the wall velocity and the tangential component of fluid velocity being zero. The right panel shows a case for slip flow with a finite slip length (Figure based on [5, 6], and assistance from Mr. M. Hansen in drawing the figure is acknowledged)
nanofluidics follows this type of slip with the slip plane (location where z potential is defined in Fig. 1) being at a finite distance from the physical channel wall. Third is the effective slip which refers to the estimate of either intrinsic or apparent slip by averaging a measurement over the length scale of the experiment. The main factors that affect slip are roughness of the surface and wettability, which presents a measure of the surface energy of the surface. Other possible factors are gases trapped in the fluid, nanobubbles, shear rates of fluid at the surface, and absorbates in the fluid [5–7], all of which can be influenced by surface modification. Recently, it was also demonstrated that surface charge can alter the drainage velocity for confined fluids indicating possibly a role for surface charge in mediating slip flow [2]. Additional use of slip phenomena has been in developing drag reduction configurations both for laminar and turbulent flows, self-cleaning superhydrophobic surfaces [8], and enhanced mixing in microfluidic devices [9]. Continuing with a discussion of applications relevant to microfluidics and nanofluidics, valving (or metering of fluids) is one critical operation for sample manipulation for lab-on-chip devices. Valving by surface-mediated flows is a passive method due to a fixed surface state and can be achieved by the tendency of the fluid to favor a channel direction based on the wettability or the surface charge of the channel. Examples include use of hydrophobic-hydrophilic interfaces to generate boundaries for directing water [10] and use of chemically modified surfaces for exploiting
Synthesis of Carbon Nanotubes
differences in surface charge for selectively driving electrokinetically driven flows in microchannels [11].
Summary
2615
S
10. Zhao, B., Moore, J.S., Beebe, D.J.: Surface-directed liquid flow inside microchannels. Science 291(5506), 1023–1026 (2001) 11. Prakash, S., Karacor, M.B.: Surface mediated flows in glass nanofluidic channels. In: Solid state sensors, actuators, and microsystems workshop. Transducer Research Foundation, Hilton Head Island, SC (2010)
Microfluidic and nanofluidic phenomena are governed by large SA/V ratios. Therefore, interactions between fluids, species, and the channel walls are critical. Toward the development of devices and systems enabling applications and related fluid control, the ability to systematically modify and control wall properties provides a powerful tool for scientists and engineers.
Surface-Plasmon-Enhanced Solar Energy Conversion
Cross-References
Synthesis of Carbon Nanotubes
▶ AC Electroosmosis: Basics and Lab-on-a-Chip Applications ▶ Applications of Nanofluidics ▶ Electrokinetic Fluid Flow in Nanostructures
Simon J. Henley, Jose´ V. Anguita and S. Ravi P. Silva Nano Electronics Center, Advanced Technology Institute, University of Surrey, Guildford, Surrey, UK
References 1. Prakash, S., Piruska, A., Gatimu, E.N., Bohn, P.W., Sweedler, J.V., Shannon, M.A.: Nanofluidics: systems and applications. IEEE Sens. J. 8, 441–450 (2008) 2. Wu, Y., Misra, S., Karacor, M.B., Prakash, S., Shannon, M.A.: Dynamic response of AFM cantilevers to dissimilar functionalized silica surfaces in aqueous electrolyte solutions. Langmuir 26(22), 16963–16972 (2010) 3. Prakash, S., Karacor, M.B., Banerjee, S.: Surface modification in microsystems and nanosystems. Surf. Sci. Rep. 64(7), 233–254 (2009) 4. Prakash, S., Long, T.M., Selby, J.C., Moore, J.S., Shannon, M.A., Shannon, M.A.: ‘Click’ modification of silica surfaces and glass microfluidic channels. Anal. Chem. 79(4), 1661–1667 (2007) 5. Lauga, E., Stone, H.A., Brenner, M.P.: Microfluidics: the no-slip boundary condition. In: Tropea, C., Yarin, A., Foss, J.F. (eds.) Handbook of Experimental Fluid Dynamics, pp. 1219–1240. Springer, New York (2007) 6. Granick, S., Zhu, Y., Lee, H.: Slippery questions about complex fluids flowing past solids. Nat. Mater. 2, 221–227 (2003) 7. Neto, C., Evans, D.R., Bonaccurso, E., Butt, H.-J., Craig, V.S. J.: Boundary slip in Newtonian liquids: a review of experimental studies. Rep. Progr. Phys. 68, 2859–2897 (2005) 8. Bhushan, B., Jung, Y.C.: Natural and artificial surfaces for superhydrophobicity, self-cleaning, low adhesion, and drag reduction. Prog. Mater. Sci. 56, 1–108 (2011) 9. Rothstein, J.P.: Slip on superhydrophobic surfaces. Annu. Rev. Fluid Mech. 42, 89–109 (2010)
▶ Plasmonic Structures for Solar Energy Harvesting
Synonyms Growth of carbon nanotubes; Growth of CNTs
Definition Carbon nanotubes (CNTs) are allotropes of graphitic carbon with a cylindrical structure and diameters 1,200 nm lmaxtr: – lmaxlong: > 1,200 nm 626 and 950
Cubes [14] Cubes [19]
Seeded growth Polyol
90 150
555 621
Octahedra [20]
Polyol
20–400
530–1,100
Decahedra [21] Decahedra [22]
Seeded growth Polyol
36–65 48–88
600–700 600–700
Icosahedra [19]
Polyol
230
613 and 950
Triangles [23] Hexagons [16]
Seeded growth Oligoamine surfactant Seeded growth Nonaqueous seeded growth
90–210 20–15000 350 60–80
Stars [14] Stars [17]
Synthesis medium H2O
Reduction agent Capping agent Ascorbic acid CTAB
H2O
Ascorbic acid
CTAB
Oleylamine
Oleylamine
Oleylamine
Ethylene glycol
PVP
Ascorbic acid Ethylene glycol
CTAB PVP
Ethylene glycol
PDDA
PVP Diethylene glycol Ethylene glycol
PVP PVP
750–1,300 550–900 broad
Ethylene glycol H2O Ethylene glycol Ethylene glycol H2O/DMF Diethylene glycol Ethylene glycol H2O/DMF H2O
Broad Broad
H2O DMF
Ascorbic acid Oligoamine surfactant Ascorbic acid PVP
CTAB Oligoamine surfactant CTAB PVP
PVP
1.2 A
a
Absorption (a.u.)
1.0
b
B
0.8 C
0.6 0.4 0.2 0.0
10μm
20nm
400
500
600
700
800
900
Wavelength (nm)
Synthesis of Gold Nanoparticles, Fig. 4 Influence of concentration of C18N3 on the size of gold nanoplates at standard conditions. SEM and TEM images of gold nano- and microplates obtained at different concentrations of C18N3, 2 mM (a) and
0.1 mM (b). On the right it is shown the UV-Vis absorption spectra of gold nanospheres (A), gold decahedrons (B), and gold microplates (C) (Reprinted with permission from [16]. Copyright 2010 American Chemical Society)
a more facile synthesis procedure for the fabrication of gold nano- and microplates was reported by Lin et al. [16]. They used a single tree-type multiple-head surfactant, bis (amidoethyl-carbamoylethyl)
octadecylamine (C18N3), which functions as both the reducing and capping agent in the reaction system. The triangle and hexagonal plate, polyhedron, and sphere morphology of the gold nanoparticles could be easily
Synthesis of Gold Nanoparticles
a
2629
b
12 L=15 nm
a=17.5 nm
10 Norm. Cross Section
S
a a=17.5 nm 20 nm
8 6
L=15 nm a a=17.5 nm
4 2 0 400
500
600
700
800
900
1000
Wavelength (nm)
Synthesis of Gold Nanoparticles, Fig. 5 Representative transmission electron microscopy (TEM) image of Au nanostars (a), experimental (solid line) and calculated (dashed line)
absorption spectra for gold nanostars; the inset shows a TEM image of a single nanostar (b) (Reprinted with permission from [17]. Copyright IOP Publishing 2008)
controlled simply by changing the molar ratio of C18N3/HAuCl4 (Fig. 4).
these particles is that each tip is a single crystal, so the challenge remains to understand the underlying growth mechanism of the selective tip growth on the seed particle’s surface. In summary, the most relevant synthetic approaches mentioned above for the preparation of different anisotropic nanoparticles are presented in Table 1. This table offers the opportunity to take a look at the wide range of synthetic possibilities for the production of anisotropic nanoparticles, but at the same time gives a good idea of the physical and chemical properties as well as limitations of these synthetic approaches for the production of particular morphologies with specific sizes. That is why further synthetic research efforts are still needed.
Branched NPs Coming from symmetric isotropic and anisotropic gold nanoparticles, another important group of gold particles are asymmetric nanoparticles, the so-called branched nanoparticles or nanostars (Fig. 5) [6]. Although these obviously more complex structures are not yet well understood, they generate interest because of their sharp edges and the correspondingly high localization of surface plasmon modes. This makes them potential candidates for a number of applications, including SERS-based detection and analysis, as well as the case of nanoplates. Sau et al. proposed the synthesis of branched nanoparticles via the seededgrowth method in the presence of silver ions by varying the seed-to-gold salt ratio and the amount of reducing agent in order to increase the rate of gold ion reduction and thereby induce branching (see Table 1). However, an easier way to synthesize gold nanostars comprises the use of a nonaqueous approach [6, 17]. Branched “nanostars” were synthesized in a concentrated solution of PVP in DMF at room temperature, where PVP acts as a stabilizer and reductant. Nanostars were not only obtained in a seed-mediated growth process, but could also be synthesized in a seedless growth process. The fascinating aspect of
Acknowledgments Authors acknowledge the financial support of the Spanish Ministerio de Ciencia e Innovacio´n (MAT2008-06126), Xunta de Galicia (INCITE09209101PR, INCITE08PXIB209007PR and 2008/077), and EU (METACHEM, grant number CP-FP 228762–2).
References 1. Eustis, S.E.-S., El-Sayed, M.A.: Why gold nanoparticles are more precious than pretty gold: noble metal surface plasmon resonance and its enhancement of the radiative and nonradiative properties of nanocrystals of different shapes. Chem. Soc. Rev. 35(3), 209–217 (2006)
S
S
2630
2. Astruc, D., Daniel, M.C.: Gold nanoparticles: assembly, supramolecular chemistry, quantum-size-related properties, and applications toward biology, catalysis, and nanotechnology. Chem. Rev. 104(1), 293–346 (2004) 3. Perez-Juste, J., et al.: Gold nanorods: synthesis, characterization and applications. Coord. Chem. Rev. 249(17–18), 1870–1901 (2005) 4. Grzelczak, M., et al.: Shape control in gold nanoparticle synthesis. Chem. Soc. Rev. 37(9), 1783–1791 (2008) 5. Treguer-Delapierre, M., et al.: Synthesis of non-spherical gold nanoparticles. Gold Bull. 41(2), 195–207 (2008) 6. Guerrero-Martı´nez, A., et al.: Nanostars shine bright for you: colloidal synthesis, properties and applications of branched metallic nanoparticles. Curr. Opin. Colloid Interf. Sci. 16(2), 118–127 (2011) 7. Chow, P.E. (ed.):Gold Nanoparticles: Properties, Characterization and Fabrication Nanotechnology Science and Technology. Nova Science Publishers, New York (2010) 8. Feldheim, D.L. (ed.): Metal Nanoparticles: Synthesis Characterization and Applications. CRC Press, Baco Raton (2001) 9. Jana, N.R., Gearheart, L., Murphy, C.J.: Evidence for seedmediated nucleation in the chemical reduction of gold salts to gold nanoparticles. Chem. Mater. 13(7), 2313–2322 (2001) 10. Oldenburg, S.J., et al.: Nanoengineering of optical resonances. Chem. Phys. Lett. 288(2–4), 243–247 (1998) 11. Kah, J.C.Y., et al.: Synthesis of gold nanoshells based on the deposition-precipitation process. Gold Bull. 41(1), 23–36 (2008) 12. Kim, F., et al.: Chemical synthesis of gold nanowires in acidic solutions. J. Am. Chem. Soc. 130(44), 14442–14443 (2008) 13. Pazos-Perez, N., et al.: Synthesis of flexible, ultrathin gold nanowires in organic media. Langmuir 24(17), 9855–9860 (2008) 14. Sau, T.K., Murphy, C.J.: Room temperature, high-yield synthesis of multiple shapes of gold nanoparticles in aqueous solution. J. Am. Chem. Soc. 126(28), 8648–8649 (2004) 15. Seo, D., et al.: Directed surface overgrowth and morphology control of polyhedral gold nanocrystals. Angew. Chem. Int. Edit. 47(4), 763–767 (2008) 16. Lin, G.H., et al.: A simple synthesis method for gold nano- and microplate fabrication using a tree-type multiple-amine head surfactant. Cryst. Growth Des. 10(3), 1118–1123 (2010) 17. Kumar, P.S., et al.: High-yield synthesis and optical response of gold nanostars. Nanotechnology 19(1), 015606 (2008) 18. Liu, M.Z., Guyot-Sionnest, P.: Mechanism of silver(I)assisted growth of gold nanorods and bipyramids. J. Phys. Chem. B 109(47), 22192–22200 (2005) 19. Kim, F., et al.: Platonic gold nanocrystals. Angew. Chem. Int. Edit. 43(28), 3673–3677 (2004) 20. Li, C.C., et al.: A facile polyol route to uniform gold octahedra with tailorable size and their optical properties. ACS Nano 2(9), 1760–1769 (2008) 21. Sanchez-Iglesias, A., et al.: Synthesis and optical properties of gold nanodecahedra with size control. Adv. Mater. 18(19), 2529–2534 (2006) 22. Seo, D., et al.: Shape adjustment between multiply twinned and single-crystalline polyhedral gold nanocrystals: decahedra, icosahedra, and truncated tetrahedra. J. Phys. Chem. C 112(7), 2469–2475 (2008)
Synthesis of Graphene 23. Millstone, J.E., et al.: Observation of a quadrupole plasmon mode for a colloidal solution of gold nanoprisms. J. Am. Chem. Soc. 127(15), 5312–5313 (2005) 24. Rodriguez-Fernandez, J., et al.: Seeded growth of submicron Au colloids with quadrupole plasmon resonance modes. Langmuir 22(16), 7007–7010 (2006) 25. Perez-Juste, J., Correa-Duarte, M.A., Liz-Marzan, L.M.: Silica gels with tailored, gold nanorod-driven optical functionalities. Appl. Surf. Sci. 226(1–3), 137–143 (2004)
Synthesis of Graphene Swastik Kar1 and Saikat Talapatra2 1 Department of Physics, Northeastern University, Boston, MA, USA 2 Department of Physics, Southern Illinois University Carbondale, Carbondale, IL, USA
Synonyms Fabrication of graphene or graphene fabrication; Graphene synthesis; How to synthesize/make/ manufacture/fabricate/produce graphene; Making graphene; Manufacturing graphene or graphene manufacturing; Production of graphene or graphene production
Definition ▶ Graphene (or monolayer graphene or single-layer graphene) is a single-atom thick, quasi-infinite, sp2-hybridized allotrope of carbon in which the atoms are packed in a planar honeycomb crystal lattice (see Fig. 1). It can be visualized as a single sheet of graphite and its lattice structure is related to those of ▶ fullerenes for drug delivery and ▶ carbon nanotubes. Fewlayered graphene, multilayered graphene, or multigraphene refers to a few layers of graphene stacked (with weak interlayer attraction) in a manner similar to graphite. Figure 1a, b are schematic representations of a sheet of single-layer graphene.
Overview Compiling a review on the diverse methods available for the synthesis of graphene is a daunting task, chiefly
Synthesis of Graphene Synthesis of Graphene, Fig. 1 Schematic representation of carbon atoms in a graphene lattice: (a) a balland-stick model and (b) a space-filling model. The arrows indicate that the sheets are quasi-infinitely extended in all directions. The edge carbon atoms are usually hydrogen terminated
2631
a
due to the explosive intensity at which this field has grown over the past few years. Numerous techniques for graphene synthesis have been developed, based on both conventional and innovative techniques, and the graphene thus produced can have vastly different properties. Quite often, this diversity stems from a need for the production of graphene that is tailored for a broad variety of experiment-specific and application-specific research. It is quite impossible to enlist all these methods into a few categories, and this article is not an attempt to individually address all of them. In general, this work attempts to track and identify some of the most popular, novel, and/or promising methods for graphene synthesis. An attempt has been made to categorize these methods into groups. Since the search for new and/or improved methods for graphene synthesis is in its ascent, the material presented here will not reflect published reports beyond May of 2011. In most cases, generic descriptions of the synthesis conditions have been presented. The aim is to provide the reader with a starting point rather than a complete recipe in each case. It should be noted that experimental conditions and parameters presented here represent typical synthesis conditions as reported in literature, since what constitutes the “best practice” is both subjective and continuously evolving. It should also be noted that this review does not deal with the many electronic, optical, thermal, mechanical or other fascinating properties of graphene, for which excellent review articles already exist in literature.
Historical Perspective The concept of graphene has been around for quite a long time, and the descriptive presence of this
S
b
material appear in a variety of reports in a number of languages. For example, “very thin layers of graphite” including “single foils of carbon” were reported as early as 1962, prepared through a partial reduction of graphite oxide [1]. Monolayer graphite/carbon was identified to form at elevated temperatures on surfaces of Pt in 1969 [2], C-doped Ni (111) crystals in 1979 [3], and other transition metal surfaces [4]. The term “graphene” was already in use in the early 1990s [5], to describe not only monolayer graphite, but also the structure of carbon fibers and fullerenes. An earnest endeavor to obtain isolated high-quality graphene from graphite started in 1999, using mechanical exfoliation of patterned graphite pillars [6, 7]. This approach matured in 2004 [8], the same year that synthesis of graphene was also reported by a high-temperature thermal annealing of SiC [9]. These two methods opened up the floodgate of graphene research and catapulted it into becoming the millennium material. In the following paragraphs, a flavor of these and some of the other popular graphene synthesis methods that have been developed since then has been captured.
Recent Advances in Graphene Synthesis Micromechanical Exfoliation of Graphene from Graphite Strictly not a synthesis process, the micromechanical exfoliation of graphene (both monolayer and few-layer graphene) from highly oriented pyrolytic graphite (HOPG), as reported by Novoselov et al. enabled, for the first time, a direct measurement of its unique 2D electronic properties. This ended the decades-long debate whether pure monatomic two-dimensional crystals can indeed be stable in an isolated state
S
S
2632
Synthesis of Graphene
a Preparation of ultrathin graphite: scotch tape (sticky side up) 1.
b
Transfer of graphene:
5. thin piece of HOPG
press gently with a gloved finger SiO2/Si substrate flat surface
fold/unfold 2.
tape (sticky side down)
c
d
repeat at different locations 3.
after repeating many times: 4.
20 μm
Synthesis of Graphene, Fig. 2 A simple recipe for obtaining single and few-layered graphene crystals using the micromechanical cleavage of graphite. (a) 1. Attach a very thin piece of HOPG onto a scotch tape. 2–4. Fold and unfold as shown to obtain ultrathin (almost invisible) flakes of graphite on the tape. (b) Select a region with very thin flakes and transfer on a suitable substrate (e.g. SiO2/Si) as shown. Carefully remove
the tape from the substrate. (c) Typical image of a few-layered flake with monolayer extensions as seen under an optical microscope (Reprinted with permission from Ref. [13]; copyright # 2010 Elsevier). (d) A typical AFM image of graphene flakes (scale bar ¼ 1 mm) (Reprinted with permission from Ref. [12]; copyright # 2005 National Academy of Sciences, U.S.A.)
[10, 11]. The few tens-of-microns sized graphene and few-layered graphene flakes could be easily transferred onto SiO2/Si and other substrates, and enabled researchers to perform a wide variety of electronic, optical, thermal and mechanical property measurements. In their first report [8], Novoselov et al. described an elaborate process that started from millimeter-thick platelets of HOPG, and involved patterning, etching, photoresist, an adhesive-tape peel off, and a subsequent ultrasonic rinse. A simplified prescription for obtaining graphite was reported by the same group a year later, whereby a freshly cleaved HOPG surface was mechanically rubbed against any solid surface. This invariably left behind debris of crystallites on the surface, and among them, one can usually find single- and few-layered graphene [12]. Owing to the simplicity of this method, a variety of alternatives have become popular, and Fig. 2a, b describes a protocol that simply uses an adhesive tape, which is favored by many. This method has become known as the scotch-tape method. As a receiving substrate, it is helpful to use a Si wafer that has 300-nm thick oxide layer: graphene itself is nearly transparent, but using this oxide thickness enables one to “see” graphene even using an optical microscope due to the suitable amount of
interferometric phase contrasting, as seen in Fig. 2c [13]. Figure 2d is an ▶ atomic force microscopy (AFM) image of a typical flake of graphene, showing the graphene layer thicknesses in flat and folded regions [12]. Epitaxial Growth of Graphene on Silicon Carbide In 2004, another technique for graphene synthesis was reported by Berger et al. [9]. In this method, graphene crystals were produced on the Si-terminated (0001) face of single-crystal 6H-SiC by thermal desorption of Si. The SiC surfaces were initially prepared by oxidation or H2 etching. The oxide was then removed by electron bombardment-assisted heating in ultrahigh vacuum (1010Torr) to 1,000 C (repeating the oxidation/deoxidation cycle improves the surface quality, and the best initial surface quality was obtained with H2 etching). The deoxidized samples were heated to temperatures ranging from 1,250 C to 1,450 C for 1–20 min; during which time, thin graphite layers were formed. Figure 3 shows a scanning tunneling microscope (STM) image of graphene produced this way by heating SiC for 8 min at 1,400 C. Over the years, this process has been fine-tuned so as to enable one to controllably produce monolayer graphene on SiC substrates [14]. It turns out that graphene grown
Synthesis of Graphene
2633
S
methods for mass production, often using liquidphase exfoliation of graphene from graphite, while the other is headed toward wafer-scale production of single- or few-layered graphene, targeted toward electrical, electromechanical, and optoelectronic applications.
Synthesis of Graphene, Fig. 3 Atomically resolved STM image of a graphene sample grown on SiC(0001) at 1,400 C for 8 min (Reprinted with permission from Ref. [9]; copyright # 2004 American Chemical Society)
from the SiC(000 1) (C-terminated) surface are of higher quality than those grown on SiC(0001) [15]. Growth on the C face renders graphene with domain sizes more than three times larger than those grown on the Si face, and with significantly reduced disorder. In 2009, it was reported that an ex situ graphitization of Si-terminated SiC (0001) in an argon atmosphere of about 1 bar produces monolayer graphene films with much larger domain sizes than those previously attained. This method for the direct synthesis of large-domain graphene (10s of microns in length, with micron sized widths that were limited by the step-sizes of the underlying substrate) on an insulating underlayer formed an important step toward realizing graphene-based nanoelectronics on a semiconductor-processable substrate. In the initial years, most samples were produced using one of the two above-mentioned methods, which produced high-quality samples of graphene appropriate for fundamental studies such as electricand magnetic-field-modulated electronic transport, ARPES studies, Raman studies, and a variety of related experiments. At the same time, as it became evident that there are several other important properties of graphene including their gigantic specific surface area, extreme mechanical strength values, and other thermal and optical properties, several other methods have been developed for large-scale production of graphene. The large-scale production can be categorized into two directions. One route is to develop
Liquid-Phase Exfoliation Reduced Graphene Oxide and Chemically Modified Graphene A significant amount of effort has gone into obtaining graphene from the exfoliation of graphite oxide into graphene oxide (GO), followed by a reduction step [16]. The resultant material is more appropriately termed reduced graphene oxide (rGO), and it belongs to the broader category of chemically modified graphenes. Typically, graphite is oxidized using techniques based on the Hummer’s method [17]. GO, as produced by the Hummers’ method, is composed of functionalized graphene sheets decorated by strongly bound oxidative debris, which acts as a surfactant to stabilize aqueous GO suspensions [18]. The oxide can hence be dispersed in water using an ultrasonic bath till it becomes a clear-colored liquid with no visible particulates. The parent graphite oxide and the dispersed graphene oxide are both insulators, and hence are unsuitable for any applications where electrical conductivity is important. One approach to overcome this problem is a liquid-or gas-phase reduction of the oxidized carbon. Among reducing agents, the use of hydrazine hydrate has become quite popular, and this can be added to the aqueous GO dispersion and heated in an oil bath at 100 C for a day, during which time, the reduced graphene oxide gradually precipitates out as a black solid. This precipitate is a high-surface-area material which can be separated out using a simple vacuum filtration technique that leaves behind a thick film of rGO. Such films can be transferred onto other substrates in the form of large-area ultrathin films of reduced graphene oxide, which are useful for the development of transparent and flexible electronic materials [19] (see also ▶ Flexible Electronics) and other applications [20, 21]. Although the reduction process restores its conductivity by several orders of magnitude compared to that of GO [22], the oxidation of graphene seems to be quite robust and can only be partially removed. Solid-state 13 C NMR spectroscopy of these materials suggests that the sp2-bonded carbon network of graphite is strongly
S
S
2634
disrupted, and a significant fraction of this carbon network is bonded to hydroxyl groups or participates in epoxide groups [23], with some carboxylic or carbonyl groups present at the edges. Further works have shown the presence of five- and six-membered-ring lactols [24]. As a result, this material has a poor electrical conductivity compared to that of pristine graphene and does not display most of the exciting quantum properties of pure graphene. These groups can be significantly removed by using a sodium borohydride and sulfuric acid treatment, followed by thermal annealing [24], which is effective in restoring the p-conjugated structure and to a significant extent the conductive nature of the graphene materials. While the effort to restore the electronic properties of rGO to a level similar to that of pristine graphene has yet to be reported, rGO-related materials can have several important properties of their own, including large specific surface area and other mechanical properties that enable them to be used as various kinds of energy storage devices (▶ Nanomaterials for Electrical Energy Storage Devices), sensors, and composite structures. Liquid-Phase Exfoliation of Graphene Directly from Graphite
In some applications, where it is important to have large volumes of chemically unmodified graphene, a different approach can be adopted, involving liquidphase exfoliation of graphene directly from graphite. The idea of liquid-phase exfoliation is essentially a combination of the two methods – mechanical exfoliation from graphite, and the liquid-phase environment that is usually applied to GO. In case of GO, the process is pretty straightforward, owing to the weaker interplanar interactions and the strong hydrophilic nature of the individual graphene sheets, and simple ultrasonication of GO causes it to easily disperse stably in water. In contrast, graphene sheets in graphite interact more strongly, and being hydrophobic, do not disperse readily into aqueous media. Early efforts to exfoliate graphite were mostly through the intercalation of graphite using a range of compounds [25]. With the resurging interest in graphene, this idea has been taken forward to produce large-scale exfoliation of graphene from commercially available expandable graphite [26]. Expandable graphite can be prepared by chemical intercalation of sulfuric acid and nitric acid. Upon heating,
Synthesis of Graphene
the volatile gaseous species released from the intercalant help exfoliate graphite very rapidly into multilayer graphene. This product is then reintercalated with oleum, (fuming sulpfuric acid with 20% free SO3), and tetrabutylammonium hydroxide (TBA, 40% solution in water) is inserted into the oleum-intercalated graphite. The TBAinserted oleum-intercalated graphite is sonicated in a N,N-dimethylformamide (DMF) solution of 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N[methoxy(polyethyleneglycol)-5000](DSPE-mPEG) for 60 min to form a homogeneous suspension. A process of centrifugation removes large pieces of material from the supernatant and results in large amounts of graphene sheets suspended in DMF. The exfoliation– reintercalation–expansion of graphite can produce high-quality single-layer graphene sheets stably suspended in organic solvents and possess high electrical conductance and can be collected as large transparent conducting films by Langmuir–Blodgett assembly in a layer-by-layer manner. In a more simple approach, stable dispersions up to 0.01 mg/ml can be directly exfoliated from graphite by ultrasonication in organic solvents such as N-methyl pyrrolidone [27]. This method is physicochemically possible because the energy required to exfoliate graphene is balanced by the solvent– graphene interaction for solvents whose surface energies match that of graphene. Typically, graphite is dispersed in a relevant solvent by sonicating in a low-power ultrasonic bath for 30 min. The resultant dispersion is then centrifuged for 90 min at 500 rpm and decanted by pipetting off the top half of the dispersion. The decanted product usually contains a mixture of single- and few-layer graphene flakes that can be made into thin films by passing them through nanoporous membranes and drying them off. A different method of obtaining stable dispersions in a liquid medium was reported by An et al. [28]. In this method, 1-pyrenecarboxylic acid (PCA) was used as a method to “wedge” out graphene sheets from graphite in suitably polarity-controlled combination of media to give rise to noncovalently functionalized graphene sheets that were stably dispersed in water. In a controlled medium, initially, PCA serves as a “molecular wedge” that cleaves the individual graphene flakes from the parent graphite pieces and then forms stable polar functional groups on the
Synthesis of Graphene
Synthesis of Graphene, Fig. 4 (a) A TEM image of a typical graphene flake exfoliated from graphite using a molecular wedging technique [28]
graphene surface via a noncovalent p–p stacking mechanism that does not disrupt its sp2 hybridization. Figure 4 shows a typical TEM (▶ Transmission Electron Microscopy) image of graphene obtained this way. The hydrophilic –COOH group of PCA facilitates the formation of stable aqueous dispersions of graphene, in a manner similar to that of graphene oxide, but without degrading the sp2 structure. The advantage of this method lies in the fact that the final product is stable in water – opening up the possibilities of experiments where an aqueous medium could be important, such as interactions with biological samples. At the same time, a simple methanol wash will remove the excess PCA from the surface and the pure graphene sheet precipitates out with time – to be utilized for experiments where pure, unfunctionalized graphene sheets are required. Similar methods have been demonstrated by other groups to obtain dispersions of graphene from graphite in surfactant water solutions [29], through which significant levels of exfoliation and monolayer formation can be achieved. Epitaxial and Large-Area Graphene on Metal Surfaces High-temperature annealing of Pt single crystals produces an atomically thin entity on their surfaces, and
2635
S
this layer has been identified as monolayer graphite back in 1969 through low-energy electron diffraction (LEED) experiments [2]. Since then, it became clear that graphene in this morphology appears either due to the surface desegregation of carbon or through decomposition of hydrocarbons. Graphene has been reported to grow on a variety of metals such as Co, Ni, Ru, Rh, Pd, Ir, and Pt, with varying degrees of grain size, epitaxial alignment with the underlying metal, defect densities and surface morphologies [30]. For example, it was shown in 2006 that ethylene, absorbed at room temperature on Ir (111) planes, could be desegregated at 1,450K to produce graphene flakes of 100 nm size that covers about 30% of the surface [31]. More recently, the desegregation method was successfully utilized to graphene on Ru (0001) surfaces, where macroscopic single-crystalline domains with linear dimensions exceeding 200 mm could be achieved [32, 33]. Figure 5a shows a high-resolution STM (▶ Scanning Tunneling Microscopy) image of graphene grown on Ru surface. By controlling the chamber pressure, growth time, and temperature, millimeter-scale graphene was reported on Ru surfaces using a controlled desegregation process in 2009 [34]. In 2008, the desegregation method received a boost when a low-pressure ▶ chemical vapor deposition (CVD) technique was utilized to synthesize graphene on an Ir (111) surface in a controlled manner over a range of temperatures from 1,120K to 1,320K [35]. Scanning tunneling microscopy of the grown structures reveals that graphene prepared using this technique exhibits a continuity of its carbon rows over terraces and step edges, enabling the possibility of large-scale growth that is not limited by surface features of metals. The thermodynamics of carbon segregation and graphene formation involves a complex interplay of carbon diffusion from the bulk to the surface of the host metal, followed by surface diffusion of C atoms that lead to nucleation and growth. In a typical CVD process, a metal substrate is placed in a flow-type furnace which is heated to a temperature between 750 C and 1,000 C in the presence of a mixture of Ar and H2 at a fixed chamber pressure P. Once the target temperature stabilizes, the flow is maintained for some time (30–60 min) to clean and prepare the surface of the metal. At the end of this preparation time, the carbon source, typically ethylene or methane is introduced for a fixed time, at the end of which the furnace is allowed to cool down. Once
S
S
2636
Synthesis of Graphene
Synthesis of Graphene, Fig. 5 (a) Atomically resolved STM image of the graphene layer grown on Ru (0001) (Reprinted with permission from Ref. [32]; copyright # (2007) by the American Physical Society). (b) A digital photograph of a graphene film
grown on a Cu foil using a CVD method, and transferred onto a SiO2/Si substrate (Reprinted with permission from Ref. [38]; copyright # (2009) by the American Association for the Advancement of Science)
cooled, the surface is found to be covered with singleand multilayered graphene. Factors that appear to control layer thickness and domain sizes of graphene obtained such as temperature, feed-gas concentration, chamber pressure and time appear to be vastly different for different metals. For example, Yu et al. used CH4:H2:Ar ¼ 0.15:1:2 with a total gas flow rate of 315 sccm and found that the type of graphene obtained on Ni foils depended strongly on the cooling rate, with multilayer graphene obtained only when the cooling rate was controlled to 10 C/s [36]. A great benefit of this technique is that thin layers of pre-patterned Ni could be utilized to fabricated large-scale patterned growth of graphene films that could be transferred to arbitrary substrates [37]. At elevated temperatures, the solubility of C atoms in Ni is much higher compared to that in Cu, and the resulting graphene formation can have different morphologies for these two metals. Centimeter-scale graphene films were shown to grow rather easily on Cu substrates in 2009 through a similar CVD process [38], with more than 95% of the surface was found to be covered with single-layer graphene that are continuous across copper surface steps and grain boundaries. Figure 5b shows an example of a macroscopic sheet of graphene grown on a Cu foil and transferred onto a SiO2/Si substrate. In case of Cu, it seems that the cooling rate is unimportant, most probably due to the low solubility of carbon in copper which help to make the growth process self-limiting. There were no discernible differences in the graphene morphology when the cooling rate was varied from >300 C/min to about
40 C/min. The ability to grow high-quality large-scale graphene on a low-cost material (Cu) at such ease has been rapidly taken advantage of to produce graphene at a roll-to-roll capability [39]. This, along with a wet chemical doping and layer-by-layer stacking process, can be used to devise many-layer macroscopic graphene films with sheet resistance at values as low as 30 Ω □1 at 90% transparency, which is superior to commercial transparent electrodes such as indium tin oxides and opens up several new possibilities in the fields of optics, optoelectronics, and photovoltaics. Other Novel Advances in Gaphene Synthesis In addition to the categories described above, there are numerous innovative methods being reported on a regular basis that are either extensions of the abovementioned categories, or are completely new methods that can become significantly useful and/or popular in future. In addition to graphene, there is a whole array of related materials, including doped and functionalized graphene and graphene nanoribbons, each of which is unique and important by its own rights. It is impossible to either enlist all or describe these methods individually within the scope of this article. Some of the more popular and/or innovative ones are enlisted in Table 1 in order to provide the reader with a starting point. It should be borne in mind that a keyword search of existing databases through available searching tools was employed to obtain these references, and omission of any significant report(s) is purely accidental and inadvertent.
Synthesis of Graphene Synthesis of Graphene, Table 1 Some recent, novel and uncategorized graphene synthesis techniques Method Solvothermal synthesis
Publication details Choucair et al., Nat. Nanotechnol. 4, 30–33 (2009) Substrate-free synthesis Dato et al., Nano Lett. 8, 2012–2016 (2008) Electrochemical 1. Liu et al., Adv Func. Mat. 18, 1518– synthesis 1525 (2008) 2. Guo et al., ACS Nano 3, 2653–2659 (2009) Microwave plasma Malesevic et al., Nanotech. 19, enhanced CVD 305604.1–305604.6 (2008) Arc discharge Wu et al., ACS Nano 3, 411–417 exfoliation (2009) MicrowaveMurugan et al., Chem. Mat. 21, 5004– solvothermal synthesis 5006 (2009) Doped graphene 1. Panchokarla et al., Adv Mat 21, 4726–4730 (2009) 2. Ci et al., Nat. Mater. 9, 430–435 (2010) 3. Li et al., Carbon 48, 255–259 (2010) Graphite exfoliation in Lu et al., ACS Nano 3, 2367–2375 ionic liquids (2009) Hydrophilic and Wang et al., Carbon 47, 1359–1364 organophilic graphene (2009) Graphene nanoribbons 1. Han et al., Phys. Rev. Lett. 98, 206805.1–206805.4 (2007) 2. Li et al., Science 319, 1229–1232 (2008) 3. Jiao et al., Nat. Nanotech. 5, 321–325 (2010) Reducing sugar Zhu et al., ACS Nano 4, 2429–2437 (2010)
Conclusion It is evident that within a very short time frame of a few years, the scientific community has witnessed a tremendous advancement in terms of various techniques of graphene synthesis. This advancement has progressed hand-in-hand with a number of extremely important fundamental discoveries related to this novel 2D crystalline material and has also opened up several avenues for groundbreaking technological directions. Despite the rapid success that this field has experienced, the interest in developing better and more cost-effective synthesis methods keeps growing. In the near term, production of high-quality graphene with reproducible electrical as well as mechanical properties will be a challenge. Similarly, techniques and parameters for size-specific, controlled synthesis
2637
S
of graphene such as graphene nanoribbons, graphene quantum dots etc., need to be well understood. Processes for large-scale synthesis of monolayer graphene (e.g., through chemical exfoliation) needs to perfected in order to utilize these materials in variety of bulk applications such as electrodes in electrical energy storage systems, fillers in composite etc. Overall, the current level of progress as well as the ongoing research efforts in graphene synthesis has been quite promising for certain niche area applications, and the scientific community is quite optimistic that in future, graphene will eventually emerge as the next ultimate engineering material.
Cross-References ▶ Atomic Force Microscopy ▶ Carbon Nanotubes ▶ Chemical Vapor Deposition (CVD) ▶ Flexible Electronics ▶ Fullerenes for Drug Delivery ▶ Graphene ▶ Nanomaterials for Electrical Energy Storage Devices ▶ Scanning Tunneling Microscopy ▶ Transmission Electron Microscopy
References 1. Boehm, H.P., Clauss, A., Fischer, G.O., Hofmann, U.: Das Adsorptions verhalten sehr d€ unner Kohlenstoffolien. Zeitschrift f€ ur anorganische und allgemeine Chemie 316, 119–127 (1962) 2. May, J.W.: Platinum surface LEED rings. Surf. Sci. 17, 267–270 (1969) 3. Eizenberg, M., Blakely, J.M.: Carbon monolayer phase condensation on Ni(111). Surf. Sci. 82, 228–236 (1979) 4. Aizawa, T., Souda, R., Otani, S., Ishizawa, Y., Oshima, C.: Anomalous bond of monolayer graphite on transition-metal carbide surfaces. Phys. Rev. Lett. 64, 768–771 (1990) 5. Dresselhaus, M.S., Dresselhaus, G., Saito, R.: Carbon fibers based on C60 and their symmetry. Phys. Rev. B 45, 6234–6242 (1992); Aizawa, T., Hwang, Y., Hayami, W., Souda, R., Otani, S., Ishizawa, Y.: Phonon dispersion of monolayer graphite on Pt(111) and NbC surfaces: bond softening and interface structures. Surf. Sci. 260, 311–318 (1992) 6. Lu, X., Yu, M., Huang, H., Ruoff, R.S.: Tailoring graphite with the goal of achieving single sheets. Nanotechnology 10, 269–272 (1999)
S
S
2638
7. Lu, X., Huang, H., Nemchuk, N., Ruoff, R.S.: Patterning of highly oriented pyrolytic graphite by oxygen plasma etching. Appl. Phys. Lett. 75, 193–195 (1999) 8. Novoselov, K.S., Geim, A.K., Morozov, S.V., Jiang, D., Zhang, Y., Dubonos, S.V., Grigorieva, I.V., Firsov, A.A.: Electric field effect in atomically thin carbon films. Science 306, 666–669 (2004) 9. Berger, C., Song, Z., Li, T., Li, X., Ogbazghi, A.Y., Feng, R., Dai, Z., Marchenkov, A.N., Conrad, E.H., First, P.N., de Heer, W.A.: Ultrathin epitaxial graphite: 2D electron gas properties and a route toward graphene-based nanoelectronics. J. Phys. Chem. B 108, 19912–19916 (2004) 10. Peierls, R.E.: Quelques proprieties typiques des corpses solides. Ann. I. H. Poincare 5, 177–222 (1935) 11. Landau, L.D.: Zur Theorie der phasenumwandlungen II. Phys. Z. Sowjetunion 11, 26–35 (1937) 12. Novoselov, K.S., Jiang, D., Schedin, F., Booth, T.J., Khotkevich, V.V., Morozov, S.V., Geim, A.K., Rice, T.M.: Two-dimensional atomic crystals. Proc. Natl. Acad. Sci. U.S. A. 102, 10451–10453 (2005) 13. Soldano, C., Mahmood, A., Dujardin, E.: Production, properties and potential of graphene. Carbon 48, 2127–2150 (2010) 14. Rollings, E., Gweon, G.-H., Zhou, S.Y., Mun, B.S., McChesney, J.L., Hussain, B.S., Fedorov, A.N., First, P.N., de Heer, W.A., Lanzara, A.: Synthesis and characterization of atomically thin graphite films on a silicon carbide substrate. J. Phys. Chem. Sol 67, 2172–2177 (2006) 15. Hass, J., Feng, R., Li, T., Li, X., Zong, Z., de Heer, W.A., First, P.N., Conrad, E.H., Jeffrey, C.A., Berger, C.: Highly ordered graphene for two dimensional electronics. Appl. Phys. Lett. 89, 143106–143108 (2006) 16. Stankovich, S., Dikin, D.A., Piner, R.D., Kohlhaas, K.A., Kleinhammes, A., Jia, Y., Wu, Y., Nguyen, S.T., Ruoff, R. S.: Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 45, 1558– 1565 (2007) 17. Hummers., W.S., Jr., Offeman, R.E.: Preparation of graphitic oxide. J. Am. Chem. Soc. 80, 1339–1339 (1958) 18. Rourke, J.P., Pandey, P.A., Moore, J.J., Bates, M., Kinloch, I. A., Young, R.J., Wilson, N.R.: The real graphene oxide revealed: stripping the oxidative debris from the graphenelike sheets. Angew. Chem. Int. Ed. 50, 3173–3177 (2011) 19. Eda, G., Fanchini, G., Manish Chhowalla, M.: Large-area ultrathin films of reduced graphene oxide as a transparent and flexible electronic material. Nat. Nanotechnol. 3, 270– 274 (2008) 20. Gilje, S., Han, S., Wang, M., Wang, K.L., Kaner, R.B.: A chemical route to graphene for device applications. Nano Lett. 7, 3394–3398 (2007) 21. Tung, V.C., Allen, M.J., Yang, Y., Kaner, R.B.: Highthroughput solution processing of large-scale graphene. Nat. Nanotechnol. 4, 25–29 (2009) 22. Si, Y., Samulski, E.T.: Synthesis of water soluble graphene. Nano Lett. 8, 1679–1682 (2008) 23. Park, S., Ruoff, R.S.: Chemical methods for the production of graphenes. Nat. Nanotechnol. 4, 217–224 (2009) 24. Gao, W., Alemany, L.B., Ci, L., Ajayan, P.M.: New insights into the structure and reduction of graphite oxide. Nat. Chem. 1, 403–408 (2009) 25. Dresselhaus, M.S., Dresselhaus, G.: Intercalation compounds of graphite. Adv. Phys. 30, 139–326 (1981)
Synthesis of Nanoparticles 26. Li, X., Zhang, G., Bai, X., Sun, X., Wang, X., Wang, E., Dai, H.: Highly conducting graphene sheets and Langmuir– Blodgett films. Nat. Nanotechnol. 3, 538–542 (2008) 27. Hernandez, Y., Nicolosi, V., Lotya, M., Blighe, F.M., Sun, Z., De, S., McGovern, I.T., Holland, B., Byrne, M., Gun’Ko, Y.K., Boland, J.J., Niraj, P., Duesberg, G., Krishnamurthy, S., Goodhue, R., Hutchison, J., Scardaci, V., Ferrari, A.C., Coleman, J.N.: High-yield production of graphene by liquid-phase exfoliation of graphite. Nat. Nanotechnol. 3, 563–568 (2008) 28. An, X., Simmons, T., Shah, R., Wolfe, C., Lewis, K.M., Washington, M., Nayak, S.K., Talapatra, S., Kar, S.: Stable aqueous dispersions of noncovalently functionalized graphene from graphite and their multifunctional highperformance applications. Nano Lett. 10, 4295–4301 (2010) 29. Lotya, M., Hernandez, Y., King, P.J., Smith, R.J., Nicolosi, V., Karlsson, L.S., Blighe, F.M., De, S., Wang, Z., McGovern, I.T., Duesberg, G.S., Coleman, J.N.: Liquid phase production of graphene by exfoliation of graphite in surfactant/water solutions. J. Am. Chem. Soc. 131, 3611–3620 (2009) 30. Wintterlin, J., Bocquet, M.-L.: Graphene on metal surfaces. Surf. Sci. 603, 1841–1852 (2009) 31. N’Diaye, A.T., Bleikamp, S., Feibelman, P.J., Michely, T.: Two-dimensional Ir cluster lattice on a graphene moire on Ir(111). Phys. Rev. Lett. 97, 215501.1–215501.4 (2006) 32. Marchini, S., Gunther, S., Wintterlin, J.: Scanning tunneling microscopy of graphene on Ru(0001). Phys. Rev. B 76, 075429.1–075429.9 (2007) 33. Sutter, P.W., Flege, J.-I., Sutter, E.A.: Epitaxial graphene on ruthenium. Nat. Mater. 7, 406–411 (2008) 34. Pan, Y., Zhang, H.G., Shi, D.X., Sun, J., Du, S., Liu, F., Gao, H.-J.: Highly ordered, millimeter-scale, continuous, single-crystalline graphene monolayer formed on Ru (0001). Adv. Mater. 21, 2777–2780 (2009) 35. Coraux, J., N’Diaye, A.T., Busse, C., Michely, T.: Structural coherency of graphene on Ir(111). Nano Lett. 8, 565–570 (2008) 36. Yu, Q., Lian, J., Siriponglert, S., Li, H., Chen, Y.P., Pei, S.-S.: Graphene segregated on Ni surfaces and transferred to insulators. Appl. Phys. Lett. 93, 113103–113105 (2008) 37. Kim, K.S., Zhao, Y., Jang, H., Lee, S.Y., Kim, J.M., Kim, K.S., Ahn, J.-H., Kim, P., Choi, J.-Y., Hong, B.H.: Large-scale pattern growth of graphene films for stretchable transparent electrodes. Nature 457, 706–710 (2009) 38. Li, X., Cai, W., An, J., Kim, S., Nah, J., Yang, D., Piner, R., Velamakanni, A., Jung, I., Tutuc, E., Banerjee, S.K., Colombo, L., Ruoff, R.S.: Large-area synthesis of high-quality and uniform graphene films on copper foils. Science 324, 1312–1314 (2009) 39. Bae, S., Kim, H., Lee, Y., Xu, X., Park, J.-S., Zheng, Y., Balakrishnan, J., Lei, T., Kim, H.R., Song, Y.I., Kim, Y.-J., ¨ zyilmaz, B., Ahn, J.-H., Hong, B.H., Iijima, S.: Kim, K.S., O Roll-to-roll production of 30-inch graphene films for transparent electrodes. Nat. Nanotechnol. 5, 574–578 (2010)
Synthesis of Nanoparticles ▶ Nanoparticles
Synthesis of Subnanometric Metal Nanoparticles
Synthesis of Subnanometric Metal Nanoparticles Javier Calvo Fuentes1, Jose´ Rivas2,3 and M. Arturo Lo´pez-Quintela2 1 NANOGAP SUB-NM-POWDER S.A., Milladoiro – Ames (A Corun˜a), Spain 2 Laboratory of Magnetism and Nanotechnology, Institute for Technological Research, University of Santiago de Compostela, Santiago de Compostela, Spain 3 INL – International Iberian Nanotechnology Laboratory, Braga, Portugal
Synonyms Atomic cluster; Cluster; Nanocluster; Quantum cluster; Quantum dot
Definition Metal atomic clusters consist of groups of atoms with well-defined compositions and one or very few stable geometric structures. They represent the most elemental building blocks in nature – after atoms – and are characterized by their size, comparable to the Fermi wavelength of an electron, which makes them a bridge between atoms and nanoparticles or bulk metals, with properties very different from both of them.
Introduction Typical metal nanoparticles with dimensions from two to several tens of nanometers show smoothly sizedependent properties. However, when particle size becomes comparable to the Fermi wavelength of an electron (0.52 nm for gold and silver), properties of metal clusters are dramatically different from what should be expected if they were due only to their high surface-to-volume ratio. In these subnanometric species, quantum effects are responsible for totally new chemical, optical, and electronic properties such as, for example, magnetism, photoluminescence, or catalytic activity. The term nanoparticle usually refers to any particle of bulk metal with dimensions in the nanoscale.
2639
S
Nanoparticles usually present a core-shell structure with a core of bulk metal surrounded by a shell of disordered atoms, and their enhanced properties are due mainly to their high surface/volume ratio. The term cluster is used in reference to subnanometric species consisting of well-defined structures of metal atoms stabilized by different types of protecting ligands, with sizes below approximately 1–2 nm. In general, clusters can be divided into (1) large clusters, consisting of a core formed by a number of metal atoms in the range 10–20 to 100–200 and a protecting shell of strong ligands such as phosphines or thiols; and (2) small clusters formed by a reduced number of atoms (2 to 10–20), which do not need any strong stabilizing ligand and have almost all their atoms on the surface. Due to quantum effects, both kinds of clusters – large and small – present discrete energy levels and an increasing band gap with decreasing size (Fig. 1). Because of this splitting of energies at the Fermi level, clusters show different properties to those of nanoparticles. Cluster properties are highly dependent on their sizes. As an example, the typical surface plasmon band observed in Ag-metal nanoparticles (Fig. 2a) disappears for clusters below approximately 1–2 nm (Figs. 2b and c corresponding to large and small clusters, respectively), indicating that all the conducting electrons are now frozen and the metal silver loses its typical metallic character. At the same time, a clear difference between large and small clusters can be observed: large clusters show a continuous decrease of the absorption band with some small bumps, similar to the absorption displayed by semiconductors (Fig. 2b), and small clusters display well-defined absorption bands indicating a molecularlike behavior (Fig. 2c). The presence of such size-dependent properties can be well represented on a 3D periodic table of elements – schematically depicted in Fig. 3 – where, between the atomic and the bulk state of every element, a whole range of new materials with their different size-dependent properties appear.
Cluster Structure It is very difficult to do precise calculations of the electronic structure of metal clusters, particularly when they are larger than just a small number of
S
S
2640
Synthesis of Subnanometric Metal Nanoparticles
Nanoparticle: Energy EF
Core Shell
Size ≈ 10 nm Core: crystalline structure = bulk Shell: disordered surface atoms stabilized by charges or ligands Large cluster: Energy Eg≤1 eV
LUMO HOMO
Size ≈ 1.1 nm Core: structure = or ≠ bulk Shell: stabilized by strong binding protecting ligands (phosphines, thiols,...) Small cluster: Energy
A Eg>1 eV
A
LUMO
corresponding to electronic shell closures that are consistent with the positions of the peaks appearing in the mass spectrum of the vaporized metal. These peaks are associated with the abundance of exceptionally stable clusters with certain number of atoms: 8, 18, 20, 34, 40, 58, 92, 138. . ., usually referred to as magic numbers. As it was mentioned above, one simple consequence of this quantum-size regime with the presence of discrete states in metal clusters is the appearance of a sizable HOMO-LUMO band gap similar to that of semiconductors. Such semiconductor-like behavior is particularly significant for smaller clusters (i.e., those with a low number of atoms) with band gaps widely exceeding 1 eV, as it is shown in Fig. 5 for some selected Ag and Cu clusters. Some efforts have been made in the last years to find prediction models for ligand-protected gold-cluster species in solution. In such models, clusters are assumed to be electronically stabilized by ligands that either withdraw electrons from the metal core or are attached as weak Lewis-base ligands coordinated to the core surface by dative bonds. In this case, the requirement for an electronically closed shell, [3] formulated as (LS·ANXM)z, is:
HOMO
A
Size ≈ 0.5 nm No core-shell structure No need for binding with strong stabilizing ligands (A=adsorbate)
Synthesis of Subnanometric Metal Nanoparticles, Fig. 1 Schematic properties of nanoparticles and subnanometric clusters
atoms, and they are in solution stabilized by some capping molecules. Such difficulty can be avoided using simple models, such as the Jellium model [1] firstly developed for clusters in the gas phase. In this model, the real cluster is replaced by an electronic shell structure consisting of a uniform, positively charged jellium sphere surrounded by valence electrons. The electrons are considered to be like moving in a meanfield potential occupying energy levels according to the Aufbau principle (1S2 | 1P6 | 1D10 | 2S2 1F14 | 2P6 1G18 | . . .) as represented in Fig. 4. This model gives a considerably good approximation, preserving many of the physicochemical characteristics of clusters. The total energy as a function of cluster size, calculated by this approximation, has discontinuities
n ¼ NnA M z
(1)
where n* represents the number of electrons for shell closing of the metallic core, which has to match one of the magic numbers corresponding to strong electron shell closures in an anharmonic mean-field potential giving rise to stable clusters; N stands for the number of core metal atoms (A); nA is the atomic valence; M is the number of electron-localizing (or electron-withdrawing) ligands X, assuming a withdrawal of one electron per ligand molecule; and z represents the overall charge of the complex. The weak ligands represented by LS may be needed for completing the steric surface protection of the cluster core. Trying to combine at the same time (1) the mentioned magic numbers required for electronic stabilization of the cluster, (2) the requirements of the Eq. 1, and (3) the fact that solution-phase clusters require a sterically complete protective ligand shell compatible with a compact atomic shell structure for the metallic core, complicates enormously the calculations of the structures satisfying such conditions. Furthermore, in cases such as some gold and silver thiolate clusters, the identity of the actual X groups is not clear due to
Synthesis of Subnanometric Metal Nanoparticles 1.2
3
2
1
0 300
400
500
Wavelength (nm)
600
700
0.6
b
0.9
Absorbance (a.u.)
a Absorbance (a.u.)
Absorbance (a.u.)
4
2641
0.6
0.3
0.0 300
400
500 Wavelength (nm)
Synthesis of Subnanometric Metal Nanoparticles, Fig. 2 Absorption spectra of (a) silver nanoparticles displaying the typical plasmon band at 400 nm, (b) surfactant-protected
600
700
S
c
0.4
0.2
0.0 200
300
400
500
Wavelength (nm)
large silver clusters showing a typical continuum decrease of the absorption, and (c) strong-ligand free small silver clusters with well-defined molecular-like absorption bands
S
Synthesis of Subnanometric Metal Nanoparticles, Fig. 3 Size-dependent 3D periodic table of elements
S
2642
Synthesis of Subnanometric Metal Nanoparticles
a n
5
4
3
2
1
0
168
156
156
112
138 112 106
138 132
70
40
8
34 20 18 8
2
92 90 68 58 40
3s 1h 2d
0
34 20 18
1f 2s 1d
6
1p
1g 2p
2
2
Na40
3p 1l 2f
106
92 70 68 58 40
20
2g
Effective potential (eV)
6
b
(V, l)
–2 2p 1f 2s 1d 1p 1s
–4
–6 0
10 Radius R (arb.units)
20
1s
Synthesis of Subnanometric Metal Nanoparticles, Fig. 4 (a) Energy-level occupations for spherical three-dimensional, harmonic, intermediate, and square-well potentials. (b) Self-consistent effective potential of a Jellium sphere
corresponding to Na40 with the electron occupation of the energy levels. Reprinted with permission from reference [2]. Copyright 1993 by the American Physical Society
ε/VSHE
E/eV –2.5
–2 3.3 2.9
–1
–3.5
2.5 1.9
Synthesis of Subnanometric Metal Nanoparticles, Fig. 5 Schematic comparison between bandgaps of some silver and copper clusters (MN, N ¼ number of atoms) and those of well-known semiconductors. Bandgaps (Eg) were calculated from the spherical Jellium model (Eg ¼ EF/N1/3; EF ¼ Fermi level), and the position of the conduction band, ECB, was estimated by the formula ECB ¼ w EF ½ Eg (w ¼ electronegativity)
1.4 1.2 Ag25
1.4
Ag10
CdTe
1 Ag6
1.9
Ag4
2.5 2
Cu25
βMnO2 GaAs GaP CuO
3.3 –5.5
3.2
3.2 3.2
3.8
3.3 Cu6 Cu4
4
2.5 1.7
Cds
2.9 Cu10
3
2.3 –4.5
0
–6.5
AgCl
WO3 ZnO
TiO2
–7.5
SnO2
–8.5
Synthesis of Subnanometric Metal Nanoparticles 8
a
Au11(PH3)7Cl3
6 4
PLDOS [1/eV]
2 0 6
b
F D P S
I H G 8
18
Au11(PH3)7(SCH3)3
4 2
8
18
0
c 10
Au39(PH3)14Cl6–
5
34
0 –2
PLDOS [1/eV]
Synthesis of Subnanometric Metal Nanoparticles, Fig. 6 Electronic structure analysis of gold clusters. The angular-momentum-projected local electron density of states (PLDOS) (projection up to the l symmetry, i.e., l ¼ 6) for the gold core in Au11(PH3)7Cl3 (a), Au11(PH3)7(SCH3)3 (b), Au39(PH3)14Cl6 (c), and Au102(p-MBA)44 (d). The zero energy corresponds to the middle of the HOMO-LUMO gap (dashed line). For plotting, each individual electron state is displayed by a Gaussian smoothing of 0.07 eV (0.03 eV in d). Shell-closing number is indicated for each case. For more details see reference [3]. Copyright 2008 National Academy of Sciences, U.S.A.
S
2643
15
–1.5
d
–1
–0.5
0 Energy [eV]
58 0.5
1
Au102 (p–MBA)44
D P S
10 5 0 –1
1.5
2
I H G F
58 –0.5
0
0.5
1
1.5
Energy [eV]
the undefined nature of the surface chemical bonds. There have been only relatively few studies so far to resolve the cluster structures by ab initio calculations. Very recently, H€akkinen’s group managed to determine, through large-scale Density Functional Theory (DFT) calculations, [3] the electronic structure of Au102(p-MBA)44 (p-MBA, para-mercaptobenzoic acid, SC7O2H5), an all-thiolate-protected cluster formed by 102 gold atoms, as well as some smaller phosphine-halide- and phosphine-thiolate-protected clusters, as it is shown in Fig. 6. In all cases, the characteristic bandgap of the clusters is clearly observed. In Table 1, it is shown how this bandgap changes for different cluster sizes and ligands. For comparison purposes, the bangap calculated from the simple spherical Jellium model is included (for clusters with sizes 25 atoms, a correction of 0.4 eV was used because of the anharmonicity observed in large clusters) [4]. It can be observed that, in general, bandgap increases when the cluster size decreases, being a nice agreement between the DFT-calculated bandgap values and those predicted by the simple Jellium model, showing at the same time that ligands play a relatively minor role.
Properties of Metal Clusters Among the properties that make metal clusters unique and useful in many applications, the following ones can be highlighted here as examples: Photoluminescence The molecular-like behavior of few-atom noble metal clusters and the mentioned bandgap similar to that of semiconductors, allows the possibility of size-tunable electronic intraband transitions [5]. The appearance of these optical transitions with energies ranging from the UV-visible to the infrared region makes noble metal quantum dots ideal potential candidates for applications such as fluorescent labeling in biology or light-emitting sources in nanoelectronics. As an example, Fig. 7 shows the fluorescence of aqueous solutions of CuN clusters (N < 14) excited at 296 nm with quantum yields over 10%, which are stable for more than 2 years. Even though, in cases where the quantum yield is lower than that cited in previous example, fluorescent noble metal clusters are a very interesting alternative to semiconductor quantum dots as biological labels, due not only to their small hydrodynamic size and inert
S
S
2644
Synthesis of Subnanometric Metal Nanoparticles
Synthesis of Subnanometric Metal Nanoparticles, Table 1 Comparison between: (1) experimentally determined bandgaps for free gas-phase gold cluster anions from photoelectron spectroscopy; (2) theoretical (Density Functional Theory) values for HOMOLUMO gaps of passivated gold cluster compounds that correspond to 8, 34, and 58 conduction-electron shell closings; and (3) bandgaps calculated through the Jellium model for clusters with the same number of gold atoms Shell closing 8e (1S21P6) 8e 8e 8e 34e (8e + 1D102S21F14) 58e (34e + 2P61G18) 58e
Experiment Cluster
Gap (eV)
Au34 Au58
1.0 0.6
Theory Cluster compound Au11(PH3)7(SMe)3 Au11(PH3)7Cl3 Au13(PH3)10Cl23+ Au25(SMe)18 Au39Cl6(PH3)14 Au102(p-MBA)44 Au102(SMe)44
Gap (eV) 1.5 2.1 1.8 1.2 0.8 0.5 0.5
Jellium model Eg ¼ 5.32/N1/3 (eV) 2.4 2.4 2.3 1.3a 1.0a 0.6a 0.6a
Adapted from reference [3]. Copyright 2008 National Academy of Sciences, U.S.A. a For these clusters, a correction of 0.4 eV was used – see text
streptavidin. Unconjugated clusters can also be nonspecifically uptaken by living cells (see Fig. 8).
Synthesis of Subnanometric Metal Nanoparticles, Fig. 7 Blue emission observed for an aqueous solution of copper clusters (N 14 atoms) under an excitation wavelength of 296 nm. Reprinted with permission from reference [6]. Copyright 2010 American Chemical Society
nature, but especially because of their biocompatibility and also they present reduced photobleaching compared to organic fluorophores. Water-soluble gold clusters have been, therefore, explored as fluorescent labels in biological experiments. Recently, for example, Parak’s group described [7] the use of stable gold clusters capped with dihydrolipoic acid (DHLA) and conjugated through EDC chemistry to biologically relevant molecules such as PEG, BSA, avidin, or
Catalysis Small clusters have been found to own a catalytic activity not observed in their bulk analogues or even nanoparticles, which makes them very attractive as new catalytic materials. Quantum chemical calculations indicate that such high reactivity is due to undercoordination of the metal atoms forming the cluster [8]. Several families of metal clusters have been found to possess surprisingly high and selective catalytic activity when immobilized on a support. For example, platinum clusters with 8–10 atoms can be used as catalysts for the oxidative dehydrogenation of propane [9], while gold clusters with 6–10 atoms have been shown to be highly active for catalyzing propene epoxidation [10]. Recently, Harding et al. [11] studied the control and tunability of the catalytic oxidation of CO by Au20 clusters deposited on MgO surfaces, finding that the active site on the cluster is characterized by enhanced electron density, which activates the adsorbed O2 molecule and promotes the bonding of CO. Small metal clusters have been also found to display electrocatalytic activities different from the material in bulk or as nanoparticles [12]. Although this is a field not too much explored so far, these electrocatalytic properties make metal clusters to be promising materials in fuel-cell applications. As another interesting example of such properties, a recent report of their use to prevent pathologies, such as fetal alcohol syndrome, [13] is highlighted here: It has been proved that some
Synthesis of Subnanometric Metal Nanoparticles
2645
Synthesis of Subnanometric Metal Nanoparticles, Fig. 8 Nonspecific uptake of unconjugated fluorescent DHLA-capped Au clusters by human endothelial cells. Cell nuclei were stained to yield blue fluorescence. Red fluorescence
S
corresponds to the clusters. See reference [7] for more information. Copyright 2009 American Chemical Society. Reprinted with permission
H
EtO Et
≥200 mV CH3CH2OH
Oxidation φout(RHE) ≥ +200 mV
Outer hemimembrane Inner hemimembrane
φtrans (RHE) ≈ –90 mV
Sinner Souter
Extracellular medium AgAQCs
φ in(RHE) ≥ +110 mV
Lintracell
Synthesis of Subnanometric Metal Nanoparticles, Fig. 9 Schematic representation of the electrocatalysis of the oxidation of ethanol by silver clusters on the cellular membrane preventing the alcohol toxicity in living cells. For more details, see reference [13]. Copyright 2010 American Chemical Society. Reprinted with permission
Lextracell
OH by ele Ag ctro clu oxi ste da rs tion
H2O2 Reduction ≤ –900 mV
Intracellular medium
clusters are able to electrocatalyze the oxidation of alcohols and prevent its cytotoxicity under physiological conditions and at very low potentials, like those found on living cells, as it is schematically represented in Fig. 9.
Synthesis of Metal Clusters As it also occurs in the case of nanoparticles, there are two main approaches to the synthesis of metal clusters: top-down and bottom-up.
S
S
2646
Synthesis of Subnanometric Metal Nanoparticles Route 1
Metal precursor Aurophilic interaction
Surface etching Nanoparticle
Cluster
+
Reductor
+ Stabilizer
Synthesis of Subnanometric Metal Nanoparticles, Fig. 11 Schematic representation of bottom-up synthesis
Route 2
+
Nanoparticle
Size reduction
Cluster
Gold-thiolate
Synthesis of Subnanometric Metal Nanoparticles, Fig. 10 Schematic representation of two possible routes for the formation of gold clusters by etching of mercaptosuccinic acid-capped gold nanoparticles. Reprinted from reference [14] with kind permission from Springer Science + Business Media B.V.
Top-Down In top-down approaches, clusters are synthesized from larger precursors such as nanoparticles or bulk metal. The most common top-down technique used for the synthesis of clusters is etching of nanoparticles (Fig. 10). In this technique, clusters are synthesized using the etching capacity of some ligands (like thiols), which remove the surface atoms of metallic nanoparticles or break them into smaller pieces leading to stable quantum clusters characterized by shellclosing magic numbers. For example, etching of mercaptosuccinic acidprotected gold nanoparticles with excess glutathione has been used to yield small photoluminescent gold clusters [14]. In such work, clusters, either with 8 or 25 gold atoms, were produced by etching of previously synthesized nanoparticles with 4–5 nm core diameter. Selection of the final cluster size was possible by only varying the etching pH from 3 (for 25 atom clusters) to 7–8 (for 8 atom clusters). Multivalent coordinating polymers such as polyethylenimine can also be used to etch preformed colloidal gold nanocrystals producing highly fluorescent water-soluble nanoclusters formed by 8 gold atoms [15]. Bottom-Up In bottom-up approaches, clusters are built from smaller structures at atomic level such as single atoms or ions (Fig. 11). In this case, the most
commonly used techniques differ depending on the size of the clusters to be synthesized. Wet chemical preparation of clusters includes the chemical reduction of metal salts in the presence of strong binding ligands [4, 16, 17] or cages [18], and by kinetic control using, e.g., microemulsions [19–21], electrochemical methods [6, 12, 22], etc. For the synthesis of large clusters, a large variety of strong, protecting ligands can be used to control the growth of the primarily formed clusters and to stabilize them. For small clusters, the kinetic control, based on slow reaction rates provided either by soft reducing agents or low concentrations (as e.g., those provided by microemulsions, electrochemical methods, etc.), is the key point to their synthesis. It is generally believed that the synthesis of small clusters is extremely difficult to achieve because of the highly precise control of experimental conditions required to stop and then isolate the clusters as soon as they are formed. That belief is based on the theory of nucleation and growth, according to which, nuclei formed during the first steps of the chemical synthesis are only stable when they grow over a specific size called critical nucleus. Below that size, nuclei dissolve because of their large Laplace pressure. Above that size, they continue growing in order to reduce their surface energy by different mechanisms like autocatalysis, Ostwald ripening, etc., until the growth is stopped by capping agents or other templates. However, such arguments, which correctly can be applied to the synthesis of nanoparticles and larger clusters having the same structure of the bulk material, are not correct when they apply to very small clusters, as those produced by kinetic control. As it was previously mentioned, numerous theoretical and experimental reports indicate that clusters can be especially stable because of their particular electronic and geometric structures different from the bulk. In such a case, the use of macroscopic thermodynamic arguments, like those used in the theory of nucleation and growth, cannot be applied. To clarify this important aspect, in Fig. 12, a scheme is shown of the variation of the free energy throughout a reaction of nanoparticle’s
Synthesis of Subnanometric Metal Nanoparticles
ΔG Reactants
C1
C2 CN C3
NP
Products Reaction coordinate
Synthesis of Subnanometric Metal Nanoparticles, Fig. 12 Schematic representation of the variation of free energy during the chemical synthesis of metal nanoparticles
formation in solution. The initial reagents at the top of the curve represent the metal ions in the presence of a reducer (or, e.g., a cathode, were reduction of metal ions takes place), while the products at the bottom represent the final stable nanoparticles (NP). In a typical synthesis of nanoparticles, a fast reduction reaction is used, which means that the driving force (which can be represented by the difference in the electrochemical potentials of the metal to be synthesized and the reducing agent) is enough to drive the system directly from a maximum energy (reactants) to a minimum energy (products), represented by the first stable nucleus (CN) having the same structure of the bulk. However, when a small reaction rate is used, stable clusters like C2, C3, etc. are produced and, with an adequate control of the reaction kinetics, it is possible to stop the reduction and to obtain stable metal clusters with the desired size. Following this scheme, silver clusters with less than 10 atoms have been synthesized, for example, in water-in-oil microemulsions consisting of a mixture of sodium bis(2-ethylhexyl) sulfosuccinate, isooctane, and water [20]. Synthesis conditions necessary for the appearance and isolation of stable clusters were achieved using small concentrations and a mild reduction agent as sodium hypophosphite monohydrate. Silver clusters produced in this way were found not only to be stable for years, but also to have a large bandgap (2.3 eV) and to present molecular-like paramagnetic properties and an intense photoluminescence.
2647
S
In the same way, using electrochemical techniques, very small clusters can be synthesized, providing a good control of the cluster size. For example, stable poly(N-vinylpyrrolidone), PVP-protected gold clusters with only 2 or 3 atoms could be synthesized through a simple electrochemical method based on the anodic dissolution of a gold electrode in the presence of the homopolymer PVP and subsequent electroreduction of the Au-PVP complexes [22]. The resulting clusters (Au2 and Au3) are the smallest ones that can be prepared and they display fluorescent (Au2) and paramagnetic (Au3) properties. By a similar procedure, highly fluorescent copper clusters stabilized with tetrabutylammonium nitrate, with less than 14 atoms, have also been reported [6]. These clusters can be dispersed in different solvents (both polar and apolar), are stable for several years and display also similar fluorescent properties than those synthesized in microemulsions [21]. Finally, it is worth mentioning the use of hard cages to inhibit the growth and synthesize small clusters, as it was demonstrated very recently by Royon et al. [18], showing that silver-containing glasses can be used for preparing fluorescent materials reducing the Ag ions in the matrix with laser pulses. In this way, they were able to prepare perennial high capacity 3D-optical recording media. In summary, it can be said that by appropriate use of the typical techniques developed for the synthesis of NPs, clusters of different sizes down to 2 atoms can be produced, isolated, and used as stable materials with novel properties, which differ very strongly from the bulk materials and nanoparticles due to their totally different electronic and geometrical structure. Acknowledgements We would like to thank the Spanish Ministry of Science and Innovation (MAT2008-06503/NAN; MAT2010-20442; NanoBioMed CONSOLIDER INGENIO) and the Regional Government of Galicia (Competitive Reference Groups 2010/41, financed with FEDER funds and Project INCITE08PXIB209049PR) for providing US financial support of different parts related to this work.
Cross-References ▶ Dry Etching ▶ Nanoparticles ▶ Optical and Electronic Properties ▶ Wet Etching
S
S
2648
Synthesized Conductance Injection
References
of gold starting from metallic nanoparticles by pHdependent ligand etching. Nano Res. 1, 333–340 (2008) Duan, H., Nie, S.: Etching colloidal gold nanocrystals with hyperbranched and multivalent polymers: a new route to fluorescent and water soluble atomic clusters. J. Am. Chem. Soc. 129, 2412–2413 (2007) Schaeffer, N., Tan, B., Dickinson, C., Rosseinsky, M.J., Laromaine, A., McComb, D.W., Stevens, M.M., Wang, Y., Petit, L., Barentin, C., Spiller, D.G., Cooper, A.I., Le´vy, R.: Fluorescent or not? Size-dependent fluorescence switching for polymer-stabilized gold clusters in the 1.1–1.7 nm size range. Chem. Commun. 34, 3986–3988 (2008) Negishi, Y., Nobusada, K., Tsukuda, T.: Glutathioneprotected gold clusters revisited: bridging the gap between gold(I)-thiolate complexes and thiolate-protected gold nanocrystals. J. Am. Chem. Soc. 127, 5261–5270 (2005) Royon, A., Bourhis, K., Bellec, M., Papon, G., Bousquet, B., Deshayes, Y., Cardinal, T., Canioni, L.: Silver clusters embedded in glass as a perennial high capacity optical recording medium. Adv. Mater. 22, 5282–5286 (2010) Lo´pez-Quintela, M.A.: Synthesis of nanomaterials in microemulsions: formation mechanisms and growth control. Curr. Opin. Colloid Interface Sci. 8, 137–144 (2003) Ledo-Sua´rez, A., Rivas, J., Rodrı´guez-Abreu, C.F., Rodrı´guez, M.J., Pastor, E., Herna´ndez-Creus, A., Oseroff, S.B., Lo´pez-Quintela, M.A.: Facile synthesis of stable subnanosized silver clusters in microemulsions. Angew. Chem. Int. Ed. 46, 8823–8827 (2007) Va´zquez-Va´zquez, C., Ban˜obre-Lo´pez, M., Mitra, A., Lo´pez-Quintela, M.A., Rivas, J.: Synthesis of small atomic copper clusters in microemulsions. Langmuir 25, 8208–8216 (2009) Santiago Gonza´lez, B., Rodrı´guez, M.J., Blanco, C., Rivas, J., Lo´pez-Quintela, M.A., Martinho, J.M.G.: One step synthesis of the smallest photoluminescent and paramagnetic PVP-protected gold atomic clusters. Nano Lett. 10, 4217–4221 (2010)
1. Chou, M.Y., Cleland, A., Cohen, M.L.: Total energies, abundances, and electronic shell structure of lithium, sodium, and potassium clusters. Solid State Commun. 52, 645–648 (1984) 2. de Heer, W.A.: The physics of simple metal clusters: experimental aspects and simple models. Rev. Mod. Phys. 65, 611–676 (1993) 3. Walter, M., Akola, J., Lopez-Acevedo, O., Jadzinsky, P.D., Calero, G., Ackerson, C.J., Whetten, R.L., Gro¨nbeck, H., H€akkinen, H.: A unified view of ligand-protected gold clusters as superatom complexes. Proc. Natl. Acad. Sci. 105, 9157–9162 (2008) 4. Zheng, J., Zhang, C., Dickson, R.M.: Highly fluorescent, water-soluble, size-tunable gold quantum dots. Phys. Rev. Lett. 93, 077402 (2004) 5. Zheng, J., Nikovich, P.R., Dickson, R.M.: Highly fluorescent noble metal quantum dots. Annu. Rev. Phys. Chem. 58, 409–431 (2007) 6. Vilar-Vidal, N., Blanco, M.C., Lo´pez-Quintela, M.A., Rivas, J., Serra, C.: Electrochemical synthesis of very stable photoluminescent copper clusters. J. Phys. Chem. C 114, 15924–15930 (2010) 7. Lin, C.-A.J., Yang, T.-Y., Lee, C.-H., Huang, S.H., Sperling, R.A., Zanella, M., Li, J.K., Shen, J.-L., Wang, H.-H., Yeh, H.I., Parak, W.J., Chang, W.H.: Synthesis, characterization, and bioconjugation of fluorescent gold nanoclusters toward biological labeling applications. ACS Nano 3, 395–401 (2009) 8. Heiz, U., Landman, U.: Nanocatalysis (Nanoscience and Technology). Springer, Berlin (2007) 9. Vajda, S., Pellin, M.J., Greeley, J.P., Marshall, C.L., Curtiss, L.A., Ballentine, G.A., Elam, J.W., Catillon-Mucherie, S., Redfern, P.C., Mehmood, F., Zapol, P.: Subnanometre platinum clusters as highly active and selective catalysts for the oxidative dehydrogenation of propane. Nat. Mater. 8, 213–216 (2009) 10. Lee, S., Molina, L.M., Lo´pez, M.J., Alonso, J.A., Hammer, B., Lee, B., Seifert, S., Winans, R.E., Elam, J.W., Pellin, M. J., Vajda, S.: Selective propene epoxidation on immobilized Au6–10 clusters: the effect of hydrogen and water on activity and selectivity. Angew. Chem. Int. Ed. 48, 1467–1471 (2009) 11. Harding, C., Habibpour, V., Kunz, S., Farnbacher, A.N.-S., Heiz, U., Yoon, B., Landman, U.: Control and manipulation of gold nanocatalysis: effects of metal oxide support thickness and composition. J. Am. Chem. Soc. 131, 538–548 (2009) 12. Rodrı´guez-Va´zquez, M.J., Blanco, M.C., Lourido, R., Va´zquez-Va´zquez, C., Pastor, E., Planes, G.A., Rivas, J., Lo´pez-Quintela, M.A.: Synthesis of atomic gold clusters with strong electrocatalytic activities. Langmuir 24, 12690–12694 (2008) 13. Selva, J., Martı´nez, S.E., Buceta, D., Rodrı´guez-Va´zquez, M.J., Blanco, M.C., Lo´pez-Quintela, M.A., Egea, G.: Silver sub-nanoclusters electrocatalyze ethanol oxidation and provide protection against ethanol toxicity in cultured mammalian cells. J. Am. Chem. Soc. 132, 6947–6954 (2010) 14. Habeeb Muhammed, M.A., Ramesh, S., Sinha, S.S., Pal, S. K., Pradeep, T.: Two distinct fluorescent quantum clusters
15.
16.
17.
18.
19.
20.
21.
22.
Synthesized Conductance Injection ▶ Dynamic Clamp
Synthesized Ionic Conductance ▶ Dynamic Clamp
Synthesized Synaptic Conductance ▶ Dynamic Clamp
Synthetic Biology
Synthetic Biology Soichiro Tsuda Exploratory Research for Advanced Technology, Japan Science and Technology Agency, Osaka, Japan
Synonyms Synthetic genomics
Definition Synthetic biology is the design and construction of biological components (e.g., enzymes, gene circuits, and whole cells) from scratch or from standardized parts.
Overview The term “Synthetic Biology” was first used in 1980 by Barbara Hobom to describe genetically engineered bacteria using recombinant DNA technology. In 2000, the term reappeared again by Eric Kool and others at the annual meeting of the American Chemical Society to refer to synthesis of unnatural chemical products (e.g., organic molecules) as an extension of synthetic chemistry [1]. Although the scope of synthetic biology has explosively expanded in the last decade, synthetic biology can be best described as “efforts to redesign life,” that is, the design and construction of biological components (e.g., enzymes, gene circuits, and whole cells) using living cells. Similar approaches have already been taken long before the advent of synthetic biology, for example, by genetic engineering, which is one of the roots of synthetic biology. However, what makes synthetic biology distinctive from other preceding biological research fields is its strong engineering perspective. While genetic engineering focused on modifying individual genes and pathways, synthetic biology does on whole systems of genes and gene products. It aims at designing and constructing the behavior of organisms to perform new tasks [2]. Computer engineering analogy is often used to explain the goal of synthetic biology: In order to build a computer,
2649
S
it involves several hierarchical layers. The most fundamental layer of computers is the electrical components, such as resistor, transistor, capacitor, etc. Based on them, a higher level structure, logic gates (AND, OR, NOT, XOR, etc) are built, which are further integrated into even upper layer, integrated circuits, and so on. Biological systems can be described similarly: DNA, RNA, and proteins are the most fundamental components of biological cells. These components form gene circuit networks, which are parts of intracellular signal pathways, and so on. Thus, synthetic biology research involves the developments of biological components in each layer for engineering new biological functions: standardized fundamental biological parts, gene circuit construction, and whole biological cells, and biological and chemical products synthesized by living cells.
Standardized Biological Parts To engineer complex systems, the standardization of components used in the systems (in this case biological systems) is necessary. Tom Knight and Drew Endy proposed the BioBrick standard biological parts as basic units for synthetic biological systems and set up an online registry of BioBrick parts where anyone can contribute in an open-source manner [3–5]. It provides a collection of well-characterized and standardized building blocks that can be used for the “plug-andplay” design and construction of unnatural biological systems. Using BioBrick standard parts, a biological engineer can program a living cell, as an electrical engineer can program an electrical circuit using electrical components. BioBrick parts are mostly DNA sequences with functionality, such as promoters, plasmids, and ribosome binding sites. Not only natural components, biologically inspired molecules can be BioBrick parts: For example, DNA analog PNA, in which the sugar-phosphate backbone is altered by N(2-aminoethyl)-glycine units linked by peptide bonds [6].
Gene Circuit Design and Construction Having set the standards for biological components, the next step is to develop minimum functional components from the standard parts. They are typically
S
S
2650
gene circuits that express particular proteins. These single circuits are connected together to construct a gene regulatory network. Network motifs, such as cascades, feedfoward, and feedback, are commonly used for the design of artificial regulatory network (and they are also commonly found in natural biological systems) [7]. Cascades are series-connected gene circuits that output (protein) of an upstream gene circuit regulates expression of its immediately downstream gene. Feedforward motif involves a master regulatory gene that influences downstream genes through non-circular pathways. When the feedforward loop is introduced in a network, for example, it is possible to construct a network that shows a nonlinear non-monotonic change in the output. For example, a sigmoid function with a steep transition can be implemented with two gene circuits connected by negative (inhibitory) feedforward motif [8]. Basu and co-workers implemented a pulse-generator system using the motif [9]. When the input stimulus is monotonously changed from low to high, the output of the pulse-generator network changes from low to high and then back to low. In addition, the pulse amplitude and delay reflect not only the input concentration but also its rate of change. Feedback is a common biological regulatory motif not only for intracellular gene networks but also for intercellular control (e.g., negative feedback regulation of hormonal control). Feedback loop allows various complex regulation over expression of genes: noise reduction [10], expression level control [11], bistable gene toggle switch [12], autonomous genetic oscillator [13], etc. In principle, any artificial regulatory scheme can be implemented with the combination of above network motifs. Although standardization and other engineering concepts, such as abstraction modularity, reliability, greatly contribute to the speed and tractability of gene circuit design, one should note the liquid nature of biological systems and components. In contrast to solid electrical components, biological components, DNA, RNA, and proteins, all operate in the liquid solution which provide an appropriate “context” (e.g., pH, temperature, materials, energy, etc) for the systems to function. This cellular context dependence and sensitivity make it difficult for biological components to be modular and interchangeable. A gene circuit from naturally occurring system is likely to be optimized to a specific cellular context through evolution. Thus, it may not function when it is placed in an artificial
Synthetic Biology
context. Ron Weiss pointed out the need to understand “how the function of a module or an entire biological system can be derived from the function of its component parts ‘and establish’ the biological rules of composition” in order to build biological components and modules [2].
Minimal Life Craig Venter and his co-workers took the synthetic gene design to its extreme. They proposed to synthesize a complete genome from scratch and create “artificial life.” In 2003, they have synthesized the whole genome of Phi-X 174, a 5386 base pair bacteriophase, from synthetic oligonucleotides, which took them only 14 days to complete the whole process [14]. The synthesized genome was injected into E. coli bacteria and confirmed to be infectious. After the success in synthesized virus, they took a further step to create a synthesized organism that has a minimal genome. A bacterial genome of Mycoplasma genitalium, which is one of the smallest known genomes in any living organisms (517 genes, 580,000 DNA base pairs), was chosen as a base for the synthesis. They removed genes in the genome that are considered redundant and injected into a M. capricolum cell in which the original genome was removed in order to test if the cell with the modified genome can be alive. In 2010, they reported that the size of the genome was eventually reduced to approximately one mega base pair that are minimal and sufficient to be considered alive. They called the cell Mycoplasma laboratorium which is the first artificial life, of which parent is a digital computer [15]. Another important aspect of minimal life in synthetic biology is “cells as chassis.” The context dependence of biological components requires the constant maintenance of appropriate cellular context. Artificially synthesized biological components and modules are not exception and therefore require cellular context and its maintenance to perform tasks. Apart from the synthetic genome mentioned above, thus, all the other parts of protein expression process (transcription, RNA processing, translation, protein folding, aminoacid modifications) can be subjects of synthetic biology research [16]. The construction of a minimal cell from scratch, as a basic unit for proving cellular context, is also an active area in this sub-research field. Lipid vesicles (liposomes) are commonly used as
Synthetic Biology
minimal bioreactors to simulate biological cells surrounded by lipid bilayer membrane. Functions of biological cells, such as membrane division, fusion, protein expression, and self-replication, have been attempted to replicate inside vesicles [17]. In recent years, microfluidic techniques, called digital microfluidics, have been applied to the production of lipid vesicles [18]. These techniques allow highthroughput production and screening of protein expression in the microfluidic devices, and are expected to facilitate the more precise control of experimental conditions.
Applied Biological and Chemical Synthesis Practical applications of synthetic biology are expected to be quite enormous because one of ultimate goals in synthetic biology is to create synthetic organisms that perform commercially useful tasks. For example, Craig Venter’s group developed the synthetic genome described above so that it could be served as a platform for future synthetic organisms on which new functions (i.e., gene circuits) can be added. Since 2009, Venter’s company, Synthetic Genomics Incorporated (SGI), has been working with Exxon Mobil Corporation to develop synthetic algae that produce next generation biofuel, such as ethanol or hydrogen [19]. They also claim it would be possible to synthesize organisms which can fix carbon dioxide using photosynthesis to mitigate climate change [20]. Another example is medical and pharmaceutical applications of synthetic biology. Microbes, typically bacteria and yeast, are engineered and used as “chemical factory” to produce chemical products that are difficult to synthesize artificially. This process inevitably involves metabolic pathway engineering of living cells and therefore requires management of complex and emergent metabolic processes [21]. Keasling and co-workers developed synthetic bacteria for the antimalarial drug production [1, 22]. The drug, artemisinin, extracted from Artemisia annua L (sweet wormwood) is known to be the most effective drug for the treatment of malaria, but expensive to produce because of short supply of the wood. The engineered bacteria incorporate several genes originally from bacteria E. coli and yeast S. cerevisiae to produce artemisinic acid, a bio-genetic precursor of artemisinin. First, MevT operon encodes enzymes
2651
S
that transform acetyl-CoA into mevalonate. Enzymes encoded by MBIS operon then transform mevalonate into farmesyl pyrophosphate (FPP). Additionally, genes for amor-phadiene synthase (transforming FPP into amorphadiene) and oxidase and redox partners (amorphadiene into artemisinic acid) were introduced. Although E. coli has a native pathway, 1-deoxy-Dxylulose 5-phosphate (DXP) pathway, to produce FPP, it has been found that the engineered mevalonate pathway produces more amorphadiene than the native pathway. Artemisinic acid purified from bacterial extracts is transformed to artemisinin using established chemistry.
Cross-References ▶ Computational Systems Bioinformatics for RNAi ▶ Liposomes ▶ Nanotechnology Applications in Polymerase Chain Reaction (PCR) ▶ Structure and Stability of Protein Materials
References 1. Benner, S.A., Sismour, A.M.: Synthetic biology. Nat. Rev. Genet. 6(7), 533–543 (2005) 2. Andrianantoandro, E., Basu, S., Karig, D.K., and Weiss, R.: Synthetic biology: new engineering rules for an emerging discipline. Mol. Syst. Biol. 2(1), 2006.0028 (2006) 3. The BioBricks Foundation http://bbf.openwetware.org/ 4. Knight, T.: Idempotent vector design for standard assembly of biobricks standard biobrick sequence interface. MIT Synthetic Biology Working Group Report, MIT Artificial Intelligence Laboratory, 1–11 (2003) 5. Endy, D.: Foundations for engineering biology. Nature 438(7067), 449–453 (2005) 6. Nielsen, P.E., Egholm, M.: An introduction to peptide nucleic acid. Curr. Issues Mol. Biol. 1(1–2), 89–104 (1999) 7. McDaniel, R., Weiss, R.: Advances in synthetic biology: on the path from prototypes to applications. Curr. Opin. Biotechnol. 16(4), 476–483 (2005) 8. Hooshangi, S., Thiberge, S., Weiss, R.: Ultrasensitivity and noise propagation in a synthetic transcriptional cascade. Proc. Natl Acad. Sci. U.S.A. 102(10), 3581–3586 (2005) 9. Basu, S., Mehreja, R., Thiberge, S., Chen, M.-T., Weiss, R.: Spatiotemporal control of gene expression with pulsegenerating networks. Proc. Natl Acad. Sci. U.S.A. 101(17), 6355–6360 (2004) 10. Becskei, A., Serrano, L.: Engineering stability in gene networks by autoregulation. Nature 405, 590–593 (2000) 11. Nevozhay, D., Adams, R.M., Murphy, K.F., Josic, K., Balzsi, G.: Negative autoregulation linearizes the
S
S 12.
13.
14.
15.
16. 17.
18.
2652 dose-response and suppresses the heterogeneity of gene expression. Proc. Natl Acad. Sci. U.S.A. 106(13), 5123– 5128 (2009) Gardner, T.S., Cantor, C.R., Collins, J.J.: Construction of a genetic toggle switch in Escherichia coli. Nature 403(6767), 339–342 (2000) Judd, E.M., Laub, M.T., McAdams, H.H.: Toggles and oscillators: new genetic circuit designs. BioEssays 22(6), 507–509 (2000) Smith, H.O., Hutchison, C.A., Pfannkoch, C., Venter, J.C.: Generating a synthetic genome by whole genome assembly: x174 bacteriophage from synthetic oligonucleotides. Proc. Natl Acad. Sci. U.S.A. 100(26), 15440–15445 (2003) Gibson, D.G., Glass, J.I., Lartigue, C., Noskov, V.N., Chuang, R.-Y., Algire, M.A., Benders, G.A., Montague, M.G., Ma, L., Moodie, M.M., Merryman, C., Vashee, S., Krishnakumar, R., Assad-Garcia, N., AndrewsPfannkoch, C., Denisova, E.A., Young, L., Qi, Z.-Q., SegallShapiro, T.H., Calvey, C.H., Parmar, P.P., Hutchison, C.A., Smith, H.O., Venter, J.C.: Creation of a bacterial cell controlled by a chemically synthesized genome. Science 329(May), 52–56 (2010) Forster, A.C., Church, G.M.: Towards synthesis of a minimal cell. Mol. Syst. Biol. 2(45), 45 (2006) Luisi, P.L., Ferri, F., Stano, P.: Approaches to semisynthetic minimal cells: a review. Naturwissenschaften 93(1), 1–13 (2006) Theberge, A.B., Courtois, F., Schaerli, Y., Fischlechner, M., Abell, C., Hollfelder, F., Huck, W.T.S.: Microdroplets in microfluidics: an evolving platform for discoveries in chemistry and biology. Angew. Chem. Int. Ed. 49(34), 5846–5868 (2010)
Synthetic Genomics 19. Synthetic Genomics Inc. Press Release http://www. syntheticgenomics.com/media/press/71409.html, July 14 2009 20. Etc group: Extreme genetic engineering: an introduction to synthetic biology, Etc Group (2007) 21. Khosla, C., Keasling, J.D.: Metabolic engineering for drug discovery and development. Nat. Rev. Drug Discov. 2(12), 1019–1025 (2003) 22. Keasling, J.: Synthetic biology for synthetic chemistry. ACS Chem. Biol. 3(1), 64–76 (2008)
Synthetic Genomics ▶ Synthetic Biology
Synthetic Lubricants ▶ Boundary Lubrication
Systems Level Data Mining for RNAi ▶ Computational Systems Bioinformatics for RNAi
T
TEM ▶ Electron Microscopy of Interactions Between Engineered Nanomaterials and Cells
Terahertz ▶ Terahertz Technology for Nano Applications
Terahertz Technology for Nano Applications Nezih Pala and Ahmad Nabil Abbas Department of Electrical and Computer Engineering, Florida International University, Miami, FL, USA
Synonyms Terahertz; THz; T-rays
Definition The terahertz (THz) region of the electromagnetic spectrum is generally defined as the frequency range of 0.1–10 THz (1012 cycles per second) corresponding to quantum energy of 0.4 meV–0.4 eV (see Fig. 1). THz electromagnetic waves (also known as T-rays) have several properties that could promote their use as sensing and imaging tool. There is no ionization
hazard for biological tissue and Rayleigh scattering of electromagnetic radiation is many orders of magnitude less for THz wavelengths than for the neighboring infrared and optical regions of the spectrum. THz radiation can also penetrate nonmetallic materials such as fabric, leather, and plastic which makes it useful in security screening for concealed weapons. The THz frequencies correspond to energy levels of molecular rotations and vibrations of DNA and proteins, as well as explosives, and these may provide characteristic fingerprints to differentiate biological tissues in a region of the spectrum not previously explored for medical use or detect and identify trace amount of explosives. THz wavelengths are particularly sensitive to water and exhibit absorption peaks which makes the technique very sensitive to hydration state and can indicate tissue condition. THz radiation has also been used in the characterization of semiconductor materials, and in testing and failure analysis of VLSI circuits. THz techniques also allowed art historians to see murals hidden beneath coats of plaster or paint in centuries-old building, without harming the artwork. The lack of efficient sources and detectors in THz frequency range compared to the relatively well-developed lower-frequency RF/microwave and higher-frequency infrared (IR)/far infrared (FIR) ranges was referred as “THz gap” in the scientific community about a decade ago. However, the unique position of THz frequencies in the electromagnetic spectrum has allowed an intense multidisciplinary approach in the development of THz emission and detection techniques as well as their applications, closing the “THz gap” from both sides. Especially rapidly evolving nanosciences and nanotechnology
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
2654
105
108
109
immensely contribute to the advancement of THz technology. These advancements lead to creation of many unique applications some of which in turn contributes to nanotechnology. Therefore, the interaction between THz and nanotechnologies is reciprocal.
Principles of THz Technology Terahertz Sources Taking advantage the above-mentioned unique spectral position THz frequencies, four major approaches have been adopted to develop THz sources. The first approach is pushing the operation frequencies of the existing microwave and millimeter-wave devices into the THz range. The second one is the free-electron-based sources like free-electron lasers and backward wave oscillators (BWOs). The third is the utilization of optical methods which has been the major technique for the demonstration of many THz applications. And the fourth one is the newly developed Quantum Cascade Lasers (QCLs). THz emission power of selected sources are shown in Fig. 2 as a function of frequency. Despite the tremendous research and development efforts, currently available THz source technologies, particularly the tunable emitters, such as free-electron lasers, backward wave oscillators, gas lasers, heterodyne photomixers, optical parametric converters are usually large, complex, require high power, and hence costly. Semiconductor electronic devices such as Gunn oscillators or Schottky diodes need frequency multipliers which severely impact the output power. Quantum cascade lasers which are the true compact solid-state THz emitters are still in their infancy and operate only at cryogenic temperatures.
Far IR
Visible
Millimeter waves, telemetry
107
High frequency Short wavelength High quantum energy
Terahertz
Microwaves, radar
106
Television, FM radio
Low frequency Long wavelength Low quantum energy
Shortwave radio
Terahertz Technology for Nano Applications, Fig. 1 A schematic showing the THz region within the electromagnetic spectrum. Although there is no strict definition of what qualify as THz waves, it is commonly considered the region between 100 GHz and 10 THz
Terahertz Technology for Nano Applications
AM radio
T
UV
X-rays, gamma rays
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 Hz
Solid-State Electronic THz Sources
Solid-state electronic emitters are limited in frequency due to the transient time of carriers which results in high-frequency roll-off. This intrinsic limitation prevented the development of a single electronic device which could oscillate in the entire THz bandwidth of 0.1–10 THz. Gunn diodes (also known as transferred electron devices), IMPATT (Impact Avalanche Transit Time) diodes, and TUNNETT (Tunnel Injection Transit Time) diodes are well-developed high-frequency sources. They take advantage of negative differential resistance (NDR) in current voltage characteristics. When a DC bias large enough to drive such a device into NDR is applied, unstable state triggers oscillations with frequencies up into the mm-wave range. Specific mechanism resulting the NDR determines the upper limit of the oscillations frequency. Current Gunn diodes and IMPATT diodes can reach oscillating frequencies of 400–500 GHz. InP Gunn devices with graded doping profiles generated RF output powers of 283 mW at 412 GHz, 203 mW at 429 GHz and the highest thirdharmonic frequency of 455 GHz with an output power of 23 mW. GaAs TUNNETs, on the other hand, produced 10 mW at 202 GHz. Resonant tunneling diodes (RTDs) are semiconductor devices with double barrier structures which forms a quantum well in between. Electrons in this quantum well are at quantized discrete energy levels. At a particular DC bias, the energy levels at the bottom of the conduction band on one side align with the lowest energy level in the well, resulting in a resonant condition in which electrons can tunnel through the barriers. As the bias increases further, the resonant condition disappears and the current falls. This NDR process combined with the inherently fast tunneling mechanism allows RTDs working at high operation
Terahertz Technology for Nano Applications
T
105 104
IMPATT
III-V laser
p-Ge laser
MMIC
DFG, parametric
1,000 Output power (mW)
Terahertz Technology for Nano Applications, Fig. 2 THz-emission power as a function of frequency. Solid lines are for the conventional THz sources; IMPATT diode stands for impact ionization avalanche transit-time diode, MMIC stands for microwave monolithic integrated circuit, TUNNET stands for tunnel injection transit time, and the multiplexer is an SBD frequency multiplier. Ovals denote recent THz sources. The values of the last two are indicated by peak power; others are by c.w. power (After [1])
2655
QCL 100
Gunn UTC-PD photomixer
10
THz-QCL
1
Lead-salt laser TUNNET
0.1 0.01 0.001 0.01
Multiplexer RTD 0.1
frequencies. RTDs based on InGaAs/AlAs material system have reached the fundamental oscillation frequency of 915 GHz with, however, quite low output level of a few tens of nanowatts. Recently, in parallel to the advancements in material- and device-processing technologies, transistors with very high cutoff frequencies (fT) have also been reported. Northrop Grumman announced an InP high electron mobility transistor (HEMT) with extrapolated fT > 1 THz. This device had sub-50 nm gate length, 25% increase in electron mobility, and reduced contact resistance. Similarly, heterojunction bipolar transistors (HBTs) based on InP materials systems achieved a cutoff frequency fT of 765 GHz at room temperature and 855 GHz at 218 K. As a newly emerged alternative approach, ballistic deflection transistors based on InGaAs-InAlAs heterostructure on an InP substrate were developed at University of Rochester with a theoretical maximum fT of approximately 1.02 THz (Fig. 3a). Another novel approach is the plasma wave instabilities in the FET channels with high sheet carrier concentrations. Such instabilities were predicted by Dyakonov and Shur in 1991. Recently, a group of researchers from Japan and Europe reported room temperature generation of radiation at 0.75 and 2.1 THz from AlGaN/GaN HEMTs (Fig. 3b).
1 10 Frequency (THz)
100
1,000
Due to the limited frequency performance of solidstate emitters, they are often used in combination with multiplication circuits to reach THz frequencies. A multiplier consists of a nonlinear electronic device, such as a Schottky varactor diode placed between an input and an output-matching network. Unfortunately, the output power is much lower than that of the input, which is a serious drawback for the THz frequency range. An input with a power of 200–300 mW at 100 GHz can be produced by HEMT amplifiers, but a multiplier with a high-order of multiplication from 100 GHz up to 1–3 THz is not feasible due to the very high losses. Much lower losses are achievable only in multipliers with a low-order of multiplication, i.e., doublers (2x) and triplers (3x), so that a THz multiplier could consist of a sequence of doublers and triplers of the frequency up to the desired THz frequency. Free-Electron-Based Sources
Free-electron lasers have quite different working principles than more traditional optical lasers such as gas lasers. They can generate either CW or pulsed highpower THz radiation, but they are very costly and have very large dimensions, functioning in large rooms containing many additional facilities which limits their availability to only limited number of facilities around the world. However, backward wave oscillators (BWO) (also called carcinotron) are based
T
T
2656
Terahertz Technology for Nano Applications
b
6 Vds = 4V Emission, u.a
2.1 THz
a
VTD
VG
VLD
3 2
0 Right drain
Left drain
Vgs = 0 V
4
1
Top drain ILD
5
0.75 THz
0 IRD VRD
Vgs = –3.5 V
1 2 3 4 Frequency, THz
Source
deflector
5 Drain
Gate
Field Plate
SiO2 / Si3Nx
Wt VLG
Left gate
Wc
Right gate
VRG
AIGaN
Source VSS
GaN
Terahertz Technology for Nano Applications, Fig. 3 (a) SEM Image of top view of ballistic deflection transistors. After [2]. (b) Schematics of the GaN/AlGaN plasmonic HEMT
with gate covered by a field plate reported in [3]. The inset shows the emission spectra of the device
on the same principles as an electron laser and are able to deliver a few milliwatts in a tunable range of 0.3–1.3 THz with a high sweeping rate. Although BWO requires a water-cooling system due to its high bias voltages of 1–6 kV their size and weight is much smaller (15 kg) which makes them convenient and affordable for many research labs [4]. Recently, there has been an intensifying research effort to develop micromachined Vacuum Electronic Devices, or “micro-VEDs (mVEDs).” Vacuum technology offers a feasible solution for efficient THz transmitters but requires a significant invention to overcome the problem of complex and difficult size scaling necessary to achieve THz operation which calls for advanced fabrication methods and stringent tolerances for interaction structures. Critical developments include the fabrication of the interaction circuit, a high current density integratable cathode structure, and achieving stable high-power electron beam transport. Fortunately, today’s microfabrication methods such as Deep Reactive Ion Etching (DRIE) and LIGA are now capable of manufacturing slow-wave interaction structures at these frequencies with the required resolution and surface roughness characteristics. Further, significant progress has been made in the development of new cathode materials and structures.
Folded waveguide (FWG) approach has emerged as promising approach to design micro-VED high-power amplifiers (HPAs) and THz-regime oscillators. FWG offers sufficient gain and bandwidth to achieve 60 mW at 670 GHz and 10 mW at 1 THz and 56 mW at 560 GHz as THz Oscillator. Another approach is the Extended Interaction Klystron (EIK) approach which applies the concept of multiple interaction structures at offset frequencies to simultaneously achieve high gain from the superposition of multiple structures interacting with the signal and high bandwidth from the gradual offset design of the same structures. This concept is expected to allow design of an HPA that would deliver output power >20 dBm and gain >20 dB at 1.03 THz. The EIK approach requires machining optical quality surfaces over macroscopic areas because the skin depth in the metal at 1 THz is comparable to the previously achievable surface roughness and will result in significant ohmic losses. The issue is complicated by the severe thermal constraints placed on the interaction structure which limits material choices. Optical THz Emission Techniques
Gas lasers are probably the most conventional CW THz sources in the frequency range of 0.9–3 THz and
Terahertz Technology for Nano Applications
with output powers in the range of 1–30 mW. They are typically pumped by a carbon dioxide laser, and the output frequency is determined by the gas in the cavity (CH4, N2, etc.). Although gas lasers are not tunable, they have several discrete emission lines. Photomixing, also known as optical heterodyne down-conversion, is a technique used to generate CW THz radiation by using ultrafast lasers. In its most commonly used version, photoconductive emitters biased at 5 kV/cm are excited by Ti-sapphire lasers, which typically deliver pulse lengths of 50–100 fs, and can be as short as 10 fs with the repetition rate of the order of 50–100 MHz and with the average power in the range of 0.2–2 W. Low-temperature grown (LT), GaAs is usually the material of choice due to its photocarrier lifetime (10 W. Quantum Cascade Lasers
One of the most exciting approaches to generate tunable THz radiation is the quantum cascade lasers (QCLs). QCL idea was proposed in 1971 as a FIR radiation source and experimentally demonstrated in 1994. In a QCL, the light produced by one carrier transition between two levels is amplified due to photon-assisted tunneling of a single type of carriers in a sequence of coupled quantum wells (superlattice) that has a staircase-like band energy. Therefore, it is a unipolar laser where the carriers can be either electrons or holes. The number of amplification stages determines the output power. The discreteness of energy levels, named sub-bands, inside the same band is a result of the spatial confinement of carriers inside the heterostructure, and the radiation frequency is determined by the energy difference of sub-bands between which radiative/lasing transitions occur. The first QCL working in the THz range was reported in 2002 [5]. This laser delivers about 2 mW power at 4.4 THz and operates at 50 K. The output power decreases dramatically with increasing temperature and becomes nearly zero at room temperature. The realization of a QCL at THz frequencies encounters a series of difficulties and limitations due to the very large values of the wavelength. Among them are very large free-carrier absorption losses and the necessity of growing a very thick heterostructure. In 2009, a resonant-phonon terahertz QCL operating up to a record braking heat-sink temperature of 186 K was demonstrated (see Fig. 4). At the lasing frequency of 3.9 THz, 63 mW of peak optical power was measured at 5 K, and approximately 5 mW could still be detected at 180 K. All these THz cascade lasers are based on n-type carriers (electrons), and the photon emission is parallel to the heterostructure plane (edge-emission). Terahertz Detection Systems Present imaging techniques in THz applications are divided between Pulsed Time Domain (PTD) and Continuous Wave (CW) modalities. THz time-domain spectroscopy (THz-TDS) has been the primary PTD
T
T
2658
Terahertz Technology for Nano Applications × 222 Injector barrier Collector barrier
hω 3′ 2′ 1′ E43 = 15.6 meV
4 3 2 1
Terahertz Technology for Nano Applications, Fig. 4 Conduction band diagram of a diagonal design quantum cascade laser with the lasing power of approximately 5 mW at 180 K. 1’ is the injector level from the preceding module. The figure shows that the upper- and lower-state wave functions are localized in separate wells with little spatial overlap and the radiative transition is from 4 ! 3 (f ¼ E43/h ffi 3.8 THz). Therefore, this scheme is called diagonal design (Adapted from [6])
method for demonstration of the potential application in THz technology. A typical THz-TDS system, as shown in Fig. 5, includes an ultrafast laser (e.g., Ti-Sapphire laser) which produces optical pulses in fs duration. Each pulse is separated into two optical paths. One path travel through a time delay stage and hits the emitter such as a photoconductive antenna or a nonlinear crystal in which the optical pulses are converted into ultrashort electromagnetic pulses. The generated EM pulses are focused onto the sample under test and collimated again to be focused onto the detector. The other part of the pulse is also delivered onto the detector directly. Amplitude of the electromagnetic waves are measured by the detector to calculate the change in the waveform due to the sample. By changing delay time between the two beams, it is possible to scan the THz pulse and construct its electric field as a function of time. Subsequently, a Fourier transform is used to extract the frequency spectrum from the time-domain data. Comparison of the waveforms with and without the sample allows the estimation of the complex refractive index of the material, which results other parameters, such as the dielectric constant, conductivity and surface impedance. Although they have been widely used in research facilities, their size and complexity prevent widespread deployment for out-of-lab applications.
On the other hand, CW imaging systems can be passive, where the terahertz or sub-terahertz energy analyzed is present in the environment or emitted by the object itself (i.e., the human body), or active, where the energy is supplied by a dedicated source. In active systems, a major concern is the presently only limited available terahertz sources. This limitation places significant demands on detector capabilities. For CW systems, two primary modalities exist for detection of signals in the terahertz range. One involves the reduction of the terahertz signal to sufficiently low frequencies to allow amplification and signal processing. This (heterodyne receiver) approach typically uses nonlinear diodes to mix the terahertz signal with a local oscillator reference. Heterodyne detection imposes many restrictions on the system design and makes it complex and expensive. These restrictions may be difficult or impossible to fulfill. The second approach is direct detection, where the terahertz signal impinging upon a detecting device results in a measurable response. Available direct detection technologies with key parameters are summarized in Table 1. As it is shown on Table 1, today the most sensitive THz detectors are Golay cells, pyroelectric detectors, bolometers, and Schottky diodes. However, they are not portable, not tunable or, most of the time, are very slow. Schottky diodes, for instance, can reach very low noise equivalent power (NEP) and high responsivity at millimeter waves. However, their responsivity drops orders of magnitude (to 20 kHz, which implies attenuation of most sounds and other laboratory disturbances.
Applications Thermal actuators are one of the key mechanisms for effecting motion in microscale and nanoscale solid structures. Other common microelectromechanical system (MEMS) mechanisms typically use electrostatic forces, hydrostatic forces, magnetic forces, or piezoactuation [6]. The advantages of thermal actuators over these other mechanisms lie in low operating voltage, high output force, and compact footprint. Disadvantages include comparatively high power consumption, comparatively slow response time, and dependence of actuation output on ambient environment. Because thermal actuators dissipate heat, thermal management of the environment surrounding the actuator may be an important design consideration. The substrate supporting a thermal actuator can experience temperature changes due to heating from the actuator. In vacuum, the substrate to which the actuator is anchored becomes the primary mechanism for dissipation of heat from the actuator. Material selection for the substrate should allow for
T
T
2684
temperature changes due to heating, and adequate thermal engineering to prevent overheating. Furthermore, placement of temperature-sensitive materials near the actuator should be carefully planned. Thermal actuators also face high temperature and low temperature limits to operation. At high temperatures, material creep and failure may occur. At low temperatures such as cryogenic environments, heating of the environment due to the actuator may disrupt the environment. Furthermore, some materials such as semiconductors may be insufficiently conductive to allow much current flow. Other materials such as many metals may develop superconductivity at cryogenic temperatures, which leads to insufficient power dissipation to create significant temperature changes for thermal expansion and actuation. In the examples discussed below, the V-type actuator design has an advantage over the hot arm, U-type design in that all beams contribute to actuation motion in the V-type design. In the U-type design, motion is generated by the hot arms, but resisted by the cold arms. As a result, a V-type design requires a smaller footprint to achieve the same output as a given U-type design. Furthermore, the inherently linear motion output of the V-type designs makes them easier to incorporate in device designs. History The hot arm, U-type actuators were originally demonstrated in the early 1990s and quickly became a staple for generation of mechanical motion in solid MEMS structures. In the late 1990s, V-type actuators were initially demonstrated and this design has been subject to much further development in the subsequent years. In particular, V-type actuators have been explored for driving compliant mechanisms, microgripper devices, and micropositioning devices. Bimorph thermal actuators (Example 3, below) have been in use for submillimeter scale applications since the late 1980s. Extensive review of the history of microfabricated thermal actuators may be found in Ref. [6]. Uses for thermal actuators have included optical and electrical switches in addition to micropositioning, microgripping, microscale mechanical testers, and microhandling and nanohandling devices [6, 11, 12]. Thermal actuators may be seen in many reports as the drivers of mechanical motion in many geared and pivoting MEMS systems, particularly in the time period of 1994–2002. Such systems are fundamentally
Thermal Actuators
limited by frictional wear, and consequently, they have not been the subject of much development in more recent years. Newer applications focus more on structures for manipulating and interacting with objects at the microscale and nanoscale.
Examples In application, three main mechanisms have been implemented as the most common thermal actuators: V-type, hot arm/U-type, and bimorph thermal actuators. Other types of thermal actuators have been created [1], but most work with thermal actuators has focused on mechanisms discussed in the examples presented here. Calculations are provided with these examples in order to walk through the basic design of common thermal actuators. These calculations are intended to help the designer to understand the range of actuator performance and the effect of device geometry on the output behavior. Final design should use multiphysics simulations and experimental characterization of fabricated devices in order to understand the real performance of any particular design. In order to facilitate design calculations, Table 1 provides some published values on the properties of polycrystalline silicon (or “polysilicon”), a common material for thermal actuators. Because it is a semiconductor, polycrystalline silicon has electrical resistivity that varies with dopant concentration, temperature, and applied strain. The values reported in this table for resistivity and other properties reflect the range of values that have been used in research literature as inputs to simulations of polysilicon layers generated by the PolyMUMPS foundry service (MEMSCAP Inc., NC, USA). In general, many of the properties of polycrystalline silicon are processdependent. Therefore, designs should ultimately be based upon measurements of the properties of material created according to a specific, defined process, and validation of device models against measured experimental output. Example 1: V-Type Actuators The first common thermal actuator mechanism has been called, variously: V-type, chevron, bent-beam, or symmetric thermal actuators. In this mechanism (Fig. 1), suspended, opposing beams are connected to a suspended shuttle structure. The ends of the actuator
Thermal Actuators
T
2685
Thermal Actuators, Table 1 Some material properties of PolyMUMPS polycrystalline silicon, approximated as an isotropic material [5, 9, 11, 13] Property Electrical resistivity Constant: Temperature dependent:
Thermal conductivity Constant value, polycrystalline Si Constant value, single crystal Si Density Heat capacity Constant value Temperature dependent: Young’s modulus Poisson’s ratio Linear coefficient of thermal expansion Constant value (at 293 K) Temperature dependent:
Symbol
Value
Units
r
1.7 105– 4 104
O-m
r ¼ r0 ½1 þ xðT T0 Þ
r0 ¼ 3.4 105 O-m x¼ K1 3 1.25 10 T0 ¼ 293 K
ky ky
29–34 150 2.33 103
W/m-K W/m-K kg/m3
c c ¼ 1:976 106 T 3 3:767 103 T 2 þ 2:623T þ 215:0 E n
705
J/kg-K J/kg-K GPa
a
2.7 106
i 3 a ¼ 3:725 1 e5:88 10 ðT125Þ þ 5:548 104 T 106 h
Thermal Actuators, Fig. 1 Examples of V-type, “chevron” thermal actuators. (a) Tilted, oblique view in scanning electron microscope (SEM) image. (b) SEM image of the top view of a fabricated V-type thermal actuator. Beams are angled 1 off perpendicular to the direction of motion of the central shuttle (Image credit: J.J. Brown and V.M. Bright)
beams are anchored to a substrate that provides electrical current through the beams. Power dissipation leads to expansion of the beams, which experience a buckling motion. When the beams are placed
162 14 0.29
K1 K1
symmetrically, linear motion is achieved in the shuttle. Engineering of the beams guides the buckling motion that they experience. The aspect ratio of the beam is defined asymmetrically in order to confine beam bending motion to one plane, typically chosen as motion in the plane of fabrication. If the beam has a cross section with thickness t, defined by the thickness of the film used for fabrication of the beam, and a width w, defined by patterning perpendicular to the film, in-plane motion is achieved when t > w. In this case, the beam has a moment of inertia that is smaller for in-plane motion than for out-of-plane motion, so in-plane motion is more probable as buckling is induced. Another technique used to guide the buckling is to angle the beams at their attachment point to the shuttle. Typically the beams are angled a few degrees off perpendicular to the shuttle, and the device is built with mirror symmetry across the shuttle in order to cancel bending moments that would cause deviations from uniaxial motion. With elongation of the actuator beams, buckling and shuttle motion is achieved in a predictable direction. The choice of angle is made dependent on whether displacement or output force is more important. Smaller angles from perpendicular provide more output displacement for
T
T
2686
Thermal Actuators
a given actuator power; larger angles provide more output force.
y u, F
x
Electrical Calculation
The goals of electrical calculations associated with thermal actuator design are the following: 1. Find the total resistance R of the actuator system in order to allow design of an appropriate power supply. 2. Understand how R relates to the actuator geometry. 3. Allow the use of I2R to determine the total power generation and by extension the thermal power that must be managed by the anchors and other structures supporting the thermal actuator. Regarding the third goal, the packaging environment surrounding the actuator must be able to remove the steady-state power generated by the actuator in order to establish stable operation. The fundamental unit of a V-type thermal actuator is a symmetric beam pair as seen in Fig. 2. For simplicity of this example calculation, power consumption in the central shuttle will be omitted from the calculations. Because the actuator is formed from symmetric beams, the calculation only necessitates examination of the power consumption in one beam. The geometry of the beam is defined as in Fig. 3. The beam has an electrical resistivity r, which is a temperature-dependent property. For the purposes of this initial calculation, r will be held constant, and the material of the beam will be defined as polycrystalline silicon, using the property values defined in Table 1. The beam resistance is calculated according to Eqs. 15, using example dimensions for 1 beam: Li ¼ 100 mm, ti ¼ 3.5 mm, wi ¼ 2 mm, Ai ¼ witi ¼ 7 mm2. Ri ¼
rLi Ai
3:4 105 Om ð100 mmÞ ¼ 490 O Ri ¼ ð7 mm2 Þ
(15a)
Anchor
Anchor
Thermal Actuators, Fig. 2 Fundamental unit of the V-type actuator, a pair of angled beams. Motion u and output force F occur in direction y. A loading force -F may oppose motion in the y direction
x′
L I
Ta
T(x)
t
· Q=0
w
x′
Thermal Actuators, Fig. 3 Basic geometry of one of the V-type actuator beams, for electrical and thermal calculations. The distance coordinate along the length of the angled beam is defined by x0
+V I Ri
Ri
Thermal Actuators, Fig. 4 Electrical circuit for the pair of beams in the thermal actuator. Resistance in the central shuttle is omitted for simplicity
(15b)
In Fig. 4, the electrical model of the pair of beams in the actuator indicates connection in series, so the resistance of the pair, Rp ¼ 2Ri. For an actuator built from Np beam pairs, the beam pair resistance adds in parallel, and the total resistance R of the actuator may be found according to Eq. 16. For Np ¼ 10 and the result from Eq. 15b, R¼98 O (Fig. 5).
R¼
Rp 2Ri ¼ Np Np
(16)
Thermal Calculation
The goals of thermal calculations for the V-type actuator are the following: 1. Find an average temperature rise DT at a given input current in order to estimate the thermal strain in subsequent mechanical calculations.
Thermal Actuators
2687
+V I
Ri
Ri
I
Thermal Actuators, Fig. 5 Electrical circuit diagram for an assembly of beam pairs such as that seen in Fig. 1. The resistances of each beam pair add in parallel. The central shuttle makes electrical connections in the center of this diagram, but it is omitted from the diagram because there is assumed to be an identical voltage drop in each beam. Therefore, there should be no current flow from one beam pair to another through the central shuttle
2. Find maximum temperature as a function of input current, in order to understand the performance limits to the actuator. 3. Find boundaries on the thermal time constant in order to understand the limit to the actuator response time. The calculation of the temperature profile in the thermal actuator can focus on the profile within a single beam, similar to the electrical calculation. As discussed earlier, for initial design purposes, it is necessary only to examine thermal conduction. Other heat transfer effects should be considered for a complete understanding of the actuator thermal behavior. Simulations based on the coupled, temperature-dependent electrical, thermal, and mechanical behavior show that V-type thermal actuators may develop nonlinear output force behavior when subjected to displacement constraints [12]. However, for basic understanding, conductive heat transfer generally suffices. As with the electrical calculation, thermal conductivity ky and heat capacitance c are temperature dependent, but for initial design, they are considered to be constant. The calculation can approximate the beam as a one-dimensional element with a temperature varying along position x. The boundary conditions are defined
T
in two locations: The anchored end maintains a constant temperature due to significant heat transfer from the anchor to the supporting substrate. The suspended end of the beam has no heat flow, due to the symmetry of the suspended structure. Consider the diagram in Fig. 3, where Ta is the temperature at point a, the beam has a temperature profile that varies along the beam with position x0 , heat leaves the beam only through point a, and a total current I flows through the beam. This diagram can be simplified into a one-dimensional model (Fig. 6) with the following boundary conditions, where Q_ is one-dimensional heat flow with units of watts (W): Tð0Þ ¼ Ta
_ iÞ ¼ 0 QðL
In this case, all variables in the thermal model are considered constant across the beam cross section at any given x0 position. Using Eqs. 17 to set up the one-dimensional model and assuming steady-state operation, the heat equation (Eq. 5) can be rewritten as Eqs. 18a and 18b, given for cases without and with air conduction, respectively. In Eq. 18a, constant thermal conductivity is also assumed. For the model with conduction through air to a nearby substrate, S is a shape factor, zg is the distance between the suspended device and the substrate, ky,air is the thermal conductivity of air, and Z is a thermal insulance term defined by the layers through which heat flows (with thickness zi, thermal conductivity ky,i, and total number p) from the suspended beam to a substrate held at a constant temperature [5–7, 13]. k y Ai
@T ¼ Q_ @x0
(17a)
@ Q_ _ i ¼ qA @x0
(17b)
2 I2 q_ ¼ r J~ ¼ r 2 Ai
(17c)
t 2zg S¼ þ1 þ1 w t
(17d)
Z¼
p X zg zi ¼ k k y;air j¼1 y;i
(17e)
T
T
2688
Thermal Actuators
x′
a
b L
I
T(x) Ta
· Q= 0
Thermal Actuators, Fig. 6 One-dimensional model of the symmetric thermal system for the V-type actuator
0 ¼ k y Ai
@2T @2T I2 _ þ qA ¼ k A þ r i y i Ai @x0 2 @x0 2
@ @T I 2 SðT Ta Þ 0 ¼ 0 ky 0 þ r 2 @x @x tZ Ai
DT ¼
I 2 L2i 2A2i ky
rI 2 L2i 3A2i ky
Li cL2i ðcLi Ai Þ ¼ ky A i ky 4 2 ð705Þð2330Þ 10 ¼ 5:5 104 s ¼ 30
ty ¼ R y C ¼ (18a)
(18b) fy ¼
The temperature profile in the beam in vacuum, Eq. 19, is found by integrating Eq. 18a twice and applying the boundary conditions to determine the constants of integration. A more precise analytical solution to Eq. 18a may be found when ky and r are allowed to vary with temperature. Ref [7] provides a solution for Eq. 18b for constant ky and temperaturedependent r. Elimination of air conduction terms in that solution provides a solution for Eq. 18a with temperature-dependent r. The maximum temperature Tmax is found at the location where the first derivative of temperature is zero, which is where heat flow Q_ is zero. In this case, heat flow is zero at x0 ¼ Li and Tmax is given by Eq. 20. The average temperature rise in the beam can be calculated by the application of Eq. 19 to Eq. 7, yielding Eq. 21. ! 02 I2 x T ¼ Ta þ r 2 L i x0 (19) 2 Ai k y Tmax ¼ Ta þ r
from the actuator without undergoing significant changes in temperature, thereby maintaining Ta at the beam anchor point. Equations 22a and 22b show the resulting time constant and the minimum bandwidth for operation in vacuum, derived from Eqs. 8, polysilicon properties from Table 1, and example dimensions for 1 beam: Li ¼ 100 mm, ti ¼ 3.5 mm, wi ¼ 2 mm, Ai ¼ witi ¼ 7 mm2. In air, the overall thermal resistance is reduced, thereby decreasing ty and increasing the bandwidth.
(20) (21)
For an upper bound on the thermal time constant, the lumped thermal resistance Ry can be calculated using the beam length Li and assuming that the substrate is an infinitely large thermal reservoir that can adequately remove the generated thermal power P
1 ¼ 290 Hz 2pty
(22a)
(22b)
Analysis of Mechanical Behavior
For basic design of a thermal actuator, modeling of the mechanical behavior can help the designer to understand the maximum force and range of displacements that are available from a given actuator design. Mechanical analysis can be performed with the following goals: 1. Find an expression for the maximum force output available from the actuator. 2. Find an expression for the maximum displacement from the actuator. 3. Develop an understanding of how device geometry impacts the force and displacement. Several contrasting mechanical analyses of V-type actuators have been published [9, 10, 14, 15]. The analysis presented here follows that of Ref. [9]. As with the electrical and thermal analyses, the symmetric device geometry allows simplification of the mechanical analysis to an examination of only one beam. Begin by simplifying the actuator design in Fig. 2 to the free body diagram seen in Fig. 7a, and consider the coordinate system presented in Fig. 7b. At point a, the beam is anchored and restricted from displacement or rotation. At point b, the end of the beam is restricted from rotation or motion in the x direction, but it can move in the y direction. A reaction force Fx,R opposes motion in the x direction. Figure 7b depicts two Cartesian coordinate systems.
Thermal Actuators
2689
a
b
uy , Fy
y′
x′
b
j
x
a
c
du
y′
Fy
ux ′ Fx ′
j
Thermal Actuators, Fig. 7 Diagrams for mechanical analysis of V-type actuators. (a) Free body diagram. (b) Coordinate systems for output motion in xy and motion in the reference frame of the beam x0 y0 . (c) Bending and extensional forces in the beam can be projected into forces in the xy coordinate system. (d) Motion in the xy coordinates is projected into motion in the x0 y0 coordinates
The xy system rests in the frame of the substrate and actuator support. The actuator output motion uy is defined in the y direction of the xy coordinate system. The rotated x0 y0 coordinate system is a local system oriented to the reference frame of the angled actuator beam. Consequentially, the x0 y0 system is rotated at an angle of ’ from the xy system. An external force Fi may be applied to point b in the y direction, affecting the uy displacement. In the reference frame of the beam, motion at point b can be decomposed to motion ux0 in the x0 direction and motion uy0 in the y0 direction (Fig. 7d). Motion in the x0 direction is pure elongation, described by Eq. 23a to include thermal strain. Motion in the y0 direction is pure bending, described by Eq. 23b for the case where rotation at the beam ends is not allowed.
u 0 EA x Fx0 ¼ EA ux0 aDTEA aDT ¼ L L 12EIz u y0 L3
(25a)
Fy;b ¼ Fy0 cos ’
(25b)
þ Fx0 cos ’ ¼ 0
(26a)
12EIz Fx;R sin ’ 3 uy0 L EA ux0 aDTEA cos ’ ¼ 0 þ L
(26b)
Fi þ Fy;b þ Fy;e ¼ Fi þ Fy0 cos ’ þ Fx0 sin ’ ¼ 0
(26c)
Fi þ cos ’
12EIz EA 0 0 u u þ aDTEA sin ’ ¼ 0 y x L L3 (26d)
The displacements ux0 and uy0 can be written in terms of ux and uy by projecting ux and uy onto the rotated coordinate system (Fig. 7d). ux0 ¼ uy sin’ ux cos’
(27a)
uy0 ¼ uy cos’ ux sin’
(27b)
However, ux¼0 at point b, so Eqs. 27 simplify to Eqs. 28. ux0 ¼ uy sin’
(28a)
uy0 ¼ uy cos’
(28b)
(23a)
(23b)
The forces Fx0 and Fy0 can be projected (Fig. 7c) to forces in x and y according to Eqs. 24 for elongation and Eqs. 25 for bending. Fx;e ¼ Fx0 cos ’
Fx;b ¼ Fy0 sin ’
Fx;R þ Fx;b þ Fx;e ¼ Fx;R Fy0 sin ’
ux
Fx
F y0 ¼
(24b)
Summing the Fx forces and Fy forces yields Eqs. 26.
uy
Fy ′ j
Fy;e ¼ Fx0 sin ’ y
Fx,R j
T
(24a)
With substitution of Eqs. 28 into Eq. 26d and some algebraic manipulation, a very useful final expression Eq. 29 can be derived that relates the average actuator temperature rise DT, the actuator output displacement uy, and the force Fi that may be produced by the actuator and which may constrain the displacement. Substitution of Eqs. 28 into Eq. 26b would yield an expression for the x direction reaction force, if desired. Ref. [9] appeared to show reasonable agreement between this model and measured unconstrained
T
T
2690
Thermal Actuators
experimental displacement output. However, an exploration of force and displacement behavior performed in Ref. [12] demonstrated that V-type actuators undergo asymmetric buckling which complicates the relationship between force and displacement at different power or temperature levels. Therefore, Eq. 29 may be useful for basic design, but it underrepresents the interaction between Fi and uy when uy is constrained near zero or when large force is applied. uy ¼ EA L
aDTEA sin ’ Fi þ EA 2 12EIz 12EIz 2 2 2 sin ’ þ L3 cos ’ L sin ’ þ L3 cos ’
(29) If the force constraint Fi is left at zero, Eq. 29 reduces to an expression for the free-moving actuator displacement. If instead the displacement uy is constrained to zero, the magnitude of the maximum force generated by the beam (Eq. 30) is seen to be the result of plane strain in the beams. A detailed thermomechanical derivation of stresses due to plane strain and thermal expansion applied to Eq. 1 may be found in Ref. [2]. The result is that, under the normal strain condition e33 ¼ 0 and no external traction or internal body forces, no stresses are present except for the compressive normal stress s33 ¼ aEDT. Fi; max ¼ aDTEAsin’
(30)
If the thermal strain aDT is held to zero, then a spring constant ki for the beam (Eq. 31) can be determined using Hooke’s Law. This spring constant reduces to Eq. 10 in the small angle approximation where ’ ! 0. For actuators that contain N angled beams, the total spring constant k can be found according to Eq. 13 by multiplying ki by N, and the maximum output force from the actuator would be found by multiplying Eq. 30 by N. ki ¼
Fi EA 2 12EIz sin ’ þ 3 cos2 ’ ¼ L uy L
(31)
A final comment about the mechanical analysis of V-type actuators is that it may sometimes be necessary to include a temperature-dependent coefficient of thermal expansion in the average thermal strain calculated as an input to Eq. 29. Because the mechanical analysis presented here only concerns
motion at the end of the actuator beam, the thermal strain can be determined as an average over the length of the beam using a modified version of Eq. 7, as shown here in Eq. 32. aDT ¼
1 L
Z
L
aðT Ta Þdx0
(32)
0
Example 2: Hot Arm Actuators The second common thermal actuation mechanism may be referred to by any of the following names: pseudo-bimorph, U-type, asymmetric, hot arm, or flexure electrothermal actuators. In these structures (Fig. 8), two parallel beams are provided. These beams are fixed to a substrate at one end, and linked to each other at the opposite end. The beams are designed to experience differential thermal expansion such that one beam expands further than the other beam. These actuators can be designed for motion out of a plane of manufacture, but they are more commonly seen for motion within planes. For in-plane motion with one common design, a “hot arm” and a “cold arm” are defined. The cold arm typically consists of a narrow flexure near the attachment point of the arm, and a wider distal portion. The hot arm is a narrow beam of uniform cross section. Electric current flows through this structure. In the wide portion of the cold arm, the current density and thermal resistance are lower than in the other portions of this actuator due to a larger cross-section area. This region is thus subject to less temperature change and less thermal expansion than the other regions. With the flow of electric current, the cold arm elongates less than the hot arm. Because the two arms are constrained to move together, the two arms sweep an arc of motion due to the difference in thermal expansion between them. By symmetrically placing these U-type actuators around a central shuttle or yoke, and linking them to the shuttle with compliant structures, linear motion can be generated.
Electrical Analysis
Electrical and thermal analysis of these actuators can be performed by dividing the actuator into four segments, as seen in Fig. 9. Ignoring small discrepancies at the boundaries to the segments, the actuator can be converted into a one-dimensional model similar to
Thermal Actuators
T
2691
Thermal Actuators, Fig. 8 SEM images of U-type, “hot arm” thermal actuators. (a) Tilted, oblique view of an actuator array. (b) Magnified view of one of the actuators in this array. The actuator is anchored at left and greatest motion occurs in the parts at right. (c) Top view of an array of fabricated U-type thermal actuators (Image credit: J.J. Brown and V.M. Bright)
that of Example 1, but constructed from a series connection of the four distinct segments, as seen in the electrical circuit diagram, Fig. 10. The total electrical resistance can be calculated, Eq. 33, by summing in series the electrical resistance for each segment and assuming constant resistivity. The resistance for each segment can be found according to Eq. 15a. In most cases, the thickness will be identical for all segments due to fabrication from a surface-micromachined material layer, so the thickness term can be moved outside of the summation along with the resistivity. For a polycrystalline silicon device with typical dimensions of L1 ¼ 100 mm, L2 ¼ 5 mm, L3 ¼ 80 mm, L4 ¼ 20 mm, w1 ¼ w4 ¼ 2 mm, w2 ¼ 5 mm, w3 ¼ 20 mm and t ¼ 3.5 mm, and with r ¼ 3.4 105 Om, the total resistance R is 632 O. R¼
n X i¼1
Ri ¼
n X rLi i¼1
Ai
¼r
n X Li w t i¼1 i i
L1 a
b L2 c
d
e L4
L3
Thermal Actuators, Fig. 9 Schematic of one hot arm actuator
+V I
R2
R1
R4
R3
Thermal Actuators, Fig. 10 Simplified electrical schematic for a hot arm actuator
T
(33)
I
Thermal Analysis
The thermal analysis of the hot arm actuator can proceed similarly to the electrical analysis, dividing the actuator into four segments according to the schematic in Fig. 11. The four segments are coupled together with temperature Ti and heat flow Q_ i in each segment i, and with temperature and heat flow boundary conditions described in Table 2. The mechanical anchor points are maintained at constant temperature Ta. Uniform
Tb b
Ta a 1 L1 A1
· Qb
Tc c 2 L2 A2
· Qc
Ta e
Td d 3 L3 A3
· Qd
4 L4 A4
Thermal Actuators, Fig. 11 Schematic for thermal analysis of the hot arm actuator
T
2692
Thermal Actuators
Thermal Actuators, Table 2 Boundary conditions for the hot arm actuator thermal model Temperature boundary conditions T1(0) ¼ Ta
Heat flow boundary conditions Q_ 1 (L1) ¼ Q_ 2 (0)
T1(L1) ¼ Tb ¼ T2(0)
Q_ 2 (L2) ¼ Q_ 3 (0) Q_ 3 (L3) ¼ Q_ 4 (0)
T2(L2) ¼ Tc¼ T3(0) T3(L3) ¼ Td ¼ T4(0) T4(0) ¼ Ta
current flow I and constant electrical resistivity are assumed through the actuator. Solution of Eq. 18a following a similar approach as in Example 1 leads to equations for temperature and heat flow in each segment, given by Eqs. 34, with constants C1i and C2i determined for each ith segment by algebraic manipulation of the boundary conditions. The constants are reported in Table 3. This analysis is similar to the approach of Ref. [7], with the difference that Ref. [7] also examines heat transfer to air in solving the heat equation for the hot arm actuator. Ti ¼
rI þ C1i xi þ C2i A2i ky 2 2
x2i
@T rI Q_ i ¼ ky Ai ¼ ky Ai C1i þ xi @xi Ai
(34a)
2
(34b)
The maximum temperature Tmax and its location xmax in the hot arm segment 1 can be found by setting the first derivative of temperature equal to zero. The first derivative of the temperature is proportional to the heat flow, given in Eq. 34b using C11. The resulting temperature and position are given in Eq. 35 and Eq. 36, respectively. Tmax ¼ T1 ðxmax Þ ¼ Ta þ ¼ Ta þ
I 2 r x2max 2ky A21
I 2 r 1 L1 L2 L3 L4 2 þ þ þ (35) 2ky 4 A1 A2 A3 A4
xmax ¼
A1 L1 L2 L3 L4 þ þ þ 2 A1 A2 A 3 A4
(36)
The average temperature rise DT and the thermal time constant can be calculated as described earlier,
subject to the discretion of the designer in choosing the complexity of the system to be represented. One option is to focus solely on the hot arm because this is the segment that experiences the highest temperatures and the most significant thermal strain. In the hot arm, the thermal behavior is most important for the actuator operation. In this case, the thermal capacitance C is calculated using the volume of the hot arm Vm ¼ L1A1. The thermal resistance can be calculated using the location xmax of maximum temperature or the total length L1 of the hot arm, and the hot arm cross-section area A1. For the average temperature rise DT, the average is taken over the length L1 of the hot arm only, giving Eq. 37. 1 DT ¼ L1 ¼
Z
L1
T1 dx Ta
0
I 2 r L1 L1 L2 L3 L4 þ þ þ 4ky A1 3A1 A2 A3 A4
(37)
Analysis of Mechanical Behavior
Few mechanical analyses of hot arm actuators have been published. The results of analyses based on energy methods may be found in Refs. [7, 10]. The analysis presented here uses a more straightforward technique based on superposition. Begin by dividing the device depicted in Fig. 9 into three sections: two cantilevers and one rigid frame. As seen in Fig. 12, section 1 becomes a cantilever with clamped attachment at left and applied force F1y and moment M1 at right. Similarly, section 4 becomes a cantilever with shorter length, and with clamped attachment at left and applied force F4y and moment M4 at right. Sections 2 and 3 become an L-shaped rigid frame that links the two cantilevers together with moment and force reactions at the joining surfaces. Additionally, an external force FY is applied at point b of this L-frame, and the frame also experiences a bending moment M due to thermal stresses in the cantilevers. In order to maintain simplicity in calculating the coupled bending behavior of the cantilevers for this example, this thermal bending moment is calculated separately rather than combining the axial deformation behavior of the cantilevers with the rotation of the rigid frame and the bending of the cantilevers. To calculate M from the thermal strain in cantilever 1, use the sum of forces in the x direction, the sum
Thermal Actuators
2693
Thermal Actuators, Table 3 Algebraic constants for Eqs. 34 in each segment of the hot arm actuator
C1i C11 C12 C13 C14
a
qb
x FY
uyb
1
2 e
4
d
c
b
3 M Fy
Thermal Actuators, Fig. 12 Diagram for mechanical analysis of the hot arm actuator. The lower part of this diagram illustrates the matching of moments and forces at points b and d where the rigid frame is cut from the two cantilevers
of moments at point c, and Hooke’s law applied to elongation of the two cantilevers in order to generate Eqs. 38. For simplicity, thermal strain is assumed only to be present in cantilever 1. Algebraic manipulation leads to an expression for the displacement of the rigid frame, Eq. 39a, and an expression, Eq. 39b, for the thermal bending moment that drives operation of the hot arm actuator. Fxb þ Fxd ¼ 0
(38a)
M Fxb L2 ¼ 0
(38b)
Fxb ¼ k1 e1 ¼
A1 E ðux aDTL1 Þ L1
(38c)
A4 Eux L4
(38d)
Fxd ¼ k4 e4 ¼
C2i C21 ¼ Ta
I2 r L1 L2 L3 L4 ¼ þ þ þ 2ky A1 A1 A2 A3 A4
I2 r L1 L2 L3 L4 ¼ þ þ þ 2ky A2 A1 A2 A3 A4
I2 r L1 L2 L3 L4 ¼ þ þ 2ky A3 A1 A2 A3 A4
I2 r L1 L2 L3 L4 ¼ þ 2ky A4 A1 A2 A3 A4
y
aDTA1 ux ¼
A1 A4 þ L1 L4
(39a)
T
I 2 r L1 L2 L3 L4 þ þ 2ky A1 A2 A3 A4
I 2 r L1 L2 L3 L4 ¼ Ta þ þ þ 2ky A1 A2 A3 A4
I 2 r L1 L2 L3 L4 ¼ Ta þ þ þ 2ky A1 A2 A3 A4
C22 ¼ Ta þ C22 C24
aDTA1 L2 E M¼
1 þ AA14 LL41
(39b)
In order to calculate the actuator motion, begin by examining the rigid frame and recording a set of force and moment equilibrium equations, and compatibility equations that describe the relationship between the rotation and displacements of the cantilever ends (Eqs. 40). Note that the small angle approximation is used to simplify Eq. 40d to the relation in Eq. 40e. FY F1y F4y ¼ 0 M4 M1 þ M þ L3 ðFY F1 Þ ¼ 0
(40a) (40b)
yb ¼ yd ¼ y
(40c)
uyd þ L3 sin y ¼ uyb
(40d)
uyd ¼ uyb L3 y
(40e)
The bending of cantilevers is a well-characterized problem, and nodal equations may be written that relate the force and moment applied to the free end of the cantilever to the displacement and rotation that are present at that end [8]. For the cantilevers of sections 1 and 4 of the hot arm actuator, these nodal equations are provided in Eqs. 41. EI1 12uyb 6L1 y 3 L1
(41a)
EI1 6L1 uyb þ 4L21 y L31
(41b)
F1y ¼ M1 ¼
T
T
2694
Thermal Actuators
EI4 12uyd 6L4 y L34
(41c)
EI4 6L4 uyd þ 4L24 y 3 L4
(41d)
F4y ¼
M4 ¼
By combining Eqs. 40 and 41, expressions may be derived that relate the force and moment applied in the rigid frame to the displacement and rotation experienced by the end of the actuator. These relations, Eqs. 42, may be further manipulated to provide an expression for the actuator displacement as a function of applied force and applied moment, Eq. 43. Using the equation for M found in Eq. 39b, the expression in Eq. 43 relates input temperature and output force and displacement for the hot arm actuator. Similar to Eq. 29 in Example 1, Eq. 43 may be further manipulated to find a spring constant for the actuator and a maximum output force.
FY I1 I4 ¼ 12uyb 3 þ 3 E L1 L4
I1 2L3 I4 I4 6y 2 þ 3 þ 2 L1 L4 L4
(42a)
M I1 2L3 I4 I4 ¼ 6uyb 2 þ 3 þ 2 E L1 L4 L4
I1 I4 L3 L23 þ 2y 2 þ þ 6I4 2 þ 3 L1 L4 L4 L4
h
(42b) i
L2 I1 þ LI42 þ 2LL33I4 þ F6EY 2 LI11 þ LI44 þ 6I4 LL32 þ L33 L21 4 4 4 4 uyb ¼
2
2
12L3 I1 I4 1 L3 I1 I4 4I1 I4 1 1 1 þ L2 þ L1 L4 L1 L4 þ L2 L2 L1 L4 þ L1 L4 L21 4 1 4 M 2E
(43) Experimental verification of this or related hot arm actuator models has not been clearly performed, as most hot arm actuator experiments have recorded data in air environments, which require different thermal analysis from that performed in this example. In Ref. [7], experimental displacement measurements appear to correspond fairly well with a curve predicted from input current, but with DT calculated to include the effects of heat loss to air. A final point about this
mechanical analysis of hot and cold arm actuators is that variations of these actuators exist, but similar analytic approaches can apply to these variations. For instance, if L3 is set equal to zero, then Eq. 43 describes the motion of a two arm actuator that has no wide region but does have different length cantilevers and different temperature profiles in each cantilever due to the length difference [16]. The out-of-plane pseudobimorph actuator could be analyzed similarly [6]. Example 3: Thermal Bimorph Actuators The bimorph thermal actuator predates the designs explored in the previous two examples. The principle of operation of this device is the generation of bending due to a differential in thermal expansion between two materials stacked together to form the actuator. The bimorph actuator is typically constructed as shown in Fig. 13, as a cantilevered beam composed of two (bimorph) or more (multimorph) dissimilar materials, and with a resistive heater patterned on top of the beam or formed from one of the beam materials. Advantages of bimorph structures are high force, simple design, and low operating voltage. The major disadvantages of these structures are relatively high electrical power consumption, curling motion, low operating frequency, and laminated construction. Many examples of implementation of bimorph thermal actuators, including more discussion of simulation and analytical models, may be found in Ref. [6]. Power consumption and average temperature rise can be calculated by applying concepts from Example 1. Others have extensively described the mechanical behavior of bimorph thermal actuators [6, 10, 17]. This example will review the relevant mechanical concepts and results. According to Fig. 13, assume the lengths of the two layers in the actuator are identical length L. Using the small angle approximation, deflection due to uniform curvature in the cantilever is given by Eq. 44, where r is a radius of curvature in the cantilever that is generated due to the thermal expansion mismatch of the two materials. Superimposed on this displacement is the deflection due to force applied at the end of a cantilever, Eq. 45, where EIi is the average flexural rigidity of the composite cantilever. Together, these give a total displacement described by Eq. 46, analogous to Eqs. 29 and 43 presented earlier.
Thermal Actuators
2695
Thermal Actuators, Fig. 13 Illustration of bimorph thermal actuators. (a) Bimorph cantilever consisting of two materials with identical length, but different thicknesses, widths, and composition. (b) Diagram of radius of curvature r and cantilever tip displacement uy used in analysis of the output motion of
L2 2r
(44)
FL3 3EIi
(45)
L2 FL3 þ 2r 3EIi
(46)
uy1 ¼ uy2 ¼
uy ¼
r¼
E1 w1 t21
2
2 þ E2 w2 t22 þ 2E1 w1 t1 E2 w2 t2 2t21 þ 3t1 t2 þ 2t22 6E1 w1 t1 E2 w2 t2 ða1 a2 ÞDT
(47) EIi ¼ E1 I1 þ E2 I2
the bimorph actuator. (c) SEM image of a bimorph actuator 300 mm length, with a heater trace patterned on its top surface (Image credit: B. Read and V.M. Bright, Air Force Institute of Technology, 1994)
EIi ¼
E1 w1 t21
2
2 þ E2 w2 t22 þ 2E1 w1 t1 E2 w2 t2 2t21 þ 3t1 t2 þ 2t22 12ðE1 w1 t1 þ E2 w2 t2 Þ
(49)
The radius of curvature r is determined (Eq. 47) using force and moment equilibrium in the composite beam, and strain compatibility at the bonded surface between the two materials [10, 17]. The combined flexural rigidity EIi is found for the nonhomogeneous beam according to Eq. 48 after the moments I1 and I2 are first calculated for the different layers in relation to the neutral surface of the beam [8, 10]. The result is Eq. 49. In Ref. [17], Eqs. 47 and 44 are derived and shown to lead to agreement with previously reported experimental data. Eqs. 45, 46, and 49 are derived through continuum solid mechanics analysis [8].
T
(48)
Future Directions Miniaturization The phenomena of electrical power dissipation, heat flow, and thermal expansion are valid concepts at submicron size scales. Continued miniaturization may require consideration of additional phenomena such as the effects of dimensional constraints on the motion of phonons and electrons, increased importance of surface forces with smaller sizes, and the Thomson effect in systems where electrical current is flowing [18]. It may be expected that thermal actuator designs will be demonstrated in submicron sizes as fabrication technology improves. Ultimately, thermal actuator designs formed from engineered macromolecular structures may be possible. As molecular-scale thermal actuators are developed, such devices may not be subject to the same friction conditions as current microscale technology, and thus pinned connections and sliding surfaces may be within the range of designs allowed at the nanoscale.
T
T
2696
Environmental Robustness As discussed earlier, the temperature profile of the thermal actuator, and the corresponding output displacement or force, can vary due to the environment in which the actuator is operated. In vacuum environments at ambient temperature, this may be a small effect, but operation in ambient air and gaseous environments can require significant modification to the actuator power. For reliable and predictable operation across a range of environments, improved control systems for thermal actuators are needed. Additionally, many possible studies of environment effects comparing experiment with simulations and analytical models remain to be explored for thermal actuators. These may, for instance, include effects of vacuum, liquid, high temperature, high radiative heat transfer, low temperature, packaging design, and environmental transients.
Thermal Actuators
(MEMS and NEMS), it may be expected that a major thrust of thermal actuator development will track the overall development and use of MEMS and NEMS. Thermal actuators are an important part of the toolkit available to microsystem designers. As new designs proliferate, it may be expected that a certain portion of these designs incorporate thermal actuators. As discussed earlier, current usages include optical and electrical switches, microgrippers, nano manipulation structures, and tensile test devices. As these tools move into broader use, thermal actuators are likely to remain key components of these devices. Other key opportunities for new application development using thermal actuators include control of optical gratings, lock and security systems, and controlled motion in microscale positioning stages. Many MEMS and NEMS applications may yet be conceived requiring low actuation voltage, compact footprint, and high force output, and thermal actuators provide a useful design option for all such applications.
Device Control and Sensor Integration One approach to improved displacement control and robustness would be implementation of closed-loop control systems that rely on data from a displacement sensor integrated to the actuator. In practice, integration of accurate in-plane displacement sensors for actuator control presents a significant design and fabrication challenge. Capacitive sensors have been integrated to thermal actuators, but these sensors typically require a large footprint to achieve useful sensitivity at small displacements [9]. Operation of capacitive sensors must be driven by sophisticated electronics placed physically close to the sensor in order to avoid signal losses due to parasitic capacitance. Piezoresistive sensors require smaller device footprints than capacitive sensors, but may be subject to distorted readings due to temperature variation resulting from heat dissipation from the thermal actuator. Furthermore, readings from piezoresistive sensors can be distorted by the electrical operation of the thermal actuator if the sensor is monolithically linked and thus electrically connected to the thermal actuator. Strategies for electrical isolation of piezoresistive sensors are needed. Optical sensors provide the most accurate options for displacement measurement, but they are not yet easily integrated with other chip scale devices for in-plane displacement measurements.
Cross-References
Application Development As one of the key mechanisms for generating motion in microscale and nanoscale electromechanical systems
▶ Basic MEMS Actuators ▶ Compliant Mechanisms ▶ Thermally Actuated Silicon Resonators
Force Output Measurement and Calibration The use of thermal actuators in applications where force is applied to microscale and nanoscale objects raises the question, “Exactly how much force is applied?” Answering this question requires calibration and measurement of the force applied with the actuator. No standardized, traceable approaches exist for microscale force calibration, and current best practices rely on measurement of the displacement of a well-characterized spring. Such an approach relies on the characterization of the spring, which itself is not a traceable, standardized practice for microscale and nanoscale devices. Furthermore, many analytical and finite element mechanical models have been presented in literature, but more experimental verification of these models is needed, as the observations of Ref. [12] show. As thermal actuators and other actuators are employed for generation of motion in microscale and nanoscale devices, further strategies are needed for understanding the magnitude of forces within these devices.
Thermal Cancer Ablation Therapies Using Nanoparticles
References 1. Lu, S., Dikin, D.A., Zhang, S., Fisher, F.T., Lee, J., Ruoff, R.S.: Realization of nanoscale resolution with a micromachined thermally-actuated testing stage. Rev. Sci. Instrum. 75, 2154–2162 (2004) 2. Fung, Y.C., Tong, P.: Classical and Computational Solid Mechanics. Advanced Series in Engineering, vol. 1. World Scientific, Singapore (2001) 3. Jackson, J.D.: Classical Electrodynamics, 2nd edn. Wiley, New York (1975) 4. Mills, A.F.: Heat Transfer, 2nd edn. Prentice Hall, Upper Saddle River (1999) 5. Lott, C.D., McLain, T.W., Harb, J.N., Howell, L.L.: Modeling the thermal behavior of a surface-micromachined linear-displacement thermomechanical microactuator. Sens. Actuators A Phys. 101, 239–250 (2002) 6. Geisberger, A.A., Sarkar, N.: Techniques in MEMS microthermal actuators and their applications. In: Leondes, C.T. (ed.) MEMS/NEMS Handbook: Techniques and Applications, Sensors and Actuators, vol. 4, pp. 201–261. Springer, New York (2006) 7. Huang, Q.-A., Lee, N.K.S.: Analysis and design of polysilicon thermal flexure actuator. J. Micromech. Microeng. 9, 64–70 (1999) 8. Craig R.R. Jr.,: Mechanics of Materials, 2nd edn. Wiley, Hoboken (2000) 9. Zhu, Y., Corigliano, A., Espinosa, H.D.: A thermal actuator for nanoscale in situ microscopy testing: design and characterization. J. Micromech. Microeng. 16, 242–253 (2006) 10. Lobontiu, N., Garcia, E.: Mechanics of Microelectromechanical Systems. Kluwer, New York (2005) 11. Brown, J.J., Suk, J.W., Singh, G., Baca, A.I., Dikin, D.A., Ruoff, R.S., Bright, V.M.: Microsystem for nanofiber electromechanical measurements. Sens. Actuators A Phys. 155, 1–7 (2009) 12. Wittwer, J.W., Baker, M.S., Howell, L.L.: Simulation, measurement, and asymmetric buckling of thermal microactuators. Sens. Actuators A Phys. 128, 395–401 (2006) 13. Howell, L.L., McLain, T.W., Baker, M.S., Lott, C.D.: Techniques in the design of thermomechanical actuators. In: Leondes, C.T. (ed.) MEMS/NEMS Handbook: Techniques and Applications, Sensors and Actuators, vol. 4, pp. 187–200. Springer, New York (2006) 14. Gianchandani, Y.B., Najafi, K.: Bent-beam strain sensors. J. Microelectromech. Syst. 5, 52–58 (1996) 15. Que, L., Park, J.-S., Gianchandani, Y.: Bent-beam electrothermal actuators – part I. Single beam and cascaded devices. J. Microelectromech. Syst. 10, 247–254 (2001) 16. Huang, Q.-A., Lee, N.K.S.: Analytical modeling and optimization for a laterally-driven polysilicon thermal actuator. Microsyst. Technol. 5, 133–137 (1999) 17. Chu, W.-H., Mehregany, M., Mullen, R.L.: Analysis of tip deflection and force of a bimetallic cantilever microactuator. J. Micromech. Microeng. 3, 4–7 (1993) 18. Jungen, A., Pfenniger, M., Tonteling, M., Stampfer, C., Hierold, C.: Electrothermal effects at the microscale and their consequences on system design. J. Micromech. Microeng. 16, 1633–1638 (2006)
2697
T
Thermal Cancer Ablation Therapies Using Nanoparticles Steven Curley Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA Department of Mechanical Engineering and Materials Science, Rice University, Houston, TX, USA
Synonyms Heat; Magnetic electromagnetic radiation; Near infrared; Radiofrequency
Definition Nanoparticle-sized materials have interesting physical, chemical, and electronic properties. Among these is the release of heat upon exposure to various types of electromagnetic radiation. For example, iron and other magnetic nanoparticles can be injected directly into malignant tumors and then release heat in response to frictional heating in an alternating magnetic field. The intrinsic plasmon resonance of gold has been utilized to produce near-infrared laser-induced thermal activation of gold nanoparticles to treat malignant lesions on the skin or mucosal surfaces. Finally, several types of conducting and semiconducting nanoparticles have been conjugated successfully to antibodies or peptides to target the nanoparticles to malignant tumor cells or to the tumor neovasculature. The malignant cells are then exposed to a noninvasive shortwave radiofrequency field which results in intracellular heating and thermal cytotoxicity to the malignant cells. Cancer cells generally are more sensitive to thermal injury compared to normal cells; thus, nanoparticle-based thermal tumor ablation has the potential to be an important and hopefully less toxic addition to current anticancer treatments.
Overview Hyperthermic cancer therapy was first described in Egyptian papyri over 4,000 years ago by applying hot
T
T
2698
oil or cautery to tumors. Modern hyperthermic treatments are now applied to both premalignant and malignant tumors in various locations such as the peritoneal cavity, liver, kidneys, lungs, and prostate. Hyperthermia also potentiates the effects of cytotoxic chemotherapeutic agents and improves the response of tumors to ionizing radiation. Thus, regional or systemic hyperthermia has been induced as part of a treatment regimen in patients with certain types of malignancies. Hyperthermic isolated limb perfusion with melphalan with or without dactinomycin is performed to treat in-transit melanoma metastases and is being investigated as a treatment for advanced extremity soft tissue sarcoma [1]. In the last several years, there have been a growing number of reports on hyperthermic intraperitoneal chemotherapy combined with cytoreductive surgery for patients with diffuse peritoneal spread of colorectal cancer, gastric cancer, mesothelioma, or pseudomyxoma peritoneii [2]. Although effective at treating several cancer types, recent studies have shown that some mammalian cancer cells are sensitive, whereas others are significantly more resistant to hyperthermia compared to healthy cells [3]. The reasons for the variable sensitivity to hyperthermia in cancer are multifactorial, but are primarily due to the role of heat shock proteins (HSPs) in cancer’s resistance to the physical stress of increasing temperature [3]. Mammalian cancer cells that are heat sensitive frequently have mutated or underexpressed HSPs, whereas deregulation in some malignant cells leads to increased expression of HSPs which confers increased thermal protection and reduces apoptosis following thermal stress. Many hyperthermic devices have been designed to produce low-level total-body hyperthermia (42–46 C) and have been used to treat a wide variety of advanced malignancies combined with systemic cytotoxic agents. Both regional and systemic hyperthermic treatment systems elevate tissue temperatures from 3 C to 7 C above normal over a 1–12-h period. Some populations of malignant cells are sensitive to relatively long durations of exposures to low-level hyperthermia, but results to date suggest that these regional and systemic hyperthermic treatments have produced little benefit in overall patient outcomes at the expense of major regional and systemic toxicities and side effects. Interestingly, exposure of mammalian cells to temperatures of 55–60 C for periods of a few seconds
Thermal Cancer Ablation Therapies Using Nanoparticles
to less than a minute has been shown to produce apoptosis and necrosis, so a treatment system that can achieve such brief periods of heating in cancer cells would be highly desirable.
Invasive Thermal Ablation Therapies Thermal tumor ablation is routinely practiced, particularly for malignant liver tumors, through direct intratumoral insertion of radiofrequency (RF) needle electrodes, microwave-emitting probes, or end- and side-firing laser fibers [4]. RF ablation therapy is an invasive treatment that is implemented by inserting needle electrodes directly into the tumor(s) to be treated and applying RF current through the wire into the tumor, resulting in local tumor and adjacent tissue necrosis (Fig. 1). The basis for the observed heating phenomena is understood by the rotational response of the charged ions and proteins in the tissue under the RF field. This rapid molecular rotation results in local frictional heating of the aqueous environment of the tumor and death of tumor cells through protein denaturation, melting of lipid bilayers, and destruction of nuclear DNA. Although invasive RFA has demonstrated efficacy in cancer therapy, there are several limitations to this approach. First, clinical studies have shown incomplete tumor destruction in 5–40% of the treated lesions, allowing significant probability of local cancer recurrence and metastasis. Another limitation of RFA is that the treatment is not targeted (i.e., nonspecific). Placement and insertion of the needle electrode is usually guided by an imaging system with poor spatial resolution (e.g., ultrasonography mm), and therefore, this method is dependent on the experience of the physician that implements RFA. Furthermore, RFA treatments result in significant complications in approximately 10% of patients due to thermally induced necrosis of healthy noncancerous tissues. The final and the most problematic limiting factor with respect to wire-based RFA is that it is a clinical treatment that is indicated for just a few select organ sites and specific tumor types (e.g., liver, kidney, breast, lung). On the other hand, tissue penetration by noninvasive RF energy fields is known to be excellent. Theoretically, noninvasive RF treatment of malignant tumors at any site in the body should be possible if agents that preferentially respond to RF energy with
Thermal Cancer Ablation Therapies Using Nanoparticles
2699
T
Thermal Cancer Ablation Therapies Using Nanoparticles, Fig. 1 Schematic illustration demonstrating invasive radiofrequency ablation of malignant liver tumors. A multiple array needle electrode is inserted directly into the tumor to be treated using ultrasound guidance to assure proper placement of the needle. An alternating current is then passed across the array resulting in frictional ionic heating in the tumor and tissue surrounding the array
intracellular release of heat can be delivered to the malignant cells.
Roles for Nanotechnology in Cancer Diagnosis and Therapy Nanotechnology is a term used to describe the engineering of materials that utilize the unique physical properties of nanometer scale matter. Nanomaterials have been used across all the major scientific disciplines including chemistry, physics, and medicine. It is in the medical realm that nanotechnology is widely heralded as one of the most promising and important approaches to diagnose and treat cancer [5]. Some nanoscale materials are already FDA approved and are used clinically. For example, iron oxide nanoparticles (Feridex™) have been shown to enhance the diagnostic capability of magnetic resonance imaging (MRI) to detect cancer [6]. Conjugating iron nanoparticles to antibodies that target proteins expressed on the surface of human cancer cells may further enhance the accuracy of MRI to diagnose early stage cancer [6]. Other nanoparticles such as carbon or polymeric nanoparticles labeled with fluorine-18 deoxyglucose have been studied in preclinical models to enhance tumor diagnosis and detection rates using positron emission tomography. Surface modification of quantum dots, which are semiconducting nanocrystals that fluoresce upon absorption of visible light, is being investigated to improve detection of lymph nodes and
other sites of metastases during surgical procedures [7]. Additionally, conjugation of quantum dots with tumor-specific peptides or antibodies may improve targeting of cancer cells and thus improve the diagnostic accuracy of this technique. Optimal imaging techniques using fluorescent nanoparticles targeted to a variety of types of human cancer with immunoconjugation to antibodies are being studied to permit in vivo localization of malignant cells. It is hoped that such imaging techniques will improve the diagnostic accuracy in numerous types of imaging modalities used to detect and follow patients with cancer. It is possible that these techniques may also allow earlier detection of cancer in high-risk populations and guide the duration and type of therapy in patients with more advanced stages of malignant disease. Finally, immunocomplexes consisting of gold nanoparticles and labeled antibodies have been demonstrated to improve the detection of several known serum tumor markers, including carcinoembryonic antigen, carcinoma antigen 125, and carbohydrate antigen 19–9, in a more rapid and accurate fashion than currently available techniques. Nanoparticles are being used as intravascular carriers to enhance delivery of anticancer agents to malignant cells. Given that most anticancer drugs have a narrow therapeutic index with significant acute and cumulative toxicities in normal tissues, a variety of types of nanoparticles are being studied in preclinical and in vitro experiments to increase the delivery of cytotoxic chemotherapy drugs to cancer cells while
T
T
2700
reducing toxicity by limiting exposure of the drug to normal tissues. For example, nanoparticles loaded with hydroxycamptothecin, 5-fluorouracil, docetaxel, and gemcitabine have been used in preclinical studies to improve chemotherapy-induced cytotoxicity in lung, colon, squamous, and pancreas cancers. Colloidal gold nanoparticles bearing tumor necrosis factor-a (TNF-a) molecules are being used in early phase human clinical trials to treat several types of cancer. Preclinical studies demonstrated that delivery of TNF-a to malignant tumors was enhanced using the nanoparticle delivery system, while avoiding the systemic toxicities that usually limit the clinical utility of this biologic agent. Nanoparticles conjugated with targeting agents are also being investigated to deliver gene therapy payloads to malignant cells. Carbon nanotubes have been used to deliver genes or proteins through nonspecific endocytosis in cancer cells. Similar to some cytotoxic chemotherapy drugs, gold or other metal nanoparticles have been shown to improve the therapeutic efficacy of external beam ionizing radiation in preclinical models. In addition to being used to deliver pharmaceutical agents, nanoparticles may be used as intrinsic chemotherapeutic agents. Gold nanoparticles 5–10 nm in diameter have been shown to have antiangiogenic properties in solid tumors and appear to induce apoptosis in chronic lymphocytic B cell leukemias through mechanisms not elucidated entirely.
Nanoparticle-Based Thermal Tumor Ablation Strategies Recent studies have shown that nanoscale materials are capable of enhancing thermal deposition in cancer cells by absorbing optical energy. Gold-silica nanoshells 150 nm in diameter have a particular plasmon resonance in near-infrared (NIR) wavelengths (650–950 nm) of light and release significant heat when exposed to NIR laser sources (Fig. 2) [8]. In a murine colon carcinoma subcutaneous tumor model, it was demonstrated that intravenous injection of polyethylene glycol-coded gold-silica nanoshells followed 6 h later by illumination of the tumor with a diode laser (8.0 nm, 4 W/cm2, 3 min) led to complete destruction of the malignant tumors. Animals treated with laser treatment alone without prior intravenous injection of the nanoshells continued to grow at a rate
Thermal Cancer Ablation Therapies Using Nanoparticles
similar to control animals who received neither nanoshell injection or laser treatment. The authors point out that this is a nontargeted delivery of the gold-silica nanoshells to tumors, and the size of the nanoparticles (130–150 nm) corresponds to the size of the enlarged pores and fenestrations in the neovasculature of tumors. The gold-silica nanoshells circulate and a significant number arrest in these pores, thus allowing thermal destruction of the vasculature of the tumor under laser irradiation. A subsequent study of prostate cancer cells using 110-nm gold nanoshells injected intravenously demonstrated that there was a nanoshell dose-related response. The tumors were again treated with a near-infrared laser at 4 W/cm2 for 3 min. Tumors treated with laser therapy alone continued to grow at a rate similar to untreated control tumors, while tumors treated at the highest dose of gold nanoshells were over 93% necrotic following a single laser treatment. A clinical trial using gold-silica nanoshell hyperthermia following NIR light exposure has been initiated for patients with oropharyngeal malignancies. Thermal therapy with NIR energy is limited to the treatment of superficial tumors (less than 1–2-cm deep) due to the significant attenuation of NIR light by biologic tissues. Iron oxide nanoparticles are also being evaluated to produce alternating current magnetic field-based hyperthermia in tumors, but the limited thermal enhancement of these nanoparticles in a magnetic field requires extremely high concentrations of iron oxide [9]. Human breast carcinomas implanted into mice underwent intratumoral injection of colloidal suspensions of iron oxide particles 10–200 nm in diameter. Twenty minutes after the injection, the animals were exposed to an alternating current magnetic field (6.5kA/m; frequency, 400 kHz) for 4 min. Real-time temperature monitoring revealed that temperatures increased from 12 C to 73 C at various areas in the tumor. Subsequent analysis of the tumor showed some areas of coagulative necrosis, but other areas of viable tumor remained. Subsequent studies have attempted to address the issue of variable distribution of the superparamagnetic iron oxide nanoparticles through different regions of the tumor. Tumor cells near blood vessels are more difficult to treat with the magnetic nanoparticles because of the heat-sink phenomena of heat being carried away by blood flow in adjacent vessels coupled with the diffusion of nanoparticles in the tumor microenvironment into the
Normalized Extinction (arb.)
Thermal Cancer Ablation Therapies Using Nanoparticles
2701
T
1.0 0.8 0.6 0.4 0.2 0.0 500
600 700 800 Wavelength (nm)
900
1000
Thermal Cancer Ablation Therapies Using Nanoparticles, Fig. 2 Left column – Transmission electron micrographs of 10nm diameter solid gold nanoparticles (top) and 150-nm diameter gold nanoshells (bottom). TEM images are 20,000 magnification and 500,000 (inset) magnification, and scale bars ¼ 1,000 nm (main) and 100 nm (inset). Right column –
Diameter-dependent visible spectra of solid gold nanoparticles (black ¼ 10 nm, red ¼ 100 nm, blue ¼ 150 nm) and the redshifted gold nanoshell spectrum (green ¼ 150 nm). Nanoparticle illustrations above each corresponding spectra are drawn to scale. Note, spectra have been normalized, and background subtracted to ease viewing of relative peak positions
blood vessels. One of the principle issues and limitations is that the iron oxide particles must be injected directly into the tumor. While this allows high concentrations of the magnetic nanoparticles for subsequent magnetic field thermal treatment, it is an invasive technique and allows for only treatment of tumors that can be identified through imaging techniques or during an operation. The magnetic particles can be used as imaging vehicles for tumors, but there are problems with diffusion of the nanoparticles out of the tumor and into the surrounding normal tissues. These high concentrations and the difficulty targeting the iron oxide nanoparticles to only cancer cells lead to destruction of normal (nonmalignant) cells surrounding the tumor, and diffusion of nanoparticles out of some areas of the tumor may cause insufficient heating to produce thermal cytotoxicity. It was recently demonstrated that gold or carbon nanoparticles and a confined, noninvasive shortwave RF field produces significant local heat release that can be used to produce thermal cytotoxicity in cancer cells in vitro and in vivo with RF field treatments of 2 min or
less [10, 11]. The heat release from gold nanoparticles exposed to shortwave RF fields is particularly impressive and is dependent on both the concentration and diameter of gold nanoparticles (Fig. 3). Solid gold nanoparticles smaller in diameter (5–10 nm) heated at a more rapid rate than gold nanoparticles 50–250 nm in diameter. In addition, this method does not suffer the limitations of other nanoparticle-based hyperthermia systems given that shortwave radiofrequency (RF) energy has excellent deep tissue penetration, low tissue specific absorption rates (SAR), and documented safety for brief exposures of humans to this form of electromagnetic radiation [12]. Concerns have been expressed about possible mutagenic alterations in cells exposed to nonionizing electromagnetic radiation in the GHz range. However, studies of shortwave RF from 10 to 100 MHz has not been associated with any cytogenetic or cytotoxic damage [13]. Exposure of volunteer test subjects to prolonged periods of shortwave RF irradiation can produce minimal thermoregulatory effects (vasodilation, sweating) but has not
T
T
2702
Thermal Cancer Ablation Therapies Using Nanoparticles
Heating Rate (ºC s−1)
3
2 150 nm nanoshell 5 nm gold 10 nm gold 50 nm gold 100 nm gold 150 nm gold 250 nm gold
1
0
0
1e-6
2e-6
3e-6
Gold Volume Fraction
Thermal Cancer Ablation Therapies Using Nanoparticles, Fig. 3 At gold nanoparticle concentrations less than 10 ppm, the heating rate of deionized water in a 13.56 MHz RF field was almost 3 C/s. This remarkable heating rate at very low concentrations of gold nanoparticles represents a heating efficiency of >30,000 W/g of nanoparticles and indicates the generation of elevation temperatures in the nanoscale environment surrounding the gold nanoparticles sufficient to denature proteins, melt lipid bilayers, and produce irreparable damage to intracellular structures and organelles
been found to cause any acute or long-term damage to normal cells and tissues. Thus, the finding that charged gold nanoparticles release heat in response to shortwave RF exposure for durations of seconds to a few minutes has important implications. The physical mechanism of this phenomenon is being investigated, and data suggests that conductive polarizations of metal nanoparticles may explain the high thermal efficiency and tremendous heat release at low concentrations of nanoparticles exposed to the RF field. Importantly, since tissue penetration by shortwave RF fields is excellent with low attenuation of the signal, it should be possible to treat tumors at almost any site in the body. The crucial component of this treatment will be successful delivery of adequate numbers of gold nanoparticles to cancer cells to produce intracellular temperatures of at least 55 C following an RF field exposure of no more than a few minutes duration, while ideally limiting uptake of the nanoparticles by normal cells. It has shown that by directly conjugating 5-nm gold nanoparticles to cetuximab to target epidermal growth factor receptors (EGFRs) expressed on cancer cells, there is enhanced uptake of the gold nanoparticles (Fig. 4). In contrast, cells expressing little or no
Thermal Cancer Ablation Therapies Using Nanoparticles, Fig. 4 (a). Transmission electron microscope (TEM) image of Panc-1 human pancreatic adenocarcinoma cells ( 7,500 magnification) that were incubated for 30 min with unconjugated, nontargeted 5-nm gold nanoparticles (control group). Almost no gold nanoparticles are seen within the cytoplasm of the cell, and it is only at a magnification of 100,000 (inset upper right) that an intracytoplasmic vesicle containing a small number of gold nanoparticles is seen. Incubation of Panc-1 cells for 60 or 120 min with the unconjugated gold nanoparticles also produced only a few scattered vesicles in the cytoplasm containing gold nanoparticles. (b). TEM image of Panc-1 cells incubated for 30 min with nonspecific IgG-conjugated gold nanoparticles (control group). Similar to the unconjugated gold nanoparticles, the nonspecific IgG-conjugated gold nanoparticles were not readily taken into the Panc-1 cells. The larger image ( 7,500 magnification) demonstrates few or no obvious intracytoplasmic vesicles containing gold nanoparticles. At higher magnification ( 100,000, inset upper right), an occasional intracytoplasmic vesicle with a few gold nanoparticles is evident. (c). TEM image of Panc-1 pancreatic adenocarcinoma cells incubated for 30 min with C225-conjugated 5-nm gold nanoparticles. At 30 min, a number of intracytoplasmic vesicles (dense black vesicles) containing gold nanoparticles are evident in the larger image ( 7,500 magnification). At a higher magnification ( 100,000), numerous vesicles containing dense collections of gold nanoparticles are found throughout the cytoplasm. At 60 and 120 min of incubation with the C225-targeted gold nanoparticles, the number of intracytoplasmic vesicles containing gold nanoparticles was seen to increase. (d). A higher magnification TEM image ( 100,000) showing details of the cell surface of Panc-1 cells incubated for 30 min with C225-conjugated gold nanoparticles. Numerous gold nanoparticles are seen at the surface of the cell consistent with binding of the C225 to epidermal growth factor receptor (EGFR). Pretreatment of the cells with C225 alone completely blocked any binding of targeted gold nanoparticles to the cell surface and led to no gold nanoparticles appearing in intracytoplasmic vesicles 30, 60, or 120 min after incubation
Thermal Cancer Ablation Therapies Using Nanoparticles
EGFR should have no enhanced uptake of cetuximabconjugated gold nanoparticles and, thus, should not be significantly affected by treatment in the RF field. Targeting gold nanoparticles approximately 5–10 nm in diameter to cancer cells to create RF-induced hyperthermic cytotoxicity has several advantages: gold nanoparticles are simple and inexpensive to synthesize; they are easily characterized due to the presence of a characteristic surface plasmon resonance band; their surface chemistry permits manipulation of charge and shape relatively easily; attaching cancer cell targeting molecules, including antibodies, peptides, or pharmacologic agents, is easily achieved; and they are biocompatible and not associated with any acute or chronic toxicities in preclinical studies [14]. In addition to producing significant heat upon exposure to shortwave RF fields, 5–10-nm gold charged nanoparticles easily penetrate through pores and fenestrations in the neovasculature of solid tumors. This method may be promising for focused local or regional application of a noninvasive RF field to treat primary or established metastatic malignant tumors overexpressing EGFR, but would be problematic for whole-body RF therapy to treat diffuse micrometastatic disease. Constitutive expression of EGFR at normal levels occurs throughout many tissues in the body, and the results in SN12PM6 renal cancer cells (low-level expression of EGFR) show that even with low levels of EGFR expression, there is enough uptake of cetuximab-targeted gold nanoparticles to produce low levels of thermal apoptosis following exposure to the RF field. It may be possible to modulate or minimize effects on normal tissues expressing a target molecule like EGFR by use of brief duration pulsed RF therapy or repeated short duration exposures, but these effects must be studied in greater detail to confirm this theoretical consideration. Nonetheless, these results indicate that it is possible to use overexpression of a specific cell surface moiety, e.g., EGFR, to enhance delivery and uptake of gold nanoparticles to produce thermal cytotoxicity in the malignant cells following brief exposure to a noninvasive RF field. There are numerous other possibilities for targeted delivery of nanoparticles into malignant cells. Radiolabeled carbon nanotubes have been conjugated with rituximab and lintuzumab to diagnose and potentially treat lymphoma cells [15]. Recently, aptamers, nucleic acid ligands much smaller than monoclonal
2703
T
antibodies, have been shown to target specific molecules in the neovasculature of tumors or on the surface of different types of cancer cells [16]. These aptamers have been conjugated to gold nanoparticles initially for use as an improved diagnostic technique. However, co-conjugation of cytotoxic agents, or use of our noninvasive RF field treatment technique, could be used as an actual anticancer therapy using this targeting technique. So-called cell-penetrating peptides, generally less than 100 amino acids in length, also have been shown to target certain types of cancer cells. The 86-amino acid HIV-1 Tat protein has been conjugated to gold nanoparticles, with subsequent rapid intracellular uptake of the nanoparticles and localization to the nucleus [17, 18]. The monoclonal antibody-conjugated gold nanoparticles used by our group are found in the cytoplasm of cells within 30 min of adding these nanoparticle conjugates to cancer cells in culture. Localization of the nanoparticles to the nucleus with thermal destruction of DNA following RF field exposure must be investigated as a technique to enhance cancer cell destruction. Chlorotoxin is another small peptide (36 amino acids) cancer cell agent that can be easily conjugated with gold nanoparticles and is currently being investigated as a cancer cell targeting agent in several human malignant cell lines [19]. Identification of cancer-specific ligands not expressed on normal cells will permit targeting of gold nanoparticles to only malignant cells for treatment in our noninvasive RF field generator and should allow treatment of measurable tumors and micrometastatic disease with minimal therapy-related toxicities. Targeting ligands, like EGFR, that are overexpressed but not unique to cancer cells are feasible but will require careful evaluation to determine acute and longterm thermal effects in normal cells and organs following RF field therapy. Nanoscale thermal therapy of targeted cancer cells is a promising new weapon in the battle against cancer. By combining nonionizing radiation and targeted metallic nanoparticles, noninvasive targeted thermal therapy has the promise of benefiting cancer patients that need a nontoxic alternative to the debilitating effects of conventional chemotherapeutics. Using noninvasive RF fields, magnetic fields, or laser irradiation to heat nanoscale materials that are internalized into cancer cells or the tumor microvasculature has the promise of reducing toxicity and maximizing therapeutic benefit for cancer patients.
T
T
2704
Thermal Chemical Vapor Deposition
References
functionalized, radiolabeled carbon nanotubes. J. Nucl. Med. 48, 1180–1189 (2007) Javier, D.J., Nitin, N., Levy, M., Ellington, A., RichardsKortum, R.: Aptamer-targeted gold nanoparticles as molecular-specific contrast agents for reflectance imaging. Bioconjug. Chem. 19, 1309–1312 (2008) Berry, C.C.: Intracellular delivery of nanoparticles via the HIV-1 tat peptide. Nanomed 3, 357–365 (2008) Berry, C.C., de la Fuente, J.M., Mullin, M., Chu, S.W., Curtis, A.S.: Nuclear localization of HIV-1 tat functionalized gold nanoparticles. IEEE Trans. Nanobioscience 6, 262–269 (2007) Mamelak, A.N., Jacoby, D.B.: Targeted delivery of antitumoral therapy to glioma and other malignancies with synthetic chlorotoxin (TM-601). Expert Opin. Drug Deliv. 4, 175–186 (2007)
1. Beasley, G.M., Ross, M.I., Tyler, D.S.: Future directions in regional treatment strategies for melanoma and sarcoma. Int. J. Hyperthermia 24, 301–309 (2008) 2. Stewart, J.H.T., Shen, P., Russell, G., Fenstermaker, J., McWilliams, L., Coldrun, F.M., Levine, K.E., Jones, B.T., Levine, E.A.: A phase I trial of oxaliplatin for intraperitoneal hyperthermic chemoperfusion for the treatment of peritoneal surface dissemination from colorectal and appendiceal cancers. Ann. Surg. Oncol. 15, 2137–2145 (2008) 3. Kampinga, H.H.: Cell biological effects of hyperthermia alone or combined with radiation or drugs: a short introduction to newcomers in the field. Int. J. Hyperthermia 22, 191–196 (2006) 4. Arciero, C.A., Sigurdson, E.R.: Diagnosis and treatment of metastatic disease to the liver. Semin. Oncol. 35, 147–159 (2008) 5. Hartman, K.B., Wilson, L.J., Rosenblum, M.G.: Detecting and treating cancer with nanotechnology. Mol. Diagn. Ther. 12, 1–14 (2008) 6. Neumaier, C.E., Baio, G., Ferrini, S., Corte, G., Daga, A.: MR and iron magnetic nanoparticles. Imaging opportunities in preclinical and translational research. Tumori 94, 226–233 (2008) 7. Misra, R.D.: Quantum dots for tumor-targeted drug delivery and cell imaging. Nanomed 3, 271–274 (2008) 8. Gobin, A.M., Lee, M.H., Halas, N.J., James, W.D., Drezek, R.A., West, J.L.: Near-infrared resonant nanoshells for combined optical imaging and photothermal cancer therapy. Nano Lett. 7, 1929–1934 (2007) 9. Kalambur, V.S., Longmire, E.K., Bischof, J.C.: Cellular level loading and heating of superparamagnetic iron oxide nanoparticles. Langmuir 23, 12329–12336 (2007) 10. Gannon, C.J., Cherukuri, P., Yakobson, B.I., Cognet, L., Kanzius, J.S., Kittrell, C., Weisman, R.B., Pasquali, M., Schmidt, H.K., Smalley, R.E., Curley, S.A.: Carbon nanotube-enhanced thermal destruction of cancer cells in a noninvasive radiofrequency field. Cancer 110, 2654–2665 (2007) 11. Gannon, C.J., Patra, C.R., Bhattacharya, R., Mukherjee, P., Curley, S.A.: Intracellular gold nanoparticles enhance non-invasive radiofrequency thermal destruction of human gastrointestinal cancer cells. J. Nanobiotechnol. 6, 2 (2008) 12. Adair, E.R., Blick, D.W., Allen, S.J., Mylacraine, K.S., Ziriax, J.M., Scholl, D.M.: Thermophysiological responses of human volunteers to whole body RF exposure at 220 MHz. Bioelectromagnetics 26, 448–461 (2005) 13. Klima, J., Scehovic, R.: The field strength measurement and SAR experience related to human exposure in 110 MHz to 40 GHz. Meas. Sci. Rev. 6, 40–44 (2006) 14. Daniel, M.C., Astruc, D.: Gold nanoparticles: assembly, supramolecular chemistry, quantum-size-related properties, and applications toward biology, catalysis, and nanotechnology. Chem. Rev. 104, 293–346 (2004) 15. McDevitt, M.R., Chattopadhyay, D., Kappel, B.J., Jaggi, J.S., Schiffman, S.R., Antczak, C., Njardarson, J.T., Brentjens, R., Scheinberg, D.A.: Tumor targeting with antibody-
16.
17. 18.
19.
Thermal Chemical Vapor Deposition ▶ Chemical Vapor Deposition (CVD)
Thermal Conductance ▶ Thermal Conductivity and Phonon Transport
Thermal Conductivity and Phonon Transport Yee Kan Koh Department of Mechanical Engineering, National University of Singapore room: E2 #02-29, Singapore, Singapore
Synonyms Heat conduction; Heat conductivity; Nanoscale heat transport; Thermal conductance; Thermal resistance; Thermal resistivity; Thermal transport
Definition Thermal conductivity (symbols: k or L) is an empirical material property that relates local heat flux (J) across
Thermal Conductivity and Phonon Transport
a real or imaginary surface to the temperature gradient (∂T/∂x) at the same point in direction x normal to the surface. Normally, the concept of thermal conductivity is used for heat conduction not mediated by massive advection of atoms and molecules, or propagation of electromagnetic waves; such heat transport is traditionally referred to as convective and radiative heat transfer, respectively. Thermal conductivity is defined according to the Fourier’s law of heat conduction given below, J ¼ k
@T @x
The negative sign indicates that heat flow against temperature gradient, i.e., from the hot side to the cold side. The units commonly used for thermal conductivity include watts per meter-Kelvin (W m1 K1) and watts per centimeter-Kelvin (W cm1 K1). The reciprocal of thermal conductivity is called the thermal resistivity.
Overview As summarized by the second law of thermodynamics, Nature seeks to approach equilibrium in an isolated system, be it a mechanical, chemical, or thermal system. In the context of heat transfer, heat flows from the hot side to the cold side whenever the system is not in thermal equilibrium (i.e., temperature is not uniform), until all temperature difference is ultimately eliminated. Although the second law of thermodynamics is very general and applies to any thermal system, it does not however stipulate how equilibrium is reached. In other words, although heat transfer occurs whenever temperature difference exists according to the second law of thermodynamics, the rate of heat transfer differs from systems to systems depending on the properties of materials or mediums in the thermal system. Historically, heat transfer is classified as conduction, convection, and radiation, depending on the mediums involved and the mechanisms in which heat is transported across the mediums. Conduction usually refers to heat transfer in solids, in which atoms and molecules are practically immobile and diffusion of atoms and molecules is relatively slow (as long as the solids are not at highly elevated temperatures). In such
2705
T
cases, heat is predominantly carried by vibration of atoms, or other quasiparticles in the crystals such as electrons and holes (see more discussions on quasiparticles below). Also, in still fluids, heat transfer is due to vibration and collisions of atoms and molecules, and thus is also categorized as conduction. On the other hand, the term convection is mainly used for heat transfer from a solid surface to fluids, mediated by diffusion or advection (macroscopic motion) of molecules and atoms. In convection, fluids are driven either by external means such as pumps (forced convection) or by buoyancy forces (natural convection). Lastly, when heat is carried by electromagnetic waves, the heat transfer is called radiation. Unlike conduction and convection, no mediums are required for radiation, since electromagnetic waves can travel through vacuum. For example, heat is transported from the Sun to the Earth by visible or infrared light through radiation. Of course, radiation could also occur through transparent mediums (e.g., glass, air etc.). For heat conduction, the rate of thermal energy transfer per unit time and unit area is proportional to temperature gradient in the respective direction. This empirical relationship is known as the Fourier’s law of heat conduction. Mathematically, J ¼ k
@T @x
where J is the heat flux across a real or imaginary interface (SI unit: W m2) and @T=@x is the local temperature gradient (SI unit: K m1). Thermal conductivity (k) is defined as the proportional constant in the Fourier’s law of heat conduction and has the SI unit of W m1 K1. Thermal conductivity for fully condensed solids spans for more than five orders of magnitude, from k >1,000 W m1 K1 for carbon allotropes [1, 2], to k 200 W m1 K1, amorphous polymer PMMA with k 0.2 W m1 K1, and semiconductor indium arsenide (InAs) with k 27 W m1 K1, being heated uniformly with a constant heat flux of J. The red solid line is the temperature profile in the hypothetical solid. Different slopes of the temperature profile are due to different thermal conductivity of each layer. Temperature drops at the interfaces are due to finite Kapitza thermal conductance of the interfaces
flux applied on a hypothetical multilayered solid as in Fig. 1, temperature drop across each layer is inversely proportional to the thermal conductivity of the layer. For a high thermal conductivity material (e.g., aluminum), the temperature drop is minimal since heat could be easily transported across the material. In other words, thermal energy could easily spread across high thermal conductivity materials, resulting in a near-uniform temperature distribution. On the other hand, low thermal conductivity impedes heat transport, and thus higher temperature difference could be maintained in the materials. This simple understanding of thermal conductivity is crucial for many modern applications (such as thermal barrier coatings and thermoelectrics), in which high temperature differences enhance the performance of the devices. Researchers in these fields seek to modify the existing materials, often through nanostructuring, to reduce the thermal conductivity so that a higher temperature difference could be maintained in the devices. While the thermal conductivity is a bulk property, interfacial or Kapitza thermal conductance [4] (symbol: G) is a property of interfaces. As heat flows across an interface, a finite temperature drop is observed at the interface, as illustrated in Fig. 1, due to resistance of the interface to thermal transport. Kapitza thermal conductance (G) relates this temperature drop (DT) to heat flux across the interface; J ¼ G DT. Typically,
thermal conductance ranges from G > 1 GW m2 K1 for metal/metal interfaces [5] to G 10 MW m2 K1 for interfaces with highly dissimilar materials [6]. It is important to realize that the Kapitza thermal conductance is not solely due to imperfections of interfaces (e.g., imperfect contact as a result of surface roughness, diffuse interfaces due to intermixing, etc.). Even for atomically perfect interfaces, heat flow is still impeded by finite thermal conductance because differences in the properties (e.g., density of phonon states) of two mediums reduce the probability for heat carriers to transmit across the interfaces. In other words, since interfaces break the reflection and translational symmetry, even perfect interfaces are imperfections by definition. Thus, heat transport is retarded at interfaces due to this break of symmetry engendered by the interfacial imperfections.
Heat Transport by Phonons in Nonmetallic Solids In a microscopic view, heat conduction in solids is due to quasiparticles that carry heat across the solids. In simple words, quasiparticles (especially those important for heat conduction in solids) are quantized excitation modes resulting from quantum-mechanical interaction of atoms and electrons in the periodic lattices. Strictly speaking, some of these quasiparticles are only theoretically defined for perfect crystals, but this restriction is usually relaxed for imperfect crystals. Quasiparticles behave like real particles with properties and behavior modified from their real particle counterparts. Common quasiparticles that are important for heat conduction in solids include phonons, electron quasiparticles and holes. Phonons are quantized lattice vibration modes due to interaction of sound waves in solids and are responsible for heat conduction in most nonmetallic solids (dielectrics and semiconductors). Electron quasiparticles are quantized excitation modes due to interaction between electrons and periodic electrical potential in the lattices. Electron quasiparticles are the main carriers for both heat and charges in metals, and as a result of electron quasiparticles being the common carriers, the thermal conductivity and electrical conductivity of metals are related by a fundamental relationship called the
Thermal Conductivity and Phonon Transport
Wiedemann–Franz law. Electron quasiparticles also contribute significantly to heat conduction in highly n-doped semiconductors and semimetals. Holes, on the other hand, are empty states available in the crystals that are not filled by electrons. Holes behave similar to electron quasiparticles, except that they carry positive charges and have different effective masses. Holes can contribute significantly to heat conduction in highly p-doped semiconductors and semimetals. Note that electron quasiparticles have different behavior (e.g., different effective masses) from stand-alone electrons, even though both carry negative charges and both are fermions. Consistent with the literature and for simplicity, electron quasiparticles are thereafter referred to as “electrons” throughout the text. To understand heat conduction by phonons in crystalline dielectrics and semiconductors, it is important to realize that phonons do not carry same amounts of heat due to different polarizations (longitudinal or transverse) and a broad distribution of frequencies (o) and lifetimes (t) of phonons. Heat conduction by phonons depends on three properties of phonons: 1. The volumetric heat capacity (C) of phonon modes, which affects the amount of heat carried by phonons. The volumetric heat capacity is a function of phonon frequency, polarization, and temperature. At high temperatures, the heat capacity of each phonon mode approaches a classical value of kB, where kB ¼ 1.38 1023 J K1 is the Boltzmann constant. 2. The group velocity (u) of phonons, which determines how fast phonons propagate in the materials and is defined as u¼
@o @k
where k is the wavenumber of phonons. The group velocity depends on the slope of phonon dispersion and thus is a function of phonon frequency (o) and polarization. At low frequencies, the group velocity of acoustic phonons approach the speed of sounds of the materials, while at high frequencies, the group velocity of acoustic phonons approach zero due to interference of reflected waves near the Brillouin zone boundaries. 3. The mean-free-path (‘) of phonons, which is the average distance phonons propagate without being
2707
T
scattered. In crystals, phonons are scattered by imperfections including impurities, anharmonicity, and grain boundaries. This scattering of phonons is the source of retardation for heat conduction in crystals, since perfect crystals have infinitely high thermal conductivity. Normally, the scattering process is quantified by the relaxation time (t) or the mean-free-path (‘) of phonons; the relaxation time and mean-free-path are related by ‘ ¼ ut, where u is the group velocity of phonons. Both the relaxation time and the mean-free-path strongly depend on the frequency of phonon modes. At high temperatures, most phonon modes are excited, but heat is carried mainly by a small fraction of these excited modes (long-wavelength acoustic phonons). Acoustic phonons near the edge and the center of the Brillouin zone, and optical phonons are inefficient in carrying heat due to their low group velocity or heat capacity. The mean-free-paths of these heat-carrying acoustic phonons, however, are not constant but span more than an order of magnitude in most crystals depending on phonon frequency. For example, calculations [7] show that 90% of heat could be carried by phonons with 0.1 ‘(o) 1 mm in Si. The mean-free-paths of phonons are also sensitive to the purity and harmonicity of materials, since imperfections, such as impurities, point defects and anharmonicity, scatter phonons. In fact, the disparity in the thermal conductivity of different materials is mainly due to differences in the mean-free-paths (instead of the heat capacity or the group velocity) of phonons in these materials, since the heat capacity and phonon velocity of different materials do not vary by more than an order of magnitude. Despite the importance, understanding of the meanfree-paths of phonons is still incomplete, mainly due to lack of a direct method to measure the mean-free-paths of phonons [8]. The mean-free-paths of phonons are usually inferred from systematic measurements of the thermal conductivity by comparing the experimental results to theoretical calculations. Figure 2 shows an example of calculations of the mean-free-paths of acoustic phonons in bulk silicon, a 200 nm silicon thin film and bulk Si0.1Ge0.9 alloy, as a function of phonon frequency. Note that different imperfections are efficient in scattering phonons of different frequency. At room temperature, phonons in bulk Si are predominantly scattered by anharmonicity, which is
T
T
2708
Thermal Conductivity and Phonon Transport
Thermal Conductivity Reduction in Nanostructures
10 300 K
1
(μm)
Bulk Si
Si Thin Film
0.1
Si0.9Ge0.1 T 0.01 1
10 ω (1012 rad s−1)
L 100
Thermal Conductivity and Phonon Transport, Fig. 2 Mean-free-path (‘) of longitudinal (red) and transverse (blue) acoustic phonons in bulk Si (solid lines), a 200 nm Si thin film (dashed lines) and bulk Si0.9Ge0.1 alloy (dash-dotted lines) calculated from a Callaway-type model [8]. o is the frequency of phonons. The dominant scattering mechanisms in bulk Si, the SiGe alloy, and the Si thin film are three-phonon process (anharmonicity), Rayleigh scattering, and boundary scattering, respectively
effective to scatter high-frequency phonons but not low-frequency phonons, see Fig. 2. Similarly, Rayleigh scattering, which is the dominant scattering mechanism for phonons in alloys, is more effective to scatter high-frequency phonons than low-frequency phonons. On the other hand, boundary scattering is effective to scatter low- and medium-frequency phonons. Over the past decade, nanostructures, such as thin films, superlattices, and nanowires, have been widely used to scatter the low-energy phonons to further reduce the thermal conductivity. It is stressed that the scattering rate and mean-free-paths of phonons in most materials are not known – calculations as in Fig. 2 are at best just an approximation. In fact, polarization of the dominant heat carriers in silicon was determined as recent as a decade ago [9].
Over the past decade, researchers advanced the understanding of heat conduction in nanostructures through systematic measurements of the thermal conductivity of nanostructures. This understanding on nanoscale heat transport facilitates the design of more efficient thermoelectric materials with reduced thermal conductivity. In this section, prior experimental studies on heat conduction by phonons in different nanostructures, including crystalline thin films, superlattices, nanowires, and semiconductor embedded with nanoscale precipitates, are briefly and selectively review. Goodson and coworkers [9, 10] reported the inplane thermal conductivity of Si thin films over a wide film thickness (741.6 mm) and temperature (20–320 K) range. The in-plane thermal conductivity was measured by the 3o method [11], which is discussed below. Heat was previously perceived to be carried by transverse acoustic phonons in Si, but Ju and Goodson [9] find from the in-plane thermal conductivity measurements of Si thin films that longitudinal acoustic phonons with mean-free-path ‘ 300 nm are the dominant carriers in Si at 300 K. Yao [12] first reported the cross-plane thermal conductivity of AlAs/GaAs superlattices, measured using an ac calorimetric method. He finds that the thermal conductivity of a (AlAs)5 nm(GaAs)5 nm superlattice approaches the thermal conductivity of AlGaAs alloys at room temperature. The observation that the thermal conductivity of superlattices approaches the thermal conductivity of the alloys is not universal. For example, Lee and coworkers [13] reported the thermal conductivity of fully strained Si/Ge superlattices, measured by the 3o method. They find that for period 10 nm, however, the defect density is so high that the thermal conductivity approaches the amorphous limit. Further measurements [14, 15] on other Si/Ge nanostructures are consistent with measurements by Lee et al. Venkatasubramanian [16] reported the thermal conductivity of Bi2Te3/Sb2Te3 superlattices as a function of superlattice period and find that the thermal conductivity of short-period superlattice is minimum when the period is 4 nm. Subsequent measurements on similar samples by Touzelbaev and coworkers [17], however,
Thermal Conductivity and Phonon Transport
do not present similar minimum thermal conductivity. The minimum thermal conductivity is also not observed on AlN/GaN superlattices [18], in which the AlN/GaN interfaces are atomically sharp and chemically abrupt. Li and coworkers [19] first measured the thermal conductivity of Si nanowires using a microfabricated device. Li et al. find that the thermal conductivity of nanowires is reduced by more than an order of magnitude when the diameter of the nanowire is 22 nm. The approach has subsequently been applied to study the thermal conductivity of other nanostructures, including rough Si nanowires [20]. The reduction of thermal conductivity, in most cases, is due to enhanced boundary scattering. Harman and coworkers [21] estimated the thermal conductivity of PbTe/PbSe nanodot superlattices from their thermoelectric devices. They found that the thermal conductivity is 0.33 W m1 K1, a factor of 3 lower than the thermal conductivity of the corresponding alloys. More complete and systematic measurements [22] on similar materials, however, indicate that the thermal conductivity of PbTe/PbSe nanodot superlattices is on the order of 1 W m1 K1, comparable with the thermal conductivity of PbTeSe alloys. The thermal conductivity of InGaAs with ErAs nanoparticles randomly distributed in the InGaAs matrix was first reported by Kim and coworkers [23]. The thermal conductivity of the nanostructured materials is found to be reduced from the thermal conductivity of the alloys by up to a factor of 2, even when the concentration of the ErAs nanoparticles is relatively small (0.3%). This reduction of thermal conductivity is explained by enhanced Rayleigh scattering when the size of the nanoparticles better match the wavelength of phonons. Subsequent measurements [24] on InGaAs with higher concentration of ErAs nanoparticles indicate that the thermal conductivity could be further reduced with higher concentration of ErAs nanoparticles.
Modern Techniques for Measurements of the Thermal Conductivity of Nanostructures For measurements of heat conduction on submicron or nanometer length scales, temperature profile has to be monitored within the similar characteristic length
2709
T
scales (1 mm) or equivalently the temperature decay profile in nanosecond time scales. Conventional approaches to measure the thermal conductivity are incapable to reach these small length and time scales, and thus are not suitable for measurements of the thermal conductivity of nanostructures. In this section, two important techniques that are capable of characterizing the thermal properties of nanostructures, namely, the 3o method and the time-domain thermoreflectance (TDTR), are introduced. The 3o method [11] is widely used to measure the thermal conductivity of bulk solids and thin films. In 3o measurements, a narrow metal line (often Au or Pt with Cr or Ti as the adhesion layer) of 400 nm thickness and 30 mm width is patterned on a sample. Electrical current i of frequency o is applied to the metal line with electrical resistance r, generating joule heating of i2r within the metal line with a frequency component at 2o. As a result of this oscillating heat source, a temperature oscillation and a corresponding resistance oscillation at frequency 2o are induced in the metal line. Hence, a component of the voltage oscillation (v ¼ ir) across the metal line contains a third harmonic, 3o. The thermal conductivity of the sample can be deduced from this 3o voltage oscillation. The cross-plane thermal conductivity of thin films can be measured by the differential 3o method [16]. A reference sample without the thin film of interest is prepared simultaneously with the sample containing the film of interest such that the metal line patterned on both samples has the same thickness and width. Both the thin film sample and the reference are measured using similar heating power and the same range of heater frequencies. The temperature drop across the thin film DTf can then be derived from the difference in the amplitude of the temperature oscillation of the sample and the reference. Assuming one-dimensional (1D) heat conduction, the cross-plane thermal conductivity of the thin film is derived from DTf. If the thin film is semiconducting, a thin dielectrics layer is required to electrically insulate the metal lines from the thin films. For semiconducting crystalline thin films, the differential 3o method is usually only suitable to measure the thermal conductivity of the films if the films are >1 mm thick [24]. Another important technique for measurements of heat conduction on nanometer length scale is the timedomain thermoreflectance (TDTR) [25]. Figure 3
T
T
2710
Thermal Conductivity and Phonon Transport
Ti:Sapphire
Nd:YVO
Wave plates
PBS
Delay stage
Isolator
EO modulator
Mechanical chopper Optical filters Photo diode
Sample
Objective
PBS
BS
Shortpass filter
Thermal Conductivity and Phonon Transport, Fig. 3 A schematic [26] of the TDTR setup. The optical filters and the shortpass filter are optional and can eliminate the artifacts due to diffusely scattered pump beam leaked into the Si photodiode during TDTR measurements of rough samples (Reprinted with permission from [26]. Copyright 2008, American Institute of Physics)
shows a schematic diagram [26] of the TDTR setup. Before TDTR measurements, the samples are usually coated with a thin layer (80 nm) of metal with high thermoreflectance (e.g., Al) by magnetron sputtering or thermal evaporation. The metal film serves as a transducer to absorb the heating pump beam and to convert the temperature excursions at the surface into changes in the intensity of the reflected probe beam. Usually, Al is used as the metal transducer due to the high thermoreflectance. However, for high temperature measurements, Al is not suitable due to its low melting point; Pt, Ta, or AuPd could be used for high temperature measurements, see [27] for further discussion. The thickness of the metal transducer layer is normally simultaneously determined during TDTR measurements by picosecond acoustics. In TDTR measurements, the output of a modelocked Ti:sapphire laser oscillator is split into a pump beam and a probe beam, with the relative delay time between the pump and probe pulses being adjusted via a mechanical stage. Samples are heated by the pump beam, which is modulated by an electro-optic modulator at frequency f, 0.1 < f < 10 MHz. Cooling of the surface after being heated by pump pulses is then monitored through changes in the intensity of the reflected probe beam using a Si photodiode and a radio-frequency (rf) lock-in amplifier. If the samples are rough, TDTR measurements should be performed
using a two-tint configuration [26], see Fig. 3. Normally, heat waves only penetrate 5 mm. Hence, heat flow in TDTR is predominantly one dimensional. Data analysis in TDTR is considerably complicated. The changes in reflected intensity at frequency f have both an in-phase Vin and out-of-phase Vout components. Normally, the ratio Vin/Vout is used for the analysis (instead of just the Vin component) to make use of the additional information in the out-of-phase signal and eliminate artifacts created by unintended variations in the diameter or position of the pump beam created by the optical delay line. Measurements of the ratios Vin/Vout as a function of delay time are compared to numerical solutions of the diffusion equation in cylindrical coordinates using a thermal model [28]. The thermal model normally has two free parameters: the thermal conductance of the Al/sample interface and L of the sample. For most cases, these two parameters can be separated from the fitting of the model calculations to the measurements [29].
Future Directions for Research Detailed understanding of heat transport in crystalline solids, especially on how phonons are scattered in nanostructures, is crucial for the design of new thermoelectric materials and thermal management of electronic devices. In this regard, future research will focus on experimental and theoretically studies on the details of how phonons carry heat across different kinds of nanostructures. One example is recent efforts by researchers in the field to investigate how heat is transported along [1, 2] and across [30] monolayer and few-layer graphenes. By carefully comparing the experimental results to theoretical calculations, these studies advance our knowledge on how phonons are scattered by impurities and nanostructures and thus facilitate future design of more efficient devices.
Cross-References ▶ Finite Element Methods for Computational Nano-optics ▶ Nanostructured Thermoelectric Materials ▶ Nanostructures for Energy ▶ Thermoelectric Heat Convertors
Thermally Actuated Nanoelectromechanical Resonators
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
15. 16. 17. 18. 19. 20.
21. 22. 23.
24.
25. 26. 27. 28. 29. 30.
Balandin, A.A., et al.: Nano Lett. 8, 902 (2008) Seol, J.H., et al.: Science 328, 213 (2010) Chiritescu, C., et al.: Science 315, 351 (2007) Swartz, E.T., Pohl, R.O.: Rev. Mod. Phys. 61, 605 (1989) Gundrum, B.C., Cahill, D.G., Averback, R.S.: Phys. Rev. B 72, 245426 (2005) Lyeo, H.-K., Cahill, D.G.: Phys. Rev. B 73, 144301 (2006) Koh, Y. K.: Ph.D thesis, University of Illinois at UrbanaChampaign (2010) Koh, Y.K., Cahill, D.G.: Phys. Rev. B 76, 075207 (2007) Ju, Y.S., Goodson, K.E.: Appl. Phys. Lett. 74, 3005 (1999) Asheghi, M., Leung, Y.K., Wong, S.S., Goodson, K.E.: Appl. Phys. Lett. 71, 1798 (1997) Cahill, D.G., Pohl, R.O.: Phys. Rev. B 35, 4067 (1987) Yao, T.: Appl. Phys. Lett. 51, 1798 (1987) Lee, S.-M., Cahill, D.G., Venkatasubramanian, R.: Appl. Phys. Lett. 70, 2957 (1997) Huxtable, S.T., Abramson, A.R., Tien, C.-L., Majumdar, A., LaBounty, C., Fan, X., Zeng, G., Bowers, J.E., Shakouri, A., Croke, E.T.: Appl. Phys. Lett. 80, 1737 (2002) Yang, B., Liu, J.L., Wang, K.L., Chen, G.: Appl. Phys. Lett. 80, 1758 (2002) Venkatasubramanian, R.: Phys. Rev. B 61, 3091 (2000) Touzelbaev, M.N., Zhou, P., Venkatasubramanian, R., Goodson, K.E.: J. Appl. Phys. 90, 763 (2001) Koh, Y.K., Cao, Y., Cahill, D.G., Jena, D.: Adv. Funct. Mater. 19, 610 (2009) Li, D., Wu, Y., Kim, P., Shi, L., Yang, P., Majumdar, A.: Appl. Phys. Lett. 83, 2934 (2003) Hochbaum, A.I., Chen, R., Delgado, R.D., Liang, W., Garnett, E.C., Najarian, M., Majumdar, A., Yang, P.: Nature 451, 163 (2008) Harman, T.C., Taylor, P.J., Walsh, M.P., LaForge, B.E.: Science 297, 2229 (2002) Koh, Y.K., Vineis, C.J., Calawa, S.D., Walsh, M.P., Cahill, D.G.: Appl. Phys. Lett. 94, 153101 (2009) Kim, W., Zide, J., Gossard, A., Klenov, D., Stemmer, S., Shakouri, A., Majumdar, A.: Phys. Rev. Lett. 96, 045901 (2006) Koh, Y.K., Singer, S.L., Kim, W., Zide, J.M.O., Lu, H., Cahill, D.G., Majumdar, A., Gossard, A.C.: J. Appl. Phys. 105, 054303 (2009) Paddock, C.A., Eesley, G.L.: J. Appl. Phys. 60, 285 (1986) Kang, K., Koh, Y.K., Chiritescu, C., Zheng, X., Cahill, D. G.: Rev. Sci. Instrum. 79, 114901 (2008) Wang, Y., Park, J.-Y., Koh, Y.K., Cahill, D.G.: J. Appl. Phys. 108, 043507 (2010). Accepted Cahill, D.G.: Rev. Sci. Instrum. 75, 5119 (2004) Costescu, R.M., Wall, M.A., Cahill, D.G.: Phys. Rev. B 67, 054302 (2003) Koh, Y.K., Bae, M.-H., Cahill, D.G., Pop, E.: Nano Lett. 10, 4363 (2010)
2711
T
Thermal Infrared Detector ▶ Insect Infrared Sensors
Thermal Management Materials ▶ Modeling Thermal Properties of Carbon Nanostructure Composites
Thermal Resistance ▶ Thermal Conductivity and Phonon Transport
Thermal Resistivity ▶ Thermal Conductivity and Phonon Transport
Thermal Transport ▶ Thermal Conductivity and Phonon Transport
Thermally Actuated MEMS Resonators ▶ Thermally Actuated Silicon Resonators
T Thermally Actuated Micromechanical Resonators ▶ Thermally Actuated Silicon Resonators
Thermal Evaporation
Thermally Actuated Nanoelectromechanical Resonators
▶ Physical Vapor Deposition
▶ Thermally Actuated Silicon Resonators
T
2712
Thermally Actuated Nanomechanical Resonators ▶ Thermally Actuated Silicon Resonators
Thermally Actuated Resonators ▶ Thermally Actuated Silicon Resonators
Thermally Actuated Silicon Resonators Siavash Pourkamali Department of Electrical and Computer Engineering, University of Denver, Denver, CO, USA
Synonyms Thermally actuated MEMS resonators; Thermally actuated micromechanical resonators; Thermally actuated nanoelectromechanical resonators; Thermally actuated nanomechanical resonators; Thermally actuated resonators; Thermal-piezoresistive resonators
Definition Thermally actuated electromechanical resonators are micro- to nanoscale structures whose specific mechanical resonant modes can be actuated by periodic thermal expansion forces resulting from joule heating in parts of their structure.
Introduction Thermal actuation based on joule-heating-induced thermal expansion of solids is a well-known mechanism that can be conveniently implemented at microand nanoscale. An electrothermal actuator simply consists of a conductive element that heats up as a result of electrical current flowing through it and therefore can be implemented without any fabrication challenges or
Thermally Actuated Nanomechanical Resonators
the need for material integration. Electrothermal actuators also have great properties such as large actuation force, low operating voltage, and simplicity of design and integration. On the downside, their considerable power consumption makes them an undesirable choice for many applications. Furthermore, thermal actuators are generally considered slow actuators only suitable for DC or very-low-frequency applications. This is mainly based on the general perception emanating from the more familiar marco-world where thermal phenomena have relatively large time constants in the few seconds or even minutes and hours range. However, as will be shown in this section, thermal actuation along with piezoresistive sensing provides an interesting option for transduction of high-frequency resonant modes of micro- and nanostructures. Micro- and nanoscale electromechanical resonant devices have received a lot of attention over the past decade as a potential emerging technology for next generation integrated frequency references, electronic filters, and resonant sensors. As a result, a wide variety of high-frequency microelectromechanical resonator technologies have been developed over the past few years, most of which use piezoelectric [1, 2] or electrostatic (capacitive) [3–5] transduction. Despite significant improvements in both piezoelectric and capacitive microresonator technologies, each approach still has certain challenges to overcome. Piezoelectric microresonators require integration of piezoelectric and metallic thin films generally resulting in lower quality factors and frequency and quality control issues. In case of air-gap capacitive resonators, the weak electromechanical coupling necessitates deep submicron transduction gaps leading to fabrication and reliability challenges, power handling limitations, and excessive squeezed film damping when operating in air. Solid dielectric resonators [5] use electrostatic forces in dielectric thin films (preferably a high-K dielectric) to take advantage of the larger dielectric coefficients and improve the electromechanical coupling strength. The fully solid structure of such device relaxes several problems associated with aggressive reduction of the air gaps such as stiction and pull-in issues and allows reduction of the dielectric gap sizes to a few nanometers. However, similar to the piezoelectric resonators, such devices require integration and physical contact of dielectric thin films and conductive electrodes with the resonant structures which
Thermally Actuated Silicon Resonators
defeats some of the main purposes of moving away from piezoelectric devices. Moreover, large dielectric constants of the very thin dielectric films result in large parasitic capacitors to be associated with such devices. Frequency tuning and compensation of the temperature-induced frequency drift are among other formidable challenges for both piezoelectric and capacitive resonators. Finally, in order to achieve a strong enough electromechanical transduction and consequently a large signal to noise ratio, both capacitive and piezoelectric resonators require relatively large electrode areas (thousands of mm2) consuming costly real estate on the chip and limiting the maximum number of devices that can be integrated on a single chip. Electrothermal actuation presents a third option for actuation of resonant modes of micro- and nanoscale structures that has been neglected to some extent so far, especially for high-frequency applications. Different versions of lower frequency thermally actuated micromechanical resonant devices have been utilized as mass sensors for chemical [6] or particulate [7] sensing. Utilization of thermal actuation for highfrequency resonant modes has been gaining more attention lately. A 10-MHz MEMS oscillator was demonstrated in [8] via electrothermal actuation of resonant modes in a suspended polysilicon membrane. In [9], thermal actuation and piezoresistive detection of flexural resonant modes of a silicon-carbide nanobeam with frequencies up to 1.1 GHz has been demonstrated. Recent studies and experiments have shown promising results and plenty of unexplored potentials for thermally actuated high-frequency resonators especially as the structural dimensions are scaled down into the nanoscale [10]. Both modeling and experimental measurements suggest that as opposed to the general presumption, thermal actuation becomes a more efficient actuation approach for higher-frequency resonant modes provided that higher frequency is achieved by shrinking the structural dimensions down. Unlike electrostatic and piezoelectric resonators, thermalpiezoresistive resonators perform better as their size is shrunk down and therefore could be a much stronger candidate for realization of highly integrated nanomechanical arrays of resonant signal processors. Self-sustained oscillation capability [11, 12] and robustness and design flexibility for sensory applications [7] are among other unique properties that thermal-piezoresistive transduction offers.
2713
T
High-Frequency Thermal-Piezoresistive Resonators: Principles of Operation and Modeling High-Frequency Thermal Actuation: As mentioned earlier, based on the perception from the macroworld, it is hard to imagine a thermal system with small enough thermal time constants to respond to high-frequency excitations. Similar to electrical RC circuits, thermal time constant of a thermal actuator, for example, a suspended conducting beam, is the product of its thermal capacitance and thermal resistance: tth ¼ RthCth. Thermal resistance can be calculated using a very similar equation to electrical resistance, that is, Rth ¼ sth–1L/A, where sth is the thermal conductivity of the structural material and L and A are the length and cross-sectional area of the element. Similar to electrical resistance, thermal resistance increases proportionally as all the dimensions of the element are scaled down. This implies that if a thermal actuator is scaled by a factor of S, its thermal resistance changes by a factor of S–1. On the other hand, thermal capacitance is proportional to the mass and consequently the volume of the element and scales by a factor of S3. Hence, overall, the thermal time constant of the element scales by a factor of S2. On the other hand, if a mechanical structure is scaled by a factor S, its mechanical resonant frequency (om) changes by a factor of S-1, that is, its mechanical time constant changes by a factor of S. The overall conclusion is that if the dimensions of a thermally actuated resonant structure are scaled down, its thermal time constant shrinks faster than its mechanical time constant. To be more specific, both thermal and mechanical responses of the system become faster upon scaling the dimensions down; however, the increase in the speed of the thermal response is sharper (proportional to S–2) than the increase in the mechanical resonant frequency (proportional to S–1). Therefore, as the structural dimensions are scaled down, the performance of thermally actuated mechanical resonators is expected to improve (more actuation force for the same actuation power) as the thermal response of the system can catch up with the mechanical vibrations more effectively. Thermal-Piezoresistive Transduction: Piezoresistive sensing is the most convenient sensing mechanism to be integrated and utilized along with thermal actuation.
T
T
2714
Similar to thermal actuation, piezoresistive sensing requires only a conductive element preferably with a high piezoresistive coefficient. To add to the simplicity, the same elements can be used for both thermal actuation and piezoresistive sensing simultaneously. Thermal-piezoresistive resonators can come in different shapes and sizes. Figure 1 shows the 3D schematic view of a resonant structure, known as I-shaped bulk acoustic wave resonator or IBAR (also known as dog-bone resonator), which is very suitable for thermal-piezoresistive transduction while being capable of providing relatively high resonance frequencies. In the in-plane extensional resonance mode, the two blocks hanging on the two sides of the structure move back and forth in opposite directions subjecting the two beams connecting them together to periodic tensile and compressive stress. Such resonance mode can simply be actuated by applying the actuation voltage (or current) between the two support pads on the two sides of the structure. The ohmic loss caused by the current flow is maximized in the narrower parts of its path, that is, the extensional pillars in the middle of the structure (the two beams connecting the two blocks to each other). Upon application of a fluctuating actuation voltage, the fluctuating ohmic power generation results in a fluctuating temperature gradient and therefore periodic thermal expansion of the extensional actuator beams. The alternating extensional force resulting from the fluctuating temperature in the beams can actuate the resonator in its in-plane extensional resonance mode. If the ohmic power and consequently the resulting temperature fluctuations have the same frequency as the mechanical resonance frequency of the structure, the mechanical vibration amplitude is amplified by the mechanical quality factor (Q) of the resonator. The amplified alternating stress in the pillars leads to considerable fluctuations in their electrical resistance due to the piezoresistive effect. When biased with a DC voltage, such resistance fluctuations modulate the current passing through the structure resulting in an AC current component known as the motional current. Thermo-electromechanical modeling: Operation of a thermal-piezoresistive resonator involves phenomena in three different domains of thermal, mechanical, and electrical nature. Ohmic loss turns the electrical input voltage into a temperature, thermal expansion turns the temperature into a mechanical force that causes a mechanical displacement, and finally the
Thermally Actuated Silicon Resonators
Thermally Actuated Silicon Resonators, Fig. 1 3D schematic view of a thermally actuated IBAR (dog-bone resonator) showing the qualitative distribution of temperature fluctuation amplitude when subject to a fluctuating current flow (red shows the maximum and blue shows the minimum)
piezoresistive effect turns the mechanical displacement back into an electrical signal (motional current). Figure 2 shows the equivalent electrical circuit models for the three domains. In the thermal domain, the actuator beams act like a low-pass RC combination. The equivalent circuit consists of a current source representing the alternating heating power along with a capacitance (Cth) and a resistance (Rth) representing the effective thermal capacitance and thermal resistance of the thermal actuators, respectively (Fig. 2a). The resulting voltage across the parallel RC combination (Tac) represents the temperature fluctuation amplitude. Typically, the fluctuating actuation voltage is provided by a combination of a DC and an AC voltage component. Due to the square relationship between ohmic power generation and electrical voltage (or current), application of only an AC actuation voltage with frequency of fa results in an ohmic power component with frequency of 2fa. In order to have the same frequency as the input AC actuation voltage for the thermal actuation force, a combination of AC and DC voltages (Vdc and vac) needs to be applied between the two terminals of the resonators. The resulting power loss component at the same frequency as vac in this case will be Pac ¼ 2VRdcAvac , where RA is the electrical resistance of the actuator elements (extensional beams
Thermally Actuated Silicon Resonators Thermally Actuated Silicon Resonators, Fig. 2 Equivalent electrical circuit for a thermalpiezoresistive resonator: (a) the thermal domain; (b) the mechanical domain; (c) the electrical domain
T
2715
a
b
Tac
Pac
Cth
b + –
F=
Rth
i
Xth =Q =
i S
1 K
2aAETac M Current = Thermal Power Voltage = Temperature
c
Voltage = Force Current = Velocity Charge = Displacement nac
+
–
RA
iac
Rs im =
Xth plEIdc L
Resonator Terminals
in IBARs) and Vdc and vac are the applied DC and AC actuation voltages, respectively. The RLC combination of Fig. 2b represents the mechanical resonance behavior of the structure. The voltage source represents the thermally generated mechanical force which is the product of thermal expansion coefficient (a) of the structural material, thermal actuator cross-sectional area (A), and Young’s modulus (E) of the thermal actuator along its length. The coefficient “2” has been added since there are two actuator beams in each resonator contributing to the actuation force. The inductor, capacitor, and resistor represent the effective mechanical mass (M), stiffness (K), and damping (b) of the resonator, respectively. Consequently, current represents velocity and electriR cal charge (Q ¼ i.dt) represents displacement (Xth), which is the elongation of the thermal actuators (resonator vibration amplitude). The vibration amplitude, which is the output of the mechanical subsystem, acts as an input to the electrical subsystem (Fig. 2c) causing a change in the resistance of the thermal actuators due to the piezoresistive effect. This change in resistance modulates the current flow through the resonator which is shown as a current source in the equivalent circuit. Rs in Fig. 2c represents the parasitic electrical resistances in series with the actuator resistance which mainly includes resistance of resonator support beams. The overall transfer function that relates the AC motional current (output current) to the AC input voltage is the product of the three transfer functions from
the thermal, mechanical, and electrical domains. The simplified transfer function at resonance frequency is HT s¼jo0 ¼ gm ¼ 4aE2 pl Q
2 AIdc KLCth o0
(1)
where pl is the longitudinal piezoresistive coefficient of the structural material, L is the thermal actuator length, and Q and o0 are the quality factor and angular resonant frequency of the resonator. For IBAR structures, knowing that K ¼ 2EA/L, Eq. 1 can be further simplified to
gm ¼ 2aEpl Q
2 Idc : Cth o0
(2)
A more detailed version of the modeling and derivations above is available in [10]. The transfer function in Eqs. 1 and 2 can be referred to as small signal voltage to current gain, or motional conductance (gm) of the resonator at its resonance frequency, which is one of the most important parameters of a thermalpiezoresistive resonator when utilized as an electronic circuit component. Resonator equivalent electrical circuit: For the resonators discussed here, where the extensional beams acting as thermal actuators are also utilized for piezoresistive readout, the physical resistance of the resonator connects the input and output of the devices.
T
T
2716
Thermally Actuated Silicon Resonators RA RS
Rm = 1/gm Cm =
QRm Lm = w0 1 QRmw 0
Thermally Actuated Silicon Resonators, Fig. 3 Compact overall equivalent electrical circuit for one-port thermally actuated resonators with piezoresistive readout
The equivalent electrical circuit in this case includes a resistance RA + Rs connected between the two terminals of the resonator, which is the overall resistance in the current path between the two pads of the resonator. In parallel to the static resistance of the actuators, there is a series RLC combination that represents the mechanical resonant behavior of the structure. The value of Rm in the RLC has to be set so that at resonance, a motional current of im ¼ gm.vac ¼ vac/Rm is added to the feedthrough current passing through RA. Therefore, Rm ¼ gm–1 and Lm and Cm values can be calculated based on the value of Rm as shown in Fig. 3 according to the resonance frequency and quality factor of the resonator. To calculate the motional conductance (gm) of a thermal-piezoresistive IBAR, which is the most important parameter to be extracted from the model using Eq. 2, all the parameters except the effective thermal capacitance of the actuators (Cth) are known for every set of resonator dimensions and structural material. Due to the distributed nature of the thermal parameters including thermal generator (electrical resistance of the structure), thermal capacitance, and thermal conductance, analytical derivations or finite element analysis should be used to find an accurate value for Cth. In [10, 13], finite element analysis has been used to simulate the transient thermal response of the actuators and calculate the fluctuating temperature amplitude (Tac) at different points along the length of the actuator beams. The mean value of the small signal temperature amplitudes at different points along the actuators is the effective temperature fluctuation amplitude for the actuator from which the effective thermal capacitance value can be extracted. As shown in [13], the effective thermal capacitance for a thermal actuator beam is equal to 0.92 of the static lumped element thermal capacitance of the beam with a good approximation.
Resonator Scaling Behavior and Optimization As an electronic component, it is desirable to maximize the resonator motional conductance (gm) (for improved signal to noise ratio) while minimizing power consumption. Therefore, a figure of merit for the resonators can be defined as the ratio of the resonator gm to the overall DC power consumption: F:M: ¼
gm 2aEpl Q ¼ PDC Cth o0 ðRA þ Rs Þ
(3)
where PDC ¼ IDC2.(RA + RS) is the static power consumption of the resonator. Different parameters that can be used to maximize F.M. are as follows: Actuator Beam Dimensions: Generally, smaller actuator dimensions lead to smaller Cth improving resonator figure of merit. However, the effect of actuator dimensions on its electrical resistance (RA) should also be taken into account. Since the extensional stiffness of the actuator beams defines the resonance frequency of the IBAR, in order to maintain the same resonance frequency for an IBAR while reducing its actuator thermal capacitance, both length and width of the actuator should be scaled down simultaneously. Such scaling does not affect the actuator electrical resistance (RA). Therefore, scaling down both the length and width of the actuator beams by a scale Sa results in an improvement in the resonator F.M. by a factor of Sa2 while maintaining an almost constant resonance frequency for the device. Resonance Frequency and Resonator Scaling: At a first glance at Eq. 3, higher resonance frequencies seem to have a deteriorating effect on resonator figure of merit. However, if higher resonant frequencies are achieved by shrinking the resonator size, at the same time Cth will be shrinking sharply. If all resonator dimensions are scaled down proportionally by a factor S, Cth / S3, o0 / S, and (RA + RS) / S. Therefore, F.M. / S that is, F.M. increases proportionally if the resonator dimensions are scaled down, while its resonance frequency increases at the same time. This confirms the validity of the explanation provided previously claiming suitability of thermal actuation for high-frequency applications. It can be concluded from the last two discussions that the capability to fabricate resonator actuator beams with very small (nanoscale) dimensions is the key to achieving very high
Thermally Actuated Silicon Resonators Thermally Actuated Silicon Resonators, Fig. 4 SEM view of single crystalline silicon thermal-piezoresistive IBARs (a) a 61-MHz 15-mmthick IBAR fabricated on a low-resistivity p-type SOI substrate; (b) a 41-MHz 3-mmthick IBAR fabricated on a low-resistivity n-type SOI substrate followed by thermal oxidation to thin down its thermal actuator beams
2717
a
T
b
10μm
10μm
a −13 Tunning range = %1.39
b
−23
−38 60.8
Tunning range = %0.91
–19 –23 –27
60.98
61.16 61.34 61.52 Frequency (MHz)
61.7
Q = 12000 Current = 100mA
–31 60.83 60.85 60.87 Frequency (MHz)
gm (dB)
−28 −33
–24
–15 gm (dB)
gm (dB)
−18
–28 –32 –36
Q = 14000 Current = 60mA
–40 61.62 61.64 61.66 Frequency (MHz)
Thermally Actuated Silicon Resonators, Fig. 5 Measured frequency response of the thermally actuated 61-MHz resonator of Fig. 4a with different bias currents. The vertical axis in the frequency response plots shows the motional conductance of the
resonators (gm) in dB with respect to 50-O network analyzer terminations. Red and blue plots refer to vacuum and air testing conditions, respectively. Current range is 45–100 mA in vacuum and 55–100 mA in air
performance levels for thermal-piezoresistive resonators, both at low and high frequencies. Electrical Resistivity and Other Material Properties: One of the most important parameters that can also be controlled for some of the most popular resonator structural materials (e.g., Si or SiC) is the electrical resistivity of the structural material. Lower electrical resistivity of the structural material improves the transduction figure of merit by lowering RA + Rs. This can be explained by the fact that, with lower electrical resistivity, the same amount of DC bias current and therefore the same motional conductance can be maintained while burning less ohmic power in the structure. A number of other structural material properties also affect the figure of merit for thermalpiezoresistive resonators. Higher thermal expansion coefficient and lower specific heat capacity improve the thermal actuation resulting in proportionally higher F.M. Higher piezoresistive coefficient improves the
piezoresistive readout having the same effect on F.M. One interesting and promising fact is the giant piezoresistive effect found in single crystalline silicon nanowires that has been observed and reported by different researchers [14]. Therefore, as the resonators are scaled down to reach higher resonant frequencies in the hundreds of MHz and GHz range, it is expected that the actuator beams with deep submicron width exhibit such strong piezoresistivity significantly improving resonator performances. Finally, using structural materials with higher Young’s modulus (E) can also improve the resonator figure of merit while increasing its resonant frequency. Figure 4 shows the scanning electron micrograph (SEM) of two different microscale single crystalline silicon IBAR structures. Figure 5 shows the motional conductance frequency plots extracted from the measured frequency response of the 61-MHz resonator in Fig. 4a. The resonator quality factor ranges from 12,000 to 14,000 under vacuum and drops to
T
T
2718
Thermally Actuated Silicon Resonators
b –45
c –30
–55
–40
dB
dB
–86
–88
dB
a –84
–65 –75
30.519
30.523
Frequency (MHz)
–60
Q = 24,400 (Vac.) gm = 43.6μA / V PDC = 1.01mW IDC = 720μA
30.52
Q = 9200 (Air) gm = 233μA / V PDC = 14.7mW IDC = 2.69mA
30.53
Frequency (MHz)
Thermally Actuated Silicon Resonators, Fig. 6 Measured motional conductance frequency plots for a 3-mm-thick 30.5MHz IBAR. The vertical axis in the plots shows the motional conductance of the resonators (gm) in dB with respect to 50-O network analyzer terminations. Due to its very narrow (750 nm
6,000–8,000 under atmospheric pressure. A largefrequency tuning range is demonstrated for this resonator by changing its DC bias current. This is mainly due to the raising temperature of the resonating body, especially its extensional beams that provide most of the structural stiffness in the in-plane extensional mode. As expected, as the DC bias current and consequently resonator motional conductance at resonance (gm) increases, due to higher static temperature and softening of the structural material, the resonance frequency decreases. Despite having high-quality factors and relaively large motional conductance, the large current requirement and power consumption for the resonator in Fig. 5 could be a major drawback against using such as electronic components. Figure 6 shows a few frequency response plots measured for a 30.5-MHz resonator with narrowed down actuator beams (750 nm wide) similar to the structure in Fig. 4b. Submicron actuator beams can be achieved by controlled oxidation of the silicon structures followed by oxide removal leaving behind a narrower beam. Because of its much thinner actuators, a resonance peak can be detected for this resonator with currents as low as 43 mA translating into DC power consumptions as low as a 3.6 mW. At the DC bias current of 720 mA (DC power of 1 mW), motional conductance of 44 mA/V was measured for this resonator (Fig. 6b). Maximum currents tolerable for such narrow actuator beams are in the few mA range leading to motional conductances as high as 0.23 mA/V in air with Q of 9,200 (Fig. 6c). According to the derived model, much higher motional conductances can be
30.425
30.445
Frequency (MHz)
wide) actuator beams, clear resonance peaks with motional conductance in the tens of mA/V have been measured for this resonator with sub-mA DC bias currents (sub-mW power consumption)
30 25 K (10−7/V2)
Q = 35,900 (Vac.) gm = 0.207μA / V PDC = 3.63μW IDC = 43μA
–50
20
15.9MHz (Air) 15.9MHz (Vac.) 31.5MHz (Air) 31.5M Hz (Vac.)
15 10 5 0 30
40
50 Current (mA)
60
70
Thermally Actuated Silicon Resonators, Fig. 7 Extracted transduction figure of merit (K) for two IBARs versus resonator bias current. The higher-frequency (31.5 MHz) resonator, which is a 2X-scaled-down version of the 15.9-MHz resonator, has a higher transduction strength. The increase in the transduction figure of merit with bias current, which is not directly predicted by the derivations, is expected to be a result of the effect of increased static temperature at higher currents on different material properties, for example, increased thermal expansion coefficient at higher temperatures
achieved by consuming the same amount of power using lower-resistivity structural material allowing higher bias currents. To verify better suitability of thermal actuation for higher-frequency resonators with smaller dimensions, Fig. 7 shows the measured transduction figure of merit for two thermal-piezoresistive resonators with the effect of quality factor taken out (K ¼ F:M: Q ) under different bias currents. The two resonators are fabricated on the same substrate and have similar horizontal
Thermally Actuated Silicon Resonators
dimension aspect ratios with the 32-MHz resonator being a 2X-scaled-down version of the 16-MHz resonator. As expected based on the previously discussed scaling trend and the derived model, the higherfrequency resonator has a higher transduction figure of merit. This further confirms that thermal actuation is not only a potentially suitable actuation mechanism for high-frequency applications, but also it can provide higher performances at higher rather than lower frequencies. Therefore, much higher transduction efficiencies are expected to be achievable for higherfrequency versions of such devices with resonant frequencies in the hundreds of megahertz and even GHz range, provided that the dimensions can be scaled down proportionally into the nanoscale. The derived resonator model backed up by measurement results shows that with sub-100-nm actuator width and submicron actuator length, gm as high as 1mA/V can be achieved for thermal-piezoresistive silicon IBARs with resonant frequencies in the gigahertz range with power consumptions as low as a few microwatts [10]. This is not including the giant piezoresistive effect in crystalline silicon that could further improve this by over an order of magnitude [14].
Self-Sustained Thermal-Piezoresistive Oscillators Generally, in order to implement an electronic oscillator with a mechanical resonator as its frequency reference, the resonator needs to be engaged in a positive feedback loop consisting of amplifying circuitry with appropriate phase shift. One of the interesting aspects of the thermal-piezoresistive mechanical resonators that use the same elements as both thermal actuators and piezoresistive sensors is the internal feedback mechanism formed between thermal actuation force and piezoresistivity. This feedback results from the fact that the piezoresistive motional current of the structure that passes through the thermal actuators also acts as an excitation for the resonator by affecting the ohmic loss in the actuator beams. Depending on the sign of the structural material piezoresistive coefficient, electrical terminations of the resonator, and the phase shift resulting from thermal delays, this could act as a positive feedback stimulating mechanical vibrations or a negative feedback limiting vibration amplitude of the structure or reducing Brownian motion of
2719
T
the structure at that specific frequency (refrigeration [11]). For example, ohmic power generation in a thermal actuator biased with a constant current (terminated with a large electrical resistance) decreases if the actuator is subjected to tensile stress, due to the decreased electrical resistance. As part of a resonant structure, when such actuator is fully stretched and is about to start to compress, its decreased ohmic power and resulting cooling assist the actuator going into compression. In the same manner, after compression of the actuator, increased ohmic power generation helps its expansion via raising its temperature. Mechanical and thermal time delays play a key role here in determining whether the feedback has an amplifying or attenuating effect. It was demonstrated by researchers at NXP semiconductors [11] that such positive feedback can be strong enough to initiate and maintain mechanical oscillations upon providing a DC electrical power source to the structure without the need for any external AC stimulation. Figure 8 shows the scanning electron micrograph of the suspended N-type single crystalline silicon structure for which such effect was demonstrated for the first time [11]. The structure is made of N-type silicon because N-type silicon has a negative longitudinal piezoresistive coefficient, which is required for such oscillation under constant current conditions. Figure 9 schematically shows the feedback loop and the transduction mechanisms connecting the three physical parameters: displacement (x), temperature (Tac), and voltage (vac) leading to self-sustained oscillation [11]. As shown in Fig. 9b, physically this can be explained as follows: upon application of a DC bias, the thermal actuator heats up and expand, pushing the mass to the right side. Due to the mass (inertia) of the plate, the nanobeam actuator experiences an overexpansion after pushing the mass and undergoes tensile stress. The negative piezoresistive coefficient turns the tensile stress into reduced electrical resistance for the nanobeam. When biased with a constant DC bias current, this means reduced ohmic power in the actuator forcing it to contract. After structural overcontraction (due to the mass of the plate), the actuator will be under compressive stress and will have increased electrical resistance. This translates into a higher thermal power forcing the structure to expand again. If the resulting driving force (due to heating and cooling) in each cycle is large enough
T
T
2720
Thermally Actuated Silicon Resonators, Fig. 8 Schematic view and SEM top view of a fully micromechanical crystalline silicon oscillator. To operate the oscillator, a DC current is applied between the terminals of the device, and the output AC voltage resulting from the resonator’s spontaneous mechanical vibrations is measured. The inset shows a zoomed-in view of the nanobeam actuator and the wide beam providing the flexural stiffness of the structure. Output electrical signal amplitude of 27 mV with frequency of 1.26 MHz was measured for this resonator when biased with a current of 1.2 mA (1.2-mW power consumption) under vacuum
Thermally Actuated Silicon Resonators
a Mechanical motion Mass
Nanobeam Rdc + rac
z y
Idc x Spring beam
Vdc + Vac Output signal (T1)
b
Idc
Vac
Ground (T2)
Nanobeam
[010] [100] T1
10 μm Spring beam T2
2 μm
T4
Idc
T2
to compensate for mechanical losses of the structure, the same sequence is repeated over and over in a selfsustained manner and the vibration amplitude keeps increasing until it is limited by nonlinearities. The 90
phase lags (Fig. 9b) resulting from the mechanical and thermal response delays lead to a perfect timing for the abovementioned sequence of physical phenomena to continue. Figure 10 shows the SEM view and time response of an extensional mode 6.7-MHz self-sustained thermalpiezoresistive oscillator operating based on the same principle. Such structure is a derivative of the IBAR structures discussed in the previous sections with relatively larger plates. It operates in an in-plane resonant mode with its two plates moving back and forth in opposite directions and the extensional actuator beams being stretched and compressed periodically. Extensional resonant modes can provide higher frequencies as well as higher-quality factors both under
T3
vacuum and in air. Dual plate oscillators with frequencies in the few megahertz range, power consumption of a few to tens of milliwatts, and operating under both vacuum and atmospheric pressure have been demonstrated [12]. The output voltage amplitudes range from tens to hundreds of millivolts. It can be shown that the requirement for selfsustained oscillation in such devices is that the absolute value of the motional conductance of the resonator becomes larger than the physical static conductance of its actuators. The minimum DC power required to achieve this condition is PdcMin ¼
KLCth om : 4aE2 jpl jQA
(4)
Again, the key to having low power consumption is scaling down the resonator dimensions that reduce the power requirement with a square relationship leading
Thermally Actuated Silicon Resonators Thermally Actuated Silicon Resonators, Fig. 9 (a) Schematic demonstration of the feedback loop and the transduction mechanisms linking the thermal, mechanical, and electrical parameters to each other. Thermal delay and mechanical resonance provide the required phase shift for self-sustained oscillation. (b) Sequence of phenomena causing an internal positive feedback loop in thermal-piezoresistive resonators biased with a constant current that lead to self-sustained oscillation
2721
a
back
d Fee
loop
Thermal domain Temperature
T ex herm pa ns al ion
Heat loss
Mechanical loss
Thermal delay e tiv sis g Re atin he
Mechanical resonance Displacement
Me
ch
an
ica
ld
om
Actuator Applies Expanding Force
90° Lag
Thermal Delay
Mechanical Delay
Idc vac ain om
d
Actuator Expands and Undergoes Tensile Stress
Heating / Expansion Half Cycle
Actuator Resistance and Thermal Power Increases
Actuator Contracts and Undergoes Compressive Stress
Cooling / Contraction Half Cycle Mechanical Delay 90° Lag
Actuator Temperature Increases
al
tric
c Ele
90° Lag
b
Resistance
Piezoresistive effect
ain
Thermally Actuated Silicon Resonators, Fig. 10 SEM view and output signal of a 6.6-MHz single crystalline silicon thermal-piezoresistive dual-plate resonator capable of self-sustained oscillation. Only a DC bias current of 3.0 mA is being applied to the resonator, and the resonator is operating in air. The power consumption of the oscillator is about 20 mW, and its output voltage amplitude is 70 mV
T
Actuator Resistance and Thermal Power Decreases Thermal Delay
90° Lag
Actuator Temperature Decreases
Actuator Applies Contracting Force
T
T
2722
to significantly lower power consumption for scaleddown versions of such devices while having higher oscillation frequencies. Another area where such devices need significant improvement is the output voltage amplitude, which is a small fraction of the DC voltage across the resonator (typically 0.01–0.1). What limits the oscillation amplitude of the resonators is nonlinearities (most likely mechanical nonlinearities) limiting the feedback loop gain to unity after reaching a certain amplitude. Therefore, a thorough study on minimization of the mechanical nonlinearities can possibly lead to fully micromechanical oscillators with higher output amplitudes. Piezoresistive coefficient of the structural material is another parameter that has a direct effect on the output voltage amplitude of such oscillators. With the same mechanical vibration amplitude and bias current, higher piezoresistive coefficient leads to proportionally higher output voltage amplitude. This further shows the significance of fabricating such devices with sub-100-nm actuator width taking advantage of the huge piezoresistivity observed in silicon nanowires [14] to increase the output amplitude. Elimination of the sustaining amplifier by the selfsustained fully micromechanical oscillators described here might not be a significant improvement as electronics can be cheap and small and readily necessary to support other functions within the systems employing the electromechanical oscillators. However, for applications where large oscillator arrays are required, for example, sensors arrays, elimination of the amplifiers and the associated interconnections could be highly beneficial.
Thermal-Piezoresistive Resonant Mass Sensors Electromechanical resonators have great potential as highly sensitive mass sensors for gas, biomolecular, or particulate sensing applications. For such sensory applications where the resonator has to be in contact with the surrounding environment, resonator robustness becomes a critical factor. This makes thermal-piezoresistive resonators made of a single monolithic structural material a great candidate for mass sensing applications. Especially for particulate sensing for air quality monitoring or environmental research,
Thermally Actuated Silicon Resonators
thermal-piezoresistive resonators seem to be the most suitable choice [7, 15]. Air-gap capacitive resonators are extremely vulnerable to contaminants or particulates entering and getting stuck in their nanogaps making it practically impossible to use them as particulate sensors. For piezoelectric resonators, it is very hard (if not impossible) to have a uniform mass sensitivity all over the resonator sensing surface or at least a large portion of it. This is due to the fact that such devices generally operate in their bulk resonance modes resulting in different vibration amplitudes and therefore different effective resonator mass at different locations on the surface. Uniform mass sensitivity is a necessity when targeting mass measurement and counting of individual particles. Furthermore, metallic and piezoelectric thin films in a piezoelectric resonant structure are typically less robust than a monolithic crystalline structure, especially when subjected to impact by high-speed airborne particles. Thermal-piezoresistive resonators can continue to work flawlessly with tens to thousands of particles deposited on them [7]. For the IBAR and dual-plate structures, the vibrating counter masses (plates) act as rigid bodies having almost the same vibration amplitude on their whole surface. This allows measuring the mass of individual deposited particles as they arrive providing the overall count and mass distribution of particles present in air samples. There is still the possibility of particles landing on the actuator beams and experiencing different mass sensitivities. However, the surface area of the actuator beams is typically much smaller than that of the plates making that probability small enough for the resulting error to be within the acceptable range for most applications. Furthermore, the miniaturization capability of thermalpiezoresistive resonators into the nanoscale allows much higher mass sensitivities to be achievable enabling detection and mass measurement of individual nanoparticles not feasible using the larger capacitive or piezoelectric resonant devices. Figure 11 shows the SEM view of an IBAR exposed to a flow of artificially generated airborne particles with the same size and composition along with its measured resonance frequency during the deposition [15]. The effect of arrival of each particle on the resonance frequency is clearly detectable and the same for all the particles. This confirms uniform mass sensitivity on the resonator sensing plates.
Thermally Actuated Silicon Resonators
2723
T
20.587 1 particle detected
2μm
Frequency (MHz)
20.586 20.585
2 particle detected
20.584
1 particle detected
20.583
1 particle detected
20.582 20.581 2 particles detected
20.58
7 particles detected overall
20.579 0 10μm
Thermally Actuated Silicon Resonators, Fig. 11 SEM view of a 20.5-MHz IBAR with seven artificially generated 1-mm diameter airborne particles deposited on it. The changes in the measured resonance frequency versus time confirm the arrival of
References 1. Piazza, G.: Integrated aluminum nitride piezoelectric microelectromechanical system for radio front ends. J. Vac. Sci. Tech. A 27(4), 776–784 (2009) 2. Rinaldi, M., Zuniga, C., Chengjie, Z., Piazza, G.: Superhigh-frequency two-port AlN contour-mode resonators for RF applications. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 57(1), 38–45 (2010) 3. Nguyen, C.T.C.: MEMS technology for timing and frequency control. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54(2), 251–270 (2007) 4. Pourkamali, S., Ho, G.K., Ayazi, F.: Low-impedance VHF and UHF capacitive silicon bulk acoustic wave resonators. IEEE Trans. Electron Dev. 54(8), 2017–2023 (2007) 5. Weinstein, D., Bhave, S.A.: Internal dielectric transduction in bulk mode resonators. J. Microelectromech. Syst. 18(6), 1401–1408 (2009) 6. Seo, J.H., Brand, O.: High Q-factor in-plane-mode resonant microsensor platform for gaseous/liquid environment. J. Microelectromech. Syst. 17(2), 483–493 (2008) 7. Hajjam, A., Wilson, J.C., Rahafrooz, A., Pourkamali, S.: Fabrication and characterization of thermally actuated
10
15 20 Time (min)
25
30
seven particles. There are five frequency reduction steps, of which two have a slope twice that of the rest which show that two particles have been deposited on the device during those periods
Cross-References ▶ Laterally Vibrating Piezoelectric Resonators ▶ NEMS Resonant Chemical Sensors ▶ Optomechanical Resonators ▶ Piezoresistivity ▶ Thermal Actuators
5
8.
9.
10.
11.
12.
13.
14. 15.
micromechanical resonators for airborne particle mass sensing, part II: device fabrication and characterization. J. Micromech. Microeng. (JMM), 20, 125019 (2010) Reichenbach, R.B., Zalalutdinov, M., Parpia, J.M., Craighead, H.G.: RF MEMS oscillator with integrated resistive transduction. IEEE Electron Dev. Lett. 27, 805–807 (2006) Bargatin, I., Kozinsky, I., Roukes, M.L.: Efficient electrothermal actuation of multiple modes of high-frequency nanoelectromechanical resonators. Appl. Phys. Lett. 90, 093116 (2007) Rahafrooz, A., Pourkamali, S.: High frequency thermally actuated electromechanical resonators with piezoresistive readout. IEEE Transactions on Electron Devices. 58(4), 1205–1214 (2011) Steeneken, P.G., Phan, K.L., Goossens, M.J., Koops, G.E.J., Verheijden, G.J.A.M., van Beek, J.T.M.: Piezoresistive heat engine and refrigerator. Nature Phys. 7, 354 (2012). doi:10.1038/nphys1871 Rahafrooz, A., Pourkamali, S.: Fully micromechanical piezo-thermal oscillators. In: Proceedings of the IEEE international Electron Device Meeting (IEDM), San Francisco, doi: 10.1109/IEDM.2010.5703314 (2010) Rahafrooz, A., Pourkamali, S.: Fabrication and characterization of thermally actuated micromechanical resonators for airborne particle mass sensing, part I: resonator design and modeling. J. Micromech. Microengin. (JMM) 20, 125018 (2010) He, R., Yang, P.: Giant piezoresistance effect in silicon nanowires. Nat. Nanotechnol. 1(1), 42–46 (2006) Hajjam, A., Wilson, J. C., & Pourkamali, S: Individual air-borne particle mass measurement using high frequency micromechanical resonators. IEEE Sensors Journal, 11(11), 2883–2890 (2011)
T
T
2724
Thermal-Piezoresistive Resonators ▶ Thermally Actuated Silicon Resonators
Thermodynamics of Small Systems ▶ Nanocalorimetry
Thermoelectric Cooler ▶ Thermoelectric Heat Convertors
Thermoelectric Device ▶ Thermoelectric Heat Convertors
Thermoelectric Generator ▶ Thermoelectric Heat Convertors
Thermoelectric Heat Convertors Joseph P. Heremans and Audrey M. Chamoire Department of Mechanical and Aerospace Engineering, The Ohio State University, E443 Scott Laboratory, Columbus, OH, USA
Synonyms Electric cooler; Peltier coefficient; Peltier cooler; Peltier couple; Peltier effect; Peltier element; Peltier module; Peltier-Seebeck effect; Seebeck coefficient; Seebeck effect; Thermoelectric cooler; Thermoelectric device; Thermoelectric generator; Thermoelectric heat pump; Thermoelectric module; Thermoelectric power; Thermopower; Thomson effect
Thermal-Piezoresistive Resonators
Definition Reversible solid-state heat engines that use heat applied between a hot and cold reservoir to generate electrical power and can be inverted to use an electrical current to pump heat out of the cold reservoir and into the hot reservoir. In both cases, they function like conventional heat engines, but use electrons as the working fluid instead of steam or refrigerants or gases. They can be used in power generation mode to recover heat wasted, for instance, in the exhaust of an automobile and convert it directly into useful electrical power. In heat pump mode, they can be used for refrigeration, and are then called Peltier elements. Thermoelectric heat converters have no moving parts, are extremely simple and robust, and have a very high specific power density, meaning that they can convert large amounts of power per gram or cubic centimeter of material. Using the thermoelectric materials commercially available until now, their efficiency is lower than that of conventional mechanical heat engines, and therefore are mostly used in low-power applications, but recent research developments promise to change that. Thermoelectric heat converters are based on two complementary effects. The generators use the Seebeck effect, which shows that a temperature difference applied to a conducting solid results in the development of a voltage; the ratio between voltage and temperature difference is the Seebeck coefficient, also known as thermoelectric power or thermopower. In heat pump mode, they are based on the Peltier effect, which shows that when an electrical current passes through a conductor, it results in the development of a heat flow; the ratio between the heat flow and the current is the Peltier coefficient. The efficiency of thermoelectric heat converters depends on the temperature differences, as it does for all heat engines, but also on a material parameter called the thermoelectric figure of merit. Recent developments in nanotechnologies have resulted in a doubling of the efficiency of modern materials compared to the state of the art a decade ago.
Introduction Thermoelectric heat convertors are solid-state heat engines, i.e., solids that convert heat to electricity or
Thermoelectric Heat Convertors Thermoelectric Heat Convertors, Fig. 1 Configuration of a thermoelectric couple as heat engine (left) or heat pump (right), and efficiency of coefficient of performance (COP) of both normalized to their Carnot value
HEAT
COLD
HOT _ __ _ _ _ _ __ _ _
+ + + + + + + ++ + +
_ __ _ __ _ _ __
+ + + + ++ + + + + +
COLD
_ HOT
0.5
0.6
ΔT= 600K 400K 200K
ΔT= 10K 20K 40K
0.4 COPmax / COPCarnot
ηmax / ηCarnot
T
2725
0.4
0.2
0.3
0.2
0.1
0
0
1
2
3
4
5
0
0
1
ZT
inversely use electrical power to pump heat. They behave like conventional steam engines and refrigeration machines [1], except that they have no moving parts. They are extremely reliable, simple, and compact, but their applications have been limited by the fact that their efficiency is lower than that of conventional mechanical systems. That efficiency is a function of a single material parameter, the thermoelectric figure of merit, ZT. Thermoelectric convertors consist of two semiconductor elements, one doped p-type and the other doped n-type, connected as shown in Fig. 1, electrically in series but thermally in parallel. When heat is applied to one side, as shown in the top left panel of Fig. 1, while the other side is kept cold, the temperature difference DT ¼ THOTTCOLD creates a voltage Vp across the p-type element that is positive on the cold side, and a voltage Vn on the n-type element that is negative on the cold side; in Fig. 1, the voltages add across the couple. Many couples can be connected in series to make a pile, which works as a thermoelectric power generator and can be connected to a load (the resistor in Fig. 1). The voltages Vp and Vn are proportional to DT. For each material, the proportionality constant is
2
3
4
5
ZT
defined as the thermoelectric power, also known as thermopower or Seebeck coefficient S: S
V DT
(1)
The “engine”in Fig. 1 can be reversed, as shown in its top right panel. If a current I passes through each element, as shown, that element pumps a heat flow Q proportional to the current. The proportionality constant is the Peltier coefficient P: P
Q q ¼ I j
(2)
and can also be written as the ratio of the thermal to electrical fluxes, q/j. The Kelvin relation relates P and S with the absolute temperature T: P ¼ S:T
(3)
The couple shown on the right therefore has the exact same construction as the couple shown on the left, and several couples can also be connected to form
T
T
2726
Thermoelectric Heat Convertors
a pile that works as a thermoelectric heat pump, cooling down its top surface and pumping the heat into its bottom surface. The efficiency, of the couple connected as a generator and the coefficient of performance, COP of the couple connected as a heat pump, like those of all heat engines, are a function first of the Carnot efficiency and COP. The ratios /Carnot and COP/COPCarnot are shown in Fig. 1 as a function of one material parameter, the thermoelectric figure of merit, ZT, defined by: ZT
S2 s T; k
(4)
where s and k are the electrical and the thermal conductivity of the material. The numerator of Eq. 4 is labeled the thermoelectric power factor P ¼ S2s, and has much the same significance as the area under the (I,V) curve of a photovoltaic solar cell. Indeed, for a given temperature gradient DT, the voltage V ¼ S.DT and the current I ¼ S/R (R ¼ resistance of the thermoelectric elements, or the source resistance of the generator) so that the power produced is proportional to the intrinsic properties S2s. It is often the case that the n-type and p-type materials have closely matched values of ZT. Maximizing ZT is the goal of research on thermoelectrics. ZT is temperature dependent, and is therefore optimized to reach maximum near the desired operating temperature. The temperature ranges, and the semiconductors used in them, are as follows [2]: 1. Materials used in thermoelectric coolers have to operate around room temperature: These materials are typically semiconductors of the tetradymite class, typified by the binary semiconductor Bi2Te3. Their general formula is (Bi1xSbx)2(Te1ySey)3. One class of material, the I-V-III2 compounds, such as AgSbTe2 that crystallize as rocksalts, can have ZT 1.3 around 200 C, a temperature range intermediate between this and the next. 2. Materials used in waste-heat recovery applications generate electrical power from combustion gases that are typically at 350–500 C. Several classes of materials are suitable for this and have been described in more specialized review articles [2]. The classical semiconductor systems used in these applications are the rocksalt lead chalcogenides, typified by the binary compound PbTe and of
general formula (Pb1xSnx)(SeyTe1y). The negative image of lead with the public at large, while not really applicable to chemically stable and insoluble compounds such as the lead chalcogenide salts mentioned here, has led to the development of n-type lead-free alternatives based on skutterudite semiconductors typified by the binary compound CoSb3, which can reach ZT 1.5, based on a reduction of the thermal conductivity by means of the creation of a special phonon mode that absorbs but does not conduct heat. P-type skutterudites exist, but to date they have much lower ZTs than p-type PbTe-based materials. Another class of materials that is being investigated now, but has not yet given ZT > 1.1, are Zintl-phase semiconductor Mg2Si1xSnx alloys. 3. At the high end, thermoelectric power generators use radioisotopes at 1,000 C as the source of heat, and the commonly useful thermoelectric semiconductors are Si1xGex alloys, often with a second intermediate-temperature (500 C) stage made from skutterudites or lead salts. Other materials, not discussed here, are semiconducting rare-earth chalcogenides and pnictides. 4. At the low end, cryogenic heat pumps operating in the range of 10–200 K would be very useful to cool diode lasers and various detectors. Here Bi1xSbx group V alloys are the semiconductors of choice, now in a cascaded configuration with tetradymite semiconductors at the higher temperature (>160– 300 K) end. Not all materials benefit from nanostructuring, in particular, not those which come with inherently low thermal conductivities such as skutterudites and AgSbTe2. Therefore, only five classes of materials based on Bi2Te3, PbTe, Mg2Si1xSnx, Si1xGex, and Bi1xSbx will be discussed here. Figure 2 shows a summary [2] of the recent great progress achieved in values for ZT, compared to the value ZT ¼ 1 that used to be the historical limit. ZT consists of mutually “counter-indicated” properties, i.e., such that most mechanisms that improve one property in ZT also are deleterious to another. Explicitly, ZT can be written as the product of two sets of counter-indicated properties shown here in parentheses: m ZT ¼ S2 n : qT: k
(5)
Thermoelectric Heat Convertors 2.0 Na0.95Pb20SbTe22 PbTe/PbS
Nano–BiSbTe 1.5
Figure of Merit ZT
Thermoelectric Heat Convertors, Fig. 2 ZT as a function of temperature for a number of classical (ZT < 1) and recently developed (ZT > 1) thermoelectric materials, after [2] (Reproduced by permission of The Royal Society of Chemistry)
T
2727
Pb0.98Tl0.02Te Pb1+xSbyTe Nano n–SiGe
1.0 PbTe BiSbTe n–SiGe Nano p–SiGe 0.5
p–SiGe 0.0
0
200
Equation 5 was derived assuming that the electrical conductivity is given by: s ¼ nqm
(6)
where n is the carrier concentration, q its charge, and m its mobility. The mobility itself is defined as the ratio between the average drift velocity of electrons in a sample under the influence of an applied electric field and that field. The ratio (m/k) of the mobility to the thermal conductivity is counter-indicated because defects and impurities that affect one of these properties usually also affect the other. The other counterindicated property is the product (S2n): Indeed, it is a general rule in doped semiconductors and metals that the higher the carrier concentration, the lower the thermopower, labeled the Pisarenko relation in Ioffe’s seminal monograph on thermoelectricity [1]. Nanotechnology can increase both (S2n) and (m/k), because lowering the dimensionality adds a new parameter that can be tuned to optimize each counterindicated pair of properties. Early work devoted in the last decade focused on improving the ratio (m/k) by reducing the lattice thermal conductivity using phonon scattering mechanisms; more recently focus shifted to (S2n). This entry will therefore be divided into two sections each addressing one of those factors.
400 600 Temperature (C)
800
1000
Improving the Mobility-to-Thermal Conductivity (m/k) Ratio Theory The thermal conductivity of semiconductors is the sum of two contributions, the lattice thermal conductivity kL due to lattice vibrations or phonons and the electronic thermal conductivity kE due to heat conducted by charge carriers: k ¼ kL þ kE
(7)
The kinetic formula developed for heat conduction through ideal gases can be applied to the phonon thermal conductivity, which is then written in terms of the specific heat C of the solid at constant pressure and per unit volume, the phonon or sound velocity, vP, and the phonon mean free path ‘P : 1 kL ¼ CvP ‘P : 3
(8)
The electronic thermal conductivity kE is directly related to the electrical conductivity, s by the Wiedemann-Franz law. The latter is derived from the fact that each electron carries not only a charge q but also an amount of heat 32 kB T, where kB is the Boltzmann constant, and is:
T
T
2728
Thermoelectric Heat Convertors
kE ¼ LsT;
(9)
where L is the Lorenz ratio, which collects universal constants and, for free electrons, is 2.45 108 V2K2. In practice, when electrons are scattered inelastically and thus lose a certain amount of energy in each collision, L < 2.45 108 V2K2. Expressing the electrical conductivity Eq. 6 in terms of the mean free path ‘E of the particles (electrons or holes) that carry the charge, ZT for metals and degenerately doped semiconductors, which most practical thermoelectric materials are, can be regrouped as: zT ¼ S2 n ‘P ‘E
1 1=2 CvP 3q2 T ð2m EF Þ
þ Ln
;
(10)
where m∗ is the effective mass and EF is the Fermi energy of the charge carriers, the difference between their electrochemical potential and the extremum of the conduction of valence band they are in. In so doing, electrons or holes are assumed to behave like quasiclassical particles at the Fermi energy in the solid, with an energy (EF) – velocity (vF) relation given by EF ¼
m v2F 2
, and ‘E ¼ vF t where t ¼ m m=q is the relaxation time, the average time between collisions. Equation 10 illustrates how nanostructures can improve the (m/k) ratio in ZT. The formula is the following: 1. Select a solid and a temperature regime where ‘E 1 is achieved at higher temperature.
2729
T
ZT of p-type Bi0.5Sb1.5Te3 has been increased from 1 for the state-of-the-art (SOA) material up to ZT ¼ 1.4 at 373 K by nanostructuring, obtained by ball milling the bulk material followed by hot pressing. Transmission electron microscopy (TEM) observation indicates mostly Sb-rich nanodots of about of 2–10 nm with fuzzy boundaries, and some Te nanoprecipitates. Compared to SOA material s is slightly higher at all temperatures and S is lower at low temperature and then higher from 433 to 523 K. k is lowered at high temperature due to the large number of interfaces (nanograins, nanoprecipitates, nanodots) that increase phonon scattering; p-type Bi0.5Sb1.5Te3 shows ZT improvements of 20–40% over SOA material, with a maximum shifted to higher temperature. • SiGe [7, 8] Boron-doped p-type doped SixGe1x was prepared from nanoparticles of the constituting element either hot pressed or using SPS. TEM shows particles of about 5–10 nm of Si in a Ge host after sintering process. Property studies as a function of compaction level show that a small change of the density induces a change of s by orders of magnitude. When kL of Si0.8Ge0.2B0.016 is compared with a 96 h ball milled sample, the latter shows a strongly lowered value kL compared to SOA material. It also exhibits a higher S and only a slight decrease of s due to strong interface scattering. This leads to an increase in the power factor P ¼ S2s over the 300– 1,000 K temperature range to a maximum of about 18 mW/cm.K2 at 800 K. Si0.8Ge0.2B0.016 finally gives a higher ZT > 0.7 at 1,000 K compared to 0.6 at 1,100 K for SOA SiGe. Phosphorous-doped n-type nanostructured Si0.8Ge0.2P0.02 was prepared by BM followed by HP (direct current–induced hot pressing method) between 1,273 and 1,473 K. ZT 1.3 at 1,173 K has been achieved. Average grain size is about 10–20 nm after HP and adjacent grains have similar crystal structure, but different crystalline orientations that could promote the phonons scattering. Both k and kL are reduced compared to SOA material due to the enhancement of phonon scattering related to the increased density of nanograin boundaries. At 300 K, k ¼ 2.5 W/m.K and kL ¼ 1.8 W/m.K. Again, s is much less reduced than k compared to the SOA alloy up to 1,023 K and
T
T
2730
shows similar values above this point, even though the electron concentration n 2.2 1020cm3 at 300 K is about the same, indicating a slightly lower electron mobility in the nanostructured samples. S is not much affected by the nanostructuring except for a slight increase in the 673–973 K temperature region. Final ZT shows an improvement of about 40% compared to the SOA sample. • Silicon [9] A huge reduction of kL by 90% is observed in BM n-type silicon (Si0.98P0.02), compared to singlecrystal (SC) material, due to high density of grain boundaries. Nanocrystalline domains from 10 to 100 nm were observed on samples hot pressed at 1,475 K for several hours. Both S and resistivity (r ¼ 1/s) increase compared to SC, the latter related to a decrease in m possibly due to the creation of a potential barrier which could originate from grainboundary scattering or defects at grain boundary or even both. The peak observed on both S and r between 800 and 1,000 K has been attributed to the precipitation of phosphorous. kL has been reduced from 80 W/m.K for heavily doped Si down to 10 W/m.K for nanostructured Si with moderate BM and even smaller value (5 W/m.K) for extended milling and higher ball-to-load ratios’ sample. The m/kL ratio increases from 0.05 for SC up to 0.24 for nanostructured bulk Si. Finally, for best sample, ZT has been increased by a factor of 3.5 compared to Si SC, reaching ZT ¼ 0.7 at 1,200 K. • PbTe [7, 8] 100-nm PbTe nanocrystals Ag-doped with Ag2Te (p-type) were synthesized via a solution phase technique and densified by SPS and compared to undoped sample. At room temperature they all show an important S 325 and 200 mV/K for undoped and Ag-doped compounds respectively which is higher than polycrystalline or SC PbTe at similar carrier concentration. s is strongly enhanced compared to undoped samples while k remains comparable. The temperature dependence of the resistivity and mobility shows a unique behavior, suggesting that in addition to the phonon scattering, grain-boundary potential barrier scattering also exists and acts as a dominant scattering mechanism. This scattering could arise from the formation of an oxide layer around nanoparticles that
Thermoelectric Heat Convertors
filter the conduction of carrier with energy lower than that of the barrier. Those results suggest that interfacial energy barrier carrier scattering is a good means to enhanced TE performance. Different results obtained by solidification from the melt will be discussed in the top-down approach section. • Mg2Si1xSnx [10, 11] Antimony-doped n-type Mg2Si0.4xSn0.6Sbx 0 x 0.015 alloys show natural nanostructuring after melting, due to compositional fluctuations and structural modulation (16 nm) and nanodots (10 nm) naturally formed by phase separation in a fashion that will be described in the following paragraph. Preparation by BM and HP compaction [10] results in a ZT 1.1 at 800 K for x ¼ 0.0075, which is similar to that of the SOA. However, a new approach is described for this system: Nanostructuring of Mg2Si [11] was achieved by using microwave heating. Nanopowders of both precursor Mg and Si were first obtained by high energy BM in dry condition, then the mixture was cold pressed and heated by microwaves. While BM influences the size of the particles and the behavior of particles under irradiation, irradiation power and time influence grain growth. Indeed the reaction is initiated by the exclusive absorption of microwaves by Si, making the size of Si particle an important parameter to control. Smaller particles means higher power which in turn induce the sublimation of Mg. Using precursor powders prepared by 2 h BM subjected to microwave power 175 W for 2 min, Mg2Si nanoparticles smaller than 100 nm can be prepared. It remains to be seen if such large particles can affect k much. Top-Down Approaches Different solidification processes can be used to achieve nanoinclusions embedded in a matrix and are mainly due to phase separation. The study of phase diagrams is essential to this approach, and examples of the metallurgical techniques used are based on the phase diagrams shown in Figs. 3 and 4. In summary, the techniques used fall into five categories: (a) The crossing from a single- (solid solution, eutectic composition) to a two-phase region is shown in Fig. 3. In the case of the solid solution, the decreasing of the solubility with temperature is an advantage to obtain fine structure.
Thermoelectric Heat Convertors
2731
1000 (a)
Temperature / °C
900
800
700
PbTe
(c) 582°C
Pb2Sb6Te11 587°C
600
a Sb2Te3
500
(b)
T
(b) Decomposition of a metastable phase: In Fig. 3, Pb2Sb6Te11 decomposes into PbTe and Sb2Te3 if annealed at appropriate temperature. (c) In Fig. 3, solidification from the melt allows control of crystallite size with the cooling rate. (d) In specific phase diagrams such as shown in Fig. 4, spinodal decomposition or nucleation and growth can be used in specific regions. Both phase separation processes occur in the solid state. (e) Liquid encapsulation techniques involve solubility of both phases at high temperature (liquid state) and no or very low solubility at room temperature (solid state). During cooling, the matrix solidifies first (higher melting point) and traps the nanoinclusion that solidify later.
400 0
10
20
30
PbTe
40 Sb2Te3
Sb concentration (at.%)
Thermoelectric Heat Convertors, Fig. 3 A pseudo-binary phase diagram showing the different cooling schemes that can result in nanoparticle formation from the melt
α
α2
α1 α1 + α2 Spinodal Decomposition A
B at%
B
Nucleation
Thermoelectric Heat Convertors, Fig. 4 A pseudo-binary phase diagram showing the solid regions where anneals can result in nanoparticle formation
• PbTe Nanostructuring • PbTe/Tetradymites [8, 12] Different methods were explored to obtain nanostructures in the system PbTe-Sb2Te3 [8] by solidification from liquid based on (a) the presence of an eutectic reaction, (b) eutectoid decomposition of Pb2Sb6Te11, and (c) precipitation. The first process was investigated using eutectic, hypereutectic (Sb2Te3 rich), and hypoeutectic (PbTe rich) compositions. Microscopic analysis indicates that PbTe-rich material shows a dendritic microstructure of PbTe in a Pb2Sb6Te11 matrix, while Sb2Te3-rich compounds have lamellar microstructure of Sb2Te3 in the same matrix. The eutectic composition shows both lamellae of Sb2Te3 and PbTe dendrites. The study of the microstructure as a function of the cooling rate indicates that a finer structure is obtained by increasing the solidification rate, and that nanometer scale structures can be reached with ultrarapid solidification techniques. Crystallographically oriented lamellar structures with alternating (PbTe/Sb2Te3) composition similar to those observed in thin-film superlattices [3] were obtained by using the decomposition of metastable Pb2Sb6Te11 into PbTe and Sb2Te3. Interlamellar spacings around 180 nm can be controlled by the temperature and time of the decomposition process. Those alloys are p-type with high carrier concentration. Widmanst€atten precipitates were observed in a PbTe matrix by precipitation of Sb2Te3 in a PbTe solid solution during unidirectional (Bridgman)
T
T
2732
solidification of Pb10.5Sb31.6Te57. The formation of those plate-like and elongated ribbon-like precipitates is due to the decrease of the solubility with decreasing temperature. Similar work has been done with the PbTe/ Bi2Te3 system [12] off and near the eutectic region of the pseudo-binary (PbTe)x(Bi2Te3)y system with x:y ¼ 2:1,1:1,1:2, and 1:3. After water quenching from the melt, phase separation occurs. The 1:1 sample is mainly PbBi2Te4 with micrometer-sized precipitates of Bi2Te3 containing lamellar PbTe. Bi2Te3-rich samples (1:2,1:3) show randomly oriented lamellar microstructures with alternating composition of PbTe-, Bi2Te3-, PbBi2Te4-, and Te-rich phases. PbTe-rich samples (2:1) present a strong PbTe dendritic microstructure in a PbBi2Te4 matrix. All these compounds have S < 0 with the lowest value S ¼ 75 mV/K at 550 K for 1:2 composition. Neither r nor k shows a strong temperature dependence, and r is lower than in binaries which was attributed to the lower resistance of PbBi2Te4. At 300 K, kL of the alloys is the same as that of Bi2Te3 but becomes smaller with increasing temperature, presumably by scattering of phonons on the increased number of heterointerfaces due to phase separation. • PbTe – M (M ¼ Sb, Bi, InSb) by Liquid Encapsulation [7, 8] PbTe-x%M systems where M ¼ Sb, Bi, or InSb and x ranges from 2% to 16% were investigated, and gave similar SEM and XRD results. No solubility of M in the matrix was observed, and nanoinclusions start to form for compositions as low as 2%M. Below x < 4%, nanostructuring is achieved with 2–6-nm inclusions evenly dispersed and probably coherently embedded, with a size that increases with x. At higher x-values, the second phase agglomerates, thus forming larger microstructures at first, and a network at yet higher x. Measurements of kL indicate that among the PbTe-x%Sb samples, the lowest value is achieved for the 2% sample from 300 to 700 K (kL ¼ 0.8 W/m.K at 300 K). kL increases with x and at x ¼ 16%, T > 500 K, exhibits a higher value than PbTe itself. The authors attribute this high kL to a percolation phenomenon. In the PbTe-4%M series, the lowest kL is achieved for M ¼ InSb while kL is higher than
Thermoelectric Heat Convertors
that of binary PbTe for M ¼ Bi: A large contrast in average mass between matrix and guest atoms is needed to obtain strong phonon scattering. • Pb1ySbyTe1xSex [13] The series Pb9.6Sb0.2Te10xSex (x ¼ 0–10) shows Sb-rich nanophases widely distributed in the matrix. No systematic trend is observed in the variation of either S or r, suggesting the presence of inhomogeneities, microscopic cracks, or different doping levels in Pb9.6Sb0.2Te10xSex. The highest S (282 mV/K at 700 K) is found for x ¼ 2; kL of samples with x ¼ 1–8 are below 1 W/m.K for 300 K < T < 800 K and the lowest kL 0.40 W/m.K was measured for Pb9.6Sb0.2Te3Se7. ZT of this n-type compound exceeded that of PbTe around 500 K and reaches ZT ¼ 1.2 at 650 K. • PbTe-PbxSby by Liquid Encapsulation [7] An excess of Pb in PbTe prepared from the melt (p- and n-type doped with Tl and Ag respectively) create 40-nm Pb nanoinclusions: S is increased (see Section “Improving the S(n) or Pisarenko Relation”), but counterbalanced by a decrease in m and s. Using the matrix encapsulation technique co-nanostructured PbTe containing both x% Pb and y% Sb excess (0.5% x, y 3%) was prepared with a majority of 2–10 nm nanoinclusions and some microprecipitates. Above 2–3% excess, Pb and Sb precipitate as a eutectic at the grain boundaries of the matrix. The transport properties and their temperature dependence show important changes compared to SOA PbTe, that are complicated functions of the z ¼ Pb/Sb ratio. Classifying the changes in order of decreasing electrical conductivity at all temperatures: (1) for z ¼ 0.25, s decreases with increasing temperature; (2) for z ¼ 0.66, s increases up to 430 K, then decreases with temperature; (3) for z ¼ 1 (two samples), s is constant up to 450 K and then decreases with temperature; (4) finally for z ¼ 1, s increases with increasing temperature. On the other hand, S shows n-type behavior and the absolute value of jSj roughly decreases with increasing z, and increases linearly with increasing temperature, suggesting a constant carrier concentration over the temperature range; this is confirmed by Hall coefficient measurements. This carrier concentration increases with the (Pb + Sb) total concentration, consistently with the known donor behavior of
Thermoelectric Heat Convertors
Sb substitutions on Pb sites, and of Te vacancies. The best properties were found for the composition PbTe-Pb0.005Sb0.02 at z ¼ 0.25, where P 30 mW/cm.K2 (T ¼ 700 K) is increased by about 71% over SOA-PbTe. ZT is smaller than SOA-PbTe for T < 475 K but reaches ZT 1.5 at 700 K. • PbTe/Ge and PbTe/SixGe1x [7] n-doped PbTe/Ge and PbTe/SixGe1x with eutectic (20% Ge) and hypereutectic (2.5, 5, and 10% Ge) compositions were prepared. For all compositions, Ge appeared to be only slightly soluble in PbTe, inducing the formation of small Pb precipitates and rodlike Ge nanoprecipitates not coherently embedded and randomly oriented. Those provide stronger mechanical properties and reduced brittleness compared to SOA-PbTe. Hypereutectic compositions show lower kL, while eutectic compositions show kL comparable to that of SOA-PbTe. A minimum kL < 1.5 W/m.K is obtained for PbTe0.95Ge0.05. Further reductions of k were realized by alloying Ge with Si and the use of PbI2 as an n-type dopant to optimize the carrier concentration independently: ZT 1.3 at 780 K was obtained for PbTe (Ge0.8Si0.2)0.05. • PbTe/PbS by Spinodal Decompositions [7] Spinodal decompositions and the nucleation and growth phenomena (Fig. 4) were explored as a way to nanostructure (Pb0.95Sn0.05Te)1x(PbS)x with x ¼ 0.04, 0.08, 0.16, and 0.30. PbI2 was used as n-type dopant. The Sn substitution helped simplify the synthesis. The nucleation and growth occurs and increases with x up to x ¼ 0.08. The phenomenon totally disappears for x ¼ 0.3. The 4% samples tend to form solid solutions with some sign of nucleation and growth of nanodots; the 8% samples contain 3–10 nm nanocrystals, and spinodal decomposition occurs leading to compositional fluctuations shaped as periodical parallel stripes of PbS- and PbTe-rich regions on a nm scale. Spinodal decomposition becomes more prevalent with increasing x. The 16% sample shows coexisting regions with 3–10 nm PbS nanoparticles embedded in a PbTe matrix outside of the spinodal region. Transport property measurements indicate that both s and S increase with x up to 8%; kL is reduced compared to bulk material. The smallest values are obtained for 8% (kL 0.38 W/m.K) and 16% (kL 0.40 W/m.K) samples where dots from the
2733
T
nucleation and growth and stripes from spinodal decomposition coexist. It is further suggested that larger interface densities and a larger composition variation help to enhance acoustic phonon scattering. The highest ZT (1.50 at 640 K) is achieved by the 8% sample. • PbTe-SrTe [14] Endotaxy can be defined as the bulk version of epitaxy, and consists of the growth of oriented precipitates with aligned crystallographic planes and directions embedded in a substrate. Na-doped p-type PbTe containing 0.5–2% endotaxially arranged SrTe nanoprecipitates has been shown to have a kL reduced by selective phonon scattering, while charge carriers are much less affected. The 1–15 nm nanoprecipitates are spherical or ellipsoidal in shape; the smaller ones are coherently strained while the larger ones exhibit interfacial dislocation. m, s, k and S are not really affected by increasing the SrTe content. The Hall coefficient is positive with a maximum at about 430 K ascribed to the appearance of a second heavy-hole valence band in PbTe that seems rather independent of the SrTe content, which in turn suggests that the Sr does not go into a solid solution. The presence of this second band is also confirmed by a change of slope in m(T) and the high value of S at high temperature. Only the sample with 0.5% SrTe shows a slight increase of m(T ¼ 300 K) from 350 cm2/V.s to 500 cm2/V.s. The 1% and 2% SrTe samples show a small decrease of S at high temperature. The highest P (25 mW/cm.K2) is reached by the 2% sample at 560 K. SrTe strongly affects kL, which decreases with increasing nanoprecipitate proportion, almost dividing kL by 2 over the whole temperature range. The lowest kL 0.45 W/m.K2 is reached by the 2% sample at 800 K; it also shows the highest ZT ¼1.7 at 815 K, and ZT > 1 from 550 K upward. • (PbTe)1x(Ag2Te)x [15] Large (50–200 nm) Ag2Te plate-like precipitates uniformly dispersed in a PbTe matrix were obtained by stepped solidification from the melt in the (PbTe)1x(Ag2Te)x system with x ¼ 1.3, 2.7, 4.1, and 5.5. Ag2Te is soluble in the PbTe matrix up to 1%, above which the proportion of Ag2Te precipitate increases with x while the size remains constant. The series shows consistent semiconducting
T
T
2734
behavior with a high r 40mΩ.cm at 300 K that reaches a maximum at about 400 K, indicating a change from an extrinsic to an intrinsic behavior. This change is confirmed by the change of sign of S and Hall coefficient at the same temperature and could be related to the phase transition of Ag2Te from a monoclinic to a cubic structure that occurs at the same temperature. k decreases with increases in both concentration of Ag2Te and temperature; at higher temperature, this is due to the behavior of the kE contribution and the dependence of r(T). A reduction of kL is observed, due to both Ag alloy scattering in PbTe matrix and to phonons scattering on Ag2Te nanoprecipitates. Thermoelectric properties have been optimized by doping using La as a donor. The nanostructure remains and La is homogeneously dispersed in the matrix. Compositions near (Pb1yLayTe)0.945(Ag2Te)0.055 with y ¼ 0.094, 0.0186, 0.0276, 0.0364 all exhibit heavily n-type doped semiconducting behavior with a strongly reduced r 700 K. kL remains low and reaches a minimum of 0.4 W/m.K at T > 650 K for y ¼ 0.0276. Finally a high ZT ¼ 1.6 at 775 K is obtained for z ¼ 0.0276 and carrier density n 3 1019cm3. • AgPbmSbTe2+m (LAST-m) and Derived Compounds [7, 16] LAST alloys have an average NaCl structure where two Pb2+ atoms are isoelectronically substituted by one Ag+ and one Sb3+, thus giving an average charge of +2 counterbalanced by a 2 charge of chalcogens. These materials are excellent n-type thermoelectrics. Ag deficiency allows to tune the electron concentration and acts as a n-type dopant. By cooling the sample from the melt, spontaneous nanostructuring occurs that consists of coherent and semicoherent nanocrystals of Ag-Sb embedded in a PbTe matrix and compositional fluctuations at nanoscopic level is observed. Those materials are stable up to their melting point ( 100 K, the density of both electrons and holes increases through thermal activation, and this gives a negative temperature coefficient to R(T). In the semiconductor regime, the negative slope prevails over the entire temperature regime. The resistance of the 15 nm wires and the hightemperature resistance of the 9 nm wires can be fit to an activated behavior. The 4 nm wires, and the 9 nm wires below 200 K, have a resistance that follows a T1/2 law; magnetoresistance measurements on narrow wires show that this behavior is due to weak localization effects. The thermopower of bulk Bi (along the binary direction) and of the 200 nm wires is similar. The thermopower of the 15 nm wires has
Thermoelectric Heat Convertors
2739
1x102
1x105
T
9 nm, Al2O3
15nm, SiO2 Eg = 0.19 eV 1x104
15nm, SiO2
9nm, Al2O3
Eg = 0.39 eV
1x101
|S| (μV/K)
R(T) / R (300 K)
1x103
48 nm, anodic 4 nm Vycor
1x100
4 nm Vycor 1x102
Bulk Bi
1x101
70 nm, anodic
1x100
Bulk Bi
1x10−1
200 nm, anodic
0 0
100
200 T (K)
300
Thermoelectric Heat Convertors, Fig. 7 Normalized resistivity of Bi nanowires of the diameters indicated, after [21]
a very pronounced temperature dependence, which was attributed to two-carrier conduction and a compensation between the electron and hole partial thermopowers. The thermopower of the 9 nm wires is much enhanced over that of the bulk, and follows a T1 dependence shown as a dashed line, as expected from an intrinsic semiconductor. Finally, the lowtemperature Seebeck coefficient of the 9 nm wires, and that of the 4 nm wires, is strongly decreased, due to localization effects. While Bi is the material that is most easily subject to size quantization, the nanostructures prepared from Bi nanowires in porous hosts have the inherent disadvantage that the host material constitutes a thermal short: The spectacular thermopowers have not yet resulted in high overall effective ZT of the filled composite. Metallic Wires Ideal and practical “bulk-like” nanowires systems would need to be prepared using a top-down approach similar to that in section “Improving the Mobility-toThermal Conductivity (m/k) Ratio” for nanodots, and
100
200
300
T(K)
Thermoelectric Heat Convertors, Fig. 8 Thermopower of Bi nanowires of the diameters indicated, showing a 1,000-fold increase of S over bulk that of Bi due to quantum confinement (After [21])
should constitute of wires grown naturally inside bulk materials. One recent example has been published by Sugihara et al. [22], where Cu-Ni alloys were deposited by rf magnetron sputtering between Ta and Au layers. Figure 9 shows an atom-probe tomographic analysis of the Cu-Ni layer. Columnar growth of one particular composition is in evidence. Cu and Ni phase form solid solutions throughout the composition range. Among these alloys is constantan, a thermocouple alloy with very large S 40 mV K1 near 300 K. Large values of thermopower are observed over an extended Cu-Ni composition range. The origin for this is that in these alloys, the Fermi level lies close to the edge of the bands that have their origin in the d-levels of Ni. Preliminary thermoelectric measurements were performed on the electrochemically grown material, and at first sight seem to indicate a thermopower value one order of magnitude larger than for bulk alloys. The measurements are very indirect, as the authors measured the heat induced by passing a current through the sample. This heat is the
T
T
2740
Thermoelectric Heat Convertors, nanocolumns in Ni-Cu alloys (After [22])
Thermoelectric Heat Convertors
Fig.
9 Ni-rich
sum of resistively induced Joule heat, and the Peltier heat given by Eq. 2. Joule heating experimentally overwhelms the Peltier effect in all but the very best thermoelectric materials: The data the authors report for the Peltier coefficient are therefore the difference between two large numbers and are in need of further verification. Still, the approach is promising because it is simple.
Conclusions Nanostructuring has proven to be one of the most versatile and effective tools to increase the thermoelectric figure of merit ZT which governs the efficiency of solid-state heat-to-electricity converters. It thus promises to be a major enabler to a major energy-saving technology usable in waste-heat recovery systems. This entry outlines the mechanisms through which nanotechnologies reduce the thermal conductivity of the thermoelectric materials, thus minimizing the heat
losses due to conduction. It also shows how in quantum-confined system, i.e., nanowires of diameters smaller than the wavelength of the electron, the thermopower is strongly enhanced, which means that more electrical power can be delivered from a give temperature difference. In this concluding paragraph, it must be repeated that nanostructuring is not the only approach used: Resonant energy levels can also produce enhancements in thermopower as demonstrated in Tl-doped PbTe, and the use of local phonon modes in CoSb3 or enhanced intrinsic phonon-phonon interactions in thermoelectric solids such as AgSbTe2 can lead to strongly reduced thermal conductivity. Can these effects be additive? Some can, others no. As mentioned, there is a lower limit to the thermal conductivity (in bulk solids), so that in systems where that limit is nearly reached with phonon-phonon interactions or local modes, additional nanostructuring will be of less value. Often thermopower-enhancing techniques such as resonant level doping are sufficiently independent of the role of phonons that nanostructuring can lead to additional gains in ZT. Finally, there are many opportunities that remain unexplored yet. In particular, the phonon mean free paths increase with decreasing temperatures, opening the way for even larger gains to be obtained in ZT at cryogenic temperatures. New preparation techniques remain to be explored, for example, the richness and flexibility of organic chemistry can be brought to bear on the problem in organic/inorganic mixed structures. These avenues of research and many others are the objects of active programs. In the future, the field promises to generate many new and important results that will help increase the overall efficiency of heat-toelectricity conversion systems and thus help alleviate the world’s energy crisis.
References 1. Ioffe, A.F.: Physics of Semiconductors. Academic, New York (1960) (translated from Russian, “Fizika Poluprovodnikov”, Russian Academy of Sciences, Moscow, 1957) 2. Minnich, A.J., Dresselhaus, M.S., Ren, Z.F., Chen, G.: Bulk nanostructured thermoelectric materials: current research and future prospects. Energy Environ. Sci. 2, 466–479 (2009)
Thermometry 3. Venkatasubramanian, R., Siivola, E., Colpitts, T., O’Quinn, B.: Thin-film thermoelectric devices with high roomtemperature figures of merit. Nature 413, 597–602 (2001) 4. Harman, T.C., Taylor, P.J., Walsh, M.P., LaForge, B.E.: Quantum dot superlattice thermoelectric materials and devices. Science 297, 2229–2232 (2002) 5. Koh, Y.K., Vineis, C.J., Calawa, S.D., Walsh, M.P., Cahil, D.G.: Lattice thermal conductivity of nanostructured thermoelectric materials based on PbTe. Appl. Phys. Lett. 94, 153101–153103 (2009) 6. Poudel, B., Hao, Q., Ma, Y., Lan, Y., Minnich, A., Yu, B., Yan, X., Wang, D., Muto, A., Vashaee, D., Chen, X., Liu, J., Dresselhaus, M.S., Chen, G., Ren, Z.: High-thermoelectric performance of nanostructured bismuth antimony telluride bulk alloys. Science. 320, 634–638 (2008) 7. Kanatzidis, M.G.: Nanostructured thermoelectrics: the new paradigm? Chem. Mater. 22, 648–659 (2010) 8. Medlin, D.C., Snyder, G.J.: Interfaces in bulk thermoelectric materials A review for current opinion in colloid and interface science. Curr. Opin. Coll. Int. Sci. 14, 226–235 (2009) 9. Bux, S.K., Blair, R.G., Gogna, P.K., Lee, H., Chen, G., Dresselhaus, M.S., Kaner, R.B., Fleurial, J.-P.: Nanostructured bulk silicon as an effective thermoelectric material. Adv. Funct. Mater. 19, 2445–2452 (2009) 10. Zaitsev, V.K., Fedorov, M.I., Gurieva, E.A., Eremin, I.S., Konstantinov, P.P., Samunin, A.Y., Vedernikov, M.V.: Highly effective Mg2Si1xSnx thermoelectric. Phys. Rev. B 74, 045207–045211 (2006) 11. Savary, E., Gascoin, F., Marinel, S.: Fast synthesis of nanocrystalline Mg2Si by microwave heating: a new route to nano-structured thermoelectric materials. Dalton Trans. 39, 1–7 (2010) 12. Yim, J.-H., Jung, K., Kim, H.-J., Park, H.-H., Park, C., Kim, J.-S.: Effect of composition on thermoelectric properties in PbTe-Bi2Te3 composites. J. Elec. Mater. 40(5), 1010–1014 (2011) 13. Poudeu, P.F.P., D’Angelo, J., Kong, H.J., Downey, A., Short, J.L., Pcionek, R., Hogan, T.P., Uher, C., Kanatzidis, M.G.: Nanostructures versus solid solutions: low lattice thermal conductivity and enhanced thermoelectric figure of merit in Pb9.6Sb0.2Te10-xSex bulk materials. J. Am. Chem. Soc. 128, 14347–14355 (2006) 14. Biswas, K., He, J., Zhang, Q., Wang, G., Uher, C., Dravid, V.P., Kanatzidis, M.G.: Strained endotaxial nanostructures with high thermoelectric figure of merit. Nat. Chem. 3, 160– 166 (2011) 15. Pei, Y., Lensch-Falk, J., Toberer, E.S., Medlin, D.L., Snyder, G.J.: High thermoelectric performance in PbTe due to large nanoscale Ag2Te precipitates and La Doping. Adv. Func. Mater. 21, 241–249 (2011) 16. Hsu, K.F., Loo, S., Guo, F., Chen, W., Dyck, J.S., Uher, C., Hogan, T., Polychroniadis, E.K., Kanatzidis, M.G.: Cubic AgPbmSbTe2m: bulk thermoelectric materials with high figure of merit. Science. 303, 818–821 (2004) 17. Mahan, G.D., Sofo, J.O.: The best thermoelectric. Proc. Nat. Acad. Sci. USA 93, 7436 (1996) 18. Heremans, J.P., Jovovic, V., Toberer, E.S., Saramat, A., Kurosaki, K., Charoenphakdee, A., Yamanaka, S., Snyder, G.J.: Enhancement of thermoelectric efficiency in PbTe by
2741
19.
20. 21.
22.
T
distortion of the electronic density of states. Science 321, 554–558 (2008) Hicks, L.D., Dresselhaus, M.S.: Thermoelectric figure of merit of a one-dimensional conductor. Phys. Rev. B 47, 16631 (1993) (1–4) Heremans, J.P.: Low-dimensional thermoelectricity. Acta. Phys. Polon. A 108, 609–634 (2005) Heremans, J.P., Thursh, C.M., Morelli, D.T., Wu, M.-C.: Thermoelectric power of bismuth nanocomposites. Phys. Rev. Lett. 88, 216801 (2002) (1–4) Sugihara, A., Kodzuka, M., Yakusiji, K., Kubota, H., Yuasa, S., Yamamoto, A., Ando, K., Takanashi, K., Ohkubo, T., Hono, K., Fukushima, A.: Giant peltier effect in a submicron-sized Cu–Ni/Au junction with nanometerscale phase separation. Appl. Phys. Express 3, 065204 (2010) (1–3)
Thermoelectric Heat Pump ▶ Thermoelectric Heat Convertors
Thermoelectric Module ▶ Thermoelectric Heat Convertors
Thermoelectric Nanomaterials ▶ Nanostructured Thermoelectric Materials
Thermoelectric Power ▶ Thermoelectric Heat Convertors
T Thermomechanical Actuators ▶ Thermal Actuators
Thermometry ▶ Nanocalorimetry
T
2742
Thermopower ▶ Thermoelectric Heat Convertors
Thermostability ▶ Structure and Stability of Protein Materials
Thickness ▶ Interfacial Investigation of Protein Films Using Acoustic Waves
Thiols ▶ Nanostructures for Surface Functionalization and Surface Properties
Third Generation Vectors ▶ Multistage Vectors
Thomson Effect ▶ Thermoelectric Heat Convertors
Three-Dimensional Imaging of Human Tissues ▶ Imaging Human Body Down to Molecular Level
THz ▶ Terahertz Technology for Nano Applications
Thermopower
TiO2 Nanotube Arrays: Growth and Application Karthik Shankar Department of Electrical and Computer Engineering, W2-083 ECERF, University of Alberta, Edmonton, AB, Canada
Definition TiO2 nanotube arrays (TNAs) denote self-organized tubular nanostructures of titanium dioxide grown by the process of electrochemical anodization. When it is titanium metal which is subjected to electrochemical anodization, the TNAs grow directly in a single step without the assistance of a template. When it is aluminum metal which is subjected to anodization, the formation of TNAs is said to be templated and requires multiple steps including (1) the formation of the nanoporous alumina template, (2) a process to coat the walls of the template with TiO2 or infiltrate a TiO2 precursor into the template followed by heating, etc. (3) release of the alumina template by selective chemical etching to form TNAs. TNAs produced by the one-step process shall henceforth be referred to as anodic TiO2 nanotube arrays and TNAs produced by the multistep process as templated TiO2 nanotube arrays. In either process, the titanium dioxide nanotubes grow in arrays with their long axis aligned orthogonal to the substrate, due to which they are commonly referred to as vertically oriented TiO2 nanotube arrays.
Introduction The formation of anodic porous titania by using fluoride-based electrochemistry was first reported by Zwilling in 1999 [1]. However, it was Gong et al. [2] in 2001 that first formed nanoporous TiO2 with a clearly identifiable nanotubular structure. Pioneering contributions to the field of TNAs were made by the research groups of Craig Grimes [3] at Penn State University and Patrik Schmuki [4] at the University of Erlangen. Initially, anodic TiO2 nanotube arrays could only be grown on titanium foil substrates. In 2005, Mor et al. reported the successful formation of anodic TNAs on
TiO2 Nanotube Arrays: Growth and Application
Ti-coated glass substrates. Since then, anodic TNAs have been formed on silicon wafers and on transparent conducting oxide (TCO) coated glass substrates. Correspondingly, the first report on templated TNAs emerged in 1996 [5], but more detailed studies commenced in late 2001 [6]. TiO2 is a highly versatile compound with applications as a pigment for paints, nontoxic biocompatible material for orthopedic implants, gas sensor, germicide, photocatalyst, n-type semiconductor scaffold for sensitized solar cells, optical coating, structural ceramic, electrical circuit varistor, spacer material for magnetic spin-valve systems, and more. TiO2 is also relatively inexpensive owing to its stability and the plentiful abundance of Ti and O in the earth’s crust. TiO2 occurs in three allotropes: anatase, rutile, and the much less common brookite. The metastable anatase phase has superior electronic properties. The key motivation behind the development of TNAs is to enhance and exploit the versatility inherent in TiO2 using the specific nanostructured morphology of TNAs. The oriented and aligned one-dimensional TNA architecture offers access to the anatase phase, a high surface area and directed electron percolation pathways for electronic and optoelectronic applications. For other applications of TNAs such as their use in stem-cell differentiators, antibiotic elution and energy transfer-based light harvesting, the specific properties of TiO2 are less important; these applications derive instead from size selectivity, confinement in nanochannels, and other geometrical attributes of the nanoscale TNA architecture. The research area of TNAs has witnessed an explosion of interest in the last decade with more than 1,000 research papers on this topic. Apart from the versatility of TiO2 referenced above, another reason for the intense interest is the location of self-organized vertically oriented TiO2 nanotube arrays at the intersection of three major trends in nanotechnology research over the last decade as shown in the schematic of Fig. 1. One such trend has been the drive to form onedimensional structures such as nanorods and nanotubes in semiconductors, inspired by the unique properties of carbon nanotubes. Another trend, which is also one of the driving motivations underlying nanotechnology, is the exploitation of self-assembly and self-organization to form complex structures from the “bottom-up” as an alternative to patterning and fabricating structures from the “top down.” The third trend has been the
2743
T
Anodic nanoporous aluminum oxide; Templated growth of aligned nanomaterials
TiO2 nanotube arrays
Non-lithographic nanofabrication
1-D semiconductors Carbon nanotubes, ZnO nanowires, etc.
TiO2 Nanotube Arrays: Growth and Application, Fig. 1 TNAs sit at the intersection of three major nanotechnology trends
use of ordered anodic nanoporous alumina as a template to form a variety of oriented and aligned nanomaterials. The use of electrochemical anodization to form nanoporous alumina has been known since the 1950s. However nanoporous alumina is an insulator which limits its device applications. Anodic aluminum oxide (AAO) is widely used as a template to form functional nanomaterials. Since crystalline TiO2 is a semiconductor, anodically formed TiO2 nanotube arrays can be directly used as a functional material after the induction of crystallinity. A major difference between AAO and TNAs is that a wider range of morphologies are achievable in TNAs. Ordered AAO has been formed in a close-packed hexagonal pore architecture and in square and triangular lattices. In TNAs, the anodization electrolyte has a decisive role in determining the morphology of the resulting structure. By anodization in ethylene glycol-based electrolytes, TNAs may be formed in a close-packed hexagonal architecture similar to AAO. At the same time, nanotube arrays with a distinctive cylindrical cross section may be achieved in other electrolytes. The wall thickness can be varied to change the porosity of the architecture. In comparison with nanochannels in AAO or in track-etched polycarbonate membranes, the nanotubular structure has a significantly higher
T
T
2744
TiO2 Nanotube Arrays: Growth and Application
surface area due to the availability of both the inner and outer surfaces of the nanotube walls. In the course of the last 10 years 2001–2010, a high degree of control has been achieved over the geometrical dimensions of the anodic TNA architecture by optimizing the anodization process. Potentiostatic anodization is most commonly used either by direct application of a constant anodization voltage or though a linear sweep of the potential to a final constant value. The key parameters involved in the fabrication process are the composition of the anodization electrolyte, the anodization voltage applied, the temperature of the anodization bath, and the anodization duration. For applications involving increased pattern order of the TNAs and for TNA growth on nonnative substrates such as glass, Si, etc., the nature of the Ti film or Ti foil specifically its density, roughness, grain size, and crystallographic texture (if any) are also critical. By tuning the various anodization parameters listed above, control over the geometrical dimensions of TNAs has been demonstrated over remarkably broad length scales [4, 7–9]; the diameter of nanotubes has been varied from 16 to 900 nm, the tube length from 250 to 1 mm, the wall thickness from 5 to 45 nm and the inter-tube spacing from 0 (close packed nanotubes) to several micrometers (widely separated nanotubes). Of these geometrical features, only the diameter and the tube length are independently controllable to date. The wall thickness and inter-tube spacing depend on the diameter and the tube length.
Growth of Anodic TiO2 Nanotube Arrays When a Ti foil or a Ti film coated electrode is subjected to anodization in highly conductive electrolytes, the anodization current experiences a steep drop in the first several seconds of the process. The drop in current occurs due to the rapid formation of a passivating oxide layer (called the barrier layer) through which electronic conduction can only occur through tunneling. In the simplest case, the anodization current density J for Fowler-Nordheim tunneling through the oxide barrier layer (BL) may be expressed as:
DU J¼C t
2
bFN t
e DU
(1)
where DU is the portion of the anodization voltage which drops across the barrier layer, t is the oxide thickness and C, bFN are constants. Equation 1 shows that the tunneling current reduces exponentially with increasing BL thickness. As the passivating layer increases in thickness, tunneling weakens and the formation of nascent gases at the anode due to electron transfer reactions is suppressed. Visually, this is indicated by the cessation of bubbling activity at the anode. At this point, current flowing through the oxide consists almost entirely of ions. The solid state migration of ions through the oxide is described by the high-field transport model in which the anodization current density J is given by: J ¼ Ae
bHF DU t
(2)
where A and bHF are constants. Later in the anodization process, the growth process is under mass-transport control as the anodization current is controlled by diffusion of fluoride ions through the tubes to the pore bottoms and results in an inverse dependence of the anodization current on the tube length [4]. The current density is given as: J ¼ pD
@c ð0 x LÞ @x
(3)
where D, c, x, and L are the diffusion coefficient and concentration of the ionic species, thickness of the diffusion layer, and length of the nanotubes, respectively, and p is the ratio of the weight of dissolved oxide to the weight of produced oxide. The anodic formation of TNAs is a highly nonequilibrium process which occurs at high overpotentials as a result of the interplay between three competing and essential processes (see Fig. 2): Field-assisted oxide dissolution and cation migration, field-assisted oxidation of Ti and chemical etching. It is chemical etching which is responsible for the development of nanotubes as opposed to mere nanopores. In contrast, the formation of nanoporous alumina is controlled mainly by the field-assisted processes. The field-assisted reactions occur on either side of the barrier layer at the bottom of the nanotubes and are responsible for driving the Ti/TiO2 interface deeper into the Ti foil or Ti film, a process which increases the length of the nanotubes. Chemical
TiO2 Nanotube Arrays: Growth and Application TiO2 Nanotube Arrays: Growth and Application, Fig. 2 Schematic representation of the Ti anodization (a) in absence of fluorides (results in flat layers), and (b) in presence of fluorides (results in the tube growth) [4] (Reprinted with permission from Elsevier)
T
2745
a
b Electrolyte H2O H+
ΔU F= t
t
O2−
Ti(OH)xOy
Electrolyte TiO2 F−
Ti4+ TiO2
O2−
[TiF6]2−
F−
etching Ti4+ oxidation
Ti
etching shortens the length of the nanotubes. The reactions involved in the form ation of TNAs are: Field assisted oxidation: Ti þ 2H2 O ! TiO2 þ 4Hþ þ 4e Field assisted migration:
Ti4þ þ 6F ! ½TiF6 2
(4) (5)
Field assisted dissolution: TiO2 þ 6F þ 4Hþ ! ½TiF6 2 þ 2H2 O
(6)
Chemical dissolution: TiO2 þ 6HF ! ½TiF6 2 þ 2H2 O þ 2Hþ
(7)
The as-anodized nanotube arrays are amorphous and non-insulating. Manifestation of the semiconductor properties of TiO2 requires induction of crystallinity, which is typically accomplished by annealing the nanotube arrays in air or in an oxygen-rih ambient at elevated temperatures of 350–580 C. For annealing temperatures of up to 500 C, the nanotubes are transformed into polycrystalline anatase alone; at higher temperature, increasing amounts of rutile are present as well. Annealing at temperatures higher than 580 C results in the destruction of the nanotubular architecture. The evolution of different fabrication recipes for anodic TNAs has occurred in four generations to date [3, 7, 8]. The first three generations were defined by the maximum length of nanotubes that was achieved and they all used fluoride ions to effect the necessary chemical etching. The fourth generation was defined by its nonuse of fluoride ions. First-generation nanotubes were up to 500 nm long and were grown
Ti
by anodization in strongly acidic aqueous electrolytes containing 0.1–0.5 wt% HF (pH 1). Other additives such as acetic acid were sometimes added to improve the mechanical strength and quality of the nanotubes. The tube length in first-generation TNAs was limited by the strong chemical etching by HF. Typical morphology obtained in first-generation nanotubes is shown in Fig. 3a. The steady-state TNA film thickness (corresponding to the tube length) occurred when the rate of movement of the Ti/ TiO2 interface deeper into the metal was exactly balanced by the chemical etching at the mouths of the nanotubes. In the second generation, fluoride ion-bearing aqueous electrolytes were still used but the rate of chemical etching was tightly controlled by adjusting the pH by adding buffers to the anodization electrolyte. Typically, trisodium citrate and sodium dihydrogen sulfate were used as buffering agents in the electrolyte and salts such as ammonium fluoride, sodium fluoride, and potassium fluoride were used as sources for fluoride ions. As the pH was varied from 1 upward toward 5, the nanotubes grew correspondingly longer due to the reduction in chemical etching. The maximum nanotube length in aqueous electrolytes of 6.6 mm was achieved at a pH of 5. The maximum growth rate was 0.25 mm/h. When the pH was increased beyond 5, nanotube formation did not occur. Nanotube formation in both the first and second generations occurred in aqueous electrolytes in a narrow window of anodization voltages ranging from 10 to 27 V, which restricted the inner diameters of the nanotubes to the range 22–100 nm. In the third generation (Fig. 3b, c, d), fluoride-bearing organic electrolytes were used to vastly expand the morphological parameter space achievable. Nanotubes as long as 1 mm were grown by complete conversion of the underling Ti foil to
T
T
2746
TiO2 Nanotube Arrays: Growth and Application
TiO2 Nanotube Arrays: Growth and Application, Fig. 3 Scanning electron micrographs showing the effect of anodization electrolyte on the morphology of TNAs: (a) Typical morphology obtained in aqueous acidic fluoride or HF electrolytes, (b) glycerol/ fluoride electrolytes, (c) ethylene glycol/fluoride electrolytes, and (d) DMSO/ fluoride electrolytes (Images a through c were reprinted from [4] with permission from Elsevier. Image d was reprinted from [9] with permission from Nature)
TNAs. Thus, the nanotube growth is limited in length only by the thickness of the underlying substrate and in principle can be extended even longer by choosing thicker substrates. The inner diameter of the nanotubes could be varied from 17 to 900 nm. When chemical etching is minimized, the slow solid state ionic transport of reactants through the barrier layer is the rate-limiting step preventing further increase in tube length. From Eq. 7, it is clear that the high-field ionic current is increased by a thinner barrier layer. This key insight motivated the use of organic electrolytes in the third generation which provided more reducing conditions than aqueous electrolytes and ensured the formation of a thinner barrier layer. The thinner barrier layer allows growth rates as high as 15 mm/h to be achieved. Third generation electrolytes consist of a fluoride-ion bearing salt dissolved in an amphiprotic organic solvent of moderate to high viscosity such as formamide, N-methyl formamide, ethylene glycol, glycerol, etc. About 1–5% of water is typically added to form organic electrolytes in order to provide oxygen-bearing hydroxyl ions during anodization for oxide formation. Nanotubes do not grow well when low-viscosity protic solvents such as methanol or ethanol are used as the base solvent to form the organic electrolyte (although mixtures with water and HF have been found to result in anodic TNAs). The same is true
of aprotic solvents such as propylene carbonate and acetonitrile. An exception is dimethylsulfoxide, an aprotic solvent which has been used to form high-quality TNAs. By detaching the nanotube arrays from the foil substrate, TNA membranes similar to nanoporous alumina membranes may be produced. The nanotube arrays need to be at least 10 mm long for the detached film to be mechanically robust enough to a form free-standing membrane. Some of the methods used to detach the TNAs from the native Ti foil substrate include the use of natural stresses in the thick TNA films through ultrasonication, selective etching of the Ti substrate by immersion in a solution of CH3OH/Br2, and the use of induced stresses by double-step anodization. The membranes have a strong tendency to curl up due to capillary forces. The membranes can be kept planar by processing with low surface tension solvents or by supercritical CO2 drying. Fourth-generation nanotubes were formed in fluoride-free electrolytes (Table 1).
Growth of Templated TiO2 Nanotube Arrays Both hard templates such as anodic nanoporous aluminum oxide (AAO) and soft templates such as polycarbonate membrane filters have been used to form
TiO2 Nanotube Arrays: Growth and Application TiO2 Nanotube Arrays: Growth and Application, Table 1 Common recipes for TNAs and the resulting morphological parameters
2747
Electrolyte Aqueous 0.5% HF Aqueous 0.5% HF Aqueous buffered 0.1 M KF (pH 5) 0.1 M NH4F in 1:1 DMF:H2O 0.15 M NH4F + 5% H2O in formamide 0.15 M NH4F + 5% H2O in formamide 0.15 M NH4F + 5% H2O in formamide 0.22 M NH4F + 5% H2O in formamide 0.27 M NH4F + 5% H2O in formamide 0.3% NH4F in EG + 1% H2O 0.3% NH4F in EG + 1% H2O 0.3% NH4F in EG + 1% H2O
Anodic potential (V) 10 10–23 (nanocones) 25 15 porous 10 15 20 35 20 30 40 65
Inner diameter (nm) 22
T
110
Length (mm) 0.2 0.4 6.6
12 29 70 180 65 45 70 135
3.6 5.4 14.4 30 19.6 9 12 30
Reprinted with permission from [7]
TNAs. Soft templates have two major disadvantages: (1) an elevated temperature (450–500 C) thermal annealing step is frequently to induce or improve the crystallinity of the templated TNAs. However, most polymers are not able to withstand temperatures above 200 C. Consequently, the polymer template pyrolyzes before the phase transition in TiO2 thus eliminating the mechanical support provided by the template and/or result in the undesirable incorporation of carbonaceous materials in the TNAs; (2) the pore densities are smaller and the nanochannels in polymeric membrane filters are frequently tilted from the surface normal and sometimes intersect within the membrane [10]. Therefore, discussion shall be restricted to AAO-templated TNAs. AAO templates are formed by electrochemical anodization of aluminum metal in three common electrolytes based on sulfuric acid, oxalic acid, and phosphoric acid. The choice of electrolyte is determined by the desired pore size. The Al metal maybe a native substrate such as aluminum foil or could be a film deposited by vacuum evaporation or sputtering on nonnative substrates such as glass and silicon. High-ordered templates approaching ideal hexagonal, square, or triangular lattices may be prepared by techniques such as two-step anodization and various imprinting methods to guide pore formation [11]. Methods to synthesize TNAs in AAO templates can be classified into liquid phase and gas-phase methods. Sol-gel hydrolysis, electrochemical hydrolysis and electrophoretic deposition of nanoparticles are examples of liquid phase methods. Atomic layer deposition (ALD)
and chemical vapor deposition (CVD) are gas-phase methods. The growth of nanotubes as opposed to nanorods in nanoporous alumina requires preferential formation of TiO2 on the walls of the template. This is accomplished by preferential wetting of the template walls by TiO2 precursors in sol-gel methods, by preferential reduction or oxidation at the walls in electrochemical methods, by preferential concentration of the electric field lines at the pore walls in electrophoretic deposition, and by the non-line-of-sight conformal deposition on the walls in gas-phase methods. ALD is a self-limiting film growth process in which alternate saturative surface reactions are made to occur by pulsing the vapors of two or more precursors in the reaction chamber alternately [12]. ALD is capable of producing highly conformal coatings even on substrates with complex features and offers exceptional control in tuning the wall thickness of the nanotubes. Consequently, ALD is the most widely used method to form the templated growth of TNAs. Typically, a flow reactor configuration is employed and the precursors used to grow TiO2 films are a titanium alkoxide, whose generic formula is Ti(OR)4 and water (H2O). The deposition temperatures used in ALD of TiO2 films for templated TNA growth vary widely, from 100 C to 325 C. The exposure times of the template to each precursor pulse are typically limited to a few seconds and the chamber is evacuated prior to the introduction of the next pulse. In addition, in some recipes, an inert gas is used to purge the chamber before the next pulse. The main disadvantages of ALD are the high cost and complexity of ALD equipment and
T
T
2748
TiO2 Nanotube Arrays: Growth and Application
Biological Applications of TiO2 Nanotube Arrays
TiO2 Nanotube Arrays: Growth and Application, Fig. 4 Scanning electron micrograph of templated TNAs formed by sequential infiltration and oxidation of titanium isopropoxide, followed by removal of the alumina matrix in which the tubes were formed (Reprinted with permission from [6])
the extremely slow film growth rate. A brief overview of other processes used to form templated TNAs is now provided. The operating principle of the sol-gel method is the use of capillary forces to drive a TiO2 sol into the nanochannels of the template which transforms into TNAs upon thermal annealing. A sol-gel process based on TiF4 hydrolysis with hydroxypropylcellulose added as a gelling agent, has been used to form 200 nm diameter TNAs in an AAO template. However, the process was highly nonuniform over large areas with irregular deposition of the TiO2 and poor penetration of the sol particles into the template nanochannels. Much better results were obtained using the pressure-assisted infiltration of liquid titanium isopropoxide (TiO2 precursor) into the AAO nanochannels followed by thermal decomposition method of the precursor at 500 C in an oxygen stream to yield TiO2 nanotube arrays [6]. In the electrophoretic sol-gel method, an electric field is used to provide significantly larger forces than mere capillary action to drive the charged particles created by sol-gel processing into the nanochannels [13]. A triple step templating process has been used to form TiO2 by the anodic oxidative hydrolysis of TiCl3 [5]. In this process, the AAO template morphology was transferred to a polymer mold which contained a negative replica of the template pattern. Gold nanorods were formed by electroless deposition on to the polymer mold. Then the gold nanorods served as a conducting template to form TNAs (Fig. 4).
Biosensing Applications Biosensors are devices that combine a biological component (a recognition layer) and a physicochemical detector component (a transducer). The majority of biosensors are targeted to proteins but biosensors for cells and for metabolically important small molecules are also being investigated. Molecular labels have enabled the construction of biosensors of high sensitivity and specificity. However, labels often disrupt the accurate measurement of kinetic constants, particularly binding equilibria, and problems, such as antibody cross reaction or impure solutions, can occur. Due to these drawbacks, label-free biosensors are highly desired for a variety of biotechnology applications such as disease diagnostics, drug discovery, environmental monitoring, and food safety. For label-free optical biosensors, the transduction methods can be divided into two categories: optical interferometric and surface plasmon methods [14]. In interferometric sensors, sensing is achieved by extracting the refractive index of the porous matrix from the Fabry-Perot interference fringes in the reflectance spectrum. Some of the prerequisites for an effective optical interferometric sensing material are a large internal surface area, ease of fabrication, and a chemically modifiable surface. TiO2 nanotube arrays satisfy the above requirements but in addition possess a high refractive index (n ¼ 2.5 for TNAs) and superior chemical stability over a wide pH range in comparison with other candidate materials such as porous silicon and nanoporous alumina. The larger index contrast between the porous host and the aqueous matrix in which the biomolecular binding measurement is carried out is expected to provide greater contrast in the interferometric spectrum, leading to lower noise and higher sensitivity [14]. Mun et al. [14] demonstrated the highly specific label-free interferometric sensing of rabbit immunoglobulin G (IgG) using protein A (derived from Staphylococcus aureus bacteria) immobilized on the walls on the nanotubes as the capture probe (Fig. 5). Orthopedic Implant Applications Because of their mechanical strength, chemical inertness, and biocompatibility, TiO2 and its alloys (particularly Ti-6Al-4V) have been used extensively
TiO2 Nanotube Arrays: Growth and Application
a
T
3500 TiO2 nanotubes 3000
2500 Reflected intensity
TiO2 Nanotube Arrays: Growth and Application, Fig. 5 TNA-based interferometric biosensors: (a) representative reflectivity spectrum of anodic TNAs showing the characteristic interference fringes and (b) time-dependent optical thickness measurements showing sequential binding of Protein A and rabbit IgG within TiO2 nanotubes (Reprinted with permission from [14])
2749
2000
1500
1000
500
0 400
500
600
700
800
900
1000
Wavelength, nm
b 14.76
Optical Thickness, nm, x103
14.74 Rabbit IgG
Protein A 14.72
PBS
PBS
Rabbit IgG Protein A TiO2 Nanotube
14.70 PBS 14.68
14.66 0
20
in orthopedic and dental implants since 1970 [15]. Osteogenesis is the process of formation of new bone tissue by cells called osteoblasts. Ideal implants are those whose surfaces will promote in vitro and in vivo osteogenesis. However, the influence of surface microand nanotopography on osteogenesis is still not well understood. High surface area porous materials are used in implants to mimic bone, which is a naturally
40
60
100 80 Time, min
120
140
160
occurring porous ceramic material composed of nano-sized organic and mineral phases that form a large macrostructure. Proteins in extracellular bone matrix and calcium phosphate, important constituents of the bone matrix, are nanostructured but the porosities in human bone are predominantly in the range 1–100 mm [15]. Spark anodization is a process which leads to the formation of a disordered oxide structure
T
T
2750
with porosities in the same regime as bone and is therefore commonly used to increase the biocompatibility of titanium and its alloys. However, recent studies using nanoporous alumina and nanotubular titania have revealed that their much smaller pores (20–200 nm) allow bone ingrowth. TNAs, in particular, have been demonstrated to be favorable templates for bone cell growth and differentiation, and also shown to significantly enhance osteoblast activity [8, 15]. The nanoporous architecture of TNAs provides a framework in which osteoblasts produce new bone. The anatase phase of TiO2 is much more efficient for nucleation and growth of bone than rutile. Another advantage of TNAs is their strong adherence to the underlying Ti substrate. In cell culture experiments, increased adhesion to TNAs of cells which produce and maintain cartilage (chondrocytes) was observed and in animal trials, chronic inflammation was found to be absent in TNAs implanted below the skin in rats. Furthermore, the calcium and phosphorous concentrations were 50% higher on these surfaces, suggesting that deposition of the bone matrix was accelerated on TNAs [8, 15]. SEM images have been used to show that the microspike extensions (filopodia) of migrating osteoblast cells go into the nanotube pores and in general progagate much faster on TNA surfaces. It has been speculated that the nanotube pore and the gap between the nanotubes serve as a useful pathway for continuous supply of ions, nutrients, and proteins, which is likely to positively contribute to the health of the growing cells [8]. Drug Eluting Coatings for Medical Implants Bacterial infection and acute inflammation at the site of the implant are the most common postsurgical complications of bone implants, dental implants, and vascular stents. To reduce the chance of complications and to prevent early stage implant failure, antibiotic therapy is indicated 6–8 weeks after surgery [15]. Systemic administration of drugs such as antibiotics to avoid or treat the infection is often ineffective at reaching infected tissues near the implant and in addition, can cause toxicity and other adverse side effects. On the other hand, localized drug delivery from the site of an implant has more efficient therapeutic effects and fewer side effects. In general, an ideal drug eluting coating must have the ability to incorporate a drug, preserve it, and deliver it gradually over the time to a specific target site [16]. TiO2 nanotube arrays are
TiO2 Nanotube Arrays: Growth and Application
being actively researched for localized drug delivery for implants because they combine the appropriate biointegration and biocompatibility with the ability to controllably release drugs (a process known as elution) confined in their channels to their surroundings. Drug loading is performed through capillary action by either immersing the templates in the concentrated drug solution or dropping the solution slowly on the template surfaces. In addition, simulated body fluid (SBF)-assisted physical adsorption of antibiotics and anti-inflammatory agents was found to be a superior method of drug loading which also resulted in improved prolonged release of both types of drugs [16]. The elution of gentamicin (a powerful aminoglycoside antibiotic used against gram-negative bacteria) from titania nanotubes has been studied [8]. In TNAs of 80 nm pore diameter and 400 nm length loaded with gentamicin (70–85% loading efficiency), drug release kinetics were found to be dependent on initial loading; and there was a sustained release, for 45, 90, and 150 min for loadings of 200, 400, and 600 mg, respectively. In addition, it was found that bacterial adhesion on the surface of gentamicin-loaded titania nanotubes was reduced significantly (in comparison with titanium surface and unloaded nanotubes), whereas normal osteoblast adhesion and proliferation is retained [8, 15]. It has also been shown that TiO2 nanotubes can control small molecule delivery in the order of weeks, and larger molecules in the order of months. The total drug loading as well as elution was mostly affected by nanotube length, and maximum small molecule elution was reached at 2 weeks. TNA-based drug eluting coatings are superior to the conventional strategy of drug elution through a drug-loaded polymer coating, because even though the polymer coating is effective in delivering a drug for long periods, it has been known to induce an inflammatory response due to eventual degradation of the polymer. Such inflammation is implicated in late-stage implant failure. TiO2 Nanotube Arrays for Stem-Cell Differentiation Recent studies on cell interactions with TNAs have demonstrated that stem cell fate is dictated by the diameter of the nanotubes. Using amorphous (unannealed) TNAs, Schmuki et al. found the adhesion, proliferation, migration, and differentiation of rat bone marrow mesenchymal stem cells to be highest
TiO2 Nanotube Arrays: Growth and Application
on nanotubes 15 nm in diameter and dramatically decreased on 70- and 100-nm nanotubes. In contrast, Jin et al. found using crystalline (annealed) TNAs and human mesenchymal stem cells that the optimum length scale for cell vitality and differentiation was 100 nm. More recent studies seem to indicate that nanotube size plays a more important role than crystallinity and nanotube chemical composition although more research is necessary to clarify the nature of topographical cues for cell behavior in this emerging field.
Applications of TiO2 Nanotube Arrays in Light Harvesting Devices One of the factors limiting maximum achievable efficiencies in conventional light harvesting devices including photovoltaic cells, photoelectrochemical cells, and photocatalytic films is the ever-present trade-off between light absorption and charge generation on the one hand, and charge separation and carrier collection on the other. This trade-off is best illustrated by considering the case of a planar p-n junction solar cell where only electron-hole pairs generated close to the depletion region are swept apart by the inbuilt electric field and efficiently collected. On the one hand, the light absorbing semiconductor layer needs to be thick enough to absorb a majority of the incident photons and on the other hand, the large thickness means that a portion of the minority charge carriers generated away from the junction are lost to recombination as they diffuse toward the junction. The vertically oriented TiO2 nanotube array architecture overcomes this trade-off by orthogonalizing the processes of light absorption/charge generation and charge separation/carrier collection so that the tradeoff is no longer as severe. The process of charge separation now occurs in the radial direction while light absorption occurs in the longitudinal direction along the axis of the nanotubes. To facilitate charge separation, the diameter of the nanotubes can be made smaller or the nanotube walls may be made thinner while simultaneously making the nanotubes longer to maximize light absorption. The porosity of the ordered structure allows the incident photons to be more effectively absorbed, due to scattering effects, than on a flat electrode. For these reasons, the vertically oriented nanotube array architecture has
2751
T
been posited as nearly ideal for light harvesting applications. TNAs for the Photocatalytic Reduction of CO2 to Methane The free energy for overall methane formation through the chemical reaction is about 801 kJ/mol, a process requiring eight photons: CO2 þ 2H2 OðgÞ ! CH4 þ 2O2
(8)
Hydrogen formation from water involves a free energy change (DG0) of 237 kJ/mol and an enthalpy change (DH0) of 285 kJ/mol; the corresponding values for CO formation from CO2 are 257 and 283 kJ/mol at 25 C (1 atm.). Hence, the minimum energy required for water and CO2 splitting processes are, respectively, 1.229 and 1.33 eV (per photon). In theory, the bandgap of a photocatalyst used for co-splitting of CO2 and water should be at least 1.33 eV, which corresponds to absorption of solar photons of wavelengths below about 930 nm. Considering the energy loss associated with entropy change (about 87 J/mol·K) and other losses involved in the CO2 splitting process (forming CO and O2), a bandgap between 2 and 2.4 eV is optimal [17]. Using nitrogen-doped TiO2 nanotube arrays, with a wall thickness low enough to facilitate effective carrier transfer to the adsorbing species, surface-loaded with nano-sized islands of platinum and copper cocatalysts, efficient sunlight-driven conversion of CO2 into CH4 and other hydrocarbons was achieved, with hydrocarbon production rates as high as 160 mL/(g h). The outstanding nature of TNAs as a platform to perform the photocatalytic reduction of CO2 is shown by the fact that the above rate of CO2 to hydrocarbon production obtained under outdoor sunlight is at least 20 times higher than previous published reports [8, 17]. TNAs for Oxidative Photochemistry Due to the large oxidizing power of photogenerated valence band holes in titania, TiO2 nanotube arrays are excellent catalysts for oxidative photochemistry. OH radicals and superoxide (O2) anions produced by photogenerated charge carriers in TiO2 can oxidize essentially all organic molecules present in the solution into carbon dioxide and water [4]. It has been shown that crystalline TiO2 nanotubes show considerably higher efficiency for the decomposition
T
T
2752
of organic small molecules than a compacted Degussa P25 layer (20–30 nm diameter nanopowder composed of anatase and rutile) under comparable conditions [4]. The TNA architecture is also exceptionally stable with no degradation observed in the samples with exposure to UV/visible light over a course of a several months. For sunlight-driven water photolysis, the titania nanotube array architecture results in a large effective surface area in close proximity with the electrolyte thus enabling diffusive transport of photogenerated holes to oxidizable species in the electrolyte. In the TNA architecture, most minority carriers generated within a “retrieval” length from the material surface, i.e., a distance from the surface equal to the sum of the depletion layer width and the diffusion length, and such carriers escape recombination and reach the electrolyte redox species. Consequently, bulk recombination is greatly reduced and the quantum yield (incident photon-to-electron current efficiency) is close to unity. Under visible light illumination, the efficiency of TNAs is limited by the large bandgap of TiO2 (3.0–3.2 eV), which renders the TNAs sensitive only to ultraviolet photons. A photoconversion efficiency for photoelectrolysis of 12.25% was obtained under UV (320–400 nm) illumination using secondgeneration nanotubes 6 mm long annealed at 580 C. High quantum yields >80% were obtained indicating that the incident UV photons were effectively utilized by the nanotube arrays for charge carrier generation. The corresponding hydrogen evolution rate was 76 mL/h W. The water-splitting reaction was confirmed by the 2:1 ratio of evolved hydrogen to oxygen. Using third-generation nanotubes 200 nm in diameter and 30 mm long formed by anodization at 35 V in a formamide-based electrolyte, the watersplitting efficiency improved to 16.25% [7]. TiO2 Nanotube Arrays as Scaffolds in Excitonic Solar Cells In excitonic solar cells, photons of energy larger than the absorption threshold of the excitonic absorber do not create free electron-hole pairs as in a conventional inorganic semiconductor. Instead excitons (electron-hole pairs bound by Coulombic attraction) are generated. Unless these excitons are dissociated at a suitable interface, they are lost to recombination. Therefore, the trade-off in excitonic solar cells is between light absorption and exciton diffusion.
TiO2 Nanotube Arrays: Growth and Application
As alluded to before, vertically oriented TNAs orthogonalize the processes of exciton generation and exciton diffusion overcoming the trade-off. Also, TNAs provide electron percolation pathways for the directed (nonrandom) movement of electrons in the structure. TNAs provide flexibility in device design since the excitonic absorbers may be coated on the nanotube walls thus exploiting the large internal surface area of the TNA architecture or the absorbers may be volumetrically filled inside the hollow cylindrical spaces of the nanotubes. Due to these reasons, TNAs are very attractive materials for use as n-type semiconducting scaffolds for excitonic solar cells. TNAs are used in two different kinds of excitonic photovoltaic devices: dye sensitized solar cells, where the absorber is a broadly absorbing dye molecule anchored to the surface of the TNAs and ordered bulk heterojunction solar cells, where the absorber is filled inside the nanotubes. Templated TNAs formed by a sol-gel method were found to have very large roughness factors which enabled them to adsorb >500 nmol cm2 of N-719. Using 17.6 mm long transparent anodic TNAs grown on conductive glass substrates, Varghese et al. obtained N-719 sensitized solar cells with an efficiency of 6.9% under AM 1.5 1 sun illumination and quantum yields higher than 65% in the spectral range 400–670 nm [9]. Using a freestanding membrane consisting of 63 mm long TNAs transferred to a FTO: glass substrate, dye-sensitized solar cells subjected to 1 sun AM.15 illumination demonstrated an open circuit photovoltage of 0.77 V, a short-circuit photocurrent of 18.5 mA cm2 and a fill factor of 0.64, resulting in an overall efficiency of 9.1% [18]. There is also increasing research interest in sensitized solar cells where the sensitizers are quantum dots of CdS or CdSe or CdTe instead of molecular dyes. Quantum yields as high as 55% have been demonstrated in quantum dotsensitized TNA solar cells [19]. An ordered double heterojunction solar cell was constructed by infiltrating a blend of regioregular poly (3-hexylthiophene) (RR-P3HT) and a methanofullerene (PCBM) into transparent anodic TNAs on FTO coated glass. The configuration used is termed a double heterojunction because of the availability of two interfaces for charge separation, namely, the TiO2/P3HT and PCBM/P3HT interfaces. The resulting solar cell had an overall conversion efficiency under 1 sun AM 1.5 illumination of 4.1% [8] (Fig. 6).
TiO2 Nanotube Arrays: Growth and Application TiO2 Nanotube Arrays: Growth and Application, Fig. 6 Schematics of TNA-based solar cells (a) Liquid junction dye-sensitized solar cells and (b) ordered bulk heterojunction solar cells (Reprinted with permission from [8])
T
2753
a Conducting glass sputtered with 0.8 nm platinum
Illumination
I−/I3− electrolyte
N-719 sensitized TiO2 nanotubes
e–
TiO2 barrier layer
Ti metal sub strate
b e– PEDOT:PSS Load Gold e– TiO2 nanotubes filled with P3HT/PCBM mixture
FTO Glass Light
Chemical Sensing Applications of TiO2 Nanotube Arrays Chemical sensors often suffer from unwanted contamination, or poisoning which limits the sensor’s operational lifetime. Volatile organic compounds are a frequent source of sensor contamination. In this respect, an important advantage of TNA-based chemical sensors is their self-cleaning ability in the presence of ultraviolet radiation [8]. The strong oxidizing power of photogenerated holes in crystalline TiO2 oxidizes organic contaminants and results in self-cleaning.
TNA-Based Ultrasensitive H2 gas Sensors One of the first applications of TNAs was in hydrogen gas sensing by measuring a change in the electrical resistance of the nanotubes upon exposure to hydrogen. A metal-semiconductor-metal (MSM) structure was employed for resistive H2 gas sensing wherein Ti metal underlying the nanotube arrays was used as the ohmic contact and a high work function metal (mainly platinum) was used as the Schottky contact. Initial work used first-generation nanotubes at high operating temperatures (e.g., 290 C) to obtain a three-order change in resistivity when exposed to a few ppm of hydrogen. By sputtering a thin overlayer of Pd on the nanotubes to
T
T
2754
enable facile dissociation of molecular hydrogen into atomic form, the operating temperature of the resistive gas sensors were brought down to room temperature. By using second-generation nanotubes anodized in aqueous electrolytes at pH 4.5 in combination with Pd sputtering, an unprecedented high sensitivity of 8.7 orders of magnitude was achieved. Room temperature operation coupled with the high sensitivities ushered in applications such as transcutaneous hydrogen sensors. The quantification of ppm-ppb hydrogen gas concentrations escaping from blood through the skin has medical relevance as an indicator of lactose intolerance, fructose malabsorption, microbial activity, bacterial growth, fibromyalgia, diabetic gastroparesis, and neonatal necrotizing enterocolitis. The prevailing explanation for the remarkable change in resistivity is that chemisorption of hydrogen atoms on the nanotube surface results in partial charge transfer to the TiO2 resulting in an electron accumulation layer on the nanotube surface which drastically improves the electrical conductance of the MSM structure. In addition, hydrogen adsorption reduces the work functions of Pd and Pt. Consequently, the height of the potential barriers at the TiO2/Pt and TiO2/Pd interfaces are lowered, thus again increasing the conductance of the MSM structure. TNA-Based H2O2 Sensors TNAs on which horseradish peroxidise (HRP) and thionine were co-immobilized have been used as electrochemical sensors to detect and quantify hydrogen peroxide [8]. Due to the high electrocatalytic activity of the immobilized HRP for thionine oxidation in the presence of H2O2, a large increase of reduction current occurs for the Th/HRP/TiO2 electrode due to peroxide. The detection limit for H2O2 was estimated to be as low as 1.2 mM. TNA-Based Amine Sensors Amines are one of several chemical pollutants released in oil and gas spills which contaminate fresh and seawater. In the light of the fact that many amines are proven or suspected to be carcinogenic and have been implicated in inducing bladder cancer, environmental monitoring of amine levels assumes significance. In this regard, TNAs coated with an electrostatically adsorbed layer of ruthenium tris(bipyridinium) ions [Ru(bpy)32+] have been used in an electrochemiluminescence sensor to detect and quantify the tripropylamine concentrations in water as low as 1 nM [20].
TiO2 Nanotube Arrays: Growth and Application
Cross-References ▶ Active Carbon Nanotube-Polymer Composites ▶ Atomic Layer Deposition ▶ Biosensors ▶ Hybrid Solar Cells ▶ Microfabricated Probe Technology ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanostructured Materials for Sensing
References 1. Zwilling, V., Aucouturier, M., Darque-Ceretti, E.: Anodic oxidation of titanium and TA6V alloy in chromic media. An electrochemical approach. Electrochim. Acta 45(6), 921–929 (1999) 2. Gong, D., Grimes, C.A., Varghese, O.K., Hu, W.C., Singh, R.S., Chen, Z., Dickey, E.C.: Titanium oxide nanotube arrays prepared by anodic oxidation. J. Mater. Res. 16(12), 3331–3334 (2001) 3. Mor, G.K., Varghese, O.K., Paulose, M., Shankar, K., Grimes, C.A.: A review on highly ordered, vertically oriented TiO2 nanotube arrays: fabrication, material properties, and solar energy applications. Sol. Energy Mater. Sol. Cells 90(14), 2011–2075 (2006). doi:10.1016/ j.solmat.2006.04.007 4. Macak, J.M., Tsuchiya, H., Ghicov, A., Yasuda, K., Hahn, R., Bauer, S., Schmuki, P.: TiO2 Nanotubes: self-organized electrochemical formation, properties and applications. Curr. Opin. Solid State Mat. Sci. 11(1–2), 3–18 (2007). doi:10.1016/j.cossms.2007.08.004 5. Hoyer, P., Formation of a titanium dioxide nanotube array. Langmuir 12(6), 1411–1413 (1996) 6. Michailowski, A., AlMawlawi, D., Cheng, G.S., Moskovits, M.: Highly regular anatase nanotubule arrays fabricated in porous anodic templates. Chem. Phys. Lett. 349(1–2), 1–5 (2001) 7. Shankar, K., Basham, J.I., Allam, N.K., Varghese, O.K., Mor, G.K., Feng, X.J., Paulose, M., Seabold, J.A., Choi, K.S., Grimes, C.A.: Recent advances in the Use of TiO2 nanotube and nanowire arrays for oxidative photoelectrochemistry. J. Phys. Chem. C 113(16), 6327–6359 (2009). doi:10.1021/ jp809385x 8. Grimes, C.A., Mor, G.K.: TiO2 Nanotube Arrays: Synthesis, Properties, and Applications (Springer, New York, 2009) 9. Varghese, O.K., Paulose, M., Grimes, C.A.: Long vertically aligned titania nanotubes on transparent conducting oxide for highly efficient solar cells. Nat. Nanotechnol. 4(9), 592–597 (2009). doi:http://www.nature.com/nnano/journal/ v4/n9/suppinfo/nnano.2009.226_S1.html 10. Wu, X.J., Zhu, F., Mu, C., Liang, Y.Q., Xu, L.F., Chen, Q.W., Chen, R.Z., Xu, D.S.: Electrochemical synthesis and applications of oriented and hierarchically quasi-1D semiconducting nanostructures. Coord. Chem. Rev. 254(9–10), 1135–1150 (2010). doi:10.1016/j.ccr.2010.02.014
Toward Bioreplicated Texturing of Solar-Cell Surfaces 11. Masuda, H.: Highly ordered nanohole arrays in anodic porous alumina. In: Wehrspohn, R.B. (ed.) Ordered Porous Nanostructures and Applications, pp. 37-56. Springer, New York (2005) 12. Kemell, M., Pore, V., Tupala, J., Ritala, M., Leskela, M., Atomic layer deposition of nanostructured TiO2 photocatalysts via template approach. Chem. Mat. 19(7), 1816–1820 (2007). doi:10.1021/cm062576e 13. Ren, X., Gershon, T., Iza, D.C., Munoz-Rojas, D., Musselman, K., MacManus-Driscoll, J.L.: The selective fabrication of large-area highly ordered TiO2 nanorod and nanotube arrays on conductive transparent substrates via sol–gel electrophoresis. Nanotechnology 20(36) (2009). doi:365604 10.1088/0957-4484/20/36/365604 14. Mun, K.S., Alvarez, S.D., Choi, W.Y., Sailor, M.J.: A stable, label-free optical interferometric biosensor based on TiO2 nanotube arrays. ACS Nano 4(4), 2070–2076 (2010). doi:10.1021/nn901312f 15. Losic, D., Simovic, S.: Self-ordered nanopore and nanotube platforms for drug delivery applications. Expert Opin. Drug Deliv. 6(12), 1363–1381 (2009). doi:10.1517/ 17425240903300857 16. Gultepe, E., Nagesha, D., Sridhar, S., Amiji, M.: Nanoporous inorganic membranes or coatings for sustained drug delivery in implantable devices. Adv. Drug Deliv. Rev. 62(3), 305–315 (2010). doi:10.1016/j.addr.2009.11.003 17. Roy, S.C., Varghese, O.K., Paulose, M., Grimes, C.A.: Toward solar fuels: photocatalytic conversion of carbon dioxide to hydrocarbons. ACS Nano 4(3), 1259–1278 (2010). doi:10.1021/nn9015423 18. Lin, C.J., Yu, W.Y., Chien, S.H.: Transparent electrodes of ordered opened-end TiO2-nanotube arrays for highly efficient dye-sensitized solar cells. J. Mater. Chem. 20(6), 1073–1077 (2010). doi:10.1039/b917886d 19. Baker, D.R., Kamat, P.V.: Photosensitization of TiO2 nanostructures with CdS quantum dots: particulate versus tubular support architectures. Adv. Funct. Mater. 19(5), 805–811 (2009). doi:10.1002/adfm.200801173 20. Xu, Z.H., Yu, J.G.: A novel solid-state electrochemiluminescence sensor based on Ru(bpy)(3)(2+) immobilization on TiO2 nanotube arrays and its application for detection of amines in water. Nanotechnology 21(24) (2010). doi:24550110.1088/0957-4484/21/24/245501
Titanium Dioxide ▶ Fate of Manufactured Nanoparticles in Aqueous Environment
Total Internal Reflection (Fluorescence) Velocimetry ▶ Nano(Evanescent-Wave)-Particle Velocimetry
2755
T
Toward Bioreplicated Texturing of SolarCell Anodes ▶ Toward Bioreplicated Texturing of Solar-Cell Surfaces
Toward Bioreplicated Texturing of Solar-Cell Surfaces Aditi Risbud1, Akhlesh Lakhtakia2 and Michael H. Bartl3 1 Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, CA, USA 2 Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA, USA 3 Department of Chemistry, University of Utah, Salt Lake City, UT, USA
Synonyms Toward bioreplicated texturing of solar-cell anodes
Definition Biological species have developed myriad photonic structures to efficiently interact with light. The results of these interactions include large angular fields of view, reduced surface reflection, Bragg diffraction, and multiple scattering. Bioreplicated texturing of solar cell surfaces and photoanodes takes advantage of these excellent optical properties in biology. The aim is to convert biological structures into exact replicas made out of semiconductors, metals, and polymers and incorporate these replicas into solar cells to enhance light harvesting and light–matter interactions.
Introduction The development of efficient, inexpensive, and environmentally benign energy sources is one of the biggest technological challenges of the twenty-first
T
T
2756
century. Even the most conservative predictions suggest worldwide demand for energy will triple within the next 40 years [1]. This rapidly growing energy demand, coupled with economic and ecologic impacts of fossil fuel–based energy sources, has sparked research into low-cost and environmentfriendly alternatives. Given the immense amount of energy the sun delivers to earth daily, solar power constitutes by far the most promising and attractive alternative to fossil fuels. However, to become the primary energy source for future consumption, highly efficient yet inexpensive strategies for direct conversion of solar power into thermal energy, electricity and chemical fuels are needed. A central challenge in solar energy conversion and photocatalysis is to optimize “light management” at the nanometer scale – from capture of sunlight to transfer and conversion of solar light into energy, fuels, or drivers of chemical reactions. A prime example of highly evolved light management is photosynthesis – one of the most fundamental processes in biology. Calculations show photosynthesis produces approximately 90 TW of energy annually, approximately six times the current worldwide total energy consumption per year [1]. In addition, the light-independent phase of photosynthesis fixes carbon dioxide from the atmosphere, providing a potential mechanism for carbon dioxide sequestration. Photosynthetic organisms capture sunlight and channel solar energy from light-harvesting outer units to a reaction center by transferring energy along multiple functional units and converting it to fuel. It is this elaborate, integrated energy conversion system consisting of controlled and cooperative interaction between functional nanoscale components that results in efficient capture, transfer and use of solar energy. In contrast, artificial solar conversion units are still far from achieving the sophistication of their counterparts in nature. The goal is thus to examine simple models of biological systems, to learn from nature’s complex energy conversion systems and to unlock and mimic its mechanisms [2–4]. For example, in biological materials, such as the wings of a butterfly or the exoskeleton of an insect, chitin and air are organized in ordered structures to produce optical interference leading to structural colors, camouflage and the control of solar radiative heat intake for thermal regulation. An early study calculated the effects of total solar radiation absorption
Toward Bioreplicated Texturing of Solar-Cell Surfaces
and emissivity in the forms of beam radiation, which travels directly from the sun to a surface; diffuse radiation, which is a uniform distribution of light on a surface from all possible angles; and reflected radiation, which is light reflected from the ground on an insect thin film [5]. The outcome of this study suggests minute changes in the thickness of these biological thin films, typically 100 nm thick, strongly influence intake of energy from solar radiation. For example, changes in butterfly wing scale film thickness could significantly shift absorption levels (by up to 25%) over a span of 80 nm, while diffuse solar absorption and emissivity remain relatively constant. Furthermore, iridescence has been found to be strongly correlated with higher absorption of the solar radiation [6]. In this entry, several novel concepts inspired by biological systems are introduced to enhance the first steps in solar energy conversion: the capture of sunlight through bioreplicated surfaces and light–matter interactions driving the conversion process. In the first section, strategies to increase the light-harvesting capability of conventional solar cells are discussed. These strategies are based on the excellent optical properties of insect eyes. In these structures, called compound eyes, intricate microscale and nanoscale features result in significantly enhanced angular fields of view, while minimizing surface reflectivity. Since these parameters are of utmost importance for optimizing light harvesting in conventional solid-state junction solar cells, researchers have recently started to develop methods for mass-replicating these biological structures as polymer films for deposition onto the surface of solar cells [7, 8]. In the second section, focus will be directed on methods to increase light– matter interactions in a new type of solar cell, the Gr€atzel-type electrochemical cell [9]. The heart of this low-cost alternative to solid-state junction devices is a nanocrystalline titania photoanode sensitized with visible-light absorbing dyes. Interestingly, coupling this photoanode to photonic crystal structures can increase its optical path, and therefore results in higher solar conversion efficiency [10, 11]. Since some of the most effective photonic crystal structures are found in biological systems, such as colored beetles and butterflies [12], researchers are devising ways to replicate these remarkable architectures into nanocrystalline titania [13]. These efforts
Toward Bioreplicated Texturing of Solar-Cell Surfaces
are reviewed and several examples of attempts to generate bioreplicated photoanodes are given.
Surface Structuring of Solar Cells A dominant factor in the performance of any solar conversion device is the amount of incident sunlight coupled into the active module of a given solar cell. The key considerations for efficient coupling of light into solar cells are to maximize the angular field of view of the solar absorption module and to minimize the overall reflection of light at the cell’s surface. Increasing the angular field of view aims to harvest as much light as possible, in terms of direct illumination from the sun, as well as diffuse light. Harvesting the direct illumination of sunlight is highest when it hits the solar cell surface at a 90 angle. While, in principle, this can be achieved simply by installing a mechanical device that rotates the solar cell surface in response to the sun’s position, such motion devices increase the cost of solar cells, require maintenance and use up part of the cell’s energy. In addition, these motion devices would only minimally increase harvesting of diffuse light – scattered and reflected light that hits the solar cell from any possible angle. During the course of millennia, biological structures have evolved to capture light from a wide range of directions. Indeed, houseflies have optimized their angular field of view through eye placement and underlying structure such that these insects can see 270 around them from a horizontal plane [2]. This remarkable ability stems from a compound eye structure, consisting of an array of microscale cylindrical lenses arranged on a curved surface. These tiny lenses not only capture light along the axis of the cylinder, but also can absorb light propagating in other directions through the cylinder’s sidewalls. In addition, compound eyes can have nanometer-scale surface structures, which serve both as an antireflective [7] and a superhydrophobic surface [14]. These properties are the result of a gradual change in refractive index between the air and the eye. While superhydrophobicity keeps the eye clean, reduced reflection from insect eyes protects them from being visible to predators, and it also significantly enhances their photosensitivity in dim environments. As a result, these surfaces are an important enhancement strategy that could be used to boost efficiency in a solar cell.
2757
T
Although these biological structures have complex nano-to-microscale features, recent advances in engineered biomimicry now allow scientists to replicate these structures using materials optimal for harvesting light for solar cells. Both physical and chemical nanofabrication techniques such as atomic layer deposition, conformal-evaporated-film-byrotation technique, chemical vapor deposition, soft lithography and nanocasting are now employed by researchers to generate replicas of various biological structures in technologically relevant materials. For example, a research group developed a soft lithographic patterning technique to replicate a nanometer-scale antireflective coating found on the lens of a moth’s compound eye [7]. A three-dimensional mold of this coating was designed with perfluoropolyether, an elastomeric polymer material capable of capturing nanometer-scale surface features. This mold was then used to create polyurethane replicas with antireflective behavior. As shown in Fig. 1, this approach results in high-fidelity replicas preserving topographical features as small as 100 nm. In addition, the group successfully used this technique to emboss antireflective nanoscale patterns in a widely used polymer blend for the photoactive region of an organic photovoltaic cell. In an attempt to scale up compound eye bioreplication methods to mass-produce antireflective/ large-field-of-view solar cell components, researchers developed a method to create reusable master negatives [8]. Specifically, corneas of the common blowfly were used as templates and replicated into metallic nickel. To achieve both a high-fidelity imprint of the hierarchically organized cornea surface and a stable and reusable master negative, the replication process was conducted in two steps. First, a modified version of the conformalevaporated-film-by-rotation method was used to deposit a thin layer of nickel onto the surface relief of the cornea sample. In a second step, additional nickel was deposited by electroforming to strengthen the initial thin film replica. Finally, after removal of the biotemplate, a freestanding negative copy of the original cornea was obtained (Fig. 2). This group also showed this negative master could be successfully used (and reused) to create polymer replicas by a casting technique to faithfully reproduce micrometer-sized surface features. In addition, it was suggested by using stamping instead of casting, nanometer-sized features could be transferred into mass-produced polymer replicas.
T
T
2758
Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 1 Scanning electron micrographs of (a) the Attacus atlas moth eye structure, (b) the negative polymer replica, (c) the positive replica made by polymerization in the negative mold. Inset of (c) shows curved surface of the replica. Scale bars are 100 mm for the top images (inset: 20 mm), 5 mm for the second row images, and 500 nm for the third and fourth row images. Reproduced from [7] with permission of The Royal Society of Chemistry http:// dx.doi.org/10.1039/ C1SM05302G
Toward Bioreplicated Texturing of Solar-Cell Surfaces
a
b
c
Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 2 Photographs of (left) a blowfly’s cornea and (right) a nickel replica after removal of the biotemplate. The nickel structure is a positive replica when viewed from above (as
shown), but a negative replica when viewed from below. Reproduced with permission from [8]. Copyright 2010 IOP Publishing Ltd.
Biotemplating of Photoanodes
nanostructured semiconductors, conducting polymers and various hybrid device architectures. For example, theoretical studies recently showed coupling the photoanode of a Gr€atzel-type solar cell to a photonic crystal could significantly enhance its efficiency [10, 11]. Photonic crystals are periodically ordered structures with lattice parameters comparable to the optical
Along with light-harvesting techniques, engineered biomimicry is a promising approach to enhance light–matter interactions in solar conversion systems. This is a particularly important area of research in next-generation, low-cost devices based on
Toward Bioreplicated Texturing of Solar-Cell Surfaces Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 3 Photographs of (a) the weevil L. augustus and (b) individual scales attached to its exoskeleton. Cross-sectional scanning electron microscopy images of (c) a single scale and (d) a region of a scale at higher magnification. Reproduced with permission from [16]. Copyright 2008, American Physical Society
T
2759
a
b
100 mm
c 1 cm
10 mm
d
2 mm
wavelength of interest [15]. Due to this periodic variation in the refractive index, light within a photonic crystal experiences a combination of Bragg diffraction and multiple scattering events. As a consequence, light of certain frequencies undergoes multiple reflections as it slowly percolates through this structure. This results in a greatly increased nominal optical path length, an increase in light–photoanode interactions and significantly enhances the probability of light-tocharge conversion. Interestingly, millions of years before we “discovered” photonic crystals, biological systems created such structures to produce vibrant colors, control solar absorption and enhance bio-fluorescence for a variety of reasons, ranging from camouflaging to frightening predators and attracting mates [12, 13]. Biological photonic structures thus were optimized to produce a wide range of optical effects and function under various illumination conditions. Not surprisingly, the range of photonic structures found in biological species is virtually limitless, ranging from simple one-dimensional multilayers, to two-dimensional lattices found in skeletons and hairs of sea animals and bird feathers and to various three-dimensional architectures within colored cuticle scales of many species of beetles and butterflies. For example, scientists discovered certain beetles obtain their color from diamond-based photonic crystals – one of the
most effective lattice structures [16]. What’s more, such photonic structures (as an example, Fig. 3 shows the green Brazilian beetle Lamprocyphus augustus) are still beyond our engineering capabilities. Similarly, photonic structures found in many butterfly wing scales have an astonishing hierarchy, consisting of ridges, lamellae and ribs, as well as highly periodic honeycomb and gyroid structures [12]. These elaborate hierarchical-scale structures act as optical components interacting with sunlight, such as multilayer reflectors, diffraction gratings and photonic crystals. Only recently have scientists begun to use this invaluable portfolio of biological photonic crystal lattices for engineered biomimicry approaches. The method of choice is bioreplication: here, a biological photonic structure is used as a template to create positive or negative copies out of photocatalytically active compounds. An example of this approach is shown in Fig. 4 for conversion of the diamondbased photonic structure of the beetle L. augustus. Replication is achieved by a sol–gel double-imprint method using a sacrificial silica glass intermediary [17]. Sol–gel templating is quite attractive due to its simplicity and low cost. The original biological templates were backfilled with liquid precursors of the desired inorganic compound (e.g., silica, titania, alumina). After solidification of the inorganic compound, the original template was selectively removed
T
T
2760
Toward Bioreplicated Texturing of Solar-Cell Surfaces
Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 4 Sol–gel bioreplication route of photonic weevil scales into titania replicas (top left). Scanning electron microscopy images of the original biopolymeric photonic structure (top right), the negative replica made of silica (bottom right), and the positive replica made of titania (bottom left). Scale bars are 200 mm (top left) and 1 mm (all others). Reproduced with permission from [17]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA
Sol-gel silica templating
Sol-gel titania templating
by etching or heat treatment. This approach enabled a positive copy (“inverse-of-the-inverse”) of the original structure composed of photocatalytically active nanocrystalline titania. Along with sol–gel chemistry, other physical and chemical deposition methods can be used for replication of biological structures – as long as they operate under mild deposition conditions and temperatures. Successful methods include atomic layer deposition [18] and conformal-evaporated-film-by-rotation techniques [19]. In general, these techniques can be applied under noncorrosive reaction environments, mild pH conditions, and relatively low deposition temperatures (below 100–200 C). Since these replicas are formed by deposition from a gas phase, these methods are highly efficient in infiltrating even complex threedimensional frameworks as long as the internal structure is fully accessible. Furthermore, the deposition parameters can readily be controlled, enabling the amount of deposited material to be carefully controlled. As shown in Fig. 5, control of atomic layer deposition can be used to create shell-like replicas of the original biological structures with finely tuned replica shell thickness and, as a result, well-controlled optical properties. Recently, a group of scientists took such bioreplication approaches one step closer to solar conversion applications by devising strategies to construct
a Gr€atzel-type titania photoanode coupled to arrays of replicated butterfly scales [20]. In this approach, 1 1 cm2 samples of butterfly wings were soaked in a titania sol–gel precursor and arranged on top of a Gr€atzeltype photoanode (Fig. 6). This composite structure was then heat-treated to allow the top butterfly-wing replica layer to bond to the bottom titania layer. This heat treatment (500 C in air) also induced titania crystallization into polycrystalline anatase and removed the biopolymeric template. Despite structural shrinkage during the heat-based framework densificationcrystallization treatment, these replicated materials displayed improved light absorption behavior as compared to anodes without the replicated butterfly wings. While this outcome is an important step toward bioreplicated solar cells, these researchers also note major challenges remaining in generating such biotemplated anodes, optimizing their electrical conductivity and integrating these structures into functional solar conversion devices.
Concluding Remarks Engineered biomimicry has arrived as a serious contender in materials research and engineering. While a few years ago the concepts of super-vision from bug eyes, ultrastrong fabrics from spiders, vertically
Toward Bioreplicated Texturing of Solar-Cell Surfaces
2761
T
Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 5 (a) Optical microscope image of the alumina-coated Morpho peleides butterfly wing scales; color changes from original blue to pink. (b) Higher magnification scanning electron microscopy image of alumina replicas of butterfly wing scales and two broken rib tips (inset). Reproduced with permission from [18]. Copyright 2006 American Chemical Society
Soaked wing
Dipping Ti(SO4)2 Precursor Copy Glass substrates
Calcination F:SnO2
Anatase film
Toward Bioreplicated Texturing of Solar-Cell Surfaces, Fig. 6 Schematic illustration of creating bioreplicated titania photoelectrodes. Step 1, soaking the butterfly wings in the titania
precursor; step 2, calcination of soaked butterfly wings on top of an anatase titania film. Reproduced with permission from [20]. Copyright 2009 American Chemical Society
climbing gecko-like robots, or optical devices from beetles was more the stuff of Hollywood science fiction, these ideas have found their way into research laboratories and inspire the development of new devices in electronics, optics and mechanics. Indeed, although the pure beauty of natural iridescence in the form of opal gem stones, jewel beetles and feather ornaments has fascinated and inspired mankind for thousands of years, today, these materials and optical concepts offer revolutionary new ways to manipulate light. In addition, these materials hold great potential for more efficient solar cells and photocatalysts, new optical reflectors and sensors and ultrafast switches.
An exciting area in engineered biomimicry is bioreplication, in which complex biological structures are converted into replicas made of semiconductors, metals or polymers. In this entry, a few examples of solar-related bioreplication strategies to increase light-harvesting ability and light–matter interaction efficiency in various types of solar cells were discussed. While many of these approaches are promising and first successful steps have been achieved, it should be emphasized there are also several limitations and challenges in bioreplication. For example, biological samples have an inherent size limit given by the dimensions of the eyes,
T
T
2762
scales, or feathers used as templates. Direct use and integration of biological samples, therefore, have limited applications, suggesting more efficient massduplication routes have to be developed. In addition, most natural structures have a much higher defect density than what is tolerated by current engineering standards, and no two natural structures are exactly the same. For example, the structural parameters of colored scales taken from different parts of a beetle, or even within a single scale, can vary by as much as 20%. Consequently, bioreplication must be viewed in its correct and just perspective: Although materials prepared by bioreplication will perhaps never achieve the precision, uniformity and reproducibility of their counterparts engineered by micro- or nano-fabrication methods, bioreplication opens the door to the full portfolio of highly complex hierarchical structures found in nature – structures still far from our engineering capabilities. It will therefore be of vital importance to devise strategies to use and integrate materials created by engineered biomimicry with micro- or nano-fabrication methods and combine the appealing aspects of two worlds – biology and materials science – in a concerted approach. Engineered biomimicry has found its way from science fiction into the laboratories; whether it makes the next step from fundamental research to wide-scale applications will depend on scientific creativity and imagination.
Cross-References ▶ Biomimetics of Optical Nanostructures ▶ BioPatterning ▶ Hybrid Solar Cells ▶ Moth-Eye Antireflective Structures ▶ Nanomaterials for Excitonic Solar Cells ▶ Nanostructures for Energy ▶ Nanostructures for Photonics ▶ Structural Color in Animals
References 1. Hoffert, M.I., Caldeira, K., Jain, A.K., Haites, E.F., Harvey, L.D.D., Potter, S.D., Schlesinger, M.E., Schneider, S.H., Watts, R.G., Wigley, T.M.L., Wuebbles, D.J.: Energy implications of future stabilization of atmospheric CO2 content. Nature 395, 881 (1998)
Toward Bioreplicated Texturing of Solar-Cell Surfaces 2. Chiadini, F., Fiumara, V., Scaglione, A., Pulsifer, D.P., Martı´n-Palma, R.J., Pantano, C.G., Lakhtakia, A.: Insect eyes inspire improved solar cells. Optics Photonics News 22(4), 38 (2011) 3. Paris, O., Burgert, I., Fratzl, P.: Biomimetics and biotemplating of natural materials. MRS Bull. 35, 219 (2010) 4. Jin, L., Zhai, J., Heng, L., Wei, T., Wen, L., Jiang, L., Zhao, X., Zhang, X.: Bio-inspired multi-scale structures in dye-sensitized solar cell. J. Photochem. Photobiol. C 10, 149 (2009) 5. Heilman, B.D., Miaoulis, I.N.: Insect thin film as solar collectors. Appl. Optics 33, 6642 (1994) 6. Bosi, S.G., Hayes, J., Large, M.C.J., Poladian, L.: Color, iridescence, and thermoregulation in Lepidoptera. Appl. Opt. 47, 5235 (2008) 7. Ko, D.-H., Tumbleston, J.R., Henderson, K.J., Euliss, L.E., DeSimone, J.M., Lopez, R., Samulski, E.T.: Biomimetic microlens array with antireflective “moth-eye” surface. Soft Matter 7, 6404 (2011). doi:10.1039/C1SM05302G 8. Pulsifer, D.P., Lakhtakia, A., Martı´n-Palma, R.J., Pantano, C.G.: Mass fabrication technique for polymeric replicas of arrays of insect corneas. Bioinsp. Biomim. 5, 036001 (2010) 9. Gr€atzel, M.: Photoelectrochemical cells. Nature 414, 338 (2001) 10. Mihi, A., Miguez, H.: Origin of light-harvesting enhancement in colloidal-photonic-crystal-based dye-sensitized solar cells. J. Phys. Chem. B 109, 15968 (2005) 11. Halaoui, L.I., Abrams, N.M., Mallouk, T.E.: Increasing the conversion efficiency of dye-sensitized TiO2 photoelectrochemical cells by coupling to photonic crystals. J. Phys. Chem. B 109, 6334 (2005) 12. Srinivasarao, M.: Nano-optics in the biological world: Beetles, butterflies, birds, and moths. Chem. Rev. 99, 1935 (1999) 13. Jorgensen, M.R., Bartl, M.H.: Biotemplating routes to threedimensional photonic crystals. J. Mater. Chem. 21, 10583 (2011). doi:10.1039/C1JM11037C 14. Koch, K., Bhushan, B., Barthlott, W.: Diversity of structure, morphology and wetting of plant surfaces. Soft Matter 4, 1943 (2008) 15. Joannopoulos, J.D., Villeneuve, P.R., Fan, S.H.: Photonic crystals: Putting a new twist on light. Nature 386, 143 (1997) 16. Galusha, J.W., Richey, L.R., Gardner, J.S., Cha, J.N., Bartl, M.H.: Discovery of a diamond-based photonic crystal structure in beetle scales. Phys. Rev. E 77, 050904 (2008) 17. Galusha, J.W., Jorgensen, M.R., Bartl, M.H.: Diamondstructured titania photonic bandgap crystals from biological templates. Adv. Mater. 22, 107 (2010) 18. Huang, J., Wang, X., Wang, Z.L.: Controlled replication of butterfly wings for achieving tunable photonic properties. Nano Lett. 6, 2325 (2006) 19. Martı´n-Palma, R.J., Pantano, C.G., Lakhtakia, A.: Biomimetization of butterfly wings by the conformal-evaporated-film-by-rotation technique for photonics. Appl. Phys. Lett. 93, 083901 (2008) 20. Zhang, W., Zhang, D., Fan, T., Gu, J., Ding, J., Wang, H., Guo, Q., Ogawa, H.: Novel photoanode structure templated from butterfly wing scales. Chem. Mater. 21, 33 (2009)
Toxicology: Plants and Nanoparticles
Toxicity ▶ In Vitro and In Vivo Toxicity of Silver Nanoparticles
Toxicity of Metal and Metal Oxide Nanoparticles on Prokaryotes (Bacteria), Unicellular and Invertebrate Eukaryotes ▶ Ecotoxicity of Inorganic Nanoparticles: From Unicellular Organisms to Invertebrates
Toxicity/Ecotoxicity ▶ Exposure and Toxicity of Metal and Oxide Nanoparticles to Earthworms
Toxicology: Plants and Nanoparticles Marie Carrie`re and Camille Larue Laboratoire Le´sions des Acides Nucle´iques, Commissariat a` l’Energie Atomique, SCIB, UMR-E 3 CEA/UJF-Grenoble 1, INAC, Grenoble, Cedex, France
Synonyms Ecotoxicity; Nanoparticle; Phytotoxicity
Definition Impact of engineered nanoparticles (ENP) on plant growth, development, physiology, and morphology.
Key Research Findings Three research areas are currently being investigated to define engineered nanoparticle (ENP) impacts on plant development (seed germination, root and shoot
2763
T
elongation, etc.), on plant physiology, and ENP uptake and translocation through plants. Impact on Plant Development In this research area, regulatory tests are used, such as the ones described in EPA guidelines [1]. These guidelines have been developed for toxicity testing of chemical pollutants, and their relevance for toxicity testing of ENPs is under debate, with several groups working on their implementation to render them accurate. Among these tests, seed germination, root and shoots elongation assays have been intensively used to assess ENP impact on plant development, and the obtained results are either positive, or negative, or inconsequential. For instance multi-walled carbon nanotubes (CNT) have been shown to cause plant biomass decrease, particularly on Cucurbita pepo [2], while they showed no impact on rye grass, lettuce, rape, radish, corn, and cucumber [3]. The same is true for TiO2 nanoparticles, which were shown in some studies to alter leaf growth and transpiration on maize [4], while they did not cause any alteration in the growth of willow tree plantlets [5]. Described results are difficult to compare, since model plant species are often different, and the considered nanoparticles are often different as well: Multiwalled CNTs may differ in their production process, post-synthesis surface treatment, and in the preparation of dispersed suspensions. TiO2 nanoparticles may differ in their primary diameter, specific surface area, shape (round, elongated, see Fig. 1), crystalline structure, etc. These conflicting effects are easily explained by the broad panel of tested ENPs, and their very different physicochemical properties. Moreover indirect effects such as ENP dissolution or impact of the solvent (or other environmental pollutants) which may be adsorbed on ENP surface have to be considered (Fig. 2). The best examples of the impact of ENP dissolution on plant development was demonstrated with Cu and ZnO nanoparticles. Even if the impact of dissolved ions did not explain all the toxic effects observed in the following studies, they greatly contributed to them. Indeed, Cu NPs have been demonstrated to cause inhibition of plant growth and to accumulate in mung beans and wheat [6], but also have some influence on the germination of lettuce [7]. Among all the ENPs tested for their impact on rye grass, lettuce, rape, radish, corn, and cucumber, only Zn and ZnO NPs
T
T
2764
Toxicology: Plants and Nanoparticles, Fig. 1 Titanium dioxide nanoparticles, rutile and elongated-shaped (a) or anatase and round-shaped (b)
Toxicology: Plants and Nanoparticles b
a
100 nm
50 nm
l+ / –
Toxicology: Plants and Nanoparticles, Fig. 2 Schematic representation of potential ENP evolution in the environment: ENPs may partially dissolve, releasing free ions (I+/). They may also adsorb environmental pollutants or solvent on their surface
ss l di
Su
reduced germination of ryegrass and corn, respectively. These two ENPs also reduced root elongation [3]. ZnO-NP dissolution was then considered by the same team the year after [8]. Uptake and Translocation Despite the relative impermeability of plant roots to particulate matter, due to their small size, ENPs have been demonstrated to penetrate plant root wall, even if most of the ENPs remain adsorbed on root surface. It has for instance been demonstrated for Fe3O4 NPs on pumpkins [9], for TiO2-alizarin red nanoconjugates on Arabidopsis thaliana [10], and CeO2 nanoparticles on alfalfa, cucumber, corn, and tomato [11]. The consequence of such result is that ENPs may then be transferred to the food chain. Note that the ENP content in these plants is very low: ENP uptake is not an efficient process; consequently the amount of ENPs reaching the food chain would also be low. Several techniques are available to demonstrate ENP uptake by plants. First, electron microscopy, due to its very high spatial resolution (typically 1 nm or below), is the method of choice to identify the presence of ENPs in biological samples, and particularly in
rfac
ea
dso
rpti
on
l+ / –
l+ / – + / – l
or l+ / –
ia
l+ / –
l+ / –
tion
olu
rt Pa
l+ / –
l+ / –
l+ / – l+ / –
l+ / –
or
plants. Scanning electron microscopy (SEM) permits to observe ENPs on the surface of plant tissues (roots or leaves), but does not give access to the internalized ENPs. Conversely transmission electron microscopy (TEM), achieved on dehydrated plant samples included in a resin and cut in ultrathin sections, permits the detection of nanosized materials in plant cells or extracellular compartments. TEM has been successfully used to prove that carbon-based ENPs were accumulated in rice plantlets [12]. Coupling these two techniques with energy-dispersive spectroscopy (EDS) allows a local chemical identification of the electrondense spots observed by TEM or SEM in plant samples, proving that they contain the chemical element constituting the ENP (when EDS analysis is performed, then spatial resolution ranges from 30 to 50 nm). Another valuable technique is micro-X-rayfluorescence (m-XRF), achieved either on a laboratory source or using synchrotron radiation. This technique allows local identification of chemical composition, and basically the identification of hot spots of one particular element constituting the ENP. With this technique, spatial resolution can be as high as 100 nm on some synchrotron beam lines such as ID22
Toxicology: Plants and Nanoparticles
(ESRF, Grenoble, France). On some synchrotron beam lines, this technique is coupled to X-ray absorption spectroscopy (XAS), which allows describing the local environment of the analyzed element, giving the absolute proof that the element is in the nanoparticulate form, and not present as dissolved ions. Moreover it informs on ENP crystal structure, and sometimes on their size (it is the case for TiO2NPs). For instance this technique has been used to evidence that alfalfa, cucumber, corn, and tomato plants accumulated CeO2-NPs in their roots [11]. Regarding the detection of CNTs, TEM enables the observation of single or clusters of NTCs. However, images obtained by TEM can be misinterpreted, since the shape of CNT is close to the structure of natural vegetal compounds. In order to avoid any misinterpretation, very specific techniques such as photothermal and photoacoustic detection have been developed and used for the identification of their accumulation and translocation in tomato plantlets [13]. These techniques are highly sensitive and specific, allowing detecting single MWCNT in plant tissues. Raman spectroscopy is also a very specific technique, which has been used to identify the presence of CNTs inside plant seeds thanks to their G band (1568 cm1) [14]. Another point is that, when taken up in plant roots, ENPs are in some cases transferred from the roots to the shoots (Fig. 3). This phenomenon, also called translocation, was demonstrated for instance with Au-NPs in tobacco [15] and Fe3O4 NPs on pumpkins [9]. Carbon-based NPs have also been demonstrated to translocate from the roots to the shoots and from the shoots to the seeds. The newly developed plantlets, resulting from the germination of these contaminated seeds, also contained carbon-based NPs, meaning that ENPs were transmitted to the next generation [12]. Very recently, thanks to these techniques, gold NPs were shown to be transferred to higher trophic levels: hornworms fed with Au-NP-exposed tobacco were shown to be contaminated with Au-NPs [15]. In this particular case, gold maximal concentration in tobacco reached less than 100 mg Au kg1 dry tobacco. Impact on Plant Physiology The impact of ENP exposure on plant physiology and metabolism is poorly described, underlining that the area of research of ENP phytotoxicity is at its infancy. A very recent study describes the impact of NTC on plant gene expression [13]. Many stress-related genes
2765
T
Root-to-shoot translocation
Adsorption on root surface and root uptake
Toxicology: Plants and Nanoparticles, Fig. 3 Schematic representation of ENP adsorption on root surface, root uptake, and transfer to the shoot
are activated upon exposure to NTC, and among them the gene encoding the protein constituting water channels (aquaporins) [13]. This channel regulates water permeability, and therefore is significantly implicated in seed germination and plant growth. In one study TiO2-NPs have also been demonstrated to inhibit root growth and transpiration by interference with water transport [4]. Actually, ENPs blocked apoplastic water flow in plant roots (within a plant, the apoplast is the free diffusional space outside the plasma membrane), certainly by obstructing nanosized cell wall pores, but also by reducing the exclusion diameter of these pores. However TiO2-NPs are likely to agglomerate when dispersed in a saline, neutral aqueous solution. Therefore, exposing plant roots to TiO2-NPs dispersed in a CaCl2 solution at near-neutral pH, even if it is more physiologically relevant, means that plants were exposed to highly agglomerated ENPs. It is thus not surprising that water flow through the plant was blocked, and it rather derives from root exposure to a high concentration (0.3 or 1 gl1) of agglomerated material, than by a specific effect of ENPs. Several studies reported that ENPs impacted plant photosynthetic activity. In a series of articles, TiO2NPs were shown to induce an increase in spinach germination, growth, nitrogen metabolism, and photosynthesis, and protected plants from oxidative damage
T
T
2766
induced by UV radiation [16]. Indeed TiO2-NPs promote energy transfer from chlorophyll b and carotenoids to chlorophyll a, but also to the P680 chlorophyll a of D1/D2/Cyt 559 complex, in vitro. The authors conclude that nanosized TiO2 would enter spinach cells and chloroplasts more easily due to their small size, then bind to the D1/D2/Cyt 559 complex and improve absorption, transfer, and conversion of light energy in chloroplasts [16]. Lastly, genotoxic effects of ENPs are currently investigated on several plant species. Root exposure to Ag-NPs reduce the mitotic index of exposed cells in Allium cepa, and more generally impair the cell division process, resulting in chromatin bridges, stickiness, disturbed metaphase, chromosomal breaks, and cell disintegration [17]. In the same test (A. cepa test is a standardized assay) TiO2-NPs show no genotoxic impact, neither chromosomal aberration nor micronuclei [18]. Conversely DNA damage is detected in A. cepa and Nicotiana tabaccum exposed to TiO2-NPs, by using comet assay and DNA laddering experiment. This genotoxic effect is correlated with the induction of malondialdehyde production, which is a marker of lipid peroxidation, as a result of oxidative stress [19]. Concluding Remarks As a conclusion, the impact of ENPs on plants is still poorly documented; however some effects have been reported. Some ENPs, particularly those which partly dissolve, impact plant developments, i.e., seed germination, root and shoot elongation, and plant biomass. Some ENPs are taken up by plant roots, sometimes translocated to the shoots, and even transmitted to the next generation via their transfer to seeds. When accumulated in plant leaves, some ENPs can be transferred to primary consumer, for instance hornworms, meaning that they reach the food chain. The genotoxic potential of some ENPs has been demonstrated, although other articles relate opposite results. Some ENPs would modulate photosynthesis by impacting photosystem activity. Note that only some ENPs have these properties and exert these effects, the results described herein definitely cannot be generalized to all ENPs. Note also that almost all the results were obtained on plants exposed to ENPs in vitro, but only one publication describes the impact of ENPsamended soil on plant development. Data thus need to
Toxicology: Plants and Nanoparticles
be collected on environmentally relevant systems, i.e., plants cultivated on natural soils and contaminated with low doses of ENPs since massive contamination of the environment with ENPs would improbably occur.
Cross-References ▶ Cellular Mechanisms of Nanoparticle’s Toxicity ▶ Effect of Surface Modification on Toxicity of Nanoparticles
References 1. U.S. Environmental Protection Agency: Ecological effects test guidelines (OPPTS 850.1200): seed germination/root elongation toxicity test. http://www.epa.gov/opptsfrs/ publications/OPPTS_Harmonized/850_Ecological_Effects _Test_Guidelines/Drafts/850-4200.pdf (1996) 2. Stampoulis, D., Sinha, S.K., White, J.C.: Assay-dependent phytotoxicity of nanoparticles to plants. Environ. Sci. Technol. 43(24), 9473–9479 (2009) 3. Lin, D., Xing, B.: Phytotoxicity of nanoparticles: inhibition of seed germination and root growth. Environ. Pollut. 150(2), 243–250 (2007) 4. Asli, S., Neumann, P.M.: Colloidal suspensions of clay or titanium dioxide nanoparticles can inhibit leaf growth and transpiration via physical effects on root water transport. Plant Cell Environ. 32(5), 577–584 (2009) 5. Seeger, E., et al.: Insignificant acute toxicity of TiO2 nanoparticles to willow trees. J. Soils Sediments 9(1), 46–53 (2009) 6. Lee, W.M., et al.: Toxicity and bioavailability of copper nanoparticles to the terrestrial plants mung bean (Phaseolus radiatus) and wheat (Triticum aestivum): plant agar test for water-insoluble nanoparticles. Environ. Toxicol. Chem. 27(9), 1915–1921 (2008) 7. Shah, V., Belozerova, I.: Influence of metal nanoparticles on the soil microbial community and germination of lettuce seeds. Water Air Soil Pollut. 197(1–4), 143–148 (2009) 8. Lin, D., Xing, B.: Root uptake and phytotoxicity of ZnO nanoparticles. Environ. Sci. Technol. 42(15), 5580–5585 (2008) 9. Zhu, H., et al.: Uptake, translocation, and accumulation of manufactured iron oxide nanoparticles by pumpkin plants. J. Environ. Monit. 10(6), 713–717 (2008) 10. Kurepa, J., et al.: Uptake and distribution of ultrasmall anatase TiO2 Alizarin red S nanoconjugates in Arabidopsis thaliana. Nano Lett. 10(7), 2296–2302 (2010) 11. Lopez-Moreno, M.L., et al.: X-ray absorption spectroscopy (XAS) corroboration of the uptake and storage of CeO2 nanoparticles and assessment of their differential toxicity
Transmission Electron Microscopy
12.
13.
14.
15.
16.
17.
18. 19.
in four edible plant species. J. Agric. Food Chem. 58(6), 3689–3693 (2010) Lin, S., et al.: Uptake, translocation, and transmission of carbon nanomaterials in rice plants. Small 5(10), 1128–1132 (2009) Khodakovskaya, M.V., et al.: Complex genetic, photothermal, and photoacoustic analysis of nanoparticleplant interactions. Proc. Natl Acad. Sci. U.S.A. 108(3), 1028–1033 (2011) Khodakovskaya, M., et al.: Carbon nanotubes are able to penetrate plant seed coat and dramatically affect seed germination and plant growth. ACS Nano 3(10), 3221– 3227 (2009) Judy, J.D., Unrine, J.M., Bertsch, P.M.: Evidence for biomagnification of gold nanoparticles within a terrestrial food chain. Environ. Sci. Technol. 45(2), 776–781 (2011) Mingyu, S., et al.: Promotion of nano-anatase TiO2 on the spectral responses and photochemical activities of D1/D2/ Cyt b559 complex of spinach. Spectrochim. Acta A 72, 1112–1116 (2009) Kumari, M., Mukherjee, A., Chandrasekaran, N.: Genotoxicity of silver nanoparticles in Allium cepa. Sci. Total Environ. 407(19), 5243–5246 (2009) Klancnik, K., et al.: Use of a modified Allium test with nanoTiO2. Ecotoxicol. Environ. Saf. 74(1), 85–92 (2011) Ghosh, M., Bandyopadhyay, M., Mukherjee, A.: Genotoxicity of titanium dioxide (TiO2) nanoparticles at two trophic levels plant and human lymphocytes. Chemosphere 81(10), 1253–1262 (2010)
Transcutaneous Delivery ▶ Dermal and Transdermal Delivery
Transdermal Permeation ▶ Dermal and Transdermal Delivery
Transduction ▶ Micromachined 3-D Force Sensors
Transfer ▶ Fate of Manufactured Nanoparticles in Aqueous Environment
2767
T
Transmission Electron Microscopy Helmut Kohl Physikalisches Institut, Westf€alische WilhelmsUniversit€at M€unster, Wilhelm-Klemm-Straße 10, M€unster, Germany
Definition Transmission electron microscopy is a method to image small thin objects by use of the transmitted electrons.
The Transmission Electron Microscope (TEM) Similar to the general scheme of a light microscope, a transmission electron microscope [1, 2] consists of an electron source, a condenser system, an objective lens and a projector system as shown in Fig. 1. Many transmission electron microscopes have additional instruments attached to it, such as an X-ray detector and/or an energy loss spectrometer in order to be able to perform elemental analyses. These techniques will be explained later on. The electron source, commonly called “gun” is composed of a cathode and an accelerating system. There are different types of cathodes being used: – Thermal cathodes – Schottky emitters, often misleadingly called (hot) field emission cathodes – (Cold) field emission cathodes. Thermal cathodes make use of thermal emission of electrons at high temperatures. Most often they use tungsten or LaB6 as emitting material. The emission barrier can be lowered by applying a strong electrostatic field to a tip (Schottky effect). This is the basic principle of a Schottky emitter. Using an even stronger electrostatic field at a sharp tip one can extract electrons by means of the quantum mechanical tunnel effect. The resulting field emission cathodes require ultra high vacuum to operate, because otherwise positively charged ions will be accelerated towards the tip eventually leading to its destruction. All of these cathodes are followed by an accelerating system ending with the anode. Most transmission electron microscopes usually operate at voltages of
T
T
2768
Transmission Electron Microscopy
Electron gun Specimen
Condenser 1
Si (Li)
2α
Condenser 2
Electron lens
NI F
X-rays
Specimen Objective Objective diaphragm
Fe 1 st
Selector diaphragm Intermediate lens
Image 2 nd
Projector lens
ϕ
Image z
Transmission Electron Microscopy, Fig. 2 Schematic diagram of an electromagnetic lens [1]
Image Electronenergy-loss spectrometer TEM
Transmission Electron Microscopy, Fig. 1 Schematic ray paths for a transmission electron microscope (TEM) equipped with an additional x-ray spectrometer and an energy loss spectrometer [1]
constructed by coils encapsulated by a ferromagnetic material as shown in Fig. 2. To understand their working principle, let us consider the movement of an electron in the homogeneous magnetic field of a long coil. The Lorentz force on the electron is given by *
*
*
F ¼ e½v B
(1)
100–300 kV, even though there are special low voltage TEMs operating at voltages from 20 kV upwards and high voltage TEMs operating at voltages up to 1.2 MeV. The role of the condenser system is to transfer the electrons from the gun onto the specimen. It is composed of two to five electron lenses to allow to independently vary the illumination angle and the size of the illuminated area. The transmitted electrons are then focused to a magnified intermediate image which in turn is magnified and transferred into the final image plane.
The velocity v can be separated into a component parallel vII to the magnetic induction B and one perpendicular component v? . As there is no force operating on the parallel component, vII remains constant. Perpendicular to the magnetic induction B the electron performs a circular motion with a radius r that is determined by the equivalence of the Lorentz force to the centrifugal force
Magnetic Electron Lenses
or
To focus an electron beam, we need lenses operating similar to glass lenses in light optics. These can be
mv2? ¼ ev? B r
r¼
mv? eB
(2)
(3)
Transmission Electron Microscopy
2769
An electron starting from the optic axis will cross it again after it has completed one circle – in other words, the electrons starting from the initial plane are all imaged in the latter plane. For a homogeneous magnetic field it can be shown that the focal length f is determined by 1 e ¼ B2 L f 8mf
(4)
where f ¼ f(1 + ef/(2E0)) is the (relativistically corrected) potential, f the electrostatic potential, E0 ¼ 511 keV the rest energy of the electron, and L the length of the coil. As can be seen from Eq. 4, the focal length depends strongly on the magnetic induction B. This general fact is also true for the commonly used “short magnetic lenses” that use an inhomogeneous magnetic field. The focal length of a magnetic electron lens in an electron microscope can therefore be easily modified by adjusting the current through the coil thus changing the magnetic field. Unfortunately, electron lenses suffer from strong spherical and chromatic aberrations [3, 4] that limit the spatial resolution limit for conventional electron microscopes to about 0.15 nm. Just in the very last years it has become possible to correct for aberrations so that with the most advanced instruments on can achieve a resolution limit of about 0.05 nm. As the focal length of every lens can be easily varied by changing the lens current, the ray paths in a transmission electron microscope can be chosen within a wide range. In particular, it is easily possible to change the illumination from a large area to a small spot mode. Whereas the illumination of a large area is used for conventional imaging, the small spot mode allows the investigation of small areas whose size may well approach the 0.1 nm level. By changing the excitations of the projector lenses it is easily possible to vary the magnification over a large range.
Image Formation in the Transmission Electron Microscope Figure 3 shows schematically the image formation process for a periodic specimen. The incident electron beam has to be treated quantum mechanically as a wave that is diffracted by the periodic object. Constructing the ray paths in the usual way, one finds
T
that all the rays diffracted in a particular direction are recombined in one point in the back focal plane of the objective lens. Rays emitted parallel to the optic axis are focused in the focal point. In other words, in the back focal plane the electrons are selected according to the scattering angles or, to put it differently again, one obtains a diffraction pattern in the back focal plane. Therefore the back focal plane is often called the diffraction plane. For a general, non-periodic specimen the diffraction pattern consists of a continuous distribution rather than of individual spikes. Its value is proportional to the squared modulus of the Fourier transform of the electron wave just below the specimen. In the case of ideal imaging all rays originating in one point of the specimen will be refocused into one point in the image plane. As in light optics, the image formation in an electron microscope can be viewed as a sequence of two Fourier transforms – the first one between the specimen and the back focal plane and the second one between the back focal and the image plane. Consequently any manipulation in the back focal plane will influence the final image. One example is the insertion of a circular aperture which prevents electrons scattered by an angle larger than a0 from entering into the image. If the first diffraction angle of a periodic specimen is larger than a0, only the undiffracted beam will enter into the image. In that case the diffraction amplitude beyond the aperture is identical to the amplitude without any specimen and thus one obtains an image without any structure. This example demonstrates that the resolution is then limited by the maximum angle passing through the objective aperture. Using Bragg’s law 2a sin y ¼ nl
(5)
where a is the lattice spacing and y the scattering angle for n-th order diffraction, one obtains for the lattice resolution limit d d
l 2 sin a0
(6)
Contrast Mechanisms in Transmission Electron Microscopy As the fast incident electrons pass through the thin specimens used for transmission electron microscopy
T
T
2770
Transmission Electron Microscopy
Transmission Electron Microscopy, Fig. 3 Ray paths showing the imaging by a lens. All rays emitted by the specimen in a given direction are focused to an off-axis point in the back focal plane
− F
F
object plane
z
back
image
focal
plane
plane lens
there is practically no absorption. Therefore an “ideal” image (without any aperture or any other optical element in the back focal plane) does not exhibit any contrast. In order to see anything in the image, the electron wave in the back focal plane has to be influenced in order to discern the structure of the specimen. This can be done in several ways leading to different types of contrast.
Scattering Absorption Contrast A circular diaphragm, inserted into the back focal plane, intercepts all the electrons scattered at angles larger than a0 ¼ r/f, where r is the radius of the hole in the aperture and f the focal length of the objective lens. Strong scatterers, who scatter a substantial fraction of the incident electrons into angles larger than a0, appear darker in an otherwise bright image. It has to be stressed that no electrons are absorbed in the specimen. The scattering takes place in the specimen, the absorption in the diaphragm in the back focal plane. The contrast is strongest for small aperture angles, because then a large fraction of the scattered electrons will be removed from the beam. Equation 6 shows that for small scattering angles the resolution deteriorates. Even though this method is very simple, it can only be used for low and medium resolution images (coarse structures). For practical applications a suitable compromise has to be found between contrast and resolution.
Dark Field Imaging To avoid the conflict between resolution and contrast one can use a completely different way to obtain contrast. One major problem of scattering absorption contrast is the fact that in the image one sees only a minor decrease of intensity, which is hard to discern, in an otherwise bright image. Therefore it is of great interest to visualize strong scatterers as bright spots. To form such an image the unscattered electrons have to be removed so that only the scattered electrons remain in the image. This can be done by use of a beam stop in the back focal plane intercepting all unscattered electrons that have left the specimen in a direction parallel to the optic axis. One way of doing this is to mount the beam stop on a thin wire and place it in the diaphragm in the back focal plane. It is difficult, however, to properly adjust the beam stop. The essential requirement for the dark field imaging mode is that the unscattered electrons are removed in the back focal plane. As in light microscopy, this can also be achieved by inserting a circular diaphragm in the back focal plane and using a hollow-cone illumination whose angle is larger than the objective aperture angle ao. In electron microscopy such an illumination can be most easily obtained by using deflection coils in the condenser system to tilt the incident beam and rotate the direction of incidence by properly addressing the deflection coils. A dark field image exhibits bright spots at the positions where scattering is strong on an otherwise dark background. The resolution is determined by the
Transmission Electron Microscopy
aperture angle a0. Usually only a small fraction of the incident electrons is scattered. Therefore the intensity in dark field images is rather low.
Phase Contrast Imaging In quantum mechanics, the influence of a thin specimen on the incident electron wave can be described by a spatially varying phase shift ’, which is proportional * to the projected potential Vðr; zÞ *
’ðrÞ ¼
i hv
Z1
T
2771 −g
3p 4 p 2 p 4 0 −p 4 −p 2 0
*
Vðr; zÞdz;
(7)
1
1
2
4
Cs ⋅q l
Transmission Electron Microscopy, Fig. 4 Phase shift g introduced by the spherical aberration of the objective lens
*
where r is a two-dimensional vector describing the lateral position in the object plane and h is Planck’s constant. Is it possible to obtain an image whose intensity is proportional to the phase shift? In light optics this can be achieved using Zernikes phase contrast method using a phase plate in the back focal plane to introduce a phase shift of p/2 between the unscattered and the scattered beam. To achieve this, one has to fabricate a glass plate of the proper thickness. How can this effect be utilized in electron microscopy? As mentioned earlier, electron lenses suffer from severe spherical aberrations that correspond to a phase shift increasing with scattering angle. Including the effect of a small defocus Df, this phase shift g is given by gðyÞ ¼ k
2
Cs 4 y y Df 4 2
(8)
where CS is the spherical aberration constant, k ¼ 2p/l the wave number of the electrons and y the scattering angle. By a proper choice of the defocus, one can obtain a phase shift of approximately p/2 over a large range of scattering angles. This is shown in Fig. 4. To optimize phase contrast, the defocus is chosen so that g stays close to p/2 for a large range of scattering angles. The performance of this method can best be described by use of a transfer function. Such transfer functions are well known in electronics, where they determine to what extend an electric signal of a given frequency f is transmitted through a circuit. To describe the transmission of a general electric
–sing 1
0
–1
0
1
2
4
Cs ⋅q l
Transmission Electron Microscopy, Fig. 5 Transfer function L(y) ¼ sin g(y) for phase contrast imaging
signal, the latter is decomposed into its frequencies – in other words, it is Fourier transformed. The concept of a transfer function can readily be generalized to twodimensional spatial signals. Due to the circular symmetry of an aligned microscope, the transfer function depends only on the modulus of the spatial frequency y/l. For phase contrast imaging, the “phase contrast transfer function” is given by the sine of g LðyÞ ¼ sin gðyÞ
(9)
Figure 5 shows its behavior for the optimum defocus, the so-called Scherzer focus Df 1:2
pffiffiffiffiffiffiffiffi Cs l
(10)
The first zero of the phase contrast transfer function determines the resolution limit d, which is given by
T
T
2772
Transmission Electron Microscopy
pffiffiffiffiffiffiffiffiffiffi d ¼ 0:67 4 Cs l3 . It can be shown, that spherical aberration is unavoidable for electric-magnetic round lenses, provided they are time independent and that there is no electrostatic charge within the lens. Typical values of the spherical aberration constant CS are about 1 mm. Inserting CS ¼ 1 mm and assuming an accelerating voltage of 300 kV, corresponding to a wave length l ¼ 20 pm, one obtains a resolution limit of d ¼ 0.2 nm. One can thus obtain highresolution images with good contrast.
Field-emission tip U1 U First anode Second anode
Scanning coils
High-Resolution Electron Microscopy So far the discussion has been confined to thin objects, where multiple scattering can be neglected. Frequently, high resolution electron microscopy is applied to crystals oriented so that the incident electron beam is parallel to a low index axis [5]. Often the resolution is then sufficient to discern atomic columns. In such crystals, however, multiple elastic scattering plays an important role. These processes strongly influence the intensity distribution in the image. It is therefore no longer possible to interpret high-resolution images of crystals directly. Instead, one has to simulate an image of a model crystal, compare the result with the experimental data, modify the model, recalculate, compare, . . ... until a good match between the simulated and the experimental image has been achieved.
Correctors Due to the spherical aberration of the objective lenses, the resolution limit of electron microscopes has stayed at values well above 0.1 nm for decades. The correction for this aberration has been a challenging task, which has been achieved only very recently by using lenses without rotational symmetry [4]. Nowadays several correctors using sextupole or quadrupole and octupole lenses are commercially available, some of which do not only correct for the spherical aberration, but also for chromatic aberration. In this way a resolution limit of 0.05 nm can be achieved.
Scanning Transmission Electron Microscopy It is also possible to obtain images by focusing the beam to a small spot that is scanned in a raster
Scanning coils Objective lens
Specimen Detector for elastically scattered electrons Detector for energy-loss electrons
Iel
Energy analyzer
energy-loss
Transmission Electron Microscopy, Fig. 6 Field emission scanning transmission electron microscope equipped with an energy loss spectrometer [1]
like manner over the specimen. The number of counts, e.g., of the scattered electrons, is then stored as intensity value for that pixel. The setup of a scanning transmission electron microscope (STEM) is schematically shown in Fig. 6. To obtain an image in a STEM, one commonly uses an annular detector subtending angles outside the cone of the incident rays. This produces a dark-field image, hence the detector is called an annular dark-field (ADF) or, for a large inner angle, a high-angle ADF (HAADF) detector. A major advantage of scanning transmission electron microscopy is the possibility to use other secondary signals to determine the intensity in the corresponding pixel of the image. In particular, the use of an element specific signal (e.g., X-rays of a particular transition) permits to acquire images showing the distribution of that element within the specimen. Whereas dedicated STEMs are not very widely spread, many modern TEMs have a scanning attachment permitting to
Transmission Electron Microscopy Transmission Electron Microscopy, Fig. 7 Ray diagram for a TEM for (a) bright field imaging and (b) diffraction [1]
2773
a
Source
T
b
Condenser diaphragm Condenser lens
Specimen Objective lens Objective diaphragm First diffraction pattern First intermediate image Selector diaphragm Intermediate lens Second diffraction pattern
Second intermediate image Projector lens Third diffraction pattern
Final image Screen Bright-field imaging
perform scanning transmission electron microscopy in addition to the conventional operating modes.
Electron Diffraction In the previous chapters the various possibilities to obtain an image in an electron microscope have been discussed. By adjusting the intermediate lens, which is the first lens in the projector system, it is easily possible to switch to the diffraction mode, where the back focal plane, i.e., the diffraction pattern, is imaged on the final screen [6, 7]. Essentially in this mode the second and further intermediate image planes and diffraction planes are interchanged as can be seen from Fig. 7. Using diffraction mode one can obtain a wealth of information on the structure of a crystallite with any standard TEM. To obtain the position of the diffraction spots, one can use Laues equation which states that the
Selected-area diffraction
difference between the wave vector k0 of the incident and kf of the diffracted beam must be equal to a reciprocal lattice vector g k f k0 ¼ g
(11)
Every spot in a diffraction pattern corresponds to a reciprocal lattice vector. Electron diffraction thus permits to measure the reciprocal lattice, which in turn allows determining the Bravais lattice of the illuminated crystallite. The position of the spots can most easily be visualized using Ewalds construction (Fig. 8). There are two conditions for the appearance of a diffraction spot: 1. The energy of the electrons, and therefore the magnitude of their wave vector, remains unchanged during the scattering process (elastic scattering). This can be visualized by drawing a sphere around the origin of the wave vector of the incident electron
T
T
2774
Transmission Electron Microscopy
ting
lec ref ice latt s⊥ ne pla g
M
2θB
ko =
uo λ O
k= g
u λ H
Transmission Electron Microscopy, Fig. 8 Ewald sphere construction to determine the directions of diffracted radiation [1]
with a radius corresponding to its length. The wave vectors of both the incident and the scattered electron must end on the surface of this so-called Ewald sphere. 2. Drawing the wave vector of the incident electron as to end in the origin of the reciprocal lattice, the wave vector of the scattered electron has to end in another reciprocal lattice point. Putting these two conditions together, one finds that electrons can only be diffracted if the Ewald sphere crosses a reciprocal lattice point. This condition is equally valid for the diffraction of X-rays, neutrons and electrons. Fast electrons with energies of the order of 100 keV or more have wavelengths of a few pm – much smaller than the lattice constant of a crystal (about 1 nm). Consequently, the length of the wave vector of the electron beam is two or three orders of magnitudes larger than the length of a reciprocal lattice vector. In the area near the origin of the reciprocal lattice, the Ewald sphere can then be approximated by its tangential plane. Thus, Ewalds construction states that in electron diffraction near the central spot one observes a planar cut through the reciprocal lattice. The plane through the origin is called the zero
order Laue zone. As an example, Fig. 9 shows the arrangements of spots for a face-centered cubic lattice illuminated along different zone axes. For larger scattering angles, this planar approximation is no longer valid and the Ewald sphere cuts through reciprocal lattice vectors of planes parallel to the zero order Laue zone, which are denoted as first, second, etc. order Laue zones. So far electron diffraction has been discussed for using a parallel beam for illumination. This method results in a spot pattern giving information of the illuminated crystallite. If there are many randomly oriented crystallites in the illuminated area, one obtains Debye-Scherrer rings just like in powder diffraction patterns. The determination of the crystal potential, however, is more complicated due to the fact that in electron diffraction multiple scattering effects cannot be neglected. These depend strongly on the thickness of the crystal and on the exact direction of the incident beam. Instead of varying the incident beam direction with time, one can illuminate the crystal with a range of incident beam directions all at once. This technique is called convergent beam electron diffraction (CBED) (Fig. 10). Instead of the diffraction spots for parallel illumination, one then observes circles that exhibit an internal structure. Comparing the intensity distribution within these circles with the results of simulations, one can determine the crystal potential with high accuracy. As long as the illumination angle is smaller than the Bragg angle, the circles do not overlap and each one corresponds to a specific reciprocal lattice vector.
Analytical Electron Microscopy Taking a broader view of the scattering of electrons in a thin specimen one can see, that in addition to the elastic scattering processes discussed up to now, many electrons experience inelastic scattering [1, 2, 6]. In doing so, they loose part of their energy. The various processes are shown schematically in Fig. 11. In order to excite an inner shell electron into an unoccupied state, the incident electron has to transfer an energy corresponding to at least the binding energy of the original inner shell electron (Fig. 11a). As these binding energies are characteristic for the element, they can be used as a signature of that particular element. Measuring the energy loss of the scattered
Transmission Electron Microscopy
T
2775
Transmission Electron Microscopy, Fig. 9 Diffraction pattern of a face centered cubic crystal illuminated along the [100], [110], and [111] zone axes near the central spot [1]
224 004
024
044
004
002
022
042
002
000
020
040
044
224 113 202
222
022
242
111 040
020
220
000
220
220
220
022
[100]
[110]
202
[111]
L
a K
specimen Ionisation
Ejected electron E = ΔE – EK
Energy loss ΔE ≥ EK ω
diffraction spots
α=1–ω
diffraction discs
Transmission Electron Microscopy, Fig. 10 Schematic diagram demonstrating the effect of a converging beam on the diffraction pattern (CBED)
b
c EK–EL
electrons thus permits to determine the chemical composition of the illuminated area of the specimen. This is the basis of a technique called EELS for electron energy loss spectroscopy. Shortly after its excitation, the atomic electron decays into its ground state releasing the energy via an X-ray photon (Fig. 11b) or an Auger electron (Fig. 11c). Both of these secondary processes can in principle be used to determine the chemical composition of the specimen. Due to the small mean free path length, the emission of Auger electrons is very surface sensitive. As their detection requires an UHV environment, they are frequently used in surface science, but not in transmission electron microscopy. X-ray spectroscopy, however, is a very common technique in electron microscopy.
EK–EL
hν x-ray quantum EX = EK – EL
KLL – Auger electron _ EK – 2EL EA ~
Transmission Electron Microscopy, Fig. 11 Schematic representation of (a) the ionization process, (b) the x-ray emission and (c) the Auger emission [1]
Electron Energy Loss Spectroscopy Using a monoenergetic incident beam and measuring its energy distribution after it has passed through a thin specimen, one obtains an energy loss spectrum [8].
T
T
2776
Transmission Electron Microscopy
Transmission Electron Microscopy, Fig. 12 Examples of energy loss spectra for the K-edges of (a) carbon, (b) aluminum and the L23-edges of (c) silicon and (d) chromium [1]
One example is shown in Fig. 12. At characteristic energies, corresponding to the binding energy of an inner shell, the signal shows a marked increase followed by a slow decay. The latter is due to the fact that it is also possible to excite an electron into any state above the Fermi level. The excitation probability, however, decreases for final states with relatively large energies. In order to obtain a quantitative measure of the concentration of an element in the specimen, the background stemming from the tails of edges with lower energies has to be extrapolated. This is most often done by using a power law model for the background intensity Ib as a function of the energy loss E r
I b ¼ AE
I A ¼ N A I 0 sA ða0 ; DEÞ
(13)
I B ¼ N B I 0 sB ða0 ; DEÞ
(14)
and
Dividing Eq. 13 by Eq. 14 one obtains IA NA sA ða0 ; DEÞ ¼ IB NB sB ða0 ; DEÞ
(15)
and for the concentration ratio NA IA sB ða0 ; DEÞ ¼ NB IB sA ða0 ; DEÞ
(16)
(12)
where A and r are free parameters that need to be fitted to experimental data. The signal intensity remaining after background subtraction is proportional to the area density of the element NA, the intensity of the incident beam I0, and the excitation cross section sA(a0, DE), which depends on the acceptance angle a0 and the energy loss window DE. For two elements A and B one obtains the two equations
The concentration can thus easily be determined from the background corrected intensities above the energy loss edges multiplied by the inverse of the ratio of the cross-sections. To measure an energy loss spectrum it is most convenient to attach a magnetic sector field spectrometer underneath the image plane of a transmission electron microscope. Modern commercial instruments allow measuring a large part of the spectrum in parallel with a diode array.
Transmission Electron Microscopy
T
2777
Using such a spectrometer in a scanning transmission electron microscope it is possible to obtain the spatial distribution of the elements in the specimen for each pixel. This technique results in images called elemental maps with a resolution at the atomic level in favorable cases. Alternatively one can obtain elemental maps by using an imaging energy filter, which turns the TEM into an energy filtering transmission electron microscope (EFTEM). Such a filter is inserted in the ray path after the first part of the projective system and intercepts all electrons except for those lying within a well defined energy window. Ideally such an energy filter should provide an image without any chromatic aberration, but at the same time it needs an energy dispersion to be able to select the desired energy range. These two conditions are met by a setup which images an intermediate image plane – the filter entrance plane – into an achromatic image plane and images the corresponding diffraction plane – the crossover plane – into the energy dispersive plane (Fig. 13). In the achromatic image plane the energy loss is “coded” by a tilt proportional to the energy loss resulting in a spatial displacement in the corresponding diffraction plane. The energy selection takes place in this energy dispersive plane. To obtain an elemental map in an EFTEM, one takes a series of at least three images at different energy losses – two below and one above the edge.
With the images below the edge one determines the parameters A and r on a pixel by pixel basis, from these the background intensity is computed and subtracted from the image above the edge (Fig. 14).
x
z
‘Crossover’ plane
Filter entrance plane (FEP)
Energy filter
Achromatic image plane
E – ΔE
E
Energy – dispersive plane
Transmission Electron Microscopy, Fig. 13 Schematic diagram showing the ray paths and the relevant planes in an imaging energy filter [1]
I selection of energy windows object
E Eb1
Eb2
T
ES
image recording
background subtraction
Transmission Electron Microscopy, Fig. 14 The steps necessary to acquire an elemental map
background extrapolation I(E) = AE−r
image
T
2778
Transmission Electron Microscopy, Fig. 15 The data cube depicting the complete information that can be explored in different ways. (a) A STEM acquires spectra point by point (b) In EFTEM, the information is obtained energy slice by energy slice
Transmission Electron Microscopy
a
The total information that can be obtained via EELS in a STEM or in an EFTEM is conveniently visualized as a data cube with two spatial coordinates and one energy loss coordinate (Fig. 15). In a STEM many energy channels corresponding to one pixel are measured at the same time (Fig. 15a), whereas in an EFTEM one measures the entire picture for one energy channel (Fig. 15b). The efficiency of both methods depends on the number of pixels and energy channels to be acquired. Taking a closer look at a spectrum just above an edge, one finds fine details in the structure that depend on the chemical environment. These details can be used to determine the bonding of the element. The energy loss near edge structure (ELNES) is similar to the X-ray absorption near edge structure (XANES).
b ΔE X
y
L2M1M1- Auger electron EA 0
N M5 M1
Rather than detecting the inelastically scattered electrons, it is also possible to use secondary processes for the determination of the elemental distribution [6]. This is the basis of x-ray spectroscopy for elemental analysis. The inner shell levels of an atom and the corresponding transitions are shown schematically in Fig. 16. The energy difference between the excited and the ground state can be released either by emission of an x-ray or an Auger electron. The probability for the emission of an Auger electron, called Auger yield, and the corresponding fluorescence yield are shown in Fig. 17 for K, L, and M-shell transitions. From these graphs one can clearly see that for large atomic numbers the fluorescence yield dominates. To determine the chemical composition of the specimen one needs to measure the energy spectrum of the emitted x-rays. This can be done either by using Bragg reflection from a crystal spectrometer to obtain the wavelength or by measuring the number of
Mα Lα1 β1
EL- EM L3 L2 L1
ΔE
n l
j
2 3 2 1 1 0
5/2 3/2 3/2 1/2 1/2
1 3/2 2 1 1/2 0 1/2
EK hν = EK - EL
Kα1
β1 β2
α2
K
Energy Dispersive X-Ray Spectroscopy
Coster - Kronig
1 0 1/2
Transmission Electron Microscopy, Fig. 16 Energy levels of the atomic subshells with their quantum numbers n, l, j and possible transitions to fill a hole in the K-shell including the nomenclature of the emitted x-ray lines [1]
electron–hole pairs created after absorption of the x-ray photon in a semiconductor detector. Whereas the use of the first method, called wavelength dispersive x-ray spectroscopy (WDX or WDS) is confined to electron microprobes and scanning electron microscopes, the latter method, named energy dispersive x-ray spectroscopy (EDX or EDS) is used in transmission electron microscopes as well as in scanning electron microscopes. An example of an EDX-spectrum, taken with a high-purity Ge-detector, is shown in Fig. 18. The position of the lines reveals the elements present in the illuminated area of the specimen. To determine the chemical composition, one has to establish a relation between the intensity of the x-ray line and the number of corresponding atoms. Even though there are many factors entering
Transmission Electron Microscopy
2779
into this relation (angular range subtended by the detector, fluorescence yield, detection efficiency, . . .), it is obvious, that the number of counts is proportional to the concentration of the element. For two elements A and B one can lump all these unknown factors into
αL 0.5
0.5
aK 0
15
0
30 45 60 Atomic number Z
1.0 90
75
the two parameters kA and kB thus obtaining for the counts IA and IB I A ¼ kA cA and
(17)
I B ¼ kB c B :
(18)
0
αM
X - ray fluorescence yield ω
Auger yield α
1.0
T
Dividing Eq. 17 by Eq. 18 one obtains CA IA ¼ kAB CB IB
(19)
where the “k-factor” kAB ¼ kB/kA is a parameter that needs to be determined. This can be done experimentally by use of well defined standard specimens. In practice one can measure the k-factors with respect to one element, e.g., Fe or Si, and use the relation kAB ¼ kAC =kBC :
Transmission Electron Microscopy, Fig. 17 Auger electron yield a and fluorescence yield o ¼ 1a as a function of atomic number [1]
300
W-Mα
x7
Ti-Kα
W-Kα1
2000
1500
W-Kα2 0
0
5
15
10
40
50
60
100
W-Kβ2
Ta-Kβ13
W-Lγ1
Ta-Kα2
W-Lβ1 W-Lβ2
Ta-Lβ
Ta-Lα W-Lι
Co-Kα
Ti-Kβ
Nb-L
Ti-Lα
W-Mζ
500
W-Kβ13
W-Lα
1000
C-Kα
intensity [counts]
200
70
0 80
energy [keV]
Transmission Electron Microscopy, Fig. 18 Energy dispersive x-ray spectrum of a hard metal (WC-TaC-TiC-Co) recorded with a high-purity germanium detector [1]
T
T
2780
This should be done once for every instrument and can then be used as long as the conditions remain the same. Comparing the limits of EELS and EDX for the detection of an element, one finds that the dependence of the fluorescence yield on the atomic number favors EELS for light elements. The relatively large background in EELS measurements as compared to an X-ray line indicates that the detection of low concentrations is easier with EDX than with EELS. On the other hand, the fine structure in an EELS spectrum yields additional information about the chemical bonding not accessible in an EDX spectrum. Thus both methods have their merits and their drawbacks and the choice strongly depends on the specimen and the information to be gained from it.
T-Rays
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering Keqin Yang, Dale Hitchcock, Jian He and Apparao M. Rao Department of Physics and Astronomy, Clemson University, Clemson, SC, USA
Definition The spark plasma sintering technique utilizes highly localized intensive Joule heating and discharging generated by a self-adjusted high-density current flow to alter the micromorphology of carbon nanotubes.
Cross-References
Introduction
▶ Electron Microscopy of Interactions Between Engineered Nanomaterials and Cells ▶ Field Electron Emission from Nanomaterials ▶ Scanning Electron Microscopy
Carbon comes in many forms, ranging from sp-bonded metastable carbene chains, sp2-bonded graphite, graphene, single wall and multiwall nanotubes, fullerenes, and amorphous soot to sp3-bonded diamond. In particular, carbon nanotubes (CNTs) exhibit unique electrical, magnetic, thermal, chemical, mechanical, and optical properties that stem from their tubular topology and reduced dimensionality. Topologically, a single-walled nanotube (SWNT) can be viewed as one sp2-bonded graphene sheet rolled into a seamless tube. A sibling of the SWNT is the multiwalled nanotube (MWNT), made of a few to tens of concentric shells with random chiral registries, spaced by 0.34 nm and coupled through van der Waals interactions as in graphite. In the past few decades, CNTs have been extensively studied from fundamental research and technological points of view. With a nm scale diameter and a large aspect ratio, CNTs are one of the closest material realizations of a 1-D system, offering unprecedented opportunities to study 1-D physics. Furthermore, CNTs are prime candidates for molecular electronics, one of the most promising directions of nanotechnology. The electrical properties of CNTs are sensitive not only to intrinsic parameters such as the number of shells, tube diameter, and chirality but also to extrinsic parameters such as defects, dopants, and the environment in which they reside [1]. For example, devices based on macroscopic forms of CNTs such as arrays,
References 1. Reimer, L., Kohl, H.: Transmission Electron Microscopy. Springer, New York (2008) 2. Williams, D., Carter, B.: Transmission Electron Microscopy. Plenum, New York (1996) 3. Hawkes, P., Kasper, E.: Principles of Electron Optics, vol. I & II. Academic, London (1989) 4. Rose, H.: Geometrical Charged-Particle Optics. Springer, Berlin (2009) 5. Spence, J.: High-Resolution Electron Microscopy. Oxford, Oxford (2003) 6. Loretto, M.H.: Electron Beam Analysis of Materials. Chapman & Hall, London (1994) 7. Morniroli, J.P.: Large-Angle Convergent Beam Electron Diffraction (LACBED). Socie´te´ Franc¸aise des Microscopies, Paris (2002) 8. Egerton, R.: Electron Energy-Loss Spectroscopy in the Transmission Electron Microscope. Plenum, New York (1996)
T-Rays ▶ Terahertz Technology for Nano Applications
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
yarns, and films are gaining intensive interest. In these devices, the primary factor governing the macroscopic electrical transport is the micromorphology, comprised of three basic aspects, (1) the crystal chemistry of an individual CNT such as chirality, defects, and dopants; (2) the topology of the interconnected CNTs; and (3) the density and nature of inter-tube junctions. It is often the case that aspects (2) and (3) matter more than (1) in determining the electrical properties of a macroscopic CNT network. Control of these intrinsic and extrinsic parameters remains challenging and relies on techniques capable of tailoring CNT structures on the atomic, nm, and mm scales.
Tailoring CNT Morphology to Realize New Functions Many methods have been used to alter the structure of CNTs, from simply hot pressing to more advanced techniques such as electron or ion irradiation, which gives rise to a seemingly endless number of new phenomena and potential functions. In an early attempt to densify CNTs, Ma et al. employed hot pressing at 2,000 C and 25 MPa in an argon atmosphere to form a “soft sintered” CNT network without altering the tubular morphology of CNTs [2]. In addition, Rubio et al. showed that it is possible to cut and introduce defects into CNTs with very high precision using an STM tip, which has the salient advantage of real-time monitoring of the change of structure [3]. Using Irradiation to Tailor CNT Morphology More recent research has focused on altering the structure of CNTs using electron or ion irradiation. How would CNTs respond if atoms were removed from the hexagonal lattice continuously [4]? At first glance, irradiating CNTs should lead to a detrimental effect on their crystal structure and thereby properties; however, it has been demonstrated that proper irradiation conditions can have beneficial effects on the properties of CNTs. This is due mainly to the CNT’s innate ability to self-heal. The lattice of CNTs can be rapidly rearranged to accommodate the formation of defects and restore a new local coherent structure from a state that is driven far away from equilibrium during the irradiation process. For example, vacancies in SWNTs lead to the formation of pentagons and
2781
T
heptagons while in MWNTs they result in inter-shell bonds. This extraordinary innate ability renders a way to tailor the nanostructure of CNTs and facilitates a controlled method of self-assembly [5]. The main advantage of using irradiation is to tailor the CNT structure on a nm scale, while the focused electron beam of an aberration-corrected scanning transmission ˚ scale resolution. electron microscope can even reach A The major drawback of this approach is the low throughput of beam-modified CNTs. Lastly, it should be mentioned that when dangling bonds are formed in adjacent tubes it is often energetically favorable to form a junction between the tubes rather than the reconstruction of individual tubes. The goal of this chapter is to introduce the spark plasma sintering (SPS) process as an alternate approach for controlling inter- and intratubular bonding in CNT networks, at nm and mm scale lengths. In electron beam irradiation, it is known that only a small portion of the impinging electron energy can be transferred to and displace nuclei due to conservation of momentum. As such, a very high electron energy (on the order of 10– 100 keV) is needed to alter the structure of CNTs. The SPS process utilizes a low-voltage high-density current and the accompanied localized intensive joule heating or discharging to modify CNTs. The SPS process can control the structure of CNTs in much the same way as irradiation but using macroscopic control rather than the microscopic control of irradiation. Mechanisms of Spark Plasma Sintering The SPS technique has been used widely in powder metallurgy, and densification of nanomaterials and nanocomposites [6]. The basic idea dates back to 1930s but was not put into wide use until recently, primarily due to the lack of applications and the high cost of equipment. SPS is a pulsed plasma discharge process that can generate intensive heating (up to 2,000 C and above) in a short time (few minutes). The voltage is merely on the order of 1–10 V, while the current density is typically on the order 102103 A/cm2 and is highly concentrated at the inter-granular contact or interfacial boundaries. The SPS process features a high thermal efficiency because of the direct heating of the sintering mold and sintered materials, and it results in a homogeneously sintered and densified sample because of the “self-adjusting” uniform heating mechanism [7], surface decontamination, and surface activation. The densification mechanism is related to
T
T
2782
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering FORCE
VACUUM CHAMBER
Spark Plasma Sintering (SPS) mechanism
PULSE CURRENT
Plasma heating GRAPHITE PUNCH
Joule heating
GRAPHITE DIE CASE
POWDER PULSE D.C. POWER SUPPLY
T.C. WELL
GRAPHITE PUNCH
Plasma discharge PULSE CURRENT
Electrical current
w.w.w.substech.com
FORCE
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 1 (left) The schematic of a SPS system, and (right) two major heating mechanisms in the SPS process
plastic deformation and grain rotation coalescence and sliding, aided by the pre-melting of grain boundaries. A schematic diagram of the SPS system and the primary heating mechanisms are shown in Fig. 1. The control parameters of SPS process are macroscopic: they are the sintering pressure, current density, ON/OFF ratio, and sintering time. It should be noted that heating and current flow are interdependent but are regarded as separate mechanisms in the SPS process. The sintered materials are under a combination of one or more of the following effects [6]: spark plasma, spark impact pressure, Joule heating, and electric field–induced diffusion. In particular, the surface of sintered materials are more easily degassed and chemically activated, promoting strong mass transport at atomic, nano, and macroscopic levels at a relatively lower temperature and in a shorter time than conventional processes such as hot pressing. Mass transport in SPS is commonly believed to be driven by electric field diffusion. In view of the self-healing ability of CNTs, the current density in the SPS process can be varied to control the deviation from equilibrium, yielding a controllable number of permanent defects. At present, a comprehensive detailed model of the impact of SPS process on CNTs is lacking so that most ongoing work is experimental and exploratory in nature.
Using SPS to Tailor CNT Morphology It has been shown that the SPS process may cause the bonding in CNTs to transform from van der Waals in a rope or yarn, to robust sp2, then to sp3 in microdiamonds [8], or lead to graphitization and crystalline–amorphous heterojunctions in amorphous carbon fibers [9]. Depending on whether CNTs survive the SPS process, the SPS process can be loosely categorized into two regimes. The first regime is the midpower range, where most research is done. In this regime, the tubular morphology is preserved, and there is enough energy to introduce intratubular restructuring (including defects and doping) and create coherent intertubular bonding (i.e., junctions). At present, all experimental evidence suggests that the intra-tube restructuring and inter-tube bonding occur simultaneously during the SPS process. The process of densification also occurs in the mid-power regime. Despite the complexity of inter-tube interactions, the inter-tube spacing and relative orientation of CNTs are the two most important factors which determine the nature of the resulting inter-tube junctions and the packing density. Densification of CNTs is of practical importance in a number of applications such as, the electrodes for supercapacitors, biosensors, and artificial bone scaffolds in which one takes advantage
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
of the high surface-to-volume ratio and the outstanding mechanical and electronic properties of CNTs. The other regime is the high-power regime where the temperature and current density are high enough to destroy tubes, forming other carbon allotropes. The boundary between the mid-power range and high-power range is not clear-cut. Preliminary studies suggest a boundary temperature of 1,000–1,200 C for MWNTs while the boundary temperature is lower for SWNTs and depends more on the perfectness of CNT structure and environment. This section mainly addresses the effects of SPS in the mid-power range, focusing on the SPS-mediated (1) intra-tube restructuring (defects and doping) and (2) formation of inter-tube and inter-shell junctions. These modifications are crucial for controlling the electrical and thermal properties of CNT networks. When a high-density current is pulsed though a CNT assembly, the contact resistance between tubes is much higher than the intrinsic resistance of the tubes, which leads to a highly localized intensive Joule heating or discharging, and the temperature can be sufficiently high to cause rearrangement of atoms/bonds. Inter-tube and Inter-shell Junctions It is known that inter-tube junctions dominate the electrical properties of a CNT network. An important objective of ongoing research is to study to what extent and in what aspects the transport properties of a SPSed network are inherited from individual CNTs. Therefore, the manipulation and formation of inter-tube junctions, especially coherent bonding, is of utmost importance in nanoelectronics (molecular electronics) and transparent conductors [10]. A direct impact of the formation of coherent inter-tube bonding is the change of the effective dimensionality of a CNT assembly. In particular, phonons and electrons may have different effective dimensionality in a SPSed CNT network. With coherent inter-tube junctions, the CNT assembly offers a flexible network architecture, opening up opportunities for several applications. The modification of inter-tube bonding from van der Waals type to robust sp2 inter-tube bonding is a nontrivial task, since the topologically complicated arrangement of sp2 bonds in the inter-tube bonding is not energetically favored. For this reason, high-energy nonequilibrium processes are needed. One example of inter-tube junctions is the coalescence of single- or double-wall CNTs to form a single SWNT of larger
2783
T
diameter [4]. Charlier et al. have proposed a mechanism for the coalescence of tubes using molecular dynamics simulations. The model begins with unzipping of a nanotube to form a carbene chain, and if the metastable carbene chain comes in contact with a neighboring nanotube, it will initiate a reaction to minimize its energy and form a coalesced nanotube. Three steps in Charlier’s model are shown in Figs. 2a–c. In sparkplasma-sintered aligned MWNTs, Yang et al. have observed that coherent inter-tube bonding can be formed at a sintering temperature of 1,500 C for 7 min under 7 MPa axial pressure. The current is applied in the direction perpendicular to the tube direction. As shown in Fig. 2d, the SPS process welds part of two parallel MWNTs together [11]. Several interesting properties of SPSed MWNTs have been uncovered through electrical and thermal transport measurements [11]. First, the transverse resistivity (rT) of a low-dimensional electron system such as CNTs is expected to be much higher than the longitudinal resistivity (rL), as is the case in thin films made of aligned CNTs [12]. The MWNTs SPSed at 1,500 C, however, showed lower resistivity in the transverse direction than the longitudinal direction (Fig. 3a). This surprising observation may be due to the following reasons: the SPS process creates coherent inter-tube bonds that account for the lower rT; the SPS-induced defects within the tubes may lower rL; and finally lower rT may be due to the larger cross section of the conduction surface in the transverse direction than in the longitudinal direction due to the hollow core and inter-tube spacing of CNTs. Second, although MWNTs are typically metallic, the resistivity in both directions shows nonmetallic behavior (dr/dT < 0) as in aligned CNT think films, suggesting that inter-tube junctions dominate the overall electrical conductivity of the CNT network. The temperature dependence of the resistivity is also unusual, and cannot be satisfactorily interpreted in the frameworks of the Luttinger liquid model, variable-range hopping, or weak localization theory. Nonetheless, the temperature dependence of resistivity is reminiscent of the results of individual highly disordered MWNTs in which the T1/2 dependence has been attributed to electron–electron scattering in the presence of random impurities [11]. Third, the thermopower is not only of different magnitudes in the transverse and longitudinal directions but also shows different temperature dependences, with the longitudinal thermopower showing
T
T
2784
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 2 (a–c) Three stages from a theoretical model of the coalescence of two SWNTs. (d) shows that several MWNTs are welded with some degree of corrugations in the sidewalls
8
8
a
T
6
b
6 S (mV/K )
Longitudinal
7 r (mW -cm)
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 3 MWNTs SPSed at 1,500 C. (a) The resistivity is lower in the transverse direction than the longitudinal direction; (b) Thermopower in the longitudinal direction shows a strong phonon drag peak below 50 K, which is similar to Kish graphite
L
5
transverse Kish Graphite
4 2 0
4
−2
transverse
Longitudinal
3 0
50
100
150 200 T (K )
a pronounced phonon drag peak below 50 K very similar to that of Kish graphite (Fig. 3b). In view of electron–phonon coupling, the phonon drag features in the thermopower are due to carriers being preferentially scattered by the phonons in the direction of heat flow, when the mean free path of both electrons and phonons are long enough. For individual CNTs, the quantized electronic band structure and phonon dispersion relations lead to a reduced phase space for electron–phonon coupling. Hence the observed phonon drag feature supports the argument that the SPS process leads to an increased effective dimensionality of the CNT assembly.
250
300
0
50
100
150 200 T (K )
250
300
Subsequently, Yang et al. [13] showed that the electrical connectivity of a network made of randomly oriented MWNTs can be effectively tuned by varying the SPS temperature. An increase in the SPS temperature from 500 C to 1,600 C leads to an increase in the packing density from 0.6 g/cm3 to 1.25 g/cm3, reflecting a change in the inter-tube spacing. The concomitant changes in the magnitude (Fig. 4a) and temperature dependence (Fig. 4b) of resistivity strongly suggest a change in the nature of the intertube junctions and thereby the electrical connectivity of the network. As shown in Fig. 4a, the room temperature resistivity as a function of SPS temperature
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
follows a power-law T-dependence, suggesting a percolation behavior. Percolation phenomena are important for the operation of electronic devices based on CNT conducting networks. In Fig. 4b, it is noted that the normalized resistivity rN (T) below 100 K can be categorized into three characteristic T-dependencies. The as-prepared (AP) sample and the sample SPSed at 500 C (S500) exhibit a sub-linear T-dependence, which is labeled as type I. In traces grouped into type II (S800, S1000 and S1200) and type III (S1500) dependencies, there is an up-turn below 60 K. This up-turn may be associated with the formation of coherent inter-tube junctions. In addition, a plateau is evident below 50 K in sample S1500. The ‘S’ shape anomaly in the thermopower for the S1500 sample (Fig. 4c) resembles the thermopower behavior of Kish graphite (Fig. 3b), as noted above. It is interesting to note that this feature in the thermopower data occurs at the same temperature as a plateau in the resistivity data for the SPS1500 sample, which suggests that there is a single unified mechanism governing both phenomena. The possibility of a Kondo effect has been ruled out [13]. SPS can also induce coherent bonding between CNTs and neighboring graphitic particles. In a CNT bundle where graphitic particles are not bonded to CNTs, Zhang et al. have shown that though the graphitic particles have lower intrinsic resistivity than CNTs, their contribution to the overall conduction will be small [14]. Without bonding to CNTs, the conduction of graphitic particles is restricted to variable range hopping (VRH) conduction, which will be very small as long as the concentration of graphitic particles is small. Intra-tube Restructuring It is also possible to introduce intra-tube defects by the SPS process, including creation of vacancies, collapsing/zipping tubes (Fig. 5a), and kinks on the tube wall (Fig. 5b) [11]. The local curvature change of CNTs often indicates the removal of carbon atoms from the lattice and the subsequent reconstruction of a new coherent structure containing non-hexagonal rings and other types of defects. Defects may profoundly affect the electronic properties of CNTs by creating new localized states near the Fermi level. Electrical transport is also affected in view of the quasi-1D molecular nature of CNTs. Finally, these SPS-induced
2785
T
crystal lattice imperfections, especially vacancies, serve as a good platform for doping CNTs since the dopant atoms tend to lodge themselves at these energetically favorable positions. Doping CNTs Doping is a powerful tool for controlling the electrical properties of CNT networks because it offers a way to vary the carrier concentration and thereby tune the electron–electron and electron–phonon coupling that basically govern the electrical properties of CNTs. It has been shown that through doping the majority carrier conduction in a CNT network can be varied from p-type to n-type in a controllable way [15]. However, doping CNTs is technically challenging for several reasons. First, it is difficult to attain a uniform doping level, due to the finite length effect, the subtle differences in capping topology, the amount and distribution of defects, and the proximity environment. This situation is worsened by the low solubility limit of dopants in carbon allotropes (e.g., the solubility of B in graphite is limited to 4 at.% at 1,200 C). Second, in small-diameter CNTs, substitutional doping inevitably compromises the integrity of the hexagonal lattice. Third, electric field doping does not work well in MWNTs and thick films. Apparently, controlling crystal chemistry in the synthesis stage is not adequate for controlling the doping of CNTs, some post-synthesis treatment is needed. Ion implantation or low-energy ion bombardment has been often used to dope carbon nanostructures [5, 16], but the application is limited by expensive equipment. It is also possible that by varying the doping conditions, the nature of the doping could be varied from low-energy endohedral (filling) or exohedreal (intercalation) doping to higher energy substitutional doping. The SPS process is a promising candidate for post-growth doping of CNTs. Recently, Hitchcock et al. have shown that a post-growth treatment of SWNT bundles with boric acid solution followed by SPS process is a viable method for doping CNTs and controlling their carrier concentration. Using boric acid concentration as the control parameter, they were able to alter the carrier concentration of the SPSed CNT networks. As a result, the resistivity, thermopower, and Hall coefficient measurements show a doping-induced shift of the Fermi level and the accompanied changes in electron–phonon coupling [17].
T
T a
2786
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
b
0.12 rRT (W-cm)
As prepared
Type III
SPS 500°C
rRT µ (TSPS – TTh )-1.3
0.06
SPS 800°C SPS 900°C
0.04 0.02
rN
rRT (W-cm)
0.08
Type II
1.4
0.08
0 600
800
0.04
SPS 1000°C
TSPS
SPS 1200°C
1.2
SPS 1500°C
1000 1200 1400 TSPS (∞C)
Type I
0.00
1.0 400
800 TSPS (∞C)
c
1200
16 12
S (mV/K )
1600
200
100
300
T (K ) dS/dT (mV/K 2)
0
0.04
0.02 0
8
400 800 1200 1600 TSPS (∞C)
4 As prepared SPS 800°C SPS 1000°C
0 −4 0
100
200
SPS 500°C SPS 900°C SPS 1500°C
300
T (K )
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 4 (a) Room temperature resistivity value as a function of SPS temperature, the inset shows the power law fitting to a percolative model; and (b) The crossover in the temperature dependence of normalized resistivity, type I,
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 5 SPS-induced defects in MWNTs (a) collapse of a MWNTs (b) kink formed on the wall a MWNT
II and III, reflects the change of the nature of inter-tube junctions, the normalization is performed by dividing the r(T) by its corresponding room temperature rRT; (c) The temperature dependence of thermopower of the sample SPSed at 1,500 C shows a strong phonon drag peak below 50 K [13]
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
Altering the Morphology of Tubes (Beyond the Soft Sintering) At high SPS temperatures (>1,500 C), it is possible to alter the morphology of CNTs. In an extreme case, it is possible to “unzip” CNTs to form graphitic nanostructures (Fig. 6) [18, 19]. This has been proposed as a novel method for generating large amounts of graphene in a short amount of time. Under the proper conditions, SPS can also change the sp2 bonds of CNTs to sp3 bonds, forming diamond [8]. Shen et al. [20] have shown that using the SPS, it is possible to grow diamond particles up 100 mm in diameter at very low pressure (80 MPa). It is noteworthy that SPS under a moderate pressure of 10–100 MPa leads to the conversion of CNTs to diamonds, whereas a relatively higher pressure in the GPa range is usually required for diamond formation by other methods. It is proposed that the nucleation of diamond is from the core of carbon nano-onions, which is an intermediate product from the breakage of carbon–carbon bonds and unzipping CNTs [8]. The nano-onions behave as compression cells (“self-compression”) to help nucleate and grow diamonds. Hence the SPS process provides for a new path for diamond synthesis, which could offer many applications in industry due to the exceptional hardness and thermal conductivity of diamond.
Conclusion Since their discovery, carbon nanotubes have been a topic of much research due to their amazing physical properties. However, in order to take CNTs from the lab to applications, scalable techniques must be developed to tune their properties. While it is possible to alter the structure and morphology of CNTs with very high precision using an irradiation technique, this method involves microscopic control parameters and the throughput is very low. In this regard, the SPS technique utilizes high-density current without the need to focus or align the electron beam, the control parameters of SPS are macroscopic, and the throughput is very high. One can envisage a combination of these two techniques to give some control of the structure and properties of CNTs at multiple length scales, which is a very promising approach to move CNTs from materials used solely in research to viable materials with real-world applications.
2787
T
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering, Fig. 6 Graphitic nanostructures formed in SPS by unzipping MWNTs [11]
Acknowledgments D. H., J. H., and A. M. R. acknowledge the support from a DOE/EPSCoR Implementation Grant (No. DEFG02-04ER-46139) and the SC EPSCoR Office/Clemson University cost sharing. K. Y. and A. M. R. also acknowledge the support through U.S. AFOSR under Grant No. FA9550-09-10384. All authors thank Prof. Malcolm Skove for scientific discussions and proofreading of the manuscript.
Cross-References ▶ Carbon Nanotubes
References 1. Saito, R., Dresselhaus, G., Dresselhaus, M.S.: Physical Properties of Carbon Nanotubes. Imperial College Press, London (1998) 2. Ma, R.Z., Xu, C.L., Wei, B.Q., Liang, J., Wu, D.H., Li, D.J.: Electrical conductivity and field emission characteristics of hot-pressed sintered carbon nanotubes. Mater. Res. Bull. 34, 741–747 (1999) 3. Rubio, A., Apell, S.P., Venema, L.C., Dekker, C.: A mechanism for cutting carbon nanotubes with a scanning tunneling microscope. Eur. J. Phys. B. 17, 301–308 (2000) 4. Charlier, J.C.: Defects in carbon nanotubes. Acc. Chem. Res. 35, 1063 (2002)
T
T
2788
Tuning Electrical Properties of Carbon Nanotubes via Spark Plasma Sintering
5. Krasheninnikov, A.V., Banhart, F.: Engineering of nanostructured carbon materials with electron or ion beams. Nat. Mater. 6, 723 (2007) 6. Munir, Z.A., Anselmi-Tamburini, U., Ohyanagi, M.: The effect of electric field and pressure on the synthesis and consolidation of materials: a review of the spark plasma sintering method. J. Mater. Sci. 41, 763 (2006) 7. Song, X., Liu, X., Zhang, J.: Neck formation and selfadjusting mechanism of neck growth of conducting powders in spark plasma sintering. J. Am. Ceram. Soc. 89, 494–500 (2006) 8. Zhang, F., Shen, J., Sun, J., Zhu, Y.-Q., Wang, G., McCartney, G.: Conversion of carbon nanotubes to diamond by spark plasma sintering. Carbon 43, 1254 (2005) 9. Qi, X., Bao, Q., Li, C.M., Gan, Y., Song, Q., Pan, C., Tang, D.Y.: Spark plasma sintering-fabricated one-dimensional nanoscale “crystalline-amorphous” carbon heterojunction. Appl. Phys. Lett. 92, 113113 (2008) 10. Gruner, G.: Carbon nanotube films for transparent and plastic electronics. J. Mater. Chem. 16, 3533 (2006) 11. Yang, K., He, J., Su, Z., Reppert, J.B., Skove, M., Tritt, T. M., Rao, A.M.: Inter-tube bonding, graphene formation and anisotropic transport properties in spark plasma sintered multi-wall carbon nanotube arrays. Carbon 48, 756 (2010) 12. Hone, J., Llaguno, M.C., Nemes, N.M., Johnson, A.T., Fischer, J.E., Walters, D.A., Casavant, M.J., Schmidt, M., Smalley, R.E.: Electrical and thermal transport properties of magnetically aligned single wall carbon nanotube films. Appl. Phys. Lett. 77, 666–668 (2000)
13. Yang, K., He, J., Puneet, P., Su, Z., Skove, M.J., Gaillard, J., Tritt, T.M., Rao, A.M.: Tuning electrical and thermal connectivity in multiwall carbon nanotube buckypaper. J Phys Condens Matter 22, 334215 (2010) 14. Zhang, H.-L., Li, J.-F., Zhang, B.-P., Yao, K.-F., Liu, W.-S., Wang, H.: Electrical and thermal properties of carbon nanotube bulk materials: experimental studies for the 328–958 K temperature range. Phys. Rev. B 75, 205407 (2007) 15. Duclaux, L.: Review of doping of carbon nanotubes (multiwalled and single-walled). Carbon 40, 1751–1764 (2002) 16. Bangert, U., Bleloch, A., Gass, M.H., Seepujal, A., van der Berg, J.: Doping of few layered graphene and carbon nanotubes using ion implantation. Phys. Rev. B 81, 245423 (2010) 17. Hitchcock, D., Yang, K., He, J., Rao, A.M.: Electrical transport properties of single-walled carbon nanotube bundles treated with boric acid. Nano: Brief Rep. Rev. 6(4), 337– 341 (2011) 18. Gutirrez, H.R., Kim, U.J., Kim, J.P., Eklund, P.C.: Thermal conversion of bundled carbon nanotubes into graphitic ribbons. Nano Lett. 5(11), 2195–2201 (2005) 19. Kim, W.S., Moon, S.Y., Bang, S.Y., Choi, B.G., Ham, H., Sekino, T., Shim, K.B.: Fabrication of graphene layers from multiwalled carbon nanotubes using high dc pulse. Appl. Phys. Lett. 95, 083103 (2009) 20. Shen, J., Zhang, F.M., Sun, J.F., Zhu, Y.Q., McCartney, D.G.: Spark plasma sintering assisted diamond formation from carbon nanotubes at very low pressure. Nanotechnology 17, 2187–2191 (2006)
U
Ultrahigh Vacuum Chemical Vapor Deposition (UHVCVD) ▶ Chemical Vapor Deposition (CVD)
Ultralarge Strain Elasticity ▶ Superelasticity and the Shape Memory Effect
Ultraprecision Machining (UPM) Suhas S. Joshi Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai, Maharastra, India
Synonyms Nanomachining; Nano-mechanical machining
Definition Ultraprecision machining refers to the ultimate ability of a manufacturing process wherein processing of a material at its lowest scale that is, at the atomic scale, is achieved. It is known that the lattice distances between two atoms are of the order of 0.2–0.4 nm; therefore, the ultraprecision machining refers to processing or removal actions of a manufacturing
process in the vicinity of 1 nm. The process is also referred as “atomic bit” processing. To remove or process atomic bits, extremely large energy density is required, which is equivalent to the atomic bonding energy. The conventional cutting tools neither have high strength to sustain high specific cutting energy nor have hardness to sustain the tool wear. Therefore, ultraprecision machining refers to use of single crystal diamond (SCD) tools for ultrafine cutting or very fine abrasives for lapping or polishing. It may also refer to use of high-energy elementary particles like photons, electrons, ions, and reactive atoms to undertake removal [1].
Historical Perspective In early 1970s, Japanese researcher N. Taniguchi illustrated historical evolution of precision in manufacturing through “Taniguchi curves” [1]. It was realized that it takes about 20 years to improve precision by one decimal point; see Table 1 depicting the progression of precision. He extrapolated the curves further to postulate that the resolution of machining processes would reach to 1 nm in the year 2000 [1]. True to the expectations, today, a number of ultraprecision machining processes are available.
Applications Ultraprecision processing is required in the manufacture of high-precision block gages, diamond indenters and tools, 3D metallic mirrors, etc., in the mechanical
B. Bhushan (ed.), Encyclopedia of Nanotechnology, DOI 10.1007/978-90-481-9751-4, # Springer Science+Business Media B.V. 2012
U
2790
Ultraprecision Machining (UPM)
Ultraprecision Machining (UPM), Table 1 Evolution of ultraprecision Year Resolution (mm)
1900 >10
1920 5
1940 0.5
1960 0.05
1980 0.005
2000 0.001
domain; Si wafers, ICs memory, thin film, ULSI devices in electronics field; and optical flats, diffraction gratings, mirrors, and aspherical lenses in the optical field.
Types of UPM Processes The ultraprecision machining processes can be broadly classified into three categories [1]: 1. Nanomechanical processing: These processes use either SCD tools to perform diamond turning or very fine abrasives in the bonded or loose form to perform nanogrinding, nanolapping, or nanopolishing. The other processes in this category include progressive mechanochemical polishing for Si wafers, pitch polishing for aspherical lenses. 2. Nanophysical processing: It involves use of highenergy elementary particles like photons, electrons, ions, and reactive atoms to perform direct ablation of substrate or carry out lithography. 3. Nanochemical or electrochemical processing: These processes involve use of chemical or electrochemical principles to effect material removal. They are chemical milling, photochemical machining, and other processes. In the following sections, principles of nanometric removal in ultraprecision machining processes are elucidated along with their applications.
Nanomechanical Processing In these processes, the processing unit, defined as the size of chip generated in a single stroke of tool, should be of the order of single or subatomic magnitude. Principle of Ultraprecision Removal The removal of atomic bits in nanomechanical processes requires specific cutting energy of varying degree as the scale of the process decreases and cutting tools with certain sharpness and hardness as discussed below:
Ultraprecision Machining (UPM), Table 2 Size-effect in processes as a function of chip thickness Process Process scale Tension test Multicrystal grain Turning Subcrystal grain Precision Subcrystal milling grain Grinding Atomic cluster region
Chip thickness (mm) 300–500 40–50
Resisting shear stress (N/mm2) 300 500
5–10
1,000
0.5–1
10,000
Size-Effect Implications
As the processing scale reduces from simple tension testing to grinding, where submicrometric chips are generated, the specific shear or breaking energy becomes extremely large. This effect is called as size effect. See Table 2 showing processing scales and shear energies [2]. Extending this trend further, when the machining scale reaches an atomic-bit, the specific shear energy reaches atomic bonding energy [1] or the theoretical shear strength of a defect-free material given by tth ¼
G 2p
(1)
For carbon steel with the modulus of rigidity, G ¼ 82 GPa, the theoretical shear strength is given by – pthCarbonSteel ¼ 13GPa
(2)
The reason for increasing the specific shear energy is attributed to the kind of defects available in the crystalline structure at that processing scale. At the atomic bit scale (0.001 mm), the energy required is comparable to the theoretical shear strength (105 J/cm3) [1] (Fig. 1). At the atomic cluster scale, deformation of a material can take place only with the help of point defects (Fig. 1) and requires energy of the order of 103–104 J/cm3 [1]. Nevertheless, at some higher scale (0.1–10 mm), called as subcrystal grain size, 2D or 3D defects in the crystalline structure cause a significant reduction in the specific shear energy to 102–103 J/cm3 (Fig. 1) [1]. In the crystal grain size range, (>10 mm), all kinds of dislocations and defects at grain boundaries lower the processing energy further to 101–102 J/cm3 [1]. Unlike the case of the ductile (metallic) materials discussed above, the brittle materials have a network of
Ultraprecision Machining (UPM)
U
2791
Ultraprecision Machining (UPM), Fig. 1 Defect distribution and scale of specific shear energy required
105
Atomic cluster processing
Atomic scale processing
Sub-grain ductile processing
Sub-grain brittle processing
Multi-grain Ductile/brittle processing
1 µm
104
Specific shear energy (J/Cm3)
Vacancy Lattice atom
103
Interstitial atom
0.2-0.4 nm
Crystal grain void range > 10µm
Microcrack
Movable dislocation Fixed dislocation 1 µm
Precipitate
102 Atomic lattice range < 0.1nm
Point defects range 1nm – 0.1µm
1 µm
Movable dislocation
Dislocation Microcrack range range 0.1µm0.1µm10 µm 10 µm Cavity/crack
10
0.001µm
0.1µm
Grain boundary
10µm
Scale of crystalline structure defects
microcracks in place of the network of dislocations (Fig. 1). In the ductile materials, below the point defect scale (1 nm to 0.1 mm) (Fig. 1), and in the brittle materials below the microcracks scale (0.1–10 mm) (Fig. 1), the specific shearing energy is equal to the theoretical shear strength of the work material. Cutting Tool Material and Geometry Implications
The cutting tool edge in ultraprecision machining being subjected to extremely high cutting pressure should withstand not only the pressure but also the wear. In addition, the edge should have high degree of sharpness to perform atomic bit removal. The SCD tools, possessing the desired properties, are the most suitable tool materials for ultraprecision machining. However, their limitation stems from the achievable cutting edge sharpness, which normally is limited to hundreds of nanometers. Therefore, in UPM, several atomic layers or atomic clusters are processed than atomic bits, primarily on softer metallic materials like Al and Cu. On the other hand, on hardened steels, and other brittle materials, SCD tools undergo rapid wear, thereby necessitating the use of diamond abrasive process like nanogrinding.
Ultraprecision Machining Systems It uses time-tested principles of precision engineering along with (1) the recent advancements in the control, feedback, and drive systems, (2) CAD/FEM for designing, followed by (3) fine-tuned assembly as well as system integration. This combination results in a machine that is thermally stable, highly reliable, and flexible and at the same time faster in response [3]. Several features of an ultraprecision machine are summarized in Table 3. Ultraprecision Machines and Processes A number of ultraprecision machines and processes have evolved using nanomechanical removal principles. They are discussed below. Single Point Diamond Turning
It is a nanomechanical processing machine that was developed primarily for 3D shaping and surface finishing of soft metals such as copper and aluminum and polymers. The machine with the features listed in Table 3 uses very sharp SCD tools to perform ultraprecision cutting. The chips formed during the process are extremely thin, which indicates ductile
U
U
2792
Ultraprecision Machining (UPM)
Ultraprecision Machining (UPM), Table 3 Salient features of an ultraprecision machining system Machine elements/purpose/ types Machine base • Provides thermal and mechanical stability, damping characteristics Work spindle • Spindle motion errors significantly affect surface quality and accuracy of machined features. • Use aerostatic or hydrostatic, recent grooved air bearings
Main features • Made of cast iron, natural or epoxy granite, polymer concrete • Both the spindle types have high rotational accuracy and rotational speeds • Aerostatic spindles are for low/medium loads, hydrostatic bearings take heavy loads
Drives • Slide drives provide stiffness, • Servo drives are used in acceleration, speed, contouring operations smoothness of motion, • Small and precise motions of accuracy, and repeatability tools for tool positioning and • Spindle drives are usually fine motion are achieved by AC/DC motors piezoelectric actuators • Slides are usually provided with linear motor or friction drives Controls • Controls are required for • A multiaxes CNC controllers linear and rotary drives, are used limiting, position, and time • PC-based controls are used switches, sensors. more recently • They also control thermal, • Feedback controls have a geometrical, and tool setting resolution of nm or sub-nm errors. Measurement and inspection systems • Provides rapid and accurate • Online measurement and positioning of cutting tool error compensation towards work surface • Laser interferometer for tool • It also monitors the tool position control wear condition
deformation of the material during machining. At the same time, the process leaves extremely thin degenerated layer of the work surface as the process initiates shear slip using dislocations (Fig. 1). Applications of the process include fabrication of metallic mirrors, precision component fabrication in laser, and space and optics fields. A typical diamond turning lathe has T-axis configuration (Fig. 2a). The main machine slide moves along x-axis, and the cutting tool is moves along Z-direction. Nanogrinding
Single-point diamond turning machine is usually provided with add-on grinding attachment (Fig. 2b) for
nanogrinding. This way, harder work materials can also be processed on the machine using the grinding attachment. However, it increases the stiffness requirement of the machine. In nanogrinding process, shear slip is generated due to very low (tens of nm) depth of cut (Fig. 1). This causes crackless ductile failure of hard/brittle materials under the abrasive grains in the diamond grinding wheels. Since the working stress on the abrasives is extremely high, only fine-grained diamond abrasives can be used for this process. The process can achieve mirror like finish on hard and brittle materials like glasses and ceramics. Its other applications involve grinding of mould inserts, glass lenses, and aspherical lenses. Electrolytic In-process Dressing (ELID) Grinding
It is a type of nanogrinding process in which metalbonded diamond grinding wheels are used. The process uses an electrolytic solution as a grinding liquid. Upon application of voltage, anodic dissolution of bonding metal from the grinding wheel occurs. This ensures that worn abrasive particles on the grinding wheel are removed quickly (a wheel dressing action), thereby ensuring continuous availability of sharp particles for the grinding process. The sharp particles initiate shear slip at low depth of cut (Fig. 3) causing crackless material removal. The process therefore is capable of achieving mirror like surface finish on glasses and ceramics. Nanolapping and Nanopolishing
Unlike nanogrinding, which uses fixed abrasives, these processes use free abrasives to achieve mirror-like finish. The processes involve atomic cluster processing. Nanopolishing involves sliding of soft abrasives between a soft polishing pad and work surface (Fig. 3a). The abrasives are usually soft and dull; they include Fe2O3, Cr2O3, CeO2, or MgO [1]. The abrasives get embedded in the soft pad and originate shear slip at the point defects to effect removal [1] (Fig. 3a). This ensures smooth polishing of hard metallic surfaces. The nanolapping on the other hand uses a medium to hard pad besides extremely sharp and hard abrasives such as diamond, CBN, SiC, SiO2, or B4C [1]. The fine and hard abrasives initiate cracking at the point defects to effect removal [1] (Fig. 3b). This helps generate mirror-like finishing on hard and brittle materials.
Ultraprecision Machining (UPM) Ultraprecision Machining (UPM), Fig. 2 Typical schematics of (a) Single-point diamond turning (b) Diamond grinding machine
2793
a
U
b Workpiece
Workpiece
Grinding wheel
Single point diamond tool
Y-Table
Y-Table X-Table
X-Table
a
b
Soft and dull abrasive particles
Soft polishing plate
Ultraprecision Machining (UPM), Fig. 3 (a–b) Nanopolishing and nanolapping
Lapping plate
Hard workpiece Point defect
Sharp edged and hard abrasives
Hard workpiece Nano-cracking layer
Nano-shear slip
P-MAC
Aspherical Lens Polishing
The progressive mechanical-chemical (P-MAC) polishing is a hybrid nanopolishing process that uses mechanical as well as chemical action to effect removal to achieve mirror-like finish and damagefree polishing of Si wafers. The process uses ultrafine SiO2 abrasives (size 10 nm) suspended in an alkaline (pH 10) solution along with an artificial leather foam to perform polishing in four stages. The abrasive particle size and the polishing pressure and the stock removal are gradually reduced after each stage [1]. In P-MAC, the greater the chemical action, the higher is the polishing efficiency and the lower is the damaged layer [1] (Fig. 4). However, the increased chemical action reduces control over accuracy (flatness), Fig. 4 [1].
Aspherical profiles are required at the edges of the lenses to ensure that light beams from the lens edges converge at the focal point. Therefore, the aspherical lens polishing involves polishing over free-form surfaces. The process is often required for polishing of highly accurate X-ray optics and lenses in commercial cameras. It involves gradual removal using a pitch polishing tool that has very small contact area as compared to the dimensions of the workpiece. The polishing rate is governed by Preston’s rule of thumb [1] h ¼ a v Dt
(3)
The pitch polishing tool acts as a pad and supports the abrasives. It remains solid at room temperature but
U
U
2794
Ultraprecision Machining (UPM)
Depth of damaged layer
Flatness
Arbitrary units
Polishing rate
0
0 100 0 100 Chemical action ratio
100
Ultraprecision Machining (UPM), Fig. 4 General processing characteristics of P-MAC
flows with time when heated or under pressure. The composition of the pitch material is usually proprietary of various manufacturers but contains materials like tar, oil, wood, paraffin wax, shellac, and so on. The properties of pitch material include [1] viscosity in the range of 107–109 Pa·s), softening point, (55–7 C), penetration hardness (60–80 by Shore D). In this process, the surface roughness tends to improve as the penetration number of the pitch increases, i.e., as the pitch becomes softer. It is seen that CeO2 and ZrO2 give better surface finish than super hard diamond abrasives.
Nanophysical Processes The principle of removal in these processes involves photothermal or photochemical interactions with the work surface. The photothermal interaction causes melting and vaporization of work material due to energy of photon or laser beams. On the other hand, the photochemical interaction involves use of photon energy to break bond between work elements [1]. It is understood that usually laser beams with metallic materials perform removal action by the photothermal principle. However, when the polymeric materials interact with laser beams, photochemical reaction takes place. With the ceramics, the interaction is partly by the photothermal and partly by photochemical reactions.
The photon beams owing to their wavelength of the order of 10–0.1 mm and the photon energy of the order of 0.1–10 eV cause thermal energy transfer that is smaller than the atomic bonding energy of materials. Hence, the photon beams are not suitable for nanometric material removal. Similarly, an electron beam due to small size (2.8106 nm), mass (91031 kg), and low energy of several hundred kilovolts penetrate deep (several micrometers) into the work surface. Therefore, the electron beams are also not suitable for nanometric material removal. Unlike the electrons, most of the ions interfere with the surface atoms of a workpiece because their mean diameter is 0.1 nm and the mean atomic distance is 0.3 nm. Consequently, the projected energized ions frequently collide with the nuclei of atoms of the workpiece and knock out or sputter the surface atoms. Hence, ion beams are most suitable for atomic bit removal. The process is also called as ion sputter machining. In the process, the electrically accelerated inert-gas ions such as Ar ions with average energy of 10 keV (which corresponds to a speed of 200 km/s) are unidirectionally oriented and projected on to the workpiece surface in a high vacuum (1.3104 Pa) [1]. Hence, they adhere firmly on the target surface. A beam of focused ions can also be used for cutting of extremely hard materials like diamond or sharpening of SCD tools.
Nanochemical or Electrochemical Processing The chemical reactions inherently involve atomic bit processing. A chemical reaction is a change in the atomic combination of reacting molecules, in which the atomic bonding of a reacting molecule is broken and a new molecule is generated. In the process, a chemically reactive gas or liquid is applied to a specified position on the solid surface of the workpiece. The reacted molecules are removed or diffused into the surrounding reacting gas or liquid. If the reacted molecules are insoluble or not in vapor form, chemically reactive deposition occurs on the workpiece surface. But if the reacted or reagent molecules diffuse into the surface layers of the workpiece and react with the atoms or molecules there, they perform a chemically reactive surface treatment [1]. The dimensional accuracy obtainable in chemical reactions is in the
Ultrashort Carbon Nanotubes
nanometer range when the processing conditions are stable. In addition, the following aspects are necessary for nanometric processing [1]: 1. In-process measurement and feedback control of position of the processing point and control of the processed volume (area) are necessary. However, it is difficult to realize these in practice. In the photoresist method, control of the processing point position or area is achieved by the patterned mask. 2. The control of the processing volume or depth can be done only by adjusting the processing time and flow rate of etchants. In the electrochemical process, the basic reaction is the same, but activation energy for the reaction is given by electric field potential and differs from the ordinary activation energy of the chemical reactions based on the thermal potential energy.
Summary Ultraprecision machining uses mechanical, physical, chemical, or electrochemical sources of energy for effecting material removal to the nanometric scale. Grain size, load applied on the grains, and the type of work material govern removal in nanomechanical processes. Size of the energy beam particle governs the processing resolution in nanophysical processes. Inprocess measurement and feedback control are essential for the nanochemical or electrochemical processes.
Cross-References ▶ Electrochemical Machining (ECM) ▶ Nanochannels for Nanofluidics: Fabrication Aspects ▶ Nanotechnology
References 1. Taniguchi, N. (ed.): Nanotechnology – Integrated Processing Systems for Ultra-precision and Ultra-fine Products. Oxford University Press, New Delhi (2008) 2. Baker, W.R., Marshall, E.R., Shaw, M.C.: The size effect in metal cutting. Trans. ASME 74, 61 (1952) 3. Lou, X., Cheng, K., Webb, D., Wardle, F.: Design of ultraprecision machine tools with applications to manufacture of miniature and micro components. J. Mater. Process. Technol. 167, 515–528 (2005)
2795
U
Ultrashort Carbon Nanotubes Lesa A. Tran and Lon J. Wilson Department of Chemistry and the Richard E. Smalley Institute for Nanoscale Science and Technology, Rice University, Houston, TX, USA
Synonyms US-tubes
Definition Ultrashort carbon nanotubes (US-tubes) are 20– 100 nm segments of single-walled carbon nanotubes (SWNTs).
Overview In recent years, single-walled carbon nanotubes (SWNTs) have been extensively studied for their use in biomedical applications. Owing to their peculiar physical and chemical properties, SWNTs have become a widely used platform for the design of medical therapeutic and diagnostic agents [1, 2]. Nevertheless, the biocompatibility and biodistribution of SWNTs still remain uncertain and are under current investigation. The ideal length of SWNTs for biomedical applications has yet to be determined, but it has been suggested that discrete, individualized SWNTs of shorter ( 1 μm
Ultrashort Carbon Nanotubes
b
Ultra-Short Carbon Nanotubes (US-tubes) 20–100 nm
Ultrashort Carbon Nanotubes, Fig. 1 Illustration of (a) a single-walled carbon nanotube and (b) a collection of ultrashort carbon nanotubes (with sidewall defects) derived from a single-walled carbon nanotube
functional groups or defect sites, and (2) the removal of or the cutting at these sidewall modifications. While the predominant mechanism of chemical SWNT cutting is debatable, there are two proposed methods under consideration: (1) the “fuse burning” mechanism, with carbon etching occurring at the bondstrained tube ends, and (2) defect site propagation, with cutting occurring at the random defect sites on the SWNT walls. Both oxidizing and nonoxidizing methods to chemically cut SWNTs have been explored. Various oxidative processes using different strong oxidizing agents (HNO3, H2SO4, and H2O2) have been proposed for the cutting of SWNTs. However, the efficiency and length distributions of these synthetic methods vary with the type and concentration of the oxidizing agents used, the reaction temperature and time, the method of dispersion during the reaction (refluxing or sonication), and the length and diameter of the SWNTs being cut. Although more severe reaction conditions result in more reaction exposure of
Ultrashort Carbon Nanotubes, Fig. 2 Atomic force microscopy (AFM) images of (a) purified SWNTs and SWNTs after piranha treatment at 22 C for (b) 1 h, (c) 3 h, and (d) 9 h (Reproduced with permission from [4])
SWNTs and, thus, shorter lengths of resulting US-tubes, less severe reaction conditions are preferred in order to minimize amorphous carbon production. For instance, it has been shown that 4:1 (vol/ vol 96% H2SO4/30% H2O2) piranha solutions are able to attack existing damage sites on the nanotube sidewalls to chemically shorten SWNTs [4]. While hot (70 C) piranha solutions can induce more damage sites, room-temperature (22 C) solutions can react with existing sidewall defects but are incapable of initiating further damage sites, resulting in minimal carbon loss and slow etch rates (Fig. 2). On the other hand, high-temperature conditions produce increasingly shorter US-tubes with increasing reaction time, with approximately half of the starting material destroyed due to chemical etching. Therefore, roomtemperature piranha solutions offer the ability to exploit active damage sites in the sidewalls of the nanotube in a controlled manner without the counterproductive destruction of the nanotubes. The aggressive oxidation of SWNTs with oleum (100% H2SO4 with 20% SO3) and nitric acid can simultaneously shorten and carboxylate the SWNTs, rendering water-soluble US-tubes less than 60 nm in length [5]. Purified SWNTs are first dispersed in oleum
Ultrashort Carbon Nanotubes
to disentangle individual SWNTs from their rope-like bundles. Next, the SWNT-oleum dispersion is extensively oxidized with strong HNO3 at elevated temperatures to cut the SWNTs into US-tubes. The oxidation of the SWNTs also forms carboxylic acid groups at the sidewall defects of the SWNTs, making US-tubes that are up to 2 wt% soluble in polar organic solvents, acids, and water. One nonoxidative process to synthesize US-tubes involves the chemical shortening of SWNTs to lengths of less than 50 nm by a fluorination/pyrolysis treatment [6]. Briefly, purified SWNTs are exposed to fluorine gas, during which fluorine atoms tend to arrange around the circumference of the nanotube to form band-like regions along the nanotube sidewalls. Upon pyrolysis of the fluorinated SWNTs at 1,000 C in an inert atmosphere, the chemically bound fluorine atoms are driven off in the form of CF4 and COF2. This leaves behind chemically cut US-tubes with partial sidewall defects along the nanotube axis. Mechanical Cutting
There have been various documented methods to mechanically cut SWNTs into shorter segments, including ultrasonication, grinding, and ball milling. However, these techniques lack length control; such mechanical methods create extensive sidewall damage and are not able to achieve sub-100 nm lengths. Therefore, more precise mechanical cutting techniques have been developed. Microlithography can be used to cut SWNTs into US-tubes [7]. This technique involves a protective photoresist polymer layered on top of a SWNTcovered solid substrate in a patterned manner to protect portions of nanotubes from lithographic damage and modification. Reactive ion etching using an oxygen plasma is then used to remove the unprotected sections of the SWNTs. After etching, the newly formed nanotube ends remain, with dangling carboxylic acid and ether groups. The photoresist polymer can then be removed, leaving cut segments of SWNTs. Regardless of whether the SWNTs are previously aligned or properly dispersed prior to cutting, microlithography offers a straightforward method to shorten SWNTs with a narrow and selective length distribution. Electron beam etching has also been used to cut SWNTs with nanometer precision [8]. Using electron energy beams exceeding 100 keV, which is made
2797
U
Ultrashort Carbon Nanotubes, Fig. 3 Transmission electron microscopy (TEM) image of the complete cutting of a SWNT bundle into US-tubes using electron beam etching (Reproduced with permission from [8])
possible by currently available electron microscopes, partial and complete cuts into SWNTs can be made with great control without the need of covering portions of SWNTs with a protective layer. Similar methods have been developed using high voltagebearing scanning tunneling microscope (STM) tips. Interestingly, upon cutting into a SWNT bundle, the newly formed tube ends immediately form hemispherical caps, enhancing the stability of the cut US-tubes (Fig. 3). Another peculiar observation is that forming a second cut within 10–50 nm from a preexisting cut proves more difficult to perform. This increase in stability is thought to be a result of the generation and annealing of vacancy-interstitial atom pairs. More recently, an ultramicrotome has been used to cut frozen layers of magnetically aligned SWNT membranes, or Buckypapers [9]. Enough mechanical control was obtained to create open-ended US-tubes with minimal sidewall damage, typically seen after chemical oxidative methods. When the Buckypaper was cut into 50 nm segments, the mean length was about 87 nm, with about 60% of the tubes ranging from 50 to 150 nm. However, the large pressure exerted by the diamond knife tip in its proximity can cause deformation and cracking of the nanotubes. Individualization of US-tubes Like their SWNT precursors, US-tubes have a large tendency to bundle together, which renders it difficult to form single-tube suspensions. Although it is
U
U
2798
Ultrashort Carbon Nanotubes
Ultrashort Carbon Nanotubes, Fig. 4 AFM image of individualized UStubes after chemical reduction (Reproduced with permission from [10])
a
b
nm
2.00
1.1 nm
1.6 nm
2.1 nm
0
−2.00 0
possible for full-length SWNTs to disperse in solution with the aid of polymers and surfactants, US-tubes cannot disperse as readily in these same conditions [10]. Because of their high aspect ratios, SWNTs are relatively flexible and can “peel” more readily from each other upon introduction to surfactant wrapping. On the other hand, US-tubes have a much lower aspect ratio, therefore making them more rigid and more difficult for surfactant molecules to insert between individual US-tubes. Therefore, chemical means to debundle and individualize US-tubes have been explored to better disperse US-tubes. Reduction of US-tubes
Chemical reduction can be used to individualize, or debundle, US-tubes [10]. Upon sonication with K0 metal in THF, the negative charge imparted on the US-tubes electrostatically overcomes the bundling forces to yield individualized US-tubes (Fig. 4). It is thought that the individualization of US-tubes via chemical reduction increases US-tube dispersion in solution, thus allowing for more even chemical functionalization.
1.00 μm
2.00
Functionalization of US-tubes
The chemical functionalization of US-tubes is thought to occur at carboxylic acid end groups of the sidewall defects. This allows for the attachment of various chemical groups that would render the US-tubes more soluble in different solvent systems. One functionalization method utilizes an in situ Bingel-Hirsch (CBr4/DBU) cyclopropanation to functionalize US-tubes with the water-soluble malonic acid bis-(3-tert-butoxycarbonylaminopropyl)ester to yield four to five adducts per nm [10]. Another documented functionalization method involves cyclopropanation using diazoacetic ester in the presence of Rh6(CO)16 [11]. This allows for the US-tube to be readily reactive with primary amines, thus making possible the functionalization of US-tubes with a number of amino acid derivatives, such as water-soluble serine, with 3–19 groups per nm. Other biologically relevant moieties, such as antioxidants [12] and oligonucleotides [13], have been successfully attached onto US-tubes and are described below. Alkylation-based reduction can also be performed on US-tubes to improve nanotube dispersion in nonpolar environments [14].
Ultrashort Carbon Nanotubes
Loading of US-tubes In addition to being able to chemically functionalize the sidewalls of US-tubes, the hollow interior of the US-tubes can be loaded or filled with various ions and small molecules, such as Gd3+ ions [15], molecular I2 [16], and 211AtCl radionuclides [17]. This property is especially relevant for drug design, as the carbon structure of the US-tube serves as an encapsulating sheath to protect potentially toxic moieties from their surrounding environment. Although the loading mechanism is not completely clear, it is likely that ions and small molecules load into the US-tubes through the sidewall defects.
Phenomena Toxicology Before US-tubes are considered for biomedical applications, their toxicological profiles must be known. A recent study documented the toxicology as well as the acute and subchronic effects of high doses (50–1,000 g/kg bodyweight) of Tween-suspended US-tubes in Swiss mice [3]. Oral administration of US-tubes did not result in death or any behavioral abnormalities for up to 14 days. Intraperitoneal administration allowed for the US-tubes to reach systemic circulation and various tissues through the lymphatic system. As opposed to unfunctionalized SWNTs, wellindividualized US-tubes had the ability to pass through the reticuloendothelial system, excreting via the kidneys and bile ducts. High doses of US-tubes can induce strongly adherent granuloma formation on and inside organs due to large (>10 mm) nanotube aggregation into fiber-like structures. However, small (
Wetting Transitions, Fig. 2 Wetting transitions observed by vibration of 15 ml water drop deposited on the micrometrically rough PDMS surface: (a) The initial Cassie state. (b) The Cassie impregnating state induced by vibrations
the characteristic scale of the surface heterogeneities. The rigorous thermodynamic derivation of Eqs. 3–6 may be obtained within the general variational approach [12]. The physical mechanisms of WT on flat and rough surfaces are quite different. WT on rough surfaces are related to the penetration of liquid into the grooves constituting the relief and result in the change of the APCA (see Fig. 2) [13–20]. Wetting transitions were observed under various experimental techniques utilizing a diversity of factors: gravity, pressure, bouncing and evaporation of droplets, electric field in the electrowetting experiments, and vibration of droplets [13–18]. WT were observed with the use of various experimental techniques, including reflection interference contrast microscopy and environmental scanning electron microscopy [20].
Energetic Approach to Wetting Transitions Various wetting states featured by very different APCA can coexist on the same heterogeneous surface. The diversity of APCA could be easily understood if one takes into account that the Gibbs energy curve for a droplet on a real surface is characterized by multiple minima points. It could be shown that the Wenzel state (Fig. 1b) is energetically favorable compared to the pure Cassie air trapping one, shown in Fig. 1a, when the inequality (7) holds: fS 1 cos y > r fS
(7)
The Wenzel state will be energetically favorable when compared to the mixed trapping state depicted in Fig. 1d, when the inequality (7a) holds:
f 1 r rf f
W (7a)
Obviously for hydrophilic materials, the Wenzel state is always favorable in relation to the Cassie one, whereas for the hydrophobic materials the situation is more complicated. When the hydrophobic material is very rough (r>>fs ), the Cassie state will be favorable for any obtuse y. For single-scale hydrophobic materials, the Cassie air-trapping wetting state usually corresponds to the highest of the multiple minima of the Gibbs energy of the droplet deposited on a rough surface. Thus, for the WT, the energy barrier should be surmounted [18]. It was supposed that this energy barrier is equal to the surface energy variation between the Cassie state and the hypothetical composite state, with the almost complete filling of surface asperities by liquid, keeping the liquid-air interface under the droplet and the contact angle constant, as shown in Fig. 3. Wetting of hydrophobic side surfaces of asperities constituting the relief is energetically unfavorable; this gives rise to the energetic barrier. The collapse of the Cassie state will occur when the liquid filling the grooves will eventually touch the bottoms of the substrate pores, and the liquid-air interface will disappear. Contrasting to the equilibrium mixed wetting state, the composite state is metastable. For the simple topography depicted in Fig. 4, the energy barrier could be calculated as follows: Etrans ¼ 2pR2 hðgSL gSA Þ=p ¼ 2pR2 hgcosy=p (8) where h and p are the geometrical parameters of the relief, shown in Fig. 4, and R is the radius of the contact area. The numerical estimation of the energetic barrier according to the Formula (8) with the parameters p ¼ h ¼ 20 mm, R ¼ 1 mm, y ¼ 105 (corresponding to low density polyethylene (LDPE)), and g ¼ 72 mJm2, gives a value of Etrans ¼ 120 nJ. It should be stressed that according to (8), the energy barrier scales as Etrans R2. The validity of this assumption will be discussed below. The energetic barrier is extremely 2 large compared to thermal trans fluctuations: EkT Ra >>1, where a is an atomic scale. On the other hand, it is much less than the energy of evaporation of the droplet Q: kT< < Etrans < < Q. Actually, this interrelation between characteristic energies makes wetting transitions possible. If it were
W
W
2834
Wetting Transitions
Pillars
Water
Wetting Transitions, Fig. 3 The composite wetting state h
Wetting Transitions, Fig. 5 Pressure-induced displacement of the water front, leading to the collapse of the Cassie wetting state h p
p
Wetting Transitions, Fig. 4 Geometric parameters of the model relief illustrating the calculation of the Cassie–Wenzel transition energetic barrier
not the case, a droplet exposed to external stimuli might evaporate before the wetting transition. The energy-barrier height from the side of the metastable (higher-energy) wetting state is always much lower than that from the side of the stable one. The mentioned energy dependence is weak for large apparent angles, and strong for low ones; this leads to a high asymmetry of the energy barrier, i.e., to a large difference between barrier heights. As a result, wetting transitions on rough surfaces are irreversible for both hydrophobic and hydrophilic materials. Single and two-stage pathways of WT were observed, including Cassie (air-trapping)-Wenzel– Cassie (impregnating), Wenzel–Cassie (impregnating), and Cassie (air-trapping)-Cassie (impregnating) transitions.
The Concept of the Critical Pressure Various experimental methods used for the study of WT supplied the close values of the pressure necessary for the Cassie–Wenzel transition, which is of the order of magnitude of 100–300 Pa for 10 ml droplets deposited on micrometrically scaled rough surfaces. It is noteworthy that the Cassie air-trapping wetting regime
observed on natural objects (birds’ wings) is much more stable when compared to that on artificial surfaces. The important parameter characterizing WT is the critical pressure under which the transition takes place [19]. For a single-scaled pillar-based biomimetic surface, with pillar width a, and groove width b the critical pressure equals: pc ¼
gfs cos y ð1 fs Þl
(9)
where l ¼ AL , and A and L are the pillar crosssectional area and perimeter, respectively. For the Teflon surfacey ¼ 114 ; a ¼ 50 mm; b ¼ 100 mm yield pc ¼ 296 Pa, in agreement with experimental results, reported by various groups [15, 17]. Recalling that the dynamic pressure of rain droplets may be as high as 104–105 Pa, which is much larger than pc 300 Pa, it could be inferred that creating biomimetic reliefs with very high critical pressure is of practical importance. The concept of critical pressure leads to the conclusion that reducing the microstructural scales (e.g., the pillars diameters and spacing) is the most efficient means to enlarge the critical pressure [19]. The mechanism of WT based on the concept of critical pressure was proposed. It was supposed that as the pressure applied to the droplet increased, the meniscus will move toward the flat substrate, as shown in Fig. 5. The meniscus will eventually touch the
Wetting Transitions
2835
Wetting Transitions, Fig. 6 Geometrical air trapping on the hydrophilic reliefs
W
Water
ϕ0 O1
ϕ
O θ
Trapped air
substrate; this will cause the collapse of the Cassie wetting and, consequently, lead to a Cassie–Wenzel transition.
curved air-liquid interface, followed by filling the pore and the consequent collapse of the Cassie air trapping-wetting regime.
Wetting Transitions on Hydrophilic Surfaces
The Dynamics of Wetting Transitions
The Cassie wetting was observed on inherently hydrophilic surfaces. For the hydrophilic materials, the difference gSL gSA is negative, and liquid penetration into pores might have to proceed spontaneously, without the energy barrier. For the explanation of the roughness-induced superhydrophobicity of inherently hydrophilic materials, it was supposed that air is entrapped by cavities constituting the topography of the surface. The simple mechanism of “geometrical” trapping could be explained as follows: consider a hydrophilic surface (y < 90 ) comprised of pores as depicted in Fig. 6. It is seen that air trapping is possible only if y > ’0, where ’0 is the angle between the tangent in the highest point of the pattern and the horizontal symmetry axis O1O. Indeed, when the liquid level is descending, the angle ’ is growing (see Fig. 6), and if the condition y > ’0 is violated; the equilibrium y ¼ ’ will be impossible. In the equilibrium position, small fluctuations of the contact angle lead to the appearance of curvature on the plane air–water interface that is energetically unfavorable. Below the central plane O – O1 where ’ > 90 , the equilibrium is impossible in the case of y< p2 . For a large pore, small fluctuations of y can lead to touching the pore bottom near its centre by the
The dynamics of WT is not clearly understood. The characteristic time of Cassie–Wenzel transition was established with reflection interference contrast microscopy as less than 20 milliseconds [20]. Two regimes of droplet front displacement were observed: zipping and non-zipping. In the zipping regime, the velocity of the front in one direction (to advance to the next row of pillars) is much smaller than the velocity in the other direction (liquid filling up one row of micropillars). It was established that the velocity of the wetting front increases with increasing gap size, decreasing pillar height, or decreasing contact angle. A velocity of the wetting front as high as 1.5 m/s was registered.
The “Dimension” of Wetting Transitions One of the unsolved problems in the field of WT is the problem of the “dimension” of the transitions, or, in other words, whether all pores underneath the droplet should be filled by liquid (the “2D scenario”), or, perhaps, only the pores adjacent to the three-phase (triple) line are filled under external stimuli such as pressure, vibrations, or impact (the “1D scenario”). Indeed, the APCA is dictated by the area adjacent to
W
W
2836
Wetting Transitions
Wetting Transitions, Fig. 7 (a) 2D mechanism of WT (all pores are filled). (b) 1D scenario of WT (only pores adjacent to the triple line are filled)
b a
Water
Water
Air
the triple line and not by the total area underneath the droplet. 1D and 2D scenario of WT are illustrated by Fig. 7. The experiments carried out with vibrated drops supported the 1D scenario of wetting transitions. It has been established that the transition occurs when the condition Fcr ¼ const is fulfilled, where Fcr is the critical force acting on the unit length of the triple line, and the transition is caused by de-pinning of the triple line [17]. The critical value of the de-pinning force Fcr has been established experimentally for various microscopically structured surfaces as Fcr 200 350 mN m1 [17]. The energy barrier Etrans to be surmounted for the elementary displacement of the triple line dr could be estimated as Etrans 2pRFcr dr
(10)
which scales as R. This scaling law corresponds to results obtained with vibrated drops but contradicts the scaling law given by expression (8). The electrowetting-stimulated WT also supported the 1D mechanism of transitions. De-pinning of the triple line was observed directly with the use of reflection interference contrast microscopy [20]. The potential barrier Etrans , according to expression (10) for a drop with a radius of R 1 mm deposited on the LDPE relief, presented in Fig. 4 (Fcr 350 mN m1 , the elementary displacement dr p=2 105 m) equals Etrans 20 nJ, smaller than the value predicted by formula (10), but still much larger than thermal fluctuations. It has been also suggested that the Cassie–Wenzel transition occurs via a nucleation mechanism starting from the drop center. In spite of intensive theoretical and experimental effort expended for the study of wetting transitions, the numerous aspects of these phenomena, including the dynamics of transitions and their “dimension,” remain unclear and call for future investigations.
Cross-References ▶ Lotus Effect ▶ Nanotechnology ▶ Surface Tension Effects of Nanostructures
References 1. Koch, K., Bohn, H.F., Barthlott, W.: Hierarchically sculptured plant surfaces and superhydrophobicity. Langmuir 25, 14116–14120 (2009) 2. Bormashenko, E., Bormashenko, Ye, Stein, T., Whyman, G., Bormashenko, E.: Why do pigeon feathers repel water? Hydrophobicity of pennae, Cassie–Baxter wetting hypothesis and Cassie–Wenzel capillarity-induced wetting transition. J. Colloid Interface Sci. 311, 212–216 (2007) 3. Zheng, Yo, Gao, Xu, Jiang, L.: Directional adhesion of superhydrophobic butterfly wings. Soft Matter 3, 178–182 (2007) 4. Que´re´, D., Reyssat, M.: Non-adhesive lotus and other hydrophobic materials. Philos. Trans. R. Soc. A 366, 1539–1556 (2008) 5. Nosonovsky, M., Bhushan, B.: Superhydrophobic surfaces and emerging applications: non-adhesion, energy, green engineering. Curr. Opin. Colloid Interface Sci 14, 270–280 (2009) 6. de Gennes, P.G., Brochard-Wyart, F., Que´re´, D.: Capillarity and Wetting Phenomena. Springer, New York (2004) 7. Marmur, A.: A guide to the equilibrium contact angle maze. In: Mittal, K.L. (ed.) Contact Angle Wettability and Adhesion, vol. 6, pp. 3–18. VSP, Leiden (2009) 8. Bonn, D., Ross, D.: Wetting transitions. Rep. Prog. Phys. 64, 1085–1163 (2001) 9. Cassie, A.B.D., Baxter, S.: Wettablity of porous surfaces. Trans. Faraday Soc. 40, 546–551 (1944) 10. Wenzel, R.N.: Resistance of solid surfaces to wetting by water. Ind. Eng. Chem. 28, 988–994 (1936) 11. Marmur, A.: Wetting on hydrophobic rough surfaces: to be heterogeneous or not to be? Langmuir 19, 8343–8348 (2003) 12. Bormashenko, E.: Young, Boruvka–Neumann, Wenzel and Cassie–Baxter equations as the transversality conditions for the variational problem of wetting. Colloids Surf. A 345, 163–165 (2009) 13. Bartolo, D., Bouamrirene, F., Verneuil, E., Buguin, A., Silberzan, B., Moulinet, S.: Bouncing of sticky droplets:
Whispering Gallery Mode Resonator Biosensors
14. 15.
16.
17.
18.
19.
20.
impalement transitions on superhydrophobic micropatterned surfaces. Europhys. Lett. 74, 299–305 (2006) Lafuma, A., Que´re´, D.: Superhydrophobic states. Nat. Mater. 2, 457–460 (2003) Jung, Y.C., Bhushan, B.: Dynamic effects induced transition of droplets on biomimetic superhydrophobic surfaces. Langmuir 25, 9208–9218 (2009) McHale, G., Aqil, S., Shirtcliffe, N.J., Newton, G.M.I., Erbil, H.Y.: Analysis of droplet evaporation on a superhydrophobic surface. Langmuir 21, 11053–11060 (2005) Bormashenko, E., Pogreb, R., Whyman, G., Erlich, M.: Cassie-Wenzel wetting transition in vibrating drops deposited on rough surfaces: is dynamic Cassie-Wenzel wetting transition a 2D or 1D affair? Langmuir 23, 6501–6503 (2007) Barbieri, L., Wagner, E., Hoffmann, P.: Water wetting transition parameters of perfluorinated substrates with periodically distributed flat-top microscale obstacles. Langmuir 23, 1723–1734 (2007) Zheng, Q.-S., Yu, Y., Zhao, Z.-H.: Effects of hydraulic pressure on the stability and transition of wetting modes of superhydrophobic surfaces. Langmuir 21, 12207–12212 (2005) Moulinet, S., Bartolo, D.: Life and death of a fakir droplet: impalement transitions on superhydrophobic surfaces. Eur Phys J E24, 251–260 (2007)
Whispering Gallery Mode Biosensor ▶ Whispering Gallery Mode Resonator Biosensors
Whispering Gallery Mode Resonator Biosensors Kerry Allan Wilson1 and Frank Vollmer2 1 London Centre For Nanotechnology, University College London, London, UK 2 Laboratory of Biophotonics and Biosensing, Max Planck Institute for the Science of Light, Erlangen, Germany
Synonyms Microcavity biosensor; Monolithic total internal reflection biosensor; Optical cavity biosensor; Optical resonance biosensor; Optical ring resonator biosensor; Resonant evanescent wave biosensor; Whispering gallery mode biosensor
2837
W
Definition A whispering gallery mode resonator biosensor monitors optical resonances in microcavities for label-free detection of molecules and their interactions.
Overview Whispering Gallery mode (WGM) resonators are named for the “whispering gallery” of St. Paul’s Cathedral in London where sound waves propagate by reflection around the curved conduit of the gallery such that a word whispered at the front end of the gallery can be easily heard by a listener at the far end of the acoustic gallery. Similarly, optical Whispering gallery modes occur when light waves confined by total internal reflection in a dielectric microstructure are reflected back on the same optical path where they interfere constructively. Interference generates the optical resonance signal, which occurs whenever exactly an integer number of wavelengths fit on the confined (circular) light path (Fig. 1). Molecules binding to the microresonator and interacting with the evanescent field generated at the sensor surface are detected from a change in optical resonance frequency. The frequency response is rendered specific for molecular binding events by functionalization of the sensor surface with biorecognition elements such as antibodies or oligonucleotides. WGM biosensors achieve single particle, single virus, and potentially single molecule detection capability by monitoring changes in the optical resonance frequencies in microspheres, microrings, microcapillaries, and microtoroids. Glass microspheres and microtoroids are particularly sensitive WGM biosensors because of their high Quality (Q) factor and their small modal volume (V). However, other types of optical resonators (microcavities) of different shapes and made of different materials are similarly suitable for biosensing applications, examples of these are ring resonators (made of silica, silicon, or transparent polymers) and photonic crystals (fabricated in silicon-on-insulator or in silicon-nitrate wafers). WGM biosensors and optical resonator derivatives have emerged as the one of the most sensitive microsystem biodetection technology that does not require chemical amplification or labelling of the analyte.
W
W
2838
light orbit in the geometrical optics limit
Whispering Gallery Mode Resonator Biosensors
light orbit as a wave optics illustration
Whispering Gallery Mode Resonator Biosensors, Fig. 1 Schematic representation of light confined inside a resonant microcavity by total internal reflectance
Optical Resonator Physics: Confining Light at High Q-Factor A high-quality (Q) factor is necessary for achieving sensitive detection using optical resonance in microcavities. Optical resonance is created by confining coherent light inside a miniature dielectric structure such that the returning light wave interferes constructively [1–3]. Figure 1 illustrates the example of an optical resonance confined by total internal reflection in a glass microsphere. Ideally, optical resonators (also referred to as cavities or microcavities) would confine light indefinitely. Real-world divergence from this condition due to damping losses results in a finite resonance lifetime t. This finite resonance lifetime can be measured, i.e., by charging the microcavity with light, terminating the light source, and then measuring the time it takes for the intensity of the confined light wave to drop to 1/e of its initial value. This exponentially decaying intensity-amplitude of the optical resonator is analogous to the decreasing amplitude of a (nano) mechanical resonator, here illustrated as mass-on-aspring (Fig. 2b). To compare optical, mechanical, and other resonant systems which can exhibit very different resonance lifetimes, one defines a time-independent and dimensionless figure of merit, the quality (Q)-factor. The Q-factor is proportional to the total number of oscillations stored in the resonator, and this definition is independent from the type of resonator (optical, mechanical, etc.) and the nature of its oscillation. For example, in an optical resonator, the value of the Qfactor is 2p times the number of electromagnetic oscillations before amplitude of the confined light wave drops below 1/e of its initial value. Similarly, the value of the mechanical Q-factor is just 2p times the
number of (Pendulum-)swings before that same limit is reached. For any resonant system, the Q-factor is thus defined as Q ¼ o t, where t is the resonator lifetime and where o is 2p times the resonance frequency (Fig. 2). The resonance frequency for optical microresonators ranges from a few TeraHertz (1012 Hz) to one PetaHertz (1015 Hz), whereas nanomechanical resonators operate at a few MHz at most. Although resonance lifetimes of very large mechanical resonator (i.e., Feaucault Pendulum) exceed those of optical resonators (many hours compared to microseconds), the optical microresonators exhibit much higher Q values as compared to their large mechanical and nanomechanical counterparts: in fact, microcavities exhibit the highest measured Q-factor of almost 1010, whereas a good nanomechanical resonator reaches a Q value of a few million at best and a Pendulum operated in vacuum reaches a Q of about 100,000. In fact, the “ultimate” Q-factor has been measured in a microsphere cavity where an optical resonance (a whispering gallery mode, WGM) is efficiently confined by total internal reflection – with only residual damping due to absorption in the glass material. Apart from the time-domain response, a resonator also has a frequency-domain response (Fig. 2c) and it is important to realize that a large Q-factor is associated with a very narrow spectral linewidth where Doline ¼ o/Q (Fig. 2d). A narrow linewidth allows for precise measurements of resonance frequency and changes thereof, which is of paramount importance for achieving high sensitivity in biosensing applications. Ultimate-Q microsphere resonators are therefore the ideal choice to build a sensitive WGM biosensor [4]. The direct relationship between a high-Q factor and a narrow linewidth can be intuitively understood from the fact that the frequency-domain response of the resonant system (Fig. 2d) is just the Fourier-transform of the time-domain response (Fig. 2c). The Q-factor is not only related to the time-domain and the frequencydomain signal of a resonator, but also to the energy stored in a resonant system. The Q-factor can be calculated from the total energy Wtotal stored in the resonator divided by the energy DW lost per oscillation (multiplied by 2 Pi), Q ¼ 2p Wtotal/DWper oscillation. For example, the lifetime and Q-factor of an ultimate WGM microsphere resonator is limited only by residual absorption and calculated as Q ¼ 2p (Wtotal/ DW)per oscillation ¼ 2p n/la, where a is the extinction coefficient of glass, about 7 dB/km, l is the wavelength
Whispering Gallery Mode Resonator Biosensors
a
optical microresonator
c
W
resonator response: time-domain 1 (a.u.)
ωoptical = ~1014 Hz
b nanomechanical resonator
ωmechanical = ~106 Hz
1/e 100
200
300
t
400
500 t (a.u.)
Q = wt
d amplitude response
m
amplitude response
Whispering Gallery Mode Resonator Biosensors, Fig. 2 (a) and (b) Comparison of optical and nanomechanical resonators. (c) Definition of Q-value in the time domain. (d) Definition of Q-value in the frequency domain
2839
resonator response: frequency-domain 1 (a.u.)
Dw w Frequency ω (a.u.) Q = w/Dw
Whispering Gallery Mode Resonator Table 1 Q-factors of various resonant systems Integrated circuit filter Mechanical wristwatch Surface acoustic waves Foucault pendulum (in vacuum) Quartz crystal Microsphere optical resonator
Biosensors, 20 100–300 2,000 104–105 106 1010
(here 633 nm), and n the refractive index of the glass (here 1.45). Table 1 gives an overview of Q-factors of resonant optical, electrical, and mechanical systems.
Principles of Resonant WGM Biosensing An optical resonator transduces the binding of molecules into a resonance frequency shift signal which can be detected with great precision due to the high Q-factor of the optical detection system [5]. Optical resonators have no moving parts and maintain a high Q-factor even in a liquid biosensing environment [4] where any mechanical system would be heavily damped (low Q). The change in resonance frequency of an optical resonator occurs in response to the binding of molecules which increase the round-trip optical path length. This is different from a mechanical
resonator where the frequency shift would occur due to mass loading by the added molecules (Fig. 3). Such a frequency shift, i.e., for a nanomechanical resonator can be easily calculated: Additional molecular mass loading, Dm, on a system with a nominal mass, m, will cause a resonance frequency shift of Domechanical ¼ omechanical Dm/2m (Fig. 3), where only nanomechanical systems are useful for such measurements due to their small mass. For an optical resonator instead, the binding of molecules add additional optical path length DL causing the resonance frequency ooptical to shift by Dooptical ¼ ooptical DL/2L, where 2L is the round-trip optical path length in the microcavity (Fig. 3). For the special case of a spherical WGM optical resonator where bound molecules form an optically contiguous layer of thickness l (Fig. 3), the resonance frequency shift is calculated as Domicrosphere ¼ omicrosphere l/R, where l is the thickness of the adsorbed molecular layer and R is the radius of the microsphere. This estimate assumes an identical refractive index for microsphere and adsorbed molecular layer, nglass ¼ nmolecules, a good approximation for the case of proteins adsorbing to a glass microsphere (nglass ¼ nprotein 1.45). Further assuming the ability to detect the resonance frequency shift (Fig. 3) with a resolution of S ¼ 1/50 of the linewidth, and assuming a modest Q of 107, which can be readily obtained
W
W
2840
Whispering Gallery Mode Resonator Biosensors
Whispering Gallery Mode Resonator Biosensors, Fig. 3 Reactive sensing principle for an optical resonator and for a nanomechanical resonator
I
Δω/ωoptical = − ΔL/2L Δω/ωmicrosphere = − I/R
R
amplitude
R+I
Δm m
Δω
ω
Δω/ωmechanical = − 1/√1+Δm/m Δm 0: dn/dc ¼ lim(Dn/Dc)c->0. The excess polarizability a is then calculated from dn/dc values according to, a ¼ eO 2n
dn m dc
(3)
where n is the solvent refractive index (water), and m is the mass of the single biomolecule. Proteins such as bovine serum albumin (BSA) have a typical dn/dc value of about 0.183 cm3/g. The excess polarizability of BSA is then calculated as aBSA ¼ 4 p e0 3.85 * 1021 cm3, where the mass of a single BSA molecule is 66,432 (g/mol) / 6.02 1023 (1/mol) ¼ 1.1 1019 g. The excess polarizability a is proportional to the molecular mass, m, of a protein and may vary only slightly between different proteins of similar molecular weight but of different amino acid sequence, such as for BSA and HSA (human serum albumin). Similar arguments can be made for many other
W
W
2842
Whispering Gallery Mode Resonator Biosensors, Fig. 4 (a) Detection of homogenous binding of biomolecules in the evanescent field of a WGM. (b) Detection of single particle interacting with WGM
Whispering Gallery Mode Resonator Biosensors
a
detection of homogenous binding
Δλ =
σα ε0(n2s
− n2M) R
σ = surface density α = ex. polarizability ε0 = vacuum permittivity ns = ref. ind. microsphere nm = ref. ind. medium
λ0
R = radius microsphere λ0 = resonance wavelength Δλ = wavelength shift
b
single particle detection Dl l0
@D max
a3 R 5/2 l01/2
e−a/L
D = 2n 2m (2ns)1/2(n 2np − n 2m) / (n 2s − n 2m)(n 2np + 2n 2m) L ≈ (λ /4p)(n 2s − n 2m)−1/2 a = particle radius
biopolymers [8]. Table 2 lists typical dn/dc values of various biopolymers and biomolecules, in different solvent conditions. Note that dn/dc values depend on many additional parameters such as temperature, solvent, ionic strength, and probing wavelength (Table 2). Interestingly, one can also calculate the wavelength dependence of polarizability values from Kramers-Kronig transformations by integration over the respective absorption spectra [8]. In biosensing experiments, one often measures the resonance wavelength shift Dl instead of the frequency shift: Dl ¼ lres, after lres,before. The wavelength shift Dl can be converted in the associated frequency shift Df by: Df ¼ clight (1/lres, after 1/lres, before) ¼ clight Dl/l2, where l is the initial resonance wavelength and clight is the speed of light. In all WGM biosensing experiments, the wavelength shift, Dl, or frequency shift, Df, reports on the surface density s of molecules bound to the microsphere resonator, at a specific time point. Depending on specific cavity material or surface functionalization, the saturation surface density of bound molecules may vary between 0 and 1015/cm2. To achieve a surface coverage of more than 1015/cm2 requires molecules to form multiple layers or to bind within less than a nm distance, certainly a physical limit as the size of an atom is, 0.1 nm ¼ 1 Angstrom. High-Q microcavities are invaluable tools for studying the binding of molecules
L = evanescent field length
Whispering Gallery Mode Resonator Biosensors, Table 2 Typical dn/dc values of biopolymers and molecules Biomolecule or biopolymer Bovine serum albumin Deoxyribonucleic acid from calf thymus a-chymotrypsinogen
dn/dc (cm3/g) [at 633 nm wavelength] 0.183 0.166
0.110 0.091 0.150
0.1N tris buffer pH ¼ 8.4 at 23 C 0.15M NaCl; 0.1 pH buffer, pH ¼ 7.2 at 23 C Water at 23 C 0.1N NaCl at 23 C 0.1M NaNO3; pH ¼ 6.6 at 26 C 0.1% NaCl at 23 C Toluene at 23 C 0.05M NaNO3
0.160
Water at 23 C
0.171
Dextran 0.147 Gelatin 0.163 Lignin sulfonate – Na 0.188 Mucopolysaccharides Polydimethylsiloxane Polyvinyl alcohol 98% hydrolyzed RNA
Measurement parameter notes Water at 23 C
to the sensor surface since precise measurements of high-surface densities can reveal interactions between closely spaced molecules, and precise measurements at low surface densities c2>c1 c2
df dt
c2¥kon = df/dt c1 time t
Whispering Gallery Mode Resonator Biosensors, Fig. 5 (a) Steady-state response of WGM biosensor. (b) Time-domain response (sensogram) of a WGM biosensor
resonator surface. The simplest steady-state kinetic model which can be applied to the binding of molecules to a surface is given by the Langmuir-like model [9] which assumes equilibrium binding between molecules and a fixed number of surface binding sites, further assuming equal affinity and no interaction between the sites. To determine a Langmuir binding isotherm, one first calculates the fractional surface coverage F by dividing the equilibrium frequency shift by the shift at saturation (measured at high solution concentration). Then, one plots the measurement of fractional surface coverage F versus the different solution concentration levels (Fig. 5a). The rate equation for the Langmuir model is given by dF/dt ¼ kon*c*(1F) koff*F, so that in steady state dF/dt ¼ 0 and F/(1F) ¼ kon / koff*c. For halfmaximal surface coverage at F¼1/2, the solution concentration is just equal to the equilibrium (dissociation) constant c ¼ K ¼ koff/ kon, which can be conveniently determined from the Langmuir isotherm (Fig. 5a). The affinity (dissociation) constant is, e.g., measured at 1 nM for BSA proteins binding to an aminosilanized microsphere [4]. For more complex binding and adsorption processes, different kinetic models can be applied such as the random sequential adsorption (RSA) model, which is derived from the
W
Langmuir model by taking into account the additional exclusion volume of bound molecules. Another kinetic model considers conformational changes by modeling bound proteins as discs with radii that may change in a post-binding step. The reader is referred to standard pharmacokinetic textbooks for further analysis of dose-response curves and their importance in analytical and clinical chemistry. Obtaining such doseresponse curves is essential for relating the solution concentration of analyte molecules to the measured surface concentrations as measured by the resonator shift signal. Instead of analyzing molecular recognition and binding in the steady-state regime, the interaction of biomolecules at a resonator surface can also be analyzed in the time domain by plotting the resonance frequency or wavelength shift signal versus time (sensogram, Fig. 5b). The sensogram and its derivative report on the kinetics and the rate of adsorption or desorption, respectively. It is common to extract the derivative (rate of adsorption or desorption) at the onset of the experiment where dF/dt ¼ kon * c in a binding experiment started at c > 0 and F¼0, and dF/dt ¼ koff for an unbinding experiment started at c ¼ 0 and F>1. Since time scales for binding can be fast, it is often necessary to start the experiment rapidly using microfluidic techniques that allow a rapid change of solution concentration. In principle, optical resonators follow such a fast response without problem since the time resolution of a microcavity is principally limited only by the resonator photon lifetime t, which is on the order of micro to nanoseconds, and dependent on the Q-value. WGM microsphere cavities have a further advantage as compared to SPR and QCM techniques in that they exhibit multiple optical resonances signals, which are useful for detection. In fact, optical resonances appear throughout an optical resonator spectrum since the resonance condition states that a multiple number of resonance wavelengths have to fit the optical path. The precise calculation of the resonances requires Maxwell’s equations and has many solutions, and a more versatile sensor uses more than one resonance shift signal to analyze the binding of molecules to the resonator surface. For example, differently polarized (TE and TM) optical resonances can be used to determine the orientation of molecules at the sensor surface [5]. Each one of the two resonances (TE and TM) probe the surface of the microsphere in a different direction, and the specific arrangement of
W
W
2844
Whispering Gallery Mode Resonator Biosensors
molecules in a monolayer and changes thereof can be measured by comparing the TE and TM wavelength shifts. Different resonance wavelength shifts have already been used to determine conformational changes in photosensitive bacteriorhodopsin membrane and to measure the orientation of the photochrome retinal within a lipid bilayer [5]. It is also possible to compare the resonance wavelength shifts for resonances excited at different wavelength (e.g., at 763 nm and 1,310 nm) in parallel. Using this approach, it is possible to directly determine the thickness and the permittivity (refractive index) of an adsorbed nanolayer, and changes thereof [10]. dl
l 1 dl l 2
L1 ½1 expðl=L1 Þ L2 ½1 expðl=L2 Þ
parallel. Even without binding, but already within the evanescent field, the Brownian motion of a single particle causes resonance wavelength fluctuations [5], a signal that can be used to extract single particle diffusion constants. Furthermore, the frequency shift signal associated with a single particle trapped by the optical field of the resonator can be used to determine a chemical surface potential: If a particle of radius a is trapped in the optical field of the resonator, the resonance wavelength shift reports on the radial position of the particle measured as the distance h from the microsphere surface [12]: Dlr ðh þ aÞ exp½ðh þ aÞ=L ¼ exp½h=L ¼ Dlr ðaÞ exp½a=L
(5)
(4)
Where l is the thickness of the adsorbed layer and L1 and L2 are the evanescent field length at probing wavelength l1 and l2, with L ðl=4pÞðn2s n2m Þ1=2 .
WGM Biosensing: Label-free Single Particle Analysis On the single particle level, the frequency shift (or rather “step”) associated with a single binding event depends on the location of the particle on the resonator surface. For a microsphere WGM biosensor, the light field is confined along the equator of the sphere (Fig. 4) and the largest frequency shift is recorded for an equatorial binding event. The frequency shift associated with the equatorial binding event can be calculated from the reactive sensing principle [7] (1), where the fractional wavelength shift now depends on particle size, a, and the evanescent field length, L (Fig. 4b). Single bioparticle detection was demonstrated for a single Influenza A virion, and analysis of the discrete frequency shift signals (Fig. 4b) was used to determine the size of the particle (50 nm radius) as well as its weight (500 attograms) [7]. The size of a bound nanoparticle can also be determined independently of binding location by measuring resonance frequency shift as well as changes in Q-factor in parallel [11]. In general, single particle detection opens up the possibility to determine size and shape of nanoparticles by monitoring frequency shifts of different optical resonances (TE, TM, etc.) and associated Q-factors in
From this measurement, a histogram of equilibrium particle positions can be replotted as a Boltzmanndistribution in kT units (k ¼ Boltzmann constant, T ¼ Temperature), revealing the trapping potential consisting of a chemical potential on the resonator surface and the optical potential of the evanescent field.
Implementation and Operation of a WGM Resonator Biosensor In its simplest implementation (Fig. 6), the WGM biosensor device is build by coupling light from a tapered optical fiber to a microsphere optical resonator [4]. The microsphere itself is fabricated from an optical fiber piece where one end is melted in a hot flame, forming a sphere on-a-stem structure. The sphere is then positioned adjacent to the tapered fiber region, where the light can couple from the optical fiber to the resonator. Light from a tunable laser is coupled into one end of the optical fiber from where it is delivered to this tapered coupling region. The output end of the optical fiber is connected to a photodetector, which records the transmitted intensity. As the laser wavelength is tuned to match the resonance wavelength of the microsphere, light starts to couple from the tapered fiber region to the resonator where it remains confined as WGM optical resonance. Due to interference in the coupling region, the light then no longer reaches the photodetector, which now records a minimum in the transmitted intensity. A computer identifies and tracks the minimum in the recorded transmission spectrum and plots the
Whispering Gallery Mode Resonator Biosensors
2845
W
Protein Buffer PC
Peristaltic pump
Waveguide Flow Microsphere
DFB Laser
Laser controller
Flow cell
Photodetector
Function Generator
Whispering Gallery Mode Resonator Biosensors, Fig. 6 Schematic representation of WGM biosensor experimental setup comprised of fluidics component (dashed lines),
excitation source and detector (dotted line), and data acquisition software (dash-dot line), and sphere-on-stem resonator housed in a flow cell (inset)
resonance wavelength versus time, at ms to ms time resolution, typically limited by the speed of the acquisition system, as well as by the speed of the laser control system. The sensogram is analyzed either post hoc or in real time and contains information on the binding and interaction of molecules at the WGM biosensor surface. Several other experimental configurations exist for coupling light into WGM biosensors and for detecting resonances. Earlier studies of WGM resonances even employed “free beam” coupling methods in which the incident laser is guided through external optics and focused onto the microcavity to excite the resonance. Only a small fraction of the incident light, however, can couple radiatively to the microcavity. It is also possible to use light sources other than lasers, e.g., light sources with a broader emission spectrum, such as an ASE (amplified spontaneous emission) light source operating in the telecom band. In this case, however, a high-resolution monochrometer is necessary to resolve the narrow linewidth resonances, certainly a limitation when using a high Q cavity. In other implementations, it is possible to use spheres labeled with a fluorophore and resolve resonances in the emission spectrum, which is collected on a bead-by-bead basis. This implementation limits the operation of the biosensor to lower Q factors, but allows the use of freestanding fluorescent spheres, such as 10 umdiameter polystyrene spheres soaked with rhodamine 6G fluorophore [13].
Most common experimental setups utilize tapered fiber optic waveguides or planar waveguides to deliver the light to various resonator geometries such as discs, rings, toroids, or spheres (Fig. 7). Light sources typically consist of a tunable distributed feedback laser (DFB) with 300 kHz linewidth available at various nominal wavelengths (760–1,550 nm) and which can be tuned by a fraction of a nm simply by ramping the operating (laser diode) current [4]. A simple (silicon or InGaAs) photodetector is sufficient to detect the intensity, and a standard computer can record the spectrum and locate the resonance wavelength. Biosensing experiments often require the integration of the resonator with a microfluidic flow system. Figure 6 shows the integration of a complete microsphere WGM biosensor with a simple flow cell.
Resonator Geometry One of the primary benefits of WGM biosensors is the unprecedented sensitivity provided by high Q-factors in optical microcavities. High Q-factors are found for a variety of resonator shapes and fabricated by different procedures (Fig. 7). The decision for particular sensor geometry depends on the required sensitivity, cost, multiplexing capability, and need for integration with microfluidics and other photonic or electronic components. Often, the most important factor remains sensitivity and to date the highest Q-factor was
W
W
2846
Whispering Gallery Mode Resonator Biosensors, Fig. 7 Schematic illustration of different resonator geometries. (a) Sphere, (b) Toroid, (c) Disk, (d) Ring, (e) Cylinder, (f) Photonic crystal
Whispering Gallery Mode Resonator Biosensors
a
Q = 105-1010
b
Q = 108
c
Q = 103-105
d
Q = 103-105
e
Q = 105
f
Q = 106
reported for a microsphere resonator at Q 1010 which is thus the ideal choice for building highly sensitive WGM biosensors. The ultimate sensitivity obtained in ultrahigh-Q resonators enables sensing of single virus particle and molecules. However, more modest Q factors (Q ¼ 1104 to 1107) observed in other resonator geometries and materials (Fig. 7) are sufficient in most biosensing applications and can already achieve detection limits superior to SPR sensors (
