Functionalized Nanoscale Materials, Devices and Systems
NATO Science for Peace and Security Series This Series presents the results of scientific meetings supported under the NATO Programme: Science for Peace and Security (SPS). The NATO SPS Programme supports meetings in the following Key Priority areas: (1) Defence Against Terrorism; (2) Countering other Threats to Security and (3) NATO, Partner and Mediterranean Dialogue Country Priorities. The types of meeting supported are generally "Advanced Study Institutes" and "Advanced Research Workshops". The NATO SPS Series collects together the results of these meetings. The meetings are coorganized by scientists from NATO countries and scientists from NATO's "Partner" or "Mediterranean Dialogue" countries. The observations and recommendations made at the meetings, as well as the contents of the volumes in the Series, reflect those of participants and contributors only; they should not necessarily be regarded as reflecting NATO views or policy. Advanced Study Institutes (ASI) are high-level tutorial courses intended to convey the latest developments in a subject to an advanced-level audience Advanced Research Workshops (ARW) are expert meetings where an intense but informal exchange of views at the frontiers of a subject aims at identifying directions for future action Following a transformation of the programme in 2006 the Series has been re-named and re-organised. Recent volumes on topics not related to security, which result from meetings supported under the programme earlier, may be found in the NATO Science Series. The Series is published by IOS Press, Amsterdam, and Springer, Dordrecht, in conjunction with the NATO Public Diplomacy Division. Sub-Series A. B. C. D. E.
Chemistry and Biology Physics and Biophysics Environmental Security Information and Communication Security Human and Societal Dynamics
http://www.nato.int/science http://www.springer.com http://www.iospress.nl
Series B: Physics and Biophysics
Springer Springer Springer IOS Press IOS Press
Functionalized Nanoscale Materials, Devices and Systems
edited by
A. Vaseashta Nanomaterials Processing and Characterization Laboratories, Marshall University, Huntington, WV, U.S.A.
I.N. Mihailescu National Institute for Lasers, Plasma and Radiation Physics, “Laser-Surface-Plasma Interactions” Laboratory, Bucharest-Magurele, Romania
Published in cooperation with NATO Public Diplomacy Division
Proceedings of the NATO Advanced Study Institute on Functionalized Nanoscale Materials, Devices and Systems for Chem.-bio Sensors, Photonics, and Energy Generation and Storage Sinaia, Romania 4–15 June 2007
Library of Congress Control Number: 2008931994
ISBN 978-1-4020-8902-2 (PB) ISBN 978-1-4020-8901-5 (HB) ISBN 978-1-4020-8903-9 (e-book)
Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands. www.springer.com
Printed on acid-free paper
All Rights Reserved © 2008 Springer Science + Business Media B.V. No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
TABLE OF CONTENTS Preface................................................................................................................xi Part I. Invited Contributions Nanoscale Materials, Devices, and Systems for Chem.-Bio Sensors, Photonics, and Energy Generation and Storage A. Vaseashta...................................................................................................3 Nanostructured Thin Optical Sensors for Trace Gas Detection C. Ristoscu, I.N. Mihailescu, D. Caiteanu, C.N. Mihailescu, Th. Mazingue, L. Escoubas, A. Perrone, and H. Du....................................29 X-Ray Photoelectron Spectroscopy and Tribology Studies of Annealed Fullerene-like WS2 Nanoparticles F. Kopnov, R. Tenne, B. Späth, W. Jägermann, H. Cohen, Y. Feldman, A. Zak, A. Moshkovich, and L. Rapoport .................................51 The Development and Application of UV Excimer Lamps in Nanofabrication I.I. Liaw and I.W. Boyd ................................................................................61 Functionalization of Semiconductor Nanoparticles M.-I. Baraton ...............................................................................................77 Flexoelectricity: A Universal Sensoric Mechanism in Biomembranes and in Chem.-Biosensors A.G. Petrov...................................................................................................87 Carbon Nanotubes: From Fundamental Nanoscale Objects Towards Functional Nanocomposites and Applications W. Maser, A.M. Benito, E. Muñoz, and M. Teresa Martínez .....................101 Ultrashort Pulse PLD: A Technique for Nanofilm Fabrication T. Szörényi and Zs. Geretovszky ................................................................121 Laser Ablation and Laser Induced Plasmas for Nanomachining and Material Analysis D. Batani ....................................................................................................145
v
vi
TABLE OF CONTENTS
Photo-, Dual- and Exoelectron Spectroscopy to Characterize Nanostructures Y. Dekhtyar ................................................................................................169 Laser Interaction with Nano-Spheres: Applications in Sub-Micron Particles Removal and Nanodot Array Fabrication M. Sentis, D. Grojo, Ph. Delaporte, and A. Pereira ..................................185 Clean Fossil Fuels: Advanced Membrane Reactors T. Tran, K. Stoitsas, and J. Schoonman .....................................................199 Nanocrystalline Diamond Films for Advanced Technological Applications C. Popov and W. Kulisch ...........................................................................215 Diamond like Carbon Films: Growth and Characterization S. Tamuleviþius and Š. Meškinis ................................................................225 Fundamentals of Laser-Assisted Fabrication of Inorganic and Organic Films J. Schou ......................................................................................................241 Nanoparticles of Semiconductors in Sol Gel Glasses R. Reisfeld ..................................................................................................257 Electrochemical Sensor Technology Based on Nanomaterials for Biomolecular Recognitions A. Erdem ....................................................................................................273 The Effects of Doping with Elements from the IIA Group on the Thermal and Electronic Properties of Amorphous Selenium G. Belev, D. Tonchev, S.O. Kasap, and H. Mani .......................................279 Nanoscale Materials for Hydrogen and Fuel Cell Systems M. Suha Yazici ...........................................................................................283 Application of Fe-Nanoscale Materials Useful in the Removal of Arsenic from Waters M. Vaclavikova, K. Stefusova, S. Jakabsky, S. Hredzak, and G. Gallios ...291 Nanopatterning Using the Bioforce Nanoenabler K. Arshak, O. Korostynska, and C. Cunniffe .............................................299
TABLE OF CONTENTS
vii
Electrocatalysts and Electrode Design for Bifunctional Oxygen/Air Electrodes V. Nikolova, P. Iliev, K. Petrov, T. Vitanov, E. Zhecheva, R. Stoyanova, I. Valov, and D. Stoychev...........................................................................305 Part II. General Contributions Preparation of Magnetic Chitosan Nanoparticles for Diverse Biomedical Applications D. Kavaz, T. Çirak, E. Öztürk, C. Bayram, and E.B. Denkbaú ..................313 Anomalous Behavior of Carbon Filled Polymer Composites Based Chemical and Biological Sensors K. Arshak, C. Cunniffe, E. Moore, and A. Vaseashta ................................321 Poly(n-isopropylacrylamide) (PNIPAM) Based Nanoparticles for In Vitro Plasmid DNA Delivery N. Ozdemir, A. Tuncel, M. Duman, D. Engin, and E.B. Denkbas..............325 Rapid, Contactless and Non-Destructive Testing of Chemical Composition of Samples O. Ivanov, L. Stoychev, and A. Vaseashta..................................................331 Synthesis and Application of Metal-Containing Silicas K. Katok, V. Tertykh, and V. Yanishpolskii................................................335 Semiconducting Gas Sensors, Remote Sensing Technique and Internet GIS for Air Pollution Monitoring in Residential and Industrial Areas O. Pummakarnchana, V. Phonekeo, and A. Vaseashta..............................339 Self-Assembled System of Semiconductor and Virus like Nanoparticles Yu. Dekhtyar, A. Kachanovska, G. Mezinskis, A. Patmalnieks, P. Pumpens, and R. Renhofa......................................................................347 Thermal Stability and Optical Activity of Erbium Doped Chalcogenide Glasses for Photonics D. Tonchev, K. Koughia, S.O. Kasap, K. Maeda, T. Sakai, J. Ikuta, and Z.G. Ivanova........................................................................................351
viii
TABLE OF CONTENTS
XRD Study of Pulsed Laser Deposited AlN Films with Nanosized Crystallites S. Bakalova, A. Szekeres, A. Cziraki, E. Gyorgy, S. Grigorescu, G. Socol,and I.N. Mihailescu .....................................................................357 Functionalization of Multi-Walled Carbon Nanotubes (MWCNTs) M. Mohl, Z. Kónya, Á. Kukovecz, and I. Kiricsi ........................................365 Sonochemical Synthesis of Inorganic Nanoparticles J. Kis-Csitári, Z. Kónya, and I. Kiricsi ......................................................369 Novel Transparent Molecular Crystals of Carbon G. Kharlamova, N. Kirillova, A. Kharlamov, and A. Skripnichenko .........373 Hydrogen Microsensor Based on Nio Thin Films I. Fasaki, M. Antoniadou, A. Giannoudakos, M. Stamataki, M. Kompitsas, F. Roubani-Kalantzopoulou, I. Hotovy, and V. Rehacek...379 Design and Characterization of Styrene-Based Proton Exchange Membranes D. Ebrasu, I. Petreanu, L. Patularu, I. Stefanescu, and M. Valeanu .........383 Strontium-Substituted Hydroxyapatite Thin Films Grown by Pulsed Laser Deposition C. Capuccini, E. Boanini, A. Bigi, M. Gazzano, F. Sima, E. Axente, and I.N. Mihailescu....................................................................................389 Growing Thin Films of Charge Density Wave System Rb0.3MoO3 by Pulsed Laser Deposition D. Dominko, D. Starešiniü, K. Biljakoviü, K. Salamon, O. Milat, A. Tomeljak, D. Mihailoviü, J. Demšar, G. Socol, C. Ristoscu, I.N. Mihailescu, and J. Marcus..................................................................399 Single Cell Detection with Driven Magnetic Beads B. McNaughton, R.R. Agayan, V.A. Stoica, R. Clarke, and R. Kopelman .403 Antimicrobial Properties of Titanium Nanoparticles B.K. Erdural, A. Yurum, U. Bakir, and G. Karakas...................................409 CsHSO4/Nanooxide Polymer Membranes for Fuel Cell A. Andronie, A. Morozan, C. Nastase, F. Nastase, A. Dumitru, S. Vulpe, I. Stamatin, and A. Vaseashta ....................................................................415
TABLE OF CONTENTS
ix
IV and CV Characteristics of Multifunctional Ilmenite-Hematite 0.67FeTiO3-0.33Fe2O3 C. Lohn, W.J. Geerts, C.B. O’Brien, J. Dou, P. Padmini, R. Schad, and R.K. Pandey, .......................................................................................419 Electrodeposition of BI1-XSBX Nanowires as an Advanced Material for Thermoelectric Applications J.E. Weber, A. Kumar, and W.G. Yelton ....................................................425 A Solid State Nano-Generator: Concept, Design and Theoretical Estimations M. Vopsaroiu, M.G. Cain, V. Kuncser, and J. Blackburn..........................431
Applications of Statistical Physics to Mixing in Microchannels: Entropy and Multifractals M. Kaufman, M. Camesasca, I. Manas-Zloczower, L.A. Dudik, and C. Liu...................................................................................................437 Synthesis and Characterization of Carbon Supported Pd and PtPd Catalysts for DMFCs A. Morozan, A. Dumitru, C. Mirea, I. Stamatin, F. Nastase, A. Andronie, S. Vulpe, C. Nastase, and A. Vaseashta ... .................................................445 Theoretical Study of the Adsorbed Small Molecule on Twisted Nanotubes by Atomic Scale Simulations V. Chihaia, A. Ghita, B.-S. Seong, and S.-H. Suh .. ...................................449 The Defect Structure of Copper Indium Disulfide D. Perniu, A. Duta, and J. Schoonman .....................................................457 ASI Group Photograph....................................................................................465 ASI Group Photograph Legend.......................................................................467 Selected photographs Taken During the ASI..................................................469 List of Participants...........................................................................................475 Author Index....................................................................................................483 Keyword Index................................................................................................487
PREFACE The primary objective of the NATO Advanced Study Institute (ASI) titled “Functionalized Nanoscale Materials, Devices, and Systems for Chem.-Bio Sensors, Photonics, and Energy Generation and Storage” was to present a contemporary and comprehensive overview of the field of nanostructured materials and devices and its applications in chem.-bio sensors, nanophotonics, and energy generation and storage devices. The study has become one of the most promising disciplines in science and technology, as it aims at the fundamental understanding of new physical, chemical, and biological properties of systems and the technological advances arising from their exploration. Such systems are intermediate in size, between the isolated atoms and molecules and bulk material, where the unique transitional characteristics between the two can be understood, controlled, and manipulated. Nanotechnologies refer to the creation and utilization of functional materials, devices, and systems with novel properties and functions that are achieved through the control of matter, atom-by-atom, molecule-by-molecule, or at a micro-molecular level. Advances made over the last few years provide new opportunities for scientific and technological developments in nanostructures and nanosystems with new architectures with improved functionality. The field is very actively and rapidly evolving and covers a wide range of disciplines. Recently, various nanoscale materials, devices, and systems with remarkable properties have been developed, with numerous unique applications in chemical and biological sensors, nanophotonics, nano-biotechnology, and in-vivo analysis of cellular processes at the nanoscale. On a scientifically related note, the potential and risk for inadvertent or deliberate contamination of the environment, food and agricultural products has recently increased due to the global threats of terrorism. As a result, decentralized sensing has emerged as an important issue for several agencies. To detect the contaminants, the current trend is to make laboratory facilities more mobile and conduct clinical trials employing direct reading, portable, labon-chip systems. A nanotechnology-based sensor platform provides platform for direct electrical detection of biological and chemical agents in a label-free, highly multiplexed format over a broad dynamic range during clinical testing. This platform uses functionalized nanotubes, nanoparticles, and nanowires to detect molecular binding with high sensitivity and selectivity. The platform is capable of detecting a broad range of molecules, viz., DNA, RNA, proteins, ions, small molecules, cells, and even pH values. Detection is possible in both liquid and gas phase and is highly multiplexable, enabling the parallel detection of multiple agents. Recent progress in nanostructured materials and its potential applications in chemical and biological sensors are likely to have a significant impact on efficient data collection, processing, and recognition with minimum false positive xi
xii
PREFACE
counts. Furthermore, nucleic acid layers combined with nanomaterials-based electrochemical or optical transducers produce a new kind of affinity biosensors such as the “DNA Biosensor” or “Genosensor” for small molecular weight molecules. Genosensors are attractive devices for converting the hybridization event into an analytical signal for obtaining sequence-specific information in connection with clinical, environmental, or forensic investigations. The continued development through combined efforts in microelectronics, surface/interface chemistry, molecular biology, and analytical chemistry is expected to lead to the establishment of Genosensor technology as a major component of analytical biochemistry. The design and fabrication of DNA-modified surfaces and materials which are reproducible, stable, and selective to complementary DNA sequences are crucial in the development of emerging analytical tools such as DNA chips or simple diagnostic devices for detecting DNA sequences. These devices have been used extensively not only for the rapid, cost-effective, simple diagnosis of inherited and infectious diseases, but also for the early detection of infectious agents in various environments. In general, the scientific community is confident that nanoscience and nanotechnology will revolutionize research on, and applications in the areas of biology, medicine, and human health. The research will also provide unprecedented means to forewarn and/or protect against the potential and risk for inadvertent or deliberate contamination of the environment and food and agricultural products. Many technologists develop new tools, yet they often have limited understanding of the restrictions that biology places on the proper design of nanotools and nanosystems. Hence, we sought to adopt with this ASI an interdisciplinary approach, bringing together recognized experts in various fields while retaining a level of treatment accessible to those active in specific individual areas of research and development. The NATO ASI titled “Functionalized Nanoscale Materials, Devices, and Systems for Chem.-Bio Sensors, Photonics, and Energy Generation and Storage” was very well planned, organized, and received by the participants. Both directors spent an enormous amount of time in carefully planning the logistics of the event from both international and local organization perspectives. To maximize global participation, the organizing committee focused on a promotional strategy that garnered a tremendous response not only from NATO and partner countries but also from the Asian-Pacific rim. Such an unexpected and substantial response presented a welcome but arduous task for the organizing committee charged with selecting the most deserving participants. Realizing that a productive ASI would result from a large number of participants from multiple scientific disciplines, the organizers accepted a wide range of promising candidates who could be funded through NATO.
PREFACE
xiii
Lectures covering the basic principles and state-of-the-art applications of nanostructured and advanced materials for sensor, optoelectronic, and photovoltaic devices were conducted by 13 experts recognized for advances in nanotechnology. Focused seminar sessions, poster sessions, and interactive feedback sessions stimulated extended interactions between participants and subject matter experts. As a venue for collaborative learning, the interactive lectures and sessions drew enthusiastic response and sharing of information and ideas from all participants. The ASI was held at the Hotel Sinaia, in the eponymous resort town of Romania, which offered a tranquil, congenial setting for the participants. The facility supported formal and informal settings for structured and spontaneous learning and sharing of ideas. The meeting lasted 10 days, with a day free to visit nearby towns, shop, or simply relax. The downtown shops and pubs provided a much-needed break after lectures and in-depth discussions at the ASI. The participants also shared a sightseeing excursion to Bran, Brasov, and Poiana Brasov and a traditional Romanian dinner complete with folk dancing. The unique balance of technical and social interactions materialized in alliances between participants, which have been evidenced by continued correspondence in the months following the ASI. The co-directors interpret the ongoing interaction and positive feedback from participants as an affirmation of a successful ASI. Such a constructive ASI is the outcome of efforts by participants, lecturers, presenters, and co-directors in addition to a host of caring individuals who supported their work. Much appreciation is extended to Mr. Marian Swartz, the Manager of the Hotel Sinaia. We would like to acknowledge editorial assistance from Adina Morozan, Rodica Cristescu, and Silvia Bakalova. Our local organizers, Adina Morozan, Adrian Ghita, Rodica Cristescu, Carmen Ristoscu, Felix Sima, and Ioana Vasiliu handled our everyday logistics with the utmost consideration and efficiency. Our gratitude goes to Dr. F. Pedrazzini, the director of the NATO Scientific Affairs Division for his encouragement, expertise, and financial support. Ms. Annelies Kersbergen with the NATO Publishing Unit of the Springer Academic Publishers has provided us with much appreciated expertise in publishing our proceedings. Similarly, successful proceedings are a result of meticulous preparation of manuscripts by participants to whom thanks are extended. Several funding agencies, such as NSF and TUBITAK, provided travel support for some of the participants and are acknowledged immensely for their generous help. The organizers also acknowledge support from the Romanian Ministry of Education and Research, which provided the conference materials. Thanks are due to the Vaseashta Foundation for poster awards. The co-directors hopefully anticipate that this ASI provides continued success for all participants as they extend collaboration in the pursuit of nanotechnology advancement.
xiv
PREFACE
We would like to close our preface with the following two quotes. “Over the next ten years, the fields of chemistry, physics, material sciences, biology, and computational sciences will converge in a way that will define nanotechnology and impact almost every industry, including computers, semiconductors, pharmaceuticals, defense, health care, communications, transportation, energy, environmental sciences, entertainment, chemicals, and manufacturing. Previously distinct disciplines will also combine: medicine and engineering, law and science, art and physics, etc. This merging will result in developments that are not simply evolutionary; they will be revolutionary.” ............Jack Uldrich and Deb Newberry “The revolutionary promise of molecular nanotechnology (MNT) has become a part of society’s expectations for the future. This technology will provide nanomedicine breakthroughs that could cure cancer and extend lifespace, bring abundance without environmental harm and provide clean sources of energy. These ideas are part of the vision that launched the field of nanotechnology.” ............K. Eric Drexler
Directors Ashok Vaseashta (Washington, DC) Ion N. Mihailescu (Bucharest) April 2008 Organizing Committee Arzum Erdem Ion N. Mihailescu Joop Schoonman Ioan Stamatin Sigitas Tamelevicius Ashok Vaseashta
NANOSCALE MATERIALS, DEVICES, AND SYSTEMS FOR CHEM.-BIO SENSORS, PHOTONICS, AND ENERGY GENERATION AND STORAGE A. VASEASHTA* On detail in Washington DC, USA from Nanomaterials Processing & Characterization Labs, Graduate Program in Physical Sciences, Marshall University, Huntington WV, USA
Abstract – A comprehensive overview of ongoing research efforts and future scientific directions in nanotechnology to develop materials, devices, and systems for potential use in environmental pollution monitoring and mitigation; energy generation and storage; and chemical-biological-radiological-nuclear sensing is presented. Applications of nanomaterials in development of biodegradable, high performance yet light weight and eco-friendly materials are presented to minimize power consumption, green-house gas emissions, and land-fill volume. Societal implications and concerns associated with nanotechnology are addressed by studying fate and transport and development of guidelines for a risk-assessment model. A roadmap of the future of nanomaterials, in-terms of complexity, nexus of disciplines, and emerging green nanotechnologies is presented.
Keywords: Chem.-bio sensors, pollution, satellite, water, energy, storage.
1. Introduction Three most imminent challenges of the 21st century include abundant clean energy, a pollution free environment, and international security and safety. It is imperative that the interdisciplinary scientific community exploit knowledgedriven transformations across scientific fields to develop materials, devices, and systems to address the aforementioned challenges. The scientific study
______ *
To whom correspondence should be addressed: Professor Dr. Ir. A. Vaseashta, email:
[email protected] A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
3
4
A. VASEASHTA
of nanoscale materials and systems is a promising field. Fundamental understanding and technological advances arise from the potential of nanoscale materials to exhibit unique properties that are attributable to their small size such as surface structure, physical characteristics, and chemical composition.1 These properties drive research that serves as a catalyst for new scientific and technological innovations. Furthermore, the geometrical dimensions span those comparable to the smallest engineered entity, the largest molecules of the living systems, and fundamental physical quantities render their study quite captivating. Materials approaching nanoscale dimensions exhibit atypical characteristics with numerous unique and hitherto unexploited applications. Advances in material synthesis, device fabrication and characterization techniques have provided the means to study, understand, control, or even manipulate the transitional characteristics between isolated atoms and molecules, and bulk materials. The unique characteristics and functionalities of nanomaterials have already been utilized in cosmetic, apparel, and sports industries, while proof-of concept electronic and optical devices have been demonstrated and are largely in the developmental stages. Recently, various nanoscale materials with new architectures and improved functionality have been developed with applications in chemical and biological sensors,2 environmental pollution sensing,3 monitoring,3,4 mitigation and remediation,3 next-generation energy generation and storage devices, nanobiotechnology, nanophotonics, in-vivo analysis of cellular processes and futuristic platforms in health and clinical medicine.5 The potential and risk for inadvertent or deliberate contamination of the environment, food and agricultural products, due to global threats of terrorism make decentralized chemical and biological sensing an important research area for academic institutions and governmental agencies. A nanotechnology based sensor platform enables direct electrical detection of biological and chemical agents in a label-free, highly multiplexed format over a broad dynamic range. Nucleic acid layers combined with nanomaterials-based electrochemical or optical transducers produce affinity biosensors such as the “DNA Biosensor” or “Genosensor” that are attractive devices for converting the hybridization event into an analytical signal for obtaining sequence-specific information in connection with clinical, environmental, or forensic investigations.6,7 A perpetual increase in population and consumption of fossil fuels to meet the current energy demand has led to increased pollution worldwide. Pollution in large cities has reached an alarming level and is widely perceived to be a leading contributor to chronic and deadly health disorders and diseases affecting millions of people each year. In a recent study, the World Health Organization (WHO) reported that over 3 million people suffer from the effects of air-borne pollution. Furthermore, reports from the World Energy Congress
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
5
(WEC) suggest that continued fuel consumption at its current rate will result in pollution creating irreversible environmental damage by 2025. Clinical studies show that inhaling particulate matter (PM) is associated with increased mortality rates that are further magnified for people suffering from diabetes, chronic pulmonary and inflammatory diseases. Of general pollutants that contaminate the urban environment, fine suspended PM, Nitrous Oxide (NOx), sulphur dioxide (SO2), volatile organic compounds (VOCs), and ozone (O3) pose the most widespread and acute risks; thus driving studies and measures, such as cap-and-trade, carbon credit, pollution credit etc., to limit emission of greenhouse gases (GHG). Figure 1 illustrates a pollutions footprint of various sources of energy and a proposed “future distributed-source energy solution”. Accordingly, preliminary results of joint investigations to monitor and mitigate envienvironmental pollution are presented. In addition to air pollution, results from joint investigations into the efficacy of nanostructured materials in the detection and remediation of water pollution are presented.
Figure 1. Sources of energy, pollution level, and distributed model of energy.8
Resistance to new technology and the fact that advances in nanotechnology have progressed faster than development of standards, “voluntary code of conduct” by industry, and implementation of regulations by policymakers, has led to the studies of societal implications and concerns associated with the production and use of nanomaterials. Studies of fate and transport of nanomaterials in air, water, and soil, plume modeling and in-silico risk-assessment models using expert elicitation and statistics based decision analysis are under active investigation and are briefly presented. Roadmap of time vs. complexity,
6
A. VASEASHTA
nexus of disciplines, and emerging green nanotechnologies is conceptualized to attain a technological singularity in future. 2. Size Effects in Reduced Dimensions Reduced dimensional systems, in which one or more dimensions are reduced such that material begins to display novel quantifiable quantities hold tremendous potential. For solids, typically reduced dimensions amount to reduction of the coordination number; hence the electrons have less opportunity to hop from site to site; thus reducing kinetic energy of electrons or their bandwidth. A higher Coulomb interaction/bandwidth ratio at a site enhances electron correlation and Mott-transition i.e. the tendencies towards the appearance of magnetism. Furthermore, the symmetries of the system are lowered and the appearance of new boundary conditions lead to surface states and interface states. A change of the quantization conditions alters the eigen value spectrum and transport properties of the solid. A high surface area/volume ratio in nanoscale materials alters mechanical and other physical properties. One critical impact is that surface stresses existing in nanomaterials have a different bonding configuration as compared to bulk atoms. As an example, surface elasticity is an effect that occurs due to the lack of bonding neighbors for surface atoms. The effects of the difference between surface and bulk elastic properties become magnified as the surface area/volume ratio increases with decreasing structural dimension. Studies to calculate surface elastic constants using MD simulations, curvature effect using the Cauchy-Born rule, and electronic effects via effective nuclei-nuclei interaction using DFT calculations provide better understanding of surface and interface effects in reduced dimensions. Despite many investigations on surface elastic effects, many questions remain unanswered. Recently, sensing by nanoscale materials has been utilized for chemicalbiological investigations; primarily because of increased surface area, reactivity, and ability of selective functionalization. For instance, nanoscale resonators9 have allowed various research groups to detect mass of molecules by detecting change in resonance frequency with mass.10 Surface effect may play a role in resonant frequency shift if the thickness of the biomolecular layer becomes comparable to the resonator’s thickness; although, the surface stress effect may affect the resonant frequency shift. We have employed the surface plasmon resonance (SPR) on the optical response2,11 as a function of biological molecule interactions resulting from adsorbate-substrate bonding. As the size of nanomaterials approach fundamental physical quantities and biological molecules (shown in Figure 2(l)), many questions remain unanswered to put the broad topic of emergence of size-effect in different materials into perspective. As an
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
7
Figure 2. (l): Dimensional compatibility, (r): roadmap of nanomaterials.
example, when does the surface effect become important? Is there a critical dimension for materials? Do some processes take place only in nanoscale dimensions and do other processes occur even in micro-scale dimensions? What do we know about the interactions of nanomaterials with biological molecules vis-à-vis “spes altera-vitae”? 2.1. NANOMATERIALS-ENVIRONMENT INTERFACE
Size and surface collectively control characteristics of nanoscale materials due to the existence of large boundaries adjoining its surrounding medium and interplay of physicochemical interactions. The surface free-energy is sizedependent and hence increases almost inversely with the decreasing feature sizes of the material. Collective response of a nanomaterial-medium system that is attributable to reduced dimensions, viz. size (area and distribution), surface structure (groups, functionalization), physical (electronic, optical, photoactivation, luminescence), and chemical (crystallinity, purity, solubility) is vital to developing a scientific model that predicts its response as a sensor, bioadverse response pathways for toxicity, adsorption pathways for materials for storage, and interaction with light for optical response. It is an exceedingly complex task due to a large matrix of parameters consisting of nanomaterials (metals, metal oxides, semiconductors, macromolecules and self-assembled), the surrounding environment, and influencing mechanisms of interactions.
8
A. VASEASHTA
Further complexities arise in nanomaterials production due to variations in, e.g. fabrication with precisely-controlled surface properties, uniform dispersion of nanoparticles in aqueous or organic mediums, linkage of nanoparticles to a polymeric material, and reproducible surface modification of nanoparticles. Hence, as the production and diversity of nanoparticles increases, it will become increasingly important to understand how engineered nanoparticles and biological systems interact in terms of bio-physico-chemical properties of engineered nanoparticles (Figure 2(r)). An extensive investigation is underway in the context of specific functions ranging from bio-nanomaterials interface to toxic potential of industrial pollution. We conducted study of functionalized SAM as biosensors that use biological molecules – usually enzymes, antibodies, or nucleic acids, to recognize simple molecules of interest via hydrogen bonding, charge-charge interactions, and other biochemical interactions to provide molecular information.7 Recent work on affinity biosensors to deliver real-time information about the antibodies to antigens, cell receptors to their glands, and DNA and RNA to nucleic acid with a complimentary sequence,12 provided a multitude of applications. As an example, they can be used to measure blood glucose levels, to detect pollutants and pesticides in the environment, to monitor food-borne pathogens in the food supply, to work as chemical and biological warfare agents, and detect the presence of micro-organisms in foods. A response of a nanomaterials based gas sensor is based on reactions replacement of atoms at the sensing surface of these materials which relies on a change of the resistance of the oxide. Depending on the free electron density in the space charge layer, the depletion region is increased. Since electric properties are influenced by the depletion layer, variation in electrical conductivity indicates sensor response. Similarly, moisture can influence the resistance or conductivity of oxide materials via two pathways: first, the adsorption of monolayer/s of water molecules at the surface; and second, the process of formation of a parallel resistance path by capillary condensation of water via adsorption of the water molecules as protons and hydroxyl groups within pores. The sensitivity and response of nanomaterials of metal-oxide sensors is highly dependent on the roughness of the substrate, which is caused by the increasing surface area and porosity of the film surface modifications in the film surface morphology. The sensitivity (and selectivity) of a can be improved by parameters such as decreasing the crystallite size, the valence control, and using noble metal catalysts. The characteristics which provide beneficial aspects are also believed to be responsible for toxicity of nanomaterials. Consequently, nanoparticle toxicity is studied in context of its ability to induce tissue damage through the generation of oxygen radicals, electron-hole pairs, and oxidant stress by abiotic and cellular
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
9
responses resulting in pro-inflammatory, mitochondrial injury and pro-apoptotic cellular effects in the lung, cardiovascular system and brain.13 It is further believed that nanoparticles absorb cellular proteins which could induce protein unfolding, fibrillation, and thiol cross-linking; leading to neuro-toxicity and reduced enzymatic activity. Nanoparticles which are cationic are also believed to induce toxicity via acidifying endosomes that lead to cellular toxicity and apoptosis in epithelial lung through endosomal rupture through proton sponge mechanism (PSM), mitochondrial targeting, and cytosolic deposition. Nanomaterials composed of redox-active elements are particularly reactive and can possibly provoke potentially damaging chemical transformations. Furthermore, even chemically benign nanoparticles may become activate by light absorption. Hence, fundamental understanding of a nanomaterial-surrounding medium is vital to sustaining technological advances of nanoscale materials as catalyst for new scientific and technological avenues. 2.2. NANOPHOTONICS AND SURFACE PLASMON RESONANCE
Two theoretical models, viz.: the classical electrodynamics for the propagation of light and the solid (or liquid) state theory for the interaction of light with the particle expressed by the complex, frequency dependent dielectric function, DF, of the particle material, which can mathematically be described as,14 according to Mie’s theory: İ(Ȧ) = İ1 + i İ2 = İDrude + Ȥ interband = 1 – (n e2/ İ0 me )/(Ȧ2 + i Ȗ Ȧ) + Ȥ interband
(1)
with n, e, İ0, me, Ȗ are respectively the electron density, elementary charge, field constant, effective mass and relaxation frequency of the conduction electrons of the Drude-Lorentz-Sommerfeld theory, Ȗ = 1/IJ with IJ the Drude relaxation time and (ne2/İ0me )1/2 = Ȧp the “Drude plasma frequency”. The interband transitions exist in most metals thus causing nanomaterials to demonstrate hybrid surface plasmon polariton (SPP) excitations. They consist of collective electronic excitations in the conduction band and (complex) polarization term for deep bands. Most nanomaterials exhibit optical absorption and scattering spectra with complex multi-peak features, characterized by quantities viz., peak height, spectral peak position ȦMax and band width ī. The dielectric function (DF) includes electronic size and surface/interface effects present in a nanoparticle. Although a detailed description is described elsewhere, it suffices to state here that Mie’s theory is used analogously to Fresnels formulae which was derived for samples with planar geometry. Extension of the theory yield effective
10
A. VASEASHTA
permeability and refractive index to produce negative refractive index materials (NRIM) by plasmonic materials to image sub-O features for producing near field super lens.15 The inverted Mie theory yields the realistic DF including all electronic and optical size and surface/interface effects and resulting evanescent field strongly interacts with adsorbed molecules, thus influencing the resulting spectra. The application of inverse Mie theory is essential for structural, electronic and optical effects for selective detection substances in aqueous/biosurroundings using noble metals as plasmonic nanostructures. Integrating plasmonic elements with existing silicon based technology will likely result in plasmonic nanophotonic arrays with multifunctional capability. 3. Chem.-Bio Sensors and Biomedical Platforms International security threats such as escalation in terrorist activities and asymmetric warfare by adversaries drive needs for novel materials and innovative platforms to detect, interdict, and counter threats due to chem.-biologicalnuclear-radiological-improvised explosive (CBRNE) agents, synthetic DNA, contact-poison, and improvised explosive devices (IEDs) in a time-efficient and reliable manner at the site of event. Additional potential risks include inadvertent or deliberate contamination of the environment, food and agricultural products, or even naturally occurring threats such as avian bird flu. In a possible weapons of mass destruction and terrorism (WMDT) scenario, rapid identification of CBRNE events will allow first responders and emergency personnel to implement critical decisions concerning barricading, evacuating, or efficient decontamination, saving hundreds of lives and preventing responders from becoming victims themselves. Conventional detection methods require time-consuming steps to arrive at meaningful data. As an example, methods such as enzyme linked immuno-sorbent assay (ELISA) and polymerase chain reactions (PCR) have been employed for pathogen detection. ELISA and PCR methods require enrichment, isolation, morphological examination, biochemical, and serological testing to positively identify pathogens. In clinical medicine, decentralized laboratory facilities allow mobile facilities for clinical analysis through direct-reading, portable, lab-on-chip (LOC) systems with wireless communications capabilities to centralized servers, command and control units. Recent progress in nanostructured materials based sensor platforms will significantly impact data collection, processing, and recognition to enable the direct detection of biological and chemical agents in a label-free, parallel, multiplexed, broadly dynamic range. This platform utilizes functionalized nanotubes, nanowires, and nanoparticles to detect a broad range of molecules including DNA, RNA, proteins, ions, cells, pH values, and molecular binding with high sensitivity and selectivity.
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
11
Biosensors are intrinsically simple and inexpensive systems that use molecules – usually enzymes, antibodies, or nucleic acids – to recognize simple molecules of interest via hydrogen bonding, charge-charge interactions and other biochemical interactions to provide pertinent molecular information (Figure 3(l)). Progress in nanomaterials and advances in fabrication processes provide opportunity to modify, embed in host matrix, or even customize the nanoparticles for use as highly sensitive and selective sensing materials. Nanomaterials based sensors are used in several configurations such as SPR, electrochemical, optical, electrical transduction, and as shock-wave generators for applications ranging from homeland security to studying the environment pollution. Nanostructures will improve our capability to detect CBRNE events with sensitivity and selectivity by several orders of magnitude, protect through filtration, adsorption, mitigation or neutralization of agents, and provide site-specific in-vivo prophylaxis. Joint investigative efforts are focused towards chemical and biological agents, and radiation detection due to a radiological-dispersal device (RDD) or “dirty bomb”.
Figure 3. (l): Electrochemical sensor basics, (r): SPR based sensor.
3.1. NANOMATERIALS BASED CHEMICAL-BIOLOGICAL SENSORS
CNTs are conducting, can act as electrodes, generate electro-chemiluminescence (ECL) in aqueous solutions, and can be derivatized with a functional group that allows immobilization of biomolecules. CNTs have high surface/volume ratios for adsorption, and have surface/weight ratios ~300 m2/g. The uniform chemical functionalization of CNTs is key to the formation of biosensors. Oxidation of nanotubes with HNO3-H2SO4 leads to high concentrations of carboxylic, carbonyl, and hydroxyl groups on the surface, and removal of the tip to expose the tube interior. Carboxyl groups can readily be derivatized by a variety of reactions allowing linking of biomolecules such as proteins, enzymes, DNA, or even metal nanoparticles. The covalent modification of nanotubes facilitates the creation of well-defined probes, which are sensitive to specific intermolecular interactions of many chemical and biological systems. Integration of the
12
A. VASEASHTA
transducer transducer and probe enables quick, accurate, and reversible measurement of target analytes without the use of reagents. Using sequence-specific attachment, NT-based electronic devices with specific molecular-recognition features of DNA have been reported.16 Covalent modification of single wall CNTs (SWNTs) offers mapping of functional groups at a molecular resolution. Furthermore, chemical processes to link catalysts, such as transition-metal complexes, to the ends of CNTs are useful in creating or modifying the structures at a molecular scale, creating interconnections for electronic devices, and even developing new classes of materials. Covalent functionalization of the sidewalls of SWNTs provides stability and best accessibility; but at the expense of damaging the sidewalls; thereby diminishing the mechanical and electronic properties. However, noncovalent routes to CNTs functionalization offer ease of synthesis and minimum disruption of the tubular structure. During interaction with the polymer coatings, the electrical properties of the nanotubes are altered, enabling detection of the molecules leading to a very sensitive sensing mechanism. In addition to nanotubes, novel materials such as porous silicon17 and porous carbon,18 with porosities of comparable dimensions to those of the biomolecules have been used for biosensor applications. The mesoporous carbon matrix is used for stable immobilization of the biological molecule, and C60 serves as an electron mediator. Both C60 and NTs are good electron mediators when used with a mesoporous carbon matrix or modified metal electrodes. CNT-based transducers, however, show a significant advantage over porous silicon due to the well defined, defect free structures, and also because the NTs promote homogenous electron transfer reactions. Efforts to sort batches of CNTs by length using high-speed centrifuges and functionalization to develop sensors for the food and agriculture industry, genetic analysis, proteomics, drug screening, clinical diagnostics and bio-warfare agent detection are underway. Furthermore, investigations using SAM based SPR and Atomic Force Microscopy (AFM) techniques are in progress to detect several pathogens. The SPR detection technique is rapid, real-time, and requires no labeling, and involves immobilizing antibodies by a coupling matrix on the surface of a thin film of precious metal, such as nanoparticles of gold deposited on the reflecting surface of an optically transparent wave-guide. The precise angle at which SPR occurs depends on several factors (Figure 3(r)). A main response is the refractive index of the metal film, to which target molecules are immobilized using specific capture molecules or receptors along the surface, that cause a change in SPR angle. This can be monitored in real-time by detecting changes in the intensity of the reflected light, producing a sensorgram. The rates of change of the SPR signal can be analyzed to yield apparent rate constants for the association and dissociation phases of the reaction. When the antigens
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
13
interact with antibodies, the refractive index of the medium surrounding the sensor changes producing a shift in the angle of resonance proportional to the change in the concentration of antigens bound to the surface. Various other sensor platforms, viz.: (a) chemo-mechanical micro-cantilever array to provide a quantitative and label-free platform for high-throughput multiplexed biomolecular analysis for detection of various biomolecules based on binding, (b) microarrays chips fabricated using technique matrix assisted pulsed laser evaporation (MAPLE) and laser induced forward transfer (LIFT) for deposition of biopolymers and a variety of biomolecules to detect dangerous gases, aerosols and micro-organisms, and (c) aligned-CNT probes for biomolecular recognition based on charge transport at the CNT transducer with the accuracy down to molecular level for quantitative and selective detection of a range of metabolites including cholesterol, ascorbic acid and uric acid, in buffer solution as well as in human plasma and blood are under investigation. Several other concepts being considered are resonance based9 shock-wave generators that only could detonate military explosives but will also detect biological and chemical weapons. The shock wave can be used for targeted delivery of drugs, chemotherapy, and cure cells without affecting the whole body. Likewise, nanothermite – a composite of fuel and an oxidizer, which in turn generate combustion waves that can hit velocities ranging from mach 4–7, can safely be used for killing cancer cells. Using electrospinning, we have investigated nanofibers prepared from high performance polymer composites embedded with metal oxides, glasses coated with rare earth metals, and biocompatible compounds for sensing, and even controlling electrical, optical, and chemical and biological response (Figure 4 – both panes). Nucleic acids offer analytical chemists a powerful tool for recognition and monitoring of many important compounds. Recent advances in molecular
Figure 4. Use of electrospinning for chem.-bio sensors, health and medicine.11
14
A. VASEASHTA
biology are used to study the effects of proteins and drugs on gene expression, viz. gel mobility shift, filter binding, DNA foot-printing and fluorescence-based assays. Most of these methods; however, are indirect and require various labeling strategies. Electrochemical DNA biosensors play an important role for clinical, pharmaceutical, environmental and forensic applications, because they provide rapid, simple and low-cost point-of-care detection of specific nucleic acid sequences. In recent years, there has been a growing interest towards design of electrochemical DNA biosensors that exploit interactions between surface-confined DNA and target drugs/biological molecules for rapid screening.16,19 Binding of small molecules to DNA primarily occurs in three modes: electrostatic interactions with the negative-charged nucleic sugar-phosphate structure, binding interactions with two grooves of DNA double helix, and intercalation between the stacked base pairs of native DNA. Most electrochemical sensors use different chemistries; and employ interactions between the target, the recognition layer and an electrode. We have followed numerous approaches to electrochemical detection including direct electrochemistry of DNA and devices based on DNA-mediated charge transport chemistry. In direct electrochemical DNA sensors, the analysis is based on a guanine signal where a basepairing interaction recruits a target molecule to the sensor, allowing monitoring of drug/biological molecule-DNA interactions, which are related to the differences in the electrochemical signals of DNA binding molecules for DNA barcoding. It is vital to develop sensing strategies to maintain critical dynamics of target capture to generate a sufficient recognition signal. Standard electrochemical techniques, such as differential pulse voltammetry (DPV), potentiometric stripping analysis (PSA), square-wave voltammetry (SWV), etc. are used as genosensors. Since genosensors are compatible with existing micro and nanofabrication technologies, they enable design of low-cost, devices that offer potential for detection and diagnosis of inherited diseases and clinical potential for detecting pathogenic bacteria, tumors, genetic disease, and forensics via credit card-sized sensor arrays. In the context of security, prevention against agroterrorism is by detecting bovine spongiform encephalopathy (BSE). A possible link between BSE and a disease of humans, a variant form of Creutzfeldt-Jakob Disease (vCJD) using nano-technology based platforms is important to provide clues about the disease. 3.2. USE OF NANOMATERIALS IN HEALTH AND MEDICINE
Sensing strategies can be modified towards applications of nanomaterials in targeted drug delivery, diagnostic, and therapeutic actions,5,20 thus creating pathways for use in health and medicine. By incorporating a drug into bio-degradable polymers, a simple and convenient drug delivery system allows time-controlled
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
15
drug regimen release. We have studied nanofibers for promising medical applications including treatment of primary pulmonary hypertension (PPH), and pulmonary arterial hypertension (PAH) by time-controlled release templates impregnated with anticoagulants and calcium channel blockers, a bioscaffold that mimics extracellular matrix (ECM) topology, polyesters com-bined with phosphatidyl choline for biomimetic applications, intravascular stents from a blend of polyactide and trimethylene carbonate, cellulose based scaffold for cartilage tissue engineering, and esophageal tissue engineering. We studied dystrophin gene immobilized on nanostructured templates as a drug carrier that initiates regeneration and boosted satellite cells mediated repair mechanisms for affected muscles for treating Duchene muscular dystrophy (DMD).21 Likewise, a nanoparticles–cinobufagin-bovine serum albumin based drug delivery mechanism holds promise in treating the hepatocellular carcinoma (HCC) malignancy. Nanotechnology based platforms offer novel opportunities to sense clinical biomarkers by imaging for therapeutic intervention. The use of iron oxide nanoparticles in an MRI system provides 3D image and normal and cancerous hepatocyte cells information.22 Furthermore, nanosized constructs such as dendrimers, liposoms, nanoshells, nanotubes, nanoemulsions, quantum dots (QDs), and even viruses offer use as imaging agents intended as non-invasive probes or targeted disease biomarkers. Gold nanoparticles are used to identify the pathogenic bacteria in a DNA microarray technique,23 and QDs are used to detect the human Y chromosome,24 and for locating cancer markers in cellular imaging.25 One to ten nanometer sized QDs with unique photochemical, photophysical, and physiological properties are proposed for drug delivery or contrast agents in MRI. For medical diagnostics, QDs of varying diameter are embedded into polymeric microbeads to achieve biological assays by multi-color optical coding. Tissue engineering is a relatively new field that seeks to regenerate human tissues through the use of some combination of cells, bioactive molecules such as drugs and mechano growth factors (MGF), and a biomaterial support system or scaffold.26 Hence, biomimetically driven studies are exploring how the topography of a surface can be used to control cell behavior.27 By changing the biomaterial surfaces at nano level, we can observe different types of cell behavior, change in cell adhesion properties – especially for fat-free mass, cell orientation and motility, cytoskeletal condensation, and elastomers for vascular tissue engineering28 to monitor advanced chronic pulmonary diseases (COPD). Recent applications of nanotechnology in dental care (nanodentistry) will permit maintaining near perfect oral health via nanomaterials, biotechnology, and nanorobotics.29 In fact, nanofillers are common nano-dentistry. Additional procedures include renaturalization, permanent hyper-sensitivity cures, and orthodontic realignments. Nanomaterials provide higher mechanical strength, enhanced bioactivity, and restorability for debilitating bone fractures. Nanomaterials-based
16
A. VASEASHTA
orthopedic applications include nanopar-ticles, nanofibres, nanoscaffolds, nanotubes and nano-composites as orthopedic implants. Developing intelligent biomaterials that closely mimic the molecular composition, mechanical responsiveness, and nanoscale organizations of the natural extracellular matrices (ECM) is vital. Recent reports indicate that adding gold nanoparticles to a modified version of an HIV combating drug, TAK-779 creates a compound that prevents the virus from gaining a cellular foothold. These novel and smart biomaterials, combined with defined biophysical cues and biological factors are essential for functional tissue regeneration. 4. Energy Generation and Storage Driven by population growth and the wave of industrialization in developing countries, energy consumption worldwide is increasing relentlessly. Moving forward, an alternate means to generate and store energy is imperative to ease environmental resource constraints, and to push the economy and society to sustainable development. Reports from the European Union’s (EU) recent research framework program (FP) and the United States’ National Nanotechnology Initiative (NNI) have listed several strategies that identify nanotechnology’s role in power generation and storage. Figure 1 further suggests that future power sources will be distributed allowing reduced dependence on fossil fuels leading to pollution reductions and GHG; that are consistent with Jeremy Rifkin’s model of peer-to-peer, shared, decentralized, and distributed generation using hydrogen energy. Significant nanotechnology contributions will influence research in fuel cells, hydrogen generation and storage, improvement in photovoltaic conversion efficiency, super-strong light-weight materials to reduce power consumption, electrical and optical devices requiring less power yet providing higher lumens, and many others. 4.1. HYDROGEN GENRATION AND STORAGE
Hydrogen produces more energy/volume than any other known sources of fuels. The future of a “hydrogen economy” depends on developing clean, safe and efficient methods for producing and storing hydrogen. Hydrogen can be produced by several methods, such as thermolysis, electrolysis, photocatalysis, hybrid processes, and gasification and CO2 sequestration. For sustainable energy, fuel cells (FC) are an important enabling technology to produce clean and renewable hydrogen energy. There are several different types of fuel cells, e.g. alkali, molten carbonate (MCFC), phosphoric acid (PAFC), proton exchange membrane (PEMFC), and solid oxide fuel (SOFC). We focus on PEM based fuel cells. Technically, water splitting can be achieved with coupled solar cell,
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
17
i.e. photo-electrolysis systems. A photo-electrochemical (PEC) cell comprises a semi-conducting photo-anode, an aqueous electrolyte, and an inert counterelectrode. The main advantage of the PEC cell is that the evolving hydrogen and oxygen gasses can be collected in separate volumes. The PEC cells are elegant, however are limited in their applicability due to several practical considerations such as hydrogen and oxygen forming in the same volume; limited lifetime of photo catalysts after which the efficiency drops below acceptable levels; and slow charge transfer across the semiconductor/electrolyte interface. Most research efforts on solar water splitting are focussed on systems with photo-catalyst powder suspended in an aqueous solution. In both photoelectrodes and powder-based photo-catalysts, surface reaction kinetics plays a key role. The PECCS cells combine the advantages of a two-electrode system with simultaneous storage of hydrogen in a metal hydride (MH). Once the MH electrode is fully loaded with hydrogen, it can be withdrawn from the cell and placed in a fuel cell system. Upon heating the electrode, the hydrogen is released and can be used to generate power. A Nernst-type sensor measures the concentration of hydrogen formed in the MH. Using a second MH reference electrode with a fixed hydrogen activity, open-circuit voltage across both MHs, EOC is given by the Nernst equation:30
EOC
RTt H F
§ a ( H ) sample · ln ¨ ref ¸ © a( H ) ¹
(2)
R is the gas constant (8.314 JK–1), T is the temperature (in Kelvin), F is Faraday’s constant (96,485 Cmol–1), tH+ is the ionic transport coefficient for protons, and a(H+)sample and a(H+)ref may be substituted with the hydrogen activities in the MH counter- and reference-electrodes, respectively. The current state-of-the-art membranes are fabricated from perfluorinated sulfonic acid (PFSA) polymers, such as Nafion® and Hydrogen uranyl phosphate tetrahydrate, HUO2PO44H2O (HUP) which can be easily synthesized and show high proton conductivity at room temperature.31 These materials have useful, high proton conductivities when fully hydrated limiting their usefulness to temperatures <100°C. Understanding of parameters such as proton transport, water diffusion, and electro-osmotic drag; ion transport typically described by Grotthus mechanism and Vehicle mechanism; and water management issues are crucial to a successful fuel cell design; as water is being constantly transported from the anode to the cathode leading to flooding. We are studying a new series of polymers based on pyridine cycles originated from poly-acrylonitrile (PAN) in a thermo-oxidative process. The thermo-oxidation in air and at moderate temperature (200–400°C) converts PAN into ladder polymers made of fused pyridine cycles. The membranes are obtained by PAN-SiO2 nanoparticles
18
A. VASEASHTA
solution casting in thin films with simultaneous thermo-oxidation process.31 These PEMs have ability to operate at moderately elevated temperatures, with little to no external humidification, and to be manufactured inexpensively. Suitable photo-anode materials for efficient solar hydrogen generation should have high visible light absorption; high photochemical stability; suitable bandgap (preferably graded); efficient charge transport in the semiconductor; and the conduction and valence band edges should “straddle” the reduction and oxidation potentials of water. Virtually all photo-electrodes display a trade-off between photochemical stability and visible-light absorption. TiO2 has been the most extensively investigated material for photo-electrochemical and photo-catalytic applications. An alternative material that has an intrinsically small bandgap is D-Fe2O3 (hematite). We studied feasibility of solar photo-electrolysis with a system in which an n-TiO2 semiconductor electrode was connected though an electrical load to a platinum back counter electrode placed under near-UV light. When the surface of TiO2 electrode was irradiated with UV light of wavelengths <415 nm, photocurrent flows from the platinum counter electrode to the TiO2 electrode indicating that the oxidation reaction (oxygen evolution) occurs at the TiO2 electrode and the reduction reaction (hydrogen evolution) at the Pt electrode. Hence water splits into O2 and H2 with UV light, without any external voltage, and is described as follows: Excitation of TiO2 by light At the TiO2 electrode At the Pt electrode The overall reaction is
=> => => =>
TiO2 + 2hȞ ĺ 2e– + h+ H2O + 2h+ ĺ ½ O2 + 2H+ 2H+ + 2e– ĺ H2 H2O + 2hȞ ĺ ½ O2 +H2
Recently, hydrogen storage in MHs has been integrated into the photoelectrolysis cell, thus leading to a new type of devices termed as photo-electro chemical conversion and storage (PECCS) cells; in which the cathode is MH allowing in-situ storage of the generated hydrogen. Our efforts are focused on preparation of new materials for PEM fuel cells. Activated carbons (AC), SWNTs and metal–organic frameworks (MOFs) are being investigated as hydrogen adsorption and storage materials.32 The hydrogen storage capacity of nanofibers ranged from |15 > Ȥ > 1 wt % at moderate pressures and temperatures.33 Experiments on SWNTs yielded adsorbed amounts <10 wt % due to narrow pore size distribution of SWNTs. US Department of Energy (US DoE) has recently set performance targets by 2010 for on-board vehicular hydrogen storage systems, e.g. system gravimetric capacity (2 kWh/kg), volumetric capacity (1.5 kWh/L), durability/operability parameters, and charging/ discharge rates. Approaches such as compressed hydrogen gas, cryogenic liquid/ gas, chemical hydrogen storage, high-surface area adsorbents, and MHs are considered at the present time for on-board storage and regeneration and
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
19
off-board regeneration and refilling in terms of parameters such as thermodynamics, kinetics, potential durability/cyclability, and hydrogen discharge kinetics. In general, the materials for hydrogen storage should possess high hydrogen content, low heats of dehydrogenation, and fast desorption kinetics at FC temperature of operation. MHs in particular received a lot of attention recently due in part to inherently safe operation, low operation pressure, low cost, and high storage capacities. Early work on intermetallic hydrides showed good sorption/desorption kinetics but extremely low storage capacities. Since then, many complex MHs have successfully been used for reversible hydrogen storage. Despite several drawbacks, complex metal borohydrides M(BH4)n prepared from direct or indirect (more practical) synthesis are potential contenders for future hydrogen storage. For conditions where hydrogen release and uptake is controlled via temperature and pressure, Ammonia Borane (AB = NH3BH3) has received significant attention in light of its stability and commercial availability. 4.2. ENERGY GENERATION AND RESOURCE CONSERVATION
Nanotechnologies have opened promising new routes for making inexpensive solar cells such as 3D-nanostructured solar cells based on flexible substrates. To develop contacts, nanowires are grown directly onto contact pads to increase carrier collection efficiency. Enhancing the efficiency of solar cells by using electro-active and conjugated poly-(3-hexylthiophene) polymers, composites doped with rare earth metals to shift the incident spectrum to optimize photovoltaic response, active interfaces to reduce the distance of the photo-produced charge carriers, and increasing anti-reflection effects are under active investigation. Also, experimental efforts include nanomaterials in beta-voltaics as a source of power for satellite sub-modules drawing energy from ionizing radiation in space to convert in electrical energy. The need for high-performance and light-weight materials that reduce product weight without compromising cost, performance, or safety is growing. New or improved materials to reduce fuel-consumption, meet consumer expectations, military needs, and regulatory mandates are needed for transport industries. In addition, lightweight materials and structures for body armor and protection, building and construction, sports and leisure goods, and packaging are in demand. Nanomaterial characteristics lead research developments of lightweight materials and structures that withstand arduous operating conditions from high temperatures, dynamic loading, thermal and mechanical stresses, enhance resistance to blast and shock, and enhance ballistic performance without structural degradation. Lightweight components in transport and aviation industries will reduce fuel consumption and environmental pollution. Investigated materials include
20
A. VASEASHTA
carbon fibers, protective coatings, honeycomb, or bonded structures. Using electrospinning, we developed nanofibers using high performance polymers impregnated with CNTs. 4.2.1. Greenhouse gases sequestration The U.S. EIA predicts a 33% increase in CO2 emission by 2030 unless appropriate sequestration and mitigation technologies are instituted. The CO2 sequestration program contributes to the President’s initiative for developing technologies for reducing GHG emissions, lowering the cost and energy penalty associated with CO2 sequestration from large point sources and understanding factors affecting CO2 storage durability, capacity, and safety for geologic formations and ecosystems. Nanoporous carbon and liquid amine absorption are used to adsorb and sequester CO2; however large scale capture requires significant improvements to overcome drawbacks arising from large amounts of energy needed for regeneration, corrosiveness, and solvent degradation. Advanced membrane reactors34 use fossil fuel conversion reactions such as the steamreforming and water-gas-shift with in-situ separation of a reaction product. The fuel chemical equilibrium is hence shifted resulting in reduced thermodynamic losses since conversion is increased significantly as compared to conventional unit operations. A nanostructured ceramic membrane reactor based on alumina allows in-situ separation of H2 from CO, CO2, and possibly H2O gas mixtures. Furthermore, our efforts include sequestering CO2 using electrospun metal oxide nanofibers and propylene oxidation to propylene-epoxide (PO) using a variety of Au/TiO2 catalysts. 5. Environmental Pollution Monitoring, Detection, and Remediation In our technologically dominated society, the environment has been stressed from increased consumption and ever-rising energy demands for better quality of life. Scientific debate supporting and contradicting evidence relating pollution to the earth’s climate change are ongoing, but its impact on health is irrefutable. Nanotechnologies may improve the quality of life by reducing pollution at its origin by use of alternate sources of energy, mitigating pollutants, and curing symptoms from pollution. We have employed principles of green chemistry for producing nanomaterials – “green nanotechnology,”8 and devices that offset fuel consumption and in turn environmental pollution. Satellites play a major role in communication, navigation, climatology, surveillance, and environmental monitoring, and support many applications. Nanotechnology and micro/nano electromechanical systems (MEMS/NEMS) hold the potential to revolutionize the field of satellite design for environmental
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
21
pollution monitoring. Using image processing of the satellite data, we can extract pollution and optical data for any given city. We have conducted experiments with nanoscale metal-oxides sensors to detect pollution. The sensors are connected to a personal digital assistant (PDA) with GPS and wireless communication capability and web-server. Using data from this proptotype device, we have superimposed data from satellite to cross-reference the pollution coordinates. We expanded the investigation to mitigate pollution in air and water from nanoscale materials. The use of such interdisciplinary and transformational technologies has initiated a widespread perception-based public debate with scope ranging from favoring further research to its moratorium in advance of scientific risk assessment. This section provides a brief discussion encompassing ethical and legal considerations, fate and transport, and in-silico riskassessment. 5.1. POLLUTION MONITORING, DETECTION AND REMEDIATION
Satellite image data has traditionally been unexploited for atmospheric pollution studies. Satellite image data consists of earth radiances influenced by the temperature, emissivity of the ground surface, and the atmospheric column above, and its surroundings as observed by its sensors in different bands. Satellite image data can aid in detection, tracking, and understanding of pollutant sources and transport by providing observations over large spatial domains with 3D Models. The pollution assessment of optical atmospheric effects can also be quantified by aerosol optical thickness (AOT). The Differential Optical Absorption Spectroscopy-Method (DOAS) algorithm can analyze the data of the Global Ozone Monitoring Experiment (GOME) on ERS-2 and the SCIAMACHY on ENVISAT to retrieve NO2 columns in the 425–450 nm regions with high accuracy. City specific data can be modeled and correlated with data from ground based stations, such as the U.S. environmental protection agency (EPA), the Texas Commission on Environmental Quality (TCEQ), France’s AIRPARIF, the Netherlands’ National Institute for Public Health and Environment (RIVM), British Atmospheric Data Centre (BADC), and the Natural Environmental Research Council (NERC) among others. There is observational evidence that these aerosols can alter cloud properties. Scattering albedo increases with pollution, thus decreasing backscatter fraction. Aerosol radiative forcing depends on hygroscopicity, which in turn depends on aerosol photo-chemistry. IR absorption spectroscopy effectively identifies trace pollutants in both ambient air and smog systems. Employing Advanced Space borne Thermal Emission and Reflection Radiometer (ASTER) data, we have investigated air pollutants using satellite images 3 comprised of high spatial and spectral resolution images from the TIR and SWIR bands; and
22
A. VASEASHTA
compared the satellite data with EPA air quality and spectroscopic study data of atmospheric pollution. A conceptual model of the interplay of data from satellite and ground based stations is shown in Figure 5(l).3 ASTER is an imaging instrument flying on the Terra satellite as part of NASA’s Earth Observing System (EOS). It is the only high-spatial resolution instrument on the satellite that has 14 bands from the visible to the thermal infrared region. In the visible green and nearinfrared (V-NIR) range between 0.52–0.86 μm, there are three bands with 15 m resolution. In the SWIR range between 1.6–2.43 μm, there are six bands with 30 m resolution. In the third range between 8.125–11.65 μm, there are five bands with a resolution of 90 m. ASTER acquires data over a 60 km swath of which center is pointable. Satellite observed data is processed employing the Multivariate techniques, which are based on groupings in a multivariate data set. Remote sensing software ER-MapperTM (v.7.x) is employed to accurately model the data through feature extraction processes for pattern recognition. Several digital imaging processes such as geometric image registration, radiometric normalization, principal component analysis, and data fusion process the image for accurate feature extraction. Multidate correlations and regression analysis methods are used for individual gases with satellite-recorded reflectance of bands. Nanomaterials satellite components as visualized in Figure 5(r) will improve the resolution, composition, and library of information.
Figure 5. (l): Model of data from satellite and ground based stations, (r): nano-satellite.
A variety of nanomaterials have been studied for their abilities to detect and mitigate pollutants in air,3 water, and soil. We have investigated various ways in which nanomaterials can successfully be used to reduce/isolate atmospheric pollution, such as use of CNT filled in a polymer composite matrix to create a static discharge to remove PM from incoming air, and chemical protective clothing, and breathing filtration masks. Employing immobilized
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
23
TiO2 films, produced by sol-gel process, we explored removal of indoor odors under weak UV illumination. At 10 ppmv weak UV light of 1 μW/cm2 was sufficient to decompose such compounds in the presence of TiO2 photocatalysts. In an experiment, 150 ȝl of an E. coli suspension, containing 30,000 × cells was placed under weak UV (1 mW/cm2) illuminated TiO2-coated glass plate showing antibacterial effect. We are focusing on mitigating water contaminants such as Arsenic (As), which is found in the natural environment, and also from anthropogenic activities. Several studies have linked long-term exposure to small concentrations of As with cancer and cardiovascular, pulmonary, immunological, neurological and endocrine effects; thus necessitating studies for developing efficient methods for removing As from drinking water. We have employed a method based sorption on iron (III) oxides, such as amorphous hydrous ferric oxide (FeOOH), poorly crystalline hydrous ferric oxide (ferrihydride) and goethite (Į-FeOOH) which have been found to be effective in removing both As (V) and As (III) from aqueous solutions. Preliminary results with synthetic akaganeite (ȕ-FeOOH) have shown that the maximum capacity of the sorbents is around 75 mg of As per gm of akaganeite. Also, quite good results were obtained with synthetic magnetite/maghemite, with a maximum capacity of ca. 40 mg of As per gm of sorbent. Incorporating iron oxides/ oxyhydroxides based nanostructures in zeolites and nanoscale zero-valent iron (nZVI) technology provides a method for high capacity As removal. 5.2. FATE AND TRANSPORT OF NANOMATERIALS
The mode of transport, interaction and chemical affects on the human body of atmospheric pollutants is not well understood. Despite beneficial effects and major developments in the field of nanomaterials, there is a significant gap in our knowledge of the environmental, health, and ecological impacts associated with nanostructured materials. A comprehensive and fundamental investigation into the dynamic transport of nanomaterials in the environment and its impact on human health and ecology is needed to guard public welfare. The complex nature of naturally occurring and engineered nanomaterials and transport, either in the environment or via different exposure routes with human body necessitates an ontological modality. A matrix of parameters which govern fate and transport modeling of nanomaterials such as exposure routes, chemical composition, surface structure, solubility, size and shape effects, toxicity, absorption, distribution, metabolism, agglomeration, and excretion rate and mechanisms is under investigation via three most common routes – inhalation, ingestion, and dermal exposure. The essential aspect of employing ontology is in defining not only the item, but also its relationship to other items in the lexicon; to enable information
24
A. VASEASHTA
retrieval from various sources, databases, and expert elicitation. Biodegradation and bioaccumulation of nanomaterials should be sufficiently addressed due to impacts on human health. Studies relating to the thermodynamic properties, interfaces, and free energy of nanoparticles as a function of particle size, composition, phase and crystallinity influence particle dissolution in a biological environment. The accumulation, dispersion, and functional surface groups play an important role in cytotoxicity and in evaluating pathways of cellular uptake, subcellular localization, and targeting of subcellular organelles. This investigation in conjunction with plume modeling and predictive science approaches will assist in prioritizing, testing, and correlating with in-vivo exposure models. 5.3. SOCIAL ACCOUNTABILITY AND RISK ASSESSMENT
The understanding of environmental effects and health risks associated with nanotechnology is very limited and sometimes contradictory. At present, a well defined risk assessment modality is needed, as the “voluntary code of conduct” for using nanomaterials is somewhat trepidation based and not entirely on scientific methodology. Voluntary codes cover seven general principles, including taxonomy, significance, implementation, sustainability, precaution, inclusiveness, and accountability. Risk analysis is a process of evaluating critical assets and systems, threats, vulnerabilities, and controls for mitigating threats. Effective risk analysis involves identifying, assessing, and mitigating risk to a satisfactory level. Outcomes of risk assessment form a strategic plan for managing risks. Nanotechnology risk assessment occurs in a climate of uncertainty and change; therefore, effective decision making by participating experts is critical for successful outcomes. According to Howard,35 high quality decisions are characterized by the following elements: appropriate framing of decisions; analysis of creative alternatives; accurate information and sound models; clear preferences for future status; sound logic; and commitment to process and outcomes. Policy and decision makers are increasingly relying on expert-opinion elicitation techniques for forecasting advances, reliability, and risks related to science and technology. Structured expert-opinion elicitation techniques effectively support complex decision-making in the face of risk and uncertainty. Since nanotechnology risk assessment is in its infancy, incorporating formal expertopinion elicitation methods within the risk assessment process may help to prevent cognitive biases and faulty cognitive processes attributing to poor decision quality. The Delphi Technique is a structured interactive group communications technique effective for reaching consensus about judgments, forecasts, or
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
25
decisions from expert panels. The Delphi process occurs as follows: a facilitator distributes survey to the expert panel; the survey is answered anonymously and independently by the expert panel; a facilitator summarizes and distributes results and rationale, and the expert panel anonymously and independently reviews summarized results and rationale, and panelists are allowed to revise individual responses. The process of eliciting, summarizing, and distributing anonymous and independent responses continues until consensus, stable responses, or a given number of rounds is met. The Delphi technique has proven merit in forecasting trends and risks for many scientists, researchers, policy, and decision makers.36 6. Conclusions, Discussion, and Future Directions Nanotechnologies promise a cornucopia of new products with superior performance and characteristics, and offer solutions to some of the greatest challenges of the 21st century. As proposed in Figure 2r, further advances are necessary to entirely understand, characterize, develop, and optimize the properties of these materials, devices, and systems. A key challenge for such applications is precisely controlled growth of these materials at desired sites with a desired structure and orientation. Specific to remote sensing via satellite, advances in nanophotonics will enhance resolution, thus providing enhanced feature extraction capabilities. Future pollution remediation nanomaterials-based techniques may include pollution transport by nanoparticles which will have medical, radiological, and even national defense implications in terms of human health, safety, and environment. Ontological approaches towards studying and understanding the interaction of nanoparticles with the human body, nanoparticles dynamics in air and aquatic systems, composition-dependent disposition and dispersion of nanoparticles, short and long-term effects of nanomaterials with the human body, immunotoxocity and phototoxocity of nanoscale materials will address some of the societal issues of nanotechnology. Furthermore, this investigation aims to reconcile two visions of the future: first, the transcendence of nanomaterials and systems into nature of self-healing, brain-like computing or artificial intelligence, virtual reality, and bio-mimetic applications; and second, the process in nature produced nanomaterials and system via self-assembly, green chemistry synthesis, and other room temperature processes. The path to reach this technological singularity; specifically the technology to mimic nature and that of nature to develop technology is complex, yet attainable. A roadmap of nanomaterials, in-terms of complexity, discipline nexus, and emerging technologies is displayed in Figure 2r envisaging the invisible boundary between vision and reality.
26
A. VASEASHTA
ACNOWLEDGEMENTS
Immense acknowledgements are due to the NATO for funding for this ASI, Ms. Annelies Kersbergen for help, guidance, and patience during the preparation of this proceeding, and my colleagues Profs. Mihailescu, Reisfeld, Denkbas, Erdem, and Stamatin and students for fruitful discussions and contributions.
References 1. Vaseashta, A., Dimova-Malinovska, D., and Marshall J. (2005), Nanostructured and Advanced Materials, Springer, Dordrecht. 2. Vaseashta, A., and Stamatin, I. (2007), JOAM 9(6), 1506–1613. 3. Vaseashta, A., Vaclavikova, M., Vaseashta, S., Gallios, G., Roy, P., and Pummakarnchana, O. (2007), Science and Technology of Advanced Materials 8, 47–59. 4. Pumakaranchana, O., Phonekeo, V., and Vaseashta, A. (2008), this volume. 5. Denkbas, E., and Vaseashta, A. (June 2008), NANO: Brief Reports and Reviews. 6. Erdem, A. (2008), this volume. 7. Erdem, A., Karadeniz, H., Caliskan, A., and Vaseashta, A. (June 2008), NANO: Brief Reports & Reviews. 8. Vaseashta, A., Riesfeld, R., and Mihailescu, I. (2008), MRS 2008 Spring Proceedings. 9. Pokropivny, V., Pokropivny, P., and Vaseashta, A. (2005), In: Nanostructured and Advanced Materials, A. Vaseashta et al., eds., Springer, Dordrecht, pp. 367–370. 10. Davis, Z., Svendsen, W., and Boisen, A. (2006), Proceedings of the 32nd International Conference on Micro and Nano Engineering, MNE 2006, Spain, pp. 475–476. 11. Vaseashta, A., Erdem, A., and Stamatin, I. (2006), MRS 2006 Spring Proceedings 920. 12. Wang, J. (2003), Analytica Chemica Acta 500, 247. 13. Sahm, T., Rong, W., Bârsan, N., Mädler, L., and Weimar, U. (2007), Sensors and Actuators B: Chemical 127(1), 63–68. 14. Wiederrecht, G. (2004), European Physical Journal Applied Physics 28, 3–18. 15. Anantha Ramakrishna, S. (2005), Reports on Progress in Physics 68, 449–521. 16. Ligia, R., Furlan, A., Garrido, L., Brumatti, G., Amarante-Mendes, G., Martins, R., Cândida, M., Facciotti, R., and Padilla, G. (2002), Biotechnology Letters 24. 17. Sailor, M., Schmedake, T., Cunin, F., and Link, J. (2002), Advanced Materials 14, 1270. 18. Baizeng, F., Haoshen, Z., and Honma, T. (2006), Journal of Physical Chemistry B Condensed Matter, Materials, Surfaces, Interfaces & Biophysical 110, 4875–4880. 19. Wang, S., Zhang, Q., Wang, Q., and Yoon, S.F. (2003), Biochemical & Biophysical Research Communication 311, 572. 20. Cui, Z., and Mumper, R. (2003), Critical Reviews in Therapeutic Drug Carrier Systems 20, 103–137. 21. Vaseashta, A., Boskovic, O., Webb, Al., Ozdemir, N., and Ozturk, E. (2006), In: Functional Properties of Nanostructured Materials, R. Kassing et al., eds., Springer, Dordrecht. 22. Smith, A., Gao, X., and Nie, S. (2004), Photochemistry and Photobiology 80, 377–385. 23. Wu, X., Liu, H., Liu, J., Haley, K., Treadway, J., Larson, J., Ge, N., Peale, F., and Bruchez, M. (2002), Nature Biotechnology, 21, 41–46. 24. Jain, K. (2003), Expert Reviews in Molecular Diagnostics 3, 153–161. 25. Tan, Q., Ji, J., Barbosa, M., Fonseca, C., and Shen, J. (2003), Biomaterials 24, 4699–4705. 26. Curtis, A., and Wilkinson, C. (1999), Biochemistry Social Symposium 65, 15.
NANOSCALE CHEM.-BIO SENSORS AND ENERGY DEVICES
27
27. West, J., and Halas, N. (2000), Current Opinion Biotechnology 11, 215–217. 28. Shi, H., Tsai, W., Garrison, M., Ferrari, S., and Ratner, B. (1999), Nature 398, 593–597. 29. Öztürk, E., Denkbaú, E., and Berestein, G. (2006), 7th International Biorelated Polymers Symposium, in ACS Meeting, 10–14. 30. Wang, C., Wang, X., Lei, Y., Chen, C., and Wang, Q. (1997), International Journal of Hydrogen Energy 22(12), 1117–1124. 31. Ebrasu, D., Stamatin, I., and Vaseashta, A. (June 2008), NANO: Brief Reports & Reviews. 32. Becher, M. et al. (2003), Comptes Rendus Physique 4, 1055–1062. 33. Benard, P., and Chahine, R. (2007), Scripta Materialia 56(10), 803–808. 34. Prasad, P., and Elnashaie, S. (2004), Industrial & Engineering Chemistry Research 43(2), 494–501. 35. Howard, R. (2007), In: Advances in Decision Analysis: From Foundations to Applications, W. Edwards, R. Miles, Jr., and D. von Winterfeldt, eds., Cambridge University Press, New York, pp. 32–56. 36. Vaseashta, A. (2008), Presented at the NATO ARW “Risk, Uncertainty and Decision Analysis for Nanomaterials: Environmental Risks and Benefits and Emerging Consumer Products”, Faro, Portugal, 2008.
NANOSTRUCTURED THIN OPTICAL SENSORS FOR TRACE GAS DETECTION C. RISTOSCU1*, I.N. MIHAILESCU1, D. CAITEANU1, C.N. MIHAILESCU1, TH. MAZINGUE2, L. ESCOUBAS2, A. PERRONE3, AND H. DU4 1 Lasers Department, National Institute for Lasers, Plasma and Radiation Physics, PO Box MG-36, Bucharest – Magurele, RO-77125, ROMANIA 2 Université Paul Cézanne Aix-Marseille III Laboratoire TECSEN, Domaine Universitaire de Saint Jérôme, Marseille, FRANCE 3 University of Salento, Physics Department and INFN, Lecce, ITALY 4 Advanced Materials Research Institute, Northumbria University, UK
Abstract – We report on new features of nanostructured coatings used in trace gas detection applications based on the modification of optical parameters. We review the general physical principles underlying the operation of optical detectors and introduce the basic systems used in trace gas recognition. Pd-doped SnO2 thin films were pulsed laser deposited on Si (100) and quartz substrates in 10 Pa O2 at different substrate temperatures. Coatings on Si were investigated by scanning electron microscopy and atomic force microscopy to determine their surface morphology and by grazing incidence X-ray diffraction for films structure. Twin films deposited on quartz were investigated using the m-line technique to monitor the variation of optical properties when exposed to butane diluted in N2 down to 100 ppm concentration. The acquired expertise was used to design a miniaturized system which was tried in real conditions.
Keywords: Metal oxide films, nanostructured coatings, optical gas sensor, m-line technique.
______ *
To whom correspondence should be addressed: C. Ristoscu, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
29
30
C. RISTOSCU ET AL.
1. Introductory Remarks On-line monitoring of emission gases from different combustion and/or chemical processes has become an important issue as developing techniques in emission and environmental control. Given adequate catalysis, the gas under detection initiates electrochemical reactions that alter the electrical response of a suitably set circuit. By measuring conductivity changes and in respect with catalysts used, the concentration of the gas can be established. Several electrochemical cells are usually employed to provide the required response and selectivity. The operation requires high electrical power, elevated temperatures, and long-term calibration procedures. Over the years, solid-state versions of such electrochemical systems have been developed using semiconductor metal oxides. Their conductivity is changed as a result of gas exposure, largely through reduction-oxidization reactions. In a similar approach systems using organic polymers instead of metal oxides have also been devised. A well developed technology currently available integrates the devices electronically into complete sensor chips. The main drawback of such systems is their electrical operation requiring high power and temperature, electrical connections, etc. They are nevertheless the dominant technology in today’s market. Alternative approaches, based on molecular mass detection in surface acoustic wave or quartz microbalance schemes, are also used, but they involve greater complexity, and the difficulties they face are also larger. A challenging alternative consists in optical schemes that use refractive changes and minimal absorption alterations to enable remote operation and circumvent the need for power supply. Such schemes rely on interferometry and diffraction to provide a unique edge over other detection techniques not only in terms of the microscopic nature of the interactions, but also as concerns the final target products. Their key advantages include remote sensing on long interrogation lengths along with zero power consumption at the sensing point location and hence no need for electrical circuitry at that point or in its vicinity. Besides, they usually operate at room temperature, with zero electro-magnetic interference, and minimal power consumption and can be used in protected or physically isolated areas. Multi-parameter sensorial response is another plus. We review in this paper the basic operation principle of nanostructured optical gas sensors focusing on the example of Pd-doped SnO2 thin films obtained by pulsed laser deposition.
NANOSTRUCTURED THIN OPTICAL SENSORS
31
2. Optical Gas Sensors 2.1. BASIC EQUATIONS
As is well known,1 in the case of non-magnetic homogeneous materials, the electromagnetic field in harmonic regime can be described in terms of electric, E Z , and magnetic fields, H Z , respectively. As a result, Maxwell’s equations can be written as follows:
rot E Z P0
rot H Z
w H Z wt
w D Z J Z wt
(1)
(2)
where D Z is the displacement vector and J Z stands for current density. Equations (1) and (2) can be also written:
rot E Z
jZP0 H Z
rot H Z jZ D Z V E Z
(3) (4)
By introducing the expression
D Z H 0H r Z E Z
(5)
(with H 0 being the electrical permittivity of vacuum and H r the material’s one), Eq. (4) becomes:
§ V · ¸ E Z rot H Z jZH 0 ¨¨ H r j ZH 0 ¸¹ ©
(6)
or further:
rot H Z jZH 0 n 2 E Z 2
(7)
Here n is the complex refractive index of the material which can be expressed as:
32
C. RISTOSCU ET AL.
n2
H r'
Hr j
V ZH 0
(8)
Due to the fact than n is complex, it can be found as n
n' jn' ' . This
2 n'2 n' '2 j 2 u n'un' ' , where n” corresponds to the absorption implies n of the material.
In these conditions, Eq. (8) can further be written:
n '2 n ' '2 H r ° V ® °2 u n'un' ' ZH 0 ¯
(9)
The solutions of this system, which have a physical meaning (n’ > 1), are:
° °n ' ° °° ® °n ' ' ° ° ° °¯
1
2 ·º 2 ª § § · H V ¨ r « ¸¸ ¸¸» 1 1 ¨¨ ¨ «2¨ © 2H 0H rZ ¹ ¸¹»» ¬« © ¼ V 2H 0Z 1
2 ·º 2 ª § « H r ¨1 1 §¨ V ·¸ ¸» ¨ 2H H Z ¸ ¸¸» « 2 ¨¨ © 0 r ¹ ¹» «¬ © ¼
(10)
Equation (10) shows that a variation in the surface conductivity of a material leads1 to a change in its refractive index (the real n’ and imaginary n” parts vary 2 as V ). 2.2. MAIN EXPERIMENTAL SETUPS
In optical sensing, gas detection is achieved by optical interrogation. One has a large choice of laser illuminated interferometers to probe for refractive index variation caused by trace gases. Our option for the studies reported in this paper went to an m-line setup illuminated with a stabilized He-Ne laser source at 633 nm. This technique is suitable and frequently applied in order to measure variations of optical parameters in transparent waveguides.2–4 A typical experimental setup is presented in Figure 1.
NANOSTRUCTURED THIN OPTICAL SENSORS
33
Figure 1. Testing facility for waveguide coupling interrogation schemes.
At the center of the test setup were the prism coupler on its rotation plate and the sensing element, i.e. the thin film, which had been deposited on a transparent substrate pressed on the prism base. Physical interactions between the gas and the sensing material leading to variations in the refractive index of the latter took place at the air-sensitive material interface. Calculations based on equations in Section 2.1. showed5 that the electromagnetic field was enhanced at this interface for the transversal magnetic (TM) polarization state of light and a thin film thickness close to the cut-off thickness of the guided mode (i.e. the thickness below which the mode cannot propagate in the thin film any longer). This configuration ensured the highest sensitivity to refractive index variation under gas exposure. 2.2.1. Coupling by a prism The coupling by a prism of an incident laser beam into a planar waveguide is governed by the incident angle Ts between the beam and the prism base (Figure 2). At a particular incidence angle (called synchronism angle), the resonant coupling of the laser beam into the waveguide can be observed by watching a dark line (known as mode line or m-line) that appears in the reflected beam.
34
C. RISTOSCU ET AL.
dark line
laser beam
Ts
prism (nP)
waveguide (nW)
reflected spot in the Fresnel field region
substrate (nS) Figure 2. Prism coupler.
Consequently, the application of the method is mainly based on measuring the angles corresponding to the m-lines observed. From the propagation constants of the guided modes determined from these angles, the refractive index n and thickness h of the waveguide can be inferred. In the case of an angle measurement accuracy of 10–3, which is rather easily achievable, we have an accuracy of 10–4 for the refractive index estimation and r2 nm for thickness evaluation. We measured the refractive index and waveguiding properties of thin films for the two polarization states of light. The resonant coupling of the laser beam into the waveguide was strongly affected by perturbations that were due to the interaction of the material and surrounding medium. The effective index N of the guided mode for the resonant coupling is connected to the incident angle Ts of light by Eq. (11):
N
ª º § sin T s · ¸ Ap » n p sin «arcsin¨ ¨ n ¸ «¬ »¼ © p ¹
(11)
where np is the refractive index of the prism, and Ap its characteristic angle. From Eq. (11), we deduce the following expression of 'N:
'N
ª º § sinT s · ¸ Ap » n p cosT s cos«arcsin¨ ¨ n ¸ «¬ »¼ © p ¹ 'T s 2 n 2p sinT s
(12)
2.2.2. Coupling by a diffractive grating Light coupling can also be obtained using a coupling grating etched in the waveguide. The advantage of this system over the prism one consists in the possibility of visualizing in the transmitted beam the actual displacement of the m-line in the presence of gas, as shown in Figure 3.
NANOSTRUCTURED THIN OPTICAL SENSORS
35
Laser Beam O Ts
/ h
d N: effective index
n
transparent substrate (ns)
e
Figure 3. Light coupling by a diffractive grating engraved in a waveguide.
The angle is measured in both cases in a similar way. The corresponding relation between synchronism angle, Ts, beam wavelength, O, grating period, /, and the effective index for zeroth order, N, is:
N
sin T s
O /
From Eq. (13), we deduced the following expression of 'N:
'N
O· § 'T s 1 ¨ N ¸ /¹ ©
(13)
2
(14) Note that the results of the two methods are comparable in terms of sensitivity. In both of them, device operation is based on the m-line coupling principle, but the use of a grating coupler increases robustness and enables miniaturization (see further Section 4). 2.3. SENSING ELEMENTS
2.3.1. Optical sensors: critical review In the case of electrochemical conductivity sensors (MOS), reversible charge transfer reactions lead to a collective behavior consisting of electrical conductivity changes. Similarly, alterations of the optical properties occur by agent
36
C. RISTOSCU ET AL.
adsorption in suitably designed materials, such as metal oxides. These changes can be detected optically with very high accuracy and allow for a wide dynamic range multidimensional parametric space for work. Nanocomposite materials exploration by a two-step production and encapsulation process has also been initiated recently. The functionality of advanced composites would further be enhanced by the inclusion of special nanoparticles. The current trends encourage efforts towards the production of composites based on metals, semiconductors, metal oxides, metal-coated nanoparticles and nanotubes. Non-linear optical mechanisms are expected to offer additional innovative sensing tools that will complement the linear optical response under chemical exposure. Further innovation is foreseen by the use of organic materials incorporated in inorganic matrices to selectively adsorb specific agents. A novel approach consists in applying polymer-based sensitive receptors and including functionalized acidic pendants for nitroaromatic detection. We have to mention, though, that polymers age rapidly and are highly prone to corrosion by various aerosols. Both factors strongly influence stability and reliability in gas detection. At present, the chief concern in optical gas sensing is the reversibility of the physicochemical optical changes that take place in materials, mainly thinfilm structures, upon exposure to the chemical environment. Research efforts are aimed at the selective detection at room temperature with sensitivity in the pp-million range and in some cases (e.g., nitro-aromatics) pp-billion range. The predicted response time of the sensors is on the order of ms (depending on the sensor head), while the anticipated recovery time determined by the adsorption affinity ranges from 10 ms to 10 s. Based on these considerations, we made our choice for highly transparent thin metal oxide films, e.g., ZnO, SnO2, TiO2, WO3, as the simplest, very reliable solution for developing a cheap, easy to operate, potentially miniaturized detection system. 2.3.2. Thin metal oxide sensors: pure, doped and clustered Materials used for waveguides have to meet the basic prerequisite of providing high transmittance in the visible or infrared region. Metal-oxide optical sensors, due to their simple construction, may offer advantages in terms of resistance to severe conditions such as high temperature and corrosive environment, and the capability of being placed in close vicinity of emissions. This type of sensor may also have a short response time so as to enable the possibility of real-time control. However, modern applications of these sensors still face the problems of high cross-sensitivity to gases, i.e. rather low selectivity; high sensitivity to
NANOSTRUCTURED THIN OPTICAL SENSORS
37
ambient humidity; and long-term drift, related to oxygen bulk diffusion and transformations of the crystal structure of polycrystalline materials.6 A considerable improvement of operating parameters such as gas response selectivity, stability, and rate, can be achieved by optimizing both the bulk and surface structure of the metal oxide films applied. Efforts to optimize sensor output characteristics have so far mainly focused on finding optimal technological methods for doping metal oxides with nanoscale metal catalyst additives. Decreasing crystallites (grains) to nm dimensions has been another major goal. Dopants can boost the catalytic activity of the base oxide, stabilize a particular valence state, favor formation of active phases, stabilize the catalyst against reduction, and/or increase the electron exchange rate. Inserting active additives into base metal oxides can modify their parameters such as charge carrier concentration; chemical and physical properties of the metal oxide matrix; electronic and physicochemical properties of the surface; surface potential and inter-crystallite barriers; phase composition; sizes of crystallites, and so on. Noble metals due to their electronic state and their distribution both on the oxide surface and into the film can strongly promote gas sensor sensitivity and selectivity. Noble metal nanoclusters have been tested as either dopants or surface clusters for boosting gas-sensing efficiency by catalyzing specific reactions with the detected gases. Several catalytic additives/dopants such as Pt,7,8 Pd,7–13 Sb,14 Au,7 and Ru15,16 were chosen for enhancing sensor sensitivity. Different methods have been applied for adding metals to the films. Among them, pulsed laser deposition (PLD) provides some advantages, including reduced contamination due to the use of laser light, controlled composition of deposited structures, and in-situ doping. Moreover, PLD is a versatile, powerful tool for custom designing nanoparticles with the desired size and composition by merely varying the experimental deposition conditions.17–20 This was therefore our choice for obtaining thin metal oxide films doping with noble metal nanoclusters. 3. Optical Gas Sensor with PLD SnO2 Thin Films Doped with Pd Nanoclusters 3.1. MATERIALS AND METHODS
The experiments were performed with a PLD installation schematically depicted in Figure 4. Before each deposition event, the vacuum chamber was evacuated to a residual pressure lower than 10–5 Pa. Meanwhile, the chamber walls were heated to facilitate desorption of water vapors and other contaminants. Vacuum quality was monitored with a quadrupole mass spectrometer.
38
C. RISTOSCU ET AL.
The SnO2 (99.99%) target was ablated using an UV XeCl* excimer laser source (Lambda Physik LPX 315i, O = 308 nm, WFWHM 30 ns) operating at a frequency repetition rate of 10 Hz. To deposit one film, 20,000 subsequent laser pulses were applied. Before each deposition event, target surface was cleaned by applying 5,000 laser pulses. During this time, a shutter was interposed between the target and the collector to avoid deposition of ablated material from the first layers of the target, which usually contain contaminants. An AR-coated MgF2 lens with 300 mm focal length was used to focus the laser beam into a 1 mm2 spot on the target surface. The target was rotated at 3 Hz frequency during multipulse laser irradiation to avoid drilling. In-situ doping with noble metal of the SnO2 films was achieved by placing a thin 1 mm diameter Pd wire across the target surface (see inset in Figure 4). The laser beam was incident on the target at an angle of 45º, while incident laser fluence during deposition was set at 10 J/cm2. Films were deposited in a dynamic flux of 10 Pa O2. The Pd concentration in the SnO2 films was controlled by varying laser spot position on the target surface. The experimental conditions were optimized on the basis of previous parametric studies.13
Pd wire (I 1 ) SnO Figure 4. Experimental PLD/RPLD setup; the inset represents the SnO2 target with a thin 1 mm diameter Pd wire.
Ablated material was collected on Si(100) or quartz substrates maintained at room temperature, 350qC, or 500qC, during the deposition. They were placed 60 mm from the target parallel to it. All substrates were cleaned in an ultrasonic bath prior to deposition. Scanning electron microscopy (SEM) measurements were carried out using an electron microscope JEOL 6320F equipped with facilities for conducting energy dispersive X-ray spectroscopy (EDS) analysis. Grazing incidence X-ray
NANOSTRUCTURED THIN OPTICAL SENSORS
39
diffraction (GIXRD) analysis was carried out using a Siemens D5000 Diffractometer equipped with a Cu KĮ source (O = 0.1541 nm). All measurements were performed at a grazing incidence angle of 3q. Diffractograms were recorded from 10q to 100q at a step of 0.02q. The surface roughness of the deposited films was studied by atomic force microscopy (AFM). X-ray photoelectron spectroscopy (XPS) analysis was carried out using spectrometer system (Kratos Axis Ultra) with a monochromatic AlKD (hQ = 1486.6 eV) photon source. The system was also equipped with a 165 mm hemispherical analyzer for the acquisition of spectra and a concentric spherical mirror analyzer for XP imaging. The m-line technique was used at 633 nm laser wavelength for waveguide coupling interrogation (see also Section 2.2). Light was coupled by a prism to the material in the form of optical resonance (guided mode), and synchronism angle variations following gas introduction were measured. A He-Ne laser source emitting at 633 nm wavelength in transversal electric (TE) mode was used. Transversal magnetic (TM) polarization was obtained using a O/2 plate placed in the exit of the laser beam. The prism used for light coupling was made of TiO2, rutile phase. Its characteristics are given in Table 1. A photo of the prism setup used for light coupling can be seen in Figure 5. TABLE 1. Characteristics of the rutile prism used for light coupling. O (nm) nTE nTM
633 2.8641 2.5821
LASER
SAMPLE HOLDER PRESS
ROTATION STAGE
Figure 5. Photograph of the prism coupler.
40
C. RISTOSCU ET AL.
The sample was pressed on the prism by a polyethylene support, the manual translation of which helped control applied pressure. Different translation and rotation systems may improve the positioning of the prism coupler (eight degrees of freedom). The incident angle of the beam on the prism base can be varied in order to attain synchronism conditions. At that stage, a photodiode measured the position of the reflected beam. The system was proven to be capable of detecting hydrocarbons in the range of 100–1,000 ppm.21–25 3.2. SENSING PERFORMANCES AND EXPLOITATION CHARACTERISTICS
3.2.1. Physicochemical characterization Figure 6 shows the morphologies for the three types of Pd:SnO2 samples deposited in 10 Pa O2 at (a) RT, (b) 350qC, and (c) 500qC, respectively. The sample deposited at RT exhibited a rough surface with cracks, while the two others were rather smooth and densely packed. A quasi-ordered nanostructure can be noticed in the sample deposited at 500qC.
a
b
c
Figure 6. SEM microimages of Pd:SnO2 structures deposited in 10 Pa O2 at (a) RT, (b) 350°C, and (c) 500°C.
AFM studies (Figure 7) sustained these observations. The surface of the sample deposited at RT was porous, with large particulates (>100 nm). By contrast, samples deposited at 350qC and 500qC were denser, and their particulate dimensions were down to ~100 and ~80 nm, respectively. An even slight increase in SnO2 grain size can lead to significant changes in the film structure that is responsible for both the catalytic and gas sensing properties of metal oxides.26 Typical GIXRD spectra of the investigated samples are given in Figure 8. The films are polycrystalline without any preferential orientation. Using the diffraction database,27 the visible peaks were clearly assigned to either SnO2 or Si. As can be seen from the spectra, the film deposited at 500qC (Figure 8c) was better crystallized than those obtained at 350qC and RT.
NANOSTRUCTURED THIN OPTICAL SENSORS
a
b
41
c
Figure 7. AFM microimages of Pd:SnO2 structures deposited in 10 Pa O2 at (a) RT, (b) 350°C, and (c) 500°C.
Note that the films only consisted of SnO2 phase. The most important lines were identified, as e.g. (110) at 26.6q, (101) at 33.9q, (200) at 37.9q, and (211) at 51.7q.27 EDS results confirmed the formation of SnO2 phase including about 2% wt of Pd. 3.2.2. Detection of trace gas (hydrocarbon) All sensors were exposed at room temperature to different concentrations of butane diluted in nitrogen (1,000, 500, 200, or 100 ppm). The applied testing protocol was as follows: (1) the detection cell was evacuated to clear it from impurities (1 min 30 s); (2) the cell was filled with N2 to atmospheric pressure (2 min 30 s); (3) the N2 + butane mixture was let in (3 min); (4) the mixture inflow was stopped, and N2 was introduced to study the return to baseline (3 min); and (5) cycles 1–4 were repeated. The typical optical responses we recorded are given in Figure 9, in which variation due to the presence of butane is evident. An important feature was that the optical response was repeatable for all tested sensors. In vacuum the signal becomes stable at ~400 mV, while the mixture of butane-nitrogen made it stabilize at about 550 mV. This variation was due to butane and could not be attributed to pressure effects. Moreover, the signal always reached the same level regardless of the preceding phase of the protocol, which indicates high reproducibility and reversibility. All films showed similar responses when the concentration of butane decreased. The shape of the signal variation kept unchanged, but the kinetics was slower at lower butane concentrations. The response time was about 1 min for 1,000 ppm, 2 min for 500 ppm, and 3–4 min for 100 ppm. The response time is defined as the time required for stabilizing the signal level from the phase of carrier (nitrogen) gas flow to that of butanenitrogen mixture exposure.
42
C. RISTOSCU ET AL.
a
130 120 110 100
SnO2 (220)
30 20
Si (331)
Si (220)
40
Si (400)
50
SnO2 (211)
SnO2 (101)
60
SnO2 (200)
70
Si (311)
80
SnO2 (110)
Lin (Counts)
90
10 0 12
20
30
40
50
60
70
80
90
10
2-Theta - Scale
b)
SnO2 (211) SnO2 (110)
Si (400)
SnO2 (312)
SnO 2
Sn O2 (20 2)
(310 )
SnO2 (220)
SnO2 (321)
SnO2 (301)
SnO2 (200)
100
SnO2 (101)
Lin (Counts)
200
0 10
20
30
40
50
60
70
80
90
10
80
90
2-Theta - Scale
SnO2 (321)
SnO2 (332)
SnO2 (221)
100
SnO2 (220)
SnO2 (101)
200
SnO2 (200)
SnO2 (110)
Lin (Counts)
300
SnO2 (002)
SnO2 (211)
Si (400)
c)
400
0 10
20
30
40
50
60
70
10
2-Theta - Scale
Figure 8. GIXRD spectra of Pd:SnO2 structures deposited in 10 Pa O2 at (a) RT, (b) 350°C, and (c) 500°C, respectively (grazing incidence angle of 3º).
250
N2
N2 + butane
V (mV)
200
N2
N2
a)
50
43
N2 + butane
N2 + butane
150 100
vacuum
vacuum
300
vacuum
NANOSTRUCTURED THIN OPTICAL SENSORS
1000ppm 200 ppm
0 0
5
10
15
20
t (min) 600
500
200
vacuum
N2
N2 + butane
N2
N2 + butane
vacuum
N2 + butane
N2 300
vacuum
V (mV)
400
1000ppm 100
500ppm
b
100ppm
0 0
5
10
15
20
t (m in) 600
500
N2
N2 N2 + butane
N2
N2 + butane
vaccum
300
vaccum
vaccum
V(mV)
400 N2 + butane
200
100
1000ppm 500ppm 100ppm
c)
0 0
5
10
15
20
t (min)
Figure 9. Sensitivity of Pd:SnO2 sensor to different concentrations of butane diluted in N2. Quartz substrate temperature during PLD was (a) RT, (b) 350°C, and (c) 500°C.
From Figure 10 we notice that the higher the substrate temperature applied during PLD, the higher the gas sensitivity of the sensors. The (110) and (101) planes of the SnO2 crystal are known to be F (flat) faces. According to the periodic bond chains (PBC) theory,28 surface atoms of F-faces are strongly bound to each other in directions parallel to the face. As a result, these faces have a slight tendency to react with the arriving atoms. It has also been established that (110) and (101) surfaces of SnO2 grow via a layer-bylayer mechanism.
44
C. RISTOSCU ET AL. 600
tension (v)
500 400
vacuum
N2
N2 +butane
N2
vacuum
N2
N2 +butane
vacuum
N2 +butane Room temperature 500˚C
300 200 100 0 0
5
10
15
20
t(s)
Figure 10. Effects of Pd:SnO2 film deposition temperature on the sensitivity to 1,000 ppm of butane.
The various crystallographic planes have different distances between the Sn atoms, which can be ranked as follows: d(110) ~ d(100) < d(101) < d(001).29 Sn atoms are centers of oxygen chemisorption. Changing the indicated distance can therefore influence the rate of dissociative oxygen chemisorption, which is in many cases the main process of gas sensing phenomena.19–23 The observed change in SnO2 crystal faceting could actually be one of the reasons behind the considerable modification of gas sensing characteristics in SnO2 films. It was shown in the case of SnO2 sensors made by selective ionic layer deposition (SILD)30 that the small size of the crystallites (grains) was an essential but not sufficient condition for achieving maximum gas sensitivity together with a fast response. When comparing the gas sensing characteristics of metal oxide-based sensors, one has to primarily consider the size of the conglomerates and their porosity, and only after that the size of grains and thickness of the film.30,31 However, we have to note that the above results are only due to part of the metal oxide matrix parameters which can influence gas sensing properties. Along with geometrical parameters, the physicochemical characteristics of the gas sensing matrix have to be taken into account. They include the chemical and phase composition; the concentration of bulk and surface oxygen vacancies; the size and density of both metal catalyst particles and single atoms on the surface of metal oxides; and the type and concentration of uncontrolled impurities. 3.2.3. Gas contamination and multiple operations The potential gas contamination of the sensors as an effect of butane exposure was studied by AFM and XPS. We compared twin samples one of which had been repeatedly exposed to gas action, while another had been kept in controlled, non-polluted atmospheres.
NANOSTRUCTURED THIN OPTICAL SENSORS
45
AFM analyses (Figure 11) demonstrated that topographies in the cases of unexposed and exposed samples were similar to each other. Nanoparticulates size was in the (45–55 nm) range. The XPS studies indicate that the near surface region of the unexposed sample comprises a mixture of O, Sn, and Pd. The first peak of the Sn 3d doublet occurs at 485.8 eV BE and is assigned to emission from Sn 3d5/2 corelevel of Sn2+ ion in tin oxide. The second component occurs at 494.2 eV and is assigned to emission from the Sn 3d3/2 core level of Sn2+ ions. The 8.4 eV separation of 3/2 and 5/2 components is consistent with this interpretation of SnO/SnO2. The primary component of O 1s occurs at 529.7 eV and is assigned to emission from O2– ions in SnO/SnO2 and PdO. The second component occurs at 530.9 eV and is assigned to emission from chemisorbed OH groups, based on the 2.2 eV separation of the secondary component from the primary one. The third component occurs at 532.9 eV and is assigned to the presence of chemisorbed H2O molecules. The first peak of Pd 3d doublet occurs at 335.7 eV and is assigned to emission from the Pd 3d5/2 core-level of Pd2+ ion in PdO. The second component at 336.8 eV is assigned to emission from the Pd 3d5/2 core-level of Pd4+ ions in PdO2. The XPS spectrum of the “as received” surface of the exposed sample clearly indicates that the near-surface region comprises a mixture of Sn, O, and C. The first component of O1s occurs at 530.6 eV BE and is assigned to emission from O2– ions. The second component occurs at 532.1 eV and, based on the 1.5 eV BE shift, is assigned to emission from adsorbed OH– groups. The Sn 3d5/2 peak occurs at 486.7 eV BE, whilst the Sn 3d3/2 is centered on 495.2 eV BE. The 8.5 eV separation of the two components is consistent with the presence of SnO/SnO2 in the near surface region of the sample. High-resolution C1s spectra were recorded for the unexposed (Figure 12) and exposed (Figure 13) samples, while the main quantitative information was collected in Tables 2 and 3.
a
b
Figure 11. AFM images of Pd:SnO2 samples (a) unexposed and (b) repeatedly exposed to 1,000 ppm butane diluted in nitrogen.
46
C. RISTOSCU ET AL. C1s ZnPd25Q
x102
C1s C1s
Intensity(CPS)
80
60
40
20 290
292
288
286
284
282
Binding Energy (eV)
Figure 12. High-resolution C1s XP spectrum obtained from the surface of unexposed sample. C 1s SnPd26 x 10
3
30
25
CPS
20
15
10
5 290
288
286 Binding Energy (eV)
284
282
Figure 13. High-resolution C1s XP spectrum obtained from the “as received” surface of exposed sample.
By comparing data in Tables 2 and 3, we see that the CH radical did not increase in the exposed sample; on the contrary, there was a reduction of this chemical species in the sample exposed to the butane. It should be stressed one would expect the CH radical to be the butane related species.
NANOSTRUCTURED THIN OPTICAL SENSORS
47
TABLE 2. Summary of quantitative information obtained by analysis of C1s peak as shown in Figure 12. Peak
Position BE (eV)
FWHM (eV)
Raw area (CPS)
RSF
Atomic mass
Atomic conc (%)
Mass conc (%)
C 1s
285.000
1.056
10,004.99
0.278
12.011
60.06
60.06
285.492
1.051
4,625.14
0.278
12.011
27.76
27.76
286.540 287.005
0.890 0.890
639.30 369.07
0.278 0.278
12.011 12.011
3.838 2.215
3.838 2.215
287.946 289.239
1.310 1.191
466.81 553.62
0.278 0.278
12.011 12.011
2.802 3.323
2.802 3.323
(C-C/C-H)
C 1s (C-CO2)
C 1s(C-O) C 1s (C=O)-O
C 1s(C-S) C 1s (shake up)
TABLE 3. Summary of quantitative information obtained by analysis of C1s peak shown in Figure 13. Peak
Position BE (eV)
FWHM (eV)
Raw area (CPS)
RSF
Atomic mass
Atomic conc (%)
Mass conc (%)
C 1s
285.000
0.830
4,350.3
0.278
12.011
45.12
45.12
285.517
0.800
3,052.8
0.278
12.011
31.66
31.66
C 1s(C-O)
285.906
0.728
1,025.1
0.278
12.011
10.63
10.63
C 1s
286.440
0.834
755.6
0.278
12.011
7.84
7.84
289.359
1.175
458.7
0.278
12.011
4.76
4.76
(C-C/C-H)
C 1s (C-CO2)
(C=O)-O
C 1s (shake up)
The latter observation and the results of morphological investigations suggest that the contamination effects due to exposure of the samples to polluted gases are very weak and hence hard to detect, if any. 4. Miniaturized Portable Prototype for Trace Gas Detection The dimensions of the detection system described in Section 2 can be reduced by using an etched coupling grating on the waveguide instead of the prism. We looked for a grating coupler for which the TM0 coupling should take place
48
C. RISTOSCU ET AL.
at normal incidence, i.e. Ts = 0 (see Figure 3). The new setup includes the lithographically submicron etched active material that is required to achieve waveguide coupling. The other basic components are: (i) enclosure for gas insertion, (ii) laser diode source and optics, (iii) high resolution CCD to detect the m-line, and (iv) data acquisition and image processing system for detecting variations of the dark line and reference them to the gas analysis. The new detection system redesigned and rescaled is presented in Figure 14. As a remarkable result, we emphasize that its dimensions do not exceed 15 cm length and 6 cm diameter. The system was recently tried in real conditions. Gas traces below 800 ppm, which is the maximum daily exposure to butane authorized under U.S. Federal Regulations, could be reliably detected. The use of the grating as sensing element also made it possible to cut down sharply the detection time. The system is still being worked on to improve its selectivity and sensitivity. Highly innovative schemes based on direct writing of waveguides in different materials have been recently developed, viz: in LiNbO3 in order to enable active electro-optic control of Mach-Zehender Interferometer, and in inorganic-organic hybrid materials deposited by sol-gel method. Multimode interference couplers written with a UV laser allow for the detection of very low refractive index variations.32 5. Conclusions We reviewed the basic physical principles and applied optical schemes to develop a new generation of high-performance optical gas detectors. General physical equations were introduced and the main solutions for optical sensing elements were critically analyzed. We based our decision to select nanostructured metal oxide thin films as sensing elements on their high sensitivity to gas traces and relative robustness and stability for multiple operations. As an application, we analyzed the case of Pd-doped-SnO2 thin films grown on Si (100) and quartz substrates by pulsed laser deposition in 10 Pa O2 at different substrate temperatures (RT, 350, and 500qC). Films deposited on Si were morphologically and structurally investigated. The results showed that 10 Pa O2 background pressure enabled the formation of nanoparticulates by condensation in the gas phase. Twin films deposited on quartz were investigated by the m-line technique to study the variation of optical properties when exposed to test gases (butane). Our results demonstrated that the gas sensing properties were mainly function of the structure and surface morphology of the samples, which in turn depended on substrate temperature. Moreover, it was evident that the higher the substrate temperature during PLD, the higher the gas sensitivity of the obtained SnO2 sensors. Our investigations did not evidence within measurement limits any remanent contamination of the sensing elements.
NANOSTRUCTURED THIN OPTICAL SENSORS
49
Figure 14. Photo of the miniaturized optical gas sensor prototype.
A miniaturized prototype with a sub-micrometric grating etched in the waveguide film as coupling element was successfully tried for operation in ambient atmosphere. The system is now under development to improve its selectivity and sensitivity to a wide range of detectable gases. ACKNOWLEDGMENTS
The financial support of the EU under the contract NANOPHOS IST-200139112 is acknowledged with thanks. CR also acknowledges the CNR – NATO fellowship Pos. 216.2169, Prot. N. 0015503 and NATO CBP.RIG.982424 contract. The Romanian and French authors acknowledge with thanks the support they received under the 2006–2007 Collaboration Agreement no. 19611 between CNRS and the Romanian Academy.
References 1. 2. 3. 4. 5. 6. 7. 8. 9.
Born, M., and Wolf, E. (1975) Principles of Optics, Pergamon, Oxford, Chapter II. Agan, S., Ay, F., Kocabas, A., and Aydinli, A. (2003) Applied Physics A 80(2), 341. Shi, J., Cao, A., Zhu, J., and Shen, Q. (2004) Applied Physics Letters 84(17), 3253. Luo, Y., Hall, D., Kou, L., Blum, O., and Hou, H. (1999) Applied Physics Letters 75(20), 3078. Mazingue, T. (2005) Ph.D. thesis, Universite Paul Cezane Aix Marseille II, Marseille, France. Korotcenkov, G. (2005) Sensors and Actuators B 107, 209. Wurzinger, O., and Reinhardt, G. (2004) Sensors and Actuators B 103, 104. Gaidi, M., Chenevier, B., and Labeau, M. (2000) Sensors and Actuators B 62, 43. Chatterjee, K., Chatterjee, A., Banerjee, A., Raut, M., Pal, N., Sen, A., and Maiti, H. (2003) Materials Chemistry and Physics 81, 33.
50
C. RISTOSCU ET AL.
10. Korotcenkov, G., Brinzari, V., Boris, Y., Ivanov, M., Schwank, J., and Morante, J. (2003) Thin Solid Films 436, 119. 11. Suda, Y., Kawasaki, H., Namba, J., Iwatsuji, K., Doi, K., and Wada, K. (2003) Surface and Coatings Technology 174–175, 1293. 12. Shatokhin, A., Putilin, F., Safonova, O., Rumyantseva, M., and Gas’kov, A. (2002) Inorganic Materials 38(4), 374. Translated from (2002) Neorganicheskie Materialy 38(4), 462–467. 13. Pererira, A., Cultrera, L., Dima, A., Susu, M., Perrone, A., Du, H., Volkov, A., Cutting, R., and Datta, P. (2006) Thin Solid Films, 497(1–2), 142. 14. Leite, E., Bernardi, M., Longo, E., Varela, J., and Paskocimas, D. (2004) Thin Solid Films 449, 67. 15. Niranjan, R., and Mulla, I. (2003) Materials and Engineering B 103, 103. 16. Rumyantseva, M., Safonova, O., Boulova, M., Ryabova, L., and Gas´kov, A. (2003) Russian Chemical Bulletin International Edition 52(6), 1217 17. Lowndes, D., Geohegan, D., Puretzky, A., Norton, D., and Rouleau, C. (1996) Science 273, 898 18. Chrisey, D., and Hubler G. (Eds.) (1994) Pulsed Laser Deposition of Thin Films, Wiley, New York. 19. Marine, W., Patrone, L., Luk’yanchuk, B., and Sentis, M. (2000) Applied Surface Science 154–155, 345. 20. György, E., Santiso, J., Figueras, A., Giannoudakos, A., Kompitsas, M., and Mihailescu, I. (2005) Journal of Applied Physics 98, 1. 21. Mazingue, T., Escoubas, L., Spalluto, L., Flory, F., Socol, G., Ristoscu, C., Axente, E., Grigorescu, S., Mihailescu, I., and Vainos, N. (2005) Journal of Applied Physics 98(7), 074312 22. Socol, G., Axente, E., Ristoscu, C., Sima, F., Popescu, A., Stefan, N., Mihailescu, I., Escoubas, L., Ferreira, J., Szekeres, A., and Bakalova, S. (2007) Journal of Applied Physics 102, 083103-1-6. 23. Mazingue, T., Escoubas, L., Flory, F., Jacquouton, P., Perrone, A., Kaminska, E., Piotrowska, A., Mihailescu, I., and Atanasov, P. (2006) Applied Optics 45, 1425. 24. Gyorgy, E., Socol, G., Axente, E., Mihailescu, I., Ducu, C., and Ciuca, S. (2005) Applied Surface Science 247, 429–433. 25. György, E., Socol, G., Mihailescu, I., Ducu, C., and Ciuca, S. (2005) Journal of Applied Physics 97, 093527-1_4 26. Golovanov, V., Korotcenkov, G., Brinzari, V., Cornet, A., Morante, J., Arbiol, J., and Rossyniol, E. (2002) CO–water interaction with SnO2 gas sensors: role of orientation effects, in: Proceeding of the 16th International Conference on Transducers, EUROSENSORS-XVI, Prague, Czech Republic, 926–929 (CD). 27. ASTM 21-1250. 28. Hartman P. (Ed.) (1993) Crystal Growth – An Introduction, North-Holland, Amsterdam, The Netherlands. 29. Brinzari, V., Korotcenkov, G., Golovanov, V., Schwank, J., Lantto, V., and Saukko, S. (2002) Thin Solid Films 408(1/2), 51. 30. Korotcenkov, G., Macsanov, V., Tolstoy, V., Brinzari, V., Schwank, J., and Faglia, G. (2003) Sensors and Actuators B 96(3), 602. 31. Korotcenkov, G., Brinzari, V., Cerneavschi, A., Ivanov, M., Golovanov, V., Cornet, A., Morante, J., Cabot, A., and Arbiol, J. (2004) Thin Solid Films 460(1/2), 315. 32. Mazingue, T., Kribich, R., Etienne, P., and Moreau, Y. (2007) Optics Communications 278, 312–316.
X-RAY PHOTOELECTRON SPECTROSCOPY AND TRIBOLOGY STUDIES OF ANNEALED FULLERENE-LIKE WS2 NANOPARTICLES F. KOPNOV1, R. TENNE1*, B. SPÄTH2, W. JÄGERMANN2, H. COHEN3, Y. FELDMAN3, A. ZAK4, A. MOSHKOVICH5, AND L. RAPOPORT5 1 Department of Materials and Interfaces, Weizmann Institute, Rehovot 76100, ISRAEL 2 Fachgebiet Oberflaechenforschung, Fachbereich Materialwissenschaften, Technische Universität Darmstadt, Petersenstrasse 23, 64287 Darmstadt, GERMANY 3 Department of Chemical Research Support, Weizmann Institute, Rehovot 76100, ISRAEL 4 “NanoMaterials” Ltd., Weizmann Science Park, Bldg. 18, 18 Einstein St., P.O. Box 4088, Nes Ziona 74140, ISRAEL 5 Department of Science, Holon Institute of Technology, Golomb St. 52, P.O. B 305, Holon 58102, ISRAEL
Abstract – The temporal chemical changes occurring at the surface of fullerene-like (IF) nanoparticles of WS2 were investigated using X-ray photoelectron spectroscopy (XPS) and compared to those of bulk powder (2H) of the same material. It is possible to follow the long term (surface oxidation and carbonization) occurring at defects on the outermost surface (0001) layer of the fullerene-like nanoparticles. Similar but perhaps more distinctive changes are observed on the prismatic (hk0) surfaces of the 2H powder. Vacuum annealing is shown to remove most of these changes and bring the surface close to its stoichiometric composition. In accordance with previous measurements, further evidence is obtained for the existence of water molecules which are entrapped in the hollow core and interstitial defects of the fullerene-like nanoparticles during the synthesis. They are also shown to be removed by the vacuum annealing process. Chemically resolved electrical measurements (CREM) in the XPS show that the vacuum annealed IF samples become more intrinsic. Finally, ___________ *To whom correspondence should be addressed: R. Tenne, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
51
52
F. KOPNOV ET AL.
tribological measurements show that the vacuum annealed IF samples perform better as an additive to oil than the non-annealed IF samples and the bulk (2H) platelets powder.
Keywords: IF-WS2, nanoparticles, XPS (X-ray photoelectron spectroscopy), tribology, annealing.
1. Introduction The extensive research of the structural, physical and chemical properties of layered transitional metal dichalcogenides, MX2 (M = metal, X = chalcogenide) have been going on for the past 5 decades.1 This is due to their unique structural properties, where MX2 layers are bonded by the weak van der Waals forces through relatively large (several Angstrom) van der Walls gap. The van der Waals gap allows accommodation of foreign atoms and molecules leading to the modifications of the physical properties, such as the electrical conductivity and the magnetic susceptibility. Two of these MX2 materials that have been broadly studied are 2H-MoS2 and 2H-WS2 (2H stands for hexagonal unit cell that consists of two layers); they have been shown to be good solid lubricants2 and catalysts.3 Since it is believed that 2H-WS2 can be a promising material for solar cells, much effort has been paid to the synthesis of WS2 thin films and studying their electronic and transport properties by electrical measurements and X-ray photoelectron spectroscopy (XPS).4 It was found that under special conditions layered transition metal dichalcogenides could be produced as closedcage (fullerene-like) nested nanoparticles (IF) and nanotubes (INT).5 Since the first inorganic fullerenes-like structures and nanotubes of MX2 compounds (M = Mo,W; X = Se,S) were synthesized much work has been done in developing new synthetic routs and elucidating their growth mechanisms.6 Moreover, the structural,7 optical,8 electronic,9 and mechanical10 properties of these materials have been extensively studied. In particular, the IF-WS2 phase revealed intriguing tribological11 performance that offers many potential applications where reduced friction and wear are desired. A 1H NMR study12 of the IF-WS2 nanoparticles showed that the nanoparticles entrap water molecules, or OH moieties, which are produced during the conversion of the oxide nanoparticles into the metal sulfide (IF) and perhaps also some residual hydrogen molecules. The water molecules are believed to be present in defects arising due to folding of the IF-WS2 layers. They may also adsorb in the nanopores between the agglomerated IF-WS2 onions; at the nanoparticles surfaces and occupy voids in the central part (core) of the IF nanoparticle as well. A dedicated
ANNEALED FULLERENE-LIKE WS2 NANOPARTICLES
53
vacuum annealing set-up was assembled in order to extract the remnant water molecules produced during the synthesis or adsorbed onto the nanoparticles surface afterwards. Sequential van der Pauw transport measurements13 of the IF-WS2 pellets showed that the vacuum annealed samples demonstrated higher resistivity than the non-annealed ones. However, apart from a preliminary study14 no systematic X-ray photoelectron spectroscopy (XPS) study of the IF-MX2 structures was carried out, so far. Such a study could reveal the electronic properties of the IF material, i.e. the Fermi level position, the position of the tungsten and sulfur bands, which could also help to elucidate the role of the intercalated water molecules within the IF nanostructures. The following work presents XPS studies conducted on the 2H and IF-WS2 before and after vacuum annealing. Two different types of IF powders were studied, i.e. freshly synthesized powder and another powder produced 3 years before the actual measurements. Our measurements showed that the IF material was indeed a p-type semiconductor, but to a lesser degree than bulk 2H-WS2. The main difference between the old and the fresh powders was that the later exhibited a better stoichiometric ratio. These characteristics support our previous results of transport measurements, i.e. the resistivity of annealed IF samples is higher than that of the pristine ones. Complementarily, the XPS-derived binding energies of W and S in pristine (non-annealed) and annealed IF-WS2 demonstrated systematic shifts, as compared to 2H platelets. Moreover, the most salient chemical change under annealing the IF-WS2 was the loss of oxygen, which was indeed indicative of water degassing. 2. Experimental 2.1. SAMPLE PREPARATION AND HANDLING
The synthesis of the IF-WS2 samples was described previously.14 The 2H-WS2 powder came from a commercial source (Alfa Aesar). The XPS measurements were carried out either on pellets or loose powder. 2.2. XPS
The powder from the 3 years old batch was analyzed by X-ray photoelectron spectroscopy (XPS) measurements using an Escalab 250 (by Thermo-VG) setup at a resolution of 0.3 eV. The protocol used for these measurements was as follows: first, survey spectrum was taken with the following parameters: step size 0.5 eV, pass energy 50 eV, dwell time 50 ms. Subsequently, detailed spectra were taken: step size 0.05 eV; pass energy 10 eV, dwell time 50 ms. The X-ray was running with a power of 150 W at 15 kV with a 500 Pm spot size. All the
F. KOPNOV ET AL.
54
data presented as figures and as series I (pellets) in Table 2 were done with this setup. All spectra were referred to the Fermi level of a freshly sputtered metallic sample (EF = EB = 0eV). Complementary XPS measurements on the freshly prepared samples were performed (series II in Table 1) with Kratos AXIS-HS analytical system using a monochromatized Al KD (1486.6 eV) source at relatively low power (75 W), with a hybrid magnetic and electro-static lens mode and detection pass energy of 20 eV. Control over sample charging was achieved by an electron flood gun (eFG) with typical acceleration of 2–3 eV. The annealed IF samples were analyzed using this setup 2–5 min after exposure to the ambient (see series II (loose powder) in Table 1). TABLE 1. Summary of the results (series II) of the XPS analysis of IF-WS2 (annealed and nonannealed) powders. The numbers (in bold) represent atomic percentage of the constituent elements. The numbers in parenthesis stand for the energy shifts in mV of the corresponding bands: E (annealed)-E (non-annealed) (the certainty of the measurements is §30 mV according to the shifts under electron flux from the eFG).
IF-WS2 pristine IF-WS2 annealed
W(4f)
S(2p)
28.5
57.9
31.1 62.7 (289 meV) (285 meV)
O(1s) C(1s)
8.1
5.5
1.7
4.5
VB (valence band)
(294 meV)
TABLE 2. Summary of the results (series I) of the XPS analysis of 2H-WS2 and IF-WS2 (annealed and non-annealed). The numbers (in bold) represent atomic percentage of the constituent elements. The numbers in the parenthesis stand for the energy shifts in meV of the corresponding bands: E (annealed)-E (non-annealed).
W(4f)
S(2p)
O(1s)
N(1s)
C(1s)
SOx
2H-WS2
17
38
36
4
5
5.7
IF-WS2 pristine IF-WS2 annealed
18
40
31
5
6
4.7
30 (230 ± 50 meV)
61 (240 ± 50 meV)
4
3
2
0
VB (valence band)
(230 ± 50 meV)
ANNEALED FULLERENE-LIKE WS2 NANOPARTICLES
55
2.3. VACUUM ANNEALING
A special setup13 was built for degassing the entrapped gas molecules. The system is evacuated by a turbomolecular pump up to 10–9 Torr (Leybold Turbovac 361). The sample (250–400 mg) is placed in the chamber and heated by a heating element which is controlled by a thermo controller (Eurotherm 2216e). The sample was heated up to 450°C. The usual heating treatment continued for 2 days. The outgoing gases are analyzed by a residual gas analyzer (RGA-Inficon model Transpector 2). 2.4. TRIBOLOGICAL MEASUREMENTS
Friction and wear experiments were performed using a ball-on disk set-up with sliding velocity of 0.4 ms–1 and load of 300 N. A bearing ball with a diameter of 10 mm moved against a steel disk quenched and tempered up to 44–46 HRc. Therefore the harder disk remained intact while the softer ball suffered the wear during the tribological measurements. The surfaces of the ball and the disk were polished (Ra = 0.02 Pm) prior to the experiment. 1 wt % of 2H-WS2 or IF-WS2 powder (pristine and vacuum annealed) were added to a paraffin oil with viscosity of 60 cSt at 20°C. The solid powder was mixed carefully with the oil for 1 h before the test. Few drops of the lubricant were fed to the contact area every minute during the entire duration of the experiment. The friction coefficient and the size of a wear spot on the surface of the ball were measured at a given periods of time after onset of the test. The studied surfaces were analyzed before and after the tribological tests by optical microscopy, electron scanning microscopy (SEM) and Raman spectroscopy. The Raman spectra were measured by a Renishaw micro-Raman microscope 2000 excited with a He/Ne laser (6,328 ǖ). 3. Results and Discussion The most striking observation of the XPS measurements is the shift of the S(2p) and W(4f) bands of the IF samples after annealing. Tables 1 and 2 summarize the results of the XPS analysis. The annealed samples in series II (measured with Kratos AXIS-HS) demonstrated the same band shifts (Table 1) as compared to series I (Table 2) (measured with Escalab 250). The line shifts of the IF powder after annealing are believed to reflect an upwards shift in the Fermi level – EF, i.e. the annealed material becomes less p-doped than the non-annealed one. These results are in line with the previous electrical measurements, where the resistivity of the pristine IF samples was found to be
F. KOPNOV ET AL.
56
higher than that of the 2H-WS2 ones, and furthermore the annealed IF samples had higher resistivity than the non-annealed ones. Another interesting observation is that the annealing process of the 3 years old IF powder led to an improved stoichiometric ratio between S and W (Table 2). The vacuum annealing caused also the disappearance of the SOx moieties. However, the magnitude of the sulfur peaks did not alter suggesting that oxygen atoms had evaporated during the annealing. Additionally, the carbon contamination was drastically reduced during annealing, and the C(1s) peak (before annealing) shifted from the binding energy that fits bonded carbon (probably as WC bonding (283.5 eV)) to the value typical for CH species and graphite (284.4 eV) (after annealing) in both sets of powders (I and II). Furthermore, smaller amounts of the contaminant atoms, like C, O, and N, as well as SOx moiety (less than 1% of the sulfur content ) were found in series II (fresh batch) measurements than in series I (3 years old batch). The larger amounts of SOx moieties and other contaminants found in series I may suggest that the surfaces of the pristine IF powder is not chemically homogeneous, or that some oxidation of the nanoparticle surfaces in air must have happened within 3 years. It should be born in mind, however, that transmission electron microscopy (TEM) and x-ray diffraction analyses (XRD) of the 3-years old sample (series I) did not reveal any noticeable differences compared to the fresh sample (series II). Moreover, the annealing process did not have any influence on the outcome of the TEM and XRD analysis. These observations suggest that the IF nanoparticles do not go through structural changes during the annealing process, which affect their outermost surfaces, only. The tribological measurements revealed that the annealed IF powder demonstrated lower friction coefficient than the non-annealed one (Table 3). The results of the XPS analysis can be interpreted along the following scheme: during the synthesis some oxygen from the oxide core or from the entrapped water molecules can react with sulfur atoms of WS2, consequently carbon atoms that enter the reaction as a contaminant bind to the tungsten atoms. These new chemical moieties provide additional acceptor states. Indeed, Table 2 shows that the quantity of SOx and C at the nanoparticles surface TABLE 3. The results of the tribological tests; each figure represents an average of three trials. Material
Friction coefficient
Wear spot (ȝm)
2H-WS2
0.09–0.11
512–544
IF-WS2 (pristine)
0.09–0.1
480–490
IF-WS2 (vacuum annealed)
0.07–0.08
310–312
ANNEALED FULLERENE-LIKE WS2 NANOPARTICLES
57
decreased nearly by the same amount after vacuum annealing. Vacuum annealing leads to removal of the WC and SOx species. While some of the reactions are restricted to the nanoparticle surface, others may occur throughout the entire volume of nanoparticles (120 nm in diameter), in the central void of a particle or at the defects arising due to folding of the IF-WS2 layers (Figure 1). The products of such a reaction, i.e. CO2, H2O, SO2 are pumped out during the vacuum annealing. Thus the acceptor states disappear and the material becomes more intrinsic and less conductive. Disappearance of the WC peak indicates that chemically a more complex process occurred during the vacuum annealing in the presented case. One of the possible mechanisms is that WC moiety reacts with the extracted water, and that leads to the formation of a volatile tungsten oxihydrate. Such volatile moieties have been documented in the literature.15 In addition, the XPS measurements showed that the surface of the IF nanoparticles underwent oxidation upon exposure to the ambient for a prolong period of time. Since, there are not any structural differences, according to the TEM and XRD analyses, between the (3 years) old and fresh batch of the IF powder, the oxidation can take place at the defects existing within a nanoparticle or at the outermost layer of the nanoparticle (Figure 1 (inset 3)). Noticeably, when the molecular layers fold sharply, a kink is often formed (see Figure 1) leaving a void, which can be filled with water molecules. These separated layers expose prismatic (hk0) faces, which are more prone to a chemical degradation by an oxidation process. It is a well known fact that the
1
2
3
1
3
2
(a)
(b)
Figure 1. (a) TEM image of an individual IF-WS2 nanoparticle shows clearly the hollow core in the centre of the closed nested WS2 layers. (b) Blown up zones 1, 2 and 3 of the nanoparticle exhibiting defects and voids wherever the folding of the layers induces imperfections in the lattice.
58
F. KOPNOV ET AL.
sulfur terminated (basal) (0001) surface of tungsten disulfide layer in the 2H-WS2 crystal is inert towards oxidation. In contrast to that, the prismatic (hk0) faces of the lattice are chemically reactive due to abundance of dangling bonds. The saturation of these dangling bonds with contaminating atoms leads to the appearance of impurity states that impair the performance of these crystals as solar cells and also adversely affects their tribological performance. 6 4. Conclusions XPS investigation of the pristine and vacuum annealed IF-WS2 nanoparticles was undertaken. Furthermore in the course of the measurements 3-years old and freshly prepared powders were studied. It was found that the freshly synthesized pristine powder had less contamination in the form of SOx, C and O moieties and better stoichiometric ratio than the powder of the 3 years old batch. The annealing improved the stoichiometric ratio in the case of the old batch powder and resulted in a reduced level of contamination for both types of powders. In addition, the annealing process led to a shift of the W(4f), S(2p) atomic bands and the valence band towards higher energies. This shift reflects an upward shift of the Fermi energy such that the annealed material becomes more intrinsic than the non-annealed one. The annealing also resulted in a higher energy shift of the carbon (1s) line. The changes in the XPS spectra of the IF powder and the energy shifts of the corresponding moieties indicate that the annealing process resulted in a more stoichiometric surfaces of the nanoparticles. The chemical reactions that occurred resulted in the disappearance of the acceptor states and thus made the material more intrinsic and less conductive.
References 1. Wilson, J., and Yoffe, A. (1969) Transition metal dichalcogenides discussion and interpretation of observed optical, electrical and structural properties, Adv. Phys. 18(73), 193. 2. Watanabe, S., Noshiro, J., and Miyake, S. (2004) Tribological characteristics of WS2/MoS2 solid lubricating multilayer films, Surf. Coat. Technol. 183(2–3), 347–351. 3. Alonso, G., Del Valle, M., Cruz, J., Licea-Claverie, A., Petranovskii, V., and Fuentes, S. (1998) Preparation of MoS2 and WS2 catalysts by in situ decomposition of ammonium thiosalts, Catal. Lett. 52(1–2), 55–61. 4. Gourmelon, E., Lignier, O., Hadouda, H., Couturier, G., Bernede, J., Tedd, J., Pouzet, J., and Salardenne, J. (1997) MS2 (M=W, Mo) Photosensitive thin films for solar cells, Sol. Energy Mater. Sol. Cells 46(2), 115–121.
ANNEALED FULLERENE-LIKE WS2 NANOPARTICLES
59
5. Remskar, M. (2004) Inorganic nanotubes, Adv. Mater. 16(17), 1497–1504. 6. Therese, H., Zink, N., Kolb, U., and Tremel, W. (2006) Synthesis of MoO3 nanostructures and their facile conversion to MoS2 fullerenes and nanotubes, Solid State Sci. 8(10), 1133–1137. 7. Hassanien, A., Tokumoto, M., Mrzel, A., Mihailovic, D., and Kataura, H. (2005) Structural and mechanical properties of MoS2-I-x nanotubes and Mo6SxIy nanowires, Physica E 29 (3–4), 684–688. 8. Loh, k., Zhang, h., Chen, W., and Ji, W. (2006) Templated deposition of MoS2 nanotubules using single source precursor and studies of their optical limiting properties, J. Phys. Chem. B 110(3), 1235–1239. 9. Milosevic, I., Vukovic, T., Damnjanovic, M., and Nikolic, B. (2000) Symmetry based properties of the transition metal dichalcogenide nanotubes, Eur. Phys. J. B 17(4), 707–712. 10. Schwarz, U., Komura, S., and Safran, S. (2000) Deformation and tribology of multi-walled hollow nanoparticles, Europhys. Lett. 50(6), 762–768. 11. Hu, J., Bultman, J., and Zabinski, J. (2004) Inorganic fullerene-like nanoparticles produced by arc discharge in water with potential lubricating ability, Tribol. Lett. 17(3), 543–546. 12. Panich, A., Kopnov, F., and Tenne, R. (2006) Nuclear magnetic resonance study of fullerenelike WS2, J. Nanosci. Nanotechnol. 6(6), 1678–1683. 13. Kopnov, F., Yoffe, A., Leitus, G., and Tenne, R. (2006) Transport properties of fullerene-like WS2 nanoparticles, Phys. Stat. Sol. (b) 243(6), 1229–1240. 14. Feldman, Y., Frey, G., Homyonfer, M., Lyakhovitskaya, V., Margulis, L., Cohen, H., Hodes, G., Hutchison, J., and Tenne, R. (1996) Bulk synthesis of inorganic fullerene-like MS2 (M=Mo, W) from the respective trioxides and the reaction mechanism, J. Am. Chem. Soc. 118(23), 5362–5367. 15. Sarin, V. (1975) Morphological changes occurring during reduction of WO3, J. Mater. Sci. 10(4), 593–598.
THE DEVELOPMENT AND APPLICATION OF UV EXCIMER LAMPS IN NANOFABRICATION I.I. LIAW1 AND I.W. BOYD2* Department of Electronic & Electrical Engineering and London Centre for Nanotechnology, 17–19 Gordon Street, London WC1H 0AH, UK 2 Department of Electronic & Electrical Engineering and London Centre for Nanotechnology, 17–19 Gordon Street, London WC1H 0AH, UK 1
Abstract – Photon-induced processes hold many unique advantages for thin film processing and surface modification. These include low thermal budgets, lack of ionisation, chemical selectivity, and high cleanliness levels resulting in new chemical pathways to facilitate processing at reduced geometries. For such applications, a variety of sources are available including a wide range of lasers and incoherent lamp systems. In this presentation, the principles and properties of ultraviolet (UV) and vacuum ultraviolet (VUV) radiation generated by decaying excimer complexes will be described. Excimer lamps based on this principle provide highly efficient, high intensity, narrow-band radiation at various distinct wavelengths, with no self adsorption, which can, through insightful variations of the geometric configurations of these dielectric-barrier discharges, make large-scale applications possible. For deep UV applications, sources emitting at wavelengths as low as 126 nm have been developed. The high photon energy levels generated from these sources have been demonstrated to directly photodissociate nitrogen molecules, enabling direct nitridation of surfaces. By surveying a selection of publications in the field, a range of applications of applications for these novel sources in the field of nano-fabrication is discussed. These include the photo-deposition of low- and high-dielectric constant layers, low-temperature oxidation of Si, SiGe and Ge, photo-etching and micro-structuring of polymer surfaces, photo-induced metallization, cleaning of surfaces and
______ *
To whom correspondence should be addressed: Ian Boyd, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
61
62
I.I. LIAW AND I.W. BOYD
UV-curing. From these examples, these relatively low cost lamp systems are demonstrated to be capable of providing low temperature alternatives for largescale materials processing in a wider range of nano-scale applications.
Keywords: Excimer lamps, dielectric barrier discharges, photo-chemistry, ultraviolet, low temperature.
1. Introduction Laboratory UV lasers have widely demonstrated the use of monochromatic photons in the ultraviolet spectral range to selectively initiate photochemical reactions, to dissociate gases or liquids, and on substrate surfaces. For industrial deployment, the need for wavelength selective large throughput systems at low cost is essential. Relatively cheap lamps that possess an extended operating lifetime and are easily fabricated in a variety of geometrical shapes are an intuitive choice. The high energy photons (5–12 eV) generated from these dielectric barrier discharge lamps can readily initiate a range of biological, chemical and physical processes. Initial applications to the generation of ozone for water disinfection, have now extended to photochemical degradation and synthesis, polymerization, and materials deposition at the boundaries of microand nano-science. These sources are also finding possible applications in the microtechnology industry and with the reduction in device geometries towards the nanoscale, their ability to induce specific changes in nanostructures will further highlight their potential towards new generations of nano-fabrication processes requiring low-temperature multi-layered materials. 2. Excimer Radiation from Gas Discharges Dense rare gases at pressures around or above atmospheric pressure have a special property that enables the efficient conversion of electron kinetic energy to electronic excitation energy initially stored in excited atomic and ionic states. In this pressure regime, this excitation is funneled rapidly to a few low-lying atomic and excimer energy levels.1 Excimers (excited dimmers or trimers) are unstable excited molecular complexes, which under normal conditions do not possess a stable ground state. Under the influence of short-pulsed particle bombardment these may disintegrate giving off their binding energy in the form of UV or VUV radiation. As an example, the major reactions for the formation of Xe2* excimers are:
NANOFABRICATION BY UV EXCIMER LAMPS
3
Xe*
Xe*
and
3
P1 , 3 P0 2 Xe o Xe2*
P1 , 3 P2 2 Xe o Xe2*
1
63
¦u Xe
(1)
(2)
1
¦ u Xe
Using Figure 1 to illustrate this further, Xe2* excimer formation only occurs at pressures above 0.1 bar. A three body reaction converts the excited Xe atom, Xe* to the exciplex Xe2* whose radiative decay is faster than that of the Xe* itself, which is responsible for a resonance emission at 147 nm (see Figure 2). The emission bands from the bottom two excimer states dominate and these states decay only by radiation and do not interact with the ground state. The fluorescence radiation of these decaying excimer complexes is normally
Figure 1. Photon energy generated by various excimer sources correlated to common bond dissociation energies (eV).
Figure 2. Scheme showing the generation of VUV line radiation at low pressure and the generation of VUV excimer radiation at 172 nm at about atmospheric pressure. 2
64
I.I. LIAW AND I.W. BOYD
restricted to a narrow band of wavelengths (10–15 nm) with a theoretical efficiency between 45–80%3–7 and particle densities around 1025 m-3. The second excimer continua from pure rare gas dimers are shown in Table 1. Using this principle, simple and efficient excimer lamps may be constructed utilizing different rare gases to induce different UV wavelengths. The use of vacuum ultraviolet Xe2* excimer sources has found widespread application due to the availability of high purity silica tubes from which they are constructed which transmit VUV radiation down to 170 nm with little loss. For shorter wavelengths, special window materials such as LiF, MgF2 or CaF2 must be used. At wavelengths <120 nm no known transparent window materials are available and windowless arrangements have to be deployed.8,9 Emission from rare gas mixtures can spread over a wider spectral region due to the formation of heteronuclear diatomic molecules.10,11 Mixtures of rare-gas halide exciplex molecules of the type RgX* (Rg: He, Ne, Ar, Kr or Xe; X: F, Cl, Br or I)3–7 are used to form bound-free BoX transitions which result in a spontaneous high efficiency VUV/UV emission and even narrower spectral bandwidths, typically 2–4 nm.12,13 A summary of several exciplexes and their respective peak wavelengths is shown in Table 2, while Figure 3 shows a selection of characteristic emission spectra produced by dielectric barrier discharge lamps. TABLE 1. Rare gas dimers with corresponding peak wavelengths in nm (commercially available lamps in thick print). He2* 74
Ne2* 83
Ar2* 126
Kr2* 146
Xe2* 172
XeCI* KrCI* Xe2* Kr2*
Intensity
Ar2*
308
222
172
146
126
Wavelength (nm)
Figure 3. Characteristic emission spectra of dielectric barrier discharge lamps.
NANOFABRICATION BY UV EXCIMER LAMPS
65
TABLE 2. Halogen and rare gas halide excimers with corresponding peak wavelengths in nm (commercially available lamps in bold). F2*
ArCl*
ArF*
KrCl*
KrF*
XeI*
Cl2*
XeBr*
Br2*
XeCl*
I2*
XeF*
157
175
193
222
248
253
259
282
289
308
342
354
2.1. SILENT DISCHARGE DRIVEN EXCIMER LAMPS
Excimer lamp systems (also known as Dielectric barrier discharge (DBD) lamps) based on (non-equilibrium) silent discharges can be operated at pressures up to several bar and enable gas excitation without the need for metal electrodes to be in direct contact with the aggressive plasma. For practical applications, this has proved to be much easier to implement compared to high-energy electron beam generators or pulsed capacitor discharges that also tend to be expensive and operate at low repetition rates. These lamps may be constructed from VUV transparent high purity silica (Suprasil) that can have internal protective coatings applied to resist fluorine attack or functional coating formation that would otherwise reduce the ignition voltage. Since Suprasil is an insulator, it cannot pass dc current, thus alternating voltages are required for their use in DBD’s. The dielectric constant and thickness in combination with the time derivative of the applied voltage determine the amount of displacement current that can pass through the dielectric. Additionally, the electric field must be sufficiently high to initiate gas breakdown. Switched mode power supplies operating between 20 kHz–15 MHz and voltages stepped up using transformers or resonant circuits from a few 100 V to several kilovolts (peak-topeak) provide an efficient and reliable driver of these discharges. Sealed lamps of different planar and cylindrical geometries can be designed as shown in Figure 4. Planar designs use electrodes either on both sides of the discharge space applied by screen printing technologies or embedded within the dielectric on one side of the discharge space (co-planar configuration) resulting in a surface Discharge Gap
Perforated or Mesh Electrodes
UV Transparent Dielectric (Silica)
Cooling Duct Mirror HV AC generator Electrodes
Co-Planar Electrode Configuration 13
Figure 4. Sealed cylindrical and planar dielectric-barrier discharge excimer lamp configurations.
66
I.I. LIAW AND I.W. BOYD
discharge being established between adjacent electrodes. In this way, internal phosphor layers are used to convert VUV excimer radiation to visible light within sealed glass cells, which form the basis of mercury-free fluorescent lamps and of flat plasma display panels. High power lamps often utilise the cylindrical form shown in Figure 5 with a cooling channel through the middle, these are the most common form of excimer UV sources. The width of the discharge gap ranges from 0.1 to several millimeters with filling pressures ranging from 0.1 to 50 bar. A third (buffer) gas such as He or Ne is added to the binary excimer-forming mixture to facilitate ignition and provide additional control over the electron energy distribution. For configurations with discharge gaps of a few millimeters and a gas fill of around 1bar, operating frequencies range from 50 Hz and 14 MHz with applied voltage amplitudes of a few kilovolts being required for ignition. At very high frequencies, the current limitation of the dielectric is reduced and for this reason DBDs are operated between line frequency and about 10 MHz. In the case of cylindrical designs, additional photon flux is usually obtained by using a specially mirror polished aluminum (6061-grade) reflector on the side opposite to the irradiation direction in order to reflect photons being emitted in the opposite direction. Commercial excimer lamps are available at 126 nm (Ar2*), 172 nm (Xe2*), 222 nm (KrCl*) and 308 nm (XeCl*) with typical efficiencies from 5–40%. XeCl* lamps of up to 2 m in length are available for UV curing applications. XeI* and XeBr* lamps radiating at 25315 and 282 nm16 respectively have also been reported. Room temperature operation is achieved through the use of cooling water allowing operation at high electrical input powers.
Figure 5. Cylindrical incoherent excimer UV source. 14
NANOFABRICATION BY UV EXCIMER LAMPS
67
As previously mentioned, open or windowless configurations have been used for lower wavelength systems. A good illustration of this concept is presented by Lenk and Menhert8 and shown in Figure 6. The idea is to initiate the discharge within a large chamber containing the both the excimer-forming gas and as well as the sample to be irradiated.
Figure 6. Windowless Ar2* excimer VUV irradiation system. (After M. Lenk and R. Mehnert.8)
3. Applications of Excimer Lamps 3.1. VUV CLEANING OF SURFACES
The cleaning of glass substrates in LCD manufacturing, of reticules for VUV lithography and in wafer processing is a standard industrial process, involving the removal of residual organic surface contaminants, such as aliphatic and aromatic hydrocarbons, which helps promote adhesion and improve wettability. The photon emission of Xe lamps at 7.25 eV can break most molecular bonds in organic compounds but is not strong enough to split the resulting product molecule CO2 and is used for this purpose. This cleaning process is performed in an Ar/O2 mixture at pressures in the 10–100 mbar range. Ar, which is used as a carrier gas, is transparent to the radiation whilst the O2, which absorbs the UV strongly, splits into the very reactive constituents O(3P) and O(1D). Ozone (O3) results from the recombination of these O atoms with O2 molecules and itself absorbs the radiation to photo-dissociate yielding more O(1D) atoms. The cleavage of organic compounds of the general form CxHyOz produces H2O molecules that also absorb this radiation strongly and following excitation, homolyse with the formation of hydroxyl radicals and hydrogen atoms. The result is a highly reactive gas mix containing O3, O(3P), O(1D) and OH radicals,
68
I.I. LIAW AND I.W. BOYD
which produce a powerful cleaning action on the surface contaminants. The intermediate molecular fragments are transformed into stable end-products of H2, H2O, CO and CO2, which are carried away in the gas stream. This produces a cleaner surface with a wettability contact angle of 5–10q compared with 30– 60q prior to the VUV cleaning process, depending on the nature of the substrate. Present technology uses parallel arrays of linear cylindrical Xe lamps that provide up to 100 mW/cm2 of irradiance. The chemical-free nature of such a cleaning process is advantageous when considering the reduced geometrical dimensions of nano-devices and the challenging juxtaposition of materials within the device matrix. 3.2. PHOTO-INDUCED DEPOSITION OF INSULATING LAYERS OF HIGH OR LOW PERMITTIVITY
Many candidate materials have been suggested to replace the ubiquitous SiO2 or more recent SiOxNy and SiN variations as the gate dielectric in future MOS devices, in order to avoid undesirable quantum tunneling in ever thinning gate structures.17 One of these was Ta2O5 but due to non-optimum band-offset characteristics, others are presently preferred. However, due to its compatibility with ULSI processing and its chemical and thermal stability, there are many other possible applications.18,19 Figure 7 schemes an excimer UV lamp system used for the growth of thin Ta2O5 layers on Si. It comprises two stainless steel vacuum chambers (lamp and reactor sections) separated by an MgF2 window, which is transparent to the VUV irradiation generated by the lamp source. Cylindrical excimer UV lamps of Xe2* (O = 172 nm) or KrCl* (O = 222 nm) were used to irradiate <100> orientation c-Si samples (n-type, 2–4 : cm resistivity), which were placed atop a heated (100–400qC) substrate stand, through a low pressure gas mixture with an output intensity of about 30 mW/cm2.20 The gas used was formed by first externally vaporizing a Ta Metalorganic precursor, which was then introduced into the chamber using an Ar carrier gas, and mixed with N2O at a fixed flow rate. A more complete description of this reactor is published elsewhere.21 The basic photochemistry inducing the CVD growth of Ta2O5 from tantalum ethoxide (Ta(OC2H5)5) precursor and N2O follows the reaction:
N 2O hQ o O 1D N 2 1 ¦ g
(3)
where the resultant O(1D) oxygen species reacts with the tantalum ethoxide causing its dissociation through a series of reactions leading to deposition. Figure 8 shows the Arrhenius plots of the deposition rate of photolytic versus
NANOFABRICATION BY UV EXCIMER LAMPS
69
Figure 7. Schematic diagram of photo-CVD system incorporating excimer UV lamps.
Figure 8. Arrhenius plot of the growth rate of Ta2O5 deposited by photo- and thermal-CVD on Si.
pyrolytic deposited Ta2O5. Clearly the growth rate of the thermally deposited films strongly depends on substrate temperature. It is also particularly slow and essentially negligible under 400qC with a high activation energy of 1.97 eV. The photolytic film growth is only slightly dependent upon temperature and exhibits a significantly lower activation energy of 0.078 eV, indicating a quite different set of reaction mechanisms. This reduced activation energy enables substrate temperatures as low as 100qC to be used. Following deposition, a subsequent UV annealing step has been shown to lead to the reduction of leakage currents in the as-deposited Ta2O5 layers.
70
I.I. LIAW AND I.W. BOYD
The proposed mechanism for the UV annealing effect is attributed to the active oxygen species (O(3P) and O(1D) and O3, as discussed in (Section 3.1), which are adsorbed onto the Ta2O5 surface, diffuse, accept electrons, and fill any voids, as shown schematically in Figure 9. The subsequent decrease in the number of voids leads to a reduction of leakage current. However, these active oxygen species can also react with Si either at the Ta2O5/Si interface or with any Si species that may be present at the surface, leading to the additional formation of a thin Si oxide layer. The likely contributions to the reduced leakage current provided by the UV annealing step are summarized as: x
Active oxygen species react with, and reduce or remove any suboxides present leading to improved stoichiometry.
x
The active oxygen species decrease the density of defects and oxygen vacancies.
x
A native oxide layer formed at the Ta2O5 and Si interface and/or on the Ta2O5 surface by the reaction between the active oxygen species and Si lead to improved interfacial quality.22
x
Densification of the deposited layers and removal of any impurities present in the as-deposited films.
x
Improvement of surface morphology and reduction of the as-deposited particle size.
Figure 9. Model of the mechanism for reduction of leakage current of Ta2O5 films by UV annealing.
NANOFABRICATION BY UV EXCIMER LAMPS
71
Figure 10 shows the (a) I-V and (b) C-V characteristics of Al/Ta2O5/Si MOS capacitors fabricated using the deposited films. The I-V plot shows that following UV annealing the leakage current is dramatically reduced (up to three orders of magnitude). Table 3 summarises this for various annealing times. The C-V plot of Figure 10b shows that the fixed oxide charge density also decreases with increased annealing times. The flat-band voltages (VFB) are negative indicating the presence of positive fixed charges near the Ta2O5/Si interface. Without annealing the VFB was –1.8 V. This reduced to –0.8 and –0.1 V after annealing for 0.5 and 1 h respectively indicating a reduction in the C-V hysteresis effect and positive surface charge. The dielectric constant of the films was determined from C-V measurements using:
C
H0 kA
(4)
d
10 3 10 4
1.2 250˚C
10 5
350˚C
10 6
400˚C
10 7
0.8 0.6 0.4
8
10
0
0h 0.5 h 1h
1
C /Co
Leakage current density (A/cm)2
where C is the measured capacitance, H0 the electric constant, k the dielectric constant, d the thickness and A the area. The dielectric constant value was found to be 24 (as-deposited).
1.2 0.8 0.4 1.6 Electric field (MV/cm)
2
0.2
8
6
4
2 0 2 Voltage (V)
4
6
8
Figure 10. (a) I-V and (b) C-V characteristics of MOS capacitors fabricated from the deposited Ta2O5 films.
TABLE 3. Comparison of electrical properties of photo-CVD Ta2O5 films at various annealing times. Annealing time (h) 0 0.5 1
Leakage current density at 1MV/cm (A/cm2)
VFB(V)
2.19 u 10– 5 2.55 u 10– 6 1.63 u 10– 8
– 1.8 – 0.8 – 0.1
Fixed charge density (cm–2) 2.9 u 1011 2.3 u 1011 3.7 u 1010
72
I.I. LIAW AND I.W. BOYD
Photolytic processing induced at low temperatures minimizes problems with atom diffusion, dopant redistribution and defect generation otherwise caused at higher temperatures. Further, the surface is not subjected to the damaging ionic bombardment that can occur in other physical deposition methods. The ability of UV light to induce specific gas phase reactions (e.g. the generation active oxygen species) on or at the surface considerably reduces any temperature dependence and allows deposition to occur at much lower temperatures. Reduced device dimensions make the performance at the metal interconnect level crucial. Propagation delays, cross talk, noise and power dissipation at this level becomes significant due to resistance-capacitance (RC) coupling. This delay may be reduced through the incorporation of low permittivity dielectric materials and /or high conductivity metals. Polymeric films show great promise for this purpose with their low dielectric constant and the ease with which they may be applied through patterning and their high thermal stability.23,24 UV photons can play a significant role in the curing of polyamic acid films with results demonstrating the complete transformation of the acid layer to polyimide at temperatures as low as 150qC. The UV processing at lower temperatures and shorter times compared to those for conventional furnace methods resulted in leakage currents an order of magnitude smaller.25,26 Dielectric constants between 1.7 and 3.6 were readily obtained and have clear potential applications as interlayer dielectrics for advanced ULSI devices. 3.3. LOW TEMPERATURE UV ASSISTED OXIDATION OF SI, GE AND SIGE
Continuous downscaling of linear devices in recent years has imposed a severe limitation on the thermal budget associated with the several process steps employed in fabrication. With the approaching nano-scale levels requiring even higher degrees of integration, low temperature alternatives to traditional thermal oxidation methods around 1,000qC are greatly desirable. Excimer lamp radiation can considerably enhance oxidation rates at temperatures up to half that value. Figure 11 shows the oxide thickness of oxides grown at 5 mbar as a function of exposure time compared to previously published data27 for layers grown using a low pressure mercury lamp, visible radiation, ozone and conventional furnace oxidation. The observed oxidation rate using the excimer lamp is by far the highest being about 90 times greater than thermal oxidation at 612qC. Thermal oxidation at 450qC is shown to be negligible even after oxidation times in excess of 5 h. When compared to the same process using the low pressure Hg lamp at 350qC, the excimer lamp is three times more effective. We already know that the 172 nm radiation easily dissociates O2 and that the atoms released can produce ozone which can also be decomposed by the
NANOFABRICATION BY UV EXCIMER LAMPS
73
Figure 11. Oxide thickness grown as a function of time for different photon, particle and thermal sources.
VUV light and that all these species can subsequently react with the Si surface forming SiO2. The photo-oxidation process may be explained by: (1) transport of the oxygen species to the oxide surface, (2) diffusion through the oxide, and (3) reaction at the Si-SiO2 interface. With the overall reaction rate being governed by the slowest reaction. Simple MOS capacitors with 11 nm thick UV-grown oxide layers with an evaporated Al contact of area 3.2 u 10-3 cm2 have been fabricated. Fixed charge densities of the as-grown films were shown to be 2.4 u 1011 cm-2, which are nowhere comparable to the best high temperature thermal oxide films grown in strict clean room conditions. Rapid thermal oxides grown at 950qC also contain high fixed oxide charge density in the 1011cm–2 range.28 However, an additional UV annealing step (up to 2 h in 1 atm of O2 at 400qC) significantly reduced the fixed charge density to 4.5 u 1010 cm-2, which is comparable to that for thermally grown SiO2 at 1,030qC. The stretching peak of Si-O-Si observed from FT-IR spectroscopy was also seen to move to higher wave numbers following annealing, indicating a possible increase in film density.29,30 Direct photo-oxidation of Ge at temperatures d 400qC using a Xe2* lamp shows oxidation rates of 0.1 nm/min which are significantly faster than thermal oxidation methods. According to FTIR and XPS analysis, the formed layers have an identical composition to that of thermally grown GeO2.33 SiGe layers were also shown to oxidise rapidly using 172 nm radiation in comparison to the thermal reaction. An enhanced interdiffusion of Si and Ge atoms and low temperature formation of Ge nanoclusters were observed.34,35 Through photoluminescence measurements, an increase in the size of these nanoparticles resulting from the extended UV oxidation process is observed.36 More recently, the use
74
I.I. LIAW AND I.W. BOYD
of deeper UV sources at 126 nm has been demonstrated to achieve higher deposition rates with up to 8 nm/min being achieved. Reduced leakage current densities as low as 5 u 10-8A/cm2 for 17 nm thin layers oxide at a field of 1 MV/cm were observed.31,32 3.4. PHOTO-INDUCED METAL DEPOSITION
The first report of structured deposition of metallic thin films by VUV driven CVD was reported by Calloway et al.37 Utilising a microwave discharge excited rare gas excimer continuum, trimethylaluminum (TMA) vapor was photodissociated resulting in the deposition of aluminum thin films. Since then, structured metal films obtained through the photo-decomposition of spin-on metallo-organic and organometallic precursors has been demonstrated. The nanoscale application here is for high density multichip interconnects packaged to high densities and multilayer stacks. Much of the concept of nanoscaled interconnects may be inferred from the low-temperature decomposition of palladium acetate films.38,39 A flow diagram of the process steps followed is shown in Figure 12.
precursor
1.deposition
substrate
UV 2.exposure
mask activator
3.stripping
4.electroless metal deposition
Cu,Ni,Cr,Ag,Au
Figure 12. UV-induced metal deposition process.
Electroless plating of metal films has the potential for low-cost, highvolume, high-selectivity and low thermal budget metallization suitable for large area nano-scaled processes. It is well known that the activity of catalysts for electroless metal deposition depends on the thickness, distribution and purity of the metal nuclei on the non-catalytic surface. The precursor material can be synthesized to incorporate a wide variety of metals. The different excimer lamp
NANOFABRICATION BY UV EXCIMER LAMPS
75
parameters (wavelength, UV intensity, exposure time), may be correlated to the different absorption coefficients of wavelengths of UV radiation used. The quality of the resulting coating based on the adhesion of the metal layers on the substrate depends decisively on the activator. By exposing through metallic contact masks, film patterning may also be achieved. This method of deposition offers several advantages over other metallization routes. Other than its low temperature, it is insensitive to substrate properties, and providing large-area and low cost layers. 4. Summary We have shown the simplicity, efficiency and ease of deployment of UV excimer lamps for the production of powerful narrow band incoherent radiation. These sealed sources have an extensive range of applications, more than we have presented here, but great potential for relatively cheap large area processes. Progress in the development of shorter wavelength sources or pulsed excimer lamps could lead to even more efficient and wide-ranging uses.40–42
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16.
Lorentz, D. C. (1976) Phys. C 82 19. Kogelschatz, U. (2002) Plasma Src. Sci. Technol. 11, A1–A6. Rhodes, Ch. K., Ed., Excimer Lasers (New York: Springer, 1979 and 1984). Eliasson, B., and U. Kogelschatz, U. (1998) Appl. Phys. B 46, 229. Ivanov, V. V., Klopovskii, K. S., Mankelevich, Yu. A., Rakhimov, A.T., Rakhimova, T.V., Rulev, G. B., and Saenko, V. B. (1996) Laser Phys. 6, 564. Oda, A., Sugarawa, H., Sakai Y., and Akashi H. (2000). J. Phys. D: Appl. Phys. 33, 1507. Carman, R., Ward, B., and Mildren, R. (2001) Proceedings of the XXV International Conference on Phenomena in Ionized Gases (ICIPG-XXV), Nagoya, 4, 331–332. Lenk, M., and Mehnert, R. (2001) Proceedings of the RadTech Europe, Basle, 153–158. Lomaev, M., Skakun, V., Tarasenko, V., Shitts, D., and Lisenko, A. (2006) Tech. Phys. Letts. 32, 590–592. Kubodera, S., Honda, M., Kitahara, M., Kawanaka, J., Sasaki, W., and Kurosawa, K. (1995) Jpn. J. Appl. Phys. 34, L618. Gerasimov, G., Volkova, A., Zvereva, G., and Babucke G. (Eds.) (1998) In: Proceedings of the 8th International Symposium on Science Technology Light sources (LS-8), Griefswald, Germany, 248–249. Shuaibov, A., Shimon, L., and Shervera, I. (1998) Instr. Exp. Technol. 41, 427. Shuaibov, A., Shimon, L., Dashchenko, A., and Shervera, I. (2001) Tech. Phys. 46, 207. Kogelschatz, U., Esrom, H., Zhang, J., and Boyd, I. (2000) Appl. Surf. Sci. 168, 29. Zhang, J., and Boyd, I. (1996) J. Appl. Phys. 80, 663. Zhang, J., and Boyd, I. (1996) J. Appl. Phys. 84, 1174.
76
I.I. LIAW AND I.W. BOYD
17. Mueller, D., Sorsch, T., Moccio, S., Baumann, F., Evans-Lutterodt, K., and Timp, G. (1999) Nature 399, 758. 18. Zhang, J., Lim, B., Dusastre, V., and Boyd, I. (1998) Appl. Phys. Lett. 73, 2299. 19. Chaneliere, C., Autran, J., Balland, B., and Devine, R. (1998) Mater. Sci. Eng. Rep, 22(6), 269–332. 20. Zhang, J., Lim, B., and Boyd, I. (1998) Thin Solid Films 336, 340. 21. Zhang, J., Boyd, I., Mooney, M., Hurley, P., O’Sullivan, B., Beechinor, T., Kelly, P., Crean, G., and Senateur, J. (1999) Mat. Res. Soc. Proc. 567, 397. 22. Gellert, B., and Kogelschatz, U. (1991) Appl. Phys. B 52, 14. 23. Singer, P. (1994) Semiconductor Int’l Oct, 34. 24. Murarka, S. (1996) Solid State Technol. 39, 83. 25. Zhang, J., and Boyd, I. (1998) Opt. Mater. 9, 251. 26. Zhang, J., and Boyd, I. (1999) Appl. Surf. Sci. 24, 352. 27. Boyd, I., Craciun, V., and Kazor, A. (1993) Jpn. J. Appl. Phys. 32, 6141. 28. Eftekhari, G. (1993) J. Electochem. Soc. 140, 787. 29. Pliskin, W. (1968) Thin Solid Films 2, 1. 30. Boyd, I., and Wilson, J. (1987) J. Appl. Phys. 62, 3195. 31. Zhang, J., and Boyd, I. (2002) Appl. Surf. Sci. 186, 64. 32. Fang, Q., Zhang, J., and Boyd, I. (2003) Appl. Surf. Sci., 369, 208–209. 33. Craciun, V., Hutton, B., Williams, D., and Boyd, I. (1998) Electron. Lett. 34, 71. 34. Craciun, V., Boyd, I., Reader, A., Kersten, W., Hakkens, F., Oosting, P., and Vandenhouldt, D. (1994) J. Appl. Phys. 75, 1972. 35. Craciun, V., Boyd, I., Hutton, B., and Williams, D. (1999) Appl. Phys. Lett. 73, 1261. 36. Zhang, J., Fang, Q., Kenyon, A., and Boyd, I. (2003) Appl. Surf. Sci. 364, 208–209. 37. Calloway. A., Galantowicz, T., and Ferner, W. (1983) J. Vac. Sci. Technol. A1, 534. 38. Boyd, I., and Zhang, J. (1997) Nucl. Inst. Methods Phys. B 121, 349. 39. Zhang, J. (1993) Thesis, Karlsruhe University Germany. 40. Hilbert, C., Gaurand, I., Motret, O., and Pouvesle, J. (1999) J. Appl. Phys. 85, 7070. 41. Motret, O., Hibert, C., Pellerin, S., and Pouvesle, J. (2000) J. Phys. D: Appl. Phys. 33, 1483. 42. Okazaki, K., and Nozaki, T. (2002) Pure Appl. Chem. 74, 447–452.
FUNCTIONALIZATION OF SEMICONDUCTOR NANOPARTICLES M.-I. BARATON* SPCTS-UMR CNRS 6638, University of Limoges 123 Avenue Albert Thomas, 87060-Limoges, FRANCE
Abstract – Functionalization of nanoparticles surface by attachment of organic entities is used to achieve and tailor many new properties, such as lubrication, optical response, chemical sensing, or biocompatibility. But because at the nanometer scale the surface properties significantly contribute to the overall properties, the consequences of the surface modifications have to be thoroughly evaluated. This paper demonstrates the relevance of Fourier transform infrared spectroscopy to the study of the surface reactions leading to the functionalization, and of the stability of the functionalized surface under the expected working conditions. In the case of semiconductor nanoparticles, this technique additionally allows the analysis of the impact of the functionalization on the electrical properties. This will be illustrated by the case study of tin oxide nanoparticles for chemical gas sensors. The correlation between surface chemistry and electrical properties is critical to optimize the nanoparticles functionalization for the targeted properties.
Keywords: Surface chemistry, surface chemical species, FTIR spectroscopy, gas-surface interactions, semiconductor metal oxides, chemical gas sensors.
1. Introduction An important characteristic of nanoparticles is their enormous specific surface area, that is their very large surface to bulk ratio. Obvious consequences are (i) a high surface reactivity of nanoparticles, leading to fast contamination by any surrounding atmosphere, and (ii) a significant contribution of specific surface
______ *
To whom correspondence should be addressed: M. Baraton, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
77
78
M.-I. BARATON
properties to the overall properties. Therefore, at the nanometer scale, material properties reflect the nature of unavoidable adsorbates and it becomes extremely difficult to define intrinsic properties independent of environmental factors. Attachment of organic entities to a semiconductor surface has been investigated for years in order to achieve and tailor many new properties, including lubrication, optical response, chemical sensing, or biocompatibility. Surface protection or passivation is essential for high-added value applications, and surface phenomena have always been a cornerstone of the microelectronic industry.1 In the case of semiconductor nanoparticles, it is therefore critical to control the surface chemistry and to carefully monitor the surface functionalization. Indeed, electrical and optical properties of the nanoparticles may be adversely affected by the formation of new surface chemical groups. For example, surface chemistry has been shown to impact optical properties of nanoparticles, and computationally it has been predicted that surface chemistry has a great influence on the lowest unoccupied molecular orbital (LUMO).2 The characterization of the surface layers formed during the functionalization remains a challenge, but it is not the only difficulty to overcome. Indeed, when surface functionalization of semiconductor nanoparticles becomes necessary to optimize the fabrication of a device (e.g. homogeneous dispersion of nanoparticles in an organic matrix), to modify the surface chemistry (e.g. control of the surface reactions at the interface), to tailor the electrical or optical properties (e.g. enhancement of gas sensor selectivity), the following steps have to be equally and simultaneously mastered3: 1. Chemical characterization of the first atomic layer of the nanoparticles surface before and after surface functionalization. 2. Selection of the most appropriate molecules to be used for surface functionalization. 3. Determination of the conditions under which the nanoparticles surface should be pretreated to optimize the functionalization. 4. Monitoring of the surface reactions during the functionalization. 5. Stability of the surface functionalization under the final working condi-tions of the functionalized semiconductor. 6. Analysis of the consequences on the surface reactivity and on the elec-trical properties of the semiconducting nanoparticles. The chemical composition of the surface layers on functionalized nanoparticles can be studied using Fourier transform infrared (FTIR) spectroscopy,4–6 X-ray photoelectron spectroscopy (XPS)7,8 or nuclear magnetic resonance (NMR).7,9 Surface layer thicknesses are usually evaluated by transmission electronic microscopy (TEM) but monolayers are fragile to high-energy X-ray and electron beams.10
SEMICONDUCTOR NANOPARTICLES
79
As it will be seen in the following section, the advantage of FTIR spectroscopy over other characterization methods is the possibility to not only identify the surface chemical species on nanoparticles and to investigate their surface chemistry but also to monitor in situ the surface functionalization while simultaneously evaluating the consequences on the electrical properties of semiconductor nanoparticles. Moreover, FTIR spectroscopy is a non-destructive analytical method. 2. FTIR Spectra and Electronic Properties Fundamentals of the theory of infrared spectroscopy can be easily found in many references.11–13 It is however less known that in addition to the standard bulk analysis of inorganic materials, FTIR spectroscopy allows both the characterization of the chemical entities adsorbed on nanoparticles surface and the analysis of the nanoparticle surface reactivity. The detailed surface characterization procedure of nanosized particles can be found elsewhere.4,14–16 A brief summary can be given as follows. The first and obvious step consists in observing the transmission spectrum of the nanoparticles and identifying the absorption bands which cannot be assigned to the bulk modes due to the frequency range where the absorption maxima fall. In general, these absorption bands originate from organic entities such as undissociated precursors or/and contamination by the environment during or after the synthesis. It has to be noted that hydrolysis is the most probable contamination to be found on a material surface resulting in hydroxyl OH surface groups. To check whether identified foreign chemical species are located on the surface or trapped inside the bulk, the sample is desorbed at increasing temperatures (activation). Usually, surface chemical species are either released (desorbed) or transformed by heating under vacuum. It should be noted that after activation, a surface is no longer in equilibrium (activated surface) and will therefore easily react as soon as it is in contact with surrounding molecules. The second step of the surface analysis consists in characterizing the surface reactive sites. This is achieved by adsorbing selected molecules (referred to as probe-molecules) onto the activated surface and by comparing the infrared spectrum of the adsorbed probe-molecule with that of its gas phase.4,17 Once the surface is characterized, it is possible to envisage chemical modifications by reaction of appropriate molecules with the surface reactive sites. An activation of the surface by heating under dynamic vacuum is usually necessary to clear the nanoparticle surface from the various impurities and to free surface reactive sites for further grafting of new molecules.
80
M.-I. BARATON
Even though interatomic vibrations are the most studied features in an infrared absorption spectrum, absorption due to surface electronic phenomena represents an intrinsic part of the infrared spectrum of semiconductors. Indeed, the semiconducting property originates from the mobility of free carriers which are electrons (n-type semiconductors) or holes (p-type semiconductors) and, according to the Drude-Zener theory,18–21 these free carriers contribute to the absorption by the material over the whole infrared range. In 1962, Harrick22 observed the modulation of the free carrier density in a silicon wafer by an external electric field. A broad absorption band was identified as the contribution of the free carriers to the infrared absorption spectrum. Most theories predict a On dependence (O being the infrared wavelength) where n generally equals 2 although variations often occur. The infrared absorption changes due to the free carriers are positive or negative depending on whether carriers are added to or subtracted from the surface. For example, electrons subtracted from the space-charge region of an n-type semiconductor surface lead to a decrease of the IR absorption, whereas electrons subtracted from the space-charge region of a p-type semiconductor surface lead to an increase of the IR absorption22,23 (see examples in Section 3 below). Even though in this pioneered experiment, the free carrier density was modulated by an electric field, modulation can also be obtained by adsorbing molecules on the semiconductor surface.23,24 Practically speaking, a decrease of the free carrier density, that is a decrease of the electrical conductivity, leads to a decrease of the IR absorption, whereas an increase of the free carrier density corresponding to an increase of the electrical conductivity leads to an increase of the IR absorption. Therefore, FTIR spectroscopy allows the in situ monitoring of the reactions taking place at the gas-nanoparticle interface simultaneously with the analysis of variations of the background infrared absorption which are related to the variations of the electrical conductivity. It must be stressed that FTIR spectrometry can combine the investigation of two related phenomena (surface reactions and electrical conductivity) involving different thicknesses of the material. Indeed, the surface reactivity concerns only the first atomic layer whereas the thickness variation of the depletion layer may concern up to several tens of nanometers. 3. Case Study of Tin Oxide Nanoparticles Most of the work dealing with the functionalization of semiconductors refers to silicon. But differently, we present here the example of tin oxide nanoparticles. Tin oxide (n-type semiconductor) is one of the most popular materials for the fabrication of semiconductor chemical sensors because of the large variations
SEMICONDUCTOR NANOPARTICLES
81
of its electrical conductivity occurring under adsorption of reducing gases. Moreover, the use of nanoparticles in the fabrication process of gas sensors by screen-printing technology has been demonstrated to largely increase the sensor sensitivity.25,26 However, an important drawback of such sensors is their crosssensitivity to humidity, precluding their operation in outdoor environment. In earlier work,27 we demonstrated that surface functionalization of the titanium oxide nanoparticles with selected organic molecules may be an asset to reduce the cross-sensitivity to humidity. In particular, the surface modification by grafting hexamethyldisilazane (HMDS) proved to be promising even though the sensor sensitivity may be slightly affected by new surface groups. To optimize the surface functionalization, the influence of HMDS on the tin oxide conductivity has to be studied in more details. The tin oxide nanoparticles used in this study were synthesized by Chemical Vapor Synthesis (CVS) whose advantage is the production of nanoparticles with a narrow size distribution and low degree of agglomeration.28 The produced tin oxide nanoparticles have the cassiterite structure with an average primary particle size of 11 nm and they are substoichiometric in oxygen (unpublished results). In situ FTIR experimental conditions are described in details elsewhere.4 3.1. ACTIVATION AND OXIDATION OF TIN OXIDE NANOPARTICLES
Figure 1 shows the infrared surface spectra of the tin oxide nanoparticles after desorption at different temperatures and subsequent oxidations at each desorption step. Indeed, a heat treatment up to 400°C under dynamic vacuum causes oxygen loss, thus leading to a reduced material close to the metallic state. Due to the high density of free electrons, the sample strongly absorbs the infrared radiation, as explained in the previous section. Therefore, an oxidation has to be performed in order to retrieve the tin oxide material. The observed behavior of tin oxide nanoparticles under desorption and oxidation can be related to the high mobility of oxygen in the material. It should be noted that the experiment corresponding to Figure 1 has been performed in situ and that the spectra have not been translated. We can observe that at 200°C (Figure 1d) and 400°C (Figure 1e), adsorption of oxygen leads to a shift of the spectrum baseline toward lower absorbance, thus indicating a decrease of free carrier concentration. In other words, after oxygen adsorption at high temperatures, the sample is more oxidized than the raw material. This is in agreement with the observed sub-stoichiometry of the as-synthesized tin oxide nanoparticles. The absorption bands in the highest wavenumber region of the spectra (4,000–3,000 cm–1) are due to the stretching vibrations of hydroxyl groups bonded to surface tin atoms in different coordination states.29
82
M.-I. BARATON
Figure 1. FTIR spectra of the tin oxide nanoparticles at room temperature (a), after desorption at room temperature (b), after desorption and subsequent O2 adsorption at 100°C (c), 200°C (d), 400°C and cooled down at room temperature (e).
3.2. ADSORPTION OF HMDS AND HMDSO ON TIN OXIDE NANOPARTICLES
The adsorption of HMDS has been performed at room temperature on the tin oxide nanoparticles previously activated (that is desorbed and oxidized) at 400°C. For comparison purposes, the adsorption of hexamethyldisiloxane (HMDSO) has been studied under similar conditions. HMDS and HMDSO usually react through the surface hydroxyl groups to form trimethylsilyl groups according to the reactions: (CH3)3Si-(NH)-Si(CH3)3 + 2 Sn-OH o 2 Sn-OSi(CH3)3 + NH3 (CH3)3Si-O-Si(CH3)3 + 2 Sn-OH o 2 Sn-OSi(CH3)3 + H2O Both HMDS and HMDSO adsorption experiments were conducted in the same way. The two reactants were adsorbed at room temperature followed by a quick evacuation, thus allowing the elimination of the gas phase and of the physisorbed species. Then, the grafted samples were desorbed at 400°C and, in both cases, the strong reduction of the samples was proved by their complete opacity to the infrared radiation. To oxidize the grafted samples, oxygen was adsorbed at 400°C and the spectra were then recorded after cooling the samples down to room temperature under oxygen. Figures 2 and 3 give the spectra at different steps of the HMDS and HMDSO adsorption experiments, respectively. In these two cases, the experiments have
SEMICONDUCTOR NANOPARTICLES
83
been made in situ and comparison of the absorbance values can be achieved within each of the experiments. In Figure 2, it can be seen that after evacuation at room temperature, the Q(CH3) and G(CH3) absorption bands of the trimethylsilyl groups are still very visible proving that HMDS is linked to the tin oxide surface.3 Moreover, one can note a slight shift of the spectrum baseline towards higher absorbance value, which indicates a higher density of free electrons. After desorption and oxidation at 400°C, the shift of the baseline toward higher absorbance is stronger. Differently from the non-grafted material, the HMDS-grafted tin oxide nanoparticles are not totally re-oxidized once they have been reduced by the thermal desorption, as indicated by the spectrum baseline which does not go back to its original position on the absorbance scale. In addition, the Q(CH3) and G(CH3) absorption bands of the trimethylsilyl groups have almost disappeared whereas a band appears at 3,730 cm–1 assigned to the Q(OH) stretching vibration of newly formed Si-OH silanol groups.3 These silanol groups originate from the oxidation of the trimethylsilyl groups and are bonded to the surface oxygen as proved by the band at 990 cm–1 assigned to the Q(O-Si) vibration in Sn-O-Si groups. Figure 3 depicts the effects of HMDSO adsorption. Formation of trimethylsilyl groups bonded to the tin oxide surface is again observed and these groups resist a desorption at room temperature. The formed H2O molecules, as mentioned in the above reaction, are responsible for a broader and more intense absorption band than in the case of HMDS (centered around 3,200 cm–1). This
Figure 2. Infrared spectra of the tin oxide nanoparticles recorded at room temperature: after desorption and O2 adsorption at 400°C (a); after HMDS adsorption and subsequent evacuation at room temperature (b); after desorption and O2 adsorption at 400°C (c).
84
M.-I. BARATON
Figure 3. Infrared spectra of the tin oxide nanoparticles recorded at room temperature: after desorption and O2 adsorption at 400°C (a); after HMDSO adsorption and subsequent evacuation at room temperature (b); after desorption and O2 adsorption at 400°C (c).
band is assigned to the stretching vibration of H-bonded OH groups. Differently from HMDS, HMDSO adsorption significantly reduces the tin oxide nanoparticles. However, a heat treatment at 400°C followed by an oxidation leads to an almost complete recovery of the oxidation state while silanol groups are formed by oxidation of the trimethylsilyl groups. At the end of the two experiments, the surface spectra of the HMDS-grafted (Figure 2c) and HMDSO-grafted (Figure 3c) tin oxide nanoparticles are similar proving that both functionalizations similarly modify the surface chemistry but, in the case of HMDS, the sample is significantly reduced. The difference in the tin oxide oxidation states at the end of the functionalization experiments can be explained by the different gaseous molecules produced during the grafting reactions. While H2O is able to reduce tin oxide nanoparticles at room temperature without preventing an oxidation at higher temperature, NH3 molecules first coordinate on the surface at room temperature (bands at 3,380 and 1,609 cm-1 respectively corresponding to stretching and bending vibrations of coordinated NH3 species).30 Then, as the desorption temperature is increased, the NH3coordinated species are released from the surface. Because NH3 is a strong reducing gas, the free carrier concentration is drastically and irreversibly affected. It has to be noted that this heat treatment under oxygen (or air) of the modified nanoparticles corresponds to the pretreatment undergone by nanoparticlesbased chemical sensors before operation. Therefore, this heat treatment gives an
SEMICONDUCTOR NANOPARTICLES
85
excellent picture of the evolution and subsequent stability of the functionalization under working conditions of the sensors. In our previous works, it has been proved that surface modification by HMDS reduces the cross-sensitivity to humidity.27 Although the surface chemistry of the tin oxide nanoparticles appears to be identical after modification with HMDSO, further experiments currently are in progress to investigate the consequences of both types of tin oxide modification on the sensitivity of the gas sensors based on these modified nanoparticles. 4. Conclusion and Outlook To summarize, surface functionalization is a powerful method which can simultaneously allow modification of the surface chemistry, tailoring of the electrical properties, and optimization of the fabrication of a device. However, surface functionalization of semiconductor nanoparticles by attaching chemical groups requires a chemical investigation of the surface before, during and after modification, and an analysis of the functionalization consequences on the electrical properties. It is known for example that, in the development of molecular electronics, the electrical properties will depend upon the structural properties of the surface. Depending on the envisaged applications, molecular entities to be attached must be adequately selected, their stability must be checked under the expected working conditions, and post-treatment must be determined to properly tune the final characteristics while retaining the benefit from the functionalization. ACKNOWLEDGEMENT
The author is grateful to Prof. Horst Hahn’s group (FZK, Karlsruhe, Germany) for providing her with tin oxide nanopowder.
References 1. Bent, S. (2002) Surf. Sci. 500, 879–903. 2. Veinot, J. (2006) Chem. Commun. 4160–4168. 3. Baraton, M. (2004) Surface functionalization of semiconducting nanoparticles, in H.S. Nalwa (ed.), Encyclopedia of Nanoscience and Nanotechnology, American Scientific, Stevenson Ranch, CA, Vol. 10, pp. 267–281.
86
M.-I. BARATON
4. Baraton, M. (1999) FT-IR surface spectrometry of nanosized particles, in H.S. Nalwa (ed.), Handbook of Nanostructured Materials and Nanotechnology, Academic, San Diego, CA, pp. 89–153. 5. Buriak, J. (2002) Chem. Rev. 102(5), 1271–1308. 6. Rogozhina, E., Belomoin, G., Smith, A., Abuhassan, L., Barry, N., Akcakir, O., Braun, P., and Nayfeh, M. (2001) Appl. Phys. Lett. 78, 3711–3713. 7. Zou, J., Baldwin, R., Pettigrew, K., and Kauzlarich, S. (2004) Nano Lett. 4, 1181–1186. 8. Legrand, J., Taleb, A., Gota, S., Guittet, M., and Petit, C. (2002) Langmuir 18(10), 4131– 4137. 9. Hua, F., Swihart, M., and Ruckenstein, E. (2005) Langmuir 21(13), 6054–6062. 10. Liao, Y., and Roberts, J. (2006) J. Amer. Chem. Soc. 128(28), 9061–9065. 11. Herzberg, G. (1962) Molecular Spectra and Molecular Structure, Van Nostrand, Princeton, NJ. 12. Wilson, Jr., E., Decius, J., and Cross, P. (1955) Molecular Vibrations. The Theory of Infrared and Raman Vibrational Spectra, Dover, NewYork. 13. Griffiths, P., and de Haseth, J. (1986) Fourier Transform Infrared Spectrometry, Wiley, Chichester. 14. Hair, M. (1967) Infrared Spectroscopy in Surface Chemistry, Marcel Dekker, New York. 15. Boehm, H., and Knözinger, H. (1983) Nature and estimation of functional groups on solid surfaces, in J.R. Anderson and M. Boudart (eds.), Catalysis, Springer, Berlin, Vol. 4, 39–207. 16. Davydov, A. (1984) Infrared Spectroscopy of Adsorbed Species on the Surface of Transition Metal Oxides, Wiley, New York. 17. Lavalley, J. (1996) Catal. Today 27, 377–401. 18. Harrick, N. (1960) J. Phys. Chem. Solids 14, 60–71. 19. Harrick, N. (1963) Ann. N.Y. Acad. Sci. 101, 928–959. 20. Harrick, N. (1967) Internal Reflection Spectroscopy, Interscience, Wiley, New York. 21. Gibson, A. (1958) J. Sci. Instrum. 35, 273–278. 22. Harrick, N. (1962) Phys. Rev. 125(4), 1165–1170. 23. Baraton, M., and Merhari, L. (2005) Synth. React. Inorg., Metal-Org. Nano-Metal Chem. 3, 733–742. 24. Chabal, Y. (1988) Surf. Sci. Rep. 8, 211–357. 25. Baraton, M. and Merhari, L. (2001) Mater. Trans. 42(8), 1616–1622. 26. Williams, G., and Coles, G. (1998) J. Mater. Chem. 8(7), 1657–1664. 27. Baraton, M., and Merhari, L. (2004) J. Europ. Ceram. Soc. 24, 1399–1404. 28. Hahn, H., Winterer, M., Seifried, S., and Srdic, V. (2000) in G.M. Chow, I.A. Ovid’ko, and T. Tsakalakos (eds.), Nanostructured Films and Coatings, NATO Science Series, Kluwer, Dordrecht, Vol. 78, pp. 1–10. 29. Tribout, J., Chancel, F., Baraton, M., Ferkel, H., and Riehemann, W. (1997) in Key Engineering Materials, Euro Ceramics V, Trans Tech, Zurich, Vol. 132–136, pp. 1341–1344. 30. Chancel, F., Tribout, J., and Baraton, M. (1997) in Key Engineering Materials, Euro Ceramics V, Trans Tech Publications, Zurich, Vol. 132–136, pp. 1341–1344.
FLEXOELECTRICITY: A UNIVERSAL SENSORIC MECHANISM IN BIOMEMBRANES AND IN CHEM.-BIOSENSORS A.G. PETROV* Institute of Solid State Physics, Bulgarian Academy of Sciences, 72 Tzarigradsko chaussee, 1784 Sofia, BULGARIA
Abstract – Flexoelectricity provides a reciprocal relationship between electricity and mechanics in membranes, i.e., between membrane curvature and polarization. Experimental evidence of biomembrane flexoelectricity (including direct and converse flexoelectric effect) is reviewed. Biological implications of flexoelectricity in mechanosensitivity, electromotility and hearing is underlined. Flexoelectricity enables membrane structures to function like soft micro- and nano-machines, sensors and actuators, thus giving important input to sensoric applications.
Keywords: Biomembrane, flexoelectricity, mechanosensitivity, biosensors.
1. Introduction This article presents an account on some basic mechanoelectric properties of membranes: flexoelectricity, mechanocapacitance and mechanoconductivity. Their investigation became possible thanks to the BLM model, namely to the fact that a BLM represent a freestanding bilayer that is supported along its periphery only. Experimentally, the way to probe BLM mechanoelectricity is to subject a membrane to a pressure differential, either static or dynamic, or both. BLM models used by us comprise the classic, celebrated Mueller-Rudin-TienWescott discovery of BLM painting a over a hydrophobic hole,1 as well as the tip-dip method where a microsize BLM is patch-clamped inside a hydrophilic borosilicate glass micropipette.2 In the present review we concentrate on the first system.
______ *
To whom the correspondence should be addressed: A. Petrov, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
87
88
A.G. PETROV
Flexoelectricity3 stands for curvature-induced membrane polarization (direct flexoeffect) or voltage-induced membrane curvature (converse flexoeffect), mechanocapacitance is lateral stress-induced capacitance variation, while mechanoconductivity means lateral stress-sensitive ion conduction of a membrane. Stress sensitivity is usually mediated by the presence of ion channels in the lipid bilayer, synthetic or native ones. On the other hand, flexoelectricity and mechanocapacitance can be observed even in unmodified lipid bilayers. All these properties are closely related to some vital function of native membranes, mechanosensitivity and electromotility. They involve two generalized degrees of the membrane, mechanical and electrical ones. Strictly speaking, mechanical degrees of freedom involved are two: curvature change and lateral stretching. Furthermore, in photoactive BLM flexoelectricity becomes light-dependent, i.e. a photoflexoelectricity phenomenon can be demonstrated. This effect involves a further degree of freedom of the membrane, the optical one. In the experiments it is easy to apply controlled amounts of pressure difference, either static or dynamic ones, and measure the current or voltage response of BLM. Quite often, though, these properties are superimposed in the resulting response. In the experimental arrangement of a BLM (and especially in a patch clamped membrane) the two mechanical degrees of freedom are coupled: inducing a curvature of the membrane by its blowing is inevitably related to a lateral stress variation. Nevertheless, since both degrees of freedom are of different symmetry (curvature variation has a polarity, while lateral stretch has not), usually by applying a phase-sensitive analyser it is possible to discriminate between the two contributions to the electric response (see below). 1.1. THEORETICAL REMARKS
The first experimental hints about the generation of a.c. currents by BLMs subjected to oscillating gradients of pressure were obtained in the early 1970s. When a BLM was clamped to certain voltage, these currents were related to mechanoconductivity4 or mechanocapacitance,5 and were found to be 2nd harmonics of the driving pressure’s frequency. Subsequently, we pointed out that the membrane curvature is also a variable in these experiments. Consequently, we related the accidental appearance of “subharmonic” current under zero voltage clamp to the oscillating flexoelectric polarization of the curved membrane3,6: PS = f (c1+c2),
(1)
(c1+c2) is the total membrane curvature, f is flexoelectric coefficient, measured in Coulombs [C] and PS is flexopolarization per unit area.6–8
FLEXOELECTRICITY IN BIOMEMBRANES
89
Phenomenologically, the generation of a “subharmonic noise”, i.e. a 1st harmonic a.c. current by a zero voltage-clamped BLM, follows from the appearance of flexopolarization. Assuming (for simplicity) oscillating spherical curvature and representing it in the form
c1 c 2 c t 2c m sinZ t ,
(2)
Z being angular frequency of oscillations and cm
1 Rm the maximal curvature, we will obtain time-dependent flexopolarization as well. Let us assume that polarization variations instantaneously follow curvature variations3 (for the case of non-zero relaxation time see10): PS t f c t
f 2c m sinZ t ,
(3)
Surface polarization leads to a transmembrane voltage difference (Helmholtz equation, Eq. (4)) that can be measured in an external circuit of two electrodes immersed in the bathing electrolyte on each side of the membrane, connected to a very high impedance electrometer (open circuit conditions, zero current clamp):
U f t
PS
H0
f
H0
2 cm sin Zt
U Z sin Z t .
(4)
The flexoelectric voltage is a first harmonic with respect to curvature oscillation and its amplitude is UZ f H 0 2 cm . When the two electrodes are effectively shorted out via low impedance ammeter (shorted circuit conditions, zero voltage clamp), the displacement current through the meter can be calculated by adopting an equivalent circuit (Figure 1), containing an a.c. voltage generator U f (describing the oscillating flexoelectric voltage according to Eq. (4)) and a capacitor C (where C is the membrane capacitance):
I f t
d (CU f ) dt
dU f . dC Uf C dt dt
(5)
Since PS is by definition the t o t a l membrane polarization due to both permanent and induced dipoles, the relevant membrane capacitance is simply C0 H 0 S0 d , where S0 is the flat membrane area and d is the capacitive thickness of the membrane. When the membrane curvature oscillates, the membrane area will oscillate as well7:
S S 0 ' S sin 2 Z t , where 'S
S0 r 2 cm2 4
(6)
for spherical curvature, with r being the membrane radius. The membrane capacitance will then also oscillate like
90
A.G. PETROV Measuring system
Membrane
RL E out
flexo
Rm $R
Vhold
E
a
in
E
Cm$C In flexo
out
E
in flexo
out flexo
b Figure 1. (a) Equivalent circuit of an oscillating membrane connected to a measuring system; (b) graph of the potential distribution across a planar and a curved membrane. E out and E in are two e.m.f. generators that modulate the surface potentials of the two surfaces of a curved membrane for the case of a positive flexoelectric coefficient, negative surface charge and zero intramembrane field, i.e., zero current clamp. Since the two surfaces’ generators operate in counterphase they can be combined in one flexoelectric generator Uf, as in text. (From Petrov3, with permission from the Publisher.)
C
C0 'C sin 2 Zt .
(7)
The relationship between ' C C0 and ' S S0 is, in general, frequencydependent.9,11,12 With flexoelectricity we are mainly interested in the first harmonic of the response. Capacitance oscillations will contribute to higher harmonics. Therefore, from (4) and (5) we have for the 1st harmonic amplitude of the membrane flexoelectric current IZ f C0 H 0 2cm Z . In this way, measuring U Z and cm , or IZ , C0 and cm , we can determine experimentally the value of the phenomenologically introduced flexocoefficient f in Eq. (1). We have shown3 that several electric multipoles (charges, dipoles, quadrupoles) of membrane components (lipids and proteins) contribute to the flexoelectric coefficient. Experimental elucidation of the importance of different multipoles by their independent variation (especially the partial electric charge per lipid head, E, expressed in percent of proton charge) is of substantial
FLEXOELECTRICITY IN BIOMEMBRANES
91
interest. Another important subject is the influence of surface adsorbed modifiers with electric multipoles that differ from these of the lipid matrix. The variations of the flexoelectric response by such modifiers could then be utilized for sensoric purposes. 2. Flexoelectric Sensors Using Current Response of a BLM
Our first experiments, aimed specifically at registering the flexoelectricity of BLMs from biological lipids (egg yolk lecithin (EYPC),8 bacterial phosphatidylethanolamine (PE)9) were performed in the current registration regime by using a low impedance current-to-voltage converter. The influence of some modifiers of the surface charge and surface dipole, as well as of the membrane conductivity, upon the value of the effect was also studied. 2.1. LIPIDS SURFACE CHARGE AND DIPOLE MODIFIERS
BLMs were formed by the classical method 1 on a 1 mm conical hole in a Teflon cup, immersed in a spectrophotometric glass cuvette, at room temperature. Bacterial phosphatidylethanolamine (PE) ex. E. coli (Koch Light) dissolved in n-decane (20 mg/ml) was used as a membrane-forming solution. As a bathing electrolyte either unbuffered KCl, 10 mM to 1 M (pH 6.2) or buffered KCl with higher or lower pH values was used (KH2PO4 pH 3.0; MES pH 6.0; TRIS pH 8.5; CHES pH 9.3; KH2PO4 pH 10.4). The surface charge of the BLM was varied either by pH variation or by adsorption of the ionic detergent cetyltrimethylammonium bromide (CTAB, Chemapol) added to the electrolyte as a 1 mM water solution. The surface dipole was varied by adsorption of the dipole modifier phloretin (Serva), added as a 10 mM solution in ethyl alcohol. 2.2. METHOD FOR EXCITATION OF MEMBRANE CURVATURE
Membrane curvature oscillations were excited by the method first employed in Ref. 4. The experimental set-up is shown in Figure 2. The Teflon cup with a very small internal volume (0.25 cm2) was closed by a Teflon cap. Two reversible Ag/AgCl electrodes were used for current measurements. One of them was mounted in the cup; another one was immersed in the glass cuvette. Through an opening in the cap connected to a flexible pneumatic pipeline, oscillating air pressure was applied to the electrolyte surface and was transferred in this way to the membrane. An electrically shielded earphone generated
92
A.G. PETROV
Figure 2. Scheme of the experimental set-up for current registration: (1) glass cuvette; (2) Teflon cup; (3) Teflon cap; (4) electrolyte: 10 mM to 1 M KCl, pH 6.2; (5) reversible electrodes, Ag/AgCl; (6) plastic pipeline; (7) ear-phone; (8) function generator, Textronics FG504; (9) currentto-voltage converter, Keithley 427; (10) oscilloscope, C1-78; (11) phase sensitive analyzer, PAR 5204; (12) switch; (13) d.c. voltage source; (14) Boxcar averager, PAR Model 162; (15) X-Y recorder. (From Petrov and Sokolov,9 with permission from the Publisher.)
air pressure. A function generator fed the earphone. The applied frequency was in the range of 10–1,000 Hz. Using a T-pipe, an open branch filled with cotton wool was introduced into the pipeline in order to equilibrate any static pressure difference between the two membrane sides while transferring the dynamic air pressure. 2.3. EVALUATION OF MEMBRANE CURVATURE BY MECHANO-CAPACITANCE
After spreading the membrane, its planarity was adjusted by varying the solution level in the cuvette, and membrane capacitance was measured. The first important point in the study of curvature-electric effects is to evaluate membrane curvature and to assure its constancy at each frequency applied, in order to study the frequency dependence of the effects. To this aim, we made use of the “capacitance microphone effect”.5 Briefly, a small voltage difference 'M (100 mV) was applied to the membrane and the displacement current due to the oscillations of the membrane capacitance was recorded. From the amplitude of the capacitance oscillation the curvature was evaluated. The radius of curvature, Rm 1 cm , was about 2 mm. This is a quite substantial curvature, if compared to the flat membrane radius.
FLEXOELECTRICITY IN BIOMEMBRANES
93
2.4. FLEXOELECTRIC RESPONSE OF BACTERIAL ETHANOLAMINE MEMBRANES
Boxcar-averaged traces of the curvature-electric currents and “capacitance microphone” currents of oscillating membranes demonstrate the doubled frequency of the capacitance response (recorded with 100 mV membrane voltage), the fundamental frequency of the flexoelectric response (recorded with zero membrane voltage) and the phase shift between them. The frequency characteristics of the flexoelectric effect for several membranes in 50 mM KCl are shown in Figure 3. Experimentally, the existence of two frequency regions (corresponding to free and blocked lipid exchange) is evident. The median frequency between them is about 150 Hz. IO (mV) 15
C˚ =2,2 nF PE 50mM KCI pH 6,22 C˚ =2,4 nF
10
5
PE + phloretin 0 0
200
400
600
800
1000 O(Hz)
Figure 3. Frequency dependence of the amplitude of flexoelectric effect. I Q represents the RMS value of the flexoelectric current, voltage-converted with 107 V/A amplification factor. BLM is formed from E. coli PE/n-decane. The electrolyte is unbuffered KCl, 50 mM, pH 6.22. Experimental points: (x) flat BLM capacitance C 0 = 2.4 nF; ({) C 0 = 2.2 nF; u, C 0 = 2.0 nF; (') phloretin-modified BLM: 40 nM phloretin in the bathing electrolyte. (From Petrov and Sokolov,9 with permission from the Publisher.)
In the low frequency region a slight increase of the response from 10 to 50 Hz is followed by a plateau up to 120 Hz and by a marked drop of the response with some membranes around 150 Hz. In the high frequency region a more or less linear growth of the response is observed. Extrapolation of the high-frequency straight line goes approximately to zero. Qualitatively, the shapes of the characteristics correspond to the theory. Phase measurements of capacitance currents from pre-curved membranes were performed with the aim of determining the sign of the flexoelectric coefficient in the two frequency regions (at 70 and 400 Hz). Both capacitance and
94
A.G. PETROV
flexoelectric currents should be in phase when curvature-induced polarization is directed the same way as the transmembrane electric field. For both frequencies this took place when the electric field pointed towards the centre of the membrane curvature. This means that the flexoelectric coefficient was negative in both frequency regions.9 In order to study the influence of the ionic strength, electrolyte solutions of non-buffered KCl 10 mM, 50 mM and 1 M were employed. In the low frequency range the only noticeable influence was upon the median frequency, which seemed to increase with decreasing ionic strength. In the high frequency region a marked correlation between the slope of the frequency characteristic (i.e., the value of the flexocoefficient) and the inverse ionic strength was found. The frequency dependence with 10 mM electrolyte was actually nonlinear, with gradually increasing slope. For comparison, the slope of 1 M curve was the lowest and the generated responses were strictly sinusoidal. The dipolar modifier phloretin is known to decrease the surface potential of lipid head dipoles by about 100 mV because of its oppositely directed dipole moment. With the symmetric addition of 40 Pl of 10 mM ethanol solution to 10 ml bathing membrane electrolyte before BLM formation, its amplitudefrequency characteristic demonstrated a drastic drop of the flexoresponse in the high frequency region and reduction of the overall slope by more than seven times (Figure 3). Unilateral addition to a pre-formed membrane did not produce any definite effect on the first harmonic, but led to the appearance of the second harmonic; this was clearly a capacitance microphone effect from the surface potential difference of the asymmetric membrane. Thus, the ability of this set-up to serve as a flexoelectric sensor for dipolar surface modifiers was demonstrated. 3. Flexoelectric Sensors Using Voltage Response of a BLM
Further experiments on membrane flexoelectricity were performed in the regime of voltage registration (open external circuit) by using a high impedance selective nanovoltmeter. These experiments provided a check of surface charge contribution in the presence of univalent or divalent ions.13 The approach was further elaborated by using stroboscopic interferometry for precise determination the of dynamically excited membrane curvature.14–17 3.1. LIPIDS AND MODIFIERS
BLMs were formed from chromatographically pure lecithin extracted from egg yolk (EYPC). The membrane-forming solution was made by dissolving lecithin in n-decane, in concentration 40 mg/ml. A drop of this solution was applied by
FLEXOELECTRICITY IN BIOMEMBRANES
95
a glass pipette to a 1 mm pretreated orifice in a Teflon plate, partitioning a Teflon chamber into two compartments. In some experiments mixed BLMs of lecithin with negatively charged phosphatidyl serine (PS, Serva, 20 mg/ml chloroform solution) were formed. The necessary amount of PS solution was dried on a piece of glass coverslip and then dipped in the membrane-forming solution EYPC/n-decane until completely dissolved. EYPC BLMs modified by uranyl acetate (UA, Merck) were prepared by adding different amounts of concentrated UA solution in water to the bathing electrolyte. As a bathing solution either unbuffered saline (0.1 M NaCl in distilled water) or a saline with phosphate buffer (pH 6.0) was used. Strong adsorption of UO22 ions on the bilayer surface is expected, thus resulting in a large positive surface charge depending on the UA concentration. The purity of EYPC was controlled by thin layer chromatography on Merck plates. Surface charge was estimated by measurements of the electrophoretic mobility of large multilamellar liposomes prepared after Bangham from EYPC in 0.1 M NaCl. Measurements were performed using Mark II (Rank Brothers) electrophoretic equipment. 3.2. EXPERIMENTAL SET-UP
The scheme of the experimental set-up is shown in Figure 4. A two-compartment Teflon chamber (after Ref.,5 with some modifications) was used. The front compartment was open to the air and had a glass window on its front face, enabling the observation of membrane formation and thinning in reflected light. The rear compartment was closed by a Teflon cap carrying the two reversible Ag/AgCl electrodes and a pipeline transmitting oscillating air pressure generated by a compression loudspeaker. A Teflon plate with a 1 mm cone-shaped orifice was pressed between the two compartments and used as a BLM support. Membrane planarity was controlled by Teflon leveling piston immersed in the front compartment and attached to a micrometer screw. It was also used to deliberately impose a membrane curvature for sign determination of the flexocoefficient. Membrane thinning was followed electrically by the membrane capacitance increase over time. Since the dynamics of membrane oscillations depends on D, the ratio of the torus elasticity to the sum of both BLM and torus elasticities the aim was to study BLMs with possible smaller tori. Such membranes displayed capacitances of more than 2.5 nF; therefore, BLMs of lesser capacitance were discarded. Flexoelectric potentials generated by the oscillating BLMs were registered by a lock-in nanovoltmeter (Figure 4).
96
A.G. PETROV
12 K1.1
11
5
6
7
1 K1.2
4
8
9 10
3
Figure 4. Scheme of the experimental set-up for voltage registration: (1) d.c. voltage source; (2) coupled switches (K1.1 and K1.2); (3) Teflon measuring cell; (4) levelling piston (5) plastic pipeline; (6) amplifier; (7) function generator; (8) selective nanovoltmeter Unitra 237; (9) lock-in nanovoltmeter Unitra 232; (10) oscilloscope, Textronics 2246A; (11) compression loudspeaker; (12) valve. (From Derzhanski et al.,13 with permission from the Publisher.)
3.3. FLEXOELECTRIC RESPONSE OF EGG YOLK LECITHIN MEMBRANES
From electrophoretic measurements of the surface charge of EYPC bilayers yield a negative sign of the surface charge and a rather small value of the electrophoretic charge density. We could estimate a value of partial charge per lipid head E = 0.4% for a freshly opened lecithin ampoule. A partial charge E = 4.6% was found for an older sample, three months post opening. The frequency dependence of the amplitude of the flexoelectric response of pure EYPC membranes is shown in Figure 5. The graph represents the data average over 21 membranes. One can see that above 160 Hz the response is essentially frequency-independent. From the plateau value of the response of each individual membrane we calculated the blocked flexocoefficient. The mean value from the 21 measurements is f = (26.5 r 5.5)10–19 C. The sign of this coefficient turned out to be positive (see below). The data from mixed EYPC + 2 mol% PS BLMs also are given in Figure 6. The points are averaged over four measurements. These measurements were complicated by the shorter lifetime of the BLM in the presence of PS, so a high frequency plateau was not revealed. However, the increased response by a charge modifier at higher frequencies is evident. The data for UA-modified EYPC membranes are shown in Figure 6 for various UA concentrations in the bathing electrolyte, ranging from 1 to 10 mM. In the upper part of this range BLMs were markedly stabilized and measurements could be extended up to higher frequencies. The results with 1 and 3 mM UA differ from those with 5 and 10 mM. In any case, modified BLMs give a higher
FLEXOELECTRICITY IN BIOMEMBRANES
97
UN / MV 2
60 50 40 1
30 20 10 0 40
120
200
280
360
440
N/Hz Figure 5. Frequency dependence of the flexoelectric voltage of BLMs. Curve 1: egg yolk lecithin/n-decane BLM (averaged data from 21 BLMs). Curve 2: egg yolk lecithin + 2% mol phosphatidyl serine/n-decane BLM (averaged data from 4 BLMs). Electrolyte: unbuffered 0.1 M NaCl. (From Derzhanski et al.,13 with permission from the Publisher.)
Figure 6. Frequency dependence of the flexoelectric voltage of BLMs modified by various concentrations of uranyl acetate (UA) in the electrolyte (0.1 M NaCl). Curve 1: no UA (same as curve 1, Figure 5). Curve 2: 1 mM UA. Curve 3: 3 mM UA. Curve 4: 5 mM UA. Curve 5: 10 mM. (From Derzhanski et al.,13 with permission from the Publisher.)
response than unmodified ones for all frequencies. From the plateau values above 300 Hz flexocoefficients f in the range of 12–20010–19 C were calculated, with a negative sign (see below); the minimum value was at 1 mM UA, the maximum at 3 mM. Finally, we describe the results for the sign of the flexocoefficient. Measurements were performed at 300 Hz and ' M = 40 mV (rear electrode positive).
98
A.G. PETROV
With EYPC membranes the comparison of the flexoelectric amplitude to that of the sum of the flexoelectric and capacitance amplitudes revealed an increase at switching ' M > 0 on and a positive (outward) initial curvature, or a decrease at switching ' M > 0 on and a negative (inward) initial curvature. This means that at positive static curvature both currents have the same phase, while at negative static curvature the phases are opposite. Since in the first case ' M ' C ! 0 , while in the second case ' M ' C 0 , from both experiments it follows that ' P ! 0 , i.e., f B ! 0 . This positive sign is opposite to the sign of the surface charge established electrophoretically. Repeating the experiment with the UA-modified BLM where recharging of the surface takes place, we see just the opposite behavior, showing that f B 0 in the whole UA concentration range studied. This negative sign is also opposite the positive surface charge due to the strong UO 22 ion adsorption. Another example of a flexoelectric sensor employing patch-clamped membranes of soy lecithin is provided.21 The effect of heavy metal ions Cd2+ and Hg2+ was investigated. While Cd2+ ions resulted in a biphasic effect, firstly increasing and then decreasing the flexoresponse, the effect of Hg2+ (Figure 7) amounted to a monotonic decrease of the flexoresponse in the millimolar range of concentrations.
Figure 7. Flexoelectric sensor effect of heavy metal ions. Hg2+ concentration dependence of the normalized flexocurrents of 4 soy lecithin membranes in patch pipettes excited at 410 Hz. (From Zheliaskova et al.,19 with permission from the Publisher.)
4. Conclusion
In conclusion, possible applications of flexoelectric BLMs as sensors for ion and dipolar species follows from the great sensitivity of flexoresponse to such adsorbed molecules (e.g. Figure 3). First prototypes of such sensors using BLM
FLEXOELECTRICITY IN BIOMEMBRANES
99
containing lipophilic ions as sensititvity amplifiers18 or patch-clamped bilayers for microscopic usage19 have already been demonstrated. The first observation of the converse flexoeffect in BLMs15,16 permits the potential use of stabilized BLM systems as microtransducers, micro sound wave generators, and microactuators in molecular electronics. Indeed, flexoelectrically induced displacements of a membrane surface that is only nano-meters thick represent (on a molecular scale) a huge, coordinated motion in space of a whole molecular assembly. This effect may, then, find some interesting applications. ACKNOWLEDGEMENTS
This work has been supported by a grant from the National Science Fund of Bulgaria (NP1-03/2004).
References 1. Mueller, P., Rudin, D., Ti Tien, H., and Wescott, W. (1962) Reconstitution of cell membrane structure in vitro and its transformation into an excitable system, Nature 194, 979–980. 2. Coronado, R., and Latorre, R. (1983) Phospholipid bilayers made from monolayers on patchclamp pipettes, Biophys. J. 43, 231–236. 3. Petrov, A. (1999) The Lyotropic State of Matter. Molecular Physics and Living Matter Physics, Gordon & Breach, New York. 4. Passechnik, V., and Sokolov, V. (1973) Permeability change of modified bimolecular phospholipid membranes accompanying periodical expansion, Biofizika 18, 655–660. 5. Ochs, A., and Burton, R. (1974) Electrical response to vibration of a lipid bilayer membrane, Biophys. J. 14, 473–489. 6. Petrov, A. (1975) Flexoelectric model of active transport, in J. Vassileva (ed.), Physical and Chemical Bases of Biological Information Transfer, Plenum, New York, pp. 111–125. 7. Petrov, A., and Derzhanski, A. (1976) On some problems in the theory of elastic and flexoelectric effects in bilayer lipid membranes and biomembranes, J. Phys. Suppl. 37, C3-155–C3-160. 8. Derzhanski, A., Petrov, A., and Pavloff, Y. (1981) Curvature induced conductive and displacement currents through lipid bilayers, J. Phys. Lett. 42, L-119–L-122. 9. Petrov, A., and Sokolov, V. (1986) Curvature-electric effect in black lipid membranes, Eur. Biophys. J. 13, 139–155. 10. Hristova, K., Bivas, I., and Derzhanski, A. (1992) Frequency dependence of the membrane flexoelectric voltage response. Adsorption of multivalent counterions on the surface of curved lipid bilayer, Mol. Cryst. Liq. Cryst. 215, 237–244. 11. Szekely, J., and Morash, B. (1980) The effect of temperature on capacitance changes in an oscillating model membrane, Biochim. Biophys. Acta 599, 73–80. 12. Wobschall, D. (1971) Bilayer membrane elasticity and dynamic response, J. Colloid Interface Sci. 36, 385–396.
100
A.G. PETROV
13. Derzhanski, A., Petrov, A., Todorov, A., and Hristova, K. (1990) Flexoelectricity of lipid bilayers, Liq. Cryst. 7, 439–449. 14. Todorov, A., Petrov, A., Brandt, M., and Fendler, J. (1991) Electrical and real-time stroboscopic interferometric measurements of bilayer lipid membrane flexoelectricity, Langmuir 7, 3127–3137. 15. Todorov, A., Petrov, A., and Fendler, J. (1994) Flexoelectricity of charged and dipolar BLM studied by stroboscopic interferometry, Langmuir 10, 2344–2350. 16. Todorov, A. (1993) Experimental investigations of direct and converse flexoelectric effect in bilayer lipid membranes, Ph.D. thesis, Syracuse University, NewYork. 17. Todorov, A., Petrov, A., and Fendler, J. (1994) First observation of the converse flexoelectric effect in bilayer lipid membranes, J. Phys. Chem. 98, 3076–3079. 18. Sun, K. (1997) Toward molecular mechanoelectric sensors: flexoelectric sensitivity of lipid bilayers to structure, location and orientation of bound amphiphilic ions, J. Phys. Chem. 101, 6327. 19. Zheliaskova, A., Naidenova, S., Marinov, Y., Mellor, I., Usherwood, P., and Petrov, A. (2001) Detection of heavy metal ions (Cd2+ and Hg2+) by their influence on flexoelectricity of patch clamped membranes, C.R. Acad. Bulg. Sci. 51(12), 53–56.
CARBON NANOTUBES: FROM FUNDAMENTAL NANOSCALE OBJECTS TOWARDS FUNCTIONAL NANOCOMPOSITES AND APPLICATIONS W. MASER*, A.M. BENITO, E. MUÑOZ, AND M. TERESA MARTÍNEZ Department of Nanotechnology, Instituto de Carboquímica (C.S.I.C.), C/Miguel Luesma Castán 4,E-50018 Zaragoza, SPAIN
Abstract – In this article we give a general introduction into the field of carbon nanotubes. On one side we describe carbon nanotubes as fundamental nanoscale objects and explain their high application potential. On the other side, we focus on carbon nanotubes as non-homogeneous materials as obtained from different sources. Methods for production, characterization, purification and functionalization and dispersions are presented explaining chances, challenges and limitations. Finally, we deal with the broad field of applications for carbon nanotubes with special emphasis onto high performance carbon nanotube composite materials.
Keywords: Carbon nanotubes, production, characterization, purification, functionalization, dispersions, composites, applications.
1. Introduction Carbon nanotubes1 were first described and brought into context by S. Iijima in 1991. It was quickly realized by the scientific community that his findings had wide ranging implications for science and technology. Driven by early theoretical studies, the very first experiments already confirmed the unique structure-property relationship. Since then, these findings boosted the development of a new and highly exciting field of research. Today, more than 15 years after Iijima’s observation, novel and often highly surprising results on carbon nanotubes are still continuously reported in the literature and in international ______ *
To whom correspondence should be addressed: W. Maser, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
101
102
W. MASER ET AL.
patent databases. The large number of highly promising results, quickly understood as real-world business opportunities, boosted by now the creation of numerous spin-off companies and also contributed to broaden-up core sectors of already well-established companies. With this article we provide the reader with the necessary background knowledge on carbon nanotubes to understand on one hand, their unique properties and opportunities, and on the other hand, the various challenges to be overcome in this broad and fascinating field of research. 2. Carbon Nanotubes: Fundamental Nanoscale Objects 2.1. STRUCTURE
A carbon nanotube (CNT) can be described as a seamlessly rolled-up sheet of graphene, resulting in an open tubular structure composed of carbon atoms arranged in a hexagonal network. The tubular structure is closed by the inclusion of six pentagonal defects into the hexagonal network at each end of the open cylinder to form the corresponding semi-fullerene2 caps. Figure 1 explains the basic structures of carbon nanotubes.
a1 Q
a2
tube axis
armchair A Q C B
zig-zag
A
B
chiral
C Figure 1. The structures of a graphene sheet (A), the family of single-wall carbon nanotubes (B), and a multi-wall carbon nanotube (C).
CARBON NANOTUBES
103
There are three different ways in which a graphene sheet can be rolled up onto itself thus leading to the three basic forms of CNTs, i.e. the family of single-wall carbon nanotubes (SWNTs): (i) Armchair tubes: These are obtained when rolling-up the sheet in such a way that carbon – carbon bounds are perpendicular to the cylinder axis. Looking at the cross-section of such a tube one observes an “armchair” – like structure of the carbon atoms. (ii) Zig-zag tubes: Here, carbon – carbon bounds are parallel to the tube axis and the cross-section results in a “zig-zag”-like pattern of carbon atoms. (iii) Chiral nanotubes: Here, the hexagons of the original graphene sheet are winding-up around tube axis in a helical way letting form the carbon atoms a helical (chiral) pattern along the length of the tube. In any case, the structure of the nanotubes easily can be described by so-called chiral vector C. It is composed of a two-dimensional pair of integer numbers (n, m) corresponding to the numbers of the hexagonal unit cell vectors a1 and a2, respectively, needed to roll-up the graphene sheet onto itself from on point to another one. Therefore, the (n, m) pair directly reveals the diameter as well as the chiral character for every type of tube, e.g. all armchair tubes are characterized by (n, n) indices. Typically, SWNTs have diameters of about 1 nm and lengths of about 1 Pm to several micrometers, according to experimental observations. Often, due to their small diameters, SWNTs like to arrange themselves into bundles which can adapt diameters even up to 100 nm. Additionally, there exists the family of multi-wall carbon nanotubes (MWNTs): They are composed of individual SWNTs concentrically placed inside each other and separated by an interlayer distance of about 0.34 nm slightly larger than in graphite due to curvature. While length and diameter of the most inner tubes are similar to SWNTs, the outer diameters, depending on the number of individual nanotubes easily can reach values up to 20 or 30 nm. Among the MWNTs, double-wall carbon nanotubes (DWNTs) may play an important intermediate role between SWNTs and MWNTs. 2.2. PROPERTIES
Carbon nanotubes are characterized by a very close and unique structure – property relationship. Composed out of only carbon atoms, carbon nanotubes are very light objects with a very low density of around 1.5 g/cm3. Having all atoms at the surface confers CNTs very high specific surface areas of about 1,400 m2/g (outer surface) or even 2,800 m2/g (if inner surface is considered as well), comparable or even better than highly activated carbons. Furthermore, carbon – carbon bounds are one of the strongest in nature. Therefore, very small individual SWNTs may have Young moduli up to 1,800 GPa (100 times stronger than steel), asymptotically reaching values of graphite for very large diameter
104
W. MASER ET AL.
tubes.3 For SWNTs arranged into bundles, due to sliding effects between individual SWNTs, the Young moduli may lower down to values of about 200 GPa. The maximum tensile strength can reach values of up to 30 GPa. Apart of the high stiffness, individual SWNTs show surprisingly high mechanically flexibility.4 Upon bending to very large angles, SWNTs will not break. Instead, they show a reversible buckling behaviour conferring SWNTs an extremely high toughness. Having micrometer dimensions along their axis and nanometersized diameters, CNTs are nano-scale objects with aspect ratios (ratio length to diameter) as high as 1,000. This is an important property whenever it comes to applications where percolation issues play a critical role. On the other side, depending of the point of view, individual SWNTs might be considered as a large molecule or already as a one-dimensional solid state material. The electronic properties5 easily can be derived applying density functional theory for graphene and zone-folding concepts taking into account additionally periodic boundary conditions for the wave function along the nanometer-sized circumference. These boundary conditions reduce the number of allowed states in the two-dimensional Brillouin zone and introduce one-dimensional van-Hove singularities in the density of states. It can be shown that one third of all SWNTs behave as metals, i.e. all (n, m) tubes with n-m = 3i (I = interger), e.g. all armchair tubes, while two thirds of all SWNTs are semiconductors, i.e. all (n, m) tubes with n-m 3i. It is really remarkable that the structural arrangement of carbon atoms in SWNTs so clearly defines the electronic properties and that these are precisely defined for each for each type of nanotube with a specific (n, m) index. In this sense it has been shown that for semi-conducting SWNTs of a given chirality, the energy gap is inversely proportional to the tube diameter.6–8 Additionally, the one-dimensional electronic character of SWNTs, as expressed by the one-dimensional van-Hove singularities in the density of states, provides these nano-scale objects fundamental electronic, optical and vibrational resonance properties of great importance to spectroscopic characterization, e.g. Raman,9 NIR-absorption,10 photoluminescence,11 and scanning tunnelling spectroscopy,12 as well for applications in the fields of quality control and sensors. Furthermore, experimentally it has been found that SWNTs show ballistic transport13 even at room temperature, have current densities14 as high as 1010 A/cm2 (copper and aluminium show values between 107 and 1010 A/cm2), and behave as excellent electron emitters15,16 with low turn-on fields of 1.5–5 V/ȝm at 1 mA/cm2 and low energy spread of 0.25 eV. Finally, it is worthwhile mentioning that CNTs possess an excellent thermal conductivity17 of up to 3,000 W/mK, superior of diamond, one of the best thermal heat conductors. With this unusual close structure-property relationship and a whole bunch of fascinating combined properties, – light, strong, flexible,
CARBON NANOTUBES
105
metal, semiconductor, heat-conductor, electron emitter –, it becomes clear that CNTs bear a great potential for applications in various sectors of technological interest as shown in Figure 2. The fields range from nanoelectronics (CNT based electronic device structures), flexible plastic electronics (photovoltaic and organic light emitting devices), functional composite materials (reinforced structures, tough materials, thermal conductivity, charge dissipation, electromagnetic shielding, intelligent textiles, and coatings, functional adhesives), energy (electrochemical energy storage devices, fuel-cell membranes), nanobio (sensors, drug delivery, cell proliferation, tissue healing), and catalysis (nanocatalyst dispersions). For a more detailed overview on structure and general properties we refer to the book of Dresselhaus.18
Figure 2. An overview of carbon nanotube’s broad application potential.
3. Carbon Nanotubes: Challenging Materials Before profiting from carbon nanotube’s unique properties and using them for various purposes, one has to be aware that carbon nanotubes are not produced as the individual objects described above. Instead, they are obtained as soot materials which are composed of CNTs of all different kinds of characteristics as well as of additional undesired carbonaceous and inorganic byproducts. This is quite challenging since no CNT soot material is like the other and no standard CNT materials exist today. The characteristics of CNT materials largely depends on the production technique being employed: Materials may contain different types of CNTs (SWNTs, MWNTs, DWNTs, …) with different
106
W. MASER ET AL.
Figure 3. Different types of CNT materials: (a) MWNTs from arc, (b) SWNTs from arc/laser, (c) thin MWNT bundles from supported CVD, (d) aligned MWNTs from floating CVD, (e) filled MWNTs from floating CVD, (f) DWNTs from supported CVD, (g) helical tubes from supported CVD.
diameter and chirality distributions, various defect structures, bundle agglomerations, and impurities such as amorphous carbon, graphite particles and attached catalytic nanoparticles. Figure 3 puts in evidence the different CNT materials characteristics. 3.1. PRODUCTION METHODS
To understand better the obtained materials a brief introduction into the most common production methods are given. Basically, one distinguishes between high-temperature and low-temperature processes.19 In high temperature processes, a solid carbon feedstock (e.g. graphite and eventually catalytic particles) is evaporated at temperatures above 3,000ºC by resistive heating or highly concentrated light. In an adequate inert atmosphere carbon species in the gas phase self-assemble to form carbon nanotubes. Electric arc discharge processes20 and laser evaporation methods21,22 are representative for this process. The soot materials obtained are: (i) straight and highly graphitic MWNTs (Øi = 1–3 nm, Øo = 2–35 nm, l = 1 ȝm) with additional polyhedral graphitic nanoparticles but without any catalytic particle (electric arc), (ii) entangled bundles of SWNTs (Øbundle = 20–100 nm, ØSWNT = 1–2 nm, l = 1 ȝm) with amorphous carbon and embedded spherical catalytic nanoparticle, typical Ni, Co, Ni/Y (electric arc, laser evaporation). Low temperature processes are based on the decomposition of a hydrocarbon feedstock at temperatures between 500ºC and 1,000ºC over catalytic particles. Carbon diffuses into the catalytic particle and at lower temperatures precipitates, under the right experimental conditions, in the form of carbon
CARBON NANOTUBES
107
nanotubes. Chemical vapor deposition (CVD) techniques are typical methods being used.19,23–29 Depending on the production parameters (type of hydrocarbon gas, flow conditions, temperature, catalysts, catalyst introduction, support materials, additives etc.) usually various different types of MWNTs are obtained in bended form, as a direct consequence of low temperature processes leading to the inclusion of structural defects. More or less homogeneous materials are composed of thick and thin MWNTs, different numbers of layers, more or less bended, organised in bundles, or as individual objects, in various yields, with various amounts of inorganic impurities (between 2 and 40 wt %) from catalyst (even as filling inside CNTs) and catalyst support as well as low or high amounts of amorphous carbon. Sometimes, soot materials with enhanced fractions of SWNTs25 and DWNTs26 are obtained. CNTs also can be grown in aligned or in patterned form.27–29 A variation is the so-called HiPCO method (high pressure disproportionation of carbon monoxide) developed by Rice University30 and commercialized by CNI company which results in SWNT materials (ØSWNT = 2 nm, l = several micrometers) without any amount of amorphous carbon but with considerable amounts of catalytic iron nanoparticles (up to 40 wt % in as produced materials). Producing an almost “monolithic” CNT material composed only of CNTs of well defined diameters, chiralities, number of walls, without any by-products and in large quantities remains the greatest challenge in this field. Today, as a general rule of thumb one can say that today’s low volume production methods (few gram-scale) result in relatively high quality materials in what concerns CNT characteristics while the more recent large scale – low price production methods (tons-scale, 100 Euro/kg) do not aim yet at welldefined materials characteristics but on the availability of CNTs on industrial scale and on applications based on the overall CNT materials characteristics rather than on the individual CNT properties. 3.2. CHARACTERIZATION METHODS
Depending on the further use and application in mind, one type of CNT material might be more suitable than another one. Therefore it is highly recommendable to collect as much information as possible on structure and properties of the used CNT materials. In the following useful standard lab-techniques and their information content respective to bulk CNT materials, as produced, are briefly described. Scanning electron microscopy (SEM): A few milligrams of powder CNT material is deposited on the SEM sample holder. Eventually, depending on the conductivity degree, a very thin conducting metal layer is sputtered on. This technique generally gives highly valuable structural information on a scale of 100 nm to a few millimeters. Although CNTs can not be directly observed, it
108
W. MASER ET AL.
reveals CNT materials morphology, the presence and organization of the different sample products, density, homogeneity, fibre lengths. It also gives hints on the conductivity state (charging and contrast effects). SEM is a rapid, efficient and relatively economic technique for analysing CNT bulk material. Transmission electron microscopy (TEM): CNT materials are suspended in ethanol and ultrasonicated. A drop is deposited on a TEM grid. Solvent is evaporated and the sample can be analysed whenever the deposited material density is transparent for the electron beam. This technique gives structural information from the sub-nanometer to the 100 μm range. Diameter, layer structure, degree of graphitization, defects, bending and inorganic impurities can be observed. Combined with element mapping, selected area electron diffraction and EELS information on type, crystallization and distribution of catalytic nanoparticles as well as CNT chirality can be obtained. TEM gives valuable information on the level of individual CNTs, however, since only a very small (transparent) volume is probed good statistics is needed to get representative results. Induced coupled plasma spectroscopy (ICPS) and elemental analysis: A few tens of milligrams of powder CNT material is dissolved in acids and the precipitate analysed by flame-spectroscopy. It is a completely quantitative technique and reveals the content of metals, as well of other elements such as oxygen, sulphur and halogens in the bulk sample. ICPS allows the calculation of the composition of a sample and also might reveal ideas on the production yield. Thermogravimetric Analysis (TGA): A few tens of milligrams of CNT powders (or pellets) are deposited in a suspended crucible whose weight-loss is measured as a function of temperature increase. TGA gives information on the oxidation temperatures and residual mass. It may reveal CNT materials composition (presence of different carbon materials) and the amount of catalyst material in the sample. TGA is a valuable qualitative and comparative technique. Very easily different types of CNT materials can be distinguished from each other and information on further treatment temperatures can be obtained. If used for qualitative analysis, a strict protocol has to be employed (oxidation temperatures and metal rest content strongly depend on various experimental parameters. Slow temperature ramps between 3–5ºC/min, low gas flows are recommended). Furthermore, various error sources have to be taken into account (buoyancy effects, sample weight, sample morphology, “fluffy”sample or pellet effect, spontaneous combustion). Powder X-ray diffraction (XRD. About 100 mg of sample material is deposited on a flat-bed sample holder. XRD reveals the sample content, i.e. the type of carbon CNT, Camorph, Graphite, the degree of graphitization, of bundling
CARBON NANOTUBES
109
and alignment, as well as the presence of catalysts or other elements. Easily different CNT samples can be distinguished from each other. Raman spectroscopy: a few milligrams of CNT powder material is placed on a sample holder. Four characteristic features can be obtained: low frequency CNT radial breathing modes (RBM) around 100–200 cm–1 (frequencies are inversely proportional to SWNT diameters), tangential shear mode (G-line) at about 1,580–1,590 cm–1, defect mode (D-line) at about 1,350–1,450 cm–1 and double resonant G’ mode at around 2,700–2,800 cm–1. Therefore, Raman gives information on the carbon structures and the quality of the CNT material. It easily allows to distinguish between different types of CNTs (SWNTs, MWNTs, DWNTs) and reveals the presence of carbonaceous defects and of graphite. Diameters of SWNTs and their distribution as well as the degree of bundling and alignment can be calculated. The conducting character as well as doping and charge transfer effects may be revealed. Due to the one-dimensional electronic structure and the presence of van-Hove singularities, resonance effects upon tuning the laser excitation frequency are observed. Resonance Raman spectroscopy is used to reveal the (n, m) structure of individual SWNTs. Raman spectroscopy is one of the most powerful standard characterization technique for CNTs with a high information content on the sample characteristics and on individual objects.9 The homogeneity degree of the samples has to be probed by mapping different sample areas. Only limitation might be eventual sample degradation effects when using elevated powers which, on the other hand, reveal information about CNT’s thermal stability. Four-point conductivity: Pellets or films are contacted using pressure contacts (most easy approach). This technique establishes the conducting behaviour of bulk CNT networks. It is a good comparative approach to distinguish between MWNTs, SWNTs or modified CNTs. However, absolute values for conductivity strongly depend on the exact calculation of the geometry factor as well as the protocol for pellet pressing and contacting the sample. Nitrogen adsorption isotherms (BET): Bulk technique using a few tens of milligrams of CNT powder (or pellets or films) deposited in the corresponding sample holder. BET gives information on the gas-adsorption process (physisorption or chemisorption), gas kinetics, CNT pore size and specific pore-size area. Different types of CNTs easily can be distinguished. The above mentioned set of techniques to characterize bulk CNT materials reveals sufficient information for further materials processing and also may give first hints on the suitability of CNT materials for certain kinds of applications. Most of them are use in a qualitative and comparative way. It still remains critically distinguishing between different carbon components and quantifying their respective content in the CNT material. An approach to
110
W. MASER ET AL.
solve this problem may by the use of NIR spectroscopy11 carried out in diluted CNT suspensions. 3.3. PURIFICATION AND TREATMENT
As discussed above, CNT materials are not pure and thus there is great interest in improving the samples’ quality for many purposes ranging from the development of proper characterization standards to further use and applications. Improvements refer to: (i) removal of metal particles, (ii) removal of carbonaceous non-CNT contributions, (iii) separation of bundle structures, (iv) removal defective CNTs (v) separation according to length, diameter and chirality. Main difficulty here is that CNTs (and the whole CNT materials) are chemically quite inert and non-soluble. Therefore, purification and treatment methods, in a first step, focus onto the removal of metals and Camorph, and the improvement of CNT structure by oxidation approaches, as shown in Figure 4.
Figure 4. A typical purification scheme for CNT materials.
Typically two types of treatments can be distinguished: On one hand, chemical oxidation in liquid phase using acids (e.g. HNO3,31,32 HCl,31,33 HCl/HNO3,31 HF/HNO3,34 H2SO4/HNO3), under reflux and, on the other hand, thermal treatments, such as air-oxidation processes at temperatures between 300–450ºC. Chemical oxidation focuses onto the removal of metal nanoparticles in the CNT materials. The efficiency strongly depends on the concentration and concentration ratio of the used acids, the reaction time, and the amount of CNTs. Thermal treatments are directed towards the removal of amorphous carbon as well as the creation of pore-structures (activation process). The efficiency depends on the temperature applied (determined beforehand by TGA), the gas-flow, the amount of CNT materials and the macroscopic sample distribution, i.e. the exposed materials surface to the heat and gas-flow. If necessary, both processes can complement each other, even in several repetitive cycles.
CARBON NANOTUBES
111
Typically, a final annealing step is employed in order to heal structural CNT defects, and improve the bundle recrystallization in case of SWNTs. There exists a wide range of recipes which strongly depend on the CNT materials used and on the further use of the material. The following general observations can be made32: Chemical treatments lead to a significant reduction of metal particles up to 95 wt %. This is accompanied by significant material losses (20–60 wt %) including non-metal material losses (5–10 wt%) due to attacked carbon materials. Additionally, basic functional groups, especially carboxylic groups are attached to defect sites in attacked CNTs and also intercalation effects may be observed. The obtained material is highly compacted with drastically reduced specific surface areas of about 2 m2/g and nearly untreatable. Airoxidation does not affect the metal content, but removes defective carbon material, resulting in further material losses and in a relative increase of the metal content. On the other hand, it also results in highly activated materials with drastically increased specific surface areas as high as 750 m2/g. It is to mention that also this process creates defects at the tips and sidewall of CNTs to which carboxylic groups may be attached. The final annealing step contributes to heal defect structures, to recover the bundle organization, to remove created carboxylic groups and intercalated molecules. As one can see, purifycation is a multi-step approach which can be efficiently controlled using the above mentioned bulk techniques in order to judge the materials quality after each step. However, due to serious material losses, one has to find a compromise between resulting CNT quality, process yields, and final materials needs. 3.4. FUNCTIONALIZATION AND DISPERSIONS
Due to the high chemical inertness of CNTs and their insoluble character, chemistry on carbon nanotubes came later into the game and only created increased interest when it became clear that simple pre-oxidation steps as taken out in purification treatments lead to the creation of structural defects in CNTs (tip-opening and side-wall) to which various oxygen containing groups (mainly carboxyl groups) are attached as a function of the strength of the oxidation process (acid strength, and oxidation time) as illustrated in Figure 5. The oxidatively introduced carboxyl groups represent useful sites for further modifications, as they enable the covalent coupling of molecules through the creation of amide and ester bonds. By this method the nanotubes can be provided with a wide range of functional moieties such as dendrimers, nucleic acids, enzymes, metal complexes, nanoparticles etc. Furthermore, the presence of (modified) carboxyl groups leads to a reduction of van der Waals interactions between the CNTs, which strongly facilitates the separation of nanotube bundles
112
W. MASER ET AL.
Figure 5. A nanotube after oxidative treatment with defect sites at tip (open) and at side-walls to which oxidative functional groups (mainly carboxyl groups) are attached.
into individual tubes. Additionally, the attachment of suitable groups renders CNTs soluble in aqueous or organic solvents, opening the possibility of further modifications through subsequent solution-based chemistry. In addition to this “defect” functionalization, direct side-wall addition reactions through thermally activated reactions (preferentially on more reactive small diameter tubes) are documented in the literature. The functionalization degree is low and may vary between 3% and 15%. Fluorination35 and cyclo-addition36 are one of the most prominent side-wall reaction which further can be used for additional substitution reactions with alcohols, amines, Grignard reagents, alkyl lithium compounds and other specific linker molecules.37 Beside covalent functionalization (defect and side-wall addition), noncovalent functionalization is becoming of increased importance. It is based on the highly delocalized ʌ-electron system of CNTs to which many long-chain molecules and polymers have a strong affinity and get non-covalently linked to the CNT surface via van der Waals forces or ʌ-ʌ interactions. Surfactants such as triton X-100 or sodium dodecyl sulfate (SDS) lead to the formation of micelles and help to separate and disperse CNTs.38 Long-chain molecules/ polymers, such as polyelectrolytes,39 peptides,40 lipids,41 DNA,42 ʌ-conjugated polymers43–45 may even fully wrap around the CNTs by undergoing corresponding conformational changes to adapt to the underlying CNT network. These noncovalent approaches bear several advantages: (i) CNT structures are not disturbed and thus the original CNT properties are maintained, (ii) they render CNTs soluble in various solvents and thus is the base for the development of stable homogeneous CNT dispersions and for spectroscopic characterization, (iii) they facilitate selective SWNT separation according to diameter, chirality, length, and conductivity via adequately designed molecules, (iv) they add
CARBON NANOTUBES
113
further functionality to CNTs, (v) they improve interface interactions with matrix materials for the development of high-performance composite materials. Both, covalent functionalization and, even more promising, non-covalent supramolecular functionalization are the base for the development of stable and homogeneous dispersions. This is a very important step towards CNT homogenization, separation, improved solution characterization (NIR,10 photoluminescence,11 dynamic light scattering46 and optical spectroscopy), solution processing of CNTs as a whole material (the deposition and incorporation techniques, transformation technology (buckypaper and fibre fabrication)) and achieving compatibility and synergetic interactions with many molecules and polymer systems. Therefore, stable and homogeneous dispersions are considered as a key-stone to transfer carbon nanotube’s high application potential into realworld applications. 4. Carbon Nanotube Applications With their bunch of fascinating properties carbon nanotubes are of great interest to technological applications in several fields, as already indicated in Section 2.2. In the following, we will not group applications into different sectors, but instead will classify the various applications with respect to the appearance of CNTs in a certain form. This means we distinguish between applications based on individual CNTs, on assemblies of CNTs and on CNT-hybrid materials, such as CNT based composites. This type of classification will allow to better understand the chances and challenges common to CNTs in a certain form of appearance. Hereby, special emphasis will be put onto the field high performance CNT composite materials. 4.1. INDIVIDUAL CARBON NANOTUBES
An individual carbon nanotube with a well defined structure and thus, electronic properties is a highly promising object for the development of nanoelectronic devices. A metal type SWNT can be imagined as a metal wire of great use as vertical interconnector in classical semi-conducting devices. On the other hand, a semi-conducting SWNT can be seen as important logic element in field effect transistor device (FET). It has been demonstrated by several groups around the world that semi-conducting SWNTs can be incorporated successfully as active element into classical silicon-based FET devices. Here they have proven (superior) operational functionality as logic gate,47,48 as gas or bio sensing devices,49,50 or as nanoelectromechanical device.51 Key issues towards comercialization of these products mainly relate to time-costly assembly technologies, which directly is related to the lack of pure SWNT materials composed of only
114
W. MASER ET AL.
one type of SWNTs. Today, three assembling approaches are followed: (i) selection of an individual SWNT of a specific electronic characteristics out of a bulk SWNT material (usually requires coupled AFM, Raman, SEM technology steps), placing the selected SWNT onto a pre-fabricated electronic structure using a nanomanipulator (AFM-based technology), contacting (usually lithographic steps). (ii) dropping a diluted SWNT dispersion onto a prefabricated electronic structure, checking if SWNTs got deposited onto (between) the right contact structures, probing the electronic characteristics of these SWNTs (requires combined AFM (SPM) and Raman techniques), contacting (lithography steps). The limitations for both techniques refer to exhaustive selection and characterization processes for having the right SWNT at the right place. Alternatively, a third approach (iii) is to grow individual SWNTs in CVD processes directly in prefabricated channels of semi-conducting device structures into which catalyst seed particles are incorporated.52 Since no control of CNT characteristics during growth can be achieved yet, these device structures show only functionality at random places. Finally, it remains to mention the dream for purely nanotube-based electronic devices achieved when connecting SWNTs of different specific characteristics. While theoretically easily possible, again, limitations refer to efficient SWNT selection and assembly technologies. Further progress strongly relates to improved SWNT production technology as well as to efficient chemical separation and reliable assembly techniques. 4.2. ASSEMBLIES OF CARBON NANOTUBES
There are several electronic applications which are based on a more statistical approach using as grown assemblies of CNTs. Most prominent example are flat panel devices making use of the electron emission effect of CNTs. As mentioned in Section 3.2, CNTs can be grown in a pixel-like pattern on a conducting substrate.27–29 The micrometer-sized pixel areas contain millions of parallel aligned MWNTs. Placing a top-electrode and applying an electric field, statistically sufficient electron emission can be obtained (although some individual CNTs may fail) to fully satisfy conditions for building commercial flatscreen devices,53 and other devices based on field emission, such as X-Ray cathodes,16 and SEM/TEM tips.54 Aligned grown assemblies of CNTs also have been used for the development of various kinds of filter membranes55 for hydrocarbon separation and waste-water treatment. As a next step, CNT dispersions can be used to obtain statistical networks of CNTs. This can result in micrometer thick CNT networks (buckypapers) or even in ultra thin and even transparent networks of either conducting or semiconducting CNTs which can be coated onto flexible transparent substrates
CARBON NANOTUBES
115
resulting in transparent (vis-IR) and conducting CNT film coatings.56 These further can be assembled into flexible and transparent thin film transistors57 maintaining their electrical performance even upon strong bending. CNT networks (properly functionalized) also serve as transducing material for gas58,59 and bio60 sensing applications as well as biological scaffold materials for cell proliferation.61 CNT dispersions also are used to produce CNT fibre materials by spinning technologies as base for the development of intelligent textiles.62,63 Incorporated into textiles they operate as electrochemical element using the proper sweat of the human body. They may act as supercapacitors (storage of electrochemical energy), detect molecules (sensors), change color (electrochromic device), actuate (close and open pores) and absorb shock-energy. 4.3. CARBON NANOTUBE BASED COMPOSITE MATERIALS
The development of high performance nanocomposite materials is one of the most promising fields for CNT applications impacting on a broad range of technological sectors. The basic idea is to incorporate nanotubes into a matrix material and transfer their unique properties to the host system resulting in a material with enhanced functional, structural and processing properties. Essentially two types of matrix materials are considered: Polymeric and ceramic materials. Since only few works have been reported in the field of ceramics, in the following we will focus only on the class of polymer-CNT composites. Here, most research has been performed in the fields of thermoplastics and thermosets (both of great technological relevance in daily applications) as well in the area of electroactive polymers (of increasing importance for the development of plastic electronics). Independent of the matrix system, there are two key issues of paramount importance to obtain highly functional CNT-based composite materials: (i) Homogeneous CNT dispersion in the host system contributing to achieve maximum stable separation of CNTs and to avoid CNT agglomerations thus resulting in a homogeneous composite material. (ii) Efficient transfer of CNT properties to the host matrix. The development of nanocomposites clearly is an issue at the interface where the formation of a proper interface layer acting as link between CNT and the rest of the host material determines the degree of CNT property transfer and thus the final characteristics of the whole composite material. The interface layer simply may lead to a reorganized polymer structure in the vicinity of the CNTs,64,65 enough to observe already some first significant changes in the overall materials characteristics, or, in the best cases, even allow highly favorable CNT property transfer leading to completely new materials characteristics.45 The development of a favorable interface layer impacting on the whole materials characteristics
116
W. MASER ET AL.
directly is linked to a high degree of CNT dispersion. Both, CNT dispersion and load-transfer are highly complex issues strongly depending on the type of CNT material and the host matrix system, both defining the way towards successful compounding. Latest at this point, it should become clear why a thorough knowledge on all aspects around CNTs, as mentioned in the former chapters is critical towards the development of functional CNT-based nanocomposites. Incorporation of nanotubes into plastics can potentially provide light structural materials with dramatically increased modulus and strength. Thermoplastics, such as polypropylene,66 polystyrene,67 as well as thermosets,68 such as epoxies are widely used for this purpose. While for low CNT loadings, typically in the region between 0.1 and 2 wt %, moderate improvements, due to the development of a proper interface layer, are observed (elastic moduli improvements in the range of a factor or 3–4, increase of thermal conductivity by a factor of 1%), at higher loading rates, agglomeration takes place and the developed materials get highly inhomogeneous. This indicates that far more work is needed to achieve mechanical properties closer to the ones of the proper CNTs. On the other hand, due to CNTs high aspect ratio and conducting (electrical and thermal) properties, conducting filler networks can be obtained already at very low percolation values. Ultra-low percolation thresholds in the range 0.001 and 0.01 wt % of CNT filler ratio (a factor of 1,000 loser than for spherical carbon black particles) have been observed to reach conductivity values around 0.1–1 S/cm.69 Therefore, the formation of percolated CNT networks in different kinds of polymer matrices for the development of lightweight materials able to dissipate electrostatic charge, to shield electromagnetic radiation, and to dissipate heat, is of increased technological interest for commercial applications. Finally, it has been shown that CNTs may interact in a very favorable way with electroactive polymer matrix systems, such as polyaniline, polythiophene, polypyrrole and respective derivatives. Their highly conjugated backbone structure with a highly delocalized ʌ-electron system is highly compatible with the CNTs extended ʌ-electron system leading to composite materials with drastically improved characteristics of special interest to optoelectronic applications. Composites of CNT/poly(p-phenylenevinylene) (CNT/PPV),70–72 CNT/poly(3,4-ethylenedioxythiophene) (CNT-PEDOT)73 have proven their functionality in light-emitting diodes and photovoltaic cells, CNT/polyaniline (CNT(PANI) have been used as printable conductors for thermal-imaging techniques.74 Especially when in-situ polymerization processes are carried out, i.e. polymerization in the presence of CNTs, as shown for CNT/PANI,45,75 highly synergetic effects can be obtained ranging from increased conductivity, thermal stability, deprotonation stability, optical activity, combined with improved processing characteristics. All these are highly promising results towards the development of future plastic electronic devices including improved flexible
CARBON NANOTUBES
117
organic light emitting devices (OLEDs), photovoltaic cells, circuits, and sensing devices. Additionally, these electroactive polymer/CNT composites are of great interest to the development of energy storage devices, such as supercapacitors.76 5. Conclusions and Outlook In this article we have explained the unique properties of carbon nanotubes and shown that these are fundamental nanoscale objects of interdisciplinary, intersectorial interest and of great educative value. They are impacting on broad fields of science and technology. We have shown that carbon nanotubes as a material is highly complex and bears several challenging key issues ranging from proper characterization and purification to functionalization and the development of homogeneous stable dispersions and highly functional CNT-composite materials. While carbon nanotubes certainly will proof advantageous for certain applications, other ones probably will not develop at all. In any case, more than 15 years-time of research in this field has taught us that carbon nantoubes are highly enabling materials and good for many further surprises. ACKNOWLEDGEMENTS
The authors gratefully acknowledge funding from Regional Government of Aragon (DGA) under its “Group of Excellence” programme (DGA-T66). W.M. gratefully acknowledges funding from Spanish Ministry of Education and Science (MEC) and European Regional Development Fund (ERDF) under project NANOPOLICOND (MAT 2006-13167-C02-02). E.M. and A.M.B gratefully acknowledge support by MEC and ERDF under project SENAGAS (TEC200405098-C02-02). A.M.B. and M.T.M. gratefully acknowledge funding from DGA under Priority Research Line Projects (PIPO21/2005 and PIPO15/2005).
References 1. Iijima, S. (1991), Nature 354, 56–58. 2. Kroto, H., Heath, J., O’Brien S., et al. (1985), Nature 318, 162–163. 3. Mintmire, J., Robertson, D., and White, C. (1993), Phys. Chem. Solids 54(12), 1835– 1840. 4. Yakobson, B., Brabec, C., and Bernholc, J. (1996), Phys. Rev. Lett. 76, 2511–2514. 5. Charlier, J., Blasé, X., and Roche, S. (2007), Rev. Mod. Phys. 79, 677–731. 6. Mintmire, J., and White, C. (1995), Carbon 33, 89–902. 7. Hamada, N., and Sawada, S. (1992), Phys. Rev. Lett. 68, 1579–1581.
118
W. MASER ET AL.
8. Saito, R., Fujita, M., Dresselhaus, G., and Dresselhaus, M. (1992), Appl. Phys. Lett. 60, 2204–2206. 9. Dresselhaus, M., Dresselhaus, G., Saito, R., et al. (2005), Phys. Rep. Rev. Sect. Phys. Lett. 409, 47–99. 10. Itkis, M., Perea, D., Niyogi, S., et al. (2003), Nano Lett. 3, 309–314. 11. O’Connell, M., Bachilo, S., Huffman, C., et al. (2002), Science 293, 593–597. 12. Wildoer, J., Venema, L., Rinzler, A., Smalley, R., and Dekker, C. (1998), Nature 391, 59–62. 13. White, C., and Todorov, T. (2001), Nature 411, 649–651. 14. Frank, S., Poncharal, P., Wang, Z., and de Heer, W. (1998), Science 280, 1744–1746. 15. Zhu, Z., Bower, C., Zhou, O., Kochanski, G., and Jin, S. (1999), Appl. Phys. Lett. 75, 873–875. 16. Yue, G., Qiu, Q., Gao, B., et al. (2002), Appl. Phys. Lett. 81, 355–357. 17. Ruoff, R., and Lorents, D. (1995), Carbon 33, 925–930. 18. Dresselhaus, M., Dresselhaus, G., and Avouris, P. (Eds.) (2001), Topics in Applied Physics, vol. 80, Springer, Berlin, Germany. 19. Benito, A., Maser, W., and Martínez, M. (2005), Int. J. Nanotechnol. 2, 71–89. 20. Journet, C., Maser, W., Bernier, P., et al. (1997), Nature 388, 756–758. 21. Thess, A., Lee, R., Nikolaev, P., et al. (1997), Science 273, 483–487. 22. Maser, W., Muñoz, E., Benito, A., et al. (1998), Chem. Phys. Lett. 292, 587–593. 23. Pérez-Mendoza, M., Vallés, C., Maser, W., Martínez, M., and Benito, A. (2005), Nanotechnology 16, S224–S229. 24. Vallés, C., Pérez-Mendoza, M., Castell, P., et al. (2006), Nanotechnology 17, 4292– 4299. 25. Colomer, J., Stephan, C., Lefrant, S., et al. (2000), Chem. Phys. Lett. 317, 83–89. 26. Vallés, C., Pérez-Mendoza, M., Martínez, M., Maser, W., and Benito, A. (2007), Diamond Relat. Mater. C 16, 1087–1090. 27. Fan, S., Chapline, M., Franklin, N., et al. (1999), Science 283, 512–514. 28. Hata, K., Futaba, D., Mizuno, K., et al. (2004), Science 306, 1362–1364. 29. Terrado, E., Redrado, M., Muñoz, E., Maser, W., Benito, A., and Martinez, M. (2006), Diamond Relat. Mater. 15, 1059–1063. 30. Nikolaev, P. (2004), J. Nanosci. Nanotechnol. 4, 307–316. 31. Chiang, I., Brinson, B., Smalley, R., et al. (2001), Phys. Chem. B 105, 1157–1161. 32. Martínez, M., Callejas, M., Benito, A., et al. (2003), Carbon 41, 2247–2256. 33. Jeong, T., Kim, W., and Hahn, Y. (2001), Chem. Phys. Lett. 344, 18–22. 34. Chattopadhyay, D., Galeska, I., and Papadimitrakopoulos, F. (2002), Carbon 40, 985–988. 35. Bettinger, H. (2003), Chem. Phys. Chem. 4, 1283–1289. 36 Tagmatarchis, N., and Prato, M. (2004), J. Mater. Chem. 14, 437–439. 37. Khabashesku, V., Billuops, W., and Margrave, J. (2002), Acc. Chem. Res. 35, 1087– 1095. 38. Islam, M., Rojas, E., Bergey, D., et al. (2003), Nano Lett. 3, 269–273. 39. Correa-Duarte, M., and Liz-Marzán, L. (2006), J. Mater. Chem. 16, 22–25. 40. Dieckmann, G., Dalton, A., Johnson, P., et al. (2003), J. Am. Chem. Soc. 125, 1770.
CARBON NANOTUBES
119
41. Qiao, R., and Ke, P. (2006), J. Am. Chem. Soc. 128, 13656–13657. 42. Heller, D., Jeng, E., Yeung, T., et al. (2006), Science 311, 508–511. 43. Panhuis, M., Maiti, A., Dalton, A., et al. (2003), J. Phys. Chem. B 107, 478–482. 44. Star, A., Stoddart, J., Steuermann, D., et al. (2001), Angew. Chem., Int. Ed. 40, 1721–1725. 45. Sainz, R., Benito, A., Martínez, M., et al. (2005), Adv. Mater. 17, 278–281. 46. Badaire, S., Poulin, P., Maugey, M., and Zakri, C. (2004), Langmuir 20, 10367– 10370. 47. Tans, S., Verschueren, A., and Dekker, C. (1998), Nature 393(6680), 49–52. 48. Avouris, P. (2004), MRS Bull. 29, 403–410. 49. Kong, J., Franklin, N., Zhou, C., et al. (2000), Science 287, 622–625. 50. Star, A., Gabriel, J., Bradley, K., et al. (2003), Nano Lett. 3, 459–463. 51. Bourlon, B., Glattli, D., Miko, C., Forró, L., and Bachtold, A. (2004), Nano Lett. 4, 709–712. 52. Duesberg, G., Graham, A., Liebau, M., et al. (2003), Nano Lett 3, 257–259. 53. Choi, Y., Cho, Y., Kang, J., et al. (2003), Appl. Phys. Lett. 82, 3565–3567. 54. de Jonge, N., and van Druten, N. (2003), Ultramicroscopy 95, 85–91. 55. Srivastava, A., et al. (2004), Nat. Mater. 3, 610. 56. Wu, X., Chen, Z., Du, X., et al. (2004), Science 305, 1273–1276. 57. Cao, Q., Hur, S.-H., Zhu, Z.-T., et al. (2006), Adv. Mater 18, 304–309. 58. Sippel-Oakley, J., Wang, H., Kang, B., et al. (2005), Nanotechnology 16, 2218–2221. 59. Sayago, I., Terrado, E., Lafuente, E., et al. (2005), Synth. Met. 148, 15–19. 60. Liu, G., and Lin, Y. (2006), J. Nanosci. Nanotechnol. 6, 948–953. 61. Zanello, L., Zhao, B., Hu, H., and Haddon, R. (2006), Nano Lett. 6, 562–567. 62. Dalton, A., Collins, S., Muñoz, E., et al. (2003), Nature 423, 703. 63. Bauhgman, R., Zakhidov, A., and de Heer, W. (2002), Science 297, 787–792. 64. Sandler, J., Broza, G., Nolte, M., et al. (2003), J. Macromol. Sci.-Phys. B 42, 479–488. 65. Chen, W., and Tao, X. (2005), Macromol. Rapid Commun. 26, 1763–1767. 66. Castell, P., Maser, W., Benito, A., et al. (2007), Polymer (submitted). 67. Qian, D., Dickey, E., Andrews, R., et al. (2000), Appl. Phys. Lett. 76, 2868–2870. 68. Biercuk, M., Llaguno, M., Radosavljecvic, M., et al. (2002), Appl. Phys. Lett. 80, 2767–2769. 69. Sandler, J., Kirk, J., Kinloch, I., et al. (2003), Polymer 44, 5893–5899. 70. Curran, S., Ajayan, P., Blau, W., et al. (1998), Adv. Mater. 10, 1091–1093. 71. Coleman, J., Curran, S., Dalton, A., et al. (1998), Phys. Rev. B 58, R7492–R7495. 72. Star, A., Stoddart, J., Steurman, D., et al. (2001), Angew. Chem. Int. Ed. 40, 1721–1725. 73. Woo, H., Czerw, R., Webster, S., et al. (2001), Synth. Met. 116, 369–372. 74. Blanchet, G., Fincher, C., and Gao, F. (2003), Appl. Phys. Lett. 82, 1290–1292. 75. Sainz, R., Small, W., Young, N., et al. (2006), Macromolecules 39, 7324–7332. 76. Picó, F., Rojo, J., Sanjuán, M., et al. (2004), J. Electrochem. Soc. 151, A831–A837.
ULTRASHORT PULSE PLD: A TECHNIQUE FOR NANOFILM FABRICATION T. SZÖRÉNYI1* AND Zs. GERETOVSZKY2 1 Research Group on Laser Physics, Hungarian Academy of Sciences, University of Szeged, PO Box 406, H-6701 Szeged, HUNGARY 2 Department of Optics and Quantum Electronics, University of Szeged, PO Box 406, H-6701 Szeged, HUNGARY
Abstract – In the present contribution the peculiarities of laser ablation are discussed with special emphasis on the differences in the mechanisms of nanoparticle formation when ablating materials with pulses of nanosecond vs. femtosecond duration. In the case of ablation using nanosecond pulses the dominating species leaving the target surface are principally atoms and ions. Cluster formation and growth mainly take place, via nucleation and condensation, from the plasma plume within the surroundings. The principal control parameter is the ambient pressure. When the major goal is not the production of colloids (either in form of an aerosol or a sol) but layer growth instead, nanostructured films can be made at pressures higher than a few pascals. On the other hand, ablation with ultrashort pulses produces a plasma plume of biphasic character: its leading edge, consisting of ionic and atomic components is followed by a spatially and temporally well separated cloud of nanoparticles. In this case nanoparticle formation is a direct consequence of the interaction of the ultrashort laser pulse with the target material. This process even works in high vacuum, which provides an additional proof for that here, contrary to nanosecond-ablation, those are the laser parameters that control the characteristics of the nanoparticles produced.
Keywords: Ablation, laser processing, clusters, nanoparticles, thin films
______ *
To whom correspondence should be addressed: T. Szörényi, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
121
122
T. SZÖRÉNYI AND Z. GERETOVSZKY
1. Pulsed Laser Ablation: General Characteristics In laser ablation the energy of the laser pulse is absorbed within a small volume of the surface layer of the target material and converted into heat almost instantaneously. The fast and confined heating builds up extreme pressures, induces phase transitions and leads to material ejection from the heated volume. While the technique is conceptually simple, the strongly nonlinear and far-fromequilibrium nature of the involved processes make it rather complex. Though being clearly an over-simplification, ablation is often envisaged as a sequence of steps, starting with the interaction of laser radiation with the target, followed by the absorption of energy and localised heating within a thin surface layer, phase transitions, and subsequent (very much forward directed) material ejection in form of a plume, which either expands into vacuum or gas, or may even interact with a liquid environment. Depending on the duration of the laser pulse and the nature of the environment, the properties of the components of the ablation plume may further change, due to collisions within the plume and/or between the constituents of both the plume and the ambient, in certain cases even through laser-plume interactions.1,2 Historically, production of nanoparticles by laser ablation dates back to the mid-1980s, when the first papers on fullerenes were published.3 For about 2 decades nanosecond lasers remained the only tools of nanoparticle production. In the third Millennium ultrashort pulse lasers moved to the forefront of laser ablation research, as well. Nevertheless, it became evident only very recently, that nanosecond and femtosecond ablation proceed along fundamentally different routes.4 In this contribution the peculiarities of nanoparticle production by laser ablation will be reviewed by following the life story of species ejected from the target while focusing on the differences between nanosecond vs. femtosecond processing. Special emphasis will be given to that particular setup in which the species are collected on a substrate, i.e. when we grow a layer. First a brief overview of the techniques available for in- and ex-situ analysis of the properties of the ablated species and their interaction within and outside the plume will be given. Then the characteristics of nanoparticle formation will be revealed by analyzing the results of selected case studies. 2. Analytical Toolbox Techniques – The Life Story of the Ablated Species The range of applicable analytical techniques is determined by the ambient the ablated species expand into. In vacuum or in a low pressure gas ambient in-situ plasma spectroscopies offer the most direct way to identify and characterize the just born species and their interaction with the surroundings during plume
ULTRASHORT PULSE PLD
123
expansion. Gated intensified CCD photography is routinely used to record the time evolution of the spatial distribution of the plume. Differentiation between various plume components is possible by working in carefully chosen spectral domains. Once the spatial and temporal evolutions of the plume have been identified via imaging, optical emission spectroscopy can be used to identify and quantify species in their excited states in different locations within the plume. On the other hand, laser-induced fluorescence or optical absorption spectroscopy can probe the “dark” part of the plume, i.e. measure unexited atoms and molecules, along with small clusters. Scattering can also be used to reveal the presence of NPs. Rayleigh scattering imaging is effective for particles larger than just 2–10 nm, so it can be used to study nanoparticle growth. Blackbody radiation can also be imaged, or spectroscopically analysed to reveal the presence and temperature of NPs. Time-of-flight techniques are also routinely applied for spaceand time-resolved analysis of the dynamics of plasma species. In vacuum or relatively low pressures (up to several hundred pascals) the plume components can be collected for (in- and ex-situ) analyses and/or layers can be grown by immersing a substrate into the expanding plume. By following the process of film growth in-situ or by the ex-situ characterization of film properties, information on the characteristics of the ablation products can be derived. It is worth noting, that this technique offers a relatively simple, yet extremely versatile method of thin film production, also known as Pulsed Laser Deposition, PLD. Due to the popularity of PLD, gaining insight into the behaviour of plasma species by means of indirect film-based techniques has a well studied basis. Contrary to the plethora of techniques available in low pressure environments the analytical toolbox is rather empty in denser media. The main reason of that stems from the reduced mean free path at high pressures or in liquid ambients. In certain cases characterization of detonation waves, cavitation bubble formation by shadowgraphy does work, but in the general case the range of analytical techniques reduces to ex-situ characterization of the ablation products. Colloidal systems formed in liquids or in gas atmospheres at pressures above several hundred pascals can be analyzed using the standard techniques of nanoscience/nanotechnology. While stabilizing a nano-dispersed phase in an aerosol is rather problematic, gases offer the cleanest environment and therefore allow for the fabrication of nanoparticles of the highest purity. On the other hand, ablation into a liquid has the advantage of providing an easy means to hinder agglomeration and preserving the original size distribution of the individual nanoparticles via protecting their surface.
124
T. SZÖRÉNYI AND Z. GERETOVSZKY
3. Ablation with Nanosecond Pulses 3.1. SINGLE-WALL CARBON NANOTUBE SYNTHESIS BY LASER VAPORIZATION
Synthesis of carbon nanotubes by laser ablation is worth mentioning not only because it was the first method used to produce fullerenes in the gas phase,3 but also because this technique remained one of the best methods to grow high-quality, high-purity single wall carbon nanotubes, SWNT. A single laser pulse (of typically nanosecond duration) is sufficient to produce a vapor cloud of carbon and catalyst atoms, from which, inside a hot oven and in inert gas atmosphere, the species self-assemble to form SWNTs. The yield is high and the process is relatively insensitive to the diameter of the metal catalyst nanoparticle.5–7 SWNT can also be grown at process temperatures well below 1,000–1,100ºC, even at RT, where high-repetition-rate8 or long-pulse9 lasers provide the sufficient heating of the target and the plasma species. A necessary prerequisite of exploiting the full potential of carbon nanotube growth by laser evaporation is the understanding of the process of SWNT formation. The potential of spectroscopies in following the time evolution of the plasma processes has convincingly been demonstrated by the spatial and temporal mapping of SWNT synthesis in a series of papers from Dave Geohegan’s group at ORNL.10–12 From the results of these studies the timeline of the process has been constructed which depicts, in Figure 1, the birth of laser generated SWNTs. The plume initially consists of atomic and molecular species. Condensation of carbon starts within 0.2 ms after ablation. The small C clusters begin to aggregate and form swirling vortex rings appearing as smoke in the gas flow. As time passes the plasma temperature continuously decreases and the excited metal atoms relax into their ground states. In the next millisecond, also the ground state metal atoms condense, and so by t | 2 ms the plume becomes almost entirely composed of clusters, trapped within swirling smoke rings. This mixture of carbon clusters and metal catalyst nanoparticles forms the feedstock for nanotube formation. The growth is however a relatively long process. After 20 ms, only SWNTs of about 200 nm length are found. Growth up to 10 μm lengths occurs over much longer times – from 100 ms to s – indicating growth rates of 0.5–5 Pm/s. While the conclusion of general interest of these time-resolved growth studies is that SWNTs grow through the conversion of condensed-phase carbon clusters and metal catalyst nanoparticles, in the context of our analysis which aims to compare nanosecond ablation to femtosecond one, the main message is that the clusters, acting as the feedstock of SWNT growth, are formed from atomic species of the plasma by condensation in an inert ambient.
ULTRASHORT PULSE PLD
125
Figure 1. The stages of SWNT formation from laser generated plasma as a function of time and plume temperature.11
3.2. THE ROLE OF PRESSURE IN DETERMINING GAS PHASE PROCESSES; FORMATION OF CARBON NANOCLUSTER FILMS
When the plume expands into a gas, the fate of its species is determined by the frequency and the nature of collisions with the surrounding atoms/molecules. Up to approximately 1 Pa the behaviour of the plume is essentially vacuumlike. With increasing background pressure the frequency of collisions increases, which reduces the initially high kinetic energy of the species. At sufficiently high pressures the expansion eventually stops, and the species become thermalized.12,13 Fast imaging techniques provide insights into the details of these events.14–16 In Figure 2 the effect of confinement is exemplified by comparing the position and shape of carbon plasma plumes in vacuum and in 20.5 Pa Ar. This latter pressure lies already within the pressure domain where the ablated atoms and ions have slowed down so much that cluster formation becomes possible. Now, the size of the clusters is jointly controlled by the laser parameters and the actual pressure. This pressure control of the kinetic energy of plasma species materializes in profound changes in the morphology of films built by collecting the ablated material on substrates.17–19 As an example, Figure 3 shows the evolution of the surface morphology of films deposited by ablating a graphite target with 6 J cm2 pulses of a KrF excimer laser (15 ns, 248 nm) in argon at different background pressures.18 At 0.67 Pa hard, scratch resistant films of smooth morphology are produced. As the Ar pressure increases to 5.32 Pa the films become much
126
T. SZÖRÉNYI AND Z. GERETOVSZKY
Figure 2. Wavelength filtered ICCD images of a carbon plume expanding into vacuum (a)–(c), and into 20.5 Pa Ar (d)–(f). Twenty nanoseconds gate with a delay after the laser pulse of 120 ns for images in the first column, 250 ns for the second column and 400 ns for the third column. The length of each frame is 30 mm.16
Figure 3. SEM images of carbon films deposited at Ar pressures of (a) 0.67, (b) 5.32, (c) 13.3 and (d) 45.2 Pa. Note the different scale bars.18
rougher due to nanoscale clustering. The cluster size derived from the SEM picture is approximately 15 ± 3 nm. At 13.3 Pa clustering becomes more pronounced, and leads to clusters of §45 nm in size. These layers can be easily scratched off, indicating much reduced hardness. By 45.2 Pa the film morphology changes from nanoscale clusters to more filamentary growth together with a further reduction in film density. The appearance of the material resembles very much that of the low density diamond-like carbon nanofoam produced by high repetition rate Nd:YAG laser in argon atmospheres of pressures exceeding §13 Pa.20 When decorated by surface-enhanced Raman-active components,
ULTRASHORT PULSE PLD
127
e.g. metal nanoparticles, such large surface area carbon nanoscaffolds may serve as efficient, low-cost sensor platforms.21 The evolution of film morphology with pressure is very similar in different materials systems, though with different threshold values depending on the material ablated and the nature of the ambient gas used.22 Since the changes in growth mechanism materialize in the variation of all film properties, from the proper analysis of these, information on the underlying mechanisms can also be derived. The comparative analysis of the growth characteristics of carbon and carbon nitride films fabricated by ablating identical graphite targets with an ArF excimer laser (Ȝ = 193 nm) in the same configuration in Ar and N2 atmospheres, respectively, is another example for such a film-based approach.23 The pressure dependence of the apparent growth rates, defined as measured film thickness per number of pulses, is shown in Figure 4. Up to approximately 0.5 Pa the nature of the atmosphere has no effect on the growth rate. It is still in line with the expectations, that in Ar the growth rate drops between 0.5 and 5 Pa. The unexpected increase between 5 and 100 Pa, however, pinpoints a change in the growth mechanism. When ablating in N2, the domain where the increase of pressure has apparently no influence on the growth rate extends up to approximately 50 Pa, suggesting contribution of more than one process to film growth. In the changes in the apparent growth rate both the variation of the number of the constituting atoms and the changes in film microstructure manifest themselves. The two contributions can be separated by recording the change in the number of constituting atoms deposited over unit film area per pulse, within the same pressure domain (Figure 5). A close-up of the change in carbon growth rate in Ar atmosphere reveals that the decrease in the apparent growth rate between 0.5 and 5 Pa, shown in Figure 4, is a consequence of the decrease in the number of carbon atoms reaching the substrate due to collisions with background gas particles. However, the threefold decrease in the number of film building atoms from 0.6 to 0.2 × 1015 atoms cm–2 pulse–1 is only accompanied by an approximately twofold decrease in the apparent growth rate, suggesting a concomitant decrease in the compactness of the films at higher pressures. In the 5–100 Pa domain the carbon growth rate keeps decreasing, while the apparent growth rate increases, revealing that the decrease in film density accelerates. In nitrogen atmospheres, up to approximately 5 Pa, the incorporation of nitrogen nearly compensates for the loss in carbon deposition, resulting in a practically constant apparent growth rate. When exceeding 5 Pa this trend discontinues. The number of carbon atoms incorporated into the films keeps decreasing while that of nitrogen atoms starts to decrease, as well. While the
T. SZÖRÉNYI AND Z. GERETOVSZKY
128
dependence of carbon and nitrogen deposition rates on N2 pressure accounts for the practically constant N/C ~ 0.35 ratios measured in the 5–50 Pa domain,24 it can not any further account for the unchanged apparent growth rate (cf. Figure 4).
-1
Apparent growth rate [nm pulse ]
0.06
0.05
Figure 4. The apparent growth rate of carbon (Ƒ) and carbon nitride (Ŷ) films as a function of pressure, fabricated by ablating identical graphite targets in Ar and N2 atmospheres, respectively.
0.04
0.03
0.02 -2
-1
0
1
2
0.6 0.5
Figure 5. The deposition rate of carbon (Ƒ and Ŷ) and nitrogen (ż) atoms in carbon (Ƒ) and carbon nitride (Ŷ and ż) films, as a function of pressure.
0.4 0.3
Growth rate [10
15
atoms cm-2 pulse-1]
lg (Pressure [Pa])
0.2 0.1 0.0 -2
-1
0
1
2
lg (Pressure [Pa])
3.0
2.6
Figure 6. The density of carbon (Ƒ) and carbon nitride (Ŷ and ż) films as a function of pressure.
-3
Mass density [gcm ]
2.8
2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 -2
-1
0
lg (Pressure [Pa])
1
2
ULTRASHORT PULSE PLD
129
The comparative analysis reveals that the pressure dependence of the absolute number of film constituting atoms (Figure 5) alone can not explain the variations in the apparent growth rates (Figure 4). The mass density vs. ambient pressure plots, derived by dividing the sum of the mass of all constituting atoms over unit area by the measured thickness, shown in Figure 6, display similar dependences for both carbon and carbon nitride, and indicate profound structural changes, indeed. In particular, these results confirm that the macroscopic density of carbon nitride films starts to decrease when the composition exceeds N/C ~ 0.2.25–28 The sudden decrease in mass density above 5 Pa explains why practically no change has been recorded in the apparent deposition rate in the 5–50 Pa domain (Figure 4) in spite of the continuous decrease in both C and N arrival rates (Figure 5). Atomic force microscope17,29,30 and TEM31 studies (Figure 7) verified, that the decrease in mass density of both DLC and carbon nitride films was a direct consequence of increased surface roughness and porosity.
Figure 7. TEM micrographs taken on films deposited with 10 J cm–2 pulses in 5 Pa N2 (left) and with 7.5 J cm–2 pulses in 50 Pa N2 (right). Each picture is 1 × 1 ȝm2 in size.31
The above results confirm that in the particular case of carbon and carbon nitride film growth the critical pressure is at around 5 Pa, indeed. Cluster growth starts at pressures above 5 Pa due to gas-phase collisions of carbon atoms/ions. The thermalized species build a layer of loose, porous network when reaching a room temperature substrate. The corollary of these and any other studies alike is that the primary products of nanosecond-ablation are atoms and ions. Cluster formation is the consequence of the interaction of the expanding plume and the ambient gas species. In inert atmospheres, the contribution of atomic vs. cluster growth to film formation is controlled by pure mechanics, i.e. by the number of collisions.
T. SZÖRÉNYI AND Z. GERETOVSZKY
130
Increasing pressure promotes cluster formation. When ablating in reactive atmospheres, a chemical effect should also be added to the kinetics: the collisions may result in formation of species composed of components of both the plasma and the atmosphere. 3.3. A UNIQUE APPROACH: BACKWARD DEPOSITION OF LASER GENERATED NANOCLUSTERS
As the mean free path of plasma species at pressures higher than a few pascals falls below a centimetre, in the absence of a forced gas flow the nanoclusters formed remain in the vicinity of the source, i.e. close to the ablation spot. A genuine approach is the collection of backward-propagating clusters on substrates placed in the target plane near to the ablated area.32,33 By tuning the laser parameters and the nature and pressure of the surrounding gas, nanoparticles of a great variety of sizes can be produced. The size distribution of nanoclusters deposited by ablating a Si wafer in He (Figure 8) illustrates well the potential of this geometry.
1
2
3
4
5
6
Number of clusters
1J/cm2
Figure 8. Size distribution of Si nanoclusters deposited in He atmosphere at 532 Pa at various energy densities of the ablating ArF laser pulses.32
1.4 J/cm2
3 J/cm2
3.9 J/cm2
1
2
3
4
5
6
Size (nm)
3.4. NANOPARTICLE PRODUCTION BY ABLATING UNDER LIQUIDS
Nanoparticles intended for chemical- and bio-applications in liquid environments should (i) be directly formed in the particular liquid most appropriate for the planned application, (ii) possess a relatively narrow size distribution, (iii) be free of surface impurities, and (iv) have reactive chemical groups on their
ULTRASHORT PULSE PLD
131
surfaces to simplify further attachment of (bio)molecules, if necessary.34 Backed by 100 years of experience colloid chemistry can produce a wide spectrum of nanoparticles, partly meeting these requirements. The peculiarity of ablation is that performed in purified and filtered deionized water a great variety of ultrapure nanomaterials can be produced, and after being in the solution, surface engineering of the nanoparticles becomes possible. Since systematic studies on the physical aspects of ablation in liquid environment have only been carried out for about 7 years, the full potential and the limitations of this technique is far from being explored. Technical problems like the formation of relatively long-living (300–500 ȝs) cavitation bubbles, absorption of laser light by and reflection on the nanoparticles in the liquid, self-focusing (time dependent) inhomogeneities in the refractive index limit the possibilities of following the events in-situ and make the analysis and description of the complicated processes involved rather hard. The state-of-the-art is the following: When ablating with nanosecond pulses in chemically inert liquids (e.g. in pure water), the size of the nanoparticles produced is relatively large, since coagulation and aggregation can hardly be overcome. In practice strongly dispersed particles of 10–300 nm mean diameter have been obtained with limited control by the laser parameters.34 When ablating in chemically active solutions chemistry helps to stabilize the primary nanoparticles, which leads to smaller mean diameters and narrower size distributions, as compared to the inert liquid case. Significant progress has been achieved by the use of aqueous solutions of surfactants. The surfactant covers the just born NPs and prevents them from agglomeration. Sodium dodecyl sulfate (SDS) proved to be one of the most efficient surfactants to stabilize gold and silver nanoparticles at around 5 and 12 nm mean sizes (Figure 9), respectively, in best cases.35–37 3.5. NANOPARTICLES BY NANOSECOND-ABLATION – SUMMARY
When ablating with pulses of nanosecond duration the material leaves the target in form of atoms, ions, and to a much less extent as small clusters. To generate nanoparticles, an environment of relatively high pressure is a prerequisite in order to create an efficient confinement of the plasma and increase its density. The formation and the evolution of the properties of the nanoparticles are essentially determined by the interaction of ejected (atomic) species and the environment. When ablating in gases, the key parameter is the pressure. Here the material can also be collected on substrates, leading to the formation of nanostructured films. Ablation in a liquid ambient results in the formation of colloidal nanoparticle solutions, but small sizes are available only when further chemical treatments are applied.
132
T. SZÖRÉNYI AND Z. GERETOVSZKY
Figure 9. Electron micrograph and the derived size distribution of silver nanoparticles produced by ablation with the second harmonic (532 nm) of a Nd:YAG laser in 0.05 M SDS aqueous solution.35
4. Ablation with Ultrashort Pulses 4.1. INTRODUCTION
In the past decade, femtosecond lasers emerged as unique tools allowing processing of practically any material with unprecedented precision. The advantages of ultrashort pulse processing over conventional laser machining with pulses of nanosecond duration originate in the 4–6 orders of magnitude shorter timescale of energy deposition. The energy injected into the material within picoseconds is absorbed instantaneously by the electrons of a very thin surface layer (skin depth), while the transfer of energy to the lattice takes longer time, typically tens of picoseconds. Although the details of the interaction are different for metals and dielectrics, the net results are the same: clean removal of a small, but well defined volume of material with minimal thermal load to the surrounding material. Historically, motivated by the interest of industrial applications which required high-precision patterning, most efforts have been devoted to the optimization of material removal with much less attention to the properties of the material removed.38 As a consequence, the potentials of femtosecond ablation in nanoparticle production remained hidden until very recently. Ultrashort pulse lasers produce a hot, highly ionized plasma plume that expands in a well-defined fashion and serves as an excellent source for thin film deposition. The plume does not contain micron-sized particulates and droplets,
ULTRASHORT PULSE PLD
133
which is often the case when ablating with nanosecond lasers, enabling thereby the production of particulate-free films. The high intensity pulses produce plumes with large concentrations of ionic species with high kinetic energies (>100 eV) which has been considered beneficial e.g. in the production of high quality diamond-like carbon films.39 However, from the data available to date, one may conclude that ablation with high intensity femtosecond pulses does not necessarily lead to improvement in film properties Instead of striving to achieve maximum intensities, ablation with pulses of lower intensities, conveniently attainable with commercial femtosecond systems, seems to be a more pragmatic approach.40 Moreover, recent results revealed that plasmas generated in low pressure environments by ultrashort pulses having intensities not far above ablation threshold were ideal sources of nanoparticles.41–46 Following the same outline we used to present the peculiarities of nanosecond ablation, in this second part we will highlight those features of the ultrashort ablation process which differ most from that of the nanosecond case. Our approach will again address both the characteristics of the plume, and the properties of the ablation products. This will be achieved by summarizing the results of selected case studies, rather than trying to give a full coverage of all aspects of ablation with ultrashort pulse lasers. 4.2. ABLATION IN VACUUM: PLUME PROPERTIES
In a series of papers published in the last few years S. Amoruso and coworkers convincingly demonstrated that – contrary to the nanosecond case where the material leaves the target surface in a single package – in the plasma generated by ultrashort pulses of intensities within the 1011–1013 W cm–2 domain three populations could be identified, each of which is characterized by different expansion dynamics.41–43 Snapshots of a silicon plume and especially the intensity profiles derived by integrating the emission along directions parallel to the target surface, and shown on the right of the images (Figure 10), clearly prove the presence of three populations in the expanding plasma. From the distances the two fast components traveled in 30 ns a decent estimate can be given to their expansion velocities, while for the determination of the velocity of the third, slow component, which remains very close to the target surface on this timescale, delays of the order of microseconds are necessary (Figure 11). The expansion velocities of the three populations derived from these images are ~107, ~106 and ~5 × 104 cms–1, respectively. Spectral analysis has shown that the fastest population consists of ions, while the second component of the plasma contains both neutral and ionized species. At time delays below 0.5 ȝs the spectra, shown here for the particular
134
T. SZÖRÉNYI AND Z. GERETOVSZKY
Figure 10. The early stage of expansion of a silicon plume produced by ablation with 0.9 ps @1,055 nm pulses of 8.4 × 1011 W cm–2 intensity into vacuum. Gate time: 10 ns.41
Figure 11. Evolution of the slow component of a silicon plume produced by ablation with 0.9 ps @1,055 nm pulses of 8.4 × 1011 W cm-2 intensity into vacuum. Gate times : 1 ȝs at 5.5 ȝs and 5 ȝs at 22.5 ȝs delays, respectively.41
case of Au (Figure 12), are dominated by emissions from atoms. At longer delays, however, structureless, broad black-body spectra appear, indicating the arrival of hot nanoparticles. Detailed analysis of the spectra yields initial temperatures in the order of few thousands Kelvin, continuously decreasing with time due to radiative cooling.42,44,45 The comparison of results obtained on ablating Si, Ni, Fe, Au, Ag, Ti and TiC with pulses of different durations (80–900 fs) and intensities (1011–2 × 1013 W cm–2) suggests that the behaviour of the plasma, outlined above, represent a common, general feature of femtosecond-ablation, at least within the parameter window examined.41–46
ULTRASHORT PULSE PLD
135
Figure 12. Time evolution of the emission spectra of a gold plume recorded 1 mm above the target surface. Ablation with 120 fs pulses of 5 × 1012 W cm–2 intensity of a Ti:sapphire laser @780 in vacuum.42
4.3. NANOPARTICLE FABRICATION IN VACUUM
Similar to the behaviour of the plasma, the size and the size distribution of the nanoparticles produced from different materials are akin, as well. AFM analysis of nanoparticle films of Ag, Au, Ni, Si and Fe of less than a monolayer coverage yielded mean nanoparticle radii ranging from 8 to 25 nm with standard deviations between 5 and 20 nm, as illustrated for the case of Ag in Figure 13.
Figure 13. Left: AFM image of nanoparticles deposited on mica by ablating a silver target with 120 fs pulses of 5.0 × 1012 W cm–2 intensity in high vacuum. Right: size histogram.42
136
T. SZÖRÉNYI AND Z. GERETOVSZKY
Size control of nanoparticles is possible by tuning the laser parameters. An increase in intensity within the 1012–1013 W cm–2 range leads to a slight increase in the mean radii.42 The relatively broad size distribution, characteristic of ablation in the visible, can be narrowed by using UV wavelengths.47 The increase of pulse duration up until the electron-phonon relaxation time, IJe-ph of the material to be ablated has apparently no significant effect on the ablation products. There was very little difference in the morphology of Ag (IJe-ph ~20 ps) and Ni (IJe-ph ~7 ps) films deposited in high vacuum using 10 ps and 200 fs pulses.48 In both cases the films comprised of almost spherical nanoparticles of 50–300 nm in diameter, with apparently more aggregation when ablating with 10 ps pulses. Nevertheless, more experimental work is required to completely characterise the ablation and deposition processes at different pulse durations. In the light of potential future applications it is important to note, that synthesis of nanocrystals of different materials is also possible with this technique. An example: The analysis of X-ray diffraction patterns of Ni and Fe nanofilms indicated that the films consisted of crystallites with sizes of few tens of nanometers, whereas AFM imaging gave exactly the same nanoparticle sizes, suggesting that each and every particle has the same single-crystalline domain, i.e. the films are composed of randomly oriented, but identical nanocrystals.49 The analysis of the magnetic behaviour of Ni nanoparticle films led to the conclusion that the film behaved as a system of isolated magnetic particles, i.e. the individual nanocrystals of 40 ± 19 nm size approached fairly well the ideal, single-domain behaviour.50 4.4. PLUME EXPANSION IN A REACTIVE GAS. A POSSIBLE ROUTE TO COATED NANOPARTICLE FABRICATION
The differences in the mechanisms and the products of ablation performed with pulses of nanosecond vs. femtosecond duration have, to the first glimpse, surprising consequences when performing the experiments in reactive atomspheres. The synthesis of carbon nitride is a good example. The prediction of the unique properties of the ȕ-C3N4 phase by Liu and Cohen in the late 1980s51 initiated tremendous efforts to synthesize this “magic” compound. By ablating carbonaceous targets in nitrogen containing atmospheres carbon nitrides, CNx with × 0.7, i.e. well below its nominal 57 atm % N content could only be obtained.52,53 There were speculations that when producing more energetic carbon species the efficiency of CN formation could be promoted. The experiments performed using femtosecond lasers had puzzling results: Films with maximum nitrogen contents less than 15 atm %, i.e. much lower than those fabricated using nanosecond pulses, could be produced.54,55
ULTRASHORT PULSE PLD
137
Recalling that compound formation materializes as a result of collisions of the carbon plasma with the N2 molecules, it is plausible that the highest nitrogen content can be reached when single C atoms/ions meet N2 molecules or (excited) nitrogen atoms. The interaction of C clusters with nitrogen results in a compound with smaller N content. Meeting of carbon nanoparticles and nitrogen species is the worst case scenario in terms of carbon nitride chemistry. The differences in the plasmas produced by nanosecond and femtosecond pulses therefore explain the apparently surprising result. While the ablation of carbonaceous targets with femtosecond pulses can not produce species rich in N, it may well be that their nitrogen content originates from the outer surface of the nanoparticles. If that is the case, ablation by femtosecond pulses in reactive atmospheres could be a dedicated technique for the production of core-shell nanoparticles. 4.5. SOLID NANOPARTICLE FILM FROM A LIQUID TARGET
Very recently Szörényi and coworkers fabricated Si-doped amorphous carbon nanofilms by ablating a commercial silicone oil with extremely clean 700 fs @248 nm pulses of 4–5 × 1011 W cm–2 intensity in high vacuum.56 AFM and SEM unambiguously proved that the surface of the films, deposited between RT and 250ºC, consisted of nanoparticles (Figure 14). The formation of nanoparticles suggests that the ablation mechanism of a liquid target might be very similar to that of a solid one. The results of a parallel TEM study confirmed that on >100 nm-scale the films were morphologically inhomogeneous, indeed (left panel of Figure 15).
Figure 14. a-C:Si nanoparticle films deposited onto silicon substrate at RT (left) and 200ºC (right) by ablating silicone oil with 700 fs @248 nm pulses of 12.2 and 14.7 mJ total energies, respectively, focused onto 4.1 mm2 areas.
138
T. SZÖRÉNYI AND Z. GERETOVSZKY
Figure 15. Top-view HRTEM images of a free standing a-C:Si film. The blurred contrast is due to defocus and noise filtering.
Even at the highest magnification used (right panel of Figure 15) no ordering could be detected, revealing that at the nanoscale the material is homogeneous and isotropic, possessing a perfectly amorphous structure. 4.6. ABLATION IN VACUUM. EXPERIMENT VERSUS THEORY
The results analysed above suggest that ablation with ultrashort laser pulses is characterized by a number of general features, independently of the nature of the target material. In particular, ablation at intensities not much higher than the plasma formation threshold, more specifically within the 3 × 1011 – 2 × 1013 W cm–2 domain, inevitably leads to the generation of nanoparticles of the target material. This characteristic seems to be true not only for elemental, but also for multicomponent (e.g. binary or even more complex) materials. These general features are consistent with the physical picture of the process, as outlined by the predictions of recent theoretical analyses.57,58 At laser intensities in the range of 1012–1013 W cm–2, corresponding to initial temperatures of a few electron volts, the (nearly) adiabatic cooling drives the material into a metastable phase, and results in the production of a relatively large fraction of nanoparticles through phase decomposition processes. On the other hand, at larger intensities (>1014 W cm–2) the material can never reach the metastable phase, resulting in an almost fully atomized plume. Thus the most promising laser intensity range in the context of nanoparticle production is theoretically predicted to be 1012–1013 W cm–2. Though further comparative studies are inevitably necessary for judging how general this statement could
ULTRASHORT PULSE PLD
139
be, the intensity values predicted by the theoretical approaches turn out to be in very good quantitative agreement with the limited experimental works’ that led to an efficient generation of nanoparticles. 4.7. NANOPARTICLE PRODUCTION IN LIQUID ENVIRONMENT
As it was shown in the case of nanosecond ablation, ablation in liquids offers a clean and efficient way for colloidal nanoparticle production. Now, the limited number of studies and the immature state of femtosecond ablation in liquid ambient do not allow us to draw solid and general conclusions.34,59 One can state though, that with femtosecond pulses – as compared to the nanosecond case – nanoparticles with smaller mean sizes and lower dispersion can be produced, but at much lower production yield.34,59–61 It sounds feasible that in liquid environment two ablation mechanisms compete, depending on the intensity.59,61 The first mechanism manifests itself at relatively low intensities (e.g. at I < 3.6 × 1015 W/cm–2 in the case of ablation of a gold target in pure deionized water) and produces almost monodisperse gold colloids of very small (~3–10 nm) sizes, as shown in the inset of Figure 16. The second process becomes active at high intensities and produces particles of much larger size and broader size distribution.
Figure 16. Size distribution of gold nanoparticles prepared by ablating a gold target under deionized water with 110 fs pulses of 5.5 × 1014 W cm–2 intensity.61 Inset shows a respective TEM micrograph.
140
T. SZÖRÉNYI AND Z. GERETOVSZKY
Similarly to the nanosecond case, in chemically active liquids enhanced stabilization can be established and leads to more efficient control of nanoparticle size.62,63 The first results are quite impressive and support the assumption that ablation with ultrashort pulses in both low pressure and liquid ambients could be a competitive approach to nanoparticle production. 5. Conclusions-Outlook Laser ablation is a general and practical route for the fabrication of nanoparticles and nanoparticle films of both elemental and multicomponent materials in vacuum, gas or liquid environments. From the point of view of the ablation products the principal difference between ablation with traditional, nanosecond and ultrashort, femtosecond lasers is that while in the former case a condensed ambient is indispensable for nanoparticle production (either as a liquid or in the form of a gas atmosphere at pressures above approximately 1 Pa), in the latter one this can even be accomplished under vacuum conditions. In the case of nanosecond ablation the dominating species leaving the target surface are atoms and ions, and cluster formation and growth take place in the plume. When ablating with femtosecond pulses of intensities not far above the threshold of plasma formation, the ablated material leaves the target in form of nanoparticles, i.e. ablation with femtosecond pulses directly produces nanoparticles and therefore independent from the nature and the pressure of the ambient. The properties of nanoparticles can be further tuned by choosing an appropriate environment, making in-situ engineering possible, and thereby opening up the way to produce more exotic (e.g. coated or functionalized) nanoparticles.
References 1. Chrisey, D., and Hubler, G. (1994), Pulsed Laser Deposition of Thin Films, Wiley, New York. 2. Schou, J. (2006), Laser beam-solid interactions: fundamental aspects, in Y. Pauleau, Materials Surface Processing by Directed Energy Techniques, Elsevier, Amsterdam. 3. Kroto, H., Heath, J., OcBrien, S., Curl, R., and Smalley, R. (1985), C60: Buckminsterfullerene, Nature 318, 162–163. 4. Bulgakova, N., Burakov, I., Meshcheryakov, Y., Stoian, R., Rosenfeld, A., and Hertel, I. (2007), JLMN-J. Laser Micro/Nanoeng. 2(1), 76–86. 5. Yudasaka, M., Ichihashi, T., Komatsu, T., and Iijima, S. (1999), Chem. Phys. Lett. 299, 91–96. 6. Thess, A., Lee, R., Nikolaev, P., Dai, H., Petit, P., Robert, J., Xu, C., Lee, Y., Kim, S., Rinzler, A., Colbert, D., Scuseria, G., Tomanek, D., Fisher, J., and Smalley, R. (1996), Science 273, 483–487.
ULTRASHORT PULSE PLD
141
7. Rinzler, A., Liu, J., Dai, H., Nikolaev, P., Huffman, C., Rodriguez-Macias, F., Boul, F., Lu, A., Heymann, D., Colbert, D., Lee, R., Fisher, J., Rao, A., Eklund, P., and Smalley, R. (1998), Appl. Phys. A 67, 29–37. 8. Dillon, A., Parilla, P., Jones, K., Riker, G., and Heben, M. (1998), A comparison of singlewall carbon nanotube production using continuous wave and pulsed laser vaporization, in R. K. Singh, D. H. Lowndes, D. B. Chrisey, E. Fogarassy, and J. Narayan (eds.), Advances in Laser Ablation of Materials. Mater. Res. Soc., Warrendale, PA, USA, pp. 403–408. 9. Kokai, F., Takahashi, K., Yudasaka, M., Yamada, R., Ichihashi, T., and Iijima, S. (1999), J. Phys. Chem. B 103, 4346–4351. 10. Puretzky, A., Geohegan, D., Fan, X., and Pennycook, S. (2000), Appl. Phys. A 70, 153–160. 11. Puretzky, A., Schittenhelm, H., Fan, X., Lance, M., Allard, L., Jr., and Geohegan, D. (2002), Phys. Rev. B 65, 245425. 12. Puretzky, A., Geohegan, D., Fan, X., and Pennycook, S. (2000), Appl. Phys. Lett. 76, 182. 13. Amoruso, S., Toftmann, B., and Schou, J. (2004), Phys. Rev. E 69, 056403. 14. Geohegan, D. (1992), Thin Solid Films 220, 138–145. 15. Amoruso, S., Toftmann, B., Schou, J., Velotta, R., and Wang, X. (2004), Thin Solid Films, 453/454, 562–572. 16. Henley, S., Carey, J., Silva, S., Fuge, G., Ashfold, M., and Anglos, D. (2005), Phys. Rev. B 72, 205413. 17. Geretovszky, Z., Haraszti, T., Szörényi, T., Antoni, F., and Fogarassy, E. (2003), Appl. Surf. Sci. 208–209, 566–574. 18. Henley, S., Carey, J., and Silva, S. (2004), Appl. Phys. Lett. 85(25), 6236–6238. 19. Bolgiaghi, D., Miotello, A., Mosaner, P., Ossi, P., and Radnóczi, G. (2005), Carbon 43, 2122. 20. Rode, A., Gamaly, E., and Luther-Davies, B. (2000), Appl. Phys. A 70, 135–144. 21. Henley, S., Carey, J., and Silva, S. (2006), Appl. Phys. Lett. 89, 183120. 22. Kabashin, A., Sylvestre, J., Patskovsky, S., and Meunier, M. (2002), J. Appl. Phys. 91, 3248. 23. Szörényi, T., and Fogarassy, E. (2003), J. Appl. Phys. 94, 2097–2101. 24. Szörényi, T., Antoni, F., Fogarassy, E., and Bertóti, I. (2000), Appl. Surf. Sci. 168, 248–250. 25. Hu, J., Yang, P., and Lieber, C. (1998), Phys. Rev. B 57, R3185. 26. Alvarez, F., dos Santos, M., and Hammer, P. (1998), Appl. Phys. Lett. 73, 3521–3523. 27. Walters, J., Kühn, M., Spaeth, C., Dooryhee, E., and Newport, R. (1998), J. Appl. Phys. 83, 3529–3534. 28. Spaeth, C., Kühn, M., Richter, R., Falke, U., Hietschold, M., Kilper, R., and Kreissig, U. (1998), Diamond Relat. Mater. 7, 1727–1733. 29. Fogarassy, E., Szörényi, T., Antoni, F., Stoquert, J., Pirio, G., Olivier, J., Legagneux, P., Boher, P., and Pons-Y-Moll, O. (2002), Appl. Surf. Sci. 197–198, 316–320. 30. Fogarassy, E., Szörényi, T., Antoni, F., Pirio, G., Olivier, J., Legagneux, P., and Boher, P. (2003), Appl. Phys. A 76, 15–19. 31. Geszti, O, Radnóczi, G., Bertóti, I., Szörényi, T., Antoni, F., and Fogarassy, E. (2002), Appl. Surf. Sci. 186, 502–506. 32. Marine, W., Patrone, L., Luk'yanchuk, B., and Sentis, M. (2000), Appl. Surf. Sci. 154–155, 345–352. 33. Patrone, L., Nelson, D., Safarov, V., Sentis, M., and Marine, W. (1999), J. Lumin. 80, 217– 221. 34. Kabashin, A., and Meunier, M. (2006), Laser ablation based synthesis of nanomaterials and references therein, in J. Perriere, E. Millon, and E. Fogarassy (eds.), Recent Advances in Laser Processing of Materials, Elsevier, Amsterdam, pp. 1–36.
142
T. SZÖRÉNYI AND Z. GERETOVSZKY
35. Mafuné, F., Kohno, J., Takeda, Y., Kondow, T., and Sawabe, H. (2000), J. Phys. Chem. B 104(35), 9111–9117. 36. Mafuné, F., Kohno, J., Takeda, Y., Kondow, T., and Sawabe, H. (2000), J. Phys. Chem. B 104(35), 8333–8337. 37. Mafuné, F., Kohno, J., Takeda, Y., Kondow, T., and Sawabe, H. (2001), J. Phys. Chem. B, 105(22), 5114–5120. 38. Szorenyi, T., et al., Proceedings of the 4th International Conference on Laser Ablation: Appl. Surf. Sci. 127–129 and (2004), Proceedings of the 7th International Conference on Laser Ablation: Appl. Phys. A 79, 1373–1376. 39. Banks, P., Dinh, L., Stuart, B., Feit, M., Komashko, A., Rubenchik, A., Perry, M., and McLean, W. (1999), Appl. Phys. A 69, S347–S353. 40. Szörényi, T. (2006), Carbon-based materials by pulsed laser deposition: from thin films to nanostructures and references therein, in J. Perriere, E. Millon and E. Fogarassy (eds.), Recent Advances in Laser Processing of Materials, Elsevier, Amsterdam, pp. 76–104. 41. Amoruso, S., Ausanio, G., Barone, A., Bruzzese, R., Gragnaniello, L., Vitiello, M., and Wang, X. (2005), J. Phys. B: At. Mol. Opt. Phys. 38, L329–L338. 42. Amoruso, S., Ausanio, G., Bruzzese, R., Vitiello, M., and Wang, X. (2005), Phys. Rev. B 71, 033406. 43. Amoruso, S., Ausanio, G., Bruzzese, R., Gragnaniello, L., Lanotte, L., Vitiello, M., and Wang, X. (2006), Appl. Surf. Sci. 252, 4863–4870. 44. Albert, O., Roger, S., Glinec, Y., Loulergue, J., Etchepare, J., Boulmer-Leborgne, C., Perrière, J., and Millon, E. (2003), Appl. Phys. A: Mater. Sci. Process. 76, 319–323. 45. Scuderi, D., Benzerga, R., Albert, O., Reynier, B., and Etchepare, J. (2006), Appl. Surf. Sci. 252, 4360–4363. 46. Teghil, R., D`Alessio, L., De Bonis, A., Galasso, A., Villani, P., and Santagata, A. (2006), Thin Solid Films 515, 1411–1418. 47. Amoruso, S., Bruzzese, R., Wang, X., Nedialkov, N., and Atanasov, P. (2007), Nanotechnology 18, 145612. 48. Mannion, P., Favre, S., Ivanov, W., O’Connor, G., and Glynn, T. (2005), Experimental investigation of micromachining on metals with pulse durations in the range of the electronphonon relaxation time (pico to sub-picosecond), Proceedings of the Third International WLT-Conference on Lasers in Manufacturing, Munich. 49. Amoruso, S., Ausanio, G., Bruzzese, R., Lanotte, L., Scardi, P., Vitiello, M., and Wang, X. (2006), J. Phys.: Condens. Matter 18, L49–L53. 50. Amoruso, S., Ausanio, G., de Lisio, C., Iannotti, V., Vitiello, M., Wang, X., and Lanotte, L. (2005), Appl. Surf. Sci. 247, 71–75. 51. Liu, A., and Cohen, M. (1989), Science 245, 841–842. 52. Muhl, S., and Méndez, J. (1999), Diamond Relat. Mater. 8, 1809–1830. 53. Perrone, A. (2002), Jpn. J. Appl. Phys. 41, 2163–2170. 54. Acquaviva, S., Perrone, A., Zocco, A., Klini, A., and Fotakis, C. (2000), Thin Solid Films 373, 266–272. 55. Geretovszky, Z., Kántor, Z., Bertóti, I., and Szörényi, T.(2000), Appl. Phys. A 70, 9–11. 56. T. Csákó et al., Appl. Surf. Sci., to be published. 57. Eliezer, S., Eliaz, N., Grossman, E., Fisher, D., Gouzman, I., Henis, X., Pecker, S., Horovitz, Y., Fraenkel, M., Maman, S., and Lereah, Y. (2004), Phys. Rev. B 69, 144119. 58. Vidal, F., Johnston, T., Laville, S., Barthélemy, O., Chaker, M., Le Drogoff, B., Margot, J., and Sabsabi, M. (2001), Phys. Rev. Lett. 86, 2573. 59. Barcikowski, S., Hahn, A., Kabashin, A., and Chichkov, B. (2007), Appl. Phys. A 87, 47–55.
ULTRASHORT PULSE PLD 60. 61. 62. 63.
143
Tsuji, T., Kakita, T., and Tsuji, M. (2003), Appl. Surf. Sci. 206 314–320. Kabashin, A., and Meunier, M. (2003), J. Appl. Phys. 94, 7941–7943. Kabashin, A., Meunier, M., Kingston, C., and Luong, J. (2003), J. Phys. Chem. B 107, 4527–4531. Sylvestre, J., Poulin, S., Kabashin, A., Sacher, E., Meunier, M., and Luong, J. (2004), J. Phys. Chem. B 108, 16864–16869.
LASER ABLATION AND LASER INDUCED PLASMAS FOR NANOMACHINING AND MATERIAL ANALYSIS D. BATANI* Università di Milano Bicocca, ITALY
Abstract – The goal of this paper is to introduce the basics of laser ablation. After introducing some general concepts for the description of lasers, and their usefulness in laser micromachining, and of the materials with which they interact, we describe in more details the femtosecond and nanosecond pulse duration regimes.
Keywords: Laser ablation, spatial coherence, ablation threshold flux, plasma frequency, Spitzer’s law, hydrodynamics, collisional absorption, crater formation
1. Introduction Soon after the invention of lasers, these systems begun to be used for applications in machining. Initially CO2 lasers were successfully used for efficient laser cutting of metals (and other materials) in various shapes. CO2 lasers are still probably the most used system at the industrial level. Laser cutting is nowadays so used that it deserved a voice in Wikipedia. More recently lasers have been used in applications like micro-hole drilling, creation of surface structures, introduction of calibrated leaks in pharmaceutical packaging (so to have a prolonged and controlled release of medicaments), selective material removal (e.g. polymer jacket from optical fibers), engraving for code marking or decoration, 2.5D micro milling (i.e. creation of structures in depth in materials), 3D engraving of glass and other transparent materials, laser cleaning in art conservation, guaranteeing high precision and high reproducibility in all cases.
______ *
To whom correspondence should be addressed: D. Batani, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
145
146
D. BATANI
Advantages of using lasers include economic advantages (reduction of work times; increase in production quality) but above all technological advantages: laser machining is precise, clean and silent; the beam can be focused on extremely small areas; the area in proximity to the edge may have very low heat alteration; the laser cut has the capacity of operating on complex profiles and with very small rays of curvature. Also, the laser is a non-contact instrument that guarantees absence of mechanical pressure on the piece (unlike water and traditional cutting systems); working capability independent of hardness of the material; capability of cutting coated or surface treated materials; no contact contamination of materials; no wear on the laser (which affects precision) Laser cutting also has a high degree of automation and flexibility able to offer ease of integration with other automated systems; and capability of adapting immediately to changes in production requirements. Finally, in many cases, laser cutting can produce finished pieces that do not require further processing. All these applications of lasers are based on ablation i.e. the capability to remove matter from a substrate by irradiating it with sufficiently high intense radiation. Focusing a high intensity laser on a solid target brings many effects: (1) ablation (removal) of material (hole drilling, micromachining, surface modifications), (2) emission of radiation (this can be used for plasma and material diagnostics but also for developing pulsed radiation sources, in the XUV and soft X-ray range), and (3) redeposition of the ablated material on a substrate (thin film production, PLD, pulsed laser deposition). In this paper, I will present only the first point and describe the basics of laser ablation in the femtosecond and nanosecond pulse duration regime. For PLD please see the contribution by J. Schou in this book. 2. The Laser System Why are lasers good for micromachining? This depends on characteristics of laser vs. “normal” light. Usual light is characterized by: low directionality, low mono-chromaticity, low coherence, low power (thermal source). Laser light is instead characterized by: high directionality, high mono-chromaticity, high degree of spatial coherence, high power (being non-thermal sources, lasers are not constrained by Kirchoff’s law, i.e. the emitted power may exceed the black body limit at a given wavelength). As it is well known, the peculiarities of laser light comes from two “ingredients”, i.e. (i) the different emission mechanism (stimulated vs. spontaneous emission), and (ii) the role of the optical cavity. Lasers appear as an ideal tool for concentrating energy in space and in time thereby realizing huge irradiation intensities with little energy, and higher intensities imply larger effects on materials. The intensity on target is given by
LASER ABLATION
147
I(W/cm2) = E(J) / (S(cm2) W(s))
(1)
where E is the laser pulse energy, S the focal spot area and W the pulse duration. Focusability is related to spatial coherence and hence to beam quality: Lasers are spatially coherent therefore the beam can be focused to a very small spot. This implies a precise cut but also saving of energy because intensity is inversely proportional to focal spot area. Due to diffraction, focal spot diameter decreases with the wavelength O. With perfect beams and aberration-free focusing optics, we get d = 2.44 O f / D = 2.44 O/F#
(2)
where F# = D/f (the constant 2.44 holds is flat-top laser beam profile in the near field, otherwise a different number appears, always of the order of unity). Provided their optical quality is good, short wavelengths offer the possibility of improved resolution and also of allowing higher intensities on target I v (F#/O)2
(3)
The other way of increasing laser intensity is by reducing the pulse duration W. With this respect after the introduction of Q-switching and mode-locking in the 1970s, which allowed pulses in the picosecond-regime to be produced, the recent breakthrough is the well known introduction of the Chirped Pulse Amplification (CPA) technique1 in the 1990s, which has allowed the generation of laser intensities as high as 1021 W/cm2. Such values are of course not directly interesting for laser-ablation and here we will limit ourselves to much lower irradiation values. In any case CPA has made possible to extend laser ablation to the femtosecond-pulse duration regime. TABLE 1. Laser parameters, in red the useful range for laser ablation.
Wavelength Repetition frequency Average power Energy per pulse Pulse duration Peak power Intensity (irradiance)
Visible, near IR, near UV CW, 100 MHz, MHz, kHz , Hz, single shot MW, W, kW pJ, nJ, mJ, J, kJ μs, ns, ps, fs MW, GW, TW mW/cm2, 109, 1016, 1021 W/cm2
O Q P E = P/Q W Po = E/W I = P/S
Several parameters are useful for a discussion of laser performance in ablation, and in all cases the variation extends over several orders of magnitude: see Table 1 (in red the useful range for laser ablation). As for the choice of the laser-type (i.e. the active medium and the pumping tool) the systems, which are more largely used in today’s laser ablation and laser micromachining, are gas
148
D. BATANI
lasers (especially excimers due to their high efficiency and short wavelength), and solid-state lasers (Nd:glass, Nd:YAG and Nd:YLF and Ti:Sa for the femtosecond regime). For such brief discussion, it appears that the “ideal” laser for ablation and micromachining should have the following characteristics: high beam quality; short wavelength; short pulse duration; high energy per pulse (and high absorption); high repetition frequency; and finally high conversion efficiency (conversion from plug-in to laser energy which implies reduced costs). 3. The Material After describing the laser, let’s now present the other principal actor of laser ablation, i.e. the material. First of all, we must notice how ablation is the removal of matter from the target (i.e. of atoms). But atoms are heavy and inertial so that laser light preferentially interacts with the electrons in the material, while the atoms are heated later due to energy exchange processes. Therefore first we need to characterize the electrons in the materials (Figure 1) and their interaction with light, and then the energy transfer to ions.
Figure 1. Electronic structure in the material.
Let’s first consider the case of metals (we will briefly speak about the case of insulators in next section). As it is well known, electrons are packed in the conduction band and only electrons at the top of the band can effectively exchange energy (the others being constrained by Pauli’s exclusion principle). Electrons in the conduction band are characterized by the value of the Fermi Energy. In practical units EF (eV) § 3 10–7 (ne(cm–3))2/3
(4)
where ne is the density of free electrons in the conduction band. Usually EF is a few electron volts, for instance for aluminum it is 11.63 eV, for copper 7 eV.
LASER ABLATION
149
Another important value is the extraction work ), i.e. the energy difference between the most energetic electrons in the conduction band and the continuum. As for the atoms in the material, they are bonded to the lattice with energies which are characteristics of the materials, but typically Ebond § 1 eV to a few electron volts. This defines an ablation (evaporation) heat: :§ (ni Ebond) = U NA Ebond/A
(5)
where ni is the atom (ion) density in the material, U the material density, NA§ 6 1023 the Avogadro number, A the atomic weight of the material. For instance in the case of Al, U/A§0.1 and assuming Ebond§1eV =1.6 10–19 J we get :§9,000 J/cm3. The corresponding value per unit mass is :U§ 3,300 J/g or, in general :U§ NA Ebond/A. Let’s also notice that the density of ions an free electrons in the material are related by ne§Z* ni, where Z* is the ionization degree or the average number of electrons in the conduction band per ion. Finally we will need to describe thermal conduction in the material and its own time scale Wth. Since Wth is finite and, as we will see, of the order of a few 10 ps, we can already infer that a fast deposition of laser energy (short pulse W, femto and picoseconds) will not allow spreading and penetration of energy to large volumes. In this regime therefore high intensity ablation (I > 1013 W/cm2) will be characterized by direct evaporation of the matter and negligible thermal effects to the surrounding material, resulting in high precision. On the contrary, for very long pulses (microseconds or sub-microseconds) ablation will take place at low laser intensities and evaporation will be accompanied by fusion of material resulting in low precision. To this we must add the fact that in general short pulse lasers are also characterized by a better beam quality, as compared to longer pulse ones. 4. Laser Ablation in the Femtosecond Regime Although of course laser ablation in the femtosecond regime is recent, follow-ing the introduction of femtosecond lasers, we start by describing it because it is a clearer process. We will also see how the description also applies to the few picosecond regime. Laser ablation is a multi step process. First, energy is deposited in a region of thickness G (which, as we’ll see soon, can be identified with the skin depth) being absorbed by the free electrons in the material (electrons in the conduction band). Second, energy is transported by electrons to a thickness l (electron thermal conduction, diffusion process). Third, we have the interaction with the atomic lattice, heating of the atoms, breaking of the chemical bonds, and finally ablation of the material and the formation of a plasma plume. In order for this third step to take place, after thermalisation, ion energy must exceed the bond-
150
D. BATANI
ing energy: this condition can be expressed as CiTi > U: where classically Ci = 3/2ni with number of atoms per unit volume (ionic density). The steps are qualitatively illustrated in Figure 2. In the following we will illustrate the basic physical mechanisms determining the depths G and l. Before that, however, let’s notice one important point. Let’s assume that energy is deposited with an exponential profile in the material with a typical scale-length l, i.e. the energy flux (energy per unit surface) is F (J/cm2) = Fo exp (-x/ l)
Figure 2. Schematic of the ablation process. Energy is absorbed to a thickness G and then diffuses to a larger penetration depth l.
Then the energy deposited between xo and xo + dx over a surface S is
dE S
dF dx dx x x 0
S Fo exp(xo / l )
dx l
(6)
For ablation to take place, this deposited energy this must exceed the ablation heat U: in the volume dV = S dx
dE dV
§x · Fo exp¨ ¸t U: l ©l ¹
(7)
The distance x = L at which the two quantities are equal defines the depth of ablated material. By solving for L we get
§F · § F · L l ln¨ o ¸ l ln¨ o ¸ ©U:l ¹ ©Fth ¹
(8)
LASER ABLATION
151
where the quantity Fth (J/cm2) = U:l is the threshold flux for ablation. This must be corrected for the absorption coefficient of laser light A = 1-R, so that Fth = U:l/A. Let’s notice two important consequences of the logarithmic dependence of L over Fo: (1) Below Fth there is no ablation, but L = l when the logarithm is 1, i.e. when Fo = e Fth § 3 Fth. (2) If you are working with a high repetition frequency laser and above the threshold flux, increasing the inward energy flux Fo (therefore the laser inten-sity I) is not advantageous. Indeed with N shots at a given laser intensity I, the total ablation depth is §NL, while if we increase the intensity N times, the ablation depth only increases as ln(N) (Moreover increasing the intensity may increase unwanted “collateral” effects). Let’s finally observe that this model is 1D. However neglecting 2D dynamics and flows and assuming a radial Gaussian profile of the energy source we can also use the above law in order to derive an equation for the radial size of the ablated region:
F(r) Fo exp(r2 / 2wo2 ) t Fth §F · 2wo2 ln¨ o ¸ ©Fth ¹
r2
From which
(9) (10)
4.1. THE OPTICAL PENETRATION DEPTH G
The initial distance G over which the laser energy is absorbed is determined by the penetration of light in the material, i.e. for a metal, metal, the skin depth. The free electrons in the conduction band form a plasma with electronic density ne = Z* ni and the dispersion relation of the electromagnetic wave (laser) in the plasma, in the collisionless approximation is
Z 2 Z 2p c 2 k 2 4Snee 2 me
Zp
(11) (12)
where Zp is the plasma frequency, which defines the plasma refractive index
n
1
Z 2p Z
2
n 1 e nc
(13)
152
D. BATANI
where nc is the critical density at which the (angular) frequency of laser light equals the plasma frequency. In practical units Zp and nc are given by
Z p (Hz) 5.64 10 4 ne (cm3 ) nc (cm3 )
1.1 10 21 ( O(Pm))2
(14) (15)
Usually Zp !!Z and then the refraction index n becomes imaginary. Therefore the wave is evanescent in the material, penetrating only over a skin depth G § c/Zp For instance for Al, Z* § 3 and ni § 6 1022 cm–3. Then Zp § 2.4 1016 Hz and G § 12 nm. In reality, in a real case, the plasma will be collisional (even strongly collisional) and the plasma frequency may not be much higher than the laser frequency. Therefore determining the skin depth may become a complex problem. Finally, from G, we can calculate the average energy in the initial plasma contained in the volume ʌwo2G, which is
Ee
A EL 2 (Swo G ) niZ *
(16)
Now, as said before, electrons in the conduction band are characterized by the value of the Fermi Energy. As a guiding rule, we can assume that if Ee EF then the characteristics of the material are not affected. We call this “cold solid approximation”. On the contrary if Ee > EF, which happens at high laser intensities, then we will need to take into account modifications in the target characteristics and, in particular, to include plasma effects. Finally the case Ee << EF practically corresponds to no ablation (since EF, Ebond, ) are all of the same order, i.e. a few electron volts). 4.2. THE THERMAL PENETRATION DEPTH l
The next step is the description of the penetration of the absorbed energy in the material at a distance l, from the initially heated region of dimension G. This step is classically described by the so-called “two temperature model”.2 There are several quantities involved in the model: the relaxation time for electrons We, the relaxation time for the ionic lattice WI, the laser pulse duration W, the equilibrium temperature Teq i.e. the common temperature finally reached by ions and electrons in the volume of thickness l, the thermal diffusivity which is classically given by D = (1/3)v2We, the electron-lattice coupling constant J, the laser source flux S, and the heat flux Q(z) at a certain depth z inside the material.
LASER ABLATION
153
The equations of the 2-T model are simply the energy conservation equations for ions and electrons
wTe wQ( z) ° °Ce wt wz J (Te Ti ) S ® °C wTi J (T T ) e i ° ¯ i wt
(17)
In the electron equation, energy balance is given by heat propagating inside the material, by the relaxation towards ions (the lattice) and the flux of energy from the initially heated region. The ion equation only contains exchange with electrons (which is now positive acting as a source term). Initially electron temperature will increase due to the term S, then heat will propagate inside; at the same time ions will remain substantially cold. Finally due to relaxation electron temperature decreases as ion temperature increase until they approach the same limiting value:
Teq
§ z · § z ·º Fo 1 ª l exp¨ ¸ G exp¨ ¸» « Ci l 2 G 2 ¬ © l ¹ © G ¹¼
(18)
Finally, the heat penetration length will be calculated as
DW
l
(19)
The square root dependence is not a surprise this being a diffusion process, and also it is maintained by the laser pulse, therefore, unless it is too short, the time that appears here is just the laser duration. Let’s also notice that in the “usual” case l >> G we get the expression which we have written before, which yields the value of the flux ablation threshold Fth
CiTeq
§ z · Fo exp¨ ¸ l © l ¹
(20)
§ z · exp¨ ¸ G © G ¹
(21)
If instead case G >> l then we get
CiTeq
Fo
Penetration of energy is still exponential but it is now governed by optical propagation, yielding a value of ablation threshold Fth = U:G.
154
D. BATANI
4.3. CALCULATION OF THE ATOMIC PARAMETERS
Let’s consider as an example the case of Cu irradiated by laser light at 532 nm (2nd harmonic of Nd:YAG). Cu has a density of atoms ni = 8.5 1022 cm-3 and Z* = 1, therefore ne = ni. This corresponds to Zp= 1.6 1016 Hz, and G = 1.8 10-6 cm = 18 nm. Also such electronic density gives EF = 7 eV and vF = 1.57 108 cm/s. This is the velocity of the electrons at the top of the conduction bands, those with energy EF and the only ones that can exchange energy. Therefore it seems appropriate to substitute v = vF in the expression for the thermal diffusivity D. The calculation of the electronic relaxation time We requires some specifications. Such time is related to the dc electrical conductivity V by the well-known expression
V
nee 2W e me
Z 2pW e 4S
(22)
therefore it is a function of material density and temperature. Figure 3 reports the behavior of V for solid density Al for temperatures between room temperature and 10 keV. Experimental measurements have been made by Milchberg et al.3 by measuring the reflectivity of femtosecond-heated Al and calculating resistivity. These are compared to the classical Drude model and to a simplified quantum calculation (Eidmann-Huller model4). All curves gives the same behavior: first conductivity decreases and then increases again when the material temperature becomes so large that the plasma regime is reached. In this case conductivity scales with temperature according to the Spitzer’s model, i.e. Vv Te3/2. Let’s notice that the simple classical Drude model, although quantitatively wrong, gives the correct qualitative behavior. All curves show a minimum value of conductivity, called “saturated” conductivity. It is possible to show that this minimum corresponds to We= a/vF, where a is the inter-ionic distance in the material (Ioffe-Regel’s limit). In other terms, in this limit, the electrons moving at velocity vF make a collision with every atom they meet on their path. In the case of copper, we get a = ni-1/3 = 2.3 10-8 cm = 2.3 Å and We = a/ vF = 2.2 10-16 s = 0.22 fs. It is easy to see that such is an incredibly fast time, much shorter than even shorter laser pulses used in ablation. The ion relaxation time is much longer because the energy exchange between very light and very heavy particles is proportional to the mass ratio, i.e. is given by:
Wi which yields § 26 ps in our case.
miW e me
(23)
LASER ABLATION
155
109
Electrical conducibilty MKS (:m)
-1
experimental data from Milchberg et al. Quantum model by Eidmann and Huller Semiclassical Model (Drude + Spitzer)
108
Vcold Al = 3.7 107 :-1m -1
107
106
T (eV)
105 0.1
1
10
100
1000
4
10
Figure 3. Behavior of electrical conductivity vs. temperature for solid state Al.
Using these values we get for the thermal diffusivity D = (1/3) vF2We = 1.8 cm /s. Now, assuming as an example W= 40 ps (at the very upper limit of the validity range for the present model of laser ablation in the femtosecond-regime) we get l = (DW § 85 nm. Finally we can calculate the threshold flux for ablation as Fth = U:l/A = 1.2 J/cm2 where we have taken into account that A = 0.3 at 532 nm. Of course this is the value calculated for a polished and clean Cu surface, which will certainly be very different from a real rough and oxidized surface; however this is not too important: A appears inside a logarithmic terms so small differences in A will not change the value of Fth too much. Coming back to the cold solid approximation, we therefore see that in such approximation (Ee§EF) no characteristics of the material is strongly affected: the electronic relaxation time remains the same, there is no ionization above the normal value of Z* therefore the electronic density doesn’t change, the values of G, A, EF, vF don’t change. On the other end, as already said, if Ee<<EF, practically there is no ablation. 2
4.4. AN EXAMPLE OF MEASUREMENTS
As an example, let’s consider the results from an ablation experiment, which was performed at the LOA Laboratory (Palaiseau, France) with the goal of drilling holes in capillary tubes for gas inlet for an electron acceleration experi-ment.5 The used laser was a dye at 620 nm, pulse duration 80 fs (FWHM), repetition rate 10 Hz, energy per pulse <3 mJ. In the experiment §60 μJ of laser energy were delivered on target while the laser focal spot diameter was varied from 13
156
D. BATANI
to 30 μm giving a laser flux F on the target surface between 2 and 40 J/cm2, corresponding to a maximum laser intensity on target 5 1014 W/cm2. The energy reproducibility from shot to shot was ±10%. The targets used in the experiments were stainless steel and fused silica of 120 μm thickness and were irradiated in air at normal incidence. A shutter placed on the beam allowed to choose the number of pulses. The ablation depth was measured by exposing the target with 1,000 pulse. Quantitative measurements of ablation depth were then performed with optical and scanning electron microscopes and the ablation depth per shot was estimated as the total ablation depth (measured value) divided by number of shots (1,000). In this experiment, precise machining was not a requirement as compared to fast operation. Hole morphology is shown in Figure 4. Figure 5 shows instead the ablation depth results obtained by varying the energy flux on target.
Figure 4. Scanning electron microscope (SEM) images. (Left) Front surface of the ablation site after exposing the target surface with 1,000 laser pulses at F = 105 J/cm2. (Right) Rear surface of the ablation site after exposing the front side with a number of pulses sufficient to get complete drilling at F = 33.6 J/cm2.
The ablation results show a reduction of ablation depth due to the presence of air, which induces re-deposition of the ablated material (but the threshold flux remains the same). 4.5. THE CASE OF DIELECTRICS
In the case of dielectrics, the valence band is full and the con-duction band is empty, there are no free electrons. Light penetrates and it is attenuated with an exponential law (Lambert-Beer law) with an attenuation coefficient D, i.e. I(z) = Io exp(-Dz). Therefore we can replace G with 1/D.
LASER ABLATION
157
Ablation Depth per Shot (Nm)
0,2
0,15
Nolte’s model (with Fth = 0.335 J/cm2 and I = 44 nm)
0,1
0,05
0
5
10
20 25 15 Flux (J / cm2)
30
35
40
Figure 5. Ablation depth (μm) per pulse vs. laser flux F on target: experimental results (black points) compared to (a) an interpolation L (nm) = 9 ln (F/Fth), and (b) to the formula L = l ln (F/Fth) where l = 44 nm is calculated with the 2T model. In both cases Fth § 0.335 J/cm2.
Figure 6. Effect of multiphoton absorption in dielectrics.
However due to the high laser intensity photons are absorbed by multiphoton absorption so that electrons cross the gap and move to the conduction band or even by multiphoton ionization and go to the continuum (free electrons), as in Figure 6. However, when the density of created free electrons reach the critical density nc, the laser can no longer propagate in the plasma and the process stops (self-regulation). Therefore with high-intensity short-pulse lasers there is no substantial difference between insulators and conductors with a free electron density nc. 4.6. HIGH INTENSITIES: PLASMA EFFECTS
At higher laser intensities (>1013 W/cm2) and short pulses, we get Ee > EF. This results in strong heating of the material and plasma effects: (1) ionization of
158
D. BATANI
material above the normal value Z: the electronic density changes inducing changes in G, A, EF, vF; (2) high temperature of the initial plasma. Formulas for heat transfer derived from solid state physics are no longer applicable. In particular there is an increase in the velocity of electrons (vF is replaced by electronic thermal velocity vth) and there is an increase in We for which we must now use Spitzer’s law giving We in a plasma
We
10 6 Te3 / 2 s 2.91 ne ln /
(24)
The plasma ionization degree should be calculated in a self consistent way. Figure 7 shows the average ionization degree Z* for gold at density U = 0.1 g/cm3 vs. temperature in various approximation or using different formulas. 80 70 60
Z*
50 40 30 20 10 0 10
1000
100
104
T (eV)
Figure 7. Average ionization degree Z* for gold at density U = 0.1 g/cm3. Squares: coronal limit; circles: Thomas Fermi model; solid lines: analytical formula for coronal equilibrium (CE); dashed line: analytical formula by Colombatnt and Tonon6 (valid in Collisional Radiative Equilibrium, CRE).
In principle, we should also calculate the energy lost to ionization by summing over the different ionization energies for each electron in Cu. (shown in Figure 8). However it is possible to show that usually the total energy lost to ionization Eion is negligible with respect to the thermal energy. The following table shows a comparison of the ablation parameters calculated in the cold solid approximation and in plasma model. The pulse duration is 400 ps. In the plasma case, values are calculated for an intensity I § 1014 W/cm2 giving a temperature Te § 500 eV.
LASER ABLATION
159
104
1000
100
10
1 0
5
10
15
20
25
30
Figure 8. Ionization energy of each electrons for Cu ions. TABLE 2. Comparison of ablation parameters calculated in the cold solid approximation and taking the plasma effects into account.
Z* Free electron density ne Electron velocity ve Electron relaxation time We Penetration depth l Ablation threshold flux Fth = U:l/A
Cold solid approximation 1 8.5 u 1022 cm–3 vF = 1.6 108 cm/s 2.2 u 10–16 s 0.1 μm 1.2 J/cm2
Plasma model 18 – 19 15 u 1023 cm–3 vth = 1.5 109 cm/s We (Spitzer) = 10-15 s 2 μm 70 J/cm2
The transition between the cold solid approximation and the plasma case is qualitatively illustrated in Figure 9. The plasma case onset corresponds to a higher ablation threshold flux, but gives larger ablation depths L. L L
§ F · o ¸ lplasma ln¨ ¸ ¨F © th plasma ¹
L
§ F · lcold ln¨ o ¸ ©Fth cold ¹
Figure 9. The transition between the cold solid approximation and the plasma case The plasma case onset corresponds to a higher ablation threshold flux, but gives larger ablation depths L.
160
D. BATANI
As an illustration of ablation measurements in the plasma case, let’s consider the following measurements performed at Milano Bicocca.7 A Nd:YAG (Quanta System) laser was used with maximum energy 60 mJ per pulse, 10 Hz repetition frequency, working in the green (532 nm) with focal spot radius wo=25 μm and pulse duration W = 40 ps giving a maximum intensity on target I = 1014 W/cm2. Figure 10 shows the depth and diameter of the ablation region on Cu targets in air and vacuum, as a function of energy flux. Figure 11 instead compares results obtained in vacuum for Copper and Lead. In both cases, results were obtained by analyzing the irradiated samples with optical microscopy. First, data show that the presence of air decreases the ablation depth of about 40%: as in the case of the previous experimental results obtained at LOA (Figure 5) this is due to the re-deposition of the ablated material (while the threshold flux is not affected). From available data, we know that for Cu U = 8.96 g/cm3 and : = 4723 J/g, while for Pb U = 11.35 g/cm3 and : = 859 J/g. Then interpolation of data yields the following penetration depths and ablation threshold: Cu, l | 5 Pm and Fth | 70 J/cm2, Pb, l | 4 Pm and Fth | 7 J/cm2. The ablation rate for lead is larger than for copper but this is due to the different in the evaporation rate while the heat penetration rate is practically the same (indeed this is mainly a function of plasma conditions, including electron temperature, which are mainly fixed by the char-acteristics of the laser pulse). Also, we see that in this experiments, the penetra-tion lengths are of the order of a couple of micrometer, as indeed it should happen in the plasma case, and not of several 10 nm, as in the cold solid approximation.
Diameter Depth (Pm)
30
(Pm)
160 140
VACUUM
25 120
VACUUM 20
100 80
15
60 10 40 5
AIR
AIR
20 0
0 0
1000
2000
3000
Flux (J/cm2)
4000
5000
0
1000
2000
3000
4000
5000
Flux (J/cm2)
Figure 10. Depth and diameter of the ablated region on Cu vs. energy flux on target. Data have been interpolated with the formulas introduced in Section 4.
Alternatively, the ablation depth per shot can be measured by counting the number of shots, which is needed to drill a clear hole in a foil with a given
LASER ABLATION
Diameter
Depth (Pm)
161
( Pm)
40
200
LEAD
35 150
LEAD
30 25
100
20 15
COPPER 50
10
COPPER
5 0
0 0
1000
2000
3000
4000
Flux (J/cm2)
5000
0
1000
2000
3000
4000
5000
Flux (J/cm2)
Figure 11. Depth and diameter of the ablated region for Cu and Pb vs. laser flux in vacuum.
Figure 12. AFM images of surface modifications induced on the irradiated Al target.
thickness. For instance, in the same experiments, we measured that 53 shots at the maximum energy flux are needed to drill a hole in a Cu foil 1-mm thick. This implies an ablation depth per shot of 19 μm, in fair agreement with the measurements reported in Figure 11. In the same conditions, the measured hole diameter was 120 μm. Finally, let’s notice that removal of material is of course not the only consequence of laser ablation. Indeed the characteristics of the material surface are also modified, an aspect that is also often used in application8,9 (e.g. surface hardening, surface cleaning, surface polishing). Let’s see for instance the analysis of the irradiated Al surface obtained with the same laser system. The Atomic Force Microscopy (AFM) pictures in Figure 12 (40 u 40 μm and 80 u 80 μm respectively) show two samples irradiated at Milano Bicocca (10 laser shots on an Al surface). The laser induced a change in the surface: the smooth surface was not irradiated, the part with “bubbles” was irradiated). The second picture shows a granular and irregular surface following laser irradiation.
162
D. BATANI
5. Laser Ablation in the Nanosecond Regime The situation for laser ablation in the nanosecond regime is very different: there is indeed one very important point, which is implicit in what we said until now. During the duration of the laser pulse W, matter has no time to move, there-fore the ionic density ni remains constant. Indeed the sound velocity is
cs
JZ * Te mi
9.8 10 5
JZ * Te (eV ) A
cm / s
(25)
Even if Te = 100 eV, we get cs = 7 106 cm/s, and then if W = 100 fs, then the typical extension of the plasma is L= cs W = 0.07 μm. This shows that there is no appreciable movement or expansion of the target material. The situation in the nanosecond regime is very different being completely dominated by hydrodynamics. Because matter has time to move, we have plasma expansion in vacuum. A low density plasma corona is formed in front of the target surface. The laser interacts with the plasma and it is absorbed in the plasma corona at the critical surface. Energy is then transported inwards by electronic thermal conduction (and/or radiation and/or suprathermal electrons). Matter is constantly removed from the solid at the ablation surface (where U = Uo of the cold solid). The plasma expands in vacuum and plasma expansion is fed by laser ablation. Beyond the ablation surface a shock is formed as a consequence of momentum conservation: the material is compressed by such ablation (shock) pressure. Shock front and ablation front moves inward in the material. The situation is evidenced in Figure 13. Ablation region
Plasma Expansion
ne ns
Overcritical region
LASER
Te ^ Kev nc corona
xs xa
xc
Figure 13. Typical density and temperature profiles established during nanosecond-laser interaction.
LASER ABLATION
163
It is possible to accurately describe the 1D hydrodynamics though analytical formulas obtaining scaling laws for the relevant quantities, the ablation pressure and the mass ablation rate.10,11 In practical units: 2/ 3 § 2 · § ·2 / 3§ A ·1 / 3 I (W / cm ) 1 ¸ ¨ P( Mbar ) 12.3¨ ¨ ¸ ¨ ¸ ©O (Pm) ¸ 14 ¹ ©2Z ¹ © 10 ¹ 1/ 3
§ · I (W / cm 2 ) ¸ m (kg / s cm 2 ) 150¨ ¨ ¸ 14 ¹ © 10
4/3
§ 1 · ¸ ¨ ©O (Pm) ¹
(26) 2/ 3
§ A · ¨ ¸ ©2Z ¹
= 32 kg/cm2s. For instance irradiation at 1012 W/cm2 with O = 1 μm yields m For a laser pulse duration W = 1 ns, this gives a mass removed for unit surface m W § 3 10-5 g/cm2. Finally for U = 2.7 g/cm3 (Al), this gives and ablated depth W/U = 0.12 μm. At 1014 W/cm2 it becomes 0.6 μm. d= m The ablation rate is a large number but the ablated mass and the ablated depth become very small because of the small focal spot and short pulse duration. However such ablated depth can be difficult to be measured because in this regime often the shock is also a very effective factor producing further ablation of material, as we will see in the following. Let’s also notice that ablation pressure can very easily reach the several 10 Mbar range. Indeed this kind of experiments are also used to compress matter to high pressures and study physical characteristics and equation of state of compressed matter.12 5.1. HYDRODYNAMICAL CODES
A number of 1D and 2D hydro codes are well adapted to describing the hydrodynamics and the ablation of nanosecond-laser irradiated targets, e.g. the code MULTI developed by R. Ramis et al. at Universidad Politecnica, Madrid.13,14 In Figure 14, we see the interaction of a 6 μm foil Al target with laser radiation with W = 3 ns at 800 nm and 1012 W/cm2. After shock breaks out on target rear side (at about 0.75 ns after the beginning of the laser pulse), the whole target is accelerated by the laser beam. At the same time a relaxation wave is generated and reaches the front side again. Since the laser pulse is still on at this time, then a second shock is generated and reaches the rear side just before 2.5 ns. The ablated mass corresponds to the rear side reaching low densities (green area). Of course if the target is massive, then the shock will never breakout on
164
D. BATANI
Figure 14. Interaction of a 6 μm Al foil with a laser with W = 3 ns at 800 nm and 1012 W/cm2.
rear side and only front side dynamics will be important. In this case relaxation will take place only at the end of the laser pulse, although some residual superficial compression may be maintained in the material (and actually it may be useful for surface hardening of the material). 5.2. THE QUESTION OF LASER ABSORPTION
The question of laser absorption is very important for an efficient laser-target coupling and then for an efficient ablation, as we have noticed in Section 2. While the absorption of femtosecond-lasers is still poorly understood, the situation is clearer for nanosecond-laser pulses In the corona laser energy is absorbed collisionally (i.e. by inverse bremsstrahlung). The absorption coefficient then reads,
D (cm1 ) 3.1 107 Z * ne2Z 2 ln / Te3/ 2 n 1
(27)
On this basis, very often we find in text books that IR lasers are absorbed more because of longer wavelengths (smaller Z). In reality this view is only partial and true only for very low intensities (or rather short pulses) for which an extended plasma corona does not develop. But as soon we get an extended corona (which is the main feature of nanosecond-interactions) this is not true. We must remember the plasma refraction index n becomes 0 as the critical density is approached, and also that the critical density is inversely proportional to the square of wavelength. Therefore, at shorter wavelengths, the laser can penetrate deeper in the plasma and reach higher intensities where absorption is higher.
LASER ABLATION
165
Experimental measurements by Fabre et al.15 performed already in the 1980s, clearly show such behavior (see Figure 15) A = absorption % 2.1010 2.1011 100
2.1012
intensity (W / cm2) 2.1013 2.1014 2.1015
0.53Mm 2ns
90 80 70 60
1.06Mm100 ps
1.06Mm 2.5ns
50 40 30 20 E(J) for 100 ps pulse 1/25E(J) for 2.5 ns pulse
10 0 10 4
10 3
10 2
10 1
1
10
Figure 15. Fraction of laser light absorbed in a plasma as a function of laser intensity on target, laser wavelength and pulse duration.
5.3. SHOCK EFFECTS AND CRATER FORMATION
At high laser irradiances in the nanosecond-regime, the shock wave becomes the dominant aspect of the interaction and leads to the formation of a crater.16 This is shown for instance in the SEM images in Figure 16. In general this regime is not interesting for micromachining, unless precision is not a requirement. However it is still very interesting from the point of view of physics.
Figure 16. SEM images showing the surface morphology of caters created on Al flat targets: (left) diameter D = 30 μm, flux § 2,200 J/cm2; (right) D = 100 μm, flux § 4,000 J/cm cm2.
166
D. BATANI
Shock dynamics is governed by Rankine-Hugoniot relations which represent the conservation laws for mass, momentum, and energy (per unit mass) across the shock front. For strong shocks, these read: Po UoUD
P
Uo D U ( D U ) 'E
E Eo
(28)
1 2 U 2
Here P, U, E, D and U are respectively the shock pressure, material density, internal energy per unit mass, shock and fluid velocity (i.e. the velocity of matter behind the shock front), and the suffix “o” refers to the unperturbed material ahead of the shock front. The condition for ablation requires the energy increase per unit mass to be larger than ablation threshold, i.e. 'E = U2/2 > :. Hydrodynamics is governed by scaling laws i.e. it is universal or scaleinvariant. Crater formation is governed by self-similar laws (i.e. power laws). The same laws concerning crater depth and crater diameter hold from micronsize laser crater to huge craters created by nuclear explosion and meteorite impacts (according to the theory formulated by Raizer in the approximation of a pointlike, in space and in time, energy release17). It is possible to derive the scaling law for crater size in the following way. The crater size Rcrat is determined by the condition 'E(Rcrat) = :. In the case of crater formation we have a spherical shock, which is expanding in the material starting from the laser focal spot, and thereby reduces its pressure according to:
§Ro ·2 Po ¨ ¸ © R ¹
P
(29)
In a material the shock and fluid velocity are usually related by the relation D = cs + SU where cs is the sound velocity in the material. For strong shocks gives D§ SU from which P = Po + UoDU §UoDU §UoSU2. Therefore
'E
U2 2
P 2UoS
2 Po §Ro · ¨ ¸ 2 U o S © R ¹
§Ro ·2 k ¨ ¸ © R ¹
and the crater size is
Rcrat
Ro Po / 2 Uo S:
Ro k / : Ro
Rcrat / k / :
We now integrate to get the total energy spent in creating the crater
(30)
LASER ABLATION
Eabs
³
Rcrat 0
167
§Ro ·2 k ¨ ¸ 2SR 2 dR 2SkRo2 Rcrat © R ¹
(31)
and by substituting the previous relation between Rcrat and Ro we finally get
Eabs
2SkRo2 Rcrat
2Sk Rcrat / k / :
Rcrat 2
3 2SkRcrat / : (32)
or in other words a cubic root dependence 1/ 3
Rcrat v Eabs
(33)
Under such conditions, the crater may (or indeed will) be much larger than the focal spot (i.e. the laser acts as a point source). 6. Conclusions We have revised the physical basics of laser ablation in the femtosecond and nanosecond pulse duration regimes. Laser ablation is a complex physical phenomenon which involves optical penetration of laser in the material and the transport of energy inside the material, a problem at the border line of solid state and plasma physics, which may imply quantum mechanical calculations. In the nanosecond-laser interaction regime, ablation is dominated by hydrodynamics: ablation feeds plasma expansion continuously and the laser interacts within the corona, producing an increased absorption at shorter wavelengths. Finally, shock dynamics and the formation of a laser crater is often a key phenomenon associated to high-intensity nanosecond-laser interactions.
References 1. 2. 3. 4. 5.
Perry, M., and Mourou, G. (1994), Science, 264, 917–924. Nolte, S. et al. (1996), J. Opt. Soc. Am. B, 14(10), 2716. Milchberg, H. et al. (1988), Phys. Rev. Lett., 61(20), 2364. Eidmann, K., Meyer-ter-Vehn, J., Schlegel, T., and Hüller, S. (2000), Phys. Rev. E, 62, 1202. Di Bernardo, A., Batani, D., Desai, T., Courtois, C., Cros, B., and Matthieussent, G. (2003), Laser Part. Beams, 21, 59–64. 6. Tonon, G., and Colombant, D. (1973), J. Appl. Phys., 44, 3524–3537.
168
D. BATANI
7. Desai, T., Batani, D., Rossetti, S., and Lucchini, G. (2005), Radiat. Eff. Defects Solids, 160(10–12), 595–600. 8. Trtica, M., Gakovic, B., Maravi, D., Batani, D., Desai, T., and Redaelli, R. (2006), Mater. Sci. Forum, 518, 167. 9. Trtica, M., Gakovic, B., Batani, D., Desai, T., Panjan, P., and Radak, B. (2006), Appl. Surf. Sci., 253, 2551–2556. 10. Lindl, J. (1995), Phys. Plasmas, 2, 3933. 11. Batani, D., Stabile, H., Ravasio, A., Desai, T., Lucchini, G., Desai, T., Ullschmied, J., Krousky, E., Juha, L., Skala, J., Kralikova, B., Pfeifer, M., Präg, A., Nishimura, H., Ochi, Y., et al. (2003), Phys. Rev. E, 68, 067403. 12. Batani, D., Stabile, H., Tomasini, M., Lucchini, G., Ravasio, A., Koenig, M., BenuzziMounaix, A., Nishimura, H., Ochi, Y., Präg, A., Hall, T., Milani, P., Barborini, E., Piseri, P., et al. (2004), Phys. Rev. Lett., 92, 065503. 13. Ramis, R., Schmalz, R., Meyer-ter-Vehn, J. (1988), Comput. Phys. Commun., 49, 475. 14. Ramis, R., and Meyer-ter-Vehn, J. (1992), MULTI2D-A computer code for two-dimensional radiation hydrodynamics, MPQ-174. 15. Garban-Labaune, C., Fabre, E., Max, C., Fabbro, R., Amiranoff, F., Virmont, J., Weinfeld, M., and Michard, A. (1982), Phys. Rev. Lett., 48, 1018–1021. 16. Bussoli, M., Batani, D., Desai, T., Milani, M., Trtica, M., Gakovic, B., and Krousky, E. (2007), Laser Part. Beams, 25, 121–125. 17. Zeldovich. Y., and Raizer, Y. (1967), Physics of Shock Waves and High Temperature Hydrodynamic Phenomena, Vol. 1, Academic, New York.
PHOTO-, DUAL- AND EXOELECTRON SPECTROSCOPY TO CHARACTERIZE NANOSTRUCTURES
Y. DEKHTYAR* Biomedical Engineering and nanotechnologies Institute, Riga Technical University, Kalku 1, Riga LV-1658, LATVIA
Abstract – Physical basics of prethreshold photo-, dual- and exo-electron spectroscopy to characterize nanostructures, as well as relevant measurement technique are described. Application cases are considered.
Keywords: Prethreshold photo-, dual- and exo-electron spectroscopy, nanostructures.
1. Introduction Advances in nanotechnologies need non-destructive characterization methods. To reach this, electron emission measurements, when energy to escape an electron from a tested object is not enough to destroy its molecular/atomic couples, are employed as such energies have values ~0.1–1 eV for non-destructive characterization. In such a case the excited electron has a mean free path (L) ~10–100 nm,1 correspondingly within the analysed emitter. These values of L fit sizes of nanostructures. This is a key point to apply electron emission for nanomaterials characterization. The paper is targeted to consider a prethreshold photo-, dual- and exo-electron emissions and their capabilities in respect with nanotechnologies. 2. Photoelectron Emission Analysis 2.1. PHYSICAL BASICS
A current (I ) of the prethreshold single photon photoelectron emission (PE) is described by the classical formula
______ *
To whom correspondence should be addressed: Y. Dekhtyar, e-mail,
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
169
Y. DEKHTYAR
170
conts(hQ M ) m
I
(1)
where hQ – energy of exciting photon, Melectron work function, m-power index. For the prethreshold emission mode (hQ t M the ҏvalues of M are equal to several electron volts,2 and therefore a mean pass of the photoelectron within the emitter is around 10–100 nm.1 To emit the electron it is excited from an initial (i) to a final (f) states that correspond to the energies Ei and Ef . The electron that leaves the specimen to the vacuum has the energy
E f Evac
W
(2)
Evac – energy of the vacuum level. Because of the energy conservation law
hQ
E f Ei
following main photoelectron spectroscopy methods could be employed. When hQ = const,
W
Ei hQ Evac ,
dI dW
dI dEi
(3)
the approach on initial states spectroscopy is provided. When Ef = const, hQ = var
dI d (hQ )
dI dE f
(4)
the spectroscopy of the final states is available. If Ei = const, however Ef = var and hQ = var with the condition 'Ef = 'hQ (' – means increment) (1) comes again to (4). When electron emission is delivered from the local states (Ei = const) and hQ = var, W
I ~ ³ N i ( E ) N f ( E , hQ )dE ,3
(5)
0
N i , N f – density of the initial and final electron states, correspondingly. Using
this way
dI ~ N i (W ) N f (W , hQ ) . dW
(6)
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
171
However,
dI dW
dI . dhQ
(7)
Because of (6) estimation of the N i , N f combination could be estimated. The electron work function in a case of emission from metals and semimetals that do not have an energy gap (Eg) is equal to an electron affinity (F). For the materials with Eg (semiconductors, dielectrics)
M Eg + F . The index m (formula (1)) characterises transition channel to excite the electron (see please the Table 1 below): TABLE 1. Index m in for different electron transitions. Material
Transition
Crystal
Direct Nondirect From the local level
Amorphous
Value of m
Source
1 2.5 1.5 >3
5 5 6 7
2.2. INSTRUMENTATION
To get the advantages (estimation of m, M and I(M) of the prethreshold PE measurements the photon energy should be as possible as close to the electron work function: hQ | t M. To supply the single photon electron emission mode and to avoid heating of a tested object, the flux of the photons should be rather weak. This stipulates small values of I. Because of this very sensitive electron detectors must be in use. Typically the secondary electron multipliers that have a noise 0.1–1 electron/second are applied. The value of F could be effected by ions/dipoles sorption/desorption on/from an emitter surface.4 This asks for stable and “passive“ electron emission measurements environment (preferably vacuum; 10–2…–7 Pa8). Typically the values of M are not distributed uniformly over a surface of the emitter. As a result a contact potential difference is induced between the surface arrears and electrical fields are induced. It decreases/increases I from the places with lower/higher M, correspondingly.4 In fact, when characterisation of M distribution is an aim of measurements, their results could become false. To avoid this, an external compensating electrical field ~102…3 V/cm directed from the surface to the detector should be provided.4
Y. DEKHTYAR
172
To measure energy of the escaped electron (for the approach by the formula (3)) an energy analyser should be in use. The approaches by (4)–(7) are available, when a monochromator of the photons is provided. 2.3. APPLICATION CASES
2.3.1.
Employing of the electron work function
2.3.1.1. Mechanical tensions of crystalline silicon surface layer Influence of any mechanical load on a negative increment of the silicon energy gap9 was in use to identify location of mechanical tensions induced within a surface layer because of its local doping with boron.10 A thickness of the doped layer was equal to 100 nm. A distribution of related to the tensions I over the surface is demonstrated in Figure 1. A crown at the boundary of the image corresponds to increased local deformation. Doping (B) 1000 Å
Si (B) |V|
Energy of electron
x Vacuum Cond. band
Cond. band
M2 Valence band
Valence band
Cond. band
M1 Valence band
Figure 1. PE Image of the mechanical tensions distribution over the Si surface.10
2.3.1.2. Thickness of nanofilm An insulator film Si3N4 was deposited with a different thickness (5–100 nm) on a Si substrate. Because of simultaneous electrons emission from the film and the substrate, both of them having differentM, their contribution to the value of I depended on a thickness of the film. Figure 21 provides the PE image of the non-uniform nanofilm. Figure 3 demonstrates a response of I during plasma chemical etching of a multi layer structure. Measurements were supplied in situ using a PE spectrometer incorporated into an etching machine. 2.3.1.3. Size depended electrical potential of nanoparticles It is known that mechanical surface tensions depend on a size of the spheri-cal like particles. This could influence of density of the surface charge, particularly in a case of ion crystals. As the result value of M should be affected.
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
173
Figure 2. PE current distribution over the nanofilm having different thickness.
Figure 3. PE current in dependence on multi layer system etching (in situ).
Figure 4 demonstrates a correlation of M on a size (X) of the hydroxyapatite nanoparticles.11 2.3.2.
Density of electron states
2.3.2.1. Electron states of the nanoparticles Computational simulation (CS) evidenced that the result in Figure 4 was stipulated because of different electron state density induced by protons.11 To verify these approaches (6) and (7) were employed.11 The achieved results demonstrated that increasing of the particle size (X) shifts the detected local
Y. DEKHTYAR
174
0,71
log M>M@ = eV
0,705 0,7 0,695 0,69 0,685 0,68 0,675 0,67 1
1,5
2
2,5
3
3,5
log X, [X]= nm Figure 4. Correlation of M on a size of the nanoparticles.11
dI/dhQ , arb. units
Shift of the local state
Tail of states
1,2 1 0,8 0,6 0,4 0,2 0
X=100 -1000 nm X=20-60 nm
-0,2 -0,4 -0,6 4,9
5,4
5,9
hQ , eV
Figure 5. Distribution of particle size.11
dI / d (hQ ) ~ N i N f
on dependence on the hydroxyapatite nano-
centre in accord with CS the and provides a tailed distribution of the electron states that is similar with disordered systems (Figure 5). The latter was evidenced because of the index m (1) increasing on 50%.11 2.3.2.2. Electron states of the bone The approach (3) was applied jointly with fluorescence and dual emission (see please below) measurements to explore electron states of the bovine bone.12 The result presented in Figure 6 evidenced that the bone has a semiconductor like electron density states distribution.12
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
E eV
vacuum level
175
Density of electron states, arb. units
-1.0 -2.0 -3.0 -4.0 -5.0 -6.0 Figure 6. Electron density of the states in bone.12
3.
Dual Electron Emission Analysis
3.1. BASIC PHYSICS
When the tested object has and energy gap, electrons and hols may be generated because of radiation by photons having an energy hQad t Eg. The excited charge carries are mobile and therefore provide an opportunity to compensate a surface charge that is resulted by the Fermi level pinning at the surface13 and adsorbed ions/dipoles.14 If the photons are switched alongside with PE measurements of I, the latter demonstrates an increment ('I). This is a dual emission mode (DE). The values of 'I may be calibrated in terms of the surface charge density, behaviour of 'I on time reflecting surface electrical potential relaxation induced by photon flux switching on/off.12 3.2. INSTRUMENTATION
The technique on DE measurements is similar that is in use to detect PE, additional light source and monochromator to select suitable hQad should incurporated. To exclude influence of hQad on direct escaping of the photoelectron, the DE is restricted by the condition hQad < hQ, where hQ is the photon energy from (1). On the other hand to minimize an effect of hQ on DE, a ratio of light fluxes must be taken into account:
Y. DEKHTYAR
176
Fad /F >> 1, where Fad and F are the fluxes of light beams that supply hQad and hQ, correspondingly; Fad is not enough for multi photon processes. To select a contribution of hQad , it should be supplied in a pulse mode. 3.3. APPLICATION CASES
3.3.1.
Detection of the energy gap
'I, arb. units
Figure 7 provides an example on Eg detection of the CdTe semiconductor. The value of 'I becomes from zero to a positive value, when hQad = Eg .
1 0.8 0.6 0.4 0.2 0
Eg
1
1.1
1.2
1.3
1.4
1.5
1.6
hQad, eV Figure 7. 'I dependence on hQad in respect to evaluate the energy gap of CdTe surface layer.
3.3.2.
Light induced time depended excitation and relaxation
The flux Fad was switched (+hQad) to radiate CdTe crystal and the value of 'I was increased in time (Figure 8 ). When Fad was switched off (-hQad), 'I had been relaxed. 4. Exoelectron Emission Analysis 4.1. BASICS OF PHYSICS
Electron emission, when its current is induced15 or modulated16 because of relaxation processes within a surface layer of the solid is named as exoelectron
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
+hQ Qad
177
-hQ Qad
1
'I, arb. units
0.95 0.9 0.85 0.8 0.75
time, minutes
0.7 0
2
4
6
8
10
Figure 8. Behavior of 'I because of Fad , in a case of CdTe.
emission (EE). To reach EE two steps should be applied. First, a tested object in advance typically has to be provided with imperfections, their concentration is not in a thermodynamically equilibrium state. After that heat is delivered to the object to induce relaxation of imperfections. As a result EE is supplied. This is a mode of thermostimulated EE (TSE). On the other hand the object tested with PE could be heated alongside to supply EE. In such a case photothermostimulated EE (PTSE) is achieved. EE current (IEE) depends on temperature (T) of the tested specimen, behaviour of IEE having a maximum (Figure 9). The latter is a subject of research and practical applications of EE for analysis. The temperatures to reach EE are not enough to provide significantly detectable thermoelectron emission current. TSE is typically delivered by following mechanisms: (a) Thermoionisation of local states trapping electrons and belonged to the imperfections15; such a mechanism is available from materials with Eg; the ionisation potential should have a little value. (b) Auger transitions of electrons,15 from local electron traps of imper-fections; this channel is available when Eg > F. (c) Field emission from the electron traps by imperfections because of heat induced electrical polarisation/depolarisation of the emitting surface layer.17
Y. DEKHTYAR
178
IEE
Ti
Tmax
Tf
T
~ (+20 … 600) o C Figure 9. Typical spectrum of EE. (Ti, ,Tf , Tmax – initial, final temperatures of the spectrum and temperature of its maximum, correspondingly).
PTSE is provided as a single photon prethreshold PE, when heat delivers: (a) Modulation of the density of the electron local states induced by imperfections and emitting electrons16 (b) Shift of the Fermi level because of (a)16 The above mechanisms fit materials having the energy gap. EE from metals is supplied perhaps from their surface oxides having properties of insulators or semiconductors.15 For the ionisation mechanism Tmax corresponds to the ionisation potential of the electron trap.18 There are experimental evidences15 that
I EE (T ) ~
dC dT
(8)
where C – concentration of relaxing imperfections. In fact the total emitted charge (Q) is directly proportional to C at Ti.15,16 An activation energy (Er) of relaxation for the simplest first order annealing reaction may be estimated from (8)15,16: Er = - kT ln [IEE(T)/N(T))]
(9)
where the k-Boltzmann constant, Tf
N (T )
³I
T
EE
(T )dT
(10)
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
179
In a case of semiconductors the EE spectrum is contributed by annealing of point type complexes of defects supplying single imperfections. Such the reaction is represented by the up-going branch of the spectrum (Ti < T < Tmax ). 16 For the first order reaction its activation energy (Eact) is available from the equation (9)16. However T f should be replaced with Tmax .16 The down-going branch of the EE spectrum (Tmax < T < Tf ) corresponds to migration of the generated single imperfection to the surface of the specimen. An activation energy (Em ) of this process in a case of uniformly distributed single defects within a specimen may be estimated from the equation:16
Em
kT ln
I EE (T max) I EE (T )
d[
I EE (Tmax ) ] I EE (T ) dT
(11)
4.2. INSTRUMENTATION
The above conditions considered to detect PE and DE (when DE is applied jointly with EE) are related to EE too. However, the temperature of the specimen should not exceed the condition, when a significant thermoelectron emission current is provided. 4.3. APPLICATION CASES
4.3.1.
Electron traps at the interface
A thin film of Si3N4 (10 nm) was deposited on a SiO2 substrate. The speci-men was irradiated with weak electrons. The latter were supplied with differ-enced energy to fill in the traps in Si3N4 , Si3N4/SiO2 , consequently. TSE was detected after each step of radiation. The TSE current demonstrated significantly different behaviour, when electron radiation reached the Si3N4/SiO2 interface (Figure 10). This is in favour to apply EE to detect interfaces of the nanostructures. 4.3.2.
Concentration of implanted atoms
A crystalline Si was irradiated by As+ (50 keV) ions. Measured total emitted charge of PTSE was proportional to the delivered fluence of ions (Figure 11).16 The value of Q was sensitive to 10–6 atomic % of As atoms concentration.
Y. DEKHTYAR
180
Si3N4
8 IEE , arb. units
Si3N4/SiO2
4
0 0
300
600
o
T, C Figure 10. TSE of the multi layer system Si3N4/SiO2.19
lg Q, arb. units
2 1.8 1.6 1.4 1.2 1 10
11
12 13 14 15 -2 lg (Flunece), cm
16
Figure 11. Emitted charge of PTSE in dependence on As+ ions fluence supplied to Si.16
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
4.3.3.
181
Activation energy of point type defects annealing and migration
4.3.3.1. Annealing of complex of single defects Tetra vacancies (V4) were generated in a monocrystalline Si because of P+ ions (100 keV) radiation. An activation energy (Er) of complexes annealing was acquired from PTSE measurements (9).16 At a high values of delivered fluence (>5.1014 cm–2) Er decreased because of interactions between V4 (Figure 12).
0.8
Er , eV
0.6 0.4 0.2 0 13.5
14
14.5
15
lg (Fluence), cm-2 Figure 12. Behavior Er of Si tetra vacancies in dependence on P+ ions fluence.16
4.3.3.2. Migration of single defects Single vacancies were generated in a monocrystallne Si owing to dissociation of the vacancy contending defect complexes. The latter were induced due to electron radiation (5 MeV). A activation energy (11) of the liberated vacancy decreased at the fluence level >4.1015 cm–2 (Figure 13), that was stipulated by interference of imperfections.16 From the experimental data, as shown in Sections 4.3.3.1 and 4.3.3.2, it is demonstrated that the photothermostimulated emission instrument can be used to estimate: (a) A thermodynamically non equilibrium concentration of defects (b) A threshold of the concentration of imperfection that provides their interaction (c) Annealing and migration activation energies of imperfections.
Y. DEKHTYAR
182
lgEm, [Em]=eV
14.8
15
15.2 15.4 15.6 15.8 16
16.2
0 -0.5 -1 -1.5 -2 lg (Fluence), cm-2
Figure 13. Em of Si single defects in dependence on electron fluence.16
5. Conclusion The review demonstrates that PE, DE and EE are capable to supply sensitive, accurate, and non-contact technique for the needs of nanotechnologies. The techniques are directed to measure the electron work function (to characterise surface potential), distribution of the electron states density (PE); energy gap and charge relaxation (DE); concentration of the point type imperfections, their annealing and migration (EE). ACKNOWLEDGMENTS
The author graciously acknowledges figures and contributions provided by the Author’s colleagues Mr. V. Noskov, Riga Technical University, Latvia, Dr. A. Balodis, Riga Technical University, LV, and Mr. V. Noskov, Riga Technical University, LV.
PHOTO-DUAL- AND EXO-ELECTRON EMISSION
183
References 1. Froitzheim H. (1977), Electron Spectroscopy for Surface Analysis. Ed. H. Ibach, Springer, New York, 315 p. 2. Fomenko V.S. (1981), Emission Properties of Materials. Naukova Dumka, Kiev, 338 p. (in Russian). 3. Kinding N.B., Spicer W.E. (1963), Band structure of cadmium sulphide photoemission studies, Phys. Rev. 138: A561–A576. 4. Dobrecov L.N., Gomoyunova M.V. (1966), Emission Electronics. Nauka, Moscow 564 p. (In Russian). 5. Kane E.O. (1962), Theory of photoelectric emission from semiconductors, Phys. Rev. 127(1): 131–141. 6. Pihtin A.N. (1983), Physical Basics of Quantum Electronics and Optoelectronics. Vsishaya shkola, Moscow, 304 p. (in Russian). 7. Dekhtyar Yu. D., Vinyarskaya Yu. A. (1994), Exoelectron analysis of amorphous silicon, J. Appl. Phys. 75(8): 4201–4207. 8. Dekhtyar Yu.D., Sagalovich G.L. (1988), Exoelectron emission of monocrystalline silicon and its practical application, Proc. USSR Academy of Sci., Physical Ser. 52: 1611–1613. (in Russian). 9. Polyakova A.L. (1979), Deformation of Semiconductors and Semiconductor Devises. Energiya, Moscow, 168 p. (in Russian). 10. Sagalovich G.L., Balodis A.Ya., Dekhtyar Yu.D. (1984), Emission test of mechanical tensions in surface layers of silicon, Electron. Tech., Ser. 8. 5(110): 25–27. 11. Bystrov V., Bystrova N., Dekhtyar Yu., Filippov S., Karlov A., Katashev A., Meissner C., Paramonova E., Patmalnieks A., Polyaka N., Sapronova A. (2006), Size depended electrical properties of hydroxyapatite nanoparticles, in: IFMBE proceedings. V. 14. CD version. Springer, Seoul, pp. 3149–3150. 12. Arvin H., Bogucharska T., Dekhtyar Yu., Hill R.M., A. Katashev, Pavlenko A., Pavlenko I., Zakaria M. (2000), Electronic transitions and structural changes in bone, Latvian J. Phys. Tech. Sci. 6(S): 50–55. 13. Nesterenko B.A., Cnitko O.V. (1983), Physical Properties of Atomic Clean Surface of Semiconductors. Naukova Dumka, Kiev, 264 p. (in Russian). 14. Volkenstein F.F. (1987), Electron Processes on Semiconductors Surface During Chemosorption. Nauka, Moscow, 430 p. (in Russian). 15. Kortov V.S., Shifrin V.P. Gaprindashvili A.I. (1975), Exoelectron spectroscopy of semiconductors and insulators, Microelectronics 8: 28–49 (in Russian). 16. Dekhtyar Yu.D. (1993), Exoelectron Spectroscopy of Point Type Defects in Semiconductors. Riga Technical University, Latvia, Riga, 59 p. (in Russian). 17. Rosenman G.I., Rez I.S., Chepelev Yu.L., Angert N.B. (1981), Exoemission of defected surface of lithium tantalite, J. Tech. Phys. 51(2): 404–408 (in Russian). 18. Nassenshtein G. (1962), Electron emission from the solid state surface after mechanical treatment, in: Exoelectron Emission, Inostrannaya literature, Moscow, pp. 72–95 (in Russian). 19. Rosenman, M., Naich, M., Molotskii, Dekhyar Yu, Noskov V. (2002), Exoelectron emission spectroscopy of silicon nitride thin films, Appl. Phys. Let. 80(15): 2743–2745.
LASER INTERACTION WITH NANO-SPHERES: APPLICATIONS IN SUB-MICRON PARTICLES REMOVAL AND NANODOT ARRAY FABRICATION M. SENTIS1*, D. GROJO1, PH. DELAPORTE1, AND A. PEREIRA2 1 Lasers Plasmas and Photonic Processing Laboratory, CNRS – Universities of Aix-Marseille, Campus de Luminy – Case 917, 13288 Marseille Cedex 9, FRANCE 2 INRS-EMT, 1650 blvd. Lionel-Boulet, Varennes (Quebec) J3X1S2, CANADA
Abstract – Laser-assisted nanoparticle removal processes like selective particle ablation, mechanical ejection, local substrate ablation, explosive evaporation are considered by experimental and theoretical approaches. Experiments are based on the estimation of particle removal efficiencies and thresholds depending on material and size of particles. Afterwards, a novel and efficient photonicbased method to synthesize porous alumina membranes (PAMs) and subsequent metal nano-dot arrays on various substrates, based on near laser field enhancement by nano spheres, is reported.
Keywords: Laser nano-particle interaction, near-field enhancement, nanodot arrays.
1. Laser-Assisted Nanoparticle Removal Processes 1.1. INTRODUCTION
Due to the emerging nanotechnologies, new cleaning solutions providing the ability to remove nanometer defects from material surfaces are now required to ambition a reliable fabrication of nanoscale devices. In this context, the dry
______ *
To whom correspondence should be addressed: M. Sentis, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
185
186
M. SENTIS ET AL.
laser cleaning (DLC) technique might be consider as a potential approach. In this method, the particles are ejected from the surfaces by the illumination of the polluted materials with nanosecond laser pulses. Some studies have already demonstrated that this technique is efficient for specific applications1 and allows eliminating particles with size below 100 nm.2,3 However, there is still a lack of knowledge of its underlying mechanisms. This may be explained by the complexity of the processes which can be combined in the studied situations. Among them, the adhesion and the laser-matter interactions involved in the study of particles at the nanometer scale exhibit singular characteristics. The literature on DLC, for more than 15 years now, shows that the removal of particles results from a combination of at least three mechanisms: (i) the mechanical force (inertia or elastic) exerted on the particles while they move with the rapid thermal expansion of materials,4,5 (ii) the local substrate ablation due to the enhancement of the laser light which can be observed underneath small particles,6,7 (iii) the explosive evaporation of the adsorbed humidity from the air at the interface between the particles and the surface of the substrate.8,9 On the basis of this latter contribution which has been evidenced recently, we may consequently wonder how dry the so-called dry laser cleaning technique is. Indeed, as we shall see later on, the humidity can play a major role on the physical processes occurring in the experiments. For this reason, we extensively investigated this aspect. The interpretation of the results of these experiments contributes to a better understanding of the humidity-based adhesion and laser-assisted removal processes involved in experiments with submicrometer particles. 1.2. EXPERIMENTAL DETAILS
Laser-induced particle ejection experiments were carried out with an ArF (Olas = 193 nm) excimer laser delivering pulses of 15 ns duration and 300 mJ energy. The samples were irradiated by single laser pulses in residual pressure conditions to avoid particle redeposition. The image relay of a metallic mask (using a lens) permits a near-uniform irradiation (S = 2 u 1.5 mm2) of the target materials. Samples were silicon (Si) (100) substrates on which 250 nm radius SiO2 spheres were spin-coated. A density of #105particles/cm2 was obtained with more than 90% of isolated particles. The samples were placed in a vacuum chamber allowing to reach a residual pressure of 10–4 Pa. Inside the chamber, the stainless steel sample holder was back-side equipped with a halogen lamp to adjust the temperature of the irradiated target up to 750K. Thus, the substrates can be subjected to different degassing step between each of the experiments and the water meniscus inevitably formed in a humid medium at the interface between the particle and surface can be gradually eliminated. In our experiments, the water meniscus may be formed due to the humid particle deposition process
LASER INTERACTION WITH NANO-SPHERES
187
or condensed from ambient (i.e. air) humidity before the sample was entered into the chamber. Software analysis of optical micrographs (dark field) of the laser cleaned substrates was used to determine the Particle Removal Efficiency (PRE ) as a function of the experimental conditions (laser energy, degassing step). The values of PRE correspond to the statistic fraction of particles which was removed by the irradiation i.e. PRE = 1–n/n0 where n and n0 are respectively the densities of particles on the surface before and after the laser irradiation. Fast imaging of the ejected particle clouds was also performed with a fast intensified CCD (ICCD) and a forward scattering-based probing technique. The temporal and spatial resolutions of these measurements were ~1 μs and 5 μm, respectively. A complete description of the experimental apparatus can be found in Ref.10 On the basis of this diagnostic, time-of-flight analyses of the ejected particles allow to determine the detachment velocity of the particles. The measurement of the initial velocity, which can be considered as an experimental signature of the different ejection mechanisms,11 is used to distinguish between them and to identify their respective validity domains. 1.3. RESULTS AND DISCUSSIONS
1.3.1. Time-of-flight (T.O.F.) analyses Detailed information on the particle cloud ejection dynamics is obtained from the distribution of the scattering signal from a CW probe beam as a function of the distance from target I(z) for different delays t between the laser pulse and the ICCD observation gate of 1 μs duration. The distance between the particles (#30 μm) being large compared to the probe wavelength (Oprobe = 532 nm) the measured signal is directly proportional to the particle density. The I(z) curves which were captured for a time delay t = 8 μs and for three different laser fluences above the particle removal threshold fluence (Fth # 130 mJ cm–2), are shown on Figure 1. Each profile was fitted with a Gaussian function G F ,t ( z ) . Consequently, according to the consistency of the fits, the complete velocity distribution f F (v ) of the ejected particles for a given laser fluence F is also well described by a Gaussian function which can be deduce from the profiles that are captured at a time delay t. This distribution is given by the expression: f F ( v ) v tG F ,t ( vt ) . The profiles shows that very different behaviors are observed as a function the laser energy deposited onto the polluted substrates. In particular, Figure 1b shows that under appropriate conditions, the ejected particle cloud is split in two distinct components with different velocities. According to the position of the detected species zmax at about 200 and 650 μm for each of these components, the corresponding propagation velocities of the particles are respectively 25 and
188
M. SENTIS ET AL.
81 m s–1. This observation means that, although we study calibrated systems, the particle detachment results from a competition between at least two different mechanisms in this given situation. For lower laser fluences (Figure1a), only the slow component is observed. For larger laser fluences (Figure1c) only the fast one remains. These observations evidenced the existence of two regimes where only one removal mechanism is responsible for the particle ejection.
Figure 1. Spatial repartition of the ejected particles (given by the forward scattering signal) with respect to the distance from target for a time delay t of 8 μs after the irradiation of the target with the nanosecond laser pulse. The profiles (and their Gaussian fitting curves) are shown for three different laser fluences: (a) Flas = 270 mJ cm–2, (b) Flas = 410 mJ cm–2 and (c) Flas = 550 mJ cm–2.
To analyze the different regimes in more details, Figure 2 shows the determined velocities of the particles as a function of laser fluence. We note that the analyses were possible for laser fluences very close to the removal threshold fluence Fth. Above Fth, particles are ejected with a characteristic velocity which gradually increases from 7.6 to 36.8 m sí1. In the range 300–500 mJ cmí2, while the scattered intensity from the particle cloud decreases a faster component appears. This transitory behavior evidenced that some particles are now removed on the basis of a physical process of different nature to that observed for the low laser energy domain. For larger fluences, only the fast component remains and particles acquire considerably more speed when increasing Flas. Propagation velocity of the particle cloud linearly increases from 66 to 231 m sí1 when fluence is varied from 415 to 680 mJ cmí2. Thus, analyzing the characteristic velocities of the ejected particles, we determined two laser energy domains where two different detachment processes dominates but that still have to be identified.
LASER INTERACTION WITH NANO-SPHERES
189
Figure 2. Velocities of the ejected particles measured by fast imaging of the scattering clouds as a function of the laser fluence. The determined velocities (above the particle removal threshold fluence) are compared with a calculation of the surface expansion dynamics for similar conditions.
Among the possible mechanisms, we first examined the likelihood of a mechanical ejection resulting from the rapid thermal expansion of the materials. For submicron particles irradiated with nanosecond laser pulses, the elasticity of particles can be neglected and the contaminants move as a whole, following the thermally-induced displacement of the surface of the substrate.10 Figure 2 presents the results of a calculation of the maximal surface velocity which is reached in the vicinity of the particles. The model consists of a numerical resolution of the 3D thermo-elasticity problem, taking into account the focusing power of the SiO2 spheres. This aspect is described using the Lorentz-Mie theory. Indeed, we already showed that, the surface of the substrate is locally illuminated by a bright spot with a Gaussian-like shape and a full width at half maximum of 160 nm12 underneath each of the spheres. The subsequent hot spots enhance the expansion dynamics of the materials (in comparison with a uniform irradiation with clean surface). However, as shown on Figure 2, the expansion velocity of the surface remains two orders of magnitude below the experimentally measured particle ejection velocities. This demonstrates that the dominant removal mechanism in these experiments is not the rapid thermal expansion of materials.
190
M. SENTIS ET AL.
1.3.2. Self-limited local ablation of the surface After the irradiation experiments, the cleaned substrates were carefully analyzed by means of SEM. For Flas corresponding to the presence of the fast component (i.e. Flas > 300 mJ cm–2), we successfully observed ablated craters in places where originally located the particles. The hole sizes were in the order of 135 nm in diameter for laser fluences less than 500 mJ cmí2. This size is relatively consistent with the laser energy distribution given by the Lorentz-Mie theory solution.12 It demonstrates that, in this fluence domain, the ejection of the particles results from the local ablation of the substrate which is induced by the optical near-field enhancement underneath the particles. This approach is not relevant in a context of extreme cleaning but could be used as a technique of nanomanufacturing. 1.3.3. Humidity dependence of the laser-induced removal of particles As stated previously, the nature of the interactions between, the laser light, the submicrometer particles and the surface may be strongly modified by the water meniscus at the interface between the particles and surface. The influence of this aspect on the particle removal with moderated laser fluences is analyzed in this section. To better understand the role of the humidity, we repeated PRE measurements as a function of laser energy after successive periods during them the polluted substrates were conserved under vacuum and at relatively high temperature. Consequently, the water meniscus under the particle was progressively reduced with respect to time in our experimental investigations. Figure3 presents the evolution of the particle removal threshold when the humidity is progressively reduced. The last value which corresponds to the driest substrate (as dry as we could get) is close to the local substrate ablation threshold reported in the previous paragraph. It means that the removal of the water meniscus leads to the disappearance of the low fluence regime mechanism which was discriminated in the TOF analyses. Thus, for dry substrates, the only way to eject particles is the substrate ablation. From this result, we propose that the explosive evaporation of the water meniscus which is induced by the laser irradiation8,9 is the particle removal mechanism in this regime. The measured ejection velocities (see Figure 2) strengthen us in our proposition. Indeed, the velocities found in the literature relating the ejection (ablation) of thin films of condensed water on metal surfaces by nanosecond laser pulse irradiations are also in the range 10–40 m s–1 for similar conditions.13 In this approach, the cleaning is still based on the momentum transfer from laser-ablated species (water) but does not require the consumption of the substrate.
LASER INTERACTION WITH NANO-SPHERES
191
Figure 3. Dependence of the particle removal threshold fluence as function of time during an experiment where the humidity trapped on the substrate is gradually reduced.
It would be over-simplified to claim that the presence of humidity transforms only the laser-matter interaction processes in our experiments. Indeed, the nature of the adhesive interaction between the particle and the surface is also strongly modified by the water meniscus which is formed between them. In particular, this aspect needs to be taken into account to explain the nonmonotonous variation of the particle removal threshold shown in Figure 3. The observation of a minimum as a function of time evidences that it exists an amount of water which optimizes the removal process. This behavior is confirmed by the ejected particle velocity measurements performed before and after that the samples was left during 24 h in a 10–2 Pa atmosphere (corresponding to the first and second points in Figure 3). Figure 4 shows that this gentle degassing step leads to a 20% decrease of the particle removal threshold Fth and a significant increase of the ejection velocity for laser fluences close to Fth It reveals that, when the humidity is slightly decreased the particle ejection becomes easier (in terms of laser energy); hence it corresponds to a better compromise between the adhesion and the cleaning force. In principle, the adhesion of submicrometer objects to surfaces is dominated by van der Waals forces. A typical value of these forces for a 250 nm silica particle is #100 nN.14 However, the presence of a water meniscus may provide an additional contribution to the adhesion by the capillary force. The capillary force can be estimated by using the standard approximation Fc = 4SJR where
192
M. SENTIS ET AL.
Figure 4. Measurement of the velocities of the ejected particles as function of the laser fluence. After that the humidity was slightly reduced by a degassing step of 24 h under a pressure P = 10–2 Pa, a decrease of the particle removal threshold fluence (# 20%) and a significant increase of the kinetic energy of the ejected particles are observed.
J is the surface tension of water (7.28 u 10–4N cm–1).15,16 For R = 250 nm, this (#200 nN) is not negligible and can easily dominate the van der Waals force. Surprisingly, on the basis of this often used approximation, the adhesion force is independent on the relative humidity RH, and consequently, on the amount of water at the interface between the particle and the surface. However, recent numerical simulations16,17 and capillary force measurements (based on Atomic Force Microscopy)18 demonstrated that while this approximation is viable for particles above 1 μm, its validity is limited for particles as small as the particles we studied. In particular, O. Pakarinen et al.16 developed a model to numerically calculate the exact (non-circular) meniscus profile from the Kelvin equation and, the subsequent capillary force in real environments. They showed that for the geometry “sphere on flat surface”, the standard approximation remains acceptable for RH # 1. However, the adhesion vanishes gradually with the humidity (with slopes increasing as the size of the particles decreases). Consequently, the non-monotonous variation of the particle removal threshold given in Figure 3 is explained by the competition between two humidity-based forces: the capillary adhesion force and the cleaning explosive evaporation (of the water meniscus) force. In the experiments, at the very beginning, when the humidity is gently reduced, the adhesion force is severely decreased but there is still enough water trapped under the particles to eject them efficiently
LASER INTERACTION WITH NANO-SPHERES
193
by the laser-produced explosive evaporation of this humidity. Thus, the removal threshold is reduced as a result of an optimization of the environment. After this time, while the sample is degassed further, the removal threshold increases progressively as a result of the reduction of the quantity of water possibly ablated and the subsequent dragging force exerted on the particles. This is observed until the quantity of water becomes sufficiently low to make impossible the ejection of the particles by momentum transfer from the evaporated water. Then, the only way to remove the particles is based on the local ablation of the substrate. 1.4. CONCLUSION OF LASER NANOPARTICLE REMOVAL
We demonstrated that the removal of submicrometer particles in nanosecond laser experiments results from a competition between two mechanisms based on the momentum transfer from laser ablated species to the particles. The local ablation of the substrate resulting from the near-field enhancement underneath the particles provides the cleaning force for large fluences. At low laser fluence, we observed that the high efficiency of the direct illumination laser cleaning technique is damage-free. However, the experimental results reveal a large dependence on the amount of humidity inevitably trapped at the interface between the particles and surface. The humidity plays a major role in the cleaning force as well as in the adhesion force exerted on the particles. This humidity dependence may explain the lack of constancy of the results of the numerous experimental studies on DLC performed in non-controlled atmospheres for more than 15 years now. The explosive evaporation of the trapped water is actually the cleaning mechanism in our studied systems. Consequently, the so-called DLC is a humidity-based process. Thus, it is rather similar to the Steam Laser Cleaning (SLC) where a liquid film is deposited on the surface before the laser irradiation. In comparison with SLC, the advantage of our approach is that the water is located only where the water is needed (at the interface between the particle and the surface). However, to optimize the technique, the relative humidity must be controlled and adjusted to lead to the optimum quantity of water trapped under the particles. 2. Preparation of Metal Nanodot Arrays by Near-Field Enhancement In this paragraph we exploit the local ablation of substrates resulting from the near-field enhancement underneath the particles to synthesize Porous Alumina Membranes (PAM) and subsequent metal nanodot arrays on various substrates.
194
M. SENTIS ET AL.
2.1. LASER PROCESS
The fabrication process is outlined in Figure5. A monolayer of self-assembled spheres is formed onto a thin alumina (Al2O3) film, which was previously coated on a silicon substrate by means of Pulsed Laser Deposition (Figure 5a). Then, pores are optically drilled in the Al2O3 film by particle-assisted near field enhancement. This is accomplished through illumination of the spheres with a single nanosecond laser pulse at the wavelength Olas = 193 nm. This leads to the local removal of the 20 nm thick Al2O3 film under each sphere. Since the spheres are arranged in a hexagonal array at the surface of the substrate, the aluminum oxide film is decorated with an ordered arrangement of holes. (Figure 5b). Using this Laser Fabricated PAM (LF-PAM) as a mask for the deposition of metal (Figure 5c), a series of ordered metal nanodots are formed at the surface of the substrate upon dissolution of the alumina layer (Figure 5d). As we shall see later on, we demonstrated the controllability of this schematic approach by forming an ordered array of gold nanodots.
Figure 5. Scheme of the nanodot array fabrication method.
Numerous studies devoted to nanostructuring by particle-assisted near-field enhancement deal with 1-μm PolyStyrene (PS) spheres. However, Piparia et al.19 demonstrated recently that, in the near UV region (<300 nm), the interaction process exhibits a self-limited character. In this spectral region, short wavelengths are strongly absorbed in most polymer materials. To avoid this limitation, we choose to use silica (SiO2) spheres. Also, spheres with radii R = 250 nm were chosen to prepare nanodot array with a larger surface density (#5 u 108dots/cm2) than accessible with 1-μm spheres. The most critical aspect of this approach is the controlled and selective nanodrilling of the alumina film. Al2O3 is a very low absorption material largely used for optical coatings in the UV region. At 193 nm, the reflectance of Al2O3
LASER INTERACTION WITH NANO-SPHERES
195
is R # 0.09 and its optical penetration depth is several orders of magnitude larger than the film thickness used in this study.20 Consequently, the laser light is mainly absorbed by the Si substrate. 2.2. DISCUSSION AND RESULTS
We carefully analyzed, by means of Atomic Force Microscopy (AFM) and Scanning Electron Microscopy (SEM), the effect of the pulse energy on the “spheres-Al2O3-Si” stacks. For laser fluence as low as 200 mJ cm–2, the silica spheres are ejected from the surface. According to the laser cleaning mechanism described in part 1. Also, craters of few nanometers deeps are formed at the surface of the Al2O3 film (AFM observations), showing that this mechanism slightly damages the surface. At laser fluences close to 340 mJ cm–2, the silica spheres and the aluminum oxide film are removed. The typical diameter of the ablated craters is 130 r 15 nm. The bottom of the crater is flat, indicating that the underlying Si substrate is not significantly affected by the process. In this situation, the ablation of the Al2O3 layer is due to the spallation mechanism induced by the confinement of the deposited energy at the alumina-silicon interface.21 For laser fluences above 400 mJ cm–2, the Si substrate is damaged. In that regime, sombrero-like structures having a diameter of about 280 nm are observed. The center of these structures protrudes the original surface of the (undamaged) aluminum oxide film by about #8 nm. Recent studies reported similar phenomenon at larger size scale with pure silicon substrates.22 The formation of such nanostructures relates to inward flows of the melted materials caused by the inhomogeneous temperature distribution in the heated layer.19,23 Figure 6 shows AFM and SEM images of gold nanodot arrays created on Si substrates by performing LF-PAM at intermediate laser fluence (340 mJ cm–2). Software analyses of the images allow us to characterize the size and shape of the gold nanostructures. From the object area Sdot, the average size of the structures Rdot = (Sdot/S)1/2is 48.2 r 2.6 nm (rSD). The circularity, given by the ratio of the square of the perimeter to 4S Sdot, is 1.47 r 0.1. Each dot has a uniform height (20 nm) and spacing (500 nm) to its neighbor. The height determined by the AFM profile (Figure6c) is consistent with the original Al2O3 film thickness. 2.3. CONCLUSION ON LASER PROCESSING TECHNIQUE FOR METAL NANO DOT ARRAY FABRICATION
We demonstrated that the fabrication of metal nanodot arrays is feasible by using exclusively laser processing techniques. The size of the dots thus generated is the same as that of the holes drilled on the PAM. This approach is quite
196
M. SENTIS ET AL.
Figure 6. SEM image (a) AFM image (b) and depth profile (c) of gold nanodots created on silicon substrates by the LF-PAM based process.
general and can be applied to wide range of metals, semiconductors or complex oxides. This technique can be extended to spheres of variable diameter, thus allowing the distance between the nanodots to be varied for a defined application. However, Mie theory calculations show that the improvement in resolution (size of the near-field enhancement region) induced by the reduction of the sphere size will not be dramatic.24 Focusing at a scale of a few nanometers is still a challenge that simple particle-masks will not solve. To improve resolution
LASER INTERACTION WITH NANO-SPHERES
197
and accuracy, one of the possible extensions of the technique will be to exploit the deterministic and non-linear character of the damage processes involved in femtosecond laser machining.25
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.
Grojo, D., Cros, A., Delaporte, P., and Sentis, M. (2006) J. Phys. IV 138, 191–201. Lizotte, T., and O’Keeffe, T. (1996) Proc. SPIE 2703, 279–287. Wu, X., Sacher, E., and Meunier, M. (1999) J. Appl. Phys. 86, 1744–1748 Luk’yanchuck, B., Wang, Z., Song, W., and Hong, M. (2004) Appl. Phys. A 79, 747. Arnold, N. (2003) Appl. Surf. Sci. 15, 208. Mosbacher, M., Munzer, H., Zimmermann, J., Solis, J., Boneberg, J., and Leiderer, P. (2001) Appl. Phys. A 72, 41. Arnold, N., Schrems, G., and Bauerle, D. (2004) Appl. Phys. A 79, 729. Vereecke, G., Rohr, E., and Heyns, M. (1999) J. Appl. Phys. 85, 3837. Mosbacher, M., Bertsch, M., Munzer, H., Dobler, V., Runge, B., Bauerle, D., Boneberg, J., and Leiderer (2002) P. Proc. SPIE 4426, 308. Grojo, D., Cros, A., Delaporte, P., and Sentis, M. (2006) Appl. Phys. B 84, 517–521. Grojo, D., Cros, A., Delaporte, P., and Sentis, M. (2006) Proc. SPIE 6261, OE. Pereira, A., Chaker, M., Guay, D., Grojo, D., Delaporte, P., and Sentis, M. Appl. Phys. Lett., submitted. Zapka, W. (2002) Laser Cleaning, B. Luk’yanchuk (Ed.), World Scientific, New Jersey, 23–48. Grojo, D. (2006) Ph.D. thesis/University Marseille, 17–32. Cappela, B., and Dietler, G. (1999) Surf. Sci. Rep. 34, 1–104. Pakarinen, O., Foster, A., Paajanen, M., Kalinainen, T., Kataienen, J., Makkonen, I., Lahtinen, J., and Nieminen, R. (2005) Model. Simul. Mater. Sci. Eng. 13, 1175–1186. Jang, G., Schatz, C., and Ratner, M. (2004) J. Chem. Phys. 120, 1157. Zitzler, L., Herminghaus, S., and Mugele, F. (2002) Phys. Rev. B 66, 155436. Piparia, R., Rothe, E., and Baird, R. (2006) Appl. Phys. Lett. 89, 223113. French, R., Mullejans, H., and Jones, J. (1998) J. Am. Ceram. Soc. 81, 2549. Xia, Z., Shao, J., Fan, Z., and Wu, S. (2006) Appl. Opt. 45, 8253. Wysocki, G., Denk, R., Piglmayer, K., Arnold, N., and Bauerle, D. (2003) Appl. Phys. Lett. 82, 692. Georgiev, D., Baird, R., Avrustsky, I., Auner, G., and Newaz, G. (2004) Appl. Phys. Lett. 84, 4881. Ahn, C., and Rabe, K. (2004) J. Mater. Sci. 303, 488. Joglekar, A., Liu, H., Meyhofer, E., Mourou, G., and Hunt, A. (2004) PNAS 101, 5856.
CLEAN FOSSIL FUELS: ADVANCED MEMBRANE REACTORS T. TRAN, K. STOITSAS, AND J. SCHOONMAN* Department DelftChemTech-Energy, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
Abstract – The development of processes with high efficiency and low carbon emissions seems to be more necessary than ever in order to face the temperature rise of our planet. In this paper preliminary results are presented of the development of membrane reactors which will be used in the steam reforming reaction so as to produce pure Hydrogen with higher efficiency than the conventional reactors. The sol-gel technique is used for the development of porous ceramic membranes that will separate Hydrogen from Carbon Monoxide or Dioxide due to the molecular sieving effect. Atomic Layer Deposition will be used for reducing the pore size of the primary membranes made by the sol-gel technique.
Keywords: Greenhouse effect, hydrogen production, carbon membrane reactors, ceramic membranes.
1. Introduction In 2003 the Nobel Laureate Richard Smalley1 presented his view on the humanity’s top-ten problems for the next 50 years. On a prominent first place Energy was presented. Every day the world uses about 85 million barrels of crude oil and nearly about 10 billion cubic meters of natural gas and demand is rising. Energy scenarios anticipate that by 2030 the total global energy demand will grow by close to 50% and by 18% in Europe – EU and non-EU. These numbers do not take into full account the rising economies in India and China. To date, fossil fuels cover about 80% of the present energy demand, with about 7% nuclear, and about 13% renewable energies, like, for example, solar energy, wind energy, biomass, hydrothermal, geothermal, and tidal wave energy. Especially the utilization of fossil fuels causes the emission of the greenhouse
______ *
To whom correspondence should be addressed: J. Schoonman, email:
[email protected]
A. Vaseashta and I.N. Mihailescu (eds.), Functionalized Nanoscale Materials, Devices and Systems. © Springer Science + Business Media B.V. 2008
199
200
T.TRAN, K. STOITSAS, AND J. SCHOONMAN
gas (GHG) carbon dioxide, and the emission of this GHC is less, when using natural gas, instead of oil, and, in particular, instead of coal. To date, the total annual CO2 emission to the atmosphere equals 3.5 Gt Carbon2 and global warming is related to the increase of the CO2 concentration. A simplified explanation of the greenhouse effect is as follows. The solar radiation passes through the clear atmosphere and some solar radiation is reflected by the Earth’s surface and the atmosphere, but most radiation is absorbed by the Earth’s surface and warms it. Infrared radiation is emitted from the Earth’s surface and some of the infrared radiation is absorbed and re-emitted by water vapor and greenhouse gases. The effect of this is to warm the surface and the lower atmosphere.3 In fact, the prediction of climate change due to human activities began with a prediction made by the Swedish chemist Svante Arrhenius in 1896. Arrhenius took note of the industrial revolution then getting underway and realized that the amount of carbon dioxide being released into the atmosphere was increasing. He predicted that if atmospheric carbon dioxide doubled, the Earth would become several degrees warmer.3 Improve energy efficiency, switch to low-carbon fuels, switch to renewable energy sources where possible and sequestration of CO2 derived from fossil fuels are options for reducing CO2 emissions into the atmosphere and attract widespread attention. In this respect hydrogen is an attractive energy carrier and in a transition towards the sustainable application of fossil fuels and towards renewable energy sources, clean fossil fuels and renewable hydrogen, i.e., hydrogen prepared using renewable energy sources will play an important role. Clean fossil fuels can be obtained using advanced membrane reactors and in this paper materials and processes for advanced membrane reactors will be presented. 2. Production of Hydrogen Hydrogen is the third most abundant element, but it is only available in bound form. It can be produced from many sources as well as a variety of technologies: natural gas by reforming and partial oxidation, coal by gasification and in-situ gasification with CO2 sequestration, and nuclear energy by electrolysis (water splitting) or for a thermo-catalytic process. Solar energy and wind energy can also be used to electrolyse water and together with gasification of biomas provide renewable hydrogen sources. The photo-electrochemical production of hydrogen has been presented by Schoonman et al.4 Hydrogen from these processes is sustainable and renewable. Hydrogen is not widely used today but it has great potential as an environmentally clean energy carrier for the near future. Energy is generated by burning directly hydrogen with oxygen producing only water as shown in reaction 1. Using hydrogen and oxygen in a fuel cell generates electrical energy.
CLEAN FOSSIL FUELS
H2(g) + ½ O2(g) l 2H2O(l) 286 kJ/mol
201
(1)
A comparison with fuels, like coal, brown coal, wood, gasoline, diesel, methanol, and natural gas reveals, that hydrogen as an energy carrier has a highenergy content per unit mass of 119.9 kJ/g and a low-energy content per unit volume of only 10 kJ/l. Therefore, the efficient storage of hydrogen is an important scientific and engineering challenge. The key reactions for the industrial chemical production of hydrogen from fossil fuels are reactions (2) and (3). 750-1,000 °C
CH4 + H2O l CO + 3H2 catalyst
ǻǾ°298 = 206 KJ/mol
(2)
ǻǾ°298 = – 41 KJ/mol
(3)
300–500°C
CO + H2O l CO2 + H2 catalyst
According to Le Chatelier’s principle, by removing H2 from the gaseous product, the equilibrium can be shifted to the product side. Effectively, this can lower the reaction temperature to 400–500°C and higher efficiencies 85–90 instead of 75% and improves the purity of the hydrogen.5 Other approaches to obtain pure hydrogen are the removal of the formed CO2 and hydrotalcite materials and Ionic Liquids are being studied for CO2-selective membranes. The removal of hydrogen or carbon dioxide is studied within the Stanford Global Climate and Energy Project (GCEP), which focuses on the fundamental and pre-commercial research on technologies that would foster the development of a global energy system with low greenhouse gas emissions. 3. Membrane Reactors McLeary et al.6 have published a recent review paper on zeolite films, membrane and membrane reactors, focussing on the progress in the field and the prospects. These authors have presented the classification of membrane reactor configurations according to membrane function and location, based on the recent literature. The authors describe the catalytic membrane reactor (CMR), the packed-bed membrane reactor (PBMR), the catalytic non-permselective membrane reactor (CNMR), the non-permselective membrane reactor, and the particle-level membrane reactor. The applications of membrane reactor processes, the reactor configuration, the type of membrane, and the operating conditions for the reaction types, i.e., based on dehydrogenation reactions, hydrogenation reactions,
202
T.TRAN, K. STOITSAS, AND J. SCHOONMAN
oxidation reactions, and examples of organic syntheses. In addition, examples of scarce industrial membrane reactor processes are presented.6 Inorganic membranes can be classified into two types, i.e., dense and porous membranes. Dense membranes prepared from Pd-Ag alloys, or oxide-ion conducting, and mixed ionic-electronic conducting perovskites only allow for hydrogen or oxygen, respectively. In the perovskites oxygen permeates via oxideion vacancies. While such membranes exhibit extremely high selectivities, they have limited permeabilities, but recent research has resulted in fluxes within reach of targets. Microporous silica membranes have proven to be promising for molecular sieving applications. Precise control of the pore size allows for separation on the basis of size by molecular filtration, or “sieving”. It is known that the hydrophilic surface of silica membranes leads to pore blocking by H2O and especially at higher temperatures this can seriously degrade the microstructure.7 Therefore, the microporous silica membranes are not stable under the steam reforming reaction of methane to separate hydrogen from CO. Here to, we have designed a reactor based on alumina and silica-alumina composite microporous membranes, which combine the steam reforming reaction, or the water-gas shift reaction with in-situ hydrogen separation in one system as illustrated in Figure 1. This type of membrane reactor will be expected to remove selectively hydrogen from the reactions and to achieve high conversion of methane at lower
Reaction catalyst
Nano-structured Layer Atomic Layer Deposition (ALD) (<0.32 nm) Intermediate Porous Layer Sol-Gel (1-5 nm)
Macroporous Support (D- Al2O3)
Figure 1. Concept of membrane reactor for hydrogen separation.
CLEAN FOSSIL FUELS
203
temperatures. The purpose of the research is to develop a microporous membrane material based on J-alumina prepared by the sol-gel route, followed by Atomic Layer Deposition. This technique allows depositing a monolayer or a multilayer inside the pores, thereby reducing the initial pore diameter to the molecular dimension of hydrogen. In this way, the separation is mainly based on the differences of the molecular diameters of the molecules involved as shown in Figure 2.
Figure 2. Kinetic diameter of H2, CO2, CO, and CH4.
4. Synthesis of Mesoporous Membranes Mesoporous membranes are developed using the sol-gel technique. Sol-gel processing is one of the most interested methods for the preparation of porous ceramics membranes. This method involves the transition of a system from a colloidal or polymeric solution (sol) into a gelatinous substance (gel). There are two kinds of sol-gel routes according to the state of the sol: colloidal route and polymeric route as illustrated in Figure 3. In general, the sol-gel route involves the hydrolysis-condensation reactions of metal alkoxides M-OR (M = Si, Ti, Zr, Al; OR = OCnH2n+1) precursors dissolved in water or an organic solvent, leading the formation of a metal-oxobased macromolecular network. The overall hydrolysis-condensation reactions of the sol-gel route can be presented as follows: Hydrolysis: M-OR + H2O ĺ M-OH + R-OH Condensation: M-OH + M-OX ĺ M-O-M + X-OH Oxolation M-OH + M-XO+H ĺ M-O+H-M + X-OH Olation (X= H or R )
(4) (5) (6)
T.TRAN, K. STOITSAS, AND J. SCHOONMAN
204
Metal salt or metalorganic precursor
Colloidal route
media
polymeric route
Sol (polymeric)
Sol (colloidal)
membrane coating
Polymeric gel
Colloidal gel
Drying (hybrid organic-inorganic membrane)
Sintering (pure inorganic membrane)
Figure 3. Diagram showing two sol-gel routes used in ceramic membrane preparation.8
The synthesis of Ȗ-Al2O3 membranes proceeds through the colloidal route. Thus a colloidal boehmite sol is prepared by hydrolysis of aluminum-tri-secbutylate at a temperature above 80oC. Two litres of water are used per mole of alkoxide. In order to peptize the sol particles 0.07 mol HNO3 per mole alkoxide is added.After the addition of alkoxide, the sol keeps boiling in the open reactor for 30 min to evaporate most of the butanol that is formed. Finally the solution is kept at 90oC under reflux conditions for about 16 h in order to form a stable boehmite sol. The concentration of the final sol is 0.92 M.9–11 For the synthesis of unsupported systems small amounts of boehmite are poured into different petri dishes and subsequently dried for 4–5 days under ambient conditions. After drying, calcination follows at 600oC for 3 h with a heating rate of 60oC/h. Supported membranes are developed on the surface of macroporous Į-alumina supports that NGK INSULATORS LTD provided to us. 12 The support is plate shaped with a diameter of 25 mm and a thickness of 2 mm. It consists of three successive layers. The final layer on the top of which the Ȗ-alumina films are developed has a thickness of 10 ȝm with a pore size of 100 nm. The preparation of the supported membranes proceeds through the slipcasting mechanism by bringing the dry support into contact with the colloidal sol of boehmite (Figure 4). Capillary forces created from the porous support
CLEAN FOSSIL FUELS
205
Support
Capillary pressure
Boehmite sol
Gel formed on the top of the support
Figure 4. Schematic representation of the dipping process.
pull the liquid into the structure. The concentration of the boehmite particles increases on the surface of the support and gelation occurs. After the deposition, drying and calcinations is performed at high temperatures to form the final microstructure. In order to improve the rep reducibility of this method an amount of dissolved poly-vinyl alcohol (PVA) is added to the boehmite sol. The final concentration of PVA in the sol is 2.5 wt%. It is possible that during the dipping procedure pinholes are formed due to the presence of irregularities or defects on the support surface or due to the presence of air bubbles in the sol. Repeating the dipping, drying and calcination steps several times can repair this type of defects. Note that multiple dipping results in an increase in membrane thickness and can be used to modify this parameter. The pore size is the important parameter for the separation performance of the membranes. For this reason unsupported Ȗ-alumina membranes are characterized by nitrogen adsorption-desorption at –196oC to calculate the pore size distribution. The samples are initially out-gassed under vacuum and at a temperature of 150oC for 16 h so as to remove moisture from the structure. The nitrogen adsorption-desorption diagram is type IV, characteristic for mesoporous solids with the typical hysteresis13 (Figure 5). The specific surface area of the membranes is estimated at 235 m2/g using the BET (Brunauer, Emmett and Teller) equation at low relative pressures (P/Po <0.4) whilst the porosity of the membranes is estimated at 48%. The pore size distribution of the Ȗ-alumina membranes is estimated by the nitrogen desorption isotherm applying the Density Functional Theory (DFT) theory, that takes into account the interactions of nitrogen with the specific material.14 The mean pore diameter of the membranes is 5.5 nm with a rather broad pore size distribution of 3–8 nm.
T.TRAN, K. STOITSAS, AND J. SCHOONMAN
206
0.01
160
0.009
140
0.008
120
0.007
Dvc/Ddp (cm 3 /g/A0 )
Volume of adsorbed gas (cm3 STP/gr)
180
100 80 60 40
Ȗ-Al2O3 calcined at 600 C o Ȗ-Alfor 2O33calcined hr in airat 600 C for 3 hr in air
0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.005 0.004 0.003 0.002
o
20
0.006
0.001 0 1
Relative pressure P/Po
0
20
40
60
80
100
Pore size dp(A0)
Figure 5. Nitrogen adsorption-desorption isotherm and pore size distribution using DFT model for unsupported Ȗ-Al2O3 thin films.
Characterization of the supported Ȗ-Al2O3 samples was performed with the permporometry technique in order to determine the pore size distribution of these membranes. Permporometry is a characterization method that can provide information of the pore size and the existence of pinholes or cracks in the supported layers. In addition, by using permporometry the discrimination between the open pores, called “active” pores, and the dead-end pores, called “inactive pores”.15 Permporometry is based on the controlled stepwise blocking of pores by condensation of a vapor, present as a component of a gas mixture and the parallel measurement of the gas flux through the membrane (Figure 6). The relative pressure of water is controlled by the flow of Helium at the feed side. By applying a pressure drop of 200–500 mbar along the membrane, a mixture of He/H2O is forced to pass through the membrane. As the relative pressure of the condensable gas increases capillary condensation follows the adsorption mode blocking the pores for the diffusion of the non-condensable gas (Figure 7). The relative pressure at which pore filling starts depends on the radius of the capillary and can be calculated from the Kelvin relation (Eq. (7)).
ln(
P ) P0
2JV cos(T ) r RT
(7)
Where P is the actual vapor pressure, Po is the saturated vapor pressure, Ȗ is the surface tension, V is the molar volume of condensate gas, R is the universal gas constant, T the temperature, r is the radius of the capillary and ș is the contact angle of the liquid on the solid.
CLEAN FOSSIL FUELS
207
Figure 6. Permporometry separation unit.
Prel = 0, All pores are open
0 < Prel < 1, Small pores are filled with condensate
Prel = 1 All pores are blocked Figure 7. Blockage of the pores as a function of the relative pressure (Prel).
The flux of He through the Ȗ-Al2O3 membranes is shown in Figure 8 during the desorption and adsorption of H2O vapor. This typical flux vs. relative pressure plot can be divided into three main areas. At low relative pressures the flux of He decreases slightly. In this region all the pores are empty and available for transport but still the available space for the diffusion of He decreases due to monolayer adsorption of H2O. At pores
T.TRAN, K. STOITSAS, AND J. SCHOONMAN
208
8.E-06 Adsorption
He Permeance (mol/m2.Pa1.S1)
7.E-06
Desorption Multilayer adsorption, capillary condensation
6.E-06 5.E-06 4.E-06 3.E-06 Monolayer adsorption
2.E-06 1.E-06 0.E+00 0
1
2
3
4
5
Kelvin diameter (nm) Figure 8. Permporometry adsorption and desorption curve of Ȗ-alumina membrane with H2O used as adsorbate.
with size 2 nm < dp < 4 nm the flux of He decreases with relative pressure because multilayer diffusion and capillary condensation takes place. For pores larger than 5 nm the flux of He is almost zero indicating that all the pores
