Nanowires and Nanobelts: Materials, Properties and Devices: Volume 2: Nanowires and Nanobelts of Functional Materials [1 ed.] 038728706X, 9780387287065

Citation preview

NANOWIRES AND NANOBELTS Materials, Properties and Devices Volume 2: Nanowires and Nanobelts of Functional Materials

NANOWIRES AND NANOBELTS Materials, Properties and Devices Volume 2: Nanowires and Nanobelts of Functional Materials

edited by

Zhong Lin Wang Georgia Institute of Technology

Library of Congress Cataloging-in-Publication Data Nanowires and nanobelts: materials properties and devices - Nanowires and Nanobelts of Functional Materials. Edited by Zhong Lin Wang. ISBN 10 1-40207443 3 (Hard Cover - 2 vol set) ISBN 13 9781402074431 ISBN 10 0-387-28706-X (Soft Cover) ISBN 13 9780387287065 ISBN 0-387-28747-7 (E-book) Printed on acid-free paper First softcover printing, 2006 © 2003 Springer Science-Business Media, Inc. All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science-Business Media, Inc., 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now know or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks and similar terms, even if the are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed in the United States of America. 9 8 7 6 5 432 I springeronline.com

SPIN 11545095

Contents

Preface

ix

List of Contributors

xi

I. Nanodevices Based on Nanowires and Nanobelts Chapter 1. Nanodevice, Nanosensors and Nanocantilevers Based on Semiconducting Oxide Nanobelts (Zhong Lin Wang) 1. 2. 3. 4. 5. 6. 7. 8. 9.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Field-Effect Transistor Based on Single Nanobelts . . . . . . . . Oxygen Sensor Using a Single Nanobelt . . . . . . . . . . . . . . . . Photoconductivity of Nanobelts . . . . . . . . . . . . . . . . . . . . . . . Gas Sensors Based on Nanobelts . . . . . . . . . . . . . . . . . . . . . . Heat Transport Through Nanobelt . . . . . . . . . . . . . . . . . . . . . Nanobelt as Nanoresonators . . . . . . . . . . . . . . . . . . . . . . . . . Nanobelts as Nanocantilever . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 2. Oxide Nanowires and Nanolasers (Peidong Yang and Haoquan Yan) 1. 2. 3. 4. 5. 6. 7. 8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis and Characterization of ZnO Nanowires . . . . . . . . Controlled Growth of ZnO Nanowires . . . . . . . . . . . . . . . . . . Photoluminescence and Lasing Properties . . . . . . . . . . . . . . . Nonlinear Optical Mixing in Single Zinc Oxide Nanowires . . Photoconductive Oxide Nanowires as Nanoscale Optoelectronic Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . Room Temperature NO2 Photochemical Sensing . . . . . . . . . . Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 3 3 8 8 9 11 11 13 16

21 21 23 25 29 35 37 39 42

II. Functional Oxide Nanowires and Nanobelts Chapter 3. Nanobelts and Nanostructures of Transparent Conducting Oxides (Zhong Lin Wang) 1. 2.

Synthesis Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxide Nanobelts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

47 47 49

vi

Contents 3. 4. 5. 6. 7. 8. 9.

Synthesis of New Materials using Nanobelts as Template . . . Complex Nanobelt Structures . . . . . . . . . . . . . . . . . . . . . . . . Nanosheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanodiskettes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Planar Defects in Oxide Nanobelts . . . . . . . . . . . . . . . . . . . . Growth Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 4. Nanomechanics and Mechanical Behavior of Nanobelts (Scott X. Mao) 1. 2. 3. 4. 5.

Nanomechanical Behavior of Semiconducting Zinc Oxide Single Nanobelt . . . . . . . . . . . . . . . . . . . . . . . . . Bending and Cutting on Nanorodes, Nanotube and Nanobelt Using AFM Tip . . . . . . . . . . . . . . . . . . . . . . . . Bending and Fundamental Resonance Frequency of Nanotube Under TEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electromechanical Behavior of Carbon Nanotube . . . . . . . . . Revolutionary Nanowires with a Twist . . . . . . . . . . . . . . . . .

Chapter 5. Ferroelectric Nanowires (Jonathan E. Spanier, Jeffrey J. Urban, Lian Ouyang, Wan Soo Yun and Hongkun Park) 1. 2. 3. 4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis of BaTiO3 and SrTiO3 Nanowires . . . . . . . . . . . . . Scanned Probe Measurements of BaTiO3 Nanowires . . . . . . . Conclusion and Future Prospects . . . . . . . . . . . . . . . . . . . . . .

Chapter 6. Growth of Oxide Nanorods through Sol Electrophoretic Deposition (Steven J. Limmer and Guozhong Cao) 1. 2. 3. 4. 5. 6.

4. 5. 6. 7.

73 73 75 75 78 80

83 83 84 87 91

93

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Sol–Gel Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 Electrophoretic Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Various Oxide Nanorods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Chapter 7. Nanowires of Functional Oxides (Guanghou Wang) 1. 2. 3.

56 59 65 65 67 68 70

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . From Nanoclusters to Nanowires of Titanium Oxides . . . . . . Layered Structures of Potassium Hexatitanate Nanowhisker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rutile Stannic Oxide Nanorods . . . . . . . . . . . . . . . . . . . . . . . Cu2O Nanowires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V2O5 Nanofibres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Potential Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

113 113 114 125 127 130 133 134

Contents Chapter 8. Controlled Growth and Optical Properties of Zinc Oxide Nanostructures (Yue Zhang and Ying Dai) 1. 2. 3. 4. 5. 6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanostructures of ZnO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Growth Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Properties of ZnO Nanostructures . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 9. One-Step Hydrothermal Synthesis and Characterizations of Titanate Nanostructures (L.-M. Peng, Q. Chen, G. H. Du, S. Zhang and W. Z. Zhou) 1. 2. 3. 4. 5. 6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . Growth Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electric and Optical Properties . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 10. Nanowires and Nanotubes of Complex Oxides (Xun Wang, Xiaoming Sun, Jian Xu and Yadong Li) 1. 2. 3. 4.

Oxides of Nanowires Based on Vapor-Transport Method . . . . Template-Confined Method to Oxides of Nanowires . . . . . . . Solution-Based Synthetic Way to Oxides 1-D Nanostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 11. Silica Nanowires/Nanotubes (Jing Zhu, W. X. Sun and Jun Luo) 1. 2.

vii

139 139 140 140 148 153 154

157 157 158 159 165 169 170

173 174 175 176 188

191

Synthesis of Silica Nanowires/Nanotubes . . . . . . . . . . . . . . . 191 Characterization of Structures and Properties of Silica Nanowires/Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . 197

III. Sulphide, Polymer and Composite Nanowires Chapter 12. Sulphide Nanowires (Shihe Yang) 1. 2. 3. 4. 5. 6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . From 0D to 1D Sulphides . . . . . . . . . . . . . . . . . . . . . . . . . . . Growth of Sulphide Nanowires . . . . . . . . . . . . . . . . . . . . . . . Structural, Electronic, Optical, and Transport Properties . . . . Potential Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . .

209 209 210 211 227 231 235

viii

Contents

Chapter 13. Generalized Solution Synthesis of Large Arrays of Extended and Oriented Nanowires (Jun Liu, Zhengrong R. Tian, James A. Voigt, Matthew J. Mcdermott and Bonnie Mckenzie) 1. 2. 3. 4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 14. Composite Nanowires (Yuegang Zhang) 1. 2. 3. 4. 5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phase Separation in Multi-Elemental Nanotubes . . . . . . . . . . Filling in Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . Coaxial Nanocables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Prospect of Composite Nanowires . . . . . . . . . .

Chapter 15. Polymer Nanowires and Nanofibers (Liming Dai and Darrell H. Reneker) 1. 2. 3. 4. 5. 6. 7.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanofibrillar Conducting Polymers . . . . . . . . . . . . . . . . . . . . Template Syntheses of Polymer Nanowires . . . . . . . . . . . . . . Syntheses of Polymer Nanowires at a Scanning Microscope Tip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrospinning of Polymer Nanofibers . . . . . . . . . . . . . . . . . Polymer Nanowires and Nanofibers with Special Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

239 239 240 243 252 257 257 257 261 264 266

269 269 270 271 273 276 279 286

Preface

Nanowires, nanobelts, nanoribbons, nanorods … , are a new class of quasi-onedimensional materials that have been attracting a great research interest in the last few years. These non-carbon based materials have been demonstrated to exhibit superior electrical, optical, mechanical and thermal properties, and can be used as fundamental building blocks for nano-scale science and technology, ranging from chemical and biological sensors, field effect transistors to logic circuits. Nanocircuits built using semiconductor nanowires demonstrated were declared a “breakthrough in science” by Science magazine in 2001. Nature magazine recently published a report claiming that “Nanowires, nanorods, nanowhiskers, it does not matter what you call them, they are the hottest property in nanotechnology” (Nature 419 (2002) 553). There is no doubt that nanowire based quasi-one-dimensional materials will the new focal point of research in the next decades. Volume 1: Metal and Semiconductor Nanowires covers a wide range of materials systems, from noble metals (such as Au, Ag, Cu), single element semiconductors (such as Si and Ge), compound semiconductors (such as InP, CdS and GaAs as well as heterostructures), nitrides (such as GaN and Si3N4) to carbides (such as SiC). The objective of this volume is to cover the synthesis, properties and device applications of nanowires based on metal and semiconductor materials. The volume starts with a review on novel electronic and optical nanodevices, nanosensors and logic circuits that have been built using individual nanowires as building blocks. Then, the theoretical background for electrical properties and mechanical properties of nanowires is given. The molecular nanowires, their quantized conductance, and metallic nanowires synthesized by chemical technique will be introduced next. Finally, the volume covers the synthesis and properties of semiconductor and nitrides nanowires. Volume 2: Nanowires and Nanobelts of Functional Materials covers a wide range of materials systems, from functional oxides (such as ZnO, SnO2 and In2O3), structural ceramics (such as MgO, SiO2 and Al2O3), composite materials (such as Si-Ge, SiC-SiO2), to polymers. This volume focuses on the synthesis, properties and applications of nanowires and nanobelts based on functional materials. Novel devices and applications made from functional oxide nanowires and nanobelts will be presented first, showing their unique properties and applications. The majority of the text will be devoted to the synthesis and properties of nanowires and nanobelts of functional oxides. Finally, sulphide nanowires, composite nanowires and polymer nanowires will be covered. The materials covered in both volumes are very diverse and rich in properties. Most of the nanowire and nanobelt structures are structurally controlled with respect to growth directions and side surfaces, resulting in controlled and tunable ix

x

Preface

electrical and optical properties, which offer huge advantages for applications in nanotechnology. The chapters were written by leading scientists worldwide whose groups have been the pioneers in the field and have done substantial work in their specific disciplines. Both volumes review the most up-to-date progress in nanowires and nanobelts. The books are intended as research books for advanced students and researchers with background in physics, chemistry, electrical engineering, mechanical engineering, chemical engineering, biology and bioengineering. Z. L. WANG

List of Contributors Chinese Academy of Sciences, China e-mail: [email protected]

Cao, Guozhong Department of Materials Science and Engineering & Center for Nanotechnology University of Washington 302 Roberts Hall, Box 352120, Seattle WA 98195, USA e-mail: [email protected]

Li, Yadong Department of Chemistry Tsinghua University, Beijing 100084, P.R. China e-mail: [email protected]

Chen, Qing Department of Electronics Peking University, Beijing 100871 China e-mail: [email protected]

Liang, Liang eVionyx, 6 Skyline Drive Hawthorne, NY10532 Limmer, Steven Department of Materials Science and Engineering & Center for Nanotechnology University of Washington 302 Roberts Hall, Box 352120 Seattle, WA 98195 USA e-mail: [email protected]

Dai, Liming Department of Polymer Engineering College of Polymer Science and Polymer Engineering The University of Akron, Akron Ohio 44325-3909, USA e-mail: [email protected]

Liu, Jun Sandia National Laboratories Albuquerque, NM 87185 e-mail: [email protected]

Dai, Ying Department of Materials Physics University of Science and Technology Beijing 30 Xueyuan Road Beijing 100083, China e-mail: [email protected]

Luo, Jun School of Materials Science and Engineering, Tsinghua University Beijing 100084, P.R. China

Du, G. H. Beijing Laboratory of Electron Microscopy Institute of Physics and Center for Condensed Matter Physics

Mao, Scott X. Department of Mechanical Engineering University of Pittsburgh, Pittsburgh e-mail: [email protected] xi

xii

List of Contributors

Mcdermott, Matthew J. Sandia National Laboratories Albuquerque, NM 87185 Mckenzie, Bonnie Sandia National Laboratories Albuquerque, NM 87185 Ouyang, Lian Department of Chemistry and Chemical Biology Harvard University, Cambridge MA 02138, USA

Sun, W. X. School of Materials Science and Engineering, Tsinghua University Beijing 100084, P.R. China Sun, Xiaoming Department of Chemistry Tsinghua University Beijing, 100084, P.R. China Tian, Zhengrong R. Sandia National Laboratories Albuquerque, NM 87185

Park, Hongkun Department of Chemistry and Chemical Biology, Harvard University Cambridge, MA 02138, USA e-mail: [email protected]

Urban, Jeffrey J. Department of Chemistry and Chemical Biology Harvard University, Cambridge MA 02138, USA

Peng, Lian-Mao Department of Electronics, Peking University, Beijing 100871, China and Beijing Laboratory of Electron Microscopy Institute of Physics and Center for Condensed Matter Physics Chinese Academy of Sciences China e-mail: [email protected], [email protected]

Voigt, James A. Sandia National Laboratories Albuquerque, NM 87185

Reneker, Darrell H. Maurice Morton Institute and Department of Polymer Science College of Polymer Science and Polymer Engineering The University of Akron, Akron Ohio 44325-3909, USA e-mail: [email protected] Spanier, Jonathan E. Department of Chemistry and Chemical Biology Harvard University, Cambridge MA 02138, USA

Wang, Guanghou Department of Physics, Nanjing University, Nanjing China e-mail: [email protected] Wang, Xun Department of Chemistry Tsinghua University Beijing 100084, P.R. China Wang, Zhong Lin (Z. L.) School of Materials Science and Engineering Georgia Institute of Technology Atlanta, GA 30332-0245, USA e-mail: [email protected] Xu, Jian Department of Chemistry Tsinghua University Beijing 100084, P.R. China

List of Contributors Yan, Haoquan Department of Chemistry University of California Berkeley, CA 94720 Yang, Peidong Department of Chemistry University of California Berkeley, CA 94720 e-mail: [email protected] Yang, Shihe Department of Chemistry Institute of Nano Science and Technology The Hong Kong University of Science and Technology, Clear Water Bay Kowloon, Hong Kong e-mail: [email protected] Yun, Wan Soo Department of Chemistry and Chemical Biology Harvard University, Cambridge MA 02138, USA Zhang, S. Department of Electronics Peking University, Beijing 100871 China e-mail: [email protected]

xiii

Zhang, Yue Department of Materials Physics University of Science and Technology Beijing, Beijing 100083 China e-mail: [email protected] Zhang, Yuegang Intel Corporation SC2-24, 2200 Mission College Blvd Santa Clara, CA 95054 e-mail: [email protected] Zhou, W. Z. School of Chemistry University of St. Andrew St. Andrews KY16 9ST, UK e-mail: [email protected] Zhu, Jing School of Materials Science and Engineering, Tsinghua University Beijing 100084, P.R. China e-mail: [email protected]

Part I Nanodevices Based on Nanowires and Nanobelts

Chapter 1 Nanodevice, Nanosensors and Nanocantilevers Based on Semiconducting Oxide Nanobelts Zhong Lin Wang School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0245, USA

1. Introduction Semiconducting transition and rare-earth metal oxides are attracting significant attention as candidates for chemical and environmental sensors [1–4]. This is because their electrical conductivity depends sensitively on the nature and concentration of adsorbed species on their surfaces. The key characteristics of these oxides are: (a) cations with mixed valence and (b) oxygen vacancies. The latter are responsible for the observed high sensitivity of the electrical properties to the presence of adsorbed molecules and allow the tuning of the conductance of the oxide. The semiconducting oxide nanobelts grown in our group [5, 6] are structurally perfect and geometrically uniform. These nanostructures are ideal materials for building nano-size devices and sensors. In this chapter, we review the field-effect transistor (FET), gas sensors and nanocantilevers that have been fabricated using ZnO/SnO2 nanobelts.

2. Field-Effect Transistor Based on Single Nanobelts [7] 2.1. Device fabrication Single-crystalline SnO2 and ZnO nanobelts of thicknesses between 10 nm and 30 nm were synthesized by thermal evaporation of oxide powders in an alumina tube without the presence of a catalyst [5] (see Chapter 3 in Volume II). Fig. 1 gives transmission electron microscopy (TEM) images of SnO2 and ZnO nanobelts. Each nanobelt is a single crystal without the presence of dislocations; its morphology and structure, such as growth direction and surface planes, are well defined, and their surfaces are clean and atomic flat. The nanobelts have a rectangular-like cross-section with an average width of ~200 nm, width-to-thickness ratios of 5–10 and lengths of

4

Wang (b) ZnO

(a) SnO2

200 nm

Fig. 1. TEM images of the as-synthesized (a) SnO2 and (b) ZnO nanobelts. Gold electrodes (50 nm thick)

(a) (10 um)2 AFM image

ZnO nanoribbon (w=285nm, h=110nm)

SiO2 (~120nm thick)

(b) V

FET configuration:

source

drain Nanoribbon Au electrodes SiO2

V

Silicon

backgate

Fig. 2. (a) AFM image of an FET device made using a single nanobelt; (b) Schematic diagram of the device.

up to a few millimeters. The contrast observed in the TEM image is due mainly from strain contrast. Large bundles of either SnO2 or ZnO nanobelts were dispersed in ethanol by ultrasonication until mostly individual nanobelts were isolated. These ethanol dispersions were dried onto a SiO2/Si substrate for imaging by non-contact mode AFM. SnO2 field-effect transistors were fabricated by depositing SnO2 nanobelt dispersions onto SiO2/Si (p⫹) substrates, followed by treatment in an oxygen atmosphere at 800 ⬚C for two hours. The SiO2/Si substrates were then spin-coated with PMMA, baked, exposed to electron-beam lithography for the definition of electrode arrays, and developed. A 30 nm thick layer of titanium was deposited by electron-beam evaporation to

Nanodevice, Nanosensors and Nanocantilevers

5

serve as the source and drain electrodes and the remaining PMMA was lifted off in hot acetone. An AFM image of FET and the schematic diagram are given in Fig. 2. The principle of this device is that controlling the gate voltage would control the current flowing from the source to the drain. An alternative way of contacting the nanostructures was applied to ZnO nanobelts. In this case, field-effect transistors were fabricated by depositing dispersed ZnO nanobelts on predefined gold electrode arrays. In both cases, the SiO2 gate dielectric thickness was 120 nm, and the back gate electrode was fabricated by evaporation of gold on the Si (p⫹) side of the substrate. Also, in both fabrication schemes, the electrode arrays were variably spaced. They included electrode gaps as small as 100 nm and as large as 6 ␮m. By forming metal electrode/nanostructure electrical contacts and capacitively coupling the nanostructure to a near by gate electrode, a field-effect transistor (FET) is produced using a nanobelt that allows the exploration of new aspects of the physical and chemical properties of the nanostructures [8–11]. 2.2. Physical characteristics of the FET Both SnO2 and ZnO nanobelts formed field effect transistors. A typical SnO2 field effect transistor pre-treated in a 1 atm oxygen atmosphere at 800 ⬚C (Fig. 3) demonstrated a gate-threshold voltage of ⫺2.5 V, a current switching ratio I(ON)/I(OFF) of nearly 10,000, and (ignoring voltage drops at the Ti contacts) a peak conductivity of 8.14 (O-cm)⫺1. Measured conductivities ranged from 4 (O-cm)⫺1 to 15 (O-cm)⫺1. The typical leakage conductance between our electrodes under ambient conditions measured between 20 pS and 80 pS and was gate-bias independent. In most cases, the

Fig. 3. Source-drain current versus gate bias for a SnO2 FET in ambient.

6

Wang

switching ratios of our FETs were limited by this background conductivity. From an analysis of the transconductance, dId/dVg, for gate biases above the threshold bias, the electron mobility of an n-type field effect transistor can be estimated if the geometry of the device is known [12]. Without subtracting voltage drops at the contacts, from analysis of Fig. 3 the electron mobility in the SnO2 nanobelt was estimated to be ~35 cm2/V-s. Other measured SnO2 nanobelt FETs exhibited electron mobility ranging from 10 cm2/V-s to 125 cm2/V-s. From four-probe measurements on other devices, the contact resistance between the Ti electrodes and the SnO2 nanobelt was found to be on the same order of magnitude as the four-probe “ON” channel resistance. Thus, our calculations underestimate both conductivity and mobility by assuming the entire source drain potential is dropped across the nanobelt. The alternative way of contacting the nanobelts by simply depositing them on top of pre-fabricated gold electrodes led to very resistive contacts. A typical ZnO field effect transistor showed a gate threshold voltage of ⫺15 V, a switching ratio of nearly 100, and a peak conductivity of 1.25 ⫻ 10⫺3 (⍀-cm)⫺1. A completely analogous behavior has been observed in the case of carbon nanotubes deposited on top of Au electrodes or covered by Ti electrodes [11]. 2.3. Sensitivity of the electrical properties of nanobelts to environment [7] Before electrical measurement, SnO2 nanobelts are annealed in a 1 atm oxygen environment at 800 ⬚C for 2 hours. Without this treatment, the as-produced nanobelts exhibit no measurable conductivity for source-drain biases from ⫺10 V to 10 V and for gate biases from ⫺20 V to 20 V, while after this treatment the SnO2 nanobelts exhibit considerable conductivity. By further annealing of the devices at lower temperatures in vacuum, oxygen, or ambient the electrical properties of the nanobelts can be tuned. After annealing of the SnO2 devices in vacuum at 200 ⬚C, the nanobelt conductivity is observed to increase along with an associated negative-shift in gate-threshold voltage. Smaller, additional increases in conductivity are observed after additional vacuum anneals. Eventually, the nanobelt behaves like a metal with the gate field being unable to affect the current flowing through the device. In contrast, annealing nanobelt devices in ambient at 200 ⬚C leads to a decrease of the conductivity, along with a shift in the gate-threshold voltage in the opposite, positive direction (Fig. 4). The source-drain conductivity at zero gate bias spans 3 orders of magnitude from 0.09 (⍀-cm)⫺1 after annealing at 200 ⬚C in ambient to 75.3 (⍀-cm)⫺1 after annealing at 250 ⬚C in vacuum. The changes in conductivity with low temperature annealing most likely result from variations in the number of oxygen species adsorbed on the SnO2 surfaces or in the number of oxygen vacancies in the SnO2 bulk with the amount of oxygen in the environment [13–17]. The number of equilibrium surface and bulk oxygen defects will be a function of the environmental oxygen partial pressure and temperature [15–18]. Annealing in vacuum should decrease the number of adsorbed oxygen species and increase the number bulk oxygen vacancies while annealing in oxygen or ambient should do the opposite. It is well established that bulk and surface oxygen vacancies in SnO2 act as electron donors, which should increase SnO2 conductivity and decrease the gate threshold voltage [15–17]. This is precisely what we observe.

Nanodevice, Nanosensors and Nanocantilevers

7

Fig. 4. SnO2 Nanobelt FET oxygen sensitivity. Source-drain current versus gate bias for a SnO2 FET after various treatments measured in this order: air, vacuum, 200 ⬚C vacuum anneal, 250 ⬚C vacuum anneal, 200 ⬚C air anneal. All measurements taken at room temperature with vacuum anneal measurements taken in vacuum and air annealing measurements in air [7].

It should be noted that the SnO2 nanobelt conductivity is observed to increase and the gate threshold voltage is observed to decrease simply by taking a nanobelt device from ambient into vacuum without annealing (Fig. 4). Since the diffusion of bulk oxygen vacancies at room temperature will be limited [14, 17], this indicates that surface oxygen desorption is most likely taking place at this temperature. Because of their small dimensions, semiconducting oxide nanobelts will have on the order of 1020 surface oxygen sites per cubic centimeter of material [16]. Thus, even for partial changes in the concentration of adsorbed oxygen species, large changes in nanobelt conductivity can be observed. As noted, annealing at 200 ⬚C and 250 ⬚C induces further changes in conductivity (Fig. 4). To account for this either the surface or bulk nanobelt composition must change as a result of these annealing steps. Mizusaki et al. [16] has argued that for temperatures below 900 ⬚C the equilibrium bulk nonstoichiometry of SnO2 will be insignificant in comparison to surface nonstoichiometry in samples of similar surface/ volume ratios. This suggests that surface oxygen vacancies should control conductivity after 200 ⬚C and 250 ⬚C anneals. Annealing might also desorb species other than oxygen from the nanobelt surfaces. Water could act as a mask on the nanobelt surface, decreasing sensitivity to environmental oxygen until removed. The strong dependence of the conductance on the oxygen deficiency in nanobelts is an important characteristic of functional oxide, using which one is capable of tuning and controlling the electrical properties of the nanodevice.

8

Wang

3. Oxygen Sensor Using a Single Nanobelt [7] The sensitivity of the FET to gas exposure can also be demonstrated by measuring the SnO2 nanobelt conductivity as a function of time after introduction of oxygen. The conductivity of a hot 200 ⬚C SnO2 nanobelt in vacuum is observed to decay to about 1/3 its initial value after exposure to oxygen at a pressure of 5 ⫻ 10⫺4 torr (Fig. 5a). The exponential time constant for this decay is 37.2 s. After removal of the oxygen, the conductivity recovers about twice as slow, with a time constant of 76.5 s. In contrast, the conductivity of the same SnO2 nanobelt at the same temperature is invariant upon exposure to nitrogen at the same pressure (Fig. 5b). Exposure of the nanobelt to 5 ⫻ 10⫺4 torr oxygen at room temperature leads to the decay of the conductivity at a much slower rate with a time constant of 156.9 s, reflecting the activated nature of the oxygen desorption processes [16]. At room temperature a smaller percentage of surface vacancy sites are likely to be affected, thus a smaller current change is produced [16].

4. Photoconductivity of Nanobelts [7] Ultraviolet light irradiation of ZnO in air is observed to result in a significant increase of the conductivity (Fig. 5). Light with a wavelength of 350 nm (E ⫽ 3.54 eV) was used, exceeding the direct band-gap of ZnO, which is of 3.2 eV [20, 22]. The increase in the conductivity results from both photogeneration of electron-hole pairs as well as doping by UV light-induced surface desorption [19–22]. These processes

Fig. 5. ZnO nanobelt response to UV light. Source-drain current versus time at zero-gate bias and a source drain potential of 6.5 V. The initial state of the nanowire in these responses was one of steady state with a 350 nm light source ON.

Nanodevice, Nanosensors and Nanocantilevers

9

could be observed by introducing a shutter between the light source and the ZnO nanobelt so that the flux of UV photons could be turned “ON” and “OFF”.

5. Gas Sensors Based on Nanobelts [24] Conductometric metal oxide semiconductor thin films are the most promising devices among solid state chemical sensors, due to their small dimension, low cost, low power consumption, on-line operation and high compatibility with microelectronic processing. The fundamental sensing mechanism of metal oxide based gas sensors relies on a change in electrical conductivity due to the interaction process ⫹ ⫺ between the surface complexes such as O⫺, O⫺ 2 , H and OH reactive chemical species and the gas molecules to be detected. Although a large number of different oxides have been investigated for their gas sensing properties, commercially available gas sensors are mainly made of SnO2 in the form of thick films, porous pellets or thin films. The effects of the microstructure, namely, ratio of surface area to volume, grain size and pore size of the metal oxide particles, as well as film thickness of the sensor are well recognized. Lack of long term stability has until today prevented a wide range application of this type of sensors. The most recent research has been devoted towards nanostructured oxides [25–29] since reactions at grain boundaries and complete depletion of carriers in the grains can strongly modify the material transport properties. Unfortunately the high temperature required for the surface reactions to take place induces a grain growth by coalescence and prevents the achievement of stable materials. Nanobelts of semiconducting oxide [5], with a rectangular cross section in a ribbon-like morphology, are very promising for sensors due to the fact that the surface to volume ratio is very high, the oxide is single crystalline, the faces exposed to the gaseous environment are always the same and the size is likely to produce a complete depletion of carriers inside the belt. Beside the deposition technique is very simple and cheap, the size and shape can be easily controlled. In the polycrystalline and thick film devices, only a small fraction of the adsorbed species adsorbed near the grain boundaries is active in modifying the device electrical transport properties. In the new sensors based on single crystalline nanobelts, almost all of the adsorbed species are active in producing a surface depletion layer. Free carriers should cross the belts bulk along the axis in a FET channel-like way. Beside, since the size of the depletion layer for tin oxide, due to oxygen ionosorption, penetrates 50 nm or more through the bulk, the belts are probably almost depleted of carriers as a pinched-off FET because of belt thickness is typically less than 50 nm. The presence of poisoning species should switch the structures from pinched-off to conductive channel, strongly modifying the electrical properties. A further reduction of belt size could envisage the development of quantum confined structures and nanodevices. For the fabrication of sensors, a platinum interdigitated electrode structure was made using lithography and metal deposition technique on an alumina substrate. A platinum heater was attached to the backside of the substrate in order to control the working temperature of the sensor. Then, a bunch of nanobelts was placed onto the electrodes for measuring their electric conductance, and a proper measure was taken to

10

Wang T 400°C

10–5

600 CO [ppm]

Current [A]

500

300

CO [ppm]

Current [A]

400

200 100 10–6

0

20

40

60

80

0 100

Time [min] Fig. 6. Response of the SnO2 nanobelts to CO at a working temperature of 400 ⬚C and 30% RH (from Comini, Faglia, Sberveglieri, Pan, Wang, Applied Physics Letters, 81 (2002) 1869).

ensure the contact of the nanobelts with the electrodes. The flow-through technique is used to determine the gas-sensing properties of the thin films. A constant flux of synthetic air equal to 0.3 l/min, mixed with the desired amount of gaseous species, flows through a stabilized sealed chamber at 20 ⬚C, atmospheric pressure and controlled humidity. Electrical characterization was carried out by a volt-amperometric technique at constant bias of 1 V, and a picoammeter measured the change of electrical current. Fig. 6 reports the isothermal response of the current flowing through the tin oxide nanobelts as two square concentration pulses of CO (250 and 500 ppm respectively) being fed into the test chamber, at a working temperature of 400 ⬚C and 30% RH (relative humidity at 20 ⬚C). The electric current increases for about 60% and 100% with the introduction of 200 and 500 ppm CO. The sensor response, defined as the relative variation in conductance due to the introduction of gas, is about ⌬G/G ⫽ 0.9. Fig. 7 shows the isothermal response of the current flowing through the nanobelts as a square concentration pulse of 0.5 ppm nitrogen dioxide is fed into the test chamber, at a working temperature of 400 ⬚C and 30% RH. The sensor response is ⌬G/G ⫽ ⫺15.5 ⫽ ⫺1550%, which is extremely high and sensitive. This means that the sensitivity of the sensor is in the level of a few ppb. In general, the selectivity of the oxide is a concern. This may be improved by fabricating sensors using several different types of nanobelts, or by functionalize the surfaces of the nanobelts. It is, however, very important to note that the CO and enthanel increases the conductivity, while NO2 decreases the conductivity of the SnO2 nanobelts.

11

Nanodevice, Nanosensors and Nanocantilevers T 200°C

10–6

0.6 Current [A]

NO2 [ppm]

0.5

0.3

10–7

NO2 [ppm]

Current [A]

0.4

0.2

0.1

10–8

0 0

5

10

15

20

25

30

35

40

Time [min]

Fig. 7. Response of the SnO2 nanobelts to NO2 at a working temperature of 200 ⬚C and 30% RH [24].

6. Heat Transport Through Nanobelt [32] Heat transport at nanoscale is a very interesting and technologically important area. With the reduce of object size, phonon modes and phonon density of states change drastically, resulting in unusual thermal transport phenomena in mesoscopic systems. Theoretical investigation of thermal conductance in a one dimensional nanowire predicted quantum effect at very low temperature: Gth⫽␲2k2BTⲐ3h [31]. Experimental measurement has proved the prediction [32]. Thermal transport along a single SnO2 nanobelt has also been carried out (Fig. 8a). Thermal contact micro-pads have been fabricated using lithography technique. The thermal conductance across the nanobelt was measured as a function of the local temperature is given in Fig. 8b [32]. An estimation from the dimension of the nanobelt gives a thermal conductivity of 2–3 times better than that of the bulk, which may due to the single crystalline structure of the nanobelt.

7. Nanobelt as Nanoresonators [33] The hardness of the nanobelt has been measured by nanoindenter to be in the order of 8–10 GPa, as described in Chapter 5 in Volume II. Another key quantity in the application of nanobelt is its Bending modulus. We have measured this quantity using a technique developed for carbon nanotubes. Based on the electric-field-induced resonant excitation, the mechanical properties of individual nanowire-like structures can be measured by in situ transmission electron microscopy (TEM) [34, 35]. Using this method, Mechanical properties of carbon nanotubes [34–37], silicon nanowires [38], and silicon carbide-silica composite nanowires [39] have been quantified.

12

Wang (a)

(b)

Nanobelt conductance

G_nanobelt(nW/K)

80 60 40 20 0

0

50

100

150 200 Temp(K)

250

300

350

Fig. 8. Measurement of thermal transport along a single SnO2 nanobelt (from Dr. Li Shi, University of Texas, Austin).

To carry out the mechanical property measurements of a nanobelt, a specimen holder for a Hitachi HF-2000 TEM (200 kV) was built for applying a voltage across a nanobelt and its counter electrode. Mechanical resonance can be induced if the applied frequency matches the natural resonance frequency of the nanobelt. Due to the mirror symmetry of the nanobelt (Fig. 9a), there are two distinct fundamental resonance frequencies corresponding to the vibration in the thickness and width directions, which are given from the classical elasticity theory as [40]

冪3␳ ␤W E ␯ ⫽ 4␲L 冪3␳ ␯x ⫽

␤21T 4␲L2 2 1

y

Ex

y

2

(1) (2)

where ␤i is a constant for the ith harmonic: ␤1 ⫽ 1.875 and ␤2 ⫽ 4.694; Ex and Ey are the bending modulus if the vibration is along x-axis (thickness direction) and y direction (width direction), respectively; ␳ is the density, L is the length, W is the width and T is the thickness of the nanobelt. The two modes are decoupled and they can be observed separately in experiments. Changing the frequency of the applied voltage, we found two fundamental frequencies in two orthogonal directions transverse to the nanobelt. Fig. 9b and c show the harmonic resonance with the vibration planes nearly perpendicular and parallel to the viewing direction, respectively. For calculating the bending modulus, it is critical to accurately measure the fundamental resonance frequency (␯1) and the dimensional sizes (L and T or W ) of the investigated ZnO nanobelts. To determine ␯1, we have checked the stability of resonance frequency to ensure one end of nanobelt is

Nanodevice, Nanosensors and Nanocantilevers (a)

W

(b)

T

13

(c)

L

y

x

Fig. 9. Measuring the Bending modulus of a ZnO nanobelt by electric field induced mechanical resonance in TEM. (a) Geometrical shape of a nanobelt; (b, c) Mechanical resonance of a nanobelt along the two orthogonal directions, respectively, closely perpendicular to the viewing direction (␯x ⫽ 1.25 MHz), and nearly parallel to the viewing direction (␯y ⫽ 1.38 MHz). Table I. The Bending modulus of individual ZnO nanobelts. Our experiments found there is no significant difference between Ex and Ey for the nanobelt grown along [0001]

Nanobelt 1 2 3 4

Length L (␮m) (⫾0.05)

Thickness T (nm) (⫾1)

Frequency ␯x1 (kHz)

Eb (GPa)

8.25 4.73 4.07 8.90

33 19 20 39

232 396 662 210

50.1 ⫾ 1.8 45.5 ⫾ 2.9 64.6 ⫾ 2.3 39.9 ⫾ 1.2

tightly fixed, and the resonant excitation have been carefully checked around the half value of the resonance frequency. The specimen holder can be rotated about its axis so that the nanobelt can be aligned perpendicular to the electron beam, so the real length (L) of the nanobelt can be measured. The projection direction along the beam is determined by electron diffraction pattern, so that the true thickness and width can -be determined because the normal direction of the nanobelt is [2110] (see Chapter 3 in Volume II). Based on the experimental data, the bending moduli of the ZnO nanobelts can be calculated using Equation (1) or (2). The experimental results are summarized in Table I [33]. The bending modulus of the ZnO nanobelts was ~52 GPa. Our experiments clearly shows that the nanobelt can be effective nanoresonators exhibiting two orthogonal resonance modes, which can be used as probes for SPM operated in tapping and scanning modes.

8. Nanobelts as Nanocantilever [41] Cantilever based scanning probe microscopy (SPM) technique is one of the most powerful approaches in imaging, manipulating and measuring nanoscale properties and phenomena. The most conventional cantilever used for SPM is based on silicon, Si3N4

14

Wang (a)

(b) F

-

+

-

+

-

+

+

-

+

-

+

-

-

+

-

+

-

+

+

-

+

-

+

-

-

+

-

+

-

+

+

-

+

-

+

-

-

+

-

+

-

+

+

-

+

-

+

-

Fig. 10. Cutting a CdO nanobelt by an electron/ion beam in TEM.

or SiC, which is fabricated by e-beam or optical lithography technique and has typically dimension of thickness of ~100 nm, width ~5 ␮m and length ~50 ␮m. Utilization of nanowire and nanotube based cantilever can have several advantages for SPM. Carbon nanotubes can be grown on the tip of a conventional cantilever and be used for imaging surfaces with a large degree of abrupt variation in surface morphology [42]. We demonstrate here the manipulation of nanobelts by AFM and its potential for as nanocantilevers. Manipulation of nanobelts is important for integrating this structurally controlled nanomaterial with microelectrical mechanical system (MEMS). The first task is to cut the nanobelt into specific length. This was done by using an AFM system, Digital Instrument 3000. Two techniques have been used to section nanobelts. One method is to effectively saw through the nanobelt by; 1) increasing the aspect ratio of the viewing screen during AFM operation, 2) minimizing the scan size to capture only the width of a nanobelt, and 3) increasing the integral and proportional gains and thus increasing the applied force on the nanobelt during scanning. This technique can be used in either Tapping Mode or Contact Mode. This technique is extremely reproducible and allows the user to record the force applied to the sample. Cutting nanobelt can also be performed by a focused electron/ion beam (Fig. 10). Two additional techniques have been used to section nanobelts [41]. One method is to effectively saw through the nanobelt by; 1) increasing the aspect ratio of the viewing screen during AFM operation, 2) minimizing the scan size to capture only the width of a nanobelt, and 3) increasing the integral and proportional gains and thus increasing the applied force on the nanobelt during scanning. This technique can be used in either Tapping Mode or Contact Mode. Images for two independent ZnO nanobelts that were fractured by the tip is given in Fig. 11. The numbered circles, within the images, correspond to multiple cutting attempts and increase with increasing cuts. It should be noted; image quality for atomic force microscopes is directly related to the tip-surface interaction. Since the atomic force probe was used as a cutting device, with high contact forces, the image quality degraded with increasing time. This observation is common during prolonged tip use or fracture. However, the intentions of this experiment were not to develop high quality AFM images, but to show the ability of an AFM to section nanobelts into specified lengths for cantilever applications. Fracturing

15

Nanodevice, Nanosensors and Nanocantilevers

(b)

(a)

2 1 1

3 4 1µm

1µm

Fig. 11. (a, b) AFM image of a ZnO nanobelt before and after it was fractured at places as indicated by the circles, respectively. Images were captured using a Dimension 3000 SPM in Tapping Mode operation.

of nanobelts is possible because the bond character is primarily ionic in nature, and thus an atomic displacement of half the lattice constant generates a cleavage due to Coulomb repulsion. Combining MEMS technology with self-assembled nanobelts we are able to produce cost effective cantilevers with much heightened sensitivity for a range of devices and applications. Force, pressure, mass, thermal, biological, and chemical sensors are all prospective devices. Semiconducting nanobelts are ideal candidates for cantilever applications. Structurally they are defect free single crystals, providing excellent mechanical properties. The reduced dimensions of nanobelt cantilevers offer a significant increase in cantilever sensitivity. The cantilevers under consideration are simple in design and practice. The two main components of the system are a silicon chip and a semiconducting nanobelt (Fig. 12a). The silicon chip is an energy wall, isolating cantilevers from local disturbances. The extremely high mechanical flexibility and a robust performance of the nanobelt (Fig. 12b) show its great potential for cantilever applications. Using a Dimension 3000 SPM in Tapping Mode, we have successfully lifted ZnO nanobelts from a silicon substrate. Capillary forces are responsible for the adhesion strength between the atomic force microscope probe and the ZnO nanobelts. Combining the aforementioned techniques with micromanipulation has led to the alignment of individual ZnO nanobelts onto silicon chips (Fig. 12c). The aligned ZnO cantilevers were manipulated to have a range of lengths. This exemplifies our ability to tune the resonance frequency of each cantilever and thus modify cantilevers for different applications such as Tapping and Contact Mode AFM. The nanobelt based nanocantilever is ~50–100 times smaller than the conventional cantilever. Decreased size in micro-optical mechanical devices corresponds to increased sensitivity. Combining the aforementioned techniques with micromanipulation has led to the horizontal alignment of individual ZnO nanobelts onto silicon chips. The aligned ZnO cantilevers, shown in Fig. 13 below, were manipulated to have a range of lengths. This exemplifies our ability to tune the resonance frequency of each cantilever and thus modify cantilevers for different applications such as Contact, Non-Contact, and Tapping Mode AFM. Periodic contrast of the ZnO cantilevers is observed as a result

16

Wang Semiconducting Oxide nanobelt

(a) Silicon Chip

(c) (b)

Silicon Cantilever

ZnO nanobelt

Fig. 12. (a) Nanobelt as nanocantilever glued on a silicon chip. (b) A ZnO nanobelt showing extremely high mechanical flexibility and toughness. (c) A ZnO nanobelt glued onto a silicon cantilever.

of electronic charge induced vibrations during SEM operation. Such contrast is absent in regions where the nanobelts are in direct contact with the silicon substrate, suggesting adequate adhesion forces between the cantilevers and the silicon chip.

9. Summary We have fabricated field-effect transistors (FETs) based on single SnO2 and ZnO nanobelts of thicknesses between 10 nm and 30 nm. Switching ratios as large as six orders of magnitude and conductivities as high as 15 (⍀-cm)⫺1 are observed. Annealing SnO2 nanobelt FETs in an oxygen-deficient atmosphere produces a negative shift in gate threshold voltage, indicating doping by the generation of surface oxygen vacancies. This treatment provides an effective way of tuning the electrical performance of the nanobelt devices. The ability of SnO2 FETs to act as gas sensors is also demonstrated. ZnO nanobelt FETs are sensitive to ultraviolet light. Both photogeneration of electron-hole pairs and doping by UV induced surface desorption contribute to the conductivity. Gas sensors have been fabricated using the single crystalline SnO2 nanobelts. Electrical characterization showed that the contacts were ohmic and the nanobelts were sensitive to environmental polluting species like CO and NO2 as well as ethanol for breath analyzers and food control applications. The sensor response, defined as the relative variation in conductance due to the introduction of the gas, is ⌬G/G ⫽ 4160% for 250 ppm of ethanol and ⌬G/G ⫽ ⫺1550% for 0.5 ppm nitrogen dioxide at 400 ⬚C and 30% RH. The results demonstrate the potential of fabricating nano-size sensors using the integrity of a single nanobelt with a sensitivity at the level of a few ppb.

Nanodevice, Nanosensors and Nanocantilevers

17

(a)

ZnO nanobelts

(b)

Fig. 13. (a) Site specific placement and alignment of ZnO nanobelts onto a silicon chip, forming nano-cantilever arrays. (b) An enlarged SEM image of the third nano-cantilever showing its shape; the width of the cantilever was measured to be 525 nm.

The bending modulus of the ZnO nanobelt has been measured to be ~50 GPa. Thermal conductance across a single SnO2 nanobelt has also been measured. Structurally controlled ZnO nanobelts have been manipulated and cut by an AFM tip for applications in MEMS technology. The re-shaped nanobelts have been demonstrated as nanoresonators and nanocantilevers, which are 50–1000 smaller in width and length than conventional cantilevers. The nano-cantilever is expected to have much improved sensitivity and greatly improved mechanical flexibility for applications in scanning probe microscopy and sensor technology.

Acknowledgment The results reviewed in this chapters were partially contributed from my collaborators and group members: William Hughes, M. S. Arnold, Ph. Avouris, E. Comini,

18

Wang

G. Faglia, G. Sberveglieri, Zhengwei Pan, Xuedong Bai, Li Shi and Enge Wang, to whom I am very grateful. Research supported by NSF, NASA, NSF China, Chinese Academy of Sciences and Georgia Tech.

References 1. S. R. Morrison, Sensors and Actuators 2 (1982) 329. 2. R. K. Sharma, P. C. H. Chan, Z. Tang, G. Yan, I.-M. Hsing and J. K. O. Sin, Sensors and Actuators B 72 (2001) 160. 3. S. H. Hahn, N. Barsan and U. Weimar, Sensors and Actuators B 78 (2001) 64. 4. H-M. Lin, S.-J. Tzeng, P.-J. Hsiau and W.-L. Tsai, NanoStruct. Mates. 10 (1998) 465. 5. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. 6. Z. L. Wang and Z. C. Kang, Functional and Smart Materials—Structural Evolution and Structure Analysis, Plenum Press, New York (1998). 7. M. S. Arnold, Ph. Avouris, Z. W. Pan and Z. L. Wang, J. Phys. Chem. B 107 (2003) 659. 8. Y. Cui and C. M. Lieber, Science 291 (2001) 851. 9. P. C. Collins, M. S. Arnold and Ph. Avouris, Science 292 (2001) 706. 10. J. Kong, N. Franklin, C. Wu, S. Pan, K. J. Cho and H. Dai, Science 287 (2000) 622. 11. Ph. Avouris, Chem. Phys. 281 (2002) 429. 12. B. G. Streetman, Solid State Electronic Devices, (4th ed.), Prentice Hall (1995). 13. G. N. Advani, P. Kluge-Weiss, R. L. Longini and A. G. Jordan, A Int. J. Electron. 48 (1980) 403. 14. B. Kamp, R. Merkle and J. Maier, Sensors and Actuators B 77 (2001) 534. 15. S. Samson and C. G. Fonstad, J. Appl. Phys. 44 (1973) 4618. 16. J. Mizusaki, H. Koinuma, J.-I. Shimoyama, M. Kawasaki and K. Fueki, J. Solid State Chem. 88 (1990) 443. 17. W. Göpel, K. Schierbaum and H.-D. Wiemhöfer, Solid State Ion. 32 (1989) 440. 18. C. P. Flynn, Point Defects and Diffusion, Clarendon Press (1972). 19. D. H. Zhang, Mater. Chem. Phys. 45 (1996) 248. 20. P. Bonasewicz, W. Hirschwald and G. Neumann, J. Electrochem. Soc. 133 (1986) 2270. 21. Y. Shapira, S. M. Cox and D. Lichtman, Surf. Sci. 54 (1976) 43. 22. T. L. Tansley and D. F. Neely, Thin Solid Films 121 (1984) 95. 22. N. Barsan, M. Schweizer-Berberich and W. Göpel, Fresenius J. Analy. Chem. 365 (1999) 287. 24. E. Comini, G. Faglia, G. Sberveglieri, Zhengwei Pan and Z. L. Wang, Appl. Phys. Lett. 81 (2002) 1869. 25. G. Sberveglieri, G. Faglia, S. Groppelli, P. Nelli and A. Camanzi, Semicond. Sci. Technol. 5 (1990) 1231. 26. M. Ferroni, V. Guidi, G. Martinelli, E. Comini, G. Sberveglieri, D. Boscarino and G. Della Mea, J. Appl. Phys. 88 (2000) 1097. 27. J. R. Morante, A. Cornet, J. Arbiol and A. Cirera, International Workshop Materials and Techmologies for Chemical Sensors (MATCHEMS), 13–14th September (2001). 28. V. E. Henrich and P. A. Cox, The Surface Science of Metal Oxides, Cambridge (1994) p. 464. 29. W. Göpel, Sensors Actuators A 56 (1996) 83. 30. L. G. C. Rego and G. Kirczenow, Phys. Rev. Letts. 81 (1998) 232; D. E. Angelescu, M. C. Cross and M. L. Roukes, Superlatt. and Microstruct. 23 (1998) 673. 31. P. Kim, L. Shi, A. Majumdar and P. L. McEuen, Phys. Rev. Lett. 87 (201) 215502; K. Schwab, E. A. Henriksen, J. M. Worlock and M. L. Roukes, Nature 404 (2000) 974.

Nanodevice, Nanosensors and Nanocantilevers

19

32. Li Shi and Z. L. Wang (to be published). 33. X. D. Bai, E. G. Wang, P. X. Gao and Z. L. Wang, Appl. Phys. Letts., in press; patent pending, Georgia Tech. 34. P. Poncharal, Z. L. Wang, D. Ugarte and W. A. de Heer, Science 283 (1999) 1513. 35. Z. L. Wang, P. Poncharal and W. A. de Heer, Pure Appl. Chem. 72 (2000) 209. 36. Z. L. Wang, P. Poncharal and W. A. de Heer, Microsc. Microanal. 6 (2000) 224. 37. R. P. Gao, Z. L. Wang, Z. G. Bai, W. A. de Heer, L. M. Dai and M. Gao, Phys. Rev. Lett. 85 (2000) 622. 38. Z. L. Wang, Adv. Mater. 12 (2000) 1295. 39. Z. L. Wang, Z. R. Dai, Z. G. Bai, R. P. Gao and J. Gole, Appl. Phys. Lett. 77 (2000) 3349. 40. L. Meirovich, Elements of Vibration Analysis, (2nd ed.) McGraw-Hill, New York (1986). 41. W. Hughes and Z. L. Wang, Appl. Phys. Lett. 82 (2003) 2886. Patent pending, Georgia Tech. 42. H. J. Dai, J. H. Hafner, A. G. Rinzler, D. T. Colbert and R. E. Smalley, Nature 384 (1996) 147.

Chapter 2 Oxide Nanowires and Nanolasers Peidong Yang and Haoquan Yan Department of Chemistry, University of California, Berkeley, CA 94720, USA

1. Introduction Nanoscale one-dimensional (1D) materials have stimulated great interest due to their importance in basic scientific research and potential technology applications [1–3]. Other than carbon nanotubes, one-dimensional (1D) nanostructures such as nanowires or quantum wires are ideal systems for investigating the dependence of electrical transport, optical and mechanical properties on size and dimensionality. They are expected to play important roles as both interconnects and functional components in the fabrication of nanoscale electronic and optoelectronic devices. Many unique and fascinating properties have already been proposed or demonstrated for this class of materials, such as superior mechanic toughness [4], higher luminescence efficiency [5], enhancement of thermoelectric figure of merit [6] and lowered lasing threshold [7]. Previously, nanowires with different compositions have been explored, using various methods including vapor phase transport process [8–10], chemical vapor deposition [11, 12], arc-discharge [13], laser ablation [12, 14], solution [5, 15] and template-based method [16, 17]. While a large part of these work has been focused on semiconductor systems such as Si [1, 12], Ge [8], GaN [9], GaAs [1, 11], it is only very recently that 1D oxide nanostructures start to emerge as very promising nanoscale building blocks because of their interesting properties, diverse functionalities, surface cleanness and chemical/thermal stability [2, 3, 7, 10, 18, 19]. One such example is the recent reports of synthesis of nanowires for oxides of zinc, tin, indium, cadmium, magnesium and gallium. Yang et al. reported the synthesis of MgO, Al2O3, ZnO, SnO2 nanowires via a carbon-thermal reduction process [18]. More recently, Wang et al. reported the synthesis of oxide nanobelts by simply evaporating the commercial metal oxide powders at high temperatures [19]. The as-synthesized oxide nanobelts are pure, structurally uniform and single crystalline, and most of them are free from defects and dislocations. They have a rectangle like cross section with typical widths of 30 to 300 nanometers, width-to-thickness ratios of 5 to 10, and lengths of up to a few

22

Yang and Yan

millimeters. The beltlike morphology appears to be a distinctive and common structural characteristic for the family of semiconducting oxides with cations of different valence states and materials of distinct crystallographic structures. These nanobelts could be an ideal system for fully understanding dimensionally confined transport phenomena in functional oxides and building functional devices along individual nanobelts. On the other hand, the interest in developing short-wavelength semiconductor lasers has culminated in the realization of room-temperature green-blue diode laser structures with ZnSe and InxGa1⫺xN as the active layers [20–22]. In this regard, zinc oxide (ZnO) is a wide band-gap (3.37 eV) semiconductor, and ultraviolet lasing action has been reported in disordered particles and thin films [23–25]. For wide band-gap semiconductors, a high carrier concentration is usually required in order to reach an optical gain that is high enough for lasing action in an electron-hole plasma (EHP) process [26]. Such an EHP mechanism, which is common for conventional laser diode operation, typically requires high lasing thresholds. As an alternative to EHP, excitonic recombination in semiconductors is a more efficient radiative process and can facilitate low-threshold stimulated emission [27, 28]. To achieve efficient excitonic laser action at room temperature, the binding energy of the exciton (E bex) must be larger than the thermal energy at room temperature (26 meV). ZnO is a good candidate for room temperature UV laser as its exciton binding energy is approximately 60 meV, significantly larger than that of ZnSe (22 meV) and GaN (25 meV). Nanostructures are expected to further lower the lasing threshold because quantum effects will result in a substantial density of states at the band edges and enhance radiative recombination due to carrier confinement. The use of semiconductor quantum well structures as low-threshold optical gain media represents a significant advancement in semiconductor laser technology [29]. Recently, light emission from semiconductor nanowhiskers has been reported in GaAs and GaP systems [30, 31]. Stimulated emission and optical gain have also been demonstrated recently in Si and CdSe nanoclusters and their ensembles [32, 33]. With these considerations, oxide, particularly ZnO nanowires are considered as interesting systems to examine their optical properties as functions of size and dimensionality [7, 10, 34–36]. Recently, we developed a simple vapor transport and condensation (CVTC) process for the synthesis of ZnO nanowires via the vapor–liquid–solid (VLS) mechanism [7, 8, 10]. In order to fully exploit the interesting optical properties of these nanowires, we have achieved several important structural control capabilities, namely, the orientation, position and diameter control. Grown in a preferred direction ⬍0001⬎, these wide band gap semiconductor nanowires form natural laser cavities with diameters 100–200 nm and lengths up to 40 ␮m. Under optical excitation, surface-emitting lasing action was observed at a near UV wavelength of 385 nm with emission line width ⬍0.3 nm. Power-dependent lasing spectra indicate the threshold of ~40 kW/cm2, which is much lower than the values reported for ZnO crystals and thin films (~300 kW/cm2). These short-wavelength nanolasers could have myriad applications including optical computing, information storage and nano-analysis. In addition, these oxide nanowires have also recently been shown to be excellent candidates as frequency converter, photodetectors and chemical or biological sensors.

Oxide Nanowires and Nanolasers

23

2. Synthesis and Characterization of ZnO Nanowires A schematic illustration of the chemical vapor transport and condensation (CVTC) system is shown in Fig. 1. Single crystalline silicon and sapphire of different orientations were used as substrates for the ZnO nanowire growth. The substrates were coated with a layer of Au thin film using thermal evaporator with a quartz crystal thickness monitor for exact Au film thickness. Equal amounts of ZnO powder and graphite powder were ground and transferred to an alumina boat. The Au-coated substrates and the alumina boat were placed into a small quartz tube. The substrates were typically placed 5–10 cm from the center of the boat. This quartz tube was then placed inside a furnace quartz tube, with the center of the alumina boat positioned at the center of the furnace and the substrates placed downstream of an argon flow. The temperature of the furnace was ramped to 800–1000 ⬚C at a rate of 50–100 ⬚C/min and typically kept at that temperature for 5–30 minutes under a constant flow of argon (20–25 sccm). After the furnace was cooled to room temperature, light or dark gray material was found on the surface of the substrates. For growing ZnO nanowires using Au nanoclusters, spin coating was used to disperse uniform Au colloids on the substrates. Although the VLS crystal growth mechanism has been widely used for semiconductor nanowire growth [1–3, 8], oxide nanowire growth through VLS mechanism could be complicated by the presence of the oxygen. In our studies, the process involves the reduction of ZnO powder by carbon to form Zn and CO/CO2 vapor at high temperature zone. The Zn vapor is transported and reacted with the Au solvent on substrates located downstream at a lower temperature to form alloy droplets (Fig. 2). As the droplets become supersaturated, crystalline ZnO nanowires are formed, possibly by the reaction between Zn and CO/CO2 at low temperature zone. The presence of small amount of CO/CO2 is not expected to significantly change the Au-Zn phase diagram, meantime they act as the oxygen source during the ZnO nanowire growth. Control experiments without graphite addition in the starting materials produce essentially nothing on the Au coated substrates, which indicates the importance of Zn vapor generated by the carbothermal reduction of ZnO.

Fig. 1. Schematic illustration of the chemical vapor transport and condensation experimental set-up for ZnO nanowire growth. Reprint with permission from [3], copyright Wiley-VCH, 2002.

24

Yang and Yan

Fig. 2. Schematic representation of ZnO nanowire growth mechanism. Reprint with permission from [3], copyright Wiley-VCH, 2002.

Fig. 3. Scanning electron microscopy images of ZnO nanowires grown on Si (100) substrate. Reprint with permission from [3], copyright Wiley-VCH, 2002.

Fig. 4. XRD pattern of ZnO nanowires on silicon substrate. Reprint with permission from [3], copyright Wiley-VCH, 2002.

Fig. 3 shows typical scanning electron microscopy (SEM) image of ZnO nanowires grown on a silicon substrate. The diameters of the nanowires normally range from 20 to 120 nm and their lengths are 5–20 ␮m. X-ray diffraction (XRD) patterns of ZnO nanowires were taken to examine the crystal structure of the nanowires.

Oxide Nanowires and Nanolasers (A)

25

(B)

Fig. 5. (A) TEM image of a ZnO nanowire with an alloy droplet on its tip, (B) High resolution transmission electron microscopy image of an individual ZnO nanowire showing its ⬍0001⬎ growth direction. Reprint with permission from [3], copyright Wiley-VCH, 2002.

All samples gave similar XRD patterns indicating the high crystallinity of nanowires. Fig. 4 shows typical XRD patterns of these ZnO nanowires. The diffraction peaks can be readily indexed to a hexagonal structure with cell constants of a ⫽ 3.24 Å and c ⫽ 5.19 Å. Additional structural characterization was carried out using transmission electron microscopy (TEM). Fig. 5A is a TEM image of a thin nanowire with an alloy tip. The presence of the alloy tip is a clear indication of the VLS growth mechanism. Fig. 5B shows a high-resolution TEM image of a single ZnO nanowire. The spacing of 2.56 ⫾ 0.05 Å between adjacent lattice planes corresponds to the distance between two (0002) crystal planes, confirming ⬍0001⬎ as the preferred growth direction for ZnO nanowires.

3. Controlled Growth of ZnO Nanowires 3.1. Orientation control Controlling the growth orientation is important for many of the proposed application of nanowires. By applying the conventional epitaxial crystal growth technique into this VLS process, it is possible to achieve precise orientation control during the nanowire growth. This technique, vapor–liquid–solid epitaxy (VLSE), is particularly powerful in controlled synthesis of nanowire arrays. Nanowires generally have preferred growth directions. For example, Si nanowires prefer to grow along ⬍111⬎ direction, ZnO nanowires prefer to grow along ⬍001⬎ direction. One strategy to grow vertically aligned nanowires is to properly select the substrate and to control the reaction conditions, so that the nanowires grow epitaxially on the substrate. For example, sapphire is an ideal substrate with lattice constant a ⫽ 4.75 Å and c ⫽ 12.94 Å. The c surface of sapphire is composed of alternate layers of six-fold symmetric oxygen and threefold symmetric Al atoms, while in the wurtzite structure of ZnO, both O and Zn are six-fold symmetric about the ZnO c axis. Since the ZnO a axis and the sapphire c are related almost exactly by a factor of 4 (mismatch

26

Yang and Yan

less than 0.08% at room temperature), ZnO nanowires can epitaxially grow from (110) plane of sapphire (Fig. 6A). Fig. 6B shows the growth of ZnO nanowires on an a-plane (110) sapphire substrate. Their diameters range from 40 to 120 nm and lengths can be adjusted between 2 and 10 microns. They are grown vertically from the substrate, as one can expect from the epitaxial growth. The hexagon end surface of the nanowires can be readily seen in Fig. 6C. This type of nanowire array growth can be also achieved using MgO (111) as substrates. Fig. 7 shows a typical XRD pattern recorded on these ZnO nanowire arrays. Only (001) diffraction peaks were observed indicating excellent (001) orientation/alignment of the nanowires on large area of the (A)

(C)

(B)

Fig. 6. The epitaxial growth of ZnO nanowires on a plane (110) sapphire can be readily seen by examination of the crystal structures of ZnO and sapphire (a ⫽ 0.4754 nm, c ⫽ 1.299 nm). (A) Schematic illustration of ZnO ab plane overlapping with the underlying (110) plane of sapphire substrate. (B–C) Arrays of ZnO nanowires grown on a-plane sapphire substrate. Reprint with permission from [3], copyright Wiley-VCH, 2002.

Fig. 7. XRD pattern of ZnO nanowires on sapphire substrate. Reprint with permission from [7], copyright American Association for the Advancement of Science, 2001.

Oxide Nanowires and Nanolasers

27

substrate. Previously, InAs nanowhiskers have been oriented on Si substrate using a fairly complicated metalorganic vapor-phase epitaxy technique [31]. 3.2. Position control It is apparent from the VLS nanowire growth mechanism that the positions of nanowires can be controlled by the initial positions of Au clusters or thin films. Various lithographical techniques including for example, soft lithography, e-beam and photolithography can be used to create patterns of Au thin film for the subsequent semiconductor nanowire growth [2–3]. SEM studies (Fig. 8) reveal extensive growth of fine, long and flexible ZnO nanowires from the edges of the hexagonally patterned Au thin film. The wire growth conforms to the hexagonal Au pattern in high fidelity. Most of the wires actually bridge the neighboring metal hexagons and form an intricate network. Fig. 8B shows an SEM image of the same sample at a higher magnification to reveal more detail. The nanowires at the edge have diameter of about 50–200 nm, and can be as long as over 50 ␮m. Fig. 9 shows the SEM images of ZnO nanowires grown from the line and square Au patterns on a-plane sapphire substrate. Selective epitaxial nanowire growth can be readily seen. It is clear that nanowires grow vertically only from the region that is coated with Au and form the designed patterns of ZnO nanowire array.

Fig. 8. SEM images of patterned ZnO nanowire network on Si substrate. Reprint with permission from [10], copyright Wiley-VCH, 2001.

Fig. 9. SEM images of epitaxial growth of line- and square-patterned ZnO nanowires on a-plane sapphire substrate. Reprint with permission from [3], copyright Wiley-VCH, 2002.

28

Yang and Yan

Fig. 10. ZnO nanowire arrays on sapphire substrate with low areal density. Reprint with permission from [3], copyright Wiley-VCH, 2002.

In addition to this positional control, it is further possible to control the nanowire areal density by modifying the thin film thickness or using the solution-made Au clusters. By dispersing different amount/density of Au clusters on the sapphire substrate, it is possible to obtain nanowire arrays with different densities (Fig. 10). We can now readily synthesize, for example, ZnO nanowire arrays with areal density spanning from 106 to 1010 cm⫺2. 3.3. Diameter control In an effort to grow ZnO nanowires with controllable diameters, substrates coated with different thickness of Au were used. The idea is based on a possible direct relationship between the sizes of the solvent particles and the resulting diameters of the nanowires. Smaller Au droplets, formed in the heating process during the reaction, should favor the growth of thinner wires. As the thickness of the Au thin film decreases, the widths of the wires also decrease. For example, the average diameters of ZnO nanowires grown on a-plane sapphire substrates at 900 ⬚C for 5 minutes are 88, 110 and 150 nm respectively when Au thin films of 0.5, 1 and 3 nm thick were used. The smallest diameter of ZnO nanowires we have achieved by using Au thin film is about 40 nm. In addition, we can also use uniformly distributed Au clusters as solvent to control the widths of the nanowires. The Au nanoclusters were dispersed on silicon substrates or within mesoporous silica films to minimize possible particle aggregation. For example, the average diameters of ZnO nanowires are 35, 46 and 54 nm respectively when Au clusters of 5, 10 and 15 nm sizes were used. Compared with the results obtained using Au thin film, the widths of the wires are much thinner when Au clusters were used, which can be as thin as 20 nm. This can be attributed to the fact that when Au thin film melts at high temperature to form Au droplets, there is a thermodynamic limit for the minimum radius of the metal liquid clusters at high temperature, Rmin ⫽ 2␴LVVL/RTlns, where ␴LV is the liquid-vapor surface free energy, VL is the molar volume of liquid and s is the vapor phase supersaturation [1–3].

29

Oxide Nanowires and Nanolasers (A)

(B)

(C)

(D)

Fig. 11. Different morphologies and superstructures based on ZnO nanowires synthesized using Zn as the vapor source. Reprint with permission from [3], copyright Wiley-VCH, 2002.

3.4. Morphological control By using different vapor source, it was recently discovered in this group that certain superstructures of the nanowires can be readily formed. For example, when Zn powder was used as vapor source, large yield of comb-like structures made of ZnO nanowires can be synthesized (Fig. 11). These nanowires have uniform diameters and are uniformly distributed on the side of the stem. Other superstructures such as tetra-pods and tapered nanowires can be also formed under different evaporation conditions. These different superstructures may have interesting physical properties, particularly optical properties.

4. Photoluminescence and Lasing Properties Photoluminescence spectra of nanowires of different diameters were measured using He-Cd laser (325 nm) as the excitation source. Fig. 12 shows the room temperature PL spectra of nanowires with an average diameter of 100, 50 and 25 nm. Strong emission at ~380 nm was observed for all nanowires. In addition, we observed increasing green emission intensity at ~520 nm when decreasing the nanowire sizes. While the UV emission corresponds to the near band-edge emission, the green emission peak is commonly referred to as deep-level or trap-state emission. The green emission has been attributed to the singly ionized oxygen vacancy in ZnO and the emission results from the radiative recombination of a photo-generated hole with an electron occupying the oxygen vacancy [37]. The progressive increase of the green emission relative to the UV emission as the wire diameter decreases suggests that there is a greater fraction of oxygen vacancies in the thinner nanowires. We believe higher surface area to volume ratio for thinner wires might favor higher level of surface and sub-surface oxygen vacancy under the current reductive synthetic environment [10]. In order to explore the possible stimulated emission from the oriented nanowires, the power dependent emission has been examined. The samples were optically

30

Yang and Yan

Fig. 12. Photoluminescence spectra of ZnO nanowires of different diameter recorded at room temperature. Spectra A, B, C correspond to nanowires with average diameters of 100, 50 and 25 nm, respectively. Reprint with permission from [10], copyright Wiley-VCH, 2001.

(A)

(B)

Fig. 13. (A) Schematic illustration of the experimental set-up for the power dependent emission studies. (B) Emission spectra from vertical ZnO nanowire arrays on a-plane sapphire substrate below and above the lasing threshold. The spectra are offset for easy comparison. Reprint with permission from [3], copyright Wiley-VCH, 2002.

pumped by the fourth harmonic of Nd : YAG laser (266 nm, 3 ns pulse width) at room temperature. The pump beam was focused on nanowires at an incidence angle 10 degree to the symmetric axis of the nanowire. Light emission was collected in the direction normal to the end surface plane (along the symmetric axis) of the nanowires (Fig. 13A). Significantly, in the absence of any fabricated mirrors, we observed lasing action in these ZnO nanowires. Fig. 13B shows the evolution of the emission spectra as we increased the pump power [7]. At low excitation intensity, the spectrum consists of a single broad spontaneous emission (Fig. 13B bottom trace) peak with a full width at half maximum (FWHM) of approximately 17 nm. This spontaneous emission is 140 meV below the band gap (3.37 eV) and is generally ascribed to the recombination of excitons through exciton-exciton collision process where one of the exciton radiatively recombines to generate a photon. As the pump power increases, the emission peak narrows due to the preferential amplification of frequencies close to the

Oxide Nanowires and Nanolasers

31

maximum of the gain spectrum. When the excitation intensity exceeds a threshold (~40 kW/cm2), sharp peaks emerge in the emission spectra (Fig. 13B upper trace). The line widths of these peaks are less than 0.3 nm, which are more than 50 times smaller than the line width of the spontaneous emission peak below the threshold. Above the threshold, the integrated emission intensity increases rapidly with the pump power. The narrow line width and the rapid increase of emission intensity indicate that stimulated emission takes place in these nanowires. The observed single or multiple sharp peaks represent different lasing/cavity modes at wavelengths between 370 and 400 nm. It’s noticed that the lasing threshold is quite low compared with previously reported values for random lasing (~300 kW/cm2) in disordered particles or thin films. The fact that we observed lasing action in these nanowire arrays without any fabricated mirror prompts us to consider these single-crystalline, well-facetted nanowires as natural resonance cavities (Fig. 14). It’s possible that the giant oscillator strength effect, which can occur in high quality nanowire crystals with dimensions larger than the exciton Bohr radius, but smaller than the optical wavelength, enables the excitonic stimulated emission in these nanowire arrays. For II-VI semiconductors, cleaved edge of the specimen is usually used as a mirror [38]. For our nanowires, one end is the epitaxial interface between the sapphire and ZnO while the other end is the sharp (0001) plane of the ZnO nanocrystals. Both can serve as good laser cavity mirrors considering the refractive indexes for sapphire, ZnO and air are 1.8, 2.45 and 1, respectively. This natural cavity/waveguide formation in nanowires suggests a simple chemical approach to form a nanowire laser cavity without cleavage and etching. In fact, when multiple lasing modes were observed for these nanowires, the observed mode spacing is about 6 nm for ~5 ␮m long wires, which agrees quantitatively well with the calculated spacing between adjacent resonance frequencies ␷F ⫽ c/2nl, where ␷F is emission mode spacing, c the light speed, n the refractive index and l the resonance cavity length. To further validate this proposed model, isolated nanowire lasing experiments were carried out since there’s argument that the lasing might originate from the amplified stimulated emission based on random scattering. Fig. 10 shows the SEM image of those isolated ZnO nanowires used in this study. The average spacing between these wires is about 2–3 ␮m, so it’s almost impossible to amplify the emission light by scattering around in such sparsely distributed nanowire array. Fig. 15 shows a lasing spectrum from these isolated nanowires, which is similar to those obtained on dense arrays. This suggests that the lasing is indeed originated from single nanowire, not by amplification of scattering in the dense nanowire array. In addition, lifetime measurements

Fig. 14. Schematic illustration of a nanowire as a resonance cavity with two naturally faceted hexagonal end faces acting as reflecting mirrors.

32

Yang and Yan

Fig. 15. Emission spectrum above the lasing threshold collected from vertical ZnO nanowire arrays of low areal density on a-plane sapphire substrate. Reprint with permission from [3], copyright Wiley-VCH, 2002.

show that the radiative recombination of the excitons is a superposition of a fast and a slow process with time constants of about 70 ps and 350 ps, respectively [7, 10]. The luminescence lifetime is mainly determined by the concentration of defects, which trap the electrons and/or holes and eventually cause their nonradiative recombination. Although the exact origin of the luminescence decay remains unclear at this stage, the long lifetime measured for these wires demonstrates the high crystal quality achieved with the nanowire growth process. Meantime, it also accounts in part for the low lasing threshold of the wires. Furthermore, single nanowire lasing experiments by near-field scanning optical microscope (NSOM) was also carried out to resolve the lasing mechanism [36]. In order to be imaged with NSOM, the wires were removed from the sapphire growth substrate by sonication and dispersed onto a quartz substrate by drop casting an ethanol-nanowire mixture. The nanowires were excited with short pulses (⬍300 fs) of 4.35 eV photons (285 nm) in order to induce photoluminescence (PL) and lasing (Fig. 16). The nanowire emission was collected by a chemically etched fiber optic probe held in constant-gap mode by the feedback electronics of the NSOM. Fig. 17 shows a topographic (A) and an NSOM PL (B) image of a single nanowire. The wire in the topographical image has a length of 5 ␮m and a diameter of about 150 nm, as determined by the height of the wire on the substrate. The corresponding optical image in Fig. 17B shows the spatially-resolved emission collected in a 70 nm wide spectral band centered at 400 nm. Strong PL emission at the end of the nanowire can be clearly seen, while only very weak emission was detected from the side surfaces of the nanowire. The PL spectrum collected at the nanowire end (Fig. 17C), corresponds well with the broad ZnO spontaneous emission. The enhanced intensity at the ends of the wire indicates increased scattering of the PL photons as they reach the end of the waveguide. Fig. 18 shows a spatially-resolved emission above the lasing threshold collected in a 70 nm wide spectral band centered at 400 nm. Strong stimulated emission can be clearly seen at the end of the nanowire, while only very weak emission was detected

Oxide Nanowires and Nanolasers

33

Fig. 16. Illustration of NSOM set-up for power dependent emission studies on single ZnO nanowire. Reprint with permission from [3], copyright Wiley-VCH, 2002.

(A)

(C)

(B)

Fig. 17. Near-field images of a single ZnO nanowire waveguide. (A) Topographical and (B) PL NSOM image size is (10 ␮m)2, with maximum topographical height of 140 nm. The sample was excited with approximately 100 nJ pulses at 285 nm, resulting in an excitation intensity of about 200 kW/cm2. (C) The PL spectrum from the wire (FWHM ~20 nm). Reprint with permission from [36], copyright American Chemical Society, 2002.

34

Yang and Yan

Fig. 18. Spatially-resolved NSOM emission image above the threshold collected on a single ZnO nanowire in a 70 nm wide spectral band centered at 400 nm. Reprint with permission from [36], copyright American Chemical Society, 2002.

from the side surfaces of the nanowire. Significant peak narrowing is observed for the emission emanating from the end of the nanowire. The power dependence of the nanowire emission exhibits a clear threshold near excitation intensity of approximately 120 kW/cm2 (using a sub-picosecond laser pulsewidth). This threshold is higher than that observed from bulk wires, though it is the same order of magnitude. The single nanolasers dispersed onto the quartz substrate are likely to exhibit a higher threshold than those attached to the growth substrate, considering the possible damage to the end faces that occurs during sample preparation. Nanowire coupling could also contribute to lower the lasing threshold for the nanowire arrays. The spectral width of the laser peak measured directly above the wire is about 3.0 nm, which is quite broad compared to previously observed ZnO lasing (⬍0.3 nm). This is also much wider than the instrumental spectral resolution (⬍0.4 nm), suggesting that the spectrum collected from above the wire originates from both lasing modes and nonlasing PL that is not sustained in the cavity, either because its direction of propagation is non-axial or because its wavelength does not correspond with one of the longitudinal cavity modes. Laser intensity along the wire axis, but far from the nanowire, is only strong if the photons are directed by the cavity, thus it is expected that the linewidth of that emission would be narrower. Another factor that could contribute to peak broadening is the low reflectivity of the nanowire end faces (18%, based on Fresnel equations), which leads to short average cavity confinement times. Thus, many modes that are poorly supported in the cavity can experience gain during the excitation pulse period but do not completely destructively interfere in only a few passes through the wire (3 reflections produce ⬎99% transmission losses). Therefore, the overall cavity confinement time is likely to be approximately equal to the excitation lifetime, since this is the only period in which the optical gain (associated with stimulated emission)

Oxide Nanowires and Nanolasers

35

will be able to counteract reflection losses. A simulation for photons taking three full passes through the cavity (about 300 fs) produced a linewidth of about 1.0 nm, which corresponds well with the observed lasing linewidth in the nanolaser beam.

5. Nonlinear Optical Mixing in Single Zinc Oxide Nanowires In addition to the interesting photoluminescence and lasing properties of these oxide nanowires, their nonlinear optical properties suggest other important potential applications as frequency converters or logic/routing elements in nanoscale optoelectronic circuitry. A linear optical property of nanowires, photoluminescence (PL) polarization, has recently been studied in single indium phosphide nanowires [30]. In that case, the PL polarization is based upon the classical electromagnetic properties of a dielectric cylinder and averages ca. 91%. In contrast, coherent nonlinear optical phenomena, such as second- and third-harmonic generation (SHG and THG, respectively), depend explicitly on the crystal lattice structure of the medium, which could yield a very high (nearly 100%) polarization dependence. In addition, the temporal response of non-resonant harmonic generation is similar to the pulsewidth of the incident laser, in some cases ~20 fs, while incoherent processes are at least 2–4 orders of magnitude slower. Moreover, non-resonant SHG is essentially independent of wavelength below the energy band gap of semiconductor materials, most often including the 1.3–1.5 ␮m wavelength region typically used in optical fiber communications. A material of particular interest is zinc oxide (ZnO). Studies of microcrystalline ZnO thin films [39, 40] have revealed a large second-order nonlinearity, characterized by (2), which determines the efficiency of a material as a converter of optical frequencies via several processes (e.g., second-harmonic generation (SHG) and sum- and difference-frequency generation (SFG, DFG)). To examine the nonlinear optical properties of the ZnO nanowires, we employ oblique collection mode NSOM, in which the sample is illuminated in the far-field, as it is preferred for nonlinear near-field imaging because of its suitability for high incident pulse intensity experiments [35]. Fig. 19 depicts the experimental laser beam geometry with respect to an arbitrarily oriented nanowire on the sapphire substrate. The hexagonal crystal structure gives the nanowires six prismatic side faces, each of which is oriented at a distinct angle to the incident beam. The face labeled “front” is nearly normal to the incident beam (for ~0). This particular laser beam/nanowire geometry produces a pattern of three parallel areas of signal along the wire. Fig. 20 shows two series of SHG images, illustrating the input polarization and nanowire orientation dependence of the SHG. This dependence arises from the two independent, non-vanishing components of (2) observed in SHG for ZnO, (2) zxx and  (2) zzz. In Fig. 20A–B, the wires are situated approximately normal to each other, with an s-polarized incident beam in Fig. 20A and p-polarization in Fig. 20B. The polarization ratio (SHGs-inc/(SHGp-inc ⫹ SHGs-inc)) for wire 1 is 0.90. Wire 2 (oriented at ~0) has an average signal that is 2.5 times that of wire 1 (oriented at ~90⬚). To quantitatively analyze the SHG polarization effect, polarization traces were taken on several wires from Fig. 20A–B, D–E. The near-field probe was maintained above each wire, and the input polarization was rotated as the SHG signal was

36

Yang and Yan

Fig. 19. Illustration of the sample/beam geometry for NSOM study of nanowire SHG. The surface normal and the fundamental beam k-vector form the angle , which was fixed during the experiment. The angle  is formed between the nanowire symmetric axis and the normal to the propagation vector of the incident beam in the sample plane. The polarization angle, , is varied during the experiment. Inset shows cross-sectional view of the fundamental (dashed) and SHG (solid) beam geometry. Two primary SHG paths are shown: one transmitted through the top face, and the other through the side face. The transmitted angle (measured from the surface normal) through the top face is approximately 24⬚. Reprint with permission from [35], copyright American Chemical Society, 2002.

monitored (Fig. 20C and F). The theoretical traces were computed. The polarization data were fit to theory [41], and nearly all the wires tested (diameters 80–100 nm) exhibited (2) a ratio (2) zzz/ zxx of approximately 2.0–2.3, while one larger wire (diameter 120–130 nm) (2) exhibited  zzz /(2) zxx ⫽ 4.2. These ratios can be compared with a value of 3.0 for bulk crystalline ZnO (22) to as high as 6.0 for polycrystalline thin films [39]. (2) zxx represents the second-order lattice response perpendicular to the symmetric axis, which is the direction most confined by the quasi-1D nature of the nanowire. Thus, one might expect (2) (2) (2) (2) the enhancement of (2) zxx to be larger than that of  zzz, given that eff ⫽  bulk ⫹ surface, (2) (2) where  surface can be enhanced relative to  bulk(24), and that surface effects have a larger (2) (2) contribution to (2) has been studied for spherically zxx than to  zzz. Enhancement of  symmetric quantum materials [42], but no previous study of the enhancement of individual components of the (2) tensor has been performed. Using a reference material, one can determine the absolute magnitude of each (2) component of (2) for individual nanowires. (2) zxx and  zzz were determined by measuring the nanowire SHG with respect to a zinc selenide (ZnSe) disk (at 1.4 ␮m excitation). Using the ZnSe reference (78 pm/V), (2) zzz for a single ZnO wire (not shown) ⫽ (2) 5.5 pm/V and (2) zxx ⫽ 2.5 pm/V. The  zzz value is considerably lower than the reported bulk value (18 pm/V) but in relatively good agreement with values reported for ZnO thin films (4–10 pm/V). One of the possible reasons for the lower value compared to bulk ZnO is that the number of ZnO molecules probed for a single nanowire is less than those probed on a solid disc. If one considers the fiber probe collection region to be roughly a cylinder (with a diameter equal to the optical aperture and a length equal to the depth of field) below the probe, then the nanowire will not necessarily fill this cylinder whereas a solid disc does. Estimates of the loss of signal from insufficient filling make the nanowire (2) zzz value greater by 1.5–3 times, resulting in better agreement with the aforementioned far-field experiments.

Oxide Nanowires and Nanolasers (A)

(D)

(B)

(E)

(C)

(F)

37

Fig. 20. Nanowire SHG polarization dependence. (A) Combined topographical and SHG image of two wires at angle of approximately 90⬚. Image size is (13 ␮m)2 and maximum topographic height is 130 nm. Beam is s-polarized, incident from the right. (B) SHG of same region as in (A) with p-polarized incident beam. (C) Polarization-dependent SHG data and theoretical predictions taken from the wires labeled in (A). The theoretical curves are calculated 2 for the SHG signal and (2) eff for a hexagonal crystal. (D) Image of (16 ␮m) region showing several nanowires. Maximum topographic height is 120 nm. The polarization  ⫽ ⫺45⬚. (E) Image of the same region as in (D) with polarization  ⫽ 45⬚. (F) Polarization traces taken from the wires labeled 1 and 2 in (C) and theoretical simulations. Reprint with permission from [35], copyright American Chemical Society, 2002.

6. Photoconductive Oxide Nanowires as Nanoscale Optoelectronic Switches Among all possible nano-devices, switches are critical for important applications like memory and logic. Electrical switching on nanometer and molecular level has been predominantly achieved through proper electrical gating configuration, as exemplified by nanotube transistors. We have demonstrated that it is possible to create highly sensitive electrical nanowire switches by exploring the photoconductivity of

38

Yang and Yan

the individual semiconductor nanowires. We found that the conductance of the ZnO nanowire is extremely sensitive to ultraviolet light exposure. The light induced insulator-to-conductor transition allows us reversibly switching the nanowires from OFF to ON states, an “optical gating” phenomena as compared to commonly used electrical gating. Four-probe measurement of individual ZnO nanowires indicates that these nanowires are essentially insulating in dark with a resistivity above 3.5 M⍀cm⫺1. When the nanowires were exposed to UV-light with wavelengths below 400 nm, it was found that the nanowire resistivity instantly decreases by typically 4 to 6 orders of magnitude [34]. In addition to the high sensitivity of the nanowire photoconductor, they also exhibit excellent wavelength selectivity. Fig. 21A shows the evolution of the photocurrent when a nanowire was exposed first to highly intense light at 532 nm (Nd : YAG, 2nd harmonics) for 200 s and then to UV-light at 365 nm. There is no photoresponse at all to the green light while exposure to less intense UV-light shows the typical change of conductivity of 4 orders of magnitude. Measurements of the spectral response show that our ZnO nanowires indeed have a cut-off wavelength of 385 nm which is expected from the bandgap of ZnO. It has been unambiguously established that oxygen chemisorption plays a profound role in enhancing photosensitivity of bulk or thin film ZnO [43, 44]. We believe that a similar photoresponse mechanism could be applied to the nanowire system with additional consideration of high surface area of the nanowire, which could further enhance the sensitivity of the device. It was generally believed that the photoresponse of ZnO consists of two parts: a solid-state process where an electron and a hole are created (h␯ → h⫹ ⫹ e⫺) and a two-step process involving oxygen species adsorbed on the surface. In the dark, oxygen molecules adsorb on the oxide surface as a negatively charged ion by capturing free electrons of the n-type oxide semiconductor (O2(g) ⫹ e⫺ → O ⫺ 2 (ad)) thereby creating a depletion layer with low conductivity near the nanowire surface. Upon exposure to UV-light, photo-generated holes migrate to the

(A)

(B)

Fig. 21. (A) Sensitivity of the photoresponse of a ZnO nanowire to light exposure at wavelengths of 532 nm and 365 nm. (B) Reversible switching of a ZnO nanowire between low and high conductivity states. Reprint with permission from [34], copyright Wiley-VCH, 2002.

Oxide Nanowires and Nanolasers

39

surface and discharge the negatively charged adsorbed oxygen ions (h⫹ ⫹ O⫺ 2 (ad) → O2(g)) through surface electron-hole recombination. Meantime, photo-generated electrons destruct the depletion layer, as a result, the conductivity of the nanowire increase significantly. The characteristics of the photoconducting ZnO nanowires leads us to the conclusion that they could be good candidates for optoelectronic switches, i.e. the insulating state as “OFF” in the dark and the conducting state as “ON” when exposed to UV-light. Fig. 21B plots the photoresponse as a function of time while the UV-lamp was switched on and off. It is evident that the nanowires can be reversibly switched between the low conductivity state and the high conductivity state. With further optimization of the nanowire composition, e.g. through proper doping, the photocurrent decay time (i.e. the response time) could be reduced to ␮s level. These highly sensitive photoconducting nanowires could serve as very sensitive UV-light detector in many applications such as microanalysis and missile plume detection, as well as fast switching devices for nanoscale optoelectronics applications where ON and OFF states can be addressed optically.

7. Room Temperature NO2 Photochemical Sensing Another major area of application for nanowires and nanotubes is likely to be the sensing of important molecules, either for medical or environmental health purposes. The ultrahigh surface to volume ratios of these structures make their electrical properties extremely sensitive to surface-adsorbed species, as recent work has shown with carbon nanotubes [45, 46], functionalized silicon nanowires and metal nanowires [47, 48]. Chemical nanosensors are interesting because of their potential for detecting very low concentrations of biomolecules or pollutants on platforms small enough to be used in vivo or on a microchip. Recently we have demonstrated the first room temperature photochemical NO2 sensors based on individual single-crystalline oxide nanowires and nanoribbons. Here the concept is illustrated using SnO2 nanowires as an example [49]. Tin dioxide is a wide bandgap (3.6 eV) semiconductor. For n-type SnO2 single crystals, the intrinsic carrier concentration is primarily determined by deviations from stoichiometry in the form of equilibrium oxygen vacancies, which are predominantly atomic defects [5]. The electrical conductivity of nanocrystalline SnO2 depends strongly on surface states produced by molecular adsorption that result in spacecharge layer changes and band modulation [51]. NO2, a combustion product that plays a key role in tropospheric ozone and smog formation, acts as an electron-trapping adsorbate on SnO2 crystal faces and can be sensed by monitoring the electrical conductance of the material. Because NO2 chemisorbs strongly on many metal oxides [52], commercial sensors based on particulate or thin-film SnO2 operate at 300–500 ⬚C to enhance the surface molecular desorption kinetics and continuously “clean” the sensors [53]. The high temperature operation of these oxide sensors is not favorable in many cases, particularly in an explosive environment. We have found that the strong photoconductive nature of individual single crystalline SnO2 nanoribbons [19] makes it possible to achieve equally favorable adsorption-desorption behavior at room

40

Yang and Yan

Fig. 22. A schematic longitudinal cross-section of a nanoribbon in the dark and in UV light. In the illuminated state, photo-generated holes recombine with trapped electrons at the surface, desorbing NO2 and other electron trapping species: h⫹ ⫹ NO2⫺(ads) → NO2(gas). The space-charge layer thins, and the nanoribbon conductivity rises. Ambient NO2 levels are tracked by monitoring changes in conductance in the illuminated state. Reprint with permission from [49], copyright Wiley-VCH, 2002.

temperature by illuminating the devices with ultraviolet (UV) light of energy near the SnO2 bandgap. The active desorption process is thus photo-induced molecular desorption (Fig. 22) [54]. We analyzed the optoelectronic response of these devices in air and NO2 environments in order to probe their chemical sensing abilities. The behavior of a representative nanoribbon is shown in Fig. 23A. In the dark and in pure air (troughs of the curve), the nanoribbons had resistances ranging from 500 M⍀ to 12 G⍀. When exposed to ultraviolet light with a wavelength of 254 nm (intensity ⫽ 0.63 mW cm⫺2; large peak of the curve), which corresponds to an energy significantly greater than the SnO2 bandgap (345 nm), the nanosensor conductance increased by 3 to 4 orders of magnitude and stabilized within 1–2 minutes. The conductance rise is due both to the generation of photocurrent, which directly increases the number of free carriers within the device, and photodesorption of surface species (mostly O2- and H2O-derived), with a concomitant thinning of the electron depletion layer near the nanoribbon surface. The effect was fully reversible when the light was turned off, with 99.9% decay of the photoresponse in 300–500 seconds. Illumination with 365 nm radiation (intensity ⫽ 0.5 mW cm⫺2; small peaks of curve Fig. 23A) also resulted in a photoresponse, typically a 10- to 100-fold increase in the nanoribbon conductance, with slightly faster rise and decay constants than in the 254 nm case. This photoswitching behavior was reproducible at both UV wavelengths indefinitely. When the nanoribbons were tested in an atmosphere of 100 ppm NO2 in air (Matheson Tri-Gas), resistance values were higher for all three states—dark, 254-exposed and 365-exposed (Fig. 23B)—compared to their respective values in pure air. The photoresponse decays were also significantly faster in 100 ppm NO2, with 99.9%

Oxide Nanowires and Nanolasers

41

Fig. 23. Photoresponse of a single nanoribbon in pure air (A) and 100 ppm NO2 (B). The large peaks of both curves correspond to 254 nm UV illumination, the small peaks to 365 nm illumination, and the troughs to dark current. The signal ratio is 45 under 365 nm light and 4 in 254 nm light. Bias is 1.0 V. Reprint with permission from [49], copyright Wiley-VCH, 2002.

Fig. 24. Cycling a nanosensor near its resolution limit under 365 nm light. NO2 concentrations are indicated. Horizontal bars are signal averages. The average signal difference for the three cycles is 16%. Bias is 0.5 V. Reprint with permission from [49], copyright Wiley-VCH, 2002.

falloffs in 35–50 seconds. This faster decay rate can be attributed to the strong adsorption and electron-trapping interaction of NO2 on SnO2 surfaces. Comparing the two curves in Fig. 23 shows that the current/conductance difference between the device operating in air and in 100 ppm NO2 was larger under 365 nm than 254 nm illumination, so that greater sensitivity to NO2 occurred using the longer wavelength. Note that UV is vital for sensing, as NO2 adsorption is irreversible in the dark. To determine the behavior of the nanosensors under realistic operating conditions, they were cycled through different NO2 concentrations under continuous 365 nm illumination and in gas flows of 150 sccm (standard cubic centimeters per minute). The device showed greater sensitivity to low NO2 concentrations (⬍15 ppm), while higher concentrations caused smaller signal changes as the nanoribbon

42

Yang and Yan

surface became saturated with NO2. In addition, the signal noise decreased at higher concentrations. The resolution limit achieved by these nanoribbons fell between 2 and 10 ppm. Fig. 24 shows the conductance response of one nanosensor cycled between pure air and 3 ppm NO2. Even with the low signal/noise ratio, current steps can be clearly distinguished as the NO2 was turned on and off. This behavior was stable for ⬎20 cycles without appreciable drift and with response times of less than one minute. The average response ratio (IAIR/INO2) at 3 ppm NO2 was 1.16.

8. Conclusions and Outlook We have surveyed the recent research activities on oxide nanowire synthesis and their interesting optical properties. The synthesis of ZnO nanowires was carried out via a VLS growth mechanism using Au as the solvent. Orientation control of the nanowires was achieved by epitaxial growth on sapphire substrates. By controlling the pattern of the Au thin film, we succeeded in growing ordered ZnO nanowire arrays. ZnO nanowires of different sizes and densities were grown by varying the thickness of Au layers or the diameter of Au clusters. Room temperature photoluminescence of the nanowires shows the near band-edge emission and the deep level emission. An enhanced deep-level emission for thinner nanowires has been observed and attributed to their larger surface area. Furthermore, room temperature ultraviolet lasing in well-oriented vertical ZnO nanowires with lasing threshold of 40 kW/cm2 was demonstrated. Grown in ⬍0001⬎ direction, these single-crystalline, well facetted nanowires form natural laser resonance cavities. This proposed lasing model has been further validated by the controlled lasing experiments on single nanowire lasing experiments by NSOM. It should be emphasized that the concept of using well-cleaved nanowires as natural optical cavities should be applicable to many other different semiconductor systems such as GaN and CdSe. Our results obtained in ZnO nanowire system suggest the feasibility of nanoscale surface-emitting lasers operating at ultraviolet or other wavelengths when the material of the nanowire cavity is altered. In addition, by creating pn junctions in these individual nanowires, one should be able to test the possibility of making electron ejection UV/blue lasers out of individual nanowires. Such miniaturized nanowire nanolasers will find applications in nano-photonics and microanalysis. In addition, the nanowire SHG was found to be primarily wavelength independent (SHG ⬎ 400 nm) and relatively efficient, with a larger (2) ⱖ 5.5 pm/V than betaeff barium borate (BBO, (2) 艐 2.0 pm/V) a commonly-used doubling crystal. The noneff linear optical properties demonstrated here suggest that ZnO nanowires could be effectively employed as frequency converters or logic components in nanoscale optoelectronics. Individual photoconductive oxide nanowires/ribbons are small, fast and sensitive devices for detecting ppm-level NO2 at room temperature in UV light. The advantages of low-temperature, potentially drift-free operation make these nanoribbons/wires good candidates for miniaturized, ultra-sensitive gas sensors in many applications.

Oxide Nanowires and Nanolasers

43

Single molecule chemical detection using nanowires may be within reach in near future. Further studies on these oxide nanowires will for sure leads to great insight in size and dimensionality controlled physical properties and important optoelectronic applications.

Acknowledgments This work was supported by the Camille and Henry Dreyfus Foundation, 3M Corporation, the National Science Foundation, and the University of California, Berkeley. P. Y. is an Alfred P. Sloan Research Fellow. Work at the Lawrence Berkeley National Laboratory was supported by the Office of Science, Basic Energy Sciences, Division of Materials Science of the U.S. Department of Energy. We thank the National Center for Electron Microscopy for the use of their facilities. P. Y. graciously acknowledges very productive and pleasant collaboration with Prof. R. Saykally and the students and postdoctoral fellows who have carried out the work described in this review, including M. Huang, J. Johnson, M. Law and H. Kind.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

P. Yang, Y. Wu and R. Fan, Inter. J. Nano. 1 (2002) 1. Y. Wu, H. Yan and P. Yang, Chemistry, Euro. J. 8 (2002) 1260. P. Yang et al., Adv. Func. Mater. 12 (2002) 323. E. W. Wang, P. E. Sheehan and C. M. Lieber, Science 277 (1997) 1971. J. D. Holmes, K. P. Johnston, R. C. Doty and B. A. Korgel, Science 287 (2000) 1471. L. D. Hicks and M. S. Dresselhaus, Phys. Rev. B47 (1996) 16631. M. Huang, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind, E. Weber, R. Russo and P. Yang, Science 292 (2001) 1897. Y. Wu and P. Yang, Chem. Mater. 12 (2000) 605; Y. Wu, B. Messer and P. Yang, Adv. Mater. 13 (2001) 1487; Y. Wu and P. Yang, J. Am. Chem. Soc. 123 (2001) 3165. C-C. Chen and C-C. Yeh, Adv. Mater. 12 (2000) 738. M. H. Huang, Y. Wu, H. Feick, E. Weber and P. Yang, Adv. Mater. 13 (2001) 113. M. Yazawa, M. Koguchi, A. Muto, M. Ozawa and K. Hiruma, Appl. Phys. Lett. 61 (1992) 2051. Y. Wu, R. Fan and P. Yang, Nano Lett. 2 (2002) 83. Y. C. Choi, W. S. Kim, Y. S. Park, S. M. Lee, D. J. Bae, H. Y. Lee, G-S. Park, W. B. Choi, N. S. Lee and J. M. Kim, Adv. Mater. 12 (2000) 746. X. F. Duan and C. M. Lieber, Adv. Mater. 279 (2000) 208; A. M. Morales, C. M. Lieber, Science 279 (1998) 208. T. J. Trentler, K. M. Hickman, S. C. Goel, A. M. Viano, P. C. Gibbons and W. E. Buhro, Science 270 (1995) 1791. M. H. Huang, A. Choudrey and P. Yang, Chem. Commun. 12 (2000) 1603; J. Zhu and S. Fan, J. Mater. Res. 14 (1999) 1175. Y. Li, G. W. Meng, L. D. Zhang and F. Philipp, Appl. Phys. Lett. 76 (2000) 2011. P. Yang and C. M. Lieber, Science 273 (1996) 1836. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. D. A. Gaul and W. S. Rees, Jr., Adv. Mater. 12 (2000) 935.

44

Yang and Yan

21. M. A. Hasse, J. Que, J. M. De Puydt and H. Cheng, Appl. Phys. Lett. 59 (1991) 1272. 22. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, H. Kiyoku and Y. Sugimoto, Jpn. J. Appl. Phys. 35 (1996) L74. 23. H. Cao, J. Y. Xu, D. Z. Zhang, S. H. Chang, S. T. Ho, E. W. Seeling, X. Liu and R. P. H. Chang, Phys. Rev. Lett. 84 (2000) 5584. 24. D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen and T. Goto, Appl. Phys. Lett. 70 (1997) 2230. 25. P. Yu, Z. K. Tang, G. K. L. Wong, M. Kawasaki, A. Ohtomo, H. Koinuma and Y. Segawa, J. Cryst. Growth 184/185 (1998) 601. 26. C. Klingshirn, J. Cryst. Growth 117 (1992) 753. 27. Y. Kayamura, Phys. Rev. B 38 (1988) 9797. 28. W. Wegscheider, L. N. Pfeiffer, M. M. Dignam, A. Pinczuk, K. W. West, S. L. McCall and R. Hull, Phys. Rev. Lett. 71 (1993) 4071. 29. D. Mehus and D. Evans, Laser Focus World 31 (1995) 117. 30. J. F. Wang, M. S. Gudiksen, X. F. Duan, Y. Cui and C. M. Lieber, Science 293 (2001) 1455. 31. K. Hiruma, M. Yazawa, T. Katsuyama, K. Ogawa, K. Haraguchi, M. Koguchi and H. Kakibayashi, J. Appl. Phys. 77 (1995) 447. 32. V. I. Klimov, A. A. Mikhailovsky, S. Xu, A. Malko, J. A. Hollingsworth, C. A. Leatherdale, H. J. Eisler and M. G. Bawendi, Science 290 (2000) 314. 33. L. Pavesi, L. D. Negro, C. Mazzoleni, G. Granzo and F. Priolo, Nature 408 (2000) 440. 34. H. Kind, H. Yan, M. Law, B. Messer and P. Yang, Adv. Mater. 14 (2002) 158. 35. J. C. Johnson, H. Yan, R. D. Schaller, P. B. Peterson, P. Yang and R. J. Saykally, Nano Lett. 2 (2002) 279. 36. J. Johnson, H. Yan, R. Schaller, L. Haber, R. Saykally and P. Yang, J. Phys. Chem. B 105 (2001) 11387. 37. K. Vanheusden, W. L. Warren, C. H. Seager, D. R. Tallant, J. A. Voigt and B. E. Gnage, J. Appl. Phys. 79 (1996) 7983. 38. B. E. A. Saleh, M. C. Teich, ‘Fundamentals of Photonics’ ed. John Wiley & Sons, New York (1991). 39. U. Griebner, R. A. Kaindl, T. Elsaesser and W. Seeber, Appl. Phys. B. 67 (1998) 757. 40. H. Cao, J. Y. Wu, H. C. Ong, J. Y. Dai and R. P. H. Chang, Appl. Phys. Lett. 73 (1998) 572. 41. B. Hecht et al., J. Chem. Phys. 112 (2000) 7761. 42. M. Jacobsohn and U. Banin, J. Phys. Chem. B 104 (2000) 1. 43. Y. Takahashi et al., Jpn. J. Appl. Phys. 33 (1994) 6611. 44. L. Liu et al., J. Electr. Mat. 29 (2000) 69. 45. J. Kong, N. Franklin, C. Wu, S. Pan, K. J. Cho and H. Dai, Science 287 (2000) 622. 46. P. G. Collins, K. Bradley, M. Ishigami and A. Zettl, Science 287 (2000) 1801. 47. Y. Cui, Q. Wei, H. Park and C. M. Lieber, Science 293 (2001) 1289. 48. F. Favier, E. C. Walter, M. P. Zach, T. Benter and R. M. Penner, Science 293 (2001) 2227. 49. M. Law, H. Kind, F. Kim, B. Messer and P. Yang, Angew. Chem. 41 (2002) 2405. 50. C. G. Founstadt and R. H. Rediker, J. Appl. Phys. 42 (1971) 2911. 51. O. V. Safonova, M. N. Rumyantseva, L. I. Ryabova, M. Labeau, G. Delabouglise and A. M. Gaskov, Mat. Sci. Eng. B 85 (2001) 43. 52. V. E. Henrich and P. A. Cox, The Surface Science of Metal Oxides, Cambridge University Press, Cambridge, U.K. (1994). 53. N. Barsan, M. S. Berberich and W. Goepel, Fresenius J. Anal. Chem. 365 (1999) 287. 54. E. Comini, A. Cristalli, G. Faglia and G. Sberveglieri, Sensors Actuators B 65 (2000) 260.

Part II Functional Oxide Nanowires and Nanobelts

Chapter 3 Nanobelts and Nanostructures of Transparent Conducting Oxides Zhong Lin Wang School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0245, USA

Transparent conducting metal oxides are fundamental to the development of smart and functional materials, which are candidates for optics, optoelectronics, and chemical, biochemical and environmental sensors as well as transducers. The oxides have two unique structural features: switchable and/or mixed cation valences, and adjustable oxygen deficiency, which are the bases for creating many novel materials with unique electronic, optical, and chemical properties. The oxides are usually made into nanoparticles or thin films in an effort to enhance their surface sensitivity, and they have recently been successfully synthesized into nanowire-like structures. Utilizing the high surface area of nanowire-like structures, it may be possible to fabricate nanoscale devices with superior performance and sensitivity. In the past several years, the preparation of one-dimensional oxide nanostructures has attracted much attention because of their interesting electrical and optical properties, and their potential applications in constructing nano-scale electronic and optoelectronic devices. To date, one-dimensional oxide nanostructures with different morphologies (such as wire-like [1–3], tube-like [4–6], belt-like [7, 8], etc.) and composition (such as ZnO [1, 7], SiO2 [2, 4], V2O5 [5], Ga2O3 [3], etc.) have been explored using various synthetic approaches including vapor-phase evaporation [1, 7], sol-gel [4, 5], template-based method [6, 9], arc-discharge [3], and laser ablation [10]. This chapter reviews the nanobelt structures of transparent conducting oxide synthesized in our laboratory.

1. Synthesis Method The vapor phase evaporation represents the simplest method for the synthesis of one-dimensional oxide nanostructures. By using this method, various kinds of onedimensional oxide nanostructures, such as nanowires of ZnO [1, 11, 12], In2O3 [13],

48

Wang O-ring

Furnace Thermal couple

View window

Carrier gas Alumina tube Substrate Source material Products

Cooling Carrier water gas

Heating component Pump out

Fig. 1. Schematic experimental setup for the growth of one-dimensional oxide nanostructures via an evaporation-based synthetic method.

SiO2 [14–18], Ga2O3 [19–23], and GeO2 [24], nanobelts of ZnO, SnO2, Ga2O3, In2O3, CdO, and PbO2 [7, 8], and nanorods of MgO [25, 26], were successfully fabricated. The syntheses were usually conducted in a tube furnace as that schematically shown in Fig. 1. In Fig. 1, the desired source oxide materials (usually in the form of powders) were placed at the center of an alumina or quartz tube that was inserted in a horizontal tube furnace, where the temperatures, pressure, and evaporation time were controlled. Before evaporation, the reaction chamber was evacuated to ~1–3 ⫻ 10⫺3 Torr by a mechanical rotary pump. At the reaction temperature, the source materials were heated and evaporated, and the vapor was transported by the carrier gas (such as Ar) to the downstream end of the tube, and finally deposited onto either a growth substrate or the inner wall of the alumina or quartz tube. For the vapor phase evaporation method, the experiments were usually carried out at a high temperature (⬎1000 ⬚C) due to the high melting point and low vapor pressure of the oxide materials. For example, the synthesis of ZnO and Ga2O3 nanobelts needs to heat the relevant ZnO and Ga2O3 powders at 1400–1450 ⬚C [7]. In order to reduce the reaction temperature, a mixed source material, in which a reduction reaction was involved, were employed. For example, Huang et al. [1, 11] obtained ZnO nanowires by heating a 1 : 1 mixture of ZnO and graphite powders at 900–925 ⬚C under a constant flow of Ar for 5–30 minutes. Wu et al. [19] prepared Ga2O3 nanowires in a high yield by heating a 1 : 1.5 mixture of Ga2O3 and graphite powders at 980 ⬚C for 2 hours. Both of them are about 500 ⬚C lower than those used in the synthesis of ZnO and Ga2O3 nanobelts [7]. In addition, the reaction temperature can be further reduced when the low melting point metal that is the cation of the final oxide compound was heated in an oxidized atmosphere. For example, Li et al. [22] obtained Ga2O3 nanowires by oxidizing Ga metal at 780 ⬚C in a quartz tube filled in air. Zhang et al. [20] prepared Ga2O3 nanowires by the reaction of Ga and SiO2 at temperature as low as 350 ⬚C. For the synthesis of SiO2 nanowires, a mixture of SiO2 and Si [16, 18] was usually used as the source materials to make the experiments done at a relative low temperature.

Nanobelts and Nanostructures

49

The growth of one-dimensional oxide nanostructures via vapor phase evaporation may be involved with or without catalyst. The feature of the catalyzed-grown nanowires is that a catalyst nanoparticle is always present at one end of the nanowires. The function of the catalyst during nanowire growth is to form a low melting point eutectic alloy with the nanowire materials, which acts as a preferential site for absorption of gas-phase reactant and, when supersaturated, the nucleation site for crystallization. During growth, the catalyst particle directs the nanowire’s growth direction and defines the diameter of the crystalline nanowires. The growth of the nanowires catalyzed by a catalyst particle follows a mechanism called vapor–liquid–solid (VLS), which was proposed by Wagner and Ellis in 1964 for silicon whisker growth [27]. The commonly used catalysts for oxide nanowire growth via a VLS process include Au [1, 11], Fe [16, 17], and Co [14, 15]. The catalysts were used either in the form of powders that were mixed with the source materials [14–17] or in the form of film that was deposited on a growth substrate [1, 11]. It should be noted that the morphology and orientation of the products were strongly affected by the catalysts. Pan et al. [7] obtained randomly distributed ZnO nanobelts (with rectangular cross-section) by simple thermal evaporation of ZnO powders at 1400–1450 ⬚C. While Huang et al. [1] prepared aligned ZnO nanowires (with round cross-section) on sapphire substrate coated with a 10 to 35 Å thick layer of Au as catalyst.

2. Oxide Nanobelts Recently, Pan et al. [7, 8] reported a new class of distinctly different semiconducting oxide nanostructures that have a rectangular cross-section, in correspondence to a belt-like morphology (so-called nanobelts). The oxides with the nanobelt morphology include ZnO, SnO2, In2O3, CdO, Ga2O3 [7, 8], PbO2 [28], ZnS [29] covering cations with different valence states and materials with different crystallographic structures. Except for the PbO2, the other five oxides have long been regarded as transparent conducting oxide (TCO) materials and have found wide applications in architectural glass, sensors, flat panel display, etc., due to their distinctive optical, electrical and thermal properties [30]. The as-synthesized nanobelts may belong to different crystallographic families, but they have a common shape that is the ribbon geometry. A summary on the oxide systems that we have successfully synthesized nanobelts is given in Table 1, and the corresponding synthesis conditions are summarized in Table 2. The details of these nanobelt structures are given in following sections. 2.1. ZnO oxide nanobelts Thermal evaporation of ZnO powders (purity: 99.99%; melting point: 1975 ⬚C) at 1400 ⬚C resulted in white wool-like products that were composed of ultra-long ZnO nanobelts in a high yield (Fig. 2). The typical lengths of the ZnO nanobelts are in the range of several tens to several hundreds of micrometers; some of them even have lengths on the order of millimeters. Energy-dispersive x-ray spectroscopy (EDS) and x-ray diffraction (XRD) measurements show that the sample is wurtzite (hexagonal)

50

Wang Table 1. Semiconductive oxide nanobelts and their growth directions and surface planes Nanobelts ZnO ZnO Ga2O3 Ga2O3 SnO2 In2O3 CdO PbO2 ZnS

Crystal structure

Growth direction/plane

Top surface

Wurtzite Wurtzite Monoclinic Monoclinic Rutile C-Rare earth NaCl Rutile Wurtzite

[0001] – [0110]* [001] [010] [101] [001] [001] [010] [0001]

⫾(2110) –– ⫾(2110) ⫾(100) ⫾(100) – ⫾(101) ⫾(100) ⫾(100) ⫾(201) –– ⫾(2110)

––

Side surfaces –

⫾(0110) ⫾(0001) ⫾(010) – ⫾(101) ⫾(010) ⫾(010) ⫾(010) – ⫾(101) – ⫾(0110)

* The nanobelts have a single stacking fault/twin parallel to the growth direction throughout the entire length.

Table 2. Synthesis conditions and morphology characteristics of oxides nanostructures

Nano-structure ZnO belt *t-SnO2 belt In2O3 belt CdO belt CdO sheet Ga2O3 belt Ga2O3 sheet PbO2 belt *t-SnO2 wire *o-SnO2 wire SnO diskette ZnS belt/sheet/rod

Source materials

Evaporation temperature ⬚C

Pressure (Torr)

Substrate temperature ⬚C

Lengths (␮m)

Width or diameter (nm)

Width-tothickness ratio

ZnO SnO SnO2 In2O3 CdO CdO

1400 1050 1350 1400 1000 1000

200–300 200–300

800–1100 800–950

⬎500 ⬎500

50–300 30–200

5–10 5–10

200–300 200–300 200–300

800–1100 700–800 700–800

50–300 ⬍100 5–10 ␮m

50–150 100–500 5–10 ␮m

Ga2O3 GaN GaN

1400 950 950

200–300

50–500

20–100

200–300

800–1100 700–850 700–850

5–10 ⬎10 20–60 nm (thickness) 5–10

5–10 ␮m

5–10 ␮m

PbO SnO Sn ⫹ SnO SnO SnO2 ZnS

950 1050 1050 1050 1350 1050

200–300 250–700 200 500–600

600–800 25–40 25–40 200–400

300

~700

⬎200

100–500

20–60 nm (thickness) 50–200 50–300 5–10 ⬎500 30–100 2–5 ⬎500 100–600 2–3 100 nm–10 ␮m 15 3–20

* t-SnO2 represents rutile structured SnO2 and o-SnO2 is orthorhombic structured SnO2.

structured ZnO with lattice constant of a ⫽ 3.249 Å and c ⫽ 5.206 Å, consistent with the standard values for bulk ZnO. TEM images (Fig. 3a–c) reveal that the geometrical shape of the ZnO nanobelts is distinct in cross section from the nanotubes and nanowires. Each nanobelt has a uniform width along its entire length, and the typical widths of the nanobelts are in the range of 50 to 300 nm. No particle was observed at the ends of the nanobelts.

51

Nanobelts and Nanostructures

2µm

200nm

Fig. 2. SEM image of ZnO nanobelts. The inset is a TEM image showing the morphological feature of the nanobelts.

(b)

[0001

]

(a)

200 nm

(c)

200 nm

(d)

50 nm

200 nm

Fig. 3. TEM and HRTEM images of ZnO nanobelts showing their unique geometrical shape. (a–c) TEM images of several straight and twisted ZnO nanobelts, displaying the shape characteristics of the belts. (d) Cross-sectional TEM image of a ZnO nanobelt, showing a rectangle-like cross-section with width-to-thickness ratio of ~9. The cross-sectional TEM specimen was prepared by slicing nanobelts embedded in epoxy using an ultramicrotone.

A ripple-like contrast appeared in the TEM image is due to strain resulted from the bending of the belt. Cross-sectional TEM image (Fig. 3d) shows that the nanobelts have a rectangular-like cross-section, with typical thickness and width-to-thickness ratios in the range of 10 to 30 nm and ~5 to 10, respectively. High-resolution TEM (HRTEM) and electron diffraction studies show that the ZnO nanobelts are structurally uniform and single crystalline but with two different growth directions.

52

Wang ––

–

The nanobelt, growing along [0001] and enclosed by ⫾(2110) and ⫾(0110) facets, shows no defect and no dislocation. The surfaces of the nanobelts are clean, atomically sharp and without any sheathed amorphous phase. 2.2. SnO2 oxide nanobelts Single crystalline SnO2 nanobelts of rutile structure were consistently synthesized by thermal evaporation of either SnO2 powders (purity: 99.9%, melting point: 1630 ⬚C) at 1350 ⬚C or SnO powders (purity: 99.9%, melting point: 1080 ⬚C) at 1000 ⬚C. After evaporation, similar white fuzz-like products were deposited on the alumina plate, whether the source material was SnO2 or SnO. SEM imaging (Fig. 5a) and energy dispersive X-ray spectroscopy (EDS) analysis show that the products are composed of large quantities of ultra-long SnO2 nanobelts (with lengths up to the order of millimeters) and a small fraction of Sn nanoparticles (as indicated by arrowheads in Fig. 4a). XRD patterns from the as-synthesized nanobelt samples prove the rutile type structure with lattice constants of a ⫽ 4.722 Å and c ⫽ 3.184 Å, which are consistent with those of bulk SnO2. TEM images (Fig. 4b–4d) display the characteristic shape of the SnO2 nanobelts (note the rectangular-like cross-section of a broken nanobelt in Fig. 4b and a side view of a nanobelt displayed in Fig. 4d). The ripple-like strain contrast can also be seen in SnO2 nanobelts. Each nanobelt is uniform in width and thickness, and the typical widths of the SnO2 nanobelts are in the range of 50 to 200 nm. Cross-sectional TEM observations show that the cross-sections of the SnO2 nanobelts are rectangular-like, with typical width-to-thickness ratios of ~5 to 10. HRTEM image (Fig. 4e) reveals that the nanobelts are single crystalline and dislocation free. Electron diffraction pattern (inset in Fig. 4e) indicates that the SnO2 nanobelt grows along – [101], and it is enclosed by ⫾(010) and ⫾(101) crystallographic facets. In addition to the normal rutile structured SnO2, it has been possible to form an orthorhombic superlattice-like structure [31]. The orthorhombic structure can form in a thin nanowire, co-exist with the normal rutile structured SnO2 in a sandwiched nanoribbon, or occur in the form of nanotubes. This result is distinct from that for bulk SnO2 where pressures in excess of 150 kbar are required to form the orthorhombic form. 2.3. In2O3 nanobelts Nanobelts of In2O3 with C-rare earth crystal structure were also synthesized by our method (Fig. 5). The evaporation of In2O3 powders (purity: 99.99%, melting point: ~1920 ⬚C) at 1400 ⬚C yields In2O3 nanobelts. TEM observations show that most of the In2O3 nanobelts have uniform width and thickness along their lengths. Typically, the In2O3 nanobelts have widths in the range of 50 to 150 nm and lengths of several tens to several hundreds of micrometers. Electron diffraction analysis shows that the In2O3 nanobelts are single crystalline, and grow along ⬍100⬎, the surfaces being enclosed by {100}. 2.4. CdO nanobelts Nanobelts of CdO with NaCl cubic structure were also synthesized by evaporating CdO powders (purity: 99.998%, melting point: 1430 ⬚C) at 1000 ⬚C. Besides CdO

53

Nanobelts and Nanostructures (a)

(b)

2 nm (e)

20 nm

402 211 020

[101]

(c)

200 nm

200 nm (d) 100 nm

0.48 nm

Fig. 4. Super-long nanobelt structure of SnO2 (with rutile crystal structure). (a) SEM image of the as-synthesized SnO2 nanobelts. (b)–(d) TEM images of SnO2 nanobelts with straight and twisted shapes. An enlargement of a broken nanobelt is inserted in (b) to display the rectangular-like cross-section of the belt. The belt-like shape is further verified by an enlargement of the boxed region in (c) as re-displayed in (d); the width-to-thickness ratio is ~5. (e) High-resolution TEM image of a SnO2 nanobelt showing that the nanobelt is single crystalline, free from dislocation and defects. Inset is the corresponding electron diffraction pattern. The SnO2 nanobelts shown in (a), (b) and (e) were obtained from thermal evaporation of SnO2 powders at 1350 ⬚C; the SnO2 nanobelts shown in (c) and (d) were obtained from thermal evaporation of SnO powders at 1000 ⬚C, but they preserve the same crystal structure and same growth morphology [7]. (Courtesy of Drs. Z. L. Wang, Z. W. Pan and Z. R. Dai, reprinted with permission from Science 2001, 291, 1947–1949. Copyright 2001 American Association for the Advancement of Science.)

200 nm

Fig. 5. Typical TEM images of In2O3 nanobelts.

54

Wang (a)

200 nm (b)

Fig. 6. TEM images of CdO nanobelts.

nanobelts, many single crystalline CdO sheets with sizes on the order of several to several tens of micrometers were also formed (Fig. 6a). These CdO sheets usually have shapes in rectangle, triangle, and parallelogram. The lengths of the CdO nanobelts are usually less than 100 ␮m, and their widths are typically 100 to 500 nm, significantly wider and shorter than those of ZnO, SnO2 and In2O3 nanobelts, respectively. As a result, the width-to-thickness ratios of CdO nanobelts are usually greater than 10. Electron diffraction pattern shows that the nanobelts grow along [100], and their surfaces are enclosed by ⫾(001) and ⫾(010) facets. Some nanobelts were broken into two halves during TEM observation due to electron beam illumination, which is likely to be caused by the easy cracking characteristic of the NaCl-type ionic structure of the nanobelt. Thus, it may be possible to cut these nanobelts with a focused electron or ion beam, so that nanobelts with specific lengths for nano-device applications could be fabricated. The structural equivalence between [100] and [010] for CdO results in the s shape growth of the nanobelt (Fig. 6b). 2.5. Ga2O3 nanobelts Nanoribbons and flat nanosheets of Ga2O3 have been synthesized by evaporating GaN at high temperature with the presence of oxygen (Fig. 7) [32]. The as-synthesized nanoribbons and nanosheets are pure, structurally uniform, single crystalline and free from dislocations. The nanoribbons and the nanosheets all have monoclinic ␤-Ga2O3

55

Nanobelts and Nanostructures (a)

010

110 12 104 0 004

(b)

500 nm

Fig. 7. (a) TEM image of long Ga2O3 nanobelts and the corresponding electron diffraction pattern. (b) TEM image of Ga2O3 nanobelts, displaying their ribbon shape.

Pb

1 µm

Fig. 8. TEM image of an individual PbO2 nanobelts.

structure (C2/m, a ⫽ 12.23, b ⫽ 3.04, c ⫽ 5.80 Å, and ␤ ⫽ 103.7⬚). The flattop and bottom surfaces for both nanoribbons and nanosheets are ⫾(100), the side surfaces – – – are ⫾(010) and ⫾(101) for nanoribbons and ⫾(010), ⫾(101) and ⫾(212) for nanosheets. The axis direction of nanoribbon growth is along either [001] or [010]. 2.6. PbO2 nanobelts ␤-PbO2 nanobelts, with a rectangular cross-section, a typical length of 10–200 ␮m, a width of 50 to 300 nm and a width-to-thickness ratio of 5 to 10, have been successfully synthesized by simple elevated evaporation of commercial PbO powders at high temperature [33] (Fig. 8). The PbO2 nanobelts are enclosed by top surfaces ⫾(201) and – side surfaces ⫾(101) and their growth direction is [010]. Each PbO2 nanobelt is found

56

Wang

Fig. 9. Wurtzite structured ZnS nanobelts synthesized by thermal evaporation of ZnS powder.

to have a large polyhedral Pb tip at one of its ends, suggesting the growth is dominated by a vapor-liquid-solid mechanism. Electron beam irradiation of the PbO2 nanobelts results in the phase transformation from PbO2 to PbO and finally to Pb. 2.7. ZnS nanobelts Zinc Sulfide has two types of crystal structures: hexagonal wurtzite ZnS (referred to hexagonal phase) and cubic zinc blend ZnS (referred to cubic phase). Typically, the stable structure at room temperature is zinc blend, with few observances of stable Wurtzite ZnS. We have synthesized stable Wurtzite structured nanobelts, nanocombs and nanowindmills, using a simple catalyst-free thermal evaporation technique [29]. Our synthesis technique is based on thermal evaporation of ZnS powders at elevated temperature. The as-synthesized sample is composed of several types of structures. The most typical structure observed is ZnS nanobelts (Fig. 9). The nanobelts have a uniform cross-section along their length, with a typical width of 2–30 ␮m, and extend to over 100 ␮m in length. Electron diffraction indicates that all three types of structures are single crystalline with the wurtzite ZnS structure (a ⫽ 3.82 Å, c ⫽ 6.26 Å). The comb-like, saw-toothed belts, and the regular belt structures were found in the same growth temperature range. The nanobelt grows along [0001], with side surfaces – –– (0110) and the top surfaces (2110).

3. Synthesis of New Materials using Nanobelts as Template Using the as-synthesized ZnO nanobelts as template, nanostructured ZnS nanocables and nantubes have been synthesized by chemical reaction [30]. Based on the geometrical shape of the nanobelt template, we anticipate to receive ZnS nanostructures based on reaction ZnO ⫹ H2S → ZnS ⫹ H2O

(1)

Fig. 10 shows an SEM image of the converted ZnS nanobelts. Indeed, besides some particle shape reaction products, nanobelts of ZnS have been formed, as indicated by

Nanobelts and Nanostructures

57

Fig. 10. SEM images of ZnO nanobelts post reaction with H2S.

arrowheads, but the nanobelts have porosity of pore sizes ~30 nm. A rolled nanobelt is presented in the image, which is possibly produced by the surface tension introduced after the reaction; such a shape was rarely observed for the as-synthesized ZnO nanobelts. TEM analysis shows that the converted ZnS nanobelts have two types of structural configurations: rectangular ZnO-ZnS nanocables and ZnS nanotubes. Fig. 11b shows a low magnification TEM image of the ZnO-ZnS nanocables, with a single crystal ZnO as the core and the nanostructured ZnS as the shell, which were converted using the ZnO template as shown in Fig. 11a. The interface between the shell and the core is fairly sharp and there appear no intermediate layer. A clearer picture of the structure is given in Fig. 11c, which displays a composite nanocable with a broken ZnS surface layer. Electron diffraction recorded from the nanocable display –– a spotted pattern that corresponds to the [2110] zone axis of ZnO and a set of ring diffraction pattern, which fits very well to zinc blend structured ZnS. This is consistent with other studies that, for small particles, low temperature synthesis forms zinc blend structure, while high temperature synthesis results in Wurtzite structure. Real space distances measured from the rings also fit the expected interplanar spacings for ZnS. EDS analysis shows that the core is mainly composed of Zn and O, and the shell is Zn and S (Cu and C signals come from the TEM grid) (Fig. 11e and f). Therefore, the core is a single crystalline ZnO and the shell is nanocrystalline ZnS. The other type of structural configuration is ZnS nanotubes (Fig. 12). The tubular shape of the ZnS nanostructured wall is apparent, and there is some porosity in the tube wall. The wall thickness can be as thick as ~100 nm and as thin as 30 nm. Electron diffraction shows the random orientations of the ZnS nanocrystallites. The conversion of ZnO nanobelts into ZnS nanocrystallites structured nanocables and nanotubes occurred in solution. The nanocable is formed by a direct reaction of H2S with the surface layer of ZnO with the presence of water as follows: ZnO ⫹ H2O → Zn2⫹ ⫹ 2(OH)⫺

(2)

H2S → H⫹ ⫹ (HS) → 2H⫹ ⫹ S2⫺

(3)

58

__

(b)

(2110)

(a)

[0001]

Wang

200 nm

250 nm

(c)

(d)

_

_

S

(0001)ZnO

[0001]

1} Zn {31 0} ZnS {22 } ZnS 1 {11

(0110)ZnO

[0110]

100 nm (e)

(f)

Fig. 11. TEM image of ZnO nanobelts (a) prio and (b) post reaction with H2S, showing the formation of ZnO-ZnS nanocable structures. (c) A ZnO-ZnS nanocable with a broken ZnS shell, and (d) a corresponding electron diffraction pattern recorded from the region, showing the presence of a single crystalline ZnO core and the nanostructured ZnS shell. (e, f) EDS spectra acquired from the regions indicated in (c), which prove the local chemical composition (The carbon and Cu lines come from the supporting films and copper grid used for TEM analysis).

A combination of (2) and (3) gives: ZnO ⫹ H2O ⫹ H2S → ZnS ⫹ 2 H2O

(4)

Due to the limited solubility of ZnO in water, the reaction is essentially a substitution reaction, thus, the nanocable still preserve the rectangular cross-section. The pores in the structures are produced due to two factors. The excess H2O produced in the reaction may present in the structure and form the pores. From the structural point of view, ZnS has the zinc blende structure (cubic) with a ⫽ 0.54109 nm, ZnO has the Wurtzite structure (hexagonal) with a ⫽ 0.3249 nm and c ⫽ 0.52065 nm; both of them

Nanobelts and Nanostructures (a)

59

{311}ZnS {220}ZnS {111}ZnS

500nm

(b)

300nm

Fig. 12. Low magnification and higher magnification TEM images of ZnS nanotubes. An electron diffraction pattern recorded from the tubes is inserted, which can be indexed to be zinc blende ZnS structure.

are incompatible in structure. Thus, the substitution reaction is unlikely to produce single crystalline ZnS. The formation of nanocrystallites is expected especially when the reaction temperature was at room temperature. The formation of nanotubes is possible if the Zn2⫹ and (OH)⫺ ions are mobile in the solution, so that they can diffuse through the porous ZnS wall and combine with the H⫹ and S2⫺ ions in the solution, resulting in the formation of an empty tubular structure.

4. Complex Nanobelt Structures 4.1. “Tadpole-like” ZnO nanostructure Using a mixture of ZnO and SnO2 powders in a weight ratio of 1:1 as the source material, a complex ZnO nanostructure was created [34]. Fig. 13a shows a lowmagnification image of the as-synthesized products with a uniform feature consisting of sets of central axial nanowires, surrounded by radial oriented “tadpole-like” nanostructures. The morphology of the string appears like a “liana”, and the axial nanowire is the “rattan”, which has a uniform cross-section with dimension in the range of a few tens of nanometer. The “tadpole-like” branches have spherical balls at the tips (Fig. 13b, c), and the branches display a ribbon shape. EDS analysis shows that the tadpole-like structure and the central nanowire are ZnO, while the ball at the tip is Sn.

60

Wang (a)

(c)

(b)

(d)

Fig. 13. (a) SEM images of the as-synthesized ZnO nanostructures, showing strings of “tadpolelike” nanostructures. (b, c) High magnification SEM images of the ZnO nanostructures showing the ribbon shape of the “tadpole-like” nanostructure. (d) Low-magnification TEM image of the as-synthesized ZnO nanowire-nanoribbon junction arrays.

The ribbon branches have a fairly uniform thickness, and their surfaces are rough with steps. The contact point between the nanoribbon with the axial nanowire is rather small in the order of a few tens of nanometers, while far away from the contacting point, the nanoribbon size is rather large in the order of 100–200 nm. Fig. 13d is a typical TEM image of the nanowire-nanoribbon junction structures. The axial nanowire is as thin as ~30 nm, while the width of the nanoribbon is rather large and increases as the growth continue. The contrast introduced by the surface steps on the ribbons is apparent. Electron diffraction shows that the growth direction – of the nanoribbons is [0110] (and equivalent directions), and the nanowires are [0001]; –– the top surfaces of the nanoribbon are ⫾(2110). In the present study, a mixture of SnO2 and ZnO powders was used as the source material. It is known that SnO2 can decompose into Sn and O2 at high temperature, thus, the growth of the nanowire-nanoribbon junction arrays is the result of vaporliquid-solid growth process, in which the Sn catalyst particles are responsible for initiating and leading the growth of ZnO nanowires and nanoribbons. Our study of ZnO – nanobelts shows that the fast growth directions of ZnO are [0001] and ⬍1010⬎. The growth of the novel structure presented in the present study can be separated into two stages. The first stage is a fast growth of the ZnO axial nanowire along [0001]

61

Nanobelts and Nanostructures [0001]

(a)

(e) –

[1010]

(b)

–

–

[1100]

[0110] [0001]

(c) –

–

[1100]

[0110] –

[1010]

–

(d)

[1010]

–

[1010]

Fig. 14. (a–d) Schematic diagram showing the two-stage growth of the ZnO junctions. (e) A cross-section model illustrating the isotropic epitaxial growth of the nanoribbons around the nanowire.

with Sn as the catalyst (Fig. 14a). The growth rate is so high that a slow increase in the size of the Sn droplet has little influence on the diameter of the nanowire, thus the axial nanowire has a fairly uniform shape along the growth direction. The second stage of the growth is the nucleation and epitaxial growth of the nanoribbons due to the arrival of the tiny Sn droplets onto the ZnO nanowire surface (Fig. 14b). This stage is much slower than the first stage because the lengths of the nanoribbons are uniform and much shorter than that of the nanowire. Since Sn is in liquid state at the growth temperature, it tends to adsorb the newly arriving Sn species and grows into a larger size particle (i.e., coalescing) (Fig. 14c). Therefore, the width of the nanoribbon increases as the size of the Sn particle at the tip becoming larger, resulting in the formation of the tadpole-like structure observed in TEM (Fig. 14d). The ZnO nanowire – – is likely to have a hexagonal cross-section bounded by ⫾(1010), ⫾(0110) and – ⫾(1100), which are six crystallographic equivalent planes. The Sn liquid droplets deposited onto the ZnO nanowire lead to the simultaneous growth of the ZnO nanorib– – – bons along the six growth directions: ⫾[1010], ⫾[0110] and ⫾[1100] (Fig. 14e). The angles between the two adjacent growth directions is 60⬚, resulting in the six-fold symmetric distribution of the nanoribbons around the nanowire. 4.2. Aligned growth of ZnO homojunctions As demonstrated in Section 4.1, Sn is an excellent catalyst for growth of ZnO nanostructures. Based on this work, we have also developed aligned ZnO nanowires on a polycrystalline alumina substrate. The SEM image shown in Fig. 15 clearly displays the reasonable alignment among the nanowires. Higher magnification SEM image shows that the nanowire has a non-uniform cross-section along its length and it becomes sharper toward the tip. The very tip has a Sn rich head.

62

Wang

Sn

Fig. 15. Aligned ZnO nanowires grown on a polycrystalline alumina substrate.

It is known that the conductivity of ZnO depends on the oxygen vacancy concentration. A stoichiometric ZnO is semiconductor, while an oxygen deficient ZnOx is metallic. It is thus possible to create a semiconductor-metal homojunction by controlling the oxygen deficiency during the growth. This experiment has been done for ZnO, as shown – in Fig. 16, where the nanowires grow along ⫾[1010]. If one examines the image carefully, a node at the region indicated by an arrowhead is visible. The crystals on the both sides are different, as proved by the contrast in the TEM image shown in the inset. This is an example of homojunction arrays. 4.3. Gallium droplet induced aligned growth of nanowires Pan et al. [35] developed a new synthetic route, in which large size (5 to 50 ␮m in diameter), low melting point Ga droplets were used as the catalyst, to the large-scale growth of highly aligned, closely packed SiO2 nanowire bunches, and many interesting new growth phenomena were observed. The experimental setup for aligned SiO2 nanowire growth is similar to that shown in Fig. 1. 1–2 grams of GaN powders were placed at the center of an alumina tube that is inserted in a horizontal tube furnace. A long silicon wafer stripe (10 ⫻ 1 cm2) was placed on the middle part of a wide alumina plate (10 ⫻ 3 cm2), which was located 10 cm away from the GaN powders at the downstream

Nanobelts and Nanostructures

63

Fig. 16. ZnO-ZnOx homojunctions grown by controlling oxygen deficiency.

end of the alumina tube. At the reaction temperature of 1150 ⬚C, the GaN powders were decomposed into a dense, hot vapor of Ga and N2. The hot Ga vapor rapidly condensed into small Ga clusters as the Ga species cooled through collision with the buffer gas. The formed Ga clusters were transferred to the downstream end of the alumina tube by the carrier gas, and then deposited onto the surface of the silicon wafer and the naked part of the alumina plate in the regions with temperatures ⬍1070 ⬚C. Ga droplets with diameters in the range of 5 to 50 ␮m were evenly distri-buted on the silicon wafer and alumina plate after 5 hours reaction. The Ga droplets deposited on the silicon wafer etched silicon to form Ga-Si alloy and thus created a dense vapor of Si species around the silicon wafer and alumina plate, which acted as Si source for the growth of SiO2 nanowires. On the silicon wafer, carrot-shaped rods (CSRs) with diameters of 10 to 50 ␮m and lengths of up to ~1 mm were grown in-groups in a high yield (Fig. 17a). For each group, several tens of CSRs radically grow upwards, forming a sisal-like structure. Each CSR terminated at its top end in a large spherical Ga ball with diameters in the range of 5 to 50 ␮m. The Ga balls are still in the liquid state at room temperature. When the CSRs were dissected along direction parallel to their long axes, the following three interesting growth phenomena were observed. First, the CSRs are not solid rods; instead, each CSR is composed of numerous, highly aligned, and closely packed nanowires (Fig. 17b). These nanowires are of uniform diameters (10 to 30 nm) and lengths (10 to 40 ␮m). Second, each Ga ball attaches to hundreds of thousands of SiO2 nanowires that grow out perpendicularly from the surface of the ball’s lower hemisphere. That means, each Ga ball can simultaneously catalyze growth of many SiO2 nanowires, which is quite different from the conventional VLS processes, in which one catalyst particle usually catalyzes growth of just one nanowire. The SiO2 nanowires connect with the molten Ga ball through a thin oxide layer composed of Ga, Si and O. Third, the CSRs have a tubular structure; that is, there is a central hole

64

Wang (a)

(b)

Ga Ball

100 µm

10 µm

(c)

200 nm

(d)

0.3 µm

Fig. 17. SEM images of the aligned SiO2 nanowires grown on the silicon wafer. (a) Low magnification SEM image of the as-grown products, showing carrot-shaped rods growing ingroups on the silicon wafer. (b) High magnification SEM image of a rod opened along its long axis, showing a densely packing of the silica nanowires. (c) TEM image of the silica nanowires. (d) TEM image of an interesting silica nanowire that has curly carlty channels.

along the long axes of the CSRs. It is very interesting that the walls of the tubular structures are composed of a large quantity of highly aligned SiO2 nanowires. TEM investigations (Fig. 17c) show that the SiO2 nanowires have a uniform diameter along their entire length and a very narrow diameter distribution of 10 to 30 nm. Electron diffraction analyses show that the nanowires are amorphous. The above results indicate that the low melting point metal Ga can serve as an effective catalyst for the growth of amorphous SiO2 nanowires via a VLS process. It is apparent that molten Ga has different catalytic behavior for nanowire growth from the commonly used metal catalysts such as Au and transition metals.

65

Nanobelts and Nanostructures (a)

(b)

(010)m 020 –

m

202

–

–

(212)m

(101)m

(010)

m

(101–)

2 µm

0.5 µm

Fig. 18. (a) and (b) Low-magnification TEM images of Ga2O3 nanosheets. The inset in (b) is the corresponding electron diffraction pattern recorded from the large nanosheet.

Some interesting structures, such as the one shown in Fig. 17d, are observed, in which there is a coiling shape hole driven inside the volume of the silica nanowire. The formation of such a structure is unclear.

5. Nanosheets Some sheet-like nanostructures can be identified from the SEM images taken from the samples of CdO and Ga2O3. Shown in Fig. 18a–b and the inset in Fig. 18b are TEM images and corresponding electron diffraction pattern of ␤-Ga2O3 nanosheets that have a monoclinic structure (space group C2/m) of lattice constants: a ⫽ 12.23, b ⫽ 3.04, c ⫽ 5.80 Å, and ␤ ⫽ 103.7⬚. Nanosheets have some straight edges with corners of specific angles, typically of 45⬚ and 90⬚ (Fig. 18a). The inserted electron diffraction pattern in Fig. 18b is close to [101] crystal zone of ␤-Ga2O3. The two perpendicular planes result in the rectangular- and L-shape structures (Fig. 18a). Beside – – the ⫾(010) and the ⫾(101) facets, a third-type of planes with a ~45⬚ angle with the two sides, as indicated by arrowheads in Fig. 18a, is also observed. This type of planes – is identified to be the (212) plane from the electron diffraction pattern. The top and bottom surfaces are also the ⫾(100) crystal planes of ␤-Ga2O3. Based on our SEM observation, the thickness of the nanoribbons and nanosheets are 20–60 nm.

6. Nanodiskettes Fig. 19a is a SEM image showing the diskette-like product that was made using either SnO or SnO2 powders under a higher pressure of 500–600 Torr comparing with the 200 Torr used for the nanobelt synthesis. The product was collected in a low temperature region of 200~400 ⬚C. The typical diameter of the diskettes in Fig. 2 is 8–10 ␮m. The thickness of the diskettes is several tens to several hundreds nanometers, which varies with the dimension of corresponding diameter. The aspect ratio of

66

Wang (a)

10 µm (b)

2 µm

(c)

(d)

2 µm 2 µm

(e)

2 µm

Fig. 19. (a) SEM image of tin oxide diskettes. (b)–(c) SEM images showing morphologies of type I tin oxide diskettes with flat surfaces. (d)–(e) SEM images showing morphologies of type II tin oxide diskettes with the growth features of terraces and spiral steps. The sample was made using SnO powders under 500–600 Torr (argon) at 1050 ⬚C and collected from the low temperature region of 200–400 ⬚C.

diameter-to-thickness is about 15. Electron diffraction and X-ray diffraction analyses [36] have determined that a diskette is a single crystal and has the tetragonal SnO structure (P4/nmm, a ⫽ 3.796 Å and c ⫽ 4.816 Å), of which the a- and b-axis are equivalent but not the c-axis. The normal direction of the diskette is parallel to [001]. The formation of circle contour is likely to lower the surface energy. Two main types of SnO diskettes (type I and type II) have been identified based on morphology. The type I SnO diskettes (Fig. 19b–c) possess a uniform thickness and flat surfaces. A perfect circle contour is adopted by the type I SnO diskettes, of which diameters are larger than 1 ␮m. The side view of a type I SnO diskette (Fig. 19c) reveals that its geometrical shape is actually a solid wheel with a drop center rim. It appears that two wedged-rim diskettes stick together by face-to-face. Fig. 19d–e show SEM images of the type II SnO diskettes, of which top surfaces are not flat. Instead, terraces and spiral steps can be identified, resulting in the formation of a cone-like peak on the top surface of the SnO diskettes. The center of the cone is not necessary to be located at the center of the diskette. The apexes of cone peaks are globules that have

67

Nanobelts and Nanostructures

been identified to be Sn rich. The drop center rim is also consistently observed for the type II SnO diskettes (e.g. as indicated by arrowhead in Fig. 19d). The formation of the SnO diskettes is likely a solidification (liquid–solid) process [36]. During the growth of the SnO diskettes, as the SnO vapor moves with the carrier gas (Ar) into the low temperature region, the SnO vapors may become super-cold liquid SnO droplets at first, provided that the flow of the carrier gas is very slow and the chamber pressure is relatively high, which are true in our case. Then the super-cold SnO droplets condense onto either the alumina substrate or the surfaces of the SnO2 nanobelts carried over by the carrier gas from the high temperature region, resulting in the nucleation and growth by continuously receiving the incoming droplets. Preference of the diskettes with a large (001) surface is possibly governed by crystal structure and surface energy. If the (001) surfaces of the crystallized SnO are constantly keep clean and the coming droplets can constantly wet and cover the entire condensed (001) surface during growth, the type I SnO diskette will form. If some impurity generated during the growth is deposited onto the diskettes, around which the cone peak will form, resulting in the formation of the type II SnO diskette. Another possibility is the liquid Sn droplets guided growth of the SnO diskettes.

7. Planar Defects in Oxide Nanobelts Generally speaking, nanostructures of oxides are single crystalline free from dislocations, but point defects of oxygen vacancies are possible. Stacking faults and (a)

0002 Stacking fault

–

0110

–

[0110] 20 nm

(b)

(c)

0.52 nm

0.52 nm

Surface

Stacking fault

–

Fig. 20. (a) TEM image of a ZnO nanobelt grows along [0110], exhibiting a single stacking fault parallel to the growth axis. (b) High-resolution TEM image from the surface of the nanobelt, showing a clean and atomically flat surface. (c) High-resolution TEM image from the stacking fault region.

68

Wang (a)

(b) L –

0110 ––

0111 –

0001 200 nm

Fig. 21. (a) TEM image of a ZnO nanobelt with a twin plane parallel to the growth axis, and –– (b) the corresponding electron diffraction pattern, beam direction [2110].

twin are occasionally observed in oxides nanobelts [37]. ZnO nanobelts with a single – stacking faults are occasionally found, and they grow along [0110] and enclosed by –– (0001) and (2110) facets. The stacking fault that is parallel to the axis and runs through out the entire length of the nanobelt (Fig. 20a). An HRTEM profile image of the nanobelt show that its surface is clean, atomically sharp and without a sheathed amorphous layer (Fig. 20b) (note the contrast showing in the image is due to the amorphous carbon film used as a substrate for TEM observation). This is natural because the nanobelts were made at a high temperature and all of the surface steps and defects may be eliminated by annealing. An enlargement of the stacking fault image is given in Fig. 20c, and the stacking plane is (0001). Twin structure is rarely seen in ZnO nanobelts. Shown in Fig. 21a is a nanobelt that has a twin structure as proved by the corresponding electron diffraction pattern in –– –– Fig. 21b. The nanobelt is oriented along [2110], the twin plane (0111) is parallel to the – side surface, and its growth direction L is perpendicular to the (0 11 1/3(c/a)2) ⫽ – (0 11 0.855) plane, where a and c are the lattice constants for ZnO. Most of the planar defects observed are in the plane parallel to the growth direction, thus, it is believed that a fast growth in the direction is due to the presence of the planar defect.

8. Growth Mechanism To understand the growth of a uniform belt structure without the presence of heterocatalysis particles, we propose a possible mechanism that may present in nanobelt growth. The source material is assumed to vaporize into molecular species at high temperature, and the molecules are composed of the stoichiometric cation-anion molecules, such as ZnO for instance (Fig. 22c). When condensed onto the substrate at a lower temperature region, the cation-anion molecules will be arranged in such a way that a proper cation-anion coordination is preserved to balance the local charge and structural symmetry (Fig. 22d), forming a small nucleus. The newly coming molecules will continue to deposit on the formed nucleus, while the surfaces that have lower energy start to form, such as the side surfaces. Due to the growth temperature is in the range of 800–1000 ⬚C, the mobility of the atoms/molecules is high enough, and the low energy surface tend to

Nanobelts and Nanostructures (a)

69

(b)

(c)

(g) Substrate

(d)

(e)

(f)

Fig. 22. (a, b) The growth fronts/ends of ZnO nanobelts, showing no visible catalytic particles at the ends. (c–g) A possible growth process about the formation of the nanobelt.

be flat, thus, preventing the accumulation of the newly coming molecules onto the surface, resulting its expansion in surface area as more molecules are sticking onto the rough growth front (Fig. 22e). The rough structure of the tip leads to a rapid accumulation of incoming molecules, resulting in the fast formation of a nanobelt (Fig. 22f), and after some time a nanobelt is formed (Fig. 22g). Fig. 22a–b show tips of ZnO nanobelts, in which the growth fronts are a rounded shape, indicating their atomic-scale roughness with the presence of steps, ledges and kinks. The newly molecules can continue to stick onto the growth front, or the side surfaces, but the smooth side surface and the high molecular mobility at the growth temperature prevents its remaining on the surface, and the molecule will randomly diffuse on the surface and finally find the lower energy sites at the growth front. The molecules are unlikely to stick to the edge of the nanobelts because of the unbalanced coordination and possibly higher energy. The size of the nanobelt cross-section is determined by the growth temperature and supersaturation ratio in kinetics of crystal growth, as mentioned above. The growth process is dominated by kinetics rather than thermodynamics. We still cannot completely rule out the possibility of self-catalyzed growth, in which a small thin layer or even in atom thickness metal layer, such as Zn, presents at the growth front, which could lead the growth via VLS. The layer could be quickly oxidized after the nanobelts are exposed to air after the growth, and our ex-situ analysis may never found this layer. The self-catalyzed growth was, however, indeed observed in the Sn-O system, which was adopted by some product formed at low temperature region.

70

Wang

Oxide nanowires or nanobelts were usually prepared by a evaporation-based synthetic method, in which only oxide vapor and solid oxide products were involved, it is likely that the growth of the 1D oxide nanostructures was governed by a vapor–solid (VS) process. In the VS process, the oxide vapor, formed at a higher temperature region, directly deposits on a substrate at a lower temperature region and grows into wire-like or belt-like nanostructures. Unlike the well-developed VLS process, the detailed VS process, for example, how can atoms or other building blocks be rationally assembled into 1D nanostructures with wire-like or belt-like morphologies, is still not fully understood. Finally, it should be noted that the absence of the nanoparticle at the end of the nanowire or nanobelt does not mean that the growth of the nanowire or nanobelt must be governed by a VS process, especially for the oxide nanostructures. This is because, in some cases, the metal catalyst, which takes effect at reaction condition, may be oxidized into oxide tip with the same composition as the nanowire during the furnace was cooled down to the room temperature.

9. Summary Functional oxides are the fundamentals of smart devices. This article reviews the novel nanostructures of functional oxides, including nanobelts, nanowires, nanosheets, nanodiskettes, etc. having been synthesized in our laboratory. Among the group of ZnO, SnO2, In2O3, Ga2O3, CdO and PbO2, which belong to different crystallographic systems and structures, a generic shape of nanobelt structure has been synthesized. The nanobelts are single crystalline and free from dislocation, and their surfaces are atomically flat. The oxides are semiconductors, which have been used for fabrication of nano-size functional devices, such as field effect transistors and gas sensors. The nanobelts and relevant nanostructures are a unique group that have important applications in nano-size electronic, optical, sensor and optoelectronic devices, which will be reviewed in Chapter 2 in Volume I.

Acknowledgments The materials reviewed in this chapters were partially contributed by my group members: Zhengwei Pan, Zurong Dai, Puxian Gao, Xudong Wang, Chris Ma, Daniel Moore, Jing Li, Will Hughes, and James Gole, to whom I am grateful. Thanks for the support from NSF, NASA and Georgia Tech.

References 1. M. H. Huang, S. Mao, H. Feick, H. Q. Yan, Y. Y. Wu, H. Kind, E. Weber, R. Russo and P. D. Yang, Science 292 (2001) 1897. 2. Z. Q. Liu, S. S. Xie, L. F. Sun, D. S. Tang, W. y. Zhou, C. Y. Wang, W. Liu, Y. B. Li, X. P. Zhou and G. Wang, J. Mater. Res. 16 (2001) 683.

Nanobelts and Nanostructures

71

3. Y. C. Choi, W. S. Kim, Y. S. Park, S. M. Lee, D. J. Bae, Y. H. Lee, G. S. Park, W. B. Choi, N. S. Lee and J. M. Kim, Adv. Mater. 12 (2000) 746. 4. M. Adachi, T. Harada and M. Harada, Langmuir 15 (1999) 7097. 5. F. Krumeich, H. J. Muhr, M. Niederberger, F. Bieri, B. Schnyder and R. Nesper, J. Am. Chem. Soc. 121 (1999) 8324. 6. B. C. Satishkumar, A. Govindaraj, E. M. Vogel, L. Basumallick and C. N. R. Rao, J. Mater. Res. 12 (1997) 604. 7. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. 8. Z. R. Dai, Z. W. Pan and Z. L. Wang, Solid State Commun. 118 (2001) 351. 9. M. J. Zheng, L. D. Zhang, X. Y. Zhang, J. Zhang and G. H. Li, Chem. Phys. Lett. 334 (2001) 298. 10. Y. H. Tang, Y. F. Zheng, N. Wang, I. Bello, C. S. Lee and S. T. Lee, Appl. Phys. Lett. 74 (1999) 3824. 11. M. H. Huang, Y. Y. Wu, H. Feick, N. Tran, E. Weber and P. D. Yang, Adv. Mater. 13 (2001) 113. 12. Y. C. Kong, D. P. Yu, B. Zhang, W. Fang and S. Q. Feng, Appl. Phys. Lett. 78 (2001) 407. 13. C. H. Liang, G. W. Meng, Y. Lei, F. Phillipp and L. D. Zhang, Adv. Mater. 13 (2001) 1330. 14. Y. Q. Zhu, W. K. Hsu, M. Terrones, N. Grobert, H. Terrones, J. P. Hare, H. W. Kroto and D. R. M. Walton, J. Mater. Chem. 8 (1998) 1859. 15. H. Takikawa, M. Yatsuki and T. Sakakibara, Jpn. J. Appl. Phys. 38 (1999) L401. 16. C. H. Liang, L. D. Zhang, G. W. Meng, Y. W. Wang and Z. Q. Chu, J. Non-Cryst. Solids 277 (2000) 63. 17. X. C. Wu, W. H. Song, K. Y. Wang, T. Hu, B. Zhao, Y. P. Sun and J. J. Du, Chem. Phys. Lett. 336 (2001) 53. 18. Z. L. Wang, R. P. Gao, J. L. Gole and J. D. Stout, Adv. Mater. 12 (2000) 1938. 19. X. C. Wu, W. H. Song, W. D. Huang, M. H. Pu, B. Zhao, Y. P. Sun and J. J. Du, Chem. Phys. Lett. 328 (2000) 5. 20. H. Z. Zhang, Y. C. Kong, Y. Z. Wang, X. Du, Z. G. Bai, J. J. Wang, D. P. Yu, Y. Ding, Q. L. Hang and S. Q. Feng, Solid State Commun. 109 (1999) 677. 21. C. H. Liang, G. W. Meng, G. Z. Wang, Y. W. Wang, L. D. Zhang and S. Y. Zhang, Appl. Phys. Lett. 78 (2001) 3202. 22. J. Y. Li, Z. Y. Qiao, X. L. Chen, L. Chen, Y. G. Cao, M. He, H. Li, Z. M. Cao and Z. Zhang, J. Alloys Compounds 306 (2000) 300. 23. C. C. Tang, S. S. Fan, M. L. Chapelle and P. Li, Chem. Phys. Lett. 333 (2001) 12. 24. Z. G. Bai, D. P. Yu, H. Z. Zhang, Y. Ding, Y. P. Wang, X. Z. Gai, Q. L. Hang, G. C. Xiong and S. Q. Feng, Chem. Phys. Lett. 303 (1999) 311. 25. P. D. Yang and C. M. Lieber, Science 273 (1996) 1836. 26. P. D. Yang and C. M. Lieber, J. Mater. Res. 12 (1997) 2981. 27. R. S. Wagner and W. C. Ellis, Appl. Phys. Lett. 4 (1964) 89. 28. Z. W. Pan, Z. R. Dai and Z. L. Wang, Appl. Phys. Lett. 80 (2002) 309. 29. C. Ma, D. Moore, J. Li and Z. L. Wang, Adv. Mater. 15 (2003) 228. 30. D. S. Ginley and C. Bright, Mater. Res. Soc. Bull. 25 (2000) 15. 31. Z. R. Dai, J. L. Gole, J. D. Stout and Z. L. Wang, J. Phys. Chem. B 106 (2001) 1274. 32. Z. R. Dai, Z. W. Pan and Z. L. Wang, J. Phys. Chem. B 106 (2002) 902. 33. Z. W. Pan, Z. R. Dai and Z. L. Wang, Appl. Phys. Lett. 80 (2001) 309. 34. P. X. Gao and Z. L. Wang, J. Phys. Chem. B 106, (2002) 12653. 35. Z. W. Pan, Z. R. Dai, C. Ma and Z. L. Wang, J. Am. Chem. Soc. 124 (2002) 1817. 36. Z. R. Dai, Z. W. Pan and Z. L. Wang, J. Am. Chem. Soc. 124 (2002) 8673. 37. Z. L. Wang, Z. W. Pan and Z. R. Dai, Microsc. Microanal. 8 (2001) 467.

Chapter 4 Nanomechanics and Mechanical Behavior of Nanobelts Scott X. Mao Department of Mechanical Engineering, University of Pittsburgh

1. Nanomechanical Behavior of Semiconducting Zinc Oxide Single Nanobelt Semiconducting oxides, such as ZnO, are now widely used as transparent conducting oxide materials and gas sensors. In particular, ZnO is regarded as a promising candidate material for flat panel displays because of its high electrical conductivity, high optical transparency as well as its low cost and easy etchability. The current studies of semiconducting oxides have been focused on two-dimensional films and zerodimensional nanoparticles, while investigations of one-dimensional semiconducting oxide nanostructures are very few. Recently, Z. L. Wang successfully synthesized the beltlike zinc oxides (so called nanobelts) by evaporating the ZnO powders at high temperatures without the presence of catalyst. HRTEM and electron diffraction show that the ZnO nanobelts are structurally uniform, single crystalline and dislocation free. Beltlike zinc oxides (so called nanobelts) were successfully synthesized by evaporating the ZnO powders at high temperatures without the presence of catalyst [1]. Morphology analysis shows the nanobelts have a rectanglelike cross section with typical widths of several hundred nanometers, width-to-thickness ratios of 5 to 10, and lengths of hundreds of micron meters. Mao’s group has made nanoindentations on the ZnO single nanobelt by using AFM indenter probe [2]. It was shown that the indentation size effect was still obvious for the indentation depth under 50 nm. Besides, the sharper the indentation tip, the higher the nanoindentation hardness. It is also demonstrated that nanomaching is possible on nanobelt using AFM tip. Fig. 1 is a 3 dimensional image of ZnO nanobelt before indentation under AFM [3]. Section analysis shows the nanobelt has a rectangular section. These values are typical of a single nanobelt. More investigations reveal that nanobelt have a rectanglelike cross section with typical widths of several hundred nanometers, width-to-thickness ratios of 5 to 10, and lengths of hundreds of micron meters. This result is in agreement with

74

Mao

Fig. 1. Three dimensional image of ZnO nanobelt by AFM indenter tip.

Fig. 2. Three dimensional image of nanoindentations [2].

the TEM study by Z. L. Wang. Fig. 2 is a local magnification of nanoindentations on nanobelt. Using the definition of hardness H ⫽ P/A, where P is load and A is the indented area, Mao’s group obtained the hardness for the indent is about 5–8 GPa under very shallow penetration (3–4 nanometer) of ZnO single nanobelt (0001) surface [2]. Hardness and reduced modulus can be derived from the unload portion of the

Nanomechanics and Mechanical Behavior of Nanobelts ZnO belt

10 hardness (GPa)

75

8 6

ZnO belt

4 2 0

0

20 40 displacement (nm)

60

Fig. 3. Indentation size effect of ZnO nanobelt.

graph using Oliver-Pharr method. The size (depth) effect on the hardness is shown in Fig. 3. The shallower the nanoindentation depth, the higher hardness the belt has. This is true in the indentation depth below 50 nm.

2. Bending and Cutting on Nanorodes, Nanotube and Nanobelt Using AFM Tip Atomic force microscopy was first used by Charles Lieber’s group [4] to determine the mechanical properties of individual, structurally isolated silicon carbide nanorodes and multiwall carbon nanotubes (MWNTs) that were pinned at one end to molybdenum disulfide surface, as shown in Fig. 4. The bending force was measured versus displacement along the unpinned lengths. The MWNTs were about two times as stiff as the SiC NRs. Continued bending of SiC NRs ultimately led to fracture, whereas the MWNTs exhibited an interesting elastic buckling process. The strength of the SiC NRs were substantially greater than those found previously for larger SiC structures, and they approach theoretical values. Because of buckling, the ultimate strengths of the stiffer MWNTs were less than those of the SiC NRs, although the MWNTs represent a uniquely tough, energy-absorbing material. ZnO nanobelt can be bent using AFM Si tip by the following way: first engage on the Si substrate just beside the nanobelt. Then withdraw the tip only once. After withdrawing, move the tip laterally to bend the nanobelt, which is shown in Fig. 5. Mao’s group can even cut the nanobelt using the AFM Si tip, as it is shown in Fig. 6. This demonstrates it is possible to nanomachine ZnO nanobelt by AFM tip.

3. Bending and Fundamental Resonance Frequency of Nanotube Under TEM Transmission electron microscopy (TEM), has been traditionally applied to the characterization of the intrinsic structures of nanomaterials. The measurement of the Young’s modulus of carbon nanotubes was first carried out by Wang’s group using TEM by quantifying the thermal vibration amplitude of the nanotube [5]. Within the framework of in-situ TEM they have recently developed a novel approach that relies

76

Mao

Fig. 4. Overview of the approach used to probe mechanical properties of NRs and nanotubes published in [4]. (A) SiC NRs or carbon nanotubes were pinned by deposition of a grid of square SiO pads. (B) Optical micrograph of a sample showing the SiO pads and the MoS2 substrate (blue). (C) An AFM image SiC NR protruding from a SiO pad. The scale bar is 500 nm. (D) Schematic of beam bending with an AFM tip. (E) Schematic of a pinned beam with a free end.

ZnO nanobelt

Fig. 5. CCD image of ZnO nanobelt bent by AFM Si tip [3].

on electric field induced mechanical resonance for measuring the properties of individual wire-like structures, such as Young’s modulus [6], electron field emission [7], tip work function [8], and electrical quantum conductance [9]. This is a new technique that provides the properties of a single nanowire with well characterized. To correlate the measured mechanical property with the intrinsic microstructure of a nanowire, Dr. Wang’s measurements were made in-situ in a transmission electron microscope (TEM), through which the morphology, defect structure and real dimensions of the object are unambiguously determined [10]. The static and dynamic properties of the nanowires can be obtained by applying a controllable static and alternating electric field. The nanowires were glued using silver past onto a gold wire,

Nanomechanics and Mechanical Behavior of Nanobelts

77

through which the electric contact was made. The counter electrode is a solid Au ball. An electric field can be applied across the two electrodes for inducing mechanical resonance. The nanowire to be used for property measurements is directly imaged under TEM (Fig. 7).

Fig. 6. Three dimensional AFM image of bent nanobelt after cutting [3]. 1= 1.21 MHz Vd = 2 V

2= 5.06 MHz Vd = 4 V

Fig. 7. Electric field induced mechanical resonance of carbon nanotubes. The resonance frequency is used to calculate the bending modulus of the nanotube [6].

78

Mao

For a beam with one end hinged and the other free, the resonance frequency is given by: f0 ⫽ (␤2/2␲) (EI/m)1/2/L2,

(1)

where f0 is the fundamental resonance frequency, ␤ ⫽ 1.875, EI is the flexural rigidity (or bending stiffness), E is the Young’s modulus, I is the moment of inertia about a particular axis of the rod, L is the length of the beam, and m is its mass per unit length. If the geometrical shape of the nanowire is known from cross-sectional TEM images, its moment of inertia can be calculated using the standard definition in mechanics, thus, Eq. (1) can be applied to the cases of carbon nanotubes, coaxial and biaxial nanowires, as long the true fundamental frequency is identified [10]. Dr. Wang’s experiment was carried out for carbon nanotubes produced by an arcdischarge technique, which are believed to be free-from defects. The carbon nanotubes have diameters 5–50 nm and lengths of 1–20 ␮m and most of them are nearly defect-free. After a systematic studies of the multi-walled carbon nanotubes, the bending modulii of the nanotubes were measured as a function of their diameters [10]. Testing the creeping property of nanomaterials is a challenge. The technique we developed may be applied for such a purpose although we still cannot perform the experiments at higher temperatures. Dr. Wang’s group also measured Young’s modulus of nanowires such as coaxial silicon nanowires sheathed by a thin layer of SiO2 and the biaxial biaxial SiC-SiOx nanowires that consist of two side-by-side sub-nanowires of silica and ␤-SiC. The tensile strengths of individual carbon nanotubes have been measured by Yu et al. [11] with a nanostressing stage located within a scanning electron microscope. A carbon nanotubes was glued at both ends to two AFM tips facing each other, and the tensile-loading experiment was prepared and observed entirely within the microscope and was recorded on video. The nanotube broke in the outermost layer and the tensile strength of this layer ranged from 11 to 63 GPa. Analysis of the stress-strain curves for individual nanotube indicated that the Young’s modulus of the outermost layer varied from 270 to 950 GPa.

4. Electromechanical Behavior of Carbon Nanotube Dr. Dai has discovered that prodding carbon nanotubes with a pointy tip can alter their ability to carry an electric current. The results, the first to demonstrate how mechanical deformations can affect a molecular wire’s electrical properties [12] as shown in Fig. 8. The discovery could be used for making tiny electromechanical devices, such as transducers that convert mechanical movements into electrical signals. Other applications include creating high-frequency telephone lines to carry voice and data and making on/off switches for nanoscale computer chips. To prod the nanotube, Dr. Dai’s group used the sharp tip of a atomic force microscope (AFM). A real-space microscopy technique, the AFM makes images of surface topography by dragging a pointy tip over a structure’s bumps and folds. The tip reads the shape like a blind person reads Braille. The textures are then translated

Nanomechanics and Mechanical Behavior of Nanobelts

79

(a)

(b)

(c)

Fig. 8. An SWNT partly suspended over a trench for electromechanical measurements published in [12]. (a) Device viewed from above. Preparation of samples involves chemical vapor deposition of SWNTs at desired surface sites using SiO2/Si substrates with patterned catalyst islands. (b) AFM image of an SWNT with suspended length l 艐605 nm. The cantilever employed for this experiment has a spring constant kc ⫽ 0.6 N m⫺1. (c) Side-view of the AFM pushing experiment. The tip is centred above the SWNT suspension by slowly zooming into the tube-suspension during real-space imaging.

into a visual image. To conduct experiments on a single nanotube, Dai’s group used AFM technology and placed an array of finely powdered metal nanoparticles on a silicon dioxide substrate, and then fed a carbon-containing gas (methane) over the substrate heated to a high temperature. The carbon infused into the metal particles, which acted as catalysts that converted carbon atoms into honeycomb-lattice nanotubes. The researchers used the technique to grow a single nanotube across a silicon dioxide trench. They then attached an electrode to each end of the tube. They used the AFM tip to push the wire down into the trench, while measuring the wire’s electrical conductance [5]. The group was initially surprised to observe that the flow of electricity dropped sharply as the nanotube bent. When the AFM tip was removed, the tube straightened and the flow of electricity returned to normal. Previous theoretical studies predicted no significant change in the conductance of nanotubes due to

80

Mao

mechanical deformation. Dai hypothesized that a dent that formed near the AFM tip could be responsible for strongly affecting the electrical flow. To make sense of the results, Drs Dai and Wu used computer simulations to show that the AFM tip dented one wall toward the other, as when a garden hose gets kinked and the flow of water is stopped. As one side of the tube is pushed closer to the other, carbon atoms form bonds across the inside of the tube. Normally, each carbon atom binds to three other carbons, leaving one electron free for use in conducting electricity. But when the walls of the tube come close together, each carbon binds to four rather than three carbons. The resulting decrease in the number of free electrons causes the electrical conductance to slow. Dr. Dai concluded that the AFM tip squashes the tube, causing each atom to bond with more atoms. And this causes the tube to turn from an electrical conductor into an insulating structure similar to that found in diamonds Remarkably, the dent disappears once the perturbing tip is removed. This high mechanical reversibility allows the full recovery of the nanotube’s electrical property.

5. Revolutionary Nanowires with a Twist Metallic single-walled carbon nanotubes (SWNTs) do not behave like normal metallic wires. In normal metallic wires the number of open conduction channels (states at the Fermi level) will increase as the cross-section of the wire increases. In contrast, the number of channels available for conduction in metallic SWNTs is independent of the tube’s diameter [13]. White and Mintmire have found how this key property of metallic SWNTs allows them to avoid the spontaneous symmetry breaking that converts other potentially excellent metallic nanowires into semiconductors [13]. They also found how this property is central to the ability of metallic and semimetallic SWNTs to sustain ballistic transport over unprecedented distances for such laterally confined systems in the presence of residual chemical and structural disorder [14]. Thus, it is the fixed number of states at the Fermi level, which when combined with their mechanical and chemical stability and the strong C–C interactions along the tube, that makes metallic and semimetallic SWNTs strong candidates for the ultimate one-dimensional conductors for use in nanoscale devices. White and Mintmire also indicated that semimetallic nanotubes will find use as electromechanical gauges [15].

References 1. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science, March 9th (2001) 1947. 2. M. Zhao, S. Mao, F. Xu, J. A Barnard and Z. L. Wang, “Nanomechanical characterization on zinc and tin oxides nanobelts”, MRS Symposium I: Nanomaterials for Structural Applications, (2002 Fall). 3. S. Mao, M. Zhao and Z. L. Wang, MRS Symposium F: Semiconductor Materials and Devices, (2002 Fall). 4. E. W. Wong, P. E. Sheehan and C. M. Lieber, Science 277 (1997) 1971. 5. M. M. Treacy, T. W. Ebbesen and J. M. Gibson, Nature 38 (1996) 678. 6. P. Poncharal, Z. L. Wang, D. Ugarte and W. A. de Heer, Science 283 (1999) 1516. 7. Z. L. Wang, P. Poncharal and W. A. de Heer, Microsc. Microanal. 6 (2000) 224.

Nanomechanics and Mechanical Behavior of Nanobelts 8. 9. 10. 11. 12.

13. 14. 15.

81

R. P. Gao, Z. L. Wang and Z. W. Pan, Appl. Phys. Letts. 78 (2001) 1757. S. Frank, P. Poncharal, Z. L. Wang and W. A. de Heer, Science 280 (1998) 1744. Z. L. Wang, P. Poncharal and W. A. De Heer, Pure Appl. Chem. 72 (2000) 209. M. F. Yu, O. Lourie, M. J. Dyer, K. Moloni, T. F. Kelly and R. S. Ruoff, Science 287 (2000) 637. T. Tombler, C. Zhou, L. Alexeyev, J. Kong, H. Dai, W. Liu, C. Jayanthi, M. Tang and S. Y. Wu, “Reversible Nanotube Electro-mechanical Characteristics Under Local Probe Manipulation”, Nature 405 (2000) 769. J. W. Mintmire, B. I. Dunlap and C. T. White, Phys. Rev. Lett. 68 (1992) 631. C. T. White and T. N. Todorov, Nature 393 (1998) 240; Nature 411 (2001) 649. C. T. White and J. W. Mintmire, 9th Forestight Conference on Molecular Nanotechnology (2002).

Chapter 5 Ferroelectric Nanowires Jonathan E. Spanier, Jeffrey J. Urban, Lian Ouyang, Wan Soo Yun and Hongkun Park Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA

1. Introduction Ferroelectric materials are compounds that exhibit a spontaneous electric polarization that can be reoriented by an external electric field [1, 2]. Over the last decade, these materials, especially the ferroelectric oxides, have received considerable attention because of the fundamental interest in their properties and their potential for a variety of technical applications. The ferroelectric oxides typically exhibit a host of other related properties such as piezoelectricity, pyroelectricity, and large nonlinear optical coefficients [1–3]. Many technical applications have been envisioned based on these unique properties of ferroelectric oxides including nonvolatile ferroelectric memory, micro-actuators, and optical switches and waveguides. Central to all these diverse properties of ferroelectric oxides is the structural phase transition of the underlying oxide lattice [1, 4]. This point is best illustrated by considering the cubic-to-tetragonal phase transition of barium titanate BaTiO3 that is representative of “displacive” ferroelectric oxides. Above ~120 ⬚C, bulk BaTiO3 exhibits the cubic perovskite structure shown in Fig. 1. As the temperature is lowered below ~120 ⬚C, this compound undergoes a phase transition to become a tetragonal structure by displacing the Ti2⫹ ion and the O2⫺ ions in opposite directions (hence the name, displacive ferroelectrics). When the opposite motions of positively and negatively charged ions coherently add up, the crystal as a whole (within the same domain) develops a spontaneous dipole moment or polarization and becomes ferroelectric. The same distortion of the unit cell, again added together coherently throughout the crystal, also results in the deformation of the whole crystal, leading to piezoelectricity. In addition, owing to the loss of the inversion symmetry, the crystal in the tetragonal phase exhibits second-order optical susceptibility responsible for the second harmonic generation. Considering the importance of the structural transition in understanding the properties of ferroelectric oxides, it is not surprising that much theoretical and

84

Spanier et al. (a)

(b)

Fig. 1. Unit cells of BaTiO3. (a) High temperature cubic phase and (b) low temperature tetragonal phase. The atoms at the corner, face, and center are Ba, O, and Ti, respectively.

experimental work has been devoted to the subject. Traditional experimental studies on phase transitions were performed on single-crystalline samples, whereas more recent studies have been centered on thin-film and nanocrystalline samples [2, 5–9]. Because of the reduced dimensionality and the large surface-to-volume ratio, thin-film ferroelectric oxides have revealed many novel phenomena that have not been observed in the bulk crystalline samples, such as the thickness-induced depression of the phase transition temperature [10–13] and the emergence of ferroelectricity from a bulk antiferroelectric material [14]. These observations then lead to the question: what happens when the dimensionality is further reduced? Here, we discuss our recent research effort aimed to address these questions [15, 16]: the synthesis and characterization of single-crystalline nanowires composed of barium titanate (BaTiO3) and strontium titanate (SrTiO3). The synthesis is based on solution-phase decomposition of bimetallic alkoxide precursors and yields wellisolated BaTiO3 nanowires with diameters ranging from 5 to 60 nm and lengths reaching up to ⬎10 ␮m [15]. Electron microscopy and diffraction measurements show that these nanowires are single-crystalline with the [001] direction preferentially aligned along the wire axis. Scanned probe microscopy investigations of individual BaTiO3 nanowires show that non-volatile electric polarization can be reproducibly induced and manipulated on these nanowires by an external electric field, demonstrating that they remain ferroelectric despite their small radial dimension [16]. Non-volatile polarization domains as small as ~100 nm2 in size can be induced on these nanowires with a retention time exceeding 5 days, suggesting that ferroelectric nanowires may be used to fabricate non-volatile memory devices with an integration density approaching 1 terabit/cm2.

2. Synthesis of BaTiO3 and SrTiO3 Nanowires As is the case with any other materials-related research, the quality of the proposed research depends critically on the quality of nanostructured materials themselves. This point is best illustrated by the “revolution” brought about in semiconductor nanocrystal research by the advent of the colloidal synthetic technique that allows preparation of nanocrystals with high crystallinity and narrow size distributions [17, 18]. Traditional synthesis of nanostructured ferroelectric oxides and thin films are performed by sol-gel [19–21], vapor deposition [22, 23], and coprecipitation methods [11]. Unfortunately, however, these methods typically yield highly agglomerated nanocrystalline solids of

Ferroelectric Nanowires

85

poor crystalline quality, making it difficult to investigate in detail the size-dependent properties of nanostructured ferroelectric oxides. This problem is partly caused by the high temperature heat treatments common to these synthetic methods that lead to rapid grain growth and annealing of nanometer-sized particles. Another common problem associated with traditional synthesis has been the formation of products with incorrect stoichiometry because simultaneous hydrolysis of molecular precursors is challenging in practice. Recently, O’Brien and coworkers and our group have developed a synthetic scheme that produces well-isolated perovskite nanocrystals [24] and nanowires [15] of homogeneous composition that produces the correct product stoichiometry and avoids high-temperature treatment steps. Specifically, our syntheses of BaTiO3 and SrTiO3 nanowires are accomplished by solution-phase decomposition of bimetallic oxide precursors in the presence of coordinating ligands, similar to the methods used to prepare metallic and semiconducting nanocrystals and nanowires [18, 25, 26]. The chosen precursor is a barium (strontium) titanium isopropoxide complex, Ba(Sr)Ti[OCH(CH3)2]6. This complex, first reported by Suyama et al. [19] is prepared by dissolving Ba (Sr) metal in a solution of titanium isopropoxide, benzene, and isopropanol. After the Ba (Sr) metal is completely dissolved, the reaction is left undisturbed for a few days to promote crystal growth, and the crystals are then separated from the solution. For the nanowire synthesis, an excess of 30% H2O2 is added at 100 ⬚C to a heptadecane solution containing a 10 : 1 molar ratio of bimetallic alkoxide precursors to oleic acid. The reaction mixture is subsequently heated to 280 ⬚C for 6 h, resulting in a white precipitate composed of nanowire aggregates. Anisotropic nanowire growth is most likely due to precursor decomposition and crystallization in a structured inverse micelle medium formed by precursors and oleic acid under these reaction conditions. Well-isolated nanowires are obtained by sonication of the reaction product followed by fractionation between water and hexane, and they are collected from the hexanes layer for further analysis. A representative transmission electron microscope (TEM) image of the reaction product is shown in Fig. 2(a), and it clearly shows that the reaction produces mostly well-isolated nanowires and some nanoparticle aggregates. Fig. 2(b) shows a scanning electron microscope (SEM) image of an isolated nanowire. Similar results are obtained for the synthesis of SrTiO3 nanowires. Extensive analysis of TEM images shows that nanowire diameters range from 5 to 60 nm and that their lengths range from a few hundred nanometers to tens of microns. The average length of nanowires is found to increase with the reaction time. Elemental analysis indicates that these nanowires have stoichiometric proportions of Ba (or Sr) and Ti. Analysis of the powder X-ray (Fig. 2(b) inset) and electron diffraction patterns (Figs. 3 and 4) demonstrates that these nanowires are composed of crystalline BaTiO3 and SrTiO3 with a cubic perovskite structure. The unit cell parameters for BaTiO3 and SrTiO3 are determined to be 4.03 Å and 3.90 Å, respectively, which agrees well with those of the bulk cubic materials. Convergent beam electron diffraction (CBED) patterns obtained from representative BaTiO3 and SrTiO3 nanowires are presented in Figs. 3 and 4, respectively. These CBED patterns exhibit sharp diffraction spots characteristic of crystalline BaTiO3 and SrTiO3, in agreement with X-ray diffraction results obtained from randomly oriented

86

Spanier et al. (a)

500 nm

(b) 500 nm

Intensity (Arb. Unit)

(110)

(100) 20

(111) (200) (211) (210) 40 2θ (degree)

(220) 60

Fig. 2. (a) Transmission electron microscopy (TEM) image of BaTiO3 nanowires, showing that the reaction produces mainly nanowires and small quantities (~10%) of nanoparticle aggregates. A TEM image of SrTiO3 nanowires is essentially indistinguishable from that of BaTiO3 nanowires. (b) Scanning electron microscopy (SEM) image of an isolated BaTiO3 nanowire. Inset: powder X-ray diffractogram of the reaction product composed of randomly oriented BaTiO3 nanowires.

Fig. 3. Left: A TEM image of a 30-nm diameter BaTiO3 nanowire. Inset: convergent beam electron diffraction (CBED) pattern obtained from the same nanowire. Right: a high-resolution TEM (HRTEM) image of the nanowire that shows lattice fringes perpendicular to the [001] direction.

Ferroelectric Nanowires

87

Fig. 4. Left: A TEM image of a 55-nm diameter SrTiO3 nanowire. Inset: CBED pattern obtained from the same nanowire. Right: a HRTEM image of the nanowire that shows lattice fringes perpendicular to the [001] direction.

nanowire ensembles in Fig. 2. Moreover, the CBED patterns taken from different positions along the nanowire are found to be identical within experimental accuracy, indicating that the entire nanowire is a single crystal. Representative high-resolution TEM (HRTEM) images of respective nanowires in Figs. 3 and 4 also confirm the singlecrystalline nature of these nanowires. Both the electron-diffraction patterns and the HRTEM image in Figs. 3 and 4 show that the principal axes of the BaTiO3 and SrTiO3 unit cells are aligned along the wire axis. Essentially all BaTiO3 and SrTiO3 nanowires prepared from over 30 independent synthesis runs possessed identical crystalline structure. At room temperature, bulk BaTiO3 exhibits tetragonal symmetry with the unit cell elongated along the c axis (c/a ⫽ 4.038 Å/3.994 Å ⫽ 1.011), while it becomes cubic above ~120 ⬚C [1]. The ferroelectric polarization lies along the tetragonal c axis in bulk BaTiO3. Unfortunately, the data in Fig. 3 alone do not establish whether BaTiO3 nanowires are tetragonal or cubic since the small tetragonal distortion cannot be discerned in the CBED data due to the size-induced diffraction broadening inherent to nanoscale materials. As will be discussed below, however, scanned probe investigations reveal that BaTiO3 nanowires exhibit ferroelectricity and that the initial polarization direction in pristine nanowires is along the wire axis. This observation suggests that BaTiO3 nanowires have tetragonal symmetry with the [001] direction (the c axis) along the wire axis. The inferred crystalline orientation agrees with the expectation that the electric polarization pointing along the wire axis is energetically favorable due to shape anisotropy [1].

3. Scanned Probe Measurements of BaTiO3 Nanowires We have investigated the nanoscale ferroelectric properties of BaTiO3 nanowires using a ultrahigh vacuum scanned probe microscope (SPM) with a conductive tip, a strategy that has been successfully applied to study ferroelectric properties of thin film samples [5, 27–33]: the topography of a sample can be examined with nanoscale

88

Spanier et al.

resolution using an SPM operating in an atomic force microscopy (AFM) mode and the ferroelectric polarization can also be evaluated and directly manipulated using electrostatic force microscopy (EFM). Compared with other characterization methods that have been used to study ferroelectric nanostructures, such as Raman scattering and X-ray diffraction, SPM has a distinct advantage because it can readily locate individual nanostructures, measure their sizes, and then directly probe and manipulate the ferroelectric polarization. A schematic diagram illustrating the experimental geometry is presented in Fig. 5(a). To induce or “write” local electric polarization on a nanowire, a voltage (Vtip) is applied to the conductive tip while holding the tip at a fixed distance above the nanowire on a gold substrate. The written polarization is then probed or “read” using EFM by measuring the shift in the resonance frequency of a SPM cantilever while scanning it with small Vtip [5, 27, 33]. The attractive and repulsive electrostatic interactions between the tip and the nanowire change the resonance frequency of the

(1) (a)

(2)

(3)

Vtip

Conductive Substrate (b)

100 nm

(c) 1 sec Δν/ν ×106

5 sec

20

10 13 nm 25 nm

30 sec 0 0

40

80

120

Fig. 5. (a) Schematic diagram illustrating the experimental geometry for reading and writing local areas of polarization on a nanowire. The panels in (a) show, left to right, a single topographic atomic force microscope (AFM) image of an 18-nm diameter BaTiO3 nanowire, and three representative electrostatic force microscopy (EFM) images from the nanowire. The first EFM image was collected from the nanowire as deposited. The second and third EFM image responses were acquired following the writing procedure for 3 min with Vtip ⫽ ⫺10 V and Vtip ⫽ ⫹10 V, respectively. Boxes (3), (1) and (2) in EFM images correspond to the resonance frequency shifts of ⫺10, 0 and ⫹10 Hz, respectively. The EFM readings were performed at Vtip ⫽ ⫺2 V. (b) EFM images taken from three different locations along a 35 nm diameter BaTiO3 nanowire following writing for different lengths of time, as denoted in the figure, with Vtip ⫽ ⫹10 V. (c) Retention characteristics of two different BaTiO3 nanowires of diameters 13 nm (circles) and 25 nm (squares). The writing was performed with Vtip ⫽ ⫹10 V for 3 min, and the reading was performed at Vtip ⫽ ⫺2 V at each time point. In (a)–(c), the distance between the tip and the top surface of the nanowire was 10 nm during the writing procedure and 35 nm during the reading procedure.

Ferroelectric Nanowires

89

cantilever in the opposite directions. The plot of the resonance frequency as a function of tip position therefore provides a spatial map of electric polarization directions on a nanowire. This force measurement is tantamount to detecting surface charge densities on the order of 10⫺2–10⫺4 C/m2; and hence it demands high level of force sensitivity. The ultrahigh vacuum (UHV) SPM instrument employed in our experiments is ideally suited for the measurement because it exhibits enhanced force gradient sensitivity due to the high Q factor of the tip. Moreover, the possible influence of adsorbates from ambient on measurements done in air is significantly suppressed in UHV. The panels in Fig. 5(a) show representative EFM images obtained from an 18-nm diameter nanowire together with a topographic image. These EFM images are obtained by subtracting an image at Vtip ⫽ ⫹2 V from that at Vtip ⫽ ⫺2 V to eliminate the capacitive contribution between the tip and the nanowire, and hence they exhibit only the contribution from the surface charges associated with a local electric polarization. The leftmost EFM image obtained from the as-deposited nanowire shows no discernible EFM contrast, indicating that pristine nanowires do not exhibit polarization perpendicular to the wire axis. The EFM images obtained after the writing procedure with Vtip ⫽ ⫺10 V and Vtip ⫽ ⫹10 V clearly show, on the other hand, that a local electric polarization perpendicular to the wire axis can be induced by an external electric field. Comparison of the two EFM images in Fig. 5(a) further shows that the shift in the cantilever resonance frequency changes sign after changing the writing procedure from Vtip ⫽ ⫹10 V to Vtip ⫽ ⫺10 V, thus indicating a reversal of the polarization direction. The EFM images obtained for a range of writing durations indicate that the original induction and reversal of polarization can be performed on the time scale of seconds. For example, Fig. 5(b) shows the EFM response from three different locations along a 35 nm diameter nanowire following writing with Vtip ⫽ ⫹10 V for 1, 5 and 30 s, respectively, demonstrating that polarization domains can be written on the time scale of seconds. Moreover, investigations on many nanowires with different diameters show that a local polarization can be induced and reversed repeatedly on a nanowire as small as 10 nm in diameter. Finally, the measurement of the retention time presented in Fig. 5(c) demonstrates that the electric polarizations induced on nanowires do not decay significantly over a period exceeding 5 days in UHV, illustrating their nonvolatile nature. The ferroelectric nature of a BaTiO3 nanowire can be further probed using the hysteresis characteristics associated with local polarization reversal shown in Fig. 6. Specifically, Fig. 6 shows that once the polarization perpendicular to the wire is created by an external field, its reversal exhibits a clear memory effect with a remanenceto-saturation ratio close to 1. Fig. 6 further shows that a coercive electric field (Ec), at which the polarization changes its direction, is ~7 kV/cm assuming that the relative dielectric constant of a BaTiO3 nanowire is ~1200 as in bulk crystals. Experiments on different nanowires show that the value of Ec remains roughly the same irrespective of nanowire diameter in the range of 10 to 50 nm. This value of Ec compares favorably to Ec ~ 10 kV/cm determined from a bulk BaTiO3 single crystal [34], but it is smaller than typical Ec ~ 30 kV/cm in polycrystalline samples [20]. Fig. 7(b)–(f) show successive EFM images of four distinct polarization domains written on a 12-nm diameter nanowire, and they demonstrate that multiple nanoscale

90

Δν/ν × 106

Spanier et al.

20

0

–20 -10

0 Vtip (V)

10

Fig. 6. Hysteresis measurement showing the fractional shift in the cantilever resonance frequency (⌬␯/␯ ⫽ (␯0 ⫺ ␯)/␯0) as a function of the writing voltage (Vtip) obtained from a 15-nm diameter BaTiO3 nanowire. Each data point was obtained by applying the writing voltage for 3 min. and subsequently measuring the shift in the cantilever resonance frequency at Vtip ⫽ ⫺2 V. The Vtip scan sequence was from 0 to ⫺10 V (▲), ⫺10 V to ⫹10 V (■ ), and ⫹10 V to ⫺10 V (●). The distance between the tip and the top surface of the nanowire was 10 nm during the writing and 35 nm during the reading.

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 7. (a) Three dimensional topographic image of a 12-nm diameter BaTiO3 nanowire. (b)–(f) Successive three dimensional EFM images showing that four distinct polarization domains can be independently manipulated by an external electric field. In these EFM images, the bright and dark colors correspond to a resonance frequency shift of ⫹10 Hz and ⫺10 Hz, respectively, and the white arrows denote the polarization directions. The upward and downward polarization spots were written with Vtip ⫽ ⫺10 V and Vtip ⫽ ⫹10 V, respectively. The distance between the tip and the top surface of the nanowire was 10 nm during the writing procedure and 35 nm during the reading procedure.

polarization domains can be induced and independently manipulated on a single BaTiO3 nanowire. Here, the second polarization domain was intentionally written over a 200-nm long wire segment, whereas the rest three domains were written with the tip position fixed over a particular spot on a nanowire. The EFM images clearly

Ferroelectric Nanowires

91

show that the second polarization domain appears 3–4 times larger compared to other circular spots, illustrating that the EFM technique can distinguish polarization domains with different lengths along the nanowire. Close inspection of Figs. 5 and 7 provides information on the smallest possible size of a polarization domain written on a nanowire. Specifically, Figs. 5 and 7 show that an EFM spot induced by a tip fixed at a particular position is roughly circular in shape and that its size is ~30 nm as determined by half width at half maximum. This size is roughly the same as the distance between the tip and the nanowire surface during the reading procedure, and it shows that the size of the smallest EFM feature is limited by the tip-sample distance, as expected from the long-ranged nature of electrostatic interactions. Considering that the size of a polarization domain perpendicular to the wire axis is limited by the nanowire diameter, the circularity of EFM spots nevertheless suggests that the spatial extent of the polarization along the nanowire length is comparable to the nanowire diameter and hence nanoscale polarization domains as small as ~100 nm2 in size can be induced on a nanowire. This represents an order of magnitude reduction in the smallest size of written polarization domains reported [29, 35].

4. Conclusion and Future Prospects Single-crystalline barium titanate (BaTiO3) and strontium titanate (SrTiO3) nanowires are synthesized using a solution-based method, and their properties are investigated by transmission electron microscopy and scanned probe microscopy. Pristine nanowires have diameters ranging from 5 to 60 nm and lengths exceeding 10 ␮m, and they exhibit crystalline structure with the [001] direction aligned along the wire axis. Scanned probe microscopy investigations of individual BaTiO3 nanowires show that local non-volatile electric polarization can be reproducibly induced and manipulated on nanowires as small as 10 nm in diameter. The coercive field for polarization reversal is determined to be ~7 kV/cm, and the retention time for the induced polarization exceeds 5 days. These nanowires should provide promising materials for fundamental investigations on nanoscale ferroelectricity, and they may also be useful in nanoscale nonvolatile memory applications. The availability of ferroelectric nanowires represents engaging new possibilities for understanding nanoscale ferroelectricity, paraelectricity and piezoelectricity. These nanowires are single-crystalline and essentially defect-free ferroelectrics. The synthetic method developed can be applied to a variety of other perovskite material systems to address a range of open scientific and technological challenges. Scanned probe investigations demonstrate that the ferroelectric polarization domains can be independently and reproducibly induced and manipulated on the scale of ~100 nm2, suggesting that ferroelectric nanowires may be used to fabricate non-volatile memory devices with integration densities approaching 1 terabit/cm2. The one-dimensional geometry of nanowires enables easy formation different self-assembled structures, such as crossbar arrays, which may prove useful in fabricating a variety of nanoscale functional devices.

92

Spanier et al.

References 1. M. E. Lines and A. M. Glass, Principles and Applications of Ferroelectrics and Related Materials, Clarendon Press, Oxford (1977). 2. J. F. Scott, Ferroelec. Rev. 1 (1998) 1. 3. T. Ikeda, Fundamentals of Piezoelectricity, Oxford University Press, Oxford (1990). 4. R. E. Cohen, Nature 358 (1992) 136. 5. C. H. Ahn et al., Science 276 (1997) 1100. 6. S. Mathews, R. Ramesh, T. Venkatesan and J. Benedetto, Science 276 (1997) 238. 7. A. V. Bune et al., Nature 391 (1998) 874. 8. B. H. Park et al., Nature 401 (1999) 682. 9. Y. Watanabe, M. Okano and A. Masuda, Phys. Rev. Lett. 86 (2001) 332. 10. W. Y. Shih, W.-H. Shih and I. A. Aksay, Phys. Rev. B 50 (1994) 15575. 11. S. Chattopadhyay, P. Ayyub, V. R. Palkar and M. Multani, Phys. Rev. B 52 (1995) 13177. 12. S. Tsunekawa et al., Phys. Rev. B 62 (2000) 3065. 13. J. Zhang, Z. Yin, M.-S. Zhang and J. F. Scott, Solid State Commun. 118 (2001) 241. 14. P. Ayyub, S. Chattopadhyay, R. Pinto and M. S. Multani, Phys. Rev. B 57 (1998) R5559. 15. J. J. Urban, W. S. Yun, Q. Gu and H. Park, J. Am. Chem. Soc. 124 (2002) 1186. 16. W. S. Yun, J. J. Urban, Q. Gu and H. Park, Nano Lett. 2 (2002) 447. 17. A. P. Alivisatos, Science 271 (1996) 933. 18. C. B. Murray, D. J. Norris and M. G. Bawendi, J. Am. Chem. Soc. 115 (1993) 8706. 19. Y. Suyama and M. Nagasawa, J. Am. Ceram. Soc. 77 (1994) 603. 20. H. B. Sharma and H. N. K. Sarma, Thin Solid Films 330 (1998) 178. 21. C. Liu, B. Zou, A. J. Rondinone and Z. J. Zhang, J. Am. Chem. Soc. 123 (2001) 4344. 22. D. L. Kaiser, M. D. Vaudin, G. Gillen, C.-S. Hwang, L. H. Robins and L. D. Rotter, J. Cryst. Growth 137 (1994) 136. 23. B. W. Wessels, Annu. Rev. Mater. Sci. 25 (1995) 525. 24. S. O’Brien, L. Brus and C. B. Murray, J. Am. Chem. Soc. 123 (2001) 12085. 25. X. Peng et al., Nature 404 (2000) 59. 26. V. F. Puntes, K. M. Krishnan and A. P. Alivisatos, Science 291 (2001) 2115. 27. X. Q. Chen et al., J. Vac. Sci. Technol. B 17 (1999) 1930. 28. C. Durkan, M. E. Welland, D. P. Chu and P. Migliorato, Phys. Rev. B 60 (1999) 16198. 29. C. Durkan, D. P. Chu, P. Migliorato and M. E. Welland, Appl. Phys. Lett. 76 (2000) 366. 30. T. Tybell, C. H. Ahn and J.-M. Triscone, Appl. Phys. Lett. 72 (1998) 1454. 31. T. Tybell, C. H. Ahn and J.-M. Triscone, Appl. Phys. Lett. 75 (1999) 856. 32. S. V. Kalinin and D. A. Bonnell, Appl. Phys. Lett. 78 (2001) 1116. 33. S. V. Kalinin and D. A. Bonnell, Phys. Rev. B 63 (2001) 125411. 34. G. A. Samara, Phys. Rev. 151 (1966) 378. 35. P. Paruch, T. Tybell and J.-M. Triscone, Appl. Phys. Lett. 79 (2001) 530.

Chapter 6 Growth of Oxide Nanorods through Sol Electrophoretic Deposition Steven J. Limmer and Guozhong Cao University of Washington, Department of Materials Science and Engineering & Center for Nanotechnology, 302 Roberts Hall, Box 352120, Seattle, WA 98195, USA

1. Introduction Metal oxides, particularly complex metal oxides, are important materials for various applications in industry and technology. This is due to their multi-faceted functional properties, their chemical and thermal stability, and their mechanical properties. Metal oxides (and particularly complex metal oxides) can have many unique physical properties including electronic and ionic conductivity, superconductivity, ferroelectricity, piezoelectricity, dielectric and magnetic properties [1]. These materials find a wide range of applications in electronic devices, sensors and actuators. For example, piezoelectrics (typically lead zirconate titanate, PZT) play a key role in many micro electro-mechanical systems (MEMS) [2]. Tin oxide doped indium oxide (ITO) films on glass substrates have been widely used as optically transparent electrodes in devices such as light-emitting diodes [3]. Sol–gel derived mesoporous titania films are being intensively studied in inorganic–organic hybrid photoelectrochemical cells [4]. Further, many of the physical properties of oxide materials are tunable through appropriate doping or substitution [5]. Zirconia that is partially stabilized through doping with materials such as calcium oxide or yttrium oxide exhibits excellent mechanical properties, particularly toughness not commonly found in other oxide materials [6]. Doped zirconia is also an excellent oxygen ionic conductor, with applications in oxygen sensors and solid oxide fuel cells (SOFC) [6]. Oxide surfaces can have special chemical properties, making them useful as catalysts and sensors [7]. Furthermore, oxide surfaces can be easily incorporated with organic functional groups through surface condensation or self-assembly [8]. For many of above-mentioned applications of metal oxides, the sensitivity or efficiency obtained is directly proportional to the surface area of the material. Nanorods or nanowires could offer a significantly larger surface area compared to that of films

94

Limmer and Cao

or the bulk material. Although nanorods or nanowires have less surface area as compared with nanoparticles, they offer a great advantage in device fabrication, particularly uniformly sized nanorods or nanowires with unidirectional alignment. Nanorods or nanowires can function as both structural and functional components. Further, nanorods and nanowires could offer different physical properties, and thus find different applications in nanotechnology. Nanorods or nanowires also offer the opportunity to study the physical properties of one-dimensional structures. Numerous researchers have studied the synthesis and fabrication of nanorods and nanowires. Many different synthesis techniques have been developed. Examples include oxidation of metallic nanorods, vapor–liquid–solid (VLS) growth under vacuum, and template filling of oxide colloidal particles. Oxidation of metallic nanorods or nanowires is one established method to create oxide nanorods. However, this method is likely to be limited to simple metal oxide nanorods [9]. VLS growth of oxide nanorods and nanowires is restricted to systems that can form a eutectic liquid with catalyst at growth temperature [10]. There is very limited information in literature about the formation of eutectic liquids in complex oxide and catalyst systems. Further, compared to other materials, oxides typically possess a high melting point and are thus likely to form a eutectic liquid only at high temperatures, requiring a high processing temperature for the growth of nanorods. Martin and his coworkers [11], along with a few other groups, have published extensively on the fabrication of oxide nanorods by filling templates with sols or colloid dispersions. Using this method, many oxide nanorods have been synthesized. Examples include nanorods of TiO2, V2O5, WO3, ZnO [11], Ga2O3 and In2O3 [12]. This method does offer several advantages over VLS technique and oxidation of metallic nanorods or nanowires. One of the advantages is the possibility of fabrication of complex oxide nanorods with precise control of stoichiometric composition. Another advantage is the simplicity of the method. Further, it offers the possibility to fabricate aligned unidirectional and uniformly sized oxide nanorods over a very large area, which is particularly attractive for device fabrication and property measurements. However, complete filling of solid inside holes could be challenging, considering the fact that typical sols or colloidal dispersions consist of 90% or more solvent. In fact, the structures synthesized by this technique are often hollow tubes rather than solid rods [11, 12]. In this chapter, we describe an approach to synthesize and fabricate nanorods of complex oxides. This approach combines several processing methods together: including sol–gel processing, electrophoretic deposition and template-based growth. This method offers the possibility of making nanorods of any complex oxides, organic– inorganic hybrids and bio-inorganic hybrids. In addition, this technique would allow the fabrication of unidirectionally aligned and uniformly sized nanorods with desired patterns for device fabrication, physical property measurements and characterization.

2. Sol–Gel Processing Sol–gel processing is a wet chemical route for the synthesis and processing of inorganic and organic–inorganic hybrid materials. Sol–gel processing offers many advantages, including low processing temperatures (typically ⬍100 ⬚C) and molecular

Oxide Nanorods through Sol Electrophoretic Deposition

95

level homogeneity. Sol–gel processing is particularly useful in making complex metal oxides and temperature sensitive organic–inorganic hybrid materials. This section will briefly summarize some fundamentals and key issues of sol–gel processing. For more details, readers may wish to consult the abundant literature in this field. For instance, the books Sol–Gel Science by Brinker and Scherer [13], Introduction to Sol–Gel Processing by Pierre [14], and Sol–Gel Materials by Wright and Sommerdijk [15] provide excellent comprehensive coverage on sol–gel processing and materials. Typical sol–gel processing consists of hydrolysis and condensation of precursors. Precursors can be either organic, such as metal alkoxides, or inorganic salts. Organic or aqueous solvents may be used to dissolve precursors, and catalysts are often added to promote and/or control both hydrolysis and condensation reactions. Hydrolysis: M(OEt)4 ⫹ x H2O → M(OEt)4⫺x(OH)x ⫹ x EtOH Condensation: M(OEt)4⫺x (OH)x ⫹ M(OEt)4⫺x (OH)x → (OEt)4⫺x(OH)x⫺1M ⫺ O ⫺ M(OEt)4⫺x(OH)x⫺1 ⫹ H2O Hydrolysis and condensation reactions are both multiple-step, reversible processes, occurring sequentially and in parallel. Condensation results in the formation of nanoscale clusters of metal oxides and hydroxides, often with organic groups attached to them. These organic groups may be due to incomplete hydrolysis, or introduced as non-hydrolyzable organic ligands. The size of the nanoscale clusters, along with the morphology and microstructure of the final product, can be tailored by controlling the hydrolysis and condensation reactions. One of the greatest advantages of sol–gel processing is the ability to synthesize and process complex oxides. This requires appropriate design and control of the hydrolysis and condensation reactions of the various constituent precursors. Ideally, the constituent materials should be homogeneously mixed at the atomic/molecular level, with the desired stoichiometric ratio in each nanoscale clusters. The challenge comes from the fact that each precursor can have different chemical reactivity, so that the hydrolysis and condensation reaction rates may differ significantly from one precursor to another. Consequently, each precursor may form nanoclusters of its own (single) metal oxide, yielding a mixture of nanoscale particles of constituent simple oxides in the sol. The resulting product would be a mixture or a composite of multiple oxide phases, instead of a single-phase complex oxide. There are several ways to avoid this homo-condensation, and achieve a homogeneous mixture of multiple components at the molecular/atomic level. Modification of precursors, the use of complex precursors, partial hydrolysis and multi-step hydrolysis and condensation are commonly the approaches utilized. Another problem in making complex oxide sols is that the constituent precursors may exert a catalytic effect on one another. As a result, the hydrolysis and condensation reaction rates in combination may be significantly different from those when the precursors are processed separately. Incorporating organic components into an oxide system by sol–gel processing makes it easy to form organic–inorganic hybrids. One approach is to co-polymerize or

96

Limmer and Cao

co-condense both the inorganic precursor(s), which lead to the formation of the inorganic component, and the organic precursor(s), which consist of non-hydrolyzable organic groups. Such organic–inorganic hybrids are a single-phase material, in which the organic and inorganic components are linked through chemical bonds. Another approach is to trap the desired organic components physically inside the inorganic or oxide network, by either homogeneously dispersing the organic components in the sol, or infiltrating the organic molecules into the gel network. Similar approaches can be applied for the incorporation of bio-components into oxide systems. Another method to incorporate bio-components into the oxide structure is to use functional organic groups to bridge inorganic and biological species. In sol preparation, not much attention has been paid to the control of crystallization or formation of crystal structure, although the formation of crystalline structure without high-temperature firing is desired for some applications. Matsuda and coworkers have demonstrated that it is possible to form the crystalline phase of BaTiO3 without high temperature sintering by carefully controlling processing conditions, including concentrations and temperature [16]. However, there is still a lack of general understanding on the control of crystallization during sol preparation. By a careful control of sol preparation and processing, monodisperse nanoscale particles of various oxides, including complex oxides, organic–inorganic hybrids and biomaterials, can be synthesized. The key issue is to promote a simultaneous, homogeneous nucleation in a diffusion-controlled system. The particle size can be varied by changing the concentration and aging time [13]. In a typical sol, nanoclusters formed by hydrolysis and condensation reactions commonly have a size ranging from 1 to 100 nm. Such nanoscale particles have many applications in nanostructured material fabrication and processing. For example, they are building blocks for the formation of photonic bandgap crystals [17]. Other researchers explore the use of these nanoscale particles for optical applications by constructing various core–shell structures [18]. In a sol, just like in other colloidal systems, gravity is a negligible factor, whereas Brownian motion plays an important role. Nanoclusters or nanoparticles possess a huge surface area/volume ratio and thus a large surface energy. There is a strong tendency for such nanoscale clusters to agglomerate. Two types of mechanisms are available to prevent such agglomeration in a sol. One is polymeric or steric stabilization, and another is electrostatic stabilization. Polymeric stabilization works by adsorbing polymeric molecules onto the nanocluster or nanoparticle surface; spatial exclusion then prevents two clusters from getting close enough to agglomerate. Electrostatic stabilization is based on the surface charge of nanoclusters or nanoparticles in a sol. Such a surface charge will interact with other charged species in the sol to form a charged structure around the particle, which in turn introduces an energy barrier to prevent two particles from approaching one another. Electrosteric stabilization is another mechanism, which combines steric and electrostatic stabilization mechanisms, where the electrostatic effect is due either to the surface charge on the solid surface, or an uneven charge distribution in the polymer molecules. A sol is a very dilute system, typically consisting of 90% or more solvent by volume. Upon drying, there can be a significant amount of shrinkage, resulting in severe cracking in sol–gel derived films and monoliths. The formation of cracks often limits sol–gel processing to the synthesis of thin films and structures less than

Oxide Nanorods through Sol Electrophoretic Deposition

97

a micron in size. This limitation, however, would not be a problem for the fabrication and process of nanostructured materials. Another potential limitation for some applications is that sol–gel-derived films or monoliths are highly porous. A sintering process at elevated temperatures is often required to obtain dense films or monoliths. Furthermore, sol–gel processing often produces amorphous oxides, potentially requiring elevated temperatures for crystallization.

3. Electrophoretic Deposition The electrophoretic deposition technique has been widely explored, particularly in film deposition from colloidal dispersions. As briefly mentioned in the previous section, nanosized particles in colloidal dispersions including sols can be stabilized by electrostatic or electrosteric mechanisms. When dispersed in a polar solvent or an electrolyte solution, the surface of nanoparticles develops an electrical charge via one or more of the following mechanisms: (1) preferential dissolution or (2) deposition of charges or charged species, (3) preferential reduction or (4) oxidation, and (5) adsorption of charged species such as polymers. Charged surfaces will electrostatically attract oppositely charged species (typically called counter-ions) in the solvent or solution. A combination of electrostatic forces, Brownian motion and osmotic forces would result in the formation of a so-called double layer structure, as schematically illustrated in Fig. 1. The figure depicts a positively charged particle surface, the

Fig. 1. Depiction of a positively charged particle surface, the concentration profiles of negative ions (counter-ions) and positive ions (surface-charge-determining-ions) and the electric potential profile. The concentration of counter-ions gradually decreases with distance from the particle surface, whereas that of charge-determining ions increases. As a result, the electric potential decreases with distance.

98

Limmer and Cao

Fig. 2. Schematic of the electrophoretic motion of a charged particle in a colloidal suspension, demonstrating the motion in the direction of the applied electric field. Some of the solvent or solution surrounding the particle will move with it, since part of the solvent or solution is tightly bound to the particle.

concentration profiles of negative ions (counter-ions) and positive ions (surfacecharge-determining-ions) and the electric potential profile. The concentration of counter-ions gradually decreases with distance from the particle surface, whereas that of charge-determining ions increases. As a result, the electric potential decreases with distance. Near to the particle surface, the electric potential decreases linearly, in the region known as the Stern layer. Outside of the Stern layer, the decrease follows an exponential relationship, and the region between Stern layer and the point where the electric potential equals zero is called the diffusion layer. Together, the Stern layer and diffusion layer are called the double layer structure in the classic theory of electrostatic stabilization. Upon application of an external electric field to a colloidal system or a sol, the constituent charged particles are set in motion in response to the electric field, as schematically illustrated in Fig. 2. This type of motion is referred to as electrophoresis. When a charged particle is in motion, some of the solvent or solution surrounding the particle will move with it, since part of the solvent or solution is tightly bound to the particle. The plane that separates the tightly bound liquid layer from the rest of the liquid is called the slip plane. The electric potential at the slip plane is known as the zeta-potential. Zeta-potential is an important parameter in determining the stability of a colloidal dispersion or a sol; a zeta potential larger than about 25 mV is typically required to stabilize a system [19]. Zeta potential is determined by a number of factors, such as the particle surface charge density, the concentration of counter-ions in the solution, solvent polarity and temperature. The zeta potential around a spherical particle can be described as [20]:

⫽

Q 4␲␧ra (1⫹␬a)

(1)

Oxide Nanorods through Sol Electrophoretic Deposition

99

with ␬⫽

冢

兺

e2 niz2i ␧r␧0kT

冣

1/2

where Q is the charge on the particle, a is the radius of the particle out to the shear plane, ␧r is the relative dielectric constant of the medium, and ni and zi are the bulk concentration and valence of the ith ion in the system, respectively. It is worthwhile to note that a positively charged surface results in a positive zeta potential in a dilute system. A high concentration of counter ions, however, can result in a zeta-potential of the opposite sign. The mobility of a nanoparticle in a colloidal dispersion or a sol is dependent on the dielectric constant of the liquid medium, the zeta potential of the nanoparticle, and the viscosity of the fluid. Several forms for this relationship have been proposed, such as the Hückel equation [20]:

⫽

2r0

3␲␩

(2)

Double layer stabilization and electrophoresis are extensively studied subjects. Readers may find additional detailed information in books on sol–gel processing [13–15] and colloidal dispersions [20–21]. Electrophoretic deposition simply uses such an oriented motion of charged particles to grow films or monoliths by enriching the solid particles from a colloidal dispersion or a sol onto the surface of an electrode. If particles are positively charged (more precisely speaking, having a positive zeta potential), then the deposition of solid particles will occur at the cathode. Otherwise, deposition will be at the anode. At the electrodes, surface electrochemical reactions proceed to generate or receive electrons. The electrostatic double layers collapse upon deposition on the growth surface, and the particles coagulate. There is not much information on the deposition behavior of particles at the growth surface. Some surface diffusion and relaxation is expected. Relatively strong attractive forces, including the formation of chemical bonds between two particles, develop once the particles coagulate. The films or monoliths grown by electrophoretic deposition from colloidal dispersions or sols are essentially a compaction of nanosized particles. Such films or monoliths are porous, i.e., there are voids inside. Typical packing densities, defined as the fraction of solid (also called green density) are less than 74%, which is the highest packing density for uniformly sized spherical particles [22]. The green density of films or monoliths by electrophoretic deposition is strongly dependent on the concentration of particles in sols or colloidal dispersions, zeta-potential, externally applied electric field and reaction kinetics between particle surfaces. Slow reaction and slow arrival of nanoparticles onto the surface would allow sufficient particle relaxation on the deposition surface, so that a high packing density is expected. Many theories have been proposed to explain the processes at the deposition surface during electrophoretic deposition. Electrochemical process at the deposition surface or electrodes is complex and varies from system to system. However, in

100

Limmer and Cao

general, a current exists during electrophoretic deposition, indicating reduction and oxidation reactions occur at electrodes and/or deposition surface. In many cases, films or monoliths grown by electrophoretic deposition are electric insulators. However, the films or monoliths are porous and the surface of the pores would be electrically charged just like the nanoparticle surfaces, since surface charge is dependent on the solid material and the solution. Furthermore, the pores are filled with solvent or solution that contains counter-ions and charge-determining ions. The electrical conduction between the growth surface and the bottom electrode could proceed via either surface conduction or solution conduction. Since films or monoliths grown by electrophoretic deposition are porous, post deposition sintering at elevated temperatures is usually required to form a dense material. However, considering the fact that the films or monoliths are a compaction of nanosized particles, sintering or densification is relatively easier than conventional ceramic sintering. If the initial solid particles were amorphous, sintering would also induce crystallization.

4. Experimental A combination of sol–gel processing and electrophoretic deposition has been used to synthesize a variety of oxide nanorods, such as TiO2, SiO2, BaTiO3, PZT and Sr2Nb2O7 [23]. An overview of the chemicals and conditions for the preparation of the various sols is given in Table 1. Briefly, the sols were prepared as follows. TiO2 sol was formed by dissolving titanium (IV) isopropoxide in glacial acetic acid, followed by the addition of DI water. Upon addition of the water, a white precipitate is instantaneously formed. However, the precipitate dissolved and the sol became a clear liquid after approximately 5 minutes of stirring. BaTiO3 sol was prepared by first dissolving barium acetate in glacial acetic acid. To this solution, titanium (IV) isopropoxide was added, followed by ethylene glycol, and the sol was stirred at 90 ⬚C for 1 hr. Sr2Nb2O7 sol was prepared as reported in Reference [24], by mixing ethanol and Table 1. An overview of the chemicals and conditions for the preparation of the various sols Sol

Precursors

Solvents/Other chemicals

pH

TiO2

Titanium (IV) isopropoxide

1.8

BaTiO3

Titanium (IV) isopropoxide, barium acetate Tetraethyl orthosilicate

Glacial acetic acid, water, lactic acid, glycerol, ethylene glycol Glacial acetic acid, ethylene glycol Ethanol, water, hydrochloric acid Ethylene glycol, ethanol, citric acid, water Glacial acetic acid, water, lactic acid, glycerol, ethylene glycol

SiO2 Sr2Nb2O7 Pb(Zr,Ti)O3

Strontium nitrate, niobium chloride Lead (II) acetate, titanium isopropoxide, zirconium n-propoxide

4.4 3 1 4.1

Oxide Nanorods through Sol Electrophoretic Deposition

101

ethylene glycol and heating to 40 ⬚C. This was followed by the addition of citric acid, strontium nitrate and niobium pentachloride while stirring, with water slowly added to the mixture to form the sol. The SiO2 sol was made by dissolving tetraethyl orthosilicate in a mixture of ethanol and DI water, with a small amount of hydrochloric to adjust the pH. Preparation of the PZT sol was described in detail in Reference [25]. Briefly, lead (II) acetate was dissolved in glacial acetic acid at 110 ⬚C, and the solution was allowed to cool back to room temperature. Titanium (IV) isopropoxide and zirconium (IV) n-propoxide were mixed, and added to the cooled lead solution. Deionized water was added to initiate and sustain hydrolysis and condensation reactions; lactic acid, glycerol and ethylene glycol were used to adjust the viscosity and stability of the sol. Nanorods of these oxides were formed by the following method. Nanorod growth occurred on a working electrode of aluminum, with a Pt mesh used as the counter electrode. The template membranes used for the growth of the nanorods were tracketched hydrophilic polycarbonate, with pore diameters of 50–200 nm, and a thickness of 10 ␮m. The polycarbonate (PC) membrane and the working electrode are placed in a polypropylene filter holder, and held in place with a silicone gasket. This assembly is placed in contact with the sol. Alternatively, the polycarbonate (PC) membrane can be attached to the working electrode with a piece of double-sided conductive (carbon) tape, and placed on top of and just in contact with the sol. The sol is drawn into the membrane pores by capillary action. A Pt counter electrode is also placed in the sol, parallel to the working electrode. Fig. 3 shows a schematic of the experimental setup. For the electrophoretic growth, a potential of 5 V is applied between the electrodes, and held for up to 60 minutes. At the end of the electrophoretic deposition, excess sol is blotted off the membrane. Samples prepared in this manner are dried at ~100 ⬚C for several hours, then placed in an oven and fired for 15–60 min. The temperature for this firing is either 500 ⬚C (SiO2, TiO2) or 700 ⬚C (BaTiO3, Sr2Nb2O7 and PZT). This is to burn off the polycarbonate membranes, make the nanorods dense and crystallize the material (for all but SiO2). Scanning electron microscopy (SEM, JEOL 840A) and transmission electron microscopy (TEM, JEOL 2010) were used to study the morphology and crystallinity of the nanorods. Samples were sputter-coated with a thin Au/Pd layer prior to observation in the SEM. For all samples that were expected to be crystalline (all except SiO2), X-ray diffraction (XRD, Phillips PW1830) was used to determine the phases and crystal structures present, and to check for the presence of texture and degree of crystallinity.

5. Various Oxide Nanorods Fig. 4 shows SEM micrographs of different sizes of TiO2 nanorods grown in a polycarbonate membrane by sol–gel electrophoresis. These nanorods have a uniform diameter throughout their entire length, with a surface that is smooth over much or all of the length. Comparing the various rods in each image, one can see that they all have roughly the same length and diameter. These images show that the rods are roughly parallel to one another over a broad area. The diameters of the TiO2 nanorods are

Fig. 3. Schematic of the growth apparatus. The polycarbonate (PC) membrane and the working electrode are placed in a polypropylene filter holder, and held in place with a silicone gasket. This assembly is placed in contact with the sol. (b)

(a)

(c)

Fig. 4. SEM micrographs of different sizes of TiO2 nanorods grown in a polycarbonate membrane by sol–gel electrophoresis. The diameters are approximately: (a) ~180 nm (for the 200 nm template) diameter rods, (b) ~90 nm (for the 100 nm template) and (c) ~45 nm (for the 50 nm template).

Oxide Nanorods through Sol Electrophoretic Deposition

103

estimated to be ~180 nm (for the 200 nm template), ~90 nm (for the 100 nm template), and ~45 nm (for the 50 nm template). This corresponds to approximately 10% shrinkage with respect to the membrane pore diameter. This size difference is most likely due to the volume shrinkage caused by densification during the heat treatment. Fig. 5a shows a TEM micrograph of a TiO2 nanorod, demonstrating that the nanorods are quite smooth and dense. Fig. 5b shows a high-resolution TEM image and electron diffraction pattern, demonstrating that the nanorods are polycrystalline, with grains that are ~5 nm in size. XRD spectra of the TiO2 rods are shown in Fig. 6, along with the spectra for a powder formed from the same sol and fired at 500 ⬚C for 60 min. From the powder XRD spectrum, it can be seen that the sample consists entirely of the

(a)

(b)

Fig. 5. Part (a) shows a TEM micrograph of a TiO2 nanorod, demonstrating that the nanorods are quite smooth and dense. Part (b) shows a high-resolution TEM image and electron diffraction pattern, demonstrating that the nanorods are polycrystalline, with grains of ~5 nm in size.

Fig. 6. XRD spectra of both the grown TiO2 nanorods and a powder derived from the same sol. Both samples consist the anatase phase, and there is no observed shift in the peak positions for the nanorod sample. In addition, the relative intensities of the peaks are the same for the nanorod sample, showing that there is no preferred orientation in the sample.

104

Limmer and Cao

anatase phase. Comparison of the two spectra shows that there are identical peaks in both samples. Further, the peak positions are the same and the intensity ratios among various peaks are similar. Note that there is a large amorphous background associated with the TiO2 nanorods. This is due to the small volume rods available for analysis, meaning that much of the X-ray beam was hitting the (amorphous) sample holder. For the TiO2 rods grown in 50 nm templates, it was found that the drying time used has a strong influence on the formation of the nanorods. When the samples were dried at 100 ⬚C for times up to 24 hours, no nanorods were observed after firing at 500 ⬚C. However, if the samples were dried for ~48 hours prior to firing, nanorods were observed. It is likely that this result is due to the increased degree of condensation that occurred in the sample dried for a longer time. By allowing the samples to undergo further condensation reactions, the rods formed were likely stronger, and thus better able to resist breakage upon firing. Fig. 7 shows SEM micrographs of BaTiO3, SiO2, Sr2Nb2O7 and PZT nanorods grown by sol–gel electrophoresis. The BaTiO3 nanorods have a diameter of about 150 nm, which is about 25% smaller than the template pores. The length of the BaTiO3 rods is about 10 ␮m, similar to the thickness of the template membrane. The SiO2 nanorods have a diameter of ~200 nm, which is the same as the template pores.

(a)

(b)

(c)

(d)

Fig. 7. SEM micrographs of other oxide nanorods grown by sol–gel electrophoresis. (a) BaTiO3 nanorods grown in a polycarbonate membrane with 200 nm diameter pores. (b) Sr2Nb2O7 grown in a polycarbonate membrane with 200 nm diameter pores. (c) SiO2 nanorods grown in a polycarbonate membrane with 200 nm diameter pores. (d) PbZr0.52Ti0.48O3 (PZT) nanorods grown in a PC membrane with 200 nm diameter pores.

Oxide Nanorods through Sol Electrophoretic Deposition

105

Under the acidic synthesis conditions used, the silica sol should be polymeric [13]. Such a polymeric sol would pack very densely in the template pores, and the silica would thus contain only very small pores. This makes it likely that the SiO2 nanorods were unable to become fully dense at the temperature used (500 ⬚C), and thus would undergo negligible shrinkage. The Sr2Nb2O7 and PZT nanorods have diameters of ~125 nm and ~150 nm, respectively. This corresponds to shrinkage of about 37% and 25% for these two materials, suggesting that the Sr2Nb2O7 did not pack as densely as the other materials. Similar to the TiO2 nanorods, these nanorods also exhibit a uniform diameter throughout their entire length, and have a relatively smooth surface. For each composition, the nanorods all have roughly the same length and diameter. These images also show that the rods are formed over a broad area on the surface, and are arranged roughly parallel to one another. Also similar to the TiO2 nanorods, the composition of a complex oxide formed by this technique (such as Sr2Nb2O7) will consist of the desired crystalline phase with the desired stoichiometry. We have recently demonstrated this fact [25] for the complex oxide PZT. Fig. 8 shows XRD spectra of the PZT nanorods and PZT powder prepared from the same sol; both PZT nanorods and powder consisted of only one crystalline phase, perovskite PZT without any detectable secondary phase. Comparison of the two spectra shows that there are identical peaks in both samples. Further, the peak positions are the same and the intensity ratios among various peaks are identical. Similar XRD results were obtained when comparing BaTiO3 powder and nanorods. This demonstrates that the electrophoretic deposition does not negatively affect the stoichiometry and compositional homogeneity that is achieved during

Fig. 8. XRD spectra of both the grown PZT nanorods and a powder derived from the same sol. Both samples show only a single perovskite phase. This demonstrates that sol–gel electrophoresis can be used to form complex oxides with the desired stoichiometry and crystal phase.

106

Limmer and Cao

the sol preparation, and thus yields complex oxide rods of the desired phase and composition. Under the experimental conditions applied in the current study, the surface charge and the zeta-potential of the nanoclusters of the sol are of the same sign. For the TiO2 sol, the pH is 1.8, well below the isoelectric point (6.0) [26], and thus the zeta potential (and particle surface) would be positive in this dilute sol. For BaTiO3, the pH is 4.4, which also leads to positively charged particles. In the SiO2 sol, the pH is about 3, which is just above the isoelectric point (~2), yielding negatively charged particles. For PZT, since the pH of the sol (4.1) is below the reported isoelectric point (~7.6) [27], the zeta potential is positive. In the case of the Sr2Nb2O7 sol, no information about the isoelectric point could be found in the literature. We assume that the sol contains positively charged particles at the given pH (~1), since oxide particles are often positively charged at such a low pH. The successful growth of Sr2Nb2O7 nanorods demonstrates that this is a reasonable assumption. Upon application of an appropriate electric potential, the charged nanoclusters in the sol will be drawn towards the anode for SiO2, and the cathode for the other oxides. The nanoclusters will fill the pores of the membrane, starting at the bottom, which is directly connected to the working electrode. In this manner, all the pores will eventually be completely filled. Prolonged deposition times lead to a layer of oxide forming on the membrane surface after the pores are filled. If we assume that the nanoclusters are uniformly sized spheres, then the highest possible packing density is 74% [22]. This would also be the highest achievable density of the nanorods before densification. If there is a range of sizes in the nanoclusters, even denser packing could be possible. Upon heating the nanorods to an elevated temperature, densification will occur along with shrinkage. This explains why the observed diameter of the nanorods is smaller than that of the membrane pores. Although we do not know how closely the nanoclusters are packed during the electrophoretic deposition, a lateral shrinkage of approximately 10–30% was observed when the nanorods were fired. Since the times and temperatures used are sufficient to form fully dense films from these sols (except SiO2), it is reasonable to assume that the nanorods are also fully dense after firing. This in turn implies that near ideal close packing of nanoclusters might be achieved by this process. The existence of broken nanorods in these figures could be explained as follows. The polycarbonate membranes templates burn off at approximately 400 ⬚C in air, but the oxide nanorods are not likely to be fully dense (or crystallized) at this temperature, and thus have very limited mechanical strength. As the membrane and rods are heated, it is expected that some nanorods would break due to the differences in thermal expansion coefficients and distortion of the membranes. As noted earlier, the use of longer drying times induces a greater degree of condensation prior to firing of the samples. This in turn should make the samples stronger and more resistant to breakage during firing. This is being investigated in the formation of smaller TiO2 nanorods, as well as the other oxide materials. Fig. 9 is a schematic drawing of the electrophoretic deposition process. It demonstrates the steps we believe occur in the growth process. At the beginning of the nanorod growth, positively charged sol particles move due to electrophoresis towards the negative electrode. They deposit at the bottom of the pore, while the negatively charged

Oxide Nanorods through Sol Electrophoretic Deposition

107

Fig. 9. Schematic of the nanorod growth process, demonstrating the electrophoretic motion of charged oxide particles into the pores of the template membrane, filling the pores from bottom up as a function of time.

counter ions move in the opposite direction. As time increases, the densely packed sol particles fill more of the pore, until the pore is completely filled. Dense cross-sections of TiO2 nanorods observed by SEM and TEM, shown in Figs. 4 and 5, suggest that the growth of oxide nanorods does indeed follow the mechanism proposed in Fig. 9. The results presented so far clearly demonstrate the ability of the sol electrophoresis technique to form nonconductive nanorods. Beyond this, however, there is no kinetic information about the process by which nanorod formation occurs. Monitoring the current that flows in the growth cell during nanorod formation can yield some insights into the growth process and kinetics of the electrophoresis. Fig. 10a is a sample of the recorded current as a function of growth time for TiO2 nanorods grown in a 100 nm template, with 5 V applied potential, and a spacing of 3 cm between the electrodes. From this data, one can see that the current rises quickly in the beginning of growth, and follows with a gradual increase after ~1500 seconds. It is postulated that the initial sharp current rise corresponds to the filling of the template pores from the bottom through the template. As the template pores are being filled with TiO2 nanoparticles or clusters, the distance that charged species, either ions or TiO2 nanoclusters, must move through the small template pores decreases. This results in the drop of total resistance of the system, increasing the current. There is another possible mechanism leading to an increase in current due to the formation of a conductance path through the deposited TiO2. Anatase is a wide band-gap semiconductor, and thus may conduct some of the current. It is possible that the resistivity of the deposited anatase is

108

Limmer and Cao (a)

(b)

Fig. 10. (a) is a sample of the recorded current as a function of growth time for TiO2 nanorods grown in a 100 nm template, with 5 V applied potential, and a spacing of 3 cm between the electrodes. (b) is a sample calculation of the current for TiO2 grown in 100 nm templates at 5 V and 3 cm separating the electrodes.

less than that of the anatase clusters moving in the sol, raising the current as the nanorods grow. After the pores are completely filled, the current rise with time is more gradual. This corresponds to the formation of a film of TiO2 on top of the template. To see how the data fits with the proposed growth model, it is useful to calculate the current vs. time behavior assuming the model, to see if it fits with the observed data. The current was calculated using a method similar to that of [28], where the current path was broken into three sections, as shown in Fig. 11a. In each section, the resistance of the circuit is considered to consist of two parallel components. In Section 1, they are counter-ions diffusing through the pores of TiO2 deposit and conduction of TiO2 deposit. In Sections 2 and 3, they are TiO2 nanoparticles and counter-ions diffusing in the sol. Fig. 11b shows the equivalent circuit used to calculate the current. The resistances are: R1,sol ⫽

␳soldl ApNp

(3a)

R1,p ⫽

␳TiO2dl ApNp

(3b)

Oxide Nanorods through Sol Electrophoretic Deposition

109

R2,sol ⫽

␳sol(s ⫺ dl) ApNp

(3c)

R2,p ⫽

d ⫺ d1 ApNp␮Qc

(3d)

R3,sol ⫽

␳sol(d ⫺ s) Ac

(3e)

d ⫺ d1 Ac␮Qc

(3f)

R3,p ⫽

where s is the thickness of the template, dl is the thickness of the deposit, d is the distance between electrodes, Ap is the cross-sectional area per template pore, Np is the total number of pores and Ac is the cross-sectional area of the growth cell. In the resistances, ␳sol is the ionic resistivity of the sol, ␳TiO2 is the resistivity of the TiO2 deposit, ␮ is the mobility of the TiO2 nanoparticles, Q is the effective charge per unit mass of the TiO2 nanoparticles and c is the concentration of the sol. Because of the small volume of TiO2 depleted from the sol (approximately ~1% for the volume of sol used) by nanorod growth, the concentration of the sol is assumed to remain constant. In this circuit, the resistances due to the electrodes and the interface reactions are neglected, as they are likely to be constant with time. This circuit only applies for times up to the moment when the template pores are completely filled, after that the system consists of three different sections as shown in Fig. 11c. For this circuit, the resistances in sections 1 and 3 are similar to those above, (a)

(c)

(b)

Fig. 11. The models used in calculation: (a) the three sections of the growth process, for dl ⬍ s, (b) the equivalent circuit for part (a), and (c) the three sections of the growth process, for dl ⬎ s.

110

Limmer and Cao

modified for the fact that dl ⬎ s. However, in section 2 the resistances are similar to those in section 1, with ApNp replaced by Ac. These equations give the current as a function of the grown thickness, dl. In order to compare the calculated currents to the observed values, it is necessary to have the current as a function of time. Assuming that the concentration and electric field (E) are constant with time, the mass deposited with time due to electrophoresis is equal to [29]: Mdeposit ⫽ ␳depositVdeposit ⫽ ␮EAdepositct

(4)

For all times where dl ⬍ s, the volume of the deposit is: Vdeposit ⫽ ApNpdl

(5)

where ApNp is the area of the deposit. Combining these two equations and solving for t, we find that: t⫽

␳depositdl(d ⫺ dl) ␮cV

(6)

where V is the voltage drop across sections 1 and 2, and can be calculated from the applied voltage (Va) and current as:

冢

V ⫽ Va ⫺ I

1 ⫺ 1 R1,sol R1,p

冣

⫺1

(7)

A similar expression can be derived for dl ⬎ s. Using Equations 3–7, one can calculate I and t as functions of dl, and combine these to plot I as a function of t. A sample of such a calculation is shown in Fig. 10b, for TiO2 grown in 100 nm templates at 5 V and 3 cm separating the electrodes. While the currents calculated by this method are much larger than those measured are, they show the identical qualitative trend. That is, the current increases sharply up to the time when the template pores are completely filled, and then increases more gradually as a film of TiO2 forms over the entire template. It is observed that samples grown for longer times have a thick film attached to the nanorods, while those grown for shorter times have little or no film. Both experimental results and calculation strongly supports the proposed model, for the growth of non-conductive oxide nanorods by electrophoresis that nanorod growth proceeds via motion of the nanoparticles to the bottom of the template pores, filling them up as time proceeds.

6. Concluding Remarks Recently, a modified version of this sol–gel electrophoretic technique has been demonstrated. Miao et al. [30] prepared single crystalline TiO2 nanowires by templatebased electrochemically induced sol–gel deposition. Titania electrolyte solution was

Oxide Nanorods through Sol Electrophoretic Deposition

111

prepared using a method developed by Natarajan and Nogami [31], in which Ti powder was dissolved into a H2O2 and NH4OH aqueous solution, forming TiO2⫹ ionic clusters. When an external electric field was applied, TiO2⫹ ionic clusters diffused to the cathode and underwent hydrolysis and condensation reactions, resulting in depositing relative amorphous TiO2 gel. After heat treatment in air at 240 ⬚C for 24 hr, nanowires of single crystal TiO2 with anatase structure and with diameters of 10, 20, and 40 nm and lengths ranging from 2 to 10 ␮m were synthesized. A number of possible applications exist for nanorods synthesized in the manner described in this chapter. One example is current research we are conducting in fabricating TiO2 nanorod photoelectrochemical cells. The higher surface area of nanorods and their relatively shorter conduction path should combine to make solar cells that are more efficient. Another application is the use of the higher surface area of nanorods for sensors, detectors and catalysts. Patterned, ordered arrays of unidirectionally aligned nanorods could serve as the foundation of two-dimensional photonic bandgap crystals. Lastly, nanorods allow one to study the physical properties of one-dimensional structures.

Acknowledgments SJL acknowledges (partial) support from the Joint Institute for Nanoscience funded by the Pacific Northwest National Laboratory (operated by Battelle for the U.S. Department of Energy) and the University of Washington.

References 1. A. J. Moulson and J. M. Herbert, Electroceramics: Materials, Properties, Applications, Chapman and Hall, London (1990). 2. A. Kholkin, Ferroelectrics 258 (2001) 209. 3. Y.-H. Tak, K.-B. Kim, H.-G. Park, K.-H. Lee and J.-R. Lee, Thin Solid Films 411 (2002) 12. 4. E. Stathatos, P. Lianos, U. Lavrencic-Stangar and B. Orel, Adv. Mater. 14 (2002) 354. 5. Y. Wu, M. J. Forbess, S. Seraji, S. J. Limmer, T. P. Chou, C. Nguyen and G. Z. Cao, J. Appl. Phys. 90 (2001) 5296. 6. D. W. Richerson, Modern Ceramic Engineering, Marcel Dekker, Inc., New York (1992). 7. M. A. Peña and J. L. G. Fierro, Chem. Rev. 101 (2001) 1981. 8. N. R. E. N. Impens, P. van der Voort and E. F. Vansant, Micropor. Mesopor. Mat. 28 (1999) 217. 9. Y. Li, G. S. Cheng and L. D. Zhang, J. Mater. Res. 15 (2000) 2305. 10. Z. W. Pan, Z. R. Dai, C. Ma and Z. L. Wang, J. Am. Chem. Soc. 124 (2002) 1817. 11. B. B. Lakshmi, C. J. Patrissi and C. M. Martin, Chem. Mater. 9 (1997) 2544. 12. B. Cheng and E. T. Samulski, J. Mater. Chem. 11 (2001) 2901. 13. C. J. Brinker and G. W. Scherer, Sol–Gel Science: The Physics and Chemistry of Sol–Gel Processing, Academic Press, San Diego (1990). 14. Alain C. Pierre, Introduction to Sol–Gel Processing, Kluwer Academic Publishers, Boston (1998). 15. J. D. Wright and N. A. J. M. Sommerdijk, Sol–Gel Materials: Chemistry and Applications, Gordon and Breach Science Publishers, Amsterdam (2001).

112

Limmer and Cao

16. H. Matsuda, N. Kobayashi, T. Kobayashi, K. Miyazawa and M. Kuwabara, J. Non-Cryst. Solids, 271 (2000) 162. 17. A. Stein and R. C. Schroden, Curr. Opin. Solid St. M. 5 (2001) 553. 18. S. J. Oldenburg, R. D. Averitt, S. L. Westcott and N. J. Halas, Chem. Phys. Lett. 288 (1998) 243. 19. J. S. Reed, Introduction to the Principles of Ceramic Processing, John Wiley & Sons, New York (1988). 20. R. J. Hunter, Zeta Potential in Colloid Science: Principles and Applications, Academic Press, London (1981). 21. D. H. Everett, Basic Principles of Colloid Science, Royal Society of Chemistry, London (1988). 22. W. D. Callister, Materials Science and Engineering: An Introduction, John Wiley & Sons, New York (1997). 23. S. J. Limmer, S. Seraji, Y. Wu, T. P. Chou, C. Nguyen and G. Z. Cao, Adv. Funct. Mater. 12 (2002) 59. 24. S. Seraji, Y. Wu, N. E. Jewell-Larson, M. J. Forbess, S. J. Limmer, T. P. Chou and G. Z. Cao, Adv. Mater. 12 (2000) 1421. 25. S. J. Limmer, S. Seraji, M. J. Forbess, Y. Wu, T. P. Chou, C. Nguyen and G. Z. Cao, Adv. Mater. 13 (2001) 1269. 26. J. Gustafsson, P. Mikkola, M. Jokinen and J. B. Rosenholm, Colloid Surface A 175 (2000) 349. 27. K. Baba, O. Nishizato, A. Takamura and M. Katsube, J. Ceram. Soc. Jpn. 101 (1993) 1370. 28. L. J. Vandeperre and O. O. Van Der Biest, Innovative Processing and Synthesis of Ceramics, Glasses, and Composites 1997, American Ceramic Society, Westerville, Ohio (1997) pp. 261–272. 29. A. Simone and P. Spinelli, Mater. Eng. 13 (2002) 33. 30. Z. Miao, D. Xu, J. Ouyang, G. Guo, Z. Zhao and Y. Tang, Nano Lett. 2 (2002) 717. 31. C. Natarajan and G. Nogami, J. Electrochem. Soc. 143 (1996) 1547.

Chapter 7 Nanowires of Functional Oxides Guanghou Wang Department of Physics, Nanjing University, Nanjing, China

1. Introduction Since the discovery of carbon nanotubes, there have been many reports on the synthesis of one-dimensional nanomaterials, such as nanorods, nanobelts, naontubes, and nanocables [1–6], including nanorods of WS2 [7–10], MoS2 [11], BN [12, 13], BC2 [14], lipids, MCM-41, and peptides [14–16]. Semiconductor one-dimensional structures, i.e., nanorods and nanowires, are known to have many interesting physical properties and great potential applications in semiconductor and electronic technologies, which have stimulated intensive, worldwide research activities. Preliminary studies of some of these one-dimensional nanostructures indicate that they may be used as microscopic probes, field emission sources [17], storage materials [18], and light-emitting devices with extremely low power consumption [19]. Among these semiconductors, some metal oxides, for instance, stannic oxide, titanium oxide, cuprous oxide and vanadium pentoxide are the most important base materials in industry for gas sensors, transistors, and electrode material as well as catalysts [20, 21]. In order to investigate their properties, many methods have been developed to synthesize one-dimensional nanomaterials, such as carbon nanotubes confined chemical reaction, vapor-liquid-solid (VLS) [22–24], solution–liquid–solid (SLS) [25], and template-based synthetic approaches [26]. However, complicated apparatus, complex process control, and special condition may be required for these approaches. Molten salt synthesis (MSS) is reported to be one of the simplest techniques for preparing ceramic powders with whiskerlike [27], needlelike [28] and platelike morphology [29–31]. In this chapter we would like to give a brief introduction to some metal oxide nanowires with relatively simple synthesis methods, structural analysis and property study.

114

Wang

2. From Nanoclusters to Nanowires of Titanium Oxides 2.1. Basic structures for TiO2 Titania has three kinds of crystal structures [32], i.e., brookite, anatase (a ⫽ 3.78 Å, c ⫽ 9.51 Å; space group of I41/amd) and rutile (a ⫽ 4.59, c ⫽ 2.96, space group of P42/mnm). Each cell of rutile consists of two TiO2 molecules while anatase has four TiO2 molecules. Fig. 1 gives crystal structures of rutile (a) and anatase (b). The common features for these structures: Ti atom is positioned at the center of six O atoms, forming a deformed Ti-O octahedron. The octahedral share top points and a rhombic side joining together and thus constructing a crystal. The differences between them exist: three sides shared for the brookite, four for anatase and two for rutile. Among them rutile is the most stable and anatase is less while brookite is not stable. Therefore, the first two are commonly concerned in nanomaterials. Fig. 2(a) shows rutile conjuncted of Ti-O octahedral sharing up-side and down-side, and forming one dimensional chains along c-axis with two kinds of arrangements; (b) is rotated 90⬚ around c-axis relative to (a). Rutile crystals are constructed by joining the chains sharing top points (Fig. 2(c) and (d) respectively). 2.2. Titania nanoclusters Siegel et al. have used high temperature evaporation and gas condensation method to obtain TiO2 nanoparticles and then nanophase materials under high vacuum condition and with high pressure [33], originally developed by Gleiter et al. [34]. Li and Wang have synthesized TiO2 nanoparticles by means of microemulsion method [35]. The latter requires no complicated machines. An inverse microemulsion system consists of an oil phase, a surfactant phase and an aqueous phase, and is a thermodynamically stable isotropic dispersion of the aqueous phase in the continuous oil phase. In order to prepare inverse micro-emulsions, a cyclohexane (analytical reagent) was

Fig. 1. Two kinds of crystal structures: (a) rutile, (b) anatase.

Nanowires of Functional Oxides (a)

(b)

(c)

(d)

115

Fig. 2. Schematics of TiO2 rutile formed by joining Ti-O octahedral: (a) and (b) one dimensional chains constructed by joining octahedral sharing the sides along [001] direction, (c) two above-mentioned chains joining through top-points, (d) projection of rutile octahedron along [001] direction.

used as the oil phase and a mixture of poly(oxyethylene), nonyl phenol ether (NPS, chemical purity) and poly(oxyethylene)-9-nonyl phenol ether (NP9, chemical purity) with weight ratio 1 : 1 as the nonionic surfactant (NP5-NP9). Two microemulsion systems were prepared, containing a 0.5 M titanium tetrachloride (TiCl4) aqueous solution and a 2.0 M ammonia as the aqueous phase, respectively. The ammonia is analytical reagent grade and the TiCl4 are of chemical purity. The aqueous TiCl4 solution was prepared by adding a little hydrochloric acid (HCl) to the distilled water before TiCl4 was dissolved in the water to avoid the formation of Ti(OH)4. The oil phase, surfactant, and aqueous phase in appropriate proportion were mixed in a beaker at 13 ⬚C in a water bath to form the microemulsion. Appropriate amounts of microemulsion I containing 0.5 M TiCl4 aqueous solution and microemulsion II containing 2.0 M ammonia were mixed together, leading to the formation of insoluble titania particles. The mixed microemulsions were then poured into acetone to precipitate TiO2 particles. Finally, the precipitates were repeatedly washed by the use of a centrifuge and acetone, followed by vacuum drying for 2 h. Fig. 3(a) is a TEM image of the crude product in which most particles are spherical and some are elongated, and Fig. 3(b) gives size distribution of the particles with the average diameter of 5.4 nm and ␴ ⫽ 1.4 nm. Raman spectrum of this product has

116

Wang (a)

(b)

Fig. 3. TEM image of TiO2 crude product containing nanoparticles (a), and their size distribution (b). (c) Raman spectrum of TiO2 nanoparticles.

shown two very broad bands at about 430 and 610 cm⫺1 [35]. The frequency of these two bands lie very close to that of A1g (612 cm⫺1) and Eg (447 cm⫺1) rutile modes, indicating that both materials contain the same localized structural group but with an absence of long order in the as-prepared product [36]. This result agrees well with that observed by Ocana et al. [37], and proves that the precursor TiO2 particles are amorphous. Fig. 4(a) and (b) show TEM images of the virgin sample but (a) calcined for 2 h at temperature of 500 ⬚C and (b) calcined for 2 h at 800 ⬚C. Fig. 4(c) and (d) give their corresponding Raman spectra, demonstrating that both are crystallized. The nanoclusters have the anatase structure with average size of 10 nm in the first case while the particles grow up to 60 nm in diameter with rutile structure for the second case, but no rod-like morphology is observed in both cases. 2.3. Single crystal TiO2 nanowires In general, the technique to prepare the inverse microemulsions for the synthesis of TiO2 nanowires or nanorods is more or less similar to that mentioned above for synthesis of TiO2 nanoparticles. However, in addition to the microemulsions I and II, the third microemulsion which contains 2.0 mol/L NaCl solution is needed, listed in Table 1 [38].

117

Nanowires of Functional Oxides (a)

(b)

(c)

(d)

Fig. 4. TEM images of the virgin sample but (a) calcined for 2 h at temperature of 500 ⬚C and (b) calcined for 2 h at 800 ⬚C. Fig. 4(c) and (d) are their corresponding Raman spectra. Table 1. Compositions of the microemulsions used in preparing TiO2 nanowires

Aqueous phase (vol %) Surfactant (vol %) Oil phase (vol %)

Microemulsion I (%)

Microemulsion II (%)

Microemulsion III (%)

0.5 mol/1 TiCl4 (2) NP5-NP9 (33) Cyclohexane (65)

2.0 mol/l NH3 (2) NP5-NP9 (33) Cyclohexane (65)

2.0 mol/l NaCl (20) NP5-NP9 (28) Cyclohexane (55)

The procedure for the synthesis of TiO2 nanowires is as follows. Firstly, appropriate amounts of microemulsion I and microemulsion II were mixed together, leading to the formation of insoluble titanium hydroxide particles within the aqueous droplets. Hereafter Ti(OH)4 was used as a convenient abbreviation for the particles. Transmission electron microscopy (TEM) observations revealed that the particle sizes

118

Wang

were about 5 nm. Secondly, this mixed microemulsion was mixed with specified amounts of the microemulsion III. Titanium hydroxide particles and NaCl solution coexist in the aqueous droplets in this system. The resultant microemulsion containing Ti(OH)4 particles and NaCl solution was then poured into sufficient amounts of acetone to precipitate Ti(OH)4 and NaCl particles. NaCl is insoluble in acetone. Therefore, these two kinds of particles will be homogeneously mixed. Finally, the precipitates were repeatedly washed by use of a centrifuge and acetone, followed by vacuum drying for 2 h. Thus a fine white powder containing NaCl and Ti(OH)4 nanoparticles was obtained as the precursor powder for nanowires. These processes were performed at 13 ⬚C in a water bath. The precursor powder was then annealed in a tube furnace for 2 h at 730, 740, 750, 760 or 780 ⬚C, respectively. After annealing, the powder containing TiO2 and the remaining NaCl was washed with distilled water to remove the NaCl. This procedure is given in the flowchart illustrated in Fig. 5. Fig. 6(a) shows the morphology of the products obtained by annealing the powder with ␥ ⫽ 400 at 750 ⬚C for 2 h, where ␥ is the molar ratio of sodium to titanium in the precursor powder. The TEM image indicates that the product mainly consists of straight, solid rod-like structures with diameters ranging from 10 to 50 nm and length typically exceeding 2 ␮m. Variation in contrast seen along the nanorod comes from the orientation variations of the rod on the TEM support. The crystal structure of the nanorods is determined by a set of elected area diffraction patterns (SADPs) obtained from a single nanowire by using a double-tilt sample holder. Fig. 6(b)–(d) show SADPs of a single rutile nanowire. These SADPs can be indexed according to the rutile TiO2 structure. No sodium titanate phase has been observed. The electron beam directions in – – – Fig. 6 (b)–(d) are the [110], [111] and [112] direction, respectively, in which elongation of spots is due to crystal lattice relaxation within the wires [39]. The trace analysis forms three TEM images of a single nanorod with different electron beam directions shows that the rod axis is parallel to the [001] direction. High-resolution TEM images recorded on individual nanowires provide further insight into the structure of this nanomaterial. Fig. 7(a) shows a low-magnification HRTEM image of a single wire viewed – along the [110] direction, indicating that the nanorod is a perfect crystal without dislocations and planar defects such as microtwins, which are frequently found in SiC whisker [15]. The growth tip of the nanowire is terminated and surrounded by planes close to {111} planes and side surface is a (110) plane. It implies that the (110) face is a flat surface of rutile nanorod with square cross section surrounded by four (110) planes (Fig. 7(b)) because the normal direction of the wire flat surface is parallel to the electron beam in this case. This has been further proved by high resolution transmission electron microscope (HRTEM), shown in Fig. 8, indicating that a nearly perfect crystal nanorod with the tip of the rod constructed by (111) planes. This shape of the tip is different from that of the rutile crystal in micrometer scale, and the tip of the thicker whisker is flat [40]. Fig. 9(a) is the powder X-ray diffraction (XRD) spectrum of products indicating that the main phase is rutile TiO2. Other weak peaks come from small amount of sodium titanates in the products. Fig. 9(b) is the XRD spectrum of the film of products deposited on an amorphous silical substrate. Only (110), (200), (210) and (200) peaks of rutile TiO2 appear in the XRD spectrum. Other peaks, including (101) and (211) peaks that are usually remarkably strong, are not observed. This demonstrates that the axes of all the nanorods are along the [001] direction.

Nanowires of Functional Oxides

119

Fig. 5. Flowsheet of procedure for the synthesis of rutile nanorods.

The annealing temperature and the density of Ti(OH)4 particles in the precursor powder influence the formation of rutile nanowires dramatically. In the same ␥ value, for instance, ␥ ⫽ 400, the product mainly consists of small spherical particles at the annealing temperature of 730 ⬚C, and only a few rutile whiskers exist. These nanoparticles are in anatase phase, which is metastable. At higher temperature of 750–760 ⬚C the nanowhiskers with rutile structure are formed. However, the whiskers with thickness of

120

Wang (a)

(b)

(c)

(d)

Fig. 6. (a) TEM image of the product obtained by annealing the precursor powder with ␥ ⫽ 400 at 750 ⬚C for 2 h, (b)–(d) SADPs of a single rutile nanorod taken along the [110], [111] and [112] direction, respectively.

(a)

Fig. 7. (a) Low magnification HRTEM image of a rutile nanowire viewed along the [110] direction. (b) Sketch of a rutile nanorod.

several hundred nanometers are found at 780 ⬚C. The most suitable temperature for the formation of thinner and longer rutile nanowires of uniform size is about 750 ⬚C. Density of Ti(OH)4 in the precursor powder greatly influences the formation of rutile nanowires. Therefore, the molar ratio of sodium to titanium, ␥, which reflects the packing density of Ti(OH)4 particles in the precursor powder, is an important parameter, closely related to the aspect ratio of the nanowires. It was calculated from the amount of chemicals used in the preparation, and can be controlled by adjusting the

Nanowires of Functional Oxides (a)

121

(b)

(c)

20

30

40 50 2θ (degrees)

(220) (211) (220)

(101) (200) (111) (210)

(210)

(110)

(a)

(200)

(b)

(110)

Intensity (a.u.)

Fig. 8. Profile HRTEM images of rutile nanorods viewed along the [010] direction. (a) and (b) show the structure of (110) side surfaces, and (c) shows the structure of the rod tip.

60

Fig. 9. X-ray diffraction (XRD) spectrum of the products: (a) powder sample; (b) the film consisting of rutile nanowhiskers.

relative amount of TiCl4, ammonia, and NaCl in the microemulsions. Table 2 gives the amounts of microemulsions used for preparation of precursor powders and average sizes of the rutile nanowires. It indicates that when  ⫽ 100 almost all the products are the rutile nanowires, but are thicker and shorter. Their average thickness is 75 nm and the 2 ␮m. The nanowires have the average thickness of 20 nm and the length up to 4 ␮m when  ⫽ 1000, however, a considerable amount of precursor particles are

122

Wang

Table 2. ␥ values of microemulsions in preparation of precursors and sizes of corresponding rutile nanorods Microemulsion I 4 ml 2 ml 1 ml

Microemulsion II

Microemulsion III

␥

Average sizes of nanorods (length ⫻ diameter)

4 ml 2 ml 1 ml

12 ml 24 nm 30 nm

100 400 1000

2 ␮m ⫻ 75 nm 2 ␮m ⫻ 30 nm 4 ␮m ⫻ 22 nm

transformed into spherical rutile particles, not rutile nanowires after annealing. Therefore, certain amount of NaCl is one of key points in the formation of rutile nanorods. For VLS [22–24] and SLS [25] approaches, wires growth requires a fluid phase (vapor in VLS and solution in SLS) in which elements of the crystal phase can easily move for a long distance. Another advantage of the fluid phase is that the materials for wire formation are homogeneously distributed Such homogeneity leads to the uniform nanorod sizes. At present, the Ti(OH)4 particles are homogeneously dispersed in NaCl particles without aggregation or agglomeration, and their densities are small in all of the mixed powder. Ti(OH)4 particles are subject to decomposition into TiO2. According to the TiO2–NaCl phase diagram [41], no liquid phase exists at temperatures less than 800 ⬚C. At the annealing temperature the vapor pressure of TiO2 is too small to grow whisker within 2 h. Mass transportation is mainly due to the movement of anatase particles. NaCl may play a role of separating the anatase particles and of promoting their diffusion. It is observed that in the sample with ␥ ⫽ 400 annealed at 750 ⬚C small titania particles with diameter of approximately 4 nm moved slowly form the side surface to the tip of a whisker and finally transformed into a rutile structure at the tip under a sufficiently intensive electron beam. Therefore, a growth process of the rutile whiskers in the powder containing NaCl and Ti(OH)4 particles during annealing is proposed as follows: (1) At 750 ⬚C, the Ti(OH)4 particles are easily decomposed into anatase TiO2 particles: Ti(OH)4 → TiO2 ⫹ 2H2O. (2) Rectangular-shaped rutile particles appear. One side of the rectangle acts as the whisker tip. (3) Anatase particles move to the tip and transform into the rutile structure. During transformation, atoms are very mobile [42], because of bond breakage, and each atom can adopt a correct position in developing the rutile whisker shape. Therefore, it is suggested that the rutile nanorods are formed in a solid system by solidstate phase transformation while the chemical reaction involved the formation process of the rutile whiskers may be simply the decomposition of Ti(OH)4. 2.4. Nanoforks and their interfaces In the above process the products mostly consist of straight, rod-like whiskers. There are also certain amount of TiO2 twinned crystals like nanoforks, shown in Fig. 10. Two kinds of twinned whiskers are formed, and each with two legs constructing

Nanowires of Functional Oxides (a)

(b)

(c)

(d)

123

Fig. 10. (a) TEM image of a TiO2 nanofork with (301) interface, and (b) selected area diffraction pattern of the twinned boundary between the two legs. (c) TEM image of a nanofork and (d) the SADP of its (101) twinned boundary.

an angle of 55⬚ and 114⬚ respectively. The legs have a common [010] direction. However, in Fig. 10(a) the two legs are related by twinning on the (301) plane with each other, and its selected area diffraction pattern (SADP) is taken in Fig. 10(b). From Fig. 10(c) and (d) the two legs of the fork are related by twinning on (101) plane and the axis of each leg is also along the [001] direction. Microstructures of these twinned boundaries have been studied by high resolution electron microscope (HREM). Fig. 11(a) shows low magnification HREM image of (301) twinned boundary when electron beam is parallel to [010] direction. A dislocation can be seen and the dislocation line parallel to the electron beam, i.e., along [010] direction. Fig. 11(b) is the magnification image of the area including the dislocation around the point B, and this dislocation is screw dislocation because Burgers vector is zero in the (010) plane, b ⫽ [010]. Fig. 11(c) is the magnification image of the area at C indicated in Fig. 11(a), and at the twinned interface a periodic structure appears with 6 distances between lattice plane (101), quite different from the interfacial structure in the equilibrium condition [43]. Fig. 12(a) is a HREM image of (101) twinned interface along [010] direction at low magnification. There is a dislocation at the C point. Fig. 12(b) is a magnified image of B point in (a), the atom arrangement is the same as the equilibrium structure. Fig. 12(c) gives a magnified image of the dislocation region in Fig. 12(a). The dislocation is embedded in the interior of the twinned whisker and fringes are parallel to the twinned interface. An extra half atomic plane appears, indicating that the Burger vector of this dislocation includes the component – of 12 [101]. Several other types of rutile TiO2 twinned nanostructures are observed, such as a straight whisker with a branch related by twinning on a (301) plane, a nanowire with three legs, among which the two adjacent legs with an angle of 55⬚ while the middle leg twinned with its two neighbors on different {301} planes [44], etc. In fact

124

Wang (a)

(b)

(c)

Fig. 11. HREM images of (301) twinned boundary: (a) low magnification; (b) amplified image of B region in (a); (c) amplified image of C region in (a).

(b)

(a)

(c)

Fig. 12. HREM images of (101) twinned boundary: (a) low magnification; (b) amplified image; (c) dislocation image at the interface.

Nanowires of Functional Oxides

125

a variety of morphologies of TiO2 twinned nanowires have been found, however, only two kinds of twins, (101) and (301) twins exist, implying that the twin can form on any one of the {101} or {301} planes. Similar situation also occurs in SnO2, which has an isomorphous crystal structure with rutile form TiO2 [45].

3. Layered Structures of Potassium Hexatitanate Nanowhisker Potassium hexatitanate (K2Ti6O13) has high chemical and thermal stability among the family of K2TixO2x ⫹ 1 compounds. It has potential applications in whisker-reinforced plastics and metal [46, 47]. The reagent grade KF and anatase form TiO2 were used as starting materials and their weight ratio was 3 : 1. After being ground in an agate mortar under acetone, the mixed powder was heated at 720 ⬚C in a tube furnace for 5 h, and then was washed with distilled water to remove the remaining KF followed by treating in boiling water for 4 h. Finally it was dried and reheated at 800 ⬚C for one hour. Fig. 13 is the TEM image of the products. The diameters of whiskers range from 20 nm to 200 nm and the average diameter is 70 nm. X-ray diffraction (XRD)

(a)

Number

(b)

22 20 18 16 14 12 10 8 6 4 2 0

20

40

60

80 100 120 140 160 180 200 Diameter (nm)

Fig. 13. (a) TEM image of K2Ti6O13 whiskers and (b) size distribution of whisker diameters.

126

Wang

Fig. 14. XRD profile of K2Ti6O13 whiskers.

(a)

(b)

(c)

(d)

Fig. 15. (a) TEM image of a K2Ti6O13 whisker composed of layers A and B, (b) SAD pattern from the area containing both A and B, (c) schematic diagram of SAD pattern of the layer A – along the [310]A zone axis, (d) schematic diagram of SAD pattern of the layer B along the – [114]B zone axis.

pattern shows all the whiskers are K2Ti6O13 with monoclinic structure. (Fig. 14). The thicker whiskers can be composed of several layers, and Fig. 15(a) shows a whisker composed of only two adjacent layers indicated by “A” and “B” respectively. Fig. 15(b) is the SAD spectrum of the area containing both layer A and B, and from which the orientation relationship between two adjacent layers in a whisker is exposed. Fig. 15(c) and (d) are the schematic diagrams of SAD patterns of layers A and B – respectively. In Fig. 15(c), the electron beam is along the [310]A direction, while it is – along the [114]B direction in Fig. 15(d). From Fig. 15(d) it can be seen that the plane

Nanowires of Functional Oxides

127

–

(001)A is parallel to (401). Therefore, the orientation relationship between layer A and – – – B should be: (001)A//(401)B; [310]A//[114]B. From the relation, it can be derived that [010]A//[010]B. This is consistent with the fact that the axis of an individual whisker is along the [010] direction. This kind of layered structure has also been observed in Mn2O3 nanowires by atomic resolution, showing that superlattices are formed with each superlattice consisting of about 5 atomic lattices, and the grain boundary of two lattices.

4. Rutile Stannic Oxide Nanorods Similar technique is employed to prepare microemulsions for SnO2 nanorods [48]. Fig. 16(a) shows the morphology of the starting powders used in the study of molten

(a)

(b)

(c)

Fig. 16. TEM images of SnO2 nanorods: (a) The precursor prepared in I microemulsion system, (b) Rods after annealing at 780 ⬚C for 2.5 h, (c) The rods after annealing at 870 ⬚C for 2.5 h.

128

Wang

salt SnO2 synthesis. These are the nanosized, but aggregated particles, and the average grain size is about 10 nm. Fig. 16(b) shows the microstructure of the powders after calcinations at 780 ⬚C for 2.5 h. The nanorods (referred to as sample I) are straight with uniform diameters range from 20–40 nm and length up to 5 ␮m, after the precursor powders was annealed. However, after annealing at 820 ⬚C for 2.5 h shown in Fig. 16(c), the SnO2 crystallites synthesized (referred as sample II) are uniform nanorods with diameters of 70–90 nm and 5–10 ␮m. The aspect ratio is up to 250. Fig. 17 shows an image of one of these single crystalline nanorods with 40 nm width and 10 ␮m, and its selected area electron diffraction(SAED) pattern, indicating that the nanorods are single crystalline with the rutile SnO2 structure. Fig. 18(a) and (b) are the powder diffraction (XRD) patterns of SnO2 sample synthesized at 780 ⬚C and 820 ⬚C for 2.5 h respectively. All peaks are indexed as tetragonal lattice of SnO2 (rutile structure). However, at the lower temperature, the diffraction peaks are broadened due to the small particle size. Fig. 19 shows the Raman spectrum of the sample

(a)

(b)

Fig. 17. (a) TEM image and (b) selected area electron diffraction (SAED) pattern of one single SnO2 nanorod.

Fig. 18. XRD patterns of SnO2 nanorods (a) after annealing at 780 ⬚C for 2.5 h, (b) after annealing at 820 ⬚C for 2.5 h.

Nanowires of Functional Oxides

129

Fig. 19. Raman spectra of SnO2 nanorods (after annealing at 780 ⬚C for 2.5 h).

I. The peaks are different from those in bulk SnO2 and nanocrystallites powder with an exception of the Eg and A1g modes [49–51]. The Eg mode is broadened, however. Such phenomena are in good agreement with the XRD results. The Raman lines at 470.63 and 628.95 cm⫺1 can be assigned to the Eg and A1g modes, respectively, of the anatase phase, whereas the Raman bands measured at 216.49 and 232.92 cm⫺1 were not detected in the nanopowders or bulk SnO2 [49, 51]. Abello et al. proposed that the relaxation of the k ⫽ 0 selection rule is progressive when the rate of disorder increases or the size decreases and infrared (IR) modes can become weakly active when the structural changes induced by disorder and size effects take place [49]. Therefore, the weak Raman bands at 216.49 and 232.92 cm⫺1 seem to correspond to IR-active Eu(2)TO and Eu(2)LO (TO is the mode of the transverse optical phonons, LO is the mode of the longitudinal optical phnonons) [50]. It is reasonable to assign the two volume modes to IR modes whose Raman activities are induced by the size effect, which are due to the smaller diameter of SnO2 nanorods. These volume modes of IR modes in GaN [52] epitaxial layers of high quality and GaN nanowires [53] were observed in the Raman spectra. Nagano [54] reported that they had grown SnO2 whiskers by VLS at 1100 ⬚C or 1300 ⬚C under nitrogen gas. Here SnO2 nanorods are synthesized by calcining the precursors, obtained by the inverse microemulsion, at relatively low temperature in air. However, other surfactants, such as a poly(oxyethylene)-5-nonyl phenolether/ poly(oxyethylene)-9-nonyl phenolether (NP5/NP9) mixture (1:1), instead of a NP5/NP9/OP (where OP is p-octyl-polyethylene glycol phenylether) mixture (1:1:1), were used to prepare the precursors. In this case, only SnO2 nanoparticles were formed with an average grain size of 80 nm, but no nanorods were formed after annealing. The reason for the preparation of the precursors with the surfactant is that water/oil (W/O) microemulsions consist of nanosized water droplets dispersed in a continuous oil medium and stabilized by surfactant molecules accumulated at the oil/water interface. The highly dispersed water pools are the ideal nanostructural reactor [55] for producing monodispersed nanoparticles, and that makes the precursors

130

Wang

easy to decompose thus forming SnO2 nanorods. If the quantity of SnCl4 in the microemulsion system is increased to 4 ml of a 0.5 M SnCl4 aqueous solution under the same experimental conditions, metallic tin was observed besides SnO2 nanorods. Therefore, the composition of the microemulsion and the quantity of SnCl4 are the key to the synthesis of SnO2 nanorods. Concerning the formation mechanism of SnO2 nanorods from molten salt synthesis after preparation of the microemulsion, the experiments demonstrate that the molten salt synthesis process is very fast reaction and the annealing time is related to the diameter and length of the nanorods when the annealing temperature is fixed. The reaction is completed in a very short time due to the short diffusion distance and the high mobility of species in the liquid state. Moreover, the melting point of the salt was lower than the firing temperature of the precursors (the melting point decreased due to nanosized NaCl). At this temperature the oxides are rearranged and then diffuse rapidly in a liquid state of the salt, and heating SnO2 nanorods are formed through nucleation and growth processes. However, the SnO2 nanorods with larger diameter and longer length growing out of the molten salt bath is strongly dependent on soaking time at the peak temperature.

5. Cu2O Nanowires Cu2O is a p-type semiconductor with a direct bandgap of 2 eV, which makes it a promising material for conversion of optical, electrical and chemical energy [56–58]. It has been reported that excitons can propagate coherently through single crystalline Cu2O. In order to obtain nanowires or nanorods of such materials various kinds of methods have been developed but most of them require high temperatures, special conditions, or tedious procedures [1, 2, 15, 59–62]. A novel reduction route has been conducted for preparing Cu2O nanowires in the presence of a suitable surfactant, polyethylene glycol (PEG; Mw ⫽ 20000), at room temperature by the following reactions [63] CuCl2 ⫹ 2NaOH → Cu(OH)2 ⫹ 2NaCl 4Cu(OH)2 ⫹ N2H4 → 2Cu2O ⫹ 6H2O ⫹ N2 where hydrazine hydrate is the reducing agent, N2H4 in the basic aqueous solution is a strong reducing agent [64], N2 ⫹ 4H2O ⫹ 4e⫺ → N2H4 ⫹ 4OH⫺ Fig. 19 shows an X-ray diffraction pattern of the Cu2O nanowires which contains five peaks perfectly indexed to crystalline Cu2O both in position and their relative intensity. Fig. 21 gives TEM images of Cu2O nanowires prepared by the reduction route in the presence of PEG at room temperature. Bulk quantities of nanowires were formed with relatively uniform diameters of about 8 nm and the lengths ranging from 10 to 20 ␮m. Fig. 21(b) and (c) are the high resolution electron microscopy (HRTEM)

Nanowires of Functional Oxides

131

Fig. 20. The X-ray diffraction pattern of a Cu2O nanowire prepared by a reduction route in the presence of PEG at room temperature.

(a)

(b)

(c)

Fig. 21. (a) TEM images of the Cu2O nanowires, (b) and (c) HRTEM images of individual Cu2O nanowires.

132

Wang

images of a nanowire with a diameter of 8 nm. The visible lattice fringes illustrate that the nanowire is crystalline rather than a single crystal within the lateral dimension. The interplanar spacing is about 0.2465 nm, which corresponds to the {111} plane of cubic Cu2O, indicating that the growth plane of the nanowires is one of the {111} planes. The two-dimensional lattice image pointed by the arrowheads in Fig. 21(c) also demonstrates that the Cu2O nanowieres are crystalline. Fig. 22(a) shows the photoelectron spectrum of Cu2pI. The peak at 932.50 eV, which was corrected with reference to C1s (284.6 eV), corresponding to the binding energy of Cu2p3/2, is in good agreement with data observed for Cu2O [65]. Fig. 22(b) presents the O1s core-level spectrum which is broad, and the two peaks (marked by a and b) are resolved by using a curve-fitting procedure. Peak a at the lower energy of 530.3 eV corresponds to O2⫺ in Cu2O [66]. Peak b at the higher energy of 531.7 eV is

(a)

(b)

Fig. 22. (a) X-ray photoelectron spectrum of Cu2pI of the Cu2O nanowire sample, (b) X-ray photoelectron spectra of O1s of Cu2O nanowire sample.

Nanowires of Functional Oxides

133

attributed to O adsorbed on the surface of Cu2O nanowires. This implies that much oxygen is adsorbed onto the surface of Cu2O nanowires because Cu2O is unstable in air and easily oxidized [66], which in turn proves that the sample is composed of Cu2O.

6. V2O5 Nanofibres Vanadium pentoxide (V2O5) fibers were synthesized from 0.2 g ammonium(meta)vanadate (Aldrich) and 2 g acidic ion exchange resin (DOWEX 50WX8-100, Aldrich) in 40 ml water. The formation of an orange sol is observed after a few hours. V2O5 fibers with lengths of a few micrometers were obtained after 3 days and aged for at 4 weeks were used the electrical measurements [69]. In order to study electric property, electrode arrays were needed and created on silicon wafers with a 1 ␮m thick thermally grown oxide layer. Gold electrodes spaced by 3 ␮m were fabricated by conventional optical lithography. AuPd (40/60) lines with a separation of about 100 nm were defined by 3-beam lithography using a two-layer resist and a modified Hitachi S2300 scanning electron microscope. V2O5 network sample was prepared by depositing a droplet of undiluted V2O5 sol on the chemically modified Si/SiO2 substrate. Fig. 23 gives a scanning force microscopy (SFM) image of the structure investigated [70]. The substrate was heavily coped by As⫹ ion implantation and served as a back gate, which was separated from the electrodes by thermally grown SiO2 layer,

Fig. 23. Scanning force microscopy image of V2O5 nanofibers below Au/Pd electrodes, separated by ~100 nm. Current–voltage characteristics were measured between the middle two electrodes, which are connected by seven V2O5 fibers.

134

Wang (a)

(b)

Fig. 24. (a) Current-voltage characteristics obtained from the middle two electrodes (Fig. 23) with zero gate voltage (VGS ⫽ 0 V) at different temperatures T ⫽ 131, 145, 160, 192, 245 K. (b) Three dimensional plot of I–V curves in dependence of gate voltage at T ⫽ 145 K.

and the two-probe resistance of the V2O5 fibers ranged between 200 and 300 M⍀. There are seven fibers between the middle two electrodes. Fig. 24 plots the currentvoltage characteristics obtained from this electrode pair at different temperatures for zero gate voltage (VG ⫽ 0). As temperature decreases, a nonohmic behavior is observed. Upon sweeping the gate voltage from ⫺20 to ⫹20 V at T ⫽ 145 K, the drain-source current (IDS) increase as shown in Fig. 24(b). The conductance increase at positive gate voltage may be attributed to the accumulation of electrons. For higher temperatures, the conductance increases and the nonohmic behavior disappears with rather small gate dependence. At low temperature, the conductance increases as the gate voltage is changed from negative to positive values, characteristic of a Fieldeffect transistor (FET) with n-type enhancement mode. The carrier mobility, estimated from the low-field regime, is found to increase from 7.7 ⫻ 10⫺5 cm2/Vs at T ⫽ 131 K to 9.6 ⫻ 10⫺3 cm2/Vs at T ⫽ 192 K with an activation energy of Ea ⫽ 0.18 eV [71]. The nonohmic current/voltage dependence at high electric fields can analyzed in the frame of small polaron hopping conduction, yielding a nearest-neighbor hopping distance of ~4 nm [70].

7. Potential Applications Metal Oxide nanowires as functional nanomaterials and nanodevices have novel properties due to size effect and surface effect and have potential applications in various kinds of fields. Following are some examples but not mature. Many problems remain to be solved and need to be further investigated. Gas sensors Water vapors absorbed on TiO2 surface are dissolved into H⫹ and ⫺ OH . OH⫺ ion is bonded with surface ions and can not move, while positive ions easily move on the surface, thus affecting conductivity of TiO2. Nanoelectrode and nanowaveguides Cu/Cu2O layered nanostructured materials with interesting optoelelctonic properties have been prepared by electrode position. Recently, Cu2O submicronspheres can be used as the negative electrode materials for lithium ion batteries [58]. It has been reported that excitons can propagate coherently through

Nanowires of Functional Oxides

135

single crystalline Cu2O. Therefore, photons can be converted into excitions, which then travel through small apertures or small dimension waveguides with little loss by scattering or diffraction, and the excitons can be converted back into photons at the end of the path [67]. Thus it is possible to use the Cu2O nanowires as nanowaveguides? Solar cell materials Stannic oxide is one of the optical conduction materials and has been extended to solar cells [68]. SnO2 nanorods show better properties compared with the bulk. Catalysts TiO2 is an important carrier of catalysts as a fine catalyst. TiO2-based catalysts consist of other compositions such as V2O5, Fe2O3, MoO3, WO3, NiO, Ru, Pt, SiO2, etc., which can be used as the most favorable catalyst for pollution controlling through reduction reaction of NOx—NH3 and decreasing NOx in the smoke. Nanodetectors and nanotips TiO2 nanorods with well-defined structures both its tip and the surfaces can be used as molecular detector and tips for nanoprobe. Nanofiber-based field-effect transistors As mentioned previously, n-type enhancement FET-like characters has been demonstrated for individual V2O5 nanofibers at low temperatures [70]. The mobility was found to be almost independent of gate voltage and to be activated by temperature. However, compared with the progress in mainstream semiconductor technology, one-dimensional wire materials have so far lacked the powerful virtues of heterostructures. By applying heterostructure technology to nanowire growth, new families of devices utilizing wire materials would emerge. Of particular interest is the fact that the very small cross section allows efficient lateral relaxation of a nanowire, thereby providing the freedom to combine materials with very different lattice constants to create heterostructures within a nanowire [72].

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

S. Ijima, Nature 354 (1991) 54. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. L. D. Zhang, G. W. Meng and F. Phillipp, Mater. Sci. Eng. A 286 (2000) 34. G. S. Chen, S. H. Chen, X. G. Zhu, Y. O. Mao and L. D. Zhang, Mater. Sci. Eng. A 286 (2000) 165. D. W. Wang, A. J. Millis and S. D. Sarma, Phys. Rev. Lett. 85 (2000) 85. M. Q. He. K. Minus, P. Z. Zhou and N. Mohammed, Appl. Phys. Lett. 77 (2000) 3731. T. Tenne, L. Margulis, M. Genet and G. Hodes, Nature 360 (1992) 444. L. Marguis, R. Tenne and S. Ijima, Microsc. Microanal. Microstruct. 7 (1996) 87. D. H. Galvan, R. Rangel and G. Alonso, Fullerene Sci. Technl. 6 (1998) 1025. Y. Feldman, E. Wasserman, D. J. Srolovitz and R. Tenne, Science 267 (1995) 222; L. Margulis, G. Salitra, R. Tenne and M. Taliander, Nature 365 (1993) 113. M. Terrones, W. K. Hsu, H. Terrones, A. K. Cheetham, H. W. Kroto and D. R. M. Walton, Chem. Phys. Lett 259 (1996) 568. N. G. Chopra, R. J. Luyken, K. Cherry, V. H. Crespi, M. L. Cohen, S. G. Louie and A. Zettl, Science 269 (1995) 966. O. Stephan, P. Bernier and P. Lefin, Science 266 (1994) 1683. Z. Sieh, K. Weng, N. G. Cherrey, X. Chopra, Y. Blasé, A. Miyampto, M. L. Rubio, S. G. Cohen, A. Louie and R. Zettl. Phys. Rev. B 51 (1995) 11299. H. Dai, E. W. Wong, Y. Z. Liu, S. S. Fan and C. M. Lieber, Nature 374 (1995) 769.

136

Wang

16. P. D. Yang and C. M. Lieber, Science 273 (1996) 1836. 17. S. Nakamura, Science 281 (1998) 956. 18. C. Liu, Y. Y. Fan, M. Liu, H. T. Cong, H. M. Cheng and M. S. Dresselhaus, Science 286 (1999) 1127. 19. X. L. Chen, J. K. Liang, Y. P. Xu, T. Xu, P. Z. Jiang, Y. D. Yu and K. Q. Lu, Mod. Phys. Lett. B 13 (1999) 286. 20. G. J. Li, X. H. Zhang and S. Kawi, Sens. Actuators B 60 (1999) 64. 21. G. Sherveglieri, Sens. Actuators B 6 (1992) 64. 22. R. S. Wagner and W. C. Ellis, Appl. Phys. Lett. 4 (1964) 89. 23. E. I. Givargizov, J. Cryst. Growth 32 (1975) 20. 24. R. S. Wanger, in Whisker Technology, Edited by A. P. Levitt, Wiley, New York (1970) p.47. 25. T. C. Yeh, K. M. Hickman, S. C. Goel, A. M. Viano, R. C. Gibbons and W. E. Buhro, Science 270 (1995) 1791. 26. S. A. Sapp. B. B. Lakshmi and C. R. Martin, Adv. Mater. 11 (1999) 402. 27. K. H. Yoon, Y. S. Cho and D. H. Kang, J. Mater. Sci. 33 (1998) 2977. 28. H. Wada, K. Sakane and T. Kitamura, J. Mater. Sci. Lett. 10 (1991) 1076. 29. S. Hashimoto and A. Yamagushi, J. Eur. Ceram. Soc. 20 (2000) 397. 30. M. Kajiwara, J. Mater. Sci. 22 (1987) 1223. 31. A. C. Tas, J. Am. Ceram. Soc. 84 (2001) 295. 32. B. A. Morales, O. Novaro, T. Lopez, S. Sanches and R. Gomez, J. Mater. Res 10 (1995) 2788. 33. R. W. Siegel, S. Ramasamy, H. Hahn, Z. Li, T. Lu and R. Gronsky, J. Mater. Res. 3 (1988) 1367. 34. R. Birringer, H. Gleiter, H. P. Klein and P. Marguarit, Phys. Lett. A 102 (1984) 365. 35. G. L. Li and G. H. Wang, Nanostruct. Materials 11 (1999) 663. 36. G. J. Exarhos, Mater. Res. Soc. Symp. Proc. 48 (1985) 873. 37. M. Ocana, J. V. Carcia-Ramos and C. J. Serna, J. Am. Ceram. Soc. 75 (1992) 2010. 38. G. L. Li, G. H. Wang and J. M. Hong, J. Mater. Res. 14(8) (1999) 3346.; Pentent (No: ZL 98 1 11329.X, 1/9/2002) 39. J. Harada, M. Takata, H. Miyatake and H. Koyama, J. Appl. Cryst. 22 (1989) 592. 40. T. Oota and I. Yamai, J. Cryst. Growth 66 (1984) 262. 41. E. M. Levin and H. F. McMuride, “Phase Diagrams for Ceramist” in American Ceramic Society, Edited by K. R. Margle Columbus, 1969, Suppl. p. 473. 42. K. N. P. Kumar, K. Keizer, A. J. Burggraaf, T. Okub, H. Nagamoto and S. Morooka, Nature 358 (1992) 48. 43. Y. Gao, K. L. Merkle, H. L. Chang, T. J. Zhang and D. J. Lam, Philos. Mag. A 65 (1992) 1103. 44. G. L. Li, G. H. Wang and J. M. Hong, J. Mater. Sci. Lett. 18 (1999) 1243. 45. H. Iwanaga, M. Egashira. K. Suzuki, M. Ichihara and S. Tacheuchi, Philos. Mag. A 58 (1988) 683. 46. Q. J. Xue, Z. Z. Zhang, W. M. Liu and W. C. Shen, J. Appl. Polymer Sci. 69 (1988) 1393. 47. J. H. Li, X. G. Xing, H. Q. Ye, J. Pan and H. Fukunaga, J. Mater. Sci. 32 (1997) 543. 48. Y. K. Liu, C. L. Zhang, W. Z. Wang, C. R. Ying and G. H. Wang, Adv. Mater. 13(24) (2001) 1883. 49. L. Aello, B. Bochu, A. Gaskov, S. Koudryaviseva, G. Lucazeau and M. Roumyantseva, J. Solid State Chem. 135 (1998) 78. 50. R. Sato and T. Asari, J. Phys Soc. Jpn. 64 (1995) 1193. 51. R. S. Katiyar, P. Dawson, M. M. Hargreave and G. R. Wilkinson, J. Phys. C: Solid State Phys. 4 (1995) 2421. 52. T. Azuhata, T. Sota, K. Suzuki and S. Nakamura, J. Phys. Condens. Matter 7 (1995) L129. 53. G. S. Cheng, S. H. Chen, X. G. Zhu, Y. Q. Mao and K. D. Zhang, Mater. Sci. Eng. A 286 (2000) 165.

Nanowires of Functional Oxides

137

54. M. Nagano, J. Cryst. Growth 66 (1984) 377. 55. G. Onori, A. Santucci, J. Phys. Chem. 57 (1999) 5430. 56. J. A. Switzer, B. M. Manue, W. R. Raub and E. W. Bohannan, J. Phys. Chem B 103 (1999) 395. 57. E. W. Bohannan, L. Y. Huang, F. S. Miller, M. G. Shumsky and J. A. Switzer, Langmuir 15 (1999) 813. 58. P. Poizot, S. Laruelle, S. Grugeon, I. Dupont and J. M. Taracon, Nature 407 (2000) 496. 59. W. Han, S. Fan, Q. Li and Y. Hu, Science 277 (1997) 1287. 60. M. Terrones, N. Grobert, J. Olivares, J. P. Zhang, H. Terrones, K. Kordaton, W. K. Hsu, J. P. Hare, P. D. Townsend, K. Prassides, A. K. Cheetham, H. W. Kroto and D. R. M. Walton, Nature 388 (1997) 52. 61. Z. F. Ren, Z. P. Huang, J. W. Xu, J. H. Wang, P. Bush, M. P. Siegal and P. N. Provencis, Science, 282 (1998) 1105. 62. A. M. Morales and C. M. Lieber, Science 279 (1998) 208. 63. W. Z. Wang, G. H. Wang, X. S. Wang, Y. J. Zhan, Y. K. Liu and C. L. Zheng, Advanced Mater. 14 (1) (2002) 67. 64. Y. D. Li, Y. Ding and Z. Y. Wang, Adv. Mater. 11 (1999) 847. 65. C. D. Wagner, W. M. Riggs, L. E. Davis, J. E. Moulder and G. E. Muilenber, Handbook of X-ray Photoelectron Spectroscopy, Perkin Elmer Corporation Physical Electronics Division, USA (1979). 66. A. O. Musa, T. Akomolafe and M. J. Carter, Sol. Energy Mater. Sol. Cells 51 (1998) 305. 67. D. Snoke, Science 273 (1996) 1351. 68. C. M. Lampert, Sol. Energy Mater. 6 (1991) 1. 69. J. Muster, G. T. Kim, V. Krstie, J. G. Park, Y. W. Park, S. Roth and M. Burghard, Adv. Mater. 12 (2000) 420. 70. G. T. Kim, J. Muster, V. Krstic, J. G. Park, Y. W. Park, S. Roth and M. Burghard, Appl. Phys. Lett. 76 (14) (2000) 1874. 71. K. Honma, M. Yoshinaka, K. Hirota, O. Yamaguchi, J. Asai and Y. Makiyama, Mater. Res. Bull. 31 (1996) 531. 72. M. T. Bjoerk, B. J. Ohisson, T. Sass, A. I. Persson, C. Thelander, M. H. Magnusson, K. Deppert, L. R. Wallenberg and L. Samuelson, Appl. Phys. Lett. 80 (6) (2002) 1058.

Chapter 8 Controlled Growth and Optical Properties of Zinc Oxide Nanostructures Yue Zhang and Ying Dai Department of Materials Physics, University of Science and Technology Beijing, Beijing 100083, China

1. Introduction An important issue in the study and application of nanomaterials is how to control the morphology of nanostructures in an effective and controllable way. One-dimensional (1D) nanostructures, such as nanotubes, nanowires and nanoribbon, have attracted extraordinary attention for their potential applications in device and interconnect integration in nanoelectronics and molecular electronics [1–5]. Over the past several years, various methods have been developed for the synthesis of one-dimensional nanomaterials including template-assisted [6], vapor–liquid–solid (VLS)-assisted [2], colloidal micellar [7], and electrochemical processes [8]. Since the novel properties of nanomaterials depend on their shape and size, the development of a new direction for synthetic methods and an understanding of the mechanism by which the shape and size of nanostructures can be easily controlled is a key issue in nanoscience. ZnO exhibits a direct bandgap of 3.37 eV at room temperature with a large exciton binding energy of 60 meV. The strong exciton binding energy, which is much larger than that of GaN (25 meV) and the thermal energy at room temperature (26 meV), can ensure an efficient exciton emission at room temperature under low excitation energy [9–10]. As a consequence, ZnO is recognized as a promising photonic material in the blue-UV region. Room temperature UV lasing properties have recently been demonstrated from ZnO epitaxial films, microcrystalline thin films, and nanoclusters [11–14]. The synthesis of one-dimensional single-crystalline ZnO nanostructures has been of growing interest owing to their promising application in nanoscale optoelectronic devices. Single-crystalline ZnO nanowires have been synthesized successfully in several groups [15–27]. Wang et al. reported the synthesis of oxide nanobelts by simply evaporating the commercial metal oxide powders at high temperatures [3, 27, 28]. The as-synthesized oxide nanobelts are pure, structurally uniform, and single crystalline, and most of them are of free from defects and dislocations. Room

140

Zhang and Dai

temperature UV lasing in ZnO nanowires has been demonstrated very recently [29]. Although different levels of growth controls for ZnO nanowires (including positional, orientational, diameter, and density control) have been achieved [30], the shape control of nanostructures is not easily obtained. Until very recently, the synthesis of complex structures of rod-based CdSe nanocrystals e.g., arrow, teardrop, and tetrapod, were formed during thermal decomposition of two precursors in a mixture of binary surfactants [31–33]. Due to the electrical and optical properties of nanomaterials depend sensitively on both shape and size, it is important to control the shape and size of nanomaterials in a controllable way and study their optical properties. In this chapter, we report a simple method to synthesize some kind of ZnO nanostructures with various shape and size by oxidation of Zn powders. Structures, growth mechanism and the optical properties for ZnO are investigated.

2. Experimental The materials discussed here were synthesized by thermal evaporation of 99.9% pure zinc powders under controlled conditions without the presence of catalyst. The zinc powders were placed in an alumina crucible or an alumina tube that was inserted in a horizontal tube furnace, where the temperature, pressure, and evaporation time were controlled. The temperature of the furnace was ramped to 850–950 ⬚C at a rate of 50–100 ⬚C/min and kept at that temperature for 1–30 min. During evaporation, various ZnO nanostructures were obtained in different reaction vessel. T-ZnO nanorods were synthesized in the alumina crucible. When instead of an alumina crucible using an alumina tube and silicon substrate was placed downstream inside the tube, ZnO nanowires or arrays were deposited onto the substrate. Structural characterization of the ZnO nanostructures were investigated by X-ray diffraction (XRD) (D/MAX-RB) with Cu Ka radiation, field-emission scanning electron microscopy (FE-SEM) (LEO1530), transmission electron microscopy (TEM) (HP-800) and High-resolution TEM (HRTEM) (JEM-2010F). The Photoluminescence (PL) measurements were carried out on a HITACHI 4500-type visible-ultraviolet spectrophotometer with a Xe lamp as the excitation light source at room temperature. The excited wavelength was at 310 nm for ZnO nanowires or ZnO arrays, 260 nm for T-ZnO nanorods. A 430 nm filter was used.

3. Nanostructures of ZnO 3.1. T-ZnO nanorods Tetrapod-like ZnO (T-ZnO) particle is a type of whiskers with unique structure. Although large sizes of T-ZnO whiskers were previously reported [34–42], the synthesis of T-ZnO whiskers in nanoscale is under developed. Until recently, T-ZnO nanorods are successfully synthesized in an alumina crucible in high yield by us [43]. A typical XRD pattern is shown in Fig. 1. The diffraction peaks can be indexed to a wurtzite structure of ZnO with cell constants of a ⫽ 0.324 nm and c ⫽0.519 nm. No diffraction peaks from Zn or other impurities are found in any of our samples.

141

20

40 60 2θ (degs.)

(203)

(110) (103) (200)(112) (201) (004) (202) (104)

(102)

(002)

Intensity (a.u.)

(100)

(101)

Zinc Oxide Nanostructures

80

Fig. 1. XRD pattern obtained from a bulk sample of T-ZnO nanorods.

Fig. 2. EDX pattern of T-ZnO nanorods.

The energy-dispersive X-ray (EDX) spectrum obtained from the nanorods, as shown in Fig. 2, also confirms that the products are of a pure ZnO phase. Fig. 3 is the typical FE-SEM images of T-ZnO nanorods synthesized. The particles are of a tetrapod shape having four legs. The low magnification image shows that uniform T-ZnO nanorods are formed with high yield. The high magnification image indicates that the surface of nanorods is smooth. The length of legs of T-ZnO nanorods is 2–3 ␮m and the edge size of centering nucleus is 70–150 nm. Very little secondary growth components are observed. The detailed structure of individual T-ZnO nanorods is characterized by TEM. Fig. 4a is the bright-field image of a T-ZnO nanorod (two of the legs overlapped). It is obvious that there are no streaking in the nanorod, which indicates a low density of structural defects, such as stacking faults and dislocations. Fig. 4b is the dark-field image of the nanorod. Single crystal nature of the nanorods is observed from the selected-area electron diffraction (SAED) as shown in the inset. The legs have a

142

Zhang and Dai (a)

(b)

Fig. 3. FE-SEM images of T- ZnO nanorods: (a) at low magnification; (b) at high magnification.

wurtzite structure with the c-axis in the lengthwise direction. Therefore, the T-ZnO nanorods are of high-quality nanocrystals. The size control of the T-ZnO nanorod is achieved by adjusting the processing condition. T-ZnO whiskers with various sizes are obtained (Fig. 5). 3.2. ZnO nanowires The ZnO nanowires synthesized in other groups were single crystalline. Although lengthwise twin boundaries have been observed in both silicon and germanium nanowires [44], the production of bicrystals with a single twin along the growth axis was fewer. Until very recently, bicrystalline silicon nanowires containing a single twin boundary along the entire length of the growth axis were obtained [45]. We demonstrate the synthesis of bicrystalline nanowires of ZnO for the first time [46].

Zinc Oxide Nanostructures

143

(a)

(b)

Fig. 4. TEM images of T-ZnO nanorods: (a) bright-field electron micrograph; (b) dark-field electron micrograph.

Instead of an alumina crucible using an alumina tube and adjusting lower partial pressure of oxygen in the reaction vessel, ZnO nanowires form on the silicon substrate. XRD pattern of nanowires is the same as the T-ZnO nanorods (Fig. 1). In contrast to the XRD spectra of the ZnO nanowires grown by VLS mechanism [16, 17] wherein the diffraction peaks of the catalyst or the zinc metal appeared with the ZnO peaks, the pure ZnO phase is synthesized on the substrates in our case. Fig. 6 shows the images of ZnO nanowires with a diameter of about 60–250 nm and a length of a few millimeters. The further structural characterization of the ZnO nanowires is performed by TEM. A low magnification image containing nanowires is shown in Fig. 7a. It is worthy to note that most ZnO nanowires in the sample appear in the form of bicrystallines (as marked by the arrows). Fig. 7b presents an enlarged view of a portion of the nanowires, in which the crystal has been slightly tilted to provide contrast between the twin variants. The twin boundary in the bicrystalline is proved by both high-resolution TEM image and electron diffraction pattern, as shown in Fig. 8. The corresponding parallel beam –– diffraction pattern along [2110] is shown in Fig. 8b. The pattern is characteristic of – a wurtzite structure that is twinned on (0114) planes, and consists of the patterns

144

Zhang and Dai (a)

(b)

Fig. 5. SEM images of T- ZnO whiskers with various sizes: (a) at 950 ⬚C; (b) at 910 ⬚C.

corresponding to the crystals on either side of the twin (as indexed). The growth axis of – the nanowires is along [0111]. Lengthwise twin boundary has been observed in the nanowire and the twin boundary paralleled to the long growth axis. Therefore, ZnO nanowires are bicrystallines containing a single twin boundary along the entire length. On the other hand, there is a small quantity of single crystalline in the sample. The direction of growth axis for the nanowire is along [0001] (Fig. 9a). Fig. 9b shows a HRTEM image of the nanowire. It reveals that the ZnO nanowire is structurally uniform single crystalline without defect and dislocation. The surfaces of the nanowires are clean, atomically sharp, and without any sheathed amorphous phase. The spacing of 0.26⫾ 0.005 nm between adjacent lattice planes corresponds to the distance between two (0002) crystal plans, which further proves [0001] is the preferred growth direction for ZnO single crystalline nanowires. This result is the same as that in previous reports [3, 29]. Another type of wire-like structure obtained at higher temperature is ZnO nanoribbon, as shown in Fig. 10a. The nanoribbons form on a silicon substrate. Fig. 10b is a

Zinc Oxide Nanostructures

145

(a)

(b)

Fig. 6. SEM images of ZnO nanowires grown on the surface of a silicon substrate: (a) at low magnification; (b) at high magnification.

SEM image of the nanobelts at high magnification, in which geometrical configuration of the nanoribbons is clearly displayed. The nanoribbons are very thin ribbons and have a typical width-to-thickness ratio of ~10 to 50. The length of ZnO nanoribbons is about several hundred micrometers. 3.3. ZnO arrays When the processing condition is controlled in a suitable partial pressure of oxygen and Zn, the ZnO arrays grow on a silicon substrate. As shown in Fig. 11, a high density of well-aligned rods with a diameter in 800 nm form uniformly over the entire substrate. Hexagon are observed at the ends of ZnO rods. The diameter of ZnO rods is somewhat bigger than the typical diameters of 50–100 nm for those prepared by other deposition methods [19, 20]. Furthermore, the ZnO crystals are well aligned vertically and show uniformity in their diameters, lengths, and densities.

146

Zhang and Dai (a)

(b)

Fig. 7. Bright-field TEM images of bicrystalline ZnO nanowires: (a) low-magnification image; (b) a portion of a nanowire at higher magnification. (b)

(a)

Fig. 8. Crystallography of the central twin observed in a bicrystalline nanowire: (a) highresolution TEM image of the twin boundary at the center of the nanowire; (b) indexed parallelbeam diffraction pattern from this bicrystalline (singly twinned) nanowire. (a)

(b)

Fig. 9. TEM images of a ZnO single crystalline nanowire: (a) low magnification TEM image, the inset is the corresponding ED pattern; (b) high-resolution TEM image.

147

Zinc Oxide Nanostructures (b)

(a)

Fig. 10. SEM images of the ZnO nanoribbons: (a) at low-magnification; (b) at highmagnification.

(b)

(a)

(004)

Intensity (a.u.)

(002)

Fig. 11. SEM images of ZnO arrays grown on silicon substrate: (a) 45 ⬚ tilted view; (b) top view.

0

20

40 60 2θ (degree)

80

100

Fig. 12. XRD pattern of ZnO arrays on a silicon substrate.

148

Zhang and Dai

The crystal structure of the arrays is examined by XRD (Fig. 12). The two peaks are indexed as (0002) and (0004) of the wurtzite structure of ZnO, indicating that the nanorods are preferentially oriented in the c-axis direction. There is a bigger difference in the XRD spectra between ZnO arrays and T-ZnO nanorods or ZnO nanowires (Figs 1 and 12).

4. Growth Mechanism Understanding the growth mechanism plays an important role in controlling and designing nanostructures. Many methodologies have been developed to synthesize one-dimensional nanostructures [47–52]. They can be categorized into two major approaches based on the reaction media during the preparation: solution phase process and gas phase process. Here, we discuss nanowire growth in gas phase. 4.1. Growth mechanism in gas phase 4.1.1. Vapor–liquid–solid growth mechanism A well-accepted mechanism of nanowire growth via gas phase reaction is the socalled vapor–liquid–solid (VLS) process proposed by Wagner in 1960s during his studies of large single-crystalline whisker growth [53]. According to this mechanism, the anisotropic crystal growth is promoted by the presence of liquid alloy-solid interface. Vapor–liquid–solid nanowire growth mechanism includes three stages: alloying, nucleation and axial growth. Recently, real-time observation of Ge nanowire growth was conducted in an in-situ high temperature transmission electron microscope [54]. Based on this mechanism a series of nanowires have been synthesized for elemental semiconductors [47], compound semiconductors [7, 16] and oxides [16, 29]. 4.1.2. Oxide-assisted growth mechanism In contrast to the well-established VLS mechanism, the Lee’s group recently proposed a new nanowire growth route called oxide-assisted nanowire growth [55–58]. This oxide-assisted method reported by the Lee group has the advantage of requiring neither a metal catalyst nor a template, which simplifies the purification and subsequent application of the wires. 4.1.3. Vapor–solid growth mechanism In addition to the above-mentioned two mechanism, the classical vapor–solid (VS) method for whiskers growth also merits attention for the growth of nanometer 1D materials [53]. In this process, the vapor is first generated by evaporation, chemical reduction or gaseous reaction. The vapor is subsequently transported and condensed onto a substrate. The VS method has been used to prepare oxide, metal whiskers with micrometer diameters. The requirements for 1D crystal growth, such as the presence of a dislocation at the vapor–solid interface, are still a matter of controversy in 1D VS growth. 4.2. Growth mechanism of ZnO nanostructures The synthesis of ZnO nanostructures in our group is based on the thermal evaporation of Zn powders under the controlled conditions without the presence of catalyst,

Zinc Oxide Nanostructures

149

which is different from ZnO nanowires grown by VLS mechanism [16].The growth mechanism of ZnO nanostructures is discussed as follows. The formation of whisker includes two stages: nucleation and growth. In our process, nucleation at the initial stage might have a crucial role on formation of ZnO nanostructures. The metallic zinc is in its vapor state at the high temperature (Zn: bp of 911 ⬚C). The gaseous zinc diffuse and immediately reoxidize in the environment of oxygen. It is known that the gaseous ZnO only exists as highly activated species with an extremely short lifetime [59]. So the oxidation reaction at our processing temperature is as follows: 2Zn(g) ⫹ O2 ⫽ 2ZnO(s). The process of the initial nucleation includes diffusion, collision of atoms and reaction between the vapor molecules (including Zn(g) and O2). When the supersaturation increased to the level at which nuclei formed, ZnO nuclei are produced and grown to a critical size. Then the nuclei grow into highly anisotropic nanostructures. The growth mechanisms for the various nanostructures are discussed as follows. 4.2.1. T-ZnO nanorods T-ZnO particles have incited many investigators to study their structure and growth mechanism. The various models have been proposed for the growth process of tetrapod ZnO particles [60–64], but the structure of the T-ZnO whiskers has not been completely elucidated because the particles observed were too large to identify by TEM. Shioziri thought that T-ZnO whiskers have the zincblende structure, four wurtzite crystals formed on their (111) faces by introducing stacking faults [60]. Iwanaga proposed the octahedral multiple twin (octa-twin) nucleus models [61–63]. Nishio proposed a new growth model based on the phase transformation from an octahedral zinc-blende ZnO crystal to a twinned wurtzite ZnO crystal [64]. Among above-mentioned models, only the Iwanaga’ model can explain the prototype angle relation [61–62]. The octa-twin nucleus model for the tetrapod particles consists of three stages: 1. octa-twin formation: the formation of octahedral embryos is composed of eight inversion-type twins 2. strain relaxation: the decohesion of some of the twin boundaries in octa-twins to relax the strain energy accumulated due to a large misfit angle around the common edges of the twins 3. leg growth: the preferential growth in the ⫹c direction of the four crystals in octa-twins This model explained the orientation relationship and the measured angle between the legs agreed exactly with those calculated for this octa-twin model. So far, there exists no complete evidence for the existence of the octa-twin nucleus. The T-ZnO nanorods synthesized by us make it possible to reveal the structure of T-ZnO nanorods by TEM and HRTEM. According to the crystal vapor nucleation mechanism, ZnO nuclei formed in the alumina crucible are homogeneous nucleation in gas phase. Firstly, octa-twins nuclei form in an atmosphere containing oxygen. The octa-twins nuclei consist of eight tetra– hedral crystals, each consisting of three {11 2 2} pyramidal faces and one {0001}

150

Zhang and Dai

basal face. The eight tetrahedral crystals are connected together by making the pyramidal faces contact one another to form an octahedron. The surfaces of the octo-twin are composed of basal planes. An important additional condition is that every twin is of the inversion type, i.e., the polarities of the twinned crystals are not mirrorsymmetric with respect to the contact plane but antisymmetric. Thus the eight basal surfaces of the octa-twin are alternately the plus surface (0001) and the minus surface (0001), as illustrated in Fig. 13. In a small nucleus, a large contribution of the surface energy to the total energy prevents the octa-twin from the crack formation. The crystal growth results from the oriented adsorption of vapor-nucleating species on certain single crystalline nucleus surface with relatively high surface energy. Due to the crystallographic polarity of ZnO crystal, legs grow preferentially in the direction perpendicular to the four plus (0001) surfaces. A preferential growth in the ⫹c direction has already been confirmed for ZnO crystal [42]. Therefore the growth mechanism for the T-ZnO nanorods might be vapor–solid mechanism. The detailed structure of individual T-ZnO nanorod is characterized by TEM. Fig. 14a is the bright-field image and dark-field image of a T-ZnO nanorod. Fig. 14b has been taken under the condition that only the central portion of the tetrapod crystal satisfies diffraction condition. It is obvious that an extra grain at the center of the particle is detected, in addition to the four legs crystals. Based on the position of the extra grain and the octa-twin model (Fig. 13), we speculate that it is one of eight tetrahedral

Fig. 13. Octa-twin composed of eight pyramidal inversion-twin crystals.

Fig. 14. TEM images of T-ZnO nanorod: (a) bright-field electron micrograph; (b) dark-field electron micrograph.

Zinc Oxide Nanostructures

151

crystals in octa-twin nuclei and the other three extra grains might exist in the center of the T-ZnO nanorod in addition to the four legs. A TEM image of another T-ZnO nanorod is shown in Fig. 15a. There exist four boundaries in the central part. It is noticed that the other two interfaces are observed in both left and right legs (as marked by the arrows), and the upper leg and lower leg are single crystals. Fig. 15b is the corresponding HRTEM image of the central part in Fig.15a. It reveals the structure of the boundaries between the element crystals. From the HRTEM image, it can be seen that the interfaces are sharp and shows no amorphous layer. Four crystals twin each other and thus four twinning boundaries form. These twins are smoothly conjugated on the boundaries and united in a coherent relationship, indicating little lattice distortion near the boundary. As discussed above, our experimental observation gives direct evidence to the octahedral multiple twin (octa-twin) nucleus model of T-ZnO nanorod. The octa-twin growth model is well substantiated. 4.2.2. ZnO nanowires ZnO nanowires form in an environment of lower partial pressure of oxygen. At the initial nucleation stage, Zn vapor might condense on the substrate in liquid and oxidize to ZnO nuclei then. The nucleation is on the silicon substrate, so it is a heterohomogeneous nucleation process. A typical ZnO nanowire tip is shown in Fig. 16. The tip is generally round and consists of a polycrystalline ZnO with twins and defects. The presence of the twin at the tip areas should result in the fast growth of ZnO nanowires. Sears proposed a vapor–solid growth model for the preferential whisker growth in many materials [65]. In accordance with this model, the defect structure and supersaturation are probably critical factors for whisker growth. As seen in the HRTEM image of a bicrystalline ZnO nanowire shown in Fig. 8a, a coherent symmetric twin is a representative characteristic in the nanowire. The twin boundary is

(a) (b)

Fig. 15. TEM images of a T-ZnO nanorod: (a) low magnification TEM image; (b) highresolution TEM image.

152

Zhang and Dai

(a) (b)

Fig. 16. TEM images of the top in a ZnO nanowire: (a) low magnification TEM image; (b) Highresolution TEM image.

parallel to the nanowire axis. This HRTEM image highlights the importance of twinning in the growth of the ZnO nanowires. Because of this peculiar configuration of twinning as in Fig. 8a, the growing frontier perpendicular to the wire axis keeps in zigzagging, which is energetically favorable, and serves as the stable sites for the rapid stacking of atoms [66]. It results in a fast axial growth rate over the radius direction. In this sense, vapor–solid phase (VS) is involved in the growth of ZnO nanowires in a way similar to that of vapor-phase epitaxy. 4.2.3. ZnO arrays mechanism Nucleation at the initial stage might also have a crucial role in both the vertical and in-plane alignments of the arrays. Compared to the ZnO nanowires, ZnO arrays formed in higher partial pressure of Zn and oxygen. At the initial nucleation stage, vapor Zn is oxidized soon and condense on the substrate to form ZnO nuclei. The ZnO nuclei might consist of a single crystalline ZnO, which is similar to the epitaxially growth of single crystalline ZnO films on silicon substrate [14]. The ZnO arrays are then assembled on the top of the particles and grew in outward continually. One unique aspect of the ZnO arrays is the feature of hexagonal prisms. The uniform hexagonal prismatic growth morphology can be simply explained by the “lowest energy” argument, i.e., the hexagonal (0001) plane of ZnO with wurtzite structure is the closest packed plane in the crystal, and the stacking along the [0001] direction therefore becomes energetically favorable. This observation of the flat ends also suggests that the rods grow in by a noncatalysis growth mechanism. In the nanorod growth process using the catalysisassisted VLS mechanism, nanosized metal clusters have a critical role as a catalyst in forming liquid droplets that absorb the gas-phase reactants where nanorod growth occurs. Hence, the metallic nanoparticles are commonly observed at the end of nanorods grown by the catalysis-assisted VLS method [16]. In our process, the flat hexagon-shaped ends are clearly shown in Fig. 11. These results exclude the possibility

153

Zinc Oxide Nanostructures

that the nanorods grow in by the catalysis-assisted growth mechanism. The growth of ZnO arrays might be controlled by VS mechanism. 4.3. Processing parameters It is obvious that the different ZnO nanostructures obtained by us are attributed to the variation of supersaturation (the partial pressure of Zn vapor and oxygen) in the surrounding region. The supersaturation in the reaction vessel (an alumina crucible or an alumina tube) strongly depends on the processing parameters, such as source materials, temperature, pressure, carrier gas (including gas species and its flow rate), substrate and evaporation time. So the control growth of ZnO nanostructures can be achieved by the controlling processing conditions. The details will be discussed elsewhere [67].

5. Optical Properties of ZnO Nanostructures PL spectra of ZnO nanostructures at room temperature are shown in Fig. 17. The typical two emission peaks of a narrow peak at 385 nm and a broad peak at 495 nm are observed in the ZnO nanostructures. They are assigned to the UV emission and green emission respectively. Whilst the UV emission corresponds to the near

(a) 600

(b) 80

495 nm

Intensity (a.u.)

Intensity (a.u.)

400 300 200 100

495 nm

70

500

60 50 40

385 nm

30 20

385 nm

10

0

0 350

250 300 350 400 450 500 550 600 650 Wavelength (nm) (c)

400

450 500 550 Wavelength (nm)

600

25 385 nm

Intensity (a.u.)

20 15 10

495 nm

5 0 350

400

450 500 550 600 Wavelength (nm)

650

Fig. 17. Photoluminescence spectra of ZnO nanostructures at room temperature: (a) T-ZnO nanorods; (b) ZnO nanowires; (c) ZnO arrays.

154

Zhang and Dai

band-edge emission, the green emission peak is commonly refered to as a deep-level or trap-state emission. It is interesting to note that green emission intensities in three nanostructures are different, while their UV emission intensities are nearly the same. Green emission intensity in nanowires is strongest, that in T-ZnO nanorods is low and the intensity in ZnO arrays is lowest. Vanheusden [68] proved that the green transition has been attributed to the singly ionized oxygen vacancy in the ZnO and the emission results from the radiative recombination of a photogenerated hole with an electron occupying the oxygen vacancy. The stronger the intensity of the green luminescence, the more singly ionized oxygen vacancies there are. Therefore, it is reasonable to believe that the strong green light emission in ZnO nanowires could be attributed to a greater fraction of oxygen vacancies in them. The weaker green light emission in T-ZnO nanorods might be related to the lower oxygen vacancies concentration compared to the ZnO nanowires. The lowest green light emission in ZnO arrays might be related to a very low level of oxygen vacancies in them. It indicates that the ZnO arrays are of excellent optical quality.

6. Conclusions In this chapter, we have demonstrated a simple vapor-phase approach to control growth of ZnO nanostructures. Three types of the distinctive morphologies: T-ZnO nanorods, nanowires and arrays, are synthesized by our method. The key for controlling the shape and size is the level of supersaturation that controls ZnO nucleation and growth. The supersaturation in the reaction vessel strongly depends on the processing parameters. The growth of ZnO nanostructures is controlled by a vapor–solid (VS) mechanism. T-ZnO nanorods grow in octa-twin model. Our experimental observation gives the direct evidence to the octahedral multiple twin nucleus model of T-ZnO nanorod. ZnO nanowires are bicrystallines containing a single twin boundary along the entire length. The defect structure acts as a key factor which greatly enhanced the nucleation and one-dimensional growth of ZnO bicrystalline nanowires. The room temperature PL spectra of the nanostructures show a narrow UV emission at 385 nm and a broad green emission at 495 nm. The green emission intensity in nanostructures is different, while their UV emission intensity is nearly the same. The difference in green emission intensity results from the different level of oxygen vacancies in ZnO. This approach also offers other advantages. Firstly, metal catalysts are not necessary in our process. As a result, the nanostructures are free of other materials. Secondly, the synthesis is under lower temperature compared to other methods. This method can also be potentially applied to the synthesis of other metal oxides, such as MgO etc. We believe that these ZnO nanostructures could be a new class of particularly useful nanostructures for fundamental studies and many technological applications.

Acknowledgment The work was supported by the National Natural Science Foundation of China (No. 50172006, 59872004, 59771050), the Fund for Returned Overseas Scholar of

Zinc Oxide Nanostructures

155

Ministry of Education of China, the Fund for Cross-Century Talent Projects of Educational Ministry of China, and the Research Fund for the Doctoral Program of Higher Education of China.

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.

S. Iijima, Nature 354 (1991) 56. J. Hu, T. W. Odom and C. M. Lieber, Acc. Chem. Res. 32 (1999) 435. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. J. R. Heath, P. J. Kuekes, G. Synder and R. S. Williams, Science 280 (1998) 1717. D. Snoke, Science 273 (1996) 1351. Y. J. Han, J. Kim and G. D. Stucky, Chem. Mater. 12 (2000) 2068. C. C. Chen, C. Y. Chao and Z. H. Lang, Chem. Mater. 12 (2000) 1516. M. B. Mohamed, K. Z. Ismail, S. Link and M. A. El-Sayed, J. Phys. Chem. 102 (1998) 9370. Y. Chen, D. M. Bagnall, H. Koh, K. Park, K. Hiraga, Z. Zhu and T. Yao, J. Appl. Phys. 84 (1998) 3912. A. Ohtomo, M. Kawasaki, Y. Sakurai, I. Ohkubo, R. Shiroki, Y. Yoshida, T. Yasuda, Y. Segawa and H. Koinuma, Mater. Sci. Eng. B 56 (1998) 263. D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen and T. Goto, Appl. Phys. Lett. 70 (1997) 2230. P. Zu, Z. K. Tang, G. K. L. Wong, M. Kawasaki, A. Ohtomo, H. Koinuma and Y. Segawa, Solid State Commun. 103 (1997) 459. H. Co, J. Y. Xu, E. W. Seelig and R. P. H. Chang, Appl. Phys. Lett. 76 (2000) 2997. D. Z. Zhang, S. H. Chang, S. T. Ho, E. W. Seelig, X. Liu and R. P. H. Chang, Phys. Rev. Lett. 84 (2000) 5584. B. D. Bao, Y. F. Chen and N. Wang, Appl. Phys. Lett. 81 (2002) 757. M. H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber and P. Yang, Adv. Mater. 13 (2001) 113. Y. C. Kong, D. P. Yu, B. Zhang, W. Feng and S. Q. Feng, Appl. Phys. Lett. 78 (2001) 407. Y. W. Wang, L. D. Zhang, G. Z. Wang, X. S. Peng, Z. Q. Chu and C. H. Liang, J. Cryst. Growth 234 (2002) 171. J. Y. Li, X. L. Chen, H. Li, M. He and Z. Y. Qiao, J. Cryst. Growth 233 (2001) 5. W. I. Park, D. H. Kim, S. W. Jung and G. C. Yi, Appl. Phys. Lett. 80 (2002) 4232. J. J. Wu and S. C. Liu, Adv. Mater. 14 (2002) 215. C. Xu, G. X. Y. Liu and G. Wang, Solid State Commun. 122 (2002) 175. L. Guo, J. X. Cheng, X. Y. Li, Y. J. Yan, S. H. Yang, C. L. Yang and J. N. Wang, Mater. Sci. Eng. C 16 (2001) 123. J. Q. Hu, Q. Li, N. B. Wong, C. S. Lee and S. T. Lee, Chem. Mater. 14 (2002) 1216. J. Zhang, L. Sun, C. Liao and C. Yan, Chem. Commun. 3 (2002) 262–263. L. Vayssieres, K. Keis, A. Hagfeldt and S. E. Lindquist, Chem. Mater. 13 (2001) 4395. Z. L. Wang, R. P. Gao, Z. W. Pan and Z. R. Dai, Adv. Eng. Mater. 3 (2001) 657. Z. R. Dai, Z. W. Pan and Z. L. Wang, Solid State Commun. 118 (2001) 351. M. H. Huang, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind, E. Weber, R. Russo and P. Yang, Science 292 (2001) 1897. P. Yang, H. Yan, S. Mao, R. Russo, J. Johnson, R. Saykally, N. Morris, J. Pham, R. He and H. J. Choi, Adv. Funct. Mater. 12 (2002) 323. X. G. Peng, L. Manna, W. D. Yang, J. Wickham, E. C. Scher, A. Kadavanich and A. P. Alivisatos, Nature 404 (2000) 59. L. Manna, E. C. Scher and A. P. Alivisatos, J. Am. Chem. Soc. 122 (2000) 12700.

156 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68.

Zhang and Dai Y. W. Jun, S. M. Lee, N. J. Kang and J. Cheon, J. Am. Chem. Soc. 123 (2001) 5150. M. L. Fuller, J. Appl. Phys. 15 (1944) 164. G. W. Sears, R. Powell and B. Donn, J. Chem. Phys. 39 (1963) 2248. K. A. Jones, J. Cryst. Growth. 8 (1971) 63. Y. Suyama, T. Tomokiyo, T. Manabe and E. Tanaka, J. Am. Cer. Soc. 71 (1988) 391. M. Kitano, T. Hamabe, S.Maeda and T. Okabe, J. Cryst. Growth 109 (1991) 277. H. Iwanaga, M. Fujii and S. Takeuchi, J. Cryst. Growth 134 (1993) 275. M. Kitano, T. Hamabe and S. Maeda, J. Cryst. Growth 102 (1990) 965. H. Iwanaga, A. Tomizuka, S. Takeuchi and T. Yoshiie, Phil. Mag. A 64 (1991) 303. H. Iwanaga, N. Shibata, O. Nittono and M. Kasuga, J. Cryst. Growth 45 (1978) 228. Y. Dai, Y. Zhang, Q. K. Li and C. W. Nan, Chem. Phys. Lett. 358 (2002) 83. Y. J. Han, J. Kim and G. D. Stucky, Chem. Mater. 12 (2000) 2068. C. C. Chen, C. Y. Chao and Z. H. Lang, Chem. Mater. 12 (2000) 1516. Y. Dai, Y. Zhang and Z. L. Wang, Chem. Phys. Lett. (in reviewing). A. M. Morales and C. M. Lieber, Science 279 (1998) 208. C. R. Martin, Science 266 (1994) 1961. I. D. Almawlawi, C. Z. Liu and M. Moskovits, J. Mater. Res. 9 (1994) 1014. W. Han, S. Fan, W. Li and Y. Hu, Science 277 (1997) 1287. T. J. Trentler, K. M. Hickman, S. C. Geol, A. M. Viano, P. C. Gibbons and W. E. Buhro, Science 270 (1995) 1791. X. Duan and C. M. Lieber, Adv. Mater. 12 (2000) 298. A. P. Levitt, Whisker Technology, Wiley-Interscience, New York (1970). Y. Wu and P. Yang, J. Am. Chem. Soc. 123 (2001) 3165. S. T. Lee, N. Wang, Y. F. Zhang and Y. H. Tang, MRS Bulletin 24 (1999) 36. W. S. Shi, H. Y. Peng, N. Wang, C. P. Li, L. Xu, C. S. Lee, R. Kalish and S. T. Lee, J. Am. Chem. Soc. 123 (2001) 11095. S. T. Lee, N. Wang and C. S. Lee, Mater. Sci. Eng. A 286 (2000) 16. N. Wang, Y. H. Tang, Y. F. Zhang, C. S. Lee, I. Bello and S. T. Lee, Chem. Phys. Lett. 299 (1999) 237. G. V. Chertihin and L. Andrews, J. Chem. Phys. 106 (1997) 3457. M. Shiojiri and C. Kaito, J. Cryst. Growth. 52 (1981) 173. M. Fujii, H. Iwanaga, M. Ichihara and S. Takeuchi, J. Cryst. Growth, 128 (1993) 1095. S. Takeuchi, H. Iwanaga and M. Fujii, Phil. Mag. A 69 (1994) 1125. H. Iwanaga, M. Fujii, M. Ichihara and S. Takeuchi, J. Cryst. Growth, 141 (1994) 234. K. Nishio and T. Isshiki, Phil. Mag. A, 76 (1997) 889. G. W. Sears, Acta. Met. 3 (1956) 268. H. Z. Zhang, Y. C. Kong, Y. Z. Wang, X. Du, Z. G. Bai, J. J. Wang, D. P. Yu, Y. Ding, Q. L. Hang and S. Q. Feng, Solid State Commun. 109 (1999) 677. Y. Dai, Y. Zhang and Z. L. Wang, Solid State Commun. Accepted. K. Vanheusden, W. L. Warren, C. H. Seager, D. R. Tallant, J. A. Voigt and B. E. Gnade, J. Appl. Phys. 79 (1996) 7983.

Chapter 9 One-Step Hydrothermal Synthesis and Characterizations of Titanate Nanostructures L.-M. Peng1,2, Q. Chen1, G. H. Du2, S. Zhang1 and W. Z. Zhou3 1

Department of Electronics, Peking University, Beijing 100871, China; 2Beijing Laboratory of Electron Microscopy, Institute of Physics and Center for Condensed Matter Physics, Chinese Academy of Sciences, China; 3School of Chemistry, University of St. Andrew, St. Andrews KY16 9ST, UK

1. Introduction Nanotubes are basically one-dimensional hollow materials [1]. In principle they can be formed by almost any compound, using a template growth mechanism, such as the sol-gel method which involves the synthesis of the desired material within the pores of a nanoporous membrane or other solid. Examples of success synthesis include TiO2, V2O5, MnO2, Co3O4, ZnO, WO3, SiO2, Al2O3 and ZrO2 nanostructures [2–4]. Most of these synthesized nanotubes made from three-dimensional crystals are, however, of either amorphous or semi-crystalline nature. In forming a nanotube by a three-dimensional compound, such as TiO2, since the nanotube is a rolled-up structure of a two-dimensional sheet, it is impossible that all the chemical bonds of the threedimensional compound will be fully satisfied on the nanotube surfaces so that to form a perfectly ordered, flawless nanotubular structure. Therefore nanotubes made from three-dimensional compounds cannot form a fully crystalline structure and the nanotube surface is normally active. Two-dimensional layered compounds are known to have fully satisfied chemical bonds on their van der Walls basal planes, which are relatively inert [5]. The atoms on the prismatic faces of these compounds are, however, not fully bonded and are therefore chemically very reactive. When nanoclusters of two-dimensional compounds are formed, the prismatic faces are full of dangling bonds which tend to destabilize the planar structure. At high enough temperature the stored chemical energy (via mainly dangling bonds) may overcome the activation barrier associated with the bending of the basal planes of the two-dimensional nanoclusters and form hollow nanostructures. These seamless and stable hollow nanostructures are particularly useful in nanoelectronics applications, since the elastic mean free path of electrons in these

158

Peng et al.

tubular materials is greatly enhanced leading to the resistance free ballistic transport of electrons in these materials [6]. In this chapter we will be concerned mainly with two titanate nanostructures, H2Ti3O7 nanotubes and K2Ti6O13 nanowires, which are made from TiO2. Interest in titania nanostructures has been enormous due to their potential applications in paint, paper, cloth fibers, cosmetics and especially in photosynthetic cells [7–9]. The first claim on the successful synthesis of crystalline TiO2 nanotubes with the anatase phase was made by Kasuga et al. [10]. These authors treated raw anatase or rutile phase TiO2 material with NaOH solution and then washed the products with HCl acqueous solution and distilled water. It was concluded from their transmission electron microscopy (TEM) observations that the crystalline raw TiO2 material was first converted to an amorphous product through alkali treatment, and subsequently titania nanotubes were formed after the treatments with distilled water and HCl acqueous solution [11]. We used a similar procedure for treating the raw TiO2 material. Instead of TiO2 nanotubes with the anatase phase, we found that our products were composed of mainly H2Ti3O7-type nanotubes [12, 13]. During the course of our investigation of these nanotubes, we found that these nanotubes may be synthesized in a single alkali treatment [14]. This simple and low cost synthetic method has also been used for synthesize other oxide nanostructures, and in this chapter we will discuss only K2Ti6O13 nanowires [15]. Both H2Ti3O7 and K2Ti6O13 belong to the titanate group with the formula A2TinO2n⫹1. These titanates form a variety of important layered structures for n ⫽ 3 to n ⫽ 6 with protons or alkali metal (A) cations in the interlayer space [16].

2. Experimental Procedures 2.1. Raw materials Both H2Ti3O7 nanotubes and K2Ti6O13 nanowires were prepared by a simple one-step hydrothermal method. Commercially bought pure (anatase) and mixed phase TiO2 prepared via a sol-gel procedure were used as the raw materials for preparing trititanate nanotubes [12] and potassium titanate nanowires [15]. Shown in Fig. 1a are the synthesized raw TiO2 particles with particle size ranging from 10 nm to a few microns, and in Fig. 1b the commercially bought TiO2 material. The synthesized TiO2 sample is composed of mixed anatase and rutile phases with large size distribution, while commercial TiO2 is monophasic with a much smaller size distribution from 80 to 300 nm. The minimum size of the TiO2 particles observed is around 1 nm. 2.2. Products In a typical synthesis, about 0.3 g raw material of TiO2 was added in a 10 M alkali aqueous solution (NaOH for preparing trititanate nanotubes and KOH for preparing potassium titanate nanowires). The specimen was transferred into a sealed teflon container and statically heated in a furnace at 130 ⬚C for three days. The final products of nanotubes and nanowires were obtained by filtering and washing with de-ionized water (or aceton and ethanol) at room temperature.

Synthesis and Characterizations of Titanate Nanostructures (a)

159

(b)

Fig. 1. Low magnification TEM images showing the raw TiO2 particles (a) prepared via a solgel method and (b) commercially bought. The TiO2 particles shown in (a) are of mixed phase, while that shown in (b) are pure anatase phase. The scale bar in Fig. 1b denotes 500 nm, and the regions indicated by arrows in Fig. 1a contain a few particles of large size.

2.3. Methods TEM samples were prepared as follows. The powder sample was dispersed in ethanol by an ultrasonic treatment (acetone was used when the product was washed by acetone). One drop of the suspension was added on a holey carbon film supported on a copper grid and allowed to dry in air before being transferred into a fieldemission gun (FEG) transmission electron microscope (Philips CM200/FEG) operating at 200 kV. The microscope equipped with a Gatan Imaging Filter system (GIF) was used for SAED, HRTEM, EDX and EELS. In EDX, nanoprobe electron beams (⬍3 nm in diameter) were applied to expose the areas of individual nanotubes under investigation. Areas smaller than 40 nm in diameter containing only one single nanotube were selected for the EELS study using the entrance aperture of the GIF system. Image simulation was performed using the MSI software CERIUS2. Adsorption-desorption measurements were obtained on a Quantachrome Autosorb-1 apparatues. The sample was degassed at 200 ⬚C in vacuum prior to absorption.

3. Structural Characterization 3.1. Basics of layered metal oxides Both H2Ti3O7 nanotubes and K2Ti6O13 nanowires are based on the alkali metal titanates with the formula A2TinO2n⫹1 which form a variety of layered structures for n ⫽ 3 to n ⫽ 6 with protons or alkali metal (A) cations in the interlayer space [16]. The common features of these layer structures are that they are composed of corrugated ribbons of edge-sharing TiO6 octahedra. The ribbons are n octahedral wide and join corners to form “stepped” layered structures. While H2Ti3O6 may be described as a “step 3” structure, meaning that the structure has three edge-sharing octahedra between corner-sharing locations (see Fig. 2a), K2Ti6O13 has a “condensed step 3” structure (see Fig. 2b). Although the “condensed step” structures are usually regarded

160

Peng et al.

(a)

(b)

Fig. 2. Atomic models for (a) “step 3” structured H2TiO7 (hydrogen atoms in the interlayer space are not shown) and (b) “condensed step 3” structured K2Ti6O13.

also as layered structures, they are not real layered structures in the sense that TiO6 octahedra of different layers are not separated. Instead every n TiO6 octahedra share oxygen with TiO6 octahedra of adjacent layers. Different layers of TiO6 octahedra are therefore not weakly bounded via van de Walls force but via strong chemical bonds. As will be shown later in this chapter that potassium hexatitanate nanostructures derived from the quasi-layered “condensed step” structures are not of tubular form, instead they prefer a three-dimensional nanowire structures. 3.2. Trititanate nanotubes Shown in Fig. 3 is a low magnification TEM image of the products resulted from the following reaction TiO2 ⫹ NaOH ⫹ H2O → (130 ⬚C for three days) → nanotubular materials This image shows clearly that large quantity (⬎90%) nanotubular materials with about 9 nm in diameter and 100 to several hundreds nanometers in length have been synthesized. The tubular nature of the materials was confirmed by in-situ tilting experiments. A single tube was firstly selected. This tube was then tilted along its tube axis and its diameter was monitore. Shown in Fig. 4 are a series of high-resolution electron microscopy (HREM) images taken with different tilting angles. The diameter of the tube was found to remain almost constant for over 34 degrees of tilting about it axis, suggesting the one-dimensional object is indeed a tube with hollow interior. Shown in Fig. 5 are a cross-sectional view of the nanotube and a model constructed based on this image. The most striking feature revealed by this image about the nanotube is that the nanotube is not a closed tube like a single-wall carbon nanotube, but of a scroll nature. It is expected that electrons would travel in these tubes more like in a strained two-dimensional layered structure rather than in a one-dimensional structure. Powder XRD was also performed with Cu K radiation on a Rigaku D/max-2400 diffractometer. Shown in Fig. 6 are experimental and simulated XRD profiles.

Fig. 3. TEM image showing nanoscale tubular materials made from raw TiO2 particles as shown in Fig. 1.

(a)

(b)

(c)

Fig. 4. Three HREM images showing a single nanotube viewing along different tilting angles when the sample is tilted along the tube axis. The three images correspond to (a) ⫹12 degrees, (b) 0 degree and (c) ⫺22 degrees of tilting angle.

(a)

(b)

Fig. 5. (a) HREM image showing that the nanotube is of the scroll type and (b) a model constructed according to (a).

162

Peng et al.

The XRD simulation used the model shown in Fig. 5b, which is based on the atomic coordinates given in Table 1 (derived from the Feist and Davies model [16]) and in Table 2 (from the Anderson and Wadsley model [17]). It should be noted that, according to the Feist and Davies model [16], the space group of H2Ti3O7 is C2/M (#12 in the International Tables) which is different from that of derived from the Na2Ti3O7 model (P21/m, space group #11) [17] and the two layers shown in Fig. 2a are not equivalent. The stacking sequence for the layer structure is “ABA” in the Feist and Davies model [16]. However, in the original paper of Feist and Davies [16] the atomic coordinates of hydrogen are not given, and these are obtained (a)

(b)

Fig. 6. (a) Experimental XRD profile taken from nanotubes and (b) simulated XRD pattern using the proposed trititanate nanotube model.

Table 1. Optimized atomic coordinates (in angstrom) for H2Ti3O7 and Feist and Davis model [16]

H1 H2 Ti1 Ti2 Ti3 O1 O2 O3 O4 O5 O6 O7

x

y

z

6.087119004 6.345664078 1.850340733 3.157948798 0.029505277 2.863489596 1.596318639 0.284913301 ⫺1.716090998 5.945283051 4.076384736 2.774444139

⫺0.000000000 1.899463548 ⫺0.000000000 ⫺0.000000000 ⫺0.000000000 ⫺0.000000000 ⫺0.000000000 ⫺0.000000000 0.000000000 ⫺0.000000000 ⫺0.000000000 ⫺0.000000000

2.775366426 7.203276390 5.003142020 2.186395998 7.881897579 0.342322955 3.038343342 5.738912276 7.850949102 1.799707063 4.107708021 6.900365424

Note: a⫽17.6373, b⫽3.7989, c⫽9.5721 Å,  ⫽ 90.0000, ␤ ⫽ 104.1442 and ␥ ⫽ 90.0000 (C2/M).

Synthesis and Characterization of Titanate Nanostructures

163

Table 2. Optimized atomic coordinates (in angstrom) for H2Ti3O7 and the Anderson and Wadsley model [17] x H1 H2 Ti1 Ti2 Ti3 O1 O2 O3 O4 O5 O6 O7

2.780141721 4.453721650 ⫺0.230182226 0.720080539 ⫺0.767072237 1.580354211 0.417303244 2.542546537 0.820692458 7.496773023 6.786206172 8.204227629

y

z

0.941070019 2.823210058 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019 0.941070019

5.108199839 7.226410680 2.733131050 6.329605835 9.013458226 1.659534188 4.336050748 6.050697043 8.252922141 7.000172340 3.021515121 0.467930338

Note: a⫽9982, b⫽3.7643, c⫽9.5453 Å,  ⫽ 90.0000, ␤ ⫽ 102.6538 and ␥ ⫽ 90.0000 (P21/M).

(a)

(b)

Fig. 7. (a) Experimental and simulated (denoted by the arrows) images of the nanotube using the model shown in (b).

in our case by geometry optimization using the ab-initio CASTEP program [18]. In the Anderson and Wadsley model [17] the layer stacking sequence is “AAA”. Atomic coordinates given in Table 2 are obtained by replacing Na with H and then fully relaxing the model. Tubular structures may be constructed via a similar procedure as widely used for constructing carbon nanotubes. The only difference here is that the sheets of TiO6 octahedra are rolled onto a scroll surface rather than a cylindrical surface. While an “AAA” stacking sequence is usually obtained when rolling a single sheet of TiO6 octahedra along the b-axis to form a tube, an “ABA” stacking sequence may easily be obtained via either introducing a small helical angle or a defect into the tube [13]. Shown in Fig. 7 is an experimental HREM image of a H2Ti3O7 nanotube, together with simulated images using the Cowley and Moodie multislice method [19] and a tube model derived from the atomic coordinates given in Table 1. An excellent agreement can be found in between the experimental and simulated images, confirming that our model of the H2Ti3O7 nanotube is correct.

164

Peng et al.

3.3. Potassium hexatitanate nanowires To understand the role of NaOH in the synthetic process of H2Ti3O7, we replaced NaOH by KOH while all other conditions were unchanged. Shown in Fig. 8 are a low magnification TEM image (Fig. 8a) and a higher magnification HREM image (Fig. 8b) of the products resulted from the following reaction TiO2 ⫹ KOH ⫹ H2O → (130 ⬚C for three days) → nanowires These images show clearly that large quantity (⬎90%) nanowires materials with about 5 nm in diameter and 100 to several hundreds nanometers in length have been synthesized. The composition of a single nanowire was analyzed using the technique of electron energy loss spectroscopy (EELS). A single nanowire was selected using a small entrance aperture of the GIF system (Fig. 9a), and EELS spectra were recorded for (a)

(b)

Fig. 8. (a) A low magnification TEM image showing many one-dimensional nanostructures and (b) corresponding HREM image of one selected object showing that they are wellcrystalline nanowires. (a)

(b) 10000 8000

Counts

K

Ti

6000 4000 O 2000 0 300

400 500 600 Energy Loss (eV)

700

Fig. 9. (a) TEM image showing that a single free-standing nanowire has been selected for EELS experiments and (b) a typical EELS spectrum obtained from a single free-standing nanowire showing that the nanowire is composed mainly of K, Ti and O.

165

Synthesis and Characterizations of Titanate Nanostructures (a)

(b) 2.60–Å (601) 2.90–Å (112)

2 nm

Fig. 10. Experimental and simulated HREM images along (a) [001] and (b) [1, 11, 6] zone axes for a K2Ti6O13 nanowire.

electrons passing through this aperture. Shown in fig. 9b is a typical EELS spectrum from the single free-standing nanowire. This EELS spectrum shows clearly the O Kedge at 532 eV, the L2,3 edges of K at 297 eV and 294 eV and that of Ti at 461 eV and 455 eV respectively, suggesting that the nanowire is composed of mainly K, Ti and O. The structure of the nanowire has been determined by combined XRD, HREM experiments and simulations. Shown in Fig. 10 are two experimental HREM images and inserted simulated images (as indicated in the figure by arrows) for two zone axes. An excellent agreement is achieved between the experimental and simulated images, suggesting that the nanowire has basically the K2Ti7O13 structure as first given by Cid-Dresdner and Buerger [20].

4. Growth Mechanism Although the two titanate nanostructures were synthesized via a very similar manner, the growth mechanisms are very different for the two materials. Kasuga et al. [11] concluded from their synthetic conditions for producing titania nanotubes that washing the alkali-treated specimen with water followed by a further treatment with HCl were two crucial steps of the nanotube formation. The argument leading to their conclusion was that there were Ti-O-Na bonds in the alkali-treated specimen and ion exchange of Na⫹ by H⫹ took place during the washing and the HCl treatment, resulting in sheet-like particles. These sheet-like particles were believed to roll into nanotubes. Shown in Fig. 11 are three TEM images obtained from three samples. These samples were recovered from the alkali-treated TiO2 solution by filtering and washing at room temperature using water (Fig. 11a), acetone (Fig. 11b) and ethanol (Fig. 11c) respectively. The acidities of the two anhydrous solvents are too weak to undertake ion exchange of possible Na⫹ in the crystals by H⫹. High yields of the nanotubes were also obtained from the two samples treated with ethanol and aceton and the washing solvents seem to be of no effect at all on the morphologies and yields of

166

Peng et al.

Fig. 11. TEM images of trititanate nanotubes obtained by washing the raw products by (a) water, (b) aceton and (c) ethanol.

the products. The nanotubes were obviously formed during the alkali treatment and no acidic or other treatments are necessary. In fact, a careful comparison of the microstructures between the products with and without a HCl treatment indicated that the acidic treatment did not enhance the formation of the nanotubes, but introducing many defects on the nanotubes instead. Specimens were also collected after the alkali treatment of 12 h, 24 h and 48 h respectively. An increase of yield of nanotubes with the reaction time was observed although the sizes of the nanotubes were similar in these specimens. It was observed that after the alkali treatment two new phases were formed. The first is amorphouslike which contains Ti, O and Na and produces diffuse rings in electron diffraction patterns. The second is a nanotubular phase containing Ti and O but not Na (or below the detection limit of EDX and EELS). Experiments have also been performed using different TiO2 materials of different or mixed phases as the raw materials as shown in Fig. 1. Nanotubes with the same size (⬎100 nm long) were produced although the original TiO2 particles were very different both in phase and in size. It is clearly shown that the formation of the nanotubes is independent of the original structure and the particle size of TiO2. The nanotubes could not form due to a direct phase transformation from the TiO2 crystals. Our experiments seem to indicate that forming the disordered intermediate phase is a crucial step in the formation of the nanotubes, and trititanate nanotubes formed in a single alkali treatment can be described as follows: TiO2 ⫹ NaOH ⫹ H2O → Disordered intermediate phase (Ti, O, Na) → Sheets of trititanate → Trititanate nanotubes Our TEM observations on the growth of K2Ti6O13 nanowires suggest that the nanowires were grown directly from the parent TiO2 particles of the anatase phase, and four TEM images showing this growth process are given in Fig. 12. Shown in Fig. 12a are the raw TiO2 particles. The particle size ranges from 80 to 200 nm. Fig. 12b is a higher magnification TEM image of the TiO2 particles after being reacted with KOH for one day. The surface of these TiO2 particles are seen to have been covered by some small particles of a few nanometers in diameter. Shown in Fig. 12c is a

Synthesis and Characterizations of Titanate Nanostructures

167

Fig. 12. TEM images showing (a) the raw TiO2 particles and (b) the particles after the reaction with KOH for one day. (c) HREM image showing an enlarged portion of (b) near the surface of a TiO2 particle and (d) the small “nuclei” shown in (b) grown into nanoscale crystal platelets.

HREM image of the interface between a TiO2 particle and the newly grown smaller particles. Detailed analysis reveals that these smaller particles are indeed K2Ti6O13 particles. These small particles may continue to grow. Shown in Fig. 12d is a TEM image after the TiO2 particles reacted with KOH for two days. The small K2Ti6O13 particles on the surface of TiO2 are seen to have grown into small crystal platelets which became nanowires as shown in Fig. 8 after three days when almost all TiO2 particles had transformed into K2Ti6O13. Detailed analysis of images like Fig. 12 and corresponding electron diffraction patterns show that the grown K2Ti6O13 nanoscale crystallites and the parent TiO2 particles satisfy the following crystallographic relationships TiO2(101)//K2Ti6O13(200),

TiO2[010]//K2Ti6O13[010],

i.e. the K2Ti6O13 nanowires grow mainly along the [010] direction on the surface of the parent TiO2 particles, with their (200) lattice plans parallel to the (101) lattice planes of TiO2.

168

Peng et al. (a)

(b)

(c)

Fig. 13. (a) [010] view of K2Ti6O13, with (200) lattice planes lying nearly horizontally. (b) [010] view of TiO2 crystal of anataze phase with (101) lattice planes lying horizontally. (c) A schematic model for the K2Ti6O13/TiO2 interface.

Shown in Fig. 13a is an atomic model of K2Ti6O13 viewing along the [010] zone axis, and in Fig. 13b is that of TiO2 crystal viewing along the same orientation. As determined experimentally, the (200) lattice planes of K2Ti6O13 are parallel to the (101) planes of TiO2, and both are shown to lie almost horizontally in Fig. 13a, b. A schematic model may be constructed as shown in Fig. 13c by joining the two (010) surfaces of K2Ti6O13 and TiO2, together. It should be pointed out, however, that the lattices of the two compounds are not perfectly matched. The (101) plane spacing of TiO2 crystal is 0.35 nm, while that of the (200) spacing of K2Ti6O13 is 0.77 nm. The misfit between the two sets of lattice planes is given by ⫽

2d101 ⫺ d200 艐 9%, (2d101 ⫹ d200)/2

i.e. there will be a large strain associated with the growth of K2Ti6O13 out of TiO2 along the [010] direction. We expect that the lateral dimensional of the grown K2Ti6O13 crystal will be restricted by the strain associated with the lattice mismatch, which provides in turn a nature selection on the diameter of the grown nanowires.

Synthesis and Characterizations of Titanate Nanostructures

169

5. Electric and Optical Properties Shown in Fig. 14 are UV-visible absorption spectra from the H2Ti3O7 and K2Ti6O13 nanostructures compared with the raw TiO2 and bulk K2Ti6O13. These raw particles are of considerable larger size. This figure shows clearly that, like TiO2, both H2Ti3O7 nanotubes and K2Ti6O13 nanowires are semiconductors and the corresponding UV-visible absorption spectra exhibit blue-shifts suggesting larger band gaps for the nanostructures. The effective band gap as derived from the absorption spectra is by definition the energy necessary to create an electron (e⫺) and hole (h⫹) pair. In principle the excited electron-hole pair forms a bound state, i.e. Wannier exciton, and the behavior of the exciton may be calculated using a “confined exciton” model [21] which gives an increase in the apparent band gap energy. The band gaps may be estimated from the absorbtion spectra by a linear fit of the square root of the absorption coefficient as a function of the photon energy near the band gap (this procedure is strictly valid only for direct band gaps) a ⫽ 兹hv ⫺ Eg/hv, h is the corresponding phonon energy. For a given wavelength  of the phonon measured in nanometer, the phonon energy is given by This procedure gives Eg ⫽ 3.22 eV for TiO2, Eg ⫽ 3.52 eV for H2Ti3O7 and Eg ⫽ 3.45 eV for K2Ti6O13 of both raw materials and nanowires. The electronic structures of H2Ti3O7 and K2Ti6O13 are similar to that of TiO2. Shown in Fig. 15 are partial density of states of these titanium oxides for s, p and d states. These results are calculated using the ab-initio CASTEP code [18] and for optimized structures of TiO2, H2Ti3O7 and K2Ti6O13, and the Fermi level is set at 0.0 eV. This figure shows clearly that for all the three oxides the top valence bands are characterized mainly by the oxygen p states, while the conduction bands immediately above the valence bands are characterized by the titanium d states. The s and p states of potassium lie considerably below the Fermi level at about ⫺27 eV and ⫺11 eV

TiO2 particles H2Ti3O7 nanotubes

(b) 2.0 1.5 Absorbance

Absorbtion

(a) 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 200

300

400 500 Wavelength (nm)

600

K2Ti6O13 nanowires K2Ti6O13 particles

1.0 0.5 0.0 200

300 400 500 Wavelength (nm)

600

Fig. 14. Experimental UV-visible absorption spectra obtained from (a) TiO2 nanoparticles and H2Ti3O7 nanotubes, and (b) bulk and nanowaires samples of K2Ti7O13.

170 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 –2 –40

(b)

s states p states d states

Ti p–state

O s–state

Density of States (electrons/eV)

Density of States (electrons/eV)

(a)

Peng et al.

Ti d–state

O p–state

–30

–20

–10 0 Energy (eV)

10

20

65 60 55 50 45 40 35 30 25 20 15 10 5 0 –5 –40

s states p states d states

H s–state

–30

–20

–10 0 Energy (eV)

10

20

60 Density of States (electrons/eV)

(c)

s states p states d states

50 40 K p–state

30 20 10

K s–state

0 –40

–30

–20

–10 0 Energy (eV)

10

20

Fig. 15. Partial density of states for s, p and d states and (a) TiO2, (b) H2Ti3O7 and (c) K2Ti6O13.

respectively, while hydrogen s states lie at ⫺17 eV. Optical properties of these oxides are therefore determined largely by the oxygen p states and titanium d states associated with the TiO6 octahedra, which are affected only indirectly via distortion of the TiO6 octahedra by potassium and hydrogen atoms in interlayer space of H2Ti3O7 and K2Ti6O13. The calculated band gaps for TiO2, H2Ti3O7 and K2Ti6O13 are 2.48 eV, 2.94 eV and 2.58 eV respectively. Although these theoretical values are much smaller than experimental values, due to the well known shortcoming of LDA [22], they nevertheless result in the right order for the band gaps, i.e. the band gap for H2Ti3O7 is the largest, and that for TiO2 is the smallest. And optical properties may be obtained by using a scissors operator displacing the empty and occupied bands relative to each other by a rigid shift to fit experimental measurements of band gaps [23].

6. Conclusions Titanate nanostructures can be synthesized via a simple one-step alkali treatment. This procedure results in rather uniform nanostructures, and the size of these nanostructures is determined by competing factors including surface tension associated with dangling bonds and the strain introduced in forming these nanostructures. While a layered structure prefers tubular form, a quasi-layered structure, such as the “condensed step” structure results in nanowires having basically a three-dimensional structure.

Synthesis and Characterizations of Titanate Nanostructures

171

Acknowledgment This work was supported by the Ministry of Science and Technology of China (Grant No. 001CB610502), Chinese Academy of Sciences and Peking University.

References 1. 2. 3. 4. 5.

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

S. Iijima, Nature 354 (1991) 56. B. Lakshmi, C. Patrissi and C. Martin, Chem. Mater. 9 (1997) 2544. H. Nakamura and Y. Matsui, J. Amer. Chem. Soc. 117 (1995) 2651. B. C. Satiskhumar, E. M. Vogel Govindaraj, L. Basumallick and C. N. R. Rao, J. Mater. Res. 12 (1997) 604. R. Tenne and A. Zettl, in “Carbon Nanotubes—Synthesis, Structure, Properties and Applications”, Edited by M. S. Dresselhaus, G. Dresselhaus and P. Avouris, Springer-Verlag 81 (2000) 81. C. Dekker, Phys. Today 52 (1999) 22. A. Fujishima and K. Honda, Nature 37 (1972) 238. A. Linsebigler, G. Lu and T. Yates, Chem. Rev. 95 (1995) 735. M. Gratzel, Nature 415 (2001) 338. T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino and K. Niihara, Langmuir 14 (1998) 2160. K. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino and K. Niihara, Adv. Mater. 11 (1999) 1307. G. H. Du, Q. Chen, R. C. Che, Z. Y. Yuan and L.-M. Peng, Appl. Phys. Lett. 79 (2001) 3702. Q. Chen, G. H. Du, S. Zhang and L.-M. Peng, Acta Cryst. B 58 (2002) 587. Q. Chen, W. Zhou, G. H. Du and L.-M. Peng, Adv. Mater. 14 (2002) 1208. G. H. Du, Q. Chen, P. D. Han, Y. Yu and L.-M. Peng, Phys. Rev B (2002). T. P. Feist and P. K. Davies, J. Solid State Chem. 101 (1992) 275. S. Andersson and A. D. Wadsley, Acta Cryst. 15 (1962) 201. M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias and J. D. Joannopoulos, Rev. Mod. Phys. 64 (1992) 1045. J. M. Cowley and A. Moodie, Acta Cryst. 10 (1957) 609. H. Cid-Dresdner and M. J. Buerger, Z. Crist. 117 (1962) 411. L. E. Brus, J. Chem. Phys. 80 (1984) 4403. R. Asahi, Y. Taga, W. Mannstadt and A. J. Freeman, Phys. Rev. B 61 (2000) 7459. G. A. Baraff and M. Schluter, Phys. Rev. B 30 (1984) 3460.

Chapter 10 Nanowires and Nanotubes of Complex Oxides Xun Wang, Xiaoming Sun, Jian Xu and Yadong Li Department of Chemistry, Tsinghua University, Beijing, 100084, P.R. China

The discovery of carbon nanotubes (C-NTs) [1] has initiated an exciting, intellectually challenging, and rapidly expanding research field for low-dimensional nanostructures [1–9]. Over the past decade, several other novel 1-D nanostructures such as noncarbon nanotubes [2], nanowires [3], nanorods [4], nanocables [7] and nanobelts [8] have also been discovered. With reduced dimensionalities, this kind of 1-D nanostructures have been regarded as the smallest dimension structures for the efficient transport of electrons, and thus would be expected to be critical to the function and integration of nanoscale devices; meanwhile, some of them have been found to show enhanced photochemical and photophysical properties different from that of bulky or nanoparticle materials [9, 10], based on which many potential applications may been explored, and downscaling materials into 1D nanostructures have become one of the most important strategies to bring novel properties to materials. 1D nanostructures of oxides are especially appealing to chemistry and materials scientists, since oxides are the commonest-seen minerals in the earth, and have now been widely used in various areas, from ceramics, catalysis, sensor, to electronics, optics and magnetics. Intense studies have been carried out on the synthesis of oxides of nanowires as well as the exploration of their novel properties, for example, ZnO nanowires have been successfully prepared through VLS growth mechanism, and well-aligned ZnO nanowire arrays may function as room-temperature ultraviolet nanolasers [10]; SnO2 nanobelts have been obtained based on simple evaporation method, and single SnO2 nanobelt have been utilized as room temperature nanosensor for NO2 [9]. However, despite the several successful cases in the synthesis of oxides 1D nanostructures, such as VLS growth mechanism, template-confined growth, etc, there still lack of a general synthetic strategy, particularly a general understanding about their formation mechanism, which may be the key to the shape control synthesis of high quality oxides or even non-oxides nanowires materials. A comparison between cases from totally different reaction systems may be helpful for us to

174

Wang et al.

understand the anisotropic growth behaviors of 1-D nanostructures in that they may provide similar phenomena, and oxides nanowires may serve as perfect models for their numerous materials and synthetic ways. Oxides are usually obtained through CVD, Sol-gel, pyrolysis, solution-based redox reaction etc, and some of them have been adjusted to prepare oxides of nanowires. Transitional metal oxides sol can be injected into the pores of templates (carbon nanotubes, AAO or zeolite, etc) to prepare polycrystalline nanowires with the action of capillary force or electronic field. CVD can be adjusted to get single crystal nanowires by properly selecting catalysts to form alloy droplets following a VLS growth mechanism. Also, based on the anisotropic growth nature, oxides of nanowires can be obtained through solution-based redox reactions. There are different ways to classify the above synthetic routes, for example, reverse micelles method can be regarded as a solution-based route, and meanwhile, the reverse micelles may be considered as a kind of soft template, too. As far as onedimensional nanostructures are concerned, the key should be focused on the way how atoms or other building blocks are rationally assembled into structure with nanometer size but much larger length, so our attention have been paid on the original driving force for the anisotropic growth phenomena, and reverse micelles have been classified as a kind of template method. In this part, the preparation of oxides nanowires will be discussed in three independent parts: VLS growth mechanism, template-confined method and solution-based template-free synthetic routes.

1. Oxides of Nanowires Based on Vapor-Transport Method CVD and PVD have been adjusted to prepare oxides of nanowires, and most of them follow a VLS growth mechanism. Based on the phase diagram, one can choose appropriate catalysts for the growth of target products. In the oxides-related growth process, metal powders or the corresponding oxides have been used as the starting materials, and the oxygen are usually provided through wet carrying gas. MgO [15], ZnO [16] and SiO2 [17], etc, have been prepared through the VLS mechanism. By properly patterning the catalysts, arrays of ZnO nanorods [10] have also been prepared. Since the catalyst droplet alloy directs the nanowires’ growth and defines the diameter of crystalline nanowires, the nanowires obtained from the VLS process typically terminates at one end in a solid catalyst nanoparticles with diameter comparable to that of the connected nanowires. As an analogy to VLS growth,Yu et al. [18] have developed a SLS (solid-liquid-solid) growth mechanism to prepare Si nanowires, in which Si are not from the vapor or liquid phase of Si sources but from the Si substrate. In a similar way, Liu et al. [19] have also reported the synthesis of SiOx nanowires. Meanwhile, Yu et al. have also reported the synthesis of GeO2 [20] and Ga2O3 [21] following VS growth mechanism. As another enrichment to VLS, Wang et al. [22] have found that, in the growth of SiO2 nanowires by employing Ga as the catalysts, an alloy droplet of Ga-Si may simultaneously serve as catalysts for hundreds of SiO2 nanowires and the diameter of

Nanowires and Nanotubes of Complex Oxides

175

the obtained nanowires is much smaller than that of the catalysts, which is apparently far different from VLS growth mechanism. Undoubtedly, these entirely new phenomena will greatly enrich the traditional VLS mechanism, and may provide a better guidance for the growth of nanowires through CVD or PVD. In the synthesis of semiconducting oxides nanobelts [8] simple evaporation method has also been applied. Since it involves no catalysts or templates, it seems that it is totally based on the anisotropic crystallization nature of corresponding samples under overcooling temperature, and meanwhile, high purity can be expected.

2. Template-Confined Method to Oxides of Nanowires Templates have also been widely used to confine the growth direction of nanowires, and the adopted templates can be integral porous materials or individual tubular nanostructures. For the template-assisted synthesis of oxide nanorods, two main synthetic methods have been developed. The first involves a two-step process: firstly, metal wires/rods grow in the templates using electrochemical methods. Then the metal wires are oxidized, either electrochemically or by heating them in air, to form the oxide nanowires/nanorods. The other method uses sol-gel process to directly fill the pores of templates with desired oxide, which is then crystallized by heating. Both the approaches have their limitations, for instance, it is very difficult to make complex metal oxides with the first method. While for the latter, capillary action is the only driving force to form nanorods from the sol, so that the packing of solids in the pores is very low. In view of this, electrophoretic sol-gel method using an electric field to induce electrophoretic motion of the nanoclusters in the sol was developed recently to overcome such limitation. This method has been successfully used to obtain lead zirconate titanate nanorods [23]. Now, the template method has been proved to be a simple and versatile approach for preparing ordered nanowires or nanotube arrays. For instance, highly ordered TiO2 single crystalline nanowire arrays could be prepared within the pores of anodic aluminum oxide (AAO) template by a cathodically induced sol-gel method [24]. Besides the porous materials with densely packed pores, individual nanotubes (e.g., carbon nanotubes) were also used as templates. Until now, as far as we know, oxides of vanadium, molybdenum, tungsten, antimony, ruthenium, iridium, and germanium have be successfully synthesized [25, 26]. As a counterpart to the above template-assisted synthesis, microemulsions, reverse micelles as well as liquid crystal have also been applied to guide the growth of nanowires, and correspondingly they might be named as soft-template method [27–34]. Hopwood JD et al., used organized reaction microenvironments of BaNaAOT reverse micelles for barium sulfate nanoparticles and nanofilaments [30]. A general mechanism for the growth of barite nanofilaments in AOT (sodium bis(2-ethylhexyl)sulfosuccinate) microemulsions, involving the irreversible fusion, unidirectional exchange, and coalescence of microemulsion droplets, followed by crystallization of an amorphous filamentous BaSO4/surfactant phase, is described.

176

Wang et al.

Li M et al. [31], described synthesis of ordered microarrays of nanocrystals of barium chromate in microemulsions of water and AOT. Guo L et al., prepared Sb2O3 and Sb2O5 nanorods by using the system of AOT–water–toluene [32]. Micelles and microemulsions usually dominate at low surfactant concentration (e.g., 0.1 M) while lyotropic liquid crystalline phase can be obtained at high surfactant concentrations. Attard and co-workers demonstrated that normal hexagonal liquid crystalline phase could template bulk mesoporous SiO2 by hydrothermal synthesis. And Cu2O nanowires have been electrodeposited from AOT reverse hexagonal crystalline liquid phases [33]. In the template-confined synthesis approaches mentioned above, it seems that, in case of hard-template synthesis the main driving force for the growth of nanowires can be attributed to the space-confined effect, however, as far as soft-template-assisted synthesis is concerned, especially microemulsion, reverse micelles, etc, there still lack of a general comprehension on the formation process of nanowires, and among most of the work presented, further investigation such as light scattering experiment were absent but seems to be quite required to show the nanostructures of micelles to clarify the detailed formation mechanism.

3. Solution-Based Synthetic Way to Oxides 1-D Nanostructures Solution-based route have been developed to prepare the one-dimensional nanostructures of transition metal oxides, especially the solution-based redox reactions. Since there is no involvement of catalysts or templates to serve as energetic favorable sites for the absorption of reactant molecules or to confine the growth direction of nanowires, it would be reasonable to imagine that they are based on the anisotropic growth nature of the corresponding products. Hence the crystal structure of the oxides or the possible intermediates may be the underlying factors that will determine the formation of one-dimensional nanostructures. Also, the influence of ion concentrations cannot be neglected, which may have different influence on the growth rate of different crystal faces. One of the most important strategies for the solution-based routes is that layer (natural or artificial) structures may roll into 1-D nanowires (nanotubes or nanorods) under elevated temperature and pressure, and the as-obtained tubular structures may serve as the original driving force for the growth of 1-D nanostructures. This inspiration may originate from the synthesis of inorganic nanotubes. Nanotubes such as Carbon, BxCyNz [35–38], MS2 (M ⫽ Mo, W, Nb, Ta etc) [2, 39], NiCl2 [40], vanadium oxide [41] and Bi [42] have been synthesized under favorable conditions. A rational assumption is now that the graphite sheets can be rolled up into tubes, other inorganic layered structure materials might also do this. It is reasonable since the layers are held together mainly by weak van der waals’ forces. A clear understanding about the bending of the graphite under high temperature or electron beam irradiation has been elucidated [43, 44] which further strengthens the possibility of the rolling process of a lamellar structure. Several layer structures of binary or ternary oxides have been investigated by Li et al., such as NaxH2 ⫺ xTi3O7, MoO3, ␦-MnO2, K4Nb6O17, etc, and artificial lamellar

Nanowires and Nanotubes of Complex Oxides

177

structures of VOx-surfactant and WO3⫺x have also been employed to prepare nanotubes or nanowires structures. Another important strategy is that crystal structures with a preferential axis may be preferable for the growth of oxides nanowires, for example, hexagonal ZnO, Mg(OH)2, lanthanide hydroxide nanowires have been successfully synthesized through a hydrothermal/solvothermal synthetic way. 3.1. Synthesis of titanate nanotubes [45] Typical sheet-like structures of titanates were shown in Fig. 1. Its delamination [46], intercalation [47–49], and layer-by-layer assembly properties [50] further confirmed the layer structures. Single crystal nanotubes of titanate have been obtained through the following reactions: 3TiO2 ⫹ 2NaOH → Na2Ti3O7 ⫹ H2O Under hydrothermal conditions, titania (rutile or anatase) will be converted into titanate first, then the as-obtained layer structures may roll into nanotubes of titanate when the cations between the layer structures have been moved away and the interlayer interaction of lamellar intercalate have been diminished from the edges. Since certain amount of Na or H cations exists between the layer structures, the functionalization of the titanate could be expected by replacing the cations with transitional metal ions, such as Co2⫹, Ag⫹, etc. In Co substituted titanate nanotubes, HRTEM revealed that the layered structure was still evident with almost the same interlayer distance. EDS spectrum of Co-NT shown in Fig. 2 indicated that the atomic ratio of Co to Ti was 11% : 38%. The amount of Co might be manipulated by adjusting the initial amount and concentration of Co ions, which explored a simple and effective route to prepare complex metal oxides nanotubes and may inspire

Fig. 1. Some typical structures of titanates: (a) Crystal structure of Cs2Ti2O5; (b) Crystal structure of K2Ti5O11; (c) Crystal structure of K2Ti6O13.

178

Wang et al.

Fig. 2. (a) HRTEM images of titanate nanotubes; (b) HRTEM images of Co-substituted titanate nanotubes; (c) EDS analysis of Co-NT: characteristic peaks for both Co and Ti are observed. The atomic ratio of Co and Ti is 11 : 38.

investigations in the chemical, physical and catalytic properties of the complex transition metal nanoporous materials. The thermal stability of the titanate nanotubes was illustrated by comparing the XRD patterns of the raw hydrous sodium titanate nanotubes (Fig. 3a) and that of the samples after calcinations at 550 ⬚C (Fig. 3b). The essentially same patterns indicated the maintenance of crystal structure of the titanate framework without degradation at 550 ⬚C. The TEM and HRTEM observations also indicated the maintenance of the tubular structures. When the nanotubes were calcined to 600 ⬚C, a new phase of titanate, Na2Ti9O19 formed (Fig. 3d), and the tubular structures disappeared. When the nanotubes were calcined to 850 ⬚C, Na2Ti9O19 transformed to a mixture of Na2Ti6O13 and TiO2 (Fig. 3e). The crystal structure of titanate framework was stable at 550 ⬚C (XRD). Since few mesoporous materials were able to keep their structure without degradation under such conditions, thus-obtained titanate nanotubes may be a perfect substitute of them. 3.2. Synthesis of manganese oxides nanowires [51–53] Another example of rolling into 1-D nanostructures has been observed in the synthesis of manganese oxides nanowires. MnO2 exists in many polymorphic forms

Nanowires and Nanotubes of Complex Oxides

179

Fig. 3. The XRD patterns of the titanate nanotubes and their derivatives after calcinations at different temperatures: (a) the raw titanate nanotubes synthesized through hydrothermal method at 140 ⬚C for two days, which could be indexed to H2Ti2O5 · H2O; (b) the sodium titanate nanotubes after calcination at 550 ⬚C, which is very near to the former, indicating the maintenance of crystal structure; (c) the Co substituted nanotubes, almost the same as plots a, in which no reflections corresponding to titanium oxide, cobalt oxide or hydroxide were observable; (d) the sodium titanate nanotubes calcinated at 600 ⬚C, which could be indexed to Na2Ti9O19 (marked with ‘^’), indicating the degradation of nanotube structure; (e) the sodium titanate nanotubes calcinated at 850 ⬚C, in which a mixture of Na2Ti6O13 and anatase TiO2 was observed and marked with ‘O’ (Na2Ti6O13) and ‘A’ (anatase), respectively; (f) the Co substituted titanate nanotubes calcinated at 850 ⬚C, in which a mixture of CoTiO3 and TiO2 (anatase and rutile) was observed and marked with ‘*’ (CoTiO3), ‘A’ (anatase), and ‘R’ (rutile).

Fig. 4. Layer structures of ␦-MnO2.

such as ␣, ␤, ␥ and ␦-type. Different forms of MnO2 are exactly based on the same structural units [MnO6] octahedral. ␣, ␤, ␥-MnO2 have [MnO6] chains in their structures, and are typical of 1 ⫻ 1, 2 ⫻ 2, 1 ⫻ 2 one-dimensional channels, respectively, while ␦-type is well known for its layer structure, which is composed of edge-sharing [MnO6] octahedral [54]. Among the several crystallographic forms of MnO2, ␦-MnO2 alone has a layer structure (Fig. 4), which is indispensable in the rolling mechanism.

180

Wang et al.

(a)

(b)

Fig. 5. (a) TEM images of ␣-MnO2 nanowires; (b) TEM images of ␤-MnO2 nanorods.

So studies have been focused on the roles ␦-MnO2 playing in the formation of the other structural MnO2 1-D nanostructures. Two strategies of Mn2⫹ oxidized into MnO2 and MnO4⫺ reduced into MnO2 have been employed: Mn2⫹ ⫹ H2O → MnO2 ⫹ 4H⫹ ⫹ 2e ⫹ MnO⫺ 4 ⫹ 4H ⫹ 3e → MnO2 ⫹ 2H2O

(E0 ⫽ 1.23 V)

(1)

(E0 ⫽ 1.695 V)

(2)

(NH4)2S2O8 has been used as the oxidizing reagent for reaction (1) to proceed. Under hydrothermal conditions, ␣, and ␤-MnO2 have been obtained by varying 2⫺ the concentration of NH ⫹ 4 and SO 4 . The final products of direct reaction of (NH4)2S2O8 (0.008 mol) and MnSO4 (0.008 mol) is ␤-MnO2 (Fig. 5b) however, if the direct reaction occurs at 90 ⬚C, 1 atm, ␥-MnO2 nanorods will be obtained. The phase will change to ␣-MnO2 (Fig. 5a) when certain amount of (NH4)2SO4 (e.g., 0.02 mol) is introduced to the reaction system. KMnO4 has also been used as oxidizing reagents of MnSO4. The molar ratio of KMnO4 and MnSO4 were varied in order to prepare MnO2 nanorods with different crystallographic forms. ␦-MnO2 (JCPDS 80-1098) nanorods have been obtained when pure KMnO4 (0.0044 mol) or a high mole ratio (around 6 : 1) KMnO4/MnSO4 mixture is hydrothermal treated at 140 ⬚C. When the mole ratio of KMnO4/MnSO4 is controlled at about 2 : 1, the products are examined to be ␣-MnO2 nanorods with diameters 20~80 nm and lengths ranging between 2 and 6 ␮m (Fig. 6a). When the mole ratio is adjusted to around 2 : 3, the reaction will result in ␤-MnO2 nanorods with diameters 40~100 nm and lengths 0.5~1.0 ␮m (Fig. 6b).

Nanowires and Nanotubes of Complex Oxides (a)

181

(b)

Fig. 6. (a) TEM images of ␣-MnO2 nanorods; (b) TEM images of ␤-MnO2 nanorods.

Fig. 7. (A) XRD patterns of the ␣-MnO2 intermediate ((NH4)2S2O8-MnSO4 system) after hydrothermal treatment for 2 hr; (B) XRD patterns of the ␣-MnO2 intermediate ((NH4)2S2O8MnSO4 system) after hydrothermal treatment for 30 min.

Although the Mn sources are different, all the MnO2 nanowires have undergone a similar growing process. As is shown in Fig. 7B, the XRD patterns of ␣-MnO2 in a (NH4)2S2O8–MnSO4 reaction system, taken after 30 min hydrothermal treatment, can be readily indexed to that of ␦-MnO2. All the samples show lamellar structure morphologies (Fig. 8a). When the reaction time is prolonged, nanotubes have been observed existing in the intermediate (Fig. 8b). However, after a further prolonged reaction time (1.5 hr), the nanotubes will disappear, and only nanowires can be obtained, and the XRD patterns (Fig. 7a) of the samples have shown a great similarity to that of ␣-MnO2 with a less crystallinity. Similar XRD patterns of ␦-MnO2 have also been obtained from the intermediate (after 1 hr) of ␣-MnO2 in a KMnO4–MnSO4 reaction system or intermediate of

182

Wang et al. (a)

(b)

Fig. 8. (a) TEM images of the ␣-MnO2 intermediate ((NH4)2S2O8–MnSO4 reaction system) after hydrothermal treatment for 30 min; (b) TEM images of the ␣-MnO2 tubular intermediate ((NH4)2S2O8–MnSO4 reaction system) after hydrothermal treatment for 45 min.

␦-MnO2 (KMnO4 hydrothermal for 2 hr). TEM images of the samples also take on a look of lamellar and curling structures. All the information seems to indicate that although ␣-MnO2 has been got from different synthetic ways, they have undergone a similar growth process. Further studies have also evidenced a rolling process for the growth of ␤-MnO2 nanorods ((NH4)2S2O8–MnSO4 system as well as KMnO4–MnSO4 reaction system), however, a different phase transformation process: ␦ → ␣ → ␤ ((NH4)2S2O8–MnSO4 system), ␦ → ␥ → ␤ (KMnO4–MnSO4 reaction system). Based on the above experiment results, all the MnO2 1-D nanostructure have undergone a common developing process, which is characteristic of rolling mechanism and phase transformation. From the synthesis of titanate nanotubes and manganese oxides nanowires, together with other non-oxides compounds such as Bi nanotubes [42], it can be deduced that rolling mechanism may be a general way for the synthesis of nanowires from natural lamellar structures. Also, as an example, the successful synthesis of manganese dioxides nanowires/ rods have shown that, through a solution-based way to oxides of nanowires, it is convenient to control the crystal structure or oxidation state of the final products by properly selecting the reducing or oxidizing reagents and the reaction conditions, which may be the major advantages of solution-based method. 3.3. From artificial lamellar structures to one-dimensional nanostructures of oxides [55] Although we can find many layer structures in the oxides, however, so far not all of them can be transformed to one-dimensional nanostructures: partly because the interaction between layer and layer may be rather strong and will hold the layer tightly. A general method has also been set up in the synthesis of VOx nanotubes and WO3⫺x nanorods by preliminarily preparing an artificial lamellar structure. The primary approach was based on self-assembling of inorganic precursors at the template-solution interface using organic molecules as structure-directing agents. The interaction between organic molecules and inorganic precursors could be coordinative

Nanowires and Nanotubes of Complex Oxides

183

interaction [56–58], electrostatic interaction [59], or even hydrogen bonding [60]. Under certain conditions, the interlayer interaction of this kind of lamellar intercalates could be diminished from the edges. Then, the rolling of the layers into the tubules would take place. Since V2O5 can be regarded as a layered structure in which VO5 square pyramids are connected by sharing corners and edges and thereby form the layers, NH4VO3 and suitable organic surfactants have been used to form lamellar structure. In these compounds, the surfactants were coordinatively (amines) or electrostaticly bound (ammoniums) via their headgroups to discrete inorganic precursor units (VO⫺3 ). Then relative stable organic-inorganic layered structures could be formed with the surfactants intercalated between VO5, which could be identified through XRD patterns as well as the TEM images of the intermediates. Different surfactants and inorganic precursors might also do this work. For example, vanadium triisopropoxide can afford V atoms (although comparatively expensive), and varieties of amine molecules and surfactants such as CTAB (cetyltrimethyl ammonium bromide) could be used as templates. Moreover, there are no such limits [61–62] in the chain lengths of surfactans as predicted. For example, n ⫽ 6 could be selected in the case of aliphatic ␣,␻-diamines templates against the vanadium alkoxide route where the chain length could only extend from n ⫽ 14 to 20. This offered perspectives for more flexible structures. In the case of octadecylamine as template, it was note worthy that there were two kinds of layered distances, which could be identified by power X-ray diffraction pattern of the vanadium oxide nanotubes (Fig. 9c). The peak with the highest

Fig. 9. (a) TEM images of VOx nanotubes; (b) HRTEM images of VOx nanotubes; (c) XRD patterns of VOx nanotubes.

184

Wang et al.

intensity at 2 ⫽ 2.15 corresponding to a d value of 4.2 nm reflects one kind of layer distance and the peak with the highest intensity at 2 ⫽ 2.51 corresponding to a d value of 3.4 nm reflects the other kind of layer distance. Two kinds of layer distances could even been found in one nanotube as marked in Fig. 9b. The layer distances of nanotubes were smaller than twice the calculated maximum lengths of the templates, which imply that overlap of the alkyl chains and/or tilt angles must be present. The difference of the layer distances may be attributed to the difference of the overlap or/and tilt angles. These interesting findings, considering the generality of the reactant and surfactants, have proved that the formation mechanism of this kind of artificial lamellar structure may be generally applied, which is critical for a general synthetic way. If, under certain conditions, the interlayer interaction of this kind of lamellar intercalates could be diminished from the edges, rolling of the layer into tubules should be expected. The above mentioned scheme for the formation of the vanadium oxide nanotube could be divided into three main steps: (1) The surfactant molecules condensed into aggregations with VO ⫺ 3 to form the structure like V2O5. (2) When treated under hydrothermal conditions, the condensation process continued and brought out more ordered lamellar assemblies. (3) These lamellar sheets began to loose at the edges and then rolled into themselves to finally form vanadium oxide nanotubes. The above process was schematized in Fig. 10.

Fig. 10. Schematic presentation of the whole rolling mechanism for the formation of the vanadium oxide nanotubes: (a) the mixture of the NH4VO3 and the surfactants; (b) layered structures formed through the hydrothermal treatment; (c) the beginning stage of the rolling process; (d) the formed nanotubes.

Fig. 11. TEM images of WOx nanowires.

Nanowires and Nanotubes of Complex Oxides

185

Following a similar process, WO3⫺x nanowires have also been obtained [63]. In a typical synthesis, meso-lamellar composites of tungsten oxide with cetyltrimethylammonium (CTA⫹) surfactant cations (WO-L) was prepared by the reaction of sodium tungstate (Na2WO4) and cetyltrimethylammonium bromide (CTAB) under mild hydrothermal conditions in 85 ⫾ 90% yield. The lamellar nature of WO-L was confirmed by the low-angle X-ray powder-diffraction (XRD) pattern. However, merely the hydrothermal treatment cannot provide enough force to broke the interaction between layers from the edges, and no nanotubes or nanowires could be obtained, so a vacuum pyrolysis process has been used to offer the energy. The temperature gradient and gas pressure during the thermal pyrolysis reaction should be properly controlled, otherwise the precursors would crack into platelets or particles. As shown in Fig. 11, pure phase nanorods of WO3 ⫺ x nanorods were obtained. Based on the above experimental results, a novel and relatively simple synthesis method for oxides nanowires have been developed and a formation mechanism is proposed, which are somewhat similar with the rolling mechanism proposed for the forming of WS2 nanotubes and W nanowires. These demonstrated that the rolling mechanism may be a general mechanism for growth of nanotubes/nanorods from layered structure and it is expected that a general synthetic method of nanotubes may be developed based on this mechanism. 3.4. Synthesis of nanobelts via a solution-based route [64, 65] Solution-based ways have also been evidenced to be an effective way to prepare oxides nanobelts, this is particularly interesting due to, for example, that the previous semiconducting oxides or other kinds of nanobelts are usually obtained from thermal evaporation process under rather high temperature conditions, which may be costly and energy-consuming. As shown in Figs. 12 and 13, nanobelts of MoO3 as well as Titanate have been prepared by properly controlling the reaction conditions.

Fig. 12. (a) TEM image showing the morphologies of the MoO3 nanobelt; (b) Crystalline structure of the orthorhombic structured MoO3. The dark balls represent atoms of Molybdenum, and the light balls correspond to the atoms of oxygen.

186

Wang et al.

Fig. 13. (a) Typical TEM image of sodium titanate nanobelts; (b) TEM image of a twisted sodium titanate nanobelts.

The synthesis of the MoO3 nanobelts was based on the preparation of MoO3·H2O solution and subsequent treatment of hydrothermal reaction at 140 for 1 day. Why can atoms be assembled into such a belt-like nanostructure? On the basis of the traditional perfect smooth face crystal growth model, a possible growth mechanism was proposed for the formation of the MoO3 nanobelts. The natural chemical character of MoO3 helps us understand the growth mechanism. The orthorhombic structured MoO3 (Fig. 12b) has a distinctive layered structure, which can seldom be found in binary oxides. One MoO3 monolayer is one layer of MoO6 distorted octahedra with corner sharing in the direction of a-axis and zig-zag edge sharing in the direction of c-axis. And two MoO3 monolayers have the structure in which one of the slabs oriented perpendicular to the b-axis. As shown in Fig. 12b, the bonding situations for MoO6 along the [100] direction are substantially different from the [001], which will result in different chemical reactivities. Two Mo-O bonds will form if the MoO6 octahedra add along the [001] direction (Fig. 12b), while only one Mo-O bond forms in case of the growth along the [100] direction. The corresponding energy release along the [001] direction was greater, although little difference do exist in the bond length. From the point of energy view, the nanobelts growing along the [001] direction are favored [66–69]. In the case of titanate nanobelts, it seems that they might have developed from the titanate nanotubes, for their synthetic routes are only different in their alkali concentration and temperature. And comparatively, the nanobelts formation conditions are a little more moderate than that for nanotubes. In a typical procedure for the synthesis of titanate nanobelts, 1.5 g TiO2 powders were dispersed in 40 Ml 4~10 mol/l KOH aqueous solution. The resulting suspension was hydrothermally treated at 160~180 ⬚C for 48 hours. This method has also been expanded to TiO2-NaOH system to generate Na2Ti4O9 nanobelts. The sodium ions in the nanobelts are exchangeable for a variety of inorganic cations with respect to structural degradation. Photoluminescence of the potassium titanate was first observed at room temperature with the excitation wavelength of 401 nm and emission wavelength of 612 nm. The chemiluminescence was detected on the potassium titatnate nanobelts at 270 ⬚C, which is lower than that of TiO2 nanopowders by 140 ⬚C. These results show that this kind of titanate nanobelt may be expected as a promising material in selective catalysis, photocatalysis, gas sensor, and so on.

Nanowires and Nanotubes of Complex Oxides

187

3.5. Synthesis of Lanthanide hydroxide nanowires [70] The synthesis of Ln(OH)3 nanowires was based on the preparation of colloidal Ln(OH)3 at room temperature, and the subsequent hydrothermal treatment at 180 ⬚C for about 12 hr. TEM images of a La(OH)3 (Fig. 14a) nanowires product shows entirely uniform nanowires morphologies with diameters 15~20 nm and lengths up to 5 ␮m. It is interesting to find that, under controlled experimental conditions, the different lighter Ln(OH)3 (La(OH)3, Pr(OH)3 (Fig. 14b), Nd(OH)3, Sm(OH)3 (Fig. 14c), Eu(OH)3, Gd(OH)3) nanowires could be prepared as high aspect ratios products with similar outlook, while, under the same conditions, heavier lanthanide hydroxide (Dy(OH)3, Tb(OH)3, Ho(OH)3, Tm(OH)3 and YbOOH) usually have lower aspect ratios or less uniform morphologies. As far as the obtained Ln(OH)3 nanowires are concerned, they all have a hexagonal crystal structure, just like that of ZnO, which is well known for its anisotropic growth nature. With the decreasing of the ion radii (from La to Lu), under the adopted experimental conditions, the crystal structure of Lanthanide hydroxides gradually changes from a typical hexagonal phase (La(OH)3) to a monoclinic one (Lu(OH)3), which results in the formation of monoclinic YbOOH instead of hexagonal Yb(OH)3 in our experiment; along with this change, the tendency to grow along certain direction has been weakened to some extent, so the heavier Lanthanide hydroxides usually have lower aspect ratios and less uniform morphologies. Further investigation can be carried out on the growth dynamics of Ln(OH)3 nanowires, which may serve as a perfect model to study the crystallization in nanoscale since they have similar and gradually changing crystal structures. 3.6. Solution-based surfactant-assisted method to oxide nanorods [71, 72] Surfactant molecules in microemulsions or reverse micelles systems can act as soft-template in the synthesis of oxides of nanowires, however, not all of them work in the same way. With the assistance of CTAB, ZnO nanorods [71] have been prepared, in which surfactant CTAB plays an important role in accelerating the

(a)

(b)

(c)

Fig. 14. (a) TEM images of La(OH)3 nanowires; (b) TEM images of Pr(OH)3 nanowires; (c) TEM images of Sm(OH)3 nanowires.

188

Wang et al.

hydrothermal oxidation and guiding the growth direction, but not serving as microreactors. Hydroxylapatite (HAP) nanorods [72] have also been obtained in presence of CTAB or SDS (sodium dodecyl sulfate). The behavior of surfactant molecules in the synthesis of inorganic nanorods is considered to correlate with the charge and stereochemistry properties of reactants. To what extent the surfactant molecules will influence the growth of nanorods is rather interesting and may deserve a further study.

4. Prospects Although oxides of nanowires have been prepared through many kinds of methods, a general method, especially a general formation mechanism, is still on the way and need to be seeking for by more studies. Most importantly, further investigation can be focused on the quality of nanowires or nanowire arrays, for example, how to produce well-aligned nanowires with uniform morphology and perfect crystallinity that can satisfy the requirement of applications in electronic or sensor field.

References 1. S. Iijima, Nature 354 (1991) 56. 2. R. Tenne, L. Margulis, M. Genut and G. Hodes, Nature 360 (1992) 444. 3. T. J. Trentler, K. M. Hickman, S. C. Geol, A. M. Viano, P. C. Gibbons and W. E. Buhro, Science 270 (1995) 1791. 4. H. J. Dai, E. W. Wong, Y. Z. Lu, S. S. Fan and C. M. Lieber, Nature 375 (1995) 769. 5. W. Q. Han, S. S. Fan, Q. Q. Li and Y. D. Hu, Science 277 (1997) 1287. 6. G. Fasol, Science 280 (1998) 545. 7. Y. Zhang, K. Suenaga, C. Colliex and S. Iijima, Science 281 (1998) 973. 8. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. 9. M. Law, H. Kind, B. Messer, F. Kim and P. D. Yang, Angew. Chem. Int. Ed. 41 (2002) 2405. 10. M. H. Huang, S. Mao, H. Feick, H. Q. Yan, Y. Y. Wu, H. Kind, E. Weber, R. Russo and P. D. Yang, Science 292 (2001) 1897. 11. X. F. Duan and C. M. Lieber, J. Am. Chem. Soc. 122 (2000) 188. 12. X. F. Duan, J. F. Wang and C. M. Lieber, Appl. Phys. Lett. 76 (2000) 1116. 13. X. F. Duan and C. M. Lieber, Adv. Mater. 12 (2000) 298. 14. A. M. Morales and C. M. Lieber, Science 279 (1998) 208. 15. P. D. Yang and C. M. Lieber, Science 273 (1996) 1836. 16. M. H. Huang, Y. Y. Wu, H. Feick, N. Tran, E. Weber and P. D. Yang, Adv. Mater. 13 (2001) 113. 17. D. P. Yu, Q. L. Hang, Y. Ding, H. Z. Zhang, Z. G. Bai, J. J. Wang, Y. H. Zhou, W. Qian, G. C. Xiong and S. Q. Feng, Appl. Phys. Lett. 73 (1998) 3076. 18. H. Z. Zhang, Y. C. Kong, Y. Z. Wang, X. Du, Z. G. Bai, J. J. Wang, D. P. Yu, Y. Ding, Q. L. Hang and S. Q. Feng, Solid State Commun. 109 (1999) 677. 19. B. Zheng, Y. Y. Wu, P. D. Yang and J. Liu, Adv. Mater. 14 (2002) 122. 20. Z. G. Bai, D. P. Yu, H. Z. Zhang, Y. Ding, Y. P. Wang, X. Z. Gal, Q. L. Hang, G. C. Xiong and S. Q. Feng, Chem. Phys. Lett. 303 (1999) 311. 21. H. Z. Zhang, Y. C. Kong, Y. Z. Wang, X. Du, Z. G. Bai, J. J. Wang, D. P. Yu, Y. Ding, Q. L. Hang and S. Q. Feng, Solid State Commun. 109 (1999) 677. 22. Z. W. Pan, Z. R. Dai, C. Ma and Z. L. Wang, J. Am. Chem. Soc. 124 (2002) 1817.

Nanowires and Nanotubes of Complex Oxides

189

23. S. J. Limmer, S. Seraji, M. J. Forbess, Y. Wu, T. P. Chou, C. Nguyen and G. Z. Cao, Adv. Mater. 13 (2001) 1269. 24. M. Zheng, D. S. Xu et al., Nano Lett. (in press). 25. Y. J. Zhang, J. Zhu, Q. Zhang, Y. J. Yan, N. L. Wang and X. Z. Zhang, Chem. Phys. Lett. 317 (2000) 504. 26. B. C. Satishkumar, A. Govindaraj, M. Nath and C. N. R. Rao et al., J. Mater. Chem. 10 (2000) 2115. 27. S. Mann, J. Chem. Soc. Dalton Trans. 7 (1997) 3953. 28. S. Oliver, A. Kuperman, N. Coombs, A. Lough and G. A. Ozin, Nature 378 (1995) 47. 29. D. Walsh, J. D. Hopwood and S. Mann, Science 264 (1994) 1576. 30. J. D. Hopwood and S. Mann, Chem. Mater. 9 (1997) 1819. 31. M. Li, H. Schnablegger and S. Mann, Nature 402 (1999) 393. 32. L. Guo, Z. H. Wu, T. Liu, W. D. Wang and H. S. Zhu, Chem. Phys. Lett. 318 (2000) 49. 33. L. M. Huang, H. T. Wang, Z. B. Zheng, A. Mitra, D. Y. Zhao and Y. S. Yan, Chem. Mater. 14 (2002) 876. 34. S. Kobayashi, N. Hamasaki, M. Sukuki, M. Kimura, H. Shirai and K. Hanabusa, J. Am. Chem. Soc. (in press). 35. N. C. Chopra, R. J. Luyken, K. Cherry, V. H. Crespi, M. I. Cohen, S. G. Louie and A. Zettl, Science 269 (1995) 996. 36. Z. W. Sieh, K. Cherrey, N. G. Chopra, X. Balse, Y. Miyamoto, A. Rubio, M. L. Cohen, S. G. Louie, A. Zettl and R. Gronsky, Phys. Rev. B 51 (1995) 11229. 37. O. Stepphan, P. M. Ajayan, C. Colliex, P. Redich, J. M. Lambert, P. Bernier and P. Lefin, Science 266 (1994) 1683. 38. A. Zettl, Adv. Mater. 8 (1996) 443. 39. T. Tenne, Adv. Mater. 7 (1995) 965. 40. Y. R. Hachohen, E. Grunbaum, J. Sloan, J. L. Hutchison and R. Tenne, Nature 395 (1998) 336. 41. M. E. Spahr, P. Bitterli, R. Nesper, M. Müller, F. Krumeich and H.-U. Nissen, Angew. Chem. Int. Ed. 37 (1998) 1263. 42. Y. D. Li, J. W. Wang, Z. X. Deng, Y. Y. Wu, X. M. Sun, D. P. Yu and P. D. Yang, J. Am. Chem. Soc. 123 (2001) 9904. 43. R. D. Heidenreich, W. M. Hess and L. L. Ban, J. Appl. Cryst. 1 (1968) 119. 44. D. Ugarte, Nature 359 (1992) 707. 45. X. M. Sun and Y. D. Li, Novel Thermally Stable and Ion-exchangeable Titanate Nanotubes (submitted to Chem Comm). 46. T. Sasaki, M. Watanabe, H. Hashizume, H. Yamada and H. Nakazawa, J. Am. Chem. Soc. 118 (1996) 8329. 47. M. Yanagisawa, S. Uchida, S. Yin and T. Sato, Chem. Mater. 13 (2001) 174. 48. C. Airoldi, L. M. Nunes and R. F. Farias, Mater. Res. Bull. 35 (2000) 2081. 49. M. W. Anderson and J. Klinowski, Inorg. Chem. 29 (1990) 3260. 50. S. W. Keller, H. N. Kim and T. E. Mallouk, J. Am. Chem. Soc. 116 (1994) 8817. 51. X. Wang and Y. D. Li, J. Am. Chem. Soc. 124 (2002) 2880. 52. X. Wang and Y. D. Li, Chem. Commun. (2002) 764. 53. X. Wang and Y. D. Li, Synthesis and Formation Mechanism of Manganese dioxide Nanowires/nanorods, Chem. Euro. J. (accepted). 54. M. M. Thackery, Prog. Solid State Chem. 25 (1997) 1. 55. X. Chen, X. M. Sun and Y. D. Li, Inorg. Chem. 41 (2002) 4524. 56. D. M. Antonelli and J. Y. Ying, Angew. Chem. Int. Ed. 34 (1996) 2014. 57. D. M. Antonelli and J. Y. Ying, Angew. Chem. Int. Ed. 35 (1996) 426. 58. D. M. Antonelli and J. Y. Ying, Inorg. Chem. 35 (1996) 3126.

190

Wang et al.

59. Q. S. Hue, D. I. Margolese, U. Ciesla, P. Y. Feng, T. E. Gier, P. Sieger, R. Leon, P. M. Petroff, F. Schuth and G. D. Stucky, Nature 368 (1994) 317. 60. T. T. Tanev and T. J. Pinnavaia, Science 267 (1995) 865. 61. F. Krumeich, H. J. Muhr, M. Niederberger, F. Bieri, B. Schnyder and R. Nesper, J. Am. Chem. Soc. 121 (1999) 8324. 62. M. E. Spahr, P. Bitterli, R. Nesper, M. Muller, F. Krumeich and H. U. Nissen, Angew. Chem. Int. Ed. 37 (1998) 1263. 63. X. L. Li and Y. D. Li, Large-scale 1D Tungsten Oxides Nanowires with High Aspect Ratio by a Novel Approach, Inorg. Chem. (accepted). 64. X. L. Li and Y. D. Li, Single crystal MoO3 nanobelts, Appl. Phys. Lett. (accepted). 65. X. M. Sun, X. Chen and Y. D. Li, Inorg. Chem. 41 (2002) 4996. 66. Z. Hussain, J. Mater. Res. 16 (2001) 2695. 67. M. Ferroni, V. Guidi, G. Martinelli, M. Sacerdoti, P. Nelli and G. Sberveglieri, Sens. Actuators B 48 (1998) 285. 68. H. C. Zeng, Inorg. Chem. 37 (1998) 1967. 69. M. Itoh, K. Hayakawa and S. Oishi, J. Phys-Condens. Mat. 13 (2001) 6853. 70. X. Wang and Y. D. Li, Angew. Chemie. Int. Ed. (accepted). 71. X. M. Sun and Y. D. Li, CTAB assisted hydrothermal orientation growth of ZnO nanorods Mater. Chem. Phy. (accepted). 72. L. Yan, Y. D. Li, Z. X. Deng, J. Zhang and X. M. Sun, Inter. J. Inorg. Mater. 3 (2001) 633.

Chapter 11 Silica Nanowires/Nanotubes Jing Zhu, W. X. Sun and Jun Luo Electron Microscope Laboratory, School of Materials Science and Engineering, Tsinghua University, Beijing 100084, P.R.China

Silica nanowire/nanotube is becoming more important among one-dimensional materials due to their promising applications in luminescence, emitting stable and high-brightness blue light, and catalyzing field. In this chapter, synthesis methods, characterizations of structure and property, applications of silica nanowires/nanotubes will be briefly introduced.

1. Synthesis of Silica Nanowires/Nanotubes Several methods have been developed to synthesize silicon oxide nanowires/ nanotubes. 1.1. Laser ablation method Excimer laser ablation for fabricating silica nanowire was, first time, introduced by Yu et al. [1, 2]. In this processing, an excimer laser high temperature deposition system was used; a disk-like target was formed by pressing a mixture of silicon powder (99% in purity) and 20 wt% of silica powder, together with 8 wt% of Fe powder added as catalysts. The target was placed inside a quartz tube pumped to 20 mTorr and heated at 1123 K for 4 hours; after 20 hours further heating at 1473 K, the target was ablated using an excimer laser of 246 nm in wavelength under flowing argon (99.999% in purity) at an ambient pressure of about 100 Torr; the laser beam was focused to a spot of 1 ⫻ 3 mm2 on the surface of the target, and the average energy was 350 mJ per pulse. The opaque-colored silica nanowires, as shown in Fig. 1, were deposited onto the interior wall of the quartz tube in the front of the water-cooled copper collector mounted downstream near the rear of the quartz tube.

192

Zhu et al.

Fig. 1. A TEM image revealing the morphology of the SiONWs. The inset shows the corresponding diffusive ring pattern of electron diffraction.

1.2. Chemical vapor deposition (CVD) and physical vapor deposition (PVD) evaporation methods [3–8, 39] Two very similar simple methods through the vapor–solid (VS) process for fabricating silica nanowires were developed by Zhu [3] and Masumoto [4] as well as Wang [39] research groups. 1.2.1. Simultaneously fabricated silicon and silica nanowires [4] An example, we introduce, is shown here. The equipment used for this purpose is schematically shown in Fig. 2. A quartz tube was mounted inside a high-temperature tube furnace. Three kinds of powder mixtures as starting materials, pure Si powder (99.999 wt%), Si and nanosized Fe powder (99.9 wt%), and Si mixed with SiO2 (99.999 wt%), were ball-milled for more than 12 hours, and then pressed into plates 13 mm in diameter at room temperature. The plate was placed in the center of the quartz tube, and then the tube was evacuated by a mechanical rotary pump to a pressure of 1 Pa. High-purity argon gas was then introduced to the quartz tube at a flow rate of 50 standard cubic centimeter per minute (sccm) and the total tube pressure was kept at about 80 Pa. The furnace temperature ranged from 1453 to 1523 K. The temperature and the Ar flow remained unchanged during the growth process of 3 hours. Silicon oxide wires with white wool alike were obtained on the surface of the mixed powder plate and the tube wall close to the plate (the region marked by A in Fig. 2), while silicon nanowires with yellow color were obtained near the cold finger of copper rod (the region marked by B in Fig. 2). The length of the SiO wire exceeds 0.8 mm and its average diameter varies from 70 nm to 1.35 ␮m. 1.2.2. Controlled growth for fabricating SiO2 and Si3N4 nanowires [3] The equipment in this processing is very similar to that schematically in Fig. 2. Here an Al2O3 tube replaced the quartz tube and Si/SiO2 mixture powder with or without Fe or Ni powder held in an Al2O3 boat replaced the pressed plate of mixed powders of Si, Si/Fe and Si/SiO2. The mixtures were heated at 1473 K for

Silica Nanowires/Nanotubes

193

Fig. 2. Schematic diagram of a high temperature oven system for synthesis of silicon and silicon oxide wires. Table 1. The compositions of the mixtures and the gases introduced [3] Experiment no. 1 2 3 4 5 6

Si (wt%)

SiO2 (wt%)

100 32 30.3 30.3 32 30.3

68 64.3 64.3 68 64.3

Fe/Ni (wt%)

5.4 5.4 5.4

Gas

Product

N2 N2 N2 Ar Ar NH3

Si3N4 nanowires and SiO2 amorphous nanowires Si3N4 nanowires and SiO2 amorphous nanowires Si3N4 nanowires and SiO2 amorphous nanowires SiO2 amorphous nanowires SiO2 amorphous nanowires Si3N4 nanowires and SiO2 amorphous nanowires, but yield is low

60 min. At the ambient pressure, a selected gas (N2 or NH3 or Ar) passed the mixtures at the rate of 500 sccm. After reactions, the products were found on the surface of the mixtures and the inner walls of the Al2O3 boats. Table 1 illustrates how to control species, crystallizability and yield of the reactive products via controlling composition of raw materials and introduced gases. It can be seen from this table that there was only a little effect of the metal catalyst on the growth of the nanowires and the metal catalyst is not necessary to add in this method. And there are no metal/Si alloy droplets on the tips of these nanowires as shown in Fig. 3, which suggests that the vapor– liquid–solid (VLS) mechanism does not work in this method and the most possible mechanism may be the vapor–solid (VS) process in this method. 1.3. Sol-gel and other chemical methods [9–22] In the preparation of silica nanotubes by sol-gel method, silica source and template materials are necessary. Tetraethoxysilicane (TEOS, Si(OC2H5)4) is mostly used for supplying high-purity silica source for synthesis of silica following the reactions: Si(OC2H5)4 ⫹ 4H2O → Si(OH)4 ⫹ 4C2H5OH Si(OH)4 → SiO2↓ ⫹ 2H2O

194

Zhu et al.

Fig. 3. The typical morphology of the nanowires.

Template materials are used to guide the formation of silica nanotubes, mainly including carbon nanotubes, filamentous carbons, surfactants and simple organic hydroxyl carboxylic acids. 1.3.1. Using carbon nanotubes as templates [9] A known quantity (3 ml) of TEOS was added to 50 mg of carbon nanotubes and stirred for 12 hours. The excess TEOS was decanted and the carbon nanotubes coated with TEOS were heated in a vacuum at 373 K, followed by further heating at 770 K for 12 hours. After this treatment, the carbon nanotubes coated by SiO2 are obtained. Then the SiO2-coated carbon nanotubes were heated in the air at 1023 K for 12 hours and the carbon was oxidized, leaving silica nanotubes. In order to prepare SiO2 nanotubes containing transition metal ions, 0.5 ml TEOS and a known quantity of the relevant transition metal compounds were added to a suspension of 50 mg carbon nanotubes in 2 ml ethanol. The resulting mixture was stirred for 12 hours and the excess solvent was decanted. The mixture was then centrifuged, and the solid obtained was dried and heated at 770 K for 12 hours. SiO2 nanotubes containing Cu and Cr (in oxidized forms) had been prepared by employing copper acetate monohydrate (50 mg) and CrO3 (60 mg), respectively. 1.3.2. Using catalytic filamentous carbons (CFC) as templates [10] TEOS was subjected to pre-hydrolysis in the presence of an acid and water in a sub-stoichiometric amount, and the obtained solution was introduced into CFC with NiO catalytic particles in the tips as a template. TEOS was condensed in CFC as the solvent was evaporated. The further thermal treatment resulted in the formation of solid silica-containing films on CFC. The carbon was eliminated (gasified as carbon dioxide) by means of an oxidative thermal treatment in the air at 873 K. SiO2-NiO nanotubes were formed. 1.3.3. Using surfactants as templates [11] A laurylamine hydrochloride (LAHC)/TEOS system was selected as an example. The experimental procedure was as follows: TEOS was added to 0.1 M LAHC aqueous solution (pH ⫽ 4.5), and the reaction was started in a stirred cell at 313 K. The

Silica Nanowires/Nanotubes

195

TEOS-to-LAHC molar ratio was adjusted to 4–12. TEOS did not dissolve in water to yield an emulsified solution in early stages of the reaction. After 2–3 hours, TEOS completely dissolved in aqueous solution due to the hydrolysis, and the solution became transparent. After 13–14 hours, the solution turned into a homogeneous gel state. Single cylindrical nano-aggregates were found and around 5 nm in diameter. The central parts of the cylindrical aggregates and the two peripheries were single silica nanotubes in which hydrocarbon chains were filled. 1.3.4. Using simple organic hydroxyl carboxylic acids as template materials [12] A citric acid (CA)/TEOS system was selected as an example. Silica nanotubes were synthesized at room temperature. Ethanol, water and TEOS, ammonia and citric acid monohydrate were used as chemical sources. In a typical synthetic procedure, 6.0 ml of TEOS was added to the mixed solution of 0.7 ml of H2O and 30 ml of EtOH; and then 12 ml of NH4OH (28% aqueous solution) containing 0.18 g of CA monohydrate were added dropwise into the homogenous solution in 2 hours under magnetic stirring with a speed of ca. 300 rpm. After the mixed gel was loaded quietly for another 2 hours, the precipitate was centrifugal washed with distilled water and dried in an oven at 1023 K; finally, the product was calcined at 773 K for 2 hours to obtain silica nanotube materials. The outer diameter of the nanotubes varied from 50 to 500 nm with the majority around 100–150 nm. Their length ranged from hundreds of nanometers up to tens of micrometers, and the thickness of their walls was ca. 30 nm. The head parts of the tubes were open. 1.3.5. Silica nanotubes from mesoporous silica materials [14–22] Much exotic morphology have been found for the mesoporous silica using acid synthesis including films, hollow spheres, fibers, hollow helicoids and gyroids [18–20]. However, the drawback of the acid-made mesoporous materials is that they are normally less stable, thermally and hydrothermally. A post-synthesis ammonia hydrothermal treatment has been introduced to improve the order and stability of the acid-made mesoporous silica while preserving the original morphology [21, 22]. The formation of silica nanotubes arises from the expansion of the mesostructure during the hydrothermal treatment. 1.4. Biaxial and coaxial silica–silicon carbide nanowires, and carbon–silica–silicon carbide nanowires [19–21] 1.4.1. Biaxial silica–silicon carbide nanowires [19] A biaxial silica–silicon carbide nanowire has been synthesized by modified high temperature (close to 1773 K for 12 hours in this work) techniques developed by Lee et al. [22]. Fig. 4 shows a biaxial SiC-SiOx nanowire, which consists of two side-byside sub-nanowires of -SiC and silica, and can be simply referred to as a composite nanowire. 1.4.2. Coaxial silica–silicon carbide nanowires [20] Coaxial silica–silicon carbide nanowires have been synthesized by a chemical method which involves three steps: (a) preparing silica xerogels containing carbon nanoparticles by a sol-gel technique based on the hydrolysis of tetraethyl or tetraethoxysilicane (TEOS) and dissolving saceharose in the sol, (b) a carbothermal

196

Zhu et al.

Fig. 4. High-resolution TEM image of a biaxially structured -SiC–SiOx nanowire (from Ref. [19]).

reduction of the silica xerogels with the carbon nanoparticles at a desired temperature (1923 K) in a short time to produce the inner -SiC nanorods, and (c) raising the temperature to a higher one (2073 K) steeply, leading the unconverted silica to decompose into SiO vapor and O2, which will combine with each other to form the outer coaxial thicker SiO2 nanorods during the cooling. In this composite nanowire, the center thinner crystalline nanowire is -SiC ranging 10–30 nm in diameter, while the outside thicker nanowire is an amorphous SiO2 with the outside diameter ranging 20–70 nm. 1.4.3. Coaxial C-silica--SiC nanowires [21] Using the same method as indicated in 1.2.2, under the following conditions: the raw materials are SiO2 powder and nanoscale C particles; the synthesis temperature is 1823 K; the carrier gas is Ar, coaxial C-silica--SiC nanowires with the diameters less than 50 nm can be obtained. The detailed characterization will be described in Section 2. 1.5. Silica nanostructure [23–26] A variety of silica nanostructures have been developed. 1.5.1. SiO2 nanowire “bundles” and “brush-like” nanofibers They have been synthesized by modified high temperature synthesis method [23]. 1.5.2. Wreath-like patterns A simple procedure to synthesize carbon-reinforced silica fibers on planar silicon substrates covered with a thin silica cap layer has been reported recently [24]. The

Silica Nanowires/Nanotubes

197

procedure contains two steps, which are first activating the substrate with ferrocene and then exposing it to methane at about 1373 K, which transforms the silica cap layer into nanowires of about 200 nm in diameter. The silica nanowires are reinforced by glassy carbon in the core, and have an optical band gap of about 3.1 eV, as well as form wreath-like patterns. 1.5.3. Comet-like patterns Through a thermal evaporation method by heating pure silicon powder at 1373 K followed by depositing silicon vapor on a quartz-glass substrate which was coated with a catalyst precursor from a 0.005 M Fe(NO3)3 aqueous solution, a bulk quantity of SiOx nanowires with a comet-like pattern can be obtained [25]. 1.5.4. Silica nanowires from liquid Ga–Si alloy [26, 40] In this method, a simple CVD system was equipped. Then a drop of liquid Ga was placed on a Si wafer freshly etched in dilute HF and placed in an alumina boat at the center of the tube furnace. The furnace was heated to 1193 K for 3 hours and slowly cooled to room temperature. In addition to the growth of GaN nanowires on the Si wafer with Fe catalyst, a silicon oxide thick layer of white hair-like with a few millimeters in thickness covered on a Ga ball as shown in Fig. 5(d), while NH3 was used as both a carrying gas and a reactant for the formation of GaN. Only SiOx nanowires were found, while both wet H2 and wet Ar but not while dry Ar were used as the carrying gas. It implies that moisture is an important factor for the growth of SiOx nanowires. The growth mechanism was discussed by Wang et al. [40]. The nanowire growth temperature is determined by the eutectic temperature of the alloy, which is usually far below the melting temperature of the metal catalyst. For example, VLS growth of Si nanowires can be carried out at temperature on the order of 673–773 K using Au (melting point, 1337 K) as the catalyst, because of the low eutectic temperature (636 K) of the Au–Si alloy. In Ga-0.006 mol % Si, the eutectic point is about 302.8 K, it is possible to synthesize Si or silica nanowires at low temperature by using Ga as catalyst which forms Ga–Si alloy with Si. The Ga-catalyzed VLS growth exhibits many interesting new growth phenomena. The use of Ga provides opportunities for development of low-temperature VLS routes for nanowire synthesis.

2. Characterization of Structures and Properties of Silica Nanowires/Nanotubes 2.1. Morphologies Either silica nanowires or silica nanotubes are amorphous with white wool alike and have smooth surface. Their diameters and shapes depend on synthesis conditions. Fig. 6(a) and (b) shows the image of SiO2 amorphous nanowires [21] and aligned SiO2 nanofibers produced on a silicon substrate, respectively. The inset in Fig. 6(b) shows an enlarged image [27].

198

Zhu et al. (a)

(b)

(c)

(d)

Fig. 5. Silica nanostructures. (a) “Brush-like” silica nanofibers [23]. (b) “Wreath-like” silica nanofibers with glassy carbon [24]. (c) “Comet-like” silica nanowires [25]. (d) Ultra-long highly oriented silica nanowires with a core of Ga ball [26]. (a)

(b)

Fig. 6. (a) SiO2 amorphous nanowires fabricated by heating Si/SiO2 solid in a flowing Ar gas at ambient pressure and 1373–1573 K [21]. (b) Aligned SiO2 nanofibers appearing like hair produced on silicon substrates [27].

199

Silica Nanowires/Nanotubes 2.2. Valences of Si and O in SiO2 nanowires—XPS analysis and composition analysis

Through XPS analysis, an evidence for the formation of SiO2 can be obtained. A set of XPS spectra from SiO2 nanowires synthesized by carbothermal reduction method is shown in Fig. 7 [7]. There are two strong peaks at 103.35 and 532.65 eV, which correspond to the binding energy of Si (2p) and O (1s) for SiO2, respectively. Usually, the atomic ratio of O to Si is larger than 2, because of the adsorption of oxygen. The survey spectrum in Fig. 7(a) also displays C (1s) peak (at 284.65 eV), which can be attributed to a small amount of the residual graphite (286.3 eV for C1s in graphite) and the contamination of surfaces. The composition detection for an individual or a bundle of SiO2 nanowires is possible to be realized by means of Energy Dispersive Spectrometer (EDS) and Electron Energy Loss Spectrometer (EELS) attached in transmission electron microscope. In a measurement for a pure sample, only silicon and oxygen can be detected. 2.3. XRD and electron diffraction measurements By using X-ray diffractions and electron diffractions, it can be found that SiO2 nanowires (or nanotubes, or nanofibers) are amorphous. The electron diffraction pattern from a SiO2 nanowire always shows the diffuse rings with the lattice parameters of SiO2. In an X-ray diffraction pattern, as shown in Fig. 8, there is a broad hump at about 2 ⫽ 21.3⬚, indicating that the silica nanowires are in an amorphous state. 2.4. Comprehensive analysis of a C/SiO2/SiC nanowire composite [21] Fig. 9 shows EELS and the structure of a C/SiO2/SiC nanowire composite. An individual nanowire composite marked by an arrow is shown in Fig. 9(a). To identify it, some measurements were performed on a STEM VG HB501 with a field emission gun operating at 100 kV and attached with a Gatan 666 paralleled-EELS spectrometer. To investigate the local chemical composition and electronic structures at different positions, a collection of spectra was acquired by scanning the sub-nanometer electron probe across the nanowire [29, 30]. Each individual spectrum was recorded with a 0.5 s acquisition time and a 0.5 eV energy resolution. The convergence angle and the collection angle were, respectively, 15 and 24 mrad, and a typical sequence of spectra was acquired by scanning the probe perpendicular to the axis of the nanowire. (a)

(b)

(c)

Fig. 7. XPS from SiO2 nanowires. (a) Survey spectrum. (b) Si2p binding energy spectrum. (c) O1s binding energy spectrum [7].

200

Zhu et al.

Fig. 8. The XRD pattern for silica nanofibers showing the characteristic form for an amorphous state [27].

(a)

(b)

(c)

Fig. 9. EELS and structure of a C/SiO2/SiC nanowire composite. (a) The core of the SiC nanowire marked by an arrow is a -SiC phase with one of the {200} planes perpendicular to its axis. (b) The profiles of C, O and Si along the direction perpendicular to a SiC nanowire’s axis. (c) A scheme of a SiC nanowire composite. The core is a -SiC crystalline phase; the interlayer and the sheath are amorphous SiO2 and carbon, respectively.

To show the chemical composition clearly, elemental profiles across a nanowire were obtained from a series of spectra; the different core-loss peaks were identified and the quantitative analysis was carried out. The K-edge weights were normalized for chemical profiling of all of the elements except for silicon for which the L-edge was used; therefore, the profiles represent the distribution of atomic concentrations (Fig. 9(b)). Once the series of the basis components were defined, the Non-Negative Linear Square (NNLS) method [31] was used for reconstructing every experimental spectrum as a linear combination of these references and to distinct the distribution of two different bonding states of silicon: SiC and SiO2 and two bonding states of C: SiC and amorphous C. Thus the structure of the nanowire composite can be illustrated as in Fig. 9(c).

Silica Nanowires/Nanotubes

201

Fig. 10. PL spectrum of the silica nanowires at room temperature under excitation at 260 nm [7].

2.5. Photo luminescence property As shown in Fig. 10 [7], in a typical photo luminescence spectrum from SiO2 nanowires at room temperature under excitation at 260 nm, a stable and strong blue light emission is revealed at 435 nm (2.85 eV) while ultraviolet and blue light emissions at 350 nm (3.54 eV), 420 nm (3.0 eV) and 465 nm (2.7 eV) can also be observed. The shoulder at 465 nm (2.7 eV) is similar to that in some bulk SiO2 ascribed to neutral oxygen vacancy [6]. The 3.0 eV (420 nm) band corresponds to some intrinsic diamagnetic defect centers [30]. The ultraviolet light emission at 350 nm (3.54 eV) was also observed in the oxidized porous silicon and the annealed SiO2 [31]. 2.6. Catalytic activity Silica nanotubes possess a high specific surface area. As the support of catalysts, silica nanotubes make sufficient surface areas of catalysts expose to reactants in order to improve activity of catalysts; thus silica nanotubes containing metal oxide seem to be a promising catalyst. 2.6.1. Catalytic activity of silica aerogels to direct oxidation of hydrogen sulfide into sulfur under stoichiometric condition [9, 10] In this section, we introduce the research work reported by a Russian research group and an Indian research group [9, 10]. They took the aerogels of SiO2–NiO nanotubes containing 7% of nickel oxide and nickel oxide-free aerogels as examples to demonstrate their catalytic activity on direct oxidation of hydrogen sulfide into sulfur. The aerogel of SiO2–NiO nanotubes with nickel constituent was found to have much higher activity than the nickel oxide-free aerogel. It can be seen from Figs. 11 and 12 where the hydrogen sulfide conversion at 2000 ⬚C is 20% for the aerogel of SiO2–NiO nanotubes against 5% for the nickel oxide-free aerogel; the 100% selectivity to sulfur is observed with both samples.

202

Zhu et al.

Fig. 11. Conversion of hydrogen sulfide (■) and selectivity to sulfur (•) versus temperature. The aerogel was synthesized with CFC as the template (SiO2–NiO nanotube containing 7% of nickel oxide); inlet gas mixture 1% H2S ⫹ 0.5% O2 ⫹ N2, consumption of gas mixture 60 cm3/min.

Fig. 12. Conversion of hydrogen sulfide (■) and selectivity to sulfur (•) versus temperature. The aerogel was synthesized with sibunit – a pure mesoporous carbon materials [10] as the template (SiO2 nanotube with nickel oxide-free); inlet gas mixture 1% H2S ⫹ 0.5% O2 ⫹ N2, consumption of gas mixture 60 cm3/min.

2.6.2. Catalytic activity of aerogels to direct oxidation of hydrogen sulfide into sulfur in excess oxygen The activities of the aerogel of SiO2–NiO nanotubes at both stoichiometric and excess oxygen conditions are much higher than the nickel oxide-free aerogel. In Fig. 13, the selectivity is 100% and the conversion is 39% for nickel oxide-free aerogel at 448 K; as to the aerogel of SiO2–NiO nanotubes, the selectivity is 100% and the conversion is 73%, respectively, at 433 K (see Fig. 14). The catalyst activity and the selectivity do not change during 20 hours operation under these conditions. It is emphasized that if the long duration of the reaction is achieved by maintaining the reaction temperature as high as to provide the evaporation of sulfur from the reaction zone, a high conversion is attained at the absence of SO2 among the reaction products. However, conventional high-active catalysts without SiO2, such as nickel oxide, copper oxide and active carbons, produce SO2 in a considerable amount at these temperatures, which is why they must work at lower temperature (⬃373 K); as a result, sulfur is deposited into the catalyst pores to cause a decrease in the catalytic

Silica Nanowires/Nanotubes

203

Fig. 13. Conversion of hydrogen sulfide (▲) and selectivity to sulfur (•) versus temperature. The SiO2 aerogel was nickel oxide-free; inlet gas mixture 0.5% H2S ⫹ 16% O2 ⫹ N2, consumption of gas mixture 60 cm3/min.

Fig. 14. Conversion of hydrogen sulfide (▲) and selectivity to sulfur (•) versus temperature. The SiO2 aerogel was SiO2–NiO nanotube containing 7% of nickel oxide; inlet gas mixture 0.5% H2S ⫹ 16% O2 ⫹ N2, consumption of gas mixture 60 cm3/min.

activity. For the case, in a SiO2–NiO system and a higher reaction temperature (⬎433 K), no sulfur deposition occurs in the catalyst pores. 2.7. Measurement of mechanical properties of nanowires by in situ (Transmission Electron Microscopy) (TEM) Here we introduce an in situ TEM method [32, 33] for measurement of mechanical properties of nanowires developed by Zhu’s research group. A reversible bending phenomenon of Si3N4 nanowires on the conductive carbonformalin micro-grid under an illumination of an electron beam was occasionally found. As shown in Fig. 15, a Si3N4 nanowire is at its original position (Fig. 15(b)); if the center of the electron beam is moved to the right upper corner marked as B, the nanowire bends away from the center of the beam (Fig. 15(a)); if the center of the electron beam is moved to the other side of this nanowire, it bends immediately to the other side (Fig. 15(c)); if the beam is moved away or the average current density is reduced to a small value, the nanowire gets back to its original straight position. This

204

Zhu et al.

phenomenon is reversible by moving the nanowires (or the center of electron beam) to and fro. And the bending cannot occur unless, at least, either the nanowire/nanotube or the supporting film is an insulator; such a phenomenon was also observed for AlN, GeO2 [34], GaN [35] etc. nanowires and carbon nanotubes placed on the polyvinyl format (formvar) coated copper [36]. Then the force that makes the nanowire/ nanotube bend in TEM can be contributed to an electrostatic force, which results from charge accumulated on the insulator. By further experiments, it is found that bending

(a)

(b)

(c)

Fig. 15. Bending of an individual Si3N4 nanowire: (a) bend to right; (b) at original position when the average density of the electron beam is small; (c) bend to left. B is marked as the center of the illumination beam [32, 33].

Silica Nanowires/Nanotubes

205

deflection (f) of an individual nanowire is approximately proportional to J2 (J: the beam current density). The calculation based on the electrostatics and the elastic mechanics was carried out to receive the same result as that from experiments f ∝ J2. The good flexibility means the high bending strength of these nanowires. Through the method mentioned above, the values of the bending strength of these nanowires can be estimated while assuming that the Young’s moduli of the nanowires are the same as those of their corresponding bulk materials, and the bending strengths of Si3N4 and AlN nanowires are about 1.8–3.6 ⫻ 103 MPa and 2.2 ⫻ 103 MPa, respectively. Compared with the bending strengths of the reaction sinter bulk Si3N4 (117.7–205.1 MPa), the hot-press sinter Si3N4 (486.5 MPa) and the bulk AlN crystal (310 MPa) [37, 38], the bending strength of these nanowires are much greater. This method can be applied to measure the mechanical property of all kinds of nanowires including silica nanowires.

References 1. D.P. Yu, Q. L. Hang, Y. Ding, H. Z. Zhang, Z. G. Bai, J. J. Wang, Y. H. Zou, W. Qian, G. C. Xiong and S. Q. Feng, Appl. Phys. Lett. 73 (1998) 3076. 2. D. P. Yu, C. S. Lee, I. Bello, X. S. Sun, Y. H. Tang, G. W. Zhou, Z. G. Bai, Z. Zhang and S. Q. Feng, Solid State Commun. 105 (1998) 403. 3. Y. J. Zhang, N. L. Wang, R. R. He, J. Liu, X. Zhang and J. Zhu, J. Cryst. Growth 233 (2001) 803. 4. J. F. Qi, T. Matsumoto and Y. Masumoto, Jpn. J. Appl. Phys. 2 40 (2001) L134. 5. C. H. Liang, L. D. Zhang, G. W. Meng, Y. W. Wang and Z. Q. Chu, J. Non-Cryst. Solids 277 (2000) 63. 6. Z. Q. Liu, S. S. Xie, L. F. Sun, D. S. Tang, W. Y. Zhou, C. Y. Wang, W. Liu, Y. B. Li, X. P. Zou and G. Wang, J. Mater. Res. 16 (2001) 683. 7. X. C. Wu, W. H. Song, K. Y. Wang, T. Hu, B. Zhao, Y. P. Sun and J. J. Du, Chem. Phys. Lett. 336 (2001) 53. 8. Y. Q. Zhu, W. K. Hsu, M. Terrones, N. Grobert, H. Terrones, J. P. Hare, H. W. Kroto and D. R. M. Walton, J. Mater. Chem. 8 (1998) 1859. 9. B. C. Statishkumar, A. Govindaraj, E. M. Vogl, L. Basumallick and C. N. R. Rao, J. Mater. Res. 3 (1997) 604. 10. M. A. Ermakova, D. Y. Ermakov, M. Y. Lebedev, N. A. Rudina and G. G. Kuvshinov, Catal. Lett. 70 (2000) 83. 11. M. Harada and M. Adachi, Adv. Mater. 12 (2000) 839. 12. L. Z. Wang, S. J. Tomura, F. H. Ohashi, M. Maeda, M. Suzuki and K. Inukai, J. Mater. Chem. 11 (2001) 1465. 13. M. Zhang, Y. Bando, L.Wada and K. Kurashima, J. Mater. Sci. Lett. 19 (1999) 1911. 14. S. M. Yang, I. Sokolov, N. Coombs, C. T. Kresge and G. A. Ozin, Adv. Mater. 11 (1999) 1427. 15. S. M. Yang, N. Coombs and G. A. Ozin, Adv. Mater. 12 (2000) 1940. 16. H. P. Lin, S. B. Liu, C. Y. Mou and C. Y. Tang, Chem. Commun. 7 (1999) 583. 17. H. P. Lin, C. Y. Mou, S. B. Liu, C. Y. Tang and C. Y. Lin, Micropor. Mesopor. Mat. 44–45 (2001) 129. 18. H. P. Lin, C. Y. Mou and S. B. Liu, Adv. Mater. 12 (2000) 103. 19. Z. L. Wang, Z. R. Dai, R. P. Gao and Z. G. Bai, Appl. Phys. Lett. 77 (2000) 3349.

206

Zhu et al.

20. G. W. Meng, L. D. Zhang, C. M. Mo, S. Y. Zhang, Y. Qin, S. P. Feng and H. J. Li, Solid State Commun. 106 (1998) 215. 21. Y. J. Zhang, N. L. Wang, S. P. Gao, R. R. He, S. Miao, J. Liu, J. Zhu and X. Zhang, Chem. Mater. 14 (2002) 3564. 22. S. T. Lee, N. Wang, Y. F. Zhang and Y. H. Tang, Mater. Res. Bull. 9 (1999) 36. 23. Z. L. Wang, R. P. Gao, J. L. Gole and J. D. Stout, Adv. Mater. 12 (2000) 1938. 24. Z. J. Zhang, B. Q. Wei and P. M. Ajayan, J. Phys.-Condens. Mat. 14 (2002) L511. 25. Y. J. Chen, J. B. Li, Y. S. Han, Q. M. Wei and J. H. Dai, Appl. Phys. A 74 (2002) 433. 26. B. Zheng, Y. Y. Wu, P. D. Yang and J. Liu, Adv. Mater. 14 (2002) 122. 27. L. Dai, X. L. Chen, T. Zhou and B. Q. Hu, J. Phys.-Condens. Mat. 14 (2002) L473. 28. R. F. Egerton, Electron Energy-Loss Spectroscopy in the Electron Microscope (2nd ed.), Plenum Press, New York (1996). 29. M. Tence, M. Quartuccio and C. Colliex, Ultramicroscopy 58 (1995) 42. 30. H. Nishikawa, T. Shiroyama, R. Nakamura, Y. Oma, K. Nagasawa and Y. Hama, Phys. Rev. B 45 (1992) 586. 31. G. G. Qiu, J. Liu, J. Q. Duan and G. Q. Yao, Appl. Phys. Lett. 69 (1996) 1689. 32. Y. J. Zhang, N. L. Wang, R. R. He, J. Zhu and Y. J. Yan, J. Mater. Res. 15 (2000) 1048. 33. Y. J. Zhang and J. Zhu, Micron 33 (2002) 523. 34. Y. J. Zhang, J. Zhu, Q. Zhang, Y. J. Yan, N. L. Wang and X. Zhang, Chem. Phys. Lett. 317 (2000) 504. 35. W. Q. Han, S. S. Fan, Q. Q. Li and Y. D. Hu, Science 277 (1997) 1287. 36. W. H. Knechtel, G. S. Dusburg and W. J. Blau, Appl. Phys. Lett. 73 (1998) 961. 37. H. Wang and T. N. Dai, Study on Producing Ultra Fine Si3N4 Powder with Rice Husks (Chinese edition), Chinese Science and Technology Press, Kunming, China (1997). 38. H. J. Zhou, Dictionary of New Materials (Chinese edition), Shanghai Science and Technology Press, Shanghai, China (1995). 39. J. L. Gole, J. D. Stout, W. L.Rauch and Z. L. Wang, Appl. Phys. Lett. 76 (2000) 2346. 40. Z. W. Wei, Z. R. Dai, C. Ma and Z. L. Wang, J. Am. Chem. Soc. 124 (2002) 1817.

Part III Sulphide, Polymer and Composite Nanowires

Chapter 12 Sulphide Nanowires Shihe Yang Department of Chemistry, Institute of Nano Science and Technology, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

1. Introduction Many metal elements can combine with sulphur to form stable crystalline semiconductor phases that exhibit a variety of unique optical and electrical properties [1]. Such metal sulphide semiconductors spend a large range of electronic energy band gap and often possess a substantial exciton binding energy. Therefore, they have attracted considerable technological and scientific interest [2, 3]. The metal sulphide semiconductors possess a variety crystalline phases depending largely on the atomic radius ratios and electronegativity differences of the constituent atoms of the semiconductors [4]. Metal sulphide quantum dots have been the subject of extensive research [5]. It has been well established that confinements of electrons and holes in the quantum dots change their physical and chemical properties in a profound way. Salient sizedependent properties have been observed, and hence the size constitutes a new parameter one can use to design, tune, and control the attributes of the so-called quantum dot molecules. During the last decade, impressive progress has been made on the controlled synthesis of the sulfide quantum dots using chemical colloidal techniques. In contrast to the conventional vacuum deposition techniques based on sophisticated instrumentation [6], the simplicity of the synthetic methodology and the possibility of large-scale chemical synthesis greatly facilitated the sulphide quantum dot research. While an extensive literature on the sulphide quantum dot is available, the synthesis of one-dimensional (1D) sulphide materials has been only a recent development of the last few years. The characterization and application of these 1D materials are even more limited at this stage. The difficulty consists in the controlled synthesis of uniform and phase pure 1D sulphide materials, which are globally unstable and thus have to be arrested in their thermodynamically metastable forms. Nevertheless, there has been impressive progress in the synthetic methodologies for 1D materials in general [7].

210

Yang

As a result, a number of sulphide nanowires have been synthesized using a variety of methods, and some of their properties have been characterized. In the following we describe the most recent development of the sulphide nanowire research with the emphasis of the synthesis.

2. From 0D to 1D Sulphides Electrons in the quantum dots are confined in all the three dimensions of the Cartesian space [8–12]. While this system of quantum dots is interesting in its own right, 1D materials are at least equally interesting and important for several reasons [13]. First, it allows us to study the effect of dimensionality on the electronic structures and phonon spectra of the materials. Second, the transport of elemental excitations can be probed as a function of the wire size. Third, with the 1D materials, one can examine the interactions between the localized and extended molecular states. What is intriguing is how localization and delocalization affect the kinetic processes along the nanowires. Moreover, new phenomena and exotic properties that are expected or have not been conceived yet may be discovered on 1D materials. Practically, nanowires, as the name suggests, are anticipated to be crucial elements of the future nanotechnology, say, for inter-connects in nano-devices [14]. Some layered metal sulphide materials such as molybdenum and tungsten disulphide can be rolled up into nanotubes [15]. In much the same way as the graphitic carbon nanotubes, these tubular structures from the 2D disulphides are perfect. In particular, the two ends of each concentric tube are closed by introducing elements of lower symmetry into the generically hexagonal tiling of the crystalline planes. Such 1D materials are not our focus in this article and excellent reviews on this topic can be found elsewhere [16]. With the development of synthetic methodologies, not only the disulphide semiconductors mentioned above but also many other sulphide semiconductors can be prepared in the form of nanowires, albeit not necessarily nanotubes. Many useful and desirable properties of the 0D sulphide nanoclusters are not only preserved but also enhanced in their 1D counterparts in terms of device applications. For example, in nanocomposite-based photovoltaics, 1D sulphides may provide paths for more efficient transport of carriers from photoinduced interfacial charge separation, and thus increase the overall photocurrent efficiency [17]. Clearly, 1D sulphide materials are not only fundamentally interesting but also highly promising in a broad range of applications, and therefore merit serious exploration. In order to study and use the 1D sulphides, some important issues concerning their synthesis have to be addressed. First, we need to understand the crystal growth mechanism and kinetics at the nanoscale. Second, we need to work out strategies to induce 1D growth by passive or active intervention in the crystallization processes. Third, one would like to have an absolute control over the sulphide nanowire growth so that the diameter, length, growth direction, and morphology can be varied at atomic precision. Finally, it would be nice that such highly controlled sulphide nanowire growth could be accomplished at the least expense of energy and efforts. This goes into the heart of rational chemical synthesis under mild conditions such as low temperature, low pressure, or ambient conditions.

Sulphide Nanowires

211

3. Growth of Sulphide Nanowires For nano-entities with isotropic constituents, spherical nanoparticles are expected to be a more thermodynamically stable form than one-dimensional (1D) and twodimensional structures owing to the minimization of the surface energy [18, 19]. Indeed, many elements and compounds have been synthesized in the form of spherical nanoparticles. It appears that an anisotropic synthesis such as the synthesis of nanowires requires either intrinsic or extrinsic constraints. The intrinsic constraints may come from the anisotropy of the crystal structures, which give rise to different reactivities and surface energies of the crystal faces [20]. On the other hand, the extrinsic constraints can be realized with templates such as zeolite channels [21], micelles [22], liquid crystals [23], etc., and 1D structural directors such as special molecules [24] and the liquid drop in the VLS/LLSS growth mechanisms [25–27]. During the last few years, various methods have been developed for the synthesis of sulphide nanowires. These methods have successfully exploited the structural constraints mentioned above. In the following, recent developments in the sulphide nanowires will be reviewed with an emphasis on the research of the author’s own laboratory, and the review is not intended to be complete and exhaustive. 3.1. Solvothermal synthesis Solvothermal synthesis is a relatively mild method for the fabrication of sulphide nanorods and nanowires. It involves dissolving inorganic and/or organometallic precursors in a suitable solvent and conducting the reactions in autoclaves normally at temperatures above the boiling point of the solvent. Qian’s group has devoted to the synthesis of many sulphide nanorods and nanowires via solvothermal processes [28–31]. For example, CdS nanorods were synthesized by simple solvothermal reactions of Cd(NO3)2 · 4H2O and NH2CSNH2 in ethylenediamine (H2NCH2CH2NH2, en) at 120 ⬚C (see Fig. 1) [30]. The CdS nanorods are ~10 m in length and ~90 nm in diameter. They are well crystallized, and grow along the c-axis. The en molecule played the role of a director for the nanorod or nanowire growth, and replacing it by other organic solvents failed to produce the CdS nanorod products [32]. It has been shown that the CdS nanorods were formed through an accordion-like folding process that was caused by the dissociation of the en molecules adsorbed on the surface of CdS. The nanorod formation was thought to proceed in several steps. First, cadmium nitrate and thiourea reacted to produce the lamellae CdS with many folds on it. After that, the folds on the lamellae agglomerated together. Then these folds broke into needlelike fragments. Finally, these needles grew to well-crystallized nanorods. It was believed that the en molecules adsorbed on the samples determined this morphology transition and the en molecules in the solution were not relevant to this transition process. Variations of the solvothermal method described above have also been reported. One case is the use of polymer matrix instead of the en solvent molecule for directing the nanowire growth. Nanorods of PbS in poly(vinyl acetate) with a core/sheath structure has been successfully synthesized using this method by the reaction of Pb(NO3)2 and thiourea and the simultaneous polymerization of vinyl acetate under

212 (a)

Yang (b)

Fig. 1. Typical CdS nanorods prepared in en at 250 ⬚C for 12 h. (a) TEM photograph of a CdS nanorod. (b) SEAD pattern of the nanorod in a [010] zone [30].

hydrothermal conditions [33]. Another interesting report is the synthesis of Bi2S3 nanorods under hydrothermal conditions even without any molecular directors [34]. This may be due to the peculiar crystal structure of the material. More recently, the application of the solvothermal method has been extended to the synthesis of ternary nanorods and nanotubes of MInS2 (M ⫽ Cu, Ag) [35–37]. These materials are important for photovoltaic applications. Both elemental constituents and organometallic precursors were experimented to fabricate the multicomponent nanorods. For the elemental solvothermal route, the elements of interest in an appropriate ratio were mixed and coaxed in en under solvothermal conditions. Using this method, nanorods of CuInS2 were produced in the tetragonal phase with sizes of ~20 nm ⫻ ~800 nm (see Fig. 2). It was found that the en solvent, the reaction temperature and time are crucial factors for the synthesis of the ternary nanorods. When the elements were used, en was also important due to its coordination capacity and its solvency for sulfur. The reaction temperature had to be above the melting point of In, indicating the solution-liquid-solid (SSL) mechanism, whereby the In liquid drop acted as the nanorod director. For the growth of ternary nanorods with the molecular director en, the mechanism is expected to be similar to that for binary nanorods. Nanorods of other ternary systems such as SbSI and Ag3CuS2 have been synthesized under hydrothermal conditions even without any molecular directors [38, 39]. 3.2. Templated fabrication Among the templates used for the sulfide nanowire synthesis, porous anodic aluminum oxide (AAO) has received special attention in the last few years. The AAO

Sulphide Nanowires

213

Fig. 2. TEM image of CuInS2 nanorods [35].

films are grown in acidic electrolytes, and possess uniform arrays of straight pores with pore diameters ranging from below 10 to 200 nm, pore lengths from 1 to 50 ␮m, and pore densities of 109–1010 cm⫺2. The regular and nearly parallel porous structures of AAO are an ideal template for the synthesis of cylindrical nanowires with a narrow diameter distribution. The advantages include the simplicity and the structural stability of the samples generated by it, resulting from the refractory nature of the oxide matrix. The optical transparency of the oxide matrix suggests the possibility of producing optoelectronic devices based on these relatively monodispersed semiconductor nanostructures. The controlled variability of the particle diameters and composition might also be used to tune in desired optical properties. Martin et al. and Chakarvarti et al. [40–42] first fabricated semiconductor nanowires using the AAO membranes as the template. Later, uniform CdS nanowires were fabricated with lengths up to 1 ␮m and diameters as small as 9 nm by ac electrochemical deposition of the semiconductor directly into the pores of anodic aluminum oxide films from an electrolyte containing Cd2⫹ and S in dimethyl sulfoxide [43]. The deposited material is hexagonal CdS with the crystallographic c-axis preferentially oriented along the length of the pore. However, structural studies have demonstrated that large numbers of stacking faults and twinned segments are present in these nanowires. This was attributed to the non-epitaxial deposition into the amorphous alumina pores. Recently, however, an aligned CdS single crystal nanowire array has been produced using dc electrolysis in AAO templates from a dimethyl sulfoxide (DMSO) solution containing cadmium chloride and elemental sulfur [44]. CdS single crystal nanowires was also prepared through electrochemically induced deposition in the pores of an AAO template from an acidic chemical bath containing cadmium chloride and thioacetamide (TAA) [45]. Electroreduction of protons on conductive substrates at small current densities has been found to trigger a growth of

214

Yang

highly crystallized hexagonal CdS thin films from acidic chemical baths containing CdCl2 and TAA. It was believed that the film growth is not controlled by the electrode kinetics and the electroreduction of protons imposes a proton gradient in the vicinity of the substrate to reduce the activation barrier for the decomposition of Cd-TAA complex at the surface of the substrate (Fig. 3). Here electrochemically induced deposition of CdS into the pores of the template has produced hexagonal CdS single crystal nanowires. This result further confirms that the CdS growth is achieved by the atom-by-atom growth of individual crystallites and not by agglomeration of particles formed in the solution phase. Interestingly, the growth rate of the CdS nanowire deposited in the pores of the template is faster than that of the CdS films deposited on the conductive substrate. Moreover, the growth rate of the nanowire increases with decreasing pore size of the template. These results were explained by the influence of diffusion. As the rate of diffusion of ions in the pores is slower than on the plane surface and the diffusion rate decreases with decreasing pore size, the pH gradient imposed by the electroreduction of protons at the bottom of the pore increases with decreasing pore size. Therefore, the growth rate of the nanowires is enhanced as the pore size decreases.

Fig. 3. TEM images of the CdS nanowires prepared by electrochemically induced deposition in the AAO templates with diameters of about 90 nm (a) and 20 nm (b) [45].

Sulphide Nanowires

215

Mineralization of hexagonal lyotropic liquid crystals has been used for the fabrication of sulfide superlattices [46, 47]. Precipitation of the organic-inorganic solid takes place within the ordered environment of the mesophase, and both the symmetry and long range order of the liquid crystal are preserved. The product contains rod-like pores of 2–3 nm diameter, spaced a few nm apart in a hexagonal lattice. Nanostructures of mixed compositions such as Cd0.5Zn0.5S can also be prepared in this way by the mineralization of a hexagonal mesophase containing 0.05 M Cd(NO3)2 and 0.05 M Zn(NO3)2 (see Fig. 4). The sulfide growth process copies the symmetry and characteristic dimensions of the original mesophase by avoiding growth of mineral within regularly spaced hydrophobic regions. The superlattice morphology is thermodynamically stable with respect to the solid lacking nanoscale features, and can be controlled through the amphiphile’s molecular structure and water content in the liquid crystal. The success of templating CdS, CdSe, and ZnS and the failure of templating Ag2S, CuS, HgS and PbS suggest that interactions of polar segments in template molecules with the precipitated mineral and with its precursor ions are necessary conditions for direct templating. Qian et al. [48] also used liquid crystals to template the sulfide nanowire synthesis. Specifically, ZnS nanowires were synthesized by a direct templating route under -ray irradiation in an inverted hexagonal liquid crystal formed by oligo(ethylene oxide)oleyl ether amphiphiles, n-hexane, n-hexanol/i-propanol (2:1), and water. The nanowires have a diameter of ca. 5 nm, and are closely packed to form a hexagonal bundle with a width of ca. 10–30 nm. Most recently, nanotubes and nanowires of CdS have been obtained from solutions containing a surfactant Triton 100-X [49]. It appears that at the high concentrations,

Fig. 4. TEM image of Cd0.5Zn0.5S grown from a hexagonal mesophase doped with 0.05 M Cd(NO3)2 and 0.05 M Zn(NO3)2 (scale bar: 50 nm) [47].

216

Yang

the surfactant molecules aggregate, providing a template for the growth of the sulfide nanotubes and nanowires. Quantum confinement in such 1D materials have been observed through photoluminescence measurements. Another way is to use the pre-formed CdS nanorods to prepare ternary nanorods of ZnxCd1-xS [50]. This was accomplished by the hydrothermal reaction of the CdS nanorods with ZnCl2 · H2O and thiourea at 180 ⬚C. The molar ratio of Zn to Cd was found to be important for the synthesis of the ternary nanorods. A similar method was used for the fabrication of nanorods of CdIn2S2 [51]. 3.3. Polymer-controlled growth Polymers can not only direct the growth of nanowires but also act as a medium for dispersing the nanowires. In addition to their own excellent mechanical, optical, and processing properties, polymers can passivate the nanowire surfaces and effect the size, size distribution, and spatial distribution of the nanowires. CdS nanowires have been encapsulated in a number of polymer matrices using different methods. CdS/polyacrylonitrile(PAN) composite nanowires have been prepared under ambient conditions, using 4 : 1 (v/v) distilled water-isopropyl alcohol solvent mixture and Na2S2O3 as sulfur source [52]. It was thought that the formation of the composite nanowires include three steps: formation of nanosized particles, polymerization of AN by photocatalysis of CdS, and regulation of the CdS nanoparticles along the polymer chains through the specific interaction between polymer and CdS nanoparticles. Upon exposure to visible light, traces of slowly formed CdS nanoparticles are capable of photoinitiating polymerization of AN monomers. In low-concentration solutions, polymer chains are apt to spread and show wire-like structures and may act as template, such that Cd2⫹ ions gather along the wires. During the homogeneous release of S2⫺ from Na2S2O3, the composite nanowires are slowly formed. Because the CN groups of the AN are partly hydrolyzed to hydrophilic amide groups, the dispersion of CdS/PAN nanowires in aqueous solution can be stable for some days and can easily be peptized by ultrasonic stirring when the coagulation is observed. Polyacrylamide is another matrix for the dispersion of single crystalline CdS nanowires with large aspect ratios [53, 54]. A typical CdS nanowire prepared by this method is shown in Fig. 5 with a diameter of ~40 nm. In this case, the solvent is a bidentate ligand ethylenediamine. After polyacrylamide with dispersed Cd2⫹ was treated with thiourea solvothermally at 170 ⬚C in ethylenediamine, a yellow gel gradually appeared, and the CdS nanowires were formed. Although ethylenediamine was believed to direct the nanorod growth, the nanorods were short without polyacrylamide. The polymer may absorb en to form a gel with many small pores. When sufficient solvent is absorbed, these pores connect and become continuous. The lateral growth of CdS is confined within the polymer matrix and axial growth along the [001] direction is favorable. CdS nanofibers have been prepared in an alternate copolymer of maleic anhydride (MA) and styrene (St) via a one-pot procedure using -irradiation [55]. The copolymer of maleic anhydride and styrene is expected to show a wide range of applications because of its high thermal stability compared with general-purpose thermoplastics. In addition, the ordered and regular polymer environment induces nucleation of CdS

Sulphide Nanowires

217

Fig. 5. TEM image of a typical CdS nanowire [53].

in such a way that its size and shape can be controlled. St monomers are activated by the radicals generated in solution upon irradiation and are converted into St radicals, which rapidly initiate copolymerization of St and MA. The formation of CdS nanofibers was accomplished by S from CS2 and Cd2⫹. A novel one-step soft solution processing route called the solvothermalcopolymerization technique was successfully developed for in situ fabrication of polystyrene (PS)/CdS nanocomposites embedded with CdS nanowires in ethylenediamine media at lower temperatures (80–140 ⬚C) [56]. In this route, the polymerization and nanocrystal formation occur simultaneously in a certain temperature range. 2,2⬘-azobisisobutyronitrile (AIBN) was used as a radical initiator and the sulfur source was from thiourea. The reaction temperature and solvent were found to play a key role in the nanocomposite synthesis, and along with the concentrations of Cd2⫹ and the monomer, they influence the spectroscopic properties of the nanocomposites. Very recently, nonionic amphiphilic triblock copolymer systems (EO)x(PO)y(EO)x have used to grow CdS nanrods at low temperature via the reaction of cadmium acetate and sodium sulfide in an aqueous phase with surfactant [57]. The size of the CdS nanorods could be controlled by varying the surfactant species, the mole ratio between the inorganic precursors and triblock copolymer, and the reaction temperature. Most polymer-encapsulated CdS nanowires possess a W-type structure and naturally grow along the c-axis of this structure. On the other hand, some semiconductors such as PbS take an isotropic rock-salt structure and it seems to be difficult to grow nanowires of these materials in polymer matrices. However, PbS nanorods (or nanostrips) have recently been encapsulated in poly(vinyl butyral) (PVB) films with the assistance of a surfactant bis(2-ethylhexyl)sulfoccinate (AOT⫺) [58, 59]. Fig. 6 shows a TEM image of the PbS nanorods (or nanostrips) after solvent treatment of the PVB matrix. The PbS nanorods have a relatively narrow size range and are highly oriented with their (100) lattice planes parallel to the substrate surface. It was thought that the precursor films contain surfactant layered domains surrounded by PVB, which curl

218

Yang

Fig. 6. TEM image of PbS nanostrips formed in a PVB polymer film [58].

towards one-dimension growth induced by the H2S gas infusion. This results in the formation of the oriented PbS nanorods in PVB films. 3.4. Vapor–liquid–solid route The vapor-liquid-solid (VLS) route for nanowire growth has advantages of good purity and relatively high yield. One of the methods is the laser-assisted catalytic growth (LCG) [60]. (see other chapters of this book???) The key feature is that equilibrium phase diagrams can be used to predict catalysts and growth conditions, thereby enabling rational synthesis of new nanowire materials. The nanowires of ZnS and CdS have been prepared by Lieber’s group with an Au catalyst at temperatures of 700–1100 ⬚C. They are single crystals with diameters as small as 3 nm, which places them in a regime of strong radial quantum confinement, and lengths exceeding 10 ␮m [61]. The CdS nanowires exhibit a stable wurtzite (W) structural phase, which is distinct from the zinc blende (ZB) structural phase. Most recently, it was demonstrated that large quantities of single crystalline CdS nanowires could be synthesized even without expensive laser systems by thermal evaporation of CdS powders under controlled conditions in the presence of Au catalyst [62]. The W-type CdS nanowires were 60–80 nm in diameter and several tens of micrometers in length, and grew along the ⬍131⬎ direction (see Fig. 7). Photoluminescence (PL) measurements show that the synthesized CdS nanowires exhibit a strong, broad, and stable red emission at room temperature, which could be attributed to their surface states. 3.5. Gas–solid reactions It is interesting that the nanowires we have described so far mostly need some sorts of template such as molecular directors, AAO membranes, surfactants, or polymer scaffold. MS2 is an exception. It has similar features as those of fullerenes such as a two-dimensional layer structure. Consequently, the layers can fold into nanotubes even without template assistance. MS2 (M ⫽ W, Mo) nanotubes and nanoparticles are known as inorganic fullerenes (IF) because of their fullerene structures. They were first observed by hydrosulfurization of a very thin film of tungsten [63]. Subsequently, IF-MoS2 was reported [64]. Negative-curvature polyhedra and nested IF clusters of these materials have also been produced through high-temperature gas–solid reaction

Sulphide Nanowires

219

(a)

(b)

Fig. 7. (a) Typical SEM image of bulk CdS nanowires. Large quantities of the nanowires were distributed homogeneously on the Si substrate. (b) XRD pattern taken on bulk CdS nanowires. The numbers above the peaks correspond to the (hkl) values of the Wurtzite CdS structure [62].

under a reducing gas atmosphere [66]. Recently, bundles of very long single-wall MoS2 nanotubes 9.6 Å in diameter were prepared [66]. Furthermore, it was reported that multiwall nanotubes of many of the layered metal dichalcogenides could be prepared by the thermal decomposition of the respective ammonium thio-metalate precursor [67]. The IF dichalcogenides were produced by chemical vapor transport, thermal decomposition into alumina membranes, and other methods. This topic has been extensively reviewed in the scope of fullerenes [68]. In the following, we will focus on the non-templated growth of another type of semiconductor nanowires. This is copper sulphide (Cu2S) nanowires fabricated simply by exposure of a copper surface to O2 and H2S at room temperature and ambient pressure. Immediately on exposure to the gas mixture, the copper surface became dark red, shinning cyan, and gray in a short time span. After 10 h reaction, the copper surface was covered with black and fluffy materials, indicating the formation of dense

220

Yang

Cu2S nanowire arrays [69–73]. Typical SEM and TEM images of the Cu2S nanowires are shown in Figs. 8 and 9, respectively [73]. The nanowires are straight and circular in cross section with diameters of a few tens of nanometers. Needle-shaped nanowires can also be produced under appropriate conditions [72]. The identification of nanowire composition was accomplished by a variety of techniques including X-ray photoelectron spectroscopy [71, 74], UV-Vis absorption spectroscopy [74], electron paramagnetic resonance spectroscopy [74], and X-ray diffraction (XRD) [69–72]. Compared to the XRD pattern of a powder sample, the XRD pattern of an as-prepared sample on a copper surface exhibits sharper diffraction peaks, and some diffraction peaks are significantly enhanced over others (see Fig. 10) [73]. This

Fig. 8. A side-view SEM image of the Cu2S nanowire arrays on a copper foil substrate. Note that the Cu2S nanowires are roughly aligned perpendicular to the copper surface [73].

Fig. 9. Transmission electron image of the copper sulphide nanowires grown in situ on a copper grid [71].

Sulphide Nanowires

221

suggests that the Cu2S nanowires are roughly oriented single crystals. In particular, – the (2 04) diffraction peak has undergone the largest increase from the powder sample – to the nanoarray sample, indicating that the (2 04) crystal plane of the Cu2S nanowires are parallel to the surface of the copper substrate. This can be seen more clearly by – – plotting the relative intensity (I(hkl)/I(2 04))/(Is(hkl) )/(Is(2 04)) of each diffraction – peak as a function of the angle between the diffraction plane and the (2 04) plane (see Fig. 11) [73]. Knowing that the nanowires are more or less perpendicular to the copper surface, it has been concluded that the growth direction of the Cu2S nanowires – is perpendicular to the (2 04) crystal plane, i.e. the c axis of the monoclinic Cu2S.

Fig. 10. XRD patterns of nanowires (a) from the as-prepared Cu2S nanowires on a copper foil substrate, and (b) from a Cu2S powder sample. The vertical dotted and dashed lines indicate the powder diffraction peak positions and intensities of monoclinic CuO and Cu2S, respectively [73].

Fig. 11. The dependence of the relative diffraction peak intensities of crystal planes (hkl) on – the angle between (hkl) and (2 04). The solid line is obtained from fitting to an exponential function [73].

222

Yang

This conclusion has been corroborated by high resolution transmission electron microscopy (TEM). As shown in Fig. 12, a Cu2S nanowire with a uniform thickness of ~45 nm displays perpendicular fringes with a regular spacing of ~3.3 Å, which is – quite close to the inter-plane distance of (2 04) [73]. The nanowire appears to be wellcrystallized although the outer portion consists of some oxide nanoparticles. The selected area electron diffraction pattern provides a further support for the growth direction as shown in Fig. 12c. We have identified some important effects of copper surface structure, reagent gas composition, and reaction temperature on the growth of Cu2S nanowires. For example, the Cu2S nanowires can only grow on certain surface of copper, i.e., Cu (220). The nanowire diameters increase with increasing temperature, and to a less extent, with increasing molar ratio of O2 to H2S. At high temperatures and high molar ratios O2 : H2S, the nanowires become thicker, less uniform in diameters, coated with nanoparticles, and even evolve into nanocones or nanoneedles. The XRD patterns of the copper foil before and after reaction are shown in Fig. 13 [73]. Clearly, the relative intensity of (200) to (111) decreased, whereas that of (220) to (111) increased after the gas-solid reaction. It appears that the relative intensity of

Fig. 12. TEM images and selected area electron diffraction pattern of a single Cu2S nanowire. (a) A low magnification TEM image; (b) high resolution TEM image and (c) selected area electron diffraction pattern along the zone axis [201] of Cu2S [73].

Sulphide Nanowires

223

Fig. 13. XRD patterns of the copper foil substrate before (a) and after (b) the formation of the Cu2S nanowires. The vertical solid lines indicate peak positions and intensities of random polycrystallites of face-center cubic Cu [72].

all the diffraction peaks decreased relative to that of the (220) diffraction peak. It was speculated that the Cu2S nanowires grew epitaxially on the (220) surface of copper along c-axis of monoclinic Cu2S [72]. However, only the Cu2S nanowires that are roughly perpendicular to the substrate surface (in this case, the (220) copper grain surface is roughly parallel to the copper substrate surface) could grow, whereas the growth of those nanowires along severely-tilted directions could not continue. The distances between neighboring Cu atoms in the face-centered cubic Cu are calculated to be 3.615 Å and 2.556 Å in the crystal plane of (220), which are comparable with the distances between Cu atoms of the copper layer of the monoclinic Cu2S (3.62 ⫾ 0.2 Å and 2.56 ⫾ 0.26 Å). The average Cu-Cu distances in the Cu-S layer is ~4.1 Å. It appears that in both the copper layer and the Cu-S layer of the monoclinic Cu2S, the Cu-Cu distances are not far from those in Cu (220) although a certain extent of angular distortion of the Cu atoms has occurred due to the S-insertion. In addition, the inter-plane distance of (220) is 1.278 Å, which is also close to the spacing between the copper layer and copper/sulfur layer of monoclinic Cu2S (1.123 Å) along the nanowire direction (the c-axis) [73]. The effect of temperature on the Cu2S nanowire growth can be appreciated from Fig. 14 [72]. Clearly, the diameters of the Cu2S nanowires increase as the reaction temperature increases at a given molar ratio of O2 to H2S. At the same time, the size distribution of the nanowire diameters also becomes broader as the reaction temperature increases. At a given temperature, the nanowire diameters increase with the molar ratio O2:H2S, and this increase is more pronounced as temperature increases. Another observation is that at higher temperatures, nanocones or nanoneedles are more readily formed, and the turning point for their formation is ~30 ⬚C. Therefore, to prepare Cu2S nanowires with uniform diameters in tens of nanometer scale, the reaction temperature should be kept low, for example, at ~10 ⬚C. However, at low temperature and low molar ratio of O2 to H2S, the yield of Cu2S nanowires decreases significantly. For example, the Cu2S nanowires produced at 0 ⬚C decreased greatly (e.g., by a factor

224

Yang

Fig. 14. Dependence of the mean diameter of the Cu2S nanowires on the molar ratio O2 :H2S at different reaction temperatures [72].

of ~10–20) when it was compared with that at 22 ⬚C given the same other reaction conditions. Furthermore, the quantity of nanowires produced at lower molar ratio (e.g., 0.667) were far less than at higher molar ratio (e.g., 2.0) at the same reaction temperature. In order to understand the intriguing nanowire growth mechanism, effort has been spent to study the early stages of the Cu2S nanowire development [75]. When the reaction time between the copper surface and the gas mixture of H2S/O2 was 10 min, mainly Cu2O layer was observed. Cu2S nuclei started to appear on the Cu2O layer for the reaction time of ⬎ 20 min. After nucleation, the Cu2S nanostructures were then evolved into nanowires and grew mainly along the direction perpendicular to the copper surface. A more extended growth period resulted in well-aligned Cu2S nanowires shown in Fig. 15a. The nanowires are not only straight but also uniform in diameter (d ⫽ ~40 nm), and have lengths of a few hundred nanometers. The nanowires grew much faster after nucleation and incipient 1D growth. It seems that the nanostructures require an induction period to prepare themselves for the smooth and rapid 1D growth. Interestingly, each nanowire consists of an inner single crystalline core and an outer shell as can be seen more clearly in Fig. 15b. By combining the selected area electron diffraction and dark-field imaging, it has been verified that the inner core of single crystalline Cu2S is enclosed by the outer shell of polycrysalline Cu2O. The single crystalline Cu2S nanowire appears to have few defects judging from Fig. 15d.

Sulphide Nanowires

225

Fig. 15. (a) TEM image showing typical morphology of copper sulphide nanowire arrays grown on copper surface. (b) Core/shell structure of a Cu2S nanowire. (c) Dark-field image of the Cu2S nanowire by using the (200)Cu2S diffraction. Bright dot contrast comes from the copper oxide shell. (d) HRTEM image taken along the [100] direction [75].

Some remarks can be made about the Cu2S nanowire growth on the basis of the experimental results presented above. First, O2 is essential to the formation of the Cu2S nanowires on a copper surface despite that the reaction Cu(s) ⫹ H2S(g) ⫽ Cu2S(s) ⫹ H2(g) is exergonic at room temperature (⌬G0 ⫽ ⫺52.8 KJ/mol) [76], the nanowire formation is believed to start with an oxidation reaction Cu(s) ⫹ 1/2O2(g) ⫽ Cu2O(s), followed by sulfidization Cu2O(s) ⫹ H2S(g) ⫽ Cu2S (s) ⫹ H2O(g). Second, the Cu2S nanowires were likely grown on the (220) surface of copper in an epitaxial fashion because this surface was found to be conducive to the nanowire formation. This is consistent with our finding that the nanowires grow along the c-axis of the monoclinic Cu2S [73]. It boils down to the question how the nanowires actually grow. Conventional wisdom may suggest the vapor-liquid-solid (VLS) growth mechanism [25–27], which normally requires high temperatures (e.g., by laser ablation) in the nucleation and growth stages. However, this is not supported by our experiments; we have never seen a drop-like feature at the nanowire tips. Nevertheless, the nanowire assembly site has to be at one of the tips, the root or the top. The biggest strength of the root-growth mechanism is that the close-by copper surface provides the necessary feeding materials for the nanowire growth because the growth site is right between the Cu2S nanowires and the copper surface. In this mechanism, however, the reaction site has

226

Yang

to be delicately structured in such a way that the grown Cu2S nanowires and the underlying copper substrate have a structural relationship. Furthermore, there is hardly any driving force for the Cu and S atoms to be inserted into the interface between Cu and Cu2S. The weakest point of the first mechanism is that it can not explain why the Cu2S nanowires thicken as the reaction goes. One is then left with the top-growth mechanism, which appears to explain all the facts observed so far. In this mechanism, the growth is on the nanowire top and the feeding stocks are from the roots. This demands an efficient transport channel for the feeding stocks across or through the nanowires. One such channel has been imagined, which consists of an illusive fluid transport layer on the Cu2S nanowire surfaces (see Fig. 16a). The shell structure in the TEM image of Fig. 15b was thought to be an evidence for this fluid transport layer, which contains species such as Cu, S, and O in the form of Cu2O nanocrystals strewn on the Cu2S nanowire surfaces. The fact that H2O is produced in the nanowire formation reaction may enhance the fluidity of the atomic transport through the nano-layer. This on-top growth mechanism can naturally explain the epitaxial relationship between the Cu2S nanowires and the Cu (220) surface (see Fig. 16a). It also accounts for the nanowire thickening with reaction time because the deposition of the Cu and S species on the nanowire surface from the fluid layer is facilitated under more reactive conditions, and sometimes nanoneedles and nanocones are formed in such cases. One unanswered question is the nature of this transport layer. In addition, a slower growth rate is expected for a thicker nanowire because the transport channel is limited to the surface layer, and this was not observed in experiments. Alternatively, the atomic transport may be accomplished through the bulk of the Cu2S nanowires. The reaction goes as follows (see Fig. 16b). O2 is adsorbed on the nanowire top and reduced, leaving electron holes and Cu⫹ vacancies on the nanowire top. Both the electron holes and the Cu⫹ vacancies migrate through the Cu2S nanowire from the top down to the root and are annihilated there. The Cu2O species thus formed on the nanowire top is then converted to Cu2S by the reaction with H2S. The nanowire assembly continues in this way on the top by interweaving the

Fig. 16. Schematic showing 1D growth on the top of the Cu2S nanowires. (a) Surface transport of copper [72]. (b) Bulk transport of holes and copper vacancies.

Sulphide Nanowires

227

sulfur and copper layers in alternation. In fact, a similar mechanism has been proposed for the growth of Cu2S thin films on copper surfaces [77]. The fact that low temperature is necessary for the synthesis of thin nanowires can be explained by noticing that (1) the atomic mobility through the nanowire increases with decreasing diameter of the nanowires and (2) the deposition of Cu and S on the top of the nanowires is much more energetically favorable than on the periphery. At low temperatures, only those Cu2S nanowires can grow, which are so thin that the atomic mobility in the nanowire is sufficiently high. The Cu and S species are selectively deposited on the top of the nanowires. As temperature increases, the atomic mobility increases, and thus the nanowire diameters and their distribution increase as observed. Furthermore, at higher temperatures, the gas-solid reaction becomes less selective, both the length and the diameter of the nanowires increase during reaction, and in some cases, Cu2S nanocones or nanoneedles can be formed. In the context of the mechanism proposed above, we examine the effect of the molar ratio O2 : H2S on the Cu2S nanowire growth. At low molar ratios of O2 to H2S at a given temperature, because the reaction is slow, only small diameter nanowires grow. However, as the molar ratio O2 : H2S increases, the reaction rate increases and therefore, the reaction becomes less selective. Specifically, the growth of the Cu2S nanowires is not only along the wire axis but also along the wire thickness. This naturally increases the nanowire diameters and broadens the size distributions. In a way similar to the high temperature effect, the reaction becomes less selective at high molar ratios of O2 to H2S, leading to the formation of nanocones or nanoneedles. In general, the effect of the molar ratio O2 : H2S on the Cu2S nanowire growth is weaker than the temperature effect because the thickness of the Cu2S nanowires is primarily determined by temperature. At low temperatures, the effect of the molar ratio O2 :H2S is not significant (see Fig. 14). It is only at high temperature that the effect of the molar ratio O2 : H2S becomes important due to the increase in the atomic mobility in the nanowires.

4. Structural, Electronic, Optical, and Transport Properties 4.1. Structures and morphologies As described in Section 3, the sulphide nanowire structures and morphologies have been characterized mainly by XRD, TEM, and SEM, which are almost essential for the characterization of nanostructures. Occasionally, EXAFS and XPS have also been used to characterize the local coordination and composition of the nanowires [28, 36, 38, 48, 51, 71, 74, 78]. The EXAFS of the Cu2S nanowires showed that the average coordination number is 2.3 with the mean Cu-S distance of 2.28 Å in the first shell [78]. This is to be compared with the corresponding values of the bulk Cu2S (3 and 2.33 Å) [78], and therefore the effect of the nanowire surfaces is evident. XPS has been very useful for the determination of the Cu2S nanowire structures (see Fig. 17) [74]. For example, the XPS in the core level region of S 2p (Fig. 17a) and Cu 2p (Fig. 17b) showed unequivocally the presence of S2⫺, Cu⫹ or Cu and trace amounts of Cu2⫹ in the nanowire sample (943.9 eV). The distinction between Cu⫹

228

Yang (a)

(b)

(c)

Fig. 17. XPS of a Cu2S nanowire sample in the core level regions of (a) S 2p and (b) Cu 2p. (c) Cu L3MM Auger spectrum of the nanowire sample [74].

and Cu can not be made from the XPS spectrum alone in Fig. 17b because they all peak at ~932.2 eV. This is where the Cu L3MM Auger spectrum (Fig. 17c) can help, which displays a characteristic peak of Cu⫹ (916.9 eV). This is consistent with the nanowires with a core of Cu2S and a shell of Cu1.96S [69–71]. In general, the techniques mentioned above provided vital structural information about the newly-synthesized sulphide nanorods and nanowires. For the nanorods prepared using the solvothermal method, the aspect ratio is usually not large although the method is simple and potentially useful. The template methods and the VLS growth can often grow high quality nanowires with a large aspect ratio at the expense of somewhat more expensive instrumentation and more tedious procedures. The crystallinity of the nanowires is mainly influenced by the processing temperature for different methods. The Cu2S nanowires, although grown at temperatures as low as room temperature or below, exhibit a single crystal structure along the c-axis of the monoclinic Cu2S. Most CdS nanowires fabricated using different methods are grown along the unique c-axis of the W-type CdS. In the PVB polymer matrix, PbS nanorods

Sulphide Nanowires

229

are grown along the [001] axis of the rock-salt PbS. Although most sulphide nanowires have a cylindrical morphology, the PbS nanorods appear have a strip shape and oriented along [001] in the polymer film. 4.2. Optical properties Linear optical properties of the metal sulphide nanowires have been extensively characterized. Especially the optical absorption has been measured in most of works involving the synthesis of metal sulphide nanowires. For the sulphide nanowires with diameters comparable to or smaller than the Bohr diameters of the corresponding semiconductors, the absorption edges are found to be significantly blue-shifted as found for the nanowires of CdS [53], PbS [58], and Cu2S [74]. For some sulphide nanorods and nanowires, photoluminescence has also been measured [31, 43, 49, 51, 53, 56, 57, 62, 69]. In general, photoluminescence of the metal sulphide nanowires displays two main features. One is due to the band gap emission and the other corresponds to the defects or surface states. It seems that polymer-wrapped sulphide nanowires usually exhibit stronger band-gap emissions with reduced intensity of the defect-related emissions. This was attributed to the passivation of the nanowire surfaces by the polymers. Fig. 18 shows an example of the photoluminescence from the CdS nanowires dispersed in the polyacrylamide gel (see Fig. 5) [53]. Only a strong

Fig. 18. Photoluminescence emission (␭ex ⫽ 310 nm) spectrum of CdS nanowires [53].

230

Yang

narrow emission at 380 nm was observed. The absence of the trap state emission suggests the stoichiometric nature of the CdS nanowires without a surface excess of Cd2⫹ and S2⫺ vacancies. For the phonon/vibration characterization of the metal sulphide nanowires, Raman spectroscopy has been very useful. Raman spectra of the CdS and Cu2S nanowires have been obtained [43–45, 62, 73, 79], showing the characteristics of their bulk counterparts. For example, the Raman spectrum of the CdS nanowires typically shows three peaks at 304, 606, and 908 cm⫺1, which correspond to the first-, second-, and third-order longitudinal optical (LO) phonon modes of CdS [79]. Some times, a forth LO mode could also be observed [43]. 4.3. Phase transitions Phase transitions in the Cu2S nanowires have been studied as a function of temperature and pressure [78]. Differential scanning calorimetry (DSC) shows an endothermic process at 101.8 ⬚C upon heating, corresponding to the phase transition from -Cu2S (monoclinic) to -Cu2S (hexagonal) in the bulk (103.5 ⬚C) [71, 80]. This lowering in the phase transition temperature may be due to the small diameter of the Cu2S nanowires. However, the reverse phase transition was found to be much lower (74.8 ⬚C) due perhaps to kinetic entropic effect. X-ray diffraction (XRD) demonstrates that the c-axis of -Cu2S is transformed to the c-axis of -Cu2S owing to the positional disorder of the Cu atoms. Two high pressure phases of Cu2S have been identified using a diamond anvil cell and the energy dispersive XRD technique. 4.4. Transport properties While the conductivity measurement of single sulphide nanowires is alluring, it is still a major challenge at this stage. The ability to fabricate semiconductor nanowires

Fig. 19. A direct I–V characteristic showing the semiconductor property of the Cu2S nanowires [81].

Sulphide Nanowires

231

on a conductive copper substrate allowed the ensemble electrical conductivity of the Cu2S nanowires to be measured [81]. Here, the copper substrate was served as the lower electrode and an Au pattern with diameter of about 0.3 mm was successfully fabricated on the other end of the nanowire arrays by inserting polyimide between the insulating spacer. Fig. 19 shows the I–V curve of the Cu2S nanowires at various temperatures. Clearly, metal-semiconductor Schottky junctions were formed at the contact points between the nanowires and the metal electrodes. At high voltages, the Schottky junctions broke through and the voltage then started to fall on the nanowires themselves. From the slopes of the I–V curve at the high voltages, one can see that the conductivity of the nanowires increases dramatically with the increase of temperature. This is due to the thermal ionization of the donors or acceptors, which give rise to free carriers in the conduction or valence band of the nanowires to form the current. The experimental results have shown that the Cu2S nanowires are highly conductive at high temperatures. The resistivity of the nanowires is estimated to be less than 10 ⍀cm at room temperature considering of the size and density of the wires.

5. Potential Applications The research on the applications of the sulphide nanowires is still very limited at present. Basically, the effort is currently on the research on the applications is at its infancy. In order to stimulate the progress in this direction, I will present some preliminary results on the application potentials of the Cu2S nanowires. One of the distinctive advantages of the Cu2S nanowires is that they can be grown in an array on a conductive surface of copper. This facilitates the studies on the applications of the Cu2S nanowires. 5.1. Field emission Field emission of carbon nanotubes (CNTs) has received a great deal of attention [82–85], because of the potential applications in the field emission flat panel display [86]. It was found that the field emission depends on the electronic property of the CNTs, in particular on whether the CNTs have metallic or semiconducting properties. Unfortunately, one is still not able to control these properties using the presently available synthetic methodologies although it is possible to modify the electronic properties of the CNTs [87, 88]. The Cu2S nanowire arrays may be a possible candidate for field emission devices because the material is a semiconductor and is naturally grown on the copper substrates by an awfully simple method under ambient conditions. Preliminary results have been obtained on the field emission properties of the Cu2S nanowire arrays. The uniformity of the field emission from the nanowire arrays and its variation with field were examined using the transparent anode technique. They consist of points of light generated by the impacting electrons onto the tin-oxide film of a transparent anode; many are tiny dots with a few large spots. With increasing field, the emission became stronger, resulting in an increasing number of bright spots. Although

232

Yang

the emission came from most areas of the film, more emission sites appear to be near the edges of the film. This is interesting because it is known that the Cu2S nanowires grown at the copper foil edges are usually longer and denser. The information about the emitting surface was obtained using a field emission microscope. Electrons passing the hole are projected to and form an emission image on a phosphor screen. Typical images consist of a number of bright diffuse spots. These spots are relatively stable, and may be attributed to emissions from the tip apexes of individual nanowires. Fig. 20 presents a plot of the emission current vs. field (I-E) for the Cu2S nanowire device. Field emission was observed at fields of ~6 MV/m, and a typical threshold field for obtaining a current density of 1 ␮A/mm2 is 11 MV/m. A current density of ~0.6 mA/cm2 may be obtained regularly from our films. The corresponding FowlerNordheim plot (inset of Fig. 23) were constructed. It is interesting to find that the F-N plot exhibits non-linearity, which could be explained by the diverse field enhancement factor of the nanowires. Specifically, the Cu2S nanowires have different radii and lengths, and therefore the field enhancements for different nanowires are varied, which would result in different turn-on fields for the individual nanowires. This conforms to the emission site imaging result, i.e., more emission sites are turned on at higher fields. The emission stability of Cu2S nanowires was recorded by measuring the current fluctuation with time at a fixed voltage (dc mode). The current fluctuation was as low as ⫾ 2% during 16 h of continuous operation at the current of about 100 ␮A (density of about 0.4 mA/cm2). In brief, these preliminary results indicate that the semiconductor nanowires of Cu2S as cold cathode have a potential future, and therefore warrant further detailed studies.

Fig. 20. Field emission current (I) vs. applied field (E) from the nanowire Cu2S film. Inset: the corresponding FN plot [81].

Sulphide Nanowires

233

5.2. Templates for the fabrication of semiconductor, metal, and conducting polymer nanowires Some sulphide nanowires have been used as templates for directing the growth of other nanowires. For example, CdS nanorods/nanowires have been used for the synthesis of ternary CdIn2S4 nanorods and CdS/CdSe core/shell nanowires [31, 51]. Because the Cu2S nanowires can be grown directly on a copper substrate in the form of array, it is possible to coat a layer of metal, semiconductor, and conducting polymer on the nanowires [89, 90]. Through chemical reactions, the Cu2S nanowires can also be converted to metal or other semiconductor nanowires [80, 91]. In particular, a gold layer has been successfully coated on the Cu2S nanowire surfaces by a simple redox deposition method. The success of this coating strategy hinges on the fact that the Cu2S nanowires are fairly good electric conductors. Basically, gold-coating on the Cu2S nanowires was accomplished via the redox reactions of galvanic cells. During the redox processes, the Cu2S nanowires act as cathodes where Au-deposition occurs, and the copper substrate serves as an anode where copper dissolution takes place. The electrolyte was the coating solution that contains HAuCl4. At least during the initial phase, the Cu2S nanowires short the circuits, ensuring the continuation of the goldcoating reactions. Later on, the gold sheath itself should constitute a much better conducting path. Because the reduction potential of AuCl4⫺/Au is much larger than that of Cu2⫹/Cu, the deposition of Au on the Cu2S nanowires and the simultaneous dissolution of Cu from the copper substrate were spontaneous. SEM images show the thickening of the nanowires at their tips after gold-coating. This is presumably due to the concentration gradient between the nanowire roots on the copper substrate and the nanowire tips towards the solution; the nanowire tips, being in touch with the bulk solution as opposed to the nanowire roots, are surrounded by more abundant Au3⫹ ions and thus grow faster. Typical TEM images of the Cu2S nanowires are shown in Fig. 21 before (a) and after the gold coating (b and c). Before coating, the nanowires are ~60 nm in diameter and have a relatively smooth surface (Fig. 21a). On immersing the nanowires in a coating solution for 1 h, the diameter of the nanowire was increased to ~100 nm (Fig. 21b, c). Although the nanowire surface became rougher after coating, it appears to be compactly covered by the gold deposit along the entire length of the nanowire. The roughness of the nanowire surface indicates that the gold coating is polycrystalline. It was believed that Au was deposited on the Cu2S nanowire surfaces in the form of Au nanocrystals, which are aggregated to form a continuous coating layer. The electron diffraction (ED) pattern of this core/sheath nanowire is shown in Fig. 21d. This ED pattern confirms the polycrystalline nature of the outer sheath of Au (ring pattern) as well as the single crystal nature of the inner core of Cu2S (discrete diffraction spots). Therefore, the monoclinic Cu2S nanowires are coated with face-center-cubic polycrystallites of Au. For the Cu2S/Au core/sheath nanowires, the core of Cu2S could be removed by etching with acid to form gold nanotubes. Specifically, this was achieved by immersing the Cu2S/Au core/sheath nanowire arrays on a copper foil into an acidic solution for an extended period of time. Fig. 22 shows the TEM picture of such a gold nanotube. It has a diameter of ~300 nm and a wall thickness of 50 nm. Clearly, one end of the gold tube is closed and the other is open. The ED pattern shows the polycrystalline fcc

234

Yang (a)

(b)

(c)

Cu2S Au

100 nm

100 nm

50 nm

(d) – Cu2S(313) Au(220)

Au(200) Au(311)

Au(111)

– Cu2S(132)

Fig. 21. TEM images of Cu2S nanowires before (a) and after (b and c) Au coating. (d) Electron diffraction pattern of a single Cu2S/Au core/sheath nanowire [89].

structure of the gold tube and the absence of the Cu2S crystals originally in the core. For thinner tubes with thinner walls, it was found that the nano-structure is easy to collapse after removing the core owing to the relatively weak aggregation forces between the gold nanoparticles. It was envisaged that such nanotubes may be used as nanoreactors or controlled drug delivery carriers. 5.3. Sensory devices, single molecule probe An obvious application of the Cu2S nanowire arrays is in the area of chemical and biological sensors. Works along this direction are being actively pursued in the author’s laboratory in collaboration with other research groups. The semiconducting

235

Sulphide Nanowires 1. Au(111) 2. Au(200) 3. Au(220) 4. Au(311)

34 12

200 nm

Fig. 22. TEM image of a gold nanotube after removal of the Cu2S core. Inset: ED pattern of the gold nanotube [89].

properties of the Cu2S nanowires, the nanoarray structure, the metal/semiconductor nanowire junction, and the flexibility for structural and surface modifications of the system will be exploited to develop practical sensors for various applications. In addition, the possibility to use photons to address individual Cu2S nanowires may lend them to single molecule probes with additional electrical controls.

6. Summary and Prospects To sum up, impressive progress has been made during the last few years in the synthesis of metal sulphide nanowires. An arsenal of different methods have been developed from the solvothermal method, VLS route, templated growth, molecular directed assembly, to the gas phase method. The successful synthesis of crystalline Cu2S nanowire arrays on a conductive copper surface using gas-solid reaction under ambient conditions has opened a new avenue for convenient synthesis and application of 1D sulphide materials. It is gratifying that measurements and applications of these novel sulphide nanowires are indeed emerging. By and large, however, the research on sulphide nanowires is still in the synthesis stage at the moment. It is reassuring that the recent rapid development in molecular electronics will facilitate the application of the sulphide nanowires that have been available due to the development of the synthetic methods [92]. Looking ahead, while further advance in the synthetic methodology for the sulphide nanowires with better control is expected, the nanowire assembly and the characterization of the electrical, magnetic, optical, and electronic properties of the sulphide nanowires appear to be desirable, challenging, and

236

Yang

fruitful endeavors. This will certainly give a great impetus to the applications of the sulphide nanowires.

Acknowledgments A group of students and visiting scholars at HKUST have done most of the works on the Cu2S nanowires. I would like to thank Dr. Z. R. Dai, Prof. Z. L. Wang, Prof. X.-Y. Li, Prof. N. Wang, Prof. J. N. Wang, Prof. W. K. Ge, Prof. Jun Chen, Prof. N. S. Xu, and many others for their collaboration on the characterization of the Cu2S nanowires. The work at HKUST was supported by a RGC grant administered by the UGC of Hong Kong.

References 1. Mukesh Jain (ed.), II–VI Semiconductor Compounds, World Scientific, Singapore (1993). 2. S. Larach (ed.), Photoelectronic Materials and Devices, Van Nostrand, Princeton, NJ, (1965); Handbook of Optics, vol. 1, McGraw-Hill (1995). 3. M. Ueta, H. Kanzaki, K. Kobayashi, Y. Toyozawa and E. Hanamura, Excitonic Processes in Solids, vol. 60, Springer Series in Solid State Sciences, Springer, Berlin (1986). 4. O. Madelung (ed.), Numerical Data and Functional Relationships in Science and Technology—Physics of Non-tetrahedrally Bonded Elements and Binary Compounds, III, vol. 17e, Springer, Berlin (1983). 5. Semiconductor Nanoclusters—Physical, Chemical, and Catalytic Aspects, Edited by P. V. Kamat and D. Meisel, Elsevier Science B. V., Amsterdam (1997). 6. P. M. Petroff, A. Lorke and A. Imamoglu, Phys. Today 54 (2001) 46. 7. J. Hu, T. W. Odom and C. M. Lieber, Acc. Chem. Res. 32 (1999) 435. 8. M. G. Bawendi, M. L. Steigerwald and L. E. Brus, Annu. Rev. Phys. Chem. 41 (1990) 477–496. 9. A. P. Alivisatos, J. Phys. Chem. 100 (1996) 13226–13239. 10. A. Eychmuller, J. Phys. Chem. B 104 (2000) 6514. 11. Ulrike Woggon, Optical Properties of Semiconductor Quantum Dots, Springer, Berlin (1996). 12. Special issue on quantum dots, Isr. J. Chem. 33 (1993). 13. M. S. Dresselhaus, D. Dresselhaus and P. C. Eklund, Science and Technology of Fullerenes and Carbon Nanotubes, Academic, New York (1996). 14. Nanotechnology: Research and Perspectives, Edited by B. C. Crandall and James Lewis, MIT Press, Massachusetts (1992). 15. R. Tenne and A. K. Zettl, Nanotubes from Inorganic Materials, in Carbon Nanotubes: Synthesis, Structure, Properties, and Applications, Edited by M. S. Dresselhaus, G. Dresselhaus and Ph. Avouris, Springer-Verlag, Heidelberg (2001) pp. 81–112. 16. R. Tenne, M. Homyonfer and Y. Feldman, Chem. Mater. 10 (1998) 3225. 17. W. U. Huynh, J. J. Dittmer and A. P. Alivisatos, Science 295 (2002) 2425–2427. 18. D. F. Evans and H. Wennerstrom (eds.), Surface energy of nanoclusters in The Colloidal Domain, Wiley-VCH, New York (1999). 19. N. Herron and D. L. Thorn, Adv. Mater. 10 (1998) 1173. 20. T. S. Ahmadi, Z. L. Wang, T. C. Green, A. Henglein and M. A. El-Sayed, Science 272 (1996) 1924–1926. 21. N. Wang, Z. K. Tang, G. D. Li and J. S. Chen, Nature 408 (2000) 50–51.

Sulphide Nanowires

237

22. K. Esumi, K. Matsuhisa and K. Torigoe, Langmuir 11 (1995) 3285. 23. P. V. Braun, P. Osenar and S. I. Stupp, Nature 380 (1996) 325. 24. W. Z. Wang, Y. Geng, P. Yan, F. Y. Liu, Y. Xie and Y. T. Qian, J. Am. Chem. Soc. 121 (1999) 4062. 25. R. S. Wagner and W. C. Ellis, Appl. Phys. Lett. 4 (1964) 89. 26. R. S. Wagner, VLS Mechanism of Crystal Growth, in Whisker Technology, Edited by A. P. Levitt, Wiley, New York (1970). 27. T. J. Trentler, K. M. Hickman, S. C. Goel, A. M. Viano, P. C. Gibbons and W. E. Buhro, Science 270 (1995) 1791. 28. Y. D. Li, H. W. Liao, Y. Ding, Y. Fan, Y. Zhang and Y. T. Qian, Inorg. Chem. 38 (1999) 1382. 29. S. H. Yu, Y. S. Wu, J. Yang, Z. H. Han, Y. Xie and Y. T. Qian, Chem. Mater. 10 (1998) 2309. 30. J. Yang, J. Y. Zeng, S. H. Yu, L. Yang, G. Zhou and Y. T. Qian, Chem. Mater. 12 (2000) 3259. 31. P. Yan, Y. Xie, Y. T. Qian and X. Liu, Chem. Comm. (1999) 1293. 32. S. H. Yu, J. Yang, Z. H. Han, Y. Zhou, R. Y. Yang, Y. Xie, Y. T. Qian and Y. H. Zhang, J. Mater. Chem. 9 (1999) 1283. 33. J. Zeng, Y. Zhu, J. Yang, Y. Liu and Y. T. Qian, Chem. Lett. (2001) 1000. 34. W. Zhang, Z. Yang, X. Huang, S. Zhang, W. Yu, Y. T. Qian, Y. Jia, G. Zhou and L. Chen, Solid State Commun. 119 (2001) 143. 35. Y. Jiang, Y. Wu, X. Mo, W. Yu, Y. Xie and Y. T. Qian, Inorg. Chem. 39 (2000) 2964. 36. Y. Jiang, Y. Wu, S. Yuan, B. Xie, S. Zhang and Y. T. Qian, J. Mater. Res. 16 (2001) 2805. 37. Y. Cui, J. Ren, G. Chen, Y. T. Qian and Y. Xie, Chem. Lett. (2001) 236. 38. J. Q. Hu, B. Deng, W. X. Zhang, K. B. Tang and Y. T. Qian, Int. J. Inorg. Mater. 3 (2001) 639. 39. C. Wang, K. Tang, Q. Yang, B. Hai, G. Shen, C. An, W. Yu and Y. T. Qian, Inorg. Chem. Comm. 4 (2001) 339. 40. J. D. Klein, R. D. Herrick II, D. Palmer, M. J. Sailor, C. J. Brumlik and C. R. Martin, Chem. Mater. 5 (1993) 902. 41. S. K. Chakarvarti and Vetter, J. Micromech. Microeng. 3 (1993) 57. 42. C. R. Martin, Science 266 (1994) 1961. 43. D. Routkevitch, T. Bigioni, M. Moskovits and J. M. Xu, J. Phys. Chem. 100 (1996) 14037. 44. D. Xu, Y. Xu, D. Chen, G. Guo, L. Gui and Y. Tang, Chem. Phys. Lett. 325 (2000) 340. 45. D. Xu, Y. Xu, D. Chen, G. Guo, L. Gui and Y. Tang, Adv. Mater. 12 (2000) 520. 46. P. V. Braun, P. Osenar and S. I. Stupp, Nature 380 (1996) 325. 47. P. V. Braun, P. Osenar, V. Tohver, S. B. Kennedy and S. I. Stupp, J. Am. Chem. Soc. 121 (1999) 7302. 48. X. Jiang, Y. Xie, J. Lu, L. Zhu, W. He and Y. T. Qian, Chem. Mater. 13 (2001) 1213. 49. C. N. R. Rao, A. Govindaraj, F. L. Deepak, N. A. Gunari and M. Nath, Appl. Phys. Lett. 78 (2001) 1853. 50. Z. Meng, Y. Peng, X. Chen and Y. T. Qian, Chem. Lett. (2001) 326. 51. J. Q. Hu, B. Deng, W. X. Zhang, K. B. Tang and Y. T. Qian, Inorg. Chem. 40 (2001) 3130. 52. M. Chen, Y. Xie, H. Chen, Z. Qiao, Y. Zhu and Y. T. Qian, J. Colloid Interface Sci. 229 (2000) 217. 53. J. Zhan, X. Yang, D. Wang, S. Li, Y. Xie, Y. Xia and Y. T. Qian, Adv. Mater. 12 (2000) 1348. 54. J. Zhan, X. Yang, S. Li, D. Wang, Y. Xie and Y. T. Qian, J. Cryst. Growth 220 (2000) 231. 55. M. Chen, Y. Xie, Z. Qiao, Y. Zhu and Y. T. Qian, J. Mater. Chem. 10 (2000) 329. 56. S. -H. Yu, M. Yoshimura, J. M. C. Moreno, T. Fujiwara, T. Fujino and R. Teranishi, Langmuir 17 (2001) 1700. 57. C. S. Yang, D. D. Awschalom and G. D. Stucky, Chem. Mater. 14 (2002) 1277. 58. S. H. Wang and S. H. Yang, Langmuir 16 (2000) 389. 59. S. H. Wang, S. H. Yang and K. K. Fung, Pure Appl. Chem. 72 (2000) 119. 60. J. Hu, T. W. Odom and C. M. Lieber, Acc. Chem. Res. 32 (1999) 435.

238 61. 62. 63. 64. 65. 66. 67. 68.

69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86.

87. 88. 89. 90. 91.

92.

Yang X. Duan and C. M. Lieber, Adv. Mater. 12 (2000) 298. Y. Wang, G. Meng, L. Zhang, C. Liang and J. Zhang, Chem. Mater. 14 (2002) 1773. R. Tenne, L. Margulis, M. Genut and G. Hodes, Nature 360 (1992) 444. Y. Feldman, E. Wasserman, D. J. Srolovitz and R. Tenne, Science 267 (1995) 222. A. Rothschild, G. L. Frey, M. Homyonfer, M. Rappaport and R. Tenne, Mater. Res. Innov. 3 (1999) 145. A. Rothschild, J. Sloan and R. Tenne, J. Am. Chem. Soc. 122 (2000) 5169. R. Rosentsveig, A. Margolin, Y. Feldman, R. Popovitz-Biro and R. Tenne, Chem. Mater. 14 (2002) 471. R. Tenne and A. K. Zettl, Nanotubes from inorganic materials, in Carbon Nanotubes: Synthesis, Structure, Properties, and Applications, Edited by M. S. Dresselhaus, G. Dresselhaus and Ph. Avouris, Springer-Verlag, Heidelberg (2001) pp. 81–112. S. H. Wang, Synthesis and characterization of nanoparticles, nanorods, nanowires, and nanocomposites, Ph. D thesis, the Hong Kong University of Science and Technology (2001). S. H. Wang and S. H. Yang, Chem. Phys. Lett. 10 (2000) 39. S. H. Wang and S. H. Yang, Adv. Mater. Opt. Electron. 322 (2000) 567. S. H. Wang and S. H. Yang, Chem. Mater. 13 (2001) 4794. S. H. Wang, S. H. Yang, Z. R. Dai and Z. L. Wang, Phys. Chem. Chem. Phys. 3 (2001) 3750. S. H. Wang and S. H. Yang, Mater. Sci. Eng. C 16 (2001) 37. N. Wang, K. K. Fung, S. H. Wang and S. H. Yang, J. Cryst. Growth 233 (2001) 226. Handbook of Chemistry and Physics (79th ed.), Edited by D. R. Lide, CRC Press, Boca Raton, FL (1998/1999). R. S. Larson, J. Electrochem. Soc. 149 (2002) B40. S. H. Wang, Lin Guo, Shihe Yang, Jing Zhao, Jing Liu and Zhonghua Wu, Mater. Chem. Phys. 75 (2002) 32. Jung Sang Suh and Jin Seung Lee, Chem. Phys. Lett. 281 (997) 384. S. H. Wang, Q. J. Huang, X. -Y. Li and S. H. Yang, Phys. Chem. Chem. Phys. 4 (2002) 3425–3429. Jun Chen, S. Z. Deng, N. S. Xu, Suhua Wang, Xiaogang Wen, Shihe Yang, Chunlei Yang, Jiannong Wang and Weikun Ge, Appl. Phys. Lett. 80 (2002) 3620. A. G. Rinzler, J. H. Hafner, P. Nikolaev, L. Lou, S. G. Kim, D. Tomanek, P. Nordlander, D. T. Colbert and R. E. Smalley, Science 269 (1995) 1550. W. A. DeHeer, A. Chatelain and D. Ugarte, Science 270 (1995) 1179. Y. Chen, D. T. Shaw and L. P. Guo, Appl. Phys. Lett. 76 (2000) 2469. S. H. Jeong, H. Y. Hwang, K. H. Lee and Y. S. Jeong, Appl. Phys. Lett. 78 (2001) 2052. N. S. Lee, D. S. Chung, I. T. Han, J. H. Kang, Y. S. Choi, H. Y. Kim, S. H. Park, Y. W. Jin, W. K. Yi, M. J. Yun, J. E. Jung, C. J. Lee, J. H. You, S. H. Jo, C. G. Lee and J. M. Kim, Diamond Relat. Mater. 10 (2001) 265. X. C. Ma, E. G. Wang, W. Z. Zhou, D. A. Jefferson, J. Chen, S. Z, Deng, N. S. Xu and J. Yuan, Appl. Phys. Lett. 75 (1999) 3105. J. M. Bonard, R. Kurt and C. Klinke, Chem. Phys. Lett. 343 (2001) 21. X. G. Wen and S. H. Yang, Nano Lett. 2 (2002) 451. W. X. Zhang, X. G. Wen and S. H. Yang, Chem. Mater. Langmuir 19 (2003) 4420. S. H. Wang, X. G. Wen and S. H. Yang, Copper nanowires prepared by the treatment of the Cu2S nanowires in a radio-frequency hydrogen plasma, in Nano Science and Technology— Novel Structures and Phenomena, Edited by Z. K. Tang and P. Sheng, Taylor & Francis Inc., London (2002) pp. 200–203. R. F. Service, Molecular electronics—breakthrough of the year, Science 294 (2001) 2442.

Chapter 13 Generalized Solution Synthesis of Large Arrays of Extended and Oriented Nanowires Jun Liu, Zhengrong R. Tian, James A. Voigt, Matthew J. Mcdermott and Bonnie Mckenzie Sandia National Laboratories, Albuquerque, NM 87185

Liang Liang eVionyx, 6 Skyline Drive, Hawthorne, NY 10532

1. Introduction Recently rational design and synthesis of extended and oriented nanostructures, such as oriented carbon nanotubes, semiconducting nanowires, oriented oxides, and metals have attracted wide attention. Nanostructured films made of oriented nanorods and high surface area architectures are desirable for many applications, including microelectronic devices, chemical and biological sensing and diagnosis, energy conversion and storage (photovoltaic cells, batteries and capacitors, and hydrogen storage devices), light emitting display devices, catalysis, drug delivery, separation, and optical storage. Many methods have been reported for the preparation of oriented nanostructures. For example, chemical vapor deposition is a widely used method to prepare oriented carbon nanotubes [1, 2], nanorods of ZnO [3, 4], Si [5], silicon carbide/nitride [6], and nanobelts of a wide range of oxide materials [7]. Besides the high temperature approach, a solution based synthesis method has been developed to prepare oriented nanorods of ZnO using a hydrothermal process [8, 9]. Oriented iron oxide nanorods were also prepared through the solution method by controlling the nucleation and growth of the minerals [10]. In contrast to synthetic materials, natural materials like seashells [11, 12] and diatoms [13], have developed sophisticated strategies to control the orientations and morphologies on multiple length scales using organic molecular directing agents to control the nucleation and growth of the inorganic minerals. These organics selectively initiate the nucleation of the desired inorganic phase, control the orientation and growth rate of the crystals, and determine the size, morphology and ordering of the inorganics. For example, the gastropod Haliotis rufescens (the red abalone) has a microstructure

240

Liu et al.

[14, 15] designed to optimize the mechanical properties of its shell [16–18]: it has an outer layer of column-like calcite, and an inter layer of laminated aragonite. The nacreous layer [19], also called the mother of pearl, are made of polygonal plates 0.5 ␮m in thickness and 15 ␮m in width, stacked together like miniature brick walls. The shell growth is mediated by two types of water soluble polyanionic proteins [20–24], those purified from the nacreous (aragonite) region, and those from the calcite region. The aragonite proteins and calcite proteins have different molecular weights, and different charge distributions on the molecules. The presence of the aragonite proteins would favor the formation of aragonite, and the presence of the calcite proteins (or the absence of any soluble proteins) would favor the formation of calcite [20–24]. Compared to inorganic nanostructures, the synthesis of soft organic nanostructured materials is a significant challenge. In particular, we are interested in oriented conducting polymer nanostructures, a very important class of electrochemically active materials [25, 26]. Electroactive and chemically active polymers with oriented nanostructures are intensely investigated for replacing Si based electronics and providing functionality and ability to interface with biomolecules. Several methods, including electrospinning [27, 28] and polymer templated electrochemical synthesis [29], have been used for preparing conducting polymer nanofibers. Highly porous conducting polymer films based on techniques like dip coating on porous supports have been widely investigated for separation and sensing [30], but the random pore structures and misalignment of the polymers are not ideal for high efficiency and faster kinetics. The controlled orientation is more critical for other applications such as in light emitting and microelectronic devices. To date, oriented conducting polymer nanostructures, including oriented polypyrrole or polyaniline nanorods or nanotubes, were mostly obtained with porous membranes as supporting templates [31–36]. In this approach, a substrate material with oriented nanoporosity is used as the mold (template) [37], which was filled with the desired polymer. Subsequently, the substrate is partially or completely removed to obtain the desired nanowires that replicate the size and the ordering of the template. The dimensions and the morphology of the polymer structures are defined (or limited) by the porous support. Recently Gao et al. [38] used oriented carbon nanotubes as the template to electrochemically deposit a thin polyaniline polymer coating on the surface of the carbon nanotubes. The formation of arrays of oriented polymer rods or tubes, using the membrane templating approach, involves carefully etching away the membrane without disturbing the conducting polymer structure. Published results indicate that oriented structures were only obtained for rods and tubes with a large diameter [39]. Etching away the membrane supports for nanorods or tubes with a diameter smaller than 100 nm caused the polymer to collapse into misoriented structures [39]. In addition, the necessity to use a porous support membrane limited the applicability of this approach to simple geometry or flat surfaces.

2. General Approach We are interested in developing a general solution synthesis route that can be used to reliably prepare oriented nanowires of a wide range of materials, including soft

Generalized Solution Synthesis

241

Fig. 1. Idealized diagram for nucleation and growth.

conductive polymers, and to be able to systematically control the orientation and morphology of these materials. This approach is based on a controlled nucleation and growth process. According to the classic theory of nucleation and growth [40], the free energy of forming stable nuclei on a substrate is determined by four factors: the degree of supersaturation S, the interfacial energy between the particle (c) and the liquid (l) ␴cl, the interfacial energy between the particle and the substrate (s) ␴cs, and the interfacial energy between the substrate and the liquid ␴sl: ⌬G ⫽ ⫺RT ln S ⫹ ␴cl ⫹ (␴cs ⫺ ␴sl) Acs

(1)

where Acs is the surface area of the particle. Fig. 1 is a schematic plot of the number of nuclei (N) as a function of degree of supersaturation (S) [40], which indicates that there is a narrow window where nucleation is favored for oriented nanostructures. Unfortunately most solution synthesis is carried out at too high a concentration so that undesirable precipitation dominates. As a result, oriented nanostructures were difficult to form. From Equation (1) and Fig. 9, we can use the following rules to design a generalized solution synthesis method for oriented nanostructures: (1) Controlling the solubility of the precursors and the degree of supersaturation so that massive precipitation is not the dominating reaction. Experimentally this is accomplished by reducing the reaction temperature, and by reducing the precursor concentrations as much as possible, while at the same time ensuring nucleation and growth can still take place. The formation of the new materials is characterized by the increase in cloudiness of the solution, which can be monitored by light scattering or turbidity measurement. A rapid increase in cloudiness is an indication of rapid precipitation and should be avoided. (2) Reducing the interfacial energy between the substrate and the particle. In literature surface functionalization is widely used to ensure that nucleation is favored on the substrate rather than in the solution. A more practical approach is to first generate large number of nuclei (seeds) of the desired material on the

242

Liu et al.

substrate surface to minimize the energy barrier for the subsequent growth of the nanostructures. (3) Ensuring the kinetic growth of oriented nanostructures is more favored over non-oriented structures. Instead of calcium carbonate that has been widely studied, here we use ZnO as a model material [41]. ZnO is a wide band gap semiconductor and has useful electronic and optical properties [4]. The novel structure discussed in this paper could lead to new approaches to control the orientation, surface area, and the defect structure that are critical in practical applications. In order to grow oriented ZnO nanowires, ZnO nanoparticles were first deposited on the substrate (Fig. 2a). Large arrays of oriented ZnO nanowires were formed upon mild heating in dilute Zn(NO3)3 solutions. Furthermore, an electrochemical deposition method is developed to prepare large arrays of oriented conductive polymer nanowires (Fig. 2b) [42, 43]. Although electrochemical deposition of conductive polymers seems to be drastically different from solution synthesis of oxides, the general principle of controlled nucleation and growth should be similar. In the first step, we used a high current density to generate the necessary nucleation centers on the substrate surfaces. After the completion of the first step, the current density was reduced twice for the polymers to grow from the nucleation sites. In order to control the morphology of the minerals, we use organic anions that selectively adsorb onto specific surfaces to promote or inhibit the growth of desired crystalline planes. The composition, structure and function of the structural directing proteins in biominerals are too complicated to be directly used for the synthesis of engineering materials. For example, in the nacreous layer of red abalone, at least four classes of soluble and insoluble proteins have been identified [44]. One of the objectives of this paper is to identify some simple organic molecules to control the morphology of the minerals, and to demonstrate in-situ morphology transitions as observed in natural materials.

Fig. 2. Schematic drawing of the steps for growing large arrays of oriented nanowires. (a) Schematics of seeded growth of ZnO. (b) Schematics of the nucleation and growth of polymer nanowires.

Generalized Solution Synthesis

243

3. Results and Discussions 3.1. Large arrays of oriented ZnO nanowires Fig. 3a through 3c show the scanning electron microscopy (SEM) images of large arrays of oriented ZnO nanorods formed by seeded growth [41]. Fig. 3a is a low magnification, face-on, image showing the uniformity of the microstructure. The hexagonal nanorods can be observed at a high magnification (Fig. 3b). The diameters of the nanorods range from 20 to 45 nm, and the length of the nanorods is about 3 ␮m. A tilted SEM image of the films reveals that most of the nanorods are highly oriented (Fig. 3c). In contrast, the same conditions without the ZnO nanoparticles as the seeds produced sparsely populated (Fig. 3d), randomly oriented ZnO rods, about 1 ␮m in diameter, and 10 ␮m in length. This result suggests that a high density of nucleation sites is critical for growing films of oriented nanorods. In order to understand how the arrays of oriented nanorods are formed, we followed the reaction kinetics by X-ray diffraction (XRD). As shown in Fig. 4, in the early stages of growth, oriented structures were not formed. The XRD patterns from 1 to 5 hours are characteristic of randomly oriented ZnO powders, showing (100) and (101) reflections as the main peaks. The (002) reflection was significantly enhanced after extended reaction, indicating that the ZnO started with randomly oriented orientation, and attained the preferred (002) orientation only after long term growth. This result is further supported by the electron microscopy images of the ZnO structures as

Fig. 3. SEM images of large arrays of oriented ZnO nanorods using seeded growth. (a) Low magnification, face-on view. (b) High magnification, face-on view. (c) Tilted SEM image. (d) Randomly oriented ZnO rods without using the seeds.

244

Liu et al.

Fig. 4. XRD patterns of the ZnO films as a function of time.

Fig. 5. Electron micrographs of ZnO on glass substrates. (a) TEM image of ZnO nanocrystals used as the seeds. (b) ZnO particles on the glass substrate after 30 minutes of reaction. (c) Roughened ZnO particles after 1 hour of growth. (d) ZnO rods after 3 hours of growth.

a function of time. Fig. 5a is the TEM image of the ZnO nanoparticles used as the seeds, containing rectangular and rod shaped ZnO particles with a wide size distribution. Fig. 5b shows the SEM image of the deposited ZnO film after 30 minutes of reaction, showing little sign of crystal growth. After one hour (Fig. 5c) of reaction, surface roughness became visible on the ZnO seeds, indicating the initiation of crystal growth. After 3 hours, short hexagonal rods were observed (Fig. 5d), although these rods were not well aligned yet.

Generalized Solution Synthesis

245

Fig. 2 through 5 confirmed the growth mechanism suggested in Fig. 2a, i.e., deposition of crystal seeds on the substrate surface, growth of randomly oriented crystals from the seeds, and growth of aligned nanorods. In the early stages of growth, ZnO crystals grow along the fastest growth orientation, the ⬍002⬎ direction, but these crystals are not aligned. However, as the rod-like crystal begin to overlap, only the crystals perpendicular to the substrate surface become kinetically favored. Randomly oriented crystals begin to overlap and further growth becomes unfavorable. This mechanism also explains why the crystals without seeds are not aligned (Fig. 3d). The low nucleation density without the seeds provides freedom for the crystals to grow in all possible orientations with respect to the substrate surface. 3.2. Large arrays of oriented conductive polymer nanowires The electrochemical deposition of conductive polymer (polyaniline) nanowires follow the same mechanism as ZnO (Fig. 2b) [42, 43]. We speculated that we would be able to grow oriented polymer nanowires if we could separate the nucleation and growth events. In the first step, we used a high current density to generate the necessary nucleation centers on the substrate surfaces. After the completion of the first step, the current density was reduced twice, in a step-wise fashion, for the polymers to grow from the nucleation sites. The morphology of the film by step-wise growth was examined under SEM. When viewed from an angle perpendicular to the surface at a low magnification (top view,

Fig. 6. SEM micrographs of oriented polyaniline on Pt. (a) Low magnification face-on. (b) High magnification face-on. (c) Tilted view, low magnification. (d) Tilted view, high magnification. The insert in Fig. 6a is an image of the oriented nanowires on a Si substrate.

246

Liu et al.

Fig. 6a), the film appears to contain uniform white spots all across the surface. At a higher magnification (Fig. 6b), it is revealed that the white spots are actually the tips of uniform nanowires, mostly oriented perpendicular with respect to the substrate. The diameters of the tips range from 50 nm to 70 nm. Some thin filament structures, 20 nm or so in diameter, can also be observed at the base of the oriented nanowires. When the sample is tilted, the morphology and orientation of the nanowires are clearly revealed (Fig. 6c and 6d). The oriented nanowires are fairly uniform in length and diameter, but the diameter of the tip is slightly smaller than that of the base. Judging from the tilt angle (about 40⬚), the nanowires are about 0.8 ␮m in length. We have also prepared similar oriented conducting polymer nanowires on other substrates, such as silicon, glass, gold, etc. The insert in Fig. 6a shows typical oriented polyaniline nanowire structures grown on Si wafers. Since the method reported in this paper does not involve porous membranes to support the polymer, it is applicable to surfaces of complex geometry. The oriented polymer nanowires will conform to the surface topologies of the substrate, and thus allow us to build hierarchical structures. Here we show two examples (Fig. 7a): the first one is oriented polymer nanowires grown on monolayers of colloidal silica spheres, and the second one is oriented polymer nanowires on a textured surface.

Fig. 7. Hierarchical nanostructures from oriented nanowires. (a) Growth of nanowires on silica spheres (top four figures) and on textured surfaces (bottom figure). (b) Radial growth of the nanowires from single silica particles. (c) The morphology of nanofibers on the edge of the colloidal silica monolayer. (d) Top view of the interconnected nanofibers. (e) Tilted view of the oriented nanofibers. (f) Low magnification of nanowires on textured surfaces. (g) High magnification of nanowires on textured surfaces.

Generalized Solution Synthesis

247

We first deposited a monolayer of closely packed silica spheres on the substrate and expected that the polymer nanowires would grow with an orientation perpendicular to the particle surfaces (not the substrate surface). In order to illustrate how the polymer grows from the silica particles, we first show the results in an area where the particles were not closely packed. Fig. 7b shows the surface roughness induced by the presence of silica particles, which follow the contours of the individual shperes. The radial growth of the polymer wires around the particles is clearly illustrated. Fig. 7c shows the morphology of the polymers across the edge of the silica monolayer. Close to the front where there are no silica particles, the polymers are oriented vertically, but on top of the silica particles the polymer orientation is disrupted and randomly connected. In the area occupied by a monolayer of densely packed silica spheres, the radial growth of polymer nanowires overlap and form three-dimensionally interconnected polymer networks (Figs. 7d to 7e). Oriented polyaniline nanowires were also prepared on textured surfaces as shown in Fig. 7f and 7g. The polyaniline nanowires on complex geometry may be useful for applications such as active filtration membranes and high surface area supports for sensors. Polarized infrared spectroscopy suggests an anisotropic molecular structure for the nanowires, indicating that not only are the polymer nanowires well oriented, the polymer molecules within the nanowires are also aligned. Here the polarized infrared spectra of the light reflected from the surface were collected in both the p- and s-polarizations, which allowed for the determination of the orientation of the polyaniline molecules by separately detecting bonds whose polarization vectors are perpendicular to the surface, rather than those that are parallel. Although there are five major features of the polyaniline spectra shown in Fig. 8a, three of them prove to be particularly useful in determining the orientation of the polymer chains. Peak A, at 3347 cm⫺1, is assigned to the N-H stretch [45, 46]. This peak is strong in the p-polarized spectrum, but almost completely missing in the s-polarized spectrum, implying that the N-H bonds are nearly perpendicular to the surface. The only structure that would allow all N-H bonds to be perpendicular to the surface is one where the polymer backbone is parallel to the surface, with the phenyl rings standing

Fig. 8. Polarized IR spectra of the oriented nanowires and schematic illustration of the growth process. (a) IR spectra. (b) Schematics of the polymer alignment.

248

Liu et al.

on-edge (Fig. 8b). This interpretation is supported by the fact that nearly all of the other absorptions are stronger in the s-polarized sample, particularly C and D, which are assigned to the C-N stretch and in-plane C-H bends respectively. In the proposed orientation, both of these peaks would be non-zero in both polarizations, but would be considerably stronger in the s-polarization. The N-H stretch has a second component at 3185 cm⫺1 that can be attributed to the hydrogen bonded amine groups. We believe that the hydrogen bonding between the polyaniline molecules disturbed the in-plane alignment of N-H stretch, and therefore this mode is allowed in both the p- and s-polarized modes. 3.3. Systematic morphology control We speculated that certain chelating agent such as citrate might have a strong influence on the crystal growth and crystal morphology of ZnO. Organic acids such as citrate play an important role in cell biology through complexing with trace element cations (Zn, Fe, etc.) [47]. We first studied the morphology of the ZnO crystals on glass surfaces, without using ZnO seeds, as a function of citrate concentrations. Without citrate ions, long ZnO rods were formed (Fig. 9a). When citrate ions were added, the ZnO rods became shorter and fatter (Fig. 9b). Further addition of citrate ions caused the formation of short and fat hexagonal crystals (Fig. 9c). In fact, the aspect ratio (crystal height/crystal width) is directly related to the citrate concentrations (Fig. 9d). These results suggest that citrate ions selectively bind to the {002}

Fig. 9. SEM images of ZnO particles as a function of citrate concentrations. (a) No citrate. (b) 0.2 mg sodium citrate in 30 ml solutions. (c) 1.5 mg sodium citrate in 30 ml solutions (d) Aspect ratio of ZnO particles as a function of citrate concentrations.

Generalized Solution Synthesis

249

surface, slowing down the crystal growth along the ⬍001⬎ orientation, and promoting the growth of other crystal planes. Therefore citrate ions provide a simple approach to control the aspect ratio of the ZnO nanorods. We have further demonstrated that much higher citrate concentrations can produce plate-like ZnO crystals, rather than rod-shaped particles. Fig. 10a is a high magnification SEM image of the ZnO crystals grown with a high citrate concentration, revealing flake like features on the (002) surfaces. We took the crystals in Fig. 10a as the seed crystal, and performed a secondary growth with a high citrate concentration. Although the new crystals in Fig. 10b and 10c retained the hexagonal shape, layered features were observed on the crystal surface, suggesting that crystal growth along ⬍001⬎ orientations was almost quenched, and the crystals were only allowed to grow side-wise in the form of thin platelets. These platelets are quite uniform in thickness (about 10 nm) throughout the sample. On the other hand, if the crystals in Fig. 10b were used as the seed crystal and placed in a solution without containing citrate ions, the (002) growth behavior was restored and the 10 nm layered features were “healed”. Figs. 9 through 10 are among the first examples to demonstrate in-situ morphology transition in synthetic ceramics, and is similar to what is observed in some biomaterials. As discussed earlier, abalone shells have a sophisticated mechanism to generate rod-to-plate transition using calcite or aragonite generating proteins. The reversible rod-to-plate transition observed in ZnO resembles the morphology transition in biogenic calcium carbonate, but without the use of a sophisticated protein molecule.

Fig. 10. Growth of plate-like ZnO structures at high citrate concentrations. (a) Fat ZnO crystals formed at a high citrate concentration, showing plate-like features on the (002) surfaces. (b) Secondary growth of plate-like ZnO crystals with citrate ions. (c) High magnification SEM image of 10 nm plate-like features on the (010) surfaces of the crystals. (d) “Healing” of the plate-like features without citrate ions.

250

Liu et al.

We used the oriented ZnO nanorods as shown in Figs. 3a through 3c as the starting materials, and performed secondary and tertiary growths in order to change the aspect ratio of the oriented nanowires. Fig. 11a through 11c confirmed continuous control of the aspect ratio using this approach. As compared with Fig. 3, the ZnO rods became much thicker (over 0.5 ␮m) after the secondary growth, but not much longer (still about 3 ␮m, as revealed by the cross-sectional SEM image in Fig. 11b), suggesting that the growth along ⬍002⬎ orientation was indeed inhibited by citrate binding. After a tertiary growth, the surface became densely populated with thick oriented ZnO rods, more than 0.8 ␮m in diameter (Fig. 11d). On the other hand, it is important to point out that if the secondary and tertiary growths were conducted without citrate, the overall crystal sizes increased, but the aspect ratio were not changed. In this case both the diameter and the length of the crystals were increased at the same time. The secondary growth at a much higher citrate concentration produced interesting by-layer structures that mimic the rod-to-plate transition in abalone sea shells (Fig. 12a). The bottom layer is made of ZnO nanorods, similar to Fig. 3, but when the secondary growth was performed with a high citrate concentration, the side-wise growth of the crystals quickly sealed off the interior of the film. However, the plate-like ZnO crystals continued to grow on top of the rods. This process produced a sandwich structure containing one layer of ZnO nanorods, and another layer of ZnO nanoplates. The bylayer structure is very similar to the column-to-plate transitions in sea shells (Fig. 12d). Large arrays of biomimetic helical ZnO nanostructures have also been obtained [48]. Extended helical or chiral nanostructures (helical nanowires, nanofibers, or nanoribbons) are usually associated with biomolecules such as proteins, polypeptides and their aggregates [49], but are not common in synthetic inorganic materials. In literature chiral crystal growth of calcium carbonate (calcite) is observed when

Fig. 11. Growth of thick ZnO nanorods by multiple growth with citrate ions. (a) Face-on view of arrays of thick ZnO rods through secondary growth. (b) Cross-sectional view of (a). (c) ZnO rods after tertiary growth in citrate solutions, tilted view. (d) ZnO rods after tertiary growth in citrate solutions, face-on view.

Generalized Solution Synthesis

251

Fig. 12. By-layer structure formed from secondary growth with high citrate concentrations. (a) Face-on view of the top ZnO plates. (b) Cross-sectional view of the bottom oriented nanorods and the top layered structures. (c) The aragonite plates in shells. (d) Column-to-plate transition in shells.

chiral molecules such as aspartic acid and polyaspartate are used to mediate the mineral growth [50, 51]. We used arrays of oriented ZnO rods (as shown in Fig. 11a) as the base material to grow the helical structures. Fig. 13a shows the scanning electron (SEM) micrograph of oriented fibrous structures grown on the (002) surface of the base ZnO crystals. The large hexagonal surfaces observed in the SEM image are the end (002) surfaces of the ZnO base crystals, indicating good alignment of the ZnO rods. At a higher magnification precisely aligned nanorods on the (002) surface of ZnO crystal are observed (Fig. 13b). Fig. 13b also indicate that these nanorods were formed from stacking of nanoplates with a uniform thickness of about 15 nm. These nanorods are about 1 ␮m in height, 500 nm in diameter at the base, and less than 30 nm in diameter at the tip. Furthermore, not only are these nanorods precisely aligned within one ZnO crystal, they are also well aligned across the whole sample because the base ZnO crystals are also aligned. Fig. 13c show that the side-width growth of the ZnO nanoplates leads to hexagonal ZnO plates that begin to overlap with one another. The inserts indicate that the structures observed here are formed by helical growth. The well-aligned column structure is also shown in Fig. 13d. The helical rod and column structures reported in this study are remarkably similar to the growth patterns of the nacreous calcium carbonate. In the early stage of growth of the nacreous layer of red abalone, large arrays of aligned calcium carbonate columns made of thin aragonite nanoplates are first observed (Fig. 13e). In the literature most theories for biomineralization of nacre emphasize the role of the organic phases [52, 53], whether these organic phases function as simple physical

252

Liu et al.

Fig. 13. SEM micrographs of helical ZnO nanostructures. (a) Large arrays of well-aligned helical ZnO nanorods on top of base ZnO rods. (b) A high magnification image of aligned helical ZnO nanorods on the (002) surface of one ZnO crystal. (c) A high magnification of the surfaces of the helical columns showing well-defined hexagonal plate-like structures. The inserts show the helical growth patterns on the surface. (d) High magnification image of oriented ZnO helical columns. (e) Nacreous calcium carbonate columns and layers near the growth tip of a young abalone.

compartments, or act to control nucleation or to terminate crystal growth by surface poisoning. In contrast, a mineral bridge mechanism was recently proposed in which the aragonite plates continuously grow from one layer to another through a central pore channel that connects the adjacent layers [54]. In the mineral bridging mechanism the role of the organic template in controlling the nucleation events as well as the crystal orientation is diminished (but not eliminated). Our results suggest that we can grow similar biomimetic structures in synthetic ceramics through a different helical growth mechanism. Here the column structures made of nanoplates grow continuously from one layer to another in a helical fashion, but an organic templating membrane is not required to grow the oriented and aligned nanoplates. The oriented structure and the morphology are direct results of the helical growth pattern.

4. Summary This Chapter discusses a generalized solution synthesis approach to prepare a range of extended and oriented nanostructures, including large arrays of oriented ZnO nanowires and soft conductive polymer nanowires. This approach involves controlling the nucleation and growth events by controlling the concentration of the soluble precursors, and increasing the number of nucleation density sites at the interfaces. The oriented ZnO and polymer nanowires were prepared through a seeded growth process. For ZnO nanowires, ZnO nanoparticles were coated onto the substrate as the nucleation seeds, while for conductive polymers a stepwise electrochemical deposition process was used. A large number of nuclei were first deposited using a large current

Generalized Solution Synthesis

253

density. The current density was subsequently reduced to grow oriented nanowires from the nucleation sites created in the first step. We further demonstrated that the solution synthesis approach can be used to systematically control the morphology of nanocrystalline materials by tailoring the surface chemistry of the crystals. As an example, morphology control of ZnO was achieved by selectively capping the fast {002} growth planes of the ZnO nanorods using citrate ions and by applying secondary and tertiary growth processes. The aspect ratio of the nanorods can be varied over a wide range. Furthermore, in-situ transition from rod growth behavior to plate growth behavior, and large arrays of oriented helical nanostructures (helical nanowires and nanocolumns) are observed. These unusual microstructures show remarkable resemblance to biogenic calcium carbonate, and thus may shed new light on biomineralization. The oriented nanoscale materials have potentials for a wide range of applications such as microsensors for chemical and biological agents.

Acknowledgment This work is partially supported by the Sandia National Laboratories (SNL) Laboratory-Directed Research and Development Program (LDRD), and by the Division of Materials Sciences and Engineering, Office of Basic Energy Sciences, U.S. Department of Energy. Sandia National Laboratories is a multi program laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the Department of Energy under Contract DE-AC04-94AL85000.

References 1. W. Z. Li, S. S. Xie, L. X. Qian, B. H. Chang, B. S. Zou, W. Y. Zhou, R. A. Zhao and G. Wang, Science 274 (1996) 1701. 2. Z. F. Ren, Z. P. Huang, J. W. Xu, J. H. Wang, P. Bush, M. P. Siegel and P. N. Provencio, Science 282 (1998) 1150. 3. M. H. Huanh, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind, E. Webber, R. Russo and P. Yang, Science 292 (2001) 1897. 4. Y. Wu, H. Yan, M. Huang, B. Messer, J. H. Song, P. Yang, Chem. Eur. J 8 (2002) 1261. 5. D. P. Yu, Y. J. Xing, Q. L. Hang, H. F. Yang, J. Xu, Z. H. Xi and S. Q. Feng, Physica E 9 (2001) 305. 6. L. C. Chen, S. W. Chang, C. S. Chang, C. Y. Wen, J.-J. Wu, Y. F. Chen, Y. S. Huang and K. H. Chen, J. Phys. Chem. Solids 62 (2001) 1567. 7. Z. W. Pan, Z. R. Dai and Z. L. Wang, Science 291 (2001) 1947. 8. L. Vayssieres, K. Keis, A. Hagfeldt and S.-E. Lingdquist, Chem. Mater. 13 (2001) 4395. 9. L. Vayssieres, K. Keis, S. E. Lingdquist, A. J. Hagfeldt, Phys. Chem. B 105 (2001) 3350. 10. L. Vayssieres, N. Beermann, S. E. Lindquist, A. Hagfeldt, Chem. Mater. 13 (2001) 233. 11. A. P. Kackson, J. F. V. Vincent and R. M. Turner, Proc. Royal Soc. London B 234 (1998) 415. 12. A. M. Belcher, X. H. Wu, R. J. Christensen, P. K. Hansma, G. D. Stucky and D. E. Morse Nature 381 (1996) 56. 13. M. Sumper, Science 295 (2002) 2430. 14. K. M. Wilbur and G. Owen, “Growth”, in Physiology of Mollusca, Edited by K. M. Wilbur, C. M. Yonge, pp. 243–282, Academic Press, New York, (1964).

254

Liu et al.

15. K. M. Wilbur, in Physiology of Mollusca, Edited by K. M. Wilbur and C. M. Yonge, Academic Press, New York (1964), pp. 211–242. 16. A. P. Kackson, J. F. V. Vincent and R. M. Turner, Proc. Royal Soc. London, B. 234 (1998) 415. 17. J. D. Currey, in The Mechanical Properties of Biological Materials, Vol. XXIV, Edited by J. F. V. Vincent, J. D. Currey, Cambridge University Press, Cambridge (1980), pp. 75–97. 18. A. G. Evans and D. A. Marshall, Acta Met. 37 (1989) 2567. 19. G. Grégoire, in Nautilus, The Biology and the Paleobiology of a Living Fossil, Edited by W. B. Saunders, N. H. Landman, Plenum Press, New York (1987) pp. 463–486. 20. M. Fritz, A. M. Belcher, M. Radmacher, D. A Walters, P. K. Hansma, G. D. Stucky, D. E. Morse and S. Mann, Nature 371 (1994) 49. 21. C. M. Zaremba, A. M. Blcher, M. Fritz, Y. L. Li, S. Mann, P. K. Hasma, D. E. Morse, J. S. Speck and G. D. Chem. Mater. 8 (1996) 679. 22. A. M. Belcher, P. K. Hansma, G. D. Stucky and D. E. Morse, Acta Mater. 46(3) (1998) 733. 23. M. Fritz and D. E. Morse, Current Opinion in Colloidal and Interface Science 3(1) (1998). 24. A. M. Belcher, X. H. Wu, R. J. Christensen, P. K. Hansma, G. D. Stucky and D. E. Morse, Nature 381(2) (1996) 56. 25. A. G. MacDiarmid, Rev. Mod. Phys. 73 (2001) 701. 26. K. Doblhofer and K. Rajeshwar, Handbook of Conducting Polymers, Marcel Dekker, Inc., New York (1998). 27. J. Doshi and D. H. Reneker, J. Electros. 35 (1999) 151. 28. D. H. Renek, A. L. Yarin, H. Fong and S. Koombhongse, J. Appl. Phys. 87 (2000) 4531. 29. C. Jérôme and R. Jérôme, Angew. Chem. Int. Ed. 37 (1998) 2488. 30. H. S. Lee and J. Hong, Syn. Met. 113 (2000) 115. 31. A. Huczko, Appl. Phys. A 70 (2000) 365. 32. C. R. Martin, Chem. Mater. 8 (1996) 1739. 33. M. Nishizwa, K. Mukai, S. Kuwabata, C. R. Martin and H. Yoneyama, J. Electrochem. Soc. 144 (1997) 1923. 34. S. De. Vito and C. R. Martin, Chem. Mater. 10 (1998) 1738. 35. S. M. Marienakos, L. C. Brousseau III, A. Jones and D. L. Feldheim, Chem. Mater. 10 (1998) 1214. 36. C.-G. Wu, T. Bein, Science 264 (1994) 1757. 37. A. Huczko, Appl. Phys. A 70 (2000) 365. 38. M. Gao, S. Huang, L. Dai, G. Wallace, R. Gao and Z. Wang, Angew. Chem. Int. Ed. 39 (2000) 3664. 39. J. Duchet, R. Legras and S. Demoustier-Champagne, Syn. Met. 98 (1998) 113. 40. B. C. Bunker, P. C. Rieke, B. J. Tarasevich, A. A. Campbell, G. E. Fryxell, G. L. Graff, L. Song, J. Liu and J. W. Virden, Science 264 (1994) 48. 41. Z. R. Tian, J. A. Voigt. J. Liu, R. T. Cygan, L. Criscenti, M. J. Mcdermott, J. Am. Chem. Soc. (2002) (submitted). 42. L. Liang, J. Liu, C. F. Windisch, Jr., G. J. Exarhos and Y. Lin, Angew. Chemie. Int. Ed. 41 (2002) 3665. 43. J. Liu, Y. Lin, L. Liang, J. A. Voigt, D. L. Huber, Z. R. Tian, E. Coker, B. Mckenzie and M. J. Mcdermott, Chem. A Eur. J. (2002) (in press). 44. X. Shen, A. M. Belcher, P. K. Hansma, G. Stucky and D. E. Morse, J. Biochem. 272(51) (1997) 32472. 45. A. J. Milton and A. P. Monkman, J. Phys. D:Appl. Phys. 26 (1993) 1468. 46. M. Angelopoulos, R. Dipietro, W. D. Zheng, A. G. MacDiarmid, A. J. Epstein, Syn. Met. 84 (1997) 35. 47. W. E. Rauser, Cell Biochem. Biophys. 31 (1999) 19.

Generalized Solution Synthesis

255

48. Z. R. Tian, J. A. Voigt, J. Liu, B. Mckenzie and M. J. Mcdermott, J. Am. Chem. Soc. (2002) (submitted). 49. B. Alberts, D. Bray, A. Johnson, J. Lewis, M. Raff, K. Roberts and P. Wlater, Essential Cell Biology, Garland Publishing Inc, New York and London (1997). 50. C. A. Orme, Nature 411 (2001) 775. 51. L. A. Gower and D. A. Tirrel, J. Cryst. Growth, 191 (1998) 153. 52. Example include: (a) S. Weiner and M. Wilbur, Phil. Trans. R. Soc. Lon., B 304 (1984) 425; (b) L. Addadi and S. Weiner, Proc. Natl. Acad. Sci. 82 (1985) 4110. (c) G. Falini, S. Albeck, S. Weiner and L. Addadi, Science 271 (1996) 67. 53. A. M. Belcher, X. H. Wu, P. K. Christensen, P. K. Hansma, G. D. Stucky and D. E. Morse, Nature 381 (1996) 56. 54. T. E. Schäffer, C. Ionescus-Zanetti, R. Proksch, M. Fritz, D. A. Walters, N. Almqvist, C. M. Zaremba, A. M. Belcher, B. L. Smith, G. D. Stucky, D. E. Morse, P. K. Hansma, P. K. Chem. Mater. 9 (1997) 1731.

Chapter 14 Composite Nanowires Yuegang Zhang Intel Corporation, SC2-24, 2200 Mission College Blvd, Santa Clara, CA 95054

1. Introduction The discovery of carbon nanotubes [1] has opened a new field of research on onedimensional nanomaterials. In addition to the nanotubes made of pure carbon, scientists also explore the possibility to make similar structure with more than one element. Nanotubes made of compounds, such as W2S [2], Mo2S [2, 3], BN [4–7], have been reported. More complicated nanotubes made of B–C–N ternary system have also been synthesized [8–12]. Although theoretical calculations suggest ternary compounds, such as BC2N, could form stable tubular structure, delicate microanalysis has revealed that phase separation exists in these ternary systems [12, 13]. The phase separation makes the nanotube a “composite” rather than a “compound”. The situation in non-tubular nanowires is very similar. In addition to the nanowires made of single elements [14–16] and compounds [17–19], multi-elemental nanowires with separated phases have also been produced. One representative structure of the composite nanowires is the coaxial nanocable which is composed of a core nanowire and one or more “sheath” layers [20]. Many composite nanowires contain one or more phases of nanotubes, and many of them are made by similar methods used to synthesize nanotubes. Therefore, in this chapter, the definition of composite nanowires covers both multi-phase solid nanowires and hollow nanotubes, or a combination of them. Identification of different phases in composite nanowires requires information of their crystal structure and elemental distribution. The former is generally obtained by high-resolution electron microscopy (HREM) and electron diffraction. The latter is by electron energy-loss spectroscopy (EELS) or energy-dispersive X-ray spectroscopy (EDX).

2. Phase Separation in Multi-Elemental Nanotubes Phase separation in multi-elemental nanotubes has so far been confirmed only in boron–nitrogen–carbon (B–C–N) system [12, 13]. B–C–N nanotubes have attracted

258

Zhang

much interests because theoretical calculations predict that their electronic properties are mainly determined by their chemical composition rather than their diameters and chirality [21, 22]. This feature could make the electronic properties of B–C–N nanotubes more controllable than their pure carbon counterpart and makes them potential candidates for nanoelectronic components. The theoretical results certainly applies only with single-phase material thus the structure determination is critical for controllable synthesis of B–C–N nanotubes. All compounds from B–C–N systems, such as BN, BC2N, BC3, however, have a very similar crystal structure with graphite. Phase separation could happen between two adjacent graphitic layers spaced only about 0.35 nm since B–C–N nanotubes are normally obtained in a multi-walled type. Determination of the structure normally requires HREM combined with elemental analytic technique with extra-high spatial resolution. A modern scanning transmission electron microscope (STEM) equipped with parallel electron energy-loss spectroscopy (PEELS) can give a spatial resolution in sub-nanometer scale and thus being widely used for characterizing B–C–N nanotubes and other multi-elemental nanowires. Earlier reports on the synthesis of B–C–N nanotubes mainly used arc-discharge method with an anode containing B, C and N [8–10]. The products contained mainly carbon with very low concentration of boron and nitrogen. Elemental analysis of the nanotubes in the products showed inhomogeneous distribution of B, N and C in the radial direction with normally B and N rich at outer surface layer [8, 10]. The B : N ratio was close to 1 : 1. The results, however, failed to provide a clear conclusion of phase separation. Suenaga and his co-workers improved the arc-discharge synthesis by using HfB anode and a carbon cathode in nitrogen atmosphere [13]. Since cathode is only slightly vaporized during arc-discharge, obtained nanotubes have higher boron and nitrogen concentration. Clear phase separation of BN and C has been observed for these nanotubes. Fig. 1A shows HREM image of a multi-walled nanotube (MWNT) obtained from the arc-discharge. Elemental profiles measured by scanning electron beam across such a tube in a STEM-PEELS system shows four maxima in the C profile and two maxima in the B and N profiles (Fig. 1B). The positions of the two maxima in the B and N profiles coincide with two minima of the C profile (Fig. 1B). In a cross-section profile of single-phase cylindrical structure, maximum only occurs at the position that corresponds to the inner edge parallel to the probing electron beam. This is because this position has the largest equivalent thickness through which electron beam transmits and loses its energy. For a single-phase nanotube, there should be only two maxima in the profile (probe size of the STEM, 0.5–1 nm, cannot resolve individual shells in a MWNT). Thus the profiles in Fig. 1B give a strong indication of a C–BN–C three-layered tubular structure as modeled in Fig. 1D. Simulated elemental profiles from the model agree with the experimental results very well and provide a strong support of the phase separation with sharp interfaces. Another commonly used method to produce B–C–N nanotubes is laser ablation [12]. The advantage of this method to arc-discharge is that it does not require the starting materials to be conductive and the ratio of vaporized elements can be easily controlled by the composition of the target. The laser ablation of a BN and C composite target with (B ⫹ N) : C ⫽ 1 : 1, however, does not give the same elemental ratio in the produced nanotubes. The results obtained by STEM-PEELS show all nanotubes are C rich, while there are a lot of BN crystallites as co-products. The tubes with a few

Composite Nanowires

259

Fig. 1. (A) HREM image of a composite nanotube. (B) Elemental profiles measured across a tube similar to that shown in (A). (C) Simulated profiles based on the model in (D). From Ref. [13].

walls are normally composed of pure carbon. Those with more walls are normally not uniform in diameter along their axis as well as the B and N distribution (Fig. 2). The distribution of B and N is rich at the outer surface layers similar to the BCN nanotubes produced by earlier arc-discharge method [8, 10]. For the tube with an elemental profile across its diameter shown in Fig. 3, the best model is composed of six inner layers of pure carbon and seven outer layers of BC7N. However, it is unable to determine whether the B and N are homogeneously doped into graphite lattice (true ternary BC7N phase) or they are mixed binary phases such as BN or BCx embedded in the graphite outer layers (sub-composite tubes). The laser ablated B–C–N nanotubes not only have similar morphologies and element distribution as those in earlier arc-discharged products, they should also share a similar growth mechanism since both methods employ a local high-temperature process to vaporized graphite and BN. From the feature that all obtained composite

260

Zhang

Fig. 2. TEM images of a pure carbon MWNT (a) and B–C–N composite nanotubes (b and c) produced by laser ablation. The thickened section indicated by the solid arrow in (b) and the section II and III in (c) have higher B concentration than other part of the tubes. From Ref. [12].

Fig. 3. Elemental profiles measured across a composite B–C–N nanotube produced by laser ablation. From Ref. [12].

nanotubes have carbon cores, a two-step growth model has been proposed [12]. The first step is the nucleation and growth of pure multi-walled carbon nanotubes. This step occurs in the high temperature plasma region where B and N can be easily diffused out of the graphite lattice and form more favorable BN crystallites. Carbon nanotubes formed in the process are all multi-walled because the presence of B and N atoms in this step could introduce defects on the graphitic wall that are preferable nucleation sites for additional layers and the foreign atoms could also serve as the bridging atoms in lip–lip growth model [23]. The second step is partial coating of B–C–N layers. The growth of additional walls could also initiate from the defects on the carbon nanotube wall. The B and N diffusion process is reduced by lower temperature when the nanotube moves out of the plasma center. The inert gas used in the arc-discharge and laser ablation processes provides an efficient heat exchange medium for cooling down the species and works as a barrier to confine the expansion of the plasma and sustain a

Composite Nanowires

261

high density of multi-element vapor for doping BN in nanotubes. The two-step model, however, is not adequate to explain the formation of the C–BN–C nanotubes where a simultaneous growth of multi-phase shells through a lip–lip-interaction is more favorable [23].

3. Filling in Carbon Nanotubes A simple and intuitive way to make composite nanowires is to fill carbon nanotubes with a different material. The hollow space in carbon nanotubes, especially in MWNTs, provides a perfect mold to form a solid nanowire inside. Theoretical simulation shows that opened nanotubes could be filled with liquid by capillary phenomenon [24]. The as-produced carbon nanotubes, however, normally have closed caps at their ends. Filling of nanotubes, therefore, always relies on a method to open the nanotubes. The first filling experiment reported by Ajayan and Iijima [25] used an oxidation method to simultaneously open and fill MWNTs. Metallic Pb was first deposited onto samples containing MWNTs produced by arc-discharge method. The samples were then heated in air to about 400 ⬚C which is above the melting point of Pb. HREM images taken after the heat treatment show that MWNTs are opened at their tips and filled with solid materials (Fig. 4). Energy-dispersive X-ray (EDX) analysis indicated the filled material contained Pb. The exact phase of the filled material in MWNTs however, was not determined. Ajayan and co-workers [26] pointed out that the temperature used in the Pb filling experiment is much lower than that necessary to open MWNTs in air without metal. The authors suggested that metal or its oxide could play a role as catalyst in the onestep process to open and fill the nanotubes. They demonstrated MWNTs could also be filled with metallic Bi using a similar method. They further compared the one-step filling process with a two-step process in which nanotubes are first opened by oxidation in air and then filled with pure metal by heating in vacuum. They found that the ratio of filled and unfilled nanotubes in the two-step process is much lower than the one-step process. A simple explanation for this difference was that the capillarity filling may be obstructed by the contaminants sucked into the cavities during opening

Fig. 4. HREM images of MWNTs filled with Pb compound (dark contrast). From Ref. [25].

262

Zhang

process in the two-step case [26]. A later study, however, indicated that it could be due to the wetting properties of the metals and nanotubes [27]. Wetting of nanotubes requires the surface tension of a liquid less than a cut-off value around 200 millinewtons per meter [27, 28]. Most pure metals (including Pb) do not fulfill this requirement and thus will not be drawn into the inner cavity of nanotubes through capillarity. In a certain environment such as high temperature and in the presence of air, the metals might have formed a compound that has a low surface tension and thus enabling the filling by capillarity. While pure metals are normally oxidized during by pre-described one-step filling process and thus resulting a different phase after filling, some oxides with a low-melting-temperature and low surface tension, such as V2O5 and MO3, can be filled into nanotubes directly without a phase change [29, 30]. V2O5 was found highly crystallized and showing a preferred orientation within MWNTs of larger inner diameters. Because of the very low surface tension, V2O5 can also form a uniform coating on the surface of nanotubes (Fig. 5) or even in between graphitic shells in MWNTs [29]. On the other hand, crystalline MO2 nanowires and outer-layer coating can be obtained through a hydrogen reduction treatment following the filling of MO3 in pre-opened nanotubes [30]. In addition to the direct filling of nanotubes with molten metals or oxides, there is another wet-chemical method that generally uses strong acids (e.g. HNO3, HF/BF3) or other oxidants (e.g. OsO4) to open nanotubes and fill the tubes with metal salts [31–34]. An ex-situ heat treatment, however, is necessary to evaporate the solvent and reduce (decompose) the salts in order to obtain solid phase metals or oxides in the nanotubes. Since the solvent could occupy a considerable volume of the inner cavity of nanotubes during filling, the resulting solid materials are normally discrete nanoparticles instead of continuous nanowires. Besides the pre-described physical and chemical methods of filling pre-produced carbon nanotubes, elements or compounds can also be simultaneously encapsulated during synthetic processes of carbon nanotubes. Such encapsulation has been observed in MWNTs synthesized by arc-discharge [35–39], electrolysis [40, 41], and field-anode activation [42]. Among these processes, the arc-discharge produces the best-graphitized sheathing MWNTs, which is also true in synthesizing hollow carbon

Fig. 5. HREM images and illustration of oxide filling in and coating on MWNTs. From Ref. [29].

Composite Nanowires

263

nanotubes, and is more versatile for encapsulation of different elements. Encapsulation abilities of more than 41 elements have been systematically investigated using an arc-discharge method in which the graphite rod for the anode was drilled and filled with a mixture of graphite and chosen element powders [37, 39, 43]. The study shows that 12 elements (Cr, Ni, Re, Au, Sm, Gd, Dy, Yb, S, Ge, Se, Sb) can completely fill long MWNTs and form continuous nanowires (Fig. 6), while others can only give partial fillings. Further study, however, reveals that the growth of nanowires in nanotubes may result from the presence of impurities in the graphite rods used in the arc-discharge [44]. STEM-EELS elemental analyses, HREM, and selected area electron diffraction (SAED) analyses showed that most nanowires were metal sulfides instead of pure metals or metal carbides (with the exception for pure Ge nanowires). Sulfur found here is from impurities in the graphite rods. For arcdischarge involving sulfur itself, obtained fillings were iron sulfides. Again, iron is from impurities in the graphite rods. Control experiments using high purity carbon rods and pure doping elements resulted in empty nanotubes without encapsulation. Intentionally adding sulfur together with metals resulted in abundant formation of nanowires encapsulated in nanotubes. The role of sulfur has been suggested as lowering solidification temperature and surface tension by forming sulfides with metals and thus facilitating nanotube fillings [43]. Other elements such as selenium, hydrogen and oxygen may have a similar effect on the filling process. The phase of the nanowires formed by filling carbon nanotubes are thus controlled by a complex catalytic process and usually different from the material we used for the filling.

Fig. 6. TEM images of filled MWNTs produced by arc-discharging using graphite together with Cr (a, e), Ni (b), Dy (c) and Yb (d). From Ref. [37].

264

Zhang

4. Coaxial Nanocables A coaxial nanocable is generally defined as a multi-layered structure with a core nanowire sheathed coaxially by one or more cylindrical layers of different materials. It is a miniature analogue of a conventional coaxial cable used for video or audio signal transmission in our daily lives. A coaxial nanocable could, however, function as an electric-transmitting cable only if it consists a conductor (semiconductor) nanowire core and two or more alternating insulating–conducting outer layers. The general definition, however, includes composite nanotubes and filled carbon nanotubes described in previous sections. We thus limit our definition to a special category of coaxial nanocables which consists a solid core nanowire and at least one layer of solid phase. This category of nanocables is usually synthesized through a simultaneous phase separation during nanowire synthesis processes. The solid phases of distinctive crystal structures other than graphitic ones make the structure characterization of the nanocables much easier by TEM than that of composite nanotubes. The direct observation of phase separation by TEM can give unambiguous evidence of uniform phase distribution along the coaxial nanocables. The first sophisticated coaxial nanocable structure was synthesized by laser ablation of a composite target containing BN, graphite, and small amount of SiO and Li3N at high temperature [20]. Fig. 7 shows the transmission electron microscope (TEM) images of the fabricated coaxial nanocables. They are normally a few tens of

Fig. 7. HREM images and selected area electron diffraction pattern of SiC–SiO2–(BN)xCy coaxial nanocables. From Ref. [20].

Composite Nanowires

265

micrometer long and 10–100 nm in their diameters. The diameters are quite uniform throughout the whole length (Fig. 7A). High-resolution TEM images in Fig. 7B and C indicate that the core of the nanowire has a crystalline phase whose electron diffraction pattern (left bottom inset of Fig. 7B) and lattice image (right bottom inset of Fig. 7C) fits well with that of cubic SiC. The intermediate layer shows an amorphous phase. The outer-most layer consists of graphite multi-walls. A detailed elemental profile analysis across the nanocable by STEM-PEELS reveals the Si : C ratio in the core wire and Si : O ratio in the intermediate layer is 1 : 1 and 1 : 2, respectively. The outer layer of the nanocable is composed of C and BN with signature of inter-shell phase separation similar to that of composite nanotubes described in the previous section. The solid phases in the core and intermediate layer region has been further confirmed by comparing the Si L-edge fine structures with those of SiC and SiO2 reference spectra (Fig. 8). The characterization, therefore, gives a well-defined composite nanowire structure with SiC–SiO2–(BN)xCy multi layers forming a semiconductor–insulator–(semi)conductor coaxial nanocable. The formation of the nanocables might involve two steps. The first is the formation and simultaneous phase separation of SiC–SiO2 nanowire from vaporized C and SiO. The second step is the coating of C and BN layers. While Li has not been detected in the nanocable, it plays a very important role in the formation of the outer graphitic layers. Without adding Li3N in the starting material, a two-layer SiC–SiO2 nanowire structure was obtained. The role of Li3N has given a clue to find a way that could lead to controllable fabrication nanocables with designed functions. Comparing to other composite nanotubes, coaxial nanocables are easy to be synthesized with relatively high purity and with high uniformity along their length.

Fig. 8. (A) A schematic illustration of a coaxial nanocable. (B) Si–L edge fine structure obtained from a nanocable with electron probe positions as indicated by arrows in (A). Spectrum c is obtained by subtracting a from b, representing a contribution from the core of the nanocable. Reference spectra from pure SiO2 and SiC phases are displayed for comparison. From Ref. [20].

266

Zhang

Following the first demonstration of SiC–SiO2–(BN)xCy coaxial nanocable, other similar structures, such as Si–SiO2–C [45], Si3N4–Si–SiO2 [46], Fe–C–BN [47], have been synthesized by carbothermal reduction, laser ablation, CVD, or a combination of these methods. It has also been demonstrated that SiO2 could easily coat SiC, Si during their fabrication process to form simple two-layer cables [14, 48, 49]. With a few exceptions [50, 51], most SiO2 layer are coaxial with the core nanowires and thus sheathing the whole core. In addition to the high temperature methods involving vapor phase reaction, several solution-based methods with relatively lower temperature have been reported to produce polymer nanocables [52–54]. A sol–gel process has also been developed to directly coat silver nanowires with silica [55]. This technique could be useful for making insulating layer when the nanowires is integrated into electronic devices.

5. Summary and Prospect of Composite Nanowires This chapter introduces composite nanowires composed of different materials. Phase separation in the radial direction makes composite nanowires different from other multielemental nanowires such as compound nanowires. The phase distribution is normally homogeneous along the axial direction of the wires. Coaxial nanocable is a representative structure of the composite nanowires. A nanocable is typically composed of a core nanowire sheathed by one or more outer layers similar to a coaxial cable used in signal transmission but with a diameter in nanometer scale. There are many ways to produce composite nanowires, from physical and chemical filling of carbon nanotubes to self-assembly using laser ablation, chemical vapor deposition, or arc-discharge methods. The sizes of the composite nanowires are normally less than a few tens of nanometers in diameter and more than several microns in length. For some structures, phase separation could be observed directly using high-resolution electron microscopy. For others, more sophisticated analytical techniques, such as electron energy-loss spectroscopy with sub-nanometer spatial resolution, have to be used to determine the structure. Composite nanowires, especially coaxial nanocables, provides a great possibility to take advantage of different function and properties of different materials within a single nanoscale component. As a reinforcement material, it could be designed to have a core of high mechanical strength and an outer layer that could form strong bonding with the matrix. As an electronic material, a properly designed conductor–insulator– conductor coaxial nanocable could be used to carry electric signal as a conventional cable, or as a quantum cable is the barrier layer is thin enough [56]. A semiconductor– insulator–conductor could be used as a metal-oxide-semiconductor field effect transistor (MOSFET), which is the main component in modern microprocessor. An outer layer such as SiO2 or BN outer layer will not only provide electrical insulation but also work as barrier to protect the functional nanowire from potential chemical or mechanical damage. Utilizing the different dielectric properties of the different layers, coaxial nanocables could also be used as optical wave guide. The application of composite nanowires, however, strongly rely on the ability to control the synthesis for a predesigned structure. For electronic and optical application, we also face the same problems as we do for carbon nanotubes, that is manipulation, integration and interconnect.

Composite Nanowires

267

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. 33.

S. Iijima, Nature 354 (1991) 56. R. Tenne, L. Margulis, M. Genut and G. Hodes, Nature 360 (1992) 444. Y. Feldman, E. Wasserman, D. J. Srolovitz and R. Tenne, Science 267 (1995) 222. N. G. Chopra, R. J. Luyken, K. Cherrey, V. H. Crespi, M. L. Cohen, S. G. Louie and A. Zettl, Science 269 (1995) 966. A. Loiseau, F. Willaime, N. Demoncy, G. Hug and H. Pascard, Phys. Rev. Lett. 76 (1996) 4737. M. Terrones, W. K. Hsu, H. Terrones, J. P. Zhang, S. Ramos, J. P. Hare, R. Castillo, K. Prassides, A. K. Cheetham, H. W. Kroto and D. R. M. Walton, Chem. Phys. Lett. 259 (1996) 568. D. Golberg, Y. Bando, M. Eremets, K. Takemura, K. Kurashima and H. Yusa, Appl. Phys. Lett. 69 (1996) 2045. O. Stephan, P. M. Ajayan, C. Colliex, P. Redlich, J. M. Lambert, P. Bemier and P. Lefin, Science 266 (1994) 1683. Z. Weng-Sieh, K. Cherrey, N. G. Chopra, X. Blase, Y. Miyamato, A. Rubio, M. L. Cohen, S. G. Louie, A. Zettl and R. Gronsky, Phys. Rev. B 51 (1995) 11229. P. Redlich, J. Loeffler, P. M. Ajayan, J. Bill, F. Aldinger and M. Ruhle, Chem. Phys. Lett. 260 (1996) 465. M. Terrones, A. M. Benito, C. Mantecadiego, W. K. Hsu, O. I. Osman, J. P. Hare, D. G. Reid, H. Terrones, A. K. Cheetham, K. Prassides, H. W. Kroto and D. R. M. Walton, Chem. Phys. Lett. 257 (1996) 576. Y. Zhang, H. Gu, K. Suenaga and S. Iijima, Chem. Phys. Lett. 279 (1997) 264. K. Suenaga, C. Colliex, N. Demoncy, A. Loiseau, H. Pascard and F. Willaime, Science 278 (1997) 633. A. M. Morales and C. M. Lieber, Science 279 (1998) 208. S. T. Lee, N. Wang, Y. F. Zhang and Y. H. Tang, MRS Bulletin 24 (1999) 36. D. P. Yu, Z. G. Bai, Y. Ding, Q. L. Hang, H. Z. Zhang, J. J. Wang, Y. H. Zou, W. Qian, G. C. Xiong, H. T. Zhou and S. Q. Feng, Appl. Phys. Lett. 72 (1998) 3458. X. Duan and C. M. Lieber, Adv. Mater. 12 (2000) 298. H. Dai, E. W. Wong, Y. Z. Lu, F. Shoushan and C. M. Lieber, Nature 375 (1995) 769. W. Q. Han, S. S. Fan, Q. Q. Li and Y. D. Hu, Science 277 (1997) 1287. Y. Zhang, K. Suenaga and S. Iijima, Science 281 (1998) 973. A. Rubio, J. L. Corkill and M. L. Cohen, Phys. Rev. B 49 (1994) 5081. Y. Miyamoto, A. Rubio, M. L. Cohen and S. G. Louie, Phys. Rev. B 50 (1994) 4976. Y. K. Kwon, Y. H. Lee, S. G. Kim, P. Jund, D. Tomanek and R. E. Smalley, Phys. Rev. Lett. 79 (1997) 2065. M. R. Pederson and J. Q. Broughton, Phys. Rev. Lett. 69 (1992) 2689. P. M. Ajayan and S. Iijima, Nature 361 (1993) 333. P. M. Ajayan, T. W. Ebbesen, T. Ichihashi, S. Iijima, K. Tanigaki and H. Hiura, Nature 362 (1993) 522. E. Dujardin, T. W. Ebbesen, H. Hiura and K. Tanigaki, Science 265 (1994) 1850. T. W. Ebbesen, J. Phys. Chem. Solids 57 (1996) 951. P. M. Ajayan, O. Stephan, P. Redlich and C. Colliex, Nature 375 (1995) 564. Y. K. Chen, M. L. H. Green and S. C. Tsang, Chem. Commun. (1996) 2489. S. C. Tsang, Y. K. Chen, P. J. F. Harris and M. L. H. Green, Nature 372 (1994) 159. D. Ugarte, A. Chatelain and W. A. Deheer, Science 274 (1996) 1897. D. Ugarte, T. Stockli, J. M. Bonard, A. Chatelain and W. A. de Heer, Appl. Phys. A 67 (1998) 101.

268

Zhang

34. B. C. Satishkumar, A. Govindaraj, J. Mofokeng, G. N. Subbanna and C. N. R. Rao, J. Phys. B At. Mol. Opt. Phys. 29 (1996) 4925. 35. S. Subramoney, R. S. Ruoff, D. C. Lorents, B. Chan, R. Malhotra, M. J. Dyer and K. Parvin, Carbon 32 (1994) 507. 36. P. M. Ajayan, C. Colliex, J. M. Lambert, P. Bernier, L. Barbedette, M. Tence and O. Stephan, Phys. Rev. Lett. 72 (1994) 1722. 37. C. Guerret-Piecourt, Y. Le Bouar, A. Loiseau and H. Pascard, Nature 372 (1994) 761. 38. M. Liu and J. M. Cowley, Carbon 33 (1995) 749. 39. A. Loiseau and H. Pascard, Chem. Phys. Lett. 256 (1996) 246. 40. K. Hsu Wen, S. Trasobares, H. Terrones, M. Terrones, N. Grober, Q. Zhu Yan, W. Z. Li, R. Escudero, J. P. Hare, H. W. Kroto and D. R. M. Walton, Chem. Mater. 11 (1999) 1747. 41. M. Terrones, W. K. Hsu, A. Schilder, H. Terrones, N. Grobert, J. P. Hare, Y. Q. Zhu, M. Schwoerer, K. Prassides, H. W. Kroto and D. R. M. Walton, Appl. Phys. A Mater. Sci. Proc. A 66 (1998) 307. 42. F. Okuyama and I. Ogasawara, Appl. Phys. Lett. 71 (1997) 623. 43. 44. N. Demoncy, O. Stephan, N. Brun, C. Colliex, A. Loiseau and H. Pascard, Euro. Phys. J. B 4 (1998) 147. 45. W. S. Shi, H. Y. Peng, L. Xu, N. Wang, Y. H. Tang and S. T. Lee, Adv. Mater. (Weinheim, Fed. Repub. Ger.) 12 (2000) 1927. 46. X. C. Wu, W. H. Song, B. Zhao, W. D. Huang, M. H. Pu, Y. P. Sun and J. J. Du, Solid State Commun. 115 (2000) 683. 47. R. Z. Ma, Y. Bando and T. Sato, Chem. Phys. Lett. 350 (2001) 1. 48. G. W. Meng, L. D. Zhang, C. M. Mo, S. Y. Zhang, Y. Qin, S. P. Feng and H. J. Li, Solid State Commun. 106 (1998) 215. 49. Q. Zhu Yan, B. Hu Wei, K. Hsu Wen, M. Terrones, N. Grobert, T. Karali, H. Terrones, J. P. Hare, P. D. Townsend, H. W. Kroto and D. R. M. Walton, Adv. Mater. 11 (1999) 844. 50. K. Suenaga, Y. Zhang and S. Iijima, Appl. Phys. Lett. 76 (2000) 1564. 51. Z. L. Wang, Z. R. Dai, R. P. Gao, Z. G. Bai and J. L. Gole, Appl. Phys. Lett. 77 (2000) 3349. 52. Y. Xie, Z. Qiao, M. Chen, X. Liu and Y. Qian, Adv. Mater. 11 (1999) 1512. 53. M. Gao, S. Huang, L. Dai, G. G. Wallace, R. Gao and Z. Wang, Angew. Chem. 39 (2000) 3664. 54. K. S. Mayya, D. I. Gittins, A. M. Dibaj and F. Caruso, Nano Lett. 1 (2001) 727. 55. N. Wang, Y. H. Tang, Y. F. Zhang, C. S. Lee and S. T. Lee, Phys. Rev. B 58 (1998) R16024. 56. Z. Y. Zeng, Y. Xiang and L. D. Zhang, J. Appl. Phys. 88 (2000) 2617.

Chapter 15 Polymer Nanowires and Nanofibers Liming Dai1 and Darrell H. Reneker2 1

Department of Polymer Engineering; 2Maurice Morton Institute and Department of Polymer Science, College of Polymer Science and Polymer Engineering, The University of Akron, Akron, Ohio 44325-0301, USA

1. Introduction As is well known, all matter is made up of atoms. The term molecule is used to describe groups of atoms that tend to exist together in a stable form. As the term implies, macromolecule (or polymer) refers to a molecule of the extraordinarily large size (typically, ⬎1,000 in molecular weight). Although polymer molecules are large, they can be defined by the basic repeat unit often termed as monomer unit. Properties of a particular macromolecule depend on the constituent monomer units and the way they are arranged in the macromolecule. Many of the overall properties of a given polymeric material depend, apart from its macromolecular characteristics, also on the spatial arrangement of the constituent macromolecules and the nature of the intermolecular interactions that hold them together [1]. This provides a broad basis for the development of various polymeric materials (e.g. polymer films, polymer fibers) with different properties from the same macromolecules. Textile fibers consume a significant fraction of the entire synthetic polymer produced. Consequently, a large number of methods ranging from the extrusion to melt blowing have been devised to commercially produce non-woven fabrics with diameters in the range of 1–50 ␮m [2]. Recently, the rapid development in nanoscience and nanotechnology has stimulated an increasing effort in studying one-dimensional nanowires and nanofibers. Since these one-dimensional nanostructures are among the smallest entities through which electrons could be transported, they are important “building blocks” for the construction of nanodevices [3]. Owing to their good processability, unique mechanical and optoelectronic properties, polymer nanowires and nanofibres have attracted growing interest in recent years. Conducting polymer nanowires and nanofibers are of particular interest for various electronic device applications, as they have been shown to possess the processing advantages of plastics and the optoelectronic properties of inorganic semiconductors or metals. This chapter provides an overview on the syntheses, properties, and

270

Dai and Reneker

potential applications of polymer nanowires and nanofibers, with an emphasis on those based on conducting polymers.

2. Nanofibrillar Conducting Polymers Polymers have long been used as insulating materials. After all, metal wires are coated in plastic to insulate them. Various conjugated polymers with alternating single and double bonds, which have been synthesized during the past two decades or so show useful levels of electrical conductivity [4]. Just as metals have high conductivity due to the free movement of electrons through their structure, in order for polymers to be electronically conducting, they must possess not only charge carriers but also an orbital system that allows the charge carriers to move. The conjugated structure can meet the second requirement through a continuous overlapping of -orbitals along the polymer backbone. Most organic polymers, however, do not have intrinsic charge carriers. The required charge carriers may be provided by partial oxidation (p-doping) of the polymer chain with electron-acceptors (e.g. I2, AsF5) or by partial reduction (n-doping) with electron-donors (e.g. Na, K). Through such a doping process, mobile charge carriers (e.g. polaron, bipolaron, and soliton) are introduced [5]. Some conjugated polymers, such as polyaniline, can acquire high conductivities through the protonation of imine nitrogen atoms. In addition to their electronic characteristics, certain conjugated polymers possess interesting optical and magnetic properties due also to the delocalization of electrons along the polymer backbone [4, 5]. Polyacetylene [-(CH⫽CH)n-], having the simplest conjugated structure, has served as the prototype for other conducting polymers. Although Natta et al. [6] were the first to polymerize acetylene into a linear conjugated macromolecule with the Ti(OBu)4/AlEt3 catalyst as early as in the late of 1950s, it was the synthesis of high-quality, free-standing polyacetylene films using the same catalyst, reported by Shirakawa and coworkers in 1974 [7], that facilitated the early research on the electronic properties of polyacetylene. Initially, the Ziegler-Natta polymerization of acetylene only produced polyacetylene films with a randomly-oriented nano-fibrillar morphology (Fig. 1a) [7]. In 1987, Naarmann and Theophilou [8] reported a method to prepare highly oriented polyacetylenes by aging the catalyst mixture of Ti(OBu)4/AlEt3 in silicone oil, which was used as a reaction medium. The resulting very pure polyacetylene film showed a metal-like appearance, which, after doping, had a conductivity up to 1.7 ⫻ 105 S/cm. This value of conductivity is significant even in comparison with the value of 106 S/cm for copper or silver, the best metallic conductors. By replacing the silicone oil with a nematic liquid crystal phase, Shirakawa et al. [9] have also produced partially aligned, highly conducting polyacetylene films. Furthermore, these authors have also obtained polyacetylene films with a highly oriented nano-fibrillar morphology (Fig. 1b) by carrying out the polymerization of acetylene on a vertical glass wall of a flask over which the catalyst-containing nematic liquid crystal solution flowed down under the influence of the gravity [9]. The values of the infrared (IR) anisotropy ranging from 2.0 to 3.8, together with the high DC conductivities obtained by doping with I2 in directions both parallel ( ll ⫽ 4600 S/cm) and perpendicular ( ⬜ ⫽ 3900 S/cm) to the orientation axis, indicate significant orientation of the polyacetylene chains along the flow direction. More recently, Shirakawa

Polymer Nanowires and Nanofibers

271

(a)

(b)

(c)

Fig. 1. (a) A typical TEM micrograph of thin polyacetylene films prepared by the Ziegler-Natta catalyst (Reproduced from [7] with permission from Wiley-VCH). (b) A typical scanning electron micrograph of a polyacetylene film prepared by polymerization on a vertical glass wall (Reproduced from [9] with permission from Marcel Dekker). (c) A typical SEM micrograph of helical polyacetylene nanofibers. (Reproduced from [10] with permission from American Association for the Advancement of Science.)

and co-workers [10] synthesized helical polyacetylene nanofibers (Fig. 1c) using chiral nematic liquid crystals as solvents for the Ziegler-Natta catalyst. I2-doping of the helical polyacetylene shows conductivities of ca.1500–1800 S/cm. This, together with the peculiar helical nanofiber structure, has implications for these materials in novel electromagnetic and optical applications.

3. Template Syntheses of Polymer Nanowires The above observed polyacetylene nanofibers consist of an intrinsic part of the polymer morphology, and cannot be separated from other features of the polymer material. In addition to the fibrillar polyacetylenes discussed above, isolated conducting

272

Dai and Reneker

polymer nanowires with promising nanometer-scale architectures have been fabricated using porous membranes or supramolecular nanostructures as templates. For instance, Smith et al. [11] have recently developed a process for producing nanocomposites with well-defined poly(p-phenylene vinylene), PPV, nanowires within self-organizing liquid-crystal matrixes. This process involves self-assembling a polymerizable liquidcrystal monomer (e.g. acrylate) into an ordered hexagonal array of hydrophilic channels (ca. 4 nm in diameter) filled with a precursor polymer of PPV, followed by photopolymerization to lock-in the matrix architecture and thermal conversion to form PPV nanofibers in the channels. As a result, a significant fluorescence enhancement was observed, most probably, due to the much reduced interchain-exciton quenching achieved by separating the PPV molecules in the polymer matrix. More generally, polymer nanowires and nanotubes with improved order and fewer structural defects can be synthesized within a template formed by the pores of a nanoporous membrane [12] or the nanochannels of a mesoporous zeolite [13]. Template synthesis often allows the production of polymeric wires or tubules with controllable diameters and lengths (Fig. 2) [14]. Template syntheses of conjugated polymers, including polyacetylene, polypyrrole, polythiophene, polyaniline, and PPV, may be achieved by electrochemical or chemical oxidative polymerization of the corresponding monomers. While the electrochemical template synthesis can be carried out within the pores of a membrane that is pre-coated with metal on one side as an anode [12], the chemical template synthesis is normally performed by immersing the membrane into a solution containing the desired monomer and oxidizing agent, with each of the pores acting as if a tiny reaction vessel. It is noted with interest that if pores in polycarbonate membranes are used as the template, highly ordered polymeric tubules are produced by preferential polymerization along the pore walls [15] due to the specific solvo-phobic and/or electrostatic interactions between the polymer and the pore wall [12]. These conducting polymer

Fig. 2. Scanning electron micrographs of polymer fibrils prepared by the template synthesis route: (A) Polystyrene. (B) Poly(vinylidene). (C) Poly(phenylene oxide). (D) Poly(methyl methacrylate). (Reproduced from [14] with permission from the American Chemical Society.)

Polymer Nanowires and Nanofibers

273

tubules show a wide range of electrical conductivities, increasing with decreasing pore diameter [16, 17]. This is because the alignment of the polymer chains on the pore wall can enhance conductivity, and the smaller tubules contain proportionately more of the ordered material. More recently, conjugated conducting polymer nanotubes were synthesized using “template-free” polymerization that involves a self-assembled supramolecular template [18]. These studies are of particular interest because hightemperature graphitization of the polymer nanotubes could transform them into carbon nanotubes of superior electronic and mechanical properties [19], as demonstrated by polyacrylonitrile (PAN) nanotubules synthesized in the pores of an alumina membrane or in zeolite nanochannels [20, 21].

4. Syntheses of Polymer Nanowires at a Scanning Microscope Tip Other related developments include the construction of conducting polymer wires by the use of a scanning tunneling microscope (STM)/scanning electrochemical microscope (SECM) for generation and manipulation of polymer structures as small as a few nanometers. The polymerization of pyrrole onto specific regions of graphite substrates at a submicron resolution was achieved by using the STM tip as an electrode [22]. Polypyrrole strips with a linewidth of 50 ␮m and length of 1 mm were produced by a SECM [23], as were micron-sized polypyrrole towers [24]. By spincoating a solution of Nafion and anilinium sulfate onto a Pt electrode in a SECM unit, Wuu et al. [25] polymerized aniline into a micron scale structure. Borgwarth et al. [26] successfully prepared a 20 ␮m wide polythiophene line by using the tip of a SECM as an electrode for region-specific oxidation of bromide into bromine, followed by the diffusion of the bromine into a conductive substrate covered with a thiophene derivative, to produce localized oxidative polymerization of thiophene monomers. Aiming for polymer structures at the nanometer scale, Nyffenegger and Penner [27] produced electrochemically active polyaniline particles with a size ranging from 10 to 60 nm in diameter and 1 to 20 nm in height by using the Pt tip of a scanning tunneling microscope as an electrode for the electropolymerization of aniline. Liu and co-workers [28] used an electrochemical reaction at an atomic force microscope tip for region-specific deposition of conducting polymer nanowires with diameters in the range from 50 to 500 nm on semiconducting and insulating substrates. Bumm et al. [29] demonstrated the use of a STM tip to probe electrical properties of individual conjugated conducting molecules (“molecular wires”) dispersed into a self-assembled monolayer film of nonconducting alkanethiolate molecules. In a closely related but separate study, Tao and co-workers [30] have electrochemically deposited a conducting polyaniline nanowire bridge between a STM tip and a gold electrode by region-selectively coating the STM tip with an insulation layer so that only a few nm2 at the tip end was exposed for localized growth of the polyaniline bridge (Fig. 3). Upon stretching the polymer nanowire by moving the STM tip away from the Au electrode, these authors observed a stepwise decrease in the conductance (Fig. 4), similar to that reported for metallic nanowires [31, 32]. The initial increase in the conductance seen in Fig. 4a is attributed to the stretching-induced

274

Dai and Reneker

Fig. 3. Schematic representation of the experimental setup used for electrochemical polymerization of a conducting polymer nanowire bridge between a scanning tunneling microscope tip and a gold substrate. In comparison to a field effect transistor, the RE, WE1, and WE2 electrodes are analogous to the gate, source and drain electrodes, respectively. (Reproduced from [30] with permission from the American Institute of Physics.)

Fig. 4. (a) The change of conductance of the polyaniline nanowire bridge during the entire stretching process. (b) A zoom in of the stepwise decrease. The substrate and tip potentials were 0.45 and 0.5 V, respectiviely. (Reproduced from [30] with permission from the American Institute of Physics.)

alignment of the polymer chains in the nanowire, since aligned conducting polymers have been demonstrated to show higher conductivities. The observed smaller conductance step height (⬍2e2/h) and wider plateaus than those of metal nanowires may indicate the occurrence of polymer chain sliding in the polyaniline nanowires during

Polymer Nanowires and Nanofibers

275

Fig. 5. (a–c) Current versus voltage (Vbias represents the bias voltage between the tip and substrate) curves for the polyaniline nanowire bridge. (d) Schematic drawing of a model used to explain the observed I–Vbias characteristics. The shaded bell shaped curves represent conductivity of the nanowire at points near the tip and substrate electrodes, respectively. (Reproduced from [30] with permission from the American Institute of Physics.)

stretching. The conductance of the polyaniline nanowires with various diameters were measured as a function of the electrochemical potential by stopping the stretching at different conductance steps. In analogy to the control of the gate voltage in semiconductor devices [33], He et al. [30] measured the current–voltage curves by sweeping the tip-substrate bias voltage (Vbias) while maintaining the electrochemical potential of the substrate electrode (or the tip electrode) at different values. The resulting I–Vbias curves show a nearly linear relationship when the substrate potential is kept between 0.30 and 0.55 V (Fig. 5a and b). As seen in Fig. 5c, however, the polyaniline nanowire exhibits rectifying characteristics at 0.25 V; the current is small at negative bias whereas it increases when the bias is positive. The observed rectifying characteristics become more pronounced when the potential is reduced to 0.2 V. The observed transition from ohmic to rectifying characteristics can be understood by considering the situation described in Fig. 5d, which shows that the polyaniline nanowire has one end attached to the tip and the other connected to the substrate. Therefore, the electrochemical potentials of the two ends are fixed at Esub and Etip. As polyaniline is in the oxidized conducting form at the electrochemical potentials close to or slightly higher than 0.3 V [4], both portions of the polyaniline nanowire near the substrate and tip are conducting at Esub ⫽ 0.40 V with a small Vbias (Etip ⫽ Esub ⫹ Vbias) so that the I–V curve is ohmic. At 0.25 V, however, a negative sweep moves the Etip to potentials lower than the redox potential of polyaniline (ca. 0.3 V) so that the polymer is in the non-conducting reduced form while a positive sweep of bias moves Etip to the highly conductive region, leading to the rectifying behavior. The controllable

276

Dai and Reneker

conductance of the conducting polymer nanowires, together with their mechanical flexibility and rectifying characteristics, should have important implications for the use of polymer nanowires and nanofibers in various micro-nano-electronic devices and many other applications.

5. Electrospinning of Polymer Nanofibers Polymer nanowires and nanofibers are attractive for a large variety of potential applications. The large-scale fabrication of polymer nanowires/nanofibers at a reasonably low cost has been a big challenge. Electrospinning, a technique which was first patented in 1934 [34], has recently been re-examined, notably by Reneker’s group [35–39], as an effective method for large scale fabrication of ultrafine polymer fibers. As schematically shown in Fig. 6, the electrospinning process involves the application of a high electric field between a droplet of polymer fluid and a metallic collection screen at a distance h from the polymer droplet. As electrically charged jet flows from the polymer droplet toward the collection screen when the voltage reaches a critical value (typically, ca. 20 KV for h ⫽ 0.2 m) at which the electrical forces overcome the surface tension of the polymer droplet. The electrical forces from the charge carried by the jet further cause a series of electrically driven bending instabilities to occur as the fluid jet moves toward the collection screen. The repulsion of charge on adjacent

Fig. 6. Schematic drawing of the electrospinning process. (Reproduced from [38] with permission from the American Institute of Physics.)

Polymer Nanowires and Nanofibers

277

segments of the fluid jet causes the jet to elongate continuously to form ultrafine polymer fibers. The solvent evaporates and the stretched polymer fiber that remains is then collected on screen as a non-woven sheet composed of one fiber many kilometers in length. The electrospun polymer fibers thus produced have diameters ranging from several microns down to 50 nm or less. This range of diameters overlaps conventional synthetic textiles and extends through diameters two or three orders of magnitude smaller. Reneker and co-workers [40–42] have made significant advances on both theoretical and experimental fronts. They have carried out detailed theoretical studies on the formation of an electrically charged polymer jet and its bending instability, and also prepared a large number of novel polymer fibers with diameters from few nanometers to microns by electrospinning various synthetic and natural polymers from either solution or melt. The electrospinning conditions for some of the polymer systems they investigated are listed in Table 1 [37, 43], with a scanning tunneling microscope image of a few segments of a polyaniline fiber shown in Fig. 7 [44]. The electrospun fiber mats possess a high surface area per unit mass, low bulk density, and high mechanical flexibility. These features make the electrospun fibers very attractive for a wide range of potential applications, including high performance filters, scaffolds in tissue engineering, wound dressings, sensors, and drug delivery systems [36, 37]. The high surface to volume ratio associated with electrospun conducting polymer fibers makes them candidates for electrode materials since the rate of electrochemical reactions is proportional to the surface area of an electrode and diffusion rate of the electrolyte. Reneker and co-workers [45] have reported the pyrolysis of electrospun polyacrylonitrile nanofibers into conducting carbon nanofibers. Srinivasan [44] prepared conducting polymer nanofibers by electrospinning polyaniline from sulfuric acid into a coagulation bath and characterized them by scanning tunneling microscopy, electron microscopy and electron diffraction.

Table 1. Some electrospun polymer fibers reported by Reneker’s group [37] Polymer class

Polymer

Solvent

High performance polymers

Polymides Polyamic acid Polyetherimide Polyaramid Poly-gamma-benzyl-glumate Poly(p-phenylene terephthalamide Nylon 6-Polyimide Polyacrylonitrile Polyethylene terephthalate

Phenol m-cresol Methylene chloride Sulphuric acid Dimethylformamide Sulphuric acid

Liquid crystalline polymers

Copolymers Textile fibre polymers

Electrically conducting polymers Biopolymers

Nylon Polyaniline

Formic acid Dimethylformamide Trifluoroacetic acid and dichloromethane Melt in vacuum Formic acid Sulphuric acid

DNA Polyhydroxybutyrate-valerate

Water Chloroform

278

Dai and Reneker

Fig. 7. Scanning tunneling microscope image of polyaniline fibers. Conducting fibers as small as 40 nm were made. (After Reference [44].)

MacDiamid et al. [46, 47] prepared camphorsulfonic acid doped polyaniline (HCSA-PANI) [48] and polyethylene oxide (PEO) composite thin fibers (ca. 0.95–2 ␮m in diameter) by the electrospinning technique from a mixed solution [46]. The fourpoint probe method [49] was used to measure the conductivity of the non-woven fiber mat and crosschecked with measurements of the conductivity of cast films produced from the same solution. Fig. 8A shows the effect of the weight percentage content of HCSA-PANI on the room temperature conductivity obtained from the HCSA-PANI/PEO electrospun fibers and cast films [46]. As can be seen in Fig. 8A, the conductivity of the electrospun fiber mat was lower than that for a cast film, though they have very similar UVvisible absorption characteristics. The lower conductivity values for the electrospun fibers than those of cast films can be attributed to the porous nature of the non-woven electrospun fiber mat (Fig. 7) as the four-probe method measures the volume resistivity rather than the conductivity of an individual fiber. Although the measured conductivity for the electrospun mat of conducting nanofibers is relatively low, their porous structures, together with the high surface-to-volume ratio, enable faster de-doping and re-doping in both liquids and vapors. In order to measure the conductivity of an individual nanofiber, MacDiarmid and co-workers [47] have also collected a single electrospun HCSA-PANI/PEO nanofiber on a silicon wafer coated with a thin layer of SiO2 and deposited two separated gold electrodes at 60.3 ␮m apart on the nanofiber (two probe method). The current–voltage (I–V) curves thus measured for single 50 wt% HCSA-PANI/PEO fibers with diameters of 600 and 419 nm are shown in Fig. 8B (a, i.e. Fiber 1 and Fiber 2, respectively), which give a more or less straight line with a conductivity of ca.0.1 S/cm. The temperature dependence of the conductivity for a single 72 wt% HCSA-PANI/PEO electrospun fiber

Polymer Nanowires and Nanofibers

279

(A)

(B) (a)

(b)

Fig. 8. (A) Electrical conductivity of the HCSA-PANI/PEO blend electrospun fiber mat and cast films. (Reproduced from [46] with permission from Elsevier.) (B) (a) Current–voltage curve for a single 50 wt% HCSA-PANI/PEO nanofiber. (Reproduced from [47] with permission from Elsevier) (b) Temperature-dependence of the conductivity for a single 72 wt% HCSAPANI/PEO nanofiber. (Reproduced from [47] with permission from Elsevier.)

with a diameter of 1.32 ␮m given in Fig. 8B(b) also shows a linear plot with a room-temperature conductivity of 33 S/cm (295 K). This value of conductivity is much higher than the corresponding value of about 0.1 S/cm for a cast film (see, Fig. 8A), suggesting a high alignment of the PANI chains in the electrospun fiber.

6. Polymer Nanowires and Nanofibers with Special Architectures Just as the microchips have revolutionized computers and electronics, nanotechnology has the potential to revolutionize many industrial sectors. However, considerable

280

Dai and Reneker

efforts are required to design, construct and operate new devices at the nanometer scale. Polymer nanowires and nanofibers possess interesting optoelectronic properties. Electrically conducting nanofibers with diameters at the nanometer scale and lengths in kilometers are both useful “building blocks” for nanodevice construction and ideal connections between the nanoscale entities and the macroscale world. Effective fabrication of ordered structures is a pre-requisite for the economical use of polymer nanowires and nanofibers, either as active materials or connecting components, in device applications. The formation of ordered structures from metal nanowires and carbon nanotubes has been an active research area for some years. Various nano- and micro-fabrication techniques have been developed. Examples include the use of electric and magnetic fields for aligning suspended metallic nanowires [50–52], the combination of fluidic alignment with patterned surface structures (e.g. microchannels) for nanowire patterning [53, 54], and the lithiographic patterning of aligned carbon nanotubes [55]. The effective fabrication of polymer nanowires or nanofibers is a more recent development. A few aligned polymer nanowires were prepared by synthesizing them at a tip, or by using a template. The possibility of forming aligned electrospun nanofibers by using a rotating cathode collector was demonstrated [56, 57]. Nonconducting electrospun polymer fibers were used as “templates” for coating with appropriate conducting polymers and for deposition of a metal layer from solution or vapor [58]. MacDiarmid and co-workers [47] reported the uniform coating of an electrospun polyacrylonitrile nanofiber with a layer of conducting polypyrrole (20 to 25 nm thick) by immersion the electrospun fiber in an aqueous solution of polymerizing polypyrrole. These authors also prepared gold coated polyacrylonitrile nanofibers through treatment of the electrospun fibers with a solution of AuS2O3 and ascorbic acid. Reneker and co-workers [59] coated electrospun poly(meta-phenylene isophthalamide) nanofibers with sub-nanometer thick coatings of various other materials (e.g. carbon, Cu, and Al) by chemical and physical vapor deposition. They also prepared nanotubes consisting of pure aluminum or mixtures of aluminum and aluminum oxide by coating the electrospun poly(meta-phenylene isophthalamide) nanofiber with an aluminum layer, followed by selectively removing the polymer nanofiber core via solvent dissolution or thermal degradation [59]. During thermal degradation of the poly(meta-phenylene isophthalamide) nanofiber cores, the aluminum coating layer was subject to a limited degree of oxidation, which produced nanotubes of mixed aluminum and aluminum oxide. The aluminum coating layers did not oxidize when the template fiber core was removed by dissolution. Fig. 9A(a–c) shows TEM images of aluminum-coated poly(meta-phenylene isophthalamide) fibers with the coating thickness increasing from ca. 10 nm to 100 nm. As seen in Fig. 9B(a–c), aluminum nanotubes remained after dissolution of the fiber cores of poly(meta-phenylene isophthalamide) with N,N-dimethylacetamide solvent. The electron diffraction pattern associated with the aluminum nanotubes shown in Fig. 9B(a) is given in Fig. 9B(d), in which all the d-spacings are characteristic of the aluminum crystal unit cell. Another interesting area closely related to the fabrication of polymer nanowires and nanofibers is the synthesis of coaxial nanowires of polymers and carbon nanotubes. Dai and co-workers [60] have developed a novel approach for chemical modification of aligned carbon nanotubes. Radio-frequency glow-discharge plasma

Polymer Nanowires and Nanofibers

281

treatment activated the surface of the nanotubes for subsequent reactions characteristic of the plasma-induced surface groups. These authors successfully grafted polysaccharide chains onto plasma activated aligned carbon nanotubes through Schiff-base formation, followed by reductive stabilization of the Schiff-base linkage with sodium cyanoborohydride (Fig. 10a). The resulting amino-dextran grafted nanotube film showed zero air/water contact angles. The (acetaldehyde) plasma treated carbon nanotube film gave relatively low advancing (90⬚), sessile (78⬚), and receding (45⬚) air/water contact angles comparing to the advancing (155⬚), sessile (146⬚), and receding (122⬚) angles for an untreated sheet of aligned carbon nanotubes. The glucose units within the surface-grafted amino-dextran chains (Figure 10a) can be converted into dialdehyde moieties by periodate oxidation (cf. Fig. 10b) [61], thereby providing considerable room for further modification (Figure 10b), including the creation of multilayer coaxial structures. (A)

(a)

(b)

(c)

Fig. 9. (A) (a–c) TEM images of aluminum-coated poly(meta-phenylene isophthalamide) electrospun nanofibers with the coating thickness increasing from about 10 nm in (a) to nearly 100 nm in (c). (After Reference [59].)

282

Dai and Reneker (B)

(a)

(b)

(c)

Fig. 9. (B) (a–c) TEM images of Al nanotubes prepared by solvent dissolution of the poly(meta-phenylene isophthalamide) core. (d) Electron diffraction pattern of nanotubes associated with the aluminum nanotubes shown in Fig. 9B(a). (After Reference [59].)

In addition to the chemical grafting of polymer chains onto the carbon nanotube surface, Dai and co-workers [62] recently used the aligned carbon nanotubes (Fig. 11a) as nanoelectrodes for making novel conducting coaxial nanowires by electrochemically depositing a concentric layer of an appropriate conducting polymer uniformly onto each of the aligned nanotubes. The SEM image for these Conducting Polymer-nanotube (CP-NT) coaxial nanowires given in Fig. 11b shows the same features as the aligned nanotube array of Fig. 11a, but the tubes had a larger diameter due to the presence of the newly-electropolymerized polypyrrole coating used in this example. The presence of the conducting polymer layer was also evident in transmission electron microscopic (TEM) images [62]. The electrochemical performance of the aligned CP-NT coaxial nanowires was evaluated by carrying out cyclic voltammetry measurements. As for polyaniline films electrochemically deposited on conventional electrodes, the cyclic voltammetric

283

Polymer Nanowires and Nanofibers (a)

Acetaldehyde Plasma

Aligned Carbon Nanotubes

Plasma Polymer Coating

OH

+ CHO

OH

Carbon Nanotube

NH2/OH Amino-dextran (23NH2/chain) =Amino-dextran backbone +H2O

HO

HO

O

OH CH

–H2O

N

O

O

OH NaBH3CN

CH2

NH

O

O

(b)

Ethylenediamine Plasma

O O OH

O Dextran-FITC (~2FITC/chain) OH/FITC

OH

(I) 2NaIO4 –HCO2H Plasma Polymer Coating NH2 Aligned Carbon Nanotube

NH2

O +

O

H O

+H2O (II)

O

H O

–H2O

O

O N

CH

N

CH

O O

NaBH3CN (III)

NH CH2 O NH CH2 O

Fig. 10. (a) Reaction scheme for the covalent attachment of amino-dextran chains onto acetaldehyde-plasma activated carbon nanotubes. For reasons of clarity, only one of the many plasma-induced aldehyde surface groups is shown for an individual nanotube. (b) Reaction scheme for the periodate oxidation of fluorescein isothiocyanate (FITC)-labeled dextran (designated as dextran-FITC), followed by covalent attachment onto ethylenediamine-plasma activated carbon nanotubes. For reasons of clarity, only two of the many plasma-induced amine surface groups are shown for an individual nanotube. (Reproduced from [60] with permission from the American Chemical Society.)

284

Dai and Reneker (a)

(b)

Fig. 11. Typical SEM images of (a) aligned carbon nanotubes after being transferred onto a gold foil (a small piece of the as-synthesized aligned nanotube film was included, at the leftbottom corner, to show the formation of a flat amorphous carbon layer); (b) the CP-NT coaxial nanowires produced by cyclic voltammetry on the aligned carbon nanotube electrode in an aqueous solution of NaClO4 (0.1 M) containing pyrrole (0.1 M). Scan rate: 25 mV/s. (Reproduced from [62] with permission from Wiley-VCH.)

response of the polyaniline coated nanotube array in an aqueous solution of 1 molar H2SO4 (Curve a of Fig. 12a) also shows oxidation peaks at 0.33 and 0.52 V, but with much higher current densities than the conventional polyaniline electrode [63]. As a control, the cyclic voltammetry measurement was also carried out on the bare aligned nanotubes under the same conditions (Curve b of Fig. 12a). In the control experiment, only capacitive current was observed with no peak attributable to the presence of any species undergoing a redox reaction. The coaxial structure allows the nanotube framework to provide mechanical stability [64, 65], and efficient thermal and electrical contact with the conducting polymer layer [66, 67]. The large interfacial surface area per unit mass obtained for the nanotube-supported conducting polymer layer is potentially useful in many optoelectronic applications, for example in sensors, organic lightemitting diodes, and photovoltaic cells where the charge injection and separation are strongly limited by the interfacial area available in more conventional devices [5, 68].

Polymer Nanowires and Nanofibers

285

(a)

(b) Response current (uA)

16 PPY/GOD/NT PPY/GOD/Au

14 12 10 8 6 4 2 0 0

1 2 3 4 5 6 7 8 9 Glucose concentration (mM) vs Ag/AgCl

Fig. 12. (a) Cyclic voltammograms of the polyaniline-coated CP-NT coaxial nanowires (Curve a) and the bare aligned carbon nanotubes (Curve b). Measured in an aqueous solution of 1 M H2SO4 with a scan rate of 50 mV/s. (Reproduced from [62] with permission from Wiley-VCH.) (b) The dependence of response current on glucose concentration for the PPY/GOx/NT electrode (solid squares) and PPY/GOx/Au electrode (open circles) at the electrode potential of 0.45 V versus Ag/AgCl (pH 7.4 buffer). The geometric area of the PPY/GOx/Au electrode is 1.5 times larger than that of the PPY/GOx/NT electrode. (After Reference [69].)

To demonstrate the potential device applications for the CP-NT coaxial nanowires, Dai and co-workers [69] have also immobilized glucose oxidase (GOX) onto the aligned carbon nanotube substrate by electro-oxidation of pyrrole (0.1 M) in the presence of glucose oxidase (2 mg/ml) and NaClO4 (0.1 M) in a pH 7.45 buffer solution. The glucose oxidase-containing polypyrrole-carbon nanotube coaxial nanowires were used to monitor the concentration change of hydrogen peroxide (H2O2) during the glucose oxidation reaction by measuring the increase in the electro-oxidation current at the oxidative potential of H2O2 (i.e. the amperometric method) [69]. As shown in Fig. 12b, a linear response of the electro-oxidation current to the glucose concentration was obtained for the CP-NT nanowire sensor. The linear relationship was found to extend to glucose concentrations as high as 20 mM [69], which is higher than 15 mM typically needed for the measurement of blood glucose in clinical practice [70]. The amperiometric response was also found to be much higher than that of more conventional flat electrodes coated with glucose oxidase-containing polypyrrole

286

Dai and Reneker

films under the same conditions. The CP-NT nanowire sensors were also demonstrated to be highly selective for glucose with their amperiometric responses being almost unchanged even in the presence of some interference species including ascorbic acid, urea, and D-fructose. Along with many other potential applications, therefore, the CPNT nanowires could be used for making new glucose sensors with a high sensitivity, selectivity, and reliability. Continued research efforts in this embryonic field could give birth to a wide range of new nanotechnologies.

7. Summary An overview of the recent progress in research and development of polymer nanowires and nanofibers was presented. Polymer nanowires and nanofibers possess interesting electronic and mechanical properties of use for many potential applications. In order to realize their commercial applications, several major problems must be solved, including the large-scale fabrication of polymer nanowires and nanofibers, effective incorporation of them into nanodevices, and proper connections of nanoscale structures and devices to the macroscale world. With the promising approaches already reviewed in this chapter and more to be developed, practical applications of polymer nanowires and nanofibers are imminent.

Acknowledgment We thank our colleagues and many others who made contributions to the work that is reviewed in this chapter.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

P. J. Flory, Principle of Polymer Chemistry, Cornell University Press, New York (1953). A. Ziabicki, Fundamentals of Fibre Formation, Wiley Interscience, New York (1976). J. Hu, T. Wang Odom and C. M. Lieber, Acc. Chem. Res. 32 (1999) 435. T. A. Skotheim, J. Reynolds and R. Elsenbaumer (eds.), Handbook of Conducting Polymers (2nd ed.) Marcel Dekker, New York (1998). L. Dai, J. Macromol. Sci. Rev. Macromol. Chem. Phys. 39 (1999) 237, and references cited therein. G. Natta, G. Mazzanti and P. Corradini, Alti Accad. Naz. Lincei Rend. Sci. Fis. Mat. Nat. 25 (1958) 2. T. Ito, H. Shirakawa and S. Ikeda, J. Polym. Sci. Polym. Chem. Ed. 12 (1974) 11. H. Naarmann and N. Theophilou Synth. Met. 22 (1987) 1. H. Shirakawa, K. Akagi, S. Katayama, K. Araya, A. Mukoh and T. J. Narahara, Macromol. Sci. Chem. A 25 (1988) 643. K. Akagi, G. Piao, S. Kaneko, K. Sakamaki, H. Shirakawa and M. Kyotani, Science 282 (1998) 1683. R. C. Smith, W. M. Fischer and D. L. Gin, J. Am. Chem. Soc. 119 (1997) 4092.

Polymer Nanowires and Nanofibers

287

12. L. S. Van Dyke and C. R. Martin, Langmuir 6 (1990) 1118; C. R. Martin, Acc. Chem. Res. 28 (1995) 61; M. Steinhart, J. H. Wendoff, A. Griner, R. B. Wehrspohn, K. Nielsch, J. Schilling, J. Choi and U. Gösele, Science 296 (2002) 1997. 13. C. G. Wu and T. Bein, Science 266 (1994) 1013. 14. V. M. Cepak and C. R. Martin, Chem. Mater. 11 (1999) 1363. 15. C. R. Martin, Adv. Mater. 3 (1991) 457. 16. Z. Cai, J. Lei, W. Liang, V. Menon and C. R. Martin, Chem. Mater. 3 (1991) 960. 17. R. V. Parthasarathy and C. R. Martin, Chem. Mater. 6 (1994) 1627. 18. M. X. Wan and J. C. Li, J. Polym. Sci., Part A: Polym. Chem. 37 (1999) 4605; H. Qiu, M. Wan, B. Matthews and L. Dai, Macromolecules 34 (2001) 675. 19. R. Saito, G. Dresselhaus, M. S. Dresselhaus, Physical Properties of Carbon Nanotubes, Imperial College Press, London (1998). 20. R. V. Parthasarathy, K. L. N. Phani and C. R. Martin, Adv. Mater. 7 (1995) 896. 21. T. Kyotani, T. Nagai, S. Inoue, A. Tomita, Chem. Mater. 9 (1997) 609. 22. R. Yang, D. F. Evans and W. A. Hendrickson, Langmuir 11 (1995) 211. 23. C. Kranz, H. E. Gaub and W. Schuhmann, Adv. Mater. 8 (1996) 634. 24. C. Kranz, M. Ludwig, H. E. Gaub and W. Schuhmann, Adv. Mater. 7 (1995) 38; 568. 25. Y.-M. Wuu, F. R. F. Fan and A. J. Bard, J. Electrochem. Soc. 136 (1989) 885. 26. K. Borgwarth, C. Ricken, D. G. Ebling and J. Heinze, Ber. dt. Bunsen. Ges. Phys. Chem. 99 (1995) 1421. 27. R. M. Nyffenegger and R. M. Penner, J. Phys. Chem. 100 (1996) 17041. 28. B. W. Maynor, S. F. Filocamo, M. W. Grinstaff and J. Liu, J. Am. Chem. Soc. 124 (2002) 522. 29. L. A. Bumm, J. J. Arnold, M. T. Cygan, T. D. Dunbar, T. P. Burgin, L. Jones II, D. L. Allara, J. M. Tour and P. S. Weiss, Science 271 (1996) 1705. 30. H. X. He, C. Z. Li, N. J. Tao, Appl. Phys. Lett. 78 (2001) 811. 31. U. Landman, W. D. Luedtke, N. A. Burman and R. J. Colton, Science 248 (1990) 454. 32. J. I. Pascual, J. Mendez, J. Gomez-Herrero, A. M. Baro, N. Garcia and V. T. Binh, Phys. Rev. Lett. 71 (1993) 1852. 33. N. C. Greenham and R. H. Friend, Solid State Phys. 49 (1995) I. 34. A. Formhals, US patent No. 1,975,504 (1934). 35. J. Doshi, G. Srinivasan and D. H. Reneker, Polym. News 20 (1995) 206. 36. J. Doshi and D. H. Reneker, J. Electrost. 35 (1995) 151. 37. D. H. Reneker and I. Chun, Nanotechnology 7 (1996) 216. 38. D. H. Reneker, A. L. Yarin, H. Fong and S. Koombhongse, J. Appl. Phys. 87 (2000) 4531. 39. D. H. Reneker, W. Kataphinan, A. Theron, E. Zussman, A. L. Yarin, Polymer 43 (2002) 6785. 40. A. F. Spivak, Y. A. Dzenis and D. H. Reneker, Mech. Res. Commun. 27 (2000) 37. 41. A. L. Yarin, S. Koombhongse and D. H. Reneker, J. Appl. Phys. 89 (2001) 3018. 42. A. L. Yarin, S. Koombhongse and D. H. Reneker, J. Appl. Phys. 90 (2001) 4836. 43. H. Fong, I. Chun and D. H. Reneker, Polymer 40 (1999) 4585. 44. G. Srinivasan, Dissertation presented to Graduate Faculty of The University of Akron, December (1994). 45. I. Chun, D. H. Reneker, H. Fong, X. Fang, J. Deitzel, N. B. Tan and K. Kearns, J. Adv. Mater. 31 (1996) 36. 46. I. D. Norris, M. M. Shaker, F. K. Ko and A. G. MacDiarmid, Synth. Met. 114 (2000) 109. 47. A. G. MacDiarmid, W. E. Jones, Jr., I. D. Norris, J. Gao, A. T. Johnson, Jr., N. J. Pinto, J. Hone, B. Han, F. K. Ko, H. Okuzaki and M. Llaguno, Synth. Met. 119 (2001) 27. 48. L. Dai, J. Lu, B. Matthews and A. W. H. Mau, J. Phys. Chem. B 102 (1998) 4049. 49. L. J. van der Pauw, Philips Res. Rep. 13 (1958) 1. 50. P. A. Smith, D. N. Christopher, N. J. Thomas, T. S. Mayer, B. R. Martin, J. Mbindyo and T. E. Mallouk, Appl. Phys. Lett. 77 (2000) 1399.

288

Dai and Reneker

51. X. Duan, Y. Huang, J. Wang and C. M. Lieber, Nature 409 (2001) 66. 52. M. Tanase, L. A. Bauer, A. Hultgren, D. M. Silevitch, L. Sun, D. H. Reich, P. C. Searson and G. J. Meyer, Nano Lett. 1 (2001) 155. 53. Y. Huang, X. Duan, Q, Wei and C. M. Lieber, Science 291 (2001) 630. 54. B. Messer, J. H. Song and P. Yang, J. Am. Chem. Soc. 122 (2000) 10232. 55. L. Dai and A. W. H. Mau, J. Phys. Chem. B 104 (2000) 1891. 56. A. Theron, E. Zussman and A. L. Yarin, Nanotechnology 12 (2001) 384. 57. J. A. Matthews, G. E. Wnek, D. G. Simpson and G. L. Bowlin, Biomacromolecules 3 (2002) 232. 58. M. Bognitzki, H. Hou, M. Ishaque, T. Frese, M. Hellwig, C. Schwarte, A. Schaper, J. H. Wendorff and A. Greiner, Adv. Mater. 12 (2000) 637; M. Bognitzki, W. Czado, T. Frese, A. Schaper, M. Hellwig, M. Steinhart, A. Greiner and J. H. Wendorff, Adv. Mater. 13 (2001) 70; H. Hou, J. Zeng, J. H. Wendorff and A. Schaper, Macromolecules 35 (2002) 2429. 59. W. Liu, M. Graham, B. V. Satola, E. A. Evans and D. H. Reneker J. Mater. Res. 17 (2002) 3206. 60. Q. Chen, L. Dai, M. Gao, S. Huang and A. W. H. Mau, J. Phys. Chem. B 105 (2001) 618. 61. L. Dai, H. A. W. St John, J. Bi, P. Zientek, R. C. Chatelier and H. J. Griesser, Surf. Interf. Anal. 29 (2000) 46. 62. M. Gao, S. Huang, L. Dai, G. Wallace, R. Gao and Z. Wang, Angew. Chem. Int. Ed. 39 (2000) 3664. 63. D. Sazou and C. Georgolios, J. Electroanal. Chem. 429 (1997) 81. 64. P. Poncharal, Z. L. Wang, D. Ugarte and W. A. de Heer, Science 283 (1999) 1513. 65. R. Gao, Z. L. Wang, Z. Bai, W. A. de Heer, L. Dai and M. Gao, Phys. Rev. Lett. 85 (2000) 622. 66. S. Frank, P. Poncharal, Z. L. Wang and W. A. de Heer, Science 280 (1998) 1744. 67. T. W. Odom, J.-L. Huang, P. Kim and C. M. Lieber, J. Phys. Chem. B 104 (2000) 2794, and references cited therein. 68. L. Dai and A. W. H. Mau, Adv. Mater. 13 (2001) 899. 69. M. Gao, L. Dai and G. G. Wallace, J. Electroanal. Chem. (in press). 70. M. Yasuzawa and A. Kunugi, Electrochem. Commun. 10 (1999) 459.

Index

2,2' -azoh isisob utyro nitri lc, 2 17 3-heam lithog raphy. 133 AAO , see anodic alu minum ox ide AF M. see atomic force microscopy agglo me ratio n. 96 A IBN. see 2,2 ' -az ohisisobutyro nitri le alumina boa t, 23, 197 cr ucib le, 140. 143, 149, 153 tube, 23 , 4R, 63, 140, 153 alumi niu m nanotubes, 280, 282 alu minium-coa ted nan ofib er s. 280 - 1 alu minum oxi de. see an odic al umi num ox ide a mino -dex tra n, 2RI , 283 amphi ph ilic trib lock co poly mer syste ms. 217 anatase (Ti02) . 107-8, 114. 158 particle s. 122. 125 phase , 103-4, 119, 129, 15R- 9. 166, 179 structure, I I I , 116 in trit ita nate prepara tion, 158 Ander son and Wadsley model. 163 anisotropic grow th. 85, 148, 174 , 187,2 11 annealing, 6-7 , 118 temperature . 119 in Sn0 2, 127-30 anodic a lum inum oxi de. 175, 212- 13. 280 AOT, see so di um bis(2-cthyl hexyl)sultt )succin atc AOT - water- to luene, 176 arag onite, 240 nanoplutes, 25 1-2 urc-d ischurge sy nthesis. 25R-60 as pect ratio. 14, 65, I R7, 22R, 24 R, 250 op tica l pro perties. 229 ato mic coordinates, 162-3 mo de ls, 160 ratio. 177, 199 tran sport . 226 ato mic force microscop y, 14-1 5, 7R, RR images. 4, 15. 76--7. 79. 8R inde nte r pro be, 73-5 mode s of, 15 pus hing experiment. 79 atomic force probe. 14-15

Au thin film . 23. 27- 8 A u-Zn alloy. 23 Ba NaAOT reverse mice lles. 175 bar ium (stro nti um) titanium isop ro poxi de complex, R5 barium titan ate, 83-9 1 scanned prob e me as ure ments, 87- 9 1 synthesis or. R5--7 BaTiO] nano wires, 83-9 1 so l, 100 BB O, see beta-bari um borat e B-e-N co mposite nanot ubes, 259- 60 bending. 75, 203- 5 and cutting, 75 instabil ity. 276 -7 modulu s. 11- 13 phen o menon, 203- 5 stiffn ess, n stre ngth . 205 beta-bariu m borate. 42 Bi 2S] nanorod s. 2 12 bias voltage , 275 195 bia xial nan ow ircs, bicrysta llinc Z nO nunowires, 14 2-·6, 154 bim et all ic oxi de precursor. R4-5 hinary/ternary oxides. 176 bis(2- cthyl hexy l)sulfo succinatc . 175. 217 blue-shifts. 169. 229 hlue 1ight emi ssion. 20 1 Bohr radius. 3 f brookite, ll4 Brownian moti on, 96-7 buckling, 75 hcnding, 75, 203-5 Burgers vector, 123 by- layer structures. 250- 1

n.

C/S i0 2/SiC nanow ire co mpos ite. 199- 200 CA, see c itric acid calcite. 240 camphorsu lfoni c acid dop ed po lyaniline , 278

2R9

290 cantilevers, 15 resonance frequency, X8-90 carbon nanotubes, 78,173,194,231,263,281-5 aligned,2XO-5 chemical grafting, 281 electromechanical behavior of, 7X-9 field emission, 231 filling in, 261 -3 growth, 14,257-61 mechanical resonance, 77 multiwalled, 75, 260 oriented, 239-40 plasma-activated, 281, 283 polypyrrole, 285 prodding, 78 role of sulfur, 263 single-walled, 80, 160 as templates, 194 tensile strength, 78 Young's modulus, 75-6, 78 carbon-silica-silicon carbide nanowires. 195 carbothermal reduction method, 199 carrier gas, 67 carrot-shaped rods, 63 CASTEP program, 163, 169 catalysis-assisted VLS method, 49, 152 catalysts, 49, 201-3 Au as, 49, 197,218 choice and patterning, 174 Fc as, 191, 197 forms of, 49 functions in growth, 49 Ga as, 62, 64, 174 high-active, 202 islands, 79 metal. 49, 60-2, 64, 70,154,174,191,193, 197,21X metal oxide, 113, 135, 148, 174,201-3,261 silica, 201-3 Sn as, 60-2, 64, 174 Ti0 2 as, 135 Ti(OHu))AIEt, as, 270 Ziegler-Natta, 271 catalyst-free thermal evaporation, 56 catalytic activity of nanotuhcs, 201-·3 filamentous carbons, 194 catalyzed-grown nanowires, 49 CBED, see convergent beam electron diffraction CdO nanobelts, 14,52,54 CdS nanorods, 2 11- 12, 2 14, 2 16- 17, 233 nanowires, 213-14, 216, 218, 228-9, 230, 233 CdSe nanocrystals, 140, 215, 233 CdS-PAN, 216

Index CdS-PS,217 CERILJS 2, 159 cetyltrirnethyl ammonium bromide, 183, 185 CFC, see catalytic lilamentous carbons chemical grafting, 281 template synthesis, 272 vapor deposition, 174, 192 vapor transport and condensation, 22-3 chiral crystal growth, 250-1 nematic liquid crystals, 270-1 citrate concentrations, 248-50 citric acid, 195 C -NTs/CNTs, see carbon nanotubes coaxial nanocables, 257, 264-6 electron di Ifraction pattem, 264 schematic diagram, 265 coaxial nanowires. 195-6, 280-2 coercive electric field, 89 complex nanobelt structures, 59-·65 aligned growth, 61-5 complex oxides, nanowires/tubes, 173-XX composite nanowires, 199-200, 259-60, 265 condensation, 22-3, 95,104,114,184 condensed step, 159-60 conducting polymer nanotube, 177-9, 282, 284 nanowires, 242, 245, 274 conductometric metal oxide semiconductor films, 9 confined exciton model, 169 continuous nanowircs. 263 convergent beam electron diffraction, X5-6 copper transport, 226, 230 Cowley and Moodie multislicc method, 163 CPNT/CP-NT, see conducting polymer nanotube crystal bulk, 84, 90, 205 CSzTi20S, 177 CU20,226 growth, 23, 25 hexagonal, 35, 215, 24X KzTi60I\, 16X, 177 perovskite, 83, 85, 105 phase, 105 platelets, 167, I X5, 249 seed, 249 structures, 9, 50, 56, 83, X5, 105, 114, 177, 185 tetrapod, 150 TiO z,168 vapor nucleation mechanism, 149 ZnO, 244, 251 crystallinity, 84, I()I, 228 crystallization, 96, 100-1, 175

Index crystallographic axis orientation. 213 forms of MnOb 180 image, 146 pattern in K 2Tib0 13, 167 planes, 61 polarity, 1)0 C-silica-{3-SiC nanowires, 196 CSRs, see carrot-shapcd rods CrAB, see eetyltrimethyl ammonium bromide Cu K a radiatiou, 160 Cu L 3MM Auger spectrum. 228 CU20 nanowircs, 130-3 CU2S nanowires, 219-35 gold coating, 233-4 cubic zinc blend ZnS, )6, 58 current-voltage curve, 134,230,27),279 relationship, 110, 134 CVD, see chemical vapor deposition CVTC, see ehcmieal vapor transport and condensation cyelic voltammograms, 28) Dai, Z.R., 78-80 deep-level emission, 154 degree of supersaturation, 241 differential scanning calorimetry, 230 transport properties, 230 diffraction peak, 221 diffusion layer, 98 Digital Instrument 3000, 14 dimethyl sulfoxide, 213 dislocation, 123 displaeive fcrroelcctrics, 83 DMSO. see dimethyl sulfoxide double layer stabilization. 99 structure, 98 DSC, see differential scanning calorimetry Ii-spacings. 280 e-beam, 14. 27 edge-sharing TiO" octahedra, 159-60, 170 EDS, see energy dispersive spectrometer EDX, see energy-dispersive X-ray EELS, see electron energy loss spectrometer EFM, see electrostatic force microscopy eight pyramidal inversion-twin crystals, 150 EHP, see electron-hole plasma electric field induced mechanical resonance. 76-7 electrochemical deposition, 242 template synthesis, 272 electron energy loss spectrometer, 164-), 199-200

291

electron-hole pairs. 8, 169 plasma, 22 electro-oxidation current. 285 electrophoresis, 98, 110 electrophoretic deposition, 97 clcctrospinning, 276-9 polymer fibers, 277 electrostatic force microscopy, 88 stabilization, 96-7 clectrosteric stabilization. 96-7 en molecule. 21 1--12 encapsulation, 262-3 energy dispersive spectrometry, )2, 178, 199 energydispersive X-ray spectroscopy, )2. 141, 257.261 epitaxial growth, 25-7, 61, 225 eutectic alloy, 49 liquid, 94 temperature, 197 exci mer laser, 191 exciton, 22, 30-1, 169 Bohr radi us, 31 extended and oriented nanowircs, 239-53 FEG, see field-emission gun Feist and Davis model, 162 Fermi level, 80, 169 ferroelectric materials, 83 nanowires, 83-91 FE-SEM, see field-emission scanning electron microscopy FET. see field-effect transistors field emission. 231-2 enhancement. 232 field-effect transistors, 3, 5, 16, 134-5 n-type enhancement mode, 134-) physical chaructcristics, )-6

field-emission gun, 1)9 scanning electron microscopy, 140-2 films/monoliths, 96-9, 100 FITC, see fluorescein isothiocyanate t1exural rigidity, 78 flow-through technique, 10 t1uorescein isothiocyanate, 283 F-N plot, 232 forrnvar, 204 lour-point probe method, 278 Fowler-Nordheim plot, 232 full width at half maximum, 30

292 functional oxide nanowires/bclts, 70, 113-35, 47-208 fundamental resonance frequencies, 12, 15, 31, 75, 78 FWHM, see full width at half maximum Ga203 monoclinic structure, 50, 54, 65 nanowires/bclts, 48, 50. 54-5, 174 nanosheets, 65 gallium droplet induced growth, 62-5 gas phase growth mechanism, 148 gas sensors, 9, 16, 134 oxygen, 8 Ga-Si alloy, 197-8 gas-solid, 218-27 gatan imaging filter system (GIF), 159 glucose oxidase, 285 gold coating in CU2S nanowires, 233-4 electrodes, 4-5, 273 nanotubcs, 233, 235 thin films, see Au thin film GO x, see glucose oxidase grafting nanotubes, 281 graphite, 23, 48, 259, 263, 273 graphitization, high-temperature, 273 green density, 99 emission, 153-4 green-blue diode laser structures, 22 growth mechanism of nanostructurcs, 148-53 in gas phase, 148 of sulfide nanowires, 211-27 growth of nanowires, 227 effect of temperature, 227 termination, 252 growth of sulfide nanowires, 211-27 gas-solid, 218-27 polymer-controlled, 216 solvothermal synthesis. 211-12 tcmplatcd fabrication, 212-16 vapor-Iiquid-solid, 218 H 2S, see hydrogen sulfide H zS04 and nanotubes, 284-5 H zTi 30 7 , 159, 160, 162-3 electric and optical properties, 169-70 HAP nanorods, see hydroxylapatite nanorods Haliotis rufescens, 239-40 hard/soft template-assisted synthesis, 176 HCSA-PANI, see camphorsulfonic acid doped polyaniline heat transport, 11-l2

Index He-Cd laser. 29 helical nanostructures, 250-3, 271 hexagonal crystal structure, 35. 215, 248 liquid crystalline phase, 176 high-resolution TEM images, 51, 53, 67, 86-7, l20-1, 124, 131, 146, 152, 161, 164-5, 167, 178, 183, 196, 225, 259, 261 -2, 264 high-temperature oven system, 193 hornocondensation, 95 homojunctions,61-3 HREM images, see high-resolution TEM images HRTEM images, see high-resolution TEM images l-liickel equation, 99 hydrogen peroxide, 285 hydrogen sulfide and catalytic activity, 201-3 O 2 : HzS, 222-4, 227 oxidation, 201-3 reaction with CuzS, 219. 226 reaction with ZnO nanobelts, 57-8 hydrolysis, 95, 195 hydrothermal synthesis of nanostructures, 157 hydroxylapatite nanorods, 188

Il-, see inorganic fullerenes dichalcogcnides, 219 in situ transmission electron microscopy, 203-5 InZ03 nanobelts, 50, 52-3 indentation size effect. 75 see also nanoindentation infrared anisotropy, 270 modes, 129 spectroscopy, 247 inorganic fullercnes, 218-19 interfacial energy, 241 isotropic epitaxial growth, 61 dispersion, 114 ITO. 93 1-V curves. see current-voltagc curves I-Vhias curves, 275 K 2Ti60 13, see potassium hexatitanate K 2Tir,o l3- rto, interface, 168 KMnOcMnS04 reaction system, 182 LAHC, see laurylamine hydrochloride lanthanide hydroxide nanowires, 187 lasers. see nanolasers laser ablation method, 191. 258, 260 laser-assisted catalytic growth, 218 lasing, 31-5 threshold, 22

Index laurylamine hydrochloride, 194 layer structures (il-MnOzJ, 179 metal oxides. 159-60 LCG, see laser-assisted catalytic growth lead zireonate titanate nanorods, 93, 104-6, 175 lip-lip growth model. 260-1 liquid crystal, 175-6,211,215,271 matrix, 272 phase, 270 polymers, 277 liquid-vapour surface free energy, 28 lithographical techniques. II, 14, 27 Ln(OHh nanowircs, 187 LO modes, see longitudinal optical phonon modes longitudinal optical phonon modes. 230 phase transitions, 230 lyotropic liquid crystals, 176,215 MA, see maleic anhydride macromolecule. 269 maleic anhydride, 216 Matheson Tri-Gas. 40 matrix crystal. 272 oxide, 213 polymer, 211, 216-17, 228, 272 mechanical resonance. 12-13 MEMS, see mieroeleetro-mechanical systems mesoporous silica. 195 metal oxide based gas sensors, 9 see also gas sensors mctaloxidc catalysts. see catalysts semiconductor field effect transistor, 266 nanowircs, 134 metal oxide-semiconductor field effect transistor. 266 microelectro-mechanical systems. 14. 93 microemulsion system. 114-15, 117, 122. 129 compositions of. 117 W/O microemulsion. see water/oil y values, 122 rnicrotwins, 118 mineral bridging mechanism, 252 n-Mn02 intermediate, 181 Mn02 nunowires, 178-82 mole/molar ratio, 85, 118. 120,216,222,224, 227.265 and nanorods, 180 molecular wires. 273 molecule. 269 molten salt synthesis, 113. 130

293

monoclinic structures CU2S/CUO, 221, 223. 225, 228, 230. 233 GazO" 50, 54, 65 K2Ti,,013,126 Lu(OHh/ YbOOH,187 monoliths, 96, 99-100 MoO] nonohelts, 185-86 MOSFET. see metal oxide semiconductor field effect transistor mother of pearl, 240, 251-2 MSS, see molten salt synthesis multi-clcmcntal nanotubcs, 257 multi wall carbon nanotubes, 75. 258, 261-3 MWNTs, see multiwall carbon nanotubes nacreous layer, 240, 251-2 nanobelts. 1-17.73-5,185 CdO. 14,52,54 complex structures. 59-65 electrical properties of, 6-7 fractured. 15 Ga20,,54 heat transport through, 11-12 In203' 50, 52, 53 isothermal response, 10-11 manipulation of, 14 mechanical behavior of, 73-80 Mo0 3,185-6 as nanocantilevcrs. 13-17 as nanoresonators. 11-13 oxide. 49-56, 59, 67-8. 73 oxygen deficiency in. 7 oxygen sensor using, 8 PbO z,55 photoconductivity of. 8 response to CO, 10 semiconducting, 15-16 single, 1-8 single. 8, 73 Sn02.52 synthesis via solution-based route, 185 titanate. 186 ZnO, 1-17,49-52 ZnS. 56 nanocables. 264-6 struet ure 0 f. 266 nanoeantilevers, 13-17, 90 resonance frequency, 89-90 nanoclusters,96, 106, 114--16. 152, 157 nanodetectors and nanotips. 135 nanodeviees based on nanowires/belts, 3-46 nanodiskettes, 65-7 nanoeleetrode and nanowaveguides, 134

294 nanofibers, 133, 196, 270-1 bundles/brush-like, 196, 198 CdS, 216-17 conducting polymers, 270-1, 278, 280 electrospinning of, 276-9 helical, 271 polymer, 271-2, 276-86 s.o, 196-8, 200, 246 V 20 s, 133 wreath-like, 196, 198 nanofibcrbascd field-effect transistor, 134-5 nanoforks, 122-3 nanoindentation, 73-5 nanoindenter, II, 74 nanolasers, 31-5 diode, 22 He-Cd, 29 laser ablation, 191, 258, 260 Nd: YAG, 30 short- wavelength, 22 single, 34 sub-picosecond width, 34 ZnO nanowire as, 173 nanorncchanics, 73-80 nanoporosity, 240 nanoribbons, 144, 147 cross-section, 40 GazO], 54 as gas sensors, 42 growth, 61 junction arrays, 60 Matheson Tri-Gas. 40 photoresponse, 41 Sn02,39 ZnO, 4, 60-1,144-5,147 nanorods. 127, 149. 151, 175, 180,211--12 Bi 2S],212 CdS, 211-12, 214 CulnSz,212-13 MInS2,212 mole ratio and, 180 PZT. 175 Sn02- 127-30 T-ZnO, 149, 151 nanoscalc optoelectronic switches. 37-9 nanoscnsors, 40-1 nanoshects, 50, 65 CdO, 53, 65 Gap], 54, 65 TiO o,163 v.o; 184 nanosized metal clusters, 152 nanostructurcs, 196 electric and optical properties, 169-70 growth mechanism, 148-53

Index nanostructures om/d. optical properties, 153-4 oxides, 9; see also oxide nanostructures silica, 148, 153, 158, 169, 196-7 tadpole-like, 59-61 rio, 158-71 nanotubes, 157, 160-1, 173, 177, 184, 191,201, 231,281 bending, 75-8 carbon, 231 co-substituted, 177-9 of complex oxides, 173-88 gold, 233, 235 grafting. 281 H 2Ti]07, 160, 163, 169 honey-comb lattice, 79 MWNTs,75 prodding, 78 rolling mechanism, 184 scroll type, 161 semiconducting, 169 Si02- 191-205 SiO z-NiO,201-2 synthesis of, 191 SWNTs,80 titanate, 177-9 trititanate, 160-3, 166 v.o, 183-4 vox' 182-3 WS 2,185 ZnS, 57-9 nanowhiskcrs, 119, 121-3, 125-6, 144, 149 K 2Tio0 13, 125-7 T-ZnO.149 T-ZnO whiskers. 144 nanowires. 21-42, 83-91,113,130,151, 173, 178, 191,194,203-5,209,211-27,239,269 aluminium, 280, 282 arrays, 220 BaTiO] and SrTiO], 83-91 bending, 204 biaxial. 195 bicrystalline, 142-6 CdS, 216-19 coaxial, 195-6 composite, 199-200, 259-60, 265 continuous, 263 CU20. 130-3 CU2S,219-35 ferroelectric, 83-91 flexibility, 205 gallium droplet induced growth, 62-5 growth. 211-27 in medical and environmental health, 39 lanthanide hydroxide, 187

Index nanowircs contd. ｾＵ mechanical properties, ＲＰＳ MnO.178-82 of complex oxides, 173-88 of functional oxides. 113 polyaniline, 24'i polymer. 269-86 pristine. 91 Si 3N 4.203-5 single crystalline, 84. 87, 91 single free-standing, 164 single, 146 sulfide, 209-36 synthesis of large arrays, 239-53 synthesis of. 191 typical morphology, 194 ZnO. 22-42, lSI Nd : YAG laser, 30 n-doping, 270 near-field scanning optical microscope, 32-4 nematic liquid crystal phase, 270-1 1 (NH4hS20X, ＱＸＰｾ (NH4hS20g-MnS04, 181 NNLS, see non-negative linear square N0 2 photochemical sensing, 39 non-negative linear square method, 200 nonstoichiometry, 7 NSOM,32 rHype enhancement, 134- 'i oxide semiconductors. 130 nucleation and growth, 241 OrH2S, 222-4, 227 octadecylaminc, 183 octahedral multiple twin nucleus model. 149-50 octa-twin nucleus model, see octahedral multiple twin nucleus model Oliver-Pharr method. 75 one-dimensional oxide nanostructures. 47-9. 176. 182. 191 from artificial lamellar structures. 182 silica nanowires/tubes. 191-205 solution-based synthesis, 176 one-dimensional sulfide materials, 209-10 one-step alkali treatment, 166, 170 optical gating phenomenon, ＳＸｾＹ lithography technique, II, 14. 133 properties of nanostructures, 153-4 opto-electronic switches. 37-9 organic/inorganic precursors. 96, 182-3. 211-12,217

295

organic hydroxyl carboxylic acids, 19'i materials and nucleation, 239-40 oriented nanowircs, 239-53 orthorhombic structure, 50 oxide nanobelts, ＴＹｾＧｩＶＬ 59, 67-8, 73 CdO, 'i2 complex structures, 59-65 GazCh 54 grow th mechanism, 68-70 [n203' 50, 52, 53 mechanical behavior of, 73--80 Pb0 2,55 planar defects in. 67..8 SnOb 1--17,52,127..30 synthesis of, 21 ZnO,49-52 ZnS,56 oxide nanorods, 93-111, 175. 187 diameter of. 104 sol-gel processing, 94-7 surfactant-assisted solution-based method, 187 synthesis of, 100 template-assisted synthesis, 175-6 oxide nanostructurcs, 9, 169, 176 binary/ternary, 176 electric and optical properties, 169-70 one-dimensional, 47-9, 176, 182. 191 oxide nanowires, 23, 25, 27, 29, 32, 34-5, 37--9, 21-42,158-9,174-5,178 coupling, 34 ethanol-nanowire mixture, 32 in medical and environmental health. 39 layered oxides, 159-60 linear optical property. 35 manganese oxide, 178-82 optoelectronic switches, 37 -9 orientation control. 25 photoluminescence and lasing, 29-35 photosensitivity, 38 position control, 27 synthesis of, 23 template-assisted synthesis, 175-6 Ti0 2,158-71 vapor-transport method, 174-5 oxide-assisted growth, 148 oxide nanostructurcs, 50 semiconductors, 9, 38, 130, 134 synthesis conditions and morphology. 50 oxidizing reagents. 180 oxygen sensor, 8 oxygen vacancy, 154

296 PAN, see polyacrylonitrile composite nanowires parallel electron energy-loss spectroscopy, 258 partial density, 170 PbO z nanobclts, 55 PbS nanostrips, 218 PC, see polycarbonate p-doping, 270 PEELS, see parallel electron energy-loss spectroscopy PEG,130 PEO, see polyethylene oxide periodate oxidation, 281, 283 pcrovskitc structure, 83-5, 105 phase separation, 257-61 phase transformation growth model, 149 phonon energy, 169 photochemical sensing, 39 photoconductive oxide nanowires, 37-9 as UY-light detectors, 39 photoelectron spectrum, 132 photoluminescence, 32-5, 140 spectra, 30, 153-4,201,229 photopolymcrization. 272 physical vapor deposition, 174, 192 PL, see photoluminescence planar defects, 67 platelets, 167, ] 85, 249 platinum interdigitated electrode structure, 9 PMMA, see polymcthyl methacrylate pn junctions, 42 p-octyl-polyethylcne glycol phenylether, 129 polarization, 35-7 ratio, 35-6 polarized infrared spectroscopy, see infrared spectroscopy poly meta-phenylene isophthalamide, 280-2 poly p-phenylene vinylene, 272 polyacctylcnc. 270-1 polyacrylonitrile nanofiber, 280 composite nanowires, 216 polyaniline, 273-5, 278 nanowire bridge, 274 nanowires. 245 spectra, 247 polycarbonate membrane, 101-2 polyethylene glycol, ]30 oxide, 278 polymer, 269 matrix, 211, 2]6-17, 228, 272

Index polymer nanowires/ libel'S, 271 -3, 276 electrospinning, 276-9 synthesis at a scanning microscope tip, 273 template synthesis, 27]-3 with special architectures, 279-86 polymer-controlled growth, 2]6 polymer-encapsulated nanowires, 217 polymeric sol, 105 stabilization. 96 polymer-wrapped sulfide nanowires, 229 polymethyl methacrylate, 4-5 polypyrrole, 240, 272-3, 280, 282, 285 polystyrene, 217 polyvinyl butyral films, 217-18 format, 204 potassium hexatitanate, 158-60. 164-70 crystal structure, 126, 168. 177 crystallographic relationships, ] 68 electric and optical properties. 169-70 image, ]25-6, ]65, ]68 nanowhisker, ] 25-7 nanowires. 159-60. 164-70 partial density and, 170 step 3 structure, 160 PPY, see poly p-phenylene vinylene Pr(OHh nanowires, 187 prodding, 78 precursors bimetallic alkoxide, 84-5 calcining, 129 condensation. 95 inorganic, 182-·3 powder for nnowircs, I 18, 120 preparation of, 122, 127 sol preparation, 100 solubility control. 241 thermal decomposition, 140 pristine nunowires, 91 PS. see polystyrene p-Iype oxide semiconductors. !30 Pt counter electrode/mesh, 10 I, 273 PYB, see polyvinyl butyral films PYD, see chemical vapor deposition pyrolysis reaction, 185 pyrrolc, 273, 284-5 polymer, see polypyrrole PZT nanorods. see lead zirconate titanate sol, 10] Quantachromc Autosorb-I, ] 59 quantum cable, 266

Index quantum contd. conductance, 76 confinement, 9, 216. 218 dots. 209-10 effect, II, 22 wires. 21 quartz. 23, 32, 34. 48,191-2,197 quasi-I D, xi, 36 quasi-layered structures, 160, 170 quenching, 272 radio-frequency glow-discharge plasma treatment. 280 Raman bands/li ncs. 129 scattering. 88 spectra. 115-17. 129.230 redox reactions, 180 resonance frequencies fundamental, 12, 15.31. 75, 78 cantilever. 88-90 reverse micelles method, 174. 187 rod-based CdSe nanocrystals, 140 room temperature photochemical sensors, 39 rutile crystal structure, 52-3.114-15,118,179 rutile nanorods. 119-22. 127 flowsbeet of synthesis. 119 stannic oxide, 127-30 y values to corresponding rod size, 122 SADPs. see selected area electron diffraction patterns SAED. see selected-area electron diffraction sapphire, 25-6, 28. 31 scanning

electrochemical microscope. 273 electron microscopy, see SEM force microscopy, 133 probe microscopy, 13-14,87-91 transmission electron microscope, 258 tunneling microscopy. 273. 278 seem, SCI' standard cuhic centimeter per minute Schottky junctions, 231 scroll-type nanotuhe, 161 SDS. see sodium dodccyl sulfate SHAD pattern, 212 SECM, see scanning electrochemical microscope second harmonic generation, 35 seeded growth, 242 seed crystal, 249 selected-area electron diffraction. 128, 141,263 patterns, 118, 123 SEM, 24. 51, 53, 57, 60, 64. 66,86,102. 144--5, 147,219-20.243,245.248.249,252, 271-2.284

297

semiconducting nanohclt, 15-16 nanowires. 213 oxide nanobelts, 50, 175 sensory devices. 234-5 SFM, see scanning force microscopy SHG, see second harmonic generation short-wavelength semiconductor lasers, 22 Si 1N4 nanowire, 203 5 f3-SiC-SiO" 196 SiC-SiOz nanowirc, 265 SiC-SiOz-(BN),C y , nanocahles, 264-5 SiC-SiO, nanowires, 78 signal ratio, 41-2 silica aerogels. 201--3 nanofibers, 196-8. 200, 246 nanostructures, 196-·7 silica nanowires/tuhes. 191-205 catalytic activity of nanotuhes, 201-3 comet-like wires, 197-8 Ga-Si alloy, 197·8 structures and properties of, 197-205 synthesis of, 191-7 valences of Si and 0. 199 silica-silicon carhide nanowires. 195 silicon chip, 15-16 simple evaporation method, 175 single alkali treatment, 166. 170 molecule probe, 234-5 single nanohclts. 8. 73 nanomcchanical behavior of, 73-5 oxygen sensor using, 8 single nanowircs, 91. 116 SaTiO] and s-no, 84-7. 91 rto, wires, 116-22 single-walled carhon nanotubes, 79-80 sintering, 96-7. 100, 203-5 SiOz nanowircs/tubes/tibers, see silica SiOrNiO nanotuhcs. 201-2 Si(OC zHj )4, see tetraethoxysilicane slip plane, 98 SLS, see solution-liquid-solid Sm(OHh nanowires, 187 SnO diskettes, 66-7 Sn02 doped In203, see ITO

seo, nanobclts, 1-17.52, 127-30 nanorods, 127--30 sodium bis(2-ethylhexyl)sulfosuccinate. 175, 217 sodium cyanoborohydride, 281 sodium dodccyl sulfate. 188 SOFC, see solid oxide fuel cells soft-template method. 175, 187

298 sol,96, 100-1, 105, 109, 133 RaTi0 3,100 electrophoretic deposition, 97-100 ionic resistivity, 109 polymeric, 105 PZT 101 Sr2Nb207, 100 rto, 100 Y 20\,133 solar cell materials, 135 sol-gel, 193-5.266 in coaxial cables. 266 sol-gel electrophoresis, 110, 175 current-voltage relationship, 110 deposition, 100 modified version, 110-11 sol-gel processing, 94-7, 100 solid oxide fuel cells. 93 solid-liquid-solid, 174 solution-based synthesis, 176-7 solution-liquid-solid, 113,212 solvothermal synthesis, 211-12 source-drain current,S, 7-8 versus gate bias after various treatments, 7 versus gatc bias in ambient, 5 versus time, 8 SPM, see scanning probe microscopy Sr2Nb207 sol, 100 St, see styrene stabilization, 96-7, 99 stacking faults, 149 standard cubic centimeter per minute, 41, 192 stannic oxide, see Sn02 and InzO] films, 93 STEM, see scanning transmission electron microscope STEM-EELS elemental analysis, 263 STEM-PEELS system, 258 Stern layer, 98 STM, see scanning tunneling microscopy strain energy relaxation, 149 strontium titanate. 84-5, 87,91 synthesis of. 85-7 styrene. 216 sulfide nanowires. 209-36 growth,211-27 optical properties, 229-30 phase transitions, 230 polymer-wrapped, 229 potential applications, 231-5 properties of, 227-31 structures and morphologies, 227, 228 transport, 230 sulfide. polymer and composite nanowires, 209-86 supersaturation, 153, 24J

Index surface area-volume, 9, 29, 39. 84,96, 277 oxygcn desorption, 7 poisoning, 252 surfactant-assisted method, 187 surfactants, 194 switching ratio. 6, 16 SWNTs, see single-walled carbon nanotubes synthesis method, 47 synthesis of silica nanowires/tuhes, 191-3 CYD and PYD evaporation methods, 192 high-temperature oven system, 193 laser ablation method, 191,258,260 sol-gel and other chemical methods, 193-5 TAA, see thioacetamide tadpole-like nanostructure, 59-61 TCO,49 TEM images, 3, 25, 51, 54-5, 58-9, 64-5, 67-8, 86-7,103,117,120,123,125-8,131,143 , 146,150-2,159,161,164,166,167. UW-7, 192,212-15,217-18,220,222,225,234-5, 244,260.263,271,281-2; see also transmission electron microscopy temperature and nanowire, 227 template-assisted synthesis, 194-5, 271-3 carbon nanotuhes, 194 CFCs, 194 surfactants (LAHC), 194 template-confined method. 175, 176 templated fabrication, 212--16 template-free polymerization, 273 TEOS. 193 TEOS-CA,195 TEOS-LAHC, 194-5 tetraethoxysilicane (Si(OC 2 H\)4), 193-5 tetrapod-like ZnO, see T-ZnO thermal evaporation technique, 56 as templates in synthesis of new materials, 56-9 various applications, 3, 49. 52, 140, 148, 185, 197,218 thermal uunsport/conductancc. see heat transport THG, see third harmonic generation thin/thick films, 9, 38 gas sensors, 9 oxygen chemisorption and photosensitivity, 38 thioacetamide, 213 thiophene, 273 third harmonic generation, 35 tin oxide doped indium oxide films, see stannic oxide Ti02 anatase. 107-8, 114, 158 as catalyst, 135 nanobelts, 186

Index TiOicontd. nanofork, 123 nanorods. J() I ｾ 5 nanotubes. 177 ｾＹ nanowires/clusters, Ｑ ＴｾＲＵ soL 100 no, nanostructures, 158. 165. 169 electric and optical properties. ＱＶＹｾ 70 growth mechanism. 165 synthesis. 158-9 Ti0 6 octahedra. 159-60, 170 titanate. see TiO z titania.114 crystal structures, 114 electrolyte solution. I J() nanoclusters, 116 titanium oxide. see TiO z transmission electron microscopy. 3-4. 11.25.75 images. see TEM images in bending nonotubes, 75-8 transparent conducting oxide, 49 transport ballistic, 80, 158 fluid. 226 heat, see heat transport properties of sulfide, 230 surface, 226 vapor, 23, 48, 148, 174 trap-state emission, 154 trititanate nanorubes, 160, 166 twin. 68 boundary. 142-4 images, 124. 146 multiple. 149-51 whiskers, 123, 125 two probe method, 278 T-ZnO nanorods, 140-5. 149, 151, 153 photoluminescence. 153-4 whiskers, 144 ultrahigh vacuum (UHY). 89 ultraviolet. see UV uniform hexagonal prismatic growth. 152 UY emission. 153-54 light-induced desorption, 8, 16 light detector», 39 light irradiation, 8 visible absorption spectra, 169

Y20S nanofibres. 133 nanosheets, 184 nanotubes, 133-4 sol. 133

299

vacuum pyrolysis. 185 van der Waul's forces, 176 vanadium oxide, see V205 vanadium triisopropoxide, 183 vapor deposition, 79. 192, 239. 280 vapor transport and condensation, 22-3. 174-5 vapor phase evaporation method, see vapor evaporation vapor-Iiquid-solid, 22, 49. 56, 60, 70, 94, 113, 148. 192-3, 218, 225 crystal growth mechanism, 23 epitax y, 25-7 nanowire growth mechanism, 27 vapor- solid, 70, 148,151-2.193 YLS, see vapor-Iiquid-solid VLSE, vapor-Iiquid-solid epitaxy volt -amperometric technique, 10 VOx nanotubcs, 183 VOx-surfactant, 177 YS, see vapor-solid Wang Z.L., 75-6, 78 Wannier exciton. 169 water/oil (W/O) microemulsions, 129 width-thickness ratio, 3, 21, 50--4, 73, 245 wet-chemical method. 262 wide band-gap semiconductors, 22. 242 W0 3 . x nanorods, 177, 182, 185 WO x nanowircs, 184 wreath-like, 196, 198 writing. 88-90 wurtzite structure CdS.219 T-ZnO,149 ZnO/ZnS, 25, 49, 50, 56-8, 142-3. 148, 152,218 XPS analysis, 199 spectrum. 228 X-ray diffraction. see XRD X-ray dilTractogram, 86 XRD patterns. 24, 128, 131, 141, 147. 179, 181, 183, 199-200,219,221,223.244 profile, 126 spectra, 103, 105, 121 Young's modulus, 75-6. 78 zeolite, 272-3 zero-gate bias, 8 zeta-potential, 98-9, 106 Ziegler-Narta polymerization, 270-1 zinc blonde. 218

300 zinc oxide, see ZnO zinc sclenide disk, 36 zirconia, 93 zirconium propoxide, lOO-1 Zn-Au alloy, 23 Zn-Cd,216 ZnO arrays, 145-8, 152-3 homojunctions, 61--3 nanobelts, 1-17, 49-52 nanoribbons, 147 reaction with nanobelts, 57-8 ZnO nanostruetures, 139-40, 148, 153,243-53 controlled growth and optical properties, 139-54 growth mechanism, 148 optical properties, 153-4 synthesis by evaporation, 140 ZnO nanowires, 25 bicrystalline image, 146 hicrystalline nanowircs, 142-6, 154

Index ZnO nanowires contd. bicrystalline synthesis, 142-5 controlled growth of, 25 diameter control, 28 direction of growth, 25 morphological control, 29 nonlinear optical mixing, 35 photoluminescence, 153--4 photoresponse of, 38 reversible switching, 38 SEM images of, 24, 27 single crystalline, 146 TEM image of, 22-42, 142, 146.151-3,243-53 ZnO thin films, 35 ZnO-ZnO, homojunctions, 63 ZnO-ZnS nanocable structure, 58 ZnS cuhic zinc blend, 56, 58 nanobelts. 56 nanotuhes, 57-9 wurtzite structure. 50, 56-7