Applied Physics in the 21st Century (Horizons in World Physics)

HORIZONS IN WORLD PHYSICS SERIES

APPLIED PHYSICS IN THE 21ST CENTURY, (HORIZONS IN WORLD PHYSICS, VOLUME 266) No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

HORIZONS IN WORLD PHYSICS SERIES Volume 266 Volume 265 Volume 264 Volume 263 Volume 262 Volume 261 Volume 260 Volume 259 Volume 258 Volume 257 Volume 256 Volume 255 Volume 254 Volume 253 Volume 252 Volume 251 Volume 250 Volume 249 Volume 248 Volume 247 Volume 246 Volume 245 Volume 244 Volume 243 Volume 242 Volume 241 Volume 240 Volume 239 Volume 238 Volume 237 Volume 236 Volume 235 Volume 234 Volume 233 Volume 232 Volume 231 Volume 230 Volume 229 Volume 228 Volume 227 Volume 226 Volume 225 Volume 224 Volume 223 Volume 222 Volume 221 Volume 220

Applied Physics in the 21st Century, Edited by Raymond P. Valencia The Physics of Quarks: New Research, Edited by Nicolas L. Watson and Theo M. Grant Lev Davidovich Landau and his Impact on Contemporary Theoretical Physics Edited by Ammar Sakaji and Ignazio Licata New Trends in Quantum Coherence and Nonlinear Optics Edited by Rafael Drampyan Astrophysics and Condensed Matter Edited Thomas G. Hardwell Superconductivity Research Advances Edited by James E. Nolan Instabilities of Relativistic Electron Beam in Plasma, Edited by Valery B. Krasovitskii Self-Focusing of Relativistic Electron Bunches in Plasma, Edited by Valery B. Krasovitskii New Topics in Theoretical Physics, Edited by Henk F. Arnoldus and Thomas George Boson’s Ferromagnetism and Crystal Growth Research, Edited by Emerson D. Seifer Horizons in World Physics, Edited by Victor H. Marselle Frontiers in General Relativity and Quantum Cosmology Research, Edited by Victor H. Marselle Optics and Electro-Optics Research, Edited by Albert V. Berzilla Trends in General Relativity and Quantum Cosmology, Edited by Albert Reimer Chemical Physics Research Trends, Edited by Benjamin V. Arnold Quantum Dots: Research Developments, Edited by Peter A. Ling Quantum Gravity Research Trends, Edited by Albert Reimer General Relativity Research Trends, Edited by Albert Reimer Spacetime Physics Research Trends, Edited by Albert Reimer New Developments in Quantum Cosmology Research, Edited by Albert Reimer Quantum Cosmology Research Trends, Edited by Albert Reimer Horizons in World Physics, Edited by Tori V. Lynch Horizons in World Physics, Edited by Albert Reimer Theoretical Physics 2002, Part 2, Edited by Henk F. Arnoldus and Thomas F. George Models and Methods of High-Tc Superconductivity: Some Frontal Aspects, Volume 2, Edited by J. K. Srivastava and S. M. Rao Models and Methods of High-Tc Superconductivity: Some Frontal Aspects, Volume 1, Edited by J. K. Srivastava and S. M. Rao Horizons in World Physics, Edited by Albert Reimer Theoretical Physics 2002, Part 1, Edited by Henk F. Arnoldus and Thomas F. George Theoretical Physics 2001, Edited by Henk F. Arnoldus and Thomas F. George Generalized functions in mathematical physics: Main ideas and concepts, Edited by A.S. Demidov Edge Excitations of Low-Dimensional Charged Systems, Edited by Oleg Kirichek On the Structure of the Physical Vacuum and a New Interaction in Nature (Theory, Experiment, Applications), Edited by Yu. A. Baurov Electron Scattering on Complex Atoms (Ions), Edited by V. Lengyel, O. Zatsarinny and E. Remeta An Introduction to Hot Laser Plasma Physics, Edited by A.A. Andreev, A.A. Mak and N.A. Solovyev Electron-Atom Scattering: An Introduction, Maurizio Dapor Theory of Structure Transformations in Non-equilibrium Condensed Matter, Edited by Alexander L. Olemskoi Topics in Cosmic-Ray Astrophysics, Edited by M. A. DuVernois Interactions of Ions in Condensed Matter, Edited by A. Galdikas and L. Pranevièius Optical Vortices, Edited by M. Vasnetsov and K. Staliunas Frontiers in Field Theory, Edited by Quantum Gravity and Strings, R.K. Kaul, J. Maharana, S. Mukhi and S. K. Rama Quantum Groups, Noncommutative Geometry and Fundamental Physical Interactions, Edited by D. Kastler, M. Russo and T. Schücker Aspects of Gravitational Interactions, Edited by S.K. Srivastava and K.P. Sinha Dynamics of Transition Metals and Alloys, Edited by S. Prakash Fields and Transients in Superhigh Pulse Current Devices Edited by G.A. Shneerson Gaussian Beams and Optical Resonators, Edited by A. N. Oraevsky Laser Cathode-Ray Tubes, Edited by Yu. M. Popov Quantum Electrodynamics with Unstable Vacuum, Edited by V.L. Ginzburg

Volume 219 Volume 218 Volume 217 Volume 216 Volume 215 Volume 214 Volume 213 Volume 212 Volume 211 Volume 210 Volume 209 Volume 208 Volume 207 Volume 206 Volume 205 Volume 204 Volume 203 Volume 202 Volume 201 Volume 200 Volume 199 Volume 198 Volume 197 Volume 196 Volume 195 Volume 194 Volume 193 Volume 192 Volume 191 Volume 190, (Supp. Vol.) Volume 190 Volume 189 Volume 188 Volume 187 Volume 186 Volume 185 Volume 184 Volume 183 Volume 182 Volume 181 Volume 180 Volume 179, (Supp. Vol. 2) Volume 179, (Supp. Vol. 1) Volume 179 Volume 178 Volume 177 Volume 176, (Supp. Vol) Volume 176

The Muon Method in Science, Edited by V.P. Smilga and Yu.M. Belousov Radiative and Collisional Characteristics of Ions in Hot Plasmas, Edited by I.I. Sobel'man Nonlinear and Quantum Optical Phenomena in Nonequilibrium Media , Edited by V.A. Shcheglov Semiconductors and Insulators: Optical and Spectroscopic …, Edited by Yu.I. Koptev Relativistic Theory of Gravity, Edited by A.A. Logunov Research in Quantum Field Theory, Edited by V.I. Man’ko Nonlinear Theory of Strong Electromagnetic Wave-plasma…, Edited by O.N. Krokhin Dynamics of Elementary Atomic-Molecular…, Edited by O. S. Scheglov Optics and Lasers, Edited by G.G. Petrash Theory and Technology of High Temperature Superconductors, Edited by Yu.I. Koptev Theory of Interaction of Multilevel Systems…, Edited by V.I. Man’ko Physical Properties of High Temperature Superconductors, Edited by Yu.I. Koptev Generalized Mayer Series in Classical Statistical Mechanics, Edited by Igor J. Ivanchik Mathematical Physics, Applied Mathematics and Informatics, Edited by Yu.I. Koptev Squeezed and Correlated States of Quantum Systems, Edited by M.A. Markov Sakharov Memorial Lectures in Physics, Edited by L.V. Keldysh and V.Ya. Fainberg Fiber Optics: Research and Development, Edited by Ye.M. Dianov Gas Dynamics, Edited by Yu.I. Koptev Generation of Nonlinear Waves and Quasistationary …, Edited by L.M. Kovrizhnykh Laser Thermonuclear Targets and Superdurable Microballoons, Edited by A.I. Isakov Spatiotemporal Characteristics of Laser Emission, Edited by M.V. Pyatakhin and A.F. Suchkov Theory of Nonstationary Quantum Oscillators, Edited by M.A. Markov Reliability Problems of Semiconductor Lasers, P.G. Eliseev Pulsars, Edited by A.D. Kuzmin Physical Characteristics and Critical Temperature of High Temperature Superconductors, Edited by M.M. Sushchinskiy Biophysical Approach to Complex Biological Phenomena, Edited by E. Volkov Research on Chemical Lasers Edited by A.N. Orayevskiy Atomic and Ionic Spectra and Elementary Processes in Plasma , Edited by I.I. Sobelman Polarization of the Vacuum and a Quantum Relativistic Gas …, A.Ye. Shabad Superconductors with A15 Lattice and Bridge Contacts Based …, Edited by M. M. Sushchinskiy Metal-Optics and Superconductivity, Edited by A.I. Golovashkin Clusters of Galaxies and Extragalactic Radio Sources, Edited by A.D. Kuz'min Integral Systems, Solid State Physics and Theory of Phase Transitions. Pt. 2. Symm. and Alg. Structures in Physics, Edited by V.V. Dodonov and V.I. Man'ko Quantum Field Theory, Quantum Mechanics and Quantum Optics: Pt. 1. Symm. and Alg. Structures in Physics, Edited by V.V. Dodonov and V.I. Man'ko Photoproduction of Pions on Nucleons and Nuclei, Edited by A.A. Komar Injection Lasers in Optical Communication and Information Processing Systems, Edited by Yu. M. Popov Electron Processes in MIS-Structure Memories, Edited by A.F. Plotnikov Invariants and the Evolution of Nonstationary Quantum Systems, Edited by M.A. Markov Luminescence of Wideband Semiconductors, Edited by M.D. Galanin Pulsed Metal and Metal Halogenide Vapor Lasers, Edited by G.G. Petrash Inelastic Light Scattering in Crystals, Edited by M.M. Sushchinskiy Electron-Excited Molecules in Nonequilibrium Plasma, Edited by N.N. Sobolev Interaction of Ultrashort Pulses with Matter , Edited by M.D. Galanin X-Ray Plasma Spectroscopy and the Properties of Multiply-Charged Ions, Edited by I.I. Sobel'man The Delfin Laser Thermonuclear Installation: Operational Complex …, Edited by G.V. Sklizkov Stoichiometry in Crystal Compounds and Its Influence on Their Physical Properties, Edited by Yu. M. Popov Quantization, Gravitation and Group Methods in Physics Edited by A.A. Komar Classical and Quantum Effects in Electrodynamics Edited by A.A. Komar

Volume 175, (Supp. Vol.) Volume 175 Volume 174, (Supp. Vol. 2) Volume 174, (Supp. Vol. 1) Volume 174 Volume 173 Volume 172 Volume 171 Volume 170 Volume 169 Volume 168 Volume 167 Volume 166 Volume 165

Luminescence and Anisotropy of Zinc Sulfide Crystals Edited by M.D. Galanin Luminescence Centers of Rare Earth Ions in Crystal Phosphors, Edited by M.D. Galanin Electron Liquid Theory of Normal Metals Edited by V.P. Silin Nonequilibrium Superconductivity, Edited by V.L. Ginzburg Thermodynamics and Electrodynamics of Superconductors, Edited by V.L. Ginzburg Quantum Mechanics and Statistical Methods , Edited by M.M. Sushchinskiy Phase Conjugation of Laser Emission , Edited by N.G. Basov Research on Laser Theory , Edited by A.N. Orayevskiy The Theory of Target Compression by Longwave Laser Emission , Edited by G.V. Sklizkov The Physical Effects in the Gravitational Field Black Holes, Edited by M.A. Markov Issues in Intense-Field Quantum Electrodynamics, Edited by V.L. Ginzburg Group Theory, Gravitation and Elementary Particle Physics, Edited by A.A. Komar The Nonlinear Optics of Semiconductor Lasers, Edited by N.G. Basov Solitons and Instantons, Operator Quantization, Edited by V.L. Ginzburg

HORIZONS IN WORLD PHYSICS SERIES

APPLIED PHYSICS IN THE 21ST CENTURY, (HORIZONS IN WORLD PHYSICS, VOLUME 266)

RAYMOND P. VALENCIA EDITOR

Nova Science Publishers, Inc. New York

Copyright © 2010 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Available upon request. ISBN: 978-1-61728-139-6 (E-Book)

Published by Nova Science Publishers, Inc.    New York

CONTENTS Preface Chapter 1

Chapter 2

Chapter 3

Chapter 4

Chapter 5

Chapter 6

ix Emerging Concepts and Challenges in Nano Metal Oxide Thin Films A. Subrahmanyam,, T. P. J. Ramesh, A. Karuppasamy, U. K. Barik, K. Jagadeesh Kumar, N. Ravichandra Raju and R. V. Muniswami Naidu Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced by High-Power IR CO2 Laser J. J. Camacho, J. M. L. Poyato, L. Díaz and M. Santos

1

63

Induction Transformer Coupled Discharges: Investigation and Application I.M. Ulanov and M.V. Isupov

113

Features on the High Frequency Dielectric Response in Ferroelectric Materials J. D. S. Guerra

169

The Principle that Generates Configuration in Animate and Inanimate Systems – A Unified View Antonio F. Miguel

195

Computational Studies on Drag Reduction Effect by Surface Grooves Haosheng Chen and Yongjian Li

217

Chapter 7

Current Limiting in Oxide Ceramic Structures Alexander Bondarchuk

Chapter 8

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology Electronics Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

245

273

viii Chapter 9

Chapter 10

Chapter 11

Chapter 12

Chapter 13

Chapter 14

Index

Contents Magnetically Modified Biological Materials as Perspective Adsorbents for Large-Scale Magnetic Separation Processes Ewa Mosiniewicz-Szablewska, Mirka Safarikova, and Ivo Safarik Optimisation of Deposition Conditions for Functional Oxide Thin Films Sharath Sriram and Madhu Bhaskaran Optical Waveguides Produced by Ion Implantation in Oxide Glasses Feng Chen, Xue-Lin Wang, Lei Wang and Ke-Ming Wang

301

319

341

Thin Film Piezoelectric Response Coefficient Estimation Techniques Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

353

Plasma Technology: An Alternative to Conventional Chemical Processes for Hydrogen Production María Dolores Calzada

365

Target-Plasma-Film Interactions in High Power Pulsed Magnetron Sputtering (HPPMS) K. Sarakinos

373 391

PREFACE Applied physics is rooted in the fundamental truths and basic concepts of the physical sciences but is concerned with the utilization of these scientific principles in practical devices and systems. This new and important book gathers the latest research from around the globe in this dynamic field. Chapter 1 - The metal oxides thin films have created a revolution in technology. The nano metal oxide thin films continue in giving a new dimension to the understanding pushing the frontiers of science with new concepts and advanced applications. The present article is an attempt to compile the emerging concepts and the novel applications in four metal oxide systems; the article also gives the details of the complexities in the understanding of the physical phenomena. The four nano metal oxide systems coverd in this article are : (i) nano titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices, (ii) nano Silver oxide thin film fluorescence in non-volatile optical memories, (iii) Nano Tungsten oxide thin film electro-chromics in efficient and smart windows and (iv) the p- type transparent conducting oxides (TCO) thin films for transparent electronics with a special reference to Zinc oxide thin films. It is interesting to note that the materials described in the four systems even though possessing a good knowledge base, still offer stiff challenges in the basic understanding. The emphasis in the present article is laid on the new applications, the related physics and chemistry. The preparation methods described are limited to those which have a scope for up-scaling (and for possible industrial adoption). This article is not a review of the state of the art for these metal oxide systems. The aim of this article is to introduce the new and emerging applications with the existing concepts. Chapter 2 - This chapter describes some fundamentals of laser-induced breakdown spectroscopy (LIBS) and experimental results obtained from ultraviolet-visible-near infrared (UV-Vis-NIR) spectra induced by laser ablation of a graphite target, developed in our laboratory. Ablation was produced by a high-power IR CO2 pulsed laser using several wavelengths (λ=9.621 and 10.591 µm), power density ranging from 0.22 to 6.31 GW cm-2 and medium-vacuum conditions (typically at 4 Pa). Spatially and time resolved analysis were carried out for the plasma plume. Wavelength-dispersed spectra of the plume reveal the emission of C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and molecular features of C2, CN, OH, CH, N2, N2+ and NH. For the assignment of molecular bands a comparison with conventional emission sources was made. The characteristics of the spectral emission intensities from the different species have been investigated as functions of the ambient pressure, laser irradiance, delay time, and distance from the target. Excitation, vibrational and rotational temperatures,

x

Raymond P. Valencia

ionization degree and electron number density for some species were estimated. Time-gated spectroscopic studies have allowed estimation of time-of-flight (TOF) and propagation velocities for various emission species. Chapter 3 - Researches in the field of low-temperature plasma provide development of devices, which are widely used by the advanced and high-tech industries. Low-temperature plasma is applied by such major industries as microelectronics, semiconductor industry, solar cell production, plasma chemistry, metallurgy, lighting engineering, etc. Among the known methods of production and use of low-temperature plasma (DC and AC arc discharges, RF plasmatrons, microwave plasmatrons) the devices based on application of induction transformer coupled toroidal discharges (TCTD) are the least studied and, thus, rarely used. Simultaneously, these discharges can be used for development of electrodeless generators of low-temperature plasma: transformer plasmatrons and new induction sources of light. The chapter deals with investigation of properties of TCTD, transformer plasmatrons and induction light sources on the basis of TCTD. Investigation results on electrophysical properties of TCTD aimed at development of transformer plasmatrons are presented in the current paper. Dependences between the strengths of TCTD electric field, discharge current and gas flow are obtained for different gases within the pressure range of 10÷105 Pa. The thermophysical characteristics of TCTD were determined: device efficiency, energy balance of a discharge (heat losses to the discharge chamber wall, plasma jet power). The stable TCTD of the atmospheric pressure in argon and in air was firstly obtained and studied by the authors of this chapter. The process of plasmachemical synthesis of NO in air plasma of TCTD was studied. The abnormally high percentage of NO ~7 % was obtained without product quenching. The transformer plasmatrons of the 10÷200 kW power, operating under the pressures of 10÷105 Pa on argon, air and argon+hydrogen, argon+oxygen mixtures, were developed on the basis of the studies performed. The schemes and constructions of these plasmatrons are presented. The electrophysical and spectral characteristics of TCTD were studied in vapors of mercury and neon. The electrodeless sources of visible and UV radiation with the power of 100 W ÷ 100 kW were developed on the basis of this research. Chapter 4 - Single crystal and/or polycrystalline ferroelectric materials show a high frequency dielectric dispersion, which has been attributed as well to a dispersive (relaxation like) as a resonant mechanism. Physical properties such as relaxation and/or resonant motion mechanisms can be investigated by analyzing the complex dielectric permittivity (real, ε’ and imaginary component, ε’’) in a broad spectral frequency range (100 MHz–13 GHz). Especially, for classical (or ‘normal’) and relaxor ferroelectric systems a dielectric response indistinguishable of dispersion or a resonance mechanism has been found in the literature. The occurrence of such common dispersion process in so different kinds of ferroelectric systems has encouraged the development of several mutually excluding models to explain this physical phenomenon. Nevertheless, the reported results are not conclusive enough to clearly distinguish each mechanism. In this work, a detailed study of the dielectric dispersion phenomenon, including the microwave frequencies, carried out in perovskite structure-type ferroelectric systems, for ‘normal’ and/or relaxor compositions, is presented. The dielectric response in “virgin” and poled state have been investigated taking into account the relative direction between the measuring direction and the orientation of the macroscopic polarization. Results revealed that the dielectric response in ferroelectric systems may be described as a

Preface

xi

general mechanism related to an “over-damped” resonant process rather than a simple relaxation-like dielectric behavior. Chapter 5 - The generation of flow configuration (shape, structure, patterns) is a phenomenon that occurs across the board, in animate and inanimate flow systems. Scientists have struggled to understand the origins of this phenomenon. What determines the geometry of natural flow systems? Is geometry a characteristic of natural flow systems? Are they following the rule of any law? Here we show that the emergence of configuration in animate flow systems is analogous to the emergence of configuration in inanimate flow systems, and that features can be put on a unifying theoretical (physics) basis, which is provided by the constructal law. All scientific endeavors are based on the existence of universality, which manifests itself in diverse ways. Here the authors also explore the idea that complex flow systems with similar functions have a propensity to exhibit similar behavior. Based on this thought relationships that characterize animate systems are tested in relation to cities and countries, and some conclusions are drawn. Chapter 6 - The drag reduction effect caused by the periodic surface grooves were studied using computational fluid dynamic method in two different flow conditions: laminar flow in a slide-disk interface, and turbulent flow on the grooved surface immerged in water. In the first part, the drag results by Reynolds equation that is commonly used in lubrication calculation is compared with the CFD result based on Navier-Stokes equation. It was validated that the Reynolds equation is not suitable when the groove depth is higher than 10% of the interface distance. Then, the drag forces on the surface with transverse and longitudinal grooves are calculated using CFD method. It was found that the ‘side wall’ effect causes the drag reduction, which means the drag reduction would appear when the loss of the drag on the groove’s bottom can not be compensated by the pressure drag or the viscous drag on the side walls of the grooves. In the second part, the drag force on the transverse rectangular grooves in turbulent flow is analyzed based on the RANS equations coupled with the RNG k-ε turbulent model. It was found that the pressure drag force makes up a large proportion of the total drag, and the turbulent vortex structure in the grooves affects the drag characteristic of the surface. The ‘side wall’ effect also functions in turbulent flows, and the drag reduction is determined by the synthesized effect of the reduction of viscous drag and the increment of pressure drag. Chapter 7 - The review of physical phenomena which lead to current limiting behavior (current is increased weaker than voltage, saturated and even decrease) in oxide ceramic structures are presented. Particular attention is given to the mechanisms of current limiting in materials whose electrical conductivity is controlled by potential barriers at the grain boundaries. In particular, the current saturation effect in the nano-grained ceramic films is examined. The consideration of physical models describing the current limiting in specific material is followed by the short posing of main experimental data. For some models, the computational modeling has been performed. Chapter 8 - Nanotechnology devices require low resistance contacts, which can be fabricated by the incorporation of silicide thin films This chapter discusses in detail the study of silicide thin films using a suite of materials characterisation tools. The silicides of interest in this study were titanium silicide (TiSi2) and nickel silicide (NiSi), given their low resistivity and low barrier heights to both n-type and p-type silicon. The silicide thin films were formed by vacuum annealing metal thin films on silicon substrates. Silicide thin films

xii

Raymond P. Valencia

formed from metal films deposited by DC magnetron sputtering and electron beam evaporation were compared. The composition, crystallographic orientation, and morphology of these thin films were studied using spectroscopy (AES, SIMS, RBS, in situ Raman spectroscopy), diffraction (Bragg-Brentano and glancing angle XRD, RHEED), and microscopy techniques (TEM, SEM, and AFM). Chapter 9 - Novel magnetically modified biological materials, containing magnetic iron oxides nanoparticles as labels, have been successfully developed and applied as magnetic affinity adsorbents for the magnetic separation of various biologically active compounds and xenobiotics. The main attention was focused on cheap and easy to get magnetic adsorbents which could be applied for large-scale processes. Among them magnetically modified plant-based materials (sawdust) and microbial cells (yeast and algae) were taken into consideration. An inexpensive, extremely simple procedure was proposed for the preparation of such magnetic adsorbents using standard water-based ferrofluids containing maghemite nanoparticles with the diameter of about 12 nm. Such ferrofluids can be prepared in a simple way (almost in any lab) and such nanoparticles can be used to prepare biocomposite materials enabling their simple magnetic separation with standard permanent magnets. Both of these properties are important for possible large-scale applications. The structural, adsorption and magnetic properties of the developed materials were studied in detail by means of scanning electron microscopy, transmission electron microscopy, spectrophotometric measurements, ESR spectroscopy and conventional magnetic methods (DC magnetization and AC susceptibility measurements). The prepared materials efficiently adsorbed selected biologically active compounds and xenobiotics (mainly different enzymes, water-soluble organic dyes and heavy metal ions). Their magnetic behavior was dominated by the superparamagnetic relaxation of isolated single domain maghemite nanoparticles, although a little amount of agglomerates was also present. However, these agglomerates were sufficiently small to show at static conditions the superparamagnetic behavior at room temperature which allows to use the developed materials as magnetic adsorbents in the magnetic separation techniques. Moreover, the prepared materials exhibit the peculiar features enabling their rapid and efficient removal not only from solutions, but also from suspensions. Such materials could be efficiently used to isolate rare biologically active compounds from difficult-to-handle materials including raw extracts, blood and other body fluids, cultivation media, environmental samples, etc. Inexpensive raw materials, extremely simple preparation method, affinity to various biologically active compounds and both organic and inorganic xenobiotics, and distinctive magnetic properties make the developed materials greatly suitable as magnetic adsorbents for large-scale magnetic separation processes. Chapter 10 - This chapter describes a systematic approach to determining the optimal conditions required to deposit thin films of complex functional oxides using RF magnetron sputtering. The motivation of this study was to attain films of designed stoichiometry and of preferentially oriented perovskite crystal structure, as the composition and structure of complex oxides determines their functionality (e.g. ferroelectricity, piezoelectricity, etc.) which is exploited by a variety of applications. The process has been demonstrated with strontium-doped lead zirconate titanate (PSZT) thin films using a combination of materials characterisation techniques – X-ray diffraction, X-ray photoelectron spectroscopy, atomic force microscopy, and crystal structure calculations. The major variables in the deposition

Preface

xiii

process using RF magnetron sputtering were identified as the deposition substrate temperature, post-deposition thermal processing, oxygen partial pressure during deposition, and the choice of metallisation on silicon substrates. Starting with these variables the influence of each was systematically determined. The results highlighted the advantages of slow cooling to promote perovskite growth, the influence of oxygen on the composition and crystal structure of the thin films, and the presence of modified unit cell structure for the PSZT thin films on gold- and platinum-coated silicon substrates. The combination of these results led to the limiting the variables in deposition to finite values and arriving at two definite sets of deposition conditions for deposition based on the dependence of microstructure on thermal conditions and the choice of substrate. The validity of the conditions chosen is then demonstrated by deposition of PSZT thin films on thermally-grown silicon dioxide and attaining nanocolumnar preferentially oriented thin films. Chapter 11 - Optical waveguides can restrict light propagation in very small size of order of several microns, reaching high optical intensities even at low pumps; consequently, some properties of the bulk materials may be considerably improved in waveguide structures, such as non-linear responses, laser actions and optical signal amplifications. As one mature technique for material property modifications, ion implantation has been applied to construct optical waveguides in many optical materials, including insulating crystals and glasses, semiconductors and organic materials. Oxide glasses receive much attention for various photonic and telecommunication applications for its competitive costs and excellent optical features. In this chapter, the authors reviewed the research results obtained for optical waveguides in oxide glasses produced by ion implantation techniques, by giving a brief introduction of basic fabrication principles and methods and a summary of interesting results obtained in this topic. The prospects of possible practical applications of ion-implanted oxide glass waveguides in photonics are also demonstrated. Chapter 12 - The response of piezoelectric materials is quantified using charge and voltage coefficients. One such coefficient is d33, which numerically describes the resulting effect for an applied cause along the same direction. There is a lack of established techniques for quantitative estimation of d33 for piezoelectric thin films. This initiated an investigation into the development of such techniques, as a consequence of which two new techniques for piezoelectric coefficient estimation, under the inverse piezoelectric effect, have been developed. One technique capitalises on the measurement accuracy of the nanoindenter in following thin film displacement, while the other uses a standard atomic force microscope in contact imaging mode to estimate d33. Both techniques were developed by rigorously testing them against standard materials and avoiding commonly reported sources of error. Full details on the development, scope, and limitations of both techniques are presented in this chapter. Chapter 13 - There is currently a strong debate on the need to look for new alternative energy sources to replace fossil fuels (particularly oil), due to atmospheric pollution derived from their use, probable reserve exhaustion and productive countries’ excessive dependence on future political events. Therefore, many countries are currently giving an incentive to research aimed at the development of new energy sources, among which hydrogen —and its use by means of fuel-cells— is to be included. Within the so-called hydrogen economy, hydrogen production —from different raw materials such as hydrocarbons, alcohols and some others— becomes an important aspect, since hydrogen is not a natural product but is obtained by means of water vapour. However, such process leads to the production of high amounts of CO2. The use of plasma technology allows generating hydrogen by decomposing organic

xiv

Raymond P. Valencia

compounds (alcohols and methane), so plasma acts as a reactive environment; this way, biogas —which can be easily found in city landfills— can also be used for hydrogen obtention. Thus, the use of plasma with this purpose arises as an alternative process not only to reforming of hydrocarbons and other substances with water vapour but also to other purely chemical processes, avoiding or minimizing CO2 emissions (involving greenhouse effects). A subproduct —obtained with H2 and which also deserves special attention— is solid carbon, which has important added value, since it may work as a base element for chemical industry and other technological fields. Given the foregoing, research in the use of plasma technology for hydrogen production is one of the research lines within Applied Physics with higher applicability potential within the energy sector. Chapter 14 - Growth of films by plasma-assisted physical vapor deposition (PVD) techniques provides means for tailoring their properties and improving their functionality in technological applications. State of the art plasma-assisted PVD techniques, like direct current magnetron sputtering (dcMS), suffer from low degree of ionization and thus, low ionto-neutral ratios in the flux of the deposited material. The implementation of external sources for the enhancement of the ionization is in many cases technically complicated and increases the end-product cost. High power pulsed magnetron sputtering (HPPMS) is a novel sputtering technique that elegantly enables the conversion of conventional sputtering source into an ion source. By applying the power in unipolar pulses of low frequency (<2 kHz) and low duty cycle (<10%) high values of peak target current density up to ~ 5 Acm-2 are achieved. This leads to the generation of ultra dense plasmas with electron densities of ~1019 m-3. These values are up to four orders of magnitude higher than those obtained by dcMS and result in a high degree of ionization of the sputtered material (up to ~90%). Moreover, due to the low duty cycles, the average target current densities are maintained at values of ~0.5 Acm-2 which are comparable to those in dcMS processes. Therefore, HPPMS allows for highly ionized deposition fluxes without excessive thermal load at the target. In the current short communication the target-plasma interactions and their effect on the deposition rate in non-reactive and reactive HPPMS are discussed. Furthermore, the interactions of the plasma species in HPPMS with the growing film and their implications on the microstructure and the phase formation of metal nitride and oxide films are demonstrated.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 1-62

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 1

EMERGING CONCEPTS AND CHALLENGES IN NANO METAL OXIDE THIN FILMS A. Subrahmanyam1,2, T. P. J. Ramesh3, A. Karuppasamy4, U. K. Barik5, K. Jagadeesh Kumar2, N. Ravichandra Raju2 and R. V. Muniswami Naidu2 1

DAAD Professor, Department of Electrical and Computer Engineering Technical University, 01062 Dresden, Germany 2 Department of Physics, Indian Institute of Technology Madras, Chennai 600036, India 3 Department of Cardiothoracic Surgery, Apollo Hospital, Chennai, India 4 Fraunhofer Institut für Elektronenstrahl und Plasmatechnik (FEP), Winterbergstraße 28, 01277 Dresden, Germany. 5 Samtel, Ghaziabad, India

SUMMARY OF THE ARTICLE The metal oxides thin films have created a revolution in technology. The nano metal oxide thin films continue in giving a new dimension to the understanding pushing the frontiers of science with new concepts and advanced applications. The present article is an attempt to compile the emerging concepts and the novel applications in four metal oxide systems; the article also gives the details of the complexities in the understanding of the physical phenomena. The four nano metal oxide systems coverd in this article are : (i) nano titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices, (ii) nano Silver oxide thin film fluorescence in non-volatile optical memories, (iii) Nano Tungsten oxide thin film electro-chromics in efficient and smart windows and (iv) the ptype transparent conducting oxides (TCO) thin films for transparent electronics with a special reference to Zinc oxide thin films. It is interesting to note that the materials described in the four systems even though possessing a good knowledge base, still offer stiff challenges in the basic understanding. The emphasis in the present article is laid on the new applications, the related physics and chemistry. The preparation methods

1

Corresponding author: email : [email protected]

2

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al. described are limited to those which have a scope for up-scaling (and for possible industrial adoption). This article is not a review of the state of the art for these metal oxide systems. The aim of this article is to introduce the new and emerging applications with the existing concepts.

A. INTRODUCTION TO METAL OXIDES The metal oxide thin film is a wide subject where one can find a vast amount of literature: books, monographs, review articles and innumerable number of research papers [A1-A3]. The metal oxides have challenges encompassing the scientist (on basic issues and understanding), the engineer (for possible applications and application oriented problems) and the technologist (for the cost effective process developments and up-scaling). The metal oxides, including the nano-metal oxides, have a variety of applications almost spanning the entire gamut of strategic to the day to day devices and absolutely mundane to novel functional materials and devices. To name a few areas of the metal oxides are: microelectronics, nonvolatile memories, spintronic devices, bio-medical devices, sensors, piezoelectric devices, magnetic devices, super conducting magnets and magnetic materials, thermal barrier and anticorrosion coatings, heterogeneous catalysis, energy conversion and storage devices etc. The metal oxide is presently a few Billion dollars industry with a consistent and rapid growth. Most metals are oxidized immediately when exposed to the ambient. One can also grow metal oxides with controlled oxygen both on the surface and in the bulk. Thus, one may address the understanding of metal oxides from the surface and bulk points of view: each having its implications, complexities and applications. Most of the applications, the bulk control the surface properties. The nature and strength of the bond between a metal and the oxygen varies with the constituent metal and the preparation or growth conditions; the oxidation process is quite complex. A number of different bonding mechanisms between the metal and the oxygen have been reported, including the weak van der Waals dispersion forces between induced surface dipoles, electrostatic polarization binding, and strong ionic and covalent bonding (the latter being more directional). There are several questions still to be answered in the understanding of the oxidation mechanism and oxidation kinetics in metal oxides. The metal oxides can adopt a vast number of structural geometries and exhibit insulating, semiconducting, metallic and even superconducting behavior. While analysing the magnetic properties, metal oxides made a significant progress. Recent years have witnessed a tremendous growth of research in the field of magneto-electronics, in view of its obvious potential for novel devices with entirely new capabilities. In this context, phenomena such as giant magneto-resistance (GMR), colossal magneto-resistance (CMR), spin-tunneling in junctions (STJ) have attracted significant attention [A4]. The oxides of transition metals and their interfaces with metals and semiconductors represent one of the most exciting and challenging areas currently studied in magnetism. The surface science of metal oxides is a relatively very young filed and there are a lot of challenges and opportunities. Single metal atoms tend to form relatively strong chemical bonds to most oxide surfaces. This is mainly the result of a severe under-coordination of the metal atom (compared to its favored bulk state), which is therefore reactive with the surface. These ionic or covalent metal–oxide bonds are correspondingly quite short, normally around 1–2 Å. Typical bond strengths range from 1 to several eV, and are normally referred to as

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

3

adsorption or desorption energies—defined as the energy required to remove a metal atom to the vacuum level. The metal–oxide bonds tend to further strengthen at various oxide surface defects, e.g., anion/cation vacancies, which can trap and immobilize diffusing metal atoms. Many of the properties of metal oxides are related to the surface and interface structure. A microscopic understanding at the atomistic level of the oxide surfaces and interfaces is of fundamental importance for the development of new materials, nano-structures and novel devices. In the emerging field of nanotechnology, one of the goals is to make the nanostructures with special or tailored properties with respect to those of the bulk materials. The oxide nanostructures or nanoparticles can exhibit unique properties owing to their limited size and high density of low-coordinated sites or can provide new extended systems with unprecedented structure, morphology and/or composition. In order to facilitate tailored nanoparticle production, the study of the largely unknown mechanisms of metal oxide nucleation and subsequent growth (which mainly depend upon the surface energy adhesion kinetics) is of pivotal importance. A key aspect in this respect is the control of defects and impurities, a field closely related to the defect engineering of materials. It is rather difficult to present a concise review of the metal oxide thin films. With all these un-limitations, the present paper attempts to describe the emerging concepts in four metal oxide thin film systems: (i) titanium, (ii) silver, (iii) tungsten and (iv) zinc. These four thin film systems have vast applications and pose a variety of complexities; the present article confines to: (i) nano Titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices, (ii) nano Silver oxide thin film fluorescence in non-volatile optical memories, (iii) Nano Tungsten oxide thin film electro-chromics in efficient and smart windows and electro-chromic displays and memories and (iv) nano Zinc oxide for p- type transparent conducting oxide (TCO) thin film for transparent electronics. This paper is a collection of ideas, and truly emerging, with potential applications. A vast amount of work is needed to understand these metal oxide systems before one realizes these devices. The aim of this paper is to introduce the problem from its fundamentals. The fond hope of the authors is to invite the researchers towards the challenges posed by these four oxide systems, ultimately leading to a better understanding of the phenomena and to develop new technologies with new and novel devices.

1. TITANIUM OXIDE THIN FILM FOR LUNG ASSISTED DEVICES 1.1. Summary Titanium oxide is a well known photocatalyst. One of the important applications of the photocatalysts is water splitting for generation of alternate fuel : hydrogen. The same concept is being used in oxygenating human blood. The water available in the blood can be the source of oxygen. Since this oxygen is generated in the immediate vicinity of the haemoglobin, this seems to be a most efficient method of oxygenating human blood. The efficacy levels are quite encouraging.

4

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

1.2. Introduction Human health care is one among the top priority in materials and bio-medical research. Nano materials, in particular, the nano metal oxides is the focus of attention for several advanced applications. Among the several ailments, respiratory failure is one of the leading causes of both mortality and morbidity world wide. With ever increasing environmental pollution, the first and most important organ being affected and afflicted is the lungs. Advances in the management of end stage lung disease have not paralled the management of end stage heart, liver or renal disease. Lungs is the organ which supplies the oxygen to the blood and removes carbon dioxide from the blood. The oxygen and carbon dioxide are attached to the Haem group of the red blood cells which is the transport carrier. Any lung damage severely affects the oxygen infusion to the haem group. The development of lung assisted devices so far are based on the concept of gas diffusion through hollow fibre membranes; this technology is constrained for long term usage. There are excellent reviews on the efforts on lung assisted devices [1.1]. The primary function of any lung assisted device (either in vitro or in vivo) is to oxygenate the blood and to remove the carbon dioxide. An emerging concept in the lung assisted devices is to split the water (by photocatalytic process) available in the blood to produce oxygen; since this oxygen is generated in the close proximity of the haemoglobin, the oxygenation process is very effective and efficient. Also, the photocatalysts are made of metal oxides, they are bio-compatible and a long term usage is feasible with almost little maintenance. The results from the preliminary experiments are very encouraging. With this application, several millions of population suffering from lung ailments are anticipated to get relief in a most cost effective way.

1.3. Introduction to Photo-Catalysis Photocatalysis is an established phenomenon shown by several metal oxide thin films: oxides of titanium, tungsten, vanadium, zinc and iron etc. This oxidation photocatalytic phenomenon is widely being used for addressing the problems of environmental pollution. The titanium dioxide (TiO2) is studied rather extensively and is ascertained to be biocompatible and it is an efficient photocatalyst. Titanium metal has been proven to be a bio-compatible material and a large number of implants and substitutes are being manufactured on commercial basis [1.2] Titanium is a highly reactive metal which forms the oxides and the oxidation kinetics are extremely fast, of the order of a few nano-seconds. Thus titanium is always covered with a thin oxidizing layer all the time (unless one works in ultra high vacuum). Titanium dioxide is a stable compound which is resistant to chemical attack (this property should not be confused with the chemical activity) and it is corrosion resistant. From among the several applications of titanium dioxide, the photocatalytic activity is a field of great technological and industrial importance. The initial work of Fujishima and Honda [1.3] on the photocatalytic splitting of water with titania without an external bias has evoked considerable interest among many researchers. Most of the efforts in the early days have been focussed to generate hydrogen as a clean fuel, but, it is still far from realization. However, the water splitting by photocatalysis, has opened up novel applications and new challenges; one of the

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

5

challenges is to understand the surface of titania. Photo active TiO2 surfaces react with aqueous solutions and they will dissociate organic molecules, can oxidize water into oxygen and hydrogen, they also show anti-bacterial activity. TiO2 is an age old (proven) industrial material used as pigment in paints. High refractive index titania is widely used in optical coatings. TiO2 is the semiconductor that facilitates charge transfer mechanism in dye sensitized solar cells. Recently, the nano-titania is emerging as a self cleaning coating and also as environmental protection and it is termed as green chemistry material. There are many applications of pure, functionalized or doped titania films or nanoparticles [1.4] (e.g. titania). Many of these applications have potential impact on our every-day life, like in solar cells, as self-cleaning surface film on glass windows [1.5], as photocatalyst, including the removal of toxic pollutants. There are excellent reviews on the photocatalytic property of titanium dioxide [1.6] The titanium dioxide (TiO2) is being explored for medical purposes only recently. The nano-phase titanium dioxide to medicine or bio-medical engineering has been developed during the past fifteen years. Compared to the bulk TiO2, the nano-TiO2 exhibits higher photocatalytic efficiency leading to pronounced phototoxic effects on bacteria, cancer cells and tumours. TiO2 is a wide band gap semiconductor; when a mono-chromatic wavelength corresponding to the band gap is incident, an electron – hole pair will be generated. The electron moves to the conduction band and the hole to the valence band. If the two charges can be separated from recombination or trapping, the electron reduces oxygen to O2- and the hole oxidizes (the harmful carbon monoxide to harmless carbon dioxide). Both oxidation and reduction cannot take place simultaneously as these chemical reactions are rate controlled. The high efficiency of the photocatalytic action depends upon (i) the quantum yield of the photogenerated carriers, (ii) the effective removal of one the photogenerated carriers, (iii) the recombination or trapping centres and (iv) the interaction of the photogenerated carrier with the lattice or surface defects.

TiO2 (Eg = 3.2 eV) Band Gap

hν(350nm)

O2 -

Conduction Band e-

O2

hν UV light

h+ Valence Band

Reduction Oxidation

CO2, H2O

Figure 1. Photocatalytic mechanism operative in Titanium oxide

It may be noted that the nano-TiO2 is more effective and efficient in photocatalysis than the bulk because the nano TiO2 has no charge present inside the particle, but do have a strong electric field, where as the larger particles contain a layer of charge underneath the surface at a distance of the space charge. For the nano TiO2, large number of photogenerated charges move on to the surface and thus more active centres exist on the surface. Moreover, due to high surface to volume ratio and large interfacial surface area, photogenerated electrons do

6

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

not recombine easily with the photo generated holes. Thus the photocatalytic activity of nanoTiO2 is much higher compared to that of the bulk TiO2.

1.4. Basics of Water Splitting by Photo-Catalysis The concept of water splitting by photocatalysis [1.7] is being adopted in the development of lung assisted devices. The photo generated holes in the presence of water can oxidize water into oxygen and hydrogen. The technological problem is to remove the photo generated electron before it combines with the hole. The catalytic process is a liquid / solid interface phenomenon and is generally categorized as “heterogeneous photocatalysis”. The quantum efficiency of the photocatalytic process (or more precisely, the electron – hole generation) is quite low in TiO2. The main issues presently being focussed are the development of visible wave length photocatalyst and enhancement of the photocatalytic efficiency. The basic reactions of TiO2 surface in the presence of light of suitable wavelength (that generated electron – hole pairs) and water (or humidity) may be summarized as :

TiO 2 + hν → h + + e − (h + = hole vacated by e- ) h + + H 2 Oabs → OH + H + h + + OH − abs → OH e − + O 2 → O•− 2 (superoxide)

1.5. Oxygenation of Blood by Photo-Catalysis There are two important issues in the oxygenation of blood: first,the supply of oxygen to the haemoglobin and the second, the mechanism of intake of oxygen by the haem group. As is well known, lungs are the organ that supply oxygen from atmosphere into the alveoli; the gas exchange/ diffusion mechanism across the alveoli is facilitated by a thin membrane (the same membrane is responsible for removing carbon dioxide from the blood). This membrane, if damaged or clogged by the particles, the efficiency of the lungs decrease significantly. With the ever increasing levels of pollution, the lung functions get diminished or deteriorated in its full activity even in a healthy human. The lung diseases are the second largest among the population in the world and very unfortunately the lung therapy has not progressed as much as the heart diseases. The basic reactions of the oxygen with the haemoglobin and the basic structure of haemoglobin are given in Notes 1 at the end of this article. Preliminary experiments conducted on the photocatalytic oxygenation of deoxygenated human blood have given positive results [1.8 – 1.11]. There are also Patents available on the subject [1.12 – 1.16]. The experiment basically consists of a photocatalytic cell (Figure 2) consisting of the photocatalytic coating of pure or doped TiO2 on the outer wall of the inner quartz tube; the inner tube is enclosed by a bigger diameter outer quartz tube with a provision

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

7

of inlet and out let for the blood. The UV light is incident on the photocatalytic coating while the blood is flowing on the surface of the inner tube and the photocatalytic action takes place. The oxygen deficient venous blood (looks blackish in color) is introduced and the photocatalytic surface is activated with suitable light to generate the electron – hole pairs. The oxidation mechanism of water available in the blood with the photogenerated holes produces dissolved oxygen and this dissolved oxygen is attached to the haem group. The blood gases are measured by a blood gas analyzer and also by an optical spectrophotometer. One of the important and critical points of this concept is that the photocatalytic action should produce only dissolved oxygen and not the reactive species oxygen. The pH, Po2 (partial pressure of dissolved oxygen), PCo2 (partial pressure of dissolved of carbon dioxide), Sao2 (percent saturation of haemoglobin) can be measured using a conventional blood gas machine (I—Stat). The haemoglobin of the subject was measured with routine lab analysis and found to be 16gm/dl. The increase in whole blood oxygen content (CaO2) in mls of oxygen/100ml of blood is calculated using the formula [1.17] [1.34 X Hb gm/dl X SaO2 (expressed as fraction)] + (0.003 X Po2 mmHg)

(1)

Figure 2. Basic concept of a photolytic lung assisted device (in vitro)

The photocatalyst prepared in the form of thin films have oxygenated the human blood with good efficacy (about 10 ml of oxygen in 100 ml of blood in 120 minutes with a surface area of 14 cm2) as shown in the Figure 3 [1.9]. The oxygenation depends upon the photocatalytic area and photocatalytic efficiency. One precaution to be exercised in prolonged use of the photocatalytic process is that the photocatalyst should not undergo photocorrosion or modification of the catalytic surface. Since the oxygen is generated within the blood, the process is highly efficient and thus this process does not demand large areas (as in the case of membrane oxygenators), means the device can be really portable. The efficacy of oxygenation shows a very high promise to develop into a small portable device which can come to the rescue of patients with various pulmonary disorders. The important feature of this new concept is that the materials employed are bio-compatible (nano- Titanium dioxide), the oxygenation is instantaneous, quantum efficiency of oxygen production is simple and the device seems to have no limitations on prolonged usage, and the UV light does not seem to damage any of the blood constituents (like the white blood cells and the DNA).

8

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

The device based on this concept requires sensors that measure the oxygen, carbon dioxide (attached to the haem group) and other blood constituents which need to be developed. As a first step, to estimate the oxygen attached to the haem group, the optical absorption measurements have been performed and the results show that the optical techniques have the potential to be used as sensors/ monitors. More details on the optical measurements of the blood can be found in the book: Visible and near infra red absorption spectra of human and animal haemoglobin [1.18]. The dissolved oxygen generated by photocatalysis can also be measured by “oxy probes” (M/s Ocean Optic Inc). The Oxy probe is based on the principle that when a fluorophor sees dissolved oxygen, the fluorescence signal decreases correspondingly(quenching). Presently, correlation is being attempted between the measurements on oxy-probes, optical techniques and the blood gas machine results.

CaO2 (ml of oxygen per 100 ml of blood)

22 20

TEST 1 TEST 2 CONTROL 1 CONTROL 2

18 16 14 12 10 8 6 0

20

40

60

80

100

120

Time (min)

Figure 3. Photolytic oxygenation of whole human blood

1.6. Challenges in the Process of Oxygenation by Photocatalysis The challenges of the present concept are to evaluate the efficacy of oxygenation of blood (or equivalently, the measurement and quantification of oxygen produced), the effect of UV light on the constituents of the blood, the quantification of the amount of water in the participating photocatalytic reaction, any change in the viscosity and pH of the blood during the oxygenation process etc. The challenge related to the photocatalytic surface is to understand the photo-corrosion that may degrade the performance over long periods of usage. One of the questions to be answered is what happens to the hydrogen produced (or hydrogen is produced at all) and how to remove the carbon dioxide from the hem group. The fundamental challenge is to set the protocols for the experiment.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films 1.80

9

414.59 414.61

1.6

1.4

414.55

1.2

414.57

1.0 A 0.8

0.6 414.35 0.4

576.56 576.51

0.2

0.00 200.0

250

300

350

400

450 nm

500

550

600

650

700.0

Figure 4. Optical absorption of photocatalytic oxygenated whole human blood

2. NANO SILVER – SILVER OXIDE THIN FILMS 2.1. Summary Two important observations have been made in nano-silver oxide: (i) p- type (hole) conductivity and (ii) the fluorescence. The fluorescence phenomenon has got potential applications in the development of ultra high density optical non-volatile memories and in Medical/Molecular imaging. The nano silver gives raise to localized surface plasmons (LSP). The LSPs have potential applications in nanoscale optics and photonics as well as in the construction of sensors and biosensors. These two observations are manifestations of the bond between the metallic silver and oxygen. .

2.2. Introduction Silver has long been used in photo-activated image processing (photography), high density alkaline batteries, household mirrors and tableware. Because of its high reflectivity, silver is an important component in optical coatings; it is used in optical solar reflectors (OSRs) in satellites. Recently, silver is used to generate localized electric fields for molecular detection and electrical field enhancement. Silver is very reactive noble metal and forms silver oxide by interacting with oxygen in the ambient. Nanosize structures of silver have, as anticipated, very different properties compared with that of the bulk silver. Some of the emerging applications of nano-silver oxide thin films are: the anti-bacterial activity [2.1], possible use in super -Resolution Near-Field Structure (RENS) for optical read – write ultra high density non-volatile memories [2.2], fluorescence imaging [2.3] and the property of Surface Enhanced Raman Scattering (SERS) [2.4, 2.5]. The decomposition of nano-silver

10

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

oxide (into oxygen and small metallic silver clusters at a relatively lower temperatures and high energy optical beams) can create a strong light-scattering centre that resolves small pits or marks beyond the diffraction limit is a useful property in ultrahigh-density optical data storage applications [2.2]. Silver, because of its d-shell electrons, exists in different oxidation states and forms several oxides: AgO, Ag2O, Ag3O, Ag2O3. The formation of these oxides depends upon the growth conditions/ reaction kinetics: availability of oxygen in the growth chamber and the energy required for the oxidation. The surface morphology and the nucleation kinetics depend upon the kinetic energy of the silver oxide molecules reaching the substrate. The heat of formation of Ag2O and AgO are -31.1 kJ/mol and 220 kJ /mol respectively. It is known that the AgO phase is relatively stable at high oxygen pressures and at low temperatures. The AgO exists in different crystal systems like cubic, monoclinic and tetragonal [2.6 – 2.8]. As it is well known that the noble metal clusters have strong electromagnetic field due to the electron gas. The local oscillations of the electron gas in the nano metal clusters are called Plasmons. The localized surface plasmons (LSPs) are charge density oscillations confined to nano metallic clusters. Excitation of LSPs by an electric field with suitable monochromatic wavelengths results in strong light scattering, appears as intense surface Plasmon absorption bands; consequently, at resonance, there will be an enhancement of the local electromagnetic field. The frequency and intensity of the surface Plasmon (SP) absorption bands are characteristic of the material. The example of SP resonance is the red colour of aqueous dispersion of colloidal gold particles (the colloidal gold is used as bio-markers). When the metal nano clusters are in the immediate vicinity of an insulator, the surface plasmons can induce defect levels in the insulator. These defect levels when photo-activated will give fluorescence emission. One example is : silver – silver oxide nano clusters. The fluorescence intensity is significantly high in these photo-activated nano clusters. The SP absorption, consequently, the fluorescence intensity, depends upon the size, size distribution and shape of the nano-structures.

2.2. Surface Plasmons and Fluorescence in Silver – Silver Oxide Nano Clusters The silver nanoparticles have the property to exhibit surface plasmons as they can sustain resonant collective electron oscillations [2.9]. These surface plasmons can be optically excited either in the near ultraviolet or visible range of wavelengths. As it is well known that the properties of the surface plasmon resonance are critically dependent on the shape, size, and spatial arrangement / close proximity of the metal and insulator particles [2.10]. The Plasmon resonance has a vast potential in nano-optics [2.10] and in bio-medical engineering [2.11]. The size-dependent physical properties in nanoparticles have a special advantage to study the evolution from small-molecule to bulk-systems. A good example is the fluorescence of noble metals. It is well-known that the quantum yield for fluorescence from noble metal films is extremely small, most of the times, below the detection levels, of the order of 10-10 [2.12]. The fluorescence mechanism is because of radiative recombination of sp-conductionband electrons with d-band holes [2.12, 2.13]. This process is explained by the plasmon enhancement of the inter-band transitions.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

11

Interestingly, Dickson et al. reported recently on nano silver clusters showing a structured luminescence band [2.3] thus exhibiting properties characteristic of distinct molecular clusters. Studies in this area are complicated by the difficulty of generating and stabilizing very small clusters. The silver oxide nano clusters can be stabilized by encapsulation in dendrimers, by thiol - ligands as well as by peptide-capping agents and by embedding in glass matrixes. Thus, the main focus for the search in silver / silver oxide is to grow nano sized metal / metal oxide clusters and on the preparation of a single stable phase oxide of silver. The present section confines to the growth of nano sized silver oxide clusters in thin film form. There are several techniques followed to grow these silver oxide thin films: RF [2.14] and DC reactive sputtering [2.5, 2.15], reactive thermal evaporation [2.16], chemical [2.17] and electro-chemical synthesis [2.18a, 2.18] and by reactive pulsed laser deposition (RPLD) [2.19]. The silver oxide thin films prepared by reactive DC Magnetron sputtering technique [2.15] have shown that (i) the oxygen pressure in the growth chamber ( ~10-3 mbar) influences the formation of different oxides of silver, (ii) at high oxygen flow rate in the growth chamber (2.01 sccm, corresponds to a chamber pressure of 7 x 10 -3 mbar) and at a sputter power density of 1.0 watts /cm2, stiochiometric Ag2O is formed, (iii) the mixed silver oxide films show P- type conductivity. The silver films are prepared at three magnetron powers: 20, 30 and 40 watts. The distance between the target and the substrate is kept at 7.5 cm. The sputter time is varied from 30s to 120s. A thickness monitor (Hind Hivac Model DTM 101) displayed the thickness of the films during the deposition. Thickness of the films grown is in the range 5.0- 40.0 nm. The oxidation of these metallic films is carried out with pure oxygen gas (99.9%) and at 520 ± 2 K for 30 minutes in a chamber (evacuated initially to a pressure of 1x10-5mbar; during oxydation the chamber pressure is at 5x10-4mbar). For plasmonic devices, it is desirable to have a single low energy phase of silver oxide (where photo-activation can be achieved more easily). The pulsed laser deposition (PLD) and the chemical modulation techniques have the potential to produce a single phase photoactivated silver oxide nano clusters. Dellasaga et al., recently reported the growth of silver oxide thin films by PLD technique, using the fourth harmonic beam (λ = 266 nm) of Nd: YAG Laser with a fluence of 1.6 J/cm2 and varying the oxygen pressure in the chamber between 4- 150 Pa range (they have not reported the electrical and optical properties) [2.19]. These experimental conditions have produced varying visible plume length and large droplets (~ 2µm to 200 nm) on the grown film. The varying visible plum length has a significant influence on the reaction and growth kinetics. In the recent study conducted in the author’s laboratory [2.20], it is reported that in the oxygen pressure range 9 Pa to 50 Pa, with the third harmonic of the laser pulse (1.06 J/cm2), it is observed that with increasing oxygen pressure in the growth chamber: (i) the hexagonal Ag2O transforms to monoclinic AgO, (ii) the grain size in the film increases from 59 nm to 200 nm, (iii) the surface roughness of the film increases from 09 nm to 42 nm, (iv) the resistivity of the films increases, (v) the activation energy (from electrical conductivity measurements) increases from 0.64 eV to 0.75 eV, (vi) the surface work function of the films increases from 5.47 eV to 5.61 eV and (vii) the optical band gap of AgO thin films decreases from 1.01 eV to 0.93 eV. The Raman spectroscopy on AgO thin films show low wave number peaks corresponding to the stretching vibration of Ag – O bonds. This study shows

12

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

that single phase AgO thin films, a requirement for plasmonic devices, can be prepared at room temperature by PLD technique with an oxygen pressure of 20 Pa. From the earlier work, two important properties (from among the several interesting properties) exhibited by silver oxide thin films are discussed in this section. They are: (i) ptype conductivity and (ii) fluorescence emission from photoactivated silver oxide.

2.3. P- type Conductivity in Mixed Phase Silver Oxide Thin Films The mixed silver oxide thin films (AgO and Ag2O) prepared by reactive DC Magnetron sputtering technique have shown P- type (hole) conductivity [2.5]. From the concepts of silicon physics, it is well known that the holes are formed if there is localization of covalent bonds in the material. How these covalent bonds are formed in the silver oxide films is the question. The p-type conductivity in these films has been explained on the basis of Sanderson’s theory of partial ionic charge; that is, the bond between the oxygen and silver contains both ionic and covalent. The mixed bonding is also supported by the theoretical calculations of electronic structure and bonding mechanism [2.21]. From the theory and empirical formulation proposed by Sanderson [2.22], it is possible to calculate partial ionic charge on the various oxides of silver. With the data given in Sanderson [2.22], calculations of the partial ionic charge of silver in Ag2O, Ag2O3, AgO and Ag3O4 yields 41, 26.7, 32 and 28.4%, respectively [2.5]. The partial ionic charge indicates the presence of covalent bonds in the material. Since the bonding between silver and oxygen is partially ionic and also since the electronic configuration after molecule formation reduces towards the non-bonding nature of the oxide ions (which is the basic assumption in Sanderson’s theory), there is a possibility of localization of the covalent bonds in these silver sub oxide thin films. The localization of the oxide ions gives raise to the p-type nature of the conduction (this concept is equivalent to the chemical modulation of the valence bond).If the density of these localized states is considerably large in the film, the holes appear as majority carriers showing p-type conduction; the p- type conduction can be confirmed with Seebeck Coefficient and in Hall Effect measurements. Thus, the silver sub oxide film possesses both n-type (due to electrons from metallic silver) and p-type (the oxide bond with silver) conduction. One possibility for understanding the covalence is through the bond energy; when the silver metal reacts with oxygen, if the bond energy for the sub-oxides (for example Ag2O3, AgO, Ag2O, Ag3O4) is less, this weak bond may give rise to the presence of an acceptor level leading to hole conduction in the film. It is known that in the sputtering process, the energy to form the molecule is taken from the plasma. More details of the p- type conduction and the complexities in the measurement are discussed in the section 4 (p-type TCOs).

2.4. Fluorescence from Photo-Activated Nano-Silver Oxide The research group of Dickson at the Georgia Institute of Technology observed photo activated fluorescence from sol- gel prepared nano scale silver oxide (Ag2O) films, exhibiting properties characteristic of distinct molecular clusters. The photoactivated fluorescence was produced from non-luminescent silver oxide thin films by photoexcitation [2.23, 2.24, 2.25,

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

13

2.26]: silver oxide decomposes into oxygen and small metallic silver clusters of several nanometer size. This dissociation of silver oxide into metallic silver and partial silver oxide is the basis for a plasmonic structure. After local photoactivation of silver oxide films with UV or blue light, the possibility for exciting the fluorescence into the visible spectra represents an attractive prospect for advanced optical data storage [2.14, 2.15]. However, several studies report that metallic interfaces play complex roles in the basic interactions of an electromagnetic field with optically active materials. For example, smooth metallic surfaces reduce the radiation from nearby organic dyes through non-radiative energy transfer, while molecules absorbed onto metallic electrodes show several orders of magnitude increase in surface-enhanced Raman signals [2.27]. Furthermore, quantum dots (QDs) located within the surface plasmon polariton (SPP) field of the metal also shows a large photoluminescence (PL) enhancement [2.28]. Then the question is: how to obtain an optimized silver oxide – silver cluster size to obtain the photoexcited fluorescence or how to control the sizes of the silver clusters. One requirement is that the silver oxide should be in a stable single phase (either Ag2O or AgO) and the phase should be such that it can be photoactivated (or the energy required to dissociate oxygen from the silver should be low).

Figure 2.1. Basic plasmonic structure of Silver – silver oxide nano clusters

14

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al. Green 525-532 nm light

Red emission Fluorescence

Blue 488514.5 nm Filter for 660 nm Filter for 514.5 nm

Scattered and fluorescent emissions

Ag2O film

Figure 2.2. Basic concept of optical ultra high density read – write non-volatile memory 0.91

fluoroscence from silver oxide film

Normalised Intensity (nm)

0.90 0.89 0.88 0.87 0.86 0.85 0.84 630

640

650

660

670

680

Wave length (nm)

Figure 2.3. Fluorescence observed in DC Magnetron sputtered silver oxide thin films

It may be mentioned here that the photoactivation of silver oxide into silver nano clusters probably depends upon the bond energy of oxygen to silver. Thus, it is quite likely that the nature of the bond : electrovalent or covalent, the oxidation energy and kinetics of nano silver are extremely important in the basic understanding of photoactivation. The mechanism for photoactivation is postulated to be initiated by absorption of a visible photon by Ag2O, a semiconductor material with a band gap of 2.25 eV . Photoexcitation of Ag2O is known to produce silver peroxide, AgO, which may photolytically form AgO. Further evidence for this mechanism was obtained from the photoactivated generation of fluorescent silver particles from AgO films. Such a light-induced reaction sequence may thus supply a nucleation site for the growth of silver clusters.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films 1.01

400 Å 250 Å 100 Å

1.00

Normalised Intensity

15

0.99 0.98 0.97 0.96 0.95 0.94 0.93 0.92 600

620

640

660

680

Wavelength (nm)

Figure 2.4. The fluorescent intensity in silver oxide thin films with different thicknesses

20 W 30 W 40 W

1.00

Normalised Intensity

0.95

0.90

0.85

0.80

0.75

0.70 600

620

640

660

680

700

Wave length (nm)

Figure 2.5. The fluorescence spectrum for silver oxide thin films grown with varying DC Magnetron power densities

More recently, it has been found that silver film can be cheaply and easily transformed into Ag nanoparticles and nanowires in a gas mixture of hydrogen and oxygen in a vacuum chamber. The particle and wire diameters are very uniform and approximately 20 -50 nm, and they can be three-dimensionally fabricated on almost all material surfaces without the need for thermal annealing. Silver nanoparticles and wires will soon appear in small and cheap molecular detection sensors by small precise adjustments for generating surface plasmon conditions.

16

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

A periodic array of silver oxide may be the first step in realizing a ultra high density (~ tera byte per square inch) [2.18]. The composite film is made from silver coated polymeric arrays on an indium tin oxide (ITO) coated glass substrate. By adjusting column diameters and lattice constants of the array to coincide with the excitation wavelength, the fluorescence was markedly enhanced. The increase is due to the efficient energy transfer from silver surface plasmon to silver oxide that is formed during the preparation of the composite film.

3. NANO TUNGSTEN OXIDE THIN FILM FOR ELECTRO-CHROMICS 3.1. Summary Electro-chromics is an emerging area for smart and energy efficient windows and displays. The well researched and reasonably understood material is tungsten oxide. However, the coloration properties of the tungsten oxide are not understood in detail yet. The nano tungsten oxides wires and rods have a vast potential not only in enhancing the coloration properties but also in giving new dimensions in display technology. The present section summarizes the efforts to understand the coloration / optical process based on the “site saturation” model. The basic experiments to prepare the nano-tungsten wires and rods also have been mentioned. The other important observation is that when tungsten oxide is irradiated with electrons, a strong coloration is produced and the coloration is retained for almost infinite time. This electron beam induced coloration can be applicable for optical memories. The details of the interaction of electrons with the tungsten oxide are described in this section.

3.2. Introduction Electrochromism is defined as the ability of a material to change its optical states in response to an applied DC voltage. In detail, the optical transmittance of the electrochromic materials can be changed in a controlled and reversible manner from a complete transparent state to a fully dark colored state by applying an electric potential across the material; this electric potential enables the charges from external circuit to intercalate with the electrochromic lattice. Tungsten oxide is the well known electrochromic material which is extensively studied for its potential application in several fields like electrochromic smart windows, displays, solar cells, fuel cells, gas sensors, automotive rear-view mirrors, sun roofs, and organic/inorganic hybrid memory devices etc, [3.1]. Tungsten oxide is also a photocatalytic material. The transmittance of WO3 films can be altered in a reversible and persistent manner by the intercalation/de-intercalation of small cations (H+, Li+, Na+) and electrons into the film. The reaction mechanism responsible for coloration in tungsten oxide thin films is widely accepted as the double injection/ extraction of electrons and ions in tungsten oxide,

WO3 + xe − + xM + ↔ M xWO3 ( M = Li, Na, H ) (Transparent)

(Blue)

3.1

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

17

There are two aspects to the electro-chromics: the material aspect (a transition metal oxide) and the device aspect consisting of the ion intercalation and de-intercalation. The electrochromic properties like coloration efficiency, cyclic durability and kinetics of coloration-bleaching process of the electro-chromic material strongly depend on its structural, morphological, and compositional characteristics, and therefore on the deposition techniques and growth parameters. The commonly employed deposition techniques are thermal evaporation, electron beam evaporation, sputtering, pulsed laser ablation, chemical vapor deposition, sol-gel coating, and electro deposition.

Figure 3.1 Schematic of an Electro-Chromic Device

Electrochromic device consists, in general, of two (transparent) electrodes, an electrolyte, and a minimum of one electrochromic active layer (Fig 3.1). A transparent glass, usually used as a substrate, is coated with a transparent and conductive electrode (ITO), which is covered with the electrochromic active material. These layers are part of an electrochemical cell completed by an electrolyte. In all-solid-state devices, an ion conductor is used instead of an electrolyte, and an ion storage medium (counter electrode) is necessary. Traditionally, current or charge is recorded as a function of applied voltage, and transmittance or reflectance is measured as a function of the potential to characterize electrochromic materials. Most investigations for the electrochromism of single film are done in liquid electrolyte experiments. In principle, only two electrodes are necessary for an electrochromic experiment, but for fundamental research a three electrode configuration is mostly used to control the potential at the electrochromic film (working electrode), relative to a reference electrode. The current -voltage graphs and transmittance-voltage graphs are typically recorded, but also the response in the electrical and optical data to a constant voltage is measured to determine the switching speed of an electrochromic film. An important property of electrochromic thin films for evaluation is the coloration efficiency (CE), expressed as cm2 C-1. In order to obtain this value, the total injected/ intercalated charge, as a function of unit area, and the change in the optical density (δOD) must be known. Both CE and OD are

18

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

wavelength dependent quantities. The change in the optical density is correlated with the transported charge Q (per unit area). From the two quantities, the coloration efficiency can be determined by the equation: CE (λ) = δOD/Q. CE is a number which is independent of the film thickness, because δOD and Q depend in the same way on the film thickness. One of the desired applications of electrochromic systems is their use in the ‘smart windows’. Compared with other window/glass constructions, up to 50 % of the energy costs for heating and cooling buildings can be saved with `smart windows [3.2]. Today, these advantages are still offset by the heavy currents the window requires for optimal operation. For example, the charge required for the color change has an order of magnitude of 100.0 Coulomb m-2, even for the best systems known today. Since the system operating voltage normally is of the order of 2.0 V, a current of several amperes must be applied for a fastswitching system via the surface electrodes at this relatively low-operating voltage. In order to obtain color-changing times in the order of minutes range for a window having a typical dimension of 2.0 m x 2.0 m, a current of approximately 10.0 A (of 2.0 V) is necessary. If there is any inhomogenity or non-uniformity, the coloration shows up: this leads to partly darkened parts of the window (near the electrodes), which may cause irreversible material deteorations. A second significant problem is the heating-up of the window by use of absorptive materials leading to impairment of the long term stability. These problems are still awaiting solutions. Two questions are addressed in the present section on electro-chromics: (i) what is the optical process of coloration in the tungsten oxide thin films and the testing of “site saturation” model for stiochiometric tungsten oxide and (ii) how to prepare the nano- wires and nano- rods of tungsten oxide for enhanced electro-chromic performance.

3.3. The Optical Absorption Process in Tungsten Oxide Thin Films The optical absorption process, in electro-chromic materials in general, and tungsten oxide in particular, is not yet clearly understood. Faughnan et al [3.3] reported that the 6+

5+

and W coloration process is primarily due to inter-valence charge transfer between W sites; Lee et al [3.4] proposed that the sub-stoichiometric tungsten oxide thin films rich in

W 4+ states exhibit better electrochromic properties where the optical absorption/ modulation 4+ 5+ during intercalation of charges is primarily due to the transitions between W and W sites. Several researchers have supported the validation of this theory and various attempts have been made to deposit sub-stoichiometric tungsten oxide thin films for optimum electrochromic performance. Recently, Berggren et al [3.5]have reported (on Li+ intercalation in tungsten oxide) that the coloration process is independent of the stoichiometry of the films and a statistical theory has been developed based on “site saturation” model: The essential features of the model (based on Li intercalation) are (i) the relative number 6+

5+

4+

of electronic transitions (P) between the states W , W , W are determined by the amount of charge intercalated in the sample i.e., x = Li / W ratio (ii) the analytical expressions for the number of possible electronic transitions are given by,

W 6 + ↔ W 5+ : P = 2 x ( 2 − x ) 3 ,

3.2

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

19

W 5+ ↔ W 4+ : P = 2 x 3 (2 − x),

3.3

W 6+ ↔ W 4+ : P = x 2 (2 − x) 2

3.4

and (iii) the absorption strength per transition depends on the type of transitions (higher for

W 6+ → W 5+ ). In the “site saturation” model, the inter-valence charge transfer between different states 6+

5+

4+

of tungsten i.e., W , W , W is governed by the amount of charge intercalated in the sample. According to Berggren et al., if the ratio of Li / W is less than 1.0, then the transition occurs primarily between W

6+

and W

5+

states and with further intercalation of charges i.e.,

1.0 < Li / W < 2.0, the transition between W 5+ and W 4+ states becomes predominant. From this theory, one may infer that at low intercalated charges, the stoichiometric or super stoichiometric tungsten oxide thin films undergo higher optical absorption (due to the 6+

larger number of available W sites) compared to sub-stoichiometric ones. From an application point of view, this is an interesting result, since the realization of larger optical modulation with small charge insertion or extraction would enhance the speed and the long term cycling stability of the electrochromic device. Then the question is how to prepare a super stoichiometric tungsten oxide thin films. From research point of view, varying the oxygen content in the tungsten oxide thin films while achieving the stoichiometry is an interesting problem. One of the techniques to grow stoichiometric or super stoichiometri tungsten oxide (WO3) thin films is by sputtering metallic tungsten target using pure oxygen as a sputter gas [3.6]. Using oxygen as sputter gas is not uncommon; it has been employed to deposit transition metal oxides like TiO2 [3.7]. The growth conditions and the properties of oxygen sputtered tungsten oxide thin films are summarized in Table 3.1. All the experimental data on oxygen sputtered WO3 thin films like the optical modulation, coloration efficiency, switching speed and cycling durability are in support of the recently reported statistical theory based on “extended site saturation model”. Table 3.1 Oxygen sputtered tungsten oxide thin films : growth conditions and properties

20

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

The x = Li / W is replaced by H / W in the present paper. In the present case, the value of x was found to be ~ 0.6 as determined from the expression [3.5],

x=

Qe , ρel

3.5

where Qe is the total effective charge passed through the device; e is the electron charge; l is the film thickness; and ρ is the density of tungsten sites present (the values of Qe and ρ are evaluated with the details given in reference [3.8]). The relative number of transitions plausible at x = 0.6 have been plotted in Figure 3.2 (inset) following equations 3.2 to 3.4. Since coloration efficiency is a thickness-dependent quantity, the coloration behavior of the films is presented in terms of thickness independent quantity ‘α ‘[Figure 3.2]. The value of x ~ 0.6 is also confirmed by the pronounced optical absorption coefficient at E ~ 1.5 eV. It is also found that the absorption coefficient is highest for the S2 samples whereas the coloration efficiency is highest for the S3 samples. Such a variation is attributed to the difference in the thickness of the samples; 553.0 nm for S2 and 621.0 nm for S3. All the above results clearly 6+

5+

indicate the predominance of W → W transitions in the oxygen sputtered stoichiometric tungsten trioxide thin films at low intercalated charge levels (x ~ 0.6). Since the relative density of W

6+

states available in the oxygen sputtered films is higher, the optical absorption 6+

5+

strength for W → W transitions per unit-inserted charge is much larger than that of the sub-stoichiometric films. Generally, the better coloration observed in the oxygen sputtered WO3 thin films is in support of the ‘extended site saturation model’. However, the large differences between the optical properties of the films S1-S3 have various reasons like ‘site saturation’, thickness variation and the slight differences in the amount of proton intercalation.

Figure 3.2. The absorption coefficient of WO3 films sputtered at different oxygen pressures as a function of energy. (Inset): The inset shows the relative number of transitions between different states of tungsten as a function of H/W ratio.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

21

The effect of temperature on the properties of WO3 films by intentionally heating the substrates during deposition has also been investigated. At lower oxygen chamber pressure, substrate heating has resulted in the formation of WO3 nanowires without any catalyst.

3.3. Kelvin Probe Studies The charge intercalation brings about a change in the surface work function of these electro chromic thin films. It is possible to estimate the surface work function of these thin films in the colored and the bleached states using Kelvin probe measurements [3.9]. The surface work function may be estimated, as an approximation, from the concepts of bands (generally the band theory is for crystalline solids with periodicity). This approximation is mainly to understand the differences between the surface work function of the thin film in the colored state (charge intercalated) and in the bleached state (the charge is de-intercalated). The concept of bands is invoked mainly to obtain a reasonable understanding of the surface work function of these amorphous thin films; equivalently, one can also understand these changes in terms of surface electrode potential. In as deposited WO3 thin films, the Fermi level lies in the band gap between the conduction band and the valence bands (wide band gap metal oxides). In principle, during the intercalation process, electrons occupy the conduction band and there will be a shift in the Fermi level position [3.10]. The coloration being a double injection: electrons flow into the tungsten oxide (from the transparent and conducting ITO) and the protons (or positive ions) from the electrolyte. The intercalated charges move the Fermi level towards the conduction band. Hashimoto et al. [3.11] have studied Li intercalated amorphous WO3 thin films by X ray Photoelectron Spectroscopy (XPS) and has shown that the movement of the Fermi level is in proportion to the amount of intercalated charge. The surface work function is the measure of the energy from the Fermi level to the vacuum level. The measurement on work function difference of these films before and after ion intercalation may be directly correlated to the movement of the Fermi level. The work function may be evaluated from the Contact Potential Difference (CPD) measured by the Kelvin probe technique. The relation between the CPD and surface work function can be given by qVCPD= ΦR– ΦF

3.6

where q is the electronic charge, VCPD is the contact potential difference measured, ΦR = 4.83 eV is the work function of the gold coated reference electrode and ΦF is the work function of the film under study. For the Kelvin probe measurements, the intercalation is done on selective area of the WO3 film by forming a hemispherical drop of electrolyte solution (0.5 M HCl dissolved in distilled water) on the film and using platinum as the counter electrode. Figure 3.3 shows the variation in CPD of the tungsten oxide electrochromic layers scanned over a selectively colored portion (0.5 cm) of the films. The counter electrode tip was placed at the centre of the hemispherical drop such that the maximum CPD value at the centre of the colored portion could be observed. The figure clearly presents the difference in the amount of charge intercalated in each sample. The maximum CPD values of colored samples S1, S2 and S3 are found to be 400.0 mV, 449.0 mV and 511.0 mV (the accuracy of the measurement is ±1.0 mV); following equation 3.6, the

22

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

work functions are: 4. 41 eV, 4.36 eV and 4.30 eV respectively. The work function of the uncolored samples is found to be ~ 4.86 eV. The difference in the CPD values indicates a change in the Fermi level in tungsten oxide with intercalation [3.12]. These values support the view that the decreasing work function is due to the enhancement of conductivity of the tungsten oxide film with intercalation. 500

colored WO3

-2

1.5 x 10 mbar -2 3.1 x 10 mbar -2 5.2 x 10 mbar

400

CPD (mV)

300

200

100

0 Decolored WO3

-100 0

1000

2000

3000

4000

5000

Distance (μm)

Figure 3.3. Contact potential difference measurements (CPD) on a selective colored portion (0.5 cm) of oxygen sputtered tungsten oxide thin films

Figure 3.4. Topography of Contact Potential Difference (CPD) for the oxygen sputtered WO3 film (S3) scanned over an area of 2.0 cm x 1.0 cm

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

23

Lower work function of S3 in comparison with that of S1 and S2 can also be explained as due to higher electronic conductivity of S3. Similar observation on the fall in electrical resistivity with coloration has been reported earlier [3.13]. Figure 3.4 shows the topography of CPD for the tungsten oxide electrochromic layer of S3 sample (scanned over an area of 2.0 cm x 1.0 cm) depicting the variation in CPD of selectively colored WO3 film. The stability of an electrochromic device with repeated color / bleach cycles depends on the amount of residual charge left on the film in each cycle after de-intercalation. The cycling durability of the films has been studied by measuring the amount of residual charge left in the sample using Kelvin probe. The bleached portion of the film has a CPD of -120.0 mVolts, which is even less than the CPD of rest of the sample. The dip in the plot clearly indicates that there are no residual charges left in the sample after bleaching; ensuring a better cycling durability of the electrochromic device based on oxygen sputtered WO3 films.

3.4. A Brief Outline on Nano-Electro-Chromics Materials and Processes In the recent days, one-dimensional (1-D) tungsten oxide nanostructures like nanorods; nanowires and nanotubes have attracted much attention due to their potential applications in various fields [3.14-3.16]. Due to large surface to volume ratio, the one-dimensional nanostructures of tungsten oxide has been used in wide range of areas like field emission, electrochromism, luminescence and gas sensing [3.17-3.20]. The significant application of 1D nanostructured tungsten oxide is its use as a structure-directing precursor for WS2 nanotube production [3.21], a useful material in tribological applications and catalyses. Many researchers have grown different forms of tungsten oxide nanostructures mostly by the oxidation of metallic tungsten in the form of powders, foils and filaments [3.22 – 3.25]. It is found that the resulting compositions, crystal structures, and shapes vary significantly with preparation techniques. Nonetheless, studies on nanoscale tungsten oxide materials are still limited by the lack of an appropriate preparation method. Therefore efforts to develop a simple, economic and efficient method for the synthesis of tungsten oxide 1-D nanostructures for device applications are still required. Nano crystalline and one dimensional WO3 nanowires have been studied for electrochromic devices due to its fast diffusion kinetics and improved reliability. Meda et al. (2002) prepared nano crystalline WO3 by organometallic chemical vapor deposition, and reported coloration efficiency of 22 cm2C-1 [3.26]. Particle size in their studies observed was about 20 – 40 nm. Deshponde et al. (2007) have reported a higher coloration efficiency of 44 cm2 C-1 for their nano structured thin films compared to 24 cm2 C-1 of amorphous thin films [3.27]. Durability of nano structured thin films tested in acidic electrolyte was superior to that of the amorphous thin films. Both studies discussed above have supported their observation of increased coloration efficiency and durability by the porous and increased surface area characteristics of the nanocrytalline materials. One dimensional nanowire has been synthesized by solvothermal technique [3.28], solution route [3.29] and thermal evaporation of WO3 powders [3.30- 3.32]. Yoo et al. (2007) investigated WO2.72 nanowires synthesized using solvothermal method [3.28]. They reported higher Li+ diffusion coefficient of 5.2×10-11 cm2S-1 and coloration and bleaching time of 3.5 s and 1.1 s respectively. Coloration efficiency was 55 cm2 C-1. They attributed this higher coloration efficiency and elevated Li diffusion coefficient to increased surface area and short diffusion path length for Li+. Similarly a fast response time has been reported for the single crystalline

24

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

WO3.H2O structure produced through solution route, using PVA-124 as a structure directing agent and glacial acetone as a stabilizer [ 3.29]. They reported a wider lattice spacing of d = 5.36 Å towards the electrolyte side which responsible for the faster kinetics. WO3.H2O was formed by the layered build up of corner sharing octahedral with water molecule in between the planes. This structure leads to an increased d spacing in this single crystalline WO3.H2O system. It is found feasible to prepare tungsten oxide (W18O49) nanorods and nanowires by the conventional dc magnetron sputtering technique in active arc suppression mode with metallic tungsten target in Ar + O2 atmosphere (at a sputtering power density of 3.0 Wcm-2 ) with substrate temperature. During deposition, the argon chamber pressure was kept fixed at 1.2 x 10-2 mbar whereas the oxygen chamber pressure was varied in the range 1.0 x 10-3 mbar to 5 .0 x 10-3 mbar. The deposition was made on glass, silicon and ITO coated glass substrates. The following results describe the nature of the dc magnetron sputtered tungsten oxide nano-wires and their dependence on the oxygen chamber pressure. The surface morphology of the substrate heated WO3 films was studied by scanning electron microscopy (SEM). Figure 3.5 shows the SEM image of tungsten oxide thin film deposited at an oxygen chamber pressure of 1.0 x 10-3 mbar. The micrograph shows the formation of randomly dispersed nanowires having a diameter of 60– 80 nm and length of ~ 10.0 µm. However, the film deposited at an oxygen chamber pressure of 2.0 x 10-3 mbar shows a non- uniform distribution of nanowires i.e., in a matrix of lengthy and horizontally dispersed nanowires. An island of quasialigned nanowires with larger diameters and shorter lengths is also found. The region surrounding the island is filled with nanowires of average length 10.0 μm and average diameter 50 – 80 nm. The maximum length of these wires extends up to 30.0 μm.

Figure 3.5. SEM image of the WO3 film deposited at an oxygen chamber pressure of 1 x 10-3 mbar with a substrate temperature of 450 °C

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

25

Figure 3.6. SEM image of the WO3 film deposited at an oxygen chamber pressure of 2 x 10-3 mbar with a substrate temperature of 450 °C

The nanowires in the island region have an average diameter of 60.0 – 100.0 nm and length of 3.0 μm. Despite the difference in the dimension and orientation of the two types of nanowires, it is found that the diameter of the nanowires is uniform throughout their length (Figure 3.6). With further increase in the oxygen chamber pressure to 3.0 x 10-3 mbar, it is found that the morphology to change from nanowires to that of mixed features with larger particulates and rectangular rods of length 1.0 to 10.0 µm [Figure 3.7]. The morphology of the films deposited at an oxygen chamber pressure of 4.0 x 10-3 mbar shows a uniform distribution of particles with an average particle size of about 500.0 nm. The surface is found to be dense, compact and is free from any pores [Figure 3.8]. However, the films deposited at an oxygen chamber pressure of 5.0 x 10-3 mbar shows the presence of pores of diameter 50.0 nm to 500.0 nm.

Figure 3.7. SEM image of the WO3 film deposited at an oxygen chamber pressure of 3 x 10-3 mbar with a substrate temperature of 450 °C

26

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.8. SEM image of the WO3 film deposited at an oxygen chamber pressure of 4 x 10-3 mbar with a substrate temperature of 450 °C

Figure 3.9. SEM image of the WO3 film deposited at an oxygen chamber pressure of 5 x 10-3 mbar with a substrate temperature of 450 °C

There are also clusters of particles distributed through out the film [Figure 3.9]. The above results clearly show the pressure dependence on the properties of substrate heated WO3 thin films.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

27

3.5. Structural Properties of WO3 Nanowires Figure 3.10 shows the X-ray diffraction pattern of the WO3 nanowire thin films deposited by DC magnetron sputtering in active arc suppression mode. The argon chamber pressure was kept constant at 1.2 x 10-2 mbar whereas the oxygen chamber pressures were maintained at 1.0 x 10-3 mbar and 2.0 x 10-3 mbar. Except intensity, we could not observe much difference in the peak positions between the two samples.

Figure 3.10. X-ray diffraction pattern of the WO3 nanowires deposited at an oxygen chamber pressure of 1 x 10-3 mbar and 2 x 10-3 mbar with a substrate temperature of 450 °C

All the main peaks can be indexed with the reflections of monoclinic tungsten oxide (W18O49) system (JCPDS card No: 05-0392). Consistent with general features of nanomaterials, the overall diffraction intensity is weak, and peak broadening is pronounced. Diffraction peaks of (010) are stronger than the rest, indicating that the major growth direction is [010]. The above result shows that these nanowires grow along the [010] direction, which is reasonable because the close-packed planes of monoclinic W18O49 crystal are (010) [3.18].

3.6. Morphology of WO3 Nanowire Thin Films Figure 3.11 shows the scanning electron microscopy image of tungsten oxide thin film deposited at an oxygen chamber pressure of 1.0 x 10-3 mbar. The micrograph shows the formation of randomly dispersed nanowires having a diameter of 60.0 – 80.0 nm and length of ~ 10.0 µm.

28

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.11. SEM image of WO3 thin films deposited at an O2 chamber pressure of 1 x 10-3 mbar showing nanowire formation over a large area of ~ 300 µm2

Figure 3.12. Low magnification SEM image of WO3 thin films deposited at an O2 chamber pressure of 1 x 10-3 mbar showing nanowires and cracks all along the surface.

As could be seen from the figure, the maximum length of the nanowires extends up to 30.0 µm. A lower magnification SEM image of the surface shows cracks all along the film [Figure 3.12]. It is evident that the surface is uniformly filled with nanowires over a large area of ~ 50000.0 µm2. Large area coatings of nanowire thin films have not been reported earlier because of the limitations in the method of deposition. Since sputtering is an industrially viable technique used for large area coatings, we have been able to achieve such a large coating of nanowire thin film. However, the cracks found all over the films gives a clue that nanowire growth could be associated with thermal induced strain in the film.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

29

Figure 3.13. SEM image of WO3 thin films deposited at an O2 chamber pressure of 2 x 10-3 mbar showing vertically grown nanowires

Figure 3.14. SEM image of WO3 thin films deposited at an O2 chamber pressure of 2 x 10-3 mbar showing cracks and nanowires

To confirm this proposition, we deposited WO3 thin films at 2.0 x10-3 mbar with a substrate heating temperature of 450 ºC. Fig 3.13 shows the surface morphology of WO3 films deposited at 2.0 x 10-3 mbar. The figure clearly shows the presence of vertically aligned nanowires along with horizontally dispersed nanowires. In fact, Figure 3.13 is the low magnification image of Figure 3.12 which clearly shows the formation of vertically aligned nanowires. The lower magnification SEM image of these samples also show cracks all over the film [Figure 3.14]. The above results give an indication that the catalyst free nanowire formation could be due to thermally induced strain in the film. In addition, the absence of nanowires at higher oxygen chamber pressures indicates that the essential conditions for the growth of nanowires are the oxygen to tungsten ratio and thermally induced strain in the film.

30

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.15. SEM image of WO3 thin films deposited at an O2 chamber pressure of 0.5 x 10-3 mbar showing the formation of nanowires within the cracks

To verify this proposition, WO3 thin films have been deposited at an oxygen chamber pressure of 0.5 x 10-3 mbar. The morphology of these films studied by SEM shows cracks along the film and within the cracks we could observe the formation of nanorods of length ~ 500.0 nm [Figure 3.15]. With decrease in O2 chamber pressure (0.5 x 10-3 mbar) we could observe significant changes in the surface morphology like (i) nanowires are formed inside the cracks and no features are observed on the top surface whereas in the other films (1.0-2.0 x 10-3 mbar) the nanowires are formed only on the top surface (ii) the density of cracks decreases with increase in oxygen chamber pressure. The difference in the location of nanowire formation can be attributed to the difference in oxygen to tungsten ratio. Similar observation of tungsten oxide nanowires grown inside the crack surface has been reported earlier. It has been explained on the basis that the surface within the crack has oxygen to tungsten ratio adequate for nanowire growth [3.33]. The importance of oxygen to tungsten ratio in catalyst free formation of nanowires has also been reported for sputtered WO3 thin films [3.34]. The pressure dependence of the density of cracks can be attributed to the difference in deposition time of the films. To maintain same thickness in films deposited at various O2 chamber pressures (Po2), the deposition time is varied between 6.0 min (Po2 – 1.0 x 10-3 mbar) to 18 min (Po2 –5.0 x 10-3 mbar). The samples deposited at Po2 of 1.0 x 10-3 mbar have a thickness of 350.0 nm as estimated from the reflectance measurements. Due to technical difficulties involved in TEM measurement of thicker films, we have evaluated the surface crystallite size of nanowires by atomic force microscopy (AFM). Figure 3.16 shows 2D and 3D representations of the surface morphology of WO3 samples deposited at Po2 of 1.0 x 10-3 mbar. The figures clearly show the formation of well-defined nanowires. In addition, it is clearly evident that the surface is smooth and dense with randomly dispersed nanowires spread throughout the film. The average diameter of the nanowire determined from the height image was found to be 70.0 nm. The detailed microstructures of tungsten oxide thin films were characterized using

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

31

transmission electron microscopy (TEM). TEM measurements were made on WO3 films deposited at 1.0 x 10-3 mbar. The thickness of the films used for TEM characterization was ~ 100.0 nm. Figure 3.17 shows transmission electron microscopy image of WO3 thin films deposited at Po2 of 1.0 x 10-3 mbar and with a substrate heating temperature of 450 ºC.

Figure 3.16. 2D and 3D representations of the surface morphology of WO3 nanowire thin films deposited at Po2 of 1 x 10-3 mbar and substrate temperature of 450 ºC

32

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.17. Transmission electron microscopy (TEM) image of WO3 thin films deposited at Po2 of 1 x 10-3 mbar and substrate temperature of 450 ºC

Figure 3.18. Higher magnification TEM image of WO3 thin films deposited at Po2 of 1 x 10-3 mbar and substrate temperature of 450 ºC

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

Figure 3.19. Selected Area Electron Diffraction Pattern (SAED) of an individual WO3 nanorod deposited at Po2 of 1 x 10-3 mbar

Figure 3.20. SAED pattern taken along the edge of a nanorod showing a superposition of single crystalline and nanoparticle pattern

33

34

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

The micrograph shows the formation of tungsten oxide nanorods with an average length of 150.0 nm and diameter in the range 40.0 – 100.0 nm. The nanorods are randomly dispersed and are closely packed. From the higher magnification TEM image we find that the individual nanorods are tapered along the edges [Figure 3.18]. In addition, between the nanorods we could find a distribution of nanoparticles with an average particle size of ~ 7.0 nm.The selected area electron diffraction pattern (SAED) of an individual nanorod exhibits a single crystalline dot pattern [Figure 3.19] with lattice spacings of 3.76 Å corresponding to [010] growth directions in W18O49 system [3.35]. Interestingly, the SAED pattern taken along the edge of a nanorod shows a single crystalline pattern superposed on the ring pattern of nanoparticles [Figure 3.20]. The above results suggest that well-crystallized W18O49 nanowires grow along the [010] direction. From the SEM, TEM and AFM measurements, we did not find any liquid droplet features at the end of tungsten oxide nanowires and nanorods and in fact there is no catalyst involved during the deposition. Thus the well-known vapor-liquid-solid (VLS) growth mechanism [3.36] cannot be applied for our nanowire growth, because a metal particle is necessary at the growth front of nanowire to act as the catalytic site. Therefore, in our samples, the vapor-solid growth process is presumed to control the formation of one-dimensional nanorods or nanowires. Sputtering of metallic tunsgten in an appropriate oxygen atmosphere at suitable substrate temperature is presumed to initiate the nucleation of nanowires. In our case, the WO3 nanowires were formed at an oxygen chamber pressure of 1.0 and 2.0 x 10-3 mbar at ~ 450 °C. In the literature [3.37], the interfacial strain between substrate and film has been reported to initiate whisker growth in tungsten oxide. In the present study, strain induced by difference in thermal expansion between the different layers of WO3 during deposition is presumed to relax as nanowires. The above results indicate that the key factors responsible for the catalyst free growth of sputtered WO3 nanowires could be (i) the appropriate oxygen chamber pressure (or O/W in film) during deposition and the (ii) thermally induced strain in the substrate heated WO3 films. The properties of the substrate heated WO3 films deposited at various oxygen chamber pressures (1.0 – 5.0 x 10-3 mbar) are presented in table 3.4.4.3. From the table, we find the nanowires deposited at Po2 of 1.0 – 2.0 x 10-3 mbar to exhibit better electrochromic properties.

3.7. Growth of Layered Tungsten Oxide Thin Films The layer structured tungsten oxide thin films were deposited on glass substrates by reactive dc magnetron sputtering (in active arc suppression mode) of metallic tungsten target in argon and oxygen atmosphere. The sputtering parameters used for the growth of layered structures are; sputtering power of 2.0 W/cm2, working pressure (argon + oxygen) of 1.4 x 102 mbar (the chamber pressure with argon gas alone is 1.0 x 10-2 mbar) and deposition time of 10 minutes. The sputtered WO3 thin films have been irradiated with electron beam in a commercial electron beam evaporation system at a vacuum of 5.0 x 10-5 mbar. For electron bombardment, the samples were placed right on the top of the slot for crucible holder in the electron beam evaporation system. In the present study, three samples of size 1.0 cm x 1.0 cm have been grown in one run of sputtering. Of the three samples, one is characterized as such and is labeled as S1. The other two samples have been subjected to electron irradiation:

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

35

sample S2 (bombarded with electron beam of energy 3.5 KeV and current of 3.0 mA for 8s) and sample S3 (bombarded with electron beam of energy 3.5 KeV and current of 6.0 mA for 4s); the sample S3 is shown in the Figure 3.21. Table 3.2. Deposition conditions and the experimentally determined optical and electrochromic parameters of tungsten oxide thin films grown with a substrate heating temperature of 450 °C

3.8. Effect of Electron Beam Irradiation on the Properties of WO3 Thin Films The tungsten oxide can switch between two optical states in response to an applied dc voltage (electrochromism), UV irradiation (photochromism) and thermal energy (thermochromism). Apart from the above-mentioned conventional methods, WO3 can also be colored by irradiation methods using electrons [3.38], ions [3.39] and lasers [3.40 ]. The chromic behavior of tungsten oxide also has a vast potential in the emerging field of ultra high density optical data storage. Once colored, the tungsten oxide has an inbuilt memory to retain the coloration information even after the source is removed. This property of tungsten oxide, as indicated in the literature [3.41, 3.42], makes it a candidate suitable for possible ultra high density data storage application. Recently, the optimization of nanoscale electron emitters (nanotube, nanowire and nanowires) gives feasibility for high-density data storage based on the electron bombardment of tungsten oxide thin films. The effect of electron bombardment on the structural, morphological, coloration and luminescent properties of dc magnetron sputtered WO3 thin films is presented. For these studies comparatively thicker films (1061.0 ± 5.0 nm) with layered structures have been grown. Due to electron bombardment, significant changes in the physical properties of WO3 have been observed. In addition, by varying the energy and flux of electrons we observed coloration and luminescence in WO3 films. Based on these results, a concept for data storage application due to electron beam induced coloration/ luminescence in tungsten oxide thin films has been proposed [3.43].

36

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.21 Photograph showing electron bombarded tungsten oxide thin film (S3)

3.8.1. Structural properties of electron bombarded WO3 films Figure 3.21 shows the X-ray diffraction pattern of the as- deposited (S1) and electron bombarded tungsten oxide thin films (S2, S3). The broad featureless pattern at lower angles clearly indicates the amorphous nature of as-deposited tungsten oxide thin films. However, the XRD pattern of electron bombarded samples (S2, S3) shows high intensity peaks corresponding to the tetragonal WO3 system (JCPDS card No: 89-1287). The samples S2 are found to be highly crystalline with sharp peaks. The highest intensity peak at 24.0° shows preferential orientation along the (100) plane. But the S3 sample shows broad and less intense peaks, characteristic of nanoscale materials. The average crystallite sizes determined by Scherrer’s formula are found to be 40.0 nm and 15.0 nm for S2 and S3 samples respectively.

Figure 3.21. X-ray diffraction pattern of the as-deposited (S1) and electron bombarded (S2, S3) WO3 thin films.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

37

3.8.2. Optical properties of electron bombarded WO3 films The transmittance spectrum of the tungsten oxide thin films (S1 - S3) in the wavelength range 250.0 nm – 850.0 nm is shown in Figure 3.22. The spectrum of glass substrate is also presented for reference. The as-deposited WO3 films have a high transmittance of ~ 85.0 % in the entire visible range of 400.0 nm – 850.0 nm. The thickness of the film is 1061.0 ± 5.0 nm. The sharp fall in transmittance at λ = 340.0 nm corresponds to the fundamental absorption edge of amorphous tungsten oxide. When these samples are bombarded with electron beams, the entire film (1.0 cm x 1.0 cm) turns deep blue in S2 sample, showing a maximum transmittance of 15.0 % at λ = 503.0 nm. In addition, the band edge is also red shifted by 32.0 nm. But in the case of S3 samples, only a portion of the film in and around the point of impingement turns blue showing a maximum transmittance of 21.0 % at λ = 477.0 nm. Interestingly, the band edge of S3 sample is blue shifted by 60.0 nm. Coloration of the film by electron impingement is attributed to the reduction of WO3 subsequently leading to the release of oxygen atoms from the sample [3.44]. Each oxygen vacancy contributes a maximum of two free electrons [3.45] and hence the extraction of oxygen atoms by electron bombardment increases the carrier concentration considerably. Thus, the blue shift in S3 sample can be ascribed to the conventional Burstein - Moss effect [3.46]. Similar results have been reported for the WO3 thin films colored by proton/ lithium intercalation [3.47, 3.1]. However, understanding of the red shift observed in S2 sample requires more investigations. (Figure 3.23 ) shows the refractive index of as-deposited WO3 thin film as a function of wavelength [400.0 – 850.0 nm]. The experimental values of n are in fair agreement with the reported values [3.39] for WO3 thin films. But, interestingly, the wavelength dependence of extinction coefficient k shows step like features as shown in Figure 3.24. The reason for such a behaviour is not clear yet; but it is presumed to be associated with the layered structure.

Figure 3.22. Spectral dependence of transmittance in the wavelength range 250-850 nm of the asdeposited (S1) and electron bombarded (S2, S3) WO3 thin films

38

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.23. Refractive index of the as-deposited sample (S1) as a function of Wavelength

Figure 3. 24. Spectral dependence of extinction coefficient for the as-deposited tungsten oxide thin films

3.8.3. Compositional studies of electron bombarded WO3 films The effect of electron bombardment on the stoiciometry of the samples has been studied by evaluating the chemical composition of the films by XPS measurements. As shown in Figure 3.25, the XPS spectrum of the as-deposited sample (S1) shows two peaks with binding energies of 35.8 and 37.9 eV, corresponding to W4f7/2 and W4f5/2, respectively. They are the valence state peaks of W6+ corresponding to the standard spectra of WO3 [3.48]. But when the sample is bombarded with electron beam, the spectrum becomes broader showing additional peaks, which create a shoulder at the lower binding energy side. In samples S2 and S3, the shoulder is found approximately at 32.3 and 33.4 eV respectively. In addition, the valley of the original doublet is also diminished considerably.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

39

Figure 3.25. XPS spectrum of the as-deposited (S1) and electron bombarded (S2, S3) WO3 films in the region of W 4f levels

These changes in the spectra are interpreted in earlier reports as a consequence of nonstoichiometric conditions on the film surface, may be due to the formation of surface oxygen vacancies [3.49]. All these results suggest the occurrence of compounds with less oxygen to tungsten ratio by electron bombardment.

3.8.4. Surface morphology of electron bombarded WO3 films The surface morphology of WO3 thin films has been studied by scanning electron microscopy (SEM) and atomic force microscopy (AFM) measurements. The surface of the as-deposited sample S1 is found to be smooth at the center but along the edges, it shows a layered structure [Figure 3.26]. Several platelets are found to be stacked uniformly one over the other. The platelets are found to have different sizes and shapes, but interestingly, all the platelets have nearly same thickness as shown in [Fig 3.27]. The possible mechanism of layered growth may be understood from the active arc suppression mechanism of the pulsing unit: Sparc-LE 20. When charges build on the target due to surface oxidation, the pulsing unit reverses the target voltage for ~10.0 μs. However, reversal of the target voltage attracts more electrons subsequently leading to the discharge of positively charged insulative layer. This action lowers the surface potential of the insulative layer and thus prevents the breakdown of an arc. Once the arc is suppressed, the Sparc-LE unit reverses the target voltage back to negative potential and normal sputtering is resumed. Thus the reduction of sputtering rate by target poisoning along with the reversal of target voltage for few microseconds is thought to initiate layer growth in the present investigation. The uniform thickness of the platelets indicates that the target is charged up at regular intervals.

40

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.26. Scanning electron microscopy (SEM) image of the as-deposited tungsten oxide thin film showing layered structure along the edge of the film

Figure 3.27. High magnification SEM image of the layered structure showing uniform thickness of the individual platelets

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

41

Figure 3.28. SEM image of the electron bombarded sample S3

Figure 3.29. High magnification SEM image of sample S3 showing cracks at the point of electron impingement

The surface morphology of electron bombarded samples S2 and S3 is alike and so a detailed analysis of S3 alone is presented here. When electron beam is irradiated on the film, cracks start developing at the point of impingement and spreads all along the film as shown in Figure 3.28. The high magnification SEM image taken at the center of the film shows flakes of different dimensions [Figure 3.29]. The cracks probably originate from the strain induced by the difference in thermal expansion between the substrate and the film. A detailed investigation on the crack surface of S3 was done by atomic force microscopy.

42

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Figure 3.30. The Atomic force microscopy images (AFM) of electron bombarded samples S3 taken at different magnifications

At the point of impingement, the surface is found to be smooth with large flakes separated by fine cracks [Figure 3.30 (e)]. However, the AFM image of a single flake shows a dip at the center with slight elevation along the edges [Figure 3.30 (f)]. In addition, high magnification AFM image of individual flakes shows randomly dispersed nanoparticles within the flake [Figure 3.30 (g)]. The surface crystallite size of the nanoparticles determined from the height image was found to be ~ 25.0 nm [Figure 3.30 (h)].

3.8.5. Luminescence in electron bombarded samples Apart from the changes in the physical properties, the electron bombardment also gives rise to luminescence. The electron bombarded sample S3 shows intense red emission at λ = 619.0 nm. The photographs of the as -deposited (S1), colored (S2), and light emitting (S3) samples are shown in Figure 3.31 (a) - (c). Interestingly, the light emission could not be ascertained to the conventional cathodoluminescence expected from electron bombardment. Rather the emission is found to be photoluminescence (PL) from the structural modifications

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

43

(tungsten oxide nanoparticles) and defect sites induced by electron bombardment. Figure 3.31 (d) shows the PL spectrum of S3 sample excited at 325.0 nm. The band to band emission was observed at 357.0 nm; for the sake of clarity, the spectrum showing red emission alone is presented here. The intense red emission is attributed to the transitions due to localized sites induced by the presence of oxygen vacancies and defects in WO3 film [3.50]. All the above results clearly indicate that the electron bombardment of WO3 thin films leads to significant changes in the structural, optical and morphological properties.

Figure 3.31. The Photographs of the (a) as-deposited sample S1; b) electron bombarded sample S2; and (c) electron bombarded sample S3. (d) PL spectrum of sample S3 excited at λ = 325 nm. (e) Outline for the data storage application based on the coloration/luminescence of WO3 films by electron bombardment

With the optimization of nanostructured electron emitters (nanorods, nanowires and nanotubes), the coloration and luminescence induced by electron bombardment can be used for data storage. The concept for data storage using electron bombardment is shown in Figure 3.31 (e). The colored /luminescent portion of the film can be treated as binary ‘1’ and the asdeposited portion as ‘0’ or vice versa. As the electrons can be focused to a fine beam (~ a few tens of nanometers), it is possible to achieve ultra high-density data storage using this method, if optimized.

44

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

4. ZINC OXIDE FOR TRANSPARENT ELECTRONICS (p-type CONDUCTING) 4.1. Summary Zinc oxide (ZnO) thin films have attracted attention mainly due to their stability and cost compared to the conventional tin doped indium oxide (ITO) thin films, it is contemplated that ZnO may replace ITO. The basic physics for all these transparent conducting oxide (TCO) remains the same. However, the difficulties encountered in producing p type transparent conducting oxides (TCOs) in general, and ZnO, in particular, is posing a challenge to the research community. There are several approaches followed to dope the ZnO for p- type conduction. The present section summarizes the basic physics of p- type TCOs and the efforts on the p- type doping in ZnO thin films.

4.2. Introduction Transparent Conducting Oxide (TCO) thin films have a history of over sixty years and a commercial production exceeding a few Billion US Dollars; and it is still a topic of intense study in several research laboratories [4.1]. The present day industry mainly concentrates on tin doped indium oxide (ITO) for TCO. But Zinc Oxide (ZnO), with abundance in nature (which makes it cost effective) and higher motilities than ITO is a possible replacement for it in near future [4.2]. It is well known that, TCO thin films are direct and wide band gap degenerate semiconductors. The stiochiometric metal oxides are all insulators; the conductivity in TCOs arise because of the oxygen vacancies (metal rich). Thus all TCOs are n- type semiconductors. The conduction carrier concentration can be enhanced by replacing the metal with a high valence element (for example, tin can replace indium); the replacement can be achieved more efficiently if the sizes of the dopant element matches with the metal. The transport mechanism in these TCO films, in general, is dominated by ionized impurity scattering. More details of the carrier generation and transport mechanism have been reviewed by Bel Hadj Tahar et al [4.3]. If p- type TCOs are prepared, a transparent p-n homojunction can be fabricated. A transparent homojunction is a good building block for efficient and new class of optoelectronic devices. There are a few reports on p type TCOs; however, the questions related to reliable and repeatable process and stability are still to be worked out [4.4, 4.5]. The requirements for a p-type transparent conducting oxide (TCO), as may be found in the literature, are: (i) chemical modulation of valence band (CMVB); and (ii) the electronic configuration of the cationic species should preferably have a d10 s0 (d10 s2 systems did not yield positive results so far) [4.6]. It may be noted that for metal oxides, the valence band is commonly the oxygen 2p band. Even though doping in this band is not difficult, the mobility of such holes is found to be quite low [4.6]. One way to obtain reasonable motilities in a ptype TCO would be to dope the valence band of the cation d or s band. The cation s band may offer the possibility of higher mobility than cation d bands and the 4d and 5d bands are still

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

45

better than the 3d bands. The question is how to achieve this selective doping in the preparation of the material, particularly, in the thin film form. The possible systems suggested for p- type TCOs are the materials with crystal structures of Delaffosite (CuAlO2 and AgInO2) and of the Spinels (ZnO). It may be noted that it is very difficult to grow thin films with these crystal structures. The fundamental issue in these ptype TCO thin films is to realize a large concentration of holes in the structure. If the hole is visualized as a localization of the covalent bond, then one has to control the oxygen reactivity with the metal ion to support the localization. In Delaffosites, it is rather difficult as the oxygen vacancies are inherent and contribute to electrons. However, holes may be realized in these materials through the multivalence of the copper and indium in the Delaffosite structures. The experimental results support this view on silver oxide [4.7] and on AgInO2 thin films prepared by reactive electron beam and DC Magnetron sputtering techniques [4.8]. The important result of the investigation is that the oxygen bonding to the indium and silver (in both indium doped silver oxide and silver doped indium oxide) is very sensitive and the localization of the valence bonds occur in a very narrow range of the growth conditions [4.9]. Thus the window for obtaining p- type behaviour seems to be very small and very sensitive to the experimental parameters. There are many questions related to the concepts of p – type TCO thin films. for example: (i) a good and confirmed recipe to prepare a p-type TCO thin films, (ii) the origin of the holes, (iii) the electrical transport mechanism (iv) the noise in the fabricated p-n junction and (v) the intricacies of measurement technique to estimate the carrier (hole) concentration and many more [4.10]. In these p- type TCO thin films there exists a vast potential for condensed matter theorist: in identifying possible material systems, theoretical modelling of the practically grown material with defects, calculation of band gaps and the nature of the band gap (direct or indirect), estimation of maximum limit for hole density, mobility and high optical transparency (in the visible spectrum) in the material systems, optimization methods to enhance the performance (the density and mobility of holes) and possible operative mechanism for these materials to possess radiation resistance property etc, [4.11].

4.3. Measurements on p- type TCOs In all the TCO thin films, particularly in the multivalent metal ion systems (Silver, Indium and copper) oxygen plays a very important role in realizing both n and p-type as shown in our earlier experiments with Silver oxide AgO [4.7]. In order to achieve the p- type conductivity in TCO films, one has to modulate the bond energy of the oxygen to the metal ion, while doing so the oxygen vacancies may still remain in the lattice. Thus, the intrinsic carrier concentration (ni) remains almost the same in the wide band gap oxide semiconductor and the electron (n) and hole (p) product remains at (thermodynamic equilibrium) n p = ni2

4.1

The holes are incorporated into the system at the expense of the electrons. At any given instant of thermodynamic equilibrium, both electrons and holes will be present in the system. For such a system, the measurement of Hall effect is rather tricky.

46

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

The recent technique: Quantitative Mobility Spectrum Analysis (QMSA)[4.12] is a powerful method for analyzing bi-polar conduction (conduction by both electrons and holes simultaneously). The basic principle of the QMSA measurement is that the mobilities of electrons and holes depend upon the magnetic fields. It may be noted that even though the contribution of electrons and holes for conductivity is independent of the applied magnetic fields (magneto resistance), the resultant combined effects, due to velocity distribution, may be dependant on the magnetic fields, provided, the magnetic field intensity is very high. Thus, the magnetic field dependant Hall measurements can effectively separate the electron and hole contributions. . The measurements done with QMSA are very well suited (and in fact QMSA gives a vast amount of information on transport) for all semiconductors with bi-polar conduction of carriers having reasonably high mobilities.

4.4. Progress in the Understanding of the Origin of Holes and p- type Conduction The origin of holes in these p- type TCO films is yet to be understood in detail. In the opinion of the authors, Sanderson’s theory of partial ionic charge may be one of the approaches to account for the holes in these p- type TCO thin films. As it is well known, the existence of holes is due to the localization of valence bands in the material. Considering the material systems: elemental metal oxides, delaffosites and spinels, it may be rather difficult, from the conventional understanding, to visualize the presence of covalent bonds. The feature of an atom which control its physical and chemical properties are: its electronic configuration and the “effective nuclear charge” felt by the valence electrons. The effective nuclear charge is the positive charge that would be felt by a foreign electron on arriving at the periphery of the atom. Many atomic properties may be correlated with effective nuclear charge; for example, (i) the ionization potentials gradually increase from left to right across the periodic table, (ii) electron affinities become increasingly negative in the same direction and (iii) electro negativities increase from left to right. An important contribution to the understanding of bond formation is the principle of electro-negativity equalization (proposed by Sanderson) which states that “when two or more atoms initially different combine chemically, they become adjusted to the same intermediate electro-negativity within the compound”. The partial charge is defined by the change in electro-negativity undergone by an atom on bond formation to the change it would have undergone on becoming completely ionic with charge + or – 1. This theory of coordinated polymeric model of Sanderson is a bridge between the ionic and covalent extremes of bond types. Though the proposed theory is empirical, experimental measurements have confirmed the existence of partial ionic charges in many compounds (examples of calculations have been given in reference 4.7). With these ideas, one may realize the p- type conduction by controlling of oxygen reactivity and the valence state of silver in the silver indium oxide thin films. The Ag: InO thin films have been prepared by two techniques: electron beam evaporation and reactive DC magnetron sputtering [4.8, 4.9, 4.13]. In the electron beam evaporation, the electron filament current is varied keeping the oxygen pressure in the growth chamber constant. In DC magnetron sputtering, the oxygen flow rate is varied keeping the magnetron power density constant. In both the techniques, the p - type conductivity with reasonable transmission in the

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

47

visible region (400 -900 nm) has been achieved. In order to notice the narrow window of the growth / process parameters, nu*merical data are presented in Tables 1 and Table 2 respectively for the reactive electron beam evaporated and reactive DC Magnetron sputtered silver indium oxide thin films.

Substrate temp.(°C)

Filament current (mA)

Evaporation rate (nm/s)

Thickness (nm)

Refractive index (at 632.8 nm)

Eg (eV) hν (αhν)2

Hall mobility (cm2 v−1 s−1)

Hall coefficient (cm3/C) and type Of charge carrier

Type of charge carrier from hot probe th d Work function (eV)

Table 4.1. Electrical and optical properties of silver doped indium oxide thin films prepared by reactive electron beam evaporation [4.8]

50

10

0.05

192.0

1.208

1.49

1.69

−1.47×10−3

P

4.686

15

0.89

186.0

1.201

2.00

0.85

−3.95×10−3(

N

4.395

20

2.67

157.0

1.192

−

5.60

−1.2210−3

30 200

154.0

1.196

–

7.45

4.672

−0.97×10

P

4.763

−3(N)

40

16.0

156.5

1.188

1.73

0.89

−6.13×10

N

4.500

10

0.05

155.5

1.217

0.96

6.03

−1.69×10−4(N)

P

4.720

1.29

−2.28×10

−3(N)

N

4.545

−0.74×10

−4(N)

P

4.623

15 20

250

5.33

P

−4(N)

0.89 2.67

160.0 164.0

1.177 1.164

1.69 –

3.13

−4(P)

30

5.33

147.0

1.205

–

1.52

0.93×10

P

4.593

40

16.0

163.5

1.235

1.21

1.59

−1.2510−4(N)

P

4.541

10

0.05

166.5

1.159

0.76

1.63

5.99(P)

15

0.89

155.0

1.192

1.98

3.02

P

4.828

−1.38×10

−3(N)

N

4.683

−4(N)

20

2.67

146.5

1.184

–

14.77

−0.78×10

P

4.740

30

5.33

138.5

1.185

–

6.36

−1.59×10−4(N)

P

4.651

4.58

−4(N)

P

4.633

40

16.0

161.5

1.191

0.81

−0.87×10

4.5. Systems with Spinel Structure As it is well known, the Spinel belongs to cubic system; the unit cell contains 32 cubic close packed Oxygen ions, 08 divalent cations of A and 16 trivalent cations of B. There are equivalent positions in this cell for eight tetrahedral cation sites and for sixteen cation sites. The total number of tetrahedral and octahedral vacancies between the oxygen layers in the unit cell are 64 and 32 respectively. Thus, 7/8 of the tetrahedral and ½ of the octahedral holes remain vacant. The Spinel structure contains rutile chains running along <110> directions giving a free path to electrons (high electronic motilities). The spinel lattice offers excellent possibilities for electron doping. The material systems belong to the spinel are : ZnAl2O4, ZnGa2O4, ZnIn2O4, CdAl2O4, CdGa2O4 and CdIn2O4. For a typical ZnAl2O4 the composition of Zn and Al are : 29.4 % and 55.6 % respectively and the oxygen concentration is about 30 35%; the crystal system is Isometric Hexoctahedral with Fd3m Space Group and has eight formula units per unit cell. These materials are nonmagnetic.

48

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

Resistivity (10-2 Ohm.cm)

Hall coefficient (10-3 cm3/coul)

Hall mobility (cm2v-1s-1)

Work function (eV)

Refrac-tive index (at 632.8 Nm)

Eg(eV) α2~hν plot

Eg (eV) (αhν)2~hν plot

0.00 0.54 1.09 1.71 2.36 2.86 Magnetron power (Watts) at 1.09sccm 40 50 60 70 80

Thick ness (nm)

Oxygen flow rates (sccm) at 40 watt

Table 4.2. Electrical and optical properties of reactive DC Magnetron sputterd Silver indium oxide thin films.

215.0 169.2 199.2 149.6 186.5 214.2

0.28 0.21 7.25 125.52 -

-9.137 -6.345 -43.70 +25.86 -

3.256 2.98 0.60 0.206 -

4.552 4.478 4.644 4.694 4.767 4.783

1.133 1.194 1.159 1.220 1.132 1.151

0.85 1.39 1.36 0.50 0.55

1.79 1.78 1.83 1.83 1.32 2.92

199.2 141.7 206.7 144.0 200.6

7.25 5.14 3.39 0.14 0.12

-43.7 -37.9 -21.1 -5.6 -1.2

0.60 0.73 0.62 4.07 1.04

4.644 4.635 4.577 4.575 4.549

1.159 1.188 1.175 1.205 1.181

1.39 1.32 1.37 1.23 1.11

1.83 2.34 2.47 1.70 1.12

4.6. Physical Properties of ZnO Zinc oxide (ZnO) is a II-VI transparent conducting oxide (TCO), with a wide band-gap of ~3.3 eV at room temperature. ZnO has a very high excitonic binding energy of 60 meV (25 meV for GaN), this property makes it so important with respect to the optoelectronic devices performance, even well above the room temperature (RT). A stoichometric ZnO has an empty zinc 4s-band and a filled valence 2p-band resulting in a band gap of 3.3 eV (the two valence 4s electrons of zinc transfer to the oxygen 2p-band). The electrical conductivity of ZnO is due to the intrinsic defects and / or external doping. In the defect structure, the ZnO is not in a stoichometric ratio and there is excess Zn or low amount of O. One of the two valence electrons of the Zn interstitial atom can be easily ionized and act as donors. Usually in a defect structure of ZnO, an increase in O does not give rise to free holes: the acceptor level corresponding to oxygen defect centres is too far from the valence band to be ionized at room temperature or the number of interstitial Zn is always higher than O. The characterization of native point defects in ZnO is still in detailed investigation. For example, experimental evidence for ZnO with an excess of Zn is inconclusive as to whether the dominant defects are metal interstitials or oxygen vacancies. This information is essential to understand the behaviour of the material and to tailor its numerous applications. The first-principles pseudopotential method was used to determine the electronic structure, atomic geometry, and formation energy of native point defects in ZnO [4.11]. The results show that both the Zn and O vacancies are the relevant defects in ZnO. It is found that the most abundant native defects to be Zn and O vacancies depending on the Zn partial pressure.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

49

In its non-stoichometric form, it is an n-type semiconductor, and could be grown with resistivity as small as ~10-6 ohm m, i.e., a resistivity only 1-2 orders of magnitude larger than that of metals. Due to its large (3.37 eV) and direct band gap, ZnO is a potential competitor for GaN-based light-emitting devices in the blue spectral range. However, like for other wide band gap semiconductors controlled p-type doping is the challenge. As grown and undoped ZnO is n-type conducting. To achieve the desired p-type conduction, one requirement is to suppress the residual donors and to avoid any deep level defects which hinder the activity of the potential p-type dopants. On this way a clear atomistic identification of the electrical active species in the material is helpful. These features make ZnO a potential candidate material for use in optoelectronic devices, such as blue LED's and laser diodes (LD's)[4.14]. ZnO has interesting luminescent properties [4.10].The reason for the high stability of ZnO in the presence of reducing plasma and high temperature has been attributed to its high surface binding energy. The important physical properties of ZnO are summarized in these two tables [4.4, 4.14] Table 4.3. Optical properties of ZnO ZnO Optical Properties Bandgap (eV) Room T: ~3.3eV @ 300K Indices of n-= 2.008 (IR) Refraction 2.029 (IR) n|| = Relative Static E-= 7.77] Dielectric Constants Optical E-= 3.70] (measured on bulk)

Low T: 3.437eV @ 2K

E||= 8.91 E||= 3.78

Table 4.4. Physical properties of ZnO Parameter Hole effective mass High-frequency dielectric constant Static dielectric constant Longitudinal lattice constant Piezoelectric coefficient Longitudinal optical phonon energy Acoustic deformation potential Debye temperature Longitudinal elastic constant

Symbol m*p ε∞ ε0 C P ω Ec T cl

Literature value 0.64 × (9.1095 × 10–31) kg 3.72 × (8.8542 × 10–12) (F m–1) 8.12 × (8.8542 × 10–12) (F m–1) 5.207 Å 0.21 72 meV 15 eV 837 K 1.6 × 1011 Pa

Thin transparent and electrically conducting ZnO films were deposited using different deposition techniques, like sputtering techniques [4.15], chemical vapour deposition (CVD) [14.16], molecular beam epitaxy (MBE) [4.17], filtered vacuum arc deposition (FVAD) [4.18], and more [4.19]. In contrast to n-type ZnO, p-type ZnO films can be produced only by doping. The major challenges in realizing p type ZnO are compensation of donor level defects

50

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

in system by the acceptors and the activation of the acceptor dopant. In some specific cases, even if the acceptor level is formed, it can’t be activated if it is deep in nature. The various methods used in growing p-type doped ZnO include standard doping, usually with nitrogen [4.20], co-doping [4.21, 4.22] or diffusion doping from the substrate material or an interface layer of the dopant, using an excimer laser [4.23]. Generally, the grown p-type films have resistivities of (1-400) x 10-3 Ω m, mobility of the order of (0.1-50) x 10-4 cm2/(V·s), and hole concentration of ~1021-1027 m-3. Ohara et al. reported on p-type ZnO thin films obtained by Sb doping, using the excimer laser diffusion method [4.24]. The ZnO layers were epitaxially grown on Si substrates (thickness ~50 nm), and p-type films with resistivity of 8·10-5 Ω m, mobility of 1.5x10-4 m2/ (V·s), and hole concentration of 5x1026 m-3 were reported, using gold electrodes as contacts. No data on the composition or microstructure of the doped films were reported. Kobayashi et al [4.25] have predicted that nitrogen is a candidate for producing a shallow p- type dopant in ZnO. The ab initio calculations of electronic band structure of ZnO by Yamamoto et al [4.26] have shown that simultaneous codoping using reactive donor co-dopants: Al, Ga and In, not only enhances the incorporation of N acceptors but also gives raise to a shallow Nitrogen acceptor levels in a band gap in p- type doped ZnO crystals. Joseph et al [4.27] found that the N doping is effective only with N2O through an ECR plasma source (but not with N2 gas). p- type ZnO films have been prepared at 650o C by chemical vapour deposition by Minegishi et al [4.28] the hole concentration obtained is 1.5 x 1016 /cm3 with a mobility of 12 cm2/V.s. By using As a dopant and employing pulsed laser deposition (PLD), Ryu et al [4.29] have synthesized p – type ZnO. Elina Kaminska et al [4.30] reported the preparation of p-type ZnO by oxidation of thin films zinc nitride. Nitrogen incorporation is seen as the most prospective way to obtain p-type conductivity. The starting material was grown by rf reactive magnetron sputtering from high purity Zn target, in Ar-N2 plasma. Crucial point during the deposition was to adjust both the N2/Ar ratio and the working pressure so as to deposit Zn3N2 without any metallic Zn inclusions. For oxidation, samples were furnace annealed in O2 flow Quartz and sapphire as well as lattice-matched GaN, ZnO and ZrB2 were used as substrates. The effect of the growth direction - determined by growth conditions - on the structural perfection of the resultant ZnO films as well as on efficiency of nitrogen incorporation was studied. Results of x-ray diffraction, secondary ion mass spectrometry and optical transmission measurements of the oxidized layers, indicating that formation of ZnO did take place. This work reports on the appropriate conditions for the growth of p-type ZnO with Hall effect mobility of 20 cm2/Vs and carrier concentration of 2x1015 cm-3. Further increase of p-type conductivity up to 5x1017 cm-3 with mobility of 5 cm2/Vs was obtained by additional doping the starting material with Cr. The window for obtaining p- type behaviour in ZnO seems to be very small and very sensitive to the experimental parameters.

4.7. ZnO Homojunction In ZnO, the n-type doping is relatively easy to accomplish with elements such as Al, Ga or In. On the other hand, reliable p-type doping is very difficult to achieve. Usually group I and group V elements are used as acceptors. So far, the most common acceptor has been nitrogen because it readily substitutes for oxygen and may not distort the lattice significantly. Besides nitrogen, phosphorous, lithium and arsenic have been successfully used to fabricate

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

51

p-type ZnO. In addition, there are a few reports on intrinsic p-type behaviour in intentionally undoped ZnO, in which the oxygen pressures were optimized. For example, intrinsic p-type ZnO thin films have been deposited by plasma-assisted metal-organic chemical vapour deposition. The optimal results reported are, resistivity of 12.7 ohm cm, Hall mobility of 2.6 cm2 /V s, and a hole concentration of 1.88 x 1017 cm−3. The origin of intrinsic p-type behaviour has been ascribed to the formation of zinc vacancy and some complex acceptor [4.31]. It is also conjectured that the oxygen chemical potential is enhanced by virtue of oxygen plasma, which can lower the formation energy of some acceptor defect, such as zinc vacancy, and thus accounts for the p-type conductivity. ZnO homojunctions deposited by three different deposition techniques are referred here. Xu et al fabricated [4.32] ZnO homojunction light-emitting diode by metal organic chemical vapour deposition. Using NO plasma, p-type ZnO thin films were grown on n-type bulk ZnO substrates. The as-grown films on glass substrates showed hole concentration of 1016–1017 cm−3 and mobility of 1–10 cm2 V−1 s−1. The typical ZnO homojunction showed rectifying behavior with a turn-on voltage of about 2.3 V and electroluminescence at room temperature with band-to-band emission at I=40 mA and defect-related emissions in the blueyellow spectrum range. Jiao et al [4.33] fabricated ZnO p-n junction light-emitting diode on a-plane Al2O3 substrate by plasma-assisted molecular-beam epitaxy. NO plasma activated by a radio frequency atomic source was used to grow the p-type ZnO layer of the LED. The currentvoltage measurements at low temperatures showed a typical diode characteristic with a threshold voltage of about 4.0 V under forward bias. With increasing temperature, the rectification characteristic was degraded gradually, and faded away at room temperature. Electroluminescence band of the ZnO p-n junction LED was located at the blue-violet region and was weakened significantly with increase of temperature. Cao et al [4.34] fabricated ZnO p-n junctions by a simple method of sol-gel spin coating. N-doped and In-N-co-doped ZnO films have been fabricated on quartz glass substrate by solgel spin coating. Their p-type conductivities were characterized by the Hall measurements, revealing low resistivities of the order of 10−1 Ω cm. Thin-film junctions comprising an undoped ZnO layer and a N-doped ZnO layer displayed the typical rectifying characteristics, suggesting formation of p-n homojunctions at the interfaces. However, it is the inexplicit ptype doping mechanism as well as the stability and reproducibility problems that become the constraint factor (as already indicated) in the development of ZnO devices.

4.8. Measurement of Charge Density by Hall Effect in p- type TCO Thin Films The conventional Hall effect experiment, based on the basic principle of the deflection of free carriers in a magnetic field produces the Hall field in a direction perpendicular to the applied magnetic field and current (passing through the sample). The Hall field (Ez) produced account for the majority carriers in the sample Ez = R Ix Bz

4.2

52

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

where Ix is the current passing through the sample, Bz is the magnetic field and Ey is the Hall field produced (the subscripts indicate the Cartesian coordinate directions) and R is the Hall coefficient related to the majority carrier concentration as R = ± (1/ nq)

4.3

n is the density of majority carriers; +ve sign is for holes and –ve sign is for electrons. While deriving the expressions (4.2) and (4.3): (i) the simultaneous contribution of electrons and holes is not considered (ii) the effect of distribution of velocities is not taken into account (iii) it is inherently assumed that the constant energy surfaces (in momentum space) are assumed to be spherical and hence no anisotropy in the effective mass of charge carriers (iv) if there is any anisotropy in the effective masses of charge carriers, the density of states (DOS) function and averaging transport properties over velocities are to be modified. When both electrons and holes are contributing to conductivity in almost competing concentrations, both the type of carriers deflect on to opposite faces of the sample in the conventional Hall experiment; under such experimental conditions, the results of the Hall effect are to be measured with great care and interpretation is to be made with utmost caution. In most of the cases of TCO thin films, the resistivity of the sample is high, then the measurement of Hall coefficient may give erroneous results (See Table 4.1 and 4.2). It may be recalled that the main difference between the well understood p-type doping in Silicon and p- type doping in TCO is that the TCO contains inherently a large number of free electrons due to oxygen deficiency . The other difference is that the Hall mobilities in TCO films are one (or two) order of magnitude less than that of Silicon. The fundamental equation governing the Hall Coefficient when both electron and holes (non extrinsic case) are present in the sample and taking into consideration the distribution of velocities (involving the solution of Boltzmann equation) and if the dominant scattering process is acoustical mode phonon scattering , is given by [4.35]. RH = ( -3π/8ce)[ (p μp2 – n μn2 ) / (p μp + n μn )2 ]

4.4

where μn and μp are the mobilities of the electrons and holes respectively, to be determined independently from other experiments. If the number densities for electrons and holes are comparable, assuming the hole mobility being smaller (because of higher effective mass), the Hall coefficient will be considerably affecting the carrier concentration values, most of the times leading to intrinsic regime, while measuring the Hall coefficient as a function of temperature in mixed conduction process, the intrinsic regime appears more often. The intrinsic property has been reported by Subrahmanyam et al. for In + Ag oxides [4.13]. This is one of the reasons (apart from the contacts to measure the Hall voltage) for several workers to encounter difficulties in the accurate measurement of Hall coefficient in the p – type TCO films.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

53

While measuring the Hall coefficient at low temperatures on the bipolar conduction semiconductors, the behaviour of mobility and the impurity scattering add to the complexity of the measurement. Some investigators followed SIMS (Secondary Ion Mass Spectrometry) to estimate the dopant concentration; however, SIMS gives the dopant (acceptor or donor) concentration in the layers of the materials, but not the active centres that effectively contribute to electrical transport.

4.9. The Future Trends of the TCO Thin Films The flexible electronics is required in several applications where weight and portability are the demands. Presently, the concept of flexible electronics is to develop the conventional circuit on a flexible plastic substrate. The recent advancement in flexible electronic displays has been revealed at Philips [4.36]. There are many projects aiming to develop "electronic paper". Such a display could, for example, be used create a fully updatable newspaper which could rolled up into a coat pocket. Flexible displays could also be used to create new mobile phones and other easily collapsible gadgets. Comparatively little work has been reported on the properties of ITO and AZO (Al doped ZnO)thin films on plastic substrates even though roll-to-roll vacuum deposited ITO films are commercially available. With the recent developments in the high refractive index plastic ophthalmic lenses, the mechanically hard ITO and AZO sputter coatings (5-7 GPa) may have a vast potential in scratch resistant and anti reflection coatings. Process temperatures with common plastic substrates are < 100 ºC, typically near room temperature (RT), which results in amorphous and poor quality of ITO. Typically, the resistivity, ρ, for ITO thin films made with a RT deposition processes is ~ 6 x 10–4 Ω–cm or higher. This high resistivity (low conductivity) is not suitable for many applications such as a transparent electrode for Organic Light Emitting Diode (OLED) displays deposited on flexible plastic film substrates. For several reasons, including performance, cost and availability, finding a TCO replacement for ITO is currently an area of very active research. However, replacement candidate TCO, e.g., Al doped ZnO (AZO) and Ga doped ZnO (GZO), exhibit similar degraded electrical performance when deposited near RT. Additionally, mechanical issues with TCO thin films on plastic substrates also are of key importance. For example, the critical strain for ITO is ~ 1.5%, which limits the useful film thickness and other processing and application parameters. The effect of the deposition method selected, and the choice of process conditions, on the ITO resistivity are to be evaluated. The efficacy of various choices in improving the electrical performance of RT deposited ITO (TCO) will be the focal theme of many investigations to come. For gas sensing applications, the nanostructured materials are recognized as essential for achieving high gas sensitivity. Numerous processing schemes have been tested successfully, albeit only on a laboratory scale. Processing techniques should be able to provide the desired oxide composition with specific dopant and requiring the least number of processing steps. In the processing of nanostructured oxides, more fundamental work is needed to understand the role of nanostructured oxide materials on gas adsorption and conductivity. In general, a detailed understanding is required on the nano structured TCO thin films for not only gas sensing but also in electromagnetic shielding and radiation resistance applications.

54

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

ACKNOWLEDGEMENTS One of the authors (AS) gratefully acknowledges the support received from Professor Eberhard Schultheiss, Director, Fraunhofer Institute fur Elektronenstrahl und Plasmatechnik (FEP), Dresden, Germany and DAAD, Germany.

Notes 1 Present Day Technology for Lung Assisted Devices Four principal development efforts have tackled the challenge of intravascular artificial lungs since the 1980s. Mortensen and colleagues at CardioPulmonics, Inc. (Salt Lake City, UT) developed the IVOX, the only intravascular artificial lung that has undergone human clinical trials.The IVOX consisted of a bundle of crimped hollow fiber membranes joined at the distal end to the inner lumen of a dual-lumen gas conduit, and at the proximal end to the outer lumen of the gas conduit, which led outside the body to a console for providing sweep gas flow through the fibers. The crimped fibers of the IVOX helped minimize fiber clumping in the vena cava and also helped disturb blood flow and diffusional boundary layers on fiber surfaces to improve overall gas exchange permeance. A total of 160 patients with severe acute respiratory distress were studied in applications that lasted up to 28 days of support. The clinically tested IVOX ranged from 0.21 m2 to 0.51 m2 in membrane area, and the average rates of O2 and CO2 transfer accomplished in the trials ranged from 40 to 70 ml/min, or about 20–30% of baseline metabolic needs. The IVOX demonstrated that intravascular artificial lungs can be implanted within the vena cava and perform for extended periods without significant complications in situ (for example, from blood thrombosis). The implantable Intrathoracic Artificial Lung (ITAL) under development at Northwestern University focuses on resting the lung in acute respiratory failure and as a bridge-to-lung transplantation in chronic lung failure. Mathematical models were developed to estimate the required surface area for 200 ml/min of oxygen transfer at a blood flow rate of 5 l/min with a pressure drop of less than 15 mm Hg. The BioLungTM total artificial lung under development at MC3, Inc. (Ann Arbor, MI) and the University of Michigan is intended for complete respiratory support as a bridge to transplant for 1–6 months. A paracorporeal total artificial lung for chronic respiratory support (Chronic Artificial Lung, or CAL) is under development at the University of Maryland. The CAL is intended as a bridge-to- transplant device with the goal of 21-day support of basal metabolic needs using a device less than 0.5 m2 in fiber membrane area. The CAL uses active mixing from a rapidly rotating disc made of microporous hollow fiber membranes that enhance gas exchange by increasing blood flow velocity past fiber surfaces and reducing diffusional boundary layers. Hb + O2 <-> HbO2 HbO2 + O2 <-> Hb(O2)2 Hb(O2)2 + O2 <-> Hb(O2)3

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

55

Hb(O2)3 + O2 <-> Hb(O2)4

Summary Reaction Hb + 4O2 -> Hb(O2)4

REFERENCES [A1] Metal Oxides: Chemistry and Applications (Chemical Industries), Fierro, J. L. G. (CRC Press, 2005). [A2] The Surface Science of Metal Oxides, Victor, E; Henrich, PA. Cox, Cambridge University Press, 1996. [A3] The Defect Chemistry of Metal Oxides, Smyth, DM, Oxford University Press, 2000. [A4] Parkin SP; et al. Giant tunnelling magnetoresistance at room temperature with MgO (100) tunnel barriers, Nature Materials, 2004, 3, 862. [1.1] Federspeil, WJ; Svitek, RG. Lung, Artificial: current research and future directions, Encyclopaedia of Biomaterials and Biomedical Engineering, Ed: Wnek, G E and. Bowling, G.L. 2008, pp 1673-1682. [1.2] Titanium in Medicine, Ed by Bruneyye, DM; Tengvall, P; Textor, M; Thomas, P. Springer, 2001. [1.3] Fujishima, A; Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode, Nature, 1972, 238, 37. [1.4] Diebold, U. The surface science of titanium dioxide, Surf. Sci. Reports, 2003, 48, 53. [1.5] Zubkov, T; Stahl, D; Thompson, TL; Panayotov, D; Diwald, O; Yates JT. Jr., Ultraviolet light-induced hydrophilicity effect on TiO2(110)(1x1). Dominant role of the photooxidation of adsorbed hydrocarbons causing wetting by water droplets, J. Phys. Chem. B, 2005, 109, 15 454. [1.6] Hashimoto, K; Irie, H; Fujishima, A. TiO2 Photocatalysis: A Historical Overview and Future Prospects, Jpn J Appl Phys Part 1, 2005, vol 44, 8269.

56

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

[1.7] Akihiko, K; Yugo, M. Heterogeneous photocatalyst materials for water splitting, Chem. Soc. Rev., 2009, Vol 38, 253. [1.8] Subrahmanyam, A; Arockia Doss, T; Paul Ramesh, T. Studies on the Oxygenation of human blood by Photocatalytic action, Artificial Organs, 2007, vol 31, pp 819. [1.9] Subrahmanyam, A; Ramesh, TPJ; Ramakrishnan, N. Oxygenation of human blood using photocatalytic reaction, ASAIO Journal, 2007, vol 53 (4), pp 434. [1.10] Martin, PM; Monzyk, BF; Burckle, EC; Bbusch, JR; Gilbert, RJ; Dasse, KA. Progress towards development of a photolytic artificial lung, Materials science & engineering. B, Solid-state materials for advanced technology, 2005, vol. 119, 246. [1.11] K.A. Dasse, B.F.Monzyk, E.C.Burckle, J.R.Busch, R.J.Gilbert, Dvelopment of a photolytic artificial lung: preliminary concept validation,., ASAIO J. vol 49, 556 (2003). [1.12] Artificial Infusion of Oxygen into human blood by photocatalytic reaction” Indian patent Application No: 427 CHE, 2006 dated 10 March 2006. [1.13] Photolytic artificial lung, US Patent 6866755B2 (2005). [1.14] Photolytic artificial lung, US Patent 7485261 B2 (2009). [1.15] Oxygen generation in whole blood by photolytic activation, US Patent 7399717B2 (2008). [1.16] Artificial Pulmonary capillary, US Patent 2004/ 0243051A1 (2004). [1.17] Marino, PL. The ICU Book, 2 Edition, Williams & Wilkins, Baltimore MD, 1998, pp 20-21. [1.18] Visible and near infrared absorption spectra of human and animal haemoglobin, W.G.Zijlstra, A. Buursma and O.W.van Assendelft (published by VSP Nederland, 2000. [2.1] Antelman, MS. US Patent, 1993, 5.211.855. [2.2] Tominaga, J., “The application of silver oxide thin films to plasmon photonic devices” J. Phys.: Condens. Matter., 2003, 15, R1101-R1122. [2.3] Peyser, LA; Vinson, AE; Bartko, AP; Dickson, RM. Photoactivated Fluorescence from Individual Silver Nanoclusters, Science, 2001, 291, 103. [2.4] Fujimaki, M; Awazu, K; Tominaga, J. Surface-enhanced Raman scattering from Ag nanoparticles formed by visible laser irradiation of thermally annealed AgOx thin films, J. Appl. Phys., 2006, 100, 074303-1. [2.5] Barik U K, Srinivasan S, Nagendra C L, and Subrahmanyam A, “Electrical and optical properties of reactive DC magnetron sputtered silver oxide thin films: role of oxygen”, Thin Solid Films, 2003, 429, 129 -134. [2.6] JCPDS card 76-1489. [2.7] JCPDS card 75-0969. [2.8] JCPDS card 84-1108. [2.9] Bohern, CF; Huffman, DR. Absorption and scattering of light by small particles, John Wiley, New York, 1983. [2.10] Anderson, PC; Rowlen KL. “illiant Optical Properties of Nanometric Noble Metal Spheres, Rods, and Aperture Arrays”, Appl. Spectrosc., 2002, 56, 24A-135A. [2.11] Nicewarner-Pena, SR; Freeman, RG; Reiss, BD; He, L; Pena, DJ; Walton, ID; Cromer, R; Keating, CD; Natan, MJ. “Submicrometer Metallic Barcodes”, Science, 2001, 294, 137-141. [2.12] Mooradian, A. “Photoluminescence of metals”, Phys. Rev. Lett., 1969, 22, 185-187.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

57

[2.13] Apell, P; Monreal, R; Lundqvist, S. “Photoluminescence of Noble Metals”, Phys. Scr., 1988, 38, 174-179. [2.14] Newman, DM; Panchmatia, P. “Nano silver oxide: a rewritable optical recording medium”, IEE Proc.-Sci. Meas. Technol., 2003, 150, 214-217. [2.15] Subrahmanyam, A; Suman Kumar, P. “Nano silver oxide thin films for optical memories : New results”, IETE Journal of Research, 2006, 55, 365-368. [2.16] Al-kuhaili, MF., “Characterization of thin films produced by the thermal evaporation of silver oxide” J. Phys. D: Appl. Phys., 2007, 40, 2847-2853. [2.17] Peyser, LA; Lee, TH; Dickson, RM. “Mechanism of Agn Nanocluster Photoproduction from Silver Oxide Films”, J. Phys. Chem. B, 2002, 106, 7725-7728. [2.18a]Varkey, AJ; Fort, AF. “Some optical properties of silver peroxide (AgO) and silver oxide (Ag2O) films produced by chemical-bath deposition”, Sol Energy Mater.Sol. Cells, 1993, 29, 253-259. [2.18] Chih Min Chuang, Wu, Wei Fang Su, MC; Cheng, KC; Chen, YF., “High intensity fluorescence of photoactivated silver oxide from composite thin film with periodic array structure”, Appl. Phys. Lett., 2006, 89, 061912-1- 061912-3. [2.19] Dellasega, D; Facibeni, A; Fonzo, FD; Russo, V, Conti, C; Ducati, C; Bassi, AL; Bottani, CE.,” Nanostructured high valence silver oxide produced by pulsed laser deposition”, Appl. Surf. Sci., 2009, 255, 5248-5251. [2.20] Ravi, N. Chandra Raju, Jagadeesh Kumar, K; Subrahmanyaml, A. “Physical Properties of silver oxide thin films by pulsed laser deposition: effect of oxygen pressure during growth”, J. Phys. D: Appl. Phys. (Manuscript under publication). [2.21] Deb, A; Chatterjee, AK, “The electronic structure and chemical bonding mechanism of silver oxide”, 1998 J. Phys.: Condens. Matter, 1998, 10, 11719- 11729. [2.22] West, AR. Solid State Chemistry and its Applications, Revised Edition, John Wiley, New York, 1984, p.290. [2.23] Abdelhamid Maali, Thierry Cardinal, Mona Treguer-Delapierre, “Intrinsic Fluorescence from individual silver nanoparticles”, Physica E, 2003, 17, 559 – 560. [2.24] Burgel, C; Mitric, R; Bonacic –Koutecky, V., “Emissive properties of silver particles at silver oxide surface defects”, Appl. Phys. A, 2004, 82, 117-123. [2.25] Gleitsmann, T; Stegemann, B; Bernhardt, TM., “Femto second- laser –activated fluorescence from silver oxide nanoparticles”, Appl. Phys. Lett., 2004, 84, 4050-4052. [2.26] Neal, T D; Okamoto, K; Scherer, A., “Surface plasmon enhanced emission from dye doped polymer layers”, Opt. Express, 2005, 13, 5522-5527. [2.27] Lynn Anne Capadona, “Photoactivated Fluorescence of small silver nano clusters and their relation to Raman spectroscopy” A Thesis, Georgia institute of Technology, July, 2004, [http://etd.gatech.edu/theses/available/etd-07022004-120153/]. [2.28] Gryczynski, I; Malicka, J; Jiang, W; Fischer, H; Chan, WCW; Gryczynski, Z; Grudzinski, W; Lakowicz, JR., “surface-plasmon-coupled emission of quantum dots”, J. Phys. Chem. B, 2005, 109(3), 1088-1093. [3.1] Granqvist, CG. Handbook of Inorganic Electrochromic Materials, Elsevier, Amsterdam, 1995, pp. 139. [3.1a] Granqvist CG. Handbook of Inorganic Electrochromic Materials, Elsevier, Amsterdam, 1995, pp. 146 [3.2] Granqvist, CG; Azens, A; Heszler, P; Kish, LB; Österlund, L. Nanomaterials for benign indoor environments: Electrochromics for “smart windows”, sensors for air

58

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

quality, and photo-catalysts for air cleaning, Sol. Energy Mater. Sol. Cells, 2007, 91, 355-365. [3.3] Faughnan, BW; Crandall, RS; Heyman, PM. The electrochromic properties of tungsten trioxide film RCA Rev., 1975, 36, 177. [3.4 Lee, SH; Cheong, HM; Tracy, CD; Mascarenhas, A; Benson, DK; Deb, SK. Raman spectroscopic studies of electrochromic a-WO3, Electrochim. Acta, 1999, 44, 3111-3115. [3.4a] Lee, SH; Cheong, HM; Tracy, CE; Mascarenhas, A; Czanderna, AW; Deb, SK. Electrochromic coloration efficiency of a-WO3-y thin films as a function of oxygen deficiency, Appl. Phys. Lett., 1999, 75, 1541-1543. [3.5] Berggren, L; Niklasson, GA. Optical charge transfer absorption in lithium-intercalated tungsten oxide thin films, Appl. Phys. Lett., 2006, 88, 081906-1-081906-3. [3.6] Subrahmanyam, A; Karuppasamy, A; Suresh kumar, C. Oxygen-sputtered Tungsten oxide Thin Films for Enhanced Electrochromic Properties, Electrochem.Solid-State Lett., 2006, 9, H111-H114. [3.6a] Subrahmanyam, A; Karuppasamy, A. Optical and electrochromic properties of oxygen sputtered tungsten oxide (WO3) thin films, Sol. Energy Mater. Sol. Cells, 2007, 91, 266-274. [3.7] Asanuma, T; Matsutani, T; Liu, C; Mihara, T; Kiuchi, M. Structural and optical properties of titanium dioxide films deposited by reactive magnetron sputtering in pure oxygen plasma, J. Appl. Phys., 2004, 95, 6011-6016. [3.8] Denesuk, M; Uhlmann, DR. Site-Saturation Model for the Optical Efficiency of Tungsten Oxide-Based Devices J. Electrochem. Soc., 1996, 143, L186-L188. [3.9] Subrahmanyam, A; SureshKumar, C; Muthu Karuppasamy, K. Sol. Energy Mater. Sol. Cells, 2007, 91, 62. [3.10] Granqvist, CG. Electrochromic tungsten oxide films: Review of progress 1993- 1998, Sol. Energy Mater. Sol. Cells, 2000, 60, 201. [3.11] Hashimoto, S; Matsuoka, H. Mechanism of electrochromism for amorphous WO3 thin films, J. Appl. Phys., 1991, 69, 933. [3.12] Hashimoto, S; Matsuoka, H. Mechanism of electrochromism for amorphous WO3 thin films, J. Appl. Phys., 1991, 69, 933-937. [3.13] Berggren, L; Ederth, J; Niklasson, GA. Electrical conductivity as a function of temperature in amorphous lithium tungsten oxide, Sol. Energy Mater. Sol. Cells, 2004, 84, 329-336. [3.14] Lee, K; Seo, WS; Park, JT. Synthesis and optical properties of colloidal tungsten oxide nanorods, J. Am. Chem. Soc., 2003, 125, 3408-3409. [3.15] Li, Y; Bando, Y; Golberg, D. Quasi-Aligned Single-Crystalline W18O49 Nanotubes and Nanowires Adv. Mater., 2003, 15, 1294-1296. [3.16] Galléa, F; Li, Z; Zhang, Z. Growth control of tungsten oxide nanostructures on planar silicon substrates Appl. Phys. Lett., 2006, 89, 193111-1 – 193111-3. [3.17] Zhou, J; Gong, L; Deng, SZ; Chen, J; She, JC; Xu, NS; Yang, R; Wang, ZL. Growth and field-emission property of tungsten oxide nanotip arrays, Appl. Phys. Lett., 2005, 87, 223108-1 – 223108-3. [3.18] Liao, CC; Chen, FR; Kai, JJ. WO3-x nanowires based electrochromic devices, Sol. Energy Mat. Sol. Cells, 2006, 90, 1147-1155.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

59

[3.19] Feng, M; Pan, AL; Zhang, HR; Li, ZA; Liu, F; Liu, HW; Shi, DX; Zou, BS; Gao, HJ. Strong photoluminescence of nanostructured crystalline tungetn oxide thin films, Appl. Phys. Lett., 2005, 86, 141901-1 – 141901-3. [3.20] Ponzoni, A; Comini, E; Sberveglieri, G; Zhou, J; Deng, SZ; Xu, NS; Ding, Y; Wang, ZL. Ultrasensitive and highly selective gas sensors using three-dimensional tungsten oxide nanowire networks, Appl. Phys. Lett., 2006, 88, 203101-1 – 203101-3. [3.21] Rothschild, A; Sloan, J; Tenne, R. Growth of WS2 Nanotubes Phases, J. Am. Chem. Soc., 2000, 122, 5169-5179. [3.22] He, Y; Wang, S; Huang, B; Liu, CT; Liaw, PK. Novel tungsten oxide microneedles with nanosized tips, Appl. Phys. Lett., 2006, 88, 223107-1 – 223107-2. [3.23] Zhao, YM; Li, YH; Ahmad, I; McCartney, DG; Zhu, YQ; Hu, WB. Two-dimensional tungsten oxide nanowire networks, Appl. Phys. Lett., 2006, 89, 133116-1 – 133116-3 [3.24] Gu, G; Zheng, B; Han, WQ; Roth, S; Liu, J. Tungsten oxide nanowires on tungsten substrates, Nano Lett., 2002, 2, 849-851. [3.25] Wang, SJ; Chen, CH; Ko, RM; Kuo, YC; Wong, CH; Wu, CH; Uang, KM; Chen, TM; Liou, BW. Preparation of tungsten oxide nanowires from sputter-deposited WCx films using an annealing/oxidation process, Appl. Phys. Lett., 2005, 86, 263103-1 -263103-3. [3.26] Meda, L; Breitkopf, RC; Haas, TE; Kirss, RU. Investigation of electrochromic properties of nanocrystalline tungsten oxide thin film, Thin Solid Films, 2002, 402, 126-130. [3.27] Deshpande, R; Lee, SH; Parilla, PA; Jones, KM; To, B; Mahan, AH; Perkins, JD; Mitra, S; Dillon, AC. “Optimization of crystalline tungsten-oxide nanoparticles for improved electrochromic applications, Solid State Ionics, 2007, 178, 895. [3.28] Yoo, SJ; Lim, JW; Sung, YE; Jung, YH; Choi, HG; Kim, DK. Fast switchable electrochromic properties of tungsten oxide nanowire bundles, Appl. Phys. Lett., 2007, 90, 173126-1-173126-3. [3.29] Seung-Wan Song and In-Su Kang, A facile fabrication route to tungsten oxide films on ceramic substrates, Sensors and Actuators B: Chemical, 2008, Vol 129, 971-976. [3.30] Baek Yunho; Kijung Yong, Controlled growth and characterization of tungsten oxide nanowires using thermal evaporation of WO3 powder, Journal of physical chemistry. C, 2007, vol. 111, pp. 1213-1218. [3.31] Gang Gu,† Bo Zheng,† WQ; Han,‡ Siegmar Roth,§ and Jie Liu*†, Tungsten Oxide Nanowires on Tungsten Substrates, Nano Letters, 2002, 2 (8), pp 849–851 [3.32] Papaefthimiou1, S; Leftheriotis1 G; Yianoulis, P. Electrochromic phenomena in transition metal oxide thin films prepared by thermal evaporation, Ionics, 1998, Vol 4, pp 321-329. [3.33] Hong, K; Xie, M; Wu, H. Tungsten oxide nanowires synthesized by a catalyst-free method at low temperature, Nanotechnology, 2006, 17, 4830-4833. [3.34] Chen, CH; Wang, SJ; Ko, RM; Kuo, YC; Uang, KM; Chen, TM; Liou, BW; Tsai, HY. The influence of oxygen content in the sputtering gas on the self-synthesis of tungsten oxide nanowires on sputter-deposited tungsten films, Nanotechnology, 2006, 17, 217-223. [3.35] Lou, XW; Zeng, HC. An Inorganic Route for Controlled Synthesis of W18O49 Nanorods and Nanofibers in Solution, Inorg. Chem., 2003, 42, 6169-6171. [3.36] Wagner, RS; Ellis, WC. Vapor-Liquid-Solid Mechanism of Single Crystal Growth, Appl. Phys. Lett., 1964, 4, 89-90.

60

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

[3.37] Klinke, C; Hannon, JB; Gignae, L; Reuter, K; Avouris, P. Tunsgten Oxide Nanowire Growth by Chemically Induced Strain, J. Phys. Chem. B, 2005, 109, 17787-17790. [3.38] Lin, TT; Lichtman, D. The effects of electron bombardment on the surface composition of WO3, oxidized W and Ta foils, J. Appl. Phys., 1979, 50, 1298-1303. [3.39] Zayim, EO; Baydogan, ND. Irradiation effect in WO3 thin films, Sol. Energy Mater. Sol. Cells, 2006, 90, 402-413. [3.40] Zhao, Y; Feng, ZC; Liang, Y; Sheng, HW. Laser-induced coloration of WO3, Appl. Phys. Lett., 1997, 71, 2227-2229. [3.41] Qiu, H; Lu, YF ; Mai, ZH. Electrochromic writing and erasing on tungsten oxide films in air by scanning tunneling microscopy, J. Appl. Phys., 2002, 91, 440-443. [3.42] Aoki, T; Matsushita, T; Suzuki, A; Tanabe, K; Okuda, M. Write-once optical recording using WO2 films prepared by pulsed laser deposition, Thin Solid Films, 2006, 509, 107-112. [3.43] Karuppasamy, A; Subrahmanyam, A. Electron beam induced coloration and luminescence in layered structure of WO3 thin films grown by pulsed dc magnetron sputtering, J. Appl. Phys., 2007, 101, 113522-113526. [3.44] Morita, H; Miura, T; Washida, H. Coloration of Tungsten Oxide Films Induced by Ion or Electron Irradiation, Jpn. J. Appl. Phys., 1981, 20, L323-L325. [3.45] Niklasson, GA; Berggren, L; Larsson, AL. Electrochromic tungsten oxide: the role of defects, Sol. Energy Mater. Sol. Cells, 2004, 84, 315-328. [3.46] Burstein, E. Anomalous Optical Absorption Limit in InSb, Phys. Rev., 1954, 93, 632-633. Moss, TS. Proc. Phys. Soc. London, 1954, Sect. B 67, 775. [3.47] Deepa, M; Singh, P; Sharma, SN; Agnihotry, SA. Effect of humidity on structure and electrochromic properties of sol-gel-derived tungsten oxide films,Sol. Energy Mater. Sol. Cells, 2006, 90, 2665-2682. [3.48] Scarminio, J; Moraes, MAB; Dias, RCE; Rouxinol, FP; Durrant, SF. Tungsten Oxide Films of High Electrochromic Efficiencies Obtained by Deposition, Electro-chem. Solid-State Lett., 2003, 6, H9-H12. [3.49] Santucci, S; Lozzi, L; Passacantando, M; Nardo, SD; Phani, AR; Cantalini, C; Pelino, M. Study of the surface morphology and gas sensing properties of WO3 thin films deposited by vacuum thermal evaporation, J. Vac. Sci. Technol. A, 1999, 17, 644-649. [3. 50]Feng, M; Pan, AL; Zhang, HR; Li, ZA; Liu, F; Liu, HW; Shi, DX; Zou, BS; Gao, HJ. Strong photoluminescence of nanostructured crystalline tungetn oxide thin films, Appl. Phys. Lett., 2005, 86, 141901-1 – 141901-3. [4.1] First world Conference on TCO, Greece, September 2006. [4.2] Second International Symposium on Transparent Oxide Thin Films for Electronics and Optics (TOEO-2), Japan, November 2001, Thin Solid Films, (2002) Vol. 411. [4.3] Bel Hadj Tahar, R; Ban, T; Ohya, Y; Takahashi, Y. Tin doped indium oxide thin films: Electrical properties, J.Appl.Phys., 1998, 83, 2631. [4.4] Look, DC. Progress in ZnO materials and devices, Journal of electronic materials, 2005, Vol. 35, pp 1299-1305. [4.5] Banerjee, AN; Chattopadhyay, K.K. Progress in Crystal Growth and Characterization of Materials, 2005, Vol.50. [4.6] Transparent Conducting Oxides, M R S Bull. 2000, Vol. 25, No 8.

Emerging Concepts and Challenges in Nano Metal Oxide Thin Films

61

[4.7] Barik, UK; Srinivasan, S; Nagendra, CL; Subrahmanyam, A. Electrical and optical properties of reactive DC magnetron sputtered silver oxide thin films: role of oxygen Thin Solid Films, 2003, Vol. 429, pp 129-134. [4.8] Subrahmanyam, A; Baraik, UK. Synthesis of P-type transparent conducting silver: indium oxide (AIO) thin films by reactive electron beam evaporation technique, J.Phys. Chem. Solids, 2005, Vol. 66 pp 817-822. [4.9] Studies on electrical and optical properties of Silver : Indium Oxide (AIO) thin films by reactive electron beam evaporation and reactive DC Magnetron sputtering technique: a Ph D Thesis by Ullash Kumar Barik, submitted to Indian Institute of Technology Madras, Chennai, India, 2005. [4.10] A study of Transparent Zinc Oxide Thin Films, a PhD Thesis by Ou Yi submitted to the Colorado School of Mines, 1993. [4.11] Kohan, AF; Ceder, G; Morgan, D; Van de Walle, CG. First-principles study of native point defects in ZnO, Phys. Rev. B, 2000, 61, 15019. [4.12] Meyer, JR; Hoffman, CA; Antoszewski, J; Faraone, L. Quantitative mobility spectrum analysis of multicarrier conduction in semiconductors Journal of Applied Physics, 1997, Vol. 812, pp709-713. [4.13] Subrahmanyam, A; Baraik, UK. Electrical and optical properties of reactive dc magnetron sputtered silver-doped indium oxide thin films: role of oxygen, Appl. Phys. A, 2006, Vol.84. [4.14] Özgür, Ü; Alivov, YAI; Liu, C; Teke, A; Reshchikov, MA; Doan, S; Avrutin, V; Cho, SJ; Morkoç, H. A comprehensive review of ZnO materials and devices, J .App. Physics, 2005, Vol. 98, 041301. [4.15] Soo Kim, H. Sang-Hun Jeong, Sang Sub Kim and Byung-Teak Lee, Magnetron sputtering growth and characterization of high quality single crystal ZnO thin films on sapphire substrates, Semicond. Sci. Technol., 2004, Vol.19 pp L29–L31. [4.16] Ataev, BM; Mamedov, VV; Omaev, AK; Magomedov, BA. Epitaxial ZnO films grown by RF-assisted low-temperature CVD method, Materials Science in Semiconductor Processing, 2003,Vol. 6, pp 535-537. [4.17] Look, DC; Reynolds, DC; Litton, CW; Jones, RL; Eason, DB; Cantwell, G. Characterization of homo epitaxial p-type ZnO thin films by molecular beam epitaxy, Appl. Phys. Lett., 2002, Vol. 81(10), pp 1830-1832. [4.18] Li, X; Yan, Y; Gessert, TA; Perkins, CL; Young, D; DeHart, C; Young, M; Coutts, TJ. Dependence of Zinc Oxide Thin Films Properties on Filtered Vacuum Arc Deposition Parameters, J. Vac. Sci. Technol. A, 2003, Vol. 21(3), pp 1342-1346. [4.19] Alam, MJ; Cameron, DC. Preparation and properties of transparent conductive aluminium-doped zinc oxide thin films by sol–gel process, J. Vac. Sci. Technol. A 2001, Vol.19 (4), pp 1642-1646. [4.20] Li, X, Yan, Y, Gessert, TA; Perkins, CL; Young, D; DeHart, C; Young, M; Coutts, TJ. p-Type ZnO Thin Films Formed by CVD Reaction of Diethyl zinc and NO Gas , J. Vac. Sci. Technol. A, 2003, Vol. 21(4), pp 1342-1346. [4.21]Joseph, M; Tabata, H; Saeki, H; Ueda, K; Kawai, T. Control of the electric and magnetic properties of ZnO films Physica B, 2001,Vol. 302-303, pp 140-148 . [4.22] Yamamoto, T. "Codoping method for solutions of doping problems in wide-band-gap semiconductors," Phys. Stat. Sol., (A) ,Vol. 193, pp.423–433.

62

A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.

[4.23] Hatanaka, Y; Niraula, M; Nakamura, A; Aoki, T. Excimer laser doping techniques for II–VI semiconductors, Appl. Surf. Sci., 2001, Vol.175-176, pp 462-467. [4.24] Ohara, K; Seino, T; Nakamura, A; Aoki, T; Kominami, H; Nakanishi, Y; Hatanaka, Y. Sb excimer-laser doping in ZnO films prepared by oxidation of sulfide on Si, Applied Surface Science, 2005, Vol.244, pp 369-372. [4.25] Kobayashi, A; Sankey, OF; Dow, JD. Deep energy levels of defects in the wurtzite semiconductors AIN, CdS, CdSe, ZnS, and ZnO, Phys Rev B, 1983, Vol. 28, pp 946 – 950. [4.26]Yamamoto, T and H.K-Yoshida, H.K, Solution Using a Codoping Method to Unipolarity for the Fabrication of p-Type ZnO, Jap. J. Appl.Phys. 1999, Vol. 38, L 166 -169. [4.27] Joseph, M; Tabata, H; Kawai, T. p-Type electrical conduction in ZnO thin films by Ga and N codoping, Jap. J. Appl.Phys., 1999, Vol. 38, pp L 1205 –L1207. [4.28] Minegishi, K; Koiwai, Y; Kikuchi, Y; Yano, K. Growth of p-type Zinc Oxide Films by Chemical Vapour Deposition, Jpn.J.Appl.Phys, 1997,Vol. 36, L 1453 -1455. [4.29] Ryu, YR; Zhu, S; Look, DC; Wrobel, JM; Jeong, HM; White, HW. Synthesis of ptype ZnO films, J.Cryst Growth, 2000, Vol.216, pp 330-334. [4.30] Kaminska, E; Piotrowska, A; Kossut, J; Butke, R; Daniel, W; Golasewska, K; Barcz, A; Jakiela, R; Dynowska, E; Przezdziecka, E.. P-type conducting ZnO: Fabrication and characterization, E-MRS Fall meeting, 2004. [4.31] Zeng, YJ; Ye, ZZ; Xu, WZ; Lu, JG; He, HP; Zhu, LP; Zhao, BH; Che, Y; Zhang, SB. p-type behavior in nominally undoped ZnO thin films by oxygen plasma growth , Appl. Phys. Lett., 2006, Vol. 88, 262103. [4.32] Xu, WZ; Ye, ZZ; Zeng, YJ; Zhu, LP; Zhao, BH; Jiang, L; Lu, JG; He, HP; Zhang, SB. ZnO light-emitting diode grown by plasma-assisted metal organic chemical vapor deposition, Appl. Phys. Lett., 2006, Vol. 88,173506. [4.33] Jiao, SJ; Zhang, ZZ; Lu, YM; Shen, DZ; Yao, B; Zhang, JY; Li, BH; Zhao, DX; Fan, XW; Tang, ZK. ZnO p-n junction light-emitting diodes fabricated on sapphire substratesAppl. Phys. Lett., 2006, Vol. 88, 03191. [4.34] Cao, Y; Miao, L; Tanemura, S; Tanemura, M; Kuno, Y; Hayashi, Y. Low resistivity p-ZnO films fabricated by sol-gel spin coating, Appl. Phys. Lett., 2006, Vol. 88, 251116. [4.35] Putley, EH. Hall effect and related Phenomenon, Butterworths, 1960. [4.36] http://www.newscientist.com/article.ns?id=dn4602.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 63-111

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 2

SPECTROSCOPIC ANALYSIS OF CHEMICAL SPECIES IN CARBON PLASMAS INDUCED BY HIGH-POWER IR CO2 LASER J. J. Camacho1*, J. M. L. Poyato1, L. Díaz2 and M. Santos2 1

Departamento de Química-Física Aplicada. Facultad de Ciencias. Universidad Autónoma de Madrid. Cantoblanco. 28049-Madrid. Spain. 2 Instituto de Estructura de la Materia, CFMAC, CSIC, Serrano 121. 28006-Madrid, Spain

ABSTRACT This chapter describes some fundamentals of laser-induced breakdown spectroscopy (LIBS) and experimental results obtained from ultraviolet-visible-near infrared (UV-VisNIR) spectra induced by laser ablation of a graphite target, developed in our laboratory. Ablation was produced by a high-power IR CO2 pulsed laser using several wavelengths (λ=9.621 and 10.591 µm), power density ranging from 0.22 to 6.31 GW cm-2 and medium-vacuum conditions (typically at 4 Pa). Spatially and time resolved analysis were carried out for the plasma plume. Wavelength-dispersed spectra of the plume reveal the emission of C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and molecular features of C2, CN, OH, CH, N2, N2+ and NH. For the assignment of molecular bands a comparison with conventional emission sources was made. The characteristics of the spectral emission intensities from the different species have been investigated as functions of the ambient pressure, laser irradiance, delay time, and distance from the target. Excitation, vibrational and rotational temperatures, ionization degree and electron number density for some species were estimated. Time-gated spectroscopic studies have allowed estimation of time-of-flight (TOF) and propagation velocities for various emission species.

*

Corresponding author: Email: [email protected]

64

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

1. INTRODUCTION The interaction of a high-energy infrared (IR) pulsed laser beam with solid materials has been investigated extensively over the past several years, due to its significance in technologies such as laser sampling for chemical analysis and pulsed laser deposition of thin film [1-5]. Laser–material interactions involve non-linear complex and collective processes that are not fully understood. Among the many factors the laser irradiance is one of the most important in controlling the mechanism of the laser–material interaction. For example, at high irradiance, a laser induced plasma can be formed above the sample surface which may absorb the incident laser energy, thereby shielding the sample and decreasing the efficiency of laser energy available for mass ablation. The properties and composition of the resulting ablation plume may evolve, both as a result of collisions between particles in the plume and through plume-laser radiation interactions. The laser-target interactions will be sensitively dependent both on the nature and condition of the target material, and on the laser pulse parameters. Subsequent laser-plume interactions will also be dependent on the properties of the laser radiation, while the evolution and propagation of the plume will also be sensitive to collisions and thus to the quality of the vacuum under which the ablation is conducted and/or the presence of any background gas. Optical emission spectroscopy (OES) is a powerful tool to get information on the laserablated species. For laser ablation of carbon, OES studies in different atmospheres are reported and these studies have yielded many interesting results [6-17]. The major parts of this work are already published by us in different journals [18-20]. The objectives of this work are: (1) to show some fundamentals of laser-induced breakdown spectroscopy (LIBS) and, (2) to review of our recent results on LIBS analysis of chemical species in carbon plasmas induced by high-power IR CO2 laser, adding some new results. OES has been used to investigate thermal and dynamical properties of a plume produced by laser ablation of a graphite target at air pressures around 4 Pa. Ablation is performed using a high-power IR CO2 pulsed laser. The emission generated by the plasma in the spectral region 200-1100 nm is due to electronic relaxation of excited C, N, H, O, ionic fragments C+, C2+, C3+, C4+, and molecular features of C2, CN, OH, NH, CH, N2+ and N2. As far as we know, a spectrum so rich in atomic lines belonging to ionized species and molecular features has not been observed previously in similar experiments. Also we analyzed these spectra at different distances from the target along the plasma expansion direction. Finally we present some new results obtained from the time resolved spectroscopic analysis of the laser ablation of a graphite target. From these results temperature, electron densities and ionization degree are obtained. Also we have studied here the spectral emission intensities from different species as functions of the ambient pressure and laser irradiance. Although OES gives only partial information about the plasma particles, this diagnostic technique helped us to draw a picture of the plasma in terms of the emitting chemical species, to evaluate their possible mechanisms of excitation and formation and to study the role of gas-phase reactions in the plasma expansion process.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

65

2. FUNDAMENTALS OF LASER INDUCED BREAKDOWN SPECTROSCOPY (LIBS) Excellent textbooks and reviews about the fundamentals of laser-induced breakdown spectroscopy (LIBS) and examples of various processes are readily available today [21-24]. LIBS, also sometimes called laser-induced plasma spectroscopy (LIPS), is a technique of atomic-molecular emission spectroscopy which utilizes a highly-power laser pulse as the excitation source. LIBS can analyze any matter regardless of its physical state, being it solid, liquid or gas. Because all elements emit light when excited to sufficiently high energy, LIBS can detect different species (atomic, ionic and molecular) and limited only by the power of the laser as well as the sensitivity and wavelength range of the spectrograph/detector. Basically LIBS makes use of OES and is to this extent very similar to arc/spark emission spectroscopy. LIBS operates by focusing the laser onto a small area at the surface of the sample; when the laser is triggered it ablates a very small amount of material which instantaneously generates a plasma plume with temperatures of about 10000–30000 K. At these temperatures, the ablated material dissociates (breakdown) into excited ionic and atomic species. At the early time, the plasma emits a continuum of radiation which does not contain any known information about the species present in the plume and within a very small timeframe the plasma expands at supersonic velocities and cools. At this point the characteristic atomic/ionic and molecular emission lines of the species can be observed. The delay between the emission of the continuum and characteristic radiation is of the order of 1 µs, this is one of the reasons for temporally gating the detector. LIBS is technically similar and complementary to a number of other laser-based techniques (Raman spectroscopy, laserinduced fluorescence etc). In fact devices are now being manufactured which combine these techniques in a single instrument, allowing the atomic, molecular and structural characterization of a sample as well as giving a deeper insight into physical properties. A typical LIBS system consists of a pulsed laser and a spectrometer with a wide spectral range and a high sensitivity, fast response rate and time gated detector. The principal advantages of LIBS over the conventional analytical spectroscopic techniques are its simplicity and the sampling speed.

2.1. Nature of the Plasmas Plasma is an ionized gas, a distinct fourth state of matter. The free electric charges (electrons and ions) make plasma electrically conductive (sometimes more than gold and copper), internally interactive, and strongly responsive to electromagnetic fields. Ionized gas is usually called plasma when it is electrically neutral (i.e., electron density is balanced by that of positive ions) and contains a significant number of the electrically charged particles, sufficient to affect its electrical properties and behaviour. Plasmas occur naturally but also can be effectively man-made in laboratory and in industry, which provides opportunities for numerous applications, including thermonuclear synthesis, electronics, lasers, fluorescent lamps, and many others. Plasma offers three major features that are attractive for applications in chemical-physics: (1) temperatures and energy density of at least some plasma components can significantly

66

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

exceed those in conventional technologies, (2) plasmas are able to produce very high concentrations of energetic and chemically active species (e.g., electrons, ions, atoms, molecules and radicals, excited states, and different wavelength photons), and (3) plasma systems can essentially be far from thermodynamic equilibrium, providing extremely high concentrations of chemically active species and keeping bulk temperature as low as room temperature. These plasma features permit significant intensification of traditional chemical processes, essential increase of their efficiency, and often successful stimulation of chemical reactions impossible in conventional chemistry. Plasmas are found in nature in various forms and are characterized normally by their electron density ne and electron temperature Te. On earth they exist in the ionosphere at height of 70-500 km (density ne = 106 cm-3, Te = 2300 K). Solar wind is another natural plasma originating from the sun with ne = 10 cm-3 and Te = 105 K. The corona which extends around the sun has an electron density ne = 108 cm-3 and its electron temperature is Te = 106 K. Finally, white dwarfs, the final state of stellar evolution, have a ne of 1030 cm-3. In plasma formation, as the temperature of material is raised, its state changes from solid to liquid and then to gas. If the temperature is elevated further, an appreciable number of gas atoms are ionized and go into a high temperature gaseous state in which the charge numbers of ions and electrons are almost the same and charge neutrality is satisfied at a macroscopic scale. When the ions and electrons move collectively, these charged particles interact via Coulomb forces which are long-range forces and decay with the inverse square of the distance between charged particles. Therefore, many charged particles interact with each other by long range forces rather then through short range collision process like in a common gas. This results in different kinds of collective phenomena such as plasma instabilities and wave phenomena [25].

2.2. Local Thermodynamic Equilibrium (LTE). Model for the Plasma Plasma description starts by trying to characterize properties of the assembly of atoms, molecules, ions and electrons rather than individual species. If thermodynamic equilibrium exits, then plasma properties can be described through the concept of temperature. Thermodynamic equilibrium is rarely complete, so physicists have settled for a useful approximation, local thermodynamic equilibrium (LTE). In LTE model it is assumed that the distribution of population densities of the electrons is determined exclusively through collisional processes and that they have sufficient rate constants so that the distribution responds instantaneously to any change in the plasma conditions. In such circumstances each process is accompanied by its inverse and these pairs of processes occur at equal rates by the principle of detailed balance. Thus, the distribution of population densities of the electrons energy levels is the same as it would be in a system in complete thermodynamic equilibrium. The population distribution is determined by the statistical mechanical law of equipartition among energy levels and does not require knowledge of atomic cross sections for its calculation. Thus, although the plasma density and temperature may vary in space and time, the distribution of population densities at any instant and point in space depends entirely on local values of density, temperature, and chemical composition of plasma. If the free electrons are distributed among the energy levels available for them, their velocities have a Maxwellian distribution

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

⎛ m dn v = ne 4π ⎜⎜ ⎝ 2πk B T e

⎞ ⎟⎟ ⎠

3/ 2

⎛ m v2 exp⎜ − ⎜ 2k B T e ⎝

⎞ 2 ⎟ v dv , ⎟ ⎠

67

(2.1)

where m is the electron mass, v is the electron velocity and kB is the Boltzmann constant. For the bound levels the distributions of population densities of neutrals and ions are given by the Boltzmann (2.2) and Saha (2.3) equations

Nj Ni

=

N z +1, k ne g z +1, k = N z,k g z,k

⎛ ( E − Ei ) ⎞ ⎟⎟ , exp⎜⎜ − j gi k T B e ⎝ ⎠

gj

⎛ 2π m k B Te ⎞ ⎟ 2⎜⎜ 2 ⎟ h ⎝ ⎠

3/ 2

(2.2)

⎛ Ip ⎞ exp⎜⎜ − z, k ⎟⎟ , ⎝ k BT e ⎠

(2.3)

where Ni, Nj, Nz+1,k and Nz,k are the population densities of various levels designated by their quantum numbers j (upper), i (lower) and k (the last for the ground level) and ionic charge z and z+1. The term gz,i is the statistical weight of the designated level, Ej and Ei are the energy of the levels j and i and Ipz,k is the ionization potential of the ion of charge z in its ground level k. Equations (2.1)-(2.3) describe the state of the electrons in an LTE plasma. For complete LTE of the populations of all levels, including the ground state, a necessary condition is that electron collisional rates for a given transition exceed the corresponding radiative rates by about an order of magnitude [26]. This condition gives a criterion [27] for the critical electron density of the level with energy ΔE

ne ,crit

5 ⎛α ⎜ ≥ 8 π ⎜⎝ a0

3

⎞ 7 ⎛ ΔE ⎟⎟ z ⎜⎜ 2 ⎠ ⎝ z EH

⎞ ⎟⎟ ⎠

3

⎛ k BT e ⎜⎜ 2 ⎝ z EH

⎞ ⎟⎟ ≅ 1.6 × 1012 Te1 / 2 (ΔE ) 3 , ⎠

(2.4)

where α is fine-structure parameter, a0 is Bohr radius, and EH is the hydrogen ionization potential. In the final form of Eq. (2.4), ne,crit is given in cm-3, Te in K and ΔE in eV. Many plasmas of particular interest do not come close to complete LTE, but can be considered to be only in partial thermodynamic equilibrium in the sense that the population of sufficiently highly excited levels are related to the next ion’s ground state population by Saha-Boltzmann relations, respective to the total population in all fine-structure levels of the ground state configuration [26]. For any atom or ion with simple Rydberg level structure, various criteria were advanced for the minimum principal quantum number ncrit for the lowest level, often called thermal or collision limit, for which partial thermodynamic equilibrium remains valid to within 10%. One criterion with quite general validity is given by Griem [27]:

ncrit

⎡ 10 z 7 ≈⎢ ⎢⎣ 2 π ne

⎛α ⎜⎜ ⎝ a0

⎞ ⎟⎟ ⎠

3

⎤ ⎥ ⎥⎦

2 / 17

⎛ k BT e ⎜⎜ 2 ⎝ z EH

⎞ ⎟⎟ ⎠

1 / 17

.

(2.5)

68

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

2.3. LIB Plasma As we mentioned previously, plasma is a local assembly of atoms, molecules, ions and free electrons, overall electrically neutral, in which the charged species often act collectively. The LIB plasma is initiated by a single laser pulse. If we consider the temporal evolution of LIB plasma, at early times the ionization grade is high. As electron-ion recombination proceeds, neutral atoms and molecules form. A recombination occurs when a free electron is captured into an ionic or atomic energy level and gives up its excess kinetic energy in the form of a photon. LIB plasmas are characterized by a variety of parameters, the most basic being the degree of ionization. A weakly ionized plasma is one in which the ratio of electrons to other species is less than 10%. At the other extreme, high ionized plasmas may have atoms stripped of many of their electrons, resulting in very high electron to atom/ion ratios. LIB plasmas typically, for low power laser intensities, fall in the category of weak ionized plasmas. At high laser power densities, LIB plasmas correspond to strong ionized plasmas.

2.3.1. Initiation mechanism: Multiphoton ionization (MPI) and electron impact ionization (EII) Plasma is initiated by electron generation and electron density growth. The conventional LIB plasma can be initiated in two ways: multiphoton ionization (MPI) and electron impact ionization (EII) both followed by electron cascade. MPI involves the simultaneous absorption of a number of photons n, required to equal the ionization potential IP(A) of an atom or molecule A nhν + A → A+ + e + IP(A); nhν ≥ IP(A),

(2.6)

where n is the number of photons needed to strip off an electron, which corresponds to the integer part of the quantity:

n=

I P + ε osc + 1. hν

(2.7)

Here εosc is the oscillation energy of a free electron in the alternating electric field. Within the classical microwave breakdown theory [28], a free electron oscillates in the alternating electric field E of the laser electromagnetic wave (with frequency ω and wavelength λ (µm)), and its oscillation energy,

e2 E 2 e2 ε osc [eV ] = = I λ2 = 4.67 × 10 −14 I W λ2 , 2 3 W 4mω 4mπc

(2.8)

remains constant. In Eq. (2.8) e is the electron charge and Iw is the laser intensity (irradiance, power density or flux density in W cm-2). The probability of MPI WMPI, by absorbing simultaneously n laser photons to strip off an electron, is expressed by the classical formula [29]

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

[ ]≅ ω n

WMPI s

−1

3/ 2

⎛ ε ⎜⎜1.36 osc IP ⎝

n

69 n

⎧ 6.35 × 10 −14 I W ⎫ ⎞ ⎟⎟ = 1.88 × 1015 λ2 n −1 n 3 / 2 ⎨ ⎬ , (2.9) IP ⎠ ⎭ ⎩

where IP is in eV. Besides, the probability of simultaneous absorption of photons decreases with the number of photons n necessary to cause ionization. EII process consists on the absorption of light photon by free or quasifree electrons, producing electrons with enough kinetic energy e* to ionize one atom or molecule e + nhν + A → e* + A → 2e + A+.

(2.10)

Two conditions must co-exist for EII to initiate: (i) an initial electron must reside in the focal volume; and (ii) the initial electron must acquire energy which exceeds the ionization energy of the material in the focus. These free or quasifree electrons can be produced by the effect of cosmic ray ionization (natural ionization), by means of MPI, or by a breakdown induced in some impurity. In air at atmospheric pressure, the natural electron density is ~103 cm-3 [30]. These electrons in the focal volume gain sufficient energy, from the laser field through inverse bremsstrahlung collision with neutrals, to ionize atoms, molecules or ions by inelastic electron-particle collision resulting in two electrons of lower energy being available to start the process again e*[ε ≥IP(A)] + A → A+ + 2e; e*[ε ≥IP(A+)] + A+ → A2+ + 2e.

(2.11)

The MPI mechanism dominates electron generation only for low exciting wavelengths. Therefore initial EII becomes a problem at a higher wavelength because neither cascade nor MPI can furnish sufficient number of electrons. At higher laser intensities, electric field of the laser is able to pull an outer shell electron out of its orbit. After the initial electron ejection the LIB plasma is commonly maintained by the absorption of optical energy and the EII. Electrons in the laser field will gain energy through electron-neutral inverse bremsstrahlung collisions and will lose energy by elastic and inelastic collisions with the neutral species through excitation of rotational and vibrational degree of freedom of molecules and excitation of electronic states. While some electrons will be lost by attachment, new electrons will be produced by ionizing collisions. At high laser intensity, few electrons will be generated with energy larger than the ionization energy. The wavelength-resolved emission spectra from the laser plasma are not expected to vary due to the plasma origin. However plasma origin may be relevant, if the enhancement is observed between UV, visible and IR excitation wavelengths. Once that LIB plasma is formed, its growth is governed by the continuity rate equation for the electron density [31]

dne 2 = ν i ne + Wn IWn N − ν a ne − ν R ne + De∇ 2 ne , dt

(2.12)

where νi is the impact ionization rate, Wn is the multiphoton ionization rate coefficient, Iw is the intensity of the laser beam, n is the number of photons required for MPI, N is the number of atoms/molecules per unit volume, νa is the attachment rate, νR is the recombination rate

70

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

and De is the electron diffusion coefficient. The term dne/dt is the net rate of change in electron concentration at a point in the focal volume at a time t after the release of initiatory electrons. On the right side of the equation (2.12), the first term is the electron generation due to impact ionization. The second term on the right is MPI rate. The third, fourth and fifth terms are sink terms which represent electron attachment, recombination and diffusion, respectively. Impact ionization is defined by multiplying the number of electrons per unit volume to the impact ionization rate νi. The impact ionization rate refers to the rate at which electrons are generated as a result of ionizing collisions. At high laser intensity, a few new electrons can be generated and gain energy larger than their ionization energy which leads to the generation of new electrons by impact ionization, thereby leading to the cascade growth.

2.3.2. Electron attachment, recombination and diffusion Electron attachment is the rate of electron attachment νa multiplied by the number of electrons per unit volume. The LIB plasma typically loose electrons to the neutral species via the attachment mechanism in the form of three-body attachment or two-body dissociative attachment. Three-body attachment is: e + AB + X → AB- + X, where X appears to be a facilitator that allows the electrons to be gained by AB even through X remains unchanged throughout the process. Two-body dissociative attachment is: e + AB → A- + B. In this mechanism the electrons must exhibit a threshold electron energy that is equal to the difference between the dissociative energy of AB and the attachment energy of A, which results in the separation of A and B. Electron recombination is the rate of electron recombination νR multiplied to the number of electrons per unit volume. When the electron density is high, such as during the last stage of cascade breakdown, the LIB plasma can lose electrons to ions through electron-ion recombination. Similar to the electron attachment, three-body recombination and two-body recombination occurs as: e + AB+ + X → AB + X, e + AB+ → A + B. The electron-ion recombination rate has been studied theoretically for a three-body recombination by Gurevich and Pitaevskii [32]

ν R = 8.8 × 10

− 27

ne2 [ s −1 ] , 3.5 Te

(2.13)

where ne is the electron density in cm-3 and Te is the electron temperature in eV. The electron diffusion term is expressed as D e ∇ 2 n e (Eq. (2.12)). This loss mechanism, more important for a small diameter laser beam, is the diffusion of electrons out of the focal volume. Morgan [33] referred to the combined effect of diffusion and cascade ionization as the responsible for top-hat intensity profile. By imposing an electron skin at the edge of the intensity profile, they found that the electron density grows exponentially as

ve =

2.408 De , a2

(2.14)

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

71

where v e is the average electron velocity, De is electron diffusion coefficient and a is the radius of the beam. The equation (2.14) is intended to be an upper boundary for diffusion losses only because laser beams typically have a radial distribution closer to the gaussian rather than top-hat distribution. In summary, two mechanisms MPI and EII can initiate a conventional LIB plasma formation. After the LIB plasma formation the temporal growth is governed by the equation (2.12). The recombination of these two source terms (MPI and EII) and three sink terms (electron attachment, electron recombination and electron diffusion) controls the development of the conventional LIB plasma. These mechanisms that directly affect the temporal development of the LIB plasma, determine the necessary spectroscopic techniques required to spectrally resolve elemental species inside the LIB plasma.

2.4. Elements of LIBS In contrast to conventional spectroscopy, where one is mainly concerned with the structure of an isolated atom and molecule, the radiation from the plasma also depends on the properties of the plasma in the intermediate environment of the atomic or molecular radiator. This dependence is a consequence of the long-range Coulomb potential effects which dominate the interactions of ions and electrons with each other and with existing neutral particles. These interactions are reflected in the characteristic radiations in several ways. They can control population densities of the discrete atomic states, spectral shift and broadening by Stark effect, decrease of ionization potentials of the atomic species, cause continuum radiation emissions and emission of normally forbidden lines. Generally, the radiation emitted from a self-luminous plasma can be divided into bound-bound, bound-free, and free-free transitions.

2.4.1. Line radiation Line radiation from plasma occurs for electron transitions between the discrete or bound energy levels in atoms, molecules or ions. In an optically thin plasma of length l along the line of sight [34], the integrated emission intensity Iji of a spectral line arising from a transition between bound levels j and i is given by

I ji =

A ji hν ji N j ds = hν ji A ji N j l , 4π ∫

(2.15)

where Nj is the population density of the upper level j, hνji is the photon energy (energy difference between levels j and i), and Aji is the spontaneous transition probability or Einstein A coefficient. The integration is taken over a depth of plasma viewed by the detector, and the intensity of radiation is measured in units of power per unit area per unit solid angle. Transition probabilities can be sometimes expressed via the oscillator strength fji. This is defined as the ratio of the number of classical oscillators to the number of lower state atoms required to give the same line-integrated absorption [35]. Its relationship to the Einstein coefficient is

72

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

f ji =

4πε 0 mc 3 g j A ji . e 2 8π 2ν ji gi

(2.16)

The usefulness of fji is that it is dimensionless, describing just the relative strength of the transition. The detailed values of Aji, gi, and gj can be obtained from reference compilations or from electronic databases, i.e by NIST [36].

2.4.2. Continuum radiation The origins of continuum radiation are both bound-free and free-free transitions. The absorption of radiation from a discrete atomic state, such that the photon has enough energy to extend above the next ionization threshold, results in a release of an electron and gives rise to the process of photoionization. The reverse process of recombination occurs when an ion and an electron recombine with emission of a photon to form an ion in the next lowest ionic state (or in the neutral atomic state). Since the upper state is continuous, the emitted or absorbed radiation in both processes is also continuous. Transitions between two free energy levels can occur in plasmas increasing the energy exchanges of charged particles. Classically, this takes place because a moving charge radiates when it is accelerated or retarded. For most cases of practical importance, these free-free transitions are classified as bremsstrahlung or cyclotron spectra. In bremsstrahlung, the acceleration of charged particle takes place via the Coulomb field of charged particles. In cyclotron radiation, the acceleration is due to the gyration of charged particles in a magnetic field. The total continuum radiation at any particular frequency I(ν) is the sum of the contributions from all such processes having components at the specified frequency. Thus

I (ν ) d ν =

1 4π

∫n ∑ N e

i

i

⎤ ⎡ ⎢γ (i , Te ,ν ) + ∑ α (i , p , Te ,ν ) ⎥h ν ds d ν , p ⎦ ⎣

(2.17)

where γ(i, Te, ν) is the atomic probability of a photon of frequency ν being produced in the field of an atom or ion (specified by i) by an electron of mean kinetic temperature Te making free-free transition; α(i, p, Te, ν) is the corresponding probability where the electron makes a free-bond transition into a level p. As before, the integration is taken over the plasma depth s.

2.4.3. Line Broadening; Determination of electron number density from Stark broadening of spectral lines The shape of the spectral lines in the LIB has been studied since the first observation of the laser-induced breakdown in early 1960s. It plays an important role for the spectrochemical analysis and quantification of the plasma parameters. The observed spectral lines are always broadened, partly due to the finite resolution of the spectrometers and partly to intrinsic physical causes. In addition, the center of the spectral lines may be shifted from its nominal central wavelength. The principal physical causes of spectral line broadening are the Doppler, resonance pressure, and Stark broadening. There are several reasons for this broadening and shift. These reasons may be divided into two broad categories: broadening due to local

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

73

conditions and broadening due to extended conditions. Broadening due to local conditions is due to effects which hold in a small region around the emitting element, usually small enough to assure local thermodynamic equilibrium. Broadening due to extended conditions may result from changes to the spectral distribution of the radiation as it traverses its path to the observer. It also may result from the combining of radiation from a number of regions which are far from each other. Natural broadening: The uncertainty principle relates the lifetime of an excited state (due to the spontaneous radiative decay) with the uncertainty of its energy. This broadening effect results in an unshifted Lorentzian profile. The full width at half maximum (FWHM) of natural broadening for a transition with a natural lifetime of τji is: ΔλNFWHM=λ2/πcτji. The natural lifetime τji is dependent on the probability of spontaneous decay: τji=1/Aji. Natural broadening is usually very small compared with other causes of broadening. Doppler broadening: The Doppler broadening is due to the thermal motion of the emitting atoms, molecules or ions. The atoms in a gas which are emitting radiation will have a distribution of velocities. Each photon emitted will be "red" or "blue" shifted by the Doppler effect depending on the velocity of the atom relative to the observer. The higher the temperature of the gas, the wider the distribution of velocities in the gas. Since the spectral line is a combination of all of the emitted radiation, the higher the temperature of the gas, the broader will be the spectral line emitted from that gas. This broadening effect is described by a Gaussian profile and there is no associated shift. For a Maxwellian velocity distribution the line shape is Gaussian, and the FWHM may be estimated as (in Å):

ΔλDFWHM = 7.16 × 10 −7 ⋅ λ ⋅ T / M ,

(2.18)

being λ the wavelength in Å, T the temperature of the emitters in K, and M the atomic mass in amu. Pressure broadening: The presence of nearby particles will affect the radiation emitted by an individual particle. There are two limiting cases by which this occurs: (i) Impact pressure broadening: The collision of other particles with the emitting particle interrupts the emission process. The duration of the collision is much shorter than the lifetime of the emission process. This effect depends on both the density and the temperature of the gas. The broadening effect is described by a Lorentzian profile and there may be an associated shift. (ii) Quasistatic pressure broadening: The presence of other particles shifts the energy levels in the emitting particle, thereby altering the frequency of the emitted radiation. The duration of the influence is much longer than the lifetime of the emission process. This effect depends on the density of the gas, but is rather insensitive to temperature. The form of the line profile is determined by the functional form of the perturbing force with respect to distance from the perturbing particle. There may also be a shift in the line center. Pressure broadening may also be classified by the nature of the perturbing force as follows: (i) Linear Stark broadening occurs via the linear Stark effect which results from the interaction of an emitter with an electric field, which causes a shift in energy which is linear in the field strength (∼E and ∼1/r2); (ii) Resonance broadening occurs when the perturbing particle is of the same type as

74

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

the emitting particle, which introduces the possibility of an energy exchange process (∼E and ∼1/r3); (iii) Quadratic Stark broadening occurs via the quadratic Stark effect which results from the interaction of an emitter with an electric field, which causes a shift in energy which is quadratic in the field strength (∼E and ∼1/r4); (iv) Van der Waals broadening occurs when the emitting particle is being perturbed by Van der Waals forces. For the quasistatic case, a Van der Waals profile is often useful in describing the profile. The energy shift as a function of distance is given in the wings by e.g. the Lennard-Jones potential (∼E and ∼1/r6). Stark broadening of spectral lines in the plasma occurs when an emitting species at a distance r from an ion or electron is perturbed by the electric field. This interaction is described by the Stark effect. The linear Stark effect exists for hydrogen and for all other atoms. Stark broadening from collisions of charged species is the primary mechanism influencing the emission spectra in LIBS. Stark broadening of well-isolated lines in the plasma can be used to determine the electron number density ne(cm-3). An estimation of the Stark width (FWHM) and line shift of the Stark broadened lines is given as [26-27,34-35,3739]: 1/ 4

Δλ

Stark FWHM

⎛ n ⎞ ⎛ n ⎞ = 2W ⎜ e16 ⎟ + 3.5 A⎜ e16 ⎟ ⎝ 10 ⎠ ⎝ 10 ⎠

⎛ n Δλ Shift = D ⎜ e16 ⎝ 10

⎞ ⎛ n ⎞ ⎟ ± 2 A⎜ e16 ⎟ ⎠ ⎝ 10 ⎠

1/ 4

(1 − BN ) W ⎛⎜ 10n −1 / 3 D

e 16

⎝

(1 − BN ) W ⎛⎜ 10n −1 / 3 D

⎝

e 16

⎞ ⎟, ⎠

⎞ ⎟, ⎠

(2.19)

(2.20)

where W is the electron impact parameter or half-width, A is the ion impact parameter both in Å, B is a coefficient equal to 1.2 or 0.75 for ionic or neutral lines, respectively, D (in Å) is the electron shift parameter and ND is the number of particles in the Debye sphere 1/ 2

N D = 1.72 × 109 T 3 / 2 / ne . The electron and the ion impact parameters are functions of temperature. The first term on the right side of Eq. (2.19) refers to the broadening due to the electron contribution, whereas the second one is the ion broadening. Since for LIB conditions Stark broadening is predominantly by electron impact, the ion correction factor can safely be neglected, and Eq. (2.19) becomes

⎛ ne ⎞ Δλ Stark ⎟. FWHM ≈ 2W ⎜ 16 ⎝ 10 ⎠

(2.21)

The coefficients W are independent of ne and slowly varying functions of electron temperature. The minus sign in Eq. (2.20) applies to the high-temperature range of those few lines that have a negative value of D/W at low temperatures. A comprehensive list of width and shift parameters W, A and D is given by Griem [27, 39].

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

75

2.4.4. Determination of excitation, vibrational and rotational temperatures The excitation temperature Texc can be calculated according to the Boltzmann equation under the assumption of LTE. The significance of this temperature depends on the degree of equilibrium within the plasma. For plasma in LTE, any point can be described by its local values of temperature, density, and chemical composition. By considering two lines λji and λnm of the same species, characterized by different values of the upper energy level (Ej≠En), the relative intensity ratio can be used to calculate the plasma excitation temperature

Texc =

E j − En ⎡I ⋅λ ⋅ g ⋅ A ⎤ k B ln ⎢ nm nm j ji ⎥ ⎣⎢ I ji ⋅ λ ji ⋅ g n ⋅ Anm ⎦⎥

.

(2.22)

When selecting a line pair, it is advisable to choose two lines as close as possible in wavelength and as far apart as possible in excitation energy. This is to limit the effect of varying the spectral response of the detection system. The use of several lines instead of just one pair leads to greater precision of the plasma excitation temperature estimation. In fact, though the precision of the intensity values can be improved by increasing the signal intensity, the transition probabilities Aji reported in the literature exhibit significance degree of uncertainty (5-50%). The excitation temperature can be calculated from the relative intensities of a series of lines from different excitation states of the same atomic or ionic species from the slope of the Bolztmann plot ln[Iji·λji/gj·Aji] versus Ej/kB

⎡ I ⋅λ ⎤ Ej ln ⎢ ji ji ⎥ = C − k B ⋅ Texc ⎢⎣ g j ⋅ A ji ⎥⎦

(2.23)

,

where Iji is the emissivity (W m-3 sr-1) of the emitted j→i spectral line, λji is the wavelength, gj=2Jj+1 is the statistical weight, Aji is the Einstein transition probability of spontaneous emission, Ej/kB is the normalized energy of the upper electronic level and C=ln(hcNj/4πQ(T)) (Q(T) is the partition function). The values of the λji, gj, Aji and Ei for selected atomic or ionic lines can be obtained from the NIST Atomic Spectral Database. A set of selected spectral lines can be chosen based on their relative strengths, accuracies and transition probabilities. The emission spectra of the diatomic species reveal a relatively complex structure which is due to the combination of the electronic transitions from the different rotational and vibrational states [40-42]. The emission intensities of the molecular bands can be analyzed in order to calculate the molecular vibrational temperature Tvib. For a plasma in LTE, the intensity of an individual vibrational v’-v” band Iv’-v” is given by

G ( v' ) h c ⎛I ⋅ λ4 ⎞ ln⎜⎜ v ' − v" v ' − v" ⎟⎟ = A − k B ⋅ Tvib ⎠ ⎝ q v ' − v"

,

(2.24)

76

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

where A is a constant, λv’-v” is the wavelength corresponding to the band head, ∞

q

v'− v"

=

2

∫ Ψv' ( R)Ψv" ( R)dR is the Franck-Condon factor and G ( v' )h c / k B is the normalized 0

energy of the upper vibrational level. A line fit to

(

)

ln I v '− v" ⋅ λ4v '− v" / q v '− v" as a function of

the upper normalized electronic-vibrational energies has a slope equal to -1/Tvib. On the other hand, the emission intensities of the rotational lines of a vibrational band can be analyzed in order to estimate the effective rotational temperature Trot. In this case it is necessary to consider the Hund´s coupling case for the both electronic states implied in the transition. From the assignment of the rotational spectrum it is possible to estimate the effective rotational temperature by considering the J value for the maximum of the band Trot=(2 Bv h c/kB)(Jmax+1/2)2, being Bv the rotational constant for upper v vibrational level and Jmax the total angular momentum at the maximum. Another method for estimating the vibrational and rotational temperatures is based on a simulation program of the spectra. Software developed in our laboratory [43] calculated the spectra of a diatomic molecule by summing the intensity of all rovibrational levels and convoluting the results with the instrumental line shape of the optical system. The emission intensity Iv’,J’-v”,J” of a molecular line can be approximated by

I v',J' − v",J" ≈

64π 4ν~ 4 v ', J ' − v ", J " 2 N v ', J ' Re qv ', v " S J ', J " , 3( 2 J '+1)

(2.25)

where ν~v ', J ' − v ", J " is the wavenumber of the transition, 2J’+1 is the rotational degeneracy of the upper state, Nv’,J’ is the population in the initial (upper) state, R e is the average electronic transition moment, qv’,v” is the Franck-Condon factor and SJ’,J” is the Hönl-London factor [44]. Spectrum simulations are based on comparison of experimental and calculated spectra for different rotational and vibrational population distributions which depend on temperature.

2.4.5. Ionization degree of the plasmas: Saha equation In plasma there is a continuous transition from gases with neutral atoms to a state with ionized atoms, which is determined by an ionization equation. The transition between a gas and a plasma is essentially a chemical equilibrium, which shifts from the gas to the plasma side with increasing temperature. Let us consider the first three different ionization equilibria of an atom A: A ↔ A+ + e + IP(A-I), A+ ↔ A2+ + e + IP(A-II), A2+ ↔ A3+ + e + IP(A-III).

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

77

For each ionization equilibrium, considering the atoms and ions in their ground electronic state, the LTE between ionization and recombination reactions at temperature T is described by the Saha equation (2.3)

(

ne ⋅ N i g e ⋅ g i 2π m k B T = N0 g0 h3

)

3/ 2

e − Ei / k B T ,

(2.26)

where ne = Ni are the electron and ion densities in the different ionization equilibria in the second member of ionization equilibria. From this equation, ionization degree ne·Ni/N0 can be estimated.

2.5. Effects of Physical Variables in LIBS The variables that can influence the LIBS measurements are mainly the laser properties i.e. wavelength, energy, pulse duration, focusing spot size, shot-to-shot energy fluctuations, ambient conditions, physical properties of the sample and the detection window (delay time and gate width). How these parameters affect the precision and accuracy of LIBS are addressed below.

2.5.1. Laser parameters In LIBS a high-power laser is used to ablate or to breakdown a gaseous sample in the form of plasma. The primary energy related parameters influencing the laser-gas interaction are the laser peak power PW (or radiant pulse energy per time, in W) and the laser peak intensity IW (power density or irradiance; energy per unit area and time, W cm-2) given by

PW = EW / τ FWHM ,

(2.27)

I W = PW / πr 2 ,

(2.28)

where EW (in J) is the pulse energy, τFWHM (in s) is the pulse duration at the FWHM and πr2 is the focal spot area (cm2). The fluence ΦW (in J cm-2) on the focused spot area, the photon flux density Fph (photon cm-2 s-1) and electric field FE (V cm-1) are given by

ΦW = EW / πr 2 ,

(2.29)

Fph = IW λ / hc ,

(2.30)

FE = IW / cε 0 ,

(2.31)

78

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

where λ is the laser wavelength, h is the Planck constant, c is the speed of light, and ε0 is the electric constant. The laser peak intensity IW, fluence, photon flux and electric field are inversely proportional to the focused spot area. For LIBS, the peak intensity IW (and the other properties ΦW, Fph, FE and PR) that can be delivered to the sample is more important than the absolute value of the laser power. For the formation of plasma, the laser fluence needs to exceed the threshold value, typically of the order of several J cm-2 for a nanosecond laser pulse [45]. If the laser energy is very close to the breakdown threshold, the pulse-to-pulse fluctuations can cause the plasma condition to be irreproducible, which results in poor measurement precision. The intensities of the emission lines are proportional to the laser energy while the laser plasma is in the optical thin region. When the laser energy increases further, it produces very dense and hot plasma that can absorb laser energy. This will lead to an increase in the continuum emission and a decrease in the signal intensity. Besides, the laser pulse duration and the shot-to-shot fluctuations can also affect the signal reproducibility and hence LIBS precision.

2.5.2. Focal properties The laser power density at the focal volume is inversely proportional to the focused spot size. For a laser beam with a Gaussian profile, the focused beam waist w0 is given by [46]

w0 =

λ f π ws

,

(2.32)

where f is the focal length of the lens and ws is the radius of the unfocused beam. The higher laser power density at the focal point can be achieved by reducing the focused beam waist using a shorter focal length lens. On the other hand, the angular spread in laser light generated by the diffraction of plane waves passing through a circular aperture consists of a central, bright circular spot (the Airy disk) surrounded by a series of bright rings. The beam divergence angle θ, measured to edges of Airy disk, is given by θ=2.44 λ/d, where λ is the laser wavelength and d is the diameter of the circular aperture. It can be shown that a laser beam, with beam divergence θi , incident on a lens of focal length f, whose diameter is several times larger than the width of the incident beam, is focused to a diffractionlimited spot of diameter approximately equal to f θi . If the focal region of the laser beam is assumed to be cylindrical in shape, the spot size in terms of length l, can be approximated as

l = ( 2 − 1)θ i f 2 / d .

(2.33)

2.5.3. Laser absorption in the plasma In LIBS the evaporation of the material begins just after the impact of the leading edge of the laser pulse on the surface. The time required for the removal of the material is less than a nanosecond, thus once the plasma is formed, a part of the laser beam will be absorbed in the

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

79

plasma by the electron-neutral or electron-ion inverse bremstrahlung (e-n IB and e-i IB), or by photoionization (PI) of the excited states. Consequently not the full laser irradiance will be able to reach the target, which is called plasma shielding. In the case of IB absorption the free electrons gain kinetic energy from the laser beam thus promoting plume ionization and excitation through collisions with excited and ground state neutrals. The IB process is usually described by the inverse absorption length αIB (cm-1) [31, 47]

α IB , e − n = σ e − n [1 − e− hν / kT ]ne N 0 ,

α IB, e −i = neσ e −i = [1 − e− hν / kT ]

4e6λ3 2π ne Ni , 3hc 4 m 3mkT

(2.34)

(2.35)

where N0 is the neutral atomic density, Ni is the ionic density, σe-n is the electron-neutral cross section of photon absorption and σe-i is the electron-ion cross section of photon absorption. The term [1 − e− hν / kT ] represents the losses due to stimulated emission. The electron number density in the plasma depends on the degree of ionization, evaporation rate and the plasma expansion velocity. Moreover, the absorption coefficient shows different temperature dependence for different energy density regimes. In the case of short wavelength lasers the photoionization of the excited atoms can play significant role. In fact, the absorption coefficient of this process σPI (cm-1) is given by [48]

α PI = N nσ PI ≈ ∑ 2.9 × 10 n

−17

( I P )5 / 2 Nn , ( hν ) 3

(2.36)

being Nn is the number density (in cm-3) of the excited state n, Ip is the ionization potential in eV and hν is the photon energy in eV. In this equation the summation is performed over the energy levels which satisfy the condition hν>En. Equation (2.36), although derived for hydrogen like atoms, can be applied to complex atomic systems.

3. EXPERIMENTAL LIBS is a plasma based method that uses instrumentation similar to that used by other spectroscopic methods (atomic emission spectroscopy, laser-induced fluorescence etc). A typical LIBS apparatus utilizes a pulsed laser that generates the powerful optical pulses used to form the plasma. Principles of laser operation in general and the operation of specific lasers are described in detail in numerous books. The discussion here will be limited to the fundamentals of the operation of the transversely excited atmospheric (TEA) carbon dioxide laser used in this work. On the other hand, it is necessary a convenient detection system. We described here the detection system used in this work.

80

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

3.1. Pulsed TEA CO2 Laser The CO2 laser is a near-infrared gas laser capable of very high power and with the highest efficiency of all gas lasers (≈10-20%) and for cw operation the highest output power. Although CO2 lasers have found many applications including surgical procedure, their popular image is as powerful devices for cutting, drilling, welding or as weapons for military applications. The linear CO2 molecule has three normal modes of vibration, labelled ν1 (symmetry stretch), ν2 (bending vibration) and ν3 (asymmetric stretch), and plotted in the upper part of Figure 1. The fundamental vibration wavenumbers are 1354, 673 and 2396 cm-1, respectively. The vibrational state of the molecule is described by the number of vibrational quanta in these modes. The bending vibrational mode is twofold degenerate and can have a vibrational angular momentum along the CO2 axis. The number of quanta of this vibrational angular momentum is stated as an upper index to the vibrational ν2 quanta. The upper laser level (0001) denotes the ground vibrational state for the mode ν1, the ground vibrational state for the mode ν2 which is doubly degenerate, and the first excited vibrational state for the mode ν3. The active medium is a gas discharge in a mixture of He, N2 and CO2. By electron impact in the discharge excited vibrational levels in the electronic ground states of N2 and CO2 are populated (Figure 1). The vibrational levels v = 1 in the N2 molecule and (ν1, ν2, ν3) = (0001) in the CO2 molecule are nearresonant and energy transfer from the N2 molecule to the CO2 molecule becomes very efficient. This populates the (0001) level in CO2 preferentially, creates inversion between the (0001) and the (0200) levels, and allows laser oscillations on many rotational transitions between these two vibrational states in the wavelength range 9.6-10.6 µm. The main laser transitions in CO2 occur between the excited states of the mode ν3(0001) and the symmetric stretching mode ν1(1000) (10.6 μm) or the bending mode ν2(0200) (9.6 μm). A single line can be selected by a Littrowgrating, forming one of the resonator end mirrors.

Figure 1. Level scheme and the three normal vibrational modes of the CO2 molecule.

Helium atoms do not take part directly in the excitation of CO2 molecules but do play an important role in heat-transfer from the gas mixture to the tube walls, as well as facilitating the depopulation of the lower vibrational levels in CO2; contributing in this way to the ,maintenance of the population inversion. The power of CO2 lasers depends on their configuration. The laser

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

81

used in these experiments was a transversely excited atmospheric (TEA) CO2 laser in which the electric discharge is transverse to the laser cavity’s axis. The pressure in the tube is close to atmospheric pressure. The CO2:N2:He mixture is exchange in a continuous way, enhancing the output power of the laser. The laser can achieve a power of 50 MW. The optical materials used in lasers emitting radiation in the infrared range are obviously different than those used in the visible range. For example, materials such as germanium (Ge) or gallium arsenide (GaAs) are completely opaque in the visible range, while being transparent in the infrared range. Some materials, such as zinc selenide (ZnSe), are transparent in both spectral ranges. Typical materials transparent in the IR range are: NaCl or CsI. Metal mirrors (copper, molybdenum, gold) are used in the IR range, owing to their small absorption (and large reflectivity) as well as their large heat capacity which enables removal of heat from the active medium.

3.2. Spectrographs and Detectors A detection system consists in a wavelength dispersing element and an electronic device as detector. Wavelength dispersing elements used for LIBS generally have to fulfilled two opposite requests: (i) need for high resolving power λ/∆λ because of highly pronounced spectral interferences in atomic-molecular emission spectroscopy; (ii) For an analysis with a single laser pulse the wide spectral region should be covered. Today, usually, four different spectrograph mountings are mainly in use for OES: Czerny-Turner, Echelle, Rowland and Paschen-Runge. Here, only the first system used by us will be described. A Czerny-Turner mount uses a plane grating. The incident radiation passes through the entrance slit and strikes a parabolic collimating mirror. This mirror produces a collimated light beam reflected onto the grating, which spatially disperses the spectral components of the incident radiation. Collimated rays of diffracted radiation strikes a second parabolic focusing mirror. The dispersed radiation is focused in the focal plane producing the entrance slit images in that plane. Because the parallel rays of a given wavelength are incident on the focusing element at a specific angle, each wavelength is focused to a slit image at a different center position on the focal plane. In this focal plane the detector is placed (see Figure 2). For the detection of plasma emissions in LIBS experiments, photomultipliers (PMT), charge-coupled and intensified charge coupled devices (CCD and ICCD) and optical multichannel analyzers (OMA) can be utilized depending on the type of spectrometer. A PMT consists of the photo-emissive cathode, several dynodes, and an anode. When the radiation hits the chatted electrons are generated. A high voltage across the dynodes leads to further multiplication of these photoelectrons. A PMT is relatively inexpensive and very sensitive detector, which allows recording of time-integrated plasma emission signals. It requires, however, a scanning mode monochromator to obtain full spectral information. Therefore, the PMT is not convenient for multielement analysis with a single laser pulse or when fast scan results are required. The charge coupled devices (CCDs) are made up of a two-dimensional array of semiconductor capacitors (pixels) that have been formed on a single silicon chip. Commercially available CCD chips usually consisted of 512×512 or 1024×1024 pixels. Performance characteristics of these instruments with respect to sensitivity, dynamic range, and signal-to-noise ratio appear to approach those of PMT’s in addition of having the multichannel advantage. A CCD system can be equipped with an intensifier (microchannel plate) which allows high photonics gain and possibility of electronic gating which is of essential importance for time resolved LIBS measurements.

82

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

Figure 2. Schematic diagram of the experimental set-up for pulsed laser ablation diagnostics.

3.3. Schematic Diagram for LIBS The experimental configuration used to study carbon by LIBS is shown in Figure 2. The laser-induced plasma was generated using a TEA CO2 laser (Lumonics model K-103) operating on an 8:8:84 mixture of CO2:N2:He, respectively. The laser is equipped with frontal Ge multimode optics (35 % reflectivity) and a rear diffraction grating with 135 lines mm-1 blazed at 10.6 μm. The CO2 laser irradiation of the target was carried out using the 9P(28) line at λ=9.621 µm (ΔE=0.1289 eV) and the 10P(20) line at λ=10.591 µm (ΔE=0.1171 eV). The temporal shape of the TEA-CO2 laser pulse, monitored with a photon drag detector (Rofin Sinar 7415), consisted in both cases in a prominent spike of a full width at half maximum (FWHM) of around 64 ns carrying ∼90% of the laser energy, followed by a long lasting tail of lower energy and about 3 μs duration. The laser pulse repetition rate was usually 1 Hz. The divergence of the emitted laser beam is 3 mrad. The pulsed CO2 laser beam was focused with a NaCl lens of 24 cm focal length onto the target. The primary laser beam was angulary defined and attenuated by a diaphragm of 17.5 mm diameter before entering to the cell. The CO2 laser average energy was measured in front of the lens with a Lumonics 20D pyroelectric detectors through a Tektronix TDS 540 digital oscilloscope. Energy losses were estimated by making pulse energy measurements with and without the NaCl window in place. The focused radius of the laser beam (0.05 cm) was measured at the target position with a pyroelectric array Delta Development Mark IV. The laser intensity (power density or irradiance) on the focal volume was varied using several calibrated CaF2 attenuators and range from 0.29 to 6.31 GW×cm-2. The high purity graphite target (~99.99 %) was placed in a low-vacuum cell equipped with a NaCl window for the laser beam and two quartz windows for optical access. The graphite target is initially at ambient temperature (298 K) and it is not water-cooled. The cell was evacuated with the aid of a rotary pump, to a base pressure of 4 Pa that was measured by a mechanical gauge. Optical emission from the plume was imaged by a collecting optical system onto the entrance slit of different spectrometers. The light emitted

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

83

from the laser-induced plasma was optically imaged 1:1, at right angles to the normal of the target surface, by a quartz lens (focal length 4 cm, f-number = f/2.3) onto the entrance slit of the spectrometer. The distance between plasma axis and entrance slit was 16 cm. The lens causes a bit chromatic aberration for OES measurements, although the geometric efficiency is barely affected. Two spectrometers were used: 1/8 m Oriel spectrometer (10 and 25 μm slits) with two different gratings (1200 and 2400 grooves mm-1) in the spectral region 2000-11000 Å at a resolution of ~1.3 Å in first-order (1200 grooves mm-1 grating), and an ISA Jobin Yvon Spex (Model HR320) 0.32 m equipped with a plane holographic grating (2400 grooves mm-1) in the spectral region 2000-7500 Å at a resolution of ~0.12 Å in first-order. Two detectors were attached to the exit focal plane of the spectrographs and used to detect the optical emissions from the laser-induced plasma: an Andor DU420-OE (open electrode) CCD camera (1024 256 matrix of 26 26 μm2 individual pixels) with thermoelectric cooling working at –30 ºC; A 1024 1024 matrix of 13 13 μm2 individual pixels ICCD (Andor iStar DH-734), with thermoelectric cooling working at –20 ºC. The low noise level of the CCD allows long integration times and therefore the detection of very low emission intensities. The intensity response of the detection system was calibrated with a standard (Osram No.6438, 6.6A, 200-W) halogen lamp and Hg/Ar pencil lamp. Several (Cu/Ne, Fe/Ne and Cr/Ar) hollow cathode lamps (HCL) were used for the spectral wavelength calibration of the spectrometers.

3.4. Timing Considerations For synchronization, the CO2 laser was operated at the internal trigger mode and the ICCD detector in external and gate modes. The external trigger signal generated by the laser is fed directly into the back of the ICCD detector head. The total insertion delay (or propagation delay) is the total length of time taken for the external trigger pulse to travel through the digital delay generator and gater so that the ICCD will switch on. This insertion delay time is 45 ± 2 ns. The time jitter between the laser and the fast ICCD detector gate was about ± 2 ns. The delay time td is the time interval between the arrival of the laser pulse on the target and the activation of the ICCD detector. The gate width delay time tw is the time interval during which the plasma emission is monitored by the ICCD. Both parameters were adjusted by the digital delay generator of the ICCD detector. The resolution of the gate pulse delay time and the gate pulse width time are 25 ps. The CO2 laser pulse picked up with the photon drag detector triggers a Stanford DG 535 pulse generator which is used as external trigger in the ICCD camera. The laser pulse and the gate monitor output were displayed in a Tektronix TDS 540 digital oscilloscope, allowing to control td eliminating the insertion time of the camera.

4. RESULTS AND DISCUSSION When a high-power laser pulse is focused on a solid surface the target becomes ablated. If the laser irradiance in the focal volume surpasses the breakdown threshold of the system formed by the vaporized atoms and residual gas, a breakdown, characterized by a brilliant flash of light accompanied by a distinctive cracking noise, is produced. At the top of figure 2 we show an image of laser-induced breakdown (LIB) plasma in graphite induced by a single

84

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

CO2 laser pulse. The plume size is around 14 cm. The laser was focused on a point at the centre of the target. The observations of the LIB geometry during the experiments indicate that the actual plasma region is not entirely spherical, but lightly elongated in the direction of the laser beam propagation. The CO2 laser pulse remains in the focal volume after the plasma formation for some significant fraction of its duration and the plasma formed can be heated to very high temperatures and pressures by inverse bremsstrahlung absorption. Since plasmas absorb radiation much more strongly than ordinary mater, plasmas can block transmission of incoming laser light to a significant degree; a phenomenon known as plasma shielding [49]. The high temperatures and pressures produced by plasma absorption can lead to thermal expansion of the plasma at high velocities, producing an audible acoustic signature, shock waves, and cavitation effects. The plasma also tends to expand back along the beam path toward the laser, a phenomenon known as moving breakdown. The shock wave heats up the surrounding gas which is instantaneously transformed in strongly ionized plasma. For the present experiments the measured focused-spot area was 7.85×10-3 cm2. This value is higher than the calculated area (2.2×10-4 cm2) obtained from the beam waist (Eq. 2.32). This fact is due to the non-gaussian profile of the CO2 laser beam. Moreover the CO2 laser beam passes through a circular aperture of diameter 17.5 mm. For this diaphragm the calculated divergence angle for the laser beams at 9.621 and 10.591 µm are 1.3 and 1.5 mrad, respectively. Thus, considering the total beam divergence (∼4.4 mrad), the calculated diameter of the focused TEA-CO2 laser (beam waist) is 1.06 mm, which is very similar to the measured value (∼1 mm). If the focal region of the laser beam is assumed to be cylindrical in shape, the spot size in terms of length l (Eq. 2.33) of the focused TEA-CO2 laser is 6.0 mm, which is similar to the measured value (∼7 mm). Table 1. Laser parameters for some LIBS experiments. Energy EW (mJ) 3161 2685 2256 1732 1209 503 324 273 242 203 171 149 131 110

Power PW (MW) 49.5 42.1 35.4 27.1 19.0 7.88 5.08 4.27 3.79 3.19 2.68 2.33 2.08 1.73

Intensity IW (GW cm-2) 6.31 5.36 4.50 3.46 2.41 1.00 0.648 0.544 0.483 0.406 0.341 0.297 0.262 0.220

Fluence ΦW (J cm-2) 403 342 287 220 154 64.0 41.3 34.7 30.8 25.9 21.8 19.0 16.7 14.0

Photon Flux, Fph (photon cm-2 s-1) 3.07×1029 2.60×1029 2.18×1029 1.67×1029 1.17×1029 4.86×1028 3.14×1028 2.64×1028 2.34×1028 1.97×1028 1.65×1028 1.44×1028 1.26×1028 1.07×1028

Electric Field FE (MV cm-1) 1.54 1.42 1.30 1.14 0.985 0.615 0.494 0.453 0.427 0.391 0.358 0.335 0.314 0.288

3000

3200

C+

3400

3600

C

2+

3

0

5200

5400

5600

3

C2: d Πg-a Πu

Δv=-2 4-6 3-5 2-4 1-3 0-2

+

C 5000

5800

C 6000

Δv=-1

C3+

C2+

CH: A2Δ-X2Πr v'=0-v"=0

C C+

+

5-2

6-3

C

+

4-1

3-0

7-3

6-2

C+

60000 6300 6400 6500 6600 6700 6800 6900 7000 7100 7200 7300 7400 7500 7600 40000 20000

Hα

0

6400

6600

6800

7000

7200

7400

7600

Air Wavelength / Å

N

8000

C

N

Relative Intensity / a. u.

10000 8000

N N

5-3

6000

N N

4000

2

10000

(f) 9000

N 2000

7000

C

C

C N

C

C C

C N

6000 5000

2 +

CN: A Πi- X Σ Δv=+1

C

2-1

O

O

N NN

1-0

O

CN: AΠi-X Σ N Δv=+2 3-1

3-0

N

80000

6200

4-2

Relative Intensity / a. u.

N N

2 +

2

2-0

14000 12000

4800

1 +

C2: AΠu-X Σg

(e)

C2+ 8-4

100000

5-1

120000

Air Wavelength / Å 1

4600

C

C C2+ C3+

4-5 3-4 2-3 1-2 0-1

C

4400

2 +

2

C2+ + C

2-2

C Hβ

4200

+

3

+

C 20000

Hβ

CN: AΠi- X Σ Δv=+3

C C C+

3

C2: d Πg-a Πu Δv=-1

Relative Intensity / a. u.

1-1 +

40000

2 +

2

Δv=+4

C

C+ CN: B2Σ+-X2Σ+

4000

160000 140000 CN:AΠ- X Σ i

3

C4+

Relative Intensity / a. u.

(d)

C2: d Πg-a Πu Δv=0

60000

C+

C+ 3800

C+

0-0 v'-v" 3

Hγ

Air Wavelength / Å

100000

80000

Δv=+1

0

2800

Air Wavelength / Å

(c)

3

4-3 3-2 1-02-1 10-9 11-10 12-11

5000

3

C2: d Πg-a Πu

C2+

10000

2-2 1-1

15000

0-0

25000 20000

2 +

C2+ C2+

2 +

CN: B Σ -X Σ Δv=0

CH: B2Σ--X2Πr v'=0-v"=0

30000

CN: B2Σ+-X2Σ+ Δv=+1 1-0 C : C1Π -A1Π 2 g u

Relative Intensity / a. u.

10-8

NH: A3Πi-X3Σ0-0 1-1 v'-v"

C

2+

3400

35000

C2+ C+

2600

45000

C2+

2400

3200

C+

5-3 6-4 9-7 11-9

C+ C

2+

+

C+

3000

C

C

2200

2800

C C+ C+

4+

C2+ 2000

2600 C3+

C

50000 0

OH: A2Σ+-X2Πi Δv=0 5-2 6-3 2 + 2 + 7-4 CN: B Σ -X Σ Δv=+3 10-7

C+

C+

C3+ C

2400

85

50000

40000

3

CH: C2Σ+-X2Πr 0-0

C4+

2200

100000

(b)

C2: e Πg-a Πu 2 2 CN: D Πi-A Πi

CC2+2+ + C

2000

Δv=+2

3

2+

150000

Δv=+2, 1, 0, -1

C2+

200000

1

C2: E Σ g-A Πu

C2+ C2+

1 +

250000

C+

300000

C

Relative Intensity / a. u.

350000

C+

400000

C2: D1Σ+u-X1Σ+g Δv=0

(a)

C2+

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

4000 3000

C

2000 1000

0

7600

7800

8000

8200

8400

Air Wavelength / Å

8600

8800

0

8800 8900 9000 9100 9200 9300 9400 9500 9600

Air Wavelength / Å

Figure 3 (a)-(f). Low-resolution PLA of carbon emission spectrum observed in the 1920-9680 Å region at an air pressure of 4 Pa, excited by the 10P(20) line at 944.20 cm-1 of the CO2 laser, and assignment of the atomic lines of C, C+, C2+, C3+, C4+, N, O and molecular bands of C2(E1Σ+g–A1Πu; Freymark system), C2(D1Σu+– X1Σg+; Mulliken system), C2(e3Πg – a3Πu; Fox-Herzberg system), CN(D2Πi–A2Πi), OH(A2Σ+– X2Π), CH(C2Σ+–X2Π), CN(B2Σ+–X2Σ+; Violet system), NH(A3Π–X3Σ-), C2(C1Πg– A1Πu; Deslandresd’Azambuja system), C2(d3Πg–a3Πu; Swan band system), CH(B2Σ-–X2Π), CH(A2Δ–X2Π), C2(A1Πu– X1Σg+; Phillips system), and CN(A2Π–X2Σ+; red system).

86

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

4.1. Identification of the Chemical Species in the Pulse Laser Ablation Plasma Plume For the different pulse laser energies measured in this work, the calculated laser peak power (Eq. 2.27), intensity (Eq. 2.28), fluence (Eq. 2.29), photon flux (Eq. 2.30), and electric field (Eq. 2.31) are tabulated in Table 1. In the scanned spectral region, from UV to NIR, OES reproduce particular emission of carbon plasmas in a low-vacuum air atmosphere (Pair=4 Pa). Typical time-integrated and spectral-resolved low-resolution OES from LIB of graphite is shown in Figure 3(a)-(f). In the recording of the spectra of Figure 3(c-f) a cutoff filter was used in order to suppress high diffraction orders. In general, the spectra of the PLA plume are dominated by emission of strong electronic relaxation of excited atomic C, ionic fragments C+, C2+ and C3+, and molecular features of C2(d3Πg–a3Πu; triplet Swan band system). The medium-weak emission is mainly due to excited atomic N, H, O, ionic fragment C4+ and molecular features of C2(E1Σ+g–A1Πu; Freymark system), C2(D1Σu+– X1Σg+; Mulliken system), CN(D2Π–A2Π), C2(e3Πg–a3Πu; Fox-Herzberg system), C2(C1Πg–A1Πu; Deslandresd’Azambuja system), OH(A2Σ+–X2Π), CH(C2Σ+–X2Π), NH(A3Π–X3Σ-), CN(B2Σ+–X2Σ+; violet system), CH(B2Σ-–X2Π), CH(A2Δ–X2Π), C2(A1Πu–X1Σg+; Phillips system) and CN(A2Π– X2Σ+; Red system). In the spectrum of Figure 3(a) in the 1920-3480 Å region, very strong atomic C, C+, C2+ and C3+ lines dominate, but also weak C4+ and molecular bands of C2(E-A; Δv=v’-v”=+2, +1, 0, -1 sequence from 200 to 222 nm), C2(D-X; Δv=0 sequence near 231.4 nm), CN(D-A; in the spectral range 223 to 260 nm), C2(e-a; in the spectral range 240 to 290 nm), CN(B–X; Δv=3 sequence from 306 to 326 nm), OH(A–X; Δv=0 sequence from 306 to 318 nm), CH(C–X; Δv=0 sequence from 314 to 317 nm), NH(A–X; Δv=0 sequence near 336 nm) and CN(B–X; Δv=2 sequence from 326 to 348 nm) are observed. In this spectrum the predominant emitting species are the C2+ 2p2 1D2 → 2s2p 1P01 atomic line at 2296.87 Å, C 3+ 1 1 2 p ( 2 P0 )3s P10 → 2 p 2 S 0 atomic line at 2478.56 Å, two lines of C at 2524.41 and 2529.98 Å, several lines of C+ at 2836.71 and 2992.62 Å and the v’=0-v”=0 band of NH(A–X) at 3360 Å. In the spectrum of Figure 3(b), the predominant emitting species are C+ (doublet 2 2 2s 2 4s S1/ 2 → 2s 2 3 p P10/ 2,3 / 2 at 3918.98 and 3920.69 Å, respectively, and multiplet 2 2 2s 2 4 f FJ0' → 2s 2 3d DJ " around 4267 Å), and the molecular bands of CN(B–X; Δv=0

sequence). Many medium intensity atomic lines of C+, C2+ and C3+, weak hydrogen lines of the Balmer series (Hβ, Hγ etc), and several molecular bands of CN, C2, and CH are also present. In the spectrum of Figure 3(c), the predominant emitting species are C+ and C2 (molecular bands: d-a; Δv=0, -1, and -2 sequences from 480 to 630 nm). Many weak lines of C, C+, C2+ and C3+ are also present. In the spectrum of Figure 3(d), the most intense lines are the doublet structure of C+ 2 s 2 3 p 2 P30/ 2,1 / 2 → 2 s 2 3s 2 S1 / 2 at 6578.05 and 6582.88 Å, respectively, C 2 s 2 2 p ( 2 P0 ) 4d 1 P10 → 2 s 2 2 p ( 2 P0 )3 p 1 P1 atomic line at 6587.61 Å, C+ +

2 2 2s 2 3d D3 / 2 → 2s 2 3 p P10/ 2 at 7231.32 Å and C

2 2 2s 2 3d D5 / 2 → 2s 2 3 p P30/ 2 at 7236.42 Å.

Also many weak lines of C, C+, C2+, Hα, N, and several bands v’-v” (5-1, 6-2, 7-3, 8-4, 3-0, 41, 5-2 and 6-3) corresponding to CN(A–X) are also present. The spectrum of Figure 3(e), shows the emission of many atomic lines of C, O, and N, the 3-0 band of C2(A-X) and several bands (2-0, 3-1, 4-2, and 5-3) of CN(A–X). Finally, in the spectrum of Figure 3(f), the

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

87

emission of many atomic lines of C and N and mainly the 1-0 and 2-1 bands of CN(A–X) can be appreciated.

a) Pulsed laser ablation of Carbon

b) Acetylene/oxygen flame 3

1

1 +

C3: Ã Πu-X Σg (000)−(000)

3 -

NH: A Πi-X Σ Δv=0

2

2 -

2

CH: B Σ -X Πr 0-0

d) Free-burning carbon arc

3600

3800

3

Δv=+1

2 +

2 +

Δv=+2

Δv=-2

Δv=0

3400

3

C2: d Πg-a Πu

1-1

CN: B Σ -X Σ Δv=-1

Δv=+1

2

CH: A Δ-X Πr 0-0

c) Propane-butane/air flame

4000

4200

4400

4600

4800

4400

4600

4800

Air Wavelength / Å Figure 4. Low-resolution emission spectra from: a) PLA of carbon at an air pressure of 4 Pa, excited by the 9P(28) line at 1039.36 cm-1 of the CO2 laser; b) Acetylene/oxygen flame; c) Propane-butane/air flame; d) Free-burning carbon arc.

For the assignment of the atomic lines of C, C+, C2+, C3+, C4+, H, N and O we used the information tabulated in NIST Atomic Spectral Database [36]. The observed emission molecular bands are identified using the spectroscopic information available in Refs. [50]. Moreover, these molecular bands were compared with the spectra obtained in our laboratory by conventional sources (free-burning carbon arc, propane-butane/air flame and acetylene/oxygen flame). As example figure 4 shows several time-integrated OES at lowresolution from: (a) PLA of carbon (air pressure of 4 Pa and CO2 laser power density IW=1.00 GW cm-2); (b) Acetylene/oxygen flame; (c) Propane-butane/air flame; (d) Free-burning carbon arc. In the Acetylene/oxygen flame around 405 nm, several bands of the ~1 ~ C3 ( A Π u − X 1Σ +g ) comet head group are observed which were not detected in the PLA of carbon. As shown in Figure 4, ionic carbon lines C+, C2+, C3+ and C4+ cannot be observed in flames [Figure 4(b,c)] or carbon electric arcs [Figure 4(d)].

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

1-1

2-2

3-3

-

3

0-0

6-3

7-4

2 + 2 +

3325

10-8

6-4

3300

8-6

5-3

Δv=+2

Q2 Q1

9-7

CN: B Σ -X Σ

7-5

5-2

0-0

3

NH: A Πi-X Σ

2

0

1

0

R2 R1

P2

P1

(b)

Δv=0

1-1

2

OH: A Σ -X Πi Δv=0

1

C2+: 2p( P )3s P 1-2s3d D2

2 +

2

0-0

2 +

CH: C Σ -X Πr Δv=0 2 + 2 + CN: B Σ -X Σ Δv=+3

C+

(a)

1-1

88

3100

3125

3150

3050

3075

3100

3125

3150

3275

Air Wavelength / Å

C2: C1Πg-A1Πu

13-12

3425

0-0

Δv=0

C+

2

2 +

C+

C+

2 -

C+C+

12-11

11-10

2 +

Δv=+1 (tail bands)

3400

2

3-2 2-1 4-3

2 +

CN: B Σ -X Σ Δv=0

C2: C1Πg-A1Πu Δv=+1

6-5 7-6 8-7 9-8 10-9

2 +

CN: BΣ -X Σ

3375

(d)

4-3 3-2 2-1 1-0

(c)

2 +

1-0

2 +

CN: B Σ -X Σ Δv=+1

3350

Air Wavelength / Å

+

3025

3075

0 C : 2s24s S1/2-2s23p P1/2 3/2

3000

3050

0-0

3025

4-4 3-3 2-2 1-1

2975

3000

2

CH: B Σ -X Πr 0-0

3750

3525

3550

3575

3600

3625

3775

3800

3825

3650

3850

3875

3900

3925

Air Wavelength / Å

Q2 Q1 2 +

2 +

0-1

1-2

3-4 2-3

4-5

CN: B Σ -X Σ Δv=-1

Hγ

4150 4175 4200 4225 4250 4275 4300 4325 4350 4375

Air Wavelength / Å

5500

5525

5550

5575

5600

5625

C+: 2s2p(3P0)3p 4S3/2-2s2p(3P0)3s 4P05/2

Δv=-1

C+: 2s2p(3P0)3p 4S3/2-2s2p(3P0)3s 4P03/2

0-1

1-2

4-2 3-1 2-0

(f)

3

C2: d Πg-a Πu 2-3

3

3-4

C2+

C+

C2+

3

3

C2: d Πg-a Πu 2 2 CH: A Δ-X Πr Δv=+2 P1 P2 0-0 R2 R1

C+

(e)

C+ C2+

Air Wavelength/ Å

5650

Air Wavelength / Å

Figure 5 (a)-(f). Measured high-resolution PLA of carbon emission spectra observed in different regions at an air pressure of 4 Pa, excited by the 9P(28) line of the CO2 laser with a laser intensity of 5.36 GW cm-2 , and assignment of some atomic lines and molecular band heads.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

89

In order to get more insight into PLA of graphite and to obtain an unambiguous assignment of the emission lines and molecular bands, we have scanned the corresponding wavelength regions with higher resolution (~0.12 Å in first-order). The spectra have been obtained with twenty-four successive exposures on the CCD camera in the spectral region 200-750 nm by a ISA Jobin Yvon Spex 0.32 m spectrometer. As examples, Figure 5(a-f) shows several spectra recorded in the PLA of carbon experiment at high-resolution. These spectra were recorded in the following experimental conditions: air pressure 4 Pa, CO2 laser excitation line 9P(28) at 9.621 μm and laser intensity 5.36 GW cm-2. The relative intensities of the observed emission lines reasonably agree with tabulated values in NIST Atomic Spectral Database [36]. In Figure 5(a-f) we have indicated with italic the position of the band heads v’-v” of violet system of CN while in regular typeface the bands of the other molecular systems. In Figure 5(a-f), a rather complex structure is observed, in consequence of the overlapping between rotational lines of different molecular band systems. Figure 5(a) displays the overlapping between CH(C-X; Δv=0 sequence), CN(B-X; Δv=3 sequence), and OH(A-X; Δv=0 sequence). The relative position of the main branches for the OH(A-X) 0-0 band is indicated. In Figure 5(b), the high intensity of the 0-0 band for NH(A-X) is observed. This fact is in agreement with the high Franck-Condon factor (q00=0.9998) for this transition. In Figure 5(c) a partial overlapping among CN(B-X; Δv=1) and C2(C-A; Δv=1) is observed. This spectrum clearly shows the reversal of the bands from v”=5, which is due to the overlap between high vibrational quantum number bands with low vibrational quantum number bands. So, the first vibrational bands (1-0, 2-1, 3-2, 4-3 and 5-4) are shaded to the violet and after reversal (6-5, 7-6, …) are shaded to the red. Figure 5(d) shows a portion of the rotational lines for the CH(B-X) 0-0 band with several single ionized carbon lines. A coincidence in the position among the CN(B-X) 4-4 and C2(C-A) 0-0 band heads is observed. Very weak emission attributable to the N2+(B2Σu+–X2Σg+) system (the most prominent v’=0-v”=0 transition appears at ∼ 391 nm) is also identifiable. In the spectrum of Figure 5(e) the CN(BX) Δv=-1 sequence, CH(A-X) 0-0 band, and C2(d-a) Δv=2 sequence were identified. Also, several C+, C2+, and atomic hydrogen lines are observed. Finally, Figure 5(f) displays the rotational structure of C2(d-a) 0-1, 1-2, and 2-3 bands. The spectral features clearly show the complexity of the relaxation process and bring out the possibility of cascading processes.

4.2. Plasma Excitation, Vibrational and Rotational Temperature Measurements The excitation temperature Texc was calculated from the relative intensities of some C+ atomic lines (250-470 nm spectral region) according to the Boltzmann equation (2.23). The estimated excitation temperature was Texc= 23000 ± 1900 K (Figure 6). The relevant spectroscopic parameters for the C+ transitions have been listed in Table 2.

90

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

0.5

C+

ln(Iki λki/gk Aki) (a.u.)

0.0

Texc=23000 ± 1900 K

-0.5 -1.0 -1.5 -2.0

190000

200000

210000

220000

230000

240000

Ek/kB (K)

Figure. 6. Linear Boltzmann plot for several C+ transition lines used to calculate plasma temperature, Texc. Plot also shows linear fit to the data with a regression coefficient of R2~0.98.

Table 2. List of C+ transition lines and their spectral database (NIST Atomic Spectra Database, 2006) used for plasma temperature calculation. Transition array

Air λ (Å)

gi

gj

Aji (s-1)

Ei (cm-1)

Ej (cm-1)

2s2p2 2P1/2-2p3 2D03/2 2s2p2 2P3/2-2p3 2D05/2 2s23p 2P01/2-2s2 4d 2 D3/2 2s2p2 2S1/2-2s23p 2P03/2 2s2p2 2S1/2-2s23p 2P01/2 2s23p 2P01/2-2s2 4s 2S1/2 2s23p 2P03/2-2s2 4s 2S1/2 2s23d 2D3/2-2s2 4f 2 0 F 5/2 2s23d 2D5/2-2s2 4f 2 0 F 7/2

2509.12 2512.06 2746.49

2 4 2

4 6 4

4.53×107 5.42×107 4.36×107

110624.17 110665.56 131724.37

150466.69 150461.58 168123.74

Rel. Int. (Arb. Uni.) 20795 41490 8500

2836.71 2837.60 3918.98 3920.69 4267.00

2 2 2 4 4

4 2 2 2 6

3.98×107 3.97×107 6.36×107 1.27×108 2.23×108

96493.74 96493.74 131724.37 131735.52 145549.27

131735.52 131724.37 157234.07 157234.07 168978.34

76920 44700 7500 16000 45000

4267.26

6

8

2.38×108

145550.70

168978.34

70000

The detection of the C2(d–a) Swan and CN(B-X) bands is of particular interest since it provides an estimation of the plasma vibrational temperature. The emission intensities of the C2 Swan Δv=-1 and CN Δv=0 band sequences were analyzed in order to calculate the molecular vibrational temperature Tvib. For a plasma in LTE, the intensity of an individual vibrational v’-v” band Iv’-v” is given by Eq. (2.24). Two Boltzmann plots of the band intensities against the vibrational energy are given in Figure 7, along with the corresponding Franck-Condon factors. For C2 and CN the estimated vibrational temperatures were Tvib=18800 ± 860 K [Figure 7(a)] and 21400 ± 900 K [Figure 7(b)], respectively. Figure 8 shows the variation of vibrational temperature at 4 Pa of air pressure with laser fluence. The vibrational temperature is maximum at a most efficient laser fluence of 287 J cm-2. These results are consistent with earlier reports on vibrational temperature by different authors [6,11,12,16].

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

91

(a) v'=1

0.30

v'=2

Franck-Condon factor,qv'-v"

ln(Iv'-v" λ4v'-v"/qv'-v") (a. u.)

40.7 40.6 40.5

0.25

v'=0

v'=3

0.20

v'=4 0.15

v'=5 0.10

v'=6 v'=7

0.05

40.4

v'=8

v'=9

0.00

1

40.3

2

3

4

5

6

7

8

9

10

v" 3

40.2

3

C2: d Πg-a Πu; Δv=−1

40.1

Tvib=18800 ± 860 K

40.0 39.9

30000 32000 34000 36000 38000 40000 42000

G(v')hc/kB (K)

34.0

0.25

0.15 0.10

v'=1

0.20

v'=11

v'=15

v'=3

0.30

v'=0

Franck-Condon factor,qv'-v"

ln(Iv'-v" λ4v'-v"/qv'-v") (a. u.)

34.1

v'=5

0.35

34.2

v'=13

34.3 v'=8

(b)

0.05

33.9

0.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

v"

33.8 2 +

2 +

CN: B Σ -X Σ ; Δv=0

33.7 33.6

Tvib=21400 ± 870 K

33.5 33.4

2000

4000

6000

8000 10000 12000 14000 16000

G(v')hc/kB (K) Figure 7. Left (a) panel: Linear Boltzmann plot of the C2 Swan Δv=-1 band sequence intensity versus the normalized energy of the upper vibrational level; Right (b) panel: Linear Boltzmann plot of the CN violet Δv=0 band sequence intensity versus the normalized energy of the upper vibrational level; Experimental conditions: laser power density of 4.5 GW cm-2 and vacuum pressure 4 Pa. Plots also show linear fit to the data and the corresponding Franck-Condon factors.

92

J. J. Camacho, J.M.L. Poyato, L. Díaz et al. 21000 20000 19000

Tvib (K)

18000 17000 16000 15000 14000 13000 12000 50

100

150

200

250

300

350

-2 Laser Fluence (J×cm )

3/2

11/2

17/2

25/2

R2e,R2f

33/2

Figure 8. The vibrational temperature Tvib calculated from the C2 Swan Δv=-1 sequence bands as a function of the CO2 laser fluence. 2

2

CH: A Δ -X Π r; v'=0-v"=0 Q2ef,Q 2fe

J"=3/2, ...

Q12ef,Q12fe P12e,P12f

4-2

2-0

5-3

Q1ef,Q1fe

J"=3/2, ...

3

3

C 2: d Π g-a Π u Δ v=+2

3-1

P21e,P21f

21/2

27/2

33/2

Q21ef,Q21fe J"=3/2, ... R1e,R1f

13/2

2+

5/2

1/2

C

9/2

3/2

+

5/2

R21e,R21f

9/2

13/2

J" 17/2

C

15/2

P2e,P2f J"=3/2, ... R12e,R12f

J"=7/2, ...

4225

4250

4275

4300

4325

4350

25/2

15/2

J"

4200

7/2

P1e,P1f

4375

4400

Wavelength / Å Figure 9. High-resolution PLA of carbon emission spectra at an air pressure of 4 Pa, excited by the 10P(20) line of the CO2 laser with a laser intensity of 6.31 GW cm-2, and assignment of some ionic carbon lines, the band heads of C2(d-a) Δv=+2 sequence and the rotational structure of the 0-0 A2Δ-X2∏ band of CH.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

93

Figure 9 presents a typical resolved LIB emission spectrum of graphite and its rotational assignment for the v’=0-v”=0 band of the A2Δ-X2Π system of CH. This electronic band is the strongest visible feature of the air, oxygen/acetylene flame and a dominant feature of all hydrocarbon combustion (see figure 4). This spectrum consists in a doublet due to a transition between a Δ upper state (Hund case (b), Ae=-1.11 cm-1) and a Π ground state intermediate between Hund cases (a) and (b) (Ae=+28.1 cm-1) depending on the J value. The fine-structure components are indicated on the branch designations by subscript: 1 ≡ 2Δ5/2-2Π3/2 and subscript 2 ≡ 2Δ3/2-2Π1/2. The upper and lower states, Λ doublets are labelled by e and f; when there are the same for both rotational levels ee is abbreviated by e and ff is abbreviated by f. The selection rules involving the parity levels (e or f) are: e ∀# f, e ΠΤ e, f ΠΤ f for ΔJ=±1 (R and P branches) and e ΠΤ f, e ∀# e, f ∀# f for ΔJ=0 (Q branches) [42]. For each value of the quantum number N (N=J-S), there are four nearly degenerate energy levels, e or f, J=N±1/2. Based on these assumptions we can expect 12 main branches (ΔJ=ΔN corresponding to R, P and Q branches) and 12 satellite branches (ΔJ≠ΔN). A partial overlapping of the 0-0 band in the region of P1 and P2 branches of the Δv=+2 of the C2 Swan band, whereas two lines of C+ and C2+ are also present. To estimate the effective rotational temperature, we consider the J value for the maximum of the 0-0 band (A-X) of CH. This effective rotational temperature is found to be Trot=2060 ± 50 K for Jmax=13/2 of the R1 branch. Figure 10 shows the simulated spectra for the 0-0 A2Δ-X2∏ band of CH calculated (Eq. 2.25) at different temperatures. A good agreement between simulated and observed spectra at a temperature of about 2000 K over the entire range 4200-4400 Å proves that self-absorption is negligible and the rotational levels follow a Boltzmann distribution.

C H : A 2 Δ -X 2 Π v '= 0 -v " = 0

T ro t = 1 0 0 0 K

4200

4250

T ro t = 2 0 0 0 K

4300

4350

4250

4250

4300

4350

4400

4300

4350

4400

T ro t = 4 0 0 0 K

T ro t = 3 0 0 0 K

4200

4 4 04 02 0 0

4300

4350

W a v e le n g th / Å

4 4 04 02 0 0

4250

W a v e le n g th / Å

Figure 10. Simulated spectra for the 0-0 A2Δ-X2∏ band of CH calculated at different temperatures.

94

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

4.3. Ionization Degree of the Plasma As the CO2 laser beam is focused on the graphite surface, the carbon material absorbs the laser energy to melt, vaporize, and excite the target material. The carbon vapor absorbs more energy and forms high temperature plasma near the surface. The plasma expands into the low-vacuum atmosphere (N2, O2, H2O, etc) and transfers its energy to it. If the pressure around the target is bigger than ~1000 Pa, the breakdown of the air takes place in a significant way. Neutral, single and highly ionized carbon emission lines are found close to the target graphite surface. The carbon clusters and the molecules of the atmosphere obtain an energy that exceeds the binding energy. In these conditions the plasma becomes a mixture of electrons, positive ions such as C+, C2+, C3+, C3+, C4+, neutral atoms such as C, N, O and H, and molecules such as C2, CN, CH, NH, and OH in excited electronic states. Let us consider the first three different ionization equilibria of carbon: 2

3

0

C(2s22p2 P0 ) ↔ C+(2s22p P1 / 2 ) + e + IP(C-I), 2

0

1

C+(2s22p P1 / 2 ) ↔ C2+(2s2 S 0 ) + e + IP(C-II), 1

2

C2+(2s2 S 0 ) ↔ C3+(2s1 S1 / 2 ) + e + IP(C-III), where the first three ionization potentials (IPs) for carbon are IP(C-I)=11.2603 eV, IP(CII)=24.3833 eV and IP(C-III)=47.8878 eV [51]. Taking into account the consideration of section 2.4.5, we can obtain the ionization degree. Figure 11 shows the ionization degree Ni/(N0+Ni) of C, C+ and C2+, plotted as a function of the gas temperature T at a constant total pressure P=(N0+ne+Ni)kBT. The graph shows that carbon is already fully ionized at thermal energies well below the first ionization-energy of 11.2603 eV (equivalent to 130670 K). If we consider a temperature of 23000 K, the ionization degrees of C, C+ and C2+ obtained by means of the Saha equation are 0.999, 0.999 and 0.28, respectively. These so high values of the ionization degrees justify the observed emission spectra. Keeping in mind these results, the temperature obtained from the relative intensity of C+ lines was chosen as the first approximation for the average excitation temperature.

4.4. Electron Number Density The electron number density was obtained by considering the discussion reported in section 2.4.3. In our experiments, for C+ lines, the Doppler line widths are 0.08-0.13 Å at 23000 K (Eq. 2.18). Stark line broadening from collisions of charged species is the primary mechanism influencing the emission spectra in these experiments. In our case, the estimation of electron density ne has been carried out by measuring the broadening of the spectral profiles of isolated lines of C+ (2174, 2747, 2837, 2993, 3877, 3920, 4267, and 5890 Å) from the time-integrated high-resolution spectra. The electron number densities of the laser-induced plasma were determined at a laser power density of IW=1 GW cm-2 and air pressure of 4 Pa. A Lorentz functions were used to fit the spectra. In order to extract the Stark broadening from the total experimentally measured line broadening, we have to previously deconvolute the

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

95

main effects that contribute to the broadening of the spectral line. Values of the electron impact half-width W were taken from the extensive tables given by Griem [27]. Electron densities in the range (0.69-5.6)×1016 cm-3, with an estimated uncertainty of 10%, were determined. At the evaluated temperature of 23000 ± 1900 K, Eq. (2.4) yields ne≈(0.392.2)×1016 cm–3. These electron densities are close to measured values. Based on these calculations, it is difficult to tell whether the plasma is in LTE or not. A possible reason for non-thermal equilibrium could be the large integration time used in the experiments. The formation of ionic species is a usual phenomenon in LIB technique. The interaction between the laser and the ablation plume is governed by EII and/or by multiphoton ionization, both followed by electron cascade. EII is the most important for the longer wavelengths used in this work. MPI (Eq. 2.10) on the other hand is relatively improbable for carbon atoms in the ground state C(2s22p2 3P0), since its high ionisation potential (11.2603 eV [51]), means that 88 photons are required for this process. The observed ionic emissions are best explained by an EII mechanism (Eq. 2.11). The free or quasifree electrons are produced by the highpower laser pulse at the target surface. These electrons gain sufficient energy from the laser field through inverse bremsstrahlung collisions with neutrals, to ionize carbon atoms or ions by inelastic electron-particle collisions resulting in two electrons of lower energy being available to start the process again (Eq. 2.11). In general, the probability of MPI is WMPI ∝ ΦWn ∝ FE2 n . Calculations of PMI probability for carbon give a negligible value of WMPI for the CO2 laser at λ=10.591 µm and IW=6.31 GW×cm-2 (n=88) (Eq. 2.9). 1.0 0.9

Ionization degree

0.8 2+

C

0.7

+

C

0.6

C

0.5

2 3

+

2

2+

2 1

3+

2

0

0.4

C(2p P0) ↔C (2p P1/2)+1e

0.3

C (2p P1/2)↔C (2s S0)+1e

+

2

2+

2 1

0

C (2s S0)↔C (2s S1/2)+1e

0.2 0.1 0.0

0

5000

10000

15000

20000

25000

30000

T/K Figure 11. Temperature dependence of the ionization degree Ni/(N0 + Ni) of carbon C, carbon singly ionized C+ and carbon doubly ionized C2+ at a constant pressure of 4 Pa.

96

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

4.5. Effect of Laser Irradiance

Air Wavelength / Å

C2: d3Πg-a3Πu Δv=+1

2-1

2º Diffraction order C2+ C2+ C3+ 4-3 3-2

CH: A2Δ-X2Πr v'=0-v"=0

C2+

C2+ C+

C+

C+ C2+

C+

0-0

0-0

3-3

(b)

10-9 1-0 12-11

C

2+

2000 2200 2400 2600 2800 3000 3200 3400 3600

-2

5.36 GWcm 4.50 3.46 2.41 1.00 -2 0.65 GWcm

CN: B2Σ+-X2Σ+ Δv=0

C+

C+

-2

5.36 GW cm 4.50 3.46 2.41 1.00 0.65 0.54 -2 0.48 GW cm

C2+

2+ C2+ C

C+

C+

(a)

C C3+

C2+

Laser-sample and laser-plasma interactions are strongly dependent on the laser beam irradiance on the target. To see the effect laser irradiance the measurements were also carried out at different laser fluences. Optical emission spectra of the carbon plasma plume in medium-vacuum (~4 Pa) as a function of the laser intensity are shown in Figs. 12(a) and 12(b). These spectra were recorded at a constant distance of 1.5 cm from the target surface along the plasma expansion direction. An increase of atomic and molecular emission intensity with increasing the laser fluence was observed. Figure 13(a) shows the emission intensity change of C(247.856 nm), C+(251.206 nm), C2+(269.775 nm), C3+(252.998 nm), C4+(227.792 nm), OH 0-0 band head (306.35 nm), and NH 1-0 band head (336.00 nm) as a function of the carbon dioxide laser fluence. The C3+, C+ and C emission intensity increases drastically with the laser fluence. Beyond ~100 J cm-2, a sharp increase of atomic (especially for C3+, C+ and C) and molecular line intensities was observed. The C2+, C4+, OH 0-0 band head, and NH 1-0 band head emission intensity increases lightly with the laser fluence. Figure 13(b) shows the emission intensity change of C+(392.07 nm), C2+(418.69 nm), NH 1-0 band head of the A-X system, CN 1-0, 2-1, 0-0, 1-1 band heads of the B-X violet system, C2 1-0 band head of the C1Πg-A1Πu Deslandres-d’Azambuja system, CH 0-0 band head of the A-X system, C2 1-0 band head of the d-a Swan system, and Hβ line as a function of the laser fluence. An increase of atomic and molecular emission intensity with increasing the laser fluence was observed. Also the background increases with the laser power. At higher laser fluences (154-342 J cm2 ), the spectral lines and molecular bands are considerably more broadened than at lower fluences as a result of the high pressure associated with the plasma. It is assumed that at higher laser fluence the PLA plasma is more energetic and more ionized so that the surrounding air can better confine the plasma; the plasma also cools down more rapidly due to the confinement.

3700 3800 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800

Air Wavelength/ Å

Figure 12. Low-resolution PLA of carbon emission spectrum observed in the (a) 2000-3640 Å and (b) 3660-4800 Å regions, at an air pressure of 4 Pa, excited by the 9P(28) line at 1039.36 cm-1 of the CO2 laser, as a function of the laser power density.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

(a)220000 160000 140000 120000

NH1-0A-X CN1-0B-X CN2-1B-X C21-0C-A

100000

Intensity / a.u.

180000

Intensity / a.u.

(b) 120000

C (247.856 nm) + C (251.206nm) 2+ C (269.775nm) 3+ C (252.998nm) 4+ C (227.792nm) OH0-0band(306.35nm) NH1-0band(336.00nm)

200000

100000 80000

80000

CN0-0B-X CN1-1B-X + C (392.07nm) 2+ C (418.69nm) CH0-0A-X C21-0d-a(Swan)

60000

40000

Hβ

60000

20000

40000 20000

97

20 30 40 50 60

0

0

50

100

150

200

250

300

350

-2

Laser Fluence/ J×cm

50

100

150

200

250

300

350

-2

Laser Fluence/ J×cm

Figure 13 (a)-(b). Emission intensity change of: (a) C(247.856 nm), C+(251.206 nm), C2+(269.775 nm), C3+(252.998 nm), C4+(227.792 nm), OH 0-0 band head (306.35 nm), and NH 1-0 band head (336.00 nm); (b) C+(392.07 nm), C2+(418.69 nm), NH 1-0 band head of the A3Π-X3Σ- system, CN 1-0, 2-1, 0-0, 1-1 band heads of the B2Σ+-X2Σ+ violet system, C2 1-0 band head of the C1Πg-A1Πu, CH 0-0 band head of the A2Δ-X2Π, C2 1-0 band head of the Swan system d3Πg-a3Πu, and Hβ line as a function of the carbon dioxide laser fluence.

4.6. Effect of Ambient Pressure on the Plasma The emission characteristics of the laser-induced plasma are influenced to a large extent by the nature and composition of the surrounding atmosphere. The pressure of the air ambient atmosphere is one of the controlling parameters of the plasma characteristics, as well as the factors related to the laser energy absorption. An interesting observation was the effect of the air pressure, studied in the range 4.6 to 63500 Pa. Figs. 14(a)-(b) show typical OES from a carbon plasma plume at different air pressures. These plasma plumes were generated by the CO2 laser intensity of 1.00 GW cm-2. In general, the spectra of the PLA plume at low pressures (P<1500 Pa) are dominated by emission of electronic relaxation of excited atomic C, N, H, O, ionic fragments C+, C2+ C3+ and C4+, and molecular features of C2(E–A), C2(D– X), C2(d–a), C2(D–X), C2(e–a), C2(C–A), C2(A-X), CN(D–A), CN(B–X), CN(A–X), OH(A–X), NH(A–X), CH(C–X), CH(B–X) and CH(A–X). The spectra of the PLA plume at high pressures (P>10000 Pa) are dominated by emission of electronic relaxation of excited atomic N, O, H, ionic fragments N+ and O+, and molecular features of CN(B–X) and CN(A–X). The intensities of the C2 1-0 band head of the C1Πg-A1Πu (3607 Å), C2 1-0 band head of the D1Σ+u–X1Σ+g (4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+-X2Σ+ violet system, CH 0-0 band head of the A2Δ-X2Π (4307 Å), C+(3919 Å), C+(4267 Å), C2+(4593 Å), C3+(4657 Å), and Hβ spectral lines increase with increasing pressure, reach a maximum at about 200 Pa, and then decrease with higher pressures. Similar results were reported in the literature [52, 53]. Figure 15 shows the evolution of the emission intensity of C2 1-0 band head of the C1Πg–A1Πu (3607 Å), C2 10 band head of the D1Σ+u–X1Σ+g (4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+–X2Σ+ violet system, CH 0-0 band head of the A2Δ–X2Π (4307 Å), N+(3437 Å), O+(3410 Å), C+(3919 Å), C+(4267 Å), C2+(4593 Å), C3+(4657 Å), and Hβ atomic lines as a function of air

98

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

pressure. From figure 15, the intensity of the C2+(4593 Å), C+(4267 Å), and C3+(4657 Å) spectral lines is found to be more sensitive to the pressure that the CH 0-0 band head of the A–X (4307 Å), C+(3919 Å), and Hβ atomic lines. The lines N+(3437 Å) and O+(3410 Å) produced in the breakdown of the air, are not observed in the PLA of carbon at low-air pressures. The intensity of CN (Δv=0 sequence) increases with increasing air pressure, reach a maximum at about 200 Pa, and then stays constant as the pressure is increased further. Beyond 200 Pa (see Figure 15), a decrease in the time-integrated emission intensities of C+, C2+, C3+, CN, C2, CH, NH and H was found. However, an increase in the emission intensities of the N+ and O+ lines was observed. We suggest that these effects are related to shielding by the air plasma, where a part of the laser energy is absorbed by the air plasma during its expansion. This result in a reduction of the atomic and ionic emission intensity of species formed from the carbon target. At low pressures (P<200 Pa), the C, C+, C2+, C3+, and CN, C2, CH, NH emissions are produced nearer to the carbon target than the N+ and O+ emissions produced nearer to the air plasma position. In general, the air ambient gas will confine the plasma near the target (produced mainly by C, C+, C2+, C3+, and CN, C2, CH) and prevent the electrons and species produced near the target escaping quickly from the laser focal volume (observation region). Therefore, the emission intensity increases with increasing pressure. However, at higher pressures (more than 200 Pa in our case), the ambient gas will hinder the plasma from penetrating the atmosphere and predictably cause a higher plasma temperature. The emission intensity of H, C, C+, C2+, C3+, and CN, C2, CH decreases because of the fact that the laser energy is absorbed by air, producing air breakdown and increasing the N, O, N+, and O+ emission intensity, in agreement with our observation in Figure 15. At lower air pressures, the absence of the shielding air plasma results in a strong increase in the intensity of the C, C+, C2+, C3+ emission from the carbon target plasma. At such lower air pressures the relative contribution of the N+, O and O+ emission diminishes, and the emission from carbon surface component becomes dominant. 3+

100000 50000 0 120000 80000 40000 0 140000 70000

+

O

+

O

+

O 2 + 2 +

CN: BΣ -XΣ Δv=+1

Pair=140Pa

+ +

+

N N N O N O + N O+ + O O+

NN O + N + O

N Pair=55000Pa

+

+

Δv=0

+

Pair=10100Pa

Δv=-1

2 + 2 +

CN: BΣ -XΣ

0

C2+ C2+ 2º order

3

Δv=+1

C C+

C Δv=0

C C+ Pair=4.6Pa 3+

2+

Hβ 3 3 C2: dΠg-aΠu Δv=-1

Pair=39.9Pa

Pair=664Pa

+

+

140000

N + N

N

70000

+

N

0

C3+

C+

Pair=4.6Pa

C

10000

C2+

0 20000

C+

Pair=35.5Pa

2+

50000

5000 0 120000 80000 40000 0 180000 120000 60000 0

+

+

3

C2: dΠg-aΠu

C+

100000

C C 2º order

(b) 1150000000

3-4 2-3 1-2 0-1

(a) 150000

+

+ +

2-2 1-1 0-0

+

0

100000 50000

O O +

0

3400 3600 3800 4000 4200 4400 4600 4800

Air Wavelength/ Å

N Pair=65300Pa

+

O

N P =9040Pa air

O O O

5000 5200 5400 5600 5800 6000 6200

AirWavelength/ Å

Figure 14 (a)-(b). Low-resolution PLA of carbon OES observed at various air pressures in: (a) 34004880 Å region and (b) 4800-6300 Å region.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

99

C2 C-A 1-0 3607 Å C2 d-a 1-0 4737 Å CN B-X 1-0 3596 Å CN B-X 0-0 3882 Å + N 3437 Å + O 3410 Å + C 3919 Å + C 4267 Å 3+ C 4657 Å 2+ C 4593 Å Hβ 4861 Å CH A-X 0-0 4307 Å

0.1

1

10

100

Air Pressure / mbar

+

9-7 10-8 11-10

2

2 +

+

C 2800

+

5-3

7-4

2

C N : B Σ -X Σ Δ v=+2

1-1

+

3

2

2

z=18 cm

5-2 6-3

z=14 cm

OH: A Σ -X Πi Δv=0

z=8 cm

+

C N: B Σ -X Σ Δ v=+3

6-4

2

0-0

z=5 cm

C

+

NH: A Πi-X Σ

3 -

0-0

Figure 15. Emission intensity change of C2 1-0 band head of the C1Πg-A1Πu (3607 Å), C2 1-0 band head of the d3Πg-a3Πu (4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+-X2Σ+ violet system, CH 0-0 band head of the A2Δ-X2Π (4307 Å), N+ (3437 Å), O+ (3410 Å), C+(3919 Å), C+(4267 Å), C2+(4593 Å), C3+(4657 Å), and Hβ line as a function of the air pressure around the carbon target.

2900

3000

3100

3200

3300

3400

Air W avelength / Å Figure 16. Optical emission spectra of the graphite ablation plume at: 5, 8, 14, and 18 cm from the target.

100

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

4.7. Spatial Characterization

C2+ 3+ 4-3 3 C 3 C2: d Πg-a Πu Δv=+1

z=0.2cm z=5cm z=8cm

10-9 1-0 12-10

2-1 3-2

C2+ C2+

C2+

C+ C2+

CH: A2Δ-X2Πr 0-0

(a)

CH: B2Σ--X2Πr 0-0

z=18cm

CN: B2Σ+-X2Σ+ Δv=0

z=14cm

C+

z=8cm

C+

z=5cm

(b)

2º Diffraction order C2+

0-0

CN: B2Σ+-X2Σ+ Δv=+1 1-0 1-0 C2: C1Πg-A1Πu

z=0.2cm

C2+

C+

C+

In this section we present experimental results on the laser ablation of a graphite target at Pair=4 Pa by using a high-power IR CO2 pulsed laser (λ=9.621 µm and laser fluence of 342 J cm-2) and distances from 0.2 up to 20 cm, from the target along the plasma expansion direction. We discuss the dynamics of the plume expansion and formation of different atomic (C, N, H and O), ionic (C+, C2+, C3+ and C4+) and molecular (C2, OH, CN, CH and NH) species. Although OES gives only partial information about the plasma particles, this diagnostic technique helped us to draw a picture of the plasma in terms of the emitting chemical species, to evaluate their possible mechanisms of excitation and formation and to study the role of gas-phase reactions in the plasma expansion process, allowing a discussion of the probable OH, NH, CH, CN, H, O and N formation by gas phase reactions during the propagation of the plasma plume. The plasma emission was recorded at several distances along the plasma expansion direction (Z axis of Figure 2) at a constant distance y=10 cm with respect to the focusing lens onto the entrance slit of the monochomator and parallel to the target surface. Figures 16-18 show time-integrated OES following nanosecond pulsed laser ablation of graphite monitored at several distances from the target in different spectral regions. One can see from these figures that the emission along the plasma Z-axis is, in the conditions of sensitivity of our detection system, about 20 cm. Also we can observe that the intensity of the C2+, C3+ and C4+ ionic emission lines decays rapidly at distances higher to 1.5 cm. On the other hand it can be see that at high distances from the graphite target (z=18 cm) atomic lines from C+, C, H, O, and N are still observed. The intensities of the molecular bands for different species (OH, NH, CH, CN and C2) generally decay with distance, being observed up to 18 cm.

z=14cm z=18cm Hβ

3500 3600 3700 3800 3900 4000 4100 4200

Air Wavelength/ Å

4200 4300 4400 4500 4600 4700 4800 4900

AirWavelength/ Å Figure 17. Optical emission spectra of the graphite ablation plume monitored at: 0.2, 5, 8, 14, and 18 cm from the target: (a) 3500-4200 Å; (b) 4200-4950 Å regions.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

(a)

z=18 cm

100

(b) 400

0

200

200

3

3

C2: d Πg-a Πu Δv=0

z=8 cm Δv=-2

Δv=-1

z=5 cm

0

4800

5000

5200

5400

5600

5800

Air Wavelength / Å

6000

v'-v"

C

Hβ

z=0.2 cm 4-6 3-5 2-4 1-3 0-2

40000

C2+

2-2 1-1 0-0

0 60000

C3+ C+

5000

20000

Emission Intensity / a. u.

0 2000 1500 1000 500 0 10000

O N

900 600

z=14 cm

400 200

4-5 3-4 2-3 1-2 +0-1

Emission Intensity / a. u.

600

6200

101

z=18 cm CC

C

O

C

z=14 cm

300 60000

z=8 cm

30000 0 60000 40000 20000 0 60000 40000 20000 0

z=5 cm

O

N

C

C

z=0.2 cm O

8600

8800

9000

9200

N 9400

CCC 9600

Air Wavelength / Å

Figure 18 (a)-(b). Optical emission spectra of the graphite ablation plume monitored at: 0.2, 5, 8, 14, and 18 cm from the target in the spectral region: (a) 4800-6250 Å and (b) 8430-9700 Å.

Figure 19(a) shows the emission intensity change of C+(3920.7 Å), C2+(4186.9 Å), Hβ(4861.36 Å), NH 1-0 band head (3365 Å), CN B-X 1-0, 0-0, and 1-1 band heads (3591, 3887, 3875 Å, respectively), C2 C-A 1-0 band head (3615 Å), CH A-X 0-0 band head (4317 Å), and C2 d-a 1-0 band head (4743 Å), as a function of the distance. The C2+ emissions are only observable close to the carbon target (z<4 cm). The intensity of CN, NH, CH bands, and Hβ line increase lightly with increasing the distance, reach a maximum at about 5 cm, and then stay constant as the distance is further increased. Beyond 8 cm, a decrease in the timeintegrated emission intensities of these species was found. The intensity of the C2 band falls continuously with the distance to the target. The continuous decrease of the C+, C2+, and C2 emission intensities with the distance indicates that these species are mainly formed from the carbon surface. This decrease is more drastic for the C2+ emission intensity probably due to the fact that the recombination with free electron is faster than for C+ species. The slight rise of the NH, CN, CH and Hβ emission intensities as a function of the distance up to zmax ≈ 5 cm indicates that these electronically excited species are mainly formed by gas phase reactions during the propagation of the plasma plume through the air gas. The NH, CN, CH and Hβ emission intensities diminish for z > 8 cm due to the decrease in density during the expansion of the plasma plume. In figure 19(b), the emission intensity change of O(8446.359 Å), O(9262.776 Å), N(8680.28 Å), C(9061.43 Å), C(9078.28 Å), C(9088.51 Å), C(9094.83 Å), C(9111.80 Å), and CN A-X 1-0 band head at 9192 Å as a function of the distance from the target are shown. The intensity of O(8446.359 Å), O(9262.776 Å), N(8680.28 Å) and CN A-X 1-0 band head increases lightly with increasing the distance from the target expansion direction, reach a maximum at about 5 cm, and then stays constant as the distance is further increased. Beyond 8 cm a decrease in the time-integrated emission intensities of these species was found. However, all the carbon emission lines decrease with increasing the distance from the target. As previously discussed in figure 19(a) the behaviour of carbon species are mainly formed from the surface target but, molecular species such as NH, CN and CH are produced in gas phase.

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

(a)

NHA-X1-0 CNB-X1-0 C2 C-A1-0

Relative Intensity / a.u.

100000

CNB-X0-0 CNB-X1-1 + C 3920.7Å 2+ C 4186.9Å CHA-X0-0 C2 d-a 1-0

10000

Hβ 1000

100000

Relative Intensity / a.u.

102

OI 8446.359Å OI 9262.776Å NI 8680.280Å CI 9061.43Å CI 9078.28Å CI 9088.51Å CI 9094.83Å CI 9111.80Å CNA-X1-0

10000

1000

(b) 0

2

4

6

8

10

12

14

16

18

Distancefromthetarget, z/ cm

0

2

4

6

8

10

12

14

16

18

Distancefromthetarget, z/ cm

Figure 19 (a)-(b). Emission intensity change of: (a) C+, C2+, Hβ, NH, CN, C2 (C-A 1-0 band head), CH, and C2 (d-a 1-0 band head) and (b) O (two lines), N, C(five lines) and CN (A-X 1-0 band head), as a function of the z distance.

I order to further identify properties of the ablation plasma plumes originated from graphite targets, we have estimated the vibrational temperatures (Eq. 2.24) of C2 molecule (da Swan Δv=-1 band sequence) as function of z-distance. The estimated vibrational temperatures were Tvib=23000±1000, 16200±900, 10800±600, 7700±500 K at 0.2, 1.5, 5 and 9 cm from the target along the plasma expansion direction, respectively compatible with a cooling stage.

4.8. Temporal Evolution of the Plasma In this section time-resolved OES analysis for the plasma plume, produced by highpower tunable IR CO2 pulsed laser ablation of graphite, at λ=10.591 µm and a laser fluence of 402 J cm-2 is presented. We focus our attention on the temporal evolution of different atomic/ionic and molecular species over a broad spectral range from 190 to 1000 nm. Excitation temperature, electron density and vibrational temperature in the laser-induced plasma were estimated from the analysis of spectral data at various times from the laser pulse incidence. In time-resolved measurements, the delay td and width tw times were varied. It was verified that the plasma was reproducible over more than 7 ablation events by recording the same spectrum several times. The temporal history of laser-induced breakdown carbon plasma is illustrated schematically in Figure 20. The time when beginning of the CO2 laser pulse is triggered is considered as the origin of the time scale (t=0). Inserts illustrate some emission spectra recorded at different delay and width times. The temporal shape of the CO2 laser pulse is also shown. Because the LIB plasma is a pulsed source the resulting spectrum evolves rapidly in time. The LIBS spectra of carbon were measured at different delay and width times. As an example figure 21(a) show OES of the graphite ablation plume at low-resolution in 2000-

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

103

C2+

C+

2800 Å region monitored at several delay times for a fixed gate width time of 0.1 µs. At early times (td<2 μs) the predominant emitting species are the C2+ 2p2 1D2 → 2s2p2 1P1 atomic line at 2296.87 Å and two lines of C3+ at 2524.41 and 2529.98 Å. The three molecular band systems observed in this spectral region are the C2(E1Σg+-A1Пu; Freymark), C2(D1Σu+- X1Σg+; Mulliken) and C2(e3Пg- a3Пu; Fox-Herzberg). These bands stay approximately constant as the delay time is increased up to ~5 μs, and decrease for higher delay times. Figure 21(b) shows the LIB emission spectrum of graphite plasma plume recorded 1, 1.2, 4 and 11 μs after the CO2 laser irradiation in the spectral region 2850-3600 Å. The main features in this region are the emission of C+ ionic species and several molecular emission bands from C2, OH and CN. The C+ emission intensities fall considerably as the delay changes from 1 to 11 μs. On the other hand, the molecular bands of C2, OH and CN emission intensities stay approximately constant as the delay time is increased up to ~5 μs, and decrease for higher delay times. From our results at low-resolution we can appreciated that the intensity of the C+, C2+ and C3+ ionic emission lines decays rapidly at delay times higher to 2 μs. On the other hand we can see that at high delay time after plasma ignition, atomic lines from C, H, O, and molecular bands of C2, CN and OH are still observed.

C+

td = 1 μs

2

td = 4 μs

2

CH : A Δ - X Πr 0-0

2 2 +

CN : B - X Δv=-1

3900

4000

4100

4200

4300

tw = 0.1 μs

4400

Wavelength / Å 2 +

3600

3700

3800

3900

4000

4100

4200

4300

2 +

CN : B Σ - X Σ Δv=0

4400

2

CH : A Δ - X Πr 0-0

2

3800

CN : B - X Δ v=-1

2 −

0.10

3700

Wavelength / Å 3600

3700

3800

3900

4000

4100

4200

4300

4400

Wavelength / Å

τFWHM= 64 ns t (gate width time) w td = 7 μs

0.05

2

tw = 0.1 μs

td (gate delay time)

2 +

2 +

CN : B Σ - X Σ Δv=0

CH : B 2Σ− - X 2Πr 0-0

Laser Power / a. u.

td = 21 μs

3600

2

CH : B Σ - X Πr 0-0

2 +

CN : B Σ - X Σ Δv=0

2 −

C2+

tw = 0.1 μs

CH : B Σ - X Πr 0-0

tw = 0.1 μs

2

CH : A Δ - X Πr 0-0

CN : B- X Δv=-1

CO2 Laser pulse 3600

3700

3800

3900

4000

4100

4200

4300

4400

Wavelength / Å

0.00

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5

Time / μs Figure 20. A schematic overview of the temporal history of laser-induced breakdown carbon plasma. Here td is the gate delay time and tw is the gate width time during which the plasma emission is monitored. Inserts illustrate some spectra observed at different delay and width times. The temporal shape of the CO2 laser pulse (recorded with the aid of the photon-drag detector) is also shown.

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

(b) C3+

td=1μs tw=0.1μs

1 + 1 +

C2: DΣu-XΣg

C+ C4+

C2+

3

3

C2: eΠg-aΠu

3

2+ 2+

CN:BΣ-XΣ 2 + 2 + Δv=+1 CN:BΣ-XΣ 2+ 2 OH:AΣ-XΠ Δv=+2 Δv=0

2+ C2+ C+C C++ C

1

C+

1 +

C2: EΣg-AΠu

3

C2:eΠg-aΠu C2+C+

(a)

td=1μs

C+

C2+

104

td=1.2μs

td=2μs td=4μs

td=4μs

td=6μs td=11μs

2000 2100 2200 2300 2400 2500 2600 2700 2800 Wavelength/ Å

td=11μs

2900 3000 3100 3200 3300 3400 3500 3600 Wavelength/ Å

Figure 21. Optical emission spectra of the graphite ablation plume monitored at different delay times for a fixed gate width time of 0.1 μs in two spectral regions.

In order to get more insight into laser ablation of graphite and to understand the laserinduced breakdown dynamics, we have scanned in the UV-Visible spectral region with higher resolution. As an example some results for the spectral region 4640-4750 Å are shown in figures 22-24. The spectral range was chosen in order to detect both double and triple ionized carbon species and C2 diatomic molecule. In figure 22, the data acquisition was performed by averaging the signal over: (a) 20 successive laser shots (td=0 and tw>>30 μs) and (b) 7 successive laser shots (td=4 μs and tw=0.02 μs). The emissions of ionized C2+(1s22s3p 3P02,1,0 → 1s22s3s 3S1) around 4650 Å, and C3+ around 4658.3 Å are considerably higher in the spectrum of Figure 22a, while the C2(d-a) Swan Δv=+1 sequence emission is similar. Figures 23a-b and 24 show the typical temporal sequence of laser-induced carbon plasma. At early times (td≤0.02 μs) emission from C2+ and C3+ is easily detected between 4645-4670 Å (see inset within figure 23-a). As seen in figure 23-b during the initial stages after laser pulse (td≤0.04 μs), C2+ emissions dominate the spectrum. As time evolves (0.04 μs≤td≤1.5 μs), C3+ emission dominate the spectrum. As the delay is increased up to 2.5 μs (1.5 μs≤td≤2.5 μs) again C2+ emission dominates the spectrum. These ionic lines decrease quickly for higher delay times, being detected up to ∼ 3 μs. Some oxygen and nitrogen ionic lines were also observed in the spectra at the gate delay from 0.02 μs to 1 μs and its emission intensities remain approximately constant in this time interval (see Figure 23a). They vanished after the delay of ∼1.5 μs. It shows that the air is ionized by the CO2 laser pulse and by the collisions with the laser induced plasma. During the time period up to ∼ 0.5 μs, no apparent C2 emissions were observed. As can be seen from Figure 24, the C2(d-a; Δv=1 band sequence) emissions were clearly observed from ∼2 μs. The C2 emission intensities increase lightly with increasing td, reach a maximum at ~5 μs, and then decrease as the time is further increased.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

105

3+ 2

3 0

2+

2

3

3

3

tw = 0.02 μs

C2: d Πg-a Πu

1-0

8-7

v'-v"

8-7

2-1

3-2

3

7-6

C : 1s 2s3s SJ"

6-5 5-4 7-6 4-3

2+

C

4-3

td = 4 μs J" 1

3

C2: d Πg-a Πu

1-0 v'-v"

2+

C : 1s 2s3p PJ'

0

2-1

2

J' 1

3-2

C

6-5 5-4

2+

C

(b)

(a) 4650

4675

4700

4725

4650

4750

4675

Wavelength / Å

4700

4725

4750

Wavelength / Å

Figure 22 (a)-(b). Measured high-resolution pulsed laser ablation of graphite emission spectra observed in the region 4645-4750 Å region. The data acquisition was performed by averaging the signal over: (a) 20 successive laser shots with td=0 and tw>>30 μs; (b) 7 successive laser shots with td=4 μs and tw=0.02 μs. The assignments of some ionic lines of C2+ and C3+ and molecular bands of C2 are indicated. The insert in (a) illustrates the rotational structure of one triplet of C2+ line.

3+

C

1000

td=0 ; tw=20 ns 100

+

500

7000

td=500 ns

6000

td= 20 ns

4650

4675 4700 Wavelength / Å

+

O

+

O O +

td=100 ns

5000 4000

tw=20 ns

td= 30 ns

200

0

O

3+

C

(b)

Relative Intensity / a. u.

1500

2+ C 2+ td=40 ns; tw=20 ns C

Relative Intensity / a. u.

Relative Intensity / a. u.

2000

2+

C

2+

C

3000 2000 1000 0

0

4650 4660 4670 4680 4690 4700 4710 4720 Wavelength / Å

4645

4650

4655

4660

4665

4670

Wavelength / Å

Figure 23(a)-(b). Time-resolved high-resolution emission spectra from laser-induced carbon plasma observed in the region: (a) 4645-4720 Å region monitored at 40 ns delay time; (b) 4645-4670 Å region monitored at 20, 30, 100, and 500 ns gate delay times for a fixed gate width time of 20 ns. The inset in (a) displays the spectrum the first 20 ns after incidence of the laser pulse.

Space-and-time resolved OESs laser-induced measurements could be used to estimate plasma expansion rate. To obtain additional time resolved information about the optical emission of the plume, wavelength resolved spectra have been recorded at different delay times at a distance of 9 mm. The temporal evolution of spectral atomic, ionic and molecular line intensities at a constant distance from the target can be used to construct the time-offlight (TOF) profile. TOF studies of the emission provide fundamental information regarding the time taken for a particular species to evolve after the laser-induced plasma has formed.

106

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

Specifically, this technique gives an indication of the velocity of the emitted species. A rough estimation of the velocity for the different species in the plume can be inferred from the time resolved spectra by plotting the intensities of selected emission lines versus the delay time, and then calculating the velocity by dividing the distance from the target by the time where the emission peaks. This method for determination of plasma velocity should be used with care due to the superposition of both expansion and forward movements of the plasma plume. 2+

C

3

8-7

1-0

td=10 μs

Relative Intensity / a. u.

30000

v'-v"

2-1

3-2

4-3

2+

C

3

C2:d Πg-a Πu

6-5 5-4

tw=0.02 μs

25000

20000

td=5 μs

3+

C

15000

td=3 μs

10000

2+

C

td=2 μs

5000

0

4640

4660

4680 4700 Wavelength / Å

4720

4740

Figure 24. Time-resolved high-resolution emission spectra from laser-induced carbon plasma observed in the region 4645-4750 Å region monitored at 2, 3, 5, and 10 μs gate delays for a fixed gate width time of 20 ns.

Figure 25 displays the TOF profile, for ablation experiments induced by CO2 laser pulses, of several C, C+ and C3+ lines intensities in UV region and C2+ in the visible as a function of delay time. However, the insert of the figure shows the time dependence of C2+ and C3+ line intensities in the visible region for ablation induced by CO2 laser pulses in which the tail has been eliminated by means of the suppression of the N2 in the gas mixture of the active laser medium. All the data are taken from high-resolution spectra and in the figures the temporal profiles of both kinds of laser pulses are also plotted. In both cases emissions from C3+ are stronger than emissions coming from the other species. All the ionic lines follow the time profile of lasers pulses lasting until four or three microseconds depending on the kind of the laser pulse. These behaviours may be related to the laser absorption processes on the target surface. Thus for “non tailed pulses” the line intensities start to growth at 400 ns while for “tailed pulses” start at 70 ns. Since the energy pumping (rise time) of each kind of pulse is different, the species reach the maximum intensity at different times: ∼1 μs for non tailed pulses and 700 ns for tailed ones, indicating that the graphite target needs some energy threshold to eject the different species. The higher intensity in the 0-400 ns time interval for the C2+ may be due to the higher sensibility of our ICCD camera in the visible region than in the UV one. The different behaviour of atomic C can be also observed in figure 25. Atomic C have a higher rise time and lasting more ( > 15 μs) than ionic species, possibly due to the

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

107

continuous recombination of ions with electrons to give excited carbon. From these results has not been observed excitation dependence on the pulse tail however the energy of the pulse intensity seems to be the pulse parameter that influence on the graphite ablation process. The peak velocities estimated for C3+, C2+ and C+ species, from figure 25, are about 7x103 m/s.

CO2 laser pulse CO2 laser pulse without tail C2+:4647.42 Å C3+: 4658.3 Å

Intensity / a.u.

Intensity / a. u.

C : 2478.56 Å + C : 2509.12 Å 2+ C : 4647.42 Å 3+ C : 2529.98 Å

0

1

2

3

4

5

Delay time / μs

0

1

2

3

4

16

18

Delay time / μs Figure 25. Emission intensity change of C(2478.56 Å), C+(2509.12 Å), C2+(4647.42 Å) and C3+(2529.98 Å) lines as a function of delay time (fixed gate width time of 20 ns) for a CO2 pulse laser with a tail of about 3 μs. The insert shows the emission intensity change of C2+(4647.42 Å) and C3+(4658.3 Å) lines as a function of delay time (fixed gate width time of 20 ns) for a CO2 pulse laser without tail.

In this section the plasma temperature was determined form the emission line intensities of several C+ lines observed in the laser-induced plasma of carbon target for a delay time of 1 μs and 0.02 μs gate width. The obtained excitation temperature was 26000 ± 3000 K. The carbon ionic multiplet line at ∼3920 Å was identified as candidate for electron-density measurements. Figure 26-a shows, the 3920 Å carbon ionic line with sufficient resolution to measure the full width at half-maximum at 8 different time delays. All the data points were fitted with Lorentzian function to determine the Stark line width. By substituting these values in Eqn. (2.21) and the corresponding value of electron impact parameter W (0.465 Å from Griem [27] at plasma temperature of 26000 K), we obtain the electron density. Figure 26-b gives the time evolution of electron density by setting the gate width of the intensifier at 0.02 μs. The initial electron density at 0.02 μs is approximately 3 1016 cm-3. Afterwards, the density increases over the period of 0.1 μs and reaches a maximum at 0.1 μs (time period of the peak CO2 laser pulse), and then decrease as the time is further increased. At shorter delay times (<0.1 μs), the line to continuum ratio is small and the density measurement is sensitive to errors in setting the true continuum level. For times >0.1 μs, the line to continuum ratio is within reasonable limits and the values of electron density shown in figure 26-b should be reliable. Initially the laser-induced plasma expands isothermically within the time of the duration of the laser pulse. After termination of the peak laser pulse (∼0.1 μs) the plasma

108

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

expands adiabatically. During this expansion the thermal energy is converted into kinetic energy and the plasma cools down rapidly. After 4 μs, the electron density is about 1.5 1016 cm-3. For a long time >4 μs, subsequent decreased C+ emission intensities result in poor signal-to-noise ratios, and there exits a limitation in the spectral resolution. The decrease of ne is mainly due to recombination between electrons and ions in the plasma. These processes correspond to the so-called radiative recombination and three-body recombination processes in which a third body may be either a heavy particle or an electron.

2

2 0

C :2s 4s S1/2→2s 3p P3/2 2

+

25000

2

16

4x10

td=3 μs +

2

2

td=1 μs

2 0

td=0.5 μs td=0.1 μs

tw=20 ns

td=0.05 μs

15000

td=0.02 μs td=0

10000

5000

Electron density / cm-3

2

C :2s 4s S1/2→2s 3p P1/2 20000

Relative Intensity / a. u.

td=4 μs

0

16

3x10

16

2x10

16

3917

3918

3919

3920

3921

1x10

3922

-3

10

Wavelength / Å

-2

10

-1

10

0

10

Delay time / μs +

Figure 26 (a). Stark-broadened profiles of the C line at 3920 Å at different delay times for a fixed gate width time of 0.02 μs. (b) Temporal evolution of electron density at different delay times from plasma ignition. In order to further identify properties of the ablation plasma plumes originated from graphite targets, we have estimated the vibrational temperatures of C2 molecule as function of delay time. The emission intensities of the C2 d-a Swan Δv=+1 band sequence were used to estimate these vibrational temperatures Tvib. The estimated vibrational temperatures were Tvib=8000±500, 8300±600, 7500±600, 4500±900 K at 3, 5, 10 and 15 μs after plasma ignition, respectively compatible with a cooling stage. Optical emission accompanying TEA-CO2 nanosecond laser ablation of carbon is very long lived (∼40 μs) relative to the average radiative lifetimes of the excited levels that give rise to the observed emission lines. At distances close to the target surface (<9 mm), all of the emission lines of C, C+, C2+ and C3+ expected in the 2000-10000 Å wavelength range are observed, illustrating that the excited species giving rise to the optical emission are produced by non-specific mechanism during the TEA CO2 laser ablation process. However, a direct excitation-de-excitation mechanism cannot explain the observed emission spectra. EII would explain the emission intensity variation with the time for C, C+, C2+, C3+ and C2 species. On the other hand, the formation of the excited molecular species would happen in gas phase by collisions between atomic or ionic species present in the plume and the residual gas at times far away from the plasma ignition. The emission process at this plasma stage is divided into two different process associated, respectively with the shock formation and the plasma cooling. During the former, the atoms, molecules and ions gushing out from the carbon target

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced...

109

are adiabatically compressed against the surrounding gas. During the latter stage the temperature of the plasma and consequently the emission intensities of atomic lines and molecular bands decrease gradually. The evolution of the TEA-CO2 laser-induced carbon plasma can be divided into several transient phases. The initial plasma (td<2 μs) is characterized by high electron and ion densities (1016-1019 cm-3), and temperatures around 2.2 eV. The emission spectrum from this early stage is characterized by emission lines from C3+, C2+ and C+ ions. Owing to the high electron density, the emission lines are broadened by Stark effect. The ionic emission lines (C3+, C2+, C+) decay rapidly being observed up to ∼3 μs. Emission lines from C atoms and molecular species (C2, CN, CH, OH) in excited electronic states can be found after about 1 μs time delay. As the plasma expands and cools, the electrons and ions recombine. After the initial plasma (td>3 μs), the molecular emissions increase slowly up to ∼5 μs and after that decay slowly up to ∼40 μs.

5. CONCLUSION This article reviews some fundamentals of LIBS and some experimental studies developed in our laboratory on the ablation of graphite using a high-power IR CO2 pulsed laser. In this experimental study we used several laser wavelengths (λ=9.621 and 10.591 µm) and laser intensity ranging from 0.22 to 6.31 GW cm-2. Ablation was produced typically at medium-vacuum conditions (∼ 4 Pa). Emissions from the resulting ablation plumes, and from the collisions with the ablated material and the background gas molecules (N2, O2, H2O, etc), have been investigated by wavelength-, space-, and time-resolved OES from UV-Vis-NIR. Wavelength-dispersed spectra of the plume reveal C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and molecular features of C2, CN, OH, CH, N2, N2+ and NH emissions corresponding to different electronic band systems. For the assignment of molecular bands a comparison with conventional emission sources was made. Excitation, vibrational and rotational temperatures, ionization degree and electron number density for some species were estimated by using different spectroscopic methods. The characteristics of the spectral emissions from the different species have been investigated as functions of the ambient pressure, laser irradiance, the distance from the target and delay time after plasma ignition. Time-gated spectroscopic studies have allowed estimation of TOF and propagation velocities for various emission species. Possible production routes for secondary emitters such as C2, CN, OH, CH, N2, N2+ and NH are discussed.

ACKNOWLEDGMENTS We gratefully acknowledge the support received in part by the DGICYT (Spain) Projects: MEC: CTQ2007-60177/BQU and MEC: CTQ2008-05393/BQU for this research.

110

J. J. Camacho, J.M.L. Poyato, L. Díaz et al.

REFERENCES [1] [2] [3] [4] [5]

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

Kroto, HW; Heath, JR; O’Brien, SC; Curl, RF; Smalley, RE. Nature, 1985, 318, 162-163. Koinuma, H; Kim, MS; Asakwa, T; Yoshimoto, M. Fuller. Sci. Technol, 1996, 4, 599-612. Ikegami, T; Nakanishi, F; Uchiyama, M; Ebihara, K. Thin Solid Films, 2004, 457, 7-11. Ashfold, MNR; Claeyssens, F; Fuge, GM; Henley, SJ. Chem. Soc. Rev., 2004, 33, 23-31. Yueh, FY; Singh, JP; Zhang, H. In Encyclopedia of Analytical Chemistry; Meyers RA. Ed. Laser-induced Breakdown Spectroscopy; Elemental Analysis; John Wiley & Sons Ltd, Chichester, 2000, pp 2066-2087. Dwivedi, RK; Thareja, RK. Phys. Rev. B, 1995, 51, 7160-7167. Demyanenco, AV; Letokhov, VS; Puretskii, AA; Ryabov, EA. Quantum Electronics 1998, 28, 33-37. Vivien, C; Hermann, J; Perrone, A; Luches, A. J. Phys. D: Appl. Phys. 1998, 31, 1263-1272. Wee, S; Park, S. M. Opt. Comm., 1999, 165, 199-205. Yamagata, Y; Sharma, A; Narayan, J; Mayo, RM; Newman, JW. J. Appl. Phys., 2000, 88, 6861-6867. Harilal, SS. Appl. Surf. Science, 2001, 172, 103-109. Acquaviva, S; Giorgi, ML. J. Phys. B: At. Mol. Opt. Phys., 2006, 35, 795-806. Saito, K; Sakka, T; Ogata, H. J. Appl. Phys., 2003, 94, 5530-5536. Zelinger, Z; Novotny, M; Bulir, J; Lancok, J; Kubat, P; Jelinek, M. Contrib. Plasma Phys., 2003, 43, 426-432. Saidane, K; Razafinimanana, M; Lange, H; Huczko, A; Baltas, M; Gleizes, A; Meunier, JL. J. Phys. D: Appl. Phys., 2004, 37, 232-239. Park, HS; Nam, SH; Park, SM. J. Appl. Phys., 2005, 97, 113103-5. Fuge, GM; Ashfold, MNR; Henley, SJ., J. Appl. Phys., 2006, 99, 14039-12. Camacho, JJ, Poyato, JML, Díaz, L; Santos, M. J. Phys. B: At. Mol. Opt. Phys., 2008, 41,105201-13. Camacho, JJ; Santos, M; Diaz, L; Poyato, JML. Appl. Phys. A., 2009, 94, 373-380. Camacho, JJ; Diaz, L; Santos, M; Juan, LJ; Poyato, JML. J. Appl. Phys., 2009, 106, 33306-11. Cremers, DA; Radziemski, LJ. Handbook of Laser-lnduced Breakdown Spectroscopy; Wiley: Chichester, England, 2006. Miziolek, AW; Palleschi, V; Schechter, I. (Eds.), Laser-lnduced Breakdown Spectroscopy; Cambrige. 2006. Singh, JP; Thakur, SN. (Eds.) Laser-lnduced Breakdown Spectroscopy; Elsevier: Oxford UK, 2007, Vol. 1, pp 1-427. Yong-Ill, L., Laser Induced Breakdown Spectrometry; Nova Science Publishers: New York 2000. Chan, F. Introduction to plasma physics and controlled fusion; Plenum Press: New York. 1984. Griem, HR. Principles of plasma spectroscopy; University Press: Cambridge. 1997.

Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced... [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53]

111

Griem, HR. Phys. Rev., 1962, 128, 515-523. MacDonald, AD. Microwave Breakdown in Gases Wiley; New York, 1966. Raizer, YP. Gas Discharge Physics; Springer: Berlin, Heidelberg. 1991. Kopiczynski, TL; Bogdan, M; Kalin, AW; Schotwau, HJ; Kneubuhl, FH. Appl. Phys. B: Photophys. Laser Chem., 1992, 54, 526-530. Radziemski, LJ; Cremers, DA; Laser-induced plasma and applications; New York: Dekker. 1989. Gurevich, A; Pitaevskii, L. Sov. Phys. JETP, 1962, 19, 870-871. Morgan, CG. Rep. Prog. Phys., 1975, 38, 621-685. Huddlestone, RH; Leonard, SL. Plasma diagnostic techniques; Academic Press: New York. 1965. Hutchinson, IH. Principles of plasma diagnostic; University Press: Cambridge. 2002. NIST Atomic Spectra Database online at http://physics.nist.gov/PhysRefData/ASD/ index.html Breene, RG. The Shift and Shape of Spectral Lines; Pergamon: London. 1961. Bengtson, RD; Tannich, JD; Kepple, P. Phys. Rev. A, 1970, 1, 532-533. Griem, HR. Spectral line broadening by plasmas; Academic Press: New York.1974. Herzberg, G. Spectra of diatomic molecules; Van Nostrand: New York. 1950. Steinfeld, JI. An introduction to modern molecular spectroscopy; MIT Press: London. 1986 Bernath, PF. Spectra of atoms and molecules; Oxford University Press: New York. 1995. Camacho, JJ; Pardo, A; Martin, E; Poyato, JML. J. Phys. B: At. Mol. Opt. Phys. 2006, 39, 2665-2679. Kovacs, I. Rotational Structure in the Spectra of Diatomic Molecules. Hilger: London. 1969. Cabalin, LM; Laserna, JJ; Spectrochim. Acta Part B, 1998, 53, 723-730. Demtröder, W. Laser Spectroscopy. Vol. 2. Experimental Techniques. Springer. Berlin 2008. Bogaerts, A; Chen, Z. Spectrochim. Acta Part B, 2005, 60,1280-1307. Zeldovich, YB; Raizer, YP. Physic of Shock waves and high temperature hydrodynamics phenomena. Academic, New York 1966. Drogoff, LB; Margotb, J; Chakera, M; Sabsabi, M; Barthelemy, O; Johnstona, TW; Lavillea, S; Vidala, F; Kaenela, VY. Spectrochim. Acta Part B, 2001, 56, 987-1002. Huber, KP; Herzberg, G. Molecular spectra and Molecular structure. IV. Constants of diatomic molecules; Van Nostrand Reinhold: New York. 1979. Martin, WC; Zalubas, R. J. Phys. Chem. Ref. Data,1983, 12, 323-380. Kim, DE; Yoo, KJ; Park, HK; Oh, KJ; Kim, DW. Appl. Spectrosc., 1997, 51, 22-29. Lu, YF; Tao, ZB; Hong, MH. Jpn. J. Appl. Phys., 1999, 38, 2958-2963.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 113-167

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 3

INDUCTION TRANSFORMER COUPLED DISCHARGES: INVESTIGATION AND APPLICATION I.M. Ulanov* and M.V. Isupov Kutateladze Institute of Thermophysics SB RAS, Novosibirsk, Russia

ABSTRACT Researches in the field of low-temperature plasma provide development of devices, which are widely used by the advanced and high-tech industries. Low-temperature plasma is applied by such major industries as microelectronics, semiconductor industry, solar cell production, plasma chemistry, metallurgy, lighting engineering, etc. Among the known methods of production and use of low-temperature plasma (DC and AC arc discharges, RF plasmatrons, microwave plasmatrons) the devices based on application of induction transformer coupled toroidal discharges (TCTD) are the least studied and, thus, rarely used. Simultaneously, these discharges can be used for development of electrodeless generators of low-temperature plasma: transformer plasmatrons and new induction sources of light. The chapter deals with investigation of properties of TCTD, transformer plasmatrons and induction light sources on the basis of TCTD. Investigation results on electrophysical properties of TCTD aimed at development of transformer plasmatrons are presented in the current paper. Dependences between the strengths of TCTD electric field, discharge current and gas flow are obtained for different gases within the pressure range of 10÷105 Pa. The thermophysical characteristics of TCTD were determined: device efficiency, energy balance of a discharge (heat losses to the discharge chamber wall, plasma jet power). The stable TCTD of the atmospheric pressure in argon and in air was firstly obtained and studied by the authors of this chapter. The process of plasmachemical synthesis of NO in air plasma of TCTD was studied. The abnormally high percentage of NO ~7 % was obtained without product quenching. The transformer plasmatrons of the 10÷200 kW power, operating under the pressures of 10÷105 Pa on argon, air and argon+hydrogen, argon+oxygen mixtures, were developed on the basis of the studies performed. The schemes and constructions of these plasmatrons are presented.

* [email protected]

114

I.M. Ulanov and M.V. Isupov The electrophysical and spectral characteristics of TCTD were studied in vapors of mercury and neon. The electrodeless sources of visible and UV radiation with the power of 100 W ÷ 100 kW were developed on the basis of this research.

INTRODUCTION The toroidal plasma–confinement devices (TOKAMAKs) are one of the most common methods for generation of high–temperature plasma [1]. This method is based on gas heating by a powerful current pulse (several mega amperes), induced by a current pulse in the transformer primary winding. This method can be also used successfully for generation of low-temperature plasma both in pulse and continuous modes. In this case we do not need the system for magnetic confinement of plasma, and a gas discharge is stabilized by the walls of the gas-discharge chamber. The use of a closed ferromagnetic core connecting the inductor (the primary coil of transformer) and toroidal plasma coil has several essential advantages over other types of RF induction gas discharges: frequency of gas discharge generation is several orders lower, the phase shift between the current in inductor and applied voltage is lower, and the level of electromagnetic noises is reduced. Owing to this, the transformer coupled toroidal discharges can be used successfully for development of efficient electrodeless generators of lowtemperature plasma and gas-discharge light sources. The first experimental works related to development and construction of electrodeless gas-discharge devices based on the principle of transformer coupled toroidal discharge were carried out in the middle of 60s [2]. The gas-discharge ion laser operating on the principle of transformer coupled toroidal discharge is described in [2]. To get a discharge in argon, the radiation yield in 7 ion lines was 50 mW, and the setup efficiency was about 0.1%. On the ground of these results the author has made a conclusion about the promising character of gas-discharge lasers, operating by the principle of transformer coupled toroidal discharge. The most important of the early works, dealt with investigations of TCTD, are the works of H.U. Eckert [3, 4]. The possibility of low-temperature plasma generation by means of TCTD in the low-frequency radio range is studied theoretically and experimentally in [3]. In particularly, possible development of the induction atmospheric-pressure plasmatron, operating at current frequencies of 60 or 180 Hz, which does not require application of RF power supplies, is considered there. According to the estimates of H. U. Eckert [3], the transformer plasmatron of atmospheric pressure, operating on argon at current frequency of 60 Hz, requires a core of 60 000 cm2 cross-section; and at this the discharge power will be 1.76 MW. With a rise of current frequency of up to 180 Hz, the core cross-section decreases to 7000 cm2, and power consumption reduces to 590 kW. H. U. Eckert has studied experimentally the transformer coupled toroidal discharge at current frequencies of 60 and 9600 Hz [3, 4]. At current frequency of 9600 Hz, the power of transformer coupled toroidal discharge was ~ 12 kW, at argon pressure of ~ 53 kPa. At a higher pressure the discharge became unstable and died out. At current frequency of 60 Hz, the power of TCTD was 0.5÷1.9 kW, at argon pressure of 0.6÷3 torr.

Induction Transformer Coupled Discharges: Investigation and Application

115

Measurements of electrical and thermal characteristics of the low-frequency (8 kHz) transformer coupled toroidal discharge in argon and air at pressures of 0.01÷32 kPa and powers of 3÷30 kW are presented in [5]. Due to development of current-convective instability of the toroidal plasma coil, V.M. Goldfarb et al. also failed to achieve the atmospheric pressure of plasma gas. For the first time the transformer coupled toroidal discharge of the atmospheric pressure was obtained by I. M. Ulanov and V. A. Kogan [6]. Application of vortex stabilization allowed the authors to get firstly a stable stationary discharge of the transformer type under the atmospheric pressure of argon and air, current frequency of 10 kHz and discharge power of up to 200 kW. Possible use of the transformer plasmatron for direct plasma-chemical synthesis of NO from air was also shown [7]. The super-equilibrium yield of NO (~7%) was reached without application of quenching methods. The transformer coupled toroidal discharge of the atmospheric pressure in argon and in the mixture of argon with nitrogen and oxygen was also studied experimentally in [8] at the discharge power of up to 8 kW. The problems of mathematical modeling of low-frequency transformer coupled toroidal discharges in argon are considered in [9, 10]. Also, a simple model that describes the temporal and spatial characteristics of the voltage and current in conductive toroidal chambers is given in [11]. Since the transformer coupled toroidal discharge allows generation of super-pure plasma even of aggressive substances and compounds, it can be used for development of various devices for etching semiconducting materials, film deposition, surface modification, etc. There are the patents [12, 13], which describe the gas-discharge devices of the transformer type for “etching” the semiconducting materials. The experiments on deposition of titanium nitride films with application of the transformer coupled toroidal discharge are known [14]. Also the transformer coupled toroidal discharges can be successfully used for creation of electrodeless light sources with a long service life. The first studies in this field were carried out by J. M. Anderson [15, 16]. The electrodeless luminescent lamp of the low pressure is described in [15] and patent [16]. However, the efficiency of this device was low because of high losses in ferrites (10÷15 W). The construction of high-intensive gas-discharge lamp on the basis of TCTD is also suggested in patent [17]. The possibility to make the efficient lamps of the transformer type appeared only after development of high-quality ferrites with low losses. The OSRAM Company has developed the electrodeless luminescent lamp ENDURA [18]. The parameters of this lamp were studied experimentally and theoretically in [19]. The authors of this chapter have firstly studied TCTD in the mixture of mercury and sulfur [20] and in pure neon [21, 22]; for the first time TCTD was studied in detail in mercury vapors under different conditions of discharge glowing [23]. As it follows from the above brief review of publications, the transformer–coupled toroidal discharges can be successfully used by various practical applications and for development of electrodeless gas-discharge devices with a long service life and high efficiency.

116

I.M. Ulanov and M.V. Isupov

1. TRANSFORMER COUPLED TOROIDAL DISCHARGES: MAIN PRINCIPLES 1.1. Principle of Operation of Transformer Coupled Toroidal Discharges The principle scheme of generation of the low-frequency transformer coupled toroidal discharge is shown in Fig. 1.

Figure 1. Principle of operation of transformer coupled toroidal discharge.1. Gas–discharge chamber.2. Ferrite core.3. Primary winding.4. 4.Power supply.

The induction discharge of the transformer type is a closed toroidal gas discharge 1, inductively connected to primary coil 3 via ferromagnetic core 2. In such a discharge plasma is heated at a release of Joule heat of closed currents. The closed currents in the toroidal induction discharge of the transformer type are caused by vortex electric field E, excited by alternating magnetic flux in the magnetic core:

Φ (t ) = Φ meiωt

(1)

and alternating magnetic flux, induced by the currents in plasma:

Φ i (t ) = ∫∫ B(t )dS

(2)

where ω is circular frequency of current, B is magnetic induction. In this case circulation of the electric field and current in the discharge chamber can be described by equations, where the bias current can be neglected:

d

∫ E (t )dl = − dt (Φ(t ) + Φi (t ))

L

rot (B ) = μ0 j

(3)

Induction Transformer Coupled Discharges: Investigation and Application

117

Neglecting active resistance of the primary coil, we will obtain that amplitude Фm does not depend on the discharge current, whose demagnetizing action is compensated by an increase in the current strength in the primary coil. To estimate a contribution of component Ei, excited by alternating magnetic flux Фi, into the discharge current, let’s introduce a current plasma coil as a closed ring of L=2πRk perimeter. Maximal induction of magnetic field of this current ring occurs in its center and equals Bk = πμ0

I . Considering ring area S=L2/(4π), L

we obtain the value of magnetic flux Фi=μ0·I·L/4. Respectively, the ratio of electric field components caused by the alternating magnetic flux in magnetic core Ф, and magnetic flux, induced by current Фi, equals [9]:

⎛ ωΦ i ⎞ ⎜ ⎟ ⎛ I ⎞ μ IL πμ I Ei ⎝ L ⎠ ⎟ = = 0 ≈ 0 ≈ 4 ⋅ 10− 6 ⎜⎜ Bm L ⎟⎠ E ⎛ ωΦ m ⎞ 4 Bm S Bm L ⎝ ⎜ ⎟ ⎝ L ⎠

(4)

The real sizes of plasma ring L varied from 0.2 to 2 m at the discharge current from 1 to 260 А and Bm from 0.1 to 1 T. Correspondingly, we obtain: Ei/E<10-2. Therefore, within the studied parametrical range of the low-temperature transformer coupled toroidal discharge, circulation of the electric field differs from zero only because of the magnetic flux changing with time in the magnetic core:

dΦ (t )

∫ E (t )dl = − dt L

(5)

According to the above estimates, we can conclude that in plasma of the induction discharge of the transformer type rot(E) = 0, E = - grad φ. With consideration of the Ohm law, from the continuous equation for current density J, we obtain:

∇σ∇ϕ = 0

(6)

To set the boundary conditions for (6), let’s take conditional surface Σ, separating a closed current coil into two parts. In this case, the boundary conditions on the Σ – surface, which reduce the doubly connected domain to the simply connected one, will be

ϕ (Σ −0 ) = 0...ϕ (Σ + 0 ) = iωΦ m i.e., the potential jump is determined by the rate of magnetic flux alteration by formula (5). According to (5), the discharge voltage and electric field strength equal: U (t ) = −iωBm Seiωt = −iωμμ0 H m Seiωt (7) E=U/L

(8)

118

I.M. Ulanov and M.V. Isupov

where Bm is the amplitude value of magnetic induction, Hm is the amplitude value of magnetic field strength, μ is magnetic conductivity of ferrite, and S is magnetic core cross-section. Then, in this chapter, U, E, I, etc. mean the mean-root-square values of voltage, field strength, discharge current, and etc. It should be noted specially that in the framework of the considered model, the model of the transformer discharge is similar to the model of electric arc with electrodes, combined into surface Σ [10]. This gives us the ground to use the term “transformer arc” and describe the transformer discharge by the same models as the other arc discharges. Since µ>>1 for ferromagnetic, electric field strength E, required for TCTD glowing, can be reached in a low-frequency discharge. In fact, the TCTD is the most low-frequent of all electrodeless discharges [3, 4]. The minimal value of current frequency, sufficient for induction discharge glowing, is determined by the electric field strength in discharge E. The main criteria for maintenance of TCTD glowing were considered firstly in the paper of Eckert [3], with regard to a particular quasistationary discharge in argon. General relationships can be also used for analysis of criteria of TCTD generation in other plasma-forming media.

1.2. Optimization of the Transformer Coupled Toroidal Discharge The correct choice of ferromagnetic material allows minimization of losses in the magnetic core and significant reduction of core sizes for the transformer gas-discharge device. According to expression 7, the required voltage of discharge glow U can be achieved by an increase in the core cross-section S or in current frequency f or in induction Bm. It is obvious that a significant increase in the core cross-section is unacceptable from the practical point of view. Thus, to achieve the required voltage, it is desirable to increase maximally the value of f·Bm, what can be only done with application of special kinds of magnetic materials. Two groups of materials with minimal volumetric losses at high frequency and induction of the magnetic field can be distinguished: amorphous soft magnetic alloys and special kinds of ferrites. Flat-strip amorphous soft magnetic alloys are characterized by high permeance, low coercitivity (Hc below 1 A/м), high specific electric resistance (ρ ~ 108 Ohm·cm), low losses for hysteresis and eddy currents (3÷5 times less than the best crystalline alloys). Since they are the semiconductors, ferrites have high specific volumetric electric resistance, exceeding the resistance of steels and alloys by factor of 50 and more. This makes it possible to use ferrites for high frequencies without a significant rise of eddy-current losses. Among disadvantages of ferrites there is relatively strong dependence of their magnetic properties on temperature. In contrast to ferrites, amorphous alloys can operate efficiently up to ~ 250–300 ºC. In case of low operation temperatures of the magnetic core, which allow application of both ferrites and amorphous alloys, the choice of the magnetic material will be determined from the condition of maximal Bm at the required current frequency and given threshold level of specific heat losses Pv (W/cm3).

Induction Transformer Coupled Discharges: Investigation and Application

119

The value of losses in the magnetic core is composed of separate components: eddycurrent losses, hysteresis, and magnetic viscosity (magnetic aftereffect). On the ground of statistical processing of abundant experimental data, the reference literature on electric engineering provides the following dependence of specific losses on frequency and induction of the magnetic field: β α PV ⎛ f ⎞ ⎛ Bm ⎞ ⎟ =⎜ ⎟ ⎜ (9) P* ⎜⎝ f* ⎟⎠ ⎜⎝ Bm* ⎟⎠ where f* = 1 kHz, Bm* = 1 T are the basic values of frequency and induction; P*, α, β are coefficients, obtained by processing of experimental dependences Pv(f, Bm). The typical dependence of specific heat losses PV on the amplitude value of magnetic induction Bm at different current frequencies is shown in Fig. 2 for the strip magnetic cores of amorphous alloy and ferrites of the Philips 3C96 grade. According to the figure, with a rise of Bm heat losses in ferrites increase faster than heat losses in amorphous steel because of lower saturation induction. However, at fixed induction of magnetic field, with a rise of frequency heat losses in amorphous alloys increase faster than in ferrites (Fig. 2). The above differences between ferrites and amorphous alloys are visually demonstrated in Fig. 3. Dependence of f·Bm on frequency f is shown there for the fixed level of specific heat losses in the magnetic core (0.5 W/cm3), for amorphous alloy and different kinds of ferrites.

Specific loss Pv, W/cm

3

Philips 3C96, 25 kHz Philips 3C96, 200 kHz Steel 5BD,25 kHz Steel 5BD,250 kHz 1

0,1

0,01 0,01

0,1

Bm, T Figure 2. Dependence of specific heat losses in the core PV on magnetic induction Bm.

1

120

I.M. Ulanov and M.V. Isupov

4

4x10

4

f·Bm, Hz·T

5x10

3x10

4

2x10

4

5BD 3С96 3С94 3С90 3F3 3F4 3F35

4

1x10

0 10

100

1000

f, kHz Figure 3. Dependence of f·Bm on current frequency f at the fixed level of specific heat losses PV =0.5 W/cm3.

According to Fig. 3, there is the optimal value of current frequency, corresponding to the maximal value of f·Bm at a constant level of heat losses, and the maximum achievable value of f·Bm depends essentially on the quality of magnetic material. In the range of low current frequencies (below 20÷30 kHz) the magnetic cores on the basis of amorphous alloys have a definite advantage over the best grades of ferrites because of a weak dependence of specific heat losses on magnetic field induction. At current frequencies above 100 kHz application of ferrites seems to be the most optimal.

1.3. Matching the Power Source and the Transformer Coupled Toroidal Discharge The transformer plasmatron is similar to inductive thermoelectric devices (induction furnaces). Hence, to match the power source with the transformer plasmatron, we can use the known calculation schemes, applied for devices of induction heating. In these devices the oscillating circuit, formed by an inductor and additional capacitor bank, serves as the load. The equivalent scheme of the power source and transformer plasmatron is shown in Fig. 4a. Here Rg and Xg are active resistance and reactance of the generator, Rtr and Xtr are active resistance and reactance of the transformer, XC1 and XC2 are capacitances for matching. Under the steady state it is convenient to present the oscillating circuit, formed by the transformer and matching vessels, as a parallel equivalent circuit, including equivalent active resistance Req and reactance Xeq (Fig. 4b). At this, Req and Xeq are determined by the following formulas [6]: Rtr ⋅ BC21 Req = (B12 + X tr ⋅ BC1 ⋅ BC 2 )2 + (Rtr ⋅ BC1 ⋅ BC 2 )2

Induction Transformer Coupled Discharges: Investigation and Application

121

( B12 + X tr ⋅ BC1 ⋅ BC 2 ) ⋅ ( X tr ⋅ BC1 − 1) + (Rtr ⋅ BC1 ⋅ BC 2 )2 X eq = (B12 + X tr ⋅ BC1 ⋅ BC 2 )2 + (Rtr ⋅ BC1 ⋅ BC 2 )2 B12 = BC1 − BC 2 X eq tgϕ = Req Z eq =

(10)

Req 1 + (tgϕ )2

where BC1, BC2 are capacitive susceptances; ϕ is phase angle, Zeq is total resistance of the circuit. The value of Zeq should correspond to the inner resistance of generator, and tgϕ of the load should fit the ratings of generator to provide the transfer of maximal active power to the transformer plasmatron.

a) ~ Ig

Rg

b)

X g X C1 X tr R tr

X C2

X eq

R eq

Figure 4. Equivalent schemes:a) Transformer plasmatron; b) Parallel equivalent circuit.

Thus, calculations are reduced to determination of Req and Xeq and, hence, XC1 and XC2 for every specific operating condition (idle running – before discharge ignition; operating condition – for specific gas and pressure). Besides, it is necessary to determine Rtr, and Xtr (active resistance and inductance of the transformer plasmatron). The equivalent circuit of the plasmatron, required for determination of Rtr and Xtr, is the equivalent circuit of the conventional transformer with active-inductive loading. This circuit is widely spread and described in electric engineering. In conclusion it should be noted once more: various electrodeless gas-discharge devices of the high and low pressure can be developed on the basis of transformer coupled toroidal discharges for efficient plasma generation with a high density of charged particles of any chemical composition. Therefore, the transformer coupled toroidal discharges can be used successfully by many practical applications, which require the use of low-temperature plasma.

122

I.M. Ulanov and M.V. Isupov

2. TRANSFORMER PLASMATRONS 2.1. Experimental Setup Initially experiments on generation and investigation of transformer coupled toroidal discharges in inert and molecular gases were carried out at setup No. 1, shown in Fig. 5a. The picture of this setup is shown in Fig. 5b.

Figure 5a. Experimental setup No. 1. 1 – primary winding; 2 – sections of magnetic core; 3 – discharge chamber; 4 – heat exchanger; 5 –main gas feed; 6 – secondary gas feed; 7 – probe No.1; 8 – probe No.2; 9 – probe No.3; 10 – spectrophotometer; 11 – mass-spectrometer.

Figure 5b. Transformer plasmatron for current frequency of 10 kHz.

The total power of setup No.1 was 180 kW. The magnetic core consisted of 8 separate sections made of cold-rolled electric sheet steel 3425 of the 80-µm thickness. The outer diameter of each section was 420 mm, the inner diameter was 160 mm, and the height was 70 mm. Each section had the own system of primary coils. The number of primary coils is 4. All primary coils were connected either in-parallel or in-series to the machine generator (10 kHz) depending on the experimental conditions. The discharge chamber was of the closed

Induction Transformer Coupled Discharges: Investigation and Application

123

“toroidal” shape. The perimeter of discharge chamber along the mean line was L=1.8 m. The chamber consisted of the metal, water-cooled, mutually insulated sections with the inner diameter Dint=(80÷100) mm and length (20÷60) mm. The plasma gas was fed to the discharge chamber through two tangential holes with the diameter of 6 mm and the swirl angle of 20°. The gas flow was controlled by rotameters. Experiments on thermal synthesis of ozone from oxygen and natural gas conversion in the presence of СО2 were carried out in the mixture with argon. For these purposes two experimental methods were used. By the first method a mixture with the known percentage of the studied gas in argon was fed into the main feed chamber. By the second method the feed of argon and required reagents was separated. Oxygen, natural gas and carbon dioxide were fed through additional vortex chamber 6, located at distance L=60 cm from the point of the main argon supply. This allowed operation in mixtures of the studied gas and argon within a small region of the discharge chamber at relatively high concentration of products in discharge. Total induced discharge voltage Udisch was measured by an additional insulated coil of wire, enveloping the magnetic core. The mean mass temperature of gas at discharge chamber outlet was measured by calorimetric study of heat exchanger 4. Distribution of the mean mass temperature of gas in the discharge chamber was measured by calorimetric (enthalpy) probe [24], whose scheme is shown in Fig. 6. The gas temperature at the outlet as well as the water temperature at the probe inlet and outlet was determined by a chromel-coppel thermocouple. The gas flow through the probe was determined by accumulation in a diaphragm evacuated volume. During the experiment the volume was evacuated constantly, and the pressure drop on the diaphragm was supersonic always. When measuring the gas flow, buffer volume evacuation was stopped, and the flow rate was determined by the time of pressure alteration in this volume. The time range was chosen to obtain a constant supersonic drop on the diaphragm, corresponding to a constant flow rate. A relative error of mean mass temperature measurements by this method depends on the ratio of the inlet velocity of the sampling gas to the velocity of the incident flow and makes up from 3% to 17%.

Figure 6. The scheme of calorimetric probe with the system of sampling and analysis

124

I.M. Ulanov and M.V. Isupov

Radial distributions of mean mass temperature of gas in the discharge chamber were measured for two distances from the point of plasma gas supply: l=60 mm (probe No.1); and l=800 mm (probe No.2). A gas sample was analyzed by spectrophotometer CF–26 and mass-spectrometer Quadrovak. Preliminary, the mass-spectrometer and spectrophotometer were calibrated by the gas mixtures with given concentrations of substances. This method of experiments allowed determination of radial distributions of product concentrations in the discharge chamber together with measurement of temperature distributions. Both tubular probes and probes with rectangular cross-section were used in the experiments. The typical rate of high-temperature gas cooling was estimated by the formulas, suggested in [25]:

(

w ⋅ T f − Tw dT = 0.107 ⋅ dt d ⋅ Re0.2

)

(11)

(where w is average gas velocity, Tf, Tw are mean mass temperatures of the flow and probe wall, d is efficient diameter of the gas-sampling channel); for initial temperature Tw 3000÷6000 K this rate was about 106÷107 К/s for the probe with d = 1.5 mm. Samples for analysis of product concentration at plasmatron gas outlet were taken by probe No.3. Probe No.З was located near the heat exchanger (Fig. 5a). The heat exchanger consisted of the water-cooled tubes with diameters of 8 and 15 cm and lengths of 1 and 1.5 m, correspondingly. According to the estimate by formula (11), the maximal average rate of gas cooling in this heat exchanger is dT/dt≈103÷104 К/s. A disadvantage of the transformer plasmatron (рис. 5b) was the fact that magnetic cores were made of ordinary transformer steel 3425 and the machine generator with current frequency of 10 kHz was used as the power source. Therefore, the magnetic cores of the transformer plasmatron were heavy (~300 kg), and the plasmatron itself was too bulky. Now it is possible to make more compact transformer plasmatrons because of development of new high-quality ferrites for power electronics and design of compact and cheap power sources with current frequency of ~100÷400 kHz [26]. The further works were aimed at reduction of transformer plasmatron sizes with preservation of high productivity. The works performed allowed development of a compact transformer plasmatron with the power of 20 kW and current frequency of 100 kHz (Fig. 7a, 7b). The magnetic cores of plasmatron 1 were assembled of EPCOS ferrite bars of N 87 grade (the analog of ferrite 3F3, Fig. 3). The total cross-section of magnetic core was 100 cm2, and the mass was ~ 20 kg. Water-cooled gas-discharge chamber 2 was made of 12 sections, mutually insulated by silastic gaskets 4. The inner diameter of gas discharge chamber was D = 45 mm and the gas discharge perimeter along the mean line was L = 1 m. To stabilize a plasma discharge, the transformer plasmatron was mounted vertically with argon input via vortex chamber 3. Plasma went out from the opposite side to the chamber of chemical reactor 6. To perform various plasma-chemical reactions, the inlet for chemical reagents 5 was made in the lower part of the discharge chamber, and branch pipes were provided below for the input of additional chemical reagents and plasma observations through a window. The

Induction Transformer Coupled Discharges: Investigation and Application

125

chamber of chemical reactor can be used both for accumulation of reaction products and allocation of items for plasma processing.

Figure 7a. The scheme of transformer plasmatron with ferrite cores.1 – ferrite cores; 2 – water-cooled chamber; 3 – vortex tangential input of argon; 4 – silastic gaskets; 5 – reagent input; 6 – outlet to chemical reactor.

Figure 7b. Transformer plasmatron with ferrite cores of the 20-kW power.

126

I.M. Ulanov and M.V. Isupov

To ignite a discharge, air was evacuated from the plasmatron and chamber to the pressure of about 10–20 Pa. After powering-on the system, argon was fed into the discharge chamber, and discharge was ignited. When the atmospheric pressure was reached, a damper was open to let argon into the atmosphere. The voltage of discharge glow was measured by an additional coil, enveloping magnetic cores 1; the discharge current was measured by the current transformer. The thermophysical characteristics of discharge were determined by calorimetric measurement of water-cooled chamber 2. The main achievement of the chapter authors should be marked out specially: for the first time the stable transformer coupled toroidal discharge of atmospheric pressure was obtained in argon, air, and argon mixtures with molecular gases (oxygen, hydrogen, carbon dioxide, and natural gas).

2.2. Electric-Physical and Thermal-Physical Characteristics of Transformer Plasmatrons The electric-physical and thermal-physical characteristics of transformer coupled toroidal discharges were studied in argon, air, and mixtures of argon with natural gas and carbon dioxide at atmospheric and reduced pressures. Experiments were carried out at setup No.1. The effect of “glow” conditions on the electric field strength of a discharge in argon was studied within the range of pressures of 10÷105 Pa; currents of 90÷450 А; and argon flow rate of 1÷30 g/s. The thermal-physical and electric-physical characteristics of the transformer coupled toroidal discharge were also studied at atmospheric pressure in the mixture of argon and hydrogen in the transformer plasmatron with ferrite cores (Fig. 7а). The effect of argon flow rate on the electric field strength in a discharge at atmospheric pressure is shown in Fig. 8.

Electric field strength, V/cm

1,5 1,4 1,3 1,2 1,1 1,0 0,9 0,8

0

5

10 15 20 Ar flow rate, g/s.

25

Figure 8. Electric field strength vs. argon flow rate.I=200 А, p=1 atm.

30

Induction Transformer Coupled Discharges: Investigation and Application

127

Electric field strength, V/cm

1,2

1,1

1,0

0,9

0,8 50

100 150 200 250 300 350 400 450 TCTD current, A

Figure 9. Current-voltage characteristic of transformer coupled toroidal discharge in argon. Argon flow rate is 4 g/s.

The initial incident region of the curve corresponds to unsteady operation, when gasvortex arc stabilization is insignificant because of a low gas flow rate. Therefore, a discharge is spatially unstable (because of development of current-convective instability), what increases heat transfer between the arc and discharge chamber wall, and, hence, this leads to a rise of applied power required for stationary discharge maintenance. At a given current this increases the electric field strength. The second constant region of the curve corresponds to stable arc “glow”. A following increase in the electric field strength relates to an increase in a degree of flow turbulence and increase in heat losses to the plasmatron wall as well as to a rise of energy, “carried out” by a gas flow. In general, this increases the power required for discharge maintenance. The current-voltage characteristic of a discharge in argon was studied for TCTD currents of 90÷450 А. The typical current-voltage characteristic of the TCTD is shown in Fig. 9 for argon flow rate G=4 g/s. According to the diagram, at the initial stage (at low currents) the current-voltage characteristic is descending. Then, after a small region of constant Е(I) there is an ascending part of the curve. This dependence E(I) is also characteristic for the DC arc discharges. Dependence of the electric field strength on the argon pressure is shown in Fig. 10. The diagram represents the experimental results both with and without vortex gas supply. According to this diagram, at low pressure (diffusion discharge) the electric field strength decreases with a rise of pressure. This is caused by the fact that the free path of electrons decreases with a rise of pressure, and this provides less electron losses to the wall of discharge chamber; on the other hand, electron concentration increases. This reduces the electric field strength in a discharge at a given current density. With the following rise of pressure the discharge contraction occurs, and current-convective instability increases. A discharge takes a wavy shape, the arc length increase, and heat transfer between the arc and the wall becomes more intensive. This increases the electric field strength, required for discharge maintenance. With the following rise of pressure considerable voltage pulsations occur in a discharge and the instantaneous voltage can exceed the maximal voltage, induced by the magnetic core.

128

I.M. Ulanov and M.V. Isupov

If the vortex supply of plasma gas is used, current-convective instability of discharge is expressed weakly. This allows a pressure increase in the discharge chamber of up to the atmospheric one and higher without significant fluctuations of the discharge voltage. 1,2

Electric field, V/cm

1,0 0,8 0,6 0,4 0,2 0,0

2

3

10

4

5

10 10 Argon pressure, Pa

10

Figure 10. Electric field strength vs. argon pressure. ■ – Without vortex supply of argon (I=100 A). + – With vortex supply of argon (I=100 A).G=3÷5 l/s 2

8

Electric field, V/cm

7 6 5 4

1

3 2 0

2

4

6 8 10 12 Air flow rate, g/s.

14

16

18

Figure 11. Average electric field strength E vs. air flow rate. Discharge current I=125 А. 1 – E at initial arc region in electric-arc plasmatron; 2 – E at developed turbulent region in electric-arc plasmatron.

Dependence of the electric field strength in air flow rate (Fig. 11) has been studied. The behavior of E=f(G) function in air is similar to the experiments with argon. It is obvious that the electric field strength in air is 4÷6 times higher than in argon. In the same diagram empirical dependences of electric field strength on air flow rate are shown at the initial region (curve No.1) and at the developed turbulent region (curve No. 2); they were derived in [27] at processing of data on the electric field strength in electric-arc plasmatrons.

Induction Transformer Coupled Discharges: Investigation and Application

129

Since there are no electrodes we could work with aggressive gas mixtures both at a reduced and atmospheric pressures. There was no any destruction of discharge chamber walls, and plasma was not contaminated by admixtures. Dependence of electric field strength in [Аг + O2] mixture on oxygen content in this mixture is shown in Fig. 12. According to the diagram, even at low oxygen concentrations in the mixture the electric field strength increases significantly. For instance, at oxygen content in argon of ~ 1 % the electric field strength increases almost twice. Dependences of electric field strength in [Аг + СО2], [Аr + СН4] and [Ar + H2O] mixtures on their percentage in argon are shown in Fig. 13. It is clear that for these gases the electric field strength is relatively high, therefore, a more powerful and efficient setup with a perfect magnetic core and power source of increased frequency (~100 kHz) is required for the work with such pure gases. 12

Electric field, V/cm

10 8 6 4 2 0

10

20 30 40 O2 concentration, vol. %

50

Electric field, V/cm

Figure 12. Electric field strength in [Ar + O2] mixture vs. oxygen content. Discharge current I=125 А, argon flow rate G=4 g/s. 12 11 10 9 8 7 6 5 4 3 2 1 0

H 2O

CO2

CH4

0

10

20

30

40

50

60

Concentration, vol. %

Figure 13. Electric field strength in [Ar + H2O, Ar + CO2, Ar + CH4] mixtures vs. substance concentration in argon. I=125 А, argon flow rate G=4 g/s

130

I.M. Ulanov and M.V. Isupov

Dependence of argon plasma temperature at the atmospheric pressure on argon flow rate is shown in Fig. 14 for the distance from the axis of discharge chamber r/R=0.3 and discharge current I=125 А. According to the diagram, with a rise of argon flow rate the hightemperature zone is localized in the center of discharge chamber, i.e., with a rise of the flow rate the mean mass temperature of gas decreases within the given radius at the same discharge current. Temperature distribution in the discharge chamber is shown in Fig. 15 for two distances from the point of gas supply and different air flow rates. According to the diagrams, temperature distributions are subjected to two main factors: the flow rate of plasma gas and the distance from the point of gas supply. A rise of the gas flow localizes the hightemperature zone in the center of discharge chamber, and the temperature gradient becomes higher with a rise of the flow rate. With an increase in the distance from the point of gas supply the temperature profile becomes smoother. 5000 4500

Temperature, K

4000 3500 3000 2500 2000 1500 1000 500 1,0

1,5

2,0

2,5

3,0

3,5

4,0

Ar flow rate, g/s.

Figure 14. Temperature of argon plasma at atmospheric pressure vs. argon flow rate for r/R=0.3, z/R=1 (probe No.1). Discharge current I=125 A 7000 6000 Temperature, K

5000 4000 3000 2000 1000 0

0

10

20

30

40

50

Distance from discharge axis, mm

Figure 15a. Temperature distributions in the discharge chamber for two distances from the point of air input into the chamber: I=125 A. L =60 mm (probe No.1).• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s

Induction Transformer Coupled Discharges: Investigation and Application

131

6000

Temperature, K

5000 4000 3000 2000 1000 0

5

10 15 20 25 30 35 40 45 50 55 Distance from discharge axis, mm

Figure 15b. Temperature distributions in discharge chamber for two points of air input into the chamber: I=125 A.L =800 mm (probe No.2).• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s

Power, kW

20

Discharge power Plasma Jet power Heat Losses

efficiency, %

40

15

30

10

20

5

10

0 0,0

0,1

0,2 0,3 H2 flow rate, g/s.

0 0,4

Figure 16. Thermal-physical characteristics of transformer plasmatron with ferrite cores. I=65 А, GAr=4 g/s.■ – discharge power;• – power in plasma jet, flowing into plasmachemical reactor;▲ – heat flux to water-cooled wall of plasmatron;♦ – plasmatron efficiency.

Then, the results of experimental investigations have been used for development of a compact transformer plasmatron with ferrite cores (рис. 7а, b) for plasmachemical reactions in a plasma jet [Ar+H2] leaving the plasmatron. Hydrogen was fed to the lower section of the gas-discharge chamber. Dependence of plasmatron efficiency, heat losses and power in the leaving plasma jet on hydrogen flow rate is shown in Fig. 16. According to the diagram, the maximal plasmatron efficiency is ~ 35%, and it can be achieved at hydrogen flow of 0.1 g/s. The following decrease in plasmatron efficiency is explained by a rise of gas flow turbulence and enlargement of heat losses to the plasmatron walls (Fig. 16).

132

I.M. Ulanov and M.V. Isupov

Dependence of the power of argon-hydrogen plasma jet on hydrogen flow rate was determined as the difference between discharge power and heat losses. According to the diagram, the maximal power was obtained at hydrogen flow rate of about 0.1 g/s, with the following increase in the flow rate the jet power decreased slightly. Analysis of thermal-physical and electric-physical characteristics of the TCTD allows us to make a conclusion that characteristics of the transformer coupled toroidal discharge are similar to characteristics of a positive DC arc column. It is shown that without vortex stabilization of a discharge current-convective discharge instability starts developing from the pressures in discharge chamber of 104 Pa, what leads to significant voltage fluctuations, providing TCTD “extinction”. If there is the vortex gas flow in the discharge chamber, the described instabilities do not develop, what allows a stable TCTD at the pressures above the atmospheric one.

2.3. Thermal Production of Ozone in Plasma of Transformer Coupled

Toroidal Discharge By now the most common plasma chemical method of ozone production is its synthesis in a barrier discharge. Ozone is produced directly in the zone of a discharge in one stage. Another plasmachemical method of ozone production is the thermal method. In this case ozone is produced in two stages. The first stage is oxygen “heating” up to the temperatures of dissociation. The second stage is the process of fast cooling of dissociated oxygen. In this process ozone is synthesized at the stage of non-equilibrium cooling of gas, consisting of oxygen atoms, molecules and radicals. Let’s consider the main processes occurring at thermal production of ozone. At the first high-temperature stage ozone concentration under the conditions of thermodynamic equilibrium is negligibly small, and ozone production by the considered method is possible only at the stage of non-equilibrium gas cooling, when ozone is formed by recombination of atomic and molecular oxygen. Let’s estimate the typical rate of cooling of dissociated oxygen (at the temperatures of 3500÷4500 К), when atomic oxygen turns mainly into ozone, but not into molecular oxygen: 1. Transition of O–atoms into molecular oxygen and ozone at cooling of dissociated gas occurs by the following main reactions: (R1) O+O2+M Æ K1 O3 + M (R2) O+O3 Æ K2 O2 + O2

(12)

(R3) O+O+M Æ K3 O2 + M 2. A change in the density of atoms, radicals and molecular oxygen in the considered processes equals:

Induction Transformer Coupled Discharges: Investigation and Application

133

d [O ] = − K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] − K 2 ⋅ [O ] ⋅ [O3 ] − K 3 ⋅ [O ]2 ⋅ [M ] dt

(13)

d [O3 ] = K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] − K 2 ⋅ [O ] ⋅ [O3 ] dt

(14)

d [O2 ] = − K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] + K 2 ⋅ [O ] ⋅ [O3 ] + K 3 ⋅ [O ]2 ⋅ [M ] dt

(15)

Thus,

d ([O ] + [O3 ]) = −2 K 2 ⋅ [O ] ⋅ [O3 ] − K 3 ⋅ [O ]2 ⋅ [M ] dt

(16)

3. A relative change in concentration of oxygen atoms and radicals during the processes of cooling equals:

1 d ([O ] + [O3 ]) K 2 ⋅ [O ] ⋅ [O3 ] K 3 ⋅ [O ]2 ⋅ [M ] 2 dt = − dt − ∫ ([O ] + [O ]) ∫ ([O ] + [O ]) ∫ ([O ] + [O ]) dt dt 3 3 3

(17)

The expression in the left part is a relative portion of atomic oxygen, converted into molecular oxygen at cooling [28]. Let’s determine the condition, when it is assumed that concentration of molecular oxygen at quenching is not changed, i.e., this is the condition, when almost all atomic oxygen turns to ozone. With this assumption we can consider that the left part of equation (17) is a small value. Considering the rate of gas mixture cooling constant (dT/dt = const), we obtain the following condition:

⎡ K 3 ⋅ [O ]2 ⋅ [M ] ⎤ dT K 2 ⋅ [O ] ⋅ [O3 ] dT ⎥ dT − ∫ >> ⎢− 2∫ ([O] + [O3 ]) ([O] + [O3 ]) ⎦⎥ dt ⎢⎣

(18)

Integrands (18) include equilibrium values of densities, determined by relationships:

[O] + [O3 ] ≈ C [O] = K (T ) [O ] [O ] [M ] 3

(19)

where С[O] is concentration of oxygen atoms at the initial temperature, K(T) is the constant of equilibrium between atomic oxygen and ozone. Integrals converge within the temperature area, where atomic oxygen turns into ozone. The sought temperature area is determined by the following condition: K(T) ~ 1, which is satisfied at T ~ 800 K. Integrals should be taken at the temperatures of 3500÷4500 K, when concentration of atomic oxygen is high. After substitution of values for T=4500 K, we obtain dT/dt>>108 K/s. It should be noted that this condition determines ozone formation in assumption of almost all transition of atomic oxygen into ozone, but not into molecular oxygen. In case of partial transition of О–atoms into ozone we can expect that formation of not-zero ozone concentration in the process of dissociated oxygen cooling is possible at dT/dt ~ 107÷108 K/s.

134

I.M. Ulanov and M.V. Isupov

Quenching with the given rates is easy to organize by the recuperative methods because the process of gas cooling in a heat exchanger is well studied, and this allows us to study experimentally the effect of the rate of “hot” oxygen cooling on kinetics of the considered reaction and final output of the product. Experiments on investigation of thermal synthesis of ozone from the mixture of argon and oxygen were carried out at setup No. 1. The power of this setup did not allowed operation with pure oxygen; therefore, the experiments were carried out in the mixture of argon and oxygen. Two methods of experiment implementation were used. In the first case, the mixture with given concentration of oxygen was fed into the discharge chamber through the main section of gas supply. In the second case argon was fed into the main section of gas supply, and oxygen was fed into the secondary section (see Fig. 6). All experiments were carried out for the discharge current I = 125A. To study experimentally the effect of dissociated oxygen quenching on ozone production, the water-cooled quenching probes of the slot type were made. These probes had the rectangular cross-sections of the gas sampling channel with the sizes of 0.2х10 and 0.5х10 mm. The geometrical scheme of this probe is shown in Fig. 17. The probes were located at the distance of 100 mm from the heat exchanger (Probe No. 2 – Fig. 6). The inlet sampling slot of the probe was at distance r/R=0.5. Visually this position of the probe corresponded to the boundary of the “glowing” conducting channel. Thus, it can be assumed that the initial temperature of gas was not lower than 4000 К (in these experiments the initial temperature of a sample was not measured).

Figure 17. Quenching probe of the slot type.

The probes were located at the distance of 100 mm from the heat exchanger (Probe No. 2 – Fig. 6). The inlet sampling slot of the probe was at distance r/R=0.5. Visually this position of the probe corresponded to the boundary of the “glowing” conducting channel. Thus, it can be assumed that the initial temperature of gas was not lower than 4000 К (in these experiments the initial temperature of a sample was not measured). Experimental results in Fig. 18 are shown for two probes, providing different rates of high-temperature gas quenching. In the diagram there are the values of volumetric ozone concentrations, which can be reached at cooling of oxygen-argon plasma in probes, depending on oxygen concentration in argon.

Induction Transformer Coupled Discharges: Investigation and Application

135

7

probe 0.2x10 mm, dT/dt~2*10 K/s. 7 probe 0.5x10 mm, dT/dt~10 K/s)

0,55 0,50 0,45 O3, Vol. %

0,40 0,35 0,30 0,25 0,20 0,15 2

4

6

8

10

12

14

O2, vol. %

Figure 18. Dependence of ozone concentration on oxygen concentration in initial mixture [Ar + O2].

According to the diagram, ozone concentration increases with a rise of gas cooling rate and with an increase in oxygen concentration in the initial mixture. At that it can be seen that ozone concentration depends non-linearly on oxygen concentration. In other words, initial concentration of atomic oxygen influences directly the kinetic processes at quenching. According to investigation of thermal synthesis of ozone in the transformer coupled toroidal discharge with the use of recuperation quenching devices, at cooling rates of ~107 K/s, from 5 to 10% of oxygen turn into ozone at oxygen content in mixture [Ar+O2] from 15 to 3%, respectively.

2.4. Synthesis of Nitrogen Monoxide in Plasma of the Transformer Coupled Toroidal Discharge One of the most examined gas-phase plasmachemical processes is synthesis of nitrogen monoxide from air. There are many publications dealt with kinetics and thermodynamics of nitrogen oxide formation in low-temperature plasma. The first detailed investigation of this plasmachemical process is the research of Ya.B. Zeldovich, P.Ya. Sadovnikov, and D.A. Frank-Kamenetsky [29], where it is shown that the bimolecular mechanism of nitrogen oxide formation and decay, used before this research, does not explain the measured times of reaction. The cited paper suggests the chain mechanism, which disposes the contradiction. This mechanism includes the following elementary reactions: 1) 2) 3) 4) 5)

O2 + M ÆK1 O + O + M…O + O + M ÆK1’ O2 + M O + N2 ÆK2 NO + N……..NO + N ÆK2’ O + N2 N + O2 ÆK3 NO + O……..NO + O ÆK3’ N + O2 N2 + M ÆK4 N + N + M….N + N + M ÆK4’ N2 + M NO + M ÆK5 N + O + M…N + O + M ÆK5’ NO + M

136

I.M. Ulanov and M.V. Isupov

In assumption of atomic oxygen stationarity (concentration of atomic oxygen is determined by equilibrium of reaction (1)), Zeldovich [29] has derived the following expression for the total rate of nitrogen monoxide formation:

⎛ 86000 ⎞ exp⎜ − ⎟ d ( NO ) RT ⎠ ⎝ = 1.5 ⋅ 1013 ⋅ ⋅ [NO ]2 − ( NO )2 mole/cm3/s dt (O2 )

{

}

(20)

where [NO] is equilibrium concentration at estimated temperature. On this basis the optimal rate of air quenching was determined in [30] within the temperature range of 3300÷1800 K; at this rate total decay of nitrogen monoxide equals 5% of the initial equilibrium value:

{

}

2 2 dT ⎛ 86000 ⎞ [NO ] − (NO ) ⋅ (T1 − T2 ) = 1.5 ⋅ 1013 ⋅ exp⎜ − ⋅ ⎟ dt RT ⎠ (O2 ) ⋅ 0.05 ⋅ (NO )1 ⎝

(21)

where T1 – initial temperature, T2 – final temperature, (NO)1 – initial concentration of NO. According to this expression, to get the maximal output of NO, it is necessary first to “set” the gas temperature, corresponding to the maximal equilibrium content of the product (T ~ 3400 K), then quenching should be carried out with the rate of ~5·106 К/s, with this rate reduction to 104 К/s (at the final temperature of ~ 1800 К). It should be noted that the described process of nitrogen oxide formation by cooling of quasi-equilibrium air within the temperature range of 1000÷4000 K is proved well by the experimental data [30]. If air is cooled from the higher temperatures (> 5000 K) there is divergence between experimental and calculation data: at air cooling from the temperatures above 5000 К, super-equilibrium concentration of nitrogen monoxide is formed (~6÷7 %) [31]. Together with a practical interest to NO production, abundance of experimental and theoretical works in this field also seemed very attractive. This allowed us to use this process as the ‘model’ one for understanding of the features of the transformer coupled toroidal discharge with regard to plasmachemical technologies. Experiments on investigation of NO synthesis from air were carried out at setup No. 1. The plasma-forming gas (air) was fed into the main vortex chamber. Temperature and concentration distributions over the discharge chamber were measured in these experiments at different distances from the point of air input into the chamber. The water-cooled tubular probes-samplers were used for this purpose (Fig. 6). Radial distributions of product temperature and concentration were studied in two different zones of the discharge chamber (Fig. 5a): Concentration of NO after the heat exchanger was also measured in these experiments (probe No.3). The mean mass temperature of gas at the discharge chamber outlet was measured via heat exchanger calorimetry. Distributions of NO concentration over the radius of discharge chamber, obtained by probes No.1 and No.2 are shown in Fig. 19(a,b) (temperature distributions under the same conditions are shown in Fig. 15(a,b)). It can be seen that near the point of air input (probe No.1) distribution of nitrogen monoxide concentration is similar to temperature distribution.

Induction Transformer Coupled Discharges: Investigation and Application

137

Far from the input point (probe No.2) distribution of nitrogen monoxide concentration does not depend on the gas temperature. It attracts attention that near the wall of discharge chamber, where the level of temperature is relatively low (Т ~ 300÷2000 К) NO concentrations, measured by probes No.1 and No.2, differ significantly. At high temperatures (Т>5000 К), NO concentrations, measured by probes No.1 and No.2 are close to each other. 8 7

NO, vol. %

6 5 4 3 2 1 0

0

10

20

30

40

50

Distance from discharge axis, mm

Figure 19а. Distribution of NO concentration over the chamber of transformer plasmatron, L =60 mm (probe No.1), I=125 A.Air flow rate.• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s

8 7

NO, vol. %

6 5 4 3 2 1 0

0

10 20 30 40 50 Distance from discharge axis, mm

Figure 19b. Distribution of NO concentration over the chamber of transformer plasmatron, L =800 mm (probe No.2). I=125 A.Air flow rate. • 3.7 g/s; ■ 6.8 g/s * 12.6 g/s

Values of NO concentration vs. the temperature of a gas sample are shown in Fig. 20(a,b) by circles and squares for probes No. 1 and 2, respectively. The solid line shows the temperature dependence of equilibrium concentrations of NO. In the same diagram the heavy

138

I.M. Ulanov and M.V. Isupov

line indicates calculated values of NO concentration, obtained in assumption of thermodynamic equilibrium of gas before quenching. 8 7

NO, vol. %

6 5 4 3 2 1 0

0

1000 2000 3000 4000 5000 6000 7000 Temperature, K

Figure 20a. NO concentration vs. air temperature.L =60 mm (probe No.1), I=125 A. Air flow rate:• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s 8 7

NO, vol. %

6 5 4 3 2 1 0

0

1000

2000

3000

4000

5000

6000

Temperature, K

Figure 20b. NO concentration vs. air temperature.L =800 mm (probe No.2). I=125 A. Air flow rate:• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s The equilibrium curve of NO content in air depending on T.

It can be seen in Fig. 20(a,b) that NO content, measured at low (Т ~ 300÷2000 К) and high temperatures (Т > 4000 К), is significantly higher than the corresponding equilibrium NO content in air. The difference between experimental and calculated concentrations of nitrogen monoxide, obtained at air cooling from the temperatures of above 5000 К, is the known fact. The explanations of this phenomenon were discussed actively in [32, 33]. According to one of these explanations [32], the additional production of NO occurs at quenching of the hightemperature gas (>4000÷5000 К) due to divergence of oscillating and translation temperatures at the initial moment of cooling. This phenomenon is not discussed in the

Induction Transformer Coupled Discharges: Investigation and Application

139

current work, but an attempt to explain a high content of NO in the peripheral gas at temperatures below 2000 К is emphasized. Dependence of NO concentrations on the mean mass temperature of gas at the outlet of a transformer plasmatron (probe No. 3) is shown in Fig. 21(a) by dots. The values of NO concentrations, obtained by the authors of [30, 34, 35] vs. the mean mass temperature of gas before the quenching devices are also shown there by the shaded area. It is necessary to note that in experiments of these authors NO was synthesized in two stages: in the first stage the air in plasmatron was heated up to some certain temperature, then high-temperature air was quenched (the cooling rate was ~ 106÷107 К/s). The calculated value of NO, reached at cooling of thermodynamically equilibrium air with experimentally determined initial temperature, is shown by the corresponding curve. 1

7 6

3

NO, vol. %

5 4 3 2 2

1 0

2000

3000 Tenperature, K

4000

Figure 21.a. Concentration of nitrogen monoxide at the heat exchanger outlet (probe No. 3) depending on the mean mass temperature of gas at the heat exchanger inlet.1 – experimental data (probe No. 3).2 – calculated value of NO concentration at the equilibrium state of air before cooling in heat exchanger;3 – the area of experimental data of different authors [30, 34, 35], obtained at the use of equilibrium gas quenching. 7,0

1

6,5 6,0 NO, vol. %

5,5 5,0 2

4,5 4,0 3,5 3,0 2,5

0

2

4

6

8

10

12

14

16

Air flow rate, g/s.

Figure 21.b. Concentration of nitrogen monoxide at the heat exchanger outlet (probe No. 3) depending on the air flow rate. 1 – experimental data (probe No. 3).2 – calculation by the model.

140

I.M. Ulanov and M.V. Isupov

Dependence of NO concentration on the air flow rate, measured at plasmatron outlet (probe No. 3), is shown in Fig. 21.b by dots. According to the diagram, at low flow rates (1÷5 g/sс) NO concentration increases almost linearly. With the following increase in the air flow, NO concentration depends on the flow rate weakly and tends to ~ 7%. Thus, a high (superequilibrium) content of nitrogen monoxide at the gas outlet from the plasmatron was determined experimentally at synthesis of nitrogen monoxide in the transformer coupled toroidal discharge at vortex arc stabilization without application of the quenching devices, which provide high rates of gas cooling. After measurements of plasma temperature and NO concentration profiles, it was determined that in the low-temperature peripheral zone of a vortex flow, which stabilizes the transformer discharge, the NO content exceeds significantly the thermodynamically equilibrium values. Let’s estimate the typical times of molecular diffusion under the considered conditions and show that the diffusion processes cannot provide significant concentration of NO at the periphery of discharge chamber. We consider a cylindrical chamber with radial temperature distribution, determined experimentally. The typical time of diffusion can be estimated as:

τ diff =

[r (T1 ) − r (T2 )]2

(22)

D

As initial temperature Т1 we will take the temperature corresponding to the maximal thermodynamic-equilibrium concentration of nitrogen monoxide (Т1 ~ 3500 К), and as final temperature Т2 we will take the temperature, when product stability is provided by a low decomposition rate of NO (Тк ~ 2000 К). It follows from experimental temperature distributions (Fig. 15a) that r(T1)-r(T2) ~ 1 cm. Taking into account that in the considered temperature range (3500÷2000 К) coefficient of diffusion is D ~ 50÷20 cm2/s, we obtain τdiff≈0.02÷0.05 s

(23)

For comparison we should indicate that time τ of nitrogen monoxide decomposition in the considered temperature range under the atmospheric pressure is [35]: T, К τ, s

3500 ~10-6

3000 8·10-5

2300 5·10-3

2000 1

Hence, at diffusion of NO molecules from the high-temperature zone (Т ~ 3500 К with equilibrium NO content of ~ 5%) to the zone with a lower temperature (to 2000 К), in every considered point gas will be in the state close to thermodynamic equilibrium. Within gas layers with the temperature below 2000 К, the time of NO decomposition is comparable or exceeds the typical times of diffusion, and concentration of NO there can correspond to the equilibrium value for Т ~ 2000 К (NO ~ 0.5 %). Therefore, at the chamber periphery, where the temperature is below 2000 К, nitrogen monoxide concentration because of diffusion should be, at its best, about 0.5%. To explain a high content of NO at Т<2000 К in the transformer coupled toroidal discharge, the authors of the current work have suggested the model of “self-quenching” [7].

Induction Transformer Coupled Discharges: Investigation and Application

141

The term “self-quenching” should be understood as the process of fast cooling of turbulent “hot” formations in peripheral “cold” gas due to turbulent transfer at the boundary between the forced and free vortices. The scheme of this process is shown in Fig. 22. In the framework of this model, the presence of nitrogen monoxide in the peripheral low-temperature zone of the discharge chamber is a result of cooling of high-temperature turbulent formations within the peripheral gas layers with the rate, sufficient for high concentration of NO.

Figure 22. The scheme of processes in discharge chamber, providing “self-quenching”: 1 – “transformer arc”; 2 – wall of discharge chamber; Dt – arc thermal size.L – size of turbulent formations; Q – heat flux to the wall.

Results of calculations by the suggested model [7] are shown in Fig. 21.b by the solid line. It is obvious that the behavior of curves is qualitatively the same, but their absolute values differ. Considering a possibility to obtain superequilibrium concentrations of NO at cooling of the high-temperature air [31], it can be assumed that the suggested model explains satisfactory the found effect. Therefore, it is shown at the example of synthesis of nitrogen monoxide from air that plasmachemical reactions, whose products are stable at the temperatures below 2000 К, distribution and final yield of the end product can be significantly effected by non-isothermal character and turbulent characteristics of a vortex flow in the discharge chamber of the lowfrequency induction transformer plasmatron.

142

I.M. Ulanov and M.V. Isupov

2.5. Natural Gas Processing in Plasma of Transformer Coupled Toroidal Discharge Recently, due to intensive development of fuel cell technologies and chemical industry, there is a great interest to hydrogen (or syngas) production from natural gas. The real industrial technologies require products consumption of tens of thousand of m3/h and, hence, the setup powers of tens of megawatts. An “appropriate” mean for generation of such flows of low-temperature plasma can be the transformer plasmatron, which has no principle limitations for the setup power. Thus, one of the purposes of the current research was the experimental modeling of real industrial processes at the existing setup. The plasmachemical technology of syngas (СО + Н2) production from СН4 requires binding of free carbon both for formation of CO molecule and undesirable product removal (soot). Therefore, synthesis should be carried out in a presence of atomic oxygen, which can be taken from the oxygen-bearing compounds. Carbon dioxide can be used as such a compound. Experiments were carried out at setup No. 1, shown in Fig. 6. Argon was fed into the main feed chamber. The mixture of СН4 and СО2 with given stoichiometry was fed into the additional feed chamber. This experimental scheme allowed us to maintain a stable discharge in the gas mixture with high contents of СО2 and СН4 at a separate part of discharge. The gas, reacted in the discharge chamber, was cooled in the heat exchanger (see Fig. 5a). Sampling was made after gas cooling in the heat exchanger (probe No. 3). Product concentration in a sample was analyzed by the mass-spectrometer. Formation of free carbon is possible in reaction CH4 + CO2 Æ 2H2 + 2CO at a lack of carbon dioxide. To get information about free carbon (soot) in the reaction products, the gas sampling line included the filters, where carbon particles were deposited. The absolute flow rates of СН4 and СО2 at the discharge chamber inlet varied from 0.25 to 0.6 m3/h and from 0.15 to 0.47 m3/h, respectively. According to mass-spectrometry results, under all studied conditions the degree of natural gas conversion in the discharge was about 100%. However, absolute concentrations of СО and Н2 changed depending on natural gas concentration in the initial mixture. Experimental results are shown as the diagrams in Fig. 23. The stoichiometric composition of supplied mixture (CH4/CO2) is plotted on axis x, and the ratio of outlet volumetric flow of a substance (Н2, СО or С2Н2) to the inlet volumetric flow of СО2 + СН4 is plotted on axis y. The “pictures” of filter surfaces are shown by circles in the same diagram. The darker the filter surface, the higher the content of free carbon in a sample. According to the diagram, with a rise of methane content relative to carbon dioxide, the contents of hydrogen and soot increase. At this, the content of carbon monoxide decreases. It should be also noted that there is acetylene in the reaction products. According to performed experiments on investigation of natural gas conversion into syngas, it is promising to use the transformer plasmatrons for processing of carbon-bearing substances.

Induction Transformer Coupled Discharges: Investigation and Application

2,0

143

H2

1,8 1,6 1,4

ηi

1,2 1,0 0,8 0,6

CO

0,4 C2H2

0,2

0,0 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 stochiometric composition CH4/CO2

Figure 23. Dependence of ηi on stoichiometric composition of initial mixture (CH4 and CO2), where ηi=Gi/(GCH4+GCO2) is the ratio of outlet volumetric flow of a substance (H2, CO or C2H2) to the inlet volumetric flow rate of CO2+CH4. Circles demonstrate the darkness degree of filters in the gassampling line.

3. TRANSFORMER LIGHT SOURCES As it is known production of low–temperature plasma by induction (electrodeless) gas discharges is one of the most promising methods to increase the efficiency and service life of gas-discharge light sources. The urgency of this problem is obvious because of the wide application of gas-discharge lamps in different fields of lighting technology (room and street illumination, biomedical and technical applications, etc). Several types of induction lamps are produced by industry; they differ both by inductor and gas-discharge chamber constructions, and by frequency of the discharge current. The radio-frequency (RF) induction discharge lamps (similar to QL induction lighting systems of Philips production), and induction lamps, operating by the principle of transformer coupled toroidal discharge (TCTD, similar to the Endura lamp, produced by Osram) are the most common now. It is necessary to note that electrodeless light sources (QL, Endura, Genura, Solara) produced now are intended only for inner and outer illumination (induction luminescent lamps). It is obvious that induction gas-discharge lamps can find wide application in other fields, especially in biomedical technologies. Thus, new efficient and ‘long-living’ ultraviolet (UV) emitters for water, air and surface disinfection can be developed on the basis of TCTD in mercury vapor. At the same time, TCTDs in mercury vapor are less studied than conventional arc discharges, and in particularly: (a) Dependence of electrical and optical characteristics of TCTD on the main conditions of discharge (discharge chamber diameter, pressure of mercury vapor, strength of discharge current) is studied insufficiently, and this does not allow prediction of a change in TCTD properties with variation of the above parameters, and correspondingly, it is impossible to choose the optimal parameters of developed electrodeless UV emitters.

144

I.M. Ulanov and M.V. Isupov

(b) Now a few theoretical models of TCTDs have been developed (e.g. [10, 11]); however, a particular TCTD in mercury vapours, considered here, has been hardly studied. Besides, it is not clear whether the available theoretical models of the arc discharges in mercury vapors can be used for calculation of TCTD characteristics in mercury vapours. For instance, in [19] the authors have calculated the characteristics of an experimental discharge cell similar to Osram Sylvania’s electrodeless fluorescent lamp in the range of ‘high-loaded’ conditions covering current densities 0.05÷0.6A/cm2 and Hg vapor pressures 1.2÷13 mtorr with application of two independent self-consistent numerical models of the positive column (GLOMAC and GLOW). The experimentally measured concentrations of excited and metastable atoms of mercury are significantly lower than predictions of the numerical code GLOMAC for the high discharge currents. Together with mercury-argon TCTD, the studies of TCTD in neon are of a particular interest because the new efficient light sources can be developed on this ground. To create the new electrodeless light sources in the visible and UV spectral regions on the basis of the TCTD principle, experimental investigations of the TCTDs were carried out in a mixture of argon and mercury vapour for a wide range of discharge parameters: mercury vapour pressure (0.1÷40 000 Pa), discharge current (1÷260 A), diameter of a gas-discharge tube (20÷75 mm), and current frequency (10÷250 kHz). The TCTD was also studied in neon within the range of neon pressures of 10÷600 Pa, diameters of gas-discharge tubes of 20÷58 mm, discharge currents of 1÷30 A, and current frequencies of 25÷250 kHz.

3.1. Experimental Setups To study the transformer coupled transformer discharge in mercury vapours, the experimental setup, shown in Fig. 24, was used. The main elements of this setup are as follows: 1 – quartz toroidal gas–discharge chamber; 2 – system of ferromagnetic cores with primary winding 3; 4 – high–frequency power source; 5 – accommodation unit between the power source and TCTD. In the experiments we have used toroidal (in the form of elongate torus) gas–discharge tubes, with the following geometrical dimensions: (a) diameter of gas– discharge tube D = 20 mm, perimeter of gas–discharge tube along the middle line L = 210 mm; (b) D = 35 mm, L = 560 mm; (c) D = 58 mm, L = 880 mm; (d) D = 75 mm, L = 1880 mm. All gas-discharge chambers were filled with buffer gas (argon) under a pressure of 100 Pa and the strictly dosed amount of pure mercury or mercury amalgam with composition of 90%In – 5%Ag – 5%Hg; then, they were soldered. To measure the strength of discharge current I, current transformer 6 was used. Simultaneously, density of discharge current J was determined as a ratio between the measured strength of discharge current and the crosssection of gas discharge chamber S: J = I/S. To measure the voltage U that drives the TCTD, measurement coil 7, enveloping the gasdischarge chamber over perimeter L was used. At that, the strength of electric field E was determined as a ratio between measured voltage U and the discharge perimeter along the middle line: E = U/L. The wall temperature of gas-discharge chamber T was controlled by thermocouples 8, which allowed us to determine the mercury vapour pressures p in the

Induction Transformer Coupled Discharges: Investigation and Application

145

discharge chamber using the known dependence between the pressure of saturated mercury vapour and the temperature. To measure the optical and spectral characteristics, the following optic probes 9 were used: •

•

•

Luxmeter, UV–radiometer TKA-PKM (model 06), intended for measuring illumination (lx) within a visible range and UV–irradiation (mW/m2) within the range 280 – 400 nm (A + B zones). UV-radiometer TKA-PKM (model 12), intended for measurements of UV– irradiation within three ranges: UVA (315–400 nm), UVB (280–315 nm) and UVC (200–280 nm). Setup for analysis of spectra on the basis of spectrometer AVASPEC–2048FT–2 and PC, used for analytical investigation in a spectral range 200–800 nm with a very high photometric sensitivity and optic resolution from 0.04 nm.

While measuring the optical characteristics of TCTD, diaphragming screen 10 with a window of size h was installed in front of the studied object. The optical detectors were located at distance l, which exceeded the size of a “window” by a factor of 10 and more, and so we considered radiation of the outlined tube region as point one. In this case, power, radiated into the studied range of the spectrum from the outlined part of the gas-discharge chamber h is calculated as the product of the light flux density (measured by a device) on the area of a sphere with radius l. The developed experimental setup was used for determination of dependence of the electric field strength, UV output, visible radiation and spectral characteristics of a discharge on the discharge conditions: discharge current strength, pressure of mercury vapour and diameter of the gas discharge chamber.

Figure 24. Scheme of experimental setup for investigation of TCTD in mercury vapours. 1 – quartz gas–discharge chamber;2 – ferromagnetic cores; 3 – primary winding; 4 – power source; 5 – accommodation unit; 6 – current transformer; 7 – measuring coil;8 – thermocouples; 9 – optic probes;10 – diaphragming screen.

146

I.M. Ulanov and M.V. Isupov

Figure 25. Scheme of experimental setup for investigation of TCTD in neon.1 – quartz gas–discharge chamber;2 – magnetic core;3 – primary winding;4 – power source;5 – exhaust tubes;6 – gas line; 7 – vacuum meter;8 – measuring coil; 9 – current transformer; 10 – optic probes;11 – vacuum pump.

To study electric-physical and radiation characteristics of the transformer coupled toroidal discharges in neon, the experimental setup, shown in Fig. 25, was used. The diameter of quartz gas-discharge tube 1 in different experiments varied from 20 to 58 mm; the crosssection of magnetic core 2 – from 5 to 20 cm2; the number of coils in primary winding 3 – from 1 to 5. To generate a discharge, power sources 4 with current frequency 25 and 250 kHz were used. Gas-discharge tube 1 was connected to gas line 6 via exhaust tubes 5, and this allowed variations in the gas pressure during the experiment. The vacuum meter ILMVAC PIZA 111 7 was used for pressure measurements; it allowed measurements of pressure in the range from 0.1 Pa to 100 kPa with an error of not higher than 10 %. The discharge voltage and current were measured by additional coil 8 and current transformer 9, respectively, with an error of not higher than 3%. To register discharge radiation in the visible range of the spectrum, the digital luxmeter ТКА–01/3 was used. To measure the total radiation flux of a discharge in the visible and long-wavelength IR spectrum range, the IMO-2N in combination with a “water” light filter (10) was used. The system was pumped by forevacuum pump 11. The software–hardware systems for control of spectrometric measurements, developed at Petrozavodsk State University, were used for spectrometric studies of plasma of an induction transformer discharge: the software–hardware system “Light” and measuring bench for highresolution spectroscopy “Spectrum”. The block-scheme of hardware system “Light” is shown in Fig. 26. The main units of this system are: 1 – system of data gathering and control on the basis of modular system CAMAC and PC; 2 – spectrometer DFS–12, operating by the principle of double monochromator; 3 – photoelectronic multiplier, connected to the outlet slot of the monochromator; 4 – optic illumination system (lens + deflecting mirror), forming an image of a light source on the diaphragmatic inlet slot of the spectrometer; 5 – unit of spatial scanning (system of mirrors,

Induction Transformer Coupled Discharges: Investigation and Application

147

controlled by a step motor in the horizontal plane); 6 – test subject (a gas-discharge light source).

Figure 26. Block-scheme of software-hardware system “Light”.1 – System of data gathering and control on the basis of the modular system CAMAC and PC; 2 – spectrometer DFS–12; 3 – photomultiplier; 4 – optic illumination system of spectrometer inlet slot; 5 – unit of spatial scanning; 6 – test subject (light source).

This setup allows registration for the spectra of spatially nonuniform emitting objects, preservation of results on detailed information about experimental conditions, and analysis of energy characteristics of radiation within the spectral range of 360–800 nm. The test subject was an induction neon lamp, operating by the principle of induction transformer discharge with the following parameters (the scheme of this lamp is shown in Fig. 26): outer radius of torus of ~ 8 cm, inner radius of torus of ~ 4 cm, inner radius of gasdischarge tube Rk ≈ 17.5 mm, perimeter of gas-discharge tube along the mean line L=2πRT≈38 cm), and neon pressure inside gas-discharge tube of 1 mm Hg. While measuring the spectral characteristics of a discharge, the current and voltage of discharge glow were also measured (with an error of not higher than 3%), together with the wall temperature of a gas-discharge tube (with an error of not higher than 1 ºС). When measuring the spectral characteristics the induction lamp was put vertically, the radiance spectra and profiles were measured in A–B cross-section of a gas-discharge tube. Point O on the torus mean line was taken as a zero point. The measurement error for the absolute values of spectral radiance was not higher than 5–10%. The scheme of a measurement setup for high-resolution spectroscopy “Spectrum”, intended for examination of spectral line circuits, is shown in Fig. 27. This setup was made by the standard scheme of spectrometer 1 with Fabry–Perot interferometer 2. Studied gas– discharge light source 3 is located within the focal plane of lens L1. Second objective L2 is focused on plane Y of diaphragmatic inlet slot of the spectrograph. The outlet slot of spectrograph is connected to photoelectronic multiplier 4, whose signal is sent to the ADC inlet and registered by PC 5. Since interferometer transmission for monochromatic wave λ, down coming at angle φ to the mirror normal, is determined by expression

148

I.M. Ulanov and M.V. Isupov

P(λ , ϕ ) =

T2

(1 − R ) + 4R sin 2

2⎛

⎜π ⎝

2d cos(ϕ ) ⎞ ⎟ λ ⎠

,

(24)

where T is mirror transmission coefficient, R is reflection coefficient, and d is distance between the mirrors; the system of interference rings, described by expression (24), where φ=r/F (r is radius, F is focal distance), is formed in plane Y. Distinguishing a light flux in the central area (φ=0) by a low-radius diaphragm and varying linearly in time parameter d/λ, we fix periodic signal P'(t), and period of this signal Δt corresponds to interferometer constant Δλ=λ2/2d. For monochromatic radiation fixed signal P' corresponds to the interferometer instrument function; if there is line broadening the registered signal is “convolution” of the spectral line circuit and instrument function. Parameter d/λ can be changed in the different manners; in the current case the “deflection” over spectral range Δλ is made by a change in the air pressure in pressure chamber 6 at fixed distance between mirrors d. For this purpose, the air is evacuated from the pressure chamber by vacuum pump 7, before the measurements, and then air leak valve 8 was opened. At the initial stage the pressure in the chamber increases almost linearly, and this causes a linear increase in air refraction coefficient n. Respectively, the wavelength with frequency ν changes also linearly: λ=c/nν. In experiments the interferometer with basis d=10 mm was used, what corresponds to operation range Δλ of about 0.1 А (for the visible range of spectrum). The coefficient of mirror reflection for this interferometer was ~74 %. The measurement error for intensity of distinguished spectral line did not exceed 2%. The discharge current and voltage were also measured together with spectral characteristics with an error of not higher than 3%.

Figure 27. Scheme of measurement setup “Spectrum”.1 – Spectrograph; 2 – Fabry–Perot interferometer, (Y– diaphragmatic inlet slot of spectrograph); 3 – light source; 4 – photomultiplier; 5 – PC with ADC; 6 – pressure chamber; 7 – pump; 8 – leak valve;L1, L2 – lenses with focal distance F.

Induction Transformer Coupled Discharges: Investigation and Application

149

3.2 Transformer Coupled Toroidal Discharge in Mercury Vapour. The dependence of electric field strength E on current density J is shown in Fig. 28 for several levels of mercury vapor pressure p (tube diameter D = 75 mm, current frequency f = 10 kHz). According to Fig. 28, the voltage–current characteristic (VCC) of TCTD is similar to the VCC of the DC arc discharges: there are a descending branch, area of minimal electric field intensity and an ascending branch. The position of VCC minimum is determined by the pressure of mercury vapour: in the range of vapour pressure of 1÷7 kPa, the VCC of TCTD has no distinct minimum. With the following increase in pressure, the electric characteristics of discharge change qualitatively: the VCC minimum shifts towards the area of lower discharge currents and becomes more distinct. Thus, for a pressure of mercury vapor of ~11 kPa, the minimum of E is in the range of 0.65÷1 A/cm2, and for pressure of 35 kPa, it is within 0.2÷0.3 A/cm2. Dependence E versus p is shown in Fig. 29, for different diameters of the discharge tube (current density is 1 A/cm2). Dependence E(p) for a DC arc discharge is shown for comparison (J = 0.35 A/cm2) [36]. According to Fig. 29, in the range of mercury vapour pressure of 3÷30 Pa, the electric field strength of TCTD has a local minimum. With a further increase in pressure, the electric field strength increases as E ~ p1/3. For the case of DC arc discharges in mercury vapour, dependence E(p) takes a qualitatively similar form [36]. A change in the discharge chamber diameter does not qualitatively effect the form of dependence E(p), decreasing the field strength with a rise in diameter. Thus, with an increase in the chamber diameter from 20 to 35 mm, the field strength decreases by a factor of 2.6, and with an increase of up to 75 mm, it decreases by a factor of 3.2.

p Hg, kPa 35 22 11 7 2 1

2,5

E, V/cm

2,0 1,5 1,0 0,5

0

1

2

3

4

J, A/cm

5

6

7

2

Figure 28. Electric field strength E versus discharge current density J, for different values of mercury vapour pressure (D = 75 mm, f = 10 kHz).

150

I.M. Ulanov and M.V. Isupov

E, V/cm

D f 20 mm, 250 kHz 35 mm, 250 kHz 75 mm, 10 kHz 32 mm [36]

1

0,1 -1 10

0

10

1

10

2

3

10 10 p Hg, Pa

4

10

5

10

Figure 29. Electric field strength E versus mercury vapour pressure p, for different diameters of chambers (J = 1 A/cm2).

For the DC arc discharges in the long cylindrical tubes with specific discharge power of 20÷50 W/cm and the middle or high pressure of mercury vapour, interaction between electric field strength, diameter of the gas–discharge tube and specific mass of evaporated mercury m is presented by the known dependence, derived by Elenbaas [37]:

ED3 / 2 ≈ 370m7 / 12

(25)

where value E is presented in V/cm, D is in millimeters, and the specific mass of evaporated mercury m is in mg/cm. The dependence of E · D3/2 on specific mass of evaporated mercury m (mg/cm) is shown in Fig. 30 for the fixed specific power of TCTD of P = 40 W/cm. Data shown in Fig. 30 corresponds to the range of mercury vapour pressures of 10÷40 kPa, and the tube diameter of 35÷75 mm. The similar dependence, obtained by Elenbaas for the mercury DC arc discharges in the long cylindrical tubes with diameters 12÷38 mm and specific amounts of mercury 0.3÷50 mg/cm (which corresponds to a pressure of mercury vapour of 17÷300 kPa) for the specific discharge power of 30 W/cm are also shown in Fig. 30 [37]. As it is seen from Fig 30, the experimental data obtained for the TCTD are laid perfectly on the line, described by expression (25). This proves the fact that, at least, in the range of mercury vapour pressure of 10÷40 kPa and diameters of gas-discharge tube of 35÷75 mm, the electrical characteristics of TCTD are similar to those of the DC arc discharges. The emission yield to non-resonance lines of the visible and UV–spectra as a function of TCTD specific power is plotted in Figs. 31 and 32 (as a percentage of input power) for mercury vapour pressure of p ≈ 25 kPa and tube diameter of 75mm (f = 10 kHz).

Induction Transformer Coupled Discharges: Investigation and Application

TCTD DC arc discharge, [37] D=35 mm, f=250 kHz D=12 mm D=58 mm, f=25 kHz D=18 mm D=75 mm, f=10 kHz D=28 mm D= 38 mm

3/2

10000

1000

3/2

E·D , V/cm·mm

151

100 0,1

1 10 m Hg, mg/cm

100

Figure 30. Interaction between electric field strength E, diameter of gas–discharge chamber D and specific amount of evaporated mercury m, for TCTD and DC arc discharges.

404.7 nm 546.1 nm

435.8 nm 577.0/579.1 nm

Emission yield, %

2,5 2,0 1,5 1,0 0,5 0,0

0

100

200 300 400 500 Specific power P, W/cm.

600

Figure 31. Dependence of emission yield to non-resonance lines of the visible spectrum range (400–600 nm) on specific discharge power. D = 75 mm, p ≈25 kPa, f = 10 kHz.

Analysis of data shown in Figs. 31 and 32 demonstrates that dependence of emission yield to the spectral line on the specific discharge power is determined by its excitation potential. Thus, the efficiency of the visible triplet 404.7, 435.8 and 546.1 nm with excitation potential of 7.73 eV reaches its maximal value with a specific discharge power of ~ 50W/cm, and with a further increase in power, decreases significantly. The feature of this triplet is the lowest excitation potential (7.73 eV) after the resonance lines (4.88 eV for line 253.7 nm). A decrease in the efficiency of the lines of the UV and visible spectrum ranges with excitation potential of 8.85 eV (312.6/313.2, 365.0/366.3, 577.0/579.0 nm) expressed weaker by an order for the vapour pressure of ~25 kPa. A decrease in emission yield with a rise of

152

I.M. Ulanov and M.V. Isupov

discharge power was not determined for the lines with excitation potential of 9.56 eV (265.2, 302.8/303.6 nm). 265.2 nm 312.6/313.2 nm

302.8/303.6 nm 365.0/366.3 nm

Emission yield, %

1,5 1,0 0,5 0,0

0

100 200 300 400 500 Specific power P, W/cm

600

Figure 32. Dependence of emission yield to non–resonance lines of the UV spectrum range (260–400 nm) on specific discharge power. D = 75 mm, p ≈25 kPa, f =10 kHz.

80 light efficiency, lm/W

70

D=20 mm, I=10 A, f=250 kHz D=35 mm, I=10 A, f=250 kHz D=75 mm, I=50 A, f=10 kHz D=27 mm, I=4 A [37]

60 50 40 30 20 10 0 -1 10

0

10

1

10

2

3

10 10 p Hg, Pa

4

10

5

10

Figure 33. Light efficiency versus mercury vapour pressure for the fixed TCTD current.

A decrease in the light efficiency of TCTD was mentioned in [20] for specific discharge power of above 80 W/cm and pressure of mercury vapour of ~ 10 kPa; however, there was no detailed information on energy redistribution in the radiation spectrum. According to published data [38], the light efficiency of mercury arc discharge should increase continuously with a rise of power (for the fixed pressure), tending to its limit value (~80 lm/W). To explain the observed effect, the authors of this paper offer the following hypothesis: since in gas discharges of the middle and high pressure, the population with excitation potential Uk is determined as Nk ~ e−Uk/kT, and the temperature of plasma increases with a rise

Induction Transformer Coupled Discharges: Investigation and Application

153

Efficiency in UV range 280-400 nm, %

in discharge power, the population of excited levels should be redistributed in favour of highly excited states, and correspondingly, the emission yield in the close IR zone of spectrum (1000–1700 nm) should also increase in the line with excitation potential of 9.5÷10 eV. Unfortunately, the used measurement system did not allow registration of the IR spectrum of discharge, and this assumption could not be checked experimentally. The dependence of TCTD light efficiency on the pressure of mercury vapour is shown in Fig. 33 for the fixed discharge current and different diameters of the gas–discharge tube. Dependence of DC arc discharge light efficiency is shown in Fig. 33 for comparison (I = 4A, D = 27 mm, [37]). According to Fig. 33, the dependence of light efficiency of TCTD and DC arc discharge on mercury vapour pressure are qualitatively similar: the first local maximum of light efficiency is reached for a pressure of ~10 Pa, and makes up ~20÷30 lm/W; the local minimum of light efficiency corresponds to the range of mercury vapour pressures of 100÷1000 Pa, and with a further increase in pressure, the light efficiency starts increasing fast. It can be also seen in Fig. 33 that with an increase in the diameter of the gas-discharge tube, the light efficiency of TCTD increases; the similar dependence is observed for the DC arc discharges [38]. The dependence of the emission yield to non-resonance lines of the UV spectrum range (280–400 nm) on the pressure of mercury vapour is shown in Fig. 34 for the fixed discharge current. According to the analysis of the results, shown in Figs. 33 and 34, the efficiency of non–resonance lines of the UV and visible ranges of the TCTD radiation spectrum similarly depend on the pressure of mercury vapour and the diameter of the gas–discharge tube.

8

D=35 mm, I=10 A, f=250 kHz D=20 mm, I=10 A, f=250 kHz

7 6 5 4 3 2 1 0 -1 10

10

0

1

10 10 p Hg, Pa

2

10

3

10

4

Figure 34. Emission yield to non-resonance lines of the UV-range 280–400 nm versus mercury vapour pressure for the fixed TCTD current.

I.M. Ulanov and M.V. Isupov Emission yield to the line 253.7 nm, %

154

D=20 mm, I=10 A, f=250 kHz D=35 mm, I=3 A, f=250 kHz D=35 mm, I=7 A, f=250 kHz D=75 mm, I=50 A, f=10 kHz

70 60 50 40 30 20 10 0 10

-1

10

0

10

1

10

2

10

3

10

4

p Hg, Pa Figure 35. Emission yield to resonance line 253.7 nm versus pressure of mercury vapour for fixed current of TCTD.

Emission yield to line 253.7 nm, %

The dependence of the emission yield to resonance line of 253.7 nm on the mercury vapour pressure is shown in Fig. 35 for the fixed discharge current. As it is clearly seen, the range of pressures of ~0.5÷1.5 Pa (this corresponds to the temperature of the cold spot of 30÷50 ◦C) is of a particular interest; up to 60% of electric power put into the resonance line there, and therefore, the mercury discharges of a low pressure are the highly efficient sources of UV radiation, which can be used in practice. Since the maximal radiation yield to resonance line 253.7 nm can be obtained in a very narrow temperature range (30÷50 ◦C), a series of experiments was carried out using a threecomponent mercury amalgam of the following composition: 90%In–5%Ag–5%Hg (Fig. 36).

Amalgam I=3 A I=10 A

70 60

Pure Hg I=3 A I=10 A

50 40 30 20 10 0

0

20

40

60

80

100

120

Tube temperature, C

Figure 36. Emission yield to line 253.7 nm versus wall temperature for the fixed current of TCTD (amalgam and pure mercury filling). Diameter of gas-discharge tube D = 35 mm, f = 150 kHz.

Induction Transformer Coupled Discharges: Investigation and Application

155

Figure 37. Electrodeless mercury high intensity discharge lamp of 3-kW power.D=60 mm, f=25 kHz.

Figure 38. Electrodeless mercury high intensity discharge lamp of 100-kW power.D=80 mm, f=10 kHz.

Figure 39. Electrodeless mercury UV lamps of 100, 200 and 50 W.F = 150 kHz, D = 35÷40 mm.

156

I.M. Ulanov and M.V. Isupov

The dependence of radiation yield to the resonance line on the temperature of a gas– discharge chamber wall is shown in Fig. 36 for the TCTD with amalgam and pure mercury filling. As it can be seen from the figure, using mercury amalgam, the maximum of resonance radiation yield moves to the area of the higher temperatures (>70 ◦C), and the temperature range corresponding to the maximal efficiency of resonance line 253.7 nm becomes significantly wider (70÷120 ◦C). The unique samples of mercury high-intensity discharge lamps of the 3÷100-kW power with efficiency of ~60 lm/W were developed on the basis of investigations of the TCTD in mercury vapors (Fig. 37, 38). On the basis of experimental studies of the TCTD with amalgam filling, the pilot samples of bactericidal induction UV lamps with the power of 50, 100, 200 and 500W and current frequency f = 150 kHz (Fig. 39) were constructed; their emission yield to line 253.7 nm is ~30÷35% of consumed electric power. 1,0 0,8 0,6 150 0,4 0,2 0,0 180 0,2 0,4 0,6 210 0,8 1,0

90 120

60

1 2 3 4 5 6 7

30

0

330

240

300 270

Figure 40. Corner diagram of directivity of the 50Wlamp.1–253.7 nm, 2–312.6/313.2 nm, 3– 365.0/366.3 nm, 4–404.7 nm, 5–435.8 nm, 6–546.1 nm, 7–577.0/9.0 nm. 90

1,0 0,8 0,6 150 0,4 0,2 0,0 180 0,2 0,4 0,6 210 0,8 1,0

120

60

30

0

1 2 3 4 5 6

330

240

300 270

Figure 41. Corner diagram of directivity of the 100W lamp. 1–253.7 nm, 2–312.6/313.2 nm, 3– 365.0/366.3 nm, 4–404.7 nm, 5–435.8 nm, 6–546.1 nm.

Induction Transformer Coupled Discharges: Investigation and Application

157

The measured corner diagrams of emission directivity of different spectral lines are shown in Figs. 40 and 41 for developed lamps with the power of 50 and 100 W (in the reduced form, the lamp position is schematically shown in the figure (vertical one), the axis of spectral characteristic measurements passes the centre of the lamp). According to Figs. 40 and 41, the corner diagram of radiation directivity of resonance line 253.7 nm differs maximally from the round one, what is possibly caused both by its intensive absorption in discharge plasma and partial absorption by the chamber walls. Before discussing the experimental results, first of all, it is necessary to emphasize the fact that in contrast to RF induction discharges, the TCTD can be considered as the analogue of an arc discharge with electrodes, combined in one plane [10]. Therefore, it is interesting to apply the standard models of arc discharges for the description of TCTD. However, it should be noted that the most standard models of arc discharges are developed for description of DC discharges in long cylindrical tubes. As a result, the question arises: whether these models are applicable for TCTD description. Nevertheless, it can be assumed that for some cases, the standard models of axisymmetric DC arc discharges will satisfactorily describe the TCTD characteristics. Let us consider the main conditions, when the TCTD can be described by the standard models: (a.) A small curvature radius of the toroidal gas-discharge chamber. Apparently, in this case deviations from cylindrical symmetry in a discharge are minimal, and the discharge itself can be shown as a discharge in a straight cylindrical tube with length L, where L is the perimeter of the gas-discharge chamber over its middle line. (b.) Fulfillment of condition f · τ > 1, where f is the current frequency and τ is the typical time of plasma “damping”. In this case, the parameters of discharge plasma (conductivity, electron temperature, concentration of excited and metastable atoms) do not change during discharge current passing through zero, and the TCTD can be considered and used as the analogue of a DC discharge. (c.) An absence of the skin-effect on the electrical characteristics of a discharge. The analysis of experimental data shown in Figs. 33–35 demonstrates that dependences of the optical characteristics of TCTD on the mercury vapour pressure, discharge current strength and diameter of the gas-discharge tube correspond qualitatively to similar dependences for the arc mercury discharges (excluding the abovementioned effect of the luminous efficiency reduction to a visible triplet with an excitation potential of 7.73 eV with an increase in discharge power). However, for some cases, there are significant quantitative differences between the radiation characteristics of TCTD and arc mercury discharges, which are possibly caused by different “glow” conditions. Thus, even for the pressure of ~ 20 kPa, the light efficiency of TCTD with D = 75mm is ~ 70 lm/W (Fig. 33), what is relatively close to the known limit of light efficiency of the arc mercury discharges with a tube diameter of 20÷30 mm ( ~ 80÷85 lm/W), obtained if the pressure of mercury vapour equals several atmospheres. The authors of the current paper consider the possibility of using the standard “channel” model of axisymmetric DC arc discharges [37], based on the assumption of LTE presence in discharge plasma, for the calculation of the TCTD parameters in mercury vapour. According to this model, electric field strength E in a mercury discharge is determined by expressions

158

I.M. Ulanov and M.V. Isupov

E=

C1 p 7 / 12 P1 / 2

(26)

C2m7 / 12 P1 / 2

(27)

(P − PT )1 / 3 D1 / 3

Or

E=

(P − PT )1 / 3 D3 / 2

where C1, C2 are constants, PT is the value of specific heat losses of discharge, P is specific discharge power, m is the specific amount of evaporated mercury and D is the tube diameter. Additionally, it is assumed by this model that PT does not depend on the discharge power and diameter of the discharge tube. In this case, minimal E is reached, if the following condition is satisfied: P = 3PT

(28)

The analysis of TCTD voltage–current characteristic (Fig. 28) with criterion (28) provides the following values of specific heat losses: PT = 16 W/cm for p ≈ 10 kPa; 7.9 W/cm for p ≈ 22 kPa and 7.5 W/cm for p ≈ 35 kPa, what coincides with the results of Elenbaas [37] (PT = 9÷10 W/cm), obtained similarly for the arc mercury lamps of high pressure in the following range of discharge parameters: P 15÷60 W/cm, D 12÷38 mm. For the pressures of mercury vapour below 10 kPa, the VCC of TCTD has no the expressed minimum, what proves the fact that formulae (26) and (27) can have high errors for calculation of electrical characteristics of the TCTD. According to formula (26), field strength E for the fixed pressure of mercury vapour depends on the tube diameter as E ~ D−1/3. The analysis of data in Fig. 29 demonstrates that an increase in the tube diameter from 35 to 75 mm really provides the reduction of the field strength by a factor of 1.3, whereas for the TCTD with a tube diameter of 20 mm, the experiment gives an excessive value of the field strength. However, it should be noted that in this experiment the ratio of perimeter and diameter L/D for the toroidal chamber was ~10, whereas, for the lamps with the tube diameter of 35÷75 mm, parameter L/D varied from 15 to 25. It can be assumed that for L/D ~ 10, the condition of axial symmetry is disturbed considerably (visually, the discharge was significantly “pressed” into the inner wall of the toroidal chamber), therefore, the standard models of axisymmetric gas discharges cannot be applied for this situation. As it can be seen from Fig. 30, for the TCTD with the chamber diameter of 35÷75 mm and fixed discharge power P ≈ 40W/cm, product E · D1/3 can be perfectly described by formula (25) with constants, obtained by Elenbaas via the analysis of characteristics of the arc axisymmetric mercury discharges. Therefore, for the pressures of mercury vapor above 10 kPa and ratio L/D > 15, the TCTD in mercury vapour can be approximately considered as the axisymmetric, and the “channel” model of axisymmetric discharges can be used for calculation and analysis of discharge parameters.

Induction Transformer Coupled Discharges: Investigation and Application

159

3.2. Transformer Coupled Toroidal Discharge in Neon

E, V/cm

Dependence of electric field strength of the TCTD on neon pressure is shown in Fig. 42. Dependence of electric field strength on discharge current is shown in Fig. 43.

D, mm; I, A 20 ; 1 [39]

2,0 1,8 1,6 1,4 1,2 1,0 0,8 0,6 0,4 0,2 10

20 ; 5 [39] 20 ; 10 35 ; 1 35 ; 8 51 ; 1 [40] 51 ; 10 [40] 58 ; 20

100061 ; 10 [39]

100 p Ne, Pa

E, V/cm

Figure 42. TCTD electric field strength vs. neon pressure.[39] – Published data, DC arc discharge.[40] – Published data, TCTD at current frequency of 450 kHz.

1,1 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2

35 mm; 25 kHz 35 mm; 250 kHz 61 mm; [39]

1

I, А

10

Figure 43. TCTD electric field strength vs. discharge current. Neon pressure of 1 torr.

According to analysis of data, shown in Figs. 42, 43, dependences E(p, D, I), observed for the TCTD in neon coincide qualitatively with similar dependences for the positive column of a gas discharge in neon. However, it was determined that the field strength increases with a rise of current frequency (Fig. 43). To analyze the observed increase in the field strength with a rise of current frequency, let’s estimate the influence of skin-effect on electric-physical characteristics of a discharge. To simplify calculations, we will represent the studied discharge as a uniform cylindrical conductor with radius r, with specific electric conductivity σ. In this case efficient depth of skin layer δ is calculated by formula

160

I.M. Ulanov and M.V. Isupov

δ=

2

(29)

2 μ0ωσ

where μ0 is magnetic constant, σ is conductivity of TCTD plasma. Plasma conductivity, averaged by the discharge cross-section, can be determined from relationship j=σE. Dependence of skin layer thickness for TCTD in neon on discharge current strength is shown in Fig. 44. Conductivity is calculated by the voltage-current characteristic, shown in Fig. 43. According to the figure, the minimal thickness of the skin layer almost twice exceeds the radius of a gas-discharge tube. In this case, the influence of skin-effect on electric-kinetic characteristics of a discharge can be described by approximated formula Rω/R0=1+k4/3 (at k<1)

(30)

where Rω is efficient resistance of a “conductor” with radius r to alternating current, R0 is resistance to direct current, k=r/2δ. It follows from formula (30) that to increase the field strength by 10% via the skin-effect, the value of k should be ~0.75, what corresponds to the skin layer thickness of ~1.2 cm. Thus, we can make a conclusion that the observed increase in the field strength with a rise of current frequency cannot be explained by the skin-effect.

δ, cm

100

25 kHz 250 kHz

10

1

1

I, А

10

Figure 44. Calculated thickness of skin-layer δ vs. strength and frequency of discharge current (p Ne=1 mmHg, D=35 mm).

Dependences of luminous efficiency of TCTD in neon on neon pressure, discharge current and diameter of the gas-discharge tube are shown in Figs. 45 and 46. Dependence of specific light flux of TCTD (radiated from 1 cm of the gas-discharge tube) on discharge current strength in visible and long-wave IR spectral region (400–1200 nm) is shown in Fig. 47. It can be seen in Fig. 45, the maximal luminous efficiency of TCTD is reached within the pressure range of 100÷200 Pa. According to publications [39], the maximal luminous efficiency of an arc discharge in neon corresponds to the pressures of 30÷50 and 30÷100 Pa for tube diameters of 60 and 20 mm, respectively.

Induction Transformer Coupled Discharges: Investigation and Application D, mm 20 20 35 58 61

Luminous efficiency Lm/W

35 30 25

161

I, A 5 А, [39] 10 А 10 А 20 А 20 А, [39]

20 15 10 5 0 10

100 p Ne, Pa

1000

Figure 45. TCTD luminous efficiency vs. neon pressure.

Observed divergence can be caused by different experimental conditions: in paper of Klarfeld [39] electric-physical and radiation characteristics of a discharge were measured in the sealed gas-discharge tubes, whereas, data in Fig. 45 was obtained at the constant neon flow through a gas-discharge tube. D, mm f, kHz 35 25 35 250 61 [39]

Luminous efficiency, Lm/W

40 35 30 25 20 15

1

I, А

10

Figure 46. Luminous efficiency vs. TCTD current strength. 1,6

25 kHz 250 kHz

specific light flux, W/cm

1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0

0

5

10

I, A

15

20

25

Figure 47. Specific light flux (W/cm) of the TCTD in visible and near IR radiation spectral ranges (400–1200 nm) vs. discharge current.D=35 mm, p=1 torr.

162

I.M. Ulanov and M.V. Isupov

Normalized intensity

For the fixed neon pressure (1 torr) and diameter of the gas-discharge chamber D=35 mm the maximal luminous efficiency of the TCTD is reached at the discharge current of 3÷5 А, and it makes up ~34÷36 lm/W (Fig. 46), what is close to the maximal value for the arc discharges in neon (~42 lm/W [39]), and exceeds significantly the efficiency of industrial arc neon gas-discharge lamps (~10÷15 lm/W) [38]. It should be noted specially that an increase in current frequency increases the light flux, radiated by the TCTD in visible and long-wave IR spectral regions (Fig. 47). According to Figs. 43 and 47, with a rise of current frequency the specific light flux of TCTD and electric field strength increase similarly, therefore, the luminous efficiency of the TCTD (Fig. 46), proportional to TCTD efficiency η=F/(IE), stays constant. 640.2 nm 1,1 540.0 nm 1,0 580.4 nm 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 -1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 x

Figure 48. Radial distribution of brightness of neon spectral lines. TCTD, p = 1 torr, D = 35 mm, J = 1 А/cm2, L/D = 11. 5

2p 3p

x=0 x=0.85

8

nk(x)/gk, cm

-3

10

7

10

5

2p 4d 6

10

5

10 18,0

18,5

19,0

19,5

20,0

20,5

21,0

21,5

Uk, eV

Figure 49. Normalized concentration of excited neon atoms nk(x)/gk vs. excitation potential of level Uk TCTD, p=1 mm Hg, D=35 mm, J=1 А/cm2.

The spectral characteristics of TCTD were firstly measured in neon: radial distributions of spectral line brightness and optic spectra of discharge in different areas of a gas discharge.

Induction Transformer Coupled Discharges: Investigation and Application

163

The shape of spectral line circuits was studied. Experimental investigations were carried out for the optimal conditions of discharge glow (p=1 torr, D=35 mm, I=5÷10 А). Gas–discharge tube perimeter L was 38 cm, thus, for the considered case L/D ≈ 11. Radial distributions of brightness F(x) for different neon lines, measured in A–B tube cross-section (Fig. 26), are shown in Fig. 48 in the normalized form. The brightness profiles of radiation transition lines 2p53p–2p53s and 2p54d–2p53p were studied. According to analysis of data, shown in Fig. 48, radial distributions of brightness are asymmetrical. The main reason for increased energy release inside the toroidal gas-discharge tube is a change in the value of longitudinal electric field E(x) with a change in coordinate x. Thus, it follows from equation (5) that E(x)=E/(1 - πxD/L)

(31)

In particular, for the considered case E(0.8)/E(-0.8)≈1.6, at that Fλ(0.8)/Fλ(-0.8)≈1.7. Thus, spectral measurements confirm the above assumption that at L/D ~ 10 TCTD is not the axisymmetric discharge. Since there is no radial symmetry, we can use the standard Abelian transformation for determination of spatial distribution of radiation intensity and, hence, distribution of excited atom concentration. Nevertheless, while analyzing the spectra of neon induction discharge, measured at different points of tube x, we can calculate concentrations of excited atoms nk(x)=Nk(x)/l(x), averaged by cross-section l(x) (Fig. 26), where Nk(x)~Fλ(x). Results of calculations (normalized to statistical weights gk of corresponding excited states k) are shown in Fig. 49. According to analysis of data in Fig. 49, the population character of excited states 2p53p and 2p54d,5s at transition from the central to peripheral regions of the TCTD in neon does not change qualitatively, only quantitative changes are observed. This proves the fact that the electron temperature is the same both in the central and peripheral regions of the TCTD. There was a series of experiments on determination of circuit shapes for neon spectral lines within the range of 585–700 nm (2p53p–2p53s radiation transitions). These experiments used the measurement setup of high-resolution spectroscopy “Spectrum”, assembled by the standard scheme of Fabry-Perot interferometer with a spectrograph (Fig. 27). According to analysis of TCTD generation conditions in neon, the main mechanisms, determining the broadening of observed spectral lines, should be the following: 1. Doppler broadening. 2. Line broadening because of self-absorption (re-adsorption) of radiation. According to the estimates performed, contribution of other mechanisms of broadening can be neglected. The shape of spectral line circuit with wavelength λ0, widened by the Doppler effect, is described by expression: ϕ (λ , λ0 ) =

c

λ0

⎛ Mc 2 (λ − λ0 ) 2 M EXP⎜⎜ − 2πRT λ20 ⎝ 2 RT

⎞ ⎟ ⎟ ⎠,

(32)

where λ0 is wavelength, corresponding to the line center, с is light speed, M is molar mass of neon, R is gas constant, and T gas temperature.

164

I.M. Ulanov and M.V. Isupov

At self-absorption, spectral radiance of the source surface b(λ,λ0) is determined by expression (in approximation of a uniform plasma layer with thickness y0)

b(λ , λ0 ) = B(1 − exp(−k (λ , λ0 ) y0 ) , k(λ,λ0) ≈(gj/gk)λ04Ajk(8πc)-1Nkφ(λ,λ0)

(33)

Thus, due to analysis of the circuit shapes for neon spectral lines (under the considered conditions of discharge glow), we can determine the following parameters of TCTD plasma: gas temperature T and population of four lower excited states 2p53s, two out of which are metastable. According to analysis performed, for the studied conditions of TCTD glow the average gas temperature is ~700–800 К, and total atomic concentration on 2p53s levels (4 levels, 2 out of them are metastable) is ~(4–8)٠1011 cm-3. The pilot samples of neon induction lamps with the power of 100÷2000 W and efficiency of 20÷30 lm/W were developed on the basis of investigations performed (fig. 50, 51).

Figure 50. Induction neon lamp of 500-W power.

Figure 51. Induction “demonstration” lamp of 2 000-W power.

Induction Transformer Coupled Discharges: Investigation and Application

165

CONCLUSIONS The integrated experimental investigations of transformer coupled toroidal discharges, aimed at development of new electrodeless generators of low-temperature plasma, plasmachemical reactors and gas-discharge light sources were carried out. One of the main achievements of this research is development of the transformer plasmatron, which can operate at the atmospheric pressures of the plasma-forming gas (argon, air) [6]. The thermal-physical and electric-physical characteristics of the transformer coupled toroidal discharge were measured in argon, air, and argon mixtures with hydrogen, oxygen, natural gas and carbon dioxide. Reactions of plasmachemical synthesis of ozone, nitrogen monoxide and syngas were studied within the gas-discharge chamber of the transformer plasmatron. Formation of high (superequilibrium) concentration of nitrogen monoxide (up to 70%) was determined within the peripheral low-temperature region of the vortex flow, what stabilizes the channel of a transformer discharge. The physical model explaining this effect was suggested [7]. The process of thermal production of nitrogen from oxygen-argon plasma was studied experimentally with the use of recuperative quenching methods. It was shown that at cooling of oxygen-argon plasma with the rates of ~107 К/s, from 5 to 10% of oxygen could transform into ozone at oxygen content in argon from 15 to 3 %, respectively. The electric-physical and spectral characteristics of the transformer coupled toroidal discharge were studied in mercury vapors and neon with the purpose of development of the new electrodeless light sources [21 – 23]. It is shown that for the pressure of mercury vapour above 10 kPa and ratio L/D > 15, the electrical characteristics of TCTD can be approximately calculated by the standard “channel” model of the DC arc discharges [23]. However, the range of the middle pressures of mercury vapor demonstrates the unknown effect of decreasing emission to the triplet with a potential of 7.73 eV for increasing discharge power. The qualitative hypothesis, explaining this effect, is suggested. Based on the obtained experimental results, we have developed and made the pilot samples of electrodeless lamps: electrodeless mercury high intensity discharge lamps with an luminous efficiency of 60 lm/W, and power of 3÷100 kW; germicide UV induction lamps of a low pressure with the power of 50,100, 200 and 500W and luminous efficiency to the resonant line 253.7 nm at the level of ~30–35% of consumed power, neon induction lamps with the power of 100÷2000 W and luminous efficiency of 20÷30 lm/W.

REFERENCES [1] [2] [3] [4] [5]

Heinrich, F. B.; Shevel’ko, V. P. Introduction to Physics of Highly Charged Ions.CRC Press, 2003. Bell, W.E. Applied Physics Letters. 1965, Vol. 7, 190–191. Eckert, H. U. AIAA Journal. 1971, Vol. 9, 1452–1456. Eckert, H. U. IEEE Transactions on Plasma Science. 1974,Vol. PS–2, 308–309. Goldfarb, V. M.; Donskoy, A.V.; Dresvin, S.V.; Rezvov, V.A. High Temperature. 1979, Vol. 17, 698–702.

166 [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

[27] [28] [29] [30] [31] [32] [33] [34]

I.M. Ulanov and M.V. Isupov Kogan, V. A.; Ulanov, I. M. High Temperature. 1993, Vol. 31, 129–135. Kolmakov, K.N.; Ulanov, I. M.; Predtechensky, M. R.; Prikhodko, V. G. Thermophysics and Aeromechanics. 2000, Vol. 7, 419–427. Shabalin, A. Plasma Sources Sci Technol. 2004, Vol. 13, 588–593. Kulumbaev, E. B.; Lelevkin, V. M. High Temperature. 1997, Vol. 35, 351–356. Kulumbaev, E. B.; Lelevkin, V. M. High Temperature. 1999, Vol. 37, 187–193. Gonzales, J. J.; Shabalin, A. Plasma Sources Sci. Technol. 2003, Vol. 12, 317–323. Reinberg, A. R. Inductively Coupled Discharge for Plasma Etching and Resist Stripping. 1984. US patent 4.431.898. Cox, M. S. Toroidal plasma source for plasma processing. 2002. US patent 6.418.874. Zhang, B. C.; Cross, R. C. J. Vac. Sci. Technol. A. 1998, Vol. 16, 2016–2020. Anderson, J. M. Illuminating Engineering. 1969, Vol. 64, 236–244. Anderson, J. M. Electrodeless Gaseous Electric Discharge Devices Utilizing Ferrite Cores. 1970. US Patent 3.500.118. Anderson, J. M. High Intensity Discharge Lamp Geometries. 1979. US patent 4.180.763. Godyak, V. A. High intensity electrodeless low pressure light source driven by a transformer core arrangement. 1998. US patent 5.834.905. Curry, J. J.; Lister, G. G.; J.E. Lawler. J. Phys. D.: Appl. Phys. 2002. Vol. 35, 2945– 2953. Didenko, A. N.; Ulanov, I. M.; Predtechensky, M. R.; Kolmakov, K. N. Phys.—Dok., 2000. Vol. 45, 155–157. Isupov, M. V.; Ulanov, I. M.; Litvintsev, A.Yu. High Temperature. 2004, Vol. 42, 682– 688. Isupov, M. V.; Ulanov, I. M. High Temperature. 2005. Vol. 43, 169–176. Ulanov, I. M.; Isupov, M. V.; Litvinsev, A. Yu. J. Phys. D.: Appl. Phys. 2007, Vol. 40, 4561–4567. Grey, J.; Jacobs, P. F.; Sherman, M. P. Rev. Sci. Instr. 1962. Vol 33, 738–741. Ambrazyavichus. Heat Transfer at Gas Quenching. Vilnius: Nauka, 1983 (in Russian). Ulanov, I. M.; Litvinsev, A. Yu.; Mischenko, P. A.; Krotov, S. V. Proc. of the Russian (International) Conference “Physics of Low-Temperature Plasma–2007”. Petrozavodsk. 2007, Vol. 1, 240–245. (in Russian). M. F. Zhukov and A.S. Anshakov. Electric-Arc Generators with Inter-Electrode Insertions. Novosibirsk: Nauka, 1981. (in Russian). O.E. Skadchenko. Investigation of Ozone Formation in a Jet of Low-Temperature Plasma. Thesis. Moscow, 1972. (in Russian). Zeldovich, Ya. B.; Sadovnikov, P.Ya.; Frank-Kamenetsky, D. A. Oxidation of Nitrogen at Combustion. Moscow. Izd. AN SSSR, 1947. (in Russian). Polak, L. S. Kinetics and Thermodynamics of Chemical Reactions in Low-Temperature Plasma. Moscow: Nauka, 1965. (in Russian). P.R. Ammann, R.S. Timmins. A. I. Ch. E. Journal. 1966, Vol. 12, 956–963. Zhivotov, V. K.; Rusanov, V. D.; Fridman, A. A. Diagnostics of Non-Equilibrium Chemically Active Plasma. Moscow: Energoatomizdat, 1985. (in Russian). Potapkin, B.V. Chemistry of High Energies. 1983, Vol. 17. 524–530. (in Russian). Polak, L. S. Application of Plasma in Chemical Reactions. Moscow: Nauka, 1970. (in Russian).

Induction Transformer Coupled Discharges: Investigation and Application

167

[35] Parkhomenko, V. D.; Soroka, P. I.; Krasnoutsky, Yu. I. Plasmachemical Technology. Vol. 4. Low-Temperature Plasma. Novosibirsk: Nauka, 1991. (in Russian). [36] Klarfeld, B. N. Zh. Tekh. Fiz. 1937, Vol. 7, 1017–1038. (in Russian). [37] Elenbaas, W. The High Pressure Mercury Vapor Discharge. Amsterdam: NorthHolland, 1951. [38] Rokhlin, G. N. Discharge Light Sources. Moscow: Energoatomizdat, 1991. (in Russian). [39] Klarfeld, B. N.; Taraskov, I. M. Technical Physics. 1934. Vol. 4, 504–514. [40] Piejak, R.; Godyak, V.; Alexandrovich, B. J. Appl. Phys. 2001, Vol. 89, 3590–3593.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 169-194

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 4

FEATURES ON THE HIGH FREQUENCY DIELECTRIC RESPONSE IN FERROELECTRIC MATERIALS J. D. S. Guerra* Grupo de Ferroelétricos e Materiais Multifuncionais, Instituto de Física, Universidade Federal de Uberlândia, 38400-902 Uberlândia-MG, Brazil.

Single crystal and/or polycrystalline ferroelectric materials show a high frequency dielectric dispersion, which has been attributed as well to a dispersive (relaxation like) as a resonant mechanism. Physical properties such as relaxation and/or resonant motion mechanisms can be investigated by analyzing the complex dielectric permittivity (real, ε’ and imaginary component, ε’’) in a broad spectral frequency range (100 MHz–13 GHz). Especially, for classical (or ‘normal’) and relaxor ferroelectric systems a dielectric response indistinguishable of dispersion or a resonance mechanism has been found in the literature. The occurrence of such common dispersion process in so different kinds of ferroelectric systems has encouraged the development of several mutually excluding models to explain this physical phenomenon. Nevertheless, the reported results are not conclusive enough to clearly distinguish each mechanism. In this work, a detailed study of the dielectric dispersion phenomenon, including the microwave frequencies, carried out in perovskite structure-type ferroelectric systems, for ‘normal’ and/or relaxor compositions, is presented. The dielectric response in “virgin” and poled state have been investigated taking into account the relative direction between the measuring direction and the orientation of the macroscopic polarization. Results revealed that the dielectric response in ferroelectric systems may be described as a general mechanism related to an “over-damped” resonant process rather than a simple relaxation-like dielectric behavior.

1. INTRODUCTION Since the late 1950s decade [1], the investigation of the high frequency dielectric properties of dielectric materials has been one of the most challenging tasks in the field of the

*

Corresponding author: E-mail: [email protected]

170

J. D. S. Guerra

Physics Condensed Matter [1]. Microwave measurements of high dielectric permittivity materials (> 100) will certainly assist in materials selection for high frequencies applications such as high-speed data transmission and dielectric resonators devices [2]. Specifically ferroelectric materials, which have relatively high dielectric permittivity, have led to a rising interest in the gigahertz region and have prompted research for microwave range application due to its excellent ferroelectric and piezoelectric properties [3]. The dielectric dispersion (frequency dependence of the dielectric parameters, such as, real, ε’ and imaginary component, ε’’ of the dielectric permittivity), as shown in the schematic representation of the Figure 1, can provide information on the dynamics of the mechanism of phase transitions and shows the frequency region where the ferroelectrics are useful for practical applications [4, 5]. The parameters εs and ε∞ in Figure 1 are the static (high frequency) and optical (high frequency) dielectric permittivities, respectively.

Figure 1. Representative curves for the high frequency dielectric response, showing the frequency dependence of the real (ε’) and imaginary (ε’’) component of the dielectric permittivity.

The study of the polarization mechanisms related to the observed behavior requires the ability to work over both large frequency and temperature ranges, were coaxial cells and network analyzers are commonly used [6]. Thus, for the study of orientation and relaxation polarization effects, automated measuring equipments are usually applied and commercially available today, using the reflectometric method by coaxial technique [7]. Most of the microwave dielectric studies have been performed in a wide temperature range from room temperature up to 900 K by using a coaxial line and a discrete varying frequency (resonant cavity-RC) and/or continuously varying frequency (CVF) method as the principal techniques [7, 8]. For instance, in the case of incipient ferroelectric materials, like Ca modified SrTiO3 [9, 10], for which a clear interpretation of the dielectric results [11] it is believed is lacking, it is interesting to investigate the dielectric relaxation mechanisms mainly at low temperatures (below room temperature) and high frequencies. However, for high permittivity and high dielectric loss materials, which is the case of ferroelectrics, the dielectric coefficients become difficult to be measured [12]. For temperature-dependent measurements, specifically at low temperature region (T < 300 K), the coaxial line of the high frequency analyzer and the sample holder must be thermally decoupled because of the thermal contraction/expansion effects of the coaxial line-sample

Features on the High Frequency Dielectric Response in Ferroelectric Materials

171

system. Thus, the dielectric response may result mainly from spurious resonance effects of the system when CVF methods are used, which may often to mask the true dielectric behavior of the sample under test. In this way, the best control of the dielectric measurements may be obtained by adjusting the pressure on the sample. Adjusting the pressure it is possible to guarantee measurement of the true dielectric response of the material under test when decreasing temperature. Within the most commonly used ferroelectric materials, the barium titanate (BaTiO3 BT) system and others perovskite-type structure ferroelectric materials have been distinguished because they presented intrinsic dispersion phenomenon near 1 GHz [3, 13]. This behavior can limit their use specifically in communication systems and electrically controlled devices, such as, phase shifters [2]. In this context, such dielectric dispersions were firstly observed in the so-called ‘normal’ ferroelectrics [14], being characterized later in relaxor ferroelectrics (frequently named as relaxors) [15] and more recently in antiferroelectrics [16] and incipient ferroelectrics [17]. However, it is well known that ‘normal’ ferroelectrics have micro-sized polar regions (domains) with long-range order [18], while relaxors and doped-incipient ferroelectrics have nanometric polar regions (nanodomains) distributed in a non-polar matrix, in which a short-range order prevails [19]. Therefore, the occurrence of such common dispersion process in so different kinds of ferroelectric systems has encouraged the development of several mutually-excluding models to explain this physical phenomenon [15, 20, 21-23], in terms of either a grain or ferroelectric domain resonance [24, 25], as well as the correlation between the ferroelectric polar structures and their respective dynamical response [26]. Previous investigations on both ceramics and single crystals of BaTiO3 [14] and others perovskite-type structure ferroelectric materials [27] suggest the presence of a large dielectric dispersion in the gigahertz region, which was reported to be close to a dipolar character (that is to say, like a Debye-type relaxation). In spite of not being still clarified, several theoretical attempts have been proposed to explain the origin of this effect. The relation between the high frequency dielectric dispersion and the microstructure of ceramics was expressed by Von Hippel [28], who attributed the piezoelectric resonance of the grain as the principal cause of the observed microwave anomalies. Kittel [29] suggested that the domain walls motion have an inertial component. He attributed the decrease of the dielectric permittivity at the microwave region to the resonance of the domain walls in ceramics. Some others models based on piezoelectric resonance of individual domains [30] or on correlated hopping of offcentered ferroelectric active ions between several potential wells [23] have been proposed in more recent years. In all the cases, the structural disorder on the atomic scale, assuming the two-minimum potential relief for some lattice ions, was used as the principal cause for the high frequency dielectric dispersion. As observed, the dielectric dispersion in ferroelectric materials has been a well investigated subject. However, more recent experimental results [31] showed evidences of a resonant dielectric behavior, rather than the previous observed dielectric dispersion. It has been shown, that the resonant behavior may coexist, in the same studied systems, together with the dispersion behavior, under certain conditions. Therefore, it seems that the high frequency dielectric response in ferroelectric materials, observed near to 1 GHz, is dictated by a universal mechanism involving a resonant response (damped or overdamped), which is until now a not fully clarified matter. The objective of the present work is to show a detailed review concerning the high frequency dielectric dispersion in ferroelectric systems, by considering the obtained results

172

J. D. S. Guerra

for ‘normal’ and relaxor ferroelectric ceramic compositions with perovskite type structure. A measuring system suitable for investigation of the high frequency dielectric dispersion, over a wide temperature range from 90 K to 450 K, is presented. Construction and correct calibration procedures are also described, including a brief description of the experimental setup. On the other hand, the present work proposes the fulfillment of the nature of the microwave dielectric dispersion in ferroelectrics materials viewing a better understanding of the dispersion and/or resonant behavior verified in ‘normal’ ferroelectric systems at high frequencies.

2. EXPERIMENTAL DESCRIPTION Dielectric measurements can be commonly carried out by using high frequency impedance analyzers. Here, will be described measurements carried out using a Network Analyzer HP-8719C in the frequency range of 50 MHz to 2 GHz. Technical reasons associated to the system configuration limited accurate measurements above the mentioned frequency range. Indeed, at higher frequencies (> 2 GHz), due to natural resonance of the experimental setup, only measurements at discrete frequencies can be obtained, although the signal in the analyzer can be generated up to higher frequencies. The dielectric response is obtained by the reflectometric method where the reflection coefficients (real and imaginary components: Γ’ and Γ’’, respectively) versus frequency are measured. The setup, currently known by coaxial probe method, is shown in the Figure 2.

Figure 2. Experimental diagram for dielectric characterization using the reflectometric technique by the coaxial probe method.

The sample is kept at rest on the end of the probe. However, in the case of very high dielectric permittivity (ε’ > 100) materials this method presents some disadvantages: a- with the increase of the frequency the modulus of the reflection coefficient ⏐Γ⏐ becomes close to 1 and its phase φ close to zero. Therefore, such materials are difficult to be measured and the

Features on the High Frequency Dielectric Response in Ferroelectric Materials

173

quality of the results largely depends on the initial calibration of the analyzer; b- air gaps formed between the sample and the inner/outer conductors cause very large measurement errors when the sample is badly tooled and c- for high temperatures different thermal expansions of the sample and inner/outer conductors also can bring errors generated by the air gaps formation. Thus, for reducing these limitations and successful characterize high dielectric permittivity materials, as is the case of ferroelectrics, an alternative coaxial line method has been considered. In order to explore a large frequency spectrum of dielectric properties of ceramic materials, CVF measurements involving the case of a coaxial line and the reflectometric technique are appropriate [7]. The coaxial line, including commercial coaxial semi-rigid connectors, provides a suitable link between the measuring port of the network analyzer and the temperature-controlled sample holder, as shown in Figure 3.

Figure 3. Experimental diagram for dielectric characterization using the reflectometric technique by the coaxial line method.

The coaxial line method can be schematically presented by a simple configuration as shown in Figure 4a. The system is based on a microwave generator signal, a signal detector and a divider system located into the network analyzer and a 50 Ω coaxial connector coupled to the coaxial line being terminated by the capacitance of a cylindrical sample. By using the reflectometric method, the coaxial line (see Figure 3) can be characterized in terms of the network theory as a terminated two-part network as shown in Figure 4b. The measured reflection coefficient Γ* (=Γ’ + jΓ’’) is fundamentally affected by the absorption and phase shift of the signal inside the line, and can be described by the S parameters (S11, S22, S12 and S21) of the network. Therefore, in order to determine the frequency dependence of the reflection coefficient of a sample, the coaxial line has to be previously calibrated. Thus, careful compensation procedures have to be carried out to (i) eliminate spurious reflections that may result by transmission line discontinuities, and also (ii)

174

J. D. S. Guerra

reduces stray resistances and stray capacitances in the sample holder. Standard terminations (open, short and 50 Ω) were used over all the operating frequency range at room temperature. Because of the special construction of the system, the calibration performed at room temperature was assumed to be valid over the whole temperature range. It is important to point out that identical dimensions of the sample and standard terminations are necessary to avoid faults by different measuring planes and different stray fields. Thus, reflection coefficients data corrected for the line effects can be frequency scanned and automatically transferred from the network analyzer to the computer. By using the relation between the reflection coefficient and the admittance at the line Y = Yo

1 − Γ* (where Υo=1/Zo) and 1 + Γ*

taking into account the complex dielectric permittivity (ε=ε’–jε’’), the real and imaginary component of dielectric permittivity are obtained by using the Eqs. (1) and (2) [12], respectively,

Figure 4. Experimental diagram of reflectometric method used by the network analyzer.

ε'=

⎤ 1 ⎡ − 2Γ ' ' ⎢ 2 2⎥ Af ⎣ (1 + Γ') + Γ' ' ⎦

(1)

ε''=

1 ⎡ 1 − Γ ' 2 − Γ ' '2 ⎤ ⎢ ⎥ Af ⎣ (1 + Γ')2 + Γ' '2 ⎦

(2)

where f is the measuring frequency; A is determined by the characteristic impedance Zo of the analyzer ( A = 2ε o

π 2r 2 t

Z o ); r and t are the radius and thickness of the sample,

Features on the High Frequency Dielectric Response in Ferroelectric Materials

175

respectively, and εo is the dielectric permittivity of the free space. To obtain more precise measurements, the sample dimensions should have (preferably) cylindrical symmetry with a diameter smaller than the diameter of the central conductor of the coaxial line, in order to avoid border effects and to ensure the electric field only in axial direction inside the sample. Heating of the sample can be achieved by a conventional resistive heater, which can be screwed onto the top of the coaxial line and coupled to a temperature control Flyever FE50RP, as shown in Figure 3. Finally, this measurement system can be inserted into a cryogenic system, which allows to perform experiments under either high vacuum level or nitrogen atmosphere to covers the low temperature ranges. The coaxial line can be inserted inside a silicate boron tube, in order to isolate the line of the liquid nitrogen. Gassy nitrogen can be placed inside the silicate boron tube by a gassy nitrogen input to account for a best good control of the cooling and heating processes. Air gaps at the interface between the coaxial line and the sample were avoided by an optimal sample polishing process, in order to guarantee parallel and flat faces. Alumina powder of 1 μm was used in this polishing process. Gold electrodes were sputtered on the faces of the discs with 2.0 mm in diameter and 0.5 mm in thickness to insure good electrical contacts, for all the investigated samples.

3. HIGH FREQUENCY DIELECTRIC MEASUREMENTS 3.1. Microwave Dielectric Characterization in Low Permittivity Materials (Al2O3)

40.0

90 K 150 K 200 K 300 K Ref. [32] Ref. [33]

20.0

1.0

40 30 20 10 0

300 K

2x10

9

4x10

9

6x10

9

0.5 ε''

ε'

30.0

ε'

Firstly, an alumina (Al2O3) ceramic disk (2.0 mm diameter and 0.5 mm thickness) was used for the dielectric characterization in the range of 50 Hz to 2 GHz at low temperatures. The Al2O3 ceramic was chosen as reference material for the measurements to successfully check the correct calibration of the experimental set-up in the low temperature region. Figure 5 shows the dielectric response of the alumina ceramic obtained at four selected temperatures (90 K, 150 K, 200 K and 300 K). As can be seen, the obtained data were characterized by a high stability of the dielectric parameters in the whole frequency interval. Also, the real component of the dielectric permittivity was found to be around 9.5, in agreement with previously reported results in the literature (ε’ ≈ 9) [32, 33].

0.0

10.0 HF

LF 4

10

5

10

6

7

8

10 10 10 10 Frequency (Hz)

9

176

J. D. S. Guerra

Figure 5. Dielectric response for alumina ceramic including the low (LF) and high frequency (HF) range (100 Hz–2 GHz), at room temperature.

For comparison, low frequency dielectric data were measured using an HP-4194A Impedance/Gain-Phase Analyzer (100 Hz–10 MHz), and the results have been added to microwave properties as shown in the same Figure 5. Since Al2O3 ceramic materials do not display a dielectric dispersion phenomenon [34] (frequency independent dielectric permittivity), the smooth continuity in the values of real and imaginary component of dielectric permittivity for low and high frequency ranges confirms the accuracy of the results obtained at the microwave frequency range. The ε’ and ε’’ values are again in agreement with those found in the literature [32, 33].

3.2. Microwave Dielectric Characterization in High Dielectric Permittivity Materials In order to demonstrate the range of applicability of the experimental setup and to illustrate its capabilities when investigating high dielectric permittivity materials, dielectric measurements were carried out in Sr1-xCaxTiO3 (x=0.1) quantum paraelectric ceramic samples, hereafter labeled as SCT–90/10, as a function of the frequency (50 MHz–2 GHz) and temperature (90 K–450 K).

10.0

90 K 150 K 200 K 250 K 300 K 360 K 430 K

2

ε' ( 10 )

8.0 6.0 4.0 2.0 0.0 2.0

SCT-90/10

2

ε'' ( 10 )

1.5 1.0 0.5 0.0 8

10

9

10

Frequency (Hz)

Features on the High Frequency Dielectric Response in Ferroelectric Materials

177

Figure 6. High frequency dependence of the real and imaginary component of dielectric permittivity (ε’ and ε’’, respectively) for the SCT quantum paraelectric system, at several temperatures above the transition temperature.

In this case, due to additional effects, related to the piezoelectric phenomenon characteristic in ferroelectric systems, a careful control of the pressure on the sample should be taken into account. The adjustment and control of the pressure on the sample was made when decreasing temperature because loss of contacts between the line and the sample holder originated by the thermal contraction of the different elements. Thus, the dielectric response of the sample was obtained at each measured temperature for a critical pressure level on the sample, sufficient to guarantee the electrical contacts. In this case, the frequency dependence of the dielectric properties (ε’ and ε’’) was carried out at various temperatures. Figure 6 shows such dependences for the temperature and frequency range of 90 K–450 K and 50 MHz–2 GHz, respectively. As can be observed, the frequency dependence of ε’ and ε’’ revealed the characteristic behavior of a dielectric dispersion, i.e., a decrease in the real component of the dielectric permittivity and a maximum in the imaginary one at the characteristic frequency (fR). The fR value lies in the order of 700 MHz, in agreement with the reported values obtained for some perovskite-type structure materials [15, 24]. Similarly to the BaTiO3 type ferroelectric system [35], the dielectric dispersion does exist for temperatures above the transition temperature, which suggests the existence of polar regions in the paraelectric state [36], and will be discussed in the next sections. Parallely, low frequency dielectric measurements were performed at the frequencies of 10 kHz, 100 kHz and 1 MHz, in the same temperature interval. The results, shown in Figure 7, do not present any dielectric dispersion in this low frequency range, evidencing a good agreement between the low and high frequency results.

3.3. Microwave Dielectric Characterization in ‘Normal’ and Relaxor Ferroelectrics The present section treats about the investigation of the high frequency dielectric dispersion processes in Pb1-xLaxTiO3 ferroelectrics ceramics for ‘normal’ (x=0.15) and relaxor (x=0.27) compositions (hereafter labeled as PLT–15 and PLT–27, respectively), in order to better understand the occurrence of such common dielectric dispersion process in so different kinds of ferroelectric systems. 16.0 90 K 200 K 300 K

2

ε' ( 10 )

12.0 8.0

LF

HF

4.0 0.0 10

4

10

5

10

6

10

7

10

8

Frequency (Hz)

10

9

178

J. D. S. Guerra

Figure 7. Low and high frequency dielectric dispersion for the SCT system for temperatures of 90 K, 200 K and 300 K, respectively.

Figure 8 shows the frequency dependence of the real, ε’ and imaginary, ε’’ component of the dielectric permittivity for the PLT–15 ceramics, at different temperatures. The Curie temperature of the samples, TC~393 K, was determined by the dielectric measurements of ε’ at low frequencies (1 kHz). A dielectric response indistinguishable from a dispersive behavior was observed in the whole analyzed temperature interval. As can be seen, for a fixed temperature (i.e. at 300 K) and in the frequency range of 70–400 MHz, ε’ decreases slightly as the frequency increases. Above 400 MHz, ε’ decreases quickly, whereas ε’’ passes through its maximum value. The frequency corresponding to the maximum of imaginary component of the dielectric permittivity (fR=700 MHz) is known as the characteristic frequency of the dispersive process, and it is associated to a polarization mechanism responsible for the dissipation. The obtained value for the characteristic frequency is in agreement with those reported for other ferroelectric materials commonly used for microwave applications [24, 37]. 1.5

0.6

3

ε' ( 10 )

PLT-15

0.0

1.2 3

300K 390K 420K 440K 445K

ε'' ( 10 )

1.2

0.9 0.6 0.3

8

10

9

10

0.0

Frequency (Hz) Figure 8. Frequency dependence of the real (ε’) and imaginary (ε’’) component of the dielectric permittivity for the PLT–15 ceramics, as a function of the temperature.

The origin of this behavior has been attributed to the ferroelectric domain walls vibrations in the ferroelectric material [38]. In this way, the domain wall motion is well known to contribute to the polarization of ferroelectric systems. The frequency of the domain walls vibration (obtained at the Gigahertz region) may be observed by applying an alternate electric field of very high frequency. For frequencies appreciably lesser than fR, the ferroelectric domains contribute their full share to the polarization so that the real component of the dielectric permittivity becomes equal to the static dielectric permittivity (εs) and, therefore, the losses (associated to the imaginary component of the dielectric permittivity) vanish. With the increase of the frequency, the domain vibrations increase and consequently the imaginary component of the dielectric permittivity starts to increase up to its maximum value. On the other hand, for a frequency higher than fR, the domain wall vibrations are no longer able to follow the field variations and the real component of dielectric permittivity approaches its clamping values (ε∞). In this frequency range, therefore, ε’’ passes through its maximum value and continuously decreases for the highest frequency values.

Features on the High Frequency Dielectric Response in Ferroelectric Materials

179

On the other hand, a closer inspection in the data of Figure 8 seems to support the picture of a distributed dipolar relaxation mechanism of the dielectric response. Indeed, the obtained value for the characteristic width (λD) at a half-height for the imaginary dielectric permittivity peak (known as full-width half maximum, FWHM), specifically at room temperature, was found to be around 1.91 decades. This value of the FWHM is higher than the FWHM of a Debye-type peak (λD=1.14 decades) [34]. Therefore, the obtained result reveals that the observed dielectric behavior cannot be associated to a Debye-type relaxation process. Results for the dispersive process obtained in the studied PLT–15 samples (in “virgin” state) clearly show a mechanism with a relaxation times distribution function. The data were fitted taking into account the most useful distribution functions for the investigation of the relaxation processes, as reported in the literature [39], and results revealed that the obtained dielectric response was found to be close to a Davidson-Cole (DC) distribution function [40]. The obtained values for the characteristic parameters, such as, the mean relaxation time (τ) and the relaxation time distribution parameter (β), which represents degrees of divergence from the ‘ideal’ Debye model, were 1.45×10-9 s and 0.75, respectively. Compared to singleexponential Debye behavior corresponding to β=1, values of 0 < β < 1 result in broadened imaginary dielectric permittivity peaks. Smaller the values of β the greater the deviation with respect to Debye-type relaxation. Such deviations from Debye behavior are commonly ascribed to a distribution of relaxation times arising from disorder [40]. The obtained mean relaxation time is in good agreement with those obtained for similar ferroelectric systems, as reported in previous works [15]. As can be seen in Figure 8, the dielectric dispersion occurs not only in the ferroelectric region (TTC) close to TC. This behavior, as pointed out in a recent work [26], suggests the existence of polar regions at temperatures higher than the transition temperature, which could be one more evidence of the contribution for an “order-disorder” type paraelectric-ferroelectric phase transition in ABO3 perovskite structures, as reported in the literature [41, 42]. The existence of an order-disorder component in the paraelectric-ferroelectric phase transition of PbTiO3 explains the persistence, although weak, of the dielectric dispersions above TC [26]. It is important to point out that, for higher temperatures, the data also revealed highest values for the FWHM, showing a distributed dipolar relaxation mechanism for all the investigated temperature range. At the same time, a decrease in the distribution parameter (β), as the temperature increases, confirms that the dielectric spectrum becomes extremely diffuse. Correspondingly, the distribution of relaxation times becomes extremely wide. As can be observed, similarly to the results reported by other authors for BaTiO3 [22, 35], PbTiO3 derived materials [8] or some tetragonal tungsten bronze (TTB) structure systems [43], the obtained dielectric response curves suggest a dispersive mechanism of the complex dielectric permittivity, without any evidence of a resonant response. Figure 9 depicts representative curves of the frequency dependence of ε’ and ε’’ for the PLT–27 (relaxor composition) measured at different temperatures below and above the respective temperature of the maximum of the dielectric permittivity, Tm. The data also reveal strong temperature dependent dielectric dispersions for this relaxor ferroelectric composition, not only for temperatures below the temperature of the maximum dielectric permittivity, but also for temperature above Tm.

J. D. S. Guerra

2.0 0.0

3.0 2.4 1.8 1.2

PLT-27

3

226K 259K 290K 325K 350K

3

ε' ( 10 )

4.0

ε'' ( 10 )

180

0.6 0.0 8

10

9

10

Frequency (Hz) Figure 9. Frequency dependence of the real (ε’) and imaginary (ε’’) component of the dielectric permittivity for the PLT–27 ceramics, as a function of the temperature.

The high frequency dielectric dispersion can be related to a microscopic polarization mechanism and, therefore, a possible explanation for the observed behavior could be found taking into account atomic level considerations. Indeed, the mechanism for this dielectric behavior is associated with a shift of Ti4+ (the B-site ion in the perovskite structure) inside the oxygen octahedron. The ion shifts from one potential to the other in a double well potential model (relaxation motion), as shown in the Figure 10a, being ΔU the high of the potential barrier [44]. In fact, such mechanism can be represented as real chains (or volumes) of correlation (named as correlation chains) including a large number of Ti4+ ion cooperative jumps, which must be considered in the relaxation mechanisms. These chains are isolated from each other by several types of defects (vacancies, impurities, point defects, etc.) (Figure 10b).

Figure 10. Double potential model (a) and correlation chains scheme (b) for the high frequency dielectric response.

Features on the High Frequency Dielectric Response in Ferroelectric Materials

181

In order to obtain additional information of the dielectric spectrum, the obtained results were fitted by using the Eqs. (3) and (4), which consider a damped harmonic oscillator model [45], where the frequency dependence of the complex dielectric permittivity is governed by the Eqs. (3) and (4).

ε ' = A+

ε '' =

BωR2 (ωR2 − ω 2 )

(ω

2 R

−ω

)

2 2

+γ ω 2

(3)

2

BωR2 γω

(4)

(ωR2 − ω 2 ) + γ 2ω 2 2

A and B are constant parameters, which are directly related with the dielectric dispersion parameters. For instance, B=εs–ε∞, defined as dielectric strength (Δε), is the contribution to the dielectric dispersion of the static dielectric permittivity, εs; A=ε∞ is the contribution to the dielectric permittivity of the higher frequency electronic processes; ω=2πf is the angular frequency and

1

τ

fR =

ωR , is the characteristic frequency of the process (defined as, 2π

= 2π f R , where τ is the mean relaxation time). The dispersion characteristic parameters,

such as, Δε, fR and γ (the damping coefficient of the dielectric response), can be obtained directly from the fitting of the obtained dielectric response (frequency dependence of ε’ and ε’’) with the Eqs. (3) and (4), which undoubtedly might be an essential feature in order to identify the mechanism involved in the dielectric anomaly. Figure 11 shows the temperature dependence of the characteristic frequency (fR) and the dielectric strength (Δε) for both PLT–15 and PLT–27 compositions. The results reveal that Δε and fR reach simultaneously a maximum and a minimum, respectively, in a temperature that coincides with its respective maximum real dielectric permittivity, TC and Tm for the PLT–15 and PLT–27 compositions, respectively. 2.8 fR: PLT-15

1.0

Δε: PLT-15

0.8 0.6 0.4 0.2

2.4 2.0 1.6

PLT-27 fR: PLT-27 Δε: PLT-27

1.2

3

PLT-15

Δε ( 10 )

fR ( GHz )

1.2

0.8 0.4

240 270 300 330 360 390 420

Temperature (K) Figure 11. Temperature dependence of the characteristic frequency (fR) and dielectric strength (Δε) for the PLT–15 and PLT–27 compositions.

182

J. D. S. Guerra

Referring to the observed results, the essential features can be summarized as follows. athe microwave dielectric dispersion process persists at temperatures higher than the temperature of the maximum dielectric permittivity for both compositions, vanishing in the temperature interval where 1/ε’ vs. T obeys the Curie law (around TB, the Burns temperature, characterized by the temperature of nucleation and slowing down of the ferroelectric domains); b- above Tm, it is noticed that fR is much more temperature dependent for the PLT– 15 than for the PLT–27 composition, vanishing earlier for the former composition; c- the magnitude of the characteristic frequencies (fR) for both compositions are comparable, although their polar structures and correlation length differ significantly [22, 46, 47]. Similar results were found for incipient ferroelectrics and antiferroelectrics [16, 17]. These experimental results allow us to interpret the microwave dielectric process according to the scenario described as follows. First, the fact that the microwave dielectric dispersions became evident even for temperatures far above Tm clearly demonstrates that the simple presence of polar regions, independently of their size, volume fraction and correlation length, is the sufficient condition for the existence of such dielectric dispersion process in perovskite ferroelectric systems. Nevertheless, the only plausible common mechanism inherent in so different ferroelectric domain structures is the boundaries of ferroelectric domains and nanodomains. Therefore, it has been proposed that the field-induced vibration of polar region boundaries of domains (domain walls) and nanodomains (interphase boundaries between the polar region and the non-polar matrix) is the common mechanism responsible for the dielectric dispersion process in the microwave range. The second intriguing question is why such dispersion occurs around the same frequency interval (fR) for all the ferroelectric systems, although their apparent distinct nature. In this context, in analogy with the oscillating membrane theory [38, 48] and in accordance with the side-way motion of the boundaries of the polar regions discussed above, it is suggested that fR is governed by the ratio between the effective force constant (κeff) and the effective mass (Meff.) of the polar regions boundaries, fR ≈

κ eff / M eff . The effective mass is the mass of

domain walls and interphase boundaries for normal ferroelectrics and relaxors, respectively, while the force constant (κeff) is dictate by elastic properties of the respective polar region boundaries [19, 38, 49]. Thus, it is believed that this ratio would have almost the same value for all perovskite ferroelectric systems, justifying the similar values found for fR. For instance, for normal ferroelectrics the higher force constant would be balanced by the higher effective mass of domain walls. On the other hand, for relaxors and incipient ferroelectrics, the relative smaller κeff is balanced by the relative low massive interphase boundaries, thus keeping the ratio reasonably constant. Finally, such proposed relation for fR is also able to explain satisfactory the behavior of fR (slope) from the paraelectric to ferroelectric phase transition. For temperatures higher than TC, the high thermal energy reduces the dipolar interactions contributing for a low κeff, and consequently, a relatively low value for fR. With the decrease of the temperature, the thermal energy decreases in favor of the formation of the polar regions, promoting the increase of Meff, and consequently fR slightly decreases. For temperatures near TC, a sudden increase in the interaction energy, and consequently in the κeff, takes place, resulting in an increase of fR. Indeed, in the case of PLT–15 the paraelectric-ferroelectric phase transition is predominantly

Features on the High Frequency Dielectric Response in Ferroelectric Materials

183

a displacive-type transition. Therefore, it is expected an abrupt crossover between the two structural phases (paraelectric-ferroelectric), which means an accelerated disappearance of the polar regions (domains and domain walls). This fact results in a sudden change in both order parameter and elastic properties, thus reflecting in the force constant. Moreover, the existence of an order-disorder component in the paraelectric-ferroelectric phase transition explains satisfactorily the existence of the dielectric dispersion above Tm. On the other hand, it is well known that relaxor ferroelectrics do not necessarily present a macroscopic structural phase transition through or bellow the temperature of maximum permittivity (Tm). Therefore, for the relaxor composition, it is proposed that the low slope of fR above Tm reflects the gradual condensation and the slowing down of polar nanoregions below TB. Furthermore, the fact the microwave dielectric dispersion is characterized at temperatures much higher than Tm, for the PLT–27 composition, confirm the existence of nanodomains at temperatures much higher than Tm. 20.0

3

ε' ( 10 )

15.0

(a)

TC = 393 K

(b)

300 K 340 K 360 K 390 K 430 K

10.0 5.0 0.0

30.0

3

ε'' ( 10 )

25.0 20.0 15.0 10.0 5.0 0.0

8

10

9

10

Frequency (Hz) Figure 12. Frequency dependence of the complex dielectric permittivity for the PLT–15 samples under a uniaxial stress, at different temperatures.

3.4. Mechanical and Electrical Driving Field Effects 3.4.1. Stress effect analysis Because the dielectric response in ferroelectric materials is strongly susceptible to the influence of electric and/or mechanical fields, the microwave dielectric properties in the PLT–15 ceramics have been now investigated under the influence of a mechanical uniaxial

184

J. D. S. Guerra

stress, applied parallel to the measurement direction. The dielectric properties were obtained over the same frequency and temperature range to that obtained for the ‘stress free’ samples described in the previous section. The results presented in the Figure 12, clearly show that at room temperature (solid line), ε’ remains essentially flat up to 700 MHz, increases traversing a maximum, and then decreases to its clamped value (see Figure 12a). As can be seen, this anomalous behavior obtained for the stressed samples and indistinguishable from a resonant response, is observed in the whole analyzed temperature range. On the other hand, the stressed samples also exhibited the highest values of the imaginary component of the dielectric permittivity, with its ε’’ peak apparently shifted to a higher frequency (Figure 12b), when compared to that obtained for the ‘stress free’ samples. This dielectric response obtained for the stressed ceramics corresponds to a true resonance rather than a dispersion process, with a characteristic frequency around 900 MHz (at room temperature) higher than the obtained for the ‘stress free’ samples (700 MHz), at the same temperature. It is interesting to point out, from the Figure 12b, that the high values of the maximum imaginary dielectric permittivity for the stressed samples, when compared to the maximum decrease of the real component (εs–ε∞), suggest that is not possible to describe the spectrum as a classical Debye’s dielectric dispersion. At the characteristic frequency (900 MHz), the value of ε’’ at room temperature, is about 2 times higher than that predicted by the Debye’s classical model, [(εs–ε∞)/2], for the stressed samples. This result has an important physical implication; indeed, it clearly confirms that the high frequency anomalies do not correspond to a real dispersion process as predicted by the Debye’s model (relaxation-like behavior). As reported in previous works [21, 22], the assumption that the observed dielectric anomaly could be, in this case, related to a piezoelectric resonance of the grains or individual domains suggests that the main influence of the uniaxial stress would be to decrease the losses related to this process, contrary with the results obtained for the PLT–15 samples. Therefore, the true mechanism responsible for the obtained anomalies remains still not clear. The dispersion characteristic parameters, were now also obtained from the fitting of the obtained dielectric response by using the Eqs. (3) and (4). In order to compare the temperature evolution of the characteristic parameters results for the stressed and ‘stress free’ samples for the PLT–15 ferroelectric ceramics are presented in the Figure 13. As can be observed, the temperature dependence of fR and ∆ε is quite similar for the stressed and the ‘stress free’ samples. So that, with the application of the uniaxial mechanic stress, although the dielectric response evidences resonance characteristics, this behavior remains modulated by the paraelectric-ferroelectric phase transition. As a result, the thermal evolution indicates that, in both cases, the characteristic frequency goes through a minimum while the dielectric strength pass through a maximum around 393 K, which coincides with the paraelectricferroelectric phase transition temperature (TC), as observed in the low frequency dielectric measurements [26]. This result shows that the maximum dielectric dispersion appears near the transition temperature, independently of the applied external mechanical driving field. On the other hand, it is important to point out that the damping coefficient values (γ), obtained from the fitting of the experimental data by using the Eqs. (3) and (4), at room temperature, were lower for the stressed samples (0.45⋅109 s-1), than those obtained for the ‘stress free’ samples (8.80⋅109 s-1). This result suggests that the high frequency dielectric anomalies, dispersion or resonance, can be described as either an over-damped or a damped resonant response, respectively, where the main contribution to the dielectric response can be

Features on the High Frequency Dielectric Response in Ferroelectric Materials

185

associated to the ferroelastic and/or ferroelectric components of the ferroelectric materials, which directly modifies the damping of the system. Referring to the obtained results, the essential aspect to be discussed can be summarized to the feature that with the application of an external mechanic driving field the dielectric response of the system clearly pass from a dispersive to a resonant dielectric behavior. In this way, considering an “over-damped” resonant response, the variations in the strength of the damping coefficient of the system originate mainly due to the contributions of either ferroelastic or ferroelectric dipolar components. For lower damping systems, as the case of the obtained resonant behavior for the stressed samples, the main contribution to the dielectric response is governed by the ferroelectric dipolar component. On the contrary, for higher damping coefficient systems, the major contribution of the dispersive behavior (as observed in the “stress free” samples) is due to the ferroelastic dipolar component, which prevails over the ferroelectric component. On the other hand, it is well known that ordinary ferroelectric materials are classic hybrids ferroics, that is to say, they present a strong coupling between the ferroelectricferroelastic dipolar components [50, 51]. When applying a uniaxial mechanic stress a reorientation of the electric and elastic dipoles takes place. In the same way, a mechanic strain in the material can be observed if applying an electric driving field. Then, as in the case of the stressed samples, applying a uniaxial stress parallel to the measurement direction promotes the dipolar reorientation in the perpendicular direction of the applied uniaxial stress, that is to say, an induced polarization may appears in the perpendicular direction to the applied uniaxial stress direction. Therefore, applying a uniaxial stress parallel to the measurement direction is equivalent to apply a poling electric field in the perpendicular direction of the measurement direction. Thus, in order to obtain a better understanding of the current analysis, the observed anomalies around 1 GHz for the studied PLT–15 ceramics, will be now investigated taking into account the influence of the relative orientation of the macroscopic polarization on the dielectric dispersion in the poled samples. 1.2

6.0 fR ; Δε - stressed fR ; Δε - stress free

4.5 3.0

0.3

1.5

0.0

300

330

360

390

420

3

0.6

Δε ( 10 )

fR (GHz)

0.9

0.0 450

Temperature (K) Figure 13. Temperature dependence of the characteristic parameters (characteristic frequency, fR and dielectric strength, Δε) for the stressed and ‘stress free’ PLT–15 samples.

186

J. D. S. Guerra

3

ε' ( 10 )

3.0

'stress free' samples

(a)

1.5

0.0

Unpoled Poled ( I I ) Poled ( )

(b)

(fR = 700 MHz) (fR = 700 MHz)

3

ε'' ( 10 )

2.0 (fR = 1.03 GHz)

1.0

0.0 8

10

9

10 Frequency (Hz)

Figure 14. Frequency dependence of the complex dielectric permittivity, at room temperature, for the PLT–15 unpoled ‘stress free’ samples, and the poled samples measured in the parallel (//) and perpendicularly (⊥) direction to the poling direction; (a)- real and (b)- imaginary component.

3.4.2. Poling field effect analysis The dielectric response was now performed in poled samples (Ep = 2 kV⋅mm-1) at room temperature, in both parallel and perpendicular directions to the poling field direction and in the same frequency interval, following the same experimental procedure previously described. Thus, the temperature dependence of dielectric permittivity (real and imaginary component) was obtained parallel (ε’//) and perpendicularly (ε’⊥) to the polarization direction. The results are shown in Figure 14. The results reveal that the dielectric response, observed in the sample measured parallel to the poling direction, presents an ‘apparent’ dielectric dispersion process, which in turn is similar to that observed for the unpoled sample (see Figure 8). However, it is observed that the maximum of the imaginary component of the dielectric permittivity (ε’’//), for the sample measured parallel to the poling direction, is around 4 times higher than (εs//–ε∞//)/2. Furthermore, a detailed inspection for the FWHM of the imaginary dielectric permittivity peak for the sample measured parallel to the poling direction revealed that the observed dielectric dispersion has not a relaxation-like character, since the FWHM of the peak was found to be around 0.85 decades. This value is lower than the characteristic width of a Debyelike process (λD=1.14 decades, the low limit value for FWHM). Indeed, it implies that it is not

Features on the High Frequency Dielectric Response in Ferroelectric Materials

187

possible to describe the observed dielectric spectrum as a classic Debye’s dielectric relaxation, as previously discussed in the section 3.4.1. Therefore, it is possible to affirm that the observed dielectric anomaly really correspond to a resonant-like dispersion rather than a simple relaxation-like dielectric behavior. Figure 14 also displays the dielectric response in the sample measured perpendicularly to the poling direction. The results show that ε’⊥ slightly increases up to 500 MHz. However, after that, it abruptly increases and subsequently decreases for higher frequencies to its clamped values. The maximum of the imaginary component of dielectric permittivity (ε”⊥) also presents a remarkable high value, now observed around 1.03 GHz. The imaginary component of the dielectric permittivity for the sample characterized perpendicular to the poling direction is about 7 times higher than the values predicted by the theoretical Debye’s model. These results show that the dielectric response obtained for the samples measured perpendicularly to the poling field is characterized by a resonant mechanism, which may be associated to a damped resonance process rather than to a simple Debye’s dielectric relaxation. The above described experimental results clearly show that in the same sample, depending on the relative orientation between the measuring direction and the macroscopic polarization direction, the dielectric dispersion in the GHz region seems to behave as either a dispersive (relaxation-like) or a resonant mechanism. This change in the dielectric response can be well described considering a general resonant behavior, which can be associated with an “over-damped” resonance process. Therefore, it can be affirmed that the microwave dielectric spectra observed in ferroelectric materials may be described in terms of an “overdamped” resonance involving either dispersive or resonant behavior, rather than a simple dispersion process (relaxation-like), which are intimately related to the variation in the damping strength of the system. At the same time, the damping strength is affected by the coupling between the ferroelectric and ferroelastic dipolar components. The values of γ were now obtained for the poled samples, by the fitting of the experimental data, and showed in the Table I, together with those obtained for the ‘stress free’ and stressed samples. As observed, the damping coefficient also shows a decrease when the dielectric behavior pass from a dispersive (relaxation like) behavior to a pure resonant response. As can be seen, the γ value for the poled samples measured in the parallel direction to the poling direction was around 8.97⋅109 s-1, similar to that obtained for the ‘stress free’ samples and, at the same time, higher than those obtained for the poled samples measured in the perpendicular direction to the poling direction. Table 1. Dispersion characteristic parameters obtained for the PLT–15 samples, at room temperature. Sample Stress free Stressed Poled // Poled ⊥

ε∞ 113 266 145 149

Δε 453 691 300 541

fR (GHz) 0.70 1.44 0.70 1.04

γ (109 s-1) 8.80 0.45 8.97 1.15

188

J. D. S. Guerra

Thus, from the obtained results, it is possible to affirm that the dispersive behavior observed for the ‘stress free’ samples may be a consequence of an increase of the damping strength, giving rise an over-damped resonance (reflected by the dispersive behavior). On the other hand, the resonant behavior observed for the poled samples measured in the perpendicular direction to the poling direction, as well as in the stressed samples, may be a consequence of a decrease of the damping strength, giving rise a damped resonance (reflected by the resonant behavior). It is important to point out that although the dielectric behavior observed for the samples measured in the parallel direction to the poling direction has a true resonant-like character, from the previously described analyses, an “apparent” dispersion behavior (relaxation-like) was observed because of the high component of the damping coefficient. Therefore, the gigahertz dielectric anomalies observed in normal ferroelectrics must be interpreted as a resonance instead of a dispersion mechanism. The question as to whether these high frequency anomalies behave as a dispersive (relaxation-like) or a resonant process, described, in general, by an “over-damped” mechanism, may be a consequence of the coupling between the ferroelectric and the ferroelastic dipolar components, whose contribution determines the character of the high frequency dielectric dispersion. This aspect will be discussed in details in the next section.

3.4.3. Discussion The ferroelectric phenomenon, in most of cases, may be follows by a ferroelastic behavior, that is to say, a ‘non-intentional’ stress can be present when an electric field E is applied. As in ferroelectric materials, a ferroelastic crystal contains two or several stable orientation states when there is no mechanical stress. It is possible to change reversibly from one state to another by applying a stress σ in defined directions. Thus, there exists a strainstress elastic hysteresis with spontaneous strain ss and a coercive stress. In this way, a transition from a ferroelastic phase to a higher temperature phase, termed paraelastic, occurs by an increase in symmetry and a change in the crystalline system [52]. The additional ‘nonintentional’ stress originated when applying an electric field may be sufficient for ferroelastic switching to occur [53]. The basic assumption of the polarization switching model is that a single ferroelectric crystallite in a polycrystalline ceramic, which is subjected to an electric field [54] or to a mechanical stress [55, 56] or both, undergoes a polarization change and a corresponding strain change. Nevertheless, either partial or complete coupling may exist between the ferroelectric and ferroelastic properties. When such coupling occurs, ss may be modified by applying an electric field and Ps may be modified by applying a mechanical stress. This leads to a set of four hysteresis cycles: polarization–electric field; strain–stress; strain–electric field; and polarization–stress, as shown in Figure 15 [44]. Crystal growth is usually accompanied by the formation of ferroelastic and/or ferroelectric domains. The crystal can be made monodomain by applying a mechanical stress (ferroelasticity), an electric field (ferroelectricity), or through one of the external actions if coupling occurs between the two properties. Depending upon the crystal structure and the order parameter, the ferroelectricity and ferroelasticity in a crystal could be fully or partially coupled.

Features on the High Frequency Dielectric Response in Ferroelectric Materials

189

Figure 15. Schematic representation of the hysteresis loops for ferroelectric-ferroelastic phases [44].

Concerning to the obtained results for the PLT–15 ceramics, it can be noted that two polarization mechanisms are always present under certain conditions. For all the samples where an external driving field was applied (a uniaxial stress and/or a poling field applied in the perpendicular and parallel direction to the measurement direction) a resonant-like dispersion has been observed. In particular, either uniaxial stress or poling field applied in the perpendicular direction to the measurement direction promotes redistribution in the orientation of the dipolar configuration over the perpendicular direction to the measurement direction. Therefore, the observed resonant behavior, for these conditions, could be associated to a slightly partial coupling between the ferroelastic and ferroelectric dipolar components, being the ferroelectric dipolar component the main responsible for the resonance dielectric response. On the other hand, for the unpoled and ‘stress free’ samples the observed behavior may be caused by the coupling between the ferroelectric and the ferroelastic dipolar components. In these conditions, the dipoles are unable to be reoriented ‘freely’ with the measurements ac electric fields. Therefore, the dielectric response is mainly governed by the contribution of the ferroelastic dipolar components, originating the dispersive dielectric behavior. In order to describe these results, the ferroelastic contributions to the piezoelectric response should be taken into account [57]. In this case, it will be considered the theoretical approach in order to investigate the influence of the external poling electric field on the dielectric response. Thus, the analyses of the data are considered when an oscillating external electric field is applied parallel or perpendicular to the poling direction. The piezoelectric effect can be represented by the Eq. (5), where c and e are the converse of the piezoelectric compliance (or elastic stiffness) and the piezoelectric tensor, respectively, defined as

190

J. D. S. Guerra

[e] = [d ][c E ] [53]. The symbols ν, μ and m correspond to the condensed index of the tensor notation (ν,μ = 1, 2,…, 6 and m = 1, 2, 3). E σ ν = cνμ sμ + emν Em

(5)

In view of the symmetry for a poled ceramic material (∞mm) and considering the respective symmetry operations, the Eq. (5) can be expanded by considering the probing electric fields applied parallel and perpendicular to the poling direction (E1 and E3, respectively) and represented by the following equations:

σ1 = c11s1 + c12 s2 + c13 s3 + e31E3

(6)

σ 2 = c12 s1 + c11s2 + c13 s3 + e31E3

(7)

σ 3 = c13 s1 + c13 s2 + c33 s3 + e33 E3

(8)

σ 5 = c55 s5 + e15 E1

(9)

Some important aspects of the piezoelectric equations (Eqs. (6) – (9)) must be highlighted when applying a probing electric field in two mutually perpendicular directions. First, when the probing electric field (E3) is applied parallel to the poling direction (P3), it generates only compressional/extensional stress (σ1, σ2, or σ3). In contrast, when an electric field (E1) is applied perpendicular to poling direction, only shear stress (σ5) is generated. The fundamental difference between both cases is that stress generated by the electric field E3 produces volumetric changes in the ceramic, while fields applied perpendicularly does not (E1⊥P3) [58]. In other words, ferroelastic dipoles are modified only by the application of an electric probing field E3, which is parallel to the macroscopic polarization. It is important to point out that the elastic stiffness c55, in the Eq. (9), is identical to the coefficient c44, after considering the respective symmetry operations. In terms of the elasticity theory, the dispersion or resonant behavior might be associated to the influence of the electric field on the elastic dipole. In this way, in order to clarify this issue it is necessary to use some basic elasticity concepts. The behavior of an elastic dipole in the presence of stress can be characterized by the relation

σν = cνμ λμ , being λ the strain tensor, which determines the interaction of

the elastic dipole with the stress field [58]. Formally, σ is defined as the negative stresses needed to maintain constant the strain per unit concentration of elastic dipoles. Since the λ tensor represents a strain tensor, it must be symmetric and, therefore, can be characterized by a strain ellipsoid with three mutually perpendicular principal axes. When expressed in the coordinate system of the principal axes, the λ tensor becomes diagonal with the three components λ1, λ2 and λ3 as the principal values. By expanding the relation of the elastic

Features on the High Frequency Dielectric Response in Ferroelectric Materials

191

dipole definition, it is not difficult to notice that results do not depend of the shear components of stress field. It suggests that the resonant behavior obtained for the perpendicular direction is governed by the electrical component of the Eq. (5), where only the shears components of the piezoelectric coefficients play the principal role, without any contribution of the components along the principal axes (σ1, σ2 and σ3). Considering that elastic dipoles in displacive ferroelectrics lie parallel to the electric ones, and that they interact only with the compressional or extensional stress [58], they are excited only by the application of the poling electric field parallel to the measurements direction. This suggests that the resonant response obtained for the perpendicular direction is governed only by ferroelectric components, because only the shear stress components are generated by piezoelectric contribution, without any ferroelastic contribution. Otherwise, the dielectric behavior (associated to an over-damped resonance) observed in the parallel direction results from a coupling between the ferroelastic and ferroelectric contributions. In this case, the ferroelastic contribution becomes the main responsible for the increase of the damping coefficient, contributing to the decrease of the characteristic frequency. It is important to point out that applying a uniaxial stress, parallel to the measurements direction, the resonant response observed can be explained considering that the ferroelectric and ferroelastic domains tend to be oriented perpendicularly to the mechanical stress and, consequently, perpendicular to the measuring electric field direction. As observed, this condition is analogous to the case of applying a poling electric field perpendicular to the measurements direction. Thus, either dispersion or resonance mechanism, are always presented without the presence of an intermediate case. This result confirm that the ferroelectric and ferroelastic contributions, to the high frequency dielectric anomalies, are always coupled each other.

4. CONCLUSIONS In summary, the dielectric microwave properties were investigated in relaxor and normal ferroelectric ceramics. It was concluded that the vibration of the boundaries of polar regions is a common mechanism responsible for the microwave dielectric dispersion process in ferroelectric systems. It was also proposed that the characteristic frequency is controlled by the ratio between the force constant and the effective mass of such boundaries and the behavior of fR above the temperature of the maximum of the permittivity reflects the ferroelectric-type phase transition. On the other hand, the microwave dielectric response of ferroelectric ceramics was investigated considering the influence of external (electric and mechanic) driving fields. Two high frequency dielectric anomalies were found in the same studied material, which were discussed in lights of an over-damped resonant process. It was confirmed that the obtained anomalies are strongly influenced by the contribution of the ferroelectric and/or ferroelastic dipolar components, characteristics of ferroelectric materials. In particular, for all the samples where an external driving field was applied (a uniaxial stress and/or a poling field applied in the perpendicular and parallel direction to the measurement direction) a resonant-like dispersion was observed. On the contrary, a dispersive (relaxationlike) behavior related to an over-damped resonance was obtained for the unpoled (and ‘stress free’) samples. The results can be well explained by considering the influence of the

192

J. D. S. Guerra

ferroelastic-ferroelectric contributions coupling on the high frequency dielectric response, which is a common feature for all ferroelectric systems.

ACKNOWLEDGMENT The author would like to thank to Ferroeletric Ceramics Group (GCFerr), of the Federal University of São Carlos (UFSCar) for experimental support, especially to Sr. F. J. Picon (GCFerr) for the technical assistance, and FAPESP (proc. No. 04/09612-0) Brazilian agency for financial support. The financial support by the ICTP to the Latin-American Network of Ferroelectric Materials (NET-43) is also gratefully acknowledged.

REFERENCES [1] [2] [3] [4] [5]

[6] [7] [8] [9] [10] [11] [12]

[13] [14] [15] [16]

Hippel, AV. Dielectric Material and Applications; Technology Press books in Science and Technology; John Wiley & Sons: New York, 1954, Vol. 1, pp 3-40. Babbitt, R; Koscica, T; Drach, W; Didomenico, L. Integrated Ferroelectrics, 1995, 8, 65-76. Zimmermang, F; Voigts, M; Weil, C; Jakoby, R; Wang, P; Menesklou, W; Ivers-Tiffée, E. J Eur Ceram Soc., 2001, 21, 2019-2023. Treece, RE; Thompson, JB; Mueller, CH; Rivkin, T; Cromar, MW. IEEE Trans Appl Supercond, 1997, 7 2363-2366. Grigas, J. Microwave Dielectric Spectroscopy of Ferroelectrics and Related Materials; Ferroelectricity and Related Phenomena Series; Gordon and Breach Publisher: Amsterdam, 1996, Vol. 9, pp 9-108. Bassora, LA; Eiras, JA. Ferroelectrics, 1999, 223, 285-292. Haraoubia, B; Meury, JL; LeTraon, A. J Phys E, 1988, 21, 456-460. Mouhsen, A; Achour, ME; Miane, JL; Ravez, J. Eur Phys J Appl Phys, 2001, 15, 97-104. Muller, KA; Burkard, H. Phys Rev B, 1979, 19, 3593-3602. Viana, R; Lunkenheimer, P; Hemberger, J; Böhmer, R; Loidl, A. Phys Rev B, 1994, 50, 601-604. Venturini, EL; Sâmara, GA; Kleemann, W. Phys Rev B, 2003, 67, 214102. Bassora, LA. Caracterização dielétrica de materiais ferroelétricos na região de microondas, Ph. D. Thesis, Federal University of São Carlos, São Paulo, Brazil, 1999, pp 10-90. Böhmer, R; Maglione, M; Lunkenheimer, P; Loidl, A. J Appl Phys, 1989, 65, 901-904. Poplavko, YM; Tsykalov, VG; Molchanov, VI. Sov Phys Solid State, 1969, 10, 2708-2710. Kamba, S; Bovtun, V; Petzelt, J; Rychetsky, I; Mizaras, R; Brilingas, A; Banys, J; Grigas, J; Kosec, M. J Phys Condens Matter, 2000, 12, 497-519. Lente, MH; Guerra, JDS; Eiras, JA; Lanfredi, S. Solid State Commun, 2004, 131, 279282.

Features on the High Frequency Dielectric Response in Ferroelectric Materials

193

[17] Lente, MH; Guerra, JDS; Eiras, JA; Mazon, T; Andreeta, MBR; Hernandes, AC. J Eur Ceram Soc, 2005, 25, 2563-2566. [18] Cao, WW; Cross, LE. Phys Rev B, 1991, 44, 5-12. [19] Glazounov, AE; Tagantsev, AK. J Phys Condens Matter, 1998, 10, 8863-8880. [20] Arlt, G; Böttger, U; Witte, S. Appl Phys Lett, 1993, 63, 602-xxx. [21] Hippel, AV. Rev Mod Phys, 1950, 22, 221-237. [22] McNeal, MP; Jang, SJ; Newnham, RE. J Appl Phys, 1998, 83, 3288-3297. [23] Zhang, L; Zhong, WL; Wang, CL; Zhang, PL; Wang, YG. Solid State Commun, 1998, 107, 769-773. [24] Maglione, M; Böhmer, R; Loidl, A; Höchli, UT. Phys Rev B, 1989, 40, 11441-11444. [25] Kazaoui, S; Ravez, J; Miane, JL. J Non-Cryst Sol, 1991, 131, 1202-1205. [26] Guerra, JDS; Lente, MH; Eiras, JA. Appl Phys Lett, 2006, 88, 102905. [27] Kersten, O; Rost, A; Schimidt, G. Phys Stat Sol (a), 1983, 75, 495-500. [28] Hippel, AVZ. Physik, 1952, 133, 158-173. [29] Kittel, C. Phys Rev., 1951, 83, 458-458. [30] Turik, AV; Shevchenko, NB. Phys Stat Sol (b), 1979, 95, 585-592. [31] Tappe, S; Böttger, U; Waser, R. Appl Phys Lett, 2004, 85, 624-626. [32] Jiang, GQ; Wong, WH; Raskovich, EY; Clark, WG. Rev Sci Instrum, 1993, 64, 1614-1621. [33] Jiang, GQ; Wong, WH; Raskovich, EY; Clark, WG. Rev Sci Instrum, 1993, 64, 1622-1626. [34] Jonscher, AK. J Phys D Appl Phys, 1999, 32, 57-70. [35] Kazaoui, S; Ravez, J; Elissalde, C; Maglione, M. Ferroelectrics, 1992, 135, 85-99. [36] Petzelt, J; Ostapchuk, T; Gregora, I; Rychetsky, I; Hoffmann-Eifert, S; Pronin, AV; Yuzyuk, Y; Gorshunov, BP; Kamba, S; Bovtun, V; Pokorny, J; Savinov, M; Porokhonskyy, V; Rafaja, D; Vanek, P; Almeida, A; Chaves, MR; Volkov, AA; Dressel, M; Waser, R. Phys Rev B, 2001, 64, 184111. [37] Böttger, U; Arlt, G. Ferroelectrics, 1992, 127, 95-100. [38] Arlt, G; Pertsev, NA. J Appl Phys, 1991, 70, 2283-2289. [39] Garcia, MF; M’Peko, JC; Ruiz, AR; Rodríguez, G; Echevarría, Y; Fernández, F; Delgado, A. J Chem Educ, 2003, 80, 1062-1073. [40] Barsoukov, E; Macdonald, JR. Impedance Spectroscopy: Theory, Experiment, and Applications; John Wiley & Son,s: New Jersey, 2005; Vol. 1, pp 27-42. [41] Baskaran, N; Ghule, A; Bhongale, C; Murugan, R; Chan, H. J Appl Phys, 2002, 91, 10038-10043. [42] Zalar, B; Laguta, VV; Blinc, R. Phys Rev Lett., 2003, 90, 037601. [43] Hornebecq, V; Elissalde, C; Porokhonskyy, V; Bovtun, V; Petzelt, J; Gregora, I; Maglione, M; Ravez, J. J Phys Chem Sol, 2003, 64, 471-476. [44] Ravez, J. Chemistry, 2000, 3, 267-283. [45] Feynman, RP; Leighton, RB; Sands, ML. The Feynman Lectures on Physics; Commemorative Issue; Addison-Wesley Publishing Co: Redwood, 1989, Vol. 2. [46] Tagantsev, AK; Glazounov, AE. Phys Rev B, 1998, 57, 18-21. [47] Bolten, D; Böttger, U; Waser, R. J Eur Ceram., Soc, 2004, 24, 725-732. [48] Fousek, J; Brezina, B. J Phys Soc Jpn, 1964, 19, 830-838. [49] Robels, U; Arlt, G. J Appl Phys, 1993, 73, 3454-3460. [50] Lu, W; Fang, D. N; Li, CQ; Hwang, KC. Acta Mater, 1999, 47, 2913-2926.

194 [51] [52] [53] [54] [55]

J. D. S. Guerra

Hwang, SC; Waser, R. Acta Mater, 2000, 48, 3271-3282. Aizu, K. J Phys Soc Jpn, 1969, 27, 387-396. Abplanalp, M; Fousek, J; Günter, P. Phys Rev Lett, 2001, 86, 5799-5802. Hwang, SC; Lynch, CS; McMeeking, RM. Acta Metall Mater, 1995, 43, 2073-2084. Hwang, SC; Huber, JE; McMeeking, RM; Fleck, NA. J Appl Phys, 1998, 84, 1530-1540. [56] Wooster, WA; Wooster, N. Nature, 1946, 157, 405-406. [57] Nye, JF. Physical Properties of Crystals: Their Representation by Tensors and Matrices; Oxford University Press: New York, 1985; Vol. 1, pp 3-168. [58] Nowick, AS; Berry, BS. Anelastic Relaxation in Crystalline Solids; Materials Science Series; Academic Press: New York, 1972; Vol. 1, pp 156-224.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 195-216

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 5

THE PRINCIPLE THAT GENERATES CONFIGURATION IN ANIMATE AND INANIMATE SYSTEMS – A UNIFIED VIEW Antonio F. Miguel Geophysics Centre of Evora, Rua Romão Ramalho 59, 7000-671 Evora, Portugal Department of Physics, University of Evora, PO Box 94, 7002-554 Evora, Portugal

ABSTRACT The generation of flow configuration (shape, structure, patterns) is a phenomenon that occurs across the board, in animate and inanimate flow systems. Scientists have struggled to understand the origins of this phenomenon. What determines the geometry of natural flow systems? Is geometry a characteristic of natural flow systems? Are they following the rule of any law? Here we show that the emergence of configuration in animate flow systems is analogous to the emergence of configuration in inanimate flow systems, and that features can be put on a unifying theoretical (physics) basis, which is provided by the constructal law. All scientific endeavors are based on the existence of universality, which manifests itself in diverse ways. Here we also explore the idea that complex flow systems with similar functions have a propensity to exhibit similar behavior. Based on this thought relationships that characterize animate systems are tested in relation to cities and countries, and some conclusions are drawn.

1. NATURE'S SHAPES From the ancient time people have struggled to explain why animals, plants, rivers, etc., are shaped the way we find it. Democritus (460-370 B.C.) attributed natural configuration (shape, structure, patterns) to "chance and necessity." [1]. Aristotle (384-322 B.C.) claims that shape is more truly than matter. He wrote: “Into what then does it grow? Not into that from which it arose but into that to which it tends. The shape then is nature” [2]. In other words, nature is the shape of a fully matured natural object. Therefore, we should understand not

196

Antonio F. Miguel

only matter but especially the shape. He exemplifies “as the teeth, for example, grow by necessity, the front ones sharp, adapted for dividing, and the grinders flat, and serviceable for masticating the food”. In 1860 Ralph Waldo Emerson (1803-1882) published the essay entitled "The Conduct of Life". He wrote [3] "Nature has her own best mode of doing each thing, and She has somewhere told it plainly, if we will keep our eyes and ears open." In Kant's (1724-1804) book entitled "Critique of Judgment " he argues that “the forms of nature are so manifold (…) that there must be laws for these forms (...), if they are to be called laws (as the concept of nature requires), they must be regarded as necessary virtue of a principle of the unity of the manifold” [4]. It has been more than 250 years since Pierre de Maupertuis (1698-1759) wrote "(…) laws of movement and of rest deduced from this principle being precisely the same as those observed in nature, we can admire the application of it to all phenomena. The movement of animals, the vegetative growth of plants ... are only its consequences; and the spectacle of the universe becomes so much the grander, so much more beautiful, the worthier of its Author, when one knows that a small number of laws, most wisely established, suffice for all movements. (…)" [5]. This “principle” - the “principle of least action” [6] – was properly stated later by scientists like William Hamilton. It came from his idea that the very perfection of the universe demands a certain economy in nature and is opposed to any needless expenditure of energy. In spite of having a wide applicability in mechanics, electricity, magnetism and quantum mechanics, this principle only accounts for point-to-point motion but cannot describe point-to-area and point-to-volume flows [6]. In 1776, Jean Meusnier's study of soap films (very popular among mathematicians of eighteen century) showed an example in geometric optimization [7]. An ordinary twodimensional plane can be twisted infinitely into a helicoid shape (minimal surface) that has the delicate balance of a soap film. Forty years later, Robert Stirling (1790-1878) patented his “Heat Economiser” and Sadi Carnot (1796-1832) wrote about the ideal configuration of a heat engine. Charles Darwin (1809–1882) busied himself with the idea that the configuration of living systems is determined by evolution and natural selection [8]. Josiah Willard Gibbs (1839-1903) in the late 19th century refers that a thermodynamic system to be in an equilibrium state it will configure its components by minimizing the energy. In 1917, D'Arcy Wentworth Thompson (1860-1948) published the book “On Growth and Form”. Thompson introduced the idea that they are principles among quite diverse forms of life. From the observation that the bones of a museum skeleton would lie in a heap on the floor without the clamps and rods pulling them together, he concluded that the tension holds the skeleton together as much as weight does. He wrote that "the form of any particle of matter, whether it be living or dead, and the changes in form which are apparent in its movements and in its growth, may in all cases be described as due to the action of force" [9]. Forces of tension, compression and shear occurred in all living structures and influenced both growth and function. Eighty later, Helbing and Molnar [10] use the concept of “force” to explain the motion of pedestrians (the so-called “social forces model”). In 1975, Benoit Mandelbrot introduced the term fractal and wrote “(…) a fractal is a shape made of parts similar to the whole in some way (…).” [11]. Fractal based description of natural systems has been widely applied. In spite of its incontestable importance, fractals do not account for dynamics hence are descriptive rather than predictive. Natural flow systems are complex and change (evolve) in many ways. Why things in nature are shaped the way they are? Why do tree-shaped designs occur in river basins, deltas,

The Principle that Generates Configuration in Animate and Inanimate...

197

lightning and lungs? Is it an optimized behavior? Why stony corals and bacterial colonies present intraspecific variability of patterns? What are the “forces” that shape their growth and form? Adrian Bejan is at the origin of the constructal paradigm, which had its start in 1996 [1214]. He arrived at this idea from a problem of minimizing the thermal resistance between an entire heat generating volume and one point [13]. The occurrence of configuration in nature is a physics phenomenon, and the constructal theory is about the physics principle from which configuration can be deduced. It is based on the following law (constructal law): “For a finitesize system to persist in time (to live), it must evolve in such a way that it provides easier access to the imposed (global) currents that flow through it.” [13]. This law states that there is a time arrow “associated with the sequence of flow configurations that constitutes the existence of the system”[14,15] (Figure 1). Besides, the system shape and structure do not develop by chance but result from the permanent struggle for better performance. Better performance means minimization and balance of the resistances (i.e., imperfection) faced by the various internal and external streams under the existing constraints. The morphim structure is the result of optimal distribution of imperfection.

Figure 1. Time arrow in plant growth [18]

Configuration plays a fundamental role in models used for “perceiving” and “understanding” nature. Here, it is shown how the constructal theory provides a unifying picture for the development of flow architectures in natural systems with internal flows. We also show that the complexity is optimized and is a result of the optimization process.

2. THE CONSTRUCTAL LAW A flow system (animate and inanimate) is a nonequilibrium thermodynamic system [1315]. Classical thermodynamics is not concerned with the configurations of its nonequilibrium systems. Basically, the first law of thermodynamics is a statement of the conservation of energy, the second law is a statement about the direction of that conservation (i.e., the

198

Antonio F. Miguel

maximization of entropy generation in an isolated system), and the third law is a statement about reaching absolute zero. The constructal law is a law of thermodynamics distinct of these three [16]. Likewise the second law it contains an arrow of time but of different things. The second law proclaims the entropy of a closed system always increases in time up to an equilibrium state. The constructal law states that the flow architecture morphs in time in the direction of flowing more easily (i.e., the maximization of flow access). The analogy between the formalism of equilibrium thermodynamics and that of constructal theory is presented by [16,17]. Constructal theory begins with the global objective(s) and the global constraint(s) of the flow system [12-17]. There are two global constraints, one external and the other internal: the external constraint is the system size, and the internal global constraint is the "amount" invested in making the flow architecture (e.g., total volume of all the ducts of the flow structure). How do we identify the configuration that brings flow architecture to the best performance? According with constructal law, the architecture of the system must “provides easier access to the imposed (global) currents that flow through it.” [13]. Consider, for example, the case of a fluid that has to be drained from a finite-size area or volume. The volume is a nonhomogeneous porous medium composed of a material of low permeability (high resistance) and various layers of higher permeabilities. The goal is the optimal arrangement of these layers such that the global flow resistance is minimal. The global constraints are the system size (area or volume to be drained) and the total size occupied by the layers. Based on this, the arrangement is optimized in order to reach maximum performance. In this way, the designer “constructs” the optimal flow architecture. A detailed description is provided by [13,14]. The acquisition of shape (architecture) is an evolutionary process not assumed in advance or postulated. Therefore, the flow architecture (shape) of the system is the result of the optimum balance between two competing trends – slow (high resistivity) and fast (low resistivity). If we increase the length of high permeability layer leads to a decrease in the resistance posed to flow in the area (volume) occupied by low permeability layers, but it also increases the resistance along the high permeability layer. Therefore, flow optimization is as a trade-off between competing trends [13-16]. This example is not unique. Behind a broad class of processes in the natural sciences is the dynamics that combines, for example, Brownian motion (diffusion) with some form of deterministic drift (convection). Here we shall consider a convection–diffusion equation of the general form

∂n + ∇.(Un ) = D∇ 2 n ∂t

(1)

where n denotes the density, U is the velocity and D is the diffusion coefficient. The spreading of a tracer or a solute, and the transport of heat or fluid are examples that can be analyzed within the framework of diffusive-convective phenomena. Diffusion is associated with a high resistivity mechanism, whereas convection is low resistivity. Why the flow architecture of systems results of the balance between both mechanisms? By applying scale analysis [13] to Eq. (1), we obtain the time scales for diffusion (tdf) and convection (tcv)

The Principle that Generates Configuration in Animate and Inanimate...

199

t df ~

L2 D

(2)

t cv ~

L u

(3)

Diffusion coefficients are usually much smaller than 1 m2 s-1 (for example, the diffusion coefficient for oxygen in air is approximately 2 × 10-5 m2 s-1). For very short distances tdf < tcv and diffusion is the “best” driving mechanism because is “faster”. On the other hand, for larger distances tdf > tcv and convection perform better. The time of transition from diffusion to organized flow is t* ~ D/u2. This simple analysis shows us that the two modes of flowing with imperfection (resistance) should exist in order to enable the best flow architecture. In summary, optimization means finding the best allocation of resistances (i.e., minimum imperfection to the global flow architecture), and therefore the configuration of the system is the one that allows best flow access. To make our point, we illustrate next the application of the constructal law to a variety of animate and inanimate flow systems.

3. CONSTRUCTAL THEORY OF INANIMATE NATURAL FLOW SYSTEMS The phenomenon of generation of flow configuration is everywhere in inanimate flow systems. Agglomerates of aerosol particles often have dendritic shapes instead of spherical shapes [18]. The reason why this occurs was not clear. Reis et al. [19] relied on the constructal law of maximization of flow access in order to construct a theory of aerosol agglomeration. Based on the idea that there are not electrically neutral surfaces in contact with air [18,19], it is assumed that the forces that make aerosol particles stick onto previously deposited particles are of the electrical type. There are two possible “modes” of agglomeration: spherical and conically. The volume of spherical agglomerates is given by

Vsph ~ K 2 t 2

(4)

while the volume of agglomerates of particles with the conical shape is 1/ 2

Vcon

⎛ q el K14 / 3 ⎞ ⎟⎟ ~ ⎜⎜ μ el ⎝ ⎠

t7 / 3

(5)

where μel is the dipole moment, qel is the charge and K is a quantity function of particle size, dipole moment, electric charge, Cunningham correction factor, electric permittivity of the air, surface density of charge and air viscosity [18]. The constructal law is simply about the physics meaning of the time direction of configuration evolution. Therefore, the architecture of the aggregate of particles evolves in time in such a way that the global rate of accumulation of the particles is maximized. In other words, the best configuration is the one that bring the entire flow system (ambient and particles) to equilibrium in the fastest way possible.

200

Antonio F. Miguel

Figure 2. Time evolution of the volume corresponding to conical and spherical agglomerates

The temporal evolution of the accumulation volume is presented in Figure 2. This plot shows that at the critical time, tct, the volume of conical agglomerates overtakes the volume of spherical agglomerates. According to Eqs. (4) and (5) 1/ 3

⎛ μ ⎞ t ct ~ ⎜⎜ 1 / 2 el 1 / 3 ⎟⎟ ⎝ q el el K ⎠

(6)

This means that the agglomerate must first grow as a sphere (t < tct) and then change to a conical shape. Existing flow configuration (spherical) is replaced by a configuration (conical, tree-shaped) that flow more easily. Experimental measurements reported in the literature confirm the main features of this constructal development [18]. The constructal law also predicted configuration generation during liquid droplet impact on a wall [20]. For small and slow enough droplets the splat comes to rest viscously, as a disk. On the other hand, for large and fast enough droplets, it splashes by developing needles that grow radially until they are arrested by viscous effects. Based on constructal theory, Bejan and Gobin [20] gave comprehensive explanation of configuration generation and also present a dimensionless group that governs the selection of geometry. This group is defined by the ratio of two lengths, the final radius of the disc that dies viscously, divided by the radius of the still inviscid ring that just wrinkles. The results of the optimization process match the observed values. A river basin is the portion of land drained by a river and its tributaries. It encompasses the entire land surface dissected and drained by many streams and creeks that flow downhill into one another, and eventually into one river. The final destination is an estuary or an ocean. A river basin is an example of an area-to-point flow. Based on constructal law, Bejan [13] has addressed this problem and optimized the channel network that minimizes the overall resistance (imperfection) to flow. Consider the area allocated to each smallest stream of the river basin. Rain falls uniformly on the elemental area with a mass flow rate. There is an optimal elemental shape defined as the ratio between the length and width such that the total flow rate collected on the elemental area flows with least global flow resistance from the area through one point on its periphery. The optimized area element becomes a building block with which larger rain plains can be covered. The tree-shaped flow architecture of the river

The Principle that Generates Configuration in Animate and Inanimate...

201

basin represents the best allocation of resistances (optimal distribution of imperfection), and therefore the configuration of the system that allows best flow access from the area to the outlet [13,18]. The flow architectures of river basins are not the result of change but they constitute the optimal configuration. River basins have geometrical features which can be measured, namely the area (measured on the vertical projection), the elongation ratio (the diameter of a circle with the same area as the drainage basin, divided by the basin maximum length), the relief (the difference of elevation between the highest and the lowest points of the drainage area) and the relief ratio (the basin relief divided by the maximum length of the basin). Based on experimental data of geometric characteristics, authors proposed several allometric scaling laws to describe feature of river basins [18]. The well known Horton, Melton and Hack allometric laws can be anticipated based on constructal law as a result of minimization of the overall resistance to flow. Reis [21] shows that the: (i) ratio of constructal lengths of consecutive streams matches Horton's law for the same ratio, while the same occurs with the number of consecutive streams that match the respective Horton's law; (ii) Hack’s scaling law is also anticipated from the constructal law and the exponent β that relates the length of mainstream, Ls, and the basin area, Ab, (Ls~Abβ, β~0.568) presented in a more accurate way (i.e., β=2τ+1/4τ, where τ is the order of the river basin); and Melton's scaling exponent is 2.45 instead of 2. Reis and Gama [22] relied also on the constructal law of maximization of flow access in order to address beachface adjustment as a response to wave swash forcing. They showed that beachface slope varies with wave height raised to the power 3/4, and sand grain size raised to the power 4/3. The largest flow system on earth (atmosphere circulation) was studied from the point of view of the constructal theory [23,24]. The sun–earth–universe assembly was viewed as a power plant the power output of which is used to force the atmosphere and hydrosphere to flow. The constructal optimization was performed in order to deliver the latitude of the boundary between the Haddley and the Ferrel cells, the boundary between the Ferrel and the Polar cells, the average temperature of the Earth surface, the convective conductance in the horizontal direction, as well as other parameters defining the circulation and the Earth surface temperature. The results of this optimization agree very well with the observed values. These results and many other examples show that the constructal law explains much of shape and structure of inanimate flow systems around us. The cracks patterns evolution during shrinking in soil [13,25], the dendritic crystals formed during rapid solidification [13], the turbulent flow structure [13], Rayleigh-Bénard convection [13], and the electrokinetic transfer through porous media [26] may also be mentioned here among other examples.

4. CONSTRUCTAL VIEW OF ANIMATE FLOW SYSTEMS Animate systems are probably the most complex and diverse system in the universe. The so-called “life” covers more than 27 orders of magnitude in mass from molecules of the genetic code to whales and sequoias [27]. To help understand the underlying configuration (architecture) behind animate systems, three different views (concepts) are employed:

202

Antonio F. Miguel (i) An animate system is made up of interacting subsystems that provide a key functional characteristic of the overall system. Subsystems such as the circulatory system, lungs, kidneys, etc. provide a function that is essential to sustain the all system. (ii) Colonies of living organisms (e.g., stony corals and bacterial colonies) refers to several individual organisms of the same species living all time closely together for mutual benefit. (iii) Living organisms that are temporally together as an ongoing group to collectively accomplish a particular purpose (e.g. pedestrians accessing shopping facilities or a stadium).

Systems at these levels of organization demonstrate the complementary nature of structure and function. Therefore, the design of animate systems is not only a matter of molecular biology but also of geometry and physics. To make our point, we review a collection of studies based on constructal theory that explain in a simple manner the configuration that occurs in (i), (ii) and (iii).

4.1. Systems View An animate system is made up of interacting subsystems. Each subsystem provides a key functional characteristic of the overall system. Are they optimized as a whole? Are the subsystems also optimized? The respiratory subsystem was assessed via the constructal principle by Reis et al. [28] and Reis and Miguel [29], with the structure of the respiratory system being the result of this appraisal. The primary function of the respiratory system is the supply of oxygen to the blood and the drainage of carbon dioxide from it, so these in turn delivers oxygen to and remove carbon dioxide from all parts of the body. There are two alternatives for accomplishing this purpose: the lung could be a duct system, or a simple single sac (volume) open to the external air from which the oxygen diffuses to the blood (and carbon dioxide diffuses after being released from the blood). This second possibility is clearly noncompetitive as compared to a duct system: the former has a small access time for duct (convective) flow (Eq. 3) of order tcv ~ 1 s ( u ~ 0.5 m/s) whereas the latter has an access time for a diffusive process (Eq. 2) of order tdf ~ 104 s (L ~ 0.5 m and D ~2 x10-5 m2/s). On the other hand, both solutions have internal imperfections. A duct system has a large friction resistance to airflow whereas the single sac has a large diffusive resistance. Tree-shaped flow architectures are the easiest way to flow between infinity of points (volume, area) and one point or vice-versa [13]. However, a cavity (alveolar sac) at the end of the tree-shaped network should exist, because for very short distances diffusion is the “best” driving mechanism (Section 2) and allows easy oxygenation of the blood from air to the tissues. What are the characteristics of lung configuration to provide the easiest access for blood to oxygen/carbon dioxide? According to the constructal principle, lung configuration must provide the easiest access for blood to oxygen/carbon dioxide under the constraints posed by the space allocated to the respiratory process (chest cavity). Reis et al. [28] and Reis and Miguel [29] show that the configuration that performs these functions at the lowest flow resistance is a tree-shaped flow structure composed of ducts with 23 levels of bifurcation that ends with alveolar sacs from

The Principle that Generates Configuration in Animate and Inanimate...

203

which oxygen diffuses into the tissues. In summary, lung configuration results of the trade-off between two competing trends. The configuration that performs best is the one that results of the “harmony” of the best allocation of resistances to allow the best flow access and not from the smash of any of the competing trends. In these studies [28,29], the dimensions of the alveolar sac, the total length of the airways, the total alveolar surface area and the total resistance to oxygen transport in the respiratory tree were also obtained. One of the most remarkable findings was that there is a length defined by the ratio of the square of the first airway diameter to its length which is constant for all individuals of the same species. Kidneys, vascularised tissues and the nervous system are also examples of optimized architectures that have been treated from the point of view of constructal theory [13,16,30]. It’s not only the subsystems that composed the animate systems that are optimized but also the whole system. In the 1940s, Max Kleiber [31] and Samuel Brody [32], based on observation of mammals and birds, established for the first time that the interspecies correlation of metabolic rate scales as 3/4 power of the body mass. Kleiber's and Brody’s works were generalized by subsequent researchers to intracellular levels, unicellular organisms, and plants, and an exponent of 3/4 is found over 27 orders of magnitude [33]. A direct explanation for the 3/4 scaling power between metabolic rate and body mass can be obtained based on constructal theory. The explanation presented by Bejan [34] was obtained by combining the tree architecture optimized for minimum pumping power and the convective heat transfer characteristics of two identical fluid trees superimposed in counterflow. Constructal theory also anticipates other important empirical allometric laws. A direct explanation for the 1/4 scaling power between breathing or heart beating time and body mass was obtained from the minimization of the pumping power required by the thorax for breathing and the heart for blood circulation. This allometric law derived in [35] is based on the following realistic assumption: the flow is intermittent (in and out) and “in” interval is of the same order of magnitude as the “out” interval. Moreover, Bejan [35] also proved that the heartbeat and the breathing time must be of the same order of magnitude, regardless of body size. The allometric laws of the design of the hair coats of animals are also projected by the constructal theory. By minimizing the combined heat loss by conduction and radiation through the hair air coat, the proportionality between the hair strand diameter and the animal body length scale raised to the power of 1/2 , and the hair coats porosity between 0.95 and 0.99 for all animal sizes were obtained [13,36]. Flow configuration (patterns) is generated in space but also in time. The modes of locomotion (flying, running, swimming) are examples that illustrate the generation of patterns in time. Bejan [13] and Bejan and Marden [37,38] showed that the constructal law (a single law) is able to describe the different modes of locomotion. Flying, running and swimming were attributed to the same principle of configuration generation for greater flow access in time (constructal law) [37,38]. Flying, for example, involves two losses. One loss is the lifting of the body weight (vertical loss due to the gravity) and the other is the horizontal loss due to drag. The total loss per distance travelled is the summation of both losses, i.e.,

Mg 3 / 2 h1 / 2 v −1 + C Dρair v 2 L2b , where M is the body mass , v is the velocity, ρair is the air density, CD is the drag coefficient, h is the vertical distance and Lb is the characteristic

204

Antonio F. Miguel

dimension of the body. The optimal flying speed is obtained by minimizing the total loss, which is provided by the constructal law. Therefore, the flying speed scales as 1/6 power of the body mass and flapping frequencies are proportional to the body mass raised to the power −1/6. These results constitute allometric laws for flying and they are in good agreement with the speeds of flying organisms ranging from the insects to the big birds. The minimization of horizontal (friction) and vertical (gravity) losses is also at origin of running and swimming. More details are provided by [37,38]. Recently, Charles and Bejan [39] collected the heights and weights of the fastest swimmers (100 m-freestyle) and sprinters (100 m-dash) for world record winners, since the beginning of the twentieth century. They then plotted the speed data (based on wining times) versus the size of these athletes. The scaling analyze showed that speed records will continue to be dominated by heavier and taller athletes. According to [39], this tendency is attributable to the scaling rules of animal locomotion [37,38], “not to the contemporary increase in the average body size of humans” [39]. The authors also suggest that if athletes of all sizes are to compete in these kinds of events, weight classes might be needed, like in certain sports, such as judo, boxing, weightlifting or wrestling. The constructal theory was applied to the design of animals but also to predict the morphology of plants. Bejan and co-authors [40] showed, among other things, that the tree length is proportional to the wood mass raised to the power 1/3, the tapered shape of the root and longitudinally uniform diameter and density of internal flow tubes, the near-conical shape of tree trunks and branches, and the existence of an optimal ratio of leaf volume divided by total tree volume. In summary, animate systems are optimized flow architectures. Besides, allometric laws can be viewed as a manifestation of the underlying dynamics that is optimized, and can be anticipated by the constructal theory.

4.2. Colonies of Living Organisms A colony refers to several individual organisms of the same species living closely together, usually for mutual benefit (e.g., formation of colonies helps escaping predators or enhance the ability to locate nutrients) despite some detrimental effects such as getting infected more easily by contagious disease [27]. The formation of dissimilar patterns inside similar colonies under different environmental conditions is especially intriguing. It seems that there is an order underlying the apparent variety of shapes. Several authors noticed that for a given set of growth conditions, the colonies experience similar patterns that can be reproduced from the experimental point of view [40-42]. The shape of bacterial colonies depends both on the nutrient level and gel (environment) hardness [41,42]. Bacterial colonies that cope with hostile environmental conditions (i.e., low level of nutrients or hard agar surface) develop branched forms. On the other hand, colonies that cope with high levels of nutrients develop a compact shape. Stony corals also consist of structure of tightly interconnected individuals (called polyps). These corals collected from exposed growth sites, where higher water currents are found, present more spherical and compact shape than corals of the same species growing in sheltered sites, which display a thin-branched patterns [43]. There are 2 questions that need to be answered: how colony

The Principle that Generates Configuration in Animate and Inanimate...

205

development displays a precise control of the configuration? why colonies develop characteristic patterns? Increasing complexity is associated with an increased use of modulatory communication signals to organize cooperative behavior. These signals do not elicit specific responses in themselves, but rather operate in a general manner to alter the probability that individuals will respond to other stimuli [41,44-46]. Ants and other social insects, for example, usually have a pheromone produced by the endocrine system for marking trails to food [44]. Therefore, pattern formation requires modulatory communication signals (chemical signals, electrical impulses or other) between interconnected individuals. According to [41], communication is on the basis of configuration (pattern) formation in bacterial colonies. This answers our first question but the following question remains: what is the reason to switch to different patterns? The answer is delivered by the constructal law in Miguel [47] and Miguel and Bejan [48]. Bacterial colonies or stony corals may develop branched (tree shaped) or spherical shapes in a differentiated response to the variability of environmental conditions. Branched and round massive patterns have different abilities to fill the same space, and thus a different ability to harvest the nutrients that are available. According to the constructal theory, the survival of flow systems calls for patterns that promote flow access. The preferred pattern is the one that allows the living system to deplete the nutrients as fast as possible. Consider s to be the characteristic external length scale of the living system, and l the branch (needle) length scale width (l<<s). The volumes of branched and spherical patterns scale as ~ l2s and ~s3, respectively. Because l2s << s3, spherical patterns are more effective for filling the space (i.e., the most effective for extracting the available nutrients). In this sense, spherical means perfection, but in reality branched patterns are also likely to occur. How can one reconcile such a contradiction? Consider, for example, stony corals growing at a rate, v, of a few cm/year (for Pocillopora damicornis the growth speed is 1–6 cm/year [47,48]). When the coral grows in a sheltered site, where water currents are practically absent, diffusion is the most important nutrient transport mechanism. The living system starts to grow at birth (t=0). Immediately after, nutrients close to the system are quickly consumed and depleted. The concentration of nutrients decreases, which triggers a wave of nutrients defined by the characteristic linear dimension ~ (Dnt)1/2 (Eq. 2) and the speed of propagation ~ (Dn/t)1/2, where Dn is the diffusion coefficient for nutrients. The initial speed of propagation is greater than any growth speed of the living system, but decreases with the inverse of the square root of time. Consequently, the speed of nutrient propagation drops below the growth speed of the living system when t > Dn/v2. Before t ≤ Dn/v2, the round (massive) shape is the most effective arrangement for filling the flow space. After this time scale, the living system begins to grow outside the nutrient diffusion region (Figure 3). Branches (bio-lanes) develop in order to generate lowresistance paths for nutrient access, in a phenomenon that resembles the respiratory tree. This channeling enables the system to continue to experience growth inside the nutrient rich region from t1critical to t2critical. At times slightly greater than t2critical the coral sticks out of the nutrient region. New branches grow forward in order to access the nutrient region until a new critical time is reached (Figure 4). Each branch generates a new group of branches, and the global feature of this scenario is the dendritic pattern. This configuration represents the most competitive configuration under these circumstances. In summary, existing flow configurations are replaced by configurations that flow more easily.

206

Antonio F. Miguel

A similar explanation can be applied to bacterial colonies. A Petri dish prepared with nutrient agar is the place for the bacteria you'll be growing. The nutrient content and the agar hardness are important parameters because control the colony pattern. Why? The survival of colony calls for patterns that promote easier flow access (constructal law). If the agar surface is soft and nutrient abundant, the colony develops a spherical pattern because, as showed before, is the more effective architecture for filling the space (i.e., the most effective for extracting the available nutrients). In a low-nutrient environment, nutrients close to the system are quickly consumed and depleted. In a hard agar surface, both the resistance of colony and nutrient to movement is high. In both cases, the contact surface of colony should be larger and the bacteria should “follow paths” of less resistance, to continue to experience growth inside the nutrient region. For both a tree shaped pattern represents the most competitive configuration [13], the optimal shape for survival in nature. In summary, corals and bacteria colonies evolve in time as the result of the continuous search for easier flowing configurations. Existing flow configurations are replaced by configurations that flow more easily. Therefore, patterns do not develop by chance when the flow configurations are able to morph in time.

Figure 3. Time evolution of characteristic dimensions of living system and nutrients: the speed of nutrient propagation falls behind the growth speed of the living system at t > Dn/v2

The Principle that Generates Configuration in Animate and Inanimate...

207

Figure 4. Time evolution of characteristic dimension of a living system and critical times: coral branches at critical times t1critical, t2critical and t3critical.

4.3. Living Organisms that are Temporally Together as an Ongoing Group The formation of dissimilar patterns inside a similar group is not an exclusive of coral and bacterial colonies. Walking is the most basic form of transportation, undertaken by almost every citizen. Systematic observations of pedestrian revealed that in standard situations (for instance, running to catch a departing bus; panic situation excluded) pedestrian motion reveals regularities. Figure 5 shows the empirical walking speed – pedestrian density relation according to data from several authors [49]. These data show that pedestrians are able to walk fast and slow. If the interpersonal distance between the pedestrians is large enough (or the density of pedestrians is small enough) the motion is not disturbed and pedestrians walk with a constant walking speed (domain I). For a density higher than ~1 pedestrian/m2, the walking speed decreases with the pedestrian density (domain II). Recently, Miguel [49] suggested that each pedestrian at domain II may be modeled as a “particle” subjected to some “effective forces” arising out of its interaction with the obstacles and the other pedestrians. He consider two sub-domains in this domain II: (i) a sub-domain which the pedestrians’ interpersonal distances are large enough and the deviation from the desired speed is due to a necessary deceleration to adjust their own speed to the speed of neighboring pedestrians, and (ii) a sub-domain which the interpersonal distances between pedestrians are small enough that “repulsive forces” will affect the walking speed. Humans (but also birds, fishes or other) in a group cannot anticipate the behavior of the other elements. The decrease of walking speed in domain II helps to maintain the so-called “territorial effect” and avoid or reduce physical contact with other pedestrians. In this study [49], the optimal flow capacity of a system (i.e., maximum pedestrian flow) and the optimal level of service provided to the pedestrian (i.e., minimum travel time) were also analyzed based on the constructal theory.

208

Antonio F. Miguel

Figure 5. Walking speed versus pedestrian density (empirical data)

Observations performed by different authors [13,50] show that in crowded spaces the movement of individuals self-organizes naturally into lanes with a specific direction. In addition, when a stationary crowd stands in the way and needs to be passed through, pedestrians organize themselves into river-like streams. They act together, more or less automatically, to accomplish a task. How and why they develop these patterns? Such organized collective behavior is based on a coordinated and orchestrated use of communication between pedestrians. Pedestrians rely on their senses to communicate. Similar to pheromones widely used by social insects such as ants, pedestrian must rely on their vision for spatial orientation. Vision guarantees a precise control of the pedestrian position in space and consequently, thereafter, a precise pattern control. But why pedestrians develop specific patterns when confronted with specific situations? Consider pedestrian that proceed from one point to every point of a finite-size territory (e.g., a square, a shopping mall, etc.). According to the constructal law, the emerging architecture will be the one that promotes the easiest flow of pedestrians. Eq. (1) reveals that are two flow mechanisms to accomplish this purpose: diffusion (slow flow or high resistivity) and streams (fast flow or low resistivity). These two competing mechanisms provide greater flow access than one mechanism alone. What is then the best mechanism that drives pedestrian movement to evolve in space and in time? To estimate diffusion and streams time scales (Eqs. 2 and 3), we need the pedestrian diffusion coefficient (D) and the pedestrian walking speed (u). Several experimental studies (see for example [50]) provide valuable data to obtain these quantities. According to field surveys the pedestrian walking speed is about 1,34 m/s. Worth noting is that pedestrians in a shopping mall or busy city street exhibit a speed that is related to the free area available around each individual. From an analogy with kinetic theory in Miguel and Bejan [48], the pedestrian diffusion coefficient is related to the walking (random) speed and the mean free interpersonal distance via the Einstein-Smoluchowski equation. On the basis of the empirical data, the relation between the pedestrian diffusion coefficient, D, and the mean interpersonal distance, λ, can be established as

D = 1.97λ − 0.61 0.31≤λ≤3.16 m and L>0.31 m

(7)

The Principle that Generates Configuration in Animate and Inanimate...

209

Here L is the distance to access. Why does pedestrian crowd motion exhibit dissimilar patterns (diffusion and streams)? When the time of transition from diffusion to organized flow (streams) is t* ~ 1.1λ-0.34, both diffusion and streams promote the easiest flow of pedestrians. However, diffusion is more effective than streams if tdf < t* (streams perform better if tcv < t*). So, diffusion is more effective when λ > 0.68 L + 0.31 or L < 1.47 λ - 0.45. This means that the diffusion is only more competitive for large mean free interpersonal distances between pedestrians and for accessing small distances. On the other hand, streams are more competitive for small mean free interpersonal distances between pedestrians and for accessing large distances. Therefore, pedestrians follow the configuration that flows more easily. These results are in agreement with empirical observations that pedestrians can diffuse among themselves only at very small pedestrian densities [48,50]. Another striking phenomenon in pedestrian dynamics is the occurrence of spontaneous lanes of pedestrians with uniform walking speed in very crowded open spaces. When facing a stationary crowd, pedestrians spontaneously self-organize in river-like streams (''rivers of people'' [13]) in order to penetrate the crowd [50]. Why do pedestrians exhibit such selforganization? The constructal approach begins with recognition of a purpose. In line with this approach, pedestrians move in configurations that provide greater access. According to Miguel and Bejan [48], the argument of minimum resistance led to the conclusion that streams/lanes provide the best flow access as long as the mean free interpersonal distance between the pedestrians is less than ~ 0.8 m (or density larger than ~ 1.6 pedestrians per square meter). Therefore, configuration evolves in the direction of greater flow access. This result explains why pedestrians self-organize into lanes when their density is high enough, as observed by Helbing et al. [51]. The occurrence of rivers of people through stationary crowds can be predicted from the constructal law in the same way as the structure of river basins [13,14,16]. The stationary crowd is the river basin, and the space vacated by the crossing pedestrians is the eroded river bed. Each pedestrian opens a small space toward the stationary crowd, thereby creating the conditions for a successor to follow. The next pedestrians follow those who are already in motion, giving rise to organized lines that form through the crowd. The channels due to the coalescence of such paths are tree-like structures similar to river branches. The physics meaning of the evolution of the configuration is the greater flow access (constructal law). In crowds that panic, the above formulation is not applicable because the lanes and the river-like streams of people are destroyed: individuals do not know the right way to escape. They strive to go forward, thereby reducing interpersonal distances, which might induce interpersonal contact (collision) or even loss of balance.

5. CITIES/COUNTRIES AND THE CONSTRUCTAL APPROACH Cities and countries are complex systems composed of animate and inanimate subsystems. To the Romans, the city was the center of culture and development [52]. It was so important that they built cities in the lands they conquered in order to spread their civilization. Since Romans, the city configuration and size was a preoccupation. A Roman city started

210

Antonio F. Miguel

with a grid layout, with major buildings in the center and the whole area surrounded by walls. During the Roman Empire, the designers tried to limit cities’ sizes because they thought that the quality of life of huge cites decreases [52]. In our days, the dynamic of city growth stills a major issue [53]. Cities are characterized by specific shape and dynamics. They increase in size and number (evolve). Cities and countries can be treated in thermodynamic terms, for these are also open structures regulating the flow of energy in and out. They have also remarkable general resemblances with living organisms. They acquire and consume resources, produce and discard wastes, all the while employing energy for a variety of tasks: transportation, communication, maintenance, and reproduction of the complexity and organization. Modern cities and countries are as much a product of evolutionary events as any organism. They are also voracious users of energy. Why do these “bodies” have the configuration that they have?

5.1. Zipf’s Law and the Constructal Law At the end of the 19th century, Vilfredo Pareto suggested that income distribution in a society is described by a power law (scaling law) [54]. In the meantime, other scaling laws were founded in biology, economy and other areas of knowledge [18,31-33,55]. In 1949, George Zipf [56] found that city sizes obey to an astonishingly simple distribution law. He attributed this feature to the “least effort principle” of human behavior similarly to the idea of Pierre de Maupertuis (Section 1). Zipf’s law has been gaining continuous interests for its accurate description of city size distributions in many different countries and at different times within a country. Zipf distribution of city ranks versus city sizes can be derived from the constructal law in the same fashion as patterns of animate and inanimate flow structures. Bejan [12] shows that this distribution can be obtained from the optimal allocation of flow paths to areas. This and others Zipf distributions have their origin on tree-shaped flow systems with patterns optimally allocated in space, in a similar way to scaling laws of river basins. These distributions are counterparts of the scaling laws of biology. In both cases, they reveal easier flowing configurations, and in this sense, scaling laws are a synonym of optimal structures. In summary, scaling laws are useful tools in studying complex systems because they reflect underlying generic features that are characteristics of the systems.

5.2. Bringing New Insights to Science: Cities/Countries and Scaling Laws The conception and performance of living organisms have long been an inspiration and subject of study in different domains [27]. From Leonardo da Vinci investigations into human flight, based on the flight of birds, to modern bionics humans created systems that resemblance working of living bodies. The constructal theory shows that the improvement of function (easier access to currents) creates configurations that characterize both animate and inanimate systems (Sections 3 and 4). Therefore, “imitate” nature's design maybe relevant because it may contribute to solve problems in an efficient and sustainable way. Organisms have evolved over thousands of years towards optimal configurations and the “existence of a scaling relationship is a hint that there is an underlying constructal

The Principle that Generates Configuration in Animate and Inanimate...

211

evolutionary process” [39]. Could cities and countries have reached a similar level of performance? All scientific efforts are based on the existence of universality, which manifests itself in diverse ways. Four decades ago, for example, studies on phase transitions and critical phenomena conducted to the development of two important concepts: scaling and universality. These developments were based both on the scaling argument of B. Widom [57] and the full theory of the renormalization group of K. G. Wilson [58]. For systems in the same class, exponents and scaling functions are the same (universal) in the vicinity of the critical point. Scaling and universality support much our current conceptualization in different subjects. Here we explore the thought that complex systems far from equilibrium with similar functions have a propensity to exhibit similar scaling correlations. The purpose is to examine and interpret the metabolism of cities and countries in the light of scaling laws for biology. These laws relate a characteristics or functions of organisms to their body mass. Therefore, we should obtain the body mass of cities and countries. Cities and countries are composed by groups of subsystems that perform specific functions towards commons and specific goals. Similar to organs of living organisms, each subsystem (entity) has a “metabolic” relationship with the rest of the body (city, country), while each subsystem has its own role in the overall metabolism. The whole-system (city, country) “metabolism” results from the summation of the average metabolic rate of all subsystems having a total effective mass. The mass of animate and inanimate subsystems must be accounted. The animate subsystem is assumed to be formed by inhabitants and livestock. Conversely, dwellings, power plants, road and rail networks, ports and airports, all compose the inanimate subsystem. The effective mass of a city/country results of the summation of the whole mass of these subsystems.

Figure 6. Metabolic power versus the mass (cities [59] and countries [60])

212

Antonio F. Miguel

The 3/4-power scaling relationship [31-33] and the metabolic power (power consumption) are depicted in Figure 6. Data for cities and countries were obtained from Isalgue et al. [59] and Miguel [60], respectively. This plot shows that the metabolic power of cities of USA and France are above the allometric law of animate systems. The same tendency is followed by the countries (USA and France). On the other hand, other countries are below the slope line of the allometric law. Unfortunately, there are no data available for cities in these countries, which make it impossible to verify this tendency. Figure 6 also shows that the measured values of the power consumption for countries are not very different from those obtained with the scaling law.

Figure 7. CO2 emission versus the country’s mass in 2002 [60]

Figure 8. Gross domestic product (GDP) versus the country’s mass in 2002 [60]

For 12 countries (Australia, Brazil, Denmark, France, Germany, Italy, Japan, Netherlands, South Africa, Turkey, UK and USA) carbon dioxide emissions based on data of 2002 and the 3/4 power scaling line are shown in Figure 7. This plot shows the existence of countries that fall on the 3/4 power line for living organisms (Netherlands, France and Italy)

The Principle that Generates Configuration in Animate and Inanimate...

213

but that there are also countries above (USA, Japan, Germany and UK) and below (Brazil, Turkey, Denmark, Australia and South Africa) this line. Except for these three countries (Netherlands, France and Italy), all countries release more and less CO2 than expected for an optimal behavior. The “metabolism” of countries like the USA is very fast and releases about 3.5 times more than the CO2 of an optimal system with similar mass. On the other hand, countries like Brazil have a “slow metabolism”, and it releases about 2.3 times less CO2 of an optimal system. To check the possible connection between the speed of “metabolism” and the economic growth, the gross domestic product (GDP) data is considered for each country (Figure 8). The 3/4 power scaling line was adjusted in order to match the Netherlands (i.e., the country that follows a 3/4 power scaling for CO2 emission). According to Figure 8, we come to the conclusion that countries have a similar behavior to that displayed in Figure 7. Except for France (perhaps due to the fact that the majority of the electrical power utilized in France comes from nuclear power plants and not from fossil fuel), the countries with a “slow metabolism” present GDP values below the 3/4 power scaling line. Therefore, it seems that there is a strong relationship between CO2 emissions and GDP which agrees with the findings of [61]. In summary, for CO2 emission there are countries that obey to different scaling laws. The scaling exponent for the countries with a “fast metabolism” (USA, Japan, Germany and UK) is 1.04 (r2=0.996) and for the countries with a “slow metabolism” (Brazil, Turkey, Denmark, Australia and South Africa) is 0.90 (r2=0.888). Notice that all countries are composed by animate and inanimate sub-systems with similar (constructal) configurations. Therefore, the existence of different scaling relationships indicates these groups of countries might have experienced different evolutionary processes. Notice that all the countries with a “fast metabolism” are member of the Group of Eight (G8) highly industrialized countries. None of the countries with a “slow metabolism” is member of this G8 group.

6. CONSTRUCTAL THEORY – A UNIFYING PERSPECTIVE Nature displays an enormous variety of shapes (patterns, configurations). Why is shape so important? Is there a principle from which shapes can be deduced? As noted by Feynman [62] “[There is a] rhythm and pattern between the phenomena of nature which is not apparent to the eye, but only to the eye of analysis; and it is these rhythms and patterns which we call Physical Law.”. Constructal law is the view that the generation of flow configuration is a physics phenomenon. The outcome of constructal law is that system shape and internal flow architecture do not develop by chance, but result from the permanent struggle for better performance. The shape of the system evolves just as the envelope of underlying dynamic processes. The constructal law is the statement that for a flow system to persist in time it must evolve in such a way that it provides easier flow access. This is a very attractive configuration principle for both animate and inanimate systems. Therefore, the constructal theory provides a common theoretical basis for the existence of design in nature. The constructal theory described in the previous section is robust and can be also applied to engineered systems [13,14]. This theory is now a fast growing field with contributions from many sources, and with leads in many directions (see for example, [12,13,14,16,18,27,63,64]).

214

Antonio F. Miguel

REFERENCES [1]

[2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

[19] [20] [21] [22] [23]

Bejan, A. (1999). How nature takes shape: extensions of constructal theory to ducts, rivers, turbulence, cracks, dendritic crystals and spatial economics. Int. J. Therm. Sci., 38, 653-663 Aristotle, Physics. Adelaide: eBooks, Adelaide; 2007 Emerson, R. W. (2006). The Conduct of Life. Lanham: University Press of America. Kant, I. (2008). Critique of Judgement. Radford: Wilder Publications. Glass, B., Temkin, O. & Straus Jr., W. (editors) (1959). Forerunners of Darwin: 1745-1859. Maryland: Johns Hopkins University Press. Reis, A. H. (2006) Constructal theory: from engineering to physics, and how flow systems develop shape and structure. Applied Mechanics Reviews, 59, 269-282. Colding, T. H. & Minicozzi, W. P. (2006) Shapes of embedded minimal surfaces. Proceedings of the National Academy of Sciences, 103, 11106-11111. Darwin, C. (1995). The Origin of Species. New York: Gramercy. Thompson, D. W. (1992). On Growth and Form. Cambridge: Cambridge University Press. Helbing, D. & Molnár, P. (1995) Social force model for pedestrian dynamics. Phys. Rev., E, 51, 4282 – 4286. Mandelbrot, B. B. (1982). The Fractal Geometry of Nature. New York: W.H. Freeman and Co. Bejan, A. (2006). Advanced Engineering Thermodynamics, 2nd ed.. Hoboken: Wiley; 1997; Advanced Engineering Thermodynamics, 3rd ed.. Hoboken: Wiley. Bejan, A. (2000). Shape and Structure, from engineering to nature. Cambridge: Cambridge University Press. Bejan, A. & Lorente, S. (2008). Design with Constructal Theory. New Jersey: Wiley. Bejan, A. & Lorente, S. (2006) Constructal theory of generation of configuration in nature and engineering. J. Applied Physics, 100, 2006. Bejan, A. & Lorente, S. (2004) The constructal law and the thermodynamics of flow systems with configuration. Int. J. Heat and Mass Transfer, 47, 3203–3214. Bejan, A. & Lorente, S., (2004). Equilibrium and nonequilibrium flow system architectures. Heat Technol., 22, 85–92. Bejan, A., Lorente, S., Miguel, A. F. & Reis, A. H. (2006). Along with Constructal Theory. Workshop series nº 1 (Hernandez & Cosinschi, editors). Lausanne: University of Lausanne Press. Reis, A. H., Miguel, A. F. & Bejan, A. (2006). Constructal theory of particle agglomeration and design of air-cleaning devices. J. Phys. D, 39, 2311–2318. Bejan, A. & Gobin, D. (2006) Constructal theory of droplet impact geometry. Int. J. Heat and Mass Transfer, 49, 2412–2419. Reis, A. H. (2006). Constructal view of scaling laws of river basins. Geomorphology, 78, 201–206. Reis, A. H. & Gama, C. (2009). Sand size versus beachface slope — an explanation based on the constructal law. Geomorphology, in press Bejan, A. & Reis, A. H. (2005). Thermodynamic optimization of global circulation and climate. Int. J. Energy Res., 29, 303-316.

The Principle that Generates Configuration in Animate and Inanimate...

215

[24] Reis, A. H. & Bejan, A. (2006). Constructal theory of global circulation and climate. Int. J. Heat Mass Transfer, 49, 1857–1875. [25] Bejan, A., Ikegami, Y. & Ledezma, G. A. (1998). Constructal theory of natural crack pattern formation for fastest cooling. Int. J. Heat Mass Transfer, 41, 1945–1954. [26] Lorente, S. (2007). Constructal view of electrokinetic transfer through porous media. J. Phys. D, 40, 2941–2947 [27] Miguel, A. F. (2009). Pattern Formation and Self-Organization in Living Systems: a unified view for coral colonies and crowd dynamics. In: "Constructal Human Dynamics, Security and Sustainability" A. Bejan, S. Lorente, A. F. Miguel, A. H. Reis (editors). Series Human and Societal Dynamics – Vol. 50. Amsterdam: IOS Press. [28] Reis, A. H, Miguel, A. F. & Aydin, M. (2004). Constructal theory of flow architecture of the lungs. Medical Physics, 31, 1135-1140. [29] Reis, A. H. & Miguel, A. F. (2006). Constructal theory and flow architectures in living systems. J. Thermal Science, 10, 57-64. [30] Bejan, A., Dincer, I., Lorente, S., Miguel, A. F. & Reis, A. H. (2004). Porous and Complex Flow Structures in Modern Technologies. New York: Springer. [31] Kleiber, M. (1947). Body size and metabolic rate. Physiological Reviews, 27, 511-541. [32] Brody, S. (1945). Bioenergetics and Growth. New York: Reinhold. [33] West, G. B., Woodruff, W. H. & Brown, J. H. (2002). Allometric scaling of metabolism from molecules and mitochondria to cells and mammals. Proc. Natl. Acad. Sci. U.S.A., 99, 2473. [34] Bejan, A. (2001). The tree of convective heat streams: its thermal insulation function and the predicted 3/4-power relation between body heat loss and body size. Int. J. Heat and Mass Transfer, 44, 699-704. [35] Bejan, A. (1997). Theory of organization in Nature: pulsating physiological processes. Int. J. Heat Mass Transfer, 40, 2097-2104. [36] Bejan, A. (1990). Optimum hair strand diameter for minimum free-convection heat transfer from a surface covered with hair. Int. J. Heat Mass Transfer, 33, 206-209. [37] Bejan, A. & Marden, J. H. (2006). Unifying constructal theory for scale effects in running, swimming and flying. J. Exp. Biol., 209, 238-248. [38] Bejan, A. & Marden, J. H. (2009). The constructal unification of biological and geophysical design. Physics of Life Reviews, 6, 85-102. [39] Charles, J. D. & Bejan, A. (2009). The evolution of speed, size and shape in modern athletics. J. Exp. Biol., 212, 2419-2425. [40] Bejan, A., Lorente, S. & Lee, J. (2008). Unifying constructal theory of tree roots, canopies and forests. J. Theoretical Biology, 254, 529–540. [41] Ben-Jacob, E., Cohen, I., Shochet, O., Aronson, I., Levine, H. & Tsimering, L. (1995). Complex bacterial patterns. Nature, 373, 566-567. [42] Matsushita, M., Wakita, J., Itoh, H., Ràfols, I., Matsuyama, T., Sakaguchi, H. & Mimura, M. (1998). Interface growth and pattern formation in bacterial colonies. Physica A, 249, 517-524. [43] Kaandorp J. A. & Sloot P. M. A. (2001). Morphological models of radiate accretive growth and the influence of hydrodynamics. J. Theor. Biol., 209, 257–274. [44] Tumlinson, J. H.,. Silverstein, R. J., Moser, J. C., Brownlee, R. G. & Ruth, J. M. (1971). Identification of the trail pheromone of a leaf-cutting ant, Atta texana. Nature, 234, 348 – 349.

216

Antonio F. Miguel

[45] Partan, S. R. & Marler, P. (1999). Communication goes multimodal. Science, 283, 1272–1273. [46] Anderson, C. & McShea, D. W. (2001). Individual versus social complexity, with particular reference to ant colonies. Biological Review, 76, 211–237. [47] Miguel, A. F. (2006). Constructal pattern formation in stony corals, bacterial colonies and plant roots under different hydrodynamics conditions. J. Theoretical Biology, 242, 954-961. [48] Miguel, A. F. & Bejan, A. (2009). The principle that generates dissimilar patterns inside aggregates of organisms. Physica A, 388, 727-731. [49] Miguel, A. F. (2009). Constructal theory of pedestrian dynamics. Physics Letters A, 373, 1734-1738 [50] Miguel, A. F. (2007). Constructal Patterns Formation in Nature, Pedestrian Motion and Epidemics Propagation. In: Constructal Theory of Social Dynamics (Bejan and Merkx, editors), New York: Springer, chapt. 5, 85-114. [51] Helbing, D., Schweitzer, F., Keltsch, J. & Molnar, P. (1997). Active walker model for the formation of human and animal trail systems. Phys. Rev. E, 56, 2527-2539. [52] Macaulay, D. (1983). City: a Story of Roman Planning and Construction. Boston: Houghton Mifflin Company. [53] Bluestone, B., Stevenson, M. H. & Williams, R. (2008). The Urban Experience: Economics, Society, and Public Policy. New York: Oxford University Press. [54] Pareto, V. (1897). Le Cours d’Economie Politique. London: Macmillan. [55] West, G. B., Brown, J. H. & Enquist, B. J. (1997). A general model for the origin of allometric scaling laws in biology. Science, 276, 122–126. [56] Zipf, G. K. (1949). Human Behavior and the Principle of Least Effort. Cambridge: Addison-Wesley. [57] Widom, B. (1965) Surface tension and molecular correlations near the critical point. J. Chem. Phys., 43, 3892-3897 [58] Wilson, K. G. (1971) Renormalization Group and Critical Phenomena. I. Renormalization Group and the Kadanoff Scaling Picture. Phys. Rev. B, 4, 3171-3183 [59] Isalgue, A., Coch, H. & Serra, R. (2007). Scaling laws and the modern city. Physica A, 382, 643–649. [60] Miguel, A. F. (2009). Quantitative study of the CO2 emission to the atmosphere from biological scaling laws. Int. J. Global Warming, 1, 129-143. [61] Shafik, N. (1994) Economic development and environmental quality: an econometric analysis, Oxford Economic Papers, 46, 757-773. [62] Feynman, R. F. (1967). The Character of the Physical Law. Cambridge: MIT press. [63] Bejan, A., Badescu, V. & de Vos, A. (2000) Constructal theory of economics. Applied Energy, 67, 37-60. [64] Bejan, A., Badescu, V. & de Vos, A. (2000) Constructal theory of economic structure generation in space and time. Energy Conversion and Management, 41, 1429-1451.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 217-243

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 6

COMPUTATIONAL STUDIES ON DRAG REDUCTION EFFECT BY SURFACE GROOVES Haosheng Chen* and Yongjian Li State Key Laboratory of Tribology, Department of precision instrument and mechnology Tsinghua University, Beijing, 100084, China

ABSTRACT The drag reduction effect caused by the periodic surface grooves were studied using computational fluid dynamic method in two different flow conditions: laminar flow in a slide-disk interface, and turbulent flow on the grooved surface immerged in water. In the first part, the drag results by Reynolds equation that is commonly used in lubrication calculation is compared with the CFD result based on Navier-Stokes equation. It was validated that the Reynolds equation is not suitable when the groove depth is higher than 10% of the interface distance. Then, the drag forces on the surface with transverse and longitudinal grooves are calculated using CFD method. It was found that the ‘side wall’ effect causes the drag reduction, which means the drag reduction would appear when the loss of the drag on the groove’s bottom can not be compensated by the pressure drag or the viscous drag on the side walls of the grooves. In the second part, the drag force on the transverse rectangular grooves in turbulent flow is analyzed based on the RANS equations coupled with the RNG k-ε turbulent model. It was found that the pressure drag force makes up a large proportion of the total drag, and the turbulent vortex structure in the grooves affects the drag characteristic of the surface. The ‘side wall’ effect also functions in turbulent flows, and the drag reduction is determined by the synthesized effect of the reduction of viscous drag and the increment of pressure drag.

*

Corresponding author: [email protected]

218

Haosheng Chen and Yongjian Li

SECTION I Drag Reduction on Groove Surface in Laminar Flow Contents Usually, Reynolds equation is used in hydrodynamic lubrication calculation, and a discrete probability distribution function is used in the equation to represent the effect of transverse and longitude grooves. Here, the drag force on the square groove surface by the Reynolds equation is validated by the Navier-Stokes equation, and it is found that the results acquired by the two methods are the same when the groove height is below 1% of the film thickness. When the height exceeds 1% of the film thickness, the forces acquired by Reynolds equations are smaller than those acquired by CFD, and the difference between them exceeds 10% when the groove height is higher than 10% of the film thickness. Side wall effect is considered to be the main reason for the difference, and Reynolds equation is believed not suitable for calculating the drag force under such conditions. Therefore, forces on side wall of the transverse and longitudinal grooves are numerically analyzed based on Navier-Stokes equation using finite volume method. The viscous force on the groove’s bottom plane is far less than the force on corresponding plane of smooth surface, but the loss of the drag force on groove’s bottom surface is compensated by either the pressure drag on side walls of transverse groove, or the viscous force on side walls of longitudinal groove. According to the numerical results for the squared grooves as calculated here, the pressure drag on side wall of transverse groove is higher than the viscous force on side wall of longitudinal groove, and drag reduction happens on the surface with longitudinal grooves.

1. CFD EVALUATION ON APPLICABILITY OF REYNOLDS EQUATION The drag reduction on groove surface has been studied for many years [1]. The pioneer work done by Dowson [2] and Shelly [3] has revealed the effect of artificially made grooves in a journal bearing lubrication. It has been proved that proper surface roughness increases the load capacity and reduces the friction coefficient [2-5], and specially designed surface roughness is also investigated in improving the head-disk interface lubrication recently [6,7]. On the other hand, MEMS technology makes it possible to design micro structures, such as micro squared grooves, on the working surface to develop the lubricating results [8,9]. However, there are some problems when using the traditional Reynolds equation to calculate the hydrodynamic lubrication under the effect of squared grooves. First, the average Reynolds equation [10, 11] is commonly used and the surface roughness effect is represented by the stochastic model, where Gaussian distribution function is used for random roughness. But the Gaussian distribution function is not suitable for the artificially designed periodical grooves. Second, as Elrod [1] has pointed out, Reynolds equation can only apply to the “Reynolds roughness” where the roughness height is an order of magnitude smaller than the film thickness. But in most of the applications where micro grooves are made on the surfaces [6,9], the groove depth is in an order of the magnitude of the lubricant film thickness, so the

Computational Studies on Drag Reduction Effect by Surface Grooves

219

work on the applicability and precision of the Reynolds equation under the effect of surface grooves still needs to be continued. In the following studies, the hydrodynamic lubrication on a surface with 50 squared transverse grooves is calculated by Reynolds equation and CFD method, respectively. When using Reynolds equation, Gaussian distribution function is replaced by a discrete probability distribution function to represent the expectancy of the film thickness. This modified Reynolds-type equation (MRE) is derived to play the same role in the analysis of groove surfaces as the ordinary Reynolds equation does. Then, the differences between the results from the less approximating CFD approach and modified Reynolds equations are compared. Based on the comparison, the applicability of Reynolds equation in calculating the effect of squared groove surface is discussed.

1.1. Geometry Models Figure 1 shows the geometry model for CFD, it is a typical slider-disk interface used in lubrication study. The upper surface is smooth and moves at a speed of U. The pitch angle of the upper surface is 1°. The lower surface is rough and stationary, and there are 50 squared transverse grooves with the wavelength λ on the lower surface. The geometry of the lubricant film is split up into two parts. One part which denoted the nominal film thickness h ( x) measures the large scale variation in film geometry. The other part is the part due to groove depth as measured from the nominal level, and it varies with the amplitude of the roughness height σ. The relationship between the two parts is σ =β (h0+h1) / 2. As a common method, a CFD software FLUENT is used as a less approximating approach to achieve more precise results. FLUENT uses a control-volume-based technique to convert the governing equations, such as Navier-Stokes equations, to algebraic equations that can be solved numerically. The continuity equation and momentum equation are adopted in the calculation, which are applicable for continuous medium. The values of the parameters used in the lubrication are listed in Table 1. The boundary conditions for CFD are defined as follows: the upper surface is a moving wall, while the lower surface is a stationary wall; the left side is the pressure inlet and the right side is the pressure outlet. Under the low Reynolds number, the lubricant is considered to perform a stable laminar flow according to Chin’s viewpoint [12], and a laminar model is adopted in CFD. z

U

λ

h0

h(x)

h(x)

h1

x

σ σ B=50λ

Figure1. Geometry model of a slider-disk with 50 squared transverse grooves

220

Haosheng Chen and Yongjian Li Table 1. Values and parameters used in the lubrication calculation Parameters U h1 h0 B η β

Value 10 m/s 2.4364e-004 m 1.5636e-004 m 5e-3 m 1 Pa.s 0.001~0.2

There are 55000 quadrangle grids total for calculation. To show the pressure and fluid flow near the wall more clearly, the grid number increases from the center to the wall and the grow factor is 1.1. The grid size is about 2×2 μm2, which is small enough in the laminar flow compared with the size of the squared transverse waves. The residual definitions are used for judging convergence. This criterion requires that the residuals decrease to 10-6 for the continuity equation and the velocity equation. During the calculation, the residuals reach the convergence criteria after 25000 iteration steps, which indicate that a quite precise CFD result is achieved.

1.2. Modified Reynolds Equations for Groove Surface A stochastic model for rough surface with transverse roughness is shown as Eq.(1), where ψi(H) is the expectancy of the film thickness that has been demonstrated detailedly by Christensen [13].

∂ ⎡ ∂p ∂ ⎤ Ψ 1 (H )⎥ = 6μU Ψ 3 (H ) ⎢ ∂x ⎣ ∂x ∂x ⎦

(1)

For random transverse roughness, the film thickness function ψi (H) was given by Christensen [13]. For surface with square grooves, the Gaussian distribution function used for random roughness is replaced by the discrete probability distribution function to calculate the expectancy of the film thickness. In the Eq.(2), E is the expectancy, xi is the variance and pi is the probability respect to xi. n

E = ∑ xi pi

(2)

ψ 1 = ( 2 β3 ) h 3 ; ψ 3 = ( β 2 β3 ) h ;

(3)

i =1

so, the expressions of the functions are:

Computational Studies on Drag Reduction Effect by Surface Grooves 2

2

3

⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ β2 = ⎜ ⎟ +⎜ ⎟ ; β3 = ⎜ ⎟ +⎜ ⎟ ⎝ 1+ β ⎠ ⎝ 1− β ⎠ ⎝ 1+ β ⎠ ⎝ 1− β ⎠

Where,

221

3

Then, the Reynolds equation is modified as following:

∂ ⎡ ∂p 3 ⎤ β 2 ∂h 6 μU h ⎥= ⎢ ∂x ⎣ ∂x ⎦ 2 ∂x

(4)

According to the geometry model shown in Figure 1, the pressure is acquired through integrating equation (4) on both sides with respect to x twice. With the boundary conditions p|h=h0=0 and p|h=h1=0, the expression of the hydrodynamic pressure is derived as Eq. (5), where K=( h1-h0)/ h0.

p = 6μU

β2 B ⎡ 1

h0 h1 1 1 ⎤ + ⎢− + ⎥ 2 2 Kh0 ⎣ h h0 + h1 h h0 + h1 ⎦

(5)

1.3. Numerical Results In the calculation, the groove depth is not higher than 20% of the film thickness, i.e. β ≤ 0.2, which indicates that the lubrication is under the full film lubricating condition. The load capacity, friction force and the friction coefficient are calculated by CFD and MRE methods, respectively. Such forces are compared with those on smooth surface using normalized percentage (F-Fs)/Fs%, where F is the force for the rough surface while Fs is the force for the smooth surface. 15

(a) MRE

40

(b) CFD 0.190

0.190 12

0.185

30

(F-Fs)Fs%

Load capacity Friction force Friction coefficient

0.175

3

0.180 0.175

20

0.170 10

Load capacity Friction force Friction coefficient

0.165 0.160

0.170

0

-3

2

Friction coefficient

0.180

6

3

1

0.185

3

Friction coefficient

9

2

(F-Fs)Fs%

1

0.155

0

1E-3

0.01

β

0.1

1

0.165

1E-3

0.01

0.1

1

0.150

β

Figure 2. Lubrication results acquired by CFD and MRE compared with those on smooth surface

As the groove depth increases, both of the load capacity and the friction force increase compared with forces on smooth surface, but the increment of the friction force is lower than that of the load capacity, which results in the decrement of the friction coefficient. The increasing process of the lubricant results can be divided into 3 characteristic stages. In stage 1, β<0.01, the effect of the surface groove on lubricating results is not obvious, the load

222

Haosheng Chen and Yongjian Li

capacity and the friction force on the groove surface is almost the same as those on smooth surface, it indicates that the terms considering the roughness effect can be omitted in the Reynolds equations. In stage 2, β<0.1, the load capacity and the friction force on the rough surface begin to increase, and the friction coefficient decreases correspondingly. However, the increments of the forces are below 5% and 15% acquired by MRE and CFD, respectively. So the groove effect should be considered in some conditions where the precision is required. In stage 3, β ≥ 0.1, the film thickness is greatly changed by the surface roughness, and the load capacity increases so rapidly that the groove effect must be included in the lubrication calculation. According to the results shown in Figure 2, it is found that there is a marked difference between the forces acquired by MRE and CFD methods, especially when β ≥ 0.01. Figure 3 shows the differences more clearly using the normalized percentage, where the subscript represents the method used in the calculation. The differences on the load capacity and the friction force both increase as the groove depth increases. Moreover, the increment speed of load capacity is higher than that of the friction force, which results in the lower friction coefficient acquired by the CFD method. As Almqvist [14] has validated, CFD is a more precise method than the Reynolds equation, so it is believed that a conservative lubrication effect is acquired by the Reynolds equation. For example, the friction coefficient and the load capacity are 10% lower and 20% higher than those acquired by the MRE method, respectively. Based on the comparison, it is considered that the Reynolds equation is good enough for calculating the hydrodynamic lubrication including the roughness effect when the roughness height is below 1% of the film thickness, while it is not suitable when the squared groove depth is in an order of the film thickness. 20

Friction force Load capacity

(FCFD-FMRE)/FMRE%

Friction coefficient 10

0

-10 1E-3

0.01

0.1

1

β

Figure 3. Comparison on lubrication results acquired by CFD and MRE

One of the important reasons that cause the different lubrication results acquired by MRE and CFD is considered to be the side wall effect caused by the squared transverse grooves. Figure 4 shows the shape of the 25th groove when β = 0.1. There are two side walls in each groove. Plane 4 is the windward of the flow direction while plane 2 is the leeward of the flow direction. Pressures on the windward plane will increase while pressures on the leeward plane will decrease, when the fluid passes the transverse groove. The contour lines of the pressures

Computational Studies on Drag Reduction Effect by Surface Grooves

223

on the two planes are marked in Figure 4(a), and the value of each line is also given in the figure. It is found that the highest pressure is about 20% higher than the lowest pressure. This result conflicts with one of the basic assumption for Reynolds equation that the pressure does not change so much across the fluid film. As Sun [15] has point out, the pressures generated in the fluid along x and z directions can be represented by Eq. (6) and (7) , respectively. β =0.1 z

3.91Mpa 6.07Mpa

A

600

4.22Mpa 5.76Mpa 4.53Mpa 5.45Mpa 4.83Mpa 5.14Mpa

Plane4

Plane2

U

Groove Plane3

x

A

Pressure along A-A cross section(kPa)

(a)

(b) β=0.1

550

500

groove

β=0.01 groove

450

β=0.001 groove 400 1E-8

1E-7

1E-6

1E-5

1E-4

1E-3

Height (m)

Figure 4. Pressure distributions on the side wall of the 25th squared transverse wave

px −

∂u ∂u =ε2 ∂z ∂x

pz = ε 2

∂v ∂v +ε4 ∂z ∂x

(6)

(7)

When the groove depth is small, the terms on the right-hand side of equations are neglected, and then the traditional Reynolds-type equation is established. However, if the groove depth is in an order of the film thickness, such terms should be retained to reveal the roughness effects. In the average Reynolds equation [10,11], the effect of Eq.(6) is compensated by the flow factors, but the effect of Eq.(7) is neglected. In the calculation results acquired by CFD, such terms are kept, and Figure 4(b) describes the pressure variation across the film thickness at the cross-section A-A when roughness height is 0.001, 0.01 and 0.1, respectively. It is proved by the results that the pressure along z direction almost keeps constant when the groove depth is small, while it changes a lot when the groove depth exceeds 1% of the fluid film thickness. Compared with the results from MRE, it is believed that the pressure variation across the film increases the hydrodynamic pressure, which directly causes the increment of the load capacity. Compared with the load capacity, the reason for the increment of the friction force is more complex. The total friction force consists of two parts. One is the viscous drag force on the surface parallel to the flow direction caused by the fluid viscosity, the other is the pressure drag caused by the side walls of the grooves shown as Figure 4(a). On the windward side wall, the pressure increases from the bottom to the top of the side wall. But on the leeward plane, the pressure variation process is reversed. The difference of pressures on the two side walls of the transverse groove causes a significant pressure drag. As the groove depth increases, the velocity of the fluid in the groove decreases and the viscous friction force on the bottom wall is then decreases. On the other hand, the pressure on

224

Haosheng Chen and Yongjian Li

the windward side wall increases as the groove depth increases, then the pressure drag is inevitably increased. So whether the friction force on the rough surface is higher or lower than that on the smooth surface is determined by whether the pressure drag generated on side walls can compensate the viscous drag loss on the bottom walls or not. The result is illustrated by Figure 5, and the speed of the pressure drags increment on the side walls is higher than that of the viscous drag loss on the bottom walls, which results in the increment of the total friction force on the rough surface. Furthermore, the increment of roughness height not only increases the pressure drag, but also decreases the viscous drag, so the total increment speed of the friction force is less than that of the load capacity, which is the reason for the decrement of the friction coefficient. The great difference between the results acquired by CFD and MRE indicates that the Reynolds-type equations are not suitable for the friction force calculation under the side wall effect. 300

Drag force (N)

250

200

150

pressure drag viscous drag

100

total drag drag on smooth surface

50

0 1E-3

0.01

0.1

0.5

β

Figure 5. Friction forces on transverse groove surface using CFD

Thus, some conclusions are drawn according to the comparison: (1) Compared with the smooth surface, surface grooves begin to show the effect when groove depth is larger than 1% of the film thickness, and the effect can not be neglected when the height exceed 10% of the film thickness. Squared transverse grooves on the surface cause higher load capacity, higher friction force, and lower friction coefficient, and the groove effect becomes stronger as the groove depth increases. (2) Compared with the CFD results, lubrication results acquired by the modified Reynolds equation is more conservative when the groove depth exceeds 1% of the film thickness, and the Reynolds equation is considered not applicable when the groove depth is higher than 10% of the film thickness because the load capacity acquired by the Reynolds equation is 20% lower than that acquired by more precise CFD method. (3) Side wall effect is believed to be one of the important reasons causing the difference between CFD and MRE results. The side wall effect causes the pressure variation across the fluid film and increases the hydrodynamic pressure, which directly

Computational Studies on Drag Reduction Effect by Surface Grooves

225

increases the load capacity compared with that acquired by MRE. On the other hand, the side wall effect both increases the pressure drags on the side walls and decreases the viscous drags on the bottom walls. Whether the pressure drags can compensate the loss of viscous drags determines the appearance of the drag reduction.

2. DRAG REDUCTION OF SURFACE GROOVES Recently, the effect of surface groove patterns on drag reduction has been studied considering the parameter γ , which represents the different surface groove directions. Their calculations show that the γ affects the lubrication results through changing the flow factors [6], it is predicted in the numerical results [5,6] that the surface with longitudinal pattern has the highest load capacity compared with the transverse and isotropic surface patterns, and it correspondingly decreases the friction coefficient. According to the previous studies, when the surface roughness height is in the same order of the lubrication film thickness, there is great difference between the results obtained from the Reynolds equation and the results obtain from the less approximating CFD-approach [14,16]. This continuous studies is to determine which kind of surface patterns is best for the drag reduction under the condition that Reynolds equation is not applicable. Forces on different planes of the groove surface are numerically calculated based on Navier-Stokes equation using finite volume method, and analysis on the drag force reduction effect of different surface roughness is also given. Same to the previous studies, all the calculations are under the condition of laminar flow.

2.1. Models and Method The geometry models of the surface with grooves are shown in Figure 6 schematically. The dimension unit is millimeter. The upper surface is smooth and it moves at a velocity U. The lower surface is stationary and grooves are on it. There are two typical kinds of patterns on stationary surface, longitudinal grooves and transverse grooves. The cross-section of the groove is rectangular and the surface pattern parameter γ is adopted here to represent the length-to-width ratio of groove [6]. γ < 1 for transverse groove, and γ >1 for longitudinal groove. The orientation of the longitudinal groove is parallel to the flow direction, while the orientation of transverse groove is perpendicular to the flow direction. The isotropic pattern can be seen as the superimposition result of the longitudinal pattern and the transverse pattern. The groove surface is divided into 5 rectangular planes as marked by c~g in Figure 6. Plane 1,3,5 are parallel to x-z plane, they are called parallel planes. Plane 2 and plane 4 are perpendicular to the x-z plane, they are called perpendicular planes. There are 5 walls in the cross-section of the groove surface as shown in Figure 6(d). Wall 1 and 5 are the top walls, while wall 2 and 4 are the side walls; wall 3 is the bottom wall. The smooth surface is also modeled to be compared with the surface pattern, it is equalspaced divided into 3 walls named wall 1, 3 and 5 respectively, which is shown in Figure 6(c). The groove is a steep-walled surface and the depth of the groove is in the same order of the fluid thickness. Therefore, the

226

Haosheng Chen and Yongjian Li

Reynolds equation is not applicable under such conditions. The effect of perpendicular planes on the fluid flow and pressure needs to be included in the calculation. The numerical method is same to that in previous studies in section 1, and the finite volume method (FVM) is adopted to calculate the force on the different planes.

2.2. Numerical Analysis Pressure drag on perpendicular planes Between the two transverse perpendicular planes, plane 4 is the windward of the flow direction while plane 2 is the leeward of the flow direction. The contour lines of the pressure on the two planes are numbered from 1~5 in Figure 5(a). The value of each line is also given in the figure, P4 represents the pressure on plane 4 and P2 represents the pressure on plane 2. The difference of pressures on the two transverse perpendicular planes causes a significant pressure drag. It is one of the important reasons that increase the drag force. A rectangular x-y coordinate is established on plane 2 and plane 3 shown in Figure 7(a). More detailed pressure along plane 4 from the bottom to the top of the interface is shown in Figure 7(b). The pressures in Figure 7(b) are normalized, they are the ratio of the pressures to the maximum pressure. The maximum pressure is marked in Figure 7(b). The center line represents the groove depth. The existence of perpendicular plane causes a severe variation of the pressure in the groove, and the pressure achieves a stable value rapidly when the height exceeds the groove depth. The pressure variation on perpendicular planes affects the pressure drag, so the pressures on side wall of grooves with different directions needs to be considered when analyzing the effect of surface patterns to drag force reduction.

1

1

1 2

4

5

U = 10 m/sec z 1

2

y

4

(a) Transverse pattern

3

1

3

3

0.5

5

21

1

3

3

x (b) Longitudinal pattern

Wall 2

Wall 4

y Wall 1 z x

Wall 5

Wall 3

(d) Walls and Mesh on cross-section A-A of groove model

Figure 6. Geometry models of surface patterns

(c) Smooth pattern

5

Computational Studies on Drag Reduction Effect by Surface Grooves (b)

1.0

1: P2=-1.90E3; P4=1.39E3 2: P2=-1.57E3; P4=1.07E3 3: P2=-1.24E3; P4=0.74E3 4: P2=-0.91E3; P4=0.41E3 5: P2=-0.58E3; P4=0.08E3

0.8

(a)

y U

5

0.4

Groove deepth 0.5 mm

0.0

-0.2 0

43 2 1

4 Groove

Plane 5 Plane 4

1 3 2 Plane 2

P / P max

5 Plane 1

0.6

0.2

227

Plane 3

x

PEO, Pmax=1.47E3

0.5

1.0

1.5

2.0

y / [mm]

Figure 7. Pressure on the perpendicular planes of transverse groove

Viscous force on perpendicular planes Not same to the effect of transverse perpendicular planes, there is no pressure drag on the longitudinal perpendicular plane. However, obvious viscous force is generated on the planes. In the same fluid, the viscous force can be represented by the strain rate of the fluid. The strain rates of the fluid on perpendicular planes of three different grooves are compared in Figure 8. All the shear rates are on plane 2 and along direction x. For smooth surface, there is no perpendicular plane and the shear rate is equal to zero. But for the later analyses on viscous drag and pressure drag on different surfaces, it is provided here and compared with the shear rate on other two surfaces. It is found from Figure 8 that there is not only pressure drag but also viscous drag on the perpendicular planes of transverse grooves. But, the strain rate on longitudinal perpendicular plane is obviously higher than the strain rate on other corresponding perpendicular plane in the groove. It is one of the important part of the drag force on the surface with longitudinal grooves.

Groove depth

2000

Shear rate / [1/s]

1500

1000

Longitudinal Smooth

500

Transverse 0 0

0.5

1.0

y / [mm]

Figure 8. Strain rate on the perpendicular planes

1.5

2.0

228

Haosheng Chen and Yongjian Li 0.005

0.016

Viscous force on transverse Viscous force on longitudinal Pressure force on transverse

Longitudinal Transverse Total force Force on plane 1,3,5 Force on plane 2 ,4

0.012

0.003

Drag / [Pa]

Drag force / [Pa]

0.004

0.002

0.008

0.004

0.001

0.000 plane1

plane 2

plane 3

plane 4

plane 5

0.000 200

Surface (a) Total force comparison when velocity is 2 m/s

400

600

800

1000

1200

Shear rate / [1/s] (b) Force comparison under different shear rates

Figure 9. Forces on planes of transverse and longitudinal grooves

Surface pattern effect Total drag force on the surface consists of pressure drag and viscous force. As discussed above, the forces on the perpendicular planes are obtained using finite volume method based on Navier-Stokes equation. The pressure drag and viscous force on five planes are shown in Figure 9(a). The sum of viscous forces on plane 1 and plane 5 of different grooves are almost the same. But the viscous force on perpendicular planes of longitudinal groove is obviously greater than the force on perpendicular planes of transverse groove. On the other side, there is no pressure drag generated on perpendicular planes of longitudinal groove, while there is significant pressure drag on the perpendicular planes of transverse groove. Moreover, the sum of pressure drag on transverse perpendicular planes is greater than the sum of viscous force on longitudinal perpendicular planes. It indicates that the transverse perpendicular planes will cause more drag force than the longitudinal perpendicular planes does. As shown in Figure 9(b), the forces on perpendicular planes and on parallel planes are both increasing as the shear rate increases. But the increment of the force difference on perpendicular planes is always greater than the increment of the force difference on parallel planes. It leads to the result that the total drag force on transverse grooves is more and more greater than that on longitudinal grooves as the shear rate increases. For smooth surface, the drag force is caused by the viscous force. Like the treatment on rough surface, the smooth surface is also divided into three planes, the shape of each plane is same to plane 1, plane 3 and plane 5 of transverse grooves, respectively. The force on each plane is shown in Figure 10(a), it is compared with the forces on corresponding planes of transverse groove and longitudinal groove. The viscous forces on plane 1 and plane 5 of grooves are greater than the viscous force on corresponding smooth surface. But on plane 3, the viscous force on smooth surface is obviously greater. It is because the flow velocity is rapidly slowed down in the grooves, and the shear rate is also slowed down. The strain rate of fluid at the center point of plane 3 from bottom to the top of the film thickness is shown as Figure 10 (b). The strain rate on smooth surface is far greater than the strain rates on rough surface. It leads to the higher viscous force on plane 3 of smooth surface compared with the forces on rough surface. So, the sum of viscous forces on three parallel planes of grooves is less than the force on smooth surface.

Computational Studies on Drag Reduction Effect by Surface Grooves 1600

Viscous force on transverse groove Viscous force on longitudinal groove Viscous force on smooth surface Pressure force on transverse groove

Shear rate / [1/s]

1200

0.007

800

400

0 0.000 plane1

plane2

plane3

plane4

plane5

all

Longitudinal

Roughness height

0.014

Drag force / [Pa]

229

0

Planes (a) Comparison of drag force on different surfaces

0.5

Smooth Transverse 1.0

1.5

2.0

y / [mm] (b) Strain rate along y direction

Figure 10. Effect of surface pattern on drag reduction compared with smooth surface

However, the decrement of viscous force on parallel planes of grooves is compensated by the pressure drags and viscous forces on perpendicular planes. Including the forces on the perpendicular planes, the sum of forces on transverse groove is greater than the force on smooth surface, and the sum of forces on longitudinal groove is a little bit less than force on smooth surface. Thus, drag reduction effect appears only on the longitudinal groove surface.

3. CONCLUSIONS In this section, the side wall effect on groove surface is numerically studied. Because of it, the Reynolds equation is not suitable and the force on transverse groove surface is different from that on longitudinal groove surface. According to the numerical results, some conclusions can be drawn as following: (1) The effect of groove surface needs to be considered when the groove depth is higher than 1% of the lubricant film thickness. (2) Reynolds equation is considered not applicable when the groove depth is higher than 10% of the film thickness, and the side wall effect causing the high pressure gradient on the wall is believed to be one of the important reasons causing the difference between CFD and MRE results. (3) Pressure drag and viscous force on the perpendicular planes play important roles in drag reduction effect. The viscous force on the bottom plane of the groove is far less than the force on the corresponding plane of smooth surface. However, drag force on groove surface is compensated by the pressure drag on transverse perpendicular plane, or by the viscous force on longitudinal perpendicular plane. Since the pressure drag appears to be always higher than the viscous force on the same perpendicular plane under different average shear rates, the longitudinal groove surface has the lowest drag force in laminar flow.

230

Haosheng Chen and Yongjian Li

SECTION II Drag Characteristic on Grooved Surface in Turbulent Flow Content In this section, the drag force on the transverse rectangular grooves in turbulent flow is analyzed by CFD methods based on the RANS equations coupled with the RNG k-ε turbulent model. The pressure drag force induced by the side wall effect could make up 80% of the total drag force when the Reynolds number is higher than 50, 000. So, the side wall effect of the transverse rectangular grooves is more significant in the turbulent flow than that in the laminar flow. The flow structure of the turbulence is more complicated and it is affected by the Reynolds number and the shape of the grooves. As the increase of the width-to-depth ratio of the grooves, the vortex structure in the grooves changes from single to double. This variation could change the pressure difference between the windward side and the leeward side of the grooves and the magnitude and direction of the viscous friction force acting on the bottom of the grooves. The pressure distribution and flow downstream of the grooves also vary as the increase of the width-to-depth ratio, the flow separation reduce the area acting by viscous drag, and also arise the pressure difference. This induces the reduction of the viscous drag and the increment of the pressure drag. As the former one is lower than the latter one, the total drag force of surface with transverse rectangular grooves is always larger than that of the smooth surface.

1. SIMULATION OF DRAG CHARACTERISTIC OF TRANSVERSE GROOVES IN TURBULENT FLOW Drag reduction by grooves (also called “riblets”) in turbulent flow is initialed by Walsh and his colleagues at NASA-Langley laboratory[17]. They had proved that the geometrical modified grooves aligned with the flow stream can reduce the drag force. Nowadays, abundant researches are still continuously carried out to deepen the understanding of the mechanism and optimize the cross-section shape and scale of these longitudinal grooves[1820]. However, the researches on the other kind of grooves – the transverse grooves, are much less than the former one. In some of the pioneer study[21] on the drag characteristic of the transverse pattern, only the height parameter is adopted to represent the effect of the patterns, but the simulation result are not adequate. In most of the industrial problems, the pattern’s height or depth is in the same order of the thickness of the boundary layer, so the dimension of the patterns in the direction aligned with the flow must be taken into consideration. So far, the drag characteristic of the surface with transverse grooves is still uncertain[22]. In the following study, the CFD methods based on the RANS equations coupled with the RNG k-ε model to analysis the drag characteristic of the surface with transverse grooves are demonstrated and the drag force composition on the grooved surface are solved. The effect of the pressure drag force is discussed.

Computational Studies on Drag Reduction Effect by Surface Grooves

231

1.1.Geometry Models Figure 1 shows the two-dimensional geometry model of a flat surface with periodic transverse rectangular grooves. Three geometric parameters are defined to demonstrate this kind of grooves, and they are the groove width (w), the groove depth (k), and the groove wavelength (λ).

Figure 1. Geometry model of surface with periodic transverse rectangular grooves

1.2. CFD Simulation Method for Grooved Surface in Turbulent Flow The Reynolds average Navier-Stocks (RANS) equation [23-25] is one of the most popular physical models for the turbulent flow. It split the motion of the flow into two parts: the average motion and the random fluctuation. The former one is calculated based on the traditional Navier-Stocks equation and the latter one is noted as the Reynolds stress term. The RANS equation is written as follows:

∂ ( ρ u i ) ∂ ( ρu i u j ) ∂p ∂ + =− + ∂t ∂x j ∂xi ∂x j where,

− ρ ui'u 'j

⎛ ∂u i ⎜μ ⎜ ∂x j ⎝

⎞ ⎟ − ρ u i′u ′j ⎟ ⎠

(1)

is known as the Reynolds stress term or turbulent stress term.

To solve the RANS equation, turbulent models for Reynolds stress term should be chosen. k-ε model is one of the turbulent models which is widely used nowadays. Here k is

k = u ′u ′ 2

i i , and ε is the turbulent the turbulent kinetic energy, which can be defined as dissipation rate of the fluctuation kinetic energy per unit mass, which represents the transfer rate of the energy of the isotropy small scale vortices from mechanical to thermal:

ε = μ (∂u i′ / ∂x j ) ⋅ (∂u i′ / ∂x j ) / ρ . k and ε could be solved by the follow two differential equations:

∂ ∂ ( ρk ) ∂ ( ρkui ) = + ∂x j ∂xi ∂t

⎡ ∂k ⎤ ⎢α k (μ + μt ) ⎥ + Gk + ρε ∂x j ⎥⎦ ⎢⎣

(2)

232

Haosheng Chen and Yongjian Li

∂ ∂ ( ρε ) ∂ ( ρεu i ) = + ∂x j ∂xi ∂t where, Gk =

⎡ ∂ε ⎤ C1∗ε ε ε2 G k − C 2ε ρ ⎢α ε (μ + μ t ) ⎥+ ∂x j ⎥⎦ k k ⎢⎣

(3)

Cμ ρk 2 ⎛ ∂ui ∂u j ⎞ ∂ui ⎜ ⎟ + , the generation term of the turbulent kinetic induced ε ⎜⎝ ∂x j ∂xi ⎟⎠ ∂x j

by the average velocity gradient;

Eij =

1 ⎛⎜ ∂u i ∂u j + 2 ⎜⎝ ∂x j ∂xi

⎞ ⎟ , η = (2 E ⋅ E )1/ 2 k / ε , ij ij ⎟ ⎠

C1∗ε = C1ε − η (1 − η / η 0 ) /(1 + βη 3 ) , Cμ = 0.0845 , C1ε = 1.42 , C2ε = 1.68 , α k = α ε = 1.39 , η 0 = 4.377 , β = 0.012 . The model constants above are for the RNG k-ε model which is short for renormalization group k-ε model. This model[26] expand the unsteady Navier-Stokes equations according to the Gauss statistic expansion, and derive the high Re number k-ε model using the method that was used to filter the wave-number section of the fluctuation spectrum. RNG k-ε model has the same form of the k and ε as the standard k-ε model, but its five model parameters are obtained theoretically but not experimentally. The values above are the newest set of parameters for RNG k-ε model being given in reference [27]. As the coefficient C1ε of the generation term of the ε equation is calculated basing on the time-averaged strain rate Ei,j in the RNG k-ε model, it is a function of the coordinates in space. At present, the RNG k-ε model is believed to be one the best two-equation models with good combination property[28-31]. After the solving of the equations for k and ε, the turbulent viscosity stress (μt) can be calculated according to Eq.(4), and finally the turbulent stress can be figured out according to the Boussinesq assumption (Eq.(5)).

μ t = C μ ρk 2 / ε ⎛ ∂u ∂u j − ρ u i′u ′j = μ t ⎜ i + ⎜ ∂x ⎝ j ∂xi

⎞ 2⎛ ⎞ ⎟ − ⎜ ρk + μ t ∂u i ⎟δ ij ⎜ ⎟ 3 ∂xi ⎟⎠ ⎝ ⎠

(4)

(5)

where, δij is the Kronecher delta. when i = j, δij = 1; and when i ≠ j, δij = 0. In the solving process using RANS methods, if the first layer mesh next to the wall locates outside the iso-stressed zone, then the wall function should be adopted at the first layer of the mesh as the boundary condition; and else if the first layer mesh can reach the

Computational Studies on Drag Reduction Effect by Surface Grooves

233

viscous sub-layer, then the RANS model should be modified to solve the near wall problems. The common method to modify the model is to introduce the damp coefficient fμ, and the vortex viscosity coefficient can be written as:

ν t = c μ f μ u 'l

(6)

' where, u is the characteristic velocity of the turbulent fluctuation, u = '

k ; l is the

characteristic length, l = k / ε . In the calculations of this article with RANS methods, the first layer mesh are all set to be within the viscous sub-layer, so Eq.(6) is conducted to modify the low Reynolds number effect in the near wall zones. Figure 3.2 shows the mesh model of the surface with periodic square cross-section grooves. The grooves are located on the lower surface of the two-dimensional channel. d is the groove depth, w is the groove width, and λ is the groove wavelength. The upper surface of the channel is assumed to be flat and smooth. H is the height of the channel. The upper and lower surfaces of the channel are set to be non-slip. Periodic boundary condition is applied and three geometric periodicities are modeled in each case. The Reynolds number of the flow is calculated according to the half-height of the channel and the main flow velocity. The mesh around the grooves was refined as the variations of the flow around those areas are significant and more accurate details are necessary. Finite volume method are used to solve the RANS equations of incompressible fluid with the commercial CFD software FLUENT. The turbulent models are based on the RNG k-ε model and pressure gradient effect is considered. QUICK format is employed in the discretization of momentum equations, PRESTO format are used to discrete the pressure terms and SimpleC algorithm is applied in the coupling of the pressure and velocity. Second order upwind format is conducted in the solving process of k and ε. Reference [32]~ [34] have introduced the details of the numerical algorithms mentioned above. 3/ 2

Figure 2. Mesh mode

Figure 3. Examine results of mesh independency

234

Haosheng Chen and Yongjian Li

Examination of the mesh independence and validation of the numerical results The fineness of the mesh is important for the numerical simulation. Mesh independence mush have been examined to validate the credibility of the numerical results. As the thickness of the boundary layer of the turbulent flow is quick small that if the mesh is not fine enough, the accurate parameters of the flow could not be obtained. As shown in Figure 3, four sets of simulation results have been obtained base on different meshes. The fineness of these four set of mesh increase gradually. The mesh number of the first set is 2688, and it is the coarsest one. The mesh of the latter one is one times finer than the former one both in the flow direction and the normal direction. Figure 3 shows the numerical simulation results based on these four sets of mesh. The pressure gradient and drag coefficient vary as the increase of the mesh number. According to the comparison of the simulation results, it can be found that for this problem when the mesh density is as fine as or finer than the second set, the calculation results are almost mesh independent. The numerical results could also be validated by comparing with the experimental results. Djenidi, et al[35] (1999) used the LDV to test and observe the flow around the transverse grooves (w/d 1, d=5mm) in the water tunnel whose cross section is 250×250mm2. They measured the velocity profile of a serious of sections located in the middle of the grooves. The lower starting point in normal direction of each measurement is the top of the grooves. The average velocity profile obtained in the test is based on the thickness of the boundary layer. For convenience, numerical results and test data are compared base on the maximum velocity location, and they are in good agreement (as shown in Figure 4).

Figure 4. Comparison of the average velocity profile between experiments and simulations

To sum up, it has been discussed above that the Reynolds averaged Navier-Stokes equations, coupled with the RNG k-ε turbulence model with the near wall modifying treatment, is adopted to investigate the turbulent flow over a plat surface with transverse grooves.

Computational Studies on Drag Reduction Effect by Surface Grooves

235

General drag characteristic of the transverse grooves in turbulent flow The objective of this research is the make further understanding of the drag characteristic of the grooved surface in turbulent flow. As mentioned in Section 1, for transverse grooves, the drag acting on the surface could be classified into two groups: the pressure drag which is acting on the grooves’ surface perpendicular to the flow, and the viscous drag, which is acting on the grooves’ surface parallel to the flow. Here, we can define the dimensionless coefficient of the drag for comparison among grooved surfaces in different scales.

Cd =

C dp =

C dv =

2( Ddp + Ddv ) 2 Dd = 2 ρU ∞ Ω t ρU ∞2 Ω t

2 Ddp

ρU Ω t 2 ∞

=

(7)

2 ∫ τ cos(t 1 , x1 )dΩ t Ωt

ρU ∞2 Ω t

2 ∫ p cos(n 1 , x1 )dΩ t 2 Ddv Ωt = 2 ρU ∞ Ω t ρU ∞2 Ω t

(8)

(9)

where, Cd, Cdp and Cdv are total drag coefficient, the pressure drag coefficient, and the viscous drag coefficient respectively; Dd, Ddp, and Ddv are the total drag force, the pressure drag and the viscous drag respectively; Ωt is the wetted perimeter; τ is the viscous shear stress, p is the pressure, t1 is the unit vector in the tangential direction, and n1 is unit vector in the exterior normal direction. For periodic square grooves (w/d = 1, λ/w = 2), the groove width could be defined as the characteristic length of the groove. As shown in Figure 5, the pressure rise induced by the grooves increases as the rise of the characteristic length of the grooves (Re = 4.5×104). This pressure fluctuation causes remarkable pressure difference between the windward side and leeward side of the grooves and forms a significant pressure drag. Figure 6(a) and (b) show the variation of the drag forces (the total drag, the pressure drag and the viscous drag) and the proportion of the pressure drag and viscous drag to the total drag. According to the curves and charts, two conclusions can be drawn for the drag characteristic of the surface with periodic transverse square grooves: (1) The drag coefficients are almost independent with the scale of the grooves; (2) The pressure drag is about 4 times as large as the viscous drag. As mentioned above the pressure drag is caused by the pressure difference acting on the grooves’ surface perpendicular to the flow, and it is a kind of side wall effect which has been discussed in Section 1. The difference between the results obtained in laminar flow and turbulent flow is that the side wall effect is more significant in the latter case. So it should be concluded that the real profile of the groove is important for the consideration of the effect of

236

Haosheng Chen and Yongjian Li

grooves on the drag characteristic of the grooved surface. In some traditional researches on drag reduction of rough surface in turbulent flow, where only the friction drag (the same as the viscous drag mentioned in this article) is taken into consideration, were not adequate. The pressure drag must be considered, and Reynolds equations should not be used to analyze the drag characteristic of the transverse grooves in turbulent flow.

Figure 5. Pressure fluctuation cause by square grooves with different characteristic length

Figure 6. Drag coefficients of square grooves with different characteristic length

2. TRANSVERSE GROOVES ON DRAG CHARACTERISTIC OF THE SURFACE IN TURBULENT FLOW 2.1. Effect of Reynolds Number The relation between the drag characteristic of the grooved surface and Reynolds number has been investigated as follows. Figure 7(a) shows variation of the drag coefficient of the

Computational Studies on Drag Reduction Effect by Surface Grooves

237

periodic transverse square grooves. As the increase of the Reynolds number, the drag coefficients drop down at the same time. When Reynolds number increases from 10, 000 to 50, 000, drop velocity of the viscous drag coefficient is much larger than that of the pressure drag coefficient. When Reynolds number is larger than 100, 000, the variation of the three drag coefficients slow down significantly. Figure 7(b) demonstrates the variation of the proportion of the pressure and viscous drag coefficient to the total drag coefficient as the increase of Reynolds number. When Reynolds number increases from 10, 000 to 50, 000, the quotient of the viscous drag decrease quickly, and when Reynolds number rises continuously, the variation of the quotient of the viscous drag coefficient are slight.

Figure 7. Drag coefficients of square grooves with different Reynolds number

Reynolds number is a parameter represents the state and structure of the flow. The results above indicate that the drag characteristic of the grooved surface is affected by the structure of the flow. For square grooves, in the range that Reynolds number varies from 10, 000 to 50, 000, the structure of the flow may change more rapidly.

2.2. Effect of Width-to-Depth Ratio Computational analysis in this part was performed to investigate the drag characteristics of the surface grooves with different geometric parameters. For the rectangular grooves, effect of the width-to-depth ratio of the grooves was focused on. The following simulation were carried out in condition that Reynolds number of the flow is 10, 000. Figure 8 shows the velocity stream line diagram and static pressure distribution on the cross-section of grooves in different width-to-depth ratio (from 1/4 to 10/1). When w/d is small (less than 2), single vortex almost takes up the whole groove. The rotating center of the vortex and the center of the groove almost coincide. This is useful to prevent the separation of the flow. But when w/d increases gradually, a negative pressure region will generate just in the downstream zone of the windward side of the groove and induce the local flow separation. Generally speaking, the separation induced by the grooves will reduce the friction drag but increase the pressure drag. So w/d could be regarded as a critical parameter to determine the flow state and drag characteristic.

238

Haosheng Chen and Yongjian Li

Figure 8. Streamlined diagram and pressure distribution rectangular grooved surfaces with different w/d ratios

Computational Studies on Drag Reduction Effect by Surface Grooves

239

When w/d rises to 7, the single vortex separates into two. These two vortices locate in the two ends of the groove respectively. When w/d is larger than 7, the main stream will almost reach the bottom of grooves. Besides the negative pressure region around the downstream zone of the windward side, the center of the vortex’s pressure is quick low too. The pressure in these two region decrease as the grooves’ width increases. The former one changes more rapidly.

Drag characteristic of the rectangular grooves Figure 9 shows the variations of the drag coefficients as the change of the width-to-depth ratio of the grooves. The variation of the total drag coefficient, the pressure drag coefficient and the viscous drag coefficient are drawn respectively. According to the curves, it can be concluded that the grooves with small width-to-depth ratio are more favorable in drag reduction design. While w/d is between 1/1.5 and 2/1, the drag coefficient rises quickly in a linear way. When w/d=3, the rise of the drag coefficient slows down. While w/d is between 6 and 7, the drag coefficient would reach it maximum value. It should be noted that when w/d equals to 7, two totally separate vortices are just going to generate.

Figure 9. Drag coefficients of square grooves with different width-to-depth ratio

When w/d<1/1.5, the viscous drag is larger than the pressure drag, and when w/d = 1/4, viscous drag makes up more than 80% of the total drag. When w/d >1/1.5, pressure drag rises quickly, while the viscous drag reduce slowly. When w/d is large than 2, the viscous drag becomes negative which means the viscous drag is in the opposite direction of the flow. This phenomenon can be explained as follows: as the increase of the grooves’ width, the separation aggravate and the area where the viscous drag is in the direction of the flow reduces, while the friction drag in the grooves whose direction is opposite to the flow direction increases, and the integral effect is that the total viscous drag becomes negative. When w/d is larger than 1/1.5, the pressure drag becomes the major part of the total drag force, which make the effect of viscous drag become less and less important. When w/d is between 6 and 7, the pressure drag coefficient reaches its maximum value. At the same time, the viscous drag coefficient arrived at it minimum point. When w/d is larger than 7, the main

240

Haosheng Chen and Yongjian Li

flow begins to reach the bottom of the grooves directly, and the viscous drag starts to rise. And the pressure drag force begins to reduce slowly after the generation of the two vortices as the continuous increase of the grooves’ width. To sum up, the drag characteristic of the rectangular grooved surface is concerned with flow influenced by the grooves. The vortices in the grooves and their variation process play an important part. The processes have direct effects on the main stream flow, the distribution of the turbulent intensity and the flow condition next to the surface. The combined effect will determine the drag force on the wall. Decreasing the width of the grooves is an effective way to reduce the drag force.

3. CONCLUSION In this section, the drag characteristic of the surface with transverse grooves in turbulent flow is numerically studied. The pressure drag plays an important role in the total drag force. The pressure drag and viscous friction drag are all relative to the turbulent flow structure. According to the numerical results, following conclusions can be drawn: (1) The side wall effect of the transverse grooves are more significant in turbulent flow than in the laminar flow, the pressure drag inducing by the side wall effect could make up about 80% of the total drag force when the Reynolds number of the flow is higher than 50, 000. (2) The turbulent flow structure in the grooves is considered to be the reason for the variation of the pressure and viscous drag. And it is affected by the shape of the grooves, such as the width-to-depth ratio. The generation and slitting of the vortexes can change the direction of the net viscous drag, and also change the pressure difference between the windward and leeward side of the grooves. (3) For rectangular grooves, as the reduction of the viscous drag is lower than the increment of the pressure drag, there is an increase in the overall surface with transverse rectangular grooves comparing with the smooth one.

NOMENCLATURE p μ h(x) h1 h0 B U

σ d

λ

= Fluid pressure = Viscosity of fluid = nominal film thickness = maximum film thickness = minimum film thickness = length of the disk = moving speed of the slider = surface roughness height = groove depth = wavelength of the squared transverse waves

Computational Studies on Drag Reduction Effect by Surface Grooves

β

241

= ratio of roughness height to the film thickness; also model constant for RNG

k-ε model

ε ψi(H)

= ratio of average film thickness to disk length; also turbulent dissipation rate = turbulent kinetic energy = expectancy function = expectancy of the film thickness function

h ( x)

= nominal film thickness

τij γ&

= stress on direction i and j = shear rate.

ρ

= density of fluid. = Coordinates directions = Velocity along direction x, y and z, respectively; w: also groove width = velocity tensor = stress tensor

k E

x,y,z u,v,w u τ

ACKNOWLEDGEMENT This work is supported by the State Key Laboratory of Tribology with the funding (No. SKLT08B03), NSFC Project (No. 50975158 and No. 50505020) and the Doctoral Program of Higher Education (No. 20070003103).

REFERENCES [1] [2] [3] [4] [5]

[6]

[7] [8]

Elrod, HG. A General theory for laminar lubrication with Reynolds roughness, ASME Journal of Lubrication Technology, 1979, 101, pp: 8-14. Dowson, D; Whomes, TL. The effect of surface roughness upon the lubrication of rigid cylindrical rollers- I. Theoretical, Wear, 1971, 18, pp: 129-150. Shelly, P; Ettles, C. Effect of transverse and longitudinal surface waviness on the operation of journal bearings, Jnl. Mech. Engrg. Sc., 1972, 14, pp: 168-172. Chang, L. Deterministic modeling and numerical simulation of lubrication between rough surfaces- a review of recent developments. Wear, 1995, 184, pp: 155-160. Chiang, HL; Hsu, CH. Lubrication performance of finite journal bearings considering effects of couple stresses and surface roughness, Tribology International, 2004, 37, pp: 297–307. Wang, LL; Cheng, IW. “An average Reynolds equation for non-Newtonian fluid with application to the lubrication of the magnetic head-disk interface,” Tribology Transaction, 1997, 40(1), pp: 111-119. Elsharkawy, AA. On the hydrodynamic liquid lubrication analysis of slider/disk interface. International journal of mechanical sciences, 2001, 43, pp: 177-192. Yang, M, Talke, F. E. Effects of gas composition, humidity and temperature on the tribology of the head/disk interface- Part I. Experiments. Tribology Transaction, 1996, 39(6), pp: 615-620.

242 [9] [10] [11]

[12]

[13] [14]

[15] [16] [17] [18] [19]

[20]

[21] [22]

[23] [24]

[25] [26] [27]

Haosheng Chen and Yongjian Li Boxenhorn, B; Greiff, P. Monolithic silicon accelerometer. Sensor and actuators, 1990, A21-A23, pp: 273-277. Christensen, H. Stochastic models for hydrodynamic lubrication of rough surfaces. Proc. Inst. Mech Engrs., 1969, 184, pp: 1013-1022. Patir, N; Cheng, HS; An average flow model for determining effects of threedimensional roughness on partial hydrodynamic lubrication. Journal of lubrication technology, 1978, 100(1), pp: 12-17. Chin, R. Numerical studies of slow viscous flow between a moving plane wall and a stationary wavy wall. Case western reserve university report, FTAS/TR, 1972,(6), PP: 72-79. Christensen, H; Tonder, K. The hydrodynamic lubrication of rough bearing surfaces of finite width. Journal of lubrication technology, 1971, (7), pp: 324-330. Almqvist, T; Larsson, R. Some remarks on the validity of Reynolds equation in the modeling of lubricant film flows of the surface roughness scale. Journal of tribology, 2004, 126(10), pp: 703. Sun, DC; Chen, KK. First effects of Stokes roughness on hydrodynamic lubrication. Journal of lubrication technology, 1977, (1), pp: 2-9. Elsharkawy, AA. On the Hydrodynamic Liquid Lubrication Analysis of Slider/Disk Interface. International journal of mechanical sciences, 2001, 43, pp: 177-192. Walsh, MJ; Lindemann, AM. Optimization and Application of Riblets for Turbulent Drag Reduction. AIAA-84-0347. Walsh, MJ. Riblets as a viscous drag reduction technique. AIAA Journal, 1983, 21(4), pp: 485-486. El-Samni, OA; Chun, HH; Yoon, HS. Drag reduction of turbulent flow over thin rectangular riblets. International Journal of Engineering Science, 2007, 45, pp: 436-454. Liu, ZH; Dong, WC; Xia, F. The effects of the tip shape of V-groove on drag reduction and flow field characteristics by numerical analysis. Journal of Hydrodynamics, 2006, 21(2), pp: 223-231. Nikuradse, K. Stromungsgesetze in Rauhen Rohren. [English Trans]. NACA Tech. Memo No. (1292). Wang, SL; Shi, XJ; Zhao, SH; and et al. The technical research and application development of groove surface in turbulent drag reduction. Journal of Southwest Petroleum University, 2008, 30(1), pp: 146-150. Versteeg, HK; Malalasekera, W. An Introduction to Computational Fluid Dynamics: the Finite Volume Method. Wiley, New York, 1995. Breuer, M; Jovicic, N; Mazaev, K. Comparison of DES, RANS and LES for the separated flow around a flat plate at high incidence. International Journal for Numerical Methods in Fluids, 2003, 41(4), pp: 357-388. Walter, DK; Leylek, JH. A new model for boundary layer transition using a singlepoint RANS approach. Journal of Turbomachinery, 2004, 126(1), pp: 193-202. Yakhot, V; Orzag, SA. Renormalization group analysis of turbulence: basic theoy. J Scient Comput, 1986, (1), pp: 3-11. Speziale, CG; Thangam, S. Analysis of an RNG based turbulence model for separated flows. Int. J. Engng. Sci., 1992, (10), pp: 1379-1388.

Computational Studies on Drag Reduction Effect by Surface Grooves

243

[28] Chen, Q. Comparison of different k-e models for indoor air flow computations. Numer Heat Transfer, Part A, 1995, 28, pp: 353-369. [29] Rodi, W. Experience with two-layer models combining the k-e model with a oneequation model near the wall. AIAA, 1991-0216. [30] Yakhot, V; Orzag, SA; Thangam, S; Gastki, TB; Speziale CG. Development of turbulence models for shear flows by a double expansion technique. Phys Fluids A, 1992, 4(2), pp: 1510-1520. [31] Kim, SW; Chen, CP. A multiple-time-scale turbulence model based on variable partitioning of the turbulent kinetic energy spectrum. Numerical Heat Transfer, Part B, 1989, 16, pp: 193-211. [32] Hayase, T; Humphrey, JAC; Greif, R. A consistently formulated QUICK scheme for fast and stable convergence using finite-volume iterative calculation procedures. Journal of Computational Physics, 1992, 98, pp: 108-118. [33] Van Doormal, JP: Raithby, GG. Enhancement of the SIMPLE method for predicting incompressible fluid flows. Numerical Heat Transfer, 1984, 7, pp: 147-163. [34] Patanker, SV; Spalding, DB. A calculation procedure for heat, mass and momentum transfer in three-dimensional parabolic flows. Int J Heat Mass Transfer, 1990, 15, pp: 1787-1806. [35] Djenidi, L; Elavarasan, R; Antonia, RA. The turbulent boundary layer over transverse square cavities. J. Fluid Mech, 1999, 395, pp: 271-294.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 245-271

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 7

CURRENT LIMITING IN OXIDE CERAMIC STRUCTURES Alexander Bondarchuk* Technological University of Mixteca, Huajuapan de Leon, Oaxaca CP. 69000, Mexico

ABSTRACT The review of physical phenomena which lead to current limiting behavior (current is increased weaker than voltage, saturated and even decrease) in oxide ceramic structures are presented. Particular attention is given to the mechanisms of current limiting in materials whose electrical conductivity is controlled by potential barriers at the grain boundaries. In particular, the current saturation effect in the nano-grained ceramic films is examined. The consideration of physical models describing the current limiting in specific material is followed by the short posing of main experimental data. For some models, the computational modeling has been performed.

INTRODUCTION The electrical properties of oxide ceramic structures are directly linked to electrically active interfaces and contacts between the adjacent grains. There are two types of contacts between grains in polycrystalline semiconductors [1]: (I) contacts modeled by the double Schottky barrier (Figure 1a) and (II) the intergranular regions in form of a neck (Figure 1b, c). Usually, the grain contacts of the first type in oxide ceramic structures can appear at the sintering in the oxidizing atmosphere. The contacts of second type (the neck) between the adjacent grains arise at the sintering of ceramics in the vacuum or in the inert atmosphere [1]. The neck, depending on its thickness, defect concentration in the intergranular region and a magnitude of the surface charge, can be (Figure 1b) or can not be (Figure 1c) a barrier for the flow of electronic current between grains. If the potential barrier in the neck exists and its

*

Corresponding author: [email protected]

246

Alexander Bondarchuk

height is larger than kT , then such intergranular region is called the closed neck [1-3]. This type of neck (Figure 1b) exists at the overlapping of the depletion layers in the intergranular region that occur at the small thickness of the neck with a large magnitude of the surface charge. If the neck thickness is sufficiently large and the overlapping of depletion layers does not take place (Figure 1c), this is the open neck [1-3]. In this case, the barrier for the flow of the electronic current does not exist.

Figure 1. Schematic layout of a contact between grains in polycrystalline n-type semiconductors: (a) modeled by the double Schottky barrier; (b) open neck; (c) closed neck.

Potential barriers make the grain boundaries (GBs) highly resistive compared to the grain interiors and therefore the current flow through the material bulk is controlled by GBs. In polycrystalline n-type semiconductors, the potential barrier is formed by capturing of electrons by localized states in the GB, resulting in a negative surface charge [4-6]. There are several ways which such grain-boundary states can arise. The lattice mismatch alone is sufficient to cause a planar array of localized states at the grain boundaries. Also, the boundary regions may contain a large number of point defects and impurities. According to [1, 7], the interface states can be generated by the adsorbed oxygen. Any of these conditions can cause a space-charge layer to form on the grain boundary. This negative interface charge is compensated by ionized, positively charged donors in a space charged region formed on each side of the grain boundary. The formed grain-boundary barrier determines the current flow between adjacent grains. Therefore the conductance of the material is controlled by grain-boundary barriers and its nonlinear electrical property (deviation from Ohm’s law) is a cumulative effect caused by individual GBs. Deviation from Ohm’s law in oxide ceramic structures can be displayed in the form of β

the superlinear current-voltage characteristic I ~ U ,

β > 1 (varistor effect) or the sublinear

Current Limiting in Oxide Ceramic Structures

247

I (U ) dependence with β < 1 (current limiting effect). The varistor effect is related to the barrier height lowering on the grains boundary with the voltage increase. A detailed description of such processes and the theory of the varistor behavior have been presented in [4-6]. In the present time, the superlinear current-voltage characteristic has been observed in ceramics on the basis of various n-type semiconductors ZnO, SnO2, TiO2, WO3 [8-20]. Such ceramics with their polycrystalline nature and a large number of intergranular barriers (i.e., a high energy absorption capability) exhibit the superlinear current-voltage characteristics and are used as active components in overvoltage protection devices, in the varistores. The varistores behave as variable resistors: they exhibit the high resistance at the low voltage (insulators) and the low resistance at the high voltage (conductors). When the voltage exceeds a certain value, for instance, during a voltage transient or surge, the varistor becomes highly conducting and draws the current through it, usually to the ground. Due to their high reliability and the high energy absorption capability, the varistors play an important role in the radio engineering and power electronics. To make such electronic devices the deep understanding of the nature of the varistor effect is required. Ceramics with the ability to limit the current might have a great potential in new overvoltage protection devices as well. However, in contrast to the varistor effect, the nature of the current limiting behavior can be different in different materials. In many cases, the mechanism of current limiting is not studied completely yet and the experimental facts of its observation are not described sufficiently in the literature or this information is fragmentary. In this chapter the short review of physical phenomena which lead to the current limiting effect in the oxide ceramic structures are presented. This work was performed with the aim to collect known experimental facts of current limiting in the oxide ceramic structures and to consider the physical models describing such current behavior. The author hopes that this can be useful for the systematical development of such materials and improvement of their engineering performance. 07.03.2009

Alexander Bondarchuk

I. CURRENT LIMITING BEHAVIOR AS A RESULT OF THE CAPTURE OF ELECTRONS AT GRAIN-BOUNDARY STATES Current limiting behavior (current is increased weaker than voltage) caused by capturing of electrons by localized states at the GBs can be observed in polycrystalline semiconductors whose conductance is controlled by grain-boundary barriers [4, 21-24]. Sublinear currentvoltage dependence ( β < 1 ) appears at intermediate voltages between Ohmic ( β = 1 ) and superlinear ( β > 1 ) regions (Figure 2) and can be explained in terms of the barrier model in following way. Let us assume that the potential barrier at a single grain boundary in polycrystalline semiconductor can be modeled by the double Schottky barrier (two back-to-back connected Schottky barriers) with zero thickness of intergranular layer (Figure 3). It is assumed too that at the surface of the adjacent grains the localized electronic states exist due to the oxygen adsorption, the presence of defects and dopants. Such grain-boundary states can catch

248

Alexander Bondarchuk

electrons moving from the interior of the grains to the boundary. Under equilibrium some of these states, i.e., roughly those below the Fermi level, are occupied by electrons, giving a negative surface charge (full circles in Figure 3a). However, some grain-boundary states (above the Fermi level) are empty (open circles in Figure 3a). The negative charge localized at the grain boundary is shielded by a region of positive space charge situated on the two sides of the grain boundary. Assume as well that this positive space charge is due to the fully ionized shallow donors. Then the condition of electroneutrality for the grain boundary can be presented in the form:

m = N D ( L1 + L2 )

(1)

where m is the surface density of electrons captured at the grain-boundary states, N D is the donor density in the grain, L1 and L2 are the thicknesses of the space-charge regions in grains.

Figure 2. Schematic of current-voltage characteristic (I— Ohmic, II— sublinear, and III– superlinear region) for single grain boundary in polycrystalline semiconductor.

Assuming that the discrete nature of the charges can be ignored, the magnitude of the potential barrier can be calculated by solving the Poisson equation for the potential, Φ (x ) , from a knowledge of the grain-boundary charge density,

ρ ( x ) [20]:

ρ ( x) d2 Φ( x) = 2 εε 0 dx where

(2)

ε is the relative permittivity and ε 0 is the permittivity of free space.

Under equilibrium at zero applied bias voltage (Figure 3a), the magnitude of the potential barrier, ϕ 0 , is [20, 23] :

Current Limiting in Oxide Ceramic Structures

ϕ0 =

q 2 L20 N D q 2 m0 2 = 2εε 0 8εε 0 N D

249

(3)

Here q is the elementary charge, L0 is the thicknesses of the space-charge region in separate grain at zero applied bias voltage ( L0 = L1 = L2 ), m0 is the surface density of electrons captured at the grain-boundary states at U = 0 .

Figure 3. Energy level diagram for a grain boundary (a) at zero applied bias voltage and (b) under applied voltage U .

From the solution of the Poisson equation, the following expressions can be written [23]:

ϕ 0 − qU1 =

q 2 L12 N D 2εε 0

(4)

ϕ 0 + qU 2 =

q 2 L22 N D 2εε 0

(5)

U = U1 + U 2 where U is a voltage applied across the grain boundary (Figure 3b).

(6)

250

ϕ

Alexander Bondarchuk

Using the equations (1) and (3)-(6), the barrier height of forward-biased Schottky barrier = ϕ 0 − qU 1 at a given voltage U (Figure 3b) can be obtained in the form 2

where b = q

2

⎞ b m2 ⎛ ⎜1 − qU ⎟ , ϕ (U ) = 4 ⎜⎝ b m 2 ⎟⎠

(2εε 0 N D ) .

(7)

Then the current density across the grain boundary (Figure 3b) in diode approximation can be presented as

j = j1 − j2 , where

j1 = 1 4 qυN D exp[− ϕ (U ) kT ] ,

j2 = 1 4 qυN D exp[− (ϕ (U ) + qU ) kT ] .

Here υ is the average thermal velocity of electrons, k is the Boltzmann constant and T is the absolute temperature. Or in compact form:

j=

q υ ND 4

⎛ ϕ (U ) ⎞ exp⎜ − ⎟ ⎝ kT ⎠

⎛ ⎛ qU ⎞ ⎞ ⎜1 − exp⎜ − ⎟⎟ , ⎝ kT ⎠ ⎠ ⎝

(8)

The surface density of trapped electrons is determined by the capture of electrons and the emission processes at grain-boundary states [23]:

dm ⎛ dm ⎞ ⎛ dm ⎞ =⎜ ⎟ −⎜ ⎟ , dt ⎝ dt ⎠1 ⎝ dt ⎠ 2

j ⎛ j ⎛ dm ⎞ ⎜ ⎟ = σ ( M − m ) ⎜⎜ 1 + 2 q ⎝ dt ⎠ 1 ⎝ q ⎛ ES ⎛ dm ⎞ ⎜ ⎟ = σ υ m N C exp⎜ − ⎝ dt ⎠ 2 ⎝ kT

⎞ ⎟⎟, ⎠

⎞ ⎟. ⎠

Here (dm dt )1 is the capture rate of electrons which is proportional to the quantity of electrons crossing the grain boundary, the capture cross-section

σ and the density of empty

grain-boundary states (M − m ) (where M is the surface density of the localized electronic states); (dm dt )1 is the emission rate of electrons from the grain-boundary states into the

conduction band, E S is the ionization energy of the grain-boundary state.

Current Limiting in Oxide Ceramic Structures

251

υ ND dm ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞ exp⎜ − = σ ( M − m) ⎟ ⎜1 + exp⎜ − ⎟⎟ 4 dt ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠ ⎛ E ⎞ − σ υ m N C exp⎜ − S ⎟. ⎝ kT ⎠

(9)

Thus,

The equation (9) describes the filling of grain-board states with voltage increase. The general picture of such processes can be given in the following way. Under equilibrium, some of the grain-board states remain empty at zero applied bias voltage (Figure 3a). Application of dc voltage causes the barrier decrease. Therefore the capture rate of electrons (dm dt )1 is increased and the surface density of trapped electrons

m is increased. The increase of negative charge localized at the grain boundary leads to some increase of the barrier height and the rate of capture process (dm dt )1 is decreased. The capture process is “autohampering” [24]. At low voltages, the change of the occupation of the grain-boundary states is negligible due to a weak change of the barrier height and Ohm’s law is performed. Further voltage increase leads to the stronger barrier decrease and the capture rate of electrons is increased. The capture of additional electrons at the grain boundary results in some rise of the barrier height that somewhat compensates the decrease of the barrier height after an application of voltage. Therefore the current density j is increased weaker than voltage, i.e. the sublinear

j (U ) dependence takes place. The voltage increase leads to the filling of the grain-boundary states and the density of empty grain-boundary states, ( M − m ) , is decreased. Therefore, the intensity of a capture process is decreased at high voltage and the occupation of the grain-boundary states is weakly changed. In these conditions, as a result of the sharp lowering of the barrier height under applied high voltage, the sublinear region at j (U ) curve is changed to the superlinear one. Calculated dependences of j and

ϕ versus U are shown in Figure 4. The calculus was

performed at the assumption that the electronic equilibrium ( dm dt = 0 ) at the grain boundary are reached at each discrete value of voltage. The next parameters were used [23]: 18

the effective density of states in the conduction band Nc = 6.0 × 10 cm thermal velocity of electrons

−3

, the average

υ = 1.0 × 107 cm s −1 , the donor density in the grain

N D = 1.0 × 1017 cm −3 , the ionization energy of grain-boundary states E S = 0.6 eV , the relative dielectric permittivity of the grain is assumed to be ε = 8.5 , the absolute temperature T = 290 K , and the value of the barrier height

ϕ 0 = 0.5 eV at U = 0 .

For convenience, the results of calculations are presented in the normalized form. The current density is normalized on the value j0 = qυN D 4 and multiplied by coefficient 10 . 10

252

Alexander Bondarchuk

The magnitude of the potential barrier and voltage is normalized on the value of the barrier height at U = 0 , ϕ 0 / q .

Figure 4. Normalized dependences of the current density (curve 1) and the barrier height (curve 2) on the voltage.

Calculated current-voltage dependences (Figure 4) are contained sublinear region (current limiting) in compliance with the qualitative consideration given above and the experiment data [21, 22].

II. CURRENT LIMITING EFFECT CAUSED BY ADSORPTION PROCESSES IN ELECTRIC FIELD Recently [25–29] it was reported about preparation of In2O3-SrO ceramics with the capability to limit the current by voltage increase. This type of current-voltage dependence in In2O3-SrO ceramics sample is shown in Figure 5 in linear (curve 1) and logarithmic (curve 2) scale. It contains linear region ab at low voltages (Ohm’s law), superlinear bc at higher voltages which is followed by the current limiting region: sublinear part cd and region de where current is even decreased (Figure 5, curve 1). Next measurements give nearly the same curve with current limiting region but it is slightly shifted at low voltages (Ohm’s law) to higher currents due to the degradation phenomena related to water adsorption [29]. The adsorption effects were studied at room temperature just after the cooling the ceramics sample from high temperature in the air [29]. The increase and subsequent decrease of the ceramics resistance on time at fixed voltage from Ohm’s region was found. During about 1–2 h just after the cooling to room temperature oxygen adsorption takes place and resistance is increased. Later on water adsorption causes the decrease of the resistance. Adsorption processes at the grain surface in the bulk of this material are likely due to its high porosity revealed by scanning electron microscopy [29]. It was found out that current saturation region is observed if measurements were carried out in air and it is not observed if measurements were performed in argon [29]. In argon

Current Limiting in Oxide Ceramic Structures

253

I (U ) dependence was nearly linear. In addition it was shown that even if current saturation region is disappeared as a result of long-term degradation, current limiting effect in In2O3-SrO ceramics can be reproduced in air again after the heat treatment of a sample in Ar at 1070 K. In that way, it was ascertained sensibility of ceramics to external environment. Current limiting region is observed only if there is enough oxygen in atmosphere and it is not observed if there is a low oxygen content in atmosphere. Therefore, this current limiting effect rather can be related to the adsorption of oxygen.

Figure 5. Current versus voltage for In2O3-SrO sample in air in linear (curve 1) and logarithmic (curve 2) scale.

However, the current limiting effect in In2O3-SrO ceramics can not be explained by the adsorption of oxygen which running independently of the applied voltage. Next experimental facts show that. The dependence of current on time in In2O3-SrO ceramics at fixed voltage is changed qualitatively with increase of voltage [25]: current is slightly increased with time at low voltages and it is decreased with time at voltages in the current limiting region ce in Figure 5. It means that processes responsible for discussed effect are controlled by voltage. In addition, the study of current-voltage dependences was always started in a few hour’s time after the cooling of a sample to room temperature, when water adsorption was dominated and the sample resistance was decreasing [29]. It confirms the fact of slight shift of I (U ) dependence in the Ohm’s law region at each measurement to higher current region. In view of these facts, the assumption about the relation of observed current limiting behavior to the oxygen adsorption (which can take place independently of the applied voltage) should be excluded. However, the oxygen adsorption can be induced and controlled by applied voltage [30, 31]. An application of voltage can cause a capture of electrons at empty grain-boundary states and, therefore, a sublinear I (U ) dependence can appear between linear and superlinear regions (see section I). However, in In2O3-SrO ceramics, on the contrary, the sublinear region

254

Alexander Bondarchuk

ce in Figure 5 is observed after the superlinear region. This fact cannot be explained in the frames of the model discussed in section I. For the explanation of the observed current limiting effect in In2O3-SrO ceramics it is useful to modify the existing barrier model. Let us assume that the density of grain-boundary states is increased with voltage due to the adsorption of oxygen: (i) the grain-boundary states are caused by the adsorbed oxygen; the capture of electrons at the grain-boundary states means the transition of adsorbed oxygen from neutral to charged form; the density of empty and filled grain-boundary states are equal to the density of adsorbed oxygen in neutral and charged forms, respectively; the correlation between the charged and neutral forms of adsorbed oxygen is controlled, according to Volkenshtein [7], by the position of the Fermi level at the surface of a sample; (ii) the adsorption equilibrium at the grain surface is kept up at the expense of adsorption and desorption oxygen only in the neutral form; oxygen in the charged form does not take part in an exchange with the atmosphere. On the basis of these assumptions the general picture of processes can be given in the following way [29]. At low voltages electronic equilibrium at the grain boundary is broken and the capture of electrons is dominated over their emission from the grain-boundary states. It leads to some increase of the barrier height and the rate of capture process is decreased. The capture process is “autohampering” [24]. Therefore, current limiting region can be observed after the Ohm’s law region (see section I). As a result of the electron capture the transformation of adsorbed oxygen to a charged form takes place, amount of adsorbed oxygen in the neutral form is decreasing and adsorption-desorption equilibrium at the grain boundaries is broken. This causes the additional oxygen adsorption. However, at low voltages it is negligible due to a weak deviation of the density of adsorbed oxygen in a neutral form from its equilibrium value. Respectively, at low voltages the additional oxygen adsorption cannot have strong influence on the number of empty grain-boundary states. With increase of voltage the barrier height is decreased and current is increased with voltage superlinearly. However, some capture processes are continued and, respectively, the density of adsorbed oxygen in a neutral form is decreased. It means that deviation from the adsorption equilibrium is being increased with voltage. This leads to a strong increase of additional oxygen adsorption. Therefore, the density of empty grain-boundary states can be gradually increased with voltage. As a result, the capture of electrons becomes stronger and superlinear region of I (U ) curve is replaced by the sublinear one or even the decrease of current could take place. The growth of the barrier height leads to the decrease of the capture rate and, therefore, it leads to the decrease of oxygen adsorption rate. Both processes are “auto-hampering” ones. In this way the equilibrium between capture and emission of electrons and between adsorption and desorption of oxygen is settled. The necessary condition for additional oxygen adsorption and realization of assumed mechanism of the barrier height growth is the presence of empty adsorption centers at the grain surface. However, during a long stay of a sample in air these centers can be filled with adsorbed water, physically adsorbed oxygen, etc. As a result, additional oxygen adsorption

Current Limiting in Oxide Ceramic Structures

255

caused by voltage application is negligible and the density of grain-boundary states is almost constant. Accordingly, I (U ) dependence of In2O3-SrO ceramic sample kept in air for a long time does not contain the current limiting region after superlinear one but the heat treatment of a sample in Ar at 1070 K again creates empty adsorption centres at the grain surface and current limiting region after superlinear one is appeared again. This can explain observed degradation behavior. Assume that the potential barrier at a single grain boundary in In2O3-SrO ceramics can be modelled by the double Schottky barrier considered in the section I. In this case, the current density j across the grain boundary in diode approximation will be determined by the following system of equations (more detail see in section I):

j=

q υ ND 4

⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞ exp⎜ − ⎟ ⎜1 − exp⎜ − ⎟ ⎟, ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠

(1)

2

⎞ b m2 ⎛ ⎜1 − qU ⎟ , ϕ (U ) = 2 4 ⎜⎝ b m ⎟⎠

υ ND dm ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞ ⎛ E ⎞ = σ ( M − m) exp⎜ − ⎟ ⎜1 + exp⎜ − ⎟ ⎟ − σ υ m N C exp⎜ − S ⎟, dt 4 ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠ ⎝ kT ⎠ where

(2)

(3)

ϕ is the magnitude of the potential barrier, m is the surface density of electrons

captured at the grain-boundary states, U is a voltage applied across the grain boundary, q is the elementary charge, M is the density of the localized electronic states, N D is the donor

ε is the relative permittivity and ε 0 is the permittivity of free space, υ is the average thermal velocity of electrons, σ is the capture cross-section, k is the 2 Boltzmann constant and T is the absolute temperature; b = q (2εε 0 N D ) . density in the grain,

According to the model suggested in [29], the application of voltage causes the additional trapping of electrons at the grain-boundary states and this leads to the increase of the density −

of adsorbed oxygen in charged form ( N ). Therefore, the surface density of adsorbed oxygen in neutral form N (t ) is decreased:

N (t ) = M − m(t ).

(4)

The density of empty grain-boundary states M − m(t ) is equal to the surface density of adsorbed oxygen in neutral form N (t ) . The decrease of N (t ) due to the capture of electrons (transition of adsorbed oxygen from neutral to charged form) breaks the adsorption equilibrium and causes the additional adsorption of oxygen. This process can be described by the equation:

256

Alexander Bondarchuk

dN dm = α (N SS − N ) − . dt dt

(5)

Here first term gives the rate of additional adsorption caused by the application of voltage, α is the adsorption rate coefficient. First term is proportional to the difference between the steady-state and variable surface densities of adsorbed oxygen in neutral form. The second term gives the rate of N (t ) decrease due to the transition of adsorbed oxygen from neutral to charged form. In a general case the barrier height and the current density are described by equations (1–5). The steady-state surface density of adsorbed oxygen in neutral form N SS can be found from equation (3) at the conditions dm dt = 0 and U = 0 :

N SS

U =0 =

where m0 = (4ϕ (0) b )

12

⎛ ϕ (0) − E S M − m0 = 2m0 N C N D−1 exp⎜ kT ⎝

⎞ ⎟, ⎠

(6)

is the equilibrium ( U = 0 ) surface density of trapped electrons.

The solution of equations (1–5) was obtained by the numerical methods. The next parameters were used: the barrier height ϕ 0 = 0.3 eV at U = 0 (taken from the temperature dependence of electrical conductivity [27]); the electron capture cross-section

σ = 1.0 × 10 −15 cm 2 , the effective density of states in the conduction band Nc = 6.0 × 1018 cm −3 , the average thermal velocity of electrons υ = 1.0 × 107 cm s −1 , the 17

donor density in the grain N D = 2.0 × 10 cm

−3

, the ionization energy of grain-boundary

states E S = 0.45 eV , the relative dielectric permittivity of the grain is assumed to be

ε = 8.5 , the absolute temperature T = 290 K , the adsorption rate coefficient α = 0.004 s −1 (this value was estimated from experimental time dependence of current[29]). In calculations of j (U ) dependence, voltage was increased by the steps

ΔU = 0.004 V with the duration of each step Δt = 3.0 s . In the experiment with ceramic sample (Figure 5) the step voltage is higher1 but ceramic sample contains a great number of grain boundaries connected in series. For convenience, the results of calculations (Figure 6 and Figure 7) are presented in the normalized form. The current density is normalized on the value j0 = qυN D 4 and 6

multiplied by coefficient 10 . The magnitude of the potential barrier and the voltage are normalized on the value of the barrier height at U = 0 ,

1

ϕ 0 / q , the density of empty and

Voltage was increased consecutively by the steps ΔU = 0.2 V with the duration of each step Δt = 3.0 s [29].

Current Limiting in Oxide Ceramic Structures

257

filled grain-boundary states at U ≠ 0 are normalized on the value m0 , the density of filled grain-boundary states at U = 0 . Calculations of j (U ) curves are performed for two cases. In the first case voltage is raised fairly slowly so the adsorption equilibrium ( dN dt = 0 ) and electronic equilibrium ( dm dt = 0 ) at the grain boundary are reached at each discrete value of voltage. To model this situation equations (1–3) at a condition dm dt = 0 were solved jointly. Equation (5) gives simply N = N SS . Obtained results are shown in Figure 6. In the second case voltage is raised relatively quickly (what frequently takes place in an experiment) so the adsorption and electronic equilibrium at the grain boundary rather are not reached. To model this situation equations (1–5) were solved jointly. Obtained results are shown in Figures 7. If voltage is raised fairly slowly (first case), j (U ) dependence is sublinear (Figure 6a, curve 1). In this case the density of empty grain-boundary states is constant (Figure 6b, curve 2) because the surface density of adsorbed oxygen in neutral form N (t ) is kept unchanged (there is enough time for the adsorption of additional amount of oxygen). This ensures fairly strong trapping of electrons (Figure 6b, curve 1). As a result, the decrease of the barrier height is slightly slowed down (Figure 6a, curve 2) and current is increased with voltage weaker than linearly (Figure 6a, curve 1). Similar behavior of j (U ) dependence was found in the experiment if current was measured nearly in the steady-state conditions [25]. If voltage is raised relatively quickly (second case), j (U ) curve is more complicated (Figure 7a, curve 1). At low voltages additional adsorption is negligible due to a weak change of the occupation of the grain-boundary states (Figure 7b, curve 1). However, there are a lot of empty grain-boundary states (Figure 7b, curve 2). The trapping of electrons leads to some not too sharp decrease of the barrier height (Figure 7a, curve 2). Therefore, at low voltages some slight sublinear region at j (U ) curve takes place (Figure 7a, curve 1). Further voltage increase leads to the filling of the grain-boundary states (Figure 7b, curve 1) and, therefore, the intensity of a capture process is decreased. In these conditions, the sublinear region at j (U ) curve is changed to even slightly superlinear one (Figure 7a, curve 1). However, after the filling of a majority of empty states some deficiency of oxygen in neutral form appears and additional adsorption becomes a substantial factor. Increase of oxygen adsorption causes the strong increase of the density of empty grainboundary states (Figure 7b, curve 2). The curve 2 in Figure 7b has minimum and the density of empty grain-boundary states raises at higher voltages. As a consequence, trapping is increased strongly (Figure 7b, curve 1) and the barrier height is passed through a minimum and then it is increased (Figure 7a, curve 2). Accordingly, superlinear region at j (U ) curve is changed to sublinear one and the current density is even decreased with voltage (Figure 7a, curve 1).

258

Alexander Bondarchuk

Figure 6. (a) Normalized dependences of the current density (curve 1) and the barrier height (curve 2), (b) the density of filled (curve 1) and empty (curve 2) grain-boundary states on the voltage. Calculations are performed for slow voltage increase (adsorption equilibrium takes place).

Figure 7. (a) Normalized dependences of the current density (curve 1) and the barrier height (curve 2), (b) the density of filled (curve 1) and empty (curve 2) grain-boundary states on the voltage. Calculations are performed for fairly rapid voltage increase (adsorption equilibrium does not take place).

Current Limiting in Oxide Ceramic Structures

259

In that way, considered model of grain-boundary controlled conduction can be applied to explain observed nonlinear current-voltage characteristics of In2O3-SrO ceramics having superlinear region which is followed by the sublinear region at higher voltages. The key point of the model and its radical difference from the known models is a relation between electronic and adsorption processes: capture of electrons at the application of voltage causes an additional oxygen adsorption and subsequent capture of electrons, respective increase of the barrier height and appearance of the current limiting region at current-voltage curve. Calculations give reasonable approximation to experimentally observed current-voltage dependences. Similar processes leading to the current limiting effect may be possible in other oxide ceramics with barrier-controlled conduction. How it has been reported in [32], in SnO2-based ceramics sintered at low temperature the current limiting effect appears if there is enough oxygen in atmosphere and it is not observed if there is a low oxygen content in atmosphere (Figure 8, curves 1-4). The low-field conductivity in argon is slightly lower than in air (Figure 8). It can be assumed that such behaviour is related to the adsorption of water molecules which are in air. In contrast, in a flux of dry argon the ceramic surface is substantially cleaned by water. The low-field conductivity in SnO2–Co3O4–Nb2O5–Cr2O3 ceramics is highly sensitive to the variation in the relative humidity of air [33] and this behavior is typical for SnO2-based materials [34, 35].

Figure 8. Current versus voltage in the SnO2–Co3O4–Nb2O5–Cr2O3 sample recorded in argon (curves 1 and 2) and in air (curves 3 and 4) at increase (1, 3) and decrease (2, 4) of voltage.

In addition, the current limiting effect in SnO2–Co3O4–Nb2O5–Cr2O3 ceramics can be transformed in oxidizing atmosphere into the varistor effect [32]. The first application of voltage to the fresh sample (which was not subjected to the electric field influence) in air always leads to the current limiting behaviour. Current is saturated and even decreased with the rise in voltage (Figure 8, curve 3). However, the decrease in the voltage gives a varistorlike superlinear curve (Figure 8, curve 4). The rise in voltage in the following measurements again gives a current limiting effect but the I (U ) curve is shifted to lower currents and the mentioned effect becomes less expressed. Further applications of voltage in air (5–10

260

Alexander Bondarchuk

measurements at a fixed voltage polarity) give a practically reproducible superlinear curve shifted slightly lower than I − U curve obtained at first measurement. Thus, the current limiting effect in tin dioxide ceramics is transformed into the varistor effect [32]. According to the barrier model considered above, such behavior of current versus voltage in tin dioxide based ceramics can be explained in following way. In SnO2–Co3O4–Nb2O5– Cr2O3 ceramics the application of the voltage in air can cause additional oxygen adsorption. As a result of it the barrier height grows and the current limiting effect takes place. At the voltage decrease additional oxygen adsorption is negligible, only electronic processes take place and therefore the varistor behaviour is observed.

III. CURRENT LIMITING IN CERAMICS BASED ON BaTO3 INDUCED BY THE CHANGE OF THE CRYSTALLINE STRUCTURE Very stable and highly reproducible current-limiting characteristics have been observed for polycrystalline ceramics prepared from sintering mixtures of coarse-grained, donor-doped BaTiO3 (tetragonal) as the major phase and ultrafine, undoped cubic perovskite such as BaSnO3, BaZrO3, SrTiO3 or BaTiO3(cubic) as the minor phase [36, 37]. The current-voltage curves of such materials have a strong initial maximum and then the current decreases to a nearly steady value as the voltage is increased. At low voltages the current increases linearly (Figure 9a) and the power dissipation through Joule heating raises the body temperature. Close to the Curie temperature ( TC ) the resistance of the material increases sharply (Figure 9b) and the current drops to a near limiting value. The Curie temperature is defined as the point at which an alteration of the crystalline morphology occurs. The crystal structure of BaTiO3 ceramics changes with heating from its initial tetragonal, non-symmetrical configuration below the Curie temperature (~120 ◦C), to a cubic perovskite symmetrical configuration above the Curie temperature. This phase transitions is accompanied by a marked peak in the relative permittivity ( ε ) vs. temperature curve [38-46]. With increasing temperature the permittivity is increased below the Curie point and is decreased above TC . Above the Curie temperature the relative permittivity in BaTiO3 ceramics follows Curie-Weiss law and can be described accurately by a simple relationship [44-46]:

ε=

C T − TC

(1)

where C is the Curie constant and T is the temperature. The anomalous increase of the resistivity in BaTiO3 ceramics is known as the PTCR effect and is used for Positive Temperature Coefficient (PTC) thermistors. The generally accepted model that describes this behavior was proposed by Heywang [40, 41] and later on was extended by Jonker [42]. According to the Heywang model, the PTCR effect in BaTiO3 ceramics originates from the potential barrier that is formed at the grain boundaries increasing

Current Limiting in Oxide Ceramic Structures

261

ρ S in the grain-boundary region above the Curie temperature depends exponentially on the magnitude of grain-boundary barrier ϕ : ⎛ϕ ⎞ ρ S ~ exp⎜ ⎟ , (2) the electrical resistance. The resistivity

⎝ kT ⎠

where

ϕ ~ 1 ε (T ) , ε (T ) is the relative permittivity of the grain-boundary region that

depends on the temperature; k is the Boltzman constant.

Figure 9. (a) Static current-voltage and (b) resistance-temperature characteristics of PTC Thermistor produced by Murata Manufacturing co., Itd.

Above the Curie temperature, the relative permittivity ε decreases with heating (see (1)) and the magnitude of grain-boundary barrier ϕ increases. According to (2), this leads to the sharp increase of resistivity at the temperature T > TC . Thus, the resistance increase of ceramics is due to the change in the dielectric constant in the boundary region between

262

Alexander Bondarchuk

BaTiO3 grains. For the first time this idea was expressed by Heywang [40, 41] and later on verified by other researchers [44, 46-48]. Unfortunately, the examined model can not explain the electrical behavior of BaTiO3 materials below the Curie temperature, in particular the low resistivity of ceramics and the influence of polarization on the resistivity which takes place at the room temperature. The explanation for electrical properties of BaTiO3 ceramics below the TC was proposed by Jonker [42]. His model is based on the ferroelectric behavior of BaTiO3 below the Curie temperature. According to Jonker, the spontaneous polarization in domains near the grain boundary regions neutralizes partly the charge localized at the grain boundaries. This affects a decrease in the potential barrier height allowing for many electrons to transport across the grain boundary. In spite of the convincing explication of the PTCR-effect base in BaTiO3 ceramics, the examined models can not interpret in detail all aspects of this phenomenon. The PTCR phenomenon is a complex interaction of mechanisms involving grain boundary effects, ferroelectric crystal effects, and defect mechanisms. Therefore the creation of a unified theory, which takes into consideration all the interdependent parameters and provides good correlation with experiment, is indeed a difficult task. The investigations in this direction are still continued. In particular, the non-linear current-voltage characteristics of such materials have been observed and modelled by Mallick and Emtage [47], Al-Allak et. al [48], Zhang and Cao [49]. The detailed consideration of these models is beyond the scope of this section and can be found in literature [47-49].

IV. CURRENT SATURATION IN NANO-GRAINED OXIDE CERAMICS As reported by Neto et al. in [51], current limiting effect (current is increased weaker than voltage and even saturated) have been observed in nano-structured ZnO ceramic films prepared by sol-gel processing. Schematic of current-voltage characteristic for such materials is presented in Figure 10. The current vs. voltage curve contains three regions: linear at low voltages, superlinear at intermediate voltages ( β > 1 ), and sublinear at higher voltages ( β < 1 ).

Figure 10. Schematic of current-voltage characteristics (I — Ohmic, II — sublinear and III superlinear region) for ZnO ceramic films obtained in [51].

Current Limiting in Oxide Ceramic Structures

263

As follows from the experimental data reported in [51], the saturation current (the region III in Figure 1) exhibits Arrhenius behavior with an activation energy on the order of a few tenths of an electron volt; the activation energy of conductivity in a strong field ( E2 ) is somewhat lower than that in a weak field ( E1 ); a specific feature of obtained ZnO ceramic films is a small grain size ( l g ~ 4 nm ) . However, unfortunately, an explanation of these facts and of the nature of current limiting behavior was not presented in [51]. The current limiting effect observed in [51] was explained by the simple grain-boundary model in [52]. This model takes into account the fundamental feature of semiconducting oxide ceramics — the presence of grain-boundary potential barriers and the small grain size of ZnO ceramics obtained in [51]. The proposed mechanism for current limiting behavior in such nano-structured oxide ceramics consists in the following [52]. Let us assume that the voltage U applied to two adjacent grains separated by grainboundary barrier (in detail see in section I). Then, the height of the forward-biased barrier ϕ is 2

⎛ qU ⎞ ⎟⎟ , ϕ (U ) = ϕ 0 ⎜⎜1 − ⎝ 4ϕ 0 ⎠ where

(1)

ϕ 0 is the equilibrium magnitude of the grain-boundary barrier at U = 0 ,

ϕ 0 = q 2 N D L20 ( 2εε ) −1 . Here q is the elementary charge, N D is the donor density in the grain, L0 is the equilibrium (at U = 0 ) thickness of the space-charge region on one side of the boundary,

ε is the relative permittivity and ε 0 is the permittivity of free space.

In a diode approximation, the current through the grain boundary is equal to the difference

between

the

forward

−1

( I 0 exp[ −ϕ ( kT ) ] )

and

reverse

−1

( I 0 exp{−[ϕ + qU ]( kT ) } ) currents. For qU >> kT , the current-voltage dependence can be represented by

⎡ ϕ ⎤ I = I 0 exp ⎢ − ⎥, ⎣ kT ⎦ where

(2)

ϕ is given by (1), k is the Boltzmann constant and T is the absolute temperature.

With increasing voltage, the thickness of the space-charge region in the lower potential grain decreases slightly, while that in the higher potential grain increases. At the saturationonset voltage VS , the thickness of the depletion region in the higher potential grain attains the grain size l g . Upon a further increase in voltage, the positive space charge in the higher potential grain cans no longer increase. This implies by virtue of the constant negative charge localized at the grain boundary and the electroneutrality condition that the height of the

264

Alexander Bondarchuk

forward-biased barrier,

ϕ , which determines the current, remains constant as a function of

voltage, and the current saturates at I S . In the current-saturation regime, the thickness of the space-charge region in the higher potential grain is constant, equal to l g and the bottom of the conduction band is lowers in this grain. The electric field in this grain, given by

1 dϕ q dx

, is higher. x =0

Thus, the current saturation results from the constant height of the current-controlling barrier at a sufficiently small grain size, l g > L0 . ~

The saturation-onset voltage VS can be found by solving jointly the equation (1) and the expression

(

)

ϕ + qU = q 2 N D 2εε 0 l g2 which was obtained at the assumption that the

space-charge region is spread with voltage over the whole grain:

US = 4

ϕ ⎛ lg

⎞ ⎜⎜ − 1⎟⎟ . q ⎝ L0 ⎠

(3)

Taking (1) and (3) into account, we obtain from (2)

⎛ E ⎞ I S = I 0 exp⎜ − ⎟, ⎝ kT ⎠

(4)

where the activation energy E at saturation is given by 2

l ⎞ ⎛ E = ϕ 0 ⎜⎜ 2 − g ⎟⎟ . L0 ⎠ ⎝

(5)

Figure 11 shows the logarithm of the saturation current as a function of the square of the mean grain size (experimental data reported by Neto et.al. [51]). The saturation current rises with mean grain size because of the increase in carrier-depletion voltage. As apparent from 2

Figure 11, log I S is proportional to l g , in full accord with relations (4) and (5). The saturation current shows thermally activated behavior (Figure 12), in accordance with (4), because the probability of carriers overcoming the barrier of fixed height rises with temperature. The activation energy derived from the data in Figure 12 is E2 = 0.16 eV . How it was reported in [51], the activation energy of conductivity in strong fields, E2 , is somewhat lower than that in weak fields, E1 . In the model [52], E1 = ϕ 0 + ς , where

ς is

the energy separation between the Fermi level and the bottom of the condition band in the

Current Limiting in Oxide Ceramic Structures

265

quasi-neutral part of the grain, and E2 = E + ς . Therefore, we obtain from (5) E2 < E1 since l g > L0 , in agreement with experiment. ~

Figure 11. Logarithm of the saturation current ( U

= 380 V

) as a function of the square of the mean

ZnO grain size [51].

Figure 12. Arrhenius plot of the saturation current ( U

= 380 V

) for ZnO film [51].

According to [52], one would expect from (3) and (5) that the model is applicable in the range 1 < l g L0 < 2 . Although the upper limit on l g L0 is probably a characteristic of the model, the proposed mechanism can be responsible for the sublinear current-voltage behavior of fine-grained ( l g > L0 ) semiconductor ceramics. In particular, the examined model [52] ~

would possibly explain the current limiting observed in nano-grained SnO2 ceramic films [53]. Unfortunately, the experimental data presented in [53] are not enough to verify this supposition.

266

Alexander Bondarchuk

V. SUBLINEAR CURRENT-VOLTAGE DEPENDENCE IN Pb (ZrXTi1-X)O3 THIN FILMS The electrical properties of lead-zirconate-titanate thin film were the subject of inquiry in many works [54-65]. This interest can be explained by a ferroelectric behavior of such materials which are promising candidates for a variety of electronic applications [65]. In addition, a significant feature of these materials is a sublinear current-voltage dependence at intermediate- and high voltage. Figure 13 presents a schematic drawing of typical I − U curves obtained at room temperature for Pb(ZrxTi1-x)O3 (PZT) thin films. In that current-voltage curve, one can distinguish a region of a steep current increase followed by a sublinear I (U ) dependence at intermediate- and high voltage. This structure of the I − U curve is observed for films prepared in the different ways: sol-gel, metalorganic chemical vapor deposition, and sputtering [65]. The current response in Pb(ZrxTi1-x)O3 thin films includes a time dependent current which is strongly dependent on the measuring technique, prehistory of the sample, and some other factors [64, 65].

Figure 13. Schematic of current-voltage characteristics (I — Ohmic and II — sublinear region) for Pb(ZrxTi1-x)O3 thin films obtained in [63].

The different samples can exhibit different types of conduction behavior depending on their thickness, grain structure and structure of interfaces [65]. In addition, it is known that the high-field conduction is strongly controlled by properties of the electrodes [56, 59]. The shape of I − V curves as well as the effects of temperature cycling, electrical pretreatment, and light pretreatment strongly suggest that electrode-controlled injection occurs in PZT thin films, which is in turn influenced by the electric field of the depletion and entrapped space charge [65]. Such features of a leakage current complicate the comprehension of the conduction mechanism in this system. Different physical models describing the leakage current behavior in PZT thin films were proposed in [55-61]. However, as it was shown in [62, 65], none of these models is able to provide an adequate explanation for the totality of the known experimental data. The space-charge influenced-injection model describing the main features of the observed current-voltage characteristics for lead-zirconate-titanate thin film was developed by

Current Limiting in Oxide Ceramic Structures

267

Stolichnov and Tagantsev. Their model is based on the following experimentally statements [65]: (i) The leakage current is determined by injection of carriers through the electrode- PZT contacts. (ii) The carriers injected are holes. According to the energy band structure of the Pt-PZTPt films, the conditions for the hole injection are much more favorable than for the electron injection because of the high work function of Pt [62]. (iii) Being in contact with Pt electrodes, the film contains a positive space charge due to the positive difference in work function δΦ between Pt and PZT [62], which corresponds to upward band bending (for the side parts with respect to the middle of the films), i.e., the built-in charge in the film is of the same sign as the carriers injected. (iv) The built-in charge is assumed to be related to deep trapping centers and oxygen vacancies, immobile during the measurement time, and homogeneous. According to Simmons [66] for a positive δΦ , this charge can be formed due to a depletion of the electrons from the states lying in an energy stripe δΦ below the Fermi level of the 18

19

−3

bulk material. For a reasonable impurity concentration of 10 − 10 cm , following Ref.63 and Ref.64, the thickness of the charged nearby electrode layer for PZT-Pt contact can be estimated as a few tenths of a micron. Model calculations taking into account oxygen vacancy migration give similar values [67]. From these considerations, it is expected that the built-in space charge is indeed homogeneous in 300 nm films which were researched in [65]. Based on these suppositions Stolichnov and Tagantsev [65] consider the band profile of the Pt-PZT-Pt system with a uniform positive space charge of depletion within the film (Figure 14). The interfacial potential barrier for holes injection Φ h is mainly determined by the difference of work functions of Pt and Pb(ZrxTi1-x)O3 [62]. However, the upward band bending caused by the space charge results in formation of the additional potential barrier ΔΦ [65]. If no external electric field is applied, the maximum of the potential barrier is located in the middle of the film (Figure 14a). Once an electric field is applied, the band profile of the system changes as shown in Figure 14b. As the electric field increases, the potential barrier decreases and its maximum moves towards the cathode [65]. This evolution of the band profile is accompanied by a strong nonlinear increase in current. At some critical field FC , the maximum of the potential barrier reaches the interface (Figure 14c) and the further current increase is only possible due to a barrier lowering according to the Schottky mechanism [65, 68]. Thus, at higher fields, the current will be controlled by the potential barrier Φ h . This barrier is mainly determined by the difference of work functions of materials, therefore the change of its height is negligible with increasing applied electric field. As a result of constant height of the current-controlling barrier at electric field F > FC , the sublinear I (U ) dependence appears.

268

Alexander Bondarchuk

According to [65], the current-temperature characteristic of the system is described by the activation energy of about 0.9 eV in the temperature range 70-200◦C. Therefore, the change of the slope of the current-temperature curve described in literature can be related to the measuring procedure because of an incomplete saturation of current-time response to a measuring voltage step [65]. The time dependence of the measured current response in experiments where measuring times are greater than 100 s and fields higher than

30 − 40 kV cm is attributed to a time drift of parameters of the conductive system of the films caused by the influence of the injected charge entrapment during measurement [65].

Figure 14. Schematic energy band profile of the PZT film [65]: (a) without external electric field, (b) at electric field F < FC applied, and (c) at electric field F > FC applied.

One can conclude that the model proposed by Stolichnov and Tagantsev reproduces fairly well the main qualitative features of current-voltage curves of Pt-PZT-Pt thin films and gives us a reasonable description of the current limiting behavior obtained in this system.

ACKNOWLEDGMENTS The author would like to thank Dr. A. Glot for his work as co-author in the papers which were used in this chapter, Dr. V. Rybak and Dr. Y. Temko for reading some sections and useful comments.

Current Limiting in Oxide Ceramic Structures

269

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

[24] [25] [26]

[27]

[28]

Myasnikov, I. A., Sukharev, V. Ya., Kupriyanov, L. Yu., & Zavjalov, S. A. ( 1991). Semiconductor sensors in physico-chemical studies; Nauka: Moscow, pp 87-88. Beekmans, N. M. J. (1978). Chem. Soc. Faraday Trans., 74, 31-45. McAleer, J. F., Moseley, P. T., Norris, J .O. & Williams, D. E. J. (1987). Chem. Soc. Far. Trans., 83, 1323 - 1331. Pike, G. E. & Seager, C. H. J. (1979). Appl. Phys., 50, 3414-3422. Mahan, G. D., Levinson, L. M. & Philipp, H. R. J. (1979). Appl. Phys., 50, 2799-2812. Vanadamme, L. K. J. & Brugman, J. C. J. (1980). Appl. Phys., 51, 4240-4244. Volkenshtein, F. F. (1987). Electron processes at semiconductor surface at chemisorption; Nauka: Moscow, pp 118-123. Matsuoka, M. & Jap, J. (1971). Appl. Phys., 10, 736-746. Glot, A. B., Chakk, A. M., Chernyi, B. K. & Yakunin, A. Y. A. (1974). Inorganic Mater., 10, 1866-1868. Einzinger, R. (1978). Appl. Surf. Sci., 1, 329-340. Clarke, D. R. J. (1978). Appl. Phys., 49, 2407-2411. Yan, J. F. & Rodes, W. W. (1982). Appl. Phys. Lett., 40, 536-537. Glot A. B., Hogarth C. A., Bulpett R. & Reynolds A. J. J. (1985). Mat. Sci. Lett., 4, 963-966. Makarov, V. O. & Trontelj, M. J. (1994). Mat. Sci., Lett., 13, 937-939. Bueno, P. R., De Cassia-Santos, M. R., Leite, E. R., Longo, E., Bisquert, J., Garcia-Belmonte, G. & Fabregat-Santiago, F. J. (2000). Appl. Phys., 88, 6545-6548. Glot, A. B. (2007). In: Ceramic Materials Research Trends; Ed.: Lin, P. B., Nova Science Publishers, Inc., New York, p. 227-273. Vijayanandhini, K. & Kutty, T. R. N. (2006). Appl. Phys. Lett., 88, 123513-3. Kutty, T. R. N. & Philip, S. (1995). Materials Science and Engineering B, 33, 58-66. Glot, A. B., Sandoval-Garcia, A. P., Gaponov, A. V., Bulpett, R., Jones, B. J. & Jimenez-Santana, G. (2008). Adv. in Tech. of Mat. and Mat. Proc. J., 10, 21-32. David, R. C. J. (1999). Am. Ceram. Soc., 82, 485–502. Mueller, R. K. J. (1961). Appl. Phys., 32, 635-639. Taylor, W. E., Odell, N. H., & Fan, H. Y. (1952). Phys. Rev. B, 88, 867-871. Gaidyalis, V. I., Markevich, N. N. & Montrimas, E. A. (1968). Fizicheskie protsessy v elektrofotograficheskikh sloyakh ZnO (Physical processes in ZnO electrophotographic layers); Vilnius: Mintis, 227-230. Goldman, E. I. & Zhdan, A. G. (1976). Sov. Phys. Semicond, 10, 1839-1845. Glot, A. B., Bondarchuk, A.N. & Mazurik, S.V. (2000). Photoelectronics, 9, 110-113. Glot, A; Behr, G. & Werner, J. (2001). In Proceedings of the 7th Conference of the European Ceramic Society, Brugge, Belgium, Key Eng. Mater., 206–213, Trans. Tech. Publications, Switzerland (2002). 1441-1444. Glot, A., Bondarchuk, A., Mazurik, S., Behr, G. & Werner, J. (2002). In Proceedings of the 7th Conference of the European Ceramic Society, Brugge, Belgium, 2001; Key Eng. Mater., 206–213, Trans. Tech. Publications, Switzerland, 1437- 1440. Bondarchuk, A. N. & Glot, A. B. (2004). Visnyk of Dniepropetrovsk Univ. Ser. Phys. Radiophys, 12, 165-167.

270

Alexander Bondarchuk

[29] Bondarchuk, A., Glot, A., Behr, G. & Werner J. (2007). Eur. Phys. J: Appl. Phys., 39, 211-217. [30] Constantinescu, M., Segal, E. & Vass, M. (1976). Rev. Roumaine Chim., 21, 503-508. [31] Kunishima, Y. N., Miyayama, M., & Yanagida, H. (1996). Jpn J. Appl. Phys., 35, 3478-3482. [32] Bondarchuk, A. N., & Glot, A. B. J. (2008). Phys. D: Appl. Phys., 41,175306 (3pp) doi: 10.1088/0022-3727/41/17/175306 (to be published). [33] Gaponov, A. V; Glot, A. B., Ivon A. I. & Chack A. M., (2007). Jimenez-Santana, G. Mater. Sci., Eng. B, 145, 76-84. [34] Korotcenkov, G., Blinov, I., Brinzari, V. & Stetter, J. R. (2007). Sensors Actuators B, 122, 519-526. [35] Nenov. T. & Yordanov, S. (1992). Sensors Actuators B, 8, 117-122. [36] Kutty, T. R. N. & Ravi, V. (1991). Journal of Materials Science: Materials in Electronics, 2, 79-88. [37] Zhang, D., Zhou, D., Jiang, S., Wang, X. & Gong, S. (2002). Sensors and Actuators A, 101, 123-131. [38] Harwood, M. G., Popper, P. & Rushman, D. F. (1947). Nature, 160, 58-59. [39] Megaw, H. D. (1957). Ferroelectricity in crystals; Methuen & Co Ltd: London, pp 15-20. [40] Heywang, W. (1961). Solid-State Electronics, 3, 51-58. [41] Heywang, W. J. (1964). Am. Ceram. Soc., 47, 484-490. [42] Jonker, G. H. (1964). Solid-State Electronics, 7, 895-903. [43] Kataoka, N., Hayashi, K., Yamamoto, T., Suguwara, Y., Ikuhara, Y. & Sakuma, T. J. (1998). Am. Ceram. Soc., 81, 1961-1963. [44] Lines, M. E. & Glass, A. M. (1977). Principles and applications of ferroelectics and related materials; Clarendon press: Oxford, pp 101-105. [45] Heywang, W. (1984). Amorphe und polykristalline Halbleiter; Springer-Verlag: Berlin, Heidelberg, New York, Tokyo, pp 101-115. [46] Vittayakorn, N. J. (2006). Appl. Sci. Res., 2, 1319-1322. [47] Mallick, G. T., Emtage, Jr., & Emtage, P. R. J. (1968). Appl. Phys., 39, 3088- 3094. [48] Al-Allak, H. M., Illingsworth, J., Brinkman, A. W. & Woods, J. (1989). J. Phys. D, 22, 1393-397. [49] Zhang, F. & Cao. Z. J. (1996). Appl. Phys., 79, 2487-2490. [50] Sinclair, D. C. & West, A. R. I. J. (1989). Appl. Phys., 66, 3850-3856. [51] Neto, J. A. D., Pulcinelli, S. H. & Santilli, C.V. (1996). In Proceedings of the International Conference on Electronic Ceramics and Applications. University of Aveiro: Aveiro, Portugal, 409-412. [52] Glot, A. B. & Bondarchuk, A. N. (1999). Inorg. Mater., 35, 532-534. [53] Leite, E. R., Lee, E. J., Ribeiro, C. & Longo, E. J. (2006). Am. Ceram. Soc., 89, 2016-2020. [54] Chen, J., Chen, H. & Lee, J. Y. (1996). Appl. Phys. Lett., 69, 4011-4015. [55] Moazzami, R., Hu, C. & Shepherd, W. (1990). Proceeding International Reliability Physics Symposium, 231-236. [56] Sudhama, C., Campbell, A. C., Maniar, P. D., Jones, R. E., Moazzami, R., Mogab, C. J. & Lee, J. C. J. (1994). Appl. Phys., 75, 1014-1019. [57] Waser, R. (1997). Integr. Ferroelectr. 15, 39-43.

Current Limiting in Oxide Ceramic Structures

271

[58] Mihara T., Watanabe, H. & Jpn. J. (1995). Appl. Phys., 34, 5664-5673. [59] Chen, X., Kingon, A. I., Al-Shareef, H. & Bellur, K. R. (1994). Ferroelectrics, 151, 133-138. [60] Hu, H. & Krupanidhi, S. B. J. (1994). Mater. Res., 9, 1484-1498. [61] Blom, P. W., Wolf, R. M., Cillessen, J. F. M. & Krijin, M. P. C. M. (1994). Phys. Rev. Lett., 73, 2107-2110. [62] Wouters, D. J., Willems, G. J., & Maes, H. E. (1995). Microelectron. Eng., 29, 249-256. [63] Takasu, H. (1997). Integr. Ferroelectr, 14, 1-10. [64] Tagantsev, A., Kholkin, A., Colla, E., Brooks, K. & Setter, N. (1995). Integr. Ferroelectr, 10, 189-204. [65] Stolichnov, I. & Tagantsev, A. J. (1998). Appl. Phys., 84, 3216-3225. [66] Simmons, J. G. J. (1971). Phys. Chem. Solids, 32, 2581-2591. [67] Brennan, C. J. (1995). Integr. Ferroelectr, 7, 93-109. [68] Sze, S. M. (1969). Physics of Semiconductor Devices; Wiley: New York, pp 301-312.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 273-300

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 8

CHARACTERISATION OF SILICIDE THIN FILMS FOR SEMICONDUCTOR AND NANOTECHNOLOGY ELECTRONICS Madhu Bhaskaran1*, Sharath Sriram1 and David R. G. Mitchell2,3 1

Microelectronics and Materials Technology Centre, School of Electrical and Computer Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia. 2 Institute of Materials Engineering, Australian Nuclear Science and Technology Organisation (ANSTO), PMB 1, Menai, New South Wales 2234, Australia. 3 Present address: Electron Microscopy Unit, University of Sydney, Sydney, New South Wales 2006, Australia.

ABSTRACT Nanotechnology devices require low resistance contacts, which can be fabricated by the incorporation of silicide thin films This chapter discusses in detail the study of silicide thin films using a suite of materials characterisation tools. The silicides of interest in this study were titanium silicide (TiSi2) and nickel silicide (NiSi), given their low resistivity and low barrier heights to both n-type and p-type silicon. The silicide thin films were formed by vacuum annealing metal thin films on silicon substrates. Silicide thin films formed from metal films deposited by DC magnetron sputtering and electron beam evaporation were compared. The composition, crystallographic orientation, and morphology of these thin films were studied using spectroscopy (AES, SIMS, RBS, in situ Raman spectroscopy), diffraction (Bragg-Brentano and glancing angle XRD, RHEED), and microscopy techniques (TEM, SEM, and AFM).

*

Corresponding author: [email protected], [email protected]

274

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

1. INTRODUCTION Advancements in nanotechnology have created the need for efficient means of communication of electrical signals to nanostructures [1]. Electrical contacts made to such nanodevices need to pose minimum possible contact resistance. In order to study and evaluate the resistance of such contacts or the resistance posed by the interface(s) in such contacts, accurate test structures and evaluation techniques need to be used. These will pave the way for the identification of new materials and/or contact architectures to develop nanoscale low resistance contacts. These low resistance contacts incorporate silicide thin films. Titanium silicide and nickel silicide thin films are very popular due to numerous advantages. These thin films are also suitable for applications in the MEMS industry as mask materials for silicon bulk-micromachining [2]. Titanium silicide (TiSi2) is widely used as a local interconnect material in complementary metal-oxide semiconductor (CMOS) technology [3,4]. One of the advantages of titanium silicide is that it can exist in a low resistivity phase, suitable for local interconnects in CMOS technology. It is also suitable for CMOS contacts, with low barrier heights to both n-type and p-type silicon [5]. Titanium silicide is also stable at high temperatures in the range of 8001000 ºC [4,6]. While substantial characterisation and application results for TiSi2 have been previously reported (examples of such work include [3,4,7-10]), the TiSi2 thin films discussed in this chapter were comprehensively analysed by a variety of spectroscopy, diffraction, and microscopy techniques. Many of these results add further to those previously published (such as those in [3,7-10]). Nickel silicide (NiSi) is a material which finds application in both CMOS and microsystems devices. NiSi thin films are used in ohmic contacts in the CMOS industry due to their low resistivity [3,6] and comparable performance for both nMOS and pMOS devices [11,12]. These advantages have made NiSi the CMOS industry standard from 2005. This chapter discusses the synthesis of nickel silicide thin films and details materials characterisation results obtained from spectroscopy, microscopy, and diffraction techniques.

2. CHARACTERISATION OF TITANIUM SILICIDE THIN FILMS This section reports on the comprehensive analysis of C54 titanium silicide thin films by a variety of spectroscopy, diffraction, and microscopy techniques. The formation of titanium silicide thin films using sputtered and evaporated titanium has been compared. The desired phase of titanium disilicide (TiSi2) is C54, as it exhibits low resistivity (14-17 μΩcm) and is more stable. The undesirable, high resistivity (60-90 μΩcm) phase of the same stoichiometry titanium silicide is termed C49. C49 TiSi2 is formed initially, which at higher temperatures (above 450-500 ºC) starts to convert to C54. The conversion of C49 to C54 is usually complete at temperatures of 650 ºC to 750 ºC (depending on the initial thickness of titanium [3,13]). In this research, the focus has been on synthesising and studying C54 titanium disilicide (TiSi2).

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

275

2.1. Silicide Thin Films from Sputtered and Evaporated Titanium The formation of titanium silicide (TiSi2) by annealing titanium thin films (deposited either by sputtering or evaporation) on silicon in vacuum is discussed in this section. The advantages of vacuum annealing (over rapid thermal processing, generally reported [3]) include (i) complete reaction of the metal with silicon, and therefore, no necessity to remove unreacted metal by etching; (ii) more uniform heating due to direct contact with the substrate heater; and (iii) lesser impurities in silicide thin films formed in vacuum. Thin films of titanium (100-110 nm) on silicon which were either sputter-deposited or electron beam evaporated were placed on a substrate heater in a vacuum chamber. Table 1. DC magnetron sputtering conditions for titanium. Target Target diameter DC power Target to substrate distance Process gas Base pressure Sputtering pressure Sputtering duration

Titanium (99.99%) 100 mm 100 W 70 mm Argon (99.999%) 1.0 x 10-5 Torr 1.0 x 10-2 Torr 7 minutes

Titanium thin films were deposited on n-type (100) silicon of resistivity 1-10 Ωcm. The silicon samples were cleaned, prior to metal deposition, by etching in buffered hydrofluoric acid. Thin films of Ti were deposited either by DC magnetron sputtering (using conditions in Table 1) or by electron beam evaporation, from a 99.99 % pure source in vacuum (2.0 x 10-7 Torr). The resulting metal thin films were about 100 nm thick, with a 6 % or 15 % variation in thickness across a three-inch silicon wafer (for evaporation or sputtering, respectively). Thin films of Ti (100-110 nm) on silicon, which were either sputter-deposited or electron beam evaporated, were placed on a substrate heater in a vacuum chamber. The anneal process was carried out at a temperature of 800 ºC for 60 minutes in vacuum (1.0 x 10−5 Torr). The temperature was ramped up at 15 °C/min and ramped down at 10 °C/min. An AES depth profile of a titanium silicide thin film formed by vacuum annealing sputtered titanium is shown in Figure 1. There is complete reaction of titanium and silicon to form a uniform TiSi2 thin film, as is apparent from the 1:2 titanium-silicon ratio. Though there is slight oxygen contamination on the surface, the oxygen signal drops rapidly indicating an oxygen-free silicide-to-silicon interface. The average surface roughness of titanium silicide thin films formed from annealing sputtered titanium was 26 nm, as obtained from AFM scans (an example of which is shown in Figure 2). Evaporated titanium thin films vacuum annealed at 800 ºC for 60 minutes also formed TiSi2, with the composition of the silicide (Figure 3) being 33-34% titanium and 66-67% silicon. When compared to the AES depth profile shown in Figure 2 (TiSi2 formed from sputtered titanium), these films do have a considerable amount of oxygen contamination on the surface. As with the titanium silicide formed from sputtered titanium, the oxygen signal drops to a very low concentration giving an oxygen-free silicide-to-silicon interface.

276

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

Figure 1. AES depth profile of titanium silicide thin film (formed from sputtered titanium).

Figure 2. AFM surface scan of titanium silicide (formed from sputtered titanium) thin film over a 5 μm x 5 μm area.

Figure 3. AES depth profile of titanium silicide thin film (formed from evaporated titanium). (Reprinted with permission from Ref. [14].)

The average surface roughness of these thin films was also studied by AFM, and was found to be lower (Ra = 20 nm). These results are discussed in sub-section 2.2.

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

277

Both sputtered and evaporated titanium formed TiSi2 under the same conditions, but oxygen contamination was higher on the surface of the silicide film obtained by annealing evaporated titanium. The oxygen contamination on the surface of TiSi2 thin films formed from evaporated titanium thin films is due to the snow-ploughing of oxygen from native oxide on the silicon substrate (discussed in detail in sub-section 2.2); this could be removed by etching in concentrated hydrochloric acid (HCl) as it was a form of titanium oxide. The physical bombardment by material being deposited while sputtering will remove the thin native oxide on silicon; and is probably the reason for lower oxygen contamination on the TiSi2 film surface. This oxide formed on the surface of TiSi2 thin films from sputtered titanium was a Ti-Si-O compound, which could not be removed by HCl. Both types of films had oxygen-free silicide-to-silicon interfaces and reasonably rough surfaces, though the average surface roughness of films formed from evaporated titanium was lower than those from sputtered titanium thin films. Due to ease of removal of surface oxide and uniformity in thickness of titanium deposited by electron beam evaporation, these films were chosen over TiSi2 thin films formed from sputtered titanium. Further detailed analysis was carried out on TiSi2 thin films formed by vacuum annealing evaporated titanium. The results of these analyses are discussed in the rest of this section.

2.2. Composition and Surface Morphology Analysis Titanium silicide (TiSi2) thin films were studied by spectroscopy techniques – Auger electron spectroscopy (AES) and secondary ion mass spectrometry (SIMS) – to study the composition, uniformity, and oxygen contamination through the films. Atomic force microscopy (AFM) was used to study the surface morphology, in order to observe the degree of regularity of grains (with respect to structure and grain size) and to estimate the average surface roughness of the films. Experimental conditions are detailed in Ref. [14]. Both depth profiles (AES and SIMS, discussed later) indicated the presence of a thin layer of titanium oxide on the silicide surface. This can be removed by etching in concentrated hydrochloric acid (HCl). Hydrofluoric acid (HF) cannot be used to remove this oxide as HF attacks titanium silicide. Silicon wafers were placed face-down on samples during the anneal process, to minimise oxygen interaction with the samples. The upper silicon wafers (prior to use during the anneal process) were dipped in buffered hydrofluoric acid (BHF) to remove native oxide. These silicon wafers showed evidence of titanium diffusion after the anneal process. This titanium diffusion was verified be performing AES depth profiles at different spots. The silicon wafers which covered patterned titanium, showed titanium diffusion in regions with identical dimensions as the patterned titanium [see Figure 4(a)]. This proved to be a novel method of selectively diffusing titanium into silicon. In addition, for a continuous titanium thin film, when a weight exerting a contact pressure of about 620 N/m2 is placed on the top face-down silicon, the upper and lower silicon samples fuse together, forming an unbreakable bond [Figure 4(b)]. Silicon wafers coated with titanium nitride (TiN) to prevent titanium diffusion were experimented with, and did reduce the extent to which titanium diffused into the upper wafer. However, this did not reduce the amount of oxide on the surface of the TiSi2 formed.

278

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

(a)

(b)

Figure 4. Schematic images depicting (a) titanium diffusion into the upper BHF-dipped silicon wafer placed over a patterned titanium sample and (b) the unbreakable bond formed when the upper BHFdipped silicon wafer is placed over a continuous titanium thin film. (Not to scale.)

TiSi2 formation is not affected by native oxide present on silicon (which occurs at the interface of titanium and silicon). Previous studies clearly show that the diffusing species (Si, from substrate) starts the Ti-Si reaction through the native oxide layer, and the oxygen atoms are slowly driven out towards the surface, resulting in an oxygen-free interface. The silicon diffusion is accompanied by the disintegration of the oxide layer. This indicates that a layer of oxide on the silicide surface will occur in most cases, as has been observed [10,15,16]. The AES depth profile indicates very uniform composition for titanium and silicon, and shows an ideal 1:2 ratio of titanium to silicon (Figure 3). SIMS depth profiling was carried out to study the uniformity in composition of titanium and silicon through the thin film. The result (Figure 5) indicates a very homogenous thin film, and no observable oxygen contamination at the interface.

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

279

Figure 5. SIMS depth profile of titanium silicide thin film. (Reprinted with permission from Ref. [14].)

Figure 6. AFM surface scan over a 5 μm × 5 μm area of titanium silicide thin film. (Reprinted with permission from Ref. [14].)

AFM surface scans of the TiSi2 formed from evaporated titanium show a regular grain structure with an average grain size between 100 and 125 nm (Figure 6), and the thin film exhibits an average surface roughness of 20 nm. The surface also shows the presence of a regular array of well-defined crystallites.

280

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

2.3. Cross-Section Transmission Electron Microscopy Analysis

TiSi2 grain ~1.5 μm in size

Cross-section transmission electron microscopy (XTEM) studies of the thin films was carried out to extract a variety of results, and to verify certain results obtained by other analyses. The specimen preparation details are given in [17]. XTEM analysis of these samples indicates a uniform thin film with crystallites on the surface at regular intervals (Figure 7). These crystallites are 100-150 nm sized on large titanium silicide grains (flat plate-like grains approximately 1.5 μm in diameter). The thickness of the titanium silicide thin film was determined to be 244±10 nm. Jump ratio maps (Figure 8) have shown that the crystallites have the same composition as the thin film. The ratio maps for titanium, silicon, oxygen, and carbon were obtained for the region corresponding to the bright field image shown in Figure 8(a). These maps show the presence of titanium and silicon only in the thin film and the crystallites [see Figure 8(b) and 8(c)]. The titanium map [in Figure 8(b)] shows a high titanium signal at the outer surface of the silicide, which is also rich in oxygen [Figure 8(d)]. This corresponds to the oxide pile-up on the titanium silicide surface due to the ‘snowplough’ effect (as seen in AES and SIMS depth profiles). The bright field image contrast [Figure 8(a)] and the silicon map [Figure 8(c)] show that some re-deposition of silicon during ion milling has occurred. This has oxidized and results in a high silicon and oxygen signal at the outer edge of the epoxy. Figure 8(e) indicates the presence of carbon in the hydrocarbon-based epoxy used for TEM specimen preparation. Energy dispersive X-ray analysis also confirmed that the crystallites had the same composition as the titanium silicide thin film. These results show that the crystallites are of the desired silicide composition, and are unaffected by the oxide pile-up.

Figure 7. Surface crystallites on titanium silicide thin films formed by vacuum annealing observed in an energy-filtered bright field XTEM image. The grain boundaries are apparent due to diffraction contrast. (Reprinted with permission from Ref. [17].)

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

281

Figure 8. (a) Reference bright field XTEM image for jump ratio maps. (b-e) Jump ratio maps for titanium, silicon, oxygen, and carbon, respectively. (Reprinted with permission from Ref. [17].)

Figure 9. High resolution cross-sectional transmission electron microscopy image showing the interface between titanium silicide (left) and silicon (right). The dark (~7 nm wide) band in the TiSi2 at the interface is a diffraction artefact. The image highlights the abrupt interface and the absence of amorphous oxide. Inset: SAED patterns for two phases [-112] TiSi2 and [110] silicon. (Reprinted with permission from Ref. [14].)

High-resolution XTEM images highlight the orthorhombic arrangement in the titanium silicide thin film (Figure 9) and the presence of a well-defined interface between the thermally formed silicide and the silicon substrate. No amorphous oxide layer was present at the interface, supporting the AES and SIMS results.

2.4. Orientation Analysis The titanium silicide thin films formed were also studied by diffraction techniques. XRD analysis was carried out on thin film samples, and only the desired C54 peaks were detected

282

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

[as shown in Figure 10(a)] [18]. The peaks due to the silicide thin films (approximately 250 nm thick) were overshadowed by the peak due to the silicon substrate, under the BraggBrentano XRD conditions. To overcome this, the films were analysed using glancing angle XRD (GA-XRD, at an incidence angle of 5º), and more pronounced peaks at identical 2θ positions were obtained [Figure 10(b)]. The results shown in Figure 10 have been shifted to 2θ positions corresponding to the wavelength of copper kα1 radiation (λ = 0.154056 nm). This was done to enable comparison with the standard diffraction files from the International Centre for Diffraction Data. The background variations in Figure 10(b) are an artefact of the diffractometer (observed in all glancing angle XRD analysis on that system), and are possibly due to reflections from the sample holder.

(a)

(b) Figure 10. Diffractograms obtained for titanium silicide from (a) Bragg-Brentano X-ray diffraction and (b) glancing angle X-ray diffraction. (Reprinted with permission from Ref. [14].)

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

283

Figure 11. Reflection high-energy electron diffraction pattern obtained for titanium silicide with an accelerating voltage of 100 kV. (Reprinted with permission from Ref. [14].)

Surface orientation analysis was performed using RHEED. Prior to analysis, the samples were dipped in concentrated hydrochloric acid, to remove surface oxide. The resulting pattern (as shown in Figure 11) indicated a polycrystalline surface orientation, as expected from XRD. The rings obtained correspond to the XRD results, and to expected values of lattice spacing for C54 titanium silicide. RHEED analyses the top few monolayers of the thin film, while GA-XRD analyses a few hundred nanometres of the film (based on the films X-ray absorption properties). A comparison of these results indicates that the orientations presented in the upper monolayers of the film (from RHEED) correspond to the averaged orientation results (obtained from GA-XRD) for the bulk of the thin film. SAED patterns of the silicide obtained during the XTEM analysis were calibrated using the silicon SAED patterns (zone axis [011]), both of which were obtained under identical conditions. The silicide SAED corresponded to the C54 phase of titanium silicide as shown in Figure 12 [19], and the obtained lattice spacings corresponded to those listed in the standard used for XRD data [18].

Figure 12. Indexed selected-area electron diffraction pattern of a region of the titanium silicide thin film with the zone axis of [-112]. (Reprinted with permission from Ref. [14].)

284

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

Figure 13. Energy filtered hollow cone dark field image confirming that the crystallites have the same orientation as the underlying grain. Inset: SAED patterns along TiSi2 [011] for the thin film and a crystallite. (Reprinted with permission from Ref. [17].)

Dark field imaging with hollow cone illumination was carried out to study the crystallographic orientation of titanium silicide crystallites at a grain boundary in the titanium silicide thin film. This indicates that the crystallites have the same orientation as the underlying titanium silicide grain. This is shown in Figure 13. This was also confirmed by SAED patterns (inset images in Figure 13).

3. CHARACTERISATION OF NICKEL SILICIDE THIN FILMS This section presents results from extensive materials characterisation of nickel silicide thin films by a variety of spectroscopy, diffraction, and microscopy techniques. Nickel silicide thin films formed from sputtered and evaporated nickel have been compared. The section also discusses Raman spectroscopy results which compares nickel silicide thin film formation on (100) and (110) silicon. The focus has been on synthesising low resistivity monosilicide (NiSi) thin films.

3.1. Silicide Thin Films from Sputtered and Evaporated Nickel Nickel silicide growth involves the consumption of silicon, when it is formed by thermally reacting nickel thin films deposited on silicon substrates. Thin films (50 nm) of nickel on (100) n-type silicon substrates, with resistivity between 1 and 10 Ωcm, were subjected to different vacuum annealing conditions in order to investigate the formation of nickel silicide of the desired monosilicide composition (NiSi, 1:1 nickel and silicon).

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

285

The native oxide on silicon substrates was removed with buffered hydrofluoric acid, prior to nickel (Ni) deposition. Ni thin films were either deposited by DC magnetron sputtering or by electron beam evaporation. Sputtering was carried out under the conditions listed in Table 2 and evaporation of 50 nm of Ni was performed from 99.99% pure nickel sources, after pumping down to a base pressure of 2 x 10-7 Torr. Table 2. DC magnetron sputtering conditions for nickel. Target Target diameter DC power Target to substrate distance Process gas Base pressure Sputtering pressure Sputtering duration

Nickel (99.99%) 100 mm 80 W 50 mm Argon (99.999%) 1.0 x 10-5 Torr 1.0 x 10-2 Torr 1 minute

Nickel thin films coated on (100) n-type silicon samples were subjected to a one-step or two-step contact anneal process. The following temperatures were among those used to anneal both evaporated and sputtered nickel films (50 nm) on silicon: (i) 200 ºC for 1 hour, followed by 350 ºC for 3 hours (ii) 250 ºC for 1 hour, followed by 350 ºC for 3 hours (iii) 250 ºC for 1 hour, followed by 400 ºC for 3 hours (iv) 350 ºC for 1 hour (v) 350 ºC for 30 minutes These temperatures were chosen because the formation of nickel silicide (NiSi) from silicon takes place through a series of stoichiometric transformations [20]. Nickel thin films on silicon react to form Ni2Si at about 250 ºC, NiSi at 350 ºC, and NiSi2 above 650-700 ºC. The samples were placed on a substrate heater in a vacuum chamber. The anneal process was started under vacuum of 1.0 x 10-5 Torr. The composition of thin films formed as a result of reacting sputtered nickel and evaporated nickel on silicon was compared using AES depth profiles. Depth profiles show that the sputtered nickel films annealed under the above conditions did not react completely with silicon, giving predominantly Ni2Si films (Figure 14). The AES depth profile of sputtered nickel films vacuum annealed at 600 ºC for 90 minutes (Figure 15) shows more complete reaction with silicon, giving NiSi with a composition of 48 % nickel to 52 % silicon. AES depth profiles of other sputtered films annealed under conditions (i) and (iii) were found to be similar to Figure 14. Evaporated nickel films reacted using any of the conditions (i)-(iii) formed NiSi films, with the composition of the silicide (Figure 16) being 49-50% nickel and 51-50% silicon. Evaporated nickel films were also annealed in vacuum at 350 ºC for different time periods of 30 minutes and 1 hour. Depth profiles of these films (formed using 50 nm evaporated nickel) showed them to be NiSi (with nickel and silicon ratios of almost 50:50). Figure 16 depicts the

286

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

depth profile obtained for a NiSi film which was formed after 30 minutes of vacuum annealing at 350 ºC.

Figure 14. AES depth profile of sputtered nickel thin film vacuum annealed at 250 ºC for 1 hour, followed by 350 ºC for 3 hours.

Figure 15. AES depth profile for sputtered nickel thin film vacuum annealed at 600 ºC for 90 minutes.

Figure 16. AES depth profile for evaporated nickel thin film vacuum annealed at 350 ºC for 30 minutes.

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

287

Compared to evaporated nickel films, it was found that more thermal effort was required to form NiSi using sputtered nickel films. Only sputtered nickel films which were annealed at a high temperature of 600 ºC appeared to have completely reacted. This could be attributed to variations in the diffusivity of metals in silicon, based on the metal thin film deposition technique [21]. Further materials characterisation using a suite of tools were carried out for NiSi thin films formed by vacuum annealing evaporated nickel thin films at 350 ºC for 30 minutes.

3.2. Composition and Surface Morphology Analysis The composition of nickel silicide thin films has been analysed using AES depth profiles, SIMS, and RBS. AFM has been used to study the surface morphology of the nickel silicide thin films. Details on experimental conditions are given in [22]. The atomic percentage versus nickel silicide depth was obtained using a combination of sputter etch and AES. The AES depth profile (Figure 16) shows a uniform NiSi film, with a composition of 49-50 % nickel and 51-50 % silicon. No oxygen was observed, either on the surface or at the silicide-silicon interface. SIMS depth profiles were used to study the composition, uniformity, and potential oxygen contamination of the nickel silicide thin films. The profile (Figure 17) indicates a homogenous thin film with no trace of oxygen contamination on its surface. The presence of an abrupt interface is indicated by the sharp drop in the nickel signal on approaching the silicide-silicon interface (also verified using XTEM).

3 Figure 17. SIMS depth profile of a nickel silicide thin film. (Reprinted with permission from Ref. [22].)

288

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

RBS results indicate a nickel silicide layer of thickness 114 nm, with the simulated spectrum in close agreement with obtained data (Figure 18). The simulated spectrum was obtained by considering a multi-layered structure of nickel silicide thin films with nickel and silicon composition varying between 48-52 % for both elements. The two significant layers used in the simulation were a 40 nm layer of 51.0 % nickel and 49.0 % silicon and a 74 nm layer of 48.5 % nickel and 51.5 % silicon. The 114 nm thick layer with a stoichiometry close to NiSi agrees well with profilometry results.

Figure 18. Spectrum from Rutherford back-scattering for a nickel silicide thin film. (Reprinted with permission from Ref. [22].)

Figure 19. AFM 1 μm × 1 μm scan in tapping mode of a nickel silicide thin film. (Reprinted with permission from Ref. [22].)

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

289

AFM has been used to study the surface morphology of the nickel silicide thin films. The average grain size and average surface roughness of these films was found to be 35 nm and 0.67 nm, respectively (Figure 19). The films were found to be very smooth with the roughness values in agreement with those reported in literature [23]. The film surface was studied using a FEG-SEM, but the very smooth film surface (which is as smooth as silicon) prevented good resolution and contrast in the images of the film surface.

3.3. Transmission Electron Microscopy Analysis This section discusses the transmission electron microscopy (TEM)-based investigation of nickel silicide thin films: energy filtered imaging to study the thin film cross-sections, jump ratio maps to confirm the composition uniformity of the thin films, high resolution imaging (HRTEM) to characterise the silicide-silicon interface, and selected-area electron diffraction (SAED) to determine the orientation of the thin films. Cross-sectional transmission electron microscopy analysis was carried out along the length of the films, to study the uniformity in thickness and grain structure of the nickel silicide thin films. The film thickness was estimated to be about 110 nm (agreeing with the profilometry results) and remained reasonably uniform with an equiaxed grain structure evident along the length of the film. It was significant that a repetitive and uniform crystalline structure, with well defined equiaxed grains about 50-60 nm in size, had resulted from the reaction of the nickel film deposited on silicon. The elastic (zero loss) image in Figure 20(a) shows that the grains had an equiaxed grain structure, with grain diameters of about 60nm. Moiré fringes were present at oblique grain boundaries in the film [as indicated in Figure 20(b)]. The interface between the NiSi and Si was not flat but showed some evidence of weak faceting beneath individual NiSi grains. This extended in the plane perpendicular to the image in Figure 20(a, b), giving rise to an oblique interface. This resulted in the interface appearing slightly blurred. This effect was more pronounced in the thicker regions of the TEM foil, as there was a greater projected thickness and consequently a greater pile-up of features in projection. In the very thinnest regions of the foil (those used for HRTEM – discussed later), the interface appeared much sharper. Plan view images highlighted the existence of small (~60 nm) crystalline grains [Figure 20(c)], comparable to the equiaxed grains seen in cross-section. The compositional uniformity was studied using jump ratio maps. The intensity in these images is a function of the areal concentration of the element of interest. The ratio maps for nickel and silicon were obtained for the region corresponding to the bright field image shown in Figure 20(a). These maps (Figure 21) show no significant variation in the Ni and Si compositions. Some modulation due to diffraction contrast effects was present, but these were readily identified as such by their sensitivity to specimen tilt. No unreacted Ni was found. The bright line on the outer surface is an EFTEM artefact induced by specimen drift during the long acquisition and ratio map processing, since no equivalent features were present in the bright field image [Figure 20(a)]. High resolution transmission electron microscopy (HRTEM) imaging indicates the presence of a very abrupt nickel silicide to silicon interface (Figure 22). The silicon lattice appears to abruptly transform into the nickel silicide lattice in the span of one atomic layer. On studying this interface at different points along the nickel silicide film, the faceting

290

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

introduced by the silicide grains appears to have introduced a curvature to the silicide-silicon interface.

Figure 20. Results from TEM analysis of NiSi thin films: (a) XTEM highlighting equiaxed grains in the NiSi thin film in which Moiré (interference) fringes due to orientation differences between grains can be observed; (b) Notable features in the as-obtained image (a) are indicated; (c) Plan view, elastic hollow cone dark field image of the film, highlighting individual grains with dimensions of 60-200 nm; and (d) Plan view TEM image showing polygonal NiSi grains. (Reprinted with permission from Ref. [24].)

Figure 21. Jump ratio maps corresponding to content of (a) nickel and (b) silicon. (Reprinted with permission from Ref. [24].)

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

291

Figure 22.HRTEM of the nickel silicide to silicon interface depicting the abrupt silicide-silicon interface. (Reprinted with permission from Ref. [24].)

3.4. Orientation Analysis The orientation of the thin films was studied using a combination of Kikuchi patterns (or EBSPs) obtained by electron back scatter diffraction (EBSD) and glancing angle X-ray diffraction (GA-XRD). The Kikuchi patterns, also called electron backscatter patterns (EBSPs), have been used to verify the existence of preferential orientations across the thin film surface. (TiSi2 thin films had significant surface roughness, making them unsuitable for this analysis.) Table 3. Nickel silicide unit cell parameters for EBSD. Composition Space group Laue group Unit cell lengths Unit cell angles Unit cell volume

Silicon 50%, Nickel 50% 62, Pnma 3, mmm 5.18 Å, 3.34 Å, 5.62 Å 90º, 90º, 90º 97.23 Å3

The nickel silicide crystal phase used for the indexing the EBSPs is shown in Table 3. This crystal phase was obtained from [25] and also corresponds to the International Committee for Diffraction Data (ICDD) nickel silicide powder diffraction file 85-0901 [26]. The multi-part figure (Figure 23) shows a sample set of the simulations generated and results obtained. A three-dimensional simulation of the EBSPs is shown in Figure 23(a) overlaid on a 2 × 2 × 2 nickel silicide unit cell array. An example of an experimentally obtained EBSP is shown in Figure 23(b). The ultra-fine grain size of the nickel silicide thin films prevents better contrast in these patterns and also limits the number of locations on the film surface which

292

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

produce usable patterns. Due to this, indexing of selected patterns was performed manually to compare the results with the simulated pattern. Based on simulations, recorded patterns were indexed, as shown by the combination of Figs. 23(b, c). An additional example of a recorded EBSP is shown in Figure 23(d). EBSD analysis was used to investigate the uniformity in crystallographic orientation across the thin film. The orientations resulting from the indexing of EBSPs compared very well with the results obtained from GA-XRD, and these indicate the existence of preferential orientations. There is no available literature showing that EBSD has been used to analyse nickel silicide thin films, and the results obtained indicate that further work is required in refining the EBSP acquisition parameters to account for the ultra-fine grain size of the nickel silicide thin films before detailed orientation maps can be generated.

Figure 23. (a) Simulated three-dimensional EBSD pattern corresponding to the crystal phase defined in Table 1. Inset is a 2 × 2 × 2 array of the nickel silicide unit cell. (b, c) Example of a recorded EBSP and the simulation used for indexing this EBSP. (d) Example of an EBSP showing a different orientation to the EBSP in (b). The low contrast in the EBSPs is due to the ultra-fine grain structure of the nickel silicide thin films. (Reprinted with permission from Ref. [22].)

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

293

Figure 24. Glancing angle X-ray diffraction results obtained for a nickel silicide thin film sample indicate preferential orientation. (Reprinted with permission from Ref. [24].)

GA-XRD (Figure 24) of the nickel silicide thin films indicates the films to be polycrystalline with preferential orientation. The peaks obtained in the diffractograms were compared with the International Committee for Diffraction Data powder diffraction file (ICDD PDF) 85-0901 (Cu kα) for indexing [26]. These results also indicate that the films are of the NiSi (monosilicide) phase, and not undesirable Ni2Si or NiSi2 phases.

3.5. In Situ XRD and Raman Spectroscopy Analysis The in situ techniques employed for this study include micro-Raman spectroscopy and Xray diffraction (XRD); in both cases the variations for temperatures up to 350 ºC has been studied. The evolution of NiSi from Ni2Si is discussed and the influence of substrate orientation is investigated. These techniques have also been used to study variations in nickel silicide formation between n-type and p-type silicon. Experimental details can be found in [27].

3.5.1. Raman analysis of reference NiSi sample A Raman spectrum obtained for the reference nickel silicide thin film (formed by vacuum annealing at 350 ºC) is shown in Figure 25. The NiSi thin film used as a reference was extensively characterised to verify it was of the required low resistivity monosilicide phase, and so, the corresponding Raman spectrum was used to identify transformation of Ni to NiSi in the nickel films under in situ study. NiSi belongs to the MnP-type orthorhombic structure (space group Pnma, D2h16); therefore, the film is Raman active. In accordance with group theory (see, for example [28] and [29]), 12 Raman active optical phonons exist for NiSi. Each of these 12 optical phonons can be observed in the Raman spectra of NiSi single crystal collected in a specific measurement geometry. Raman spectra of NiSi films, NiSi powder, and polycrystalline NiSi will contain only some of these modes. Depending on the degree of texture, some Raman

294

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

active modes may be more pronounced than others in thin film spectra. Donthu et al. [29] identifies eight phonon peaks for nickel silicide powder at 197, 214, 255, 288, 314, 332, 360, and 397 cm-1. Peaks at 197, 216, 256, 289, 314, 332, and 362 cm-1 were observed in the Raman spectra for the nickel silicide thin film shown in Figure 25. According to factor group analysis, the peaks 197, 216, 332, and 362 cm-1 can be assigned to the Ag symmetry, while peaks at 256 and 314 cm-1 may belong to either B2g or B3g symmetry [29]. The five other phonon peaks, expected from group theory and, in particular, the peak at 397 cm-1 [29], were not observed in this work.

Figure 25. Micro-Raman spectrum of nickel silicide film formed at 350 ºC by vacuum annealing (reference sample). (Reprinted with permission from Ref. [27].)

3.5.2. In situ measurement results for nickel on n-type (100) silicon Raman spectra obtained while heating nickel thin films on n-type (100) silicon are shown in Figure 26(a). The absence of peaks for the as-deposited Ni film (see spectrum at 24 ºC, for example) is typical for metallic films that have no optic lattice vibrations. Ni2Si peaks are expected at 100 and 140 cm-1, but due to the presence of strong background in this region of the spectra and relatively large noise, the appearance and disappearance of the Ni2Si phase cannot be conclusively determined. However, these two peaks are typically accompanied by a small peak at ~217 cm-1 and a very weak peak at ~ 190 cm-1 [30,31] related to the existence of the NiSi phase. Therefore, in this work, the presence of the Ni2Si phase has been determined based on the appearance of a small single peak at ~215 cm-1, while confirmation on the formation of NiSi phase was made from the appearance of a double peak at ~196 and 215 cm-1. Based on this it seems that the Ni2Si and NiSi phases start to form at 250 ºC and 290 ºC, respectively. The diffractograms in Figure 26(b) show the XRD peaks obtained while the sample was heated to 350 ºC. Nickel and silicide peaks are concentrated in the 2θ range of 40-50º; the peaks at 44.54º and 45.88º are those of nickel (from the sample) and platinum (from the heating stage), respectively [32,33]. Due to thermal expansion, peaks shift left; the shifts are more prominent in nickel than in platinum. This could be due to platinum and nickel being located at different heights with respect to the detector and the different thermal expansion

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

295

coefficients of the two metals. When the sample reaches a temperature of 250 ºC, there is a drop in intensity of the nickel peak and a small peak appears at 47.09º which corresponds to Ni2Si [34]. At 325 ºC, the nickel silicide (NiSi) peak at 45.21º becomes prominent. The diffractogram labelled ‘350 ºC (30 minute dwell)’ in Figure 26(b) corresponds to the XRD data collected after the sample was held at 350 ºC for 30 minutes and this indicates a nickel silicide thin film with a preferential (211) orientation.

(a)

(b) Figure 26. (a) Micro-Raman spectra and (b) X-ray diffractograms registered at different temperatures for nickel thin film deposited on n-type (100) silicon substrate. (Reprinted with permission from Ref. [27].)

296

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

(a)

(b) Figure 27. (a) Micro-Raman spectra and (b) X-ray diffractograms registered at different temperatures for nickel thin film deposited on p-type (100) silicon substrate. (Reprinted with permission from Ref. [27].)

3.5.3. In situ measurement results for nickel on p-type (100) silicon Raman measurements were carried out for these samples under conditions similar to those for n-type (100) silicon substrates. The spectra obtained at various temperatures are shown in Figure 27(a). The weak single peak at ~215 cm-1 (Ni2Si) again appears at 250 ºC and double peaks at 215.6 and 196.5 cm-1 (NiSi) start to appear at 290 ºC. The diffractograms

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

297

in Figure 27(b) show the XRD peaks obtained while the sample was heated to 350 ºC. The peaks at 44.54º and 45.88º are those of nickel (from the sample) and platinum (from the heating stage), respectively [32,33]. The drop in intensity of the nickel peak occurs at 325 ºC. After dwelling at 350 ºC for 30 minutes, the nickel silicide (NiSi) peak at 45.21º corresponding to (211) orientation becomes prominent. There is a small peak at 46.75º which corresponds to nickel silicide (121), and appears only during the initial 350 ºC measurements. The peak for Ni2Si at 47.09º is not prominent during this measurement cycle.

3.5.4. In situ measurement results for nickel on p-type (110) silicon Raman spectra obtained while heating nickel thin films on p-type (110) silicon are shown in Figure 28. The formation of Ni2Si and the transformation from Ni2Si to NiSi seem to occur at slightly higher temperatures on these samples when compared to the nickel deposited on (100) silicon substrates. The spectrum at 350 ºC indicates a polycrystalline film with strong NiSi peaks at 215.6 and 196.5 cm-1 and weak NiSi peaks at 255, 289, 310, 332 and 362 cm-1. These results were conclusive, overcoming necessity of performing in situ XRD measurements.

Figure 28. Micro-Raman spectra registered at different temperatures for nickel thin film deposited on ptype (110) silicon substrate. (Reprinted with permission from Ref. [27].)

298

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

4. CONCLUSION This chapter presents results from a comprehensive characterization of titanium silicide (TiSi2) and nickel silicide (NiSi) thin films formed from thin films of metal deposited on silicon. TiSi2 thin films formed from evaporated and sputtered titanium were compared, and thin films formed from evaporated titanium were subject to further analyses. The titanium silicide films formed by annealing electron beam evaporated titanium films on silicon have a thin layer of oxide on the surface, but an oxygen-free TiSi2–Si interface. The stoichiometry, uniformity in composition with depth and the surface morphology of these thin films were studied using AES and SIMS depth profiles and AFM surface scans. All these results point to a chemically uniform film of the desired composition and the presence of crystallites on the surface at regular intervals. These crystallites were shown to have the same composition and orientation as the underlying thin film. XTEM analysis of the thin films was used to verify the presence of a layer of oxide on the surface, and high resolution XTEM images have shown a well-defined silicide–silicon interface. SAED employed during the XTEM analysis confirmed the presence of C54 titanium silicide. The XRD results showed the existence of a polycrystalline film, with all orientations present corresponding to the desired C54 phase of TiSi2. Glancing angle XRD and RHEED were also used to analyse the thin films, in order to increase the intensity of reflections from the thin film (and decrease substrate effects). NiSi films formed by vacuum annealing of sputtered and evaporated nickel films have been compared using AES depth profiles. Nickel silicide of NiSi composition was formed using evaporated nickel and several annealing temperatures and time periods. Higher thermal energy was required to form NiSi using sputtered nickel films, which did not react uniformly with silicon at lower temperature conditions. Nickel silicide thin films formed by vacuum annealing of nickel thin films on silicon substrates have been studied by spectroscopy, microscopy, and diffraction techniques. Spectroscopy results from AES, SIMS, and RBS indicate (i) the presence of the desired 1:1 nickel and silicon stoichiometry, (ii) uniform composition with depth, and (iii) lack of oxygen contamination either on the film surface or at the silicide-silicon interface. AFM scan results and TEM analyses have shown that the films have an extremely smooth surface (as smooth as silicon with an average surface roughness of 0.67 nm) comprising of ultra-fine grains approximately 30-50 nm in diameter. HRTEM showed that the nickel silicide to silicon interface was atomically abrupt. GA-XRD confirmed that the phase formed was NiSi; with the thin films preferentially oriented with (211) NiSi planes parallel to the (100) planes of the silicon substrate. EBSD analysis has shown that the films are preferentially oriented, confirming results from XRD analysis. The evolution of nickel silicide (NiSi) from the nickel film was observed using both Raman and XRD in situ studies. Raman studies of nickel silicide formation on both n-type and p-type (100) silicon substrates suggest that the transformation temperatures for both are similar (~ 290 ºC); though XRD studies indicate the transformation temperature to be ~ 325 ºC for p-type (100) silicon. This discrepancy could be due to the fact that the laser used for Raman measurements analyses a much smaller area (few μm2) compared to XRD (few mm2). Raman spectra also indicate that the transformation from Ni2Si to NiSi occurs approximately at 300 ºC for (110) silicon as opposed to 290 ºC for (100) silicon.

Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology...

299

REFERENCES [1] [2] [3] [4] [5] [6] [7]

[8]

[9] [10] [11]

[12] [13] [14] [15] [16] [17] [18]

[19] [20] [21] [22]

Roukes, M. L. (2001). Sci. Amer., 285, 48-57. Bhaskaran, M., Sriram, S. & Sim, L. W. (2008). J. Micromech. Microeng, 18, 095002. Zhang, S. & Ostling, M. (2003). Crit Rev Solid State Mater Sci., 28, 1-129. Murarka, S. P. (1995). Intermetallics, 3, 173-186. Plummer, J. D., Deal, M. D. & Griffin, P. B. (2000). Silicon VLSI Technology: Fundamentals, Practice and Modeling; Prentice-Hall: NJ, p 700. Chen, J., Colinge, J. P., Flandre, D., Gillon, R., Raskin, J. P. & Vanhoenacker, D. ( 1997). J. Electrochem. Soc., 144, 2437-2441. Da Silva, A. N. R., Furlan, R. & Santiago-Aviles, J. J. (1995). In Silicide Thin Films – Fabrication, Properties, and Applications; Tung, R. T., Maex, K., Pellegrini, P. W., Allen, L. H., Eds.; Materials Research Society Symposium Proceedings; Materials Research Society: Pittsburgh, PA, Vol. 402, pp 119-124. Maury, D., Regolini, J. L. & Gayet, P. (1995). In Silicide Thin Films – Fabrication, Properties, and Applications; Tung, R. T., Maex, K., Pellegrini, P. W., Allen, L. H., Eds.; Materials Research Society Symposium Proceedings; Materials Research Society: Pittsburgh, PA, Vol. 402, pp 283-294. Kim, K. H., Lee, J. J., Seo, D. J., Choi, C. K., Hong, S. R., Koh, J. D., Kim, S. C., Lee, J. Y. & Nicolet, M. A. (1992). J. Appl. Phys., 71, 3812-3815. Bender, H., Chen, W. D., Portillo, J., van den Hove, L. & Van der Vorst, W. (1989). Appl. Surf. Sci., 38, 37-47. Morimoto, T., Ohguro, T., Momose, H. S., Iinuma, T., Kunishima, I., Suguro, K., Katakabe, I., Nakajima, H., Tsuchiaki, M., Ono, M., Katsumata, Y. & Iwai, H. ( 1995). IEEE Trans. Electron Devices, 42, 915-922. Iwai, H., Ohguro, T. & Ohmi, S. I. (2002). Microelectron. Eng., 60, 157-169. Ezoe, K., Kuriyama, H., Yamamoto, T., Ohara, S. & Matsumoto, S. (1998). Appl. Surf. Sci., 130-132, 13-17. Bhaskaran, M., Sriram, S., Short, K. T., Mitchell, D. R. G., Holland, A. S. & Reeves, G. K. (2007). J. Phys. D: Appl. Phys., 40, 5213-5219. Dexin, C. X., Harrison, H. B. & Reeves, G. K. (1988). J. Appl. Phys., 63, 2171-2173. Wee, A. T. S., Huan, A. C. H., Osipowicz, T., Lee, K. K., Thian, W. H., Tan, K. L., & Hogan, R. (1996). Thin Solid Films, 283, 130-134. Bhaskaran, M., Sriram, S., Mitchell, D. R. G., & Holland, A. S. (2008). Semicond. Sci. Technol., 23, 035021. Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 35-0785. Andrews, K. W., Dyson, D. J. & Keown, S. R. (1971). Interpretation of Electron Diffraction Patterns; Ed. 2; Plenum: NY, p 152. Clevenger, L. A. & Mann, R. W. (1995). In Properties of Metal Silicides; Maex, K., van Rossum, M., Eds., INSPEC: London, p 64. Thomä, A. (1990). J. Phys.: Condens. Matter, 2, 3167-3175. Bhaskaran, M., Sriram, S., Holland, A. S. & Evans, P. J. (2008). Micron, 40, 99-103.

300

Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell

[23] Tsuchiya, Y., Tobioka, A., Nakatsuka, O., Ikeda, H., Sakai, A., Zaima, S. & Yasuda, Y. (2002). Jpn. J. Appl. Phys., 41, 2450-2454. [24] Bhaskaran, M., Sriram, S., Mitchell, D. R. G., Short, K. T., Holland, A. S. & Mitchell, A. (2008). Micron, 40, 11-14. [25] Villars, P. & Calvert, L. D. (1991). Pearson's Handbook of Crystallographic Data for Intermetallic Phases; Ed. 2; ASM: Ohio, p 2887. [26] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 85-0901. [27] Bhaskaran, M., Sriram, S., Perova, T. S., Ermakov, V., Thorogood, G. J., Short, K. T., & Holland, A. S. (2008). Micron, 40, 89-93. [28] Hayes, W. & Loudon, R. (1978). Scattering of light by Crystals; Wiley: NY. [29] Donthu, S. K., Chi, D. Z., Tripathy, S., Wong, A. S. W. & Chua, S. J. (2004). Appl. Phys. A: Mater. Sci. Process., 79, 637-642. [30] Lee, P. S., Mangelinck, D., Pey, K. L., Shen, Z. X., Ding, J. & Osipowicz, T. (2000). See, A. Electrochem. Solid State Lett., 3, 153. [31] Nemanich, R. J., Tsai, C. C., Stafford, B. L., Abelson, J. R. & Sigmon, T. W. (1984). In Thin Films and Interfaces II; Baglin, J. E. E., Campbell, D. R., Chu, W. K., Eds.; Materials Research Society Symposium Proceedings; Elsevier Science Publishing: New York, Vol. 25, p 9. [32] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 87-0712. [33] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 04-0802. [34] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 80-2283.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 301-318

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 9

MAGNETICALLY MODIFIED BIOLOGICAL MATERIALS AS PERSPECTIVE ADSORBENTS FOR LARGE-SCALE MAGNETIC SEPARATION PROCESSES Ewa Mosiniewicz-Szablewska1, Mirka Safarikova2, and Ivo Safarik2,3 1

Institute of Physics, Polish Academy of Sciences, Al. Lotników 32/46, 02-668 Warsaw, Poland, 2 Department of Biomagnetic Techniques, Institute of Systems Biology and Ecology, Academy of Sciences, Na Sadkach 7, 370 05 Ceske Budejovice, Czech Republic 3 Department of Medical Biology, Faculty of Science, University of South Bohemia, Branisovska 31, 370 05 Ceske Budejovice, Czech Republic

ABSTRACT Novel magnetically modified biological materials, containing magnetic iron oxides nanoparticles as labels, have been successfully developed and applied as magnetic affinity adsorbents for the magnetic separation of various biologically active compounds and xenobiotics. The main attention was focused on cheap and easy to get magnetic adsorbents which could be applied for large-scale processes. Among them magnetically modified plantbased materials (sawdust) and microbial cells (yeast and algae) were taken into consideration. An inexpensive, extremely simple procedure was proposed for the preparation of such magnetic adsorbents using standard water-based ferrofluids containing maghemite nanoparticles with the diameter of about 12 nm. Such ferrofluids can be prepared in a simple way (almost in any lab) and such nanoparticles can be used to prepare biocomposite materials enabling their simple magnetic separation with standard permanent magnets. Both of these properties are important for possible large-scale applications. The structural, adsorption and magnetic properties of the developed materials were studied in detail by means of scanning electron microscopy, transmission electron microscopy, spectrophotometric measurements, ESR spectroscopy and conventional magnetic methods (DC magnetization and AC susceptibility measurements). The

302

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik prepared materials efficiently adsorbed selected biologically active compounds and xenobiotics (mainly different enzymes, water-soluble organic dyes and heavy metal ions). Their magnetic behavior was dominated by the superparamagnetic relaxation of isolated single domain maghemite nanoparticles, although a little amount of agglomerates was also present. However, these agglomerates were sufficiently small to show at static conditions the superparamagnetic behavior at room temperature which allows to use the developed materials as magnetic adsorbents in the magnetic separation techniques. Moreover, the prepared materials exhibit the peculiar features enabling their rapid and efficient removal not only from solutions, but also from suspensions. Such materials could be efficiently used to isolate rare biologically active compounds from difficult-tohandle materials including raw extracts, blood and other body fluids, cultivation media, environmental samples, etc. Inexpensive raw materials, extremely simple preparation method, affinity to various biologically active compounds and both organic and inorganic xenobiotics, and distinctive magnetic properties make the developed materials greatly suitable as magnetic adsorbents for large-scale magnetic separation processes.

1. INTRODUCTION Magnetically modified biocompatible materials, containing magnetic nanoparticles as labels, have attracted much attention because of their great potential as magnetic affinity adsorbents for various biologically active compounds. They have been successfully applied for the magnetic separation of various proteins (enzymes, antibodies, antigens, receptors, lectins), nucleic acids (DNA, RNA, oligonucleotides), low-molecular weight biologically active compounds (drugs) and xenobiotics (carcinogens, water soluble dyes, heavy metal ions, radionuclides) [1,2]. Magnetic separation techniques have many interesting applications in different areas of biosciences ranging from cells separation [3] to removal of xenobiotics from aqueous wastes [4]. They are alternative methods to gravitational, centrifugal or filtration separation techniques and enable a simple magnetic manipulation with the adsorbents using an external magnetic field. Magnetic affinity adsorbents can be efficiently used for work in difficult-to-handle materials including raw extracts, blood and other body fluids, cultivation media, environmental samples, etc. However, for large-scale applications (e.g., in biotechnology or environmental technology), the finding of relatively cheap and readily available magnetic adsorbents is necessary. That is the reason why an extremely simple procedure was here proposed for the preparation of such magnetic adsorbents using standard water-based ferrofluids (which can be prepared in any lab) and inexpensive raw materials (sawdust and microbial cells). The sawdust and microbial cells have been chosen for magnetic modification because of their well known affinity for various xenobiotics (mainly water-soluble organic dyes and heavy metal ions [5-8]) which allows to use these materials for the large-scale removal of xenobiotics from polluted water sources. A large amount of environmental contaminants (among them dyes) is produced every year in different branches of industry. A substantial part of them pollutes many water sources. Even a very small amount of dye in the water (10 – 50 mg/dm3) affects the water transparency and aesthetic values [9, 10]. There are more than 100.000 commercially

Magnetically Modified Biological Materials as Perspective Adsorbents...

303

available dyes and over 7x105 t of them are annually used in textile, paper, leather, plastics, food and cosmetics industries [11]. It is estimated that 20-50 % of these dyes are lost into wastewaters, causing environmental contaminations. Dyes may be toxic and mutagenic. They contaminate not only the environment but also traverse through the entire food chain, leading to the biomagnification. To decrease the dyes concentration, different procedures such as coagulation, flocculation, chemical degradation, oxidation, photodegradation, aerobic and anaerobic biodegradation, etc., are used [10]. However, such methods are often very expensive and cannot be used on a large scale. Moreover, many synthetic dyes are difficult to remove by the conventional wastewater systems, due to their complex chemical structure [10]. Therefore, the adsorption on appropriate adsorbents seems to be an efficient procedure for their removal. There are many adsorbents available, but the main attention is focused on cheap and easy to get materials which could be applied for large-scale processes. Among them, living or dead microorganisms (yeast, bacteria, fungi, algae) are intensively studied (see [8] and references therein). Each microorganism is able to bind or degrade several types of dyes and on the other side each dye can have an affinity to various microorganisms. In addition, the microbial cells efficiently interact with magnetic nanoparticles present both in low-pH and high-pH ionic magnetic fluids, leading to the formation of magnetically labeled cells, which could be easily separated from the system using an appropriate magnetic separator [8]. Another cheap and easy available adsorbents which could be applied for large-scale processes are plant-based materials (e.g. different types of sawdust). It has been known for a long time that these materials have an affinity for different biologically active compounds such as enzymes (e.g. trypsin [12], urokinase [13], elastase [14], cellulases [15]) or polyphenols [16]. Moreover, many organic compounds such as acid and basic dyes and oils [5, 6] and heavy metal ions [7] have been efficiently adsorbed on these materials. Therefore the sawdust seems to be also promising adsorbent for the removal of dyes. Currently, extensive studies are performed in many laboratories to find the optimal magnetic adsorbents with best magnetic and adsorption properties enabling to separate contaminating xenobiotics from large volumes of polluted water. These materials should be superparamagnetic to that they would exhibit magnetic properties when placed within a magnetic field (thus enabling their simple magnetic separation from the treated systems), but retained no residual magnetism when removed from the magnetic field. They should form stable colloidal suspensions and they should not aggregate in magnetic fields. They should also have an affinity to adequate xenobiotics. Inspired by these experiments, we directed our attention to the possibility of application of magnetically modified sawdust and microbial cells as new inexpensive and readily available magnetic adsorbents, which could be used for the separation and purification of biologically active compounds and xenobiotics. Therefore new ferrofluid-modified materials – spruce sawdust [17, 18], baker’s yeast (Saccharomyces cerevisiae) cells [19], brewer’s yeast (Saccharomyces cerevisiae subsp. uvarum) cells [20, 21], fodder yeast (Kluyveromyces fragilis) cells [22] and unicellular algae (Chlorella vulgaris) cells [23] – containing maghemite nanoparticles as magnetic labels – were prepared and tested as possible adsorbents for binding of different substances. The structural and magnetic properties of the developed materials were studied in detail by means of scanning electron microscopy, transmission electron microscopy, ESR spectroscopy and conventional magnetic methods (DC magnetization and AC susceptibility measurements). These studies are of considerable

304

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

interest for development of new inexpensive magnetic affinity adsorbents, which could exhibit the peculiar features enabling their rapid and efficient removal not only from solutions, but also from suspensions. Such materials could be efficiently used to isolate rare biologically active compounds from difficult-to-handle materials including raw extracts, blood and other body fluids, cultivation media, environmental samples, etc. They are also greatly suitable as magnetic adsorbents for large-scale magnetic separation processes.

2. MAGNETIC MODIFICATION OF BIOLOGICAL MATERIALS An absolute majority of biological materials is diamagnetic (i.e. is not attracted to a magnetic field). Therefore, when a magnetic separation technique is applied, these materials have to be first magnetically modified, usually by forming complexes with magnetic particles. Different procedures are available to convert diamagnetic biological materials into their magnetic derivatives [8]. However, for large-scale applications (e.g. in biotechnology or environmental technology), relatively cheap and readily available magnetic adsorbents are necessary. That is the reason why the new, extremely simple procedures for magnetic modification of biological materials were developed and presented here. According to these procedures the biological materials are magnetically modified by a contact with water-based ionic magnetic fluids stabilized with perchloric acid or tetramethylammonium hydroxide. The ferrofluids are prepared using the standard Massart procedure [24, 25] and contain magnetic iron oxide nanoparticles (usually in the form of maghemite) with the diameter of about 12 nm. Figure 1 shows exemplary TEM image of the obtained perchloric acid stabilized magnetic fluid and the corresponding log-normal distribution of particle diameters.

Number of particles

60 = 12.6 nm σ = 0.31

50 40 30 20 10 0

0

5

10

15

20

25

30

D [nm] Figure 1. TEM image of perchloric acid stabilized magnetic fluid obtained according to Massart procedure (the bar corresponds to 200 nm) and the corresponding log-normal distribution of particle diameters. The vertical bars are the experimental data, whereas the solid line represents the fit to the log-normal distribution function.

The above-mentioned ionic ferrofluids can be prepared in a simple way (almost in any lab) and such nanoparticles can be used to prepare biocomposite materials enabling their

Magnetically Modified Biological Materials as Perspective Adsorbents...

305

simple magnetic separation with standard permanent magnets. Both of these properties are important for large-scale magnetic separation techniques. Having in mind possible large-scale applications, the inexpensive raw materials should be also used. Therefore spruce sawdust (waste material) and microbial cells (baker’s, brewer’s and fodder yeast and unicellular algae, all of them produced in large quantities and available at low price) were taken into consideration. In order to produce magnetically modified spruce sawdust the raw material was sieved to obtain particles with the diameter less than 0.5 mm. Then 500 mg of sawdust was suspended in 7 ml of methanol in a test tube and 1 ml of perchloric acid stabilized ferrofluid was added. The suspension was mixed on a rotary mixer (Dynal) for 1 hour. During this time almost complete adsorption of ferrofluid nanoparticles on the sawdust occurred. Then the magnetic sawdust was repeatedly washed with water and the suspension was stored at 4oC [18]. Alternatively, the modified sawdust was washed with methanol and air-dried.

Figure 2. Raw (SEM image) and magnetically modified (optical micrograph) spruce sawdust particles. The scale bars correspond to 50 μm.

Figure 2 shows microscope images of spruce sawdust particle before and after magnetic modification. It is seen that maghemite nanoparticles are attached to the sawdust particle surface as the individual magnetic particles or agglomerates of these particles. The prepared magnetic sawdust can be easily manipulated by means of magnetic field and therefore it can be used as the magnetic adsorbent in magnetic separation techniques. A different procedure was used when working with dried fodder yeast (Kluyveromyces fragilis) and unicellular algae (Chlorella vulgaris) cells. The microbial cells were first washed six to eight times with an excess of 0.1 M acetic acid in order to remove substantial portion of soluble macromolecules which otherwise caused spontaneous precipitation of magnetic fluid. Then 1 ml of perchloric acid stabilized ferrofluid was added to 3 ml of the suspension of washed cells in acetic acid (1 + 3, v/v) and the suspension was mixed at room temperature for 1 h on a Dynal MX1 sample mixer (Invitrogen, USA). The adsorption of maghemite nanoparticles onto the microbial cells was fast; majority of nanoparticles was adsorbed within several minutes. The residual ferrofluid was removed by washing with 0.1 M acetic acid and then by repeated washing with water, until the supernatant was clear. The magnetized cells were captured using an appropriate magnetic separator. The resultant magnetic adsorbents were stored in water suspensions at 4 oC [19, 20].

306

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

Figure 3. Cross-sections of fodder yeast cells (before and after magnetic modification) observed by TEM. The scale bars correspond to 200 nm.

Figure 4. TEM picture of original dried Chlorella vulgaris cell (left) and magnetically modified cell (right). The bar lines correspond to 1 μm.

Figures 3 and 4 show TEM images of original dried and magnetically modified microbial cells. The drying process caused damage to the cell walls, often followed by the release of the intracellular components. Magnetic modification influenced the whole cells not the cell fragments. Analysis of TEM micrographs showed the presence of both isolated magnetic nanoparticles and their agglomerates on the cell surface. The nanoparticles were roughly spherical in shape and externally attached to the microbial cells walls. The outer cell surface preferentially accumulated magnetic nanoparticles even in the case of ruptured cells; only negligible binding of magnetic nanoparticles on the inner cell wall surface was observed. Obtained this way magnetically modified microbial cells could be easily separated using commercially available magnetic separators or strong permanent magnets and therefore they can be used as cheap magnetic adsorbents. These biocomposite materials were stable even after one year storage of the suspension at 4 ºC. It was found recently that the procedure used for sawdust modification (methanol / acid ferrofluid) can be also successfully used for dried microbial cells modification (unpublished results). To prepare magnetically responsive baker’s yeast cells with highly active intracellular enzymes, tetramethylammonium hydroxide stabilized ferrofluid in glycin-NaOH buffer was used; such cells have been used as non-toxic and efficient magnetically responsive whole cell biocatalysts for hydrogen peroxide decomposition and invert sugar formation [26]. As mentioned above the new developed procedures of obtaining of magnetically modified biological materials are cheap and extremely simple and may be performed almost in any lab. Therefore they allow to produce also other new magnetic adsorbents which can be used for large-scale magnetic separation.

Magnetically Modified Biological Materials as Perspective Adsorbents...

307

3. MAGNETIC CHARACTERIZATION OF MAGNETICALLY MODIFIED BIOLOGICAL MATERIALS Magnetically modified biological materials obtained according to above-mentioned preparation procedure were characterized by means of EPR spectroscopy and conventional magnetic methods in order to test if they can be used as magnetic affinity adsorbents in magnetic separation procedures. As mentioned above, such magnetic adsorbents should be superparamagnetic to that they would exhibit magnetic properties when placed within a magnetic field, but retained no residual magnetism when removed from the field. They should also form stable colloidal suspensions and they should not aggregate in magnetic fields.

3.1. DC Magnetization Measurements DC magnetization was measured by means of an extraction magnetometer MagLab 2000 System (Oxford Instruments Ltd.) in the applied magnetic field ± 3 kOe and wide temperature range 4 – 300 K. Figure 5 shows temperature dependencies of magnetization measured in zero field cooled – field cooled (ZFC-FC) regime at the applied magnetic field of 50 Oe for magnetically modified spruce sawdust and two kind of magnetically modified microbial cells (Kluyveromyces fragilis and Chlorella vulgaris). The ZFC curves were obtained by first cooling the samples in zero magnetic field from 300 to 4 K. Then the magnetic field H = 50 Oe was applied and the magnetization was measured with increasing temperature. The FC curves were obtained in a similar manner except that the samples were cooled in the same measuring field H = 50 Oe. It is seen that the ZFC-FC curves split at T < TB (TB = 190K, 210 K and 250 K for fodder yeast cells, algae cells and spruce sawdust, respectively), indicating the existence of the irreversible processes. The observed behavior is reminiscent of a blocking process of small single domain particles, which turn to a superparamagnetic state with increasing temperature. The ZFC curves for all investigated materials show a maximum associated to the transition between the superparamagnetic and blocked state. Moreover these maxima are very broad and a clear Curie-Weiss law behavior is not observed above the blocking temperature TB. This indicates the existence of dipole-dipole interactions between the particles, which may lead to the formation of small agglomerates and thus a wide distribution in particle size ranging from ultrafine isolated particles up to particle aggregates. The magnetization in the FC curves for all materials does not fulfill the Curie-Weiss law, which also reveals the presence of non-negligible dipole-dipole interactions between the particles. Such behavior has been observed in several iron oxide particle systems [27 - 29]. At about 250 K and 280 K a kink is seen in both ZFC and FC curves for microbial materials and spruce sawdust, respectively, which is associated with the melting point of the solution. Above the blocking temperature, the field dependent magnetization curves of all prepared materials show the superparamagnetic behavior indicated by the absence of hysteresis (see Figure 6). At lower temperatures the magnetization curves display the hysteresis, which confirms that the maghemite nanoparticles are in the ferrimagnetic state.

308

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik 0,08 0,07

M [emu/g]

0,06

fodder yeast cells

FC FC

algae cells

FC

0,05

spruce sawdust

0,04 0,03

ZFC

0,02 0,01

ZFC

ZFC

0

50

100

150

200

250

300

T[K] Figure 5. Temperature dependencies of magnetization for magnetically modified fodder yeast and algae cells and spruce sawdust recorded at the applied magnetic field of 50 Oe. 0,4

fodder yeast cells

0,3

T = 4.2 K algae cells

M [emu/g]

0,2 0,1

spruce sawdust

0,0 -0,1 -0,2 -0,3 -0,4 -3

-2

-1

0

1

2

3

H [kOe] 0,4 0,3

fodder yeast cells

T = 300 K algae cells

M [emu/g]

0,2 0,1

spruce sawdust

0,0 -0,1 -0,2 -0,3 -0,4

-0,3

-0,2

-0,1

0,0

0,1

0,2

0,3

H [kOe]

Figure 6. Field dependent hysteresis loops for magnetically modified fodder yeast and algae cells and spruce sawdust recorded at 4.2 and 300 K.

Magnetically Modified Biological Materials as Perspective Adsorbents...

309

At T = 4.2 K, the remanence-to-saturation ratio for all prepared materials, R = MR/MS = 0.37, is smaller than the expected R = 0.5 value for non-interacting, randomly oriented particles with uniaxial symmetry [30]. It is an additional confirmation for the existence of interparticle magnetic dipole-dipole interactions.

3.2. AC Susceptibility Measurements In order to study the effects of interparticle interactions on the dynamics of the blocking process, temperature dependence of the in-phase (real) component χ' (T) and the out-of-phase (imaginary) component χ '' (T) of the AC magnetic susceptibility were measured for different frequencies ƒ of the excitation field for all magnetically modified materials. Measurements were performed by means of an extraction magnetometer MagLab 2000 System (Oxford Instruments Ltd.) using an excitation field of 10 Oe and driving frequencies in the range 36 – 2237 Hz. The experimental data (see Figure 7) for all prepared materials exhibit the expected behavior of a superparamagnetic system, i.e., the occurrence of a maximum in both components χ' (T) and χ'' (T) at different temperatures Tm' and Tm'' which shift towards higher temperatures with increasing frequency ƒ [31]. At about 260 K and 280 K a kink is seen in both χ' (T) and χ '' (T) curves for microbial materials and spruce sawdust, respectively, which is associated with the melting point of the solution. At low temperatures all χ'' (T) curves show oscillations, which are due to the imperfection of the experimental system. The real component of the AC susceptibility of all materials shows a value not equal to zero for T approaching zero (see Figure 7). It may be due to the presence of small agglomerates of particles coupled by magnetic dipolar interactions. These interactions modify the magnetic behavior of the system, introducing a collective component which has the influence on the low temperature magnetic relaxation. It has been shown [31] that the magnetic relaxation of an interacting nanosized magnetic particle system at low temperatures is extended towards longer time scales as compared to the relaxation of a non-interacting particle system. Another indication of the influence of magnetic dipole-dipole interactions on the dynamics of the samples comes from the increasing with increasing frequency of the height of the peak in χ'' (T) for all investigated materials (see Figure 7), whereas it is almost constant with frequency for a non-interacting system [31]. The empirical parameter Φ [32], which represents the relative shift of the temperature Tm per interval of frequency,

Φ=

ΔTm , Tm Δ log10 ( f )

(3.1)

calculated from the frequency dependence of the maximum in the imaginary χ'' (T) part of the AC susceptibility is equal to 0.10, 0.09 and 0.08 for spruce sawdust, yeast cells and algae cells, respectively. These values are close to the 0.1 – 0.13 values found for superparamagnetic systems [32]. The slightly lower values originate from interparticle magnetic dipole-dipole interactions.

310

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

12

χ' x 10

4 0

36 Hz 216 Hz 1117 Hz 2237 Hz

15 10

-5

8

χ" x 10 [emu/g*Oe]

20 36 Hz 216 Hz 1117 Hz 2237 Hz

16

-4

[emu/g*Oe]

20

fodder yeast cells 0

50

5 fodder yeast cells 0

100 150 200 250 300 350

0

50

100 150 200 250 300 350 T [K]

8

-5

χ' x 10

χ" x 10 [emu/g*Oe]

12

36 Hz 216 Hz 1117 Hz 2237 Hz

12

-4

[emu/g*Oe]

T [K]

4

algae cells 0

0

50

8

4 algae cells 0

100 150 200 250 300 350

36 Hz 216 Hz 1117 Hz 2237 Hz

0

50

100 150 200 250 300 350 T [K]

4

4 3

[emu/g Oe]

36 Hz 216 Hz 937 Hz 2237 Hz

36 Hz 216 Hz 937 Hz 2237 Hz

3 2

-5

2 1 0

spruce sawdust 0

50

100 150 200 250 300 350

χ" x 10

χ' x 10

-4

[emu/g Oe]

T [K]

1 spruce sawdust 0

0

50

T [K]

100 150 200 250 300 350 T [K]

Figure 7. Temperature dependencies of the real (χ') and imaginary (χ'') components of the magnetic susceptibility for magnetically modified fodder yeast and algae cells and spruce sawdust recorded at different excitation frequencies. Arrows indicate increasing frequencies. The data were taken with an external magnetic field H = 10 Oe.

3.3. Electron Spin Resonance (ESR) Measurements In order to give a deeper insight into the nature of the magnetic structures described above, ESR measurements were performed. ESR spectra were recorded by means of a

Magnetically Modified Biological Materials as Perspective Adsorbents...

311

standard X-band spectrometer (Bruker EMX – 10/12) operating at 9.46 GHz with 100 kHz field modulation. Resonance absorption was measured as a first derivative of the absorbed microwave power versus magnetic field. Figure 8 shows the ESR spectra of all magnetically modified materials plus the spectra of the concentrated ferrofluid (1:1) used for magnetic modification and diluted ferrofluid (1:100). The huge dilution (from 1:1 to1:100) of the ferrofluid sample causes a relative shift in the resonance field of about 50 Oe. However, the shift observed in the resonance field between the concentrated ferrofluid sample (1:1) and the ferrofluid-modified materials is about 196, 319 and 362 Oe for spruce sawdust, fodder yeast cells and algae cells, respectively. This remarkable resonance field shift is the signature of a strong interaction among the maghemite nanoparticles when attached to the biological materials. The ESR data confirm that maghemite nanoparticles do attach to the microbial cell wall and spruce sawdust surface after incubation of these materials with the biocompatible magnetic fluid sample as revealed by the microscope micrographs (see Figures 2, 3 and 4). The exemplary ESR spectra of magnetically modified biological materials recorded at various temperatures are presented in Figure 9. The room temperature spectra for all materials show well-defined single broad signals with an effective g value of about 2.1 and the peak-topeak line width ΔHpp = 1040, 990 and 880 Oe for spruce sawdust, fodder yeast cells and algae cells, respectively. The line widths of these signals considerably exceed the magnetocrystalline-anisotropy-determined minimum value, ΔHpp = 400 Oe, for noninteracting single domain maghemite particles [33]. It suggests the existence of nonnegligible dipole-dipole interactions between nanoparticles.

ESR signal intensity [arb. units]

15

spruce sawdust fodder yeast cells

10

algae cells

5 0 -5 -10

diluted ferrofluid 1:100

-15 ferrofluid 1:1

-20 1,5

2,0

2,5

3,0

3,5

4,0

4,5

5,0

H [kOe] Figure 8. Room temperature ESR spectra of maghemite nanoparticles in labeled fodder yeast and algae cells and spruce sawdust and maghemite nanoparticles suspended as a stable magnetic fluid samples at different concentrations.

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik ESR signal intensity [arb. units]

312

1

T = 300 K

T = 4.2 K

T = 100 K

0

-1 fodder yeast cells 0

1

2

3

4

5

6

ESR signal intensity [arb. units]

H [kOe] 2 T= 300 K

T= 4.2 K 1

T= 100 K

0 -1 -2

algae cells 0

1

2

3

4

5

6

ESR signal intensity [arb. units]

H [kOe] T = 100 K 2

T = 300 K

T = 4.2 K Cu

2+

0

-2 free radicals 0

1

2

spruce sawdust 3

4

5

6

H [kOe] Figure 9. ESR spectra of magnetically modified fodder yeast and algae cells and spruce sawdust recorded at various temperatures.

Magnetically Modified Biological Materials as Perspective Adsorbents...

313

On the spruce sawdust resonance line two more ESR signals are superimposed (see Figure 9), which are clearly seen only at low temperatures due to their paramagnetic behavior: • •

A sharp signal, at g = 2.00, which seems to be due to free radicals, usually found in biological materials [34]. A narrow (ΔBpp = 19 mT) resonance at g = 2.27, which seems to be attributed to the existence of paramagnetic Cu2+ complexes, sometimes found in wood [34].

Upon decreasing the temperature the main resonance signals for all materials shift to lower fields and gradually broaden, closely following the predictions for the ESR of superparamagnetic nanoparticles systems [35]. Therefore, it may be expected that the broadening and shift of these signals is associated with a blocking of the magnetization in maghemite nanoparticles. In general, for a superparamagnetic system of particles having a statistical distribution of shapes and sizes, the simple power relation between the resonance field shift δHres and the peak-to-peak line width ΔHpp can be expressed as [35]: δHres ~ (ΔHpp)n

(3.2)

where n = 2 (3) is predicted for partially (randomly) oriented particles. In order to test this power relation we have plotted the data of figure 9 on a double logarithmic scale (see Figure 10). It is seen that a straight line with a slope close to 3 (3.08 for spruce sawdust and 3.32 for microbial cells) can well approximate the data for all prepared materials. Thus, the broad ESR signal is shown to be due to the superparamagnetic behavior of small isolated randomly oriented maghemite particles.

200

200

spruce sawdust algae cells

δHres [Oe]

160

fodder yeast cells

140 n

100 80 60 40

δHres ~ (ΔHpp)

n = 3.08

120 1000

1050

1100

1150

ΔHpp [Oe]

1200 1250

δH

900

1000

1100

∼ (Δ H

)

pp n = 3.32

res

n

δHres [Oe]

180

20

1200 1300

ΔHpp [Oe]

Figure 10. Relation between the resonance field shift (δHres) and peak-to-peak line width (ΔHpp) of the ESR signal for magnetically modified spruce sawdust and fodder yeast and algae cells. Points correspond to the experimental data, the solid line is the best fit according to Equation 3.2.

314

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

The presented ESR results for all prepared materials are in good agreement with the magnetization, susceptibility and TEM measurements and confirm that the magnetic behavior of these materials is dominated by the superparamagnetic relaxation of isolated single domain maghemite particles, although a certain amount of agglomerates of these particles coupled by magnetic dipole-dipole interactions is also present. However, these agglomerates are sufficiently small to show at static conditions the superparamagnetic behavior at room temperature. Therefore, the obtained results are very promising from the point of view of using the prepared materials as the magnetic affinity adsorbent in the magnetic separation techniques.

4. ADSORPTION PROPERTIES OF MAGNETICALLY MODIFIED BIOLOGICAL MATERIALS Magnetically modified spruce sawdust and microbial cells were tested as possible adsorbents for binding of different substances [17-23]. They efficiently adsorbed different organic compounds, such as water-soluble organic dyes, and mercury ions. Eight different dyes belonging to four dye classes were tested, namely, crystal violet and aniline blue (triphenylmethane group), amido black 10B, congo red, Saturn blue LBRR and Bismarck brown Y (azodyes group), acridine orange (acridine group) and safranin O (safranin group). Commercially available dyes were used during the experiments. They were dissolved in distilled water without buffering the solution. Preliminary experiments also showed that adsorption properties of spruce sawdust and microbial cells were not significantly influenced by magnetic modification. The adsorption of the tested dyes reached equilibrium in approximately 30-120 min. In order to achieve equilibrium during the adsorption process, an incubation time of 3 h was used for all adsorption experiments. Measurements were performed in test tubes containing 50 μl of sedimented magnetically modified biological material, appropriate amount of tested dye solution and water to a total volume of 10 ml. The suspension was mixed for 3 h at room temperature. Then the magnetic adsorbent was separated from the suspension using a magnetic separator (Dynal MPC-1 or MPC-6, Invitrogen, USA) and the clear supernatant was used for the spectrophotometric measurement. The concentration of free (unbound) dye in the supernatant (Ceq) was determined from the calibration curve. The amount of dye bound to the unit volume of the adsorbent (qeq) was calculated using the following formula [20]:

q eq =

Dtot − 10C eq 50

(4.1)

where Dtot is the total amount of dye used in the experiment. Examples of equilibrium adsorption isotherms for the unbuffered water solutions of selected tested dyes using magnetically modified fodder yeast and algae cells as adsorbents are presented in Figure 11. These isotherms represent distribution of dyes between the

Magnetically Modified Biological Materials as Perspective Adsorbents...

315

aqueous and solid phases as the dye concentration increases. They follow the typical Langmuir adsorption pattern given by [22, 23]:

q eq =

Qmax bC eq

(4.2)

1 + bC eq

where qeq (expressed in mg/g or mg/ml) is the amount of the adsorbed dye per unit mass or sedimented volume of magnetically modified biomass and Ceq (expressed in mg/l) is the unadsorbed dye concentration in solution at equilibrium. Qmax is the maximum amount of the dye per unit mass or sedimented volume of biomass to form a complete monolayer on the surface bound at high dye concentration and b is a constant related to the affinity of the binding sites (expressed in l/mg). The Langmuir model allows simple calculation of maximum adsorption capacity Qmax, which is a very important parameter describing the adsorption process. The results obtained for the magnetically modified fodder, baker’s and brewer’s yeast and algae cells and spruce sawdust are presented in Table 1. Table 1. Comparison of maximum adsorption capacities of magnetically modified microbial cells and spruce sawdust for the tested dyes Maximum adsorption capacity of magnetically modified biomaterials [mg/g] Dye Acridine orange Amido black 10B Aniline blue Bismarck brown Congo red Crystal violet Safranin O Saturn blue LBRR

Fodder yeast cells [22] 62.2 29.9 75.7 49.7 42.9 138.2 33.0

Brewer’s yeast cells [20]

Baker’s yeast cells [19]

Algae cells

Spruce sawdust

[23]

[18]

82.8 11.6 228.0 93.1 41.7 46.6

430.2 85.9 90.3

24.1 257.9 201.9 156.7 42.9 115.7 24.2

52.1 52.4 25.0

The results show that dye sorption by ferrofluid-modified biomass follows a chemical, equilibrated and saturable mechanism. Thus, the adsorption increases when the initial dye concentration rises, as long as binding sites are not saturated. The fact that the Langmuir isotherm fits the experimental data very well may be due to the homogeneous distribution of active sites on the magnetically modified sawdust particles and microbial cells [36]. The prepared magnetically modified microbial cells were tested as possible adsorbents for binding of heavy metal ions. Heavy metal pollution represents an important environmental problem due to the toxic effects of metals, and their accumulation throughout the food chain leads to serious ecological and health problems. Magnetically modified brewer’s yeast were used to study Hg2+ biosorption-desorption process using synthetic solutions in batch system. The biosorption process was fast; 80% of biosorption occured within 60 min and equilibrium was achieved at around 90 min. The maximum Hg2+ biosorption capacity was 114.6 mg/g at 35 ºC and this value decreased to 76.2 mg/g at room temperature. The adsorption was well

316

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

fitted to the Langmuir isotherm. Results suggest that chemisorption processes could be the rate-limiting step in the biosorption process. The yeast biomass could be easily and efficiently regenerated by 0.1M HNO3. Biosorption of other heavy metal ions from artificial wastewater was also studied; the biosorption capacities were 29.9 mg/g for Cu2+, 14.1 mg/g for Ni2+ and 11.8 mg/g for Zn2+ at room temperature [21]. fodder yeast cells algae cells

160

Bismarck brown Congo red

Safranin O

qeq [mg/g]

120

Safranin O

80

Bismarck brown Congo red

40 0

0

50

100

150

200

250

3

Ceq [mg/dm ]

Figure 11. Equilibrium adsorption isotherms (calculated from experimental data using Langmuir equation (4.2)) for unbuffered water solutions of tested organic dyes using magnetically modified fodder yeast [22] and algae cells [23]. Ceq – equilibrium liquid-phase concentration of the unbound dyes (mg/dm3); qeq – equilibrium solid-phase concentration of the adsorbed dyes (mg/g).

The prepared magnetically modified spruce sawdust was additionally tested as a possible adsorbent for binding of different enzymes [17]. It was successfully used for the batch separation of proteolytic enzymes (trypsin, bacterial protease produced by a strain of Bacillus sp.) and hen egg white lysozyme. The degree of lysozyme purity increased from 65% (a technical preparation) to 96% (after the magnetic sawdust treatment) [17].

5. CONCLUSION The main aim of this work was to find cheep and easy to get magnetic adsorbents which could be used for large-scale magnetic separation procedures. Such materials should be superparamagnetic to that they would exhibit magnetic properties when placed within a magnetic field, but retained no residual magnetism when removed from the magnetic field. They should form stable colloidal suspensions and they should not aggregate in magnetic fields. They should also have an affinity to corresponding biologically active compounds and xenobiotics. We proposed an inexpensive, extremely simple procedure for the preparation of such magnetic adsorbents using standard water-based ferrofluids containing magnetic iron oxides (mainly maghemite) nanoparticles with the diameter of about 12 nm. Such ferrofluids can be prepared in a simple way and such nanoparticles can be used to prepare biocomposite

Magnetically Modified Biological Materials as Perspective Adsorbents...

317

materials enabling their simple magnetic separation with standard permanent magnets. Both of these properties are important for large-scale applications. Inexpensive and easily available raw materials – spruce sawdust, fodder yeast (Kluyveromyces fragilis), baker’s yeast (Saccharomyces cerevisiae), brewer’s yeast (Saccharomyces cerevisiae subsp. uvarum) and unicellular algae (Chlorella vulgaris) cells – were used for magnetic modification. It allows to use these materials for large-scale applications. The prepared magnetic adsorbents efficiently adsorbed several water-soluble organic dyes, belonging to different groups, and also heavy metal ions. It means that they can thus be new promising magnetic affinity adsorbents which may be used to the large-scale removal of environmental contaminants from polluted water sources. The magnetic measurements show that the magnetic behavior of the prepared materials is mainly dominated by the superparamagnetic relaxation of isolated single domain maghemite nanoparticles, although a little amount of agglomerates is also present. However, these agglomerates are sufficiently small to show at static conditions the superparamagnetic behavior at room temperature. The magnetic properties of the prepared magnetic adsorbents enable their rapid and efficient removal not only from solutions, but also from suspensions, so they could be used in the separation process performed directly in unprocessed samples such as waste water, biological fluids, fermentation media, etc. Summarizing, inexpensive raw materials, extremely simple preparation method, affinity to various biologically active compounds and both organic and inorganic xenobiotics, and distinctive magnetic properties make the developed materials greatly suitable as new magnetic adsorbents for large-scale magnetic separation processes.

REFERENCES [1]

Safarik, I; Safarikova, M. In Scientific and Clinical Applications of Magnetic Carriers; Hafeli, U; Schutt, W; Teller, J; Zborowski, M. Ed; Plenum: New York and London, 1997, pp 323-340 . [2] Safarik, I; Safarikova, M. In Encyklopedia of Separation Science; Wilson, I. D; Adlard, T. R; Poole, CF; Cool, M; Ed; AcademicPress Ltd.: London, 2000, pp 2163-2170. [3] Safarik, I; Safarikova, MJ. Chromatogr. B., 1999, 722, 33-53. [4] Ebner, AD; Ritter, JA; Ploehn, HJ; Koche, RL; Navratil, JD. Separ. Sci. Technol., 1999, 34, 1277-1300. [5] Shukla, A; Zhang, YH; Dubey, P; Margrave, J. L; Shukla, SS. J. Hazard. Mater., 2002, 95, 137-152. [6] Garg, VK; Kumar, R; Gupta, R. Dyes Pigment., 2004, 62, 1-10. [7] Yu, B; Zhang, Y; Shukla, A; Shukla, SS; Doris, KLJ. Hazard Mater, 2001, 84, 83-94. [8] Safarik, I; Safarikova, M. China Particuology, 2007, 5, 19-25. [9] Banat, I. M; Nigam, P; Singh, D; Marchant, R. Bioresour. Technol., 1996, 58, 217-227. [10] Robinson, T; McMulan, G; Marchant, R; Nigam, P. Bioresour. Technol, 2001, 77, 247-255. [11] Zollinger, H. Colour Chemistry: Synthesis, Properties and Applications of Organic Dyes and Pigments; VCH Publishers: New York, 1987.

318

Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik

[12] Safarik, IJ. Chromatogr. A, 1984, 294, 504-506. [13] Kobayashi, O; Matsui, K; Minamiura, N; Yamamoto, TJ. Chromatogr. A, 1981, 210, 180-185. [14] Fujimoto, KI; Ogawa, M; Saito, N; Kosaki, G; Minamiura, N; Yamamoto, T. Biochim. Biophys. Acta, 1980, 612, 262-267. [15] Palonen, H; Tjerneld, F; Zacchi, G; Tenkanen, M. J. Biotechnol. 2004, 107, 65-72. [16] Sakanaka, SJ. Agr. Food Chem., 2003, 51, 3140-3143. [17] Safarik, I; Safarikova, M; Weyda, F; Mosiniewicz-Szablewska, E; ŚlawskaWaniewska, AJ. Magn. Magn. Mater., 2005, 293, 371-376. [18] Safarik, I; Lunackova, P; Mosiniewicz-Szablewska, E; Weyda, F; Safarikova, M., Holzforschung, 2007, 61, 247-253. [19] Safarik, I; Ptackova, L; Safarikova, M. Eur. Cells Mater., 2002, 3 Suppl. 2, 52-55. [20] Safarikova, M; Ptackova, L; Kibrikova, I; Safarik, I. Chemosphere, 2005, 59, 831-835. [21] Yavuz, H; Denizli, A; Gungunes, H; Safarikova, M; Safarik, I. Sep. Purif. Technol., 2006, 52, 253-260. [22] Safarik, I; Rego, LFT; Borovska, M; Mosiniewicz-Szablewska, E; Weyda, F; Safarikova, M. Enz. Microb. Technol. 2007, 40, 1551-1556. [23] Safarikova, M; Rainha Pona, BM; Mosiniewicz-Szablewska, E; Weyda, F; Safarik, I. Fresenius Environ. Bull., 2008, 17, 486-492. [24] Massart, R. IEEE Trans. Magn., 1981, 17, 1247-1248. [25] Berger, P; Adelman, N. B; Beckman, K. J; Campbell, D. J; Ellis, A. B; Lisensky, G. C. J. Chem. Educ., 1999, 76, 943-948. [26] Safarikova, M; Maderova, Z; Safarik, I. Food Res., Int., in press. [27] Del Barco, E; Asenjo, J; Zhang, XX; Pieczyński, R; Julia, A; Tejada, J; Ziolo, RF; Fiorani, D; Testa, AM. Chem. Mater., 2001, 13, 1487-1490. [28] Mamiya, H; Nakatami, I; Furubayashi, T. Phys. Rev. Lett., 1998, 80, 177-180. [29] Dormann, JL; D’Orazio, F; Lucari, F; Tronc, E; Pren, P; Jolivet, J; Fiorani, D; Cherkaoui, R; Nogues, M. Phys. Rev. B, 1996, 53, 14291-14297. [30] [Denham, CR; Blackemore RP; Frankel, RB. IEEE Trans. Magn., 1980, 16, 1006-1007. [31] Jonsson, T; Norblad, P; Svedlindh, P. Phys. Rev. B, 1998, 57, 497-504. [32] Dormann, JL; Fiorani, D; Tronc, E. Adv. Chem. Phys., 1997, XCVIII, 326-330. [33] Dunlop, DJ; Ozdemir, O. Rock Magnetism: Fundamentals and Frontiers; Cambridge University Press: New York, 1997. [34] Swartz, HM; Bolton, JR; Borg, DC. Biological Applications of Electron Spin Resonance; Wiley-Interscience, a Division of John Willey & Sons, Inc: New York, 1972. [35] Nagata, K; Ishihara, AJ. Magn. Magn. Mater., 1992, 104-107, 1571-1573. [36] Ozacar M; Sengil, IA. Bioresour. Technol. 2005, 96, 791-795.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 319-340

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 10

OPTIMISATION OF DEPOSITION CONDITIONS FOR FUNCTIONAL OXIDE THIN FILMS Sharath Sriram* and Madhu Bhaskaran Microelectronics and Materials Technology Centre, School of Electrical and Computer Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia.

ABSTRACT This chapter describes a systematic approach to determining the optimal conditions required to deposit thin films of complex functional oxides using RF magnetron sputtering. The motivation of this study was to attain films of designed stoichiometry and of preferentially oriented perovskite crystal structure, as the composition and structure of complex oxides determines their functionality (e.g. ferroelectricity, piezoelectricity, etc.) which is exploited by a variety of applications. The process has been demonstrated with strontium-doped lead zirconate titanate (PSZT) thin films using a combination of materials characterisation techniques – X-ray diffraction, X-ray photoelectron spectroscopy, atomic force microscopy, and crystal structure calculations. The major variables in the deposition process using RF magnetron sputtering were identified as the deposition substrate temperature, post-deposition thermal processing, oxygen partial pressure during deposition, and the choice of metallisation on silicon substrates. Starting with these variables the influence of each was systematically determined. The results highlighted the advantages of slow cooling to promote perovskite growth, the influence of oxygen on the composition and crystal structure of the thin films, and the presence of modified unit cell structure for the PSZT thin films on gold- and platinum-coated silicon substrates. The combination of these results led to the limiting the variables in deposition to finite values and arriving at two definite sets of deposition conditions for deposition based on the dependence of microstructure on thermal conditions and the choice of substrate. The validity of the conditions chosen is then demonstrated by deposition of PSZT thin films on thermally-grown silicon dioxide and attaining nanocolumnar preferentially oriented thin films.

*

Corresponding author: [email protected], [email protected]

320

Sharath Sriram and Madhu Bhaskaran

1. INTRODUCTION In the pursuit of enhanced functionality and miniaturisation, high performance thin film coatings are increasingly incorporated into device designs. Oxides form a very large category of materials used for processes reliant on functionality. Oxide compounds can exhibit a wide variety of properties such as semiconduction, piezoelectricity, ferroelectricity, gas sensitivity, etc. Oxides can exhibit one or more functional properties, with properties often enhanced in thin films, leading to the widespread use of the terminology ‘multifunctional oxide thin films’. The ability to exhibit multifunctional behaviour by an oxide is determined by its chemical composition and crystal structure. With thin film deposition often done from pure source material of desired composition, the resulting structure is the primary concern (though the stoichiometry of the thin films will need to be verified). Most multifunctional oxides often have complex crystal structure symmetries, such as tetragonal or rhombohedral phases [1]. This structure is often collectively termed ‘perovskite’ [2,3]. Such compounds generally have the ABO3 chemical structure. An example of such a multifunctional oxide with perovskite structure is lead titanate (PbTiO3) which exhibits piezoelectric, ferroelectric, and electro-optic properties. In this case lead (Pb) is the A-site compound and titanium (Ti) is the B-site compound. The properties exhibit by such multifunctional oxide compounds can be further enhanced by the addition of dopants. These dopants can replace either a small percentage of A- or B-site atoms or both. This results in ‘complex multifunctional oxides’. One of the most popular compounds in this category, with a B-site substituent is lead zirconate titanate [PZT: Pb(Zr1xTix)O3], in which the ferroelectric and piezoelectric properties are enhanced, resulting also in a more stable compound. Additional substitution of A-site lead atoms of PZT results in compounds with either new properties or further enhancement in existing ones: (i) lanthanumdoped PZT (PLZT) is known for enhanced electro-optic behaviour [4]; (ii) stannum-doped PZT and niobium-doped PZT (PNZT) are known to exhibit shape memory behaviour [5] (not exhibited by PZT); and (iii) strontium-doped PZT (PSZT) has been reported to have improved piezoelectric, ferroelectric, dielectric tenability, and pyroelectric (infra-red radiation sensors by thermal effects) behaviour [3,6-8]. The possible multidisciplinary applications of oxide thin films are apparent, and the multifunctionality displayed is dependent on the chemical and crystal structure. Accurate control of deposition process and understanding of variables which control the resulting films is vital to harnessing the capabilities multifunctional oxide thin films present. This chapter uses a selected complex oxide and demonstrates the process of optimisation of deposition conditions. The compound chosen is strontium-doped PZT (PSZT) which has both A- and Bsite dopants. The composition chosen of (Pb0.92Sr0.08)(Zr0.65Ti0.35)O3 is reported to exhibit enhanced properties [3,6-8] and is expected to have the complex rhombohedral crystal structure at room temperature [9]. The deposition process of choice is RF magnetron sputtering, which enables large (wafer-) scale deposition and is suitable for commercial production.

Optimisation of Deposition Conditions for Functional Oxide Thin Films

321

2. EXPERIMENTAL DETAILS This chapter focuses on determination of optimised deposition conditions for thin films by RF magnetron sputtering. PSZT thin films were deposited under the conditions shown in Table 1, which highlights the variables in this process. Table 1. Variables in PSZT thin film deposition conditions. Target Target diameter RF power Target to substrate distance Base pressure Sputtering pressure Process gas Substrate temperature Temperature ramp-up rate Temperature ramp-down rate

(Pb0.92Sr0.08)(Zr0.65Ti0.35)O3 100 mm 100 W 70 mm 9.0 x 10-6 Torr variable 10% oxygen in argon variable 10 ºC/min variable

Sputtering parameters principally determining the deposition rate of the thin films were kept constant. These parameters (deposition power and target-substrate distance) were chosen so as to achieve a compromise between a slow deposition rate to enable grain growth and crystallisation, while achieving a reasonable deposition rate to achieve micron-thick films. Initial tests showed that the choice of forward RF power of 100 W (maintaining the reflected power at ~0 W) and a target-substrate distance of 70 mm, resulted in a deposition rate of 300400 nm/hour. This variation is over 3-inch substrates due to the absence of substrate rotation option in the sputtering system. This deposition rate was considered satisfactory to achieve micron-thick films, and as a result these parameters were kept constant during further study. Sputtering parameters such as post-deposition cooling rate, sputtering pressure, and substrate temperature which are expected to have a significant influence on the composition, crystal structure, and microstructure of the thin films were varied during this study.

3. INFLUENCE OF POST-DEPOSITION COOLING RATE While the substrate temperature during deposition is a critical variable in determining the crystallography of these complex perovskite-structured PSZT thin films (the influence of which is discussed in Ref. [10]), post-deposition thermal processing was expected to influence the degree of grain growth and crystallisation. Extensive X-ray diffraction analysis of samples deposited at high temperatures and those deposited at lower temperatures followed by annealing was carried out. This work found that slower post-deposition cooling results in thin films exhibiting more pronounced perovskite structure. The optimal post-deposition cooling rate was 5 ºC/min. An example of the results obtained is presented in Fig. 1, where the switch in the intensity of the pyrochlore and perovskite peaks at around 2θ of 30º is apparent.

322

Sharath Sriram and Madhu Bhaskaran

Figure 1. X-ray diffraction results for PSZT thin films deposited at 600 ºC: (a) as deposited and cooled at 15 ºC/min, (b) as deposited and cooled at 5 ºC/min, (c) cooled at 15 ºC/min and post-deposition annealed at 1000 ºC for 30 min in argon. (Reprinted with permission from Ref. [10].)

The process of slower cooling was also found to be better than post-deposition annealing, as it: (i) prevented the occurrence of microcracks in the thin films and (ii) resulted in stronger preferential orientation, rather than a collection of perovskite orientations. This study [10] supported results previously presented for other ferroelectric material by Wasa et al. [11], and helped limit the main deposition variables to two.

4. INFLUENCE OF OXYGEN PARTIAL PRESSURE The level of piezoelectric response exhibited by PZT compounds depends on the thin films having a composition closely corresponding to the ideal ABO3 structure and on having a preferential perovskite orientation. Lead loss and lead excess have been reported to have a

Optimisation of Deposition Conditions for Functional Oxide Thin Films

323

marked influence on the materials properties of PZT type compounds [12,13], with the composition variations affecting perovskite orientation. Deposition of PSZT thin films was carried out in an atmosphere of 10% oxygen in argon. Argon is an inert gas and exerts no influence on thin film composition. Changes in sputtering pressure are directly related to changes in the partial pressure of oxygen present in the sputtering environment. This section reports on the influence of the oxygen partial pressure on the chemical composition, crystal structure, and surface morphology of the deposited PSZT thin films. Deposition was carried out on platinum-coated silicon substrates at 650 ºC – the substrate temperature was kept constant to study the influence of partial pressure. The results obtained from films deposited at oxygen partial pressures of 1, 2, 3, 4, and 5 mTorr (relating to sputtering pressures from 10-50 mTorr) are discussed.

4.1. Composition Analysis X-ray photoelectron spectroscopy (XPS) analysis was carried out to study the composition variations in the PSZT thin films deposited at varying oxygen partial pressures. Peaks for lead, zirconium, titanium, and oxygen expected at around 137 eV, 181 eV, 454 eV, and 531 eV, respectively, were observed. Examples of the spectra recorded are shown in Fig. 2. Peaks for the dopant (strontium) were not pronounced due to its low concentration, and possible detected peaks at ~133 eV for strontium could not be resolved satisfactorily due to their vicinity to the peaks of lead.

Figure 2. XPS spectra for PSZT thin films deposited at sputtering pressures of 10, 20, and 30 mTorr. The spectra for 10 mTorr and 30 mTorr overlap, corresponding to the similar stoichiometries estimated. (Also see Table 2. Reprinted with permission from Ref. [14].)

XPS analysis highlights the marked influence of variations in oxygen partial pressure during sputtering, observed by variations in oxygen (O) concentration in the thin films, and in some cases by the undesirable decrease in lead (Pb) concentration in the thin films (Table 2). Films deposited at sputtering pressures of 10 and 30 mTorr (oxygen partial pressures of 1 and 3 mTorr, respectively) have resulting compositions which are acceptably close to the PSZT

324

Sharath Sriram and Madhu Bhaskaran

8/65/35 stoichiometry of the sputtering target. The composition of films deposited at sputtering pressures of 20 mTorr and 50 mTorr indicates insufficient and excess lead, respectively. The zirconium (Zr) to titanium (Ti) ratio of films deposited at 10, 30, and 40 mTorr was reasonably constant (the ratio varying between 1.90 and 2.15), while the lead and oxygen content varied relative to the other. As can be observed from Table 2, the films deposited at 10 mTorr appears to have the composition closest to the desired ABO3, albeit mildly lead deficient. Table 2. Variations in composition of PSZT thin films with sputtering pressure. Sputtering pressure (mTorr)

O2 partial pressure (mTorr)

10 20 30 40 50

1 2 3 4 5

Elemental composition (%) Pb

Zr

Ti

O

18.18 12.80 22.21 25.83 26.41

14.72 17.94 14.28 14.23 13.14

7.73 9.19 6.81 6.63 5.40

59.36 60.08 56.70 53.24 55.05

The trend observed in the composition (with the exception of films deposited at 20 mTorr) agrees with earlier studies on diffusive transport regime assisted deposition of heavier atoms [15,16]. With increasing sputtering pressure, lighter atoms experience more collisions (with decreased mean free path), as a result of which the deposited films become richer in the heavier species which constitute the film. A combination of the previous studies on transport regimes during sputtering and the results presented here provide a technique for optimising lead content (by altering the sputtering pressure) in lead-based or PZT-type films, where leadloss during sputtering often occurs, resulting in poor electrical performance.

4.2. Crystallographic Orientation Analysis Glancing angle X-ray diffraction (GA-XRD) has been used to study the crystallographic orientation of the deposited PSZT thin films. PSZT film peaks in diffractograms obtained using Bragg-Brentano XRD were over-shadowed by substrate peaks; this was overcome by using GA-XRD. All deposited thin films were poly-crystalline with preferential orientation(s). Figures 3(a) and 3(b) show expected perovskite peaks at 2θ of 29.6º and 34.3º and these correspond to (104) and (006) orientations for a rhombohedral PSZT unit cell [17]. Figure 3(c) shows another expected peak for PSZT thin films corresponding to the (108) orientation. The relative shifts of the perovskite peaks with respect to variation in oxygen partial pressure are evident in the diffractograms shown in Fig. 3. Films deposited at 10 and 30 mTorr, and which show similar composition, show similar orientation and least variations from expected 2θ positions. High levels of lead (as in films deposited at 50 mTorr) results in an increase in the exhibited d-spacings which has caused the peaks to shift left. High levels of oxygen (as in films deposited at 20 mTorr) results in a decrease in d-spacing from expected values which has caused the peaks to shift right. This result corresponds well with the radius of the atoms, with lead having a larger atomic radius than oxygen.

Optimisation of Deposition Conditions for Functional Oxide Thin Films

(a)

(b)

325

(c)

Figure 3. Comparison of glancing angle X-ray diffraction results for PSZT thin films deposited under different sputtering pressures. Parts (a-c) of the image show significant 2θ positions for PSZT. (Reprinted with permission from Ref. [14].)

4.3. Surface Morphology Analysis The surface morphology of PSZT thin films deposited at different oxygen partial pressures has been analysed using atomic force microscopy (AFM). The average surface roughness (Ra) values and grain sizes measured are listed in Table 3. An AFM scan of a PSZT thin film deposited at 10 mTorr is shown in Fig. 4(a); these films had the least surface roughness and had uniformly sized grains. Figure 4(b) shows an AFM scan of a PSZT thin film deposited at 30 mTorr illustrating the presence of crystalline agglomerates. Films deposited at 20 and 50 mTorr were similar to this, also with crystalline agglomerates. Films deposited at 50 mTorr contain the maximum amount of lead, and are extremely rough. The observable effect that composition had on orientation is not apparent in the case of the AFM results, with no obvious trend being present in this case.

(a)

(b)

Figure 4. AFM contact mode 5 μm × 5 μm surface scans of PSZT thin films deposited at sputtering pressures of (a) 10 mTorr and (b) 30 mTorr. (Reprinted with permission from Ref. [14].)

326

Sharath Sriram and Madhu Bhaskaran

Table 3. Average surface roughness (Ra) and grain size of PSZT films deposited under different sputtering pressures. Sputtering pressure (mTorr)

O2 partial pressure (mTorr)

Ra (nm)

Grain size (nm)

10 20 30 40 50

1 2 3 4 5

10.93 44.82 37.04 19.80 43.93

170 170 140 210 140

4.4. Section Summary This section summarises the results obtained from the analyses of PSZT thin films deposited under different sputtering pressures. The composition, orientation, and surface roughness of thin films sputtered at oxygen partial pressures of 1, 2, 3, 4, and 5 mTorr have been characterised and compared. XPS and GA-XRD results complement each other, with observable effects due to either excess lead or oxygen. Increasing the sputtering pressure increased lead concentration due to diffusion-based transport of species (and this can be used to prevent lead loss in deposited films). High levels of lead resulted in an increase in the exhibited d-spacings and high levels of oxygen resulted in a decrease in d-spacing, from expected values. Surface morphology studies do not indicate an identifiable trend, but highlights the presence of agglomerates in some cases. The PSZT thin films deposited at a sputtering pressure of 10 mTorr (oxygen partial pressure 1 mTorr) have given the best results – desired composition (close to the desired ABO3) and orientation, with low surface roughness and uniformly sized grains.

5. IDENTIFICATION OF MODIFIED PSZT UNIT CELL Having identified suitable deposition conditions in terms of post-deposition cooling rate and sputtering pressure, the choice of temperature of deposition of the PSZT thin films was addressed. Substrate temperatures of 500 to 700 ºC are normally used for the deposition of thin films of PZT compounds, to attain desired perovskite structure [10,11]. In order to incorporate these piezoelectric thin films into standard devices, deposition on silicon substrates is desirable. Deposition of PZT has been conventionally carried out on platinumcoated silicon substrates [10,11]. Considering the inert properties of gold and that its lattice spacings are close to those of PSZT [18], deposition on gold-coated silicon substrates was also investigated. As the gold-silicon eutectic point is 363 ºC, deposition on gold coated substrates was limited to 300 ºC. Considering these preliminary choices, deposition at 300 ºC on gold and at 600-700 ºC on platinum was investigated. This section reports on the influence of multi-layered bottom electrodes (with gold or platinum) on the orientation of deposited PSZT thin films of the composition (Pb0.92Sr0.08)(Zr0.65Ti0.35)O3. For this composition, the films are expected to have a rhombohedral unit cell at room temperature (corresponding to space group R3c) [9]; and so,

Optimisation of Deposition Conditions for Functional Oxide Thin Films

327

the corresponding powder diffraction file was used as a reference [19]. Bragg-Brentano X-ray diffraction (XRD) was used to experimentally determine the d-spacings in the deposited PSZT thin films, in order to calculate the unit cell parameters. These results have been used to index X-ray diffractograms for PSZT thin films deposited on different multi-layer bottom electrode configurations, and to identify those that promote perovskite orientation. The XRD analysis carried out showed that a majority of the PSZT peaks were shifted from the expected 2θ positions [19]; X-ray photoelectron spectroscopy was used to confirm that composition of the films was very close to that of the reference used [14,19]. As the diffractometer was regularly calibrated using a quartz standard, errors introduced by the system were ruled out. The as-deposited metal films (gold and platinum) at room temperature were expected to be nanocrystalline; but the temperatures at which PSZT deposition was carried out resulted in grain growth with preferential orientation, manifested as strong (111) metal peaks. These (111) metal peaks were at the expected 2θ positions [18,20], again verifying that the PSZT peaks were shifted from the expected. From these results, unit cell parameters for PSZT deposited on gold and platinum were calculated, which indicated that two modified unit cells existed based on deposition temperature and substrate.

5.1. Modified PSZT Unit Cell on Gold at 300 ºC Using d-spacings obtained from XRD (for the 2θ range of 20º-60º) the unit cell parameters for the PSZT thin films deposited on gold were estimated. The results obtained are summarised in Table 4. In a similar manner to the above, the unit cell parameters for the PSZT thin films deposited on platinum were estimated. The results obtained are summarised in Table 5. These results show the equivalent d-spacings increase or decrease based on the dominating unit cell parameters for the chosen orientation. The resulting changes in the rhombohedral unit cell parameters a and c were –4.92 % and +9.49 %, respectively; this change is from the expected a of 5.732 Å and c of 14.317 Å to 5.450±0.010 Å and 15.675±0.015 Å, respectively. This corresponds to a decrease in the unit cell volume by about 1 %. These results show that the majority of the equivalent (i.e. calculated) d-spacings show an increase of about 5.26-5.31 %. The resulting change in both the rhombohedral unit cell parameters a and c is +5.29 %; from the expected a of 5.732 Å and c of 14.317 Å to 6.035±0.010 Å and 15.075±0.015 Å, respectively. This corresponds to an increase in the unit cell volume by about 16 %.

328

Sharath Sriram and Madhu Bhaskaran Table 4. Comparison between expected, experimental, and equivalent lattice spacings for PSZT on gold.

Orientation

(012) (004) (104) (110) (113) (006) (202) (024) (211) (116) (122) (018) (214) (300) (125) (034) (028)

Expected d-spacings based on ICDD PDF 04-002-5985 Å 4.0793 2.9033 2.8661 2.4570 2.3862 2.3451 2.0397 1.8604 1.8338 1.8150 1.6836 1.6618 1.6547 1.5694 -

Experimental d-spacing values from XRD Å 3.767 3.058 2.623 2.503 2.152 1.932 1.809 1.774 1.586 1.531

Equivalent d-spacings calculated from experimentally derived unit cell Å 4.295 3.769 3.057 3.018 2.587 2.513 2.469 2.147 1.959 1.931 1.911 1.773 1.750 1.742 1.652 1.581 1.528

Difference in equivalent d-spacings from expected values % +5.288 +5.294 +6.661 +5.291 +5.314 +5.283 +5.261 +5.300 +5.300 +5.289 +5.310 +5.308 +5.276 +5.263 -

Table 5. Comparison between expected, experimental, and equivalent lattice spacings for PSZT on platinum.

Orientation

(012) (104) (110) (113) (006) (202) (024) (211) (116) (122) (108) (214) (300) (125) (028)

Expected d-spacings based on ICDD PDF 04-002-5985 Å 4.0793 2.9033 2.8661 2.4570 2.3862 2.3451 2.0397 1.8604 1.8338 1.8150 1.6836 1.6618 1.6547 1.5694 -

Experimental d-spacing values from XRD Å 4.055 3.014 2.784 2.614 2.106 2.027 1.956 1.849 1.577 1.510

Equivalent d-spacings calculated from experimentally derived unit cell Å 4.043 3.015 2.725 2.416 2.613 2.260 2.022 1.772 1.886 1.739 1.810 1.624 1.573 1.550 1.508

Difference in equivalent d-spacings from expected values % –0.890 +3.847 –4.923 –1.669 +9.505 –3.346 –0.868 –4.752 +2.847 –4.187 +7.508 –2.275 –4.937 –1.236 -

Optimisation of Deposition Conditions for Functional Oxide Thin Films

329

5.2. Modified PSZT Unit Cell on Platinum at 650 ºC In order to overcome this undesirable reaction, a 200 nm silicon dioxide layer was introduced to isolate the metals from silicon. This increased the preferential (111) orientation, relative to the (200) peak, of the underlying gold layer [18] and also the relative intensity of the perovskite (104) PSZT orientation as shown in Fig. 5(b). Cross-sectional transmission electron microscopy studies have confirmed that this added silicon dioxide layer prevents the reaction between gold and silicon. Deposition duration did not influence the orientation of the PSZT thin films, with the ratio of relative intensities of the preferential orientations remaining unaffected. However, the extent of preferential orientation in the underlying gold layer increased, due to continued grain growth.

5.3. Orientation of PSZT Thin Films on Gold at 300 ºC PSZT thin films were deposited on silicon samples coated with 150 nm of gold with a 15 nm titanium adhesion layer. Figure 5(a) shows the resulting diffractogram for such films; here two perovskite PSZT peaks can be observed. Cross-sectional transmission electron microscopy (JEOL 2010F TEM) of these samples indicated that interdiffusion had occurred between the bottom electrode layers and silicon [21]. This could be attributed to the fact that the PSZT deposition on these samples was carried out at 300 ºC, which is close to the eutectic point of gold and silicon – the eutectic point could have been reached as the vapour pressure at the sputtering pressure would have been lower.

(a) Figure 5. (Continued).

330

Sharath Sriram and Madhu Bhaskaran

(b) Figure 5. X-ray diffractograms obtained for PSZT thin films deposited on silicon samples coated with 150 nm Au using a 15 nm Ti adhesion layer without and with an intermediate 200 nm SiO2 layer are shown in (a) and (b), respectively. (Reprinted with permission from Ref. [17].)

5.4. Orientation of PSZT Thin Films on Platinum at 650 ºC PSZT thin films were deposited on a number of platinum-based bottom electrode configurations on silicon substrates, as listed below: (i) 200 nm Pt / 20 nm Ti / Si (ii) 200 nm Pt / 20 nm Ti / 200 nm SiO2 / Si (iii) 300 nm Pt / 70 nm Ti / 200 nm SiO2 / Si (iv) 200 nm Pt / 20 nm TiO2 / Si PSZT thin films on platinum require heating to temperatures of 600 ºC and above to enhance the perovskite structure; hence, all depositions of PSZT on platinum were carried out at 650 ºC [10,11]. Films deposited on electrode configuration (i) were found to have high surface roughness due to the formation of platinum-titanium silicide at these temperatures (verified by Auger electron spectroscopy), but perovskite PSZT peaks were observed [Fig. 6(a)]. A silicon dioxide layer was introduced to inhibit this silicide formation [configurations (ii) and (iii)], along with experimenting with thicker platinum and titanium layers [configuration (iii)], in order to preserve the smooth platinum surface. PSZT thin films deposited on electrode configurations (ii) and (iii) were not usable, as the underlying layers delaminated due to stress imbalance between the various layers. This is attributed to the differences in coefficients of thermal expansion and ratio of thicknesses of the different layers [22]. Titanium dioxide layers (known to be stable and inert) were experimented with, and PSZT thin films deposited on electrode configuration (iv) were very smooth, as the reaction between platinum and titanium was prevented, and preferential perovskite PSZT orientation was obtained [Figure 6(b)].

Optimisation of Deposition Conditions for Functional Oxide Thin Films

331

(a)

(b) Figure 6. X-ray diffractograms obtained for PSZT thin films deposited on (a) 200 nm Pt on 20 nm Ti on Si and (b) 200 nm Pt on 20 nm TiO2 on Si. (Reprinted with permission from Ref. [17].)

Deposition of PSZT thin films on platinum was also carried out at 600 ºC and 700 ºC, and there were no significant differences in the preferential orientations with temperature. Deposition durations of two and four hours also resulted in almost identical diffractograms, with just the substrate peaks suppressed for samples from longer depositions.

5.5. Section Summary This section discusses the orientation dependence of PSZT thin films on bottom electrode architectures. XRD results highlighted variations in obtained 2θ peak positions from expected positions, indicating the existence of modified unit cells for PSZT thin films on gold and platinum. The modified rhombohedral unit cell parameters and the equivalent d-spacings are

332

Sharath Sriram and Madhu Bhaskaran

discussed. The variations in unit cell parameters from expected values could be attributed to guiding effects from the crystalline bottom electrode layers; and variations appear consistent with the larger lattice spacings for gold in comparison to platinum. The structure obtained on gold-coated silicon at 300 ºC is unique to such gold-coated substrates, apparently due to the guiding effect exhibited by (111) gold on (104) PSZT. No preferential orientation was observed on other materials (such as LiNbO3, Pt, etc.) at 300 ºC. Considering the second variable of deposition temperature, additional work was carried out with PSZT deposition on silicon substrates with thermal silicon dioxide.

6. DEPOSITION ON THERMAL SILICON DIOXIDE PSZT thin films were deposited either at room temperature followed by an annealing process or at a substrate temperature of 700 ºC. PSZT thin films deposited at room temperature were subject to post-deposition furnace annealing at 700 ºC for 3 hours in the presence of high purity argon. In the case of samples deposited at temperatures of 700 ºC, the samples were heated to deposition substrate temperature at a ramp rate of 10 ºC/min and cooled, subsequent to deposition, at 5 ºC/min; previous work has shown that these conditions improve the degree of perovskite orientation in the thin films, in line with previous results (Section 3 and Ref. [10]).

6.1. X-Ray Diffraction Analysis The X-ray diffractogram of PSZT thin films deposited at room temperature and subjected to post-deposition annealing is shown in Fig. 7(a), and consists of peaks at expected 2θ positions [17]. These peaks are characterised by weak reflections and broad peak widths; indicative of weak preferential orientation, limited grain growth, and a nanocrystalline structure. Figure 7(b) is representative of diffractograms obtained for PSZT thin films deposited at substrate temperature of 700 ºC. Expected perovskite peaks at 2θ of 29.6º and 49.3º, for (104) and (108) orientations, for a rhombohedral PSZT unit cell [17] were observed. Smaller peaks obtained at 34.3º and 58.5º correspond to (006) and (300) orientations, respectively. The temperature at which these thin films were deposited was chosen to encourage thermallydriven grain growth and the diffractogram [Fig. 7(b)] confirms that this resulted in uniform crystal growth, manifested as strong and sharp peaks in the diffractogram. These results, with strong c-axis preference, also promise strong columnar growth in the thin films (discussed in sub-section 6.3). These results indicate that though post deposition annealing encouraged grain growth, only thin film deposition at high temperatures (in situ substrate heating) results in strong preferential orientation. The PSZT thin film samples deposited at 700 ºC, with promising XRD results, were subject to further analyses.

Optimisation of Deposition Conditions for Functional Oxide Thin Films

333

(a)

(b) Figure 7. X-ray diffractograms obtained for PSZT thin films: (a) deposited at room temperature with subsequent furnace annealing at 700 ºC for 3 hours and (b) deposited at a substrate temperature of 700 ºC for 3 hours.

6.2. Atomic Force Microscopy Analysis AFM surface scan results for a PSZT thin film deposited at 700 ºC are shown in Fig. 8. The topography images in Figs. 8(a) and 8(b) depicts tightly packed nanocrystalline grains with an average grain size of 80-100 nm. The average surface roughness (Ra) of these films, measured to be about 9-11 nm, indicates that the pronounced grain structure is regular. The

334

Sharath Sriram and Madhu Bhaskaran

accentuated faceting of the grain faces is only conveyed by the peak roughness (Rp) values measured to be ~90 nm. The polygonal structure of the grains is apparent in the deflection image shown in Fig. 8(c).

(a)

(b)

(c)

Figure 8. Atomic force microscopy scan results for PSZT thin film deposited at 700 ºC: (a) threedimensional representation of film surface showing facetted tightly packed grains; (b) topography image obtained over an area of 1 μm × 1 μm; and (c) the deflection image corresponding to the topography in (b).

6.3. Electron Microscopy Analysis Cross-sectional scanning electron microscopy (SEM) showed that the films had nanocolumnar grains extending through the thickness of the PSZT thin films (Fig. 9). The structure and width of these grains matched those expected from XRD and AFM analyses. There are regions where the PSZT grains ends abruptly [especially in Fig. 9(b)]; this was due to the brittle nature of the ceramic PSZT thin films which prevented better sectioning.

Optimisation of Deposition Conditions for Functional Oxide Thin Films

335

(a)

(b) Figure 9. Cross-sectional scanning electron micrographs in (a) and (b) showing the regular nanocolumnar grains spanning the thickness of the film. (Observed at specimen tilt of 9º.)

Plan view TEM analysis results for PSZT thin films deposited on silicon dioxide are presented in Fig. 10. The nanocrystals observed in the images correspond to grain sizes of 80100 nm for the columnar preferentially oriented grains in the structure. A majority of the grains appear to be triangular in shape. AFM scans confirmed that the grains in the PSZT thin film are densely packed, and this can be observed in the top-left of Fig. 10(a). In the other region, the ion milling process for the TEM sample preparation has resulted in numerous regions devoid of material. This is a by-product of the very rough crystalline film surface, on which milling from the back results in the majority of the material being removed, with only the capping regions of the grains left behind.

336

Sharath Sriram and Madhu Bhaskaran

The thin region of the specimen gives us valuable information regarding the nanostructure of the thin films. Figure 10(b) shows that the grains vary in size from 80 to 100 nm, with well defined crystalline and polygonal structure. Strong Bragg diffraction from many grains in Fig. 10 indicates that they share the same orientation. These nanocrystals extend all the way through the thickness of the specimen forming the columnar structure observed in the cross-sectional analysis. Selected area electron diffraction of the plan view specimen showed that the nanocrystals exhibited the expected perovskite structure, but were randomly distributed – there was no preferential orientation along the substrate surface (XRD indicated preferential orientation perpendicular to the substrate).

(a)

(b) Figure 10. Plan view transmission electron microscopy results obtained for PSZT thin film deposited at 700 ºC at two different magnifications are shown in (a) and (b).

Optimisation of Deposition Conditions for Functional Oxide Thin Films

337

Figure 11. Schematic representation of the PSZT thin films deposited on thermal SiO2, showing the nanocolumnar polygonal grain structure (Not to scale.)

This section presents results from the first instance of deposition of preferentially oriented, nanocrystalline, and nanocolumnar PSZT thin films directly on thermal silicon dioxide. A schematic depicting the overall structure of the deposited films is shown in Fig. 11. No intermediate seed or activation layers were used between PSZT and SiO2. A substrate temperature of 700 ºC was found to be suitable for obtaining the desired perovskite structure. This relates to the summary of the previous section and helps arrive at the conclusion that at higher temperatures, thermal effort dominates the resulting microstructure in such complex oxide thin films.

7. CONCLUSION This chapter demonstrates a systematic approach in determining optimal deposition conditions for complex oxide thin films. A material with the ABO3 perovskite structure, with both A- and B-site dopants, was chosen for this study. This was strontium-doped lead zirconate titanate. The influence of oxygen partial pressure, bottom metal electrode choice, and substrate temperature on the composition and crystal structure of the resulting thin films was analysed. The potential variables identified in Table 1 have now been replaced with optimised parameters (Table 6). During the course of identification of optimal deposition conditions, the following key results have been obtained: A. Conditions to achieve desired PSZT thin film composition have been identified.

338

Sharath Sriram and Madhu Bhaskaran B. In studying the influence of oxygen partial pressure on thin film composition, conditions for deposition of controlled lead-rich films have been identified. This will serve as a technique to design film composition in order to overcome lead loss due to subsequent processing. C. Suitable bottom electrode architectures for device designs and piezoelectric response measurements have been identified. D. Two different modified unit cell structures of PSZT, based on the guiding effect of gold (at 300 ºC) or as a consequence of temperature (at 650 ºC), have been identified. This result demonstrates an approach for tailoring the crystal structure of the material. E. The first instance of deposition of perovskite piezoelectric thin films directly on thermal silicon dioxide (without using intermediary seed layers) is demonstrated [2325]. Table 6. Optimised PSZT thin film deposition conditions. Target Target diameter RF power Target to substrate distance Base pressure Sputtering pressure Process gas Substrate temperature Temperature ramp-up rate Temperature ramp-down rate a b

(Pb0.92Sr0.08)(Zr0.65Ti0.35)O3 100 mm 100 W 70 mm 9.0 x 10-6 Torr 1.0 x 10-2 Torr 10% oxygen in argon 300 ºCa or 600-700 ºCb 10 ºC/min 5 ºC/min

Suitable for deposition on gold Suitable for deposition on platinum

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9]

Voigt, W. Lehrbuch der Kristallphysik; B. G. Teubner: Leipzig and Berlin, 1910. Randall, CA; Kim, N; Kucera, JP; Cao, W; Shrout, TR. J. Am. Ceram. Soc., 1998, 81, 677-688. Yu, Y; Tu, J; Singh, RN. J. Am. Ceram. Soc., 2001, 84, 333-340. Tunaboylu, B; Harvey, P; Esener, SC. Integr. Ferroelectr. 1998, 19, 11-32. Yu, Y; Singh, RN. J. Appl. Phys., 2000, 88, 7249-7257. Zheng, H; Reaney, IM; Lee, WE; Jones, N; Thomas, H. J. Am. Ceram. Soc., 2002, 85, 207-212. Bedoya, C; Muller, Ch; Baudour, JL; Madigou, V; Anne, M; Roubin, M. Mater. Sci., Eng. B, 200, 75, 43-52. Araujo, EB; Eiras, JA. J. Phys. D: Appl. Phys., 2003, 36, 2010-2013. Jaffe, B; Cook, WR; Jaffe, H. Piezoelectric Ceramics; Academic: New York, 1971, p 136.

Optimisation of Deposition Conditions for Functional Oxide Thin Films

339

[10] Sriram, S; Bhaskaran, M; Holland, AS. Semicond. Sci. Technol., 2006, 21, 1236-1243. [11] Wasa, K; Kitabatake, M; Adachi, H. Thin Film Materials Technology: Sputtering of Compound Materials; Springer-Verlag: Heidelberg, 2004. [12] Miclea, C; Tanasoiu, C; Miclea, CF; Amarande, L; Gheorghiu, A; Spanulescu, I; Plavitu, C; Miclea, CT.; Cioangher, MC; Trupina, L; Iuga, A. J. Eur. Ceram. Soc., 2007, 27, 4055-4059. [13] Garg, A; Agrawal, DC. Mater. Sci. Eng. B, 1999, 56, 46-50. [14] Sriram, S; Bhaskaran, M; Du Plessis, J; Short, KT; Sivan, VP; Holland, AS. Micron 2009, 40, 104-108. [15] Rossnegal, SM; Yang, I; Cuomo, JJ. Thin Solid Films, 1991, 199, 59-69. [16] Castaldi, L; Gibbs, MRJ; Davies, HA. J. Appl. Phys., 2003, 93, 9165-9169. [17] Bhaskaran, M; Sriram, S; Short, KT; Mitchell, DR. G; Holland, AS. Thin Solid Films 2008, 516, 8101-8105. [18] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 04-0784. [19] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 04-002-5985. [20] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD, formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA 19073, Card 04-0802. [21] Sriram, S; Bhaskaran, M; Mitchell, DRG; Short, KT; Holland, AS; Mitchell, A. Microscopy Microanal, 2009, 15, 30-35. [22] Lu, DX; Pun, EYB; Wong, EMW; Chung, PS; Lee, ZY. IEEE T. Ultrason. Ferr. Freq. Contr., 1997, 44, 675-680. [23] Ramesh, R. International Patent WO/1994/013471. Available online at http://www.wipo.int/pctdb/en/wo.jsp?IA=US1993010387 [24] Zhao, J; Lu, L; Thompson, CV; Lu, Y; Song, WD. Proc. SPIE, 2002, 4426, 221-224. [25] Sriram, S; Bhaskaran, M; Mitchell, A; Mitchell, DRG; Kostovski, G. Nanoscale Res. Lett., 2009, 4, 29-33.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 341-352

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 11

OPTICAL WAVEGUIDES PRODUCED BY ION IMPLANTATION IN OXIDE GLASSES Feng Chen∗, Xue-Lin Wang, Lei Wang and Ke-Ming Wang School of Physics, Shandong University, Jinan, 250100, People’s Republic of China

ABSTRACT Optical waveguides can restrict light propagation in very small size of order of several microns, reaching high optical intensities even at low pumps; consequently, some properties of the bulk materials may be considerably improved in waveguide structures, such as non-linear responses, laser actions and optical signal amplifications. As one mature technique for material property modifications, ion implantation has been applied to construct optical waveguides in many optical materials, including insulating crystals and glasses, semiconductors and organic materials. Oxide glasses receive much attention for various photonic and telecommunication applications for its competitive costs and excellent optical features. In this chapter, we reviewed the research results obtained for optical waveguides in oxide glasses produced by ion implantation techniques, by giving a brief introduction of basic fabrication principles and methods and a summary of interesting results obtained in this topic. The prospects of possible practical applications of ion-implanted oxide glass waveguides in photonics are also demonstrated.

Keywords: Optical waveguides; ion implantation; glasses.

I. INTRODUCTION An optical waveguide structure is defined as a region with higher refractive index surrounded by lower-index mediums, which allows the light confinement within its boundaries by total internal reflection. The restricted inside dimensions of the light are

∗

Corresponding author. Tel.: +86-531-88364655; fax: +86-531-88565167; E-mail: [email protected].

342

Feng Chen, Xue-Lin Wang, Lei Wang et al.

typically with order of several microns, which is somehow comparable with the light wavelength [1,2]. Planar or slab waveguides restrict light propagation in one dimension (1D waveguide), whilst channel or ridge waveguides confine the light in two dimensions (2D waveguide). The small-size cross sections of waveguide structures offer high light intensities produced by even very low powers; consequently, the nonlinearities or laser actions in waveguides may be considerably improved with respect to those in bulk materials [3-6]. In addition, the components based on waveguides usually have very small sizes, which make the related optical chips very compact. As a result, they may be connected together with optical fibers quite simply because the dimensions of waveguides are directly compatible with fibers used in telecommunications, in which waveguides are often used as amplifiers, signal switches, modulators, etc [7]. Another advantage of waveguide technology is the low costs of the guiding devices, which makes components with diverse functions easily introduced into the market for practical applications. In principle, many optical materials can be used as the substrates for optical waveguide formation, such as insulating crystals, glasses, semiconductors, and organic materials. Glasses, with their many attractive features, e.g. large wafers, low costs, diverse compositions, and various optical properties with excellent performances, have been extensively used for production of passive as well as active waveguide devices [8-10]. Among this large glass family, the group that has been mostly investigated for guiding structures so far is oxide glass, which receives most extensive applications in integrated photonics and modern telecommunications. By certain chemical methods, such as ion exchange [11,12], optical waveguides in both 1D (planar) and 2D (channel) configurations can be constructed. Ion implantation is a well-known technique for modifications of chemical, physical and optical properties of material surface [13]. In 1968, this technique was applied to form waveguide structures in fused silica, which was the first successful attempt for ion implantation on the waveguide fabrication [14]. Since the ion implantation is an unequilibrium process, which does not depended on the chemical properties of the substrates, in principle, this technique can be applicable to most optical materials in solid states, which is one of the most advantageous characteristics [13,15]. Up to now, waveguides have been so far fabricated in more than 100 optical materials (including insulating crystals, glasses, semiconductors and organic polymers) by implantation of various ions at the energies of several hundred kilo-electron-volt (keV) up to several mega-electron-volt (MeV) or even higher [15]. In addition, the ion technique offers precise determination of the refractive index in selected regions with considerable modulations. One can easily control the RI profiles of the substrates by adjusting the implantation parameters (species, energies, fluences of the incident ions) and geometries, and further optimize waveguide properties by postimplantation treatments (e.g. conventional or rapid thermal annealing, pulsed laser annealing, etc.) [13,15]. Ion implantation has been used to produce waveguides in several glass families with diverse compositions. In this chapter, we will give a review of the ion implanted optical waveguides in glasses. In details, in Section II, a brief introduction of fundamental models of ion implanted glass waveguides will be given; Section III summarizes the fabrication methods for both 1D and 2D waveguides in glasses; research results of the related topics that have been obtained so far for several glass families are overviewed in Section IV; finally, possible applications such waveguides are raised in Section V.

Optical Waveguides Produced by Ion Implantation in Oxide Glasses

343

II. FUNDAMENTAL MODELS There is no a unique explanation on ion-implanted waveguide formation for all materials, because of the diversity of configuration and properties. However, there is one simple model, i.e. so-called “barrier”-type model, which was firstly raised by Townsend in 1976 as a prediction, which seems to be a reasonable demonstration [13,15,16]. When implantation of light ions at energies of several MeV are performed, a majority of the damage occurs at the end of the ion track inside the substrates through nuclear collisions, which results in a decrease of physical density by means of volume expansion and hence a reduced refractive index of an optical barrier. Such a barrier confines the light in a narrow layer with relatively high refractive index between itself and the sample surface, forming an optical waveguide. This model is applicable to some glass wafers. Figure 1 (a) shows the typical refractive index profile of the barrier-confined waveguide, which is produced by 2.8 MeV He+ ions into a phosphate glass at dose of 1×1016 ions/cm2.

Figure 1. Refractive index profile of (a) barrier-confined waveguide in phosphate glass and (b) “well + barrier” restricted waveguide in silicate glass.

344

Feng Chen, Xue-Lin Wang, Lei Wang et al.

In some cases, the ion implantation may induce positive changes of refractive index in the surface regions of the substrate, forming enhanced well that confines the light propagation inside the waveguide in a non-leaky manner (referring to “enhanced well” model). The physical mechanism may be due to the compressing effect in the implanted regions, where the physical density may increase because of the volume reduction. In most cases, the barrier may also happen at the end of the ion track, which bring out a “well + barrier” model. Figure 1 (b) shows the refractive index profile of a proton implanted silicate glass waveguide. Note that the different refractive index alternations may happen for glasses wafers belonging to one same glass family, since the glasses are amorphous solids with no fixed compositions.

III. FABRICATION METHODS Ion implantation is a mature technology for semiconductor production, and has been widely used in the optical communication devices. In the research area of the ion-implanted waveguides, accelerators are more often used because they offer high energies of specific implanted ions at acceptable doses. The implanted ions, normally with positive charges, are extracted out from the sources, experiencing acceleration and mass/energy selection, and bombarded into the target materials by beam scanning technique, by which uniform irradiation is ensured over the sample surface. This method has been adopted for waveguide fabrication in a large number of optical materials. An improved method, a so-called focused ion beam (FIB) implantation, could provide an ion beam with diameters ranging from several microns to 100 nm or even much less, which provides directly selective implantation to regions with small dimensions, forming directly-writing waveguides [17,18]. The planar waveguides can be produced in glass wafers by means of simple ion implantation processing. The waveguide may be confined by a buried damage layer (optical barrier), or an enhanced refractive index well, or both of them together. One may select suitable implantation parameters to control the refractive index alternations of the glasses, which in principle offer possibility of designable guide devices.

Figure 2. Schematic plots of ridged waveguide formation: (a) planar waveguide layer construction by single ion implantation; and (b) selected milling of planar waveguide by ion beam etching method.

Optical Waveguides Produced by Ion Implantation in Oxide Glasses

345

Figure 3. Selected ion implantation for channel waveguide formation with a stripe mask.

Figure 4. Focused ion beam writing for channel waveguide formation.

For 2D waveguide formation in glass by means of ion implantation, several solutions are available. If the waveguide is confined by optical barrier, one may form ridge waveguide structures by milling selected parts of the planar waveguide substrate, see Figure 2 (a) and (b). If the waveguide is confined by enhanced refractive index well, single implantation may be efficient with a stripe mask, as shown in Figure 3. Alternatively, one may create highrefractive-index stripes by using focused ion beam writing, which avoids the lithographic masking (Figure 4). All these techniques are applied to fabricate waveguides in oxide glasses. Since the implantation creates color centers (optical absorption sites) by electronic energy deposition, and destroys somehow the equilibrium of the origin structures by nuclear collisions, which both inevitably increase the waveguide losses, therefore annealing treatment is commonly necessary for all the ion-implanted waveguides in glasses as well crystals. Such anneals are often useful for considerable improvement of waveguide qualities [13,15].

346

Feng Chen, Xue-Lin Wang, Lei Wang et al.

IV. MATERIALS In this section, the development of ion-implanted waveguides in oxide glasses will be summarized. The progress of non-oxide glass waveguides produced by ion implantation is also mentioned briefly.

1) Fused Silica Fused silica is a high purity synthetic amorphous silicon dioxide. The excellent properties, such as near zero thermal expansion, good chemical inertness, low dielectric constant and losses, make fused silica extensively used in the optical fields. In addition, since its refractive index is very close to that of the commercial fibers, fused silica is also very attractive for telecommunication applications. Optical waveguides in this material could be fabricated by means of directly femtosecond fiber laser with high repetition rate [19]. The earliest waveguide in fused silica was reported by Schineller et al., who used proton implantation [14], and later work was performed on the investigation of He+, N+, O+, Si+, P+, Ti+, or Ge+ ions [20-28]. The refractive index of the fused silica increases due to the compaction under the influence of the nuclear and electronic damage, forming indexenhanced wells. This non-tunneling index potential often means fine confinement of the light propagation within the waveguides and, as a result, low propagation losses. Johnson et al. investigated the compaction effects of Si+ ion implanted fused silica waveguides, where a minimum loss coefficient of 0.15 dB/cm was measured for the samples with postimplantation annealing at 500ºC [22]. In addition, Leech et al. studied the formation and the properties of the MeV Si, P or Ge ion implanted channel waveguides in fused silica [24]. The propagation losses of the annealed channels were only of 0.1-0.2 dB/cm at the telecommunication wavelength of 1.3 and 1.5 μm. More recently, Wang et al. reported on planar waveguides in fused silica by 3.0 MeV O+ ion implantation at the dose of 1×1015 ions/cm2, and attenuation coefficient of 0.14 dB/cm was obtained for the thermal treated samples [23]. In addition, Drouard et al. performed Ti ion implantation into bulk fused silica, and both planar and channel waveguides were fabricated. The mode fields of the waveguides may match those of the fibers by adjusting the doses of Ti ions [20]. It is also possible to form buried waveguides in fused silica because of the index increase induced by the compaction effects of the implantation. Von Bibra et al. reported on buried waveguides with graded depths fabricated by a focused 3.0 MeV proton ion beam writing in fused silica through a tapered film varying in thickness of 5-40 μm [26]. The technique also provides possibility of the formation of photonic lattices in fused silica. Moreover, Chen et al. studied second order nonlinearity in a 5 MeV Ge2+-ion implanted waveguide after thermal poling treatment [27]. The results show that, with the periodical UV erasure of the second order nonlinearity, the quasi-phase-matched SHG from 1064 to 532 nm in the silica waveguide arrays could be realized. Recently Chao et al. reported the channel waveguide formation in similar silica hosts by Ge2+ ion implantation combined with lithographic masking [28]. All of the above results show that ion implantation is a practical method for effective waveguide formation in fused silica comparable with other techniques.

Optical Waveguides Produced by Ion Implantation in Oxide Glasses

347

2) Silicate Glasses Silicate glasses are excellent candidates for waveguide amplifiers and lasers because of their high solubility for rare-earth ions, such as Nd3+, Er3+, and Yb3+, as well as the chemical robustness. By the exchange of K+/Na+ or Ag+/Na+ ions, planar and channel waveguides could be fabricated in silicate wafers [29,30]. Ion implantation is also effective for the modulation of refractive index of silicate glass. However, the silicate is amorphous, particularly with no fixed composition; different silicate may have different ratio for the same composition. Earlier work on the ion-implanted waveguides in silicate glasses was done by Kakarantzas et al. in 1992 [31]. From their results, standard optical barrier-type waveguides were formed by He+ ion implantation at the doses of ~1016 ions/cm2. A considerable large index decrease happened in the surface region, which degraded the waveguide quality [32]. Because silicate is a mixture, refractive index behaviors may be different from other silicate glasses. We reported on optical waveguide formation in one silicate glass, which exhibited index increase in the surface region, resulting in non-tunneling waveguide structures [33-35]. The results show that, for light ion-implanted silicate waveguides, a refractive-index decreased barrier exists at the end of ion range, whilst in heavier-ion implanted case, no such barrier is built up. Waveguides with large index decrease exhibit high propagation losses. Even for annealed samples, the loss could be ~15dB/cm, which is too high to be acceptable for practical applications. However, low loss may be obtained for silicate waveguides confined by indexincreased wells due to the non-leaky properties. We reported on a single-mode Nd-doped silicate waveguide, which was formed by 3.0 MeV Si+-ion implantation at low dose of ~1014 ions/cm2, with propagation loss of only ~0.6 dB/cm after annealing treatment at 218ºC for 60 min [36]. For another silicate waveguide fabricated by 6 MeV carbon ion implantation, the losses of the first two modes were 0.42 and 0.91 dB/cm after moderate annealing treatments, respectively [35]. Low propagation attenuation suggests the potential usage in related fields. In addition, more recently, Wang et al. reported the first ion implanted channel waveguides in a Nd doped silicate glass by 6 MeV O3+ at dose of 1×1015 ions/cm2 [37]. Figure 5 shows the 3D near-field intensity distribution of the quasi-TE modes of channel waveguides.

Figure 5. 3D near field intensity distribution of the quasi-TE modes of the channel waveguide by O ion implanted into Nd-doped silicate glass: from the left to right, TE00, TE10, and TE20.

3) Phosphate Glasses Er3+/Yb3+ co-doped phosphate glasses are extensively used in the field of telecommunications to form erbium-doped waveguide amplifiers (EDWA). The most

348

Feng Chen, Xue-Lin Wang, Lei Wang et al.

common method for waveguide formation in these glasses is ion exchange. Alternatively, implantation of light as well as heavy ions could be used to fabricate waveguides in phosphate glasses [38-40]. The earlier work from Kakarantzas et al. showed that similar refractive index behaviors were observed for the He+ ion implanted waveguides in their phosphate and silicate glass samples [31]. Recently we studied the waveguide properties in an Er3+/Yb3+ co-doped phosphate glass formed by means of He or Si ion implantation [33]. It was found that the refractive indices in the surface and the barrier regions of the waveguide were reduced by 0.012 and 0.021 after the He ion implantation at the dose of 5×1015 ions/cm2, while in the Si implanted case (at the dose of 9×1014 ions/cm2), these index decrease in the two regions were 0.018 and 0.029, respectively. Such waveguides usually have high losses (~5dB/cm for the He- and ~20 dB/cm for the Si-implanted one) due to the relatively poor confinement of the light inside the guide layers. Tan et al. reported ridge waveguide formation in the He-implanted phosphate glass, which shows a designable configuration of waveguide devices [38]. Figure 6 shows the comparison of the experimental near-field mode intensity distribution with the calculated modal profile. Sum et al. investigated the channel waveguides by focused proton beam writing technique [39]. In their work, 2.0 MeV proton beams at the doses of 0.4-1.0×1015 ions/cm2 with spot size of 1.0 μm were used to write channel waveguides in the phosphate wafers. The atomic force microscopy (AFM) investigations of the waveguides revealed that positive rather than negative changes of the refractive index occurred near the end of the ion range, which means the different mechanisms between the formation of low and high dose light ion implanted waveguides [40]. The propagation losses of the waveguides could be reduced from 3.2 to 0.8 dB/cm after the annealing at 220 ºC.

Figure 6. Comparison of (a) the calculated modal profile and (b) the experimental near-field mode intensity distribution in a He-implanted Er/Yb co-doped phosphate glass waveguide.

4) Tellurite Glasses Tellurite glass is proposed as a host for broadband erbium-doped fiber amplifiers because of their excellent optical and chemical properties. Berneschi et al. reported the channel waveguide formation in an Er3+-doped tellurite glass by 1.5 MeV Ni+ ion implantation at doses of 0.5~4×1016 ions/cm2, with a 75 μm-thick silicon stripe mask [41]. The Ni+ irradiation

Optical Waveguides Produced by Ion Implantation in Oxide Glasses

349

created positive RI changes; hence the guiding structures did not require barrier confinement of the guided light.

5) Germanate Glasses Germanate glass is another promising candidate for integrated photonic applications. Kakarantzas et al. studied the planar waveguide structures in lead germanate glass, in which low-temperature (at 77K) He ion implantation at energy of 2.5 MeV and doses of 4-5×1016 ions/cm2 was performed to fabricate the waveguides [32]. The surface refractive index increased 0.34%, forming an enhanced well, and the barrier was with 0.12% index lower than the substrate, constructing a “well plus barrier” refractive index profile. Because of the nonleaky mechanisms, the waveguide has a low propagation loss of 0.15-0.3 dB/cm after annealing at 200-300 °C in air. In addition, the laser action at wavelength of ~1.9μm was realized in a Tm3+-doped lead germanate glass waveguide formed by 2.9 MeV He ion implantation at dose of 4×1016 ions/cm2 [42].

6) Other Glasses Other works were performed in the topics of ion implantation into some other oxide glasses, such as soda-lime glasses [43], filter glasses [44], etc. Barrier-confined waveguides were obtained for those glasses after the implantations.

V. APPLICATIONS There are several basic applications for ion-implanted oxide glass waveguides. These waveguides can be either used as passive devices or active components, according to the diverse properties of the glass wafers. For all glass waveguides, it may be used as passive devices, which can be designed to guide light propagation at diverse wavelengths. Drouard et al. used Ti ion implantation to fabricate planar lightwave circuit in bulk silica glass [19]. The results show that the refractive index distributions can be tailored by the doses of the Ti ions. This was helpful to fit the guided mode of the standard single-mode fibers, or to allow a sharp radius of curvature of bent waveguides. The waveguide devices exhibit low propagation attenuations of 0.1-0.8 dB/cm at wavelength of 1.5 μm, which can be used for practical applications in telecommunication windows. Moreover, since the refractive index profiles of such waveguide devices can be adjustable, more flexibility is expected for components with more functions. In fact, other oxide glass can also be used as passive guide-wave devices, although there has not been too much work presented in this topic so far. The rear-earth ion doped glass waveguides can also serve as active devices, such as waveguide laser modules and waveguide amplifiers. Shepherd et al. reported the first planar waveguide laser in the 2μm region, based on a He+-implanted Tm3+:lead germanate glass [40]. The energy and the dose of the He+ ions are 2.9 MeV and 4×1016 ions/cm2. Large index

350

Feng Chen, Xue-Lin Wang, Lei Wang et al.

increase of ~0.2 happened at the most path of the incident ions, which constructed an enhanced well to confine the light. After the annealing treatment at 200ºC, the fluorescence of the waveguides was quite similar to that of the original bulk materials. With suitable pump, the threshold was expected to be as low as 10 mW. Although there were very limited literatures reported on the laser actions in ion-implanted glass waveguides, in principle many rear-earth ion doped glasses could be the waveguide hosts for this work. Erbium doped waveguide amplifiers (EDWA) receive extensively applications at telecommunication window at ~1.5 μm. By using focused proton writing, EDWA can be produced in Er3+-doped glasses, for example, Er3+/Yb3+ co-doped phosphate glass. Liu et al. reported buried channel waveguides in Er3+/Yb3+ co-doped phosphate glass, which serve as EDWA [38]. The proton doses were of 1014-1015 ions/cm2, and the energy was 2 MeV. The maximum net gain of the waveguide amplifiers at 1.534 µm wavelength was measured to be ~1.72 dB/cm with 100 mW pump power at 975 nm. Another promising application for ion-implanted waveguides in glasses is for frequency doubling (second harmonic generation, SHG). Chen and Chao et al. reported first order SHG from 1064 nm to 532 nm in planar and channel waveguides produced by Ge ion implantation into fused silica, based on a quasi-phase-matched (QPM) configuration [26]. For channel waveguide, the second-harmonic peak power of ~1.8 μW was obtained with fundamental peak pumping power of ~780 mW, which meant SH conversion efficiency was 6.1× 10-4 %/W·cm2.

CONCLUSION We reviewed the progress of ion-implanted waveguides in oxide glasses, by giving basic fabrication methods and experimental data obtained for several kinds of glasses. Applications as both passive and active devices for these waveguides were also mentioned briefly. As it was summarized, ion implantation is very promising for fabricating a number of attractive optical waveguide devices in hosts of various oxide glasses. Future work may focus on the designable waveguide structures and realization of devices with more functions.

ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of Program for New Centrary Excellent Talent in University, China (under grant No. NCET-08-0331).

REFERENCES [1] [2] [3]

Lifante, G. (2003.). Integrated Photonics: Fundamentals; John Wiley and Sons Ltd: Atrium, England, Hunsperger, R. G. (2002.). Integrated Optics: Theory and Technology, SpringerVerlag, Berlin, Kip, D. (1996). Appl. Phys. B, vol 67, 131-150.

Optical Waveguides Produced by Ion Implantation in Oxide Glasses [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31]

351

Stegeman G. I. & Seaton, C. T. (1985). J. Appl. Phys., vol 58, R57-R78. Mackenzie, J. I. (2007). IEEE J. Sel. Top. Quantum Electron., vol 13, 626-637. Murphy, E. J. (1999). Integrated Optical Circuits and Components: Design and Applications, Marcel Dekker Inc. New York. Polman, A. (1997) J. Appl. Phys., vol 82, 1-39. Yan, Y. C., Faber, A. J., De Waal, H., Kik, P. G. & A Polman, A. (1997). Appl. Phys. Lett., vol 71, 2922-2924. Takahashi, H., Suzuki, S., Kato, K. & Nishi, I. (1990). Electron. Lett., vol 26, 8788. Ramaswamy, R. V. & Srivastava R. (1988). J. Lightwave Technol., Vol 6, 984-1000. Silva, B., Melo, R. P., Falc˜o-Filho, E. L. & De Araújo, C. B. (1997). Appl. Opt., vol. 36, 5949-5950. Florea, C. & Winick, K. A. (1999). J. Lightwave Technol., vol. 17, 1593-1601. Townsend, P. D., Chandler, P. J. & Zhang, L. (1994). Optical Effects of Ion Implantation, Cambridge University, Cambridge. Schineller, E. R., Flam, R. P. & Wilmot, D. W. (1968). J. Opt. Soc. Am. vol. 58, 1171-1176. Chen, F., Wang, X. L. & Wang, K. M. (2007). Opt. Mater., Vol. 29, 1523-1542. Townsend, P. D. (1976). Ion Implantation in Optical Materials, Institute of Physics Conference Series Vol. 28 Chapter 3, 104. Fu, Y. & Ngoi, B. K. A. (2005). Opt. Eng., vol. 44, 128002. Jarupoonphol, W., Murakami, K., Sakata, K., Takai, M., Hosono, A. & Okuda, S. (2003). J. Vac. Sci. Technol. B, vol. 21, 1598-1601. Shah, L., Arai, A., Eaton, S. & Herman, P. (2005). Opt. Express, 13, 1999-2006. Drouard, E., Escoubas, L., Flory, F., Tisserand, S. & Roux, L. (2004). J. Lightwave Technol., vol. 22, 2310-2315. Webb, A. P., Houghton, A. J. & Townsend, P. D. (1976). Rad. Effects, vol. 30, 177. Johnson, C. M., Ridgway, M. C. & Leech, P. W. (1996). Appl. Phys. Lett., vol. 67, 984986. Wang, X. L., Wang, K. M., Fu, G., Li, S. L., Shen, D. Y., Ma, H. J. & Nie, R. (2004). Opt. Express, vol. 12, 4675-4680. Leech, P. W., Ridgway, M. & Faith, M. (1996). IEE Proc.-Optoelectron., vol. 143, 281-286. Rao, E. V. K. & Moutonnet, D. (1975). J. Appl. Phys., vol. 46, 955-957. Von Bibra, M. L., Roberts, A. & Dods, S. D. (2000). Nucl. Instr. Meth. B vol. 168, 47-52. Chen, H. Y., Lin, C. L., Yang, Y. H., Chao, S., Niu, H. & Shih, C. T. (2005). Appl. Phys. Lett., vol. 86, 081107. Chao, S., Chen, H. Y., Yang, Y. H., Wang, Z. W., Shih, C. T. & Niu, H. (2005). Opt. Express, vol. 13, 7190-7196. Peters, P. M., Funk, D. S., Peskin, A. P., Veasey, D. L., Sanford, N. A., Houde-Walter, S. N. & Hayden, J. S. (1999). Appl. Opt., vol. 38, 6879-6886. Ondráček, F., Salavcová, L., Míka, M., Lahodný, F., Slavík, R., Špirková J. & Čtyroký, J. (2007). Opt. Mater., vol. 30, 457-461. Kakarantzas, G., Zhang, L., & Townsend, P. D. (1992). E-MRS Symp. Proc., vol. 29, 97.

352

Feng Chen, Xue-Lin Wang, Lei Wang et al.

[32] Kakarantzas, G., Townsend, P. D. & Wang, J. (1993). Electron. Lett., vol. 29, 489-490. [33] Chen, F., Wang, X. L., Li, X. S., Hu, L. L., Lu, Q. M., Wang, K. M., Shi, B. R., & Shen, D. Y. (2002). Appl. Surf. Sci., vol. 193, 92-101. [34] Li, S. L., Wang, K. M., Chen, F., Wang, X. L., Fu, G., Lu, Q. M., Hu, L. L., Shen, D. Y., Ma, H. J., Nie R. (2005). Surf .Coat. Technol., vol. 200, 598-601. [35] Li, S. L., Wang, K. M., Chen, F., Wang, X. L., Fu, G., Lu, Q. M., Hu, L. L., Shen, D. Y. & Ma, H. J. (2005). J. Phys. D: Appl. Phys., vol. 38, 2899-2903. [36] Chen, F., Wang, K. M., Wang, X. L., Li, X. S., Lu, Q. M., Shen, D. Y., & Nie, R. (2002). J. Appl. Phys., vol. 92, 2959-2961. [37] Wang, L., Chen, F., Wang, X. L., Wang, K. M., Jiao, Y., Wang, L. L., Li, X. S., Lu, Q. M., Ma, H. J. & Nie, R. (2007). J. Appl. Phys., vol. 101, 053112. [38] Tan, Y., Chen, F., Hu, L. L., Xing, P. F., Chen, Y. X., Wang, X. L. & Wang, K. M. (2007). J. Phys. D, vol.40, 6545-6548. [39] Sum, T. C., Bettiol, A. A., Liu, K., Ren, M. Q., Pun, E. Y. B., Rao, S. V., Van Kan, J. A., & Watt, F. (2005). Nucl. Instr. Meth. B, vol. 231, 394-399. [40] Liu, K., Pun, E. Y. B., Sum, T. C., Bettiol, A. A., Van Kan, J. A. & Watt, F. (2004). Appl. Phys. Lett., vol. 84, 684-686. [41] Berneschi, S., Conti, G. N., Bányász, I., Watterich, A., Khanh, N. Q., Fried, M., Pászti, F., Brenci, M., Pelli, S. & Righini, G. C. (2007). Appl. Phys. Lett., vol. 90, 121136. [42] Shepherd, D. P., Brinck, D., Wang, J., Tropper, A. C., Hanna, D. C., Kakarantzas, G. & Townsend, P. D. (1994). Opt. Lett., vol. 19, 954-956. [43] Skelland, N. D. & Townsend, P. D. (1994). J. Non-Cryst. Solids, vol. 188, 243-253. [44] Okur, I. & Townsend, P. D. (1997). Nucl. Instr. Meth. B, vol. 124, 76-80.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 353-363

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 12

THIN FILM PIEZOELECTRIC RESPONSE COEFFICIENT ESTIMATION TECHNIQUES Sharath Sriram*, Madhu Bhaskaran and Arnan Mitchell Microelectronics and Materials Technology Centre, School of Electrical and Computer Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia.

ABSTRACT The response of piezoelectric materials is quantified using charge and voltage coefficients. One such coefficient is d33, which numerically describes the resulting effect for an applied cause along the same direction. There is a lack of established techniques for quantitative estimation of d33 for piezoelectric thin films. This initiated an investigation into the development of such techniques, as a consequence of which two new techniques for piezoelectric coefficient estimation, under the inverse piezoelectric effect, have been developed. One technique capitalises on the measurement accuracy of the nanoindenter in following thin film displacement, while the other uses a standard atomic force microscope in contact imaging mode to estimate d33. Both techniques were developed by rigorously testing them against standard materials and avoiding commonly reported sources of error. Full details on the development, scope, and limitations of both techniques are presented in this chapter.

1. INTRODUCTION The piezoelectric effect is exhibited by the category of materials in which charge is generated when the materials are under stress. The inverse or converse effect also exists; whereby, the materials undergo dimensional deformation when exposed to an electric field. The piezoelectric response is quantified using charge and voltage coefficients. This enables comparison between the piezoelectric behaviour of different piezoelectric materials. One such

*

Corresponding author: [email protected], [email protected]

354

Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

coefficient is d33, which numerically describes the resulting effect for an applied cause along the same direction. The subscript ‘3’ nominally corresponds to ‘z’ in the standard axis system. Under the direct piezoelectric effect, when a force is applied along direction ‘3’ (equivalent to stress experienced by this direction), a quantity of charge results along this direction. In this case, d33 has the units of pC/N. Under the inverse piezoelectric effect, when a voltage (or an electric field) is applied along direction ‘3’, the dimension of the material along this direction changes. In this case, d33 has the units of pm/V. A schematic representation of the direct and inverse effect, with respective d33 computations, is shown in Figure 1. Though d33 is described under these two effects, and with two different units, the numerical value in both cases has been shown to be the same based on Maxwell’s thermodynamic principles [1]. This law breaks down in the case of thin films, due to the substrate clamping effect on such two-dimensional structures [2], where the length and width dimensions are significantly greater than the thickness.

(a)

(b)

Figure 1. Two-dimensional schematic of (a) the direct and (b) the inverse piezoelectric effects. The corresponding expressions used to calculate d33 are shown.

When applying existing d33 measurement techniques to thin films, numerous challenges were faced, considering the small variations (in charge or displacement) expected from the thin films. The most commonly reported approach, with an ever increasing number of publications each year, is piezoresponse force microscopy (PFM) [3]. PFM uses atomic force microscopes (AFM) with additional options: to apply an electric voltage through the tip, which is generally metal-coated to make it conductive; to access signals from the fourquadrant laser detector; electronics to process these signals; and software configuration to map these results as images. PFM excels in relative mapping to study piezoelectric response variations, but is only semi-quantitative in d33 measurements. Approaches for quantitative measurement of d33 have been demonstrated using nanoindenters by Koval et al. [4], under the direct piezoelectric effect. Using strontium-doped lead zirconate titanate (PSZT) and reference thin film materials – amorphous silicon dioxide (non piezoelectric) and aluminium nitride (standard piezoelectric) – two new and relatively simple techniques for piezoelectric response estimation have been developed. The techniques aimed to enable piezoelectric response estimation using standard equipment, such as the nanoindenter and atomic force microscope, without the need for extensive modifications or additional options.

Thin Film Piezoelectric Response Coefficient Estimation Techniques

355

The capabilities of the techniques described in this chapter have been demonstrated using PSZT thin films deposited on gold-coated silicon substrates. PSZT is a doped variant of the popular piezoelectric lead zirconate titanate (PZT). Deposition details for the PSZT thin films can be found in [5-7].

2. SAMPLE PREPARATION FOR d33 MEASUREMENTS The need to use standard equipment necessitates a couple of additional steps in preparation of samples for d33 measurements. To estimate the piezoelectric response of the deposited thin films in terms of d33, samples were fabricated to form a sandwich electrode structure. The following sample preparation description applies for any film to be studied, but has been described with reference to PSZT. During PSZT deposition, a portion of the bottom metal electrode (gold or platinum) was covered with a piece of silicon to prevent sputtering of PSZT on the selected region of the sample; this was done to create electrical access to the bottom electrode. This form of shadow masking was performed in order to study the film in its as-deposited form, without exposing it to wet chemical etchants. The top metal electrode (600 nm aluminium) was deposited at room temperature by electron beam evaporation. In this case also, evaporation was carried out through a shadow mask with small openings to define the top electrode. On all samples, two or more electrodes were defined to study the uniformity in response of the thin films. Individual samples were firmly mounted onto glass slides using non-conductive double-sided tape. The glass slides used had gold islands defined by photolithographic patterning, with the gold having been deposited by electron beam evaporation (at room temperature) to a thickness of 150 nm, using a 15 nm titanium adhesion layer. The top and bottom electrodes were wire-bonded to individual gold islands on the glass slide using gold ribbon (76 μm wide and 12.5 μm thick). Fine gauge wire was bonded using silver epoxy to these gold islands, to serve as connectors to external devices, such as the function generator. All connections were tested to ensure that no electrical shorts were present. A schematic representation of such a sample is shown in Figure 2.

Figure 2. A schematic representation of samples prepared for piezoelectric response measurements. (Not to scale.)

356

Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

3. ESTIMATION OF d33 USING A NANOINDENTER Published work on the measurement of the piezoelectric response of thin films using a nanoindenter [8-11] are generally based on the direct piezoelectric effect, and use the nanoindenter to apply stress on the thin films. While these techniques are expected to give the same numerical value for d33, they require an in-depth understanding of indenter mechanics, dynamics, and principles. Whereas, the technique presented below uses the indenter only as a means of accurately measuring displacement. Related work by Rar et al. [11] has been carried out for piezoelectric substrates using the inverse piezoelectric effect, and highlights that the indentation approach leads to better quantification of d33, compared to the popular piezoresponse force microscopy (PFM) technique.

3.1. Experimental Details An UMIS 2000 nanoindenter (CSIRO, Australia) was used to carry out the measurements, given its precision for displacement measurements. A blunt metallic tip was used to prevent deformations to the thin film being tested, and contact of the tip to the film was made with a force of 1 mN. The indenter tip was brought into contact with the surface of the top electrode or the piezoelectric thin film. The piezoelectric thin film studied was strontium-doped lead zirconate titanate (PSZT) deposited for a duration of 4 hours at 300 ºC on gold-coated silicon substrates (PSZT on Au/Ti/Si with a PSZT thickness of ~1900 nm). A schematic of this arrangement is shown in Figure 3. The top electrode was connected to ground of the function generator. This was done to ensure that no electrostatic interaction between the tip and the sample affected the measured displacements. The indenter software (IBIS, version 1, 2004 from Fischer-Cripps Laboratories Pty. Ltd.) was used to follow the displacement of the surface of the films over time. Most measurements were made by placing the tip on the surface of the aluminium top electrode, to ensure the response to the entire electric field is measured. Low frequency (ranging from 25-500 mHz) electrical signals with peak-to-peak voltages of 10 V and 32 V were used, due to the limitations in the sampling rate of the software. Mapping options in the software were used to precisely position the nanoindenter tip and perform line scans and area scans to study piezoelectric response variations.

3.2. Results and Discussion Initial piezoelectric response measurements were done on the surface of the aluminium top electrode, in order to study the response of the piezoelectric thin film to the entire applied electric field. Applied peak-to-peak voltages of 10 V and 32 V at one particular point, resulted in peak-to-peak film thickness changes of 4.34 nm and 13.97 nm respectively. These relate to d33 values of 434 pm/V and 437 pm/V respectively, indicating a linear and consistent result. These are typical of results obtained for d33 measurements done at different points. Mapping of the thin film piezoelectric response over an area of 80 μm by 80 μm on the aluminium top electrode showed a region of high piezoelectric response, with the varying

Thin Film Piezoelectric Response Coefficient Estimation Techniques

357

response around that region (Figure 4). This is due to the variations in the alignment of domains in the region being studied. A minimum d33 value of 458 pm/V and a maximum d33 value of 608 pm/V were obtained, with an average value of 545 pm/V.

Figure 3. Schematic of the arrangement used to measure the piezoelectric response of the PSZT thin films. (Not to scale.)

Figure 4. Results from mapping the piezoelectric response of a PSZT thin film over an 80 μm × 80 μm area (on the aluminium electrode surface) for an applied peak-to-peak voltage of 32 V. (Reprinted with permission from Ref. [12].)

358

Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

The piezoelectric response was mapped along a line of length 1250 μm, at intervals of 250 μm, and the resulting data is shown in Figure 5. The response is high and reasonably consistent on the top of the electrode, reaches a maximum closer to the edge of the electrode, and drops to a lower value when measured directly on the thin film. These results follow the expected variations in electric field strength. The field is expected to be at its maximum between the top and bottom electrodes, with its strength decreasing with distance from the electrode area.

Figure 5. Results from a linear map of the piezoelectric response for an applied peak-to-peak voltage of 32 V, to study variations in response with distance from the top aluminium electrode. (Reprinted with permission from Ref. [12].)

In order to validate the measurement technique, a sample with the same overall arrangement, but with amorphous silicon dioxide instead of PSZT was tested. The resulting response under an applied electric potential for silicon dioxide was a null response, similar to those obtained when no electric potential was applied to the piezoelectric PSZT thin film. This supports the conclusion that the film thickness variations measured were a direct consequence of the piezoelectric behaviour. The range of d33 values was also verified using the technique presented in the next section. Discussion on the physical processes resulting in the measured piezoelectric response coefficient and comparison of these values to those in published literature is presented in Ref. [12].

4. ESTIMATION OF d33 USING AN ATOMIC FORCE MICROSCOPE This section describes a technique for the quantification of d33 using atomic force microscopes (AFM), without the need for modifications; to support and simplify other reported techniques, such as piezoresponse force microscopy (PFM) [13] and piezo-nano-

Thin Film Piezoelectric Response Coefficient Estimation Techniques

359

indentation (PNI) [4]. The AFM is used under the standard contact mode imaging settings, as is done for most surface topography scans. Samples are prepared (as described in Section 2) to study the piezoelectric response of the thin films under the inverse piezoelectric effect. The technique uses the scan result at a given point to extract the peak-to-peak displacement of the thin film for a specific applied peak-to-peak voltage. This results in a quantified value for d33 in pm/V.

4.1. Experimental Details PSZT thin films deposited on Au/Ti/SiO2/Si substrates at 300 ºC for 4 hours were used during the development of this technique. These films were subsequently annealed in the sputtering chamber, during the cooling down process, at 250 ºC for 1 hour. A Digital Instruments Dimension 3100 atomic force microscope with a Nanoscope IIIa controller was used for these experiments. The instrument was calibrated for ‘z’ displacements before use, using a 100 nm step height standard. The system was used in the contact mode imaging arrangement, with non-conductive silicon nitride tips. Figure 6 is a schematic representation of the testing arrangement used (similar to Figure 3). The AFM tip is placed at a chosen point on the surface of the top electrode. An electric field is applied between the top and bottom electrodes with the top electrode connected to the ground of the function generator to prevent charging up of the non-conductive tip surface (this could cause attractive or repulsive forces, and so, could introduce errors). The electric field in Figure 6 is represented using two components: Vp represents the peak value or amplitude of the signal and is half the peak-to-peak voltage value, and f(t) represents the signal type used and corresponds to the sine function. Frequencies ranging from 500 mHz to 10 Hz were experimented with, limited by the sampling rate and tip mechanics of the system. All scan results incorporated in this article were carried out with resolution settings of 512 pixels/line and 512 lines in an image. The scan area value is set to zero and the line scan frequency was set to 1 Hz. The electric field was applied using an Agilent Technologies 33220A function generator without a 50 Ω termination. In order to accurately evaluate the electric signal applied across the film, an oscilloscope was connected in parallel during all measurements.

4.2. Results and Discussion This section describes the results obtained using the proposed piezoelectric response measurement technique, and discusses the manner in which the electro-mechanical coefficient d33 can be obtained from the scan results. Figure 7 shows the scan results obtained with a scan rate of 1 Hz and for varying applied signal frequencies, from 1 to 4 Hz. Considering that the AFM tip is not scanning over any region (it is at one chosen point on the top electrode), the displacements observed in Figure 7 correspond to variations in thickness over time at a point chosen on the sample. This is different from normal AFM scan results in which displacements are mapped over a chosen area. Figure 7 [especially Figure 7(b)] clearly depicts that every scan line has sinusoidal

360

Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

variations, with the number of cycles directly corresponding to the ratio of the driving frequency (1-4 Hz) and the scan frequency (1 Hz); each sinusoid corresponds to a set of bands (one dark and one light) in the scan result. The bands in the scan result shift for each line, with an overall diagonal appearance; this is a result of delays introduced by the AFM feedback system after each scan line.

Figure 6. Schematic of the arrangement used for piezoelectric response measurements using the AFM. (Not to scale. Reprinted with permission from Ref. [14].)

This result demonstrates that the thin film displacements occur at the applied electrical signal frequency, as expected. Samples with amorphous silicon dioxide in place of PSZT were tested under the same arrangement and exhibited no displacement variations (null response – only background noise in scan results). This clearly indicates that the displacements being measured are due to the piezoelectric behaviour of the PSZT thin films. Figure 8 was used to verify the technique further. During the initial portion of the scan, no electric field was applied across the thin film. The result is just background noise in the scan result. On applying an electric field with a frequency of 1 Hz, the film starts to respond instantly, reflected by approximately one sinusoidal cycle in the middle portion of the scan. The number of sinusoids increases to two in the last section of the scan, when the frequency is doubled. Figure 9 highlights the ability of the proposed technique in identifying displacement variations with applied voltage at a constant driving frequency; larger displacements increase contrast in the image while the number of bands in the image remain constant. During the course of the scan, the amplitude of the driving voltage was changed. From an initial peak-topeak amplitude of 12 V, it was increased to 16 V, with the final scan region being recorded at 20 V. The corresponding peak-to-peak displacements are 1.3 nm, 2.7 nm, and 4.1 nm.

Thin Film Piezoelectric Response Coefficient Estimation Techniques

361

(a)

(b) Figure 7. Scan result when the frequency of the applied sinusoidal electric field was varied from 1 to 4 Hz, in steps of 1 Hz, with a fixed peak-to-peak amplitude of 20 V. The dark and bright bands seen in (a) are due to the applied sinusoidal voltage’s positive and negative cycles, respectively. The threedimensional representation of (a) shown in (b) highlights the sinusoidal nature of the displacement variations. (Reprinted with permission from Ref. [14].)

From these scan results, the piezoelectric voltage coefficient d33 can be estimated in pm/V, as the ratio of the peak-to-peak displacement (in pm) to the applied peak-to-peak voltage (in V). For example, the displacement variations in Figure 9 of 1.3 nm for 12 V, 2.7 nm for 16 V, and 4.1 nm for 20 V, correspond to d33 values of approximately 109, 169, and 205 pm/V, respectively (these values are lesser than those for films discussed in Section 3.2 due to lesser degree of preferential orientation, as described in Refs. [5,12,14]).

362

Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell

Figure 8. Scan result where three sectors of the image correspond to time periods of no applied signal and to signal frequencies of 1 Hz and 2 Hz for an applied sinusoidal electric field of fixed peak-to-peak amplitude of 20 V. (Reprinted with permission from Ref. [14].)

Figure 9. Scan result obtained under an applied 4 Hz sinusoidal voltage. The peak-to-peak amplitude of the voltage was increased at regular intervals. The applied voltage and resulting displacement are shown as peak-to-peak values. (Reprinted with permission from Ref. [14].)

5. CONCLUSION This chapter presents two new and simplified techniques for the estimation of the piezoelectric charge coefficient d33. The following key results have been obtained: A. A technique has been developed to estimate thin film d33 using a nanoindenter under the inverse piezoelectric effect.

Thin Film Piezoelectric Response Coefficient Estimation Techniques

363

B. The capabilities of using the nanoindenter to map piezoelectric response data over an area and along a line, to study influence of electrodes or domain/grain distribution, are presented. C. An atomic force microscope in the contact imaging mode has been used to estimate thin film d33. By following the displacement at one point, a representation of thin film variation as a function of frequency and amplitude of the applied electric field can be studied. The two techniques described in this chapter enable quantification of piezoelectric response by enabling accurate estimation of the piezoelectric response coefficient d33 using simplified techniques and easily accessible instrumentation.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

Cady, W. G. (1964). Piezoelectricity; Dover Publications: New York, Vol. 1. Lefki, K. & Dormans, G. J. M. (1994). J. Appl. Phys., 76, 1764-1767. Gruverman, A. & Kalinin, S. V. (2006). J. Mater. Sci., 41, 107-116. Koval, V., Reece, M. J. & Bushby, A. J. (2005). J. Appl. Phys., 97, 074301. Bhaskaran, M., Sriram, S., Short, K. T., Mitchell, D. R. G. & Holland, A. S. (2008). Thin Solid Films, 516, 8101-8105. Sriram, S., Bhaskaran, M., Du Plessis, J., Short, K. T., Sivan, V. P. & Holland, A. S. (2009). Micron, 40, 104-108. Sriram, S., Bhaskaran, M., Mitchell, A., Mitchell, D. R. G. & Kostovski, G. (2009). Nanoscale Res. Lett., 4, 29-33. Karapetian, E., Kachanov, M. & Kalinin, S. V. (2005). Philos. Mag., 85, 1017-1051. Sridhar, S., Giannakopoulos, A. E. & Suresh, S. (2000). J. Appl. Phys., 87, 8451-8456. Algueróa, M., Calzada, M. L., Bushby, A. J. & Reece, M. J. (2004). Appl. Phys. Lett., 85, 2023-2025. Rar, A., Pharr, G. M., Oliver, W. C., Karapetian, E. & Kalinin, S. V. (2006). J. Mater. Res., 21, 552-556. Sriram, S., Bhaskaran, M., Holland, A. S., Short, K. T. & Latella, B. A. (2007). J. Appl. Phys., 101, 104910. Kalinin, S. V. & Bonnell, D. A. (2004). In Nanoscale Phenomena in Ferroelectric Thin Films; Hong, S.; Ed.; Kluwer Academic: Boston, MA, pp 183-217. Sriram, S., Bhaskaran, M., Short, K. T., Matthews, G. I. & Holland, A. S. (2009). Micron, 40, 109-113.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 365-372

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 13

PLASMA TECHNOLOGY: AN ALTERNATIVE TO CONVENTIONAL CHEMICAL PROCESSES FOR HYDROGEN PRODUCTION María Dolores Calzada* Grupo de Espectroscopia de Plasmas, Edificio Einstein (C-2), planta baja, Campus de Rabanales, Universidad de Córdoba, 14071 Córdoba (Spain)

ABSTRACT There is currently a strong debate on the need to look for new alternative energy sources to replace fossil fuels (particularly oil), due to atmospheric pollution derived from their use, probable reserve exhaustion and productive countries’ excessive dependence on future political events. Therefore, many countries are currently giving an incentive to research aimed at the development of new energy sources, among which hydrogen —and its use by means of fuel-cells— is to be included. Within the so-called hydrogen economy, hydrogen production —from different raw materials such as hydrocarbons, alcohols and some others— becomes an important aspect, since hydrogen is not a natural product but is obtained by means of water vapour. However, such process leads to the production of high amounts of CO2. The use of plasma technology allows generating hydrogen by decomposing organic compounds (alcohols and methane), so plasma acts as a reactive environment; this way, biogas —which can be easily found in city landfills— can also be used for hydrogen obtention. Thus, the use of plasma with this purpose arises as an alternative process not only to reforming of hydrocarbons and other substances with water vapour but also to other purely chemical processes, avoiding or minimizing CO2 emissions (involving greenhouse effects). A subproduct —obtained with H2 and which also deserves special attention— is solid carbon, which has important added value, since it may work as a base element for chemical industry and other technological fields. Given the foregoing, research in the use of plasma technology for hydrogen production is one of the research lines within Applied Physics with higher applicability potential within the energy sector.

*

Corresponding author: Email: [email protected]

366

María Dolores Calzada

INTRODUCTION The energy sector itself constitutes an important part of the economic activity in any given country, is an indispensable service for its citizens’ everyday life and contributes an undeniable value to the remainder sectors in such economy. Studies show that 80 % of the total energy generated and consumed by the human being comes from fossil fuels, whose combustion involves the formation of CO2, which is considered a non-environmentallyfriendly and greenhouse-effect gas. The current rate of CO2 emissions from different human activities is around 25 % of the natural rate of atmospheric CO2-recycling by photosynthetic processes. Although physical and chemical exchanges with oceans should be included into the global carbon-CO2 cycle, the previous figure shows the existence of a strong artificial perturbation in a natural cycle whose atmospheric component plays a relevant role in greenhouse effect and, therefore, in global warming. Figures of molar CO2 concentration in air have increased with a progressively more marked tendency since the beginning of the 20th century, currently exceeding 380 molar ppm. On the other hand, studies completed in Norwegian fiords have shown up that atmospheric CO2 levels affect oceans, given that the absorption of carbon anhydride increases water’s acidity levels, thus producing an excessive increase of plankton and altering equilibrium in oceans. All the previous has given rise to the establishment of limits for CO2 emissions collected in different protocols such as that of Kyoto. Furthermore, additional problems related to the use of fossil fuels for energy generation should also be taken into account, such as forecasts on the exhaustion of world oil reserves within several decades and instability in domestic policy and international relations in the main oil-producer countries. Consequently, European Union member countries and other countries such as United States, Japan, China and Brazil have impelled R&D policies aimed at promoting the development of the so-called alternative energies such as solar and wind energies, the use of biomass and hydrogen as fuels, thus contributing to guarantee energy supply, reduce their dependence from oil-producer countries and contribute to diminish —drastically in some cases— CO2 emissions into the atmosphere. Particularly, hydrogen is considered as an alternative fuel to fossil ones, especially within the automotive sector, being extremely clean, since the chemical energy accumulated in H-H bonds is freed when combined with oxygen, thus producing only water as a sub-product of the reaction which takes place in the so-called fuel cells or in internal combustion engines; this energy is subsequently transformed into electric current to feed vehicle engines. Thus, hydrogen beings to look like an efficient, nonpollutant and abundant energy source at a reasonable price in medium to long term. However, hydrogen is not a natural resource and must be obtained from other raw materials (water, biomass, fossil fuels, etc.), resulting in some cases in a less-clean-than-expected production process.

HYDROGEN ECONOMY The so-called hydrogen economy tackled different issues such as production, storage (safety) and use of hydrogen as a fuel. The Issues regarding its storage (safety) and use in the automotive industry are those in which research has reached the most advanced stages so far.

Plasma Technology: An Alternative to Conventional Chemical Processes...

367

Thus, hydrogen can be stored at different pressure values both as a liquid and a gas and also in the shape of hydrides. Regarding its use, energy can be obtained from internal combustion engines or by means of the so-called fuel cells. However, one of the most important issues still demanding further in-depth research is the production of enough quantities of hydrogen to supply the current demand. Therefore, great part of the efforts in this hydrogen economy should be aimed at finding the most economic and simple method for hydrogen production. For hydrogen generation at industrial scale, water electrolysis and hydrocarbons reforming with water steam are regularly used. Water decomposition into hydrogen and oxygen is produced by the process of electrolysis. This decomposition takes place within an electrolytic cell containing two electrodes separated by a gas-proof diaphragm. Oxygen is produced in the anode and hydrogen in the cathode according to the following reactions:

1 O 2 + e − (anode) 2 − 2H 2 O + 2e → H 2 + 2OH − (cathode)

2OH → H 2 O +

(1)

1 H 2 O → H + O 2 (total reaction) 2 This process consumes a high quantity of electric energy, given that it is necessary to apply 1 amp for 24 hours to decompose a mole of water (18 grams). Thus, the production of 1 amp per second shall be 0.22 cm-3 of H2 and 0.11 cm-3 of O2, which entails a high cost in electric energy. For this reason, electrolysis is usually carried out by means of photovoltaic solar energy or wind energy. Hydrogen production from hydrocarbons reforming with water steam takes place in two stages. The first of them according to the following reaction: CmHn + mH2O → mCO + (m+n/2)H2

(2)

This reaction —independently from the used hydrocarbon— is endothermic and its enthalpy (ΔH) depends on the heat of formation of the hydrocarbon in question. Thus, for instance, such value is 206 and 260 kJ/mole in the cases of methane and ethane, respectively. The second stage comprises the following reaction: mCO + mH2O → mCO2 + mH2

(3)

In this case, the reaction is exothermic (ΔH = -m × 41.1 kJ/mole). Thus, it is observed that hydrogen production from hydrocarbons reforming with water steam entails massive production of CO2 in two steps, the first of them being CO production, also considered as a greenhouse effect gas. This means that the level of emissions is not reduced, but may even increase. For this reason, efforts have been recently aimed at finding methods for hydrogen production which enable reducing or even eliminating CO2 emissions.

368

María Dolores Calzada

PLASMA: REACTIVE MEDIUM TO OBTAIN HYDROGEN A new alternative option to that of hydrocarbons reforming with water steam is transformation of hydrocarbons into hydrogen and carbon through controlled pyrolytic processes according to the following reaction: CnH2m → nC + mH2

(3)

This is also an endothermic reaction. Thus, for instance, absorbed energy in the case of methane is 75 kJ/mole, 84 kJ/mole for ethane, 100 kJ/mole for propane and 125 kJ/mole for butane. The main inconvenient of these reactions is that they require very high temperatures, over 500 K, while it has been ascertained that, for instance, at temperatures over 1000 K practically all methane molecules enclosed in a recipient become split. A medium which can be considered as one of the most appropriate ones for Reaction (3) to take place is plasma. This statement is based on the fact that plasma is a partially-ionized gas in which electrons, ions and atoms coexist with enough energy to break up, by means of collisions, the bonds of substances introduced into it. In plasmas generated at reduced pressure (below atmospheric pressure), electron energy may reach 20000 K vs. 300 to 400 K found in heavy particles (atoms, ions and radicals), while in plasmas at atmospheric pressure, electron energy may be ≥ 7000 K and that of heavy particles ≥ 1200 K. On the other hand, this temperature difference between plasma particles is characteristic of Non-Local Thermodynamic Equilibrium (LTE) plasmas, which allows varying population density and temperature of its species by modifying the operation conditions of the discharge (applied power, gas kind and flow, etc.), influencing the reactions which take place in the discharge, in order to make them follow the desired direction. Therefore, Non-Local Thermodynamic Equilibrium plasmas may be considered as one of the most appropriate methods to produce certain reactions which cannot take place or are not effective enough through conventional methods. Thus, if a hydrocarbon flowing is conducted into plasma generated with an inert gas, the decomposition of such hydrocarbon gives rise to the production of H2 and C with no CO2 emissions. Such decomposition may be constituted by several stages depending on the operation conditions under which plasma is generated and the kind of hydrocarbon; not only electrons can take part in such reactions but also radicals themselves formed after a first stage of decomposition. However, not only hydrocarbons can be considered as a raw material from which hydrogen is to be obtained by using plasma. Other substances such as alcohols (methanol, ethanol and bioethanol), pure methane, natural gas and even biogas may be used with this purpose. The use of plasma as a reactive medium to decompose different raw materials with the aim of obtaining hydrogen involves a series of advantages, not only from an environmental viewpoint but also from an operation viewpoint. Among its advantages it is to stand out the wide interval of frequencies, from some MHz to GHz, at which plasmogen gases can be generated. These gases can be both pure and mixed, and working pressure ranges between some Torr to some atmospheres. Furthermore, there are several energy applicators. All the previous allows studying plasmas generated under different operation conditions to find those which fit best the kind of raw material from which hydrogen is extracted. However, the main

Plasma Technology: An Alternative to Conventional Chemical Processes...

369

weakness of plasmas is the fact that, in most cases, it is necessary to use gas or easilyvaporizable raw materials. The following figures are an example of the decomposition of methanol by microwave plasma generated in Ar at atmospheric pressure into which a flowing of the afore-mentioned alcohol has been introduced. The capacity of this plasma to break up methanol molecules was observed in collected spectra corresponding to the characteristic radiation emission of the plasma used in these experiments [1]. Figure 1 shows the emission spectrum of plasma generated in pure Ar, showing the lines of atomic Ar and some molecular bands corresponding to impurities of different species (at level of traces) in the gas used as plasmogen gas. Figure 2 shows not only Ar lines but also the molecular bands of radicals from the decomposition of the methanol molecules introduced into the plasma. Both the increase of the line Hα of hydrogen and the appearance of the band C2 involve the formation of molecular hydrogen and solid carbon at plasma end. This shows that, apart from hydrogen, other sub-products —which contribute value added to the process— may be obtained. Among these sub-products it is to stand out carbon, with applications within different scientific and technologic fields such as Mechanics, Electronics, Medicine, etc. 4

1.5x10

4

I (u.a)

1.0x10

3

5.0x10

0.0

Hα

NH OH

300

400

500 600 λ (nm)

700

I (u.a)

Figure 1. Spectrum of pure-Ar plasma generated at atmospheric pressure [1] 5x10

4

4x10

4

3x10

4

2x10

4

1x10

4

CN Hα OH and CH CN and C2

C2

C2

NH

C2

0 300

400

500

600

700

λ (nm)

Figure 2. Spectrum of Ar plasma generated at atmospheric pressure when a methanol flowing is introduced into the plasma [1]

370

María Dolores Calzada

Studies carried out in this line have shown that, at plasma end, the percentage of species recombination which gives rise to the primitive substance introduced into the plasma is very small, always obtaining H2 and C, as well as other sub-products depending on the raw material used [1-12]. Figures 3, 4 and 5 are an example of mass spectra at the discharge exit after methanol was decomposed by plasma at atmospheric pressure [1]. The first twenty spectra were recorded while the plasma was switched-on and the next twenty ones with the plasma switched-off. In Figure 3, one observes molecular hydrogen production from methanol decomposition by plasma. Spectra corresponding to Figure 4 show the formation of acetylene (26 amu), which is obtained from the association of CH radicals. It results in the possibility of obtaining other sub-products from alcohol decomposition.

I (a.u.)

Molecular Hydrogen

Atomic Hydrogen

Plasma off Plasma on

Mass

Figure 3. Mass spectra: production of molecular hydrogen [1]

Water t

I (a.u.) Methanol Acetylene Nitrogen

Oxygen

Plasma off

Plasma on

Mass

Figure 4. Mass spectra: production of acetylene [1]

With respect to CO2 (44 a.m.u.), Figure 5 shows the same amount when the plasma is switched-on and switched-off, which implies non-formation of CO2 by plasma. This substance comes mainly from argon, alcohol impurities and the air which is in contact with the MS fiber. This is the reason why the use of the discharge at atmospheric pressure not only simplifies its implementation on an industrial scale but also allows it to be considered a noncontaminating hydrogen molecular source.

Plasma Technology: An Alternative to Conventional Chemical Processes...

371

Argon

I (a.u.)

CO2

Plasma off

Plasma on

Mass (amu)

Figure 5. Mass spectra: CO2 due to air in contact with the MS fiber [1]

This result also shows another advantage in the use of plasma as a medium to decompose substances containing hydrogen: In case of using natural gas or biogas, whose content is not 100 % methane, the molecules of the remainder component substances also undergo decomposition, so less pollutant sub-products are obtained. This behaviour of plasma has already been observed in the purification of Kr and Xe gases from air distillation and destruction of SF6 [13-14], a gas generated during the manufacture of microelectric components and circuits.

FINAL COMMENT The use of hydrogen as a fuel in the automotive sector is currently a priority research objective for different automobile brands —General Motors, Honda, Mazda, Nissan, Renault, Volkswagen, Hyundai and Toyota among them. It is in the issue of hydrogen production where plasma may play a decisive role as a reactive medium to produce the decomposition of different substances containing hydrogen, without the emission of greenhouse effect gases as sub-products of the reactions taking place in the plasma. Finally, it should also be pointed out that the research field of hydrogen production by plasma is a multidisciplinary and still wide-open field which needs synergies from different scientific and technologic fields, among which Physics plays an essential role, since it is the field which may contribute the definitive understanding of the processes taking place in plasma and their optimization. This way, with the participation of different sectors, hydrogen shall really become the fuel of a not very far future.

REFERENCES [1] [2] [3]

Jiménez, M; Yubero, C; Calzada, MD. J. Phys D: Appl. Phys. 2008, vol. 41, 155201 (6pp). Deluga, GA; Salge, JR; Schmidt, LD. Verykios, X.E. Science, 2004, vol. 303, 993 Bromberg, L; Cohn, DR; Rabinovich, A; Alexeev, N; Samokhin, A; Ramprasad, R; Tamhankar, S. Int. J. Hydrogen Energ., 2000, vol. 25, 1157-1161.

372 [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]

María Dolores Calzada Bromberg, L; Cohn, DR; Rabinovich, A; Alexeev, N. Int. J. Hydrogen Energ. 1999, vol. 24, 1131-1137. Bromberg, L; Cohn, DR; Rabinovich, A; Heywwod, J. Int. J. Hydrogen Energ. 2001, vol. 26, 1115-1121. Lutz, AE; Bradshaw, RW; Bromberg, L.; Rabinovich, A. Int. J. Hydrogen Energ., 2004, vol. 29, 1401-1424. Deminsky, M; Jivotov, V; Potapkin, B; Rusanov, V. Pure Appl. Chem., 2002, vol. 74, 413-418. Sarmiento, B; Brey, JJ; Vera, IG; González-Elipe, AR; Cotrino, J; Rico, VJ. J. Power Sources, 2007, vol. 169, 140-143. Li, H; Zou, J; Zhang, Y; Liu, C. Chem. Lett., 2004, vol. 33, 744-745. Aubry, O; Met, C; Khacef, A; Cormier, JM. Chem. Eng. J., 2005, vol. 106, 241-247. Tanabe, S; Matsuguma, H; Okitsu, K; Matsumoto, H. Chem. Lett., 2000, vol. 10, 1116-1117. Yanguas-Gil, A; Hueso, JL.; Cotrino, J; Caballero, A; Gonzalez-Elipe, AR. Appl. Phys. Lett. 2004, vol. 85, 4004-4006. Nantel-Valiquette, M; Kabouzi,Y; Castaños-Martinez, E; Makasheva, K; Moisan, M; Rostaing, JC. Pure Appl. Chem., 2006, vol. 78, 1173-1185. Kabouzi, Y; Moisan, M; Rostaing, JC; Trassy, C; Guerin, D; Keroack, D; Zakrzewski, Z. J. Appl. Phys., 2003, vol. 93, 9483-9496. Rostaing, JC; Bryselbout, F; Moisan, M; Parent, JC. C. R. Acad. Sci. Paris, 2000, vol.1 (Série IV), 99-105.

In: Applied Physics in the 21st Century… Editor: Raymond P. Valencia, pp. 373-389

ISBN: 978-1-60876-074-9 © 2010 Nova Science Publishers, Inc.

Chapter 14

TARGET-PLASMA-FILM INTERACTIONS IN HIGH POWER PULSED MAGNETRON SPUTTERING (HPPMS) K. Sarakinos* Materials Chemistry, RWTH Aachen University, 52074, Aachen, Germany

ABSTRACT Growth of films by plasma-assisted physical vapor deposition (PVD) techniques provides means for tailoring their properties and improving their functionality in technological applications. State of the art plasma-assisted PVD techniques, like direct current magnetron sputtering (dcMS), suffer from low degree of ionization and thus, low ion-to-neutral ratios in the flux of the deposited material. The implementation of external sources for the enhancement of the ionization is in many cases technically complicated and increases the end-product cost. High power pulsed magnetron sputtering (HPPMS) is a novel sputtering technique that elegantly enables the conversion of conventional sputtering source into an ion source. By applying the power in unipolar pulses of low frequency (<2 kHz) and low duty cycle (<10%) high values of peak target current density up to ~ 5 Acm-2 are achieved. This leads to the generation of ultra dense plasmas with electron densities of ~1019 m-3. These values are up to four orders of magnitude higher than those obtained by dcMS and result in a high degree of ionization of the sputtered material (up to ~90%). Moreover, due to the low duty cycles, the average target current densities are maintained at values of ~0.5 Acm-2 which are comparable to those in dcMS processes. Therefore, HPPMS allows for highly ionized deposition fluxes without excessive thermal load at the target. In the current short communication the target-plasma interactions and their effect on the deposition rate in non-reactive and reactive HPPMS are discussed. Furthermore, the interactions of the plasma species in HPPMS with the growing film and their implications on the microstructure and the phase formation of metal nitride and oxide films are demonstrated.

*

Corresponding author: [email protected]

374

K. Sarakinos

1. INTRODUCTION Thin films are used in diverse technological applications, such as surface protection and decoration, data storage, optical and microelectronic devices. The increasing demand for new functional films has been a strong incentive for intensive research towards understanding fundamentals and technical aspects of their growth and developing new deposition techniques. Among the various methods employed for film growth, Physical Vapor Deposition (PVD) techniques, such as magnetron sputtering, are widely used [1]. Magnetron sputtering is a plasma-based technique, in which inert gas (commonly Ar) atoms are ionized and accelerated by a negative potential towards a solid material source (target) causing the ejection (sputtering) of atoms which condensate on a surface (substrate) and form a film [1, 2]. The film growth is a complex non-equilibrium process in which both thermodynamics and kinetics allow for tailoring of the film properties [2, 3]. The latter can be achieved by (i) the growth temperature [2, 4] and (ii) bombardment by energetic species [5-9], which can be employed either in conjunction with the growth temperature or as an alternative when heating of the substrate is not possible. In general, there are two kinds of bombarding species, i.e. neutrals and ions. Ions are favored over the neutrals, since their energy, flux and trajectory can be easily controlled by using external electric and/or magnetic fields [5] providing thus well defined conditions of energetic bombardment. In these terms conventional magnetron sputtering techniques, such as dc magnetron sputtering (dcMS), are not always the best solution. The dcMS plasmas exhibit a relatively low electron (plasma) density ne (ne=10141016 m-3) [10] while the penning ionization is the main ionization mechanism [11]. These features have as a consequence an ionization degree of a few percent for both the sputtered material and the Ar gas. This in turn results ion low ion-to-neutral ratios available at the substrate, while Ar+ ions are the dominant bombarding species [10, 11]. The low ion-toneutral ratios often require relatively high energies of the individual bombarding species in order for the effect of the energetic bombarding on the growing film to be significant [8]. The high bombarding energies may result in subplantation of Ar+ ions into the film [12, 13] leading to generation of lattice defects [12, 13], high residual stresses [14-18], deterioration of the quality of the film/substrate interface [16, 19], and poor film adhesion. Thus, the increase of the ion-to-neutral ratio and primarily the increase of the ionization degree of the film forming species has been an objective of many research works during the past decades. Several solutions have been demonstrated, such as the use of inductively coupled plasmas, external electron and ion sources in conjunction with a magnetron plasma [20], as well as other ionized PVD techniques such as cathodic arc [20]. These approaches have successfully served the protective coatings, microelectronic and optical industries by enabling the deposition of high quality films on complex shaped substrates. However, in the case that the use of magnetron sputtering based processes without the complexity of external sources is required the challenge is to turn a conventional magnetron source into an ion source. The latter can be achieved by increasing the plasma (electron) density and thus the frequency of the electron impact ionization collisions [11, 21]. It can be shown that the plasma density and thus the ionization degree are proportional to the power density applied to the target [22] which implies that this approach is by definition limited by the maximum thermal load that the cathode can accommodate.

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

375

In the mid-90’s Mozgrin et al. [23] Bugaev et al. [24] and Fetisov et al. [25] demonstrated that the operation of a conventional sputtering source in a pulsed mode, with a pulse duration ranging from 1 µs to 1 s and a frequency less than 2 kHz, allowed for pulse target power up to three orders of magnitude higher than the average power in dcMS, while maintaining the average target power at the dcMS level [23-25].These high values of pulse power resulted, in turn, in ultra-dense plasmas with electron densities in the order of 1018 m-3 [23-25], which is much higher than the values of 1014-1016 m-3 commonly obtained for dcMS [10, 11]. A few years later in 1999, Kouznetsov et al. [26] demonstrated the operation of a Cu target in this high power pulsed mode. They showed that high plasma densities obtained using this pulsed technique resulted in a total ion flux two orders of magnitude higher than that of a dcMS plasma, and a Cu ionization of ~70 %. The new deposition technique was called High Power Pulsed Magnetron Sputtering (HPPMS) and in the years after Kouzsetov’s report it has extensively been used as an elegant approach to increase the ionization degree in conventional magnetron sputtering devices. Later, several research groups [20] adopted the alternative name High Power Impulse Magnetron Sputtering (HiPIMS) for this technique. The high degree of ionization in HPPMS has been shown to allow for the growth of ultradense and ultra-smooth films [27], homogeneous deposition on substrates of complex geometry [26-28], engineering of the film/substrate interface [29], control over the phase composition [30, 31] and more recently exotic applications such as gasless sputtering [32]. Another issue which related to the HPPMS process and has attracted considerable attention is the deposition rate, since it has been reported that depending on the target material, the pulse configuration and whether the deposition is performed in a metallic or a reactive mode, rates ranging from 15% to 120% of the values achieved by dcMS are obtained, when the same average power is used [20, 33-40]. The achievements of HPPMS in the last decade show the great potential of this technique. The establishment of HPPMS as a mainstream magnetron sputtering technology goes through a deeper understanding of the underlying physics which would enable development of products (films and processes) unique for this technique. The key to achieve this is to consider, understand and describe the HPPMS process in light of the target-plasmafilm interactions. As shown schematically in Figure 1, the interactions of the plasma with the target determine the plasma density and ionization, the ejection (sputtering) rate of the target material and the transport of ionized and neutral species from the target to the substrate. These in turn have implications on the deposition rate and determine the properties of the growing film. Along these lines in section 2 the issue of the deposition in both non-reactive and reactive HPPMS is discussed. In section 3 the concept of the target-plasma-film interactions is used in order to understand the effect of HPPMS on the microstructure and the phase formation of metal nitride and oxide films.

2. THE DEPOSITION RATE IN HPPMS The deposition rate in HPPMS is an area that has been attracting considerable attention. It has been reported that depending on the target material, the pulse configuration and whether the deposition is performed in a metallic or a reactive mode, rates ranging from 15% to 120% of the values achieved by dcMS are obtained, when the same average power is used [20, 33-

376

K. Sarakinos

40]. Since the deposition rate is a quantity of significant technological importance, several research groups have attempted to identify mechanisms that determine its magnitude in HPPMS. In the following paragraphs, the issue of deposition rate in non-reactive and reactive HPPMS is discussed.

Figure 1. The target-plasma-film interactions during HPPMS are of key importance for the further understanding and the development of this technique.

2.1. Non-Reactive HPPMS The use of HPPMS for non reactive deposition has been shown to give lower deposition rates than those obtained by dcMS at the same average target power [20, 34]. In order to understand the mechanisms behind this loss of deposition rate, researchers have focused their attention on two stages of the film deposition process, (i) the interactions of the plasma species with the target and their effect on the target erosion (sputtering) rate, and (ii) the transport of the ionized sputtered species from the target to the substrate [41-44]. In the following the effect of the target sputtering rate on the deposition rate is discussed, In a first approximation, the deposition rate in sputtering processes is proportional to the target erosion rate ( Φ ), i.e. the number of species ejected from the target per time and area unit. The erosion rate is, in turn, equal to

Φ = ji × Y

(1)

where ji is the number (current) of ions impinging onto the target and Y designates the sputtering yield, namely the number of particles sputtered per impinging ion. In conventional sputtering processes (e.g. dcMS) the ion target current consists mostly of Ar+ species [10], i.e.

ji

dcMS

which implies that the erosion rate is equal to

≈ j Ar

+

(2)

Target-Plasma-Film Interactions in High Power Pulsed Magnetron... +

Φ dcMS ≈ j Ar × Y Ar

+

377 (3)

The total ion current is proportional to the average current applied to the target ( I Tav ). The sputtering yield, for a given ion-target combination, exhibits a power-law dependence on the energy ( E ) of the impinging ions [45], i.e.

Y ∝ Em

(4)

with m < 1 . The ion energy is, in turn, equal to

E = q × eVT

(5)

where VT is the target voltage, q stands for the charge state of the impinging ion and e designates the elementary charge. Deposition at a constant average target power ( PTav ) implies that

PTav = I Tav × VT = const.

(6)

It is known that the HPPMS discharges are driven by cathode potentials in the range of 400 to 2000 V [46, 47], exceeding the target potentials commonly employed in dcMS (~300 eV) [46, 47]. This fact in combination with Eq. (6) means that the use of a constant average target power leads to lower I Tav values in HPPMS. The lower I Tav values result in a decrease in the flux of ions to the target and thus, in a decrease in target erosion rate (Eq. (1)) [46]. This decrease cannot be compensated by the increase of the target voltage, due to the nonlinear dependence of the sputtering yield on the ion energy [46, 48], as demonstrated in Eq. (4). This argument implies that, when the target voltage is increased and the average power is maintained constant, a loss of deposition rate should be expected, irrespective of the deposition process used. For most target materials the values of the relative HPPMS-to-dcMS deposition rates reported in the literature [20] are lower than those which correspond to the sputtering process described by Eq. (3), i.e. sputtering of the target by Ar+ ions. Thus, a different mechanism has to be invoked, in order to explain these findings. A number of early studies on the deposition rate in HPPMS [20, 22] suggested that the high degree of ionization obtained with this technique leads to an increased probability for ionized sputtered species (M+) to be re-directed towards the target and self-sputter resulting in an ion current

ji

HPPMS

+

= j Ar + j M

+

(7)

The presence of M+ ions in the ion flux towards the target results in an erosion rate equal to [49], +

+

+

+

Φ HPPMS = j Ar × Y Ar ( E ) + j M × Y M ( E ) − j M

+

(8)

378

K. Sarakinos

with Y

M+

(E ) being the energy dependent sputtering yield of M+ ions (also referred to as

self-sputtering yield). The term “ − j

M+

” on the right-hand side of Eq. (8) accounts for the

self-sputtering and results in a decease of the total number of sputtered species as compared to dcMS. The sputtering yield ( Y sputtering yield ( Y

M

+

Ar +

) is for most metals higher or comparable to the self-

) [22]. This further contributes to the decrease of Φ

HPPMS

with respect

to Φ . It is evident that the understanding of the mechanisms that determine the ionization and the re-direction of ionized sputtered species is of key importance in order to better clarify the deposition rate changes observed in HPPMS. In general, an increase of the fraction of ionized species can be obtained via an increase of the efficiency of the ionization process. In HPPMS the electron impact is the dominant ionization mechanism [21]. Thus, higher electron densities and energies, as well as longer interaction times of the sputtered species with the energetic electrons would lead to an increase of the ionization [21]. This can be, for instance, achieved by increasing the target voltage, the average and peak target power (or current) during the HPPMS operation [44, 46, 47, 50], as well as by increasing the pulse-on time [34, 51]. In all these cases a decrease of the deposition rate has been demonstrated for a number of target materials [20, 46, 49, 52]. Moreover, the high values of peak target current are decisive for the plasma composition in the target’s vicinity, since they result in high ionization degrees of the sputtered species and depletion of Ar gas in front of the target, i.e. gas rarefaction [53]. Investigations of the plasma composition in a number of Ar-metal HPPMS discharges have demonstrated that the high ionization, the gas rarefaction, and the loss of the deposition rate are observed simultaneously [34]. The gas rarefaction signifies that less Ar is available for ionization during the pulse on-time, which contributes to the abundance of M+ species in the target’s vicinity and to the high re-direction probability [34, 46, 47]. In order to identify whether the rarefaction is a necessary condition for the loss of the deposition rate to occur, Alami et al. [46] studied the plasma composition and the deposition rate of Cr films grown by HPPMS and dcMS. In this study a constant average target current was used [46], which allows for an easy understanding of the underlying physics, since in a first approximation [46] the total ion target current remains constant at all deposition conditions. Thus, according to Eqs (1)-(8), changes in the target erosion rate can be assigned to changes in the composition of the ion target current and the ion energy (i.e. target voltage) [46]. In Figure 2 the deposition rate of Cr is plotted as a function of the peak target current (ITp) for two pulse on/off time configurations. It is seen that the increase of the pulse off-time from 950 to 2450 µs led to a higher peak target current and a decrease in the deposition rate. However, when the pulse on/off time configuration was held constant, the increase of the peak target current up to a value of ~28 A resulted in HPPMS rates equal to the dcMS ones (dotted lines in Figure 2). Above this value, and irrespective of the pulse on/off time configuration used, the HPPMS deposition rates deviated from the dcMS rates. Time resolved optical emission spectroscopy measurements [46] showed that at these conditions (ITp > 28 A) the ionization of the sputtered material increased significantly, while rarefaction was observed. Gas rarefaction was also observed below the characteristic value 28 A, where the ionization degree significantly lower than at peak target current values above 28 A [46]. This implies that simultaneous high ionization and Ar rarefaction are prerequisites for triggering the self-sputtering and loss of the dcMS

379

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

deposition rate [46]. Based on Eqs. (1)-(8) and using the constant average current approach Sarakinos et al. [49] derived the following expression for the HPPMS ( R dcMS rate ratio( R

dcMS

d

HPPMS

d

) to the

):

R HPPMS ∝ f R dcMS

+

Ar +

⋅

Y Ar ( E ) +

Y Ar ( E 0 )

+

+ f

M+

⋅

(Y M ( E ) − 1) +

Y Ar ( E 0 )

(9)

Figure 2. Dependence of the deposition rate of HPPMS grown Cr films on the peak target current. The films have been deposited using pulse off-times (950 and 1450 µs) at a constant pulse on-time of 50 µs. The deposition rate deviates from the dcMS one (dotted lines) at the well defined peak target current value of 28 A irrespective of the pulse on/off time configuration used. Above peak target current value significant increase of the ionization of the sputtered material, as well as rarefaction of the Ar gas takes place (data taken from [46]).

In Eq. (9) f

Ar +

and f

M+

target current. Y

Ar +

(E ) and Y M (E ) are the sputtering and the self-sputtering yields as

are the relative fractions of Ar+ and target ions (M+) in the +

functions of the energy of the sputtering ions, while Y

Ar +

( E 0 ) is the sputtering yield at the

reference dcMS conditions [49]. In order to quantify the dependence of the composition of the ion target current on the target material at the various deposition conditions, Eq. (9) was fitted to experimentally determined deposition rates for C, Cr, and Cu [49]. The corresponding f

M+

values are plotted as a function of the target voltage (i.e. ion energy) in

Figure 3. The analysis showed that in the case of C the current almost exclusively consisted of Ar+ ions at all deposition conditions [49] On the other hand, the results for Cr and Cu indicated a transition from and Ar+- to a M+-dominated current, when the target voltage or equivalently the peak target current was increased [49]. The absence of C+ ions in the ion flux to the target calculated for C could be attributed to the lower sputtering yield and the lower ionization probability of this, in comparison to Cr and Cu [49]. The lower sputtering yield results in lower fluxes of sputtered species, which in turn lead to a less pronounced gas rarefaction [53] while the lower ionization probability contributes further to the relative reduction of the local C+ population [49].

380

K. Sarakinos

Figure 3. Relative fraction of ionized sputtered species (fM+) for a number of target materials as a function of the ion energy (i.e discharge voltage) as estimated by the formalism developed by Sarakinos et al. [49]. The fM+ values decrease when the sputtering yield and the ionization probability of the sputtered material decrease (data taken from [49]).

2.2. Reactive HPPMS The addition of a reactive gas into the sputtering atmosphere leads to target coverage, i.e. chemisorption and/or implantation [54,55] of reactive gas species at the target surface and the sub-surface layers, respectively. Deposition from a fully covered target (referred to as the compound sputtering mode) allows for growth of stoichiometric compound films, i.e. compound films with sufficient incorporation of the reactive gas atoms [55]. At these conditions deposition rates lower than those obtained from an elemental (e.g. metallic) target are commonly achieved [55]. Growth of stoichiometric compound films with relatively high rates can be facilitated in the intermediate target coverage regime (referred to as transition zone) between the metallic and the compound mode [55]. In reactive dcMS the transition sputtering zone is frequently unstable, and a hysteresis of the process characteristics is often observed [55]. This is particularly pronounced during reactive dcMS of metal oxides [56]. As a consequence, stoichiometric films can only be obtained in the compound sputtering mode [55], unless a feedback system for controlling the target coverage is employed [57]. Earlier studies on the reactive HPPMS of various metal oxides (TiO2, ZrO2, Ta2O5) revealed deposition rates ranging between 25 % and 120% of dcMS rates [36, 38]. In these studies the obtained deposition rates were interpreted in light of the self-sputtering phenomenon [36, 38]. More recently, systematic investigations of the reactive HPPMS of Al2O3 [58] and ZrO2 [59] showed that, in contrary to dcMS, the HPPMS process exhibits a hysteresis free and stable transition zone, as demonstrated for the reactive deposition of ZrO2 in Figure 4. The stabilization of the transition zone allows for deposition of stoichiometric films at a lower target coverage, when compared to the compound mode in dcMS [58, 59], and has been shown to result in rates similar to [58] or up to two times higher [59] than the dcMS rates (Figure 5).

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

381

Figure 4. Target voltage versus O2 flow during deposition of ZrO2 using (a) dcMS and HPPMS with a pulse on-time of 50 µs and pulse of times (b) 450 µs and (c) 1450 µs. HPPMS allows for a stable transition zone as opposed to dcMS (data taken from [59]).

Figure 5. Normalized (with respect to the average target power) deposition rate of ZrO2 grown using (a) dcMS and HPPMS with a pulse on-time of 50 µs and pulse of times (b) 450 µs and (c) 1450 µs. The stabilization of the transition mode in HPPMS (see Figure 4) allows for deposition of transparent ZrO2 films with rates up to 2 times higher than those in dcMS (data taken from [59]).

In general, the stability of the transition zone in reactive sputtering processes is determined by the competition between the formation and the removal (sputtering) of the compound from the target surface [55]. According to the formalism developed by Berg et al. [55] the steady-state target coverage, ϑ t , is equal to,

ji Ycϑt − a 2 F (1 − ϑt ) = 0 q

(10)

where ji is the ion target current density, Yc the compound sputtering yield, q the elementary charge, a the sticking coefficient of the reactive gas and F the flux of the reactive gas molecules towards the target [55]. Stabilization of the transition mode can be

382

K. Sarakinos

achieved, if for a certain nominal partial pressure/flow of the reactive gas the term “

JYcϑt ” q

(removal of the compound) increases and/or the term “ a 2 F (1 − ϑt ) ” (formation of the compound) decreases. The pulsed character of the HPPMS discharge affects both terms in Eq. (10). The high peak currents (up to three orders of magnitude higher than the average values) result in a significantly higher instantaneous erosion rate during the pulse on-time, which may effectively clean the target surface and displace the oxidation onset at higher reactive gas flows [58]. In addition, the absence of plasma during the pulse off-time results in a limited activation of the reactive species [58, 60]. Under these conditions relatively high levels of reactive gas exposure are necessary for the formation of the compound on the target surface, i.e. this mechanism leads to lower effective values of the sticking coefficient a in Eq. (10). Furthermore, as a result of the high target voltage in HPPMS [46] higher compound sputtering yields Yc and as a consequence higher values of the compound erosion rate can be achieved [59]. The high peak target current in HPPMS results in rarefaction of the neutral species in the target’s vicinity, as discussed in Section 2.1, which affects not only the Ar but also the reactive gas [37]. The rarefaction implies that the reactive species flux F lower than the value which corresponds to the nominal partial pressure of the reactive gas should be expected [37, 59].

3. HPPMS FOR TAILORING THE FILM MICROSTRUCTURE AND PHASE FORMATION 3.1. The Effect of HPPMS on the Microstructure of Metal Nitride Films Polycrystalline films grown by PVD techniques exhibit a variety of microstructures with respect to the size, the morphology, and the relative orientation of the crystallites [8, 61]. These features have, for instance, implications on the mechanical strength [62] and the electrical conductivity [63-65] of the film. The microstructure is determined primarily by surface and bulk diffusion processes [8, 61], and may be controlled by variation of the growth temperature (Ts). In the case of metallic films, the effect of the temperature can be quantified using the so-called homologous temperature

Ts (Tm is the melting temperature of the Tm

deposited material) and the obtained microstructural features can be classified according to the structural zone models (SZM) [4, 6, 7, 9]. At

Ts < 0.2 (Zone I), the size of the grains is Tm

determined by the nucleation density, the film consists of uninterrupted fibrous columns and exhibits porous and rough morphologies and a random texture [8, 61]. In zone T ( 0.2 <

Ts < 0.4 ) the surface diffusion has considerable influence on the growth. This fact Tm

results in V-shaped grains at the interface due to competitive growth. At larger film

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

383

thicknesses faster growing grains overwhelm slower ones giving rise to a columnar dense morphology, smoother surfaces and strong texture [8, 61]. At higher temperatures (

Ts > 0.4 ; Zone II) bulk diffusion becomes significant giving rise to a dense columnar Tm

microstructure which is retained from the interface to the film surface [8, 61]. In Zone III the structure is characterized by equi-axed three-dimensional grains. This growth mode is a result of periodic interruption of the crystal growth and under the presence of contaminants in the film it can be observed at any substrate temperature [8, 61].

Figure 6. Cross sectional TEM micrographs of CrN films grown at room temperature by (a) dcMS and (b) HPPMS at a peak target current of 44 A. The dcMS grown films exhibit a columnar underdense morphology with inter-columnar porosity. The use of HPPMS results in dense films and suppression of the columnar growth (data taken from [68]).

The simultaneous use of low flux ion bombardment (typical for conventional magnetron sputtering processes) allows for the transitions between the structural zones to occur at lower temperatures [6, 7, 9] and for the morphological features predicted by the SZM’s to remain valid [6, 7, 9]. In contrast, growth of thin films during HPPMS is characterized by the high ion flux during the pulse on-time which can be up to three orders of magnitude higher than in dcMS [20]. Observations of the morphology of films grown by this technique have shown that the SZM in its classical form is no longer valid [29, 66-68]. An example is given in Figure 6 where cross sectional TEM images of CrN films grown by HPPMS and dcMS under otherwise the same conditions are shown [68]. In contrast to the underdense columnar morphology and high surface roughness obtained by dcMS (Figure 6 (a)), the use of HPPMS resulted in film densification, surface smoothening and suppression of the columnar structure , as shown in Figure 6 (b). In addition, some columns grew on the top of existing columns, i.e. renucleation is observed. As mentioned above, in films grown under low flux ion irradiation the renucleation occurs when the individual grain growth is inhibited by contaminants [61]. The interruption of the columnar growth can be enhanced when the

384

K. Sarakinos

plasma ionization and thus the substrate ion current are increased, e.g. by increasing the peak target current [68]. This can facilitate the transition from a dense polycrystalline to a featurelesss/nanocrytalline structure [68], as shown in Figure 7. It is therefore, evident that the low-energy high-flux ion irradiation during HPPMS can be used, in order to overcome the characteristically underdense and rough microstructures and obtain morphologies unique for low temperature deposition [68].

Figure 7. Cross sectional SEM micrographs of CrN films grown at room temperature by (a) dcMS and HPPMS at a peak target current values of (b) 44, (b) 74 and (d) 180 A. The increase of the peak target current results in a transition from a columnar to a featureless microstructure (data taken from [68]).

3.2. Phase Composition Tailoring of Metal Oxide Films Deposited By HPPMS The influence of the energetic bombardment on the phase formation of HPPMS deposited films can be unraveled using TiO2. In general, TiO2 films can grow in an amorphous and in two tetragonal crystalline structures; the rutile and the anatase phases [69]. Deposition and/or annealing at high temperatures (700–900 °C) have been reported to lead to the formation of the rutile phase [69]. The anatase phase is generally obtained at lower temperatures [69], while deposition at room temperature often leads to the formation of amorphous films [69]. In addition, the energetic bombardment by positively charged ions (facilitated by applying a negative bias voltage on the substrate) can promote the crystallization of TiO2 films at room temperature [70]. When HPPMS is employed, the rutile phase can be achieved even at room temperature and the increase of the flux of energetic species towards the growing films favors the formation of the rutile at the expense of the anatase phase [31, 71-73]. This can be, for instance, achieved by increasing the peak target current [72] or decreasing the working

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

385

pressure during deposition (Figure 8) [71, 73]. In general, when O2 is used as a reactive gas, the oxidized target surface is the source for the generation of negatively charged oxygen ions [74]. These ions are accelerated in the cathode sheath by the negative target potential towards the substrate [75] and it has been shown to have implications on the structure formation and the phase composition of reactively sputtered metal oxide films [76]. In HPPMS target voltages higher than those in dcMS are commonly observed, as shown in the previous sections. This implies that metal oxide films deposited by HPPMS do not only experience the high-flux low-energy bombardment by ionized sputtered species but they are also irradiated by O- species with energies higher than those during dcMS. The trajectory of the O- ions, when targets with no pronounced erosion track are used, is largely directional and perpendicular to the target surface [77]. Therefore, substrates placed perpendicular to the target surface will encounter a significantly suppressed O- flux, as compared to parallel oriented substrates, as well as high positive ion-to-neutral ratios due to an abnormal transport of ions observed in HPPMS discharges [43]. The effect of the substrate orientation on the phase composition is shown in Figure 9. Deposition on substrates facing the target leads to the growth of rutile-rich films (Figure 9) [73]. The deposition on the inclined substrates (Figure 9 (b)) also results in rutile films with a morphology that consists of crystallites embedded in an amorphous matrix [73]. These results outline an alternative pathway for the synthesis of crystalline TiO2 films; in the absence of high energy bombardment (by means of O- ions), i.e. crystalline (rutile-rich) TiO2 films can be deposited when a sufficient flux of low-energy (positive) ion irradiation is available [73].

Figure 8. X-ray diffraction (XRD) patters of TiO2 films grown at room temperature by HPPMS at working pressures of (a) 0.5, (b) 1.0, (c) 2.0, (d) 2.5 and (e) 3.0 Pa. The capitals “A” and “R” designate the position the XRD peaks in unstrained bulk anatase- and rutile-TiO2 phase, respectively. The increase of the working pressure results in a transformation from rutile rich films to film that consist of mixtures from anatase- and rutile phase (data taken from [73]).

386

K. Sarakinos

Figure 9. X-ray diffraction patters of TiO2 films grown at room temperature by and at a working pressure of 0.5 Pa by HPPMS for substrate planed at an angle of (a) O (parallel) and (b) 90° (perpendicular) with respect to the target. The capitals “A” and “R” designate the position the XRD peaks in unstrained bulk anatase- and rutile-TiO2 phase, respectively. In both cases films with a primarily rutile-TiO2 crystal structure are obtained. However, the XRD pattern for the film grown on perpendicularly oriented substrate implies a microstructure that consists of TiO2 crystals embedded in an amorphous matrix (data taken from [73]).

4. SUMMARY High power pulsed magnetron sputtering is a novel ionized physical vapor deposition technique with great potential for improving already existing and developing new functional films. One issue related to HPPMS is the lower film deposition rates as compared to conventional magnetron sputtering techniques (e.g. dcMS) for the same average target power when the discharge operates in a non-reactive mode. This is an intrinsic characteristic of the HPPMS process which can be largely attributed to the re-direction of ionized sputtered species to the target resulting in self-sputtering. Prerequisites for the self-sputtering to occur are the high density of ionized sputtered species and the depletion of Ar species (gas rarefaction) in the target’s vicinity. This implies that target materials with a low sputtering yield and low ionization probability would exhibit a less pronounced loss of the deposition rate. In addition, for a given material the loss of the rate can be suppressed by choosing discharge conditions that correspond to a relatively low degree of ionization, i.e. a compromise between deposition rate and ionization of the sputtered material can be always found depending on the requirements of the specific application. In the case of reactive deposition, operation in HPPMS mode can provide a stable transition zone. This, in turn, facilitates deposition of stoichiometric compound films at target coverage lower and higher rates as compared to dcMS. As far as it concerns film properties, the high fluxes of ionized material in HPPMS enable the room-temperature growth of metal oxide and nitride films with unique microstructure and phase composition

Target-Plasma-Film Interactions in High Power Pulsed Magnetron...

387

5. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32]

Mattox, D. M. (1998). Handbook of Physical Vapor Deposition (PVD) Processing, Noyes Publications, Westwood. Ohring, M. (1992). The Material Science of Thin Films, Academic Press, San Diego. Michely, T. & Krug, J. (2004). Islands Mounts and Atoms, Springer, Berlin. Mochan, B. A. & Demchishin, A.V. (1969). Fiz. Met. Metalloved., 28, 33. Hopwood, J. (2000). Ionized Physical Vapor Deposition, Academic Press, San Diego. Thornton, J. A. & Vac, J. 1986). Sci. Technol., A 4, 3059. Messier, R., Giri, A. P., Joy, A. R. & Vac, J. (1984). Sci. Technol., A 2. Petrov, I., Barna, P. B., Hultman, L., Greene, J. E. & Vac, J. (2003). Sci. Technol., A 21, S117. Grovenor, C. R. M., Hentzell, H. T. G. & Smith, D. A. (1984). Acta Metall., 32, 773. Christou, C., Barber, Z. H. & Vac, J. (2000). Sci. Technol., A 18, 2897. Chapman, B. (1980). Glow Discharge Processes, John Wiley & Sons. Lifshitz, Y., Kasi, S. R. & Rabalais, J. W. (1989). Phys. Rev. Lett., 62, 1290. Robertson, J. (2002). Mat. Sci. Eng., R37, 129. Janssen, G. C. A. M. & Kamminga, J. D. (2004). Appl. Phys. Lett. 85, 3086. Pauleau, Y. (2001). Vacuum, 61, 175. Koch, R. & Phys, J. (1994). C.: Condens. Matter. 6, 9519. Windischmann, H. (1992). Crit. Rev. Solid. Stat. Mater. Sci., 17, 547. Davis, C. A. (1993). Thin Solid Films, 226, 30. Hovsepian, P. (1988). Arch. Metall., 33. Helmersson, U., Lattemann, M., Bohlmark, J., Ehiasarian, A. P. & Gudmundsson, J. T. (2006). Thin Solid Films, 513, 1. Gudmundsson, J. T. (2008). J. Phys. : Conf. Ser., 100, 082013. Christie, D. J. (2005). J. Vac. Sci. Technol., A 23, 330. Mozgrin, D. V. Fetisov, I. K. Khodachenko, G. V. (1995). Plasma Phys. Rep., 21, 401. Bugaev, S. P., Koval, N. N., Sochugov, N. S., & Zakharov, A. N. (1996). XVIIth International on Discharges and Electrical Insulation in Vacuum, p. 1074. Fetisov, I. K., Filippov, A. A., Khodachenko, G. V., Mozgrin, D. V. & Pisarev, A.A. (1999). Vacuum, 53, 133. Kouznetsov, V., Macak, K., Schneider, J. M., Helmersson, U., Petrov, I., (1999). Surf. Coat. Technol, 122, 290. Alami, J., Persson, P. O. Å., Music, D., Gudmundsson, J. T., Bohlmark, J. & Helmersson, U. (2005). Vac, J. Sci. Technol., A 23, 278. Chistyakov, R., Abraham, B. & Sproul, W. D. (2006). Society of Vacuum Coaters 49th Annual Technical Conference Procedings, Washington, DC, p. 88. Ehiasarian, A. P., Münz, W. D., Hultman, L., Helmersson, U., Petrov, I. (2003). Surf. Coat. Technol., 163-164, 267. Alami, J., Eklund, P., Andersson, J. M., Lattemann, M., Wallin, E., Bohlmark, J., Persson P., Helmersson, U., (2007). Thin Solid Films, 515, 3434. Konstantinidis, S. Dauchot, J. P. & Hecq, M. (2006). Thin Solid Films, 515, 1182. Andersson, J. Anders, A. (2008). Appl. Phys. Lett., 92, 221503.

388

K. Sarakinos

[33] Sproul, W. D., Christie, D. J. & Carter, D. C., (2004). Society of Vacuum Coaters 47th Annual Technical Conference Proceedings, Dallas, TX, p. 96. [34] Konstantinidis, S., Dauchot, J. P., Ganciu, M., Ricard, A. & Hecq, M. (2006). Appl. J. Phys., 99 013307. [35] Bobzin, K., Bagcivan, N., Immich, P., Bolz, S., Alami, J., Cremer, R. & Mat. J. (2008). Proc. Technol. [36] Glocker, D. A., Romach, M. M., Christie, D. J. & Sproul, W. D. (2004). Society of Vacuum Coaters 47th Annual Technical Conference Proceedings, Dallas, TX, p. 183. [37] Sarakinos, K., Alami, J., Wuttig, M., Phys. J., (2007). D: Appl. Phys., 40, 2108. [38] Davis, J. A., Sproul, W. D., Christie, D. J., Geisler, M., (2004). Society of Vacuum Coaters 47th Annual Technical Conference Proceedings, Dallas, TX, p. 215. [39] Alami, J., Eklund, P., Emmerlich, J., Wilhelmsson, O., Jansson, U., Högberg, H., Hultman, L., Helmersson, U. (2006). Thin Solid Films, 515, 1731. [40] DeKoven, B. M., Ward, P. R., Weis, R. E., Christie, D. J., Scholl, R. A., Sproul, W. D., Tomasel, F., Anders, A., (2003). Society of Vacuum Coaters 46th Annual Technical Conference Proceedings, San Francisco, CA, p. 158. [41] Konstantinidis, S., Dauchot, J. P., Ganciu, M. & Hecq, M., (2006). Appl. Phys. Lett., 88, 021501. [42] Sevvana, S., Chennadi, S., Lakkaraju, R. K., Li, J. & Rohde, S., (2005). Society of Vacuum Coaters 48th Annual Technical Conference Proceedings, Denver, CO, p. 720. [43] Lundin, D., Larsson, P., Wallin, E., Lattemann, M., Brenning, N., Helmersson, U., (2008). Plasma Sources Sci. Technol., 17, 035021. [44] Vlcek, J., Kudlacek, P., Burcalova, K., Musil, J., Vac. J., (2007). Sci. Technol., A 25, 42. [45] Berisch, R., (1982). Sputtering by Particle Bombardment I, Springer (Berlin). [46] Alami, J., Sarakinos, K., Mark, G., Wuttig, M., (2006). Appl. Phys. Lett., 89, 154104. [47] Ehiasarian, A. P. New, R., Münz, W. D., Hultman, L., Helmersson, U. & Kouznetsov, V., (2002). Vacuum, 65, 147. [48] Emmerlich, J., Mraz, S., Snyders, R., Jiang, K. & Schneider, J. M., (2008). Vacuum, 82, 867. [49] Sarakinos, K., Alami, J., Dukwen, J., Wördenweber, J., Wuttig, M. & Phys. J., (2008). D: Appl. Phys., 41, 215301. [50] Bohlmark, J., Alami, J., Christou, C., Ehiasarian, A. P., Helmersson, U. & Vac, J. (2005). Sci. Technol., A 23, 18. [51] Anders, A., Andersson, J. & Ehiasarian, A. P. J. (2007). Appl. Phys. 102, 113303. [52] Horwat, D., Anders, A. & Phys. J., (2008). D: Appl. Phys. 41, 135210. [53] Rossnagel, S. M., Kaufman, H. R. & Vac, J., (1987). Sci. Technol., A 5, 2276. [54] Depla, D., De Gryse, R. & Vac. J. (2002). Sci. Technol., A 20, 521. [55] Berg, S. & Nyberg, T. (2005). Thin Solid Films, 476, 215. [56] Severin, D., Kappertz, O., Kubart, T., Nyberg, T., Berg, S., Pflug, A., Siemers, M. & Wuttig, M. (2006). Appl. Phys. Lett., 88, 161504. [57] Sproul, W. D., Graham, M. E., Wong, M. S., Lopez, S., Li, D. & Scholl, R.A., (1995). J. Vac. Sci. Technol., A 1188. [58] Wallin, E. & Helmersson, U. (2008). Thin Solid Films, 516, 6398. [59] Sarakinos, K., Alami, J., Klever, C. & Wuttig, M. (2008). Surf. Coat. Technol., 202, 5033.

Target-Plasma-Film Interactions in High Power Pulsed Magnetron... [60] [61] [62] [63] [64] [65] [66]

[67]

[68] [69] [70] [71] [72]

[73] [74] [75] [76] [77]

389

Depla, D. & De Gryse, R. (2004). Surf. Coat. Technol., 183, 184. Barna, P. B. & Adamik, M. (1998). Thin Solid Films, 317, 27. Koehler, J. S. (1970). Phys. Rev., B2, 547 Tellier, C. R. & Tosser, A. J. (1982.). Size Effects in Thin Films, Elsevier, Amsterdam. Munoz, R. C., Finger, R., Arenas, C., Kremer, G. & Moraga, L. (2002). Phys. Rev. B 66, 205401. Van Attekum, P. M., Woerlee, P. H., Verkade, G. C. & Hoeben, A. A. (1984). Phys. Rev., B 29, 645. Chistyakov, R., Abraham, B., Sproul, W. D., Moore, J. & Lin, J. (2007). Society of Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY, p. 139. Paulitsch, J. Mayrhofer, P.H. Mitterer, C. Münz, W.D. Schenkel, M., (2007). Society of Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY, p. 150. Alami, J., Sarakinos, K., Uslu, F., Wuttig, M. &. Phys. J. (2009), D: Appl. Phys., 42, 015304. Wiggins, M. D., Nelson, M. C., Aita, C. R. & Vac. J., (1996). Sci. Technol., A14, 772. Zhou, W., Zhong, X. X., Wu, X. C., Yuan, L. Q., Shu, Q. W., Li, W., Xia, Y. X. & Phys, J., (2007). D.: Appl. Phys., 40, 219. Stranak, V., Quaas, M., Wulff, H., Hubicka, Z., Wrehde, S., Tichy, M., Hippler, R. & Phys. J., (2008). D: Appl. Phys. 41, 055202. Bandorf, R., Vergöhl, M., Giesel, P., Wallendorf, T. & Mark, G. (2007). Society of Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY, p. 160. Alami, J., Sarakinos, K., Uslu, F., Klever, C., Dukwen, J., Wuttig, M. & Phys. J., (2009). D: Appl. Phys., 42, 115204. Tominaga, K. K., Ito, D. & Miyamoto, M. (2006). Vacuum, 80, 654. Mraz, S. & Schneider, J. M. (2006). Appl. J., Phys., 100, 023503. Ngaruiya, J. M., Kappertz, O., Mohamed, S. H. & Wuttig, M., (2004). Appl. Phys. Lett., 85, 748. Severin, D., Kappertz, O., Nyberg, T., Berg, S. & Wuttig, M. (2007). Thin Solid Films 515, 3554.

INDEX

A Abelian, 163 absolute zero, 198 absorption, 8, 9, 10, 14, 18, 19, 20, 37, 56, 58, 68, 69, 71, 72, 78, 79, 81, 84, 93, 97, 106, 157, 163, 164, 173, 247, 311, 345, 366 absorption coefficient, 20, 79 absorption spectra, 8, 56 acceleration, 72, 344 acceptor, 12, 48, 50, 53 accommodation, 144, 145 accuracy, xiii, 21, 77, 176, 353 acetic acid, 305 acetone, 24 acetylene, 87, 93, 142, 370 achievement, 126 acid, 275, 277, 283, 285, 303, 304, 305, 306 acidic, 23 acidity, 366 acoustic, 84 acoustical, 52 activation, 11, 50, 56, 83, 263, 264, 268, 337, 382 activation energy, 11, 263, 264, 268 active site, 315 actuators, 242 ADC, 147, 148 adhesion, 3, 329, 330, 355, 374 adjustment, 177, 201 adsorption, xii, 3, 53, 163, 247, 252, 253, 254, 255, 256, 257, 258, 259, 260, 301, 303, 305, 314, 315, 316 adsorption isotherms, 314, 316 aerobic, 303 aerosol, 199

AFM, xii, 30, 34, 39, 42, 273, 275, 276, 277, 279, 287, 288, 289, 298, 325, 333, 334, 335, 348, 354, 358, 359, 360 Ag, 11, 15, 46, 52, 56, 144, 154, 294, 347 agar, 204, 206 age, 5 agent, 24 agents, 11 aggregates, 216, 307 aid, 82, 103 air, x, 57, 60, 64, 69, 85, 86, 87, 88, 89, 90, 92, 93, 94, 96, 97, 98, 99, 101, 104, 113, 115, 126, 128, 130, 131, 135, 136, 138, 139, 140, 141, 143, 148, 165, 173, 199, 202, 203, 214, 243, 252, 253, 254, 259, 260, 305, 349, 366, 370, 371 air quality, 58 air-dried, 305 airports, 211 airways, 203 alcohol, 369, 370 alcohols, xiii, 365, 368 algae, xii, 301, 303, 305, 307, 308, 309, 310, 311, 312, 313, 314, 315, 316, 317 algorithm, 233 alkaline, 9 alloys, 118, 119, 120 alternative, xiii, 173, 302, 365, 366, 368, 374, 375, 385 alternative energy, xiii, 365 alternatives, 202 aluminium, 61, 354, 355, 356, 357, 358 alveoli, 6 amalgam, 144, 154, 156 ambient pressure, ix, 63, 64, 109 amorphous, 21, 23, 36, 37, 53, 58, 118, 119, 120, 281, 344, 346, 347, 354, 358, 360, 384, 386

392

Index

amplitude, 117, 118, 119, 219, 359, 360, 361, 362, 363 anaerobic, 303 analog, 124 anatase, 384, 385, 386 angular momentum, 76, 80 aniline, 314 animals, 195, 196, 203, 204 anisotropy, 52, 311 annealing, xi, 15, 59, 273, 275, 277, 280, 284, 286, 287, 293, 294, 298, 321, 322, 332, 333, 342, 345, 346, 347, 348, 349, 350, 384 anode, 81, 367 anomalous, 184, 260 anti-bacterial, 5, 9 ants, 205, 208 aqueous solutions, 5 arc plasma, 128 argon, x, 24, 27, 34, 113, 114, 115, 118, 123, 124, 125, 126, 127, 128, 129, 130, 132, 134, 144, 165, 252, 259, 321, 322, 323, 332, 338, 370 argument, 209, 211, 377 Aristotle, 195, 214 arrow of time, 198 arsenic, 50 arsenide, 81 ASD, 111 assignment, ix, 63, 76, 85, 87, 88, 89, 92, 93, 109 assumptions, 93, 254 athletes, 204 atmosphere, 6, 24, 34, 86, 94, 97, 126, 175, 201, 216, 245, 253, 254, 259, 323, 366, 380 atmospheric pressure, x, 69, 81, 113, 114, 115, 126, 129, 130, 140, 165, 368, 369, 370 atomic force microscope, xiii, 353, 354, 358, 359, 363 atomic force microscopy (AFM), 30, 39, 41, 319, 325, 333, 348 atoms, 2, 37, 46, 66, 68, 69, 71, 73, 74, 76, 77, 79, 80, 83, 94, 95, 108, 109, 111, 132, 133, 144, 157, 162, 163, 278, 320, 324, 368, 374, 380 attachment, 69, 70, 71 attacks, 277 Auger electron spectroscopy, 277, 330 automotive sector, 366, 371 availability, 10, 53 averaging, 52, 104, 105

B Bacillus, 316

back, 39, 83, 84, 247, 288, 291, 335 background noise, 360 bacteria, 5, 206, 303 bacterial, 197, 202, 204, 205, 206, 207, 215, 216, 316 band gap, 5, 11, 14, 21, 44, 45, 48, 49, 50 barium, 171 barrier, xi, 2, 132, 180, 245, 246, 247, 248, 250, 251, 252, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 267, 273, 274, 343, 344, 345, 347, 348, 349 barriers, xi, 55, 245, 246, 247, 263 batteries, 9 beams, 10, 37, 71, 84, 348 beating, 203 behavior, xi, xii, 2, 20, 35, 51, 62, 128, 141, 169, 170, 171, 172, 177, 178, 179, 180, 182, 184, 185, 187, 188, 189, 190, 191, 195, 197, 205, 207, 208, 210, 213, 245, 247, 253, 255, 257, 259, 260, 262, 263, 264, 265, 266, 268, 302, 307, 309, 313, 314, 317 behaviours, 106 bending, 80, 267 benign, 57 bias, 4, 51, 116, 248, 249, 251, 384 bifurcation, 202 binding, 2, 38, 48, 49, 94, 142, 303, 306, 314, 315, 316 binding energies, 38 binding energy, 38, 48, 49, 94 biocatalysts, 306 biocompatible, 302, 311 biocompatible materials, 302 biodegradation, 303 bioethanol, 368 biogas, xiv, 365, 368, 371 biologically active compounds, xii, 301, 302, 303, 316, 317 biomass, 315, 316, 366 biosciences, 302 biosensors, 9 biosorption, 315 biotechnology, 302, 304 bipolar, 53 birds, 203, 204, 207, 210 bleaching, 17, 23 blood, xii, 3, 4, 6, 7, 8, 9, 54, 56, 202, 203, 302, 304 blood flow, 54 body fluid, xii, 302, 304 body mass, 203, 211 body size, 203, 204, 215 body temperature, 260

Index body weight, 203 Bohr, 67 Boltzmann constant, 67, 250, 255, 261, 263 Boltzmann distribution, 93 bonding, 2, 12, 45, 57 bonds, 2, 11, 12, 45, 46, 366, 368 boundary conditions, 117, 219, 221 Boussinesq, 232 breakdown, ix, 39, 63, 64, 65, 68, 69, 70, 72, 77, 78, 83, 94, 98, 102, 103, 104 breathing, 203 bremsstrahlung, 69, 72, 84, 95 broadband, 348 Brownian motion, 198 buffer, 123, 144, 306 buildings, 18, 210 bulk materials, xiii, 3, 341, 342, 350 burning, 87 butane, 87, 368

C calculus, 251 calibration, 83, 172, 173, 174, 175, 314 calorimetry, 136 cancer cells, 5 candidates, 266, 347 capacitance, 173 capillary, 56 carbon, xiv, 4, 5, 6, 7, 8, 63, 64, 79, 82, 85, 86, 87, 88, 89, 92, 94, 95, 96, 97, 98, 99, 101, 102, 103, 104, 105, 106, 107, 108, 109, 123, 126, 142, 165, 202, 212, 280, 281, 347, 365, 366, 368, 369 carbon atoms, 95 carbon dioxide, 4, 5, 6, 7, 8, 79, 96, 97, 123, 126, 142, 165, 202, 212 carbon monoxide, 5, 142 carcinogens, 302 Carnot, 196 carrier, 4, 5, 37, 44, 45, 47, 50, 52, 264 catalysis, ix, 1, 2, 3 catalyst, 21, 29, 30, 34, 59 cathode, 81, 83, 267, 367, 374, 377, 385 cation, 3, 44, 47 cavitation, 84 cavities, 243 cell, x, xiii, 6, 17, 47, 82, 113, 142, 144, 291, 292, 306, 311, 319, 324, 326, 327, 328, 331, 332, 338, 367 cell surface, 306

393 ceramic, xi, 59, 172, 173, 175, 176, 188, 190, 245, 246, 247, 255, 256, 259, 262, 263, 265, 334 ceramics, 171, 177, 178, 180, 183, 184, 185, 189, 191, 245, 247, 252, 253, 254, 255, 259, 260, 261, 262, 263, 265 CFD, xi, 217, 218, 219, 220, 221, 222, 223, 224, 225, 229, 230, 231, 233 CH4, 129, 142, 143 channels, 209, 346 charge coupled device, 81 charge density, 10, 248 charged particle, 65, 66, 72, 121 chemical bonds, 2 chemical degradation, 303 chemical energy, 366 chemical industry, xiv, 142, 365 chemical properties, 46, 342, 348 chemical reactions, 5, 66, 124 chemical reactor, 124, 125 chemical vapor deposition, 17, 23, 62, 266 chemical vapour, 49, 50, 51 chemical vapour deposition, 49, 50, 51 chemisorption, 269, 316, 380 chest, 202 circulation, 116, 117, 201, 203, 214, 215 classes, 204, 314 classical, x, 68, 71, 169, 184, 383 cleaning, 5, 58, 214 clinical trial, 54 clinical trials, 54 clusters, 10, 11, 12, 13, 14, 26, 57, 94 CMOS, 274 Co, vii, 113, 116, 118, 120, 132, 135, 142, 149, 159, 166, 193, 214, 270 CO2, vii, ix, xiii, 54, 63, 64, 80, 82, 83, 84, 85, 87, 88, 89, 92, 94, 95, 96, 97, 100, 102, 103, 104, 106, 107, 108, 109, 129, 142, 143, 212, 213, 216, 365, 366, 367, 368, 370, 371 coagulation, 303 coatings, 2, 5, 9, 28, 53, 320, 374 coil, 114, 115, 116, 117, 123, 126, 144, 145, 146 collisions, 64, 69, 70, 74, 79, 94, 95, 104, 108, 109, 324, 343, 345, 368, 374 combined effect, 46, 70, 240 combustion, 93, 366, 367 commons, 211 communication, xiv, 171, 205, 208, 210, 274, 344, 373 communication systems, 171 community, 44 compaction, 346 compensation, 49, 173

394 competition, 381 competitor, 49 complement, 326 complex systems, 209, 210, 211 complexity, 53, 89, 197, 205, 210, 216, 374 compliance, 189, 252 complications, 54 components, 65, 72, 81, 93, 117, 119, 172, 185, 187, 188, 189, 190, 191, 196, 247, 306, 309, 310, 342, 349, 359, 371 composition, xii, 3, 38, 47, 50, 53, 60, 64, 66, 75, 97, 121, 142, 143, 144, 154, 179, 182, 183, 230, 241, 273, 275, 277, 278, 280, 284, 285, 287, 288, 289, 298, 319, 320, 321, 322, 323, 324, 325, 326, 327, 337, 338, 347, 375, 378, 379, 385, 386 compounds, xii, xiv, 39, 46, 115, 142, 301, 302, 303, 304, 314, 316, 317, 320, 322, 326, 365 comprehension, 266 Computational Fluid Dynamics, 242 computational modeling, xi, 245 concentrates, 44 concentration, 37, 44, 45, 47, 50, 51, 52, 53, 70, 123, 124, 127, 129, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 157, 162, 163, 164, 165, 190, 205, 245, 267, 275, 289, 303, 314, 315, 316, 323, 326, 366 conception, 210 conceptualization, 211 condensation, 183 condensed matter, 45 conductance, 201, 246, 247 conduction, 5, 10, 12, 21, 44, 46, 49, 52, 53, 61, 62, 203, 250, 251, 256, 259, 264, 266 conductive, 17, 61, 65, 115, 268, 354, 355, 359 conductivity, xi, 9, 11, 12, 22, 23, 44, 45, 46, 48, 50, 51, 52, 53, 58, 118, 157, 159, 160, 245, 256, 259, 263, 264, 382 conductor, 17, 159, 160, 175 configuration, xi, 12, 17, 44, 46, 67, 80, 82, 172, 173, 189, 195, 196, 197, 198, 199, 200, 201, 202, 203, 205, 206, 209, 210, 213, 214, 260, 330, 343, 348, 350, 354, 375, 378, 379 confinement, 96, 114, 341, 346, 348, 349 conservation, 197 constraints, 197, 198, 202 construction, 9, 114, 115, 174, 344 consumption, 114, 142, 212, 284 contaminants, 383 contamination, 275, 277, 278, 287, 298 continuity, 69, 176, 219, 220

Index control, 2, 3, 13, 17, 34, 45, 46, 58, 71, 83, 146, 147, 171, 175, 177, 205, 206, 208, 219, 320, 342, 344, 375 convection, 198, 199, 201, 215 convective, 115, 127, 128, 132, 198, 201, 202, 203, 215 convergence, 220, 243 convergence criteria, 220 conversion, xiv, 2, 123, 142, 274, 350, 373 cooling, xiii, 18, 83, 102, 108, 124, 132, 133, 134, 135, 136, 138, 139, 140, 141, 142, 165, 175, 215, 252, 253, 307, 319, 321, 322, 326, 359 copper, 45, 65, 81, 282 coral, 205, 207, 215 corona, 66 correlation, 8, 171, 180, 182, 203, 254, 262 correlations, 211, 216 corrosion, 2, 4, 8 cosmetics, 303 costs, xiii, 18, 341, 342 Coulomb, 18, 66, 71, 72 coupling, 76, 185, 187, 188, 189, 191, 192, 233 covalent, 2, 12, 14, 45, 46 covalent bond, 2, 12, 45, 46 covalent bonding, 2 covering, 144 CPD, 21, 22, 23 crack, 30, 41, 215 cracking, 83 CRC, 55, 165 credibility, 234 critical points, 7 cross-sectional, 281, 336 crystal growth, 332, 383 crystal structure, xii, 23, 45, 188, 260, 319, 320, 321, 323, 337, 338, 386 crystal structures, 23, 45 crystalline, 21, 23, 33, 34, 36, 59, 60, 118, 188, 260, 289, 324, 325, 332, 335, 336, 384 crystalline solids, 21 crystallisation, 321 crystallites, 279, 280, 284, 298, 382, 385 crystallization, 384 crystals, xiii, 50, 171, 201, 214, 270, 341, 342, 345, 386 cubic system, 47 cultivation, xii, 302, 304 culture, 209 current limit, xi, 245, 247, 252, 253, 254, 255, 259, 260, 262, 263, 265, 268 cycles, xiv, 23, 188, 360, 361, 373 cycling, 19, 23, 266

Index cyclotron, 72

D damping, 157, 181, 184, 185, 187, 188, 191 data gathering, 146, 147 database, 90 decay, 66, 73, 100, 109, 135, 136 decomposition, 9, 140, 306, 367, 368, 369, 370, 371 defects, 3, 5, 43, 45, 48, 49, 57, 60, 61, 62, 180, 246, 247, 374 deficiency, 52, 58, 257 definition, 191, 374 deformation, 49, 353 degenerate, 44, 80, 93 degradation, 252, 253, 255, 303 dendrimers, 11 density, ix, xiv, 3, 9, 10, 11, 12, 14, 16, 18, 20, 24, 30, 35, 43, 45, 46, 52, 63, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 78, 79, 82, 87, 91, 94, 96, 101, 102, 107, 108, 109, 117, 121, 127, 132, 144, 145, 149, 198, 199, 203, 204, 207, 208, 209, 234, 241, 248, 249, 250, 251, 252, 254, 255, 256, 257, 258, 263, 343, 344, 368, 373, 374, 375, 381, 382, 386 deposition, xii, xiv, 11, 17, 21, 23, 24, 28, 30, 34, 49, 50, 51, 53, 57, 60, 62, 64, 115, 266, 275, 280, 285, 287, 319, 320, 321, 322, 324, 326, 327, 329, 332, 337, 338, 345, 355, 373, 374, 375, 376, 377, 378, 379, 380, 381, 384, 386 deposition rate, xiv, 321, 373, 375, 376, 377, 378, 379, 380, 381, 386 derivatives, 304 desorption, 3, 254, 315 destruction, 129, 371 detection, 9, 10, 15, 75, 77, 79, 81, 83, 90, 100 deviation, 179, 207, 246, 254 diaphragm, 82, 84, 123, 148, 367 dielectric constant, 49, 261, 346 dielectric materials, 169 dielectric permittivity, x, 169, 170, 171, 172, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 186, 187, 251, 256 dielectric strength, 181, 184, 185 differential equations, 231 diffraction, xii, 10, 27, 36, 50, 78, 82, 86, 273, 274, 280, 281, 282, 283, 284, 289, 291, 293, 298, 319, 321, 322, 324, 325, 327, 336, 385, 386 diffusion, 4, 6, 23, 50, 70, 71, 127, 140, 198, 199, 202, 205, 208, 209, 277, 278, 326, 382 diffusion process, 140, 382

395 diffusion region, 205 diffusivity, 287 diodes, 49, 62 dipole, 190, 199, 307, 309, 311, 314 dipole moment, 199 discharges, x, 113, 114, 115, 118, 121, 122, 126, 127, 143, 144, 146, 149, 150, 151, 152, 153, 154, 157, 158, 162, 165, 377, 378, 385 discretization, 233 discs, 175 diseases, 6 disinfection, 143 disorder, 171, 179, 183 dispersion, x, 2, 10, 169, 170, 171, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191 displacement, xiii, 353, 354, 356, 359, 360, 361, 362, 363 dissociation, 13, 132 dissolved oxygen, 7, 8 distillation, 371 distilled water, 21, 314 distress, 54 distribution, 10, 24, 25, 34, 46, 52, 66, 71, 73, 93, 130, 136, 140, 141, 162, 163, 179, 197, 201, 210, 218, 219, 220, 230, 237, 238, 240, 304, 307, 313, 314, 315, 347, 348, 363 distribution function, 179, 218, 219, 220, 304 divergence, 78, 82, 84, 136, 138, 161, 179 diversity, 343 DNA, 7, 302 domain structure, 182 domain walls, 171, 178, 182, 183 domestic policy, 366 donor, 49, 50, 53, 248, 251, 255, 256, 260, 263 donors, 48, 49, 246, 248 dopant, 44, 50, 53, 323 dopants, 49, 50, 247, 320, 337 doped, xii, 5, 6, 44, 45, 47, 50, 51, 53, 57, 60, 61, 171, 260, 319, 320, 337, 347, 348, 349, 350, 354, 355, 356 doping, 44, 47, 48, 49, 50, 51, 52, 61, 62 Doppler, 72, 73, 94, 163 drainage, 201, 202 drugs, 302 drying, 306 durability, 17, 19, 23 duration, 73, 77, 78, 82, 84, 107, 256, 275, 285, 329, 356, 375 dyes, xii, 13, 302, 303, 314, 315, 316, 317 dynamical properties, 64

396

Index

E ears, 196 earth, 66, 201, 347, 349 EBSD, 291, 292, 298 ecological, 315 econometric analysis, 216 economic activity, 366 economic growth, 213 economics, 214, 216 education, 241 elasticity, 190 election, 344 electric arc, 87, 118 electric charge, 65, 199 electric conductivity, 159 electric current, 366 electric energy, 367 electric field, x, 5, 9, 10, 68, 69, 73, 74, 77, 78, 86, 113, 116, 117, 118, 126, 127, 128, 129, 144, 145, 149, 150, 151, 157, 159, 162, 163, 175, 178, 185, 188, 189, 190, 191, 259, 264, 266, 267, 268, 353, 356, 358, 359, 360, 361, 362, 363 electric potential, 16, 358 electric power, 154, 156 electrical conductivity, xi, 48, 245, 256, 382 electrical power, 213 electrical properties, 65, 245, 262, 266 electrical resistance, 261 electricity, 196 electrodes, 13, 17, 18, 50, 118, 129, 157, 175, 266, 267, 326, 355, 358, 359, 363, 367 electroluminescence, 51 electrolysis, 367 electrolyte, 17, 21, 23 electromagnetic, 10, 13, 53, 65, 68, 114 electromagnetic fields, 65 electromagnetic wave, 68 electron beam, xii, 16, 17, 34, 35, 37, 38, 41, 45, 46, 47, 61, 273, 275, 277, 285, 298, 355 electron charge, 20, 68 electron density, 65, 66, 67, 68, 69, 70, 94, 102, 107, 108, 109 electron diffraction, 34, 283, 289, 336 electron gas, 10 electron microscopy, xii, 24, 27, 31, 32, 39, 40, 252, 280, 281, 289, 301, 303, 329, 334, 336 electronic materials, 60 electronic structure, 12, 48, 57 electrons, 5, 10, 12, 16, 21, 35, 37, 39, 43, 45, 46, 47, 48, 52, 65, 66, 67, 68, 69, 70, 71, 79, 81, 94, 95, 98, 107, 108, 109, 127, 246, 247, 248,

249, 250, 251, 253, 254, 255, 256, 257, 259, 262, 267, 368, 378 elongation, 201 emission, ix, 10, 12, 23, 42, 51, 57, 58, 63, 64, 65, 69, 71, 72, 73, 74, 75, 76, 78, 79, 81, 82, 83, 85, 86, 87, 88, 89, 90, 92, 93, 94, 96, 97, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 150, 151, 152, 153, 154, 156, 157, 165, 212, 213, 216, 250, 254, 369, 371, 378 emission source, ix, 63, 109 emitters, 35, 43, 73, 109, 143 encapsulation, 11 endocrine, 205 endocrine system, 205 endothermic, 367, 368 energy characteristics, 147 energy density, 65, 79 energy supply, 366 energy transfer, 13, 16, 80 engines, 366, 367 enlargement, 131 entrapment, 268 entropy, 198 environment, xiv, 71, 204, 206, 253, 303, 323, 365 environmental conditions, 204, 205 environmental contaminants, 302, 317 environmental contamination, 303 environmental protection, 5 environmental technology, 302, 304 enzymes, xii, 302, 303, 306, 316 epitaxy, 49, 51, 61 epoxy, 280, 355 EPR, 307 equilibrium, ii, 45, 66, 67, 73, 75, 76, 77, 95, 115, 132, 133, 136, 137, 138, 139, 140, 196, 198, 199, 211, 248, 251, 254, 255, 256, 257, 258, 263, 314, 315, 316, 342, 345, 366, 374 equilibrium state, 139, 196, 198 erbium, 347, 348 erosion, 376, 377, 378, 382, 385 ESR, xii, 301, 303, 310, 311, 312, 313, 314 ESR spectra, 310, 311, 312 ESR spectroscopy, xii, 301, 303 estimating, 76 etching, 115, 275, 277, 344 ethane, 367, 368 ethanol, 368 evacuation, 123 evaporation, xii, 11, 17, 23, 34, 46, 47, 57, 59, 60, 61, 78, 79, 273, 275, 277, 285, 355

397

Index evolution, 10, 64, 66, 68, 97, 102, 105, 107, 108, 109, 184, 196, 199, 200, 201, 206, 207, 209, 215, 267, 293, 298 evolutionary process, 198, 211, 213 excitation, 16, 64, 65, 69, 75, 79, 80, 89, 94, 100, 107, 108, 151, 152, 157, 162, 309, 310 expansions, 173 experimental condition, 11, 52, 89, 122, 147, 161, 287 exposure, 382 external environment, 253 extinction, 37, 38, 132 extraction, 16, 19, 37, 307, 309 eye, 213 eyes, 196

F fabricate, 50, 345, 348, 349 fabrication, xiii, 59, 341, 342, 344, 350 failure, 4, 54 faults, 174 feedback, 360, 380 fermentation, 317 Fermi level, 21, 22, 248, 254, 264, 267 ferrite, 118, 124, 125, 126, 131 ferroelectrics, 170, 171, 172, 173, 177, 182, 183, 188, 191 ferrofluids, xii, 301, 302, 304, 316 ferromagnetic, 114, 116, 118, 144, 145 Feynman, 193, 213, 216 FIB, 344 fiber, 54, 346, 348, 370, 371 fiber membranes, 54 fibers, 54, 342, 346, 349 field-emission, 58 filament, 46 film formation, 284 film thickness, 18, 20, 53, 218, 219, 220, 221, 222, 223, 224, 225, 228, 229, 240, 241, 289, 356, 358, 383 filters, 142, 143 filtration, 302 financial support, 192, 350 finite volume, 218, 225, 226, 228 finite volume method, 218, 225, 226, 228 flame, 87, 93 flexibility, 349 flight, x, 63, 105, 210 flocculation, 303 flow field, 242

flow rate, 11, 46, 48, 54, 123, 126, 127, 128, 129, 130, 131, 132, 137, 138, 139, 140, 142, 143, 200 fluctuations, 77, 78, 128, 132 FLUENT, 219, 233 fluid, xi, 198, 203, 217, 220, 222, 223, 224, 225, 227, 228, 233, 240, 241, 243, 304, 305, 311 fluorescence, ix, 1, 3, 8, 9, 10, 12, 15, 16, 57, 65, 79, 350 fluorescent lamps, 65 focused ion beam, 344, 345 focusing, 65, 77, 81, 100 foils, 23, 60 food, 196, 205, 303, 315 forests, 215 fossil fuel, xiii, 213, 365, 366 fossil fuels, xiii, 365, 366 fractals, 196 free energy, 72 free radicals, 313 freedom, 69 friction, 202, 204, 218, 221, 222, 223, 224, 225, 230, 236, 237, 239, 240 fuel, xiii, 3, 4, 16, 142, 213, 365, 366, 371 fuel cell, 16, 142, 366, 367 fulfillment, 172 funding, 241 fungi, 303 furnaces, 120 fusion, 110 FWHM, 73, 74, 77, 82, 179, 186

G G8, 213 GaAs, 81 gallium, 81 gas diffusion, 4 gas exchange, 6, 54 gas phase, 100, 101, 108 gas sensors, 16, 59 gases, x, 7, 76, 113, 122, 126, 129, 368, 371 gauge, 82, 355 Gaussian, 73, 78, 218, 219, 220 GDP, 212, 213 gel, 12, 17, 51, 60, 61, 62, 204, 262, 266 General Relativity, ii generation, xi, xiv, 3, 6, 14, 44, 56, 68, 69, 70, 114, 115, 116, 118, 121, 122, 142, 163, 195, 198, 199, 200, 203, 213, 214, 216, 232, 240, 350, 366, 367, 373, 374, 385 generators, x, 113, 114, 165 genetic code, 201

398

Index

germanium, 81 glass, xiii, 5, 11, 16, 17, 18, 24, 34, 37, 51, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 355 glasses, xiii, 341, 342, 344, 345, 346, 347, 348, 349, 350 global warming, 216, 366 goals, 3, 211 gold, xiii, 10, 21, 50, 65, 81, 319, 326, 327, 328, 329, 331, 338, 355, 356 GPO, 273, 319, 353 grades, 120 grain, xi, 171, 201, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 277, 279, 280, 284, 289, 291, 292, 321, 325, 326, 327, 329, 332, 333, 335, 337, 363, 383 grain boundaries, xi, 245, 246, 254, 256, 260, 262, 280, 289 grains, 184, 245, 246, 247, 248, 262, 263, 277, 280, 289, 290, 298, 325, 326, 333, 334, 335, 336, 382 graph, 94 graphite, ix, 63, 64, 82, 83, 86, 89, 93, 94, 99, 100, 101, 102, 103, 104, 105, 106, 108, 109 gratings, 83 gravity, 203 greenhouse, xiv, 365, 366, 367, 371 grids, 220 gross domestic product, 213 groups, 118, 211, 213, 235, 317, 375, 376 growth, xiii, 2, 3, 10, 11, 14, 17, 19, 27, 28, 29, 30, 34, 39, 45, 46, 50, 57, 59, 61, 62, 68, 69, 70, 71, 106, 188, 196, 197, 204, 205, 206, 210, 213, 215, 254, 284, 319, 321, 327, 329, 332, 374, 375, 380, 382, 383, 385, 386 growth mechanism, 34 growth temperature, 374, 382

H H2, xiv, 131, 143, 365, 367, 368, 370 HA, 339 haemoglobin, 3, 4, 6, 7, 8, 56 halogen, 83 hardness, 204, 206 harm, 203 harmony, 203 harvest, 205 health, 4, 315 health care, 4 health problems, 315 heart, 4, 6, 203

heart disease, 6 heartbeat, 203 heat, x, 80, 113, 116, 118, 119, 120, 122, 123, 124, 127, 131, 132, 134, 136, 139, 141, 142, 158, 196, 197, 198, 203, 215, 243, 253, 255, 367 heat capacity, 81 heat loss, x, 113, 118, 119, 120, 127, 131, 132, 158, 203, 215 heat transfer, 127, 203, 215 heating, 18, 21, 29, 31, 35, 114, 120, 132, 175, 260, 261, 275, 294, 297, 330, 332, 374 heavy metal, xii, 302, 303, 315, 317 heavy particle, 108, 368 height, 30, 42, 66, 122, 179, 201, 218, 219, 222, 223, 224, 225, 226, 230, 233, 240, 241, 246, 247, 250, 251, 252, 254, 256, 257, 258, 259, 260, 262, 263, 264, 267, 309, 359 height growth, 254 heterogeneous, 2, 6 heterogeneous catalysis, 2 high pressure, 96, 97, 150, 152, 158, 229 high resolution, 289, 298 high temperature, 49, 66, 84, 94, 111, 137, 138, 173, 252, 274, 287, 321, 332, 368, 384 high-speed, 170 high-tech, x, 113 hip, 81, 210, 212 Hm, 118 homogenous, 278, 287 host, 348 hostile environment, 204 household, 9 HRTEM, 289, 291, 298 human, 3, 6, 7, 8, 9, 54, 56, 210, 216, 366 human behavior, 210 humans, 204, 210 humidity, 6, 60, 241, 259 hybrid, 16 hybrids, 185 hydrides, 367 hydro, xiii, 55, 365, 367, 368 hydrocarbon, 93, 280, 367, 368 hydrocarbons, xiii, 55, 365, 367, 368 hydrochloric acid, 277, 283 hydrodynamic, 218, 219, 221, 222, 223, 224, 241, 242 hydrodynamics, 111, 215, 216 hydrofluoric acid, 275, 277, 285 hydrogen, x, xiii, 3, 4, 6, 8, 15, 67, 74, 79, 86, 89, 113, 126, 131, 132, 142, 165, 306, 365, 366, 367, 368, 369, 370, 371 hydrogen peroxide, 306

Index hydrophilicity, 55 hydrosphere, 201 hydroxide, 304, 306 hypothesis, 152, 165 hysteresis, 118, 119, 188, 189, 307, 308, 380 hysteresis loop, 189, 308

I IBIS, 356 ICU, 56 id, 62, 329 identification, 49, 274, 337 illumination, 143, 145, 146, 147, 284 images, 42, 81, 278, 281, 284, 289, 298, 305, 306, 333, 335, 354, 383 imaging, xiii, 9, 284, 289, 353, 359, 363 implants, 4 implementation, xiv, 134, 370, 373 impurities, 3, 180, 246, 275, 369, 370 IMS, 277 in situ, xii, 54, 273, 293, 297, 298, 332 in transition, 59 in vitro, 4, 7 in vivo, 4 incentive, xiii, 365, 374 incidence, 102, 105, 242, 282 income, 210 income distribution, 210 incompressible, 233, 243 incubation, 311, 314 incubation time, 314 independence, 234 indexing, 291, 292, 293 indication, 29, 106, 309 indices, 348 indium, 16, 44, 45, 46, 47, 48, 60, 61 indium tin oxide (ITO), 16 induction, x, 113, 114, 116, 117, 118, 119, 120, 141, 143, 146, 147, 156, 157, 163, 164, 165 inductor, 114, 120, 143 industrial, ix, 2, 4, 142, 162, 230, 367, 370 industrialized countries, 213 industry, x, xiv, 2, 44, 65, 113, 142, 143, 274, 302, 365, 366 inelastic, 69, 95 inert, 122, 245, 323, 326, 330, 368, 374 inertness, 346 infinite, 16 infrared, ix, 56, 63, 64, 80, 81 injection, 16, 21, 266, 267 inorganic, xii, 16, 302, 317 insects, 204, 205, 208

399 insertion, 19, 83 insight, 65, 89, 104, 310 inspection, 179, 186 inspiration, 210 instabilities, 66, 132 instability, 115, 127, 128, 132, 366 instruments, 81 insulation, 215 insulators, 44, 247 integration, 71, 72, 83, 95 interaction, 5, 16, 64, 73, 74, 77, 95, 150, 182, 190, 207, 262, 277, 311, 356, 378 interactions, xiv, 13, 64, 71, 96, 182, 307, 309, 311, 314, 373, 375, 376 intercalation, 16, 17, 18, 19, 20, 21, 22, 23, 37 interface, xi, 3, 6, 50, 175, 217, 218, 219, 226, 241, 246, 267, 274, 275, 278, 281, 287, 289, 291, 298, 374, 375, 382 interference, 148, 290 internal combustion, 366, 367 international relations, 366 interpersonal contact, 209 interphase, 182 interstitial, 48 interstitials, 48 interval, 83, 104, 106, 175, 177, 178, 182, 186, 203, 309, 368 intravascular, 54 intrinsic, 45, 48, 51, 52, 72, 171, 386 ion beam, 344, 345, 346 ion bombardment, 383 ion implantation, xiii, 341, 342, 344, 345, 346, 347, 348, 349, 350 ionic, 2, 12, 46, 64, 65, 67, 68, 72, 74, 75, 79, 86, 87, 92, 95, 97, 100, 102, 103, 104, 105, 106, 107, 108, 109, 303, 304 ionization, x, xiv, 46, 63, 64, 67, 68, 69, 70, 71, 72, 76, 77, 79, 94, 95, 109, 250, 251, 256, 373, 374, 375, 377, 378, 379, 380, 384, 386 ionization energy, 69, 70, 250, 251, 256 ionization potentials, 46, 71, 94 ionosphere, 66 ions, xii, 16, 21, 35, 47, 65, 66, 67, 68, 69, 70, 71, 73, 77, 94, 95, 107, 108, 109, 171, 302, 303, 314, 315, 317, 342, 343, 344, 346, 347, 348, 349, 350, 368, 374, 376, 377, 378, 379, 384 IR spectra, 160, 162 iron, xii, 4, 301, 304, 307, 316 irradiation, 34, 35, 56, 82, 103, 145, 344, 348, 383, 385 island, 24, 25 isothermal, 141 isotherms, 314, 316

400

Index

isotropic, 225 isotropy, 231 iteration, 220 ITO, 16, 17, 21, 24, 44, 53

J Joule heating, 260 judo, 204

K K+, 347 Kant, 214 kidneys, 202 kinetic energy, 10, 68, 69, 79, 108, 231, 241, 243 kinetics, 2, 3, 4, 10, 11, 14, 17, 23, 134, 135, 374

L L1, 147, 148 L2, 117, 147, 148 LA, 56, 57, 85, 88, 92, 96, 98, 192 lamina, xi, 217, 219, 220, 225, 229, 230, 235, 240, 241 laminar, xi, 217, 218, 219, 220, 225, 229, 230, 235, 240, 241 land, 200 landfills, xiv, 365 Langmuir, 315, 316 lanthanum, 320 laser, ix, xiii, 11, 49, 50, 56, 57, 62, 63, 64, 65, 68, 69, 70, 71, 72, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 94, 95, 96, 97, 100, 102, 103, 104, 105, 106, 107, 108, 109, 114, 298, 341, 342, 346, 349, 354 laser ablation, ix, 17, 63, 64, 82, 100, 102, 104, 105, 108 Laser Induced Breakdown Spectroscopy (LIBS), 63, 64, 65, 71, 74, 77, 78, 79, 81, 82, 84, 102, 109 laser radiation, 64 lasers, 35, 65, 79, 80, 106, 114, 347 lattice, 5, 16, 24, 34, 45, 47, 49, 50, 171, 246, 283, 289, 294, 326, 328, 332, 374 leakage, 266, 267 LED, 49, 51 lens, 78, 82, 100, 146, 147 lenses, 53, 148 lifetime, 73 ligands, 11 light beam, 81

light scattering, 10 light-emitting diodes, 62 lighting systems, 143 limitation, 108 limitations, xiii, 3, 7, 28, 142, 173, 353, 356 linear, xiii, 64, 73, 74, 80, 90, 91, 148, 205, 239, 252, 253, 262, 341, 356, 358, 377 linear dependence, 377 liquid nitrogen, 175 lithium, 37, 50, 58 liver, 4 loading, 121 localization, 12, 45, 46 location, 30, 234 locomotion, 203, 204 losses, x, 71, 79, 82, 113, 115, 118, 119, 120, 127, 131, 132, 158, 178, 184, 203, 345, 346, 347, 348 low power, 68, 342 low temperatures, 10, 51, 53, 74, 170, 175, 309, 313 low-temperature, x, 61, 113, 114, 117, 121, 135, 140, 141, 142, 165, 349 lubrication, xi, 217, 218, 219, 220, 221, 222, 224, 225, 241, 242 lumen, 54 luminescence, 11, 23, 35, 42, 43, 60 lung, ix, 1, 3, 4, 6, 7, 54, 56, 202 lung disease, 4, 6 lung function, 6 lungs, 4, 6, 54, 197, 202, 215 lysozyme, 316

M macromolecules, 305 maghemite, xii, 301, 302, 303, 304, 305, 307, 311, 313, 314, 316, 317 magnetic, vi, xii, 2, 46, 51, 52, 61, 72, 114, 116, 117, 118, 119, 120, 122, 123, 124, 126, 127, 129, 146, 160, 241, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 314, 316, 317, 374 magnetic field, 46, 51, 52, 72, 117, 118, 119, 120, 302, 303, 304, 305, 307, 308, 310, 311, 316, 374 magnetic fluids, 303, 304 magnetic materials, 2, 118 magnetic particles, 304, 305 magnetic properties, xii, 2, 61, 118, 301, 302, 303, 307, 316, 317 magnetic relaxation, 309 magnetic structure, 310

Index magnetism, 2, 196, 303, 307, 316 magnetization, xii, 301, 303, 307, 308, 313, 314 magnetoresistance, 55 magnetron sputtering, xii, xiv, 24, 27, 34, 46, 50, 58, 60, 273, 275, 285, 319, 320, 321, 373, 374, 375, 383, 386 magnets, xii, 2, 301, 305, 306, 317 mainstream, 201, 375 maintenance, 4, 80, 118, 127, 210 mammals, 203, 215 management, 4 manifold, 196 manipulation, 302 man-made, 65 manners, 148 mapping, 354, 357 market, 342 mask, 171, 274, 345, 348, 355 masking, 345, 346, 355 mass spectrometry, 50, 277 material surface, 15, 342 mathematicians, 196 matrices, 194 matrix, 24, 83, 171, 182, 385, 386 MBE, 49 measurement, xiii, 8, 12, 21, 30, 45, 46, 52, 53, 78, 107, 124, 126, 144, 147, 148, 153, 163, 171, 173, 175, 184, 185, 189, 191, 234, 253, 260, 267, 268, 293, 294, 296, 297, 314, 353, 354, 356, 358, 359 measures, 219 mechanical stress, 188, 191 media, xii, 118, 201, 215, 302, 304, 317 melt, 94 melting, 307, 309, 382 melting temperature, 382 membranes, 4, 54 memory, 14, 16, 35, 320 MEMS, 218, 274 mercury, x, 114, 115, 143, 144, 145, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 165, 314 mercury, 149, 167 metabolic, 54, 203, 211, 212, 215 metabolic rate, 203, 211, 215 metabolism, 211, 213, 215 metal ions, 316 metal organic chemical vapor deposition, 62 metal oxide, ix, 1, 2, 3, 4, 11, 17, 19, 21, 44, 46, 59, 380, 385, 386 metal oxides, ix, 1, 2, 4, 19, 21, 44, 46, 380 metallurgy, x, 113 metals, 2, 10, 49, 56, 287, 295, 315, 329, 378

401 methane, xiv, 142, 365, 367, 368, 371 methanol, 305, 306, 368, 369, 370 microbial, xii, 301, 302, 303, 305, 306, 307, 309, 311, 313, 314, 315 microbial cells, xii, 301, 302, 303, 305, 306, 307, 313, 314, 315 microelectronics, x, 2, 113 microorganism, 303 microorganisms, 303 microscope, xiii, 305, 311, 353, 354, 359, 363 microscopy, xii, 24, 27, 30, 32, 39, 40, 41, 42, 60, 252, 273, 274, 277, 280, 281, 284, 289, 298, 301, 303, 319, 325, 329, 334, 336, 348, 354, 356, 358 microstructure, xiii, xiv, 50, 171, 319, 321, 337, 373, 375, 382, 384, 386 microstructures, 30, 382, 384 microwave, x, 68, 113, 169, 170, 171, 172, 173, 176, 178, 182, 183, 187, 191, 311, 369 migration, 267 military, 80 mitochondria, 215 mobility, 44, 45, 47, 48, 50, 51, 52, 53, 61 modeling, xi, 115, 142, 241, 242, 245 models, x, xi, 54, 118, 144, 157, 158, 169, 171, 197, 215, 225, 226, 231, 232, 233, 242, 243, 245, 247, 259, 262, 266, 342, 382 modulation, 11, 12, 18, 19, 44, 289, 311, 347 modules, 349 modulus, 172 mole, 136, 367, 368 molecular beam, 49, 61 molecular beam epitaxy, 49, 61 molecular biology, 202 molecular oxygen, 132, 133 molecular weight, 302 molecular-beam, 51 molecules, 5, 10, 13, 66, 68, 69, 71, 73, 80, 94, 108, 109, 111, 132, 140, 201, 215, 259, 368, 369, 371, 381 molybdenum, 81 momentum, 52, 76, 80, 219, 233, 243 monochromator, 81, 146 monolayer, 315 monolayers, 283 morbidity, 4 morphological, 17, 35, 43, 383 morphology, xii, 3, 10, 24, 25, 29, 30, 31, 39, 41, 60, 204, 260, 273, 277, 287, 289, 298, 323, 325, 326, 382, 383, 385 mortality, 4 motion, x, 73, 169, 171, 178, 180, 182, 196, 198, 207, 209, 231

402

Index

motivation, xii, 319 movement, 21, 196, 206, 208 MPI, 68, 69, 71, 95 MRS, 62, 351 multidisciplinary, 320, 371 multiplication, 81 multiplier, 146, 147 mutagenic, 303

N NA, 193, 194 Na+, 16, 347 NaCl, 81, 82 nanocrystalline, 59, 327, 332, 333, 337 nanocrystals, 335, 336 nanodevices, 274 nanomaterials, 27 nanometer, 13 nanometers, 43 nanoparticles, xii, 3, 5, 10, 15, 34, 42, 43, 56, 57, 59, 301, 302, 303, 304, 305, 306, 307, 311, 313, 316, 317 nanorods, 23, 24, 30, 34, 43, 58 nanoscale materials, 36 nanostructured materials, 53 nanostructures, 3, 23, 58, 274 nanotechnology, 3, 274 nanotube, 23, 35 nanotubes, 23, 43 nanowires, 15, 21, 23, 24, 25, 27, 28, 29, 30, 34, 35, 43, 58, 59 NASA, 230 National Academy of Sciences, 214 natural, xi, xiii, 66, 69, 73, 123, 126, 142, 165, 172, 195, 196, 197, 198, 215, 365, 366, 368, 371 natural gas, 123, 126, 142, 165, 368, 371 natural science, 198 natural sciences, 198 natural selection, 196 Navier-Stokes equation, xi, 217, 218, 219, 225, 228, 232, 234 Nd, 11, 347 negativity, 46 neon, x, 114, 115, 144, 146, 147, 159, 160, 161, 162, 163, 164, 165 nervous system, 203 nickel (Ni),, xi, 67, 77, 79, 94, 95, 273, 274, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 348, 351 niobium, 320 NIR, ix, 63, 86, 109

NIST, 72, 75, 87, 89, 90, 111 nitride, xiv, 50, 115, 277, 354, 359, 373, 375, 386 nitrogen, 50, 104, 115, 135, 136, 138, 139, 140, 141, 165, 175 noble metals, 10 noise, 45, 81, 83, 108, 294, 360 nonequilibrium, 197, 214 nonequilibrium systems, 197 nonlinearities, 342 non-Newtonian fluid, 241 nonstoichiometric, 39 non-thermal, 95 non-uniform, 18 non-uniformity, 18 normal, 39, 80, 83, 147, 182, 188, 191, 234, 235, 304, 359 normal distribution, 304 n-type, xi, 12, 49, 50, 51, 246, 247, 273, 274, 275, 284, 285, 293, 294, 295, 296, 298 nuclear, 46, 213, 343, 345, 346 nuclear charge, 46 nuclear power, 213 nuclear power plant, 213 nucleation, 3, 10, 14, 34, 182, 382 nucleic acid, 302 numerical analysis, 242 nutrient, 204, 205, 206 nutrients, 204, 205, 206

O observations, 9, 84, 124, 207, 209 observed behavior, 170, 180, 189, 307 oceans, 366 Ohmic, 247, 248, 262, 266 oil, xiii, 114, 117, 289, 365, 366 oils, 303 oligonucleotides, 302 one dimension, 23, 342 ophthalmic, 53 optical, ix, xiii, 1, 3, 5, 7, 8, 9, 11, 13, 14, 16, 17, 18, 19, 20, 35, 43, 45, 47, 48, 49, 50, 56, 57, 58, 60, 61, 69, 76, 78, 79, 81, 82, 83, 105, 108, 143, 145, 157, 170, 293, 305, 341, 342, 343, 344, 345, 346, 347, 348, 350, 374, 378 optical absorption coefficient, 20 optical coatings, 5, 9 optical density, 17 optical detectors, 145 optical fiber, 342 optical properties, 11, 20, 47, 48, 56, 57, 58, 61, 342 optical pulses, 79

Index optical transmission, 50 optics, 9, 10, 82 optimization, 35, 43, 45, 196, 197, 198, 199, 200, 201, 214, 371 optimization method, 45 optoelectronic, 44, 48, 49 optoelectronic devices, 44, 48, 49 orbit, 69 ores, 120, 124, 125, 126, 131, 144 organ, 4, 6, 23, 208 organic, xii, xiii, 5, 13, 16, 51, 62, 302, 303, 314, 316, 317, 341, 342, 365 organic compounds, xiv, 303, 314, 365 organic polymers, 342 organism, 210 organometallic, 23 orientation, x, xii, 25, 36, 169, 170, 185, 187, 188, 189, 208, 225, 273, 283, 284, 289, 290, 291, 292, 293, 295, 297, 298, 322, 324, 325, 326, 327, 329, 330, 331, 332, 336, 361, 382, 385 orientation state, 188 orthorhombic, 281, 293 oscillation, 68 oscillations, 10, 80, 309 oscillator, 71, 181 oxidation, 2, 4, 5, 7, 10, 11, 14, 23, 39, 50, 59, 62, 303, 382 oxide, ix, xi, xiii, xiv, 1, 2, 3, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 21, 22, 23, 24, 27, 30, 34, 35, 36, 37, 38, 40, 43, 44, 45, 46, 47, 48, 53, 56, 57, 58, 59, 60, 61, 135, 136, 245, 246, 247, 259, 263, 274, 277, 278, 280, 281, 283, 285, 298, 304, 307, 320, 337, 341, 342, 345, 346, 349, 350, 373, 375, 385 oxide clusters, 11 oxide nanoparticles, 43, 57, 59, 304 oxides, ix, xii, 1, 2, 3, 4, 10, 11, 12, 16, 19, 21, 44, 46, 52, 53, 301, 316, 319, 320, 380 oxygen plasma, 51, 58, 62 oxygenation, 4, 6, 7, 8, 202 ozone, 123, 132, 133, 134, 135, 165, 166

P pacing, 24, 283 paints, 5 parabolic, 81, 243 paramagnetic, 313 parameter, 67, 74, 107, 148, 158, 179, 183, 188, 225, 230, 237, 309, 315 Pareto, 210, 216 particle collisions, 95

403 particles, 5, 6, 10, 14, 25, 26, 56, 57, 64, 65, 66, 71, 72, 73, 74, 100, 121, 142, 199, 304, 305, 307, 309, 311, 313, 314, 315, 368, 376 partition, 75 passive, 342, 349, 350 patents, 115 patients, 7, 54 patterning, 355 Pb, 266, 267, 320, 323, 324 pedestrian, 207, 208, 209, 214, 216 pedestrians, 196, 202, 207, 208, 209 peptide, 11 periodic, xi, 16, 46, 57, 148, 217, 231, 233, 235, 237, 383 periodic table, 46 periodicity, 21 permeability, 198 permit, 66 permittivity, x, 169, 170, 171, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 186, 187, 191, 199, 248, 251, 255, 256, 260, 261, 263 perovskite, x, xii, 169, 171, 172, 177, 179, 180, 182, 260, 319, 320, 321, 322, 324, 326, 327, 329, 330, 332, 336, 337, 338 peroxide, 14, 57, 306 perturbation, 366 Petri dish, 206 petroleum, 242 PF, 111 pH, 7, 8, 303 phase transitions, 170, 211, 260 pheromone, 205, 215 phonon, 49, 52, 294 phonons, 293 phosphate, 343, 347, 348, 350 phosphate glasses, 347 phosphorous, 50 photocatalysis, 4, 5, 6, 8 photocatalysts, 3, 4, 5, 56 photocorrosion, 7 photodegradation, 303 photoelectron spectroscopy, xii, 319, 323, 327 photoexcitation, 12 photographs, 42 photoionization, 72, 79 photoluminescence, 13, 42, 56, 57, 59, 60 photon, 14, 68, 69, 71, 72, 73, 77, 78, 79, 82, 83, 84, 86, 103 photonic, xiii, 56, 341, 346, 349 photonic devices, 56 photonics, xiii, 9, 81, 341, 342 photons, 66, 68, 69, 95

404 photooxidation, 55 photosynthetic, 366 photovoltaic, 367 physical chemistry, 59 physical properties, 10, 35, 42, 49, 65, 77 physical sciences, ix physicists, 66 physics, ii, ix, xi, 1, 12, 44, 65, 110, 111, 195, 197, 199, 202, 209, 213, 214, 375, 378 physiological, 215 piezoelectric, xiii, 2, 170, 171, 177, 184, 189, 190, 191, 320, 322, 326, 338, 353, 354, 355, 356, 357, 358, 359, 360, 361, 362, 363 piezoelectric properties, 170, 320 piezoelectricity, xii, 319, 320 PL spectrum, 43 planar, 58, 246, 342, 344, 345, 346, 347, 349, 350 Planck constant, 78 plane waves, 78 plankton, 366 plants, 195, 196, 203, 204, 211, 213 plasma physics, 110 plasmatrons, x, 113, 124, 142 plasmons, 9, 10 plastic, 53 plastics, 303 platelets, 39, 40 platinum, xiii, 21, 294, 297, 319, 323, 326, 327, 328, 330, 331, 338, 355 play, 13, 79, 80, 191, 219, 229, 240, 247, 371 PLD, 11, 50 ploughing, 277 PMI, 95 point defects, 48, 61, 180, 246 point-to-point, 196 poisoning, 39 Poisson equation, 248, 249 polarity, 260 polarization, x, 2, 169, 170, 178, 180, 185, 186, 187, 188, 189, 190, 262 pollutant, 366, 371 pollutants, 5 pollution, xiii, 4, 6, 315, 365 polycrystalline, x, 169, 188, 245, 246, 247, 248, 260, 283, 293, 297, 298, 384 poly-crystalline, 324 polymer, 57 polymers, 342 polyphenols, 303 polyps, 204 poor, 53, 78, 108, 324, 348, 374 population, 4, 6, 66, 67, 71, 76, 80, 152, 163, 164, 368, 379

Index population density, 71, 368 pores, 25 porosity, 203, 252, 383 porous, 23, 198, 201, 215, 382 porous media, 201, 215 portability, 53 ports, 211 powder, 59, 175, 291, 293, 327 powders, 23 power plant, 201, 211 power plants, 211, 213 power-law, 377 powers, 11, 115, 142 precipitation, 305 predators, 204 prediction, 143, 343 preference, 332 press, 214, 216, 270, 318 probability, 68, 69, 71, 72, 73, 75, 95, 205, 218, 219, 220, 264, 377, 378, 379, 380, 386 probability distribution, 218, 219, 220 probe, 8, 21, 23, 47, 122, 123, 124, 130, 131, 134, 136, 137, 138, 139, 140, 142, 172 production, x, xiii, 3, 7, 23, 44, 109, 113, 132, 134, 136, 138, 142, 143, 165, 320, 342, 344, 365, 366, 367, 368, 370, 371 productivity, 124 program, 76 propagation, x, xiii, 63, 64, 83, 84, 100, 101, 109, 205, 206, 341, 342, 344, 346, 347, 348, 349 propane, 87, 368 proportionality, 203 proposition, 29, 30 protection, 5, 247, 374 protective coating, 374 proteins, 302 proteolytic enzyme, 316 protocols, 8, 366 protons, 21 pseudo, 48 p-type, xi, 12, 44, 45, 49, 50, 51, 52, 61, 62, 273, 274, 293, 296, 297, 298 pulse, 11, 64, 65, 68, 77, 78, 81, 82, 83, 86, 95, 102, 103, 104, 105, 106, 107, 114, 375, 378, 379, 381, 382, 383 pulsed laser, ix, 11, 17, 50, 57, 60, 63, 64, 65, 79, 82, 100, 102, 105, 109, 342 pulsed laser deposition, 11, 50, 57, 60, 64 pulses, xiv, 79, 106, 373 pumping, 106, 203, 285, 350 pumps, xiii, 341 purification, 303, 371 PVA, 24

Index

Q QDs, 13 quality of life, 210 quanta, 80 quantitative estimation, xiii, 353 quantum, 5, 6, 7, 10, 13, 57, 67, 89, 93, 176, 177, 196 quantum dots, 13, 57 quantum mechanics, 196 quartz, 6, 51, 82, 144, 145, 146, 327 quasi-equilibrium, 136

R R&D, 366 radial distribution, 71, 124, 162, 163 radiation, x, 13, 45, 53, 64, 65, 71, 72, 73, 81, 84, 114, 145, 146, 147, 148, 152, 153, 154, 156, 157, 161, 163, 203, 282, 320, 369 radio, 51, 114, 143, 247 radionuclides, 302 radius, 67, 71, 78, 82, 130, 136, 145, 147, 148, 157, 159, 160, 174, 200, 324, 349 rail, 211 rain, 200 Raman, xii, 9, 11, 13, 56, 57, 58, 65, 273, 284, 293, 294, 295, 296, 297, 298 Raman scattering, 56 Raman spectra, 293, 294, 295, 296, 297, 298 Raman spectroscopy, xii, 11, 57, 65, 273, 284, 293 random, 208, 218, 220, 231, 382 range, x, 2, 10, 11, 18, 23, 24, 34, 37, 45, 49, 51, 65, 66, 71, 74, 80, 81, 82, 86, 93, 95, 97, 102, 104, 108, 113, 114, 117, 120, 123, 126, 136, 140, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 156, 158, 160, 163, 165, 169, 170, 171, 172, 174, 175, 176, 177, 178, 179, 182, 184, 237, 265, 268, 274, 294, 307, 309, 327, 347, 348, 358, 377 RAS, 113 ratings, 121 raw materials, xii, xiii, 302, 305, 317, 365, 366, 368 Rayleigh, 201 reaction mechanism, 16 reactivity, 45, 46 reading, 268 reagent, 125 reagents, 123, 124 reality, 205

405 receptors, 302 recognition, 209 recombination, 5, 10, 68, 69, 70, 71, 72, 77, 101, 107, 108, 132, 370 recombination processes, 108 reconcile, 205 rectification, 51 recycling, 366 red blood cells, 4 red shift, 37 redistribution, 152, 189 refining, 292 reflection, 53, 148, 172, 173, 341 reflectivity, 9, 81, 82 refractive index, 5, 37, 53, 341, 342, 343, 344, 345, 346, 347, 348, 349 refractive indices, 348 regression, 90 regular, 39, 89, 279, 280, 298, 333, 335, 362 relationship, 71, 160, 210, 211, 212, 213, 219, 260 relationships, xi, 118, 133, 195, 213 relaxation, x, xii, 64, 86, 89, 97, 169, 170, 171, 179, 180, 181, 184, 186, 187, 188, 191, 302, 309, 314, 317 relaxation process, 89, 179 relaxation processes, 179 relaxation time, 179, 181 relaxation times, 179 reliability, 23, 247, 270 renal disease, 4 renormalization, 211, 232 reparation, 335, 355 reproduction, 210 reserves, 366 residuals, 220 resistance, xi, 2, 45, 46, 53, 117, 118, 120, 121, 160, 197, 198, 199, 200, 201, 202, 203, 205, 206, 209, 247, 252, 253, 260, 261, 273, 274 resistive, 175, 246 resistivity, xi, 23, 49, 50, 51, 52, 53, 62, 198, 208, 260, 261, 262, 273, 274, 275, 284, 293 resolution, 72, 83, 85, 86, 87, 88, 89, 92, 94, 96, 98, 102, 103, 104, 105, 106, 107, 145, 146, 147, 163, 281, 289, 298, 359 resonator, 80 resources, 210 respiratory, 4, 54, 202, 203, 205 respiratory failure, 4, 54 response time, 23 Reynolds, xi, 61, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 229, 230, 231, 233, 234, 236, 237, 240, 241, 242, 269

406

Index

Reynolds number, 219, 230, 233, 236, 237, 240 rhombohedral, 320, 324, 326, 327, 331, 332 rhythm, 213 rhythms, 213 rings, 21, 78, 148, 283 river basins, 196, 201, 209, 210, 214 rivers, 195, 209, 214 RNA, 302 robustness, 347 rods, 16, 18, 25, 196 Roman Empire, 210 room temperature, xii, 48, 51, 53, 55, 66, 170, 174, 176, 179, 184, 186, 187, 252, 253, 262, 266, 302, 305, 311, 314, 315, 317, 320, 326, 327, 332, 333, 355, 383, 384, 385, 386 room-temperature, 386 roughness, 11, 218, 219, 220, 222, 223, 224, 225, 240, 241, 242, 275, 276, 277, 279, 289, 291, 298, 325, 326, 330, 333, 383 rutile, 47, 384, 385, 386

S Saccharomyces cerevisiae, 303, 317 safety, 366 sample, 18, 19, 21, 23, 35, 36, 37, 38, 39, 41, 42, 43, 51, 52, 64, 65, 77, 78, 96, 124, 134, 137, 142, 170, 172, 173, 174, 175, 177, 186, 187, 252, 253, 254, 256, 259, 266, 278, 282, 291, 293, 294, 297, 305, 311, 335, 343, 344, 355, 356, 358, 359 sampling, 64, 65, 123, 124, 134, 142, 143, 356, 359 sand, 201 sapphire, 50, 61, 62 satellite, 93 saturation, xi, 7, 16, 18, 19, 20, 119, 245, 252, 263, 264, 265, 268, 309 sawdust, xii, 301, 302, 303, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 315, 316, 317 scaling, ix, 2, 201, 203, 204, 210, 211, 212, 213, 214, 215, 216 scaling law, 201, 210, 211, 212, 213, 214, 216 scaling relations, 210, 212, 213 scaling relationships, 213 Scanning electron, 40 scanning electron microscopy, xii, 24, 27, 39, 252, 301, 303, 334 scanning tunneling microscopy, 60 scatter, 291 scattering, 10, 44, 52, 53, 56, 288 Schottky, 245, 246, 247, 250, 255, 267 Schottky barrier, 245, 246, 247, 250, 255

search, 11, 206 seed, 337, 338 selected area electron diffraction, 34 selecting, 75 self-organization, 209 SEM, xii, 24, 25, 26, 28, 29, 30, 34, 39, 40, 41, 273, 289, 305, 334, 384 SEM micrographs, 384 semiconductor, x, 5, 14, 45, 49, 81, 113, 247, 248, 265, 269, 274, 344 semiconductors, xiii, 2, 44, 46, 49, 53, 61, 62, 118, 245, 246, 247, 341, 342 sensing, 23, 53, 60 sensitivity, 53, 65, 81, 100, 145, 289, 320 sensors, 2, 8, 9, 15, 16, 57, 59, 269, 320 separation, xii, 70, 230, 237, 239, 264, 301, 302, 303, 304, 305, 306, 307, 314, 316, 317 series, 75, 78, 86, 122, 154, 163, 214, 256, 285, 368 shape, xi, 10, 72, 73, 76, 78, 82, 84, 102, 103, 123, 127, 163, 195, 196, 197, 198, 199, 200, 201, 204, 205, 206, 210, 213, 214, 215, 222, 228, 230, 240, 242, 266, 306, 320, 335, 367 sharing, 24 shear, 190, 191, 196, 227, 228, 229, 235, 241, 243 shear rates, 227, 229 shock, 84, 108 shock waves, 84 short-range, 171 shoulder, 38 sign, 52, 74, 267 signals, 13, 81, 205, 274, 311, 313, 354, 356 signal-to-noise ratio, 81, 108 silica, 342, 346, 349, 350 silica glass, 349 silicate, 175, 343, 344, 347, 348 silicate glass, 347 silicon, xi, xiii, 12, 24, 58, 81, 242, 273, 274, 275, 277, 278, 280, 281, 282, 283, 284, 285, 287, 288, 289, 290, 291, 293, 294, 295, 296, 297, 298, 319, 323, 326, 329, 330, 332, 335, 337, 338, 346, 348, 354, 355, 356, 358, 359, 360 silicon dioxide, xiii, 319, 329, 330, 332, 335, 337, 338, 346, 354, 358, 360 silver, 3, 9, 10, 11, 12, 13, 14, 15, 16, 45, 46, 47, 56, 57, 61, 355 simulation, 76, 230, 234, 237, 241, 288, 291, 292 simulations, 76, 234, 291 sine, 359 single crystals, 171 sintering, 245, 260

Index SiO2, 330, 337, 359 sites, 3, 18, 19, 20, 43, 47, 204, 315, 345 skeleton, 196 skin, 70, 157, 159, 160 SME, 241 software, 146, 147, 219, 233, 354, 356 soil, 201 solar, x, 5, 9, 16, 113, 366, 367 solar cell, x, 5, 16, 113 solar cells, 5, 16 solar energy, 367 sol-gel, 17, 51, 60, 62, 262, 266 solid phase, 315 solid state, 342 solidification, 201 solid-state, 17 solubility, 347 soot, 142 sorption, 315 spatial, 10, 115, 146, 147, 163, 208, 214 species, ix, xiv, 7, 44, 49, 63, 64, 65, 66, 68, 69, 70, 71, 74, 75, 86, 94, 95, 98, 100, 101, 102, 103, 104, 105, 106, 108, 109, 202, 203, 204, 278, 324, 326, 342, 368, 369, 370, 373, 374, 375, 376, 377, 378, 379, 380, 382, 384, 386 specific heat, 118, 119, 120, 158 spectral component, 81 spectrophotometric, xii, 301, 314 spectroscopic methods, 79, 109 spectroscopy, ix, xii, 11, 57, 63, 64, 65, 71, 79, 81, 110, 111, 146, 147, 163, 273, 274, 277, 284, 293, 298, 301, 303, 307, 319, 327, 330, 378 spectrum, 15, 37, 38, 39, 43, 45, 51, 61, 64, 76, 85, 86, 89, 93, 96, 102, 103, 104, 105, 109, 145, 146, 148, 151, 152, 153, 173, 179, 181, 184, 187, 232, 243, 288, 293, 294, 297, 369 speed, 17, 19, 65, 78, 163, 170, 204, 205, 206, 207, 208, 209, 213, 215, 219, 222, 224, 240 speed of light, 78 spin, 2, 51, 62 spintronic devices, 2 sputtering, xiii, xiv, 11, 12, 17, 19, 24, 28, 34, 39, 45, 46, 49, 59, 61, 266, 275, 277, 319, 321, 323, 324, 325, 326, 329, 355, 359, 373, 374, 375, 376, 377, 378, 379, 380, 381, 386 stability, 18, 19, 23, 44, 49, 51, 140, 175, 381 stabilization, 115, 127, 132, 140, 380, 381 stabilize, 124 stages, 104, 132, 139, 221, 366, 367, 368, 376 standard model, 157, 158 Stark effect, 71, 73, 74, 109 statistical processing, 119

407 steady state, 120 steel, 119, 122, 124 stiffness, 189, 190 stochastic, 218, 220, 242 stochastic model, 218, 220 stoichiometry, xii, 18, 19, 142, 274, 288, 298, 319, 320, 324 storage, 2, 10, 13, 17, 35, 43, 306, 366, 374 strain, 28, 29, 34, 41, 53, 185, 188, 190, 227, 228, 232, 316 streams, 197, 200, 201, 208, 209, 215 strength, 2, 19, 20, 71, 72, 73, 117, 118, 126, 127, 128, 129, 143, 144, 145, 149, 150, 151, 157, 158, 159, 160, 161, 162, 181, 184, 185, 187, 188, 358, 382 stress, 183, 184, 185, 186, 187, 188, 189, 190, 191, 231, 232, 235, 241, 330, 353, 356 stretching, 11, 80 strikes, 81 strong interaction, 311 strontium, xii, 319, 320, 323, 337, 354, 356 structural modifications, 42 structure formation, 385 substances, xiv, 115, 124, 142, 303, 314, 365, 368, 371 substitutes, 4, 50 substitution, 133, 320 substrates, xi, xiii, 21, 24, 34, 50, 51, 53, 58, 59, 61, 273, 284, 285, 296, 297, 298, 319, 321, 323, 326, 330, 332, 342, 343, 355, 356, 359, 374, 375, 385 suffering, 4 sugar, 306 sulfur, 115 superconducting, 2 superimposition, 225 supernatant, 305, 314 superposition, 33, 106 supply, 6, 14, 116, 123, 124, 127, 128, 130, 134, 202, 366, 367 suppression, 24, 27, 34, 39, 106, 383 surface area, 5, 7, 23, 54, 203 surface component, 98 surface diffusion, 382 surface energy, 3 surface layer, 380 surface modification, 115 surface properties, 2 surface region, 344, 347 surface roughness, 11, 218, 222, 225, 240, 241, 242, 275, 276, 277, 279, 289, 291, 298, 325, 326, 330, 333, 383 surgical, 80

408

Index

survival, 205, 206 susceptibility, xii, 301, 303, 309, 310, 314 suspensions, xii, 302, 303, 304, 305, 307, 316, 317 swimmers, 204 switching, 17, 18, 19, 188 Switzerland, 269 symbols, 190 symmetry, 80, 157, 158, 163, 175, 188, 190, 294, 309 synchronization, 83 synthesis, x, 11, 23, 59, 65, 113, 115, 123, 132, 134, 135, 136, 140, 141, 142, 165, 274, 385 systems, ix, x, xi, 1, 3, 10, 18, 44, 45, 46, 47, 66, 79, 89, 103, 109, 143, 146, 169, 171, 177, 178, 179, 182, 185, 191, 195, 196, 197, 198, 199, 201, 202, 203, 204, 205, 209, 210, 211, 212, 213, 214, 215, 216, 303, 307, 309, 313

T targets, 102, 108, 385 technical assistance, 192 technology, ix, xiii, 1, 4, 16, 56, 142, 143, 218, 242, 274, 302, 304, 342, 344, 365, 375 teeth, 196 telecommunication, xiii, 341, 346, 349, 350 telecommunications, 342, 347 TEM, xii, 30, 32, 34, 273, 280, 281, 289, 290, 298, 304, 306, 314, 329, 335, 383 temperature dependence, 79, 137, 181, 184, 186, 256, 309 temperature gradient, 130 temporal, 68, 71, 82, 102, 103, 104, 105, 106, 115, 200 tension, 196, 216 test data, 234 textile, 303 thermal energy, 35, 108, 182, 298 thermal evaporation, 11, 17, 23, 57, 59, 60 thermal expansion, 34, 41, 84, 173, 294, 330, 346 thermal load, xiv, 373, 374 thermal resistance, 197 thermodynamic, 45, 66, 67, 73, 132, 138, 140, 196, 197, 210, 354 thermodynamic equilibrium, 45, 66, 67, 73, 132, 138, 140 thermodynamics, 135, 197, 214, 374 thermonuclear, 65 thorax, 203 three-dimensional, 15, 59, 242, 243, 291, 292, 334, 361, 383 three-dimensional representation, 334, 361

threshold, 51, 70, 72, 78, 83, 106, 118, 350 threshold level, 118 thrombosis, 54 tin, 16, 44, 260 tin oxide, 16 titania, 4 titanium, ix, xi, 1, 3, 4, 5, 55, 58, 115, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 298, 320, 323, 324, 329, 330, 355 titanium dioxide, 4, 5, 55, 58 titanium oxide, 3, 5 torus, 144, 147 total energy, 366 total internal reflection, 341 toxic, 5, 303, 306, 315 toxic effect, 315 trade-off, 198, 203 trajectory, 374, 385 transfer, 5, 13, 16, 18, 19, 48, 54, 58, 80, 121, 127, 141, 201, 203, 215, 231, 243 transformation, 163, 254, 293, 297, 298, 368, 385 transformations, 285 transition, 2, 17, 19, 59, 67, 71, 72, 73, 75, 76, 89, 90, 93, 133, 163, 177, 179, 182, 184, 188, 191, 199, 209, 242, 254, 255, 256, 307, 379, 380, 381, 384, 386 transition metal, 2, 17, 19, 59 transition temperature, 177, 179, 184 transitions, 10, 18, 19, 20, 43, 71, 72, 75, 80, 89, 163, 170, 211, 260, 383 translation, 138 transmission, xii, 31, 46, 50, 84, 147, 148, 170, 173, 280, 281, 289, 301, 303, 329, 336 transmission electron microscopy, xii, 31, 280, 281, 289, 301, 303, 329, 336 transparency, 45, 302 transparent, ix, 1, 3, 16, 17, 21, 44, 48, 49, 53, 61, 81, 381 transplant, 54 transplantation, 54 transport, 4, 44, 45, 46, 52, 53, 198, 203, 205, 262, 324, 326, 375, 376, 385 transportation, 207, 210 travel time, 207 tribological, 23 tribology, 241, 242 triggers, 83, 205 trypsin, 303, 316 tubular, 124, 136 tumours, 5 tungsten, 3, 4, 16, 18, 19, 20, 21, 22, 23, 24, 27, 29, 30, 34, 35, 36, 37, 38, 39, 40, 43, 58, 59, 60, 179

409

Index tunneling, 2, 60, 346, 347 turbulence, 127, 131, 214, 230, 234, 242, 243 turbulent, xi, 128, 141, 201, 217, 230, 231, 232, 233, 234, 235, 236, 240, 241, 242, 243 turbulent flows, xi, 217 two-dimensional, 81, 196, 231, 233, 354

U ultra-fine, 291, 292, 298 ultraviolet (UV), ix, x, 7, 8, 13, 35, 63, 69, 86, 104, 106, 109, 114, 143, 144, 145, 150, 151, 152, 153, 154, 155, 156, 165, 346 uncertainty, 73, 75, 95 unification, 215 uniform, 15, 24, 25, 39, 40, 159, 164, 204, 209, 267, 275, 278, 280, 287, 289, 298, 332, 344 universality, xi, 195, 211 universe, 196, 201 urokinase, 303 UV irradiation, 35 UV light, 7, 8 UV radiation, x, 114, 154 UV spectrum, 152, 153

V vacancies, 3, 39, 43, 44, 45, 47, 48, 180, 267 vacuum, ix, xi, 3, 4, 15, 21, 34, 49, 53, 60, 63, 64, 82, 86, 91, 94, 96, 109, 146, 148, 175, 245, 273, 275, 277, 280, 284, 285, 286, 287, 293, 294, 298 valence, 5, 12, 18, 19, 21, 38, 44, 45, 46, 48, 57 validation, 18, 56, 234 validity, xiii, 67, 242, 319 van der Waals, 2, 74 vanadium, 4 vapor, xiv, 17, 23, 34, 62, 94, 143, 144, 149, 158, 165, 266, 373, 386 vapor-liquid-solid, 34 variability, 197, 205 variables, xii, 77, 319, 320, 321, 322, 337 variance, 220 variation, 20, 21, 23, 90, 108, 143, 187, 219, 223, 224, 226, 230, 235, 236, 239, 240, 259, 275, 289, 321, 324, 363, 382 vector, 235 velocity, 46, 54, 67, 71, 73, 79, 106, 123, 124, 198, 203, 220, 223, 225, 228, 232, 233, 234, 237, 241, 250, 251, 255, 256 vessels, 120 vibration, 11, 80, 178, 182, 191

vibrational modes, 80 viscosity, 8, 119, 199, 223, 232, 233 visible, ix, x, 6, 10, 11, 13, 14, 37, 45, 47, 56, 63, 69, 81, 93, 106, 114, 144, 145, 146, 148, 150, 151, 153, 157, 160, 161, 162 vision, 208 VLS, 34 volumetric changes, 190 vortex, xi, 115, 116, 123, 124, 125, 127, 128, 132, 136, 140, 141, 165, 217, 230, 233, 237, 239 vortices, 141, 231, 239, 240

W walking, 207, 208, 209 wall temperature, 144, 147, 154 waste water, 317 wastes, 302 wastewater, 303, 316 wastewaters, 303 water, xi, xii, xiii, 3, 4, 6, 7, 8, 21, 24, 55, 56, 82, 123, 124, 125, 126, 131, 134, 136, 143, 146, 204, 205, 217, 234, 252, 253, 254, 259, 301, 302, 303, 304, 305, 314, 316, 317, 365, 366, 367, 368 water vapour, xiii, 365 water-soluble, xii, 302, 314, 317 wave number, 11 waveguide, xiii, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350 waveguide technology, 342 wavelengths, ix, 10, 63, 69, 95, 109, 349 weakness, 369 weapons, 80 welding, 80 wells, 171, 346, 347 wetting, 55 white blood cells, 7 wide band gap, 5, 21, 44, 45, 49 wind, 66, 366, 367 windows, ix, 1, 3, 5, 16, 18, 57, 82, 349 wires, 15, 16, 18, 24 wood, 204, 313 workers, 52 writing, 60, 344, 345, 346, 348, 350

X xenobiotics, xii, 301, 302, 303, 316, 317 x-ray absorption, 283 x-ray analysis, 280

410

Index

x-ray diffraction, 50 X-ray diffraction (XRD), xii, 27, 36, 273, 281, 282, 283, 291, 292, 293, 294, 297, 298, 319, 321, 322, 324, 325, 326, 327, 328, 331, 332, 334, 336, 385, 386 X-ray photoelectron spectroscopy (XPS), 21, 38, 39, 323, 326

Y yeast, xii, 301, 303, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 315, 316, 317

yield, 5, 10, 44, 114, 115, 141, 150, 151, 152, 153, 154, 156, 376, 377, 378, 379, 380, 381, 386

Z zinc, 1, 3, 4, 44, 48, 50, 51, 61, 62, 81 zinc oxide, 61 zirconium, 323, 324 Zn, 47, 48, 50 ZnO, 44, 45, 48, 49, 50, 51, 53, 60, 61, 62, 247, 262, 263, 265, 269